No OneTemporary
Actions

Size

7 MB

Referenced Files

None

Subscribers

None

View Options

This file is larger than 256 KB, so syntax highlighting was skipped.

	diff --git a/sys/contrib/openzfs/.github/workflows/zfs-tests-functional.yml b/sys/contrib/openzfs/.github/workflows/zfs-tests-functional.yml
	index aad3d552b236..2987cdac6d64 100644
	--- a/sys/contrib/openzfs/.github/workflows/zfs-tests-functional.yml
	+++ b/sys/contrib/openzfs/.github/workflows/zfs-tests-functional.yml
	@@ -1,82 +1,82 @@
	name: zfs-tests-functional

	on:
	push:
	pull_request:

	jobs:
	tests-functional-ubuntu:
	strategy:
	fail-fast: false
	matrix:
	os: [18.04, 20.04]
	runs-on: ubuntu-${{ matrix.os }}
	steps:
	- uses: actions/checkout@v2
	with:
	ref: ${{ github.event.pull_request.head.sha }}
	- name: Install dependencies
	run: \|
	sudo apt-get update
	sudo apt-get install --yes -qq build-essential autoconf libtool gdb lcov \
	git alien fakeroot wget curl bc fio acl \
	sysstat mdadm lsscsi parted gdebi attr dbench watchdog ksh \
	nfs-kernel-server samba rng-tools xz-utils \
	zlib1g-dev uuid-dev libblkid-dev libselinux-dev \
	xfslibs-dev libattr1-dev libacl1-dev libudev-dev libdevmapper-dev \
	libssl-dev libffi-dev libaio-dev libelf-dev libmount-dev \
	libpam0g-dev pamtester python-dev python-setuptools python-cffi \
	python-packaging python3 python3-dev python3-setuptools python3-cffi \
	libcurl4-openssl-dev python3-packaging
	- name: Autogen.sh
	run: \|
	sh autogen.sh
	- name: Configure
	run: \|
	./configure --enable-debug --enable-debuginfo
	- name: Make
	run: \|
	make --no-print-directory -s pkg-utils pkg-kmod
	- name: Install
	run: \|
	sudo dpkg -i *.deb
	# Update order of directories to search for modules, otherwise
	# Ubuntu will load kernel-shipped ones.
	sudo sed -i.bak 's/updates/extra updates/' /etc/depmod.d/ubuntu.conf
	sudo depmod
	sudo modprobe zfs
	# Workaround for cloud-init bug
	# see https://github.com/openzfs/zfs/issues/12644
	FILE=/lib/udev/rules.d/10-cloud-init-hook-hotplug.rules
	if [ -r "${FILE}" ]; then
	HASH=$(md5sum "${FILE}" \| awk '{ print $1 }')
	if [ "${HASH}" = "121ff0ef1936cd2ef65aec0458a35772" ]; then
	# Just shove a zd* exclusion right above the hotplug hook...
	sudo sed -i -e s/'LABEL="cloudinit_hook"'/'KERNEL=="zd*", GOTO="cloudinit_end"\n&'/ "${FILE}"
	sudo udevadm control --reload-rules
	fi
	fi
	# Workaround to provide additional free space for testing.
	# https://github.com/actions/virtual-environments/issues/2840
	sudo rm -rf /usr/share/dotnet
	sudo rm -rf /opt/ghc
	sudo rm -rf "/usr/local/share/boost"
	sudo rm -rf "$AGENT_TOOLSDIRECTORY"
	- name: Tests
	run: \|
	- /usr/share/zfs/zfs-tests.sh -v -s 3G
	+ /usr/share/zfs/zfs-tests.sh -vR -s 3G
	- name: Prepare artifacts
	if: failure()
	run: \|
	RESULTS_PATH=$(readlink -f /var/tmp/test_results/current)
	sudo dmesg > $RESULTS_PATH/dmesg
	sudo cp /var/log/syslog $RESULTS_PATH/
	sudo chmod +r $RESULTS_PATH/*
	# Replace ':' in dir names, actions/upload-artifact doesn't support it
	- for f in $(find $RESULTS_PATH -name ':'); do mv "$f" "${f//:/__}"; done
	+ for f in $(find /var/tmp/test_results -name ':'); do mv "$f" "${f//:/__}"; done
	- uses: actions/upload-artifact@v2
	if: failure()
	with:
	name: Test logs Ubuntu-${{ matrix.os }}
	path: /var/tmp/test_results/20*/
	if-no-files-found: ignore
	diff --git a/sys/contrib/openzfs/.github/workflows/zfs-tests-sanity.yml b/sys/contrib/openzfs/.github/workflows/zfs-tests-sanity.yml
	index 4df49461ed97..8cba6d78dde2 100644
	--- a/sys/contrib/openzfs/.github/workflows/zfs-tests-sanity.yml
	+++ b/sys/contrib/openzfs/.github/workflows/zfs-tests-sanity.yml
	@@ -1,78 +1,78 @@
	name: zfs-tests-sanity

	on:
	push:
	pull_request:

	jobs:
	tests:
	runs-on: ubuntu-latest
	steps:
	- uses: actions/checkout@v2
	with:
	ref: ${{ github.event.pull_request.head.sha }}
	- name: Install dependencies
	run: \|
	sudo apt-get update
	sudo apt-get install --yes -qq build-essential autoconf libtool gdb lcov \
	git alien fakeroot wget curl bc fio acl \
	sysstat mdadm lsscsi parted gdebi attr dbench watchdog ksh \
	nfs-kernel-server samba rng-tools xz-utils \
	zlib1g-dev uuid-dev libblkid-dev libselinux-dev \
	xfslibs-dev libattr1-dev libacl1-dev libudev-dev libdevmapper-dev \
	libssl-dev libffi-dev libaio-dev libelf-dev libmount-dev \
	libpam0g-dev pamtester python-dev python-setuptools python-cffi \
	python-packaging python3 python3-dev python3-setuptools python3-cffi \
	python3-packaging libcurl4-openssl-dev
	- name: Autogen.sh
	run: \|
	sh autogen.sh
	- name: Configure
	run: \|
	./configure --enable-debug --enable-debuginfo
	- name: Make
	run: \|
	make --no-print-directory -s pkg-utils pkg-kmod
	- name: Install
	run: \|
	sudo dpkg -i *.deb
	# Update order of directories to search for modules, otherwise
	# Ubuntu will load kernel-shipped ones.
	sudo sed -i.bak 's/updates/extra updates/' /etc/depmod.d/ubuntu.conf
	sudo depmod
	sudo modprobe zfs
	# Workaround for cloud-init bug
	# see https://github.com/openzfs/zfs/issues/12644
	FILE=/lib/udev/rules.d/10-cloud-init-hook-hotplug.rules
	if [ -r "${FILE}" ]; then
	HASH=$(md5sum "${FILE}" \| awk '{ print $1 }')
	if [ "${HASH}" = "121ff0ef1936cd2ef65aec0458a35772" ]; then
	# Just shove a zd* exclusion right above the hotplug hook...
	sudo sed -i -e s/'LABEL="cloudinit_hook"'/'KERNEL=="zd*", GOTO="cloudinit_end"\n&'/ "${FILE}"
	sudo udevadm control --reload-rules
	fi
	fi
	# Workaround to provide additional free space for testing.
	# https://github.com/actions/virtual-environments/issues/2840
	sudo rm -rf /usr/share/dotnet
	sudo rm -rf /opt/ghc
	sudo rm -rf "/usr/local/share/boost"
	sudo rm -rf "$AGENT_TOOLSDIRECTORY"
	- name: Tests
	run: \|
	- /usr/share/zfs/zfs-tests.sh -v -s 3G -r sanity
	+ /usr/share/zfs/zfs-tests.sh -vR -s 3G -r sanity
	- name: Prepare artifacts
	if: failure()
	run: \|
	RESULTS_PATH=$(readlink -f /var/tmp/test_results/current)
	sudo dmesg > $RESULTS_PATH/dmesg
	sudo cp /var/log/syslog $RESULTS_PATH/
	sudo chmod +r $RESULTS_PATH/*
	# Replace ':' in dir names, actions/upload-artifact doesn't support it
	- for f in $(find $RESULTS_PATH -name ':'); do mv "$f" "${f//:/__}"; done
	+ for f in $(find /var/tmp/test_results -name ':'); do mv "$f" "${f//:/__}"; done
	- uses: actions/upload-artifact@v2
	if: failure()
	with:
	name: Test logs
	path: /var/tmp/test_results/20*/
	if-no-files-found: ignore
	diff --git a/sys/contrib/openzfs/META b/sys/contrib/openzfs/META
	index 8dacb8082f05..aab6e2abe567 100644
	--- a/sys/contrib/openzfs/META
	+++ b/sys/contrib/openzfs/META
	@@ -1,10 +1,10 @@
	Meta: 1
	Name: zfs
	Branch: 1.0
	Version: 2.1.99
	Release: 1
	Release-Tags: relext
	License: CDDL
	Author: OpenZFS
	-Linux-Maximum: 5.14
	+Linux-Maximum: 5.15
	Linux-Minimum: 3.10
	diff --git a/sys/contrib/openzfs/cmd/zdb/zdb.c b/sys/contrib/openzfs/cmd/zdb/zdb.c
	index dff61b7bae8b..8fe926c50e85 100644
	--- a/sys/contrib/openzfs/cmd/zdb/zdb.c
	+++ b/sys/contrib/openzfs/cmd/zdb/zdb.c
	@@ -1,8878 +1,8880 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2019 by Delphix. All rights reserved.
	* Copyright (c) 2014 Integros [integros.com]
	* Copyright 2016 Nexenta Systems, Inc.
	* Copyright (c) 2017, 2018 Lawrence Livermore National Security, LLC.
	* Copyright (c) 2015, 2017, Intel Corporation.
	* Copyright (c) 2020 Datto Inc.
	* Copyright (c) 2020, The FreeBSD Foundation [1]
	*
	* [1] Portions of this software were developed by Allan Jude
	* under sponsorship from the FreeBSD Foundation.
	* Copyright (c) 2021 Allan Jude
	* Copyright (c) 2021 Toomas Soome <tsoome@me.com>
	*/

	#include <stdio.h>
	#include <unistd.h>
	#include <stdlib.h>
	#include <ctype.h>
	#include <sys/zfs_context.h>
	#include <sys/spa.h>
	#include <sys/spa_impl.h>
	#include <sys/dmu.h>
	#include <sys/zap.h>
	#include <sys/fs/zfs.h>
	#include <sys/zfs_znode.h>
	#include <sys/zfs_sa.h>
	#include <sys/sa.h>
	#include <sys/sa_impl.h>
	#include <sys/vdev.h>
	#include <sys/vdev_impl.h>
	#include <sys/metaslab_impl.h>
	#include <sys/dmu_objset.h>
	#include <sys/dsl_dir.h>
	#include <sys/dsl_dataset.h>
	#include <sys/dsl_pool.h>
	#include <sys/dsl_bookmark.h>
	#include <sys/dbuf.h>
	#include <sys/zil.h>
	#include <sys/zil_impl.h>
	#include <sys/stat.h>
	#include <sys/resource.h>
	#include <sys/dmu_send.h>
	#include <sys/dmu_traverse.h>
	#include <sys/zio_checksum.h>
	#include <sys/zio_compress.h>
	#include <sys/zfs_fuid.h>
	#include <sys/arc.h>
	#include <sys/arc_impl.h>
	#include <sys/ddt.h>
	#include <sys/zfeature.h>
	#include <sys/abd.h>
	#include <sys/blkptr.h>
	#include <sys/dsl_crypt.h>
	#include <sys/dsl_scan.h>
	#include <sys/btree.h>
	#include <zfs_comutil.h>
	#include <sys/zstd/zstd.h>

	#include <libnvpair.h>
	#include <libzutil.h>

	#include "zdb.h"

	#define ZDB_COMPRESS_NAME(idx) ((idx) < ZIO_COMPRESS_FUNCTIONS ? \
	zio_compress_table[(idx)].ci_name : "UNKNOWN")
	#define ZDB_CHECKSUM_NAME(idx) ((idx) < ZIO_CHECKSUM_FUNCTIONS ? \
	zio_checksum_table[(idx)].ci_name : "UNKNOWN")
	#define ZDB_OT_TYPE(idx) ((idx) < DMU_OT_NUMTYPES ? (idx) : \
	(idx) == DMU_OTN_ZAP_DATA \|\| (idx) == DMU_OTN_ZAP_METADATA ? \
	DMU_OT_ZAP_OTHER : \
	(idx) == DMU_OTN_UINT64_DATA \|\| (idx) == DMU_OTN_UINT64_METADATA ? \
	DMU_OT_UINT64_OTHER : DMU_OT_NUMTYPES)

	/* Some platforms require part of inode IDs to be remapped */
	#ifdef __APPLE__
	#define ZDB_MAP_OBJECT_ID(obj) INO_XNUTOZFS(obj, 2)
	#else
	#define ZDB_MAP_OBJECT_ID(obj) (obj)
	#endif

	static char *
	zdb_ot_name(dmu_object_type_t type)
	{
	if (type < DMU_OT_NUMTYPES)
	return (dmu_ot[type].ot_name);
	else if ((type & DMU_OT_NEWTYPE) &&
	((type & DMU_OT_BYTESWAP_MASK) < DMU_BSWAP_NUMFUNCS))
	return (dmu_ot_byteswap[type & DMU_OT_BYTESWAP_MASK].ob_name);
	else
	return ("UNKNOWN");
	}

	extern int reference_tracking_enable;
	extern int zfs_recover;
	extern unsigned long zfs_arc_meta_min, zfs_arc_meta_limit;
	extern int zfs_vdev_async_read_max_active;
	extern boolean_t spa_load_verify_dryrun;
	extern boolean_t spa_mode_readable_spacemaps;
	extern int zfs_reconstruct_indirect_combinations_max;
	extern int zfs_btree_verify_intensity;

	static const char cmdname[] = "zdb";
	uint8_t dump_opt[256];

	typedef void object_viewer_t(objset_t , uint64_t, void data, size_t size);

	uint64_t *zopt_metaslab = NULL;
	static unsigned zopt_metaslab_args = 0;

	typedef struct zopt_object_range {
	uint64_t zor_obj_start;
	uint64_t zor_obj_end;
	uint64_t zor_flags;
	} zopt_object_range_t;
	zopt_object_range_t *zopt_object_ranges = NULL;
	static unsigned zopt_object_args = 0;

	static int flagbits[256];

	#define ZOR_FLAG_PLAIN_FILE 0x0001
	#define ZOR_FLAG_DIRECTORY 0x0002
	#define ZOR_FLAG_SPACE_MAP 0x0004
	#define ZOR_FLAG_ZAP 0x0008
	#define ZOR_FLAG_ALL_TYPES -1
	#define ZOR_SUPPORTED_FLAGS (ZOR_FLAG_PLAIN_FILE \| \
	ZOR_FLAG_DIRECTORY \| \
	ZOR_FLAG_SPACE_MAP \| \
	ZOR_FLAG_ZAP)

	#define ZDB_FLAG_CHECKSUM 0x0001
	#define ZDB_FLAG_DECOMPRESS 0x0002
	#define ZDB_FLAG_BSWAP 0x0004
	#define ZDB_FLAG_GBH 0x0008
	#define ZDB_FLAG_INDIRECT 0x0010
	#define ZDB_FLAG_RAW 0x0020
	#define ZDB_FLAG_PRINT_BLKPTR 0x0040
	#define ZDB_FLAG_VERBOSE 0x0080

	uint64_t max_inflight_bytes = 256 * 1024 * 1024; /* 256MB */
	static int leaked_objects = 0;
	static range_tree_t *mos_refd_objs;

	static void snprintf_blkptr_compact(char , size_t, const blkptr_t ,
	boolean_t);
	static void mos_obj_refd(uint64_t);
	static void mos_obj_refd_multiple(uint64_t);
	static int dump_bpobj_cb(void arg, const blkptr_t bp, boolean_t free,
	dmu_tx_t *tx);

	typedef struct sublivelist_verify {
	/* FREE's that haven't yet matched to an ALLOC, in one sub-livelist */
	zfs_btree_t sv_pair;

	/* ALLOC's without a matching FREE, accumulates across sub-livelists */
	zfs_btree_t sv_leftover;
	} sublivelist_verify_t;

	static int
	livelist_compare(const void larg, const void rarg)
	{
	const blkptr_t *l = larg;
	const blkptr_t *r = rarg;

	/* Sort them according to dva[0] */
	uint64_t l_dva0_vdev, r_dva0_vdev;
	l_dva0_vdev = DVA_GET_VDEV(&l->blk_dva[0]);
	r_dva0_vdev = DVA_GET_VDEV(&r->blk_dva[0]);
	if (l_dva0_vdev < r_dva0_vdev)
	return (-1);
	else if (l_dva0_vdev > r_dva0_vdev)
	return (+1);

	/* if vdevs are equal, sort by offsets. */
	uint64_t l_dva0_offset;
	uint64_t r_dva0_offset;
	l_dva0_offset = DVA_GET_OFFSET(&l->blk_dva[0]);
	r_dva0_offset = DVA_GET_OFFSET(&r->blk_dva[0]);
	if (l_dva0_offset < r_dva0_offset) {
	return (-1);
	} else if (l_dva0_offset > r_dva0_offset) {
	return (+1);
	}

	/*
	* Since we're storing blkptrs without cancelling FREE/ALLOC pairs,
	* it's possible the offsets are equal. In that case, sort by txg
	*/
	if (l->blk_birth < r->blk_birth) {
	return (-1);
	} else if (l->blk_birth > r->blk_birth) {
	return (+1);
	}
	return (0);
	}

	typedef struct sublivelist_verify_block {
	dva_t svb_dva;

	/*
	* We need this to check if the block marked as allocated
	* in the livelist was freed (and potentially reallocated)
	* in the metaslab spacemaps at a later TXG.
	*/
	uint64_t svb_allocated_txg;
	} sublivelist_verify_block_t;

	static void zdb_print_blkptr(const blkptr_t *bp, int flags);

	typedef struct sublivelist_verify_block_refcnt {
	/* block pointer entry in livelist being verified */
	blkptr_t svbr_blk;

	/*
	* Refcount gets incremented to 1 when we encounter the first
	* FREE entry for the svfbr block pointer and a node for it
	* is created in our ZDB verification/tracking metadata.
	*
	* As we encounter more FREE entries we increment this counter
	* and similarly decrement it whenever we find the respective
	* ALLOC entries for this block.
	*
	* When the refcount gets to 0 it means that all the FREE and
	* ALLOC entries of this block have paired up and we no longer
	* need to track it in our verification logic (e.g. the node
	* containing this struct in our verification data structure
	* should be freed).
	*
	* [refer to sublivelist_verify_blkptr() for the actual code]
	*/
	uint32_t svbr_refcnt;
	} sublivelist_verify_block_refcnt_t;

	static int
	sublivelist_block_refcnt_compare(const void larg, const void rarg)
	{
	const sublivelist_verify_block_refcnt_t *l = larg;
	const sublivelist_verify_block_refcnt_t *r = rarg;
	return (livelist_compare(&l->svbr_blk, &r->svbr_blk));
	}

	static int
	sublivelist_verify_blkptr(void arg, const blkptr_t bp, boolean_t free,
	dmu_tx_t *tx)
	{
	ASSERT3P(tx, ==, NULL);
	struct sublivelist_verify *sv = arg;
	sublivelist_verify_block_refcnt_t current = {
	.svbr_blk = *bp,

	/*
	* Start with 1 in case this is the first free entry.
	* This field is not used for our B-Tree comparisons
	* anyway.
	*/
	.svbr_refcnt = 1,
	};

	zfs_btree_index_t where;
	sublivelist_verify_block_refcnt_t *pair =
	zfs_btree_find(&sv->sv_pair, &current, &where);
	if (free) {
	if (pair == NULL) {
	/* first free entry for this block pointer */
	zfs_btree_add(&sv->sv_pair, &current);
	} else {
	pair->svbr_refcnt++;
	}
	} else {
	if (pair == NULL) {
	/* block that is currently marked as allocated */
	for (int i = 0; i < SPA_DVAS_PER_BP; i++) {
	if (DVA_IS_EMPTY(&bp->blk_dva[i]))
	break;
	sublivelist_verify_block_t svb = {
	.svb_dva = bp->blk_dva[i],
	.svb_allocated_txg = bp->blk_birth
	};

	if (zfs_btree_find(&sv->sv_leftover, &svb,
	&where) == NULL) {
	zfs_btree_add_idx(&sv->sv_leftover,
	&svb, &where);
	}
	}
	} else {
	/* alloc matches a free entry */
	pair->svbr_refcnt--;
	if (pair->svbr_refcnt == 0) {
	/* all allocs and frees have been matched */
	zfs_btree_remove_idx(&sv->sv_pair, &where);
	}
	}
	}

	return (0);
	}

	static int
	sublivelist_verify_func(void args, dsl_deadlist_entry_t dle)
	{
	int err;
	struct sublivelist_verify *sv = args;

	zfs_btree_create(&sv->sv_pair, sublivelist_block_refcnt_compare,
	sizeof (sublivelist_verify_block_refcnt_t));

	err = bpobj_iterate_nofree(&dle->dle_bpobj, sublivelist_verify_blkptr,
	sv, NULL);

	sublivelist_verify_block_refcnt_t *e;
	zfs_btree_index_t *cookie = NULL;
	while ((e = zfs_btree_destroy_nodes(&sv->sv_pair, &cookie)) != NULL) {
	char blkbuf[BP_SPRINTF_LEN];
	snprintf_blkptr_compact(blkbuf, sizeof (blkbuf),
	&e->svbr_blk, B_TRUE);
	(void) printf("\tERROR: %d unmatched FREE(s): %s\n",
	e->svbr_refcnt, blkbuf);
	}
	zfs_btree_destroy(&sv->sv_pair);

	return (err);
	}

	static int
	livelist_block_compare(const void larg, const void rarg)
	{
	const sublivelist_verify_block_t *l = larg;
	const sublivelist_verify_block_t *r = rarg;

	if (DVA_GET_VDEV(&l->svb_dva) < DVA_GET_VDEV(&r->svb_dva))
	return (-1);
	else if (DVA_GET_VDEV(&l->svb_dva) > DVA_GET_VDEV(&r->svb_dva))
	return (+1);

	if (DVA_GET_OFFSET(&l->svb_dva) < DVA_GET_OFFSET(&r->svb_dva))
	return (-1);
	else if (DVA_GET_OFFSET(&l->svb_dva) > DVA_GET_OFFSET(&r->svb_dva))
	return (+1);

	if (DVA_GET_ASIZE(&l->svb_dva) < DVA_GET_ASIZE(&r->svb_dva))
	return (-1);
	else if (DVA_GET_ASIZE(&l->svb_dva) > DVA_GET_ASIZE(&r->svb_dva))
	return (+1);

	return (0);
	}

	/*
	* Check for errors in a livelist while tracking all unfreed ALLOCs in the
	* sublivelist_verify_t: sv->sv_leftover
	*/
	static void
	livelist_verify(dsl_deadlist_t dl, void arg)
	{
	sublivelist_verify_t *sv = arg;
	dsl_deadlist_iterate(dl, sublivelist_verify_func, sv);
	}

	/*
	* Check for errors in the livelist entry and discard the intermediary
	* data structures
	*/
	/* ARGSUSED */
	static int
	sublivelist_verify_lightweight(void args, dsl_deadlist_entry_t dle)
	{
	sublivelist_verify_t sv;
	zfs_btree_create(&sv.sv_leftover, livelist_block_compare,
	sizeof (sublivelist_verify_block_t));
	int err = sublivelist_verify_func(&sv, dle);
	zfs_btree_clear(&sv.sv_leftover);
	zfs_btree_destroy(&sv.sv_leftover);
	return (err);
	}

	typedef struct metaslab_verify {
	/*
	* Tree containing all the leftover ALLOCs from the livelists
	* that are part of this metaslab.
	*/
	zfs_btree_t mv_livelist_allocs;

	/*
	* Metaslab information.
	*/
	uint64_t mv_vdid;
	uint64_t mv_msid;
	uint64_t mv_start;
	uint64_t mv_end;

	/*
	* What's currently allocated for this metaslab.
	*/
	range_tree_t *mv_allocated;
	} metaslab_verify_t;

	typedef void ll_iter_t(dsl_deadlist_t ll, void arg);

	typedef int (zdb_log_sm_cb_t)(spa_t spa, space_map_entry_t *sme, uint64_t txg,
	void *arg);

	typedef struct unflushed_iter_cb_arg {
	spa_t *uic_spa;
	uint64_t uic_txg;
	void *uic_arg;
	zdb_log_sm_cb_t uic_cb;
	} unflushed_iter_cb_arg_t;

	static int
	iterate_through_spacemap_logs_cb(space_map_entry_t sme, void arg)
	{
	unflushed_iter_cb_arg_t *uic = arg;
	return (uic->uic_cb(uic->uic_spa, sme, uic->uic_txg, uic->uic_arg));
	}

	static void
	iterate_through_spacemap_logs(spa_t spa, zdb_log_sm_cb_t cb, void arg)
	{
	if (!spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP))
	return;

	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	for (spa_log_sm_t *sls = avl_first(&spa->spa_sm_logs_by_txg);
	sls; sls = AVL_NEXT(&spa->spa_sm_logs_by_txg, sls)) {
	space_map_t *sm = NULL;
	VERIFY0(space_map_open(&sm, spa_meta_objset(spa),
	sls->sls_sm_obj, 0, UINT64_MAX, SPA_MINBLOCKSHIFT));

	unflushed_iter_cb_arg_t uic = {
	.uic_spa = spa,
	.uic_txg = sls->sls_txg,
	.uic_arg = arg,
	.uic_cb = cb
	};
	VERIFY0(space_map_iterate(sm, space_map_length(sm),
	iterate_through_spacemap_logs_cb, &uic));
	space_map_close(sm);
	}
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	}

	static void
	verify_livelist_allocs(metaslab_verify_t *mv, uint64_t txg,
	uint64_t offset, uint64_t size)
	{
	sublivelist_verify_block_t svb;
	DVA_SET_VDEV(&svb.svb_dva, mv->mv_vdid);
	DVA_SET_OFFSET(&svb.svb_dva, offset);
	DVA_SET_ASIZE(&svb.svb_dva, size);
	zfs_btree_index_t where;
	uint64_t end_offset = offset + size;

	/*
	* Look for an exact match for spacemap entry in the livelist entries.
	* Then, look for other livelist entries that fall within the range
	* of the spacemap entry as it may have been condensed
	*/
	sublivelist_verify_block_t *found =
	zfs_btree_find(&mv->mv_livelist_allocs, &svb, &where);
	if (found == NULL) {
	found = zfs_btree_next(&mv->mv_livelist_allocs, &where, &where);
	}
	for (; found != NULL && DVA_GET_VDEV(&found->svb_dva) == mv->mv_vdid &&
	DVA_GET_OFFSET(&found->svb_dva) < end_offset;
	found = zfs_btree_next(&mv->mv_livelist_allocs, &where, &where)) {
	if (found->svb_allocated_txg <= txg) {
	(void) printf("ERROR: Livelist ALLOC [%llx:%llx] "
	"from TXG %llx FREED at TXG %llx\n",
	(u_longlong_t)DVA_GET_OFFSET(&found->svb_dva),
	(u_longlong_t)DVA_GET_ASIZE(&found->svb_dva),
	(u_longlong_t)found->svb_allocated_txg,
	(u_longlong_t)txg);
	}
	}
	}

	static int
	metaslab_spacemap_validation_cb(space_map_entry_t sme, void arg)
	{
	metaslab_verify_t *mv = arg;
	uint64_t offset = sme->sme_offset;
	uint64_t size = sme->sme_run;
	uint64_t txg = sme->sme_txg;

	if (sme->sme_type == SM_ALLOC) {
	if (range_tree_contains(mv->mv_allocated,
	offset, size)) {
	(void) printf("ERROR: DOUBLE ALLOC: "
	"%llu [%llx:%llx] "
	"%llu:%llu LOG_SM\n",
	(u_longlong_t)txg, (u_longlong_t)offset,
	(u_longlong_t)size, (u_longlong_t)mv->mv_vdid,
	(u_longlong_t)mv->mv_msid);
	} else {
	range_tree_add(mv->mv_allocated,
	offset, size);
	}
	} else {
	if (!range_tree_contains(mv->mv_allocated,
	offset, size)) {
	(void) printf("ERROR: DOUBLE FREE: "
	"%llu [%llx:%llx] "
	"%llu:%llu LOG_SM\n",
	(u_longlong_t)txg, (u_longlong_t)offset,
	(u_longlong_t)size, (u_longlong_t)mv->mv_vdid,
	(u_longlong_t)mv->mv_msid);
	} else {
	range_tree_remove(mv->mv_allocated,
	offset, size);
	}
	}

	if (sme->sme_type != SM_ALLOC) {
	/*
	* If something is freed in the spacemap, verify that
	* it is not listed as allocated in the livelist.
	*/
	verify_livelist_allocs(mv, txg, offset, size);
	}
	return (0);
	}

	static int
	spacemap_check_sm_log_cb(spa_t spa, space_map_entry_t sme,
	uint64_t txg, void *arg)
	{
	metaslab_verify_t *mv = arg;
	uint64_t offset = sme->sme_offset;
	uint64_t vdev_id = sme->sme_vdev;

	vdev_t *vd = vdev_lookup_top(spa, vdev_id);

	/* skip indirect vdevs */
	if (!vdev_is_concrete(vd))
	return (0);

	if (vdev_id != mv->mv_vdid)
	return (0);

	metaslab_t *ms = vd->vdev_ms[offset >> vd->vdev_ms_shift];
	if (ms->ms_id != mv->mv_msid)
	return (0);

	if (txg < metaslab_unflushed_txg(ms))
	return (0);


	ASSERT3U(txg, ==, sme->sme_txg);
	return (metaslab_spacemap_validation_cb(sme, mv));
	}

	static void
	spacemap_check_sm_log(spa_t spa, metaslab_verify_t mv)
	{
	iterate_through_spacemap_logs(spa, spacemap_check_sm_log_cb, mv);
	}

	static void
	spacemap_check_ms_sm(space_map_t sm, metaslab_verify_t mv)
	{
	if (sm == NULL)
	return;

	VERIFY0(space_map_iterate(sm, space_map_length(sm),
	metaslab_spacemap_validation_cb, mv));
	}

	static void iterate_deleted_livelists(spa_t spa, ll_iter_t func, void arg);

	/*
	* Transfer blocks from sv_leftover tree to the mv_livelist_allocs if
	* they are part of that metaslab (mv_msid).
	*/
	static void
	mv_populate_livelist_allocs(metaslab_verify_t mv, sublivelist_verify_t sv)
	{
	zfs_btree_index_t where;
	sublivelist_verify_block_t *svb;
	ASSERT3U(zfs_btree_numnodes(&mv->mv_livelist_allocs), ==, 0);
	for (svb = zfs_btree_first(&sv->sv_leftover, &where);
	svb != NULL;
	svb = zfs_btree_next(&sv->sv_leftover, &where, &where)) {
	if (DVA_GET_VDEV(&svb->svb_dva) != mv->mv_vdid)
	continue;

	if (DVA_GET_OFFSET(&svb->svb_dva) < mv->mv_start &&
	(DVA_GET_OFFSET(&svb->svb_dva) +
	DVA_GET_ASIZE(&svb->svb_dva)) > mv->mv_start) {
	(void) printf("ERROR: Found block that crosses "
	"metaslab boundary: <%llu:%llx:%llx>\n",
	(u_longlong_t)DVA_GET_VDEV(&svb->svb_dva),
	(u_longlong_t)DVA_GET_OFFSET(&svb->svb_dva),
	(u_longlong_t)DVA_GET_ASIZE(&svb->svb_dva));
	continue;
	}

	if (DVA_GET_OFFSET(&svb->svb_dva) < mv->mv_start)
	continue;

	if (DVA_GET_OFFSET(&svb->svb_dva) >= mv->mv_end)
	continue;

	if ((DVA_GET_OFFSET(&svb->svb_dva) +
	DVA_GET_ASIZE(&svb->svb_dva)) > mv->mv_end) {
	(void) printf("ERROR: Found block that crosses "
	"metaslab boundary: <%llu:%llx:%llx>\n",
	(u_longlong_t)DVA_GET_VDEV(&svb->svb_dva),
	(u_longlong_t)DVA_GET_OFFSET(&svb->svb_dva),
	(u_longlong_t)DVA_GET_ASIZE(&svb->svb_dva));
	continue;
	}

	zfs_btree_add(&mv->mv_livelist_allocs, svb);
	}

	for (svb = zfs_btree_first(&mv->mv_livelist_allocs, &where);
	svb != NULL;
	svb = zfs_btree_next(&mv->mv_livelist_allocs, &where, &where)) {
	zfs_btree_remove(&sv->sv_leftover, svb);
	}
	}

	/*
	* [Livelist Check]
	* Iterate through all the sublivelists and:
	* - report leftover frees (**)
	* - record leftover ALLOCs together with their TXG [see Cross Check]
	*
	* (**) Note: Double ALLOCs are valid in datasets that have dedup
	* enabled. Similarly double FREEs are allowed as well but
	* only if they pair up with a corresponding ALLOC entry once
	* we our done with our sublivelist iteration.
	*
	* [Spacemap Check]
	* for each metaslab:
	* - iterate over spacemap and then the metaslab's entries in the
	* spacemap log, then report any double FREEs and ALLOCs (do not
	* blow up).
	*
	* [Cross Check]
	* After finishing the Livelist Check phase and while being in the
	* Spacemap Check phase, we find all the recorded leftover ALLOCs
	* of the livelist check that are part of the metaslab that we are
	* currently looking at in the Spacemap Check. We report any entries
	* that are marked as ALLOCs in the livelists but have been actually
	* freed (and potentially allocated again) after their TXG stamp in
	* the spacemaps. Also report any ALLOCs from the livelists that
	* belong to indirect vdevs (e.g. their vdev completed removal).
	*
	* Note that this will miss Log Spacemap entries that cancelled each other
	* out before being flushed to the metaslab, so we are not guaranteed
	* to match all erroneous ALLOCs.
	*/
	static void
	livelist_metaslab_validate(spa_t *spa)
	{
	(void) printf("Verifying deleted livelist entries\n");

	sublivelist_verify_t sv;
	zfs_btree_create(&sv.sv_leftover, livelist_block_compare,
	sizeof (sublivelist_verify_block_t));
	iterate_deleted_livelists(spa, livelist_verify, &sv);

	(void) printf("Verifying metaslab entries\n");
	vdev_t *rvd = spa->spa_root_vdev;
	for (uint64_t c = 0; c < rvd->vdev_children; c++) {
	vdev_t *vd = rvd->vdev_child[c];

	if (!vdev_is_concrete(vd))
	continue;

	for (uint64_t mid = 0; mid < vd->vdev_ms_count; mid++) {
	metaslab_t *m = vd->vdev_ms[mid];

	(void) fprintf(stderr,
	"\rverifying concrete vdev %llu, "
	"metaslab %llu of %llu ...",
	(longlong_t)vd->vdev_id,
	(longlong_t)mid,
	(longlong_t)vd->vdev_ms_count);

	uint64_t shift, start;
	range_seg_type_t type =
	metaslab_calculate_range_tree_type(vd, m,
	&start, &shift);
	metaslab_verify_t mv;
	mv.mv_allocated = range_tree_create(NULL,
	type, NULL, start, shift);
	mv.mv_vdid = vd->vdev_id;
	mv.mv_msid = m->ms_id;
	mv.mv_start = m->ms_start;
	mv.mv_end = m->ms_start + m->ms_size;
	zfs_btree_create(&mv.mv_livelist_allocs,
	livelist_block_compare,
	sizeof (sublivelist_verify_block_t));

	mv_populate_livelist_allocs(&mv, &sv);

	spacemap_check_ms_sm(m->ms_sm, &mv);
	spacemap_check_sm_log(spa, &mv);

	range_tree_vacate(mv.mv_allocated, NULL, NULL);
	range_tree_destroy(mv.mv_allocated);
	zfs_btree_clear(&mv.mv_livelist_allocs);
	zfs_btree_destroy(&mv.mv_livelist_allocs);
	}
	}
	(void) fprintf(stderr, "\n");

	/*
	* If there are any segments in the leftover tree after we walked
	* through all the metaslabs in the concrete vdevs then this means
	* that we have segments in the livelists that belong to indirect
	* vdevs and are marked as allocated.
	*/
	if (zfs_btree_numnodes(&sv.sv_leftover) == 0) {
	zfs_btree_destroy(&sv.sv_leftover);
	return;
	}
	(void) printf("ERROR: Found livelist blocks marked as allocated "
	"for indirect vdevs:\n");

	zfs_btree_index_t *where = NULL;
	sublivelist_verify_block_t *svb;
	while ((svb = zfs_btree_destroy_nodes(&sv.sv_leftover, &where)) !=
	NULL) {
	int vdev_id = DVA_GET_VDEV(&svb->svb_dva);
	ASSERT3U(vdev_id, <, rvd->vdev_children);
	vdev_t *vd = rvd->vdev_child[vdev_id];
	ASSERT(!vdev_is_concrete(vd));
	(void) printf("<%d:%llx:%llx> TXG %llx\n",
	vdev_id, (u_longlong_t)DVA_GET_OFFSET(&svb->svb_dva),
	(u_longlong_t)DVA_GET_ASIZE(&svb->svb_dva),
	(u_longlong_t)svb->svb_allocated_txg);
	}
	(void) printf("\n");
	zfs_btree_destroy(&sv.sv_leftover);
	}

	/*
	* These libumem hooks provide a reasonable set of defaults for the allocator's
	* debugging facilities.
	*/
	const char *
	_umem_debug_init(void)
	{
	return ("default,verbose"); /* $UMEM_DEBUG setting */
	}

	const char *
	_umem_logging_init(void)
	{
	return ("fail,contents"); /* $UMEM_LOGGING setting */
	}

	static void
	usage(void)
	{
	(void) fprintf(stderr,
	"Usage:\t%s [-AbcdDFGhikLMPsvXy] [-e [-V] [-p <path> ...]] "
	"[-I <inflight I/Os>]\n"
	"\t\t[-o <var>=<value>]... [-t <txg>] [-U <cache>] [-x <dumpdir>]\n"
	"\t\t[<poolname>[/<dataset \| objset id>] [<object \| range> ...]]\n"
	"\t%s [-AdiPv] [-e [-V] [-p <path> ...]] [-U <cache>]\n"
	"\t\t[<poolname>[/<dataset \| objset id>] [<object \| range> ...]\n"
	"\t%s [-v] <bookmark>\n"
	"\t%s -C [-A] [-U <cache>]\n"
	"\t%s -l [-Aqu] <device>\n"
	"\t%s -m [-AFLPX] [-e [-V] [-p <path> ...]] [-t <txg>] "
	"[-U <cache>]\n\t\t<poolname> [<vdev> [<metaslab> ...]]\n"
	"\t%s -O <dataset> <path>\n"
	"\t%s -r <dataset> <path> <destination>\n"
	"\t%s -R [-A] [-e [-V] [-p <path> ...]] [-U <cache>]\n"
	"\t\t<poolname> <vdev>:<offset>:<size>[:<flags>]\n"
	"\t%s -E [-A] word0:word1:...:word15\n"
	"\t%s -S [-AP] [-e [-V] [-p <path> ...]] [-U <cache>] "
	"<poolname>\n\n",
	cmdname, cmdname, cmdname, cmdname, cmdname, cmdname, cmdname,
	cmdname, cmdname, cmdname, cmdname);

	(void) fprintf(stderr, " Dataset name must include at least one "
	"separator character '/' or '@'\n");
	(void) fprintf(stderr, " If dataset name is specified, only that "
	"dataset is dumped\n");
	(void) fprintf(stderr, " If object numbers or object number "
	"ranges are specified, only those\n"
	" objects or ranges are dumped.\n\n");
	(void) fprintf(stderr,
	" Object ranges take the form <start>:<end>[:<flags>]\n"
	" start Starting object number\n"
	" end Ending object number, or -1 for no upper bound\n"
	" flags Optional flags to select object types:\n"
	" A All objects (this is the default)\n"
	" d ZFS directories\n"
	" f ZFS files \n"
	" m SPA space maps\n"
	" z ZAPs\n"
	" - Negate effect of next flag\n\n");
	(void) fprintf(stderr, " Options to control amount of output:\n");
	(void) fprintf(stderr, " -b block statistics\n");
	(void) fprintf(stderr, " -c checksum all metadata (twice for "
	"all data) blocks\n");
	(void) fprintf(stderr, " -C config (or cachefile if alone)\n");
	(void) fprintf(stderr, " -d dataset(s)\n");
	(void) fprintf(stderr, " -D dedup statistics\n");
	(void) fprintf(stderr, " -E decode and display block from an "
	"embedded block pointer\n");
	(void) fprintf(stderr, " -h pool history\n");
	(void) fprintf(stderr, " -i intent logs\n");
	(void) fprintf(stderr, " -l read label contents\n");
	(void) fprintf(stderr, " -k examine the checkpointed state "
	"of the pool\n");
	(void) fprintf(stderr, " -L disable leak tracking (do not "
	"load spacemaps)\n");
	(void) fprintf(stderr, " -m metaslabs\n");
	(void) fprintf(stderr, " -M metaslab groups\n");
	(void) fprintf(stderr, " -O perform object lookups by path\n");
	(void) fprintf(stderr, " -r copy an object by path to file\n");
	(void) fprintf(stderr, " -R read and display block from a "
	"device\n");
	(void) fprintf(stderr, " -s report stats on zdb's I/O\n");
	(void) fprintf(stderr, " -S simulate dedup to measure effect\n");
	(void) fprintf(stderr, " -v verbose (applies to all "
	"others)\n");
	(void) fprintf(stderr, " -y perform livelist and metaslab "
	"validation on any livelists being deleted\n\n");
	(void) fprintf(stderr, " Below options are intended for use "
	"with other options:\n");
	(void) fprintf(stderr, " -A ignore assertions (-A), enable "
	"panic recovery (-AA) or both (-AAA)\n");
	(void) fprintf(stderr, " -e pool is exported/destroyed/"
	"has altroot/not in a cachefile\n");
	(void) fprintf(stderr, " -F attempt automatic rewind within "
	"safe range of transaction groups\n");
	(void) fprintf(stderr, " -G dump zfs_dbgmsg buffer before "
	"exiting\n");
	(void) fprintf(stderr, " -I <number of inflight I/Os> -- "
	"specify the maximum number of\n "
	"checksumming I/Os [default is 200]\n");
	(void) fprintf(stderr, " -o <variable>=<value> set global "
	"variable to an unsigned 32-bit integer\n");
	(void) fprintf(stderr, " -p <path> -- use one or more with "
	"-e to specify path to vdev dir\n");
	(void) fprintf(stderr, " -P print numbers in parseable form\n");
	(void) fprintf(stderr, " -q don't print label contents\n");
	(void) fprintf(stderr, " -t <txg> -- highest txg to use when "
	"searching for uberblocks\n");
	(void) fprintf(stderr, " -u uberblock\n");
	(void) fprintf(stderr, " -U <cachefile_path> -- use alternate "
	"cachefile\n");
	(void) fprintf(stderr, " -V do verbatim import\n");
	(void) fprintf(stderr, " -x <dumpdir> -- "
	"dump all read blocks into specified directory\n");
	(void) fprintf(stderr, " -X attempt extreme rewind (does not "
	"work with dataset)\n");
	(void) fprintf(stderr, " -Y attempt all reconstruction "
	"combinations for split blocks\n");
	(void) fprintf(stderr, " -Z show ZSTD headers \n");
	(void) fprintf(stderr, "Specify an option more than once (e.g. -bb) "
	"to make only that option verbose\n");
	(void) fprintf(stderr, "Default is to dump everything non-verbosely\n");
	exit(1);
	}

	static void
	dump_debug_buffer(void)
	{
	if (dump_opt['G']) {
	(void) printf("\n");
	(void) fflush(stdout);
	zfs_dbgmsg_print("zdb");
	}
	}

	/*
	* Called for usage errors that are discovered after a call to spa_open(),
	* dmu_bonus_hold(), or pool_match(). abort() is called for other errors.
	*/

	static void
	fatal(const char *fmt, ...)
	{
	va_list ap;

	va_start(ap, fmt);
	(void) fprintf(stderr, "%s: ", cmdname);
	(void) vfprintf(stderr, fmt, ap);
	va_end(ap);
	(void) fprintf(stderr, "\n");

	dump_debug_buffer();

	exit(1);
	}

	/* ARGSUSED */
	static void
	dump_packed_nvlist(objset_t os, uint64_t object, void data, size_t size)
	{
	nvlist_t *nv;
	size_t nvsize = (uint64_t )data;
	char *packed = umem_alloc(nvsize, UMEM_NOFAIL);

	VERIFY(0 == dmu_read(os, object, 0, nvsize, packed, DMU_READ_PREFETCH));

	VERIFY(nvlist_unpack(packed, nvsize, &nv, 0) == 0);

	umem_free(packed, nvsize);

	dump_nvlist(nv, 8);

	nvlist_free(nv);
	}

	/* ARGSUSED */
	static void
	dump_history_offsets(objset_t os, uint64_t object, void data, size_t size)
	{
	spa_history_phys_t *shp = data;

	if (shp == NULL)
	return;

	(void) printf("\t\tpool_create_len = %llu\n",
	(u_longlong_t)shp->sh_pool_create_len);
	(void) printf("\t\tphys_max_off = %llu\n",
	(u_longlong_t)shp->sh_phys_max_off);
	(void) printf("\t\tbof = %llu\n",
	(u_longlong_t)shp->sh_bof);
	(void) printf("\t\teof = %llu\n",
	(u_longlong_t)shp->sh_eof);
	(void) printf("\t\trecords_lost = %llu\n",
	(u_longlong_t)shp->sh_records_lost);
	}

	static void
	zdb_nicenum(uint64_t num, char *buf, size_t buflen)
	{
	if (dump_opt['P'])
	(void) snprintf(buf, buflen, "%llu", (longlong_t)num);
	else
	nicenum(num, buf, sizeof (buf));
	}

	static const char histo_stars[] = "****************************************";
	static const uint64_t histo_width = sizeof (histo_stars) - 1;

	static void
	dump_histogram(const uint64_t *histo, int size, int offset)
	{
	int i;
	int minidx = size - 1;
	int maxidx = 0;
	uint64_t max = 0;

	for (i = 0; i < size; i++) {
	if (histo[i] > max)
	max = histo[i];
	if (histo[i] > 0 && i > maxidx)
	maxidx = i;
	if (histo[i] > 0 && i < minidx)
	minidx = i;
	}

	if (max < histo_width)
	max = histo_width;

	for (i = minidx; i <= maxidx; i++) {
	(void) printf("\t\t\t%3u: %6llu %s\n",
	i + offset, (u_longlong_t)histo[i],
	&histo_stars[(max - histo[i]) * histo_width / max]);
	}
	}

	static void
	dump_zap_stats(objset_t *os, uint64_t object)
	{
	int error;
	zap_stats_t zs;

	error = zap_get_stats(os, object, &zs);
	if (error)
	return;

	if (zs.zs_ptrtbl_len == 0) {
	ASSERT(zs.zs_num_blocks == 1);
	(void) printf("\tmicrozap: %llu bytes, %llu entries\n",
	(u_longlong_t)zs.zs_blocksize,
	(u_longlong_t)zs.zs_num_entries);
	return;
	}

	(void) printf("\tFat ZAP stats:\n");

	(void) printf("\t\tPointer table:\n");
	(void) printf("\t\t\t%llu elements\n",
	(u_longlong_t)zs.zs_ptrtbl_len);
	(void) printf("\t\t\tzt_blk: %llu\n",
	(u_longlong_t)zs.zs_ptrtbl_zt_blk);
	(void) printf("\t\t\tzt_numblks: %llu\n",
	(u_longlong_t)zs.zs_ptrtbl_zt_numblks);
	(void) printf("\t\t\tzt_shift: %llu\n",
	(u_longlong_t)zs.zs_ptrtbl_zt_shift);
	(void) printf("\t\t\tzt_blks_copied: %llu\n",
	(u_longlong_t)zs.zs_ptrtbl_blks_copied);
	(void) printf("\t\t\tzt_nextblk: %llu\n",
	(u_longlong_t)zs.zs_ptrtbl_nextblk);

	(void) printf("\t\tZAP entries: %llu\n",
	(u_longlong_t)zs.zs_num_entries);
	(void) printf("\t\tLeaf blocks: %llu\n",
	(u_longlong_t)zs.zs_num_leafs);
	(void) printf("\t\tTotal blocks: %llu\n",
	(u_longlong_t)zs.zs_num_blocks);
	(void) printf("\t\tzap_block_type: 0x%llx\n",
	(u_longlong_t)zs.zs_block_type);
	(void) printf("\t\tzap_magic: 0x%llx\n",
	(u_longlong_t)zs.zs_magic);
	(void) printf("\t\tzap_salt: 0x%llx\n",
	(u_longlong_t)zs.zs_salt);

	(void) printf("\t\tLeafs with 2^n pointers:\n");
	dump_histogram(zs.zs_leafs_with_2n_pointers, ZAP_HISTOGRAM_SIZE, 0);

	(void) printf("\t\tBlocks with n*5 entries:\n");
	dump_histogram(zs.zs_blocks_with_n5_entries, ZAP_HISTOGRAM_SIZE, 0);

	(void) printf("\t\tBlocks n/10 full:\n");
	dump_histogram(zs.zs_blocks_n_tenths_full, ZAP_HISTOGRAM_SIZE, 0);

	(void) printf("\t\tEntries with n chunks:\n");
	dump_histogram(zs.zs_entries_using_n_chunks, ZAP_HISTOGRAM_SIZE, 0);

	(void) printf("\t\tBuckets with n entries:\n");
	dump_histogram(zs.zs_buckets_with_n_entries, ZAP_HISTOGRAM_SIZE, 0);
	}

	/ARGSUSED/
	static void
	dump_none(objset_t os, uint64_t object, void data, size_t size)
	{
	}

	/ARGSUSED/
	static void
	dump_unknown(objset_t os, uint64_t object, void data, size_t size)
	{
	(void) printf("\tUNKNOWN OBJECT TYPE\n");
	}

	/ARGSUSED/
	static void
	dump_uint8(objset_t os, uint64_t object, void data, size_t size)
	{
	}

	/ARGSUSED/
	static void
	dump_uint64(objset_t os, uint64_t object, void data, size_t size)
	{
	uint64_t *arr;
	uint64_t oursize;
	if (dump_opt['d'] < 6)
	return;

	if (data == NULL) {
	dmu_object_info_t doi;

	VERIFY0(dmu_object_info(os, object, &doi));
	size = doi.doi_max_offset;
	/*
	* We cap the size at 1 mebibyte here to prevent
	* allocation failures and nigh-infinite printing if the
	* object is extremely large.
	*/
	oursize = MIN(size, 1 << 20);
	arr = kmem_alloc(oursize, KM_SLEEP);

	int err = dmu_read(os, object, 0, oursize, arr, 0);
	if (err != 0) {
	(void) printf("got error %u from dmu_read\n", err);
	kmem_free(arr, oursize);
	return;
	}
	} else {
	/*
	* Even though the allocation is already done in this code path,
	* we still cap the size to prevent excessive printing.
	*/
	oursize = MIN(size, 1 << 20);
	arr = data;
	}

	if (size == 0) {
	(void) printf("\t\t[]\n");
	return;
	}

	(void) printf("\t\t[%0llx", (u_longlong_t)arr[0]);
	for (size_t i = 1; i * sizeof (uint64_t) < oursize; i++) {
	if (i % 4 != 0)
	(void) printf(", %0llx", (u_longlong_t)arr[i]);
	else
	(void) printf(",\n\t\t%0llx", (u_longlong_t)arr[i]);
	}
	if (oursize != size)
	(void) printf(", ... ");
	(void) printf("]\n");

	if (data == NULL)
	kmem_free(arr, oursize);
	}

	/ARGSUSED/
	static void
	dump_zap(objset_t os, uint64_t object, void data, size_t size)
	{
	zap_cursor_t zc;
	zap_attribute_t attr;
	void *prop;
	unsigned i;

	dump_zap_stats(os, object);
	(void) printf("\n");

	for (zap_cursor_init(&zc, os, object);
	zap_cursor_retrieve(&zc, &attr) == 0;
	zap_cursor_advance(&zc)) {
	(void) printf("\t\t%s = ", attr.za_name);
	if (attr.za_num_integers == 0) {
	(void) printf("\n");
	continue;
	}
	prop = umem_zalloc(attr.za_num_integers *
	attr.za_integer_length, UMEM_NOFAIL);
	(void) zap_lookup(os, object, attr.za_name,
	attr.za_integer_length, attr.za_num_integers, prop);
	if (attr.za_integer_length == 1) {
	if (strcmp(attr.za_name,
	DSL_CRYPTO_KEY_MASTER_KEY) == 0 \|\|
	strcmp(attr.za_name,
	DSL_CRYPTO_KEY_HMAC_KEY) == 0 \|\|
	strcmp(attr.za_name, DSL_CRYPTO_KEY_IV) == 0 \|\|
	strcmp(attr.za_name, DSL_CRYPTO_KEY_MAC) == 0 \|\|
	strcmp(attr.za_name, DMU_POOL_CHECKSUM_SALT) == 0) {
	uint8_t *u8 = prop;

	for (i = 0; i < attr.za_num_integers; i++) {
	(void) printf("%02x", u8[i]);
	}
	} else {
	(void) printf("%s", (char *)prop);
	}
	} else {
	for (i = 0; i < attr.za_num_integers; i++) {
	switch (attr.za_integer_length) {
	case 2:
	(void) printf("%u ",
	((uint16_t *)prop)[i]);
	break;
	case 4:
	(void) printf("%u ",
	((uint32_t *)prop)[i]);
	break;
	case 8:
	(void) printf("%lld ",
	(u_longlong_t)((int64_t *)prop)[i]);
	break;
	}
	}
	}
	(void) printf("\n");
	umem_free(prop, attr.za_num_integers * attr.za_integer_length);
	}
	zap_cursor_fini(&zc);
	}

	static void
	dump_bpobj(objset_t os, uint64_t object, void data, size_t size)
	{
	bpobj_phys_t *bpop = data;
	uint64_t i;
	char bytes[32], comp[32], uncomp[32];

	/* make sure the output won't get truncated */
	CTASSERT(sizeof (bytes) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (comp) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (uncomp) >= NN_NUMBUF_SZ);

	if (bpop == NULL)
	return;

	zdb_nicenum(bpop->bpo_bytes, bytes, sizeof (bytes));
	zdb_nicenum(bpop->bpo_comp, comp, sizeof (comp));
	zdb_nicenum(bpop->bpo_uncomp, uncomp, sizeof (uncomp));

	(void) printf("\t\tnum_blkptrs = %llu\n",
	(u_longlong_t)bpop->bpo_num_blkptrs);
	(void) printf("\t\tbytes = %s\n", bytes);
	if (size >= BPOBJ_SIZE_V1) {
	(void) printf("\t\tcomp = %s\n", comp);
	(void) printf("\t\tuncomp = %s\n", uncomp);
	}
	if (size >= BPOBJ_SIZE_V2) {
	(void) printf("\t\tsubobjs = %llu\n",
	(u_longlong_t)bpop->bpo_subobjs);
	(void) printf("\t\tnum_subobjs = %llu\n",
	(u_longlong_t)bpop->bpo_num_subobjs);
	}
	if (size >= sizeof (*bpop)) {
	(void) printf("\t\tnum_freed = %llu\n",
	(u_longlong_t)bpop->bpo_num_freed);
	}

	if (dump_opt['d'] < 5)
	return;

	for (i = 0; i < bpop->bpo_num_blkptrs; i++) {
	char blkbuf[BP_SPRINTF_LEN];
	blkptr_t bp;

	int err = dmu_read(os, object,
	i * sizeof (bp), sizeof (bp), &bp, 0);
	if (err != 0) {
	(void) printf("got error %u from dmu_read\n", err);
	break;
	}
	snprintf_blkptr_compact(blkbuf, sizeof (blkbuf), &bp,
	BP_GET_FREE(&bp));
	(void) printf("\t%s\n", blkbuf);
	}
	}

	/* ARGSUSED */
	static void
	dump_bpobj_subobjs(objset_t os, uint64_t object, void data, size_t size)
	{
	dmu_object_info_t doi;
	int64_t i;

	VERIFY0(dmu_object_info(os, object, &doi));
	uint64_t *subobjs = kmem_alloc(doi.doi_max_offset, KM_SLEEP);

	int err = dmu_read(os, object, 0, doi.doi_max_offset, subobjs, 0);
	if (err != 0) {
	(void) printf("got error %u from dmu_read\n", err);
	kmem_free(subobjs, doi.doi_max_offset);
	return;
	}

	int64_t last_nonzero = -1;
	for (i = 0; i < doi.doi_max_offset / 8; i++) {
	if (subobjs[i] != 0)
	last_nonzero = i;
	}

	for (i = 0; i <= last_nonzero; i++) {
	(void) printf("\t%llu\n", (u_longlong_t)subobjs[i]);
	}
	kmem_free(subobjs, doi.doi_max_offset);
	}

	/ARGSUSED/
	static void
	dump_ddt_zap(objset_t os, uint64_t object, void data, size_t size)
	{
	dump_zap_stats(os, object);
	/* contents are printed elsewhere, properly decoded */
	}

	/ARGSUSED/
	static void
	dump_sa_attrs(objset_t os, uint64_t object, void data, size_t size)
	{
	zap_cursor_t zc;
	zap_attribute_t attr;

	dump_zap_stats(os, object);
	(void) printf("\n");

	for (zap_cursor_init(&zc, os, object);
	zap_cursor_retrieve(&zc, &attr) == 0;
	zap_cursor_advance(&zc)) {
	(void) printf("\t\t%s = ", attr.za_name);
	if (attr.za_num_integers == 0) {
	(void) printf("\n");
	continue;
	}
	(void) printf(" %llx : [%d:%d:%d]\n",
	(u_longlong_t)attr.za_first_integer,
	(int)ATTR_LENGTH(attr.za_first_integer),
	(int)ATTR_BSWAP(attr.za_first_integer),
	(int)ATTR_NUM(attr.za_first_integer));
	}
	zap_cursor_fini(&zc);
	}

	/ARGSUSED/
	static void
	dump_sa_layouts(objset_t os, uint64_t object, void data, size_t size)
	{
	zap_cursor_t zc;
	zap_attribute_t attr;
	uint16_t *layout_attrs;
	unsigned i;

	dump_zap_stats(os, object);
	(void) printf("\n");

	for (zap_cursor_init(&zc, os, object);
	zap_cursor_retrieve(&zc, &attr) == 0;
	zap_cursor_advance(&zc)) {
	(void) printf("\t\t%s = [", attr.za_name);
	if (attr.za_num_integers == 0) {
	(void) printf("\n");
	continue;
	}

	VERIFY(attr.za_integer_length == 2);
	layout_attrs = umem_zalloc(attr.za_num_integers *
	attr.za_integer_length, UMEM_NOFAIL);

	VERIFY(zap_lookup(os, object, attr.za_name,
	attr.za_integer_length,
	attr.za_num_integers, layout_attrs) == 0);

	for (i = 0; i != attr.za_num_integers; i++)
	(void) printf(" %d ", (int)layout_attrs[i]);
	(void) printf("]\n");
	umem_free(layout_attrs,
	attr.za_num_integers * attr.za_integer_length);
	}
	zap_cursor_fini(&zc);
	}

	/ARGSUSED/
	static void
	dump_zpldir(objset_t os, uint64_t object, void data, size_t size)
	{
	zap_cursor_t zc;
	zap_attribute_t attr;
	const char *typenames[] = {
	/* 0 */ "not specified",
	/* 1 */ "FIFO",
	/* 2 */ "Character Device",
	/* 3 */ "3 (invalid)",
	/* 4 */ "Directory",
	/* 5 */ "5 (invalid)",
	/* 6 */ "Block Device",
	/* 7 */ "7 (invalid)",
	/* 8 */ "Regular File",
	/* 9 */ "9 (invalid)",
	/* 10 */ "Symbolic Link",
	/* 11 */ "11 (invalid)",
	/* 12 */ "Socket",
	/* 13 */ "Door",
	/* 14 */ "Event Port",
	/* 15 */ "15 (invalid)",
	};

	dump_zap_stats(os, object);
	(void) printf("\n");

	for (zap_cursor_init(&zc, os, object);
	zap_cursor_retrieve(&zc, &attr) == 0;
	zap_cursor_advance(&zc)) {
	(void) printf("\t\t%s = %lld (type: %s)\n",
	attr.za_name, ZFS_DIRENT_OBJ(attr.za_first_integer),
	typenames[ZFS_DIRENT_TYPE(attr.za_first_integer)]);
	}
	zap_cursor_fini(&zc);
	}

	static int
	get_dtl_refcount(vdev_t *vd)
	{
	int refcount = 0;

	if (vd->vdev_ops->vdev_op_leaf) {
	space_map_t *sm = vd->vdev_dtl_sm;

	if (sm != NULL &&
	sm->sm_dbuf->db_size == sizeof (space_map_phys_t))
	return (1);
	return (0);
	}

	for (unsigned c = 0; c < vd->vdev_children; c++)
	refcount += get_dtl_refcount(vd->vdev_child[c]);
	return (refcount);
	}

	static int
	get_metaslab_refcount(vdev_t *vd)
	{
	int refcount = 0;

	if (vd->vdev_top == vd) {
	for (uint64_t m = 0; m < vd->vdev_ms_count; m++) {
	space_map_t *sm = vd->vdev_ms[m]->ms_sm;

	if (sm != NULL &&
	sm->sm_dbuf->db_size == sizeof (space_map_phys_t))
	refcount++;
	}
	}
	for (unsigned c = 0; c < vd->vdev_children; c++)
	refcount += get_metaslab_refcount(vd->vdev_child[c]);

	return (refcount);
	}

	static int
	get_obsolete_refcount(vdev_t *vd)
	{
	uint64_t obsolete_sm_object;
	int refcount = 0;

	VERIFY0(vdev_obsolete_sm_object(vd, &obsolete_sm_object));
	if (vd->vdev_top == vd && obsolete_sm_object != 0) {
	dmu_object_info_t doi;
	VERIFY0(dmu_object_info(vd->vdev_spa->spa_meta_objset,
	obsolete_sm_object, &doi));
	if (doi.doi_bonus_size == sizeof (space_map_phys_t)) {
	refcount++;
	}
	} else {
	ASSERT3P(vd->vdev_obsolete_sm, ==, NULL);
	ASSERT3U(obsolete_sm_object, ==, 0);
	}
	for (unsigned c = 0; c < vd->vdev_children; c++) {
	refcount += get_obsolete_refcount(vd->vdev_child[c]);
	}

	return (refcount);
	}

	static int
	get_prev_obsolete_spacemap_refcount(spa_t *spa)
	{
	uint64_t prev_obj =
	spa->spa_condensing_indirect_phys.scip_prev_obsolete_sm_object;
	if (prev_obj != 0) {
	dmu_object_info_t doi;
	VERIFY0(dmu_object_info(spa->spa_meta_objset, prev_obj, &doi));
	if (doi.doi_bonus_size == sizeof (space_map_phys_t)) {
	return (1);
	}
	}
	return (0);
	}

	static int
	get_checkpoint_refcount(vdev_t *vd)
	{
	int refcount = 0;

	if (vd->vdev_top == vd && vd->vdev_top_zap != 0 &&
	zap_contains(spa_meta_objset(vd->vdev_spa),
	vd->vdev_top_zap, VDEV_TOP_ZAP_POOL_CHECKPOINT_SM) == 0)
	refcount++;

	for (uint64_t c = 0; c < vd->vdev_children; c++)
	refcount += get_checkpoint_refcount(vd->vdev_child[c]);

	return (refcount);
	}

	static int
	get_log_spacemap_refcount(spa_t *spa)
	{
	return (avl_numnodes(&spa->spa_sm_logs_by_txg));
	}

	static int
	verify_spacemap_refcounts(spa_t *spa)
	{
	uint64_t expected_refcount = 0;
	uint64_t actual_refcount;

	(void) feature_get_refcount(spa,
	&spa_feature_table[SPA_FEATURE_SPACEMAP_HISTOGRAM],
	&expected_refcount);
	actual_refcount = get_dtl_refcount(spa->spa_root_vdev);
	actual_refcount += get_metaslab_refcount(spa->spa_root_vdev);
	actual_refcount += get_obsolete_refcount(spa->spa_root_vdev);
	actual_refcount += get_prev_obsolete_spacemap_refcount(spa);
	actual_refcount += get_checkpoint_refcount(spa->spa_root_vdev);
	actual_refcount += get_log_spacemap_refcount(spa);

	if (expected_refcount != actual_refcount) {
	(void) printf("space map refcount mismatch: expected %lld != "
	"actual %lld\n",
	(longlong_t)expected_refcount,
	(longlong_t)actual_refcount);
	return (2);
	}
	return (0);
	}

	static void
	dump_spacemap(objset_t os, space_map_t sm)
	{
	const char *ddata[] = { "ALLOC", "FREE", "CONDENSE", "INVALID",
	"INVALID", "INVALID", "INVALID", "INVALID" };

	if (sm == NULL)
	return;

	(void) printf("space map object %llu:\n",
	(longlong_t)sm->sm_object);
	(void) printf(" smp_length = 0x%llx\n",
	(longlong_t)sm->sm_phys->smp_length);
	(void) printf(" smp_alloc = 0x%llx\n",
	(longlong_t)sm->sm_phys->smp_alloc);

	if (dump_opt['d'] < 6 && dump_opt['m'] < 4)
	return;

	/*
	* Print out the freelist entries in both encoded and decoded form.
	*/
	uint8_t mapshift = sm->sm_shift;
	int64_t alloc = 0;
	uint64_t word, entry_id = 0;
	for (uint64_t offset = 0; offset < space_map_length(sm);
	offset += sizeof (word)) {

	VERIFY0(dmu_read(os, space_map_object(sm), offset,
	sizeof (word), &word, DMU_READ_PREFETCH));

	if (sm_entry_is_debug(word)) {
	uint64_t de_txg = SM_DEBUG_TXG_DECODE(word);
	uint64_t de_sync_pass = SM_DEBUG_SYNCPASS_DECODE(word);
	if (de_txg == 0) {
	(void) printf(
	"\t [%6llu] PADDING\n",
	(u_longlong_t)entry_id);
	} else {
	(void) printf(
	"\t [%6llu] %s: txg %llu pass %llu\n",
	(u_longlong_t)entry_id,
	ddata[SM_DEBUG_ACTION_DECODE(word)],
	(u_longlong_t)de_txg,
	(u_longlong_t)de_sync_pass);
	}
	entry_id++;
	continue;
	}

	uint8_t words;
	char entry_type;
	uint64_t entry_off, entry_run, entry_vdev = SM_NO_VDEVID;

	if (sm_entry_is_single_word(word)) {
	entry_type = (SM_TYPE_DECODE(word) == SM_ALLOC) ?
	'A' : 'F';
	entry_off = (SM_OFFSET_DECODE(word) << mapshift) +
	sm->sm_start;
	entry_run = SM_RUN_DECODE(word) << mapshift;
	words = 1;
	} else {
	/* it is a two-word entry so we read another word */
	ASSERT(sm_entry_is_double_word(word));

	uint64_t extra_word;
	offset += sizeof (extra_word);
	VERIFY0(dmu_read(os, space_map_object(sm), offset,
	sizeof (extra_word), &extra_word,
	DMU_READ_PREFETCH));

	ASSERT3U(offset, <=, space_map_length(sm));

	entry_run = SM2_RUN_DECODE(word) << mapshift;
	entry_vdev = SM2_VDEV_DECODE(word);
	entry_type = (SM2_TYPE_DECODE(extra_word) == SM_ALLOC) ?
	'A' : 'F';
	entry_off = (SM2_OFFSET_DECODE(extra_word) <<
	mapshift) + sm->sm_start;
	words = 2;
	}

	(void) printf("\t [%6llu] %c range:"
	" %010llx-%010llx size: %06llx vdev: %06llu words: %u\n",
	(u_longlong_t)entry_id,
	entry_type, (u_longlong_t)entry_off,
	(u_longlong_t)(entry_off + entry_run),
	(u_longlong_t)entry_run,
	(u_longlong_t)entry_vdev, words);

	if (entry_type == 'A')
	alloc += entry_run;
	else
	alloc -= entry_run;
	entry_id++;
	}
	if (alloc != space_map_allocated(sm)) {
	(void) printf("space_map_object alloc (%lld) INCONSISTENT "
	"with space map summary (%lld)\n",
	(longlong_t)space_map_allocated(sm), (longlong_t)alloc);
	}
	}

	static void
	dump_metaslab_stats(metaslab_t *msp)
	{
	char maxbuf[32];
	range_tree_t *rt = msp->ms_allocatable;
	zfs_btree_t *t = &msp->ms_allocatable_by_size;
	int free_pct = range_tree_space(rt) * 100 / msp->ms_size;

	/* max sure nicenum has enough space */
	CTASSERT(sizeof (maxbuf) >= NN_NUMBUF_SZ);

	zdb_nicenum(metaslab_largest_allocatable(msp), maxbuf, sizeof (maxbuf));

	(void) printf("\t %25s %10lu %7s %6s %4s %4d%%\n",
	"segments", zfs_btree_numnodes(t), "maxsize", maxbuf,
	"freepct", free_pct);
	(void) printf("\tIn-memory histogram:\n");
	dump_histogram(rt->rt_histogram, RANGE_TREE_HISTOGRAM_SIZE, 0);
	}

	static void
	dump_metaslab(metaslab_t *msp)
	{
	vdev_t *vd = msp->ms_group->mg_vd;
	spa_t *spa = vd->vdev_spa;
	space_map_t *sm = msp->ms_sm;
	char freebuf[32];

	zdb_nicenum(msp->ms_size - space_map_allocated(sm), freebuf,
	sizeof (freebuf));

	(void) printf(
	"\tmetaslab %6llu offset %12llx spacemap %6llu free %5s\n",
	(u_longlong_t)msp->ms_id, (u_longlong_t)msp->ms_start,
	(u_longlong_t)space_map_object(sm), freebuf);

	if (dump_opt['m'] > 2 && !dump_opt['L']) {
	mutex_enter(&msp->ms_lock);
	VERIFY0(metaslab_load(msp));
	range_tree_stat_verify(msp->ms_allocatable);
	dump_metaslab_stats(msp);
	metaslab_unload(msp);
	mutex_exit(&msp->ms_lock);
	}

	if (dump_opt['m'] > 1 && sm != NULL &&
	spa_feature_is_active(spa, SPA_FEATURE_SPACEMAP_HISTOGRAM)) {
	/*
	* The space map histogram represents free space in chunks
	* of sm_shift (i.e. bucket 0 refers to 2^sm_shift).
	*/
	(void) printf("\tOn-disk histogram:\t\tfragmentation %llu\n",
	(u_longlong_t)msp->ms_fragmentation);
	dump_histogram(sm->sm_phys->smp_histogram,
	SPACE_MAP_HISTOGRAM_SIZE, sm->sm_shift);
	}

	if (vd->vdev_ops == &vdev_draid_ops)
	ASSERT3U(msp->ms_size, <=, 1ULL << vd->vdev_ms_shift);
	else
	ASSERT3U(msp->ms_size, ==, 1ULL << vd->vdev_ms_shift);

	dump_spacemap(spa->spa_meta_objset, msp->ms_sm);

	if (spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP)) {
	(void) printf("\tFlush data:\n\tunflushed txg=%llu\n\n",
	(u_longlong_t)metaslab_unflushed_txg(msp));
	}
	}

	static void
	print_vdev_metaslab_header(vdev_t *vd)
	{
	vdev_alloc_bias_t alloc_bias = vd->vdev_alloc_bias;
	const char *bias_str = "";
	if (alloc_bias == VDEV_BIAS_LOG \|\| vd->vdev_islog) {
	bias_str = VDEV_ALLOC_BIAS_LOG;
	} else if (alloc_bias == VDEV_BIAS_SPECIAL) {
	bias_str = VDEV_ALLOC_BIAS_SPECIAL;
	} else if (alloc_bias == VDEV_BIAS_DEDUP) {
	bias_str = VDEV_ALLOC_BIAS_DEDUP;
	}

	uint64_t ms_flush_data_obj = 0;
	if (vd->vdev_top_zap != 0) {
	int error = zap_lookup(spa_meta_objset(vd->vdev_spa),
	vd->vdev_top_zap, VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS,
	sizeof (uint64_t), 1, &ms_flush_data_obj);
	if (error != ENOENT) {
	ASSERT0(error);
	}
	}

	(void) printf("\tvdev %10llu %s",
	(u_longlong_t)vd->vdev_id, bias_str);

	if (ms_flush_data_obj != 0) {
	(void) printf(" ms_unflushed_phys object %llu",
	(u_longlong_t)ms_flush_data_obj);
	}

	(void) printf("\n\t%-10s%5llu %-19s %-15s %-12s\n",
	"metaslabs", (u_longlong_t)vd->vdev_ms_count,
	"offset", "spacemap", "free");
	(void) printf("\t%15s %19s %15s %12s\n",
	"---------------", "-------------------",
	"---------------", "------------");
	}

	static void
	-dump_metaslab_groups(spa_t *spa)
	+dump_metaslab_groups(spa_t *spa, boolean_t show_special)
	{
	vdev_t *rvd = spa->spa_root_vdev;
	metaslab_class_t *mc = spa_normal_class(spa);
	+ metaslab_class_t *smc = spa_special_class(spa);
	uint64_t fragmentation;

	metaslab_class_histogram_verify(mc);

	for (unsigned c = 0; c < rvd->vdev_children; c++) {
	vdev_t *tvd = rvd->vdev_child[c];
	metaslab_group_t *mg = tvd->vdev_mg;

	- if (mg == NULL \|\| mg->mg_class != mc)
	+ if (mg == NULL \|\| (mg->mg_class != mc &&
	+ (!show_special \|\| mg->mg_class != smc)))
	continue;

	metaslab_group_histogram_verify(mg);
	mg->mg_fragmentation = metaslab_group_fragmentation(mg);

	(void) printf("\tvdev %10llu\t\tmetaslabs%5llu\t\t"
	"fragmentation",
	(u_longlong_t)tvd->vdev_id,
	(u_longlong_t)tvd->vdev_ms_count);
	if (mg->mg_fragmentation == ZFS_FRAG_INVALID) {
	(void) printf("%3s\n", "-");
	} else {
	(void) printf("%3llu%%\n",
	(u_longlong_t)mg->mg_fragmentation);
	}
	dump_histogram(mg->mg_histogram, RANGE_TREE_HISTOGRAM_SIZE, 0);
	}

	(void) printf("\tpool %s\tfragmentation", spa_name(spa));
	fragmentation = metaslab_class_fragmentation(mc);
	if (fragmentation == ZFS_FRAG_INVALID)
	(void) printf("\t%3s\n", "-");
	else
	(void) printf("\t%3llu%%\n", (u_longlong_t)fragmentation);
	dump_histogram(mc->mc_histogram, RANGE_TREE_HISTOGRAM_SIZE, 0);
	}

	static void
	print_vdev_indirect(vdev_t *vd)
	{
	vdev_indirect_config_t *vic = &vd->vdev_indirect_config;
	vdev_indirect_mapping_t *vim = vd->vdev_indirect_mapping;
	vdev_indirect_births_t *vib = vd->vdev_indirect_births;

	if (vim == NULL) {
	ASSERT3P(vib, ==, NULL);
	return;
	}

	ASSERT3U(vdev_indirect_mapping_object(vim), ==,
	vic->vic_mapping_object);
	ASSERT3U(vdev_indirect_births_object(vib), ==,
	vic->vic_births_object);

	(void) printf("indirect births obj %llu:\n",
	(longlong_t)vic->vic_births_object);
	(void) printf(" vib_count = %llu\n",
	(longlong_t)vdev_indirect_births_count(vib));
	for (uint64_t i = 0; i < vdev_indirect_births_count(vib); i++) {
	vdev_indirect_birth_entry_phys_t *cur_vibe =
	&vib->vib_entries[i];
	(void) printf("\toffset %llx -> txg %llu\n",
	(longlong_t)cur_vibe->vibe_offset,
	(longlong_t)cur_vibe->vibe_phys_birth_txg);
	}
	(void) printf("\n");

	(void) printf("indirect mapping obj %llu:\n",
	(longlong_t)vic->vic_mapping_object);
	(void) printf(" vim_max_offset = 0x%llx\n",
	(longlong_t)vdev_indirect_mapping_max_offset(vim));
	(void) printf(" vim_bytes_mapped = 0x%llx\n",
	(longlong_t)vdev_indirect_mapping_bytes_mapped(vim));
	(void) printf(" vim_count = %llu\n",
	(longlong_t)vdev_indirect_mapping_num_entries(vim));

	if (dump_opt['d'] <= 5 && dump_opt['m'] <= 3)
	return;

	uint32_t *counts = vdev_indirect_mapping_load_obsolete_counts(vim);

	for (uint64_t i = 0; i < vdev_indirect_mapping_num_entries(vim); i++) {
	vdev_indirect_mapping_entry_phys_t *vimep =
	&vim->vim_entries[i];
	(void) printf("\t<%llx:%llx:%llx> -> "
	"<%llx:%llx:%llx> (%x obsolete)\n",
	(longlong_t)vd->vdev_id,
	(longlong_t)DVA_MAPPING_GET_SRC_OFFSET(vimep),
	(longlong_t)DVA_GET_ASIZE(&vimep->vimep_dst),
	(longlong_t)DVA_GET_VDEV(&vimep->vimep_dst),
	(longlong_t)DVA_GET_OFFSET(&vimep->vimep_dst),
	(longlong_t)DVA_GET_ASIZE(&vimep->vimep_dst),
	counts[i]);
	}
	(void) printf("\n");

	uint64_t obsolete_sm_object;
	VERIFY0(vdev_obsolete_sm_object(vd, &obsolete_sm_object));
	if (obsolete_sm_object != 0) {
	objset_t *mos = vd->vdev_spa->spa_meta_objset;
	(void) printf("obsolete space map object %llu:\n",
	(u_longlong_t)obsolete_sm_object);
	ASSERT(vd->vdev_obsolete_sm != NULL);
	ASSERT3U(space_map_object(vd->vdev_obsolete_sm), ==,
	obsolete_sm_object);
	dump_spacemap(mos, vd->vdev_obsolete_sm);
	(void) printf("\n");
	}
	}

	static void
	dump_metaslabs(spa_t *spa)
	{
	vdev_t vd, rvd = spa->spa_root_vdev;
	uint64_t m, c = 0, children = rvd->vdev_children;

	(void) printf("\nMetaslabs:\n");

	if (!dump_opt['d'] && zopt_metaslab_args > 0) {
	c = zopt_metaslab[0];

	if (c >= children)
	(void) fatal("bad vdev id: %llu", (u_longlong_t)c);

	if (zopt_metaslab_args > 1) {
	vd = rvd->vdev_child[c];
	print_vdev_metaslab_header(vd);

	for (m = 1; m < zopt_metaslab_args; m++) {
	if (zopt_metaslab[m] < vd->vdev_ms_count)
	dump_metaslab(
	vd->vdev_ms[zopt_metaslab[m]]);
	else
	(void) fprintf(stderr, "bad metaslab "
	"number %llu\n",
	(u_longlong_t)zopt_metaslab[m]);
	}
	(void) printf("\n");
	return;
	}
	children = c + 1;
	}
	for (; c < children; c++) {
	vd = rvd->vdev_child[c];
	print_vdev_metaslab_header(vd);

	print_vdev_indirect(vd);

	for (m = 0; m < vd->vdev_ms_count; m++)
	dump_metaslab(vd->vdev_ms[m]);
	(void) printf("\n");
	}
	}

	static void
	dump_log_spacemaps(spa_t *spa)
	{
	if (!spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP))
	return;

	(void) printf("\nLog Space Maps in Pool:\n");
	for (spa_log_sm_t *sls = avl_first(&spa->spa_sm_logs_by_txg);
	sls; sls = AVL_NEXT(&spa->spa_sm_logs_by_txg, sls)) {
	space_map_t *sm = NULL;
	VERIFY0(space_map_open(&sm, spa_meta_objset(spa),
	sls->sls_sm_obj, 0, UINT64_MAX, SPA_MINBLOCKSHIFT));

	(void) printf("Log Spacemap object %llu txg %llu\n",
	(u_longlong_t)sls->sls_sm_obj, (u_longlong_t)sls->sls_txg);
	dump_spacemap(spa->spa_meta_objset, sm);
	space_map_close(sm);
	}
	(void) printf("\n");
	}

	static void
	dump_dde(const ddt_t ddt, const ddt_entry_t dde, uint64_t index)
	{
	const ddt_phys_t *ddp = dde->dde_phys;
	const ddt_key_t *ddk = &dde->dde_key;
	const char *types[4] = { "ditto", "single", "double", "triple" };
	char blkbuf[BP_SPRINTF_LEN];
	blkptr_t blk;
	int p;

	for (p = 0; p < DDT_PHYS_TYPES; p++, ddp++) {
	if (ddp->ddp_phys_birth == 0)
	continue;
	ddt_bp_create(ddt->ddt_checksum, ddk, ddp, &blk);
	snprintf_blkptr(blkbuf, sizeof (blkbuf), &blk);
	(void) printf("index %llx refcnt %llu %s %s\n",
	(u_longlong_t)index, (u_longlong_t)ddp->ddp_refcnt,
	types[p], blkbuf);
	}
	}

	static void
	dump_dedup_ratio(const ddt_stat_t *dds)
	{
	double rL, rP, rD, D, dedup, compress, copies;

	if (dds->dds_blocks == 0)
	return;

	rL = (double)dds->dds_ref_lsize;
	rP = (double)dds->dds_ref_psize;
	rD = (double)dds->dds_ref_dsize;
	D = (double)dds->dds_dsize;

	dedup = rD / D;
	compress = rL / rP;
	copies = rD / rP;

	(void) printf("dedup = %.2f, compress = %.2f, copies = %.2f, "
	"dedup * compress / copies = %.2f\n\n",
	dedup, compress, copies, dedup * compress / copies);
	}

	static void
	dump_ddt(ddt_t *ddt, enum ddt_type type, enum ddt_class class)
	{
	char name[DDT_NAMELEN];
	ddt_entry_t dde;
	uint64_t walk = 0;
	dmu_object_info_t doi;
	uint64_t count, dspace, mspace;
	int error;

	error = ddt_object_info(ddt, type, class, &doi);

	if (error == ENOENT)
	return;
	ASSERT(error == 0);

	error = ddt_object_count(ddt, type, class, &count);
	ASSERT(error == 0);
	if (count == 0)
	return;

	dspace = doi.doi_physical_blocks_512 << 9;
	mspace = doi.doi_fill_count * doi.doi_data_block_size;

	ddt_object_name(ddt, type, class, name);

	(void) printf("%s: %llu entries, size %llu on disk, %llu in core\n",
	name,
	(u_longlong_t)count,
	(u_longlong_t)(dspace / count),
	(u_longlong_t)(mspace / count));

	if (dump_opt['D'] < 3)
	return;

	zpool_dump_ddt(NULL, &ddt->ddt_histogram[type][class]);

	if (dump_opt['D'] < 4)
	return;

	if (dump_opt['D'] < 5 && class == DDT_CLASS_UNIQUE)
	return;

	(void) printf("%s contents:\n\n", name);

	while ((error = ddt_object_walk(ddt, type, class, &walk, &dde)) == 0)
	dump_dde(ddt, &dde, walk);

	ASSERT3U(error, ==, ENOENT);

	(void) printf("\n");
	}

	static void
	dump_all_ddts(spa_t *spa)
	{
	ddt_histogram_t ddh_total;
	ddt_stat_t dds_total;

	bzero(&ddh_total, sizeof (ddh_total));
	bzero(&dds_total, sizeof (dds_total));

	for (enum zio_checksum c = 0; c < ZIO_CHECKSUM_FUNCTIONS; c++) {
	ddt_t *ddt = spa->spa_ddt[c];
	for (enum ddt_type type = 0; type < DDT_TYPES; type++) {
	for (enum ddt_class class = 0; class < DDT_CLASSES;
	class++) {
	dump_ddt(ddt, type, class);
	}
	}
	}

	ddt_get_dedup_stats(spa, &dds_total);

	if (dds_total.dds_blocks == 0) {
	(void) printf("All DDTs are empty\n");
	return;
	}

	(void) printf("\n");

	if (dump_opt['D'] > 1) {
	(void) printf("DDT histogram (aggregated over all DDTs):\n");
	ddt_get_dedup_histogram(spa, &ddh_total);
	zpool_dump_ddt(&dds_total, &ddh_total);
	}

	dump_dedup_ratio(&dds_total);
	}

	static void
	dump_dtl_seg(void *arg, uint64_t start, uint64_t size)
	{
	char *prefix = arg;

	(void) printf("%s [%llu,%llu) length %llu\n",
	prefix,
	(u_longlong_t)start,
	(u_longlong_t)(start + size),
	(u_longlong_t)(size));
	}

	static void
	dump_dtl(vdev_t *vd, int indent)
	{
	spa_t *spa = vd->vdev_spa;
	boolean_t required;
	const char *name[DTL_TYPES] = { "missing", "partial", "scrub",
	"outage" };
	char prefix[256];

	spa_vdev_state_enter(spa, SCL_NONE);
	required = vdev_dtl_required(vd);
	(void) spa_vdev_state_exit(spa, NULL, 0);

	if (indent == 0)
	(void) printf("\nDirty time logs:\n\n");

	(void) printf("\t%*s%s [%s]\n", indent, "",
	vd->vdev_path ? vd->vdev_path :
	vd->vdev_parent ? vd->vdev_ops->vdev_op_type : spa_name(spa),
	required ? "DTL-required" : "DTL-expendable");

	for (int t = 0; t < DTL_TYPES; t++) {
	range_tree_t *rt = vd->vdev_dtl[t];
	if (range_tree_space(rt) == 0)
	continue;
	(void) snprintf(prefix, sizeof (prefix), "\t%*s%s",
	indent + 2, "", name[t]);
	range_tree_walk(rt, dump_dtl_seg, prefix);
	if (dump_opt['d'] > 5 && vd->vdev_children == 0)
	dump_spacemap(spa->spa_meta_objset,
	vd->vdev_dtl_sm);
	}

	for (unsigned c = 0; c < vd->vdev_children; c++)
	dump_dtl(vd->vdev_child[c], indent + 4);
	}

	static void
	dump_history(spa_t *spa)
	{
	nvlist_t **events = NULL;
	char *buf;
	uint64_t resid, len, off = 0;
	uint_t num = 0;
	int error;
	char tbuf[30];

	if ((buf = malloc(SPA_OLD_MAXBLOCKSIZE)) == NULL) {
	(void) fprintf(stderr, "%s: unable to allocate I/O buffer\n",
	__func__);
	return;
	}

	do {
	len = SPA_OLD_MAXBLOCKSIZE;

	if ((error = spa_history_get(spa, &off, &len, buf)) != 0) {
	(void) fprintf(stderr, "Unable to read history: "
	"error %d\n", error);
	free(buf);
	return;
	}

	if (zpool_history_unpack(buf, len, &resid, &events, &num) != 0)
	break;

	off -= resid;
	} while (len != 0);

	(void) printf("\nHistory:\n");
	for (unsigned i = 0; i < num; i++) {
	boolean_t printed = B_FALSE;

	if (nvlist_exists(events[i], ZPOOL_HIST_TIME)) {
	time_t tsec;
	struct tm t;

	tsec = fnvlist_lookup_uint64(events[i],
	ZPOOL_HIST_TIME);
	(void) localtime_r(&tsec, &t);
	(void) strftime(tbuf, sizeof (tbuf), "%F.%T", &t);
	} else {
	tbuf[0] = '\0';
	}

	if (nvlist_exists(events[i], ZPOOL_HIST_CMD)) {
	(void) printf("%s %s\n", tbuf,
	fnvlist_lookup_string(events[i], ZPOOL_HIST_CMD));
	} else if (nvlist_exists(events[i], ZPOOL_HIST_INT_EVENT)) {
	uint64_t ievent;

	ievent = fnvlist_lookup_uint64(events[i],
	ZPOOL_HIST_INT_EVENT);
	if (ievent >= ZFS_NUM_LEGACY_HISTORY_EVENTS)
	goto next;

	(void) printf(" %s [internal %s txg:%ju] %s\n",
	tbuf,
	zfs_history_event_names[ievent],
	fnvlist_lookup_uint64(events[i],
	ZPOOL_HIST_TXG),
	fnvlist_lookup_string(events[i],
	ZPOOL_HIST_INT_STR));
	} else if (nvlist_exists(events[i], ZPOOL_HIST_INT_NAME)) {
	(void) printf("%s [txg:%ju] %s", tbuf,
	fnvlist_lookup_uint64(events[i],
	ZPOOL_HIST_TXG),
	fnvlist_lookup_string(events[i],
	ZPOOL_HIST_INT_NAME));

	if (nvlist_exists(events[i], ZPOOL_HIST_DSNAME)) {
	(void) printf(" %s (%llu)",
	fnvlist_lookup_string(events[i],
	ZPOOL_HIST_DSNAME),
	(u_longlong_t)fnvlist_lookup_uint64(
	events[i],
	ZPOOL_HIST_DSID));
	}

	(void) printf(" %s\n", fnvlist_lookup_string(events[i],
	ZPOOL_HIST_INT_STR));
	} else if (nvlist_exists(events[i], ZPOOL_HIST_IOCTL)) {
	(void) printf("%s ioctl %s\n", tbuf,
	fnvlist_lookup_string(events[i],
	ZPOOL_HIST_IOCTL));

	if (nvlist_exists(events[i], ZPOOL_HIST_INPUT_NVL)) {
	(void) printf(" input:\n");
	dump_nvlist(fnvlist_lookup_nvlist(events[i],
	ZPOOL_HIST_INPUT_NVL), 8);
	}
	if (nvlist_exists(events[i], ZPOOL_HIST_OUTPUT_NVL)) {
	(void) printf(" output:\n");
	dump_nvlist(fnvlist_lookup_nvlist(events[i],
	ZPOOL_HIST_OUTPUT_NVL), 8);
	}
	if (nvlist_exists(events[i], ZPOOL_HIST_ERRNO)) {
	(void) printf(" errno: %lld\n",
	(longlong_t)fnvlist_lookup_int64(events[i],
	ZPOOL_HIST_ERRNO));
	}
	} else {
	goto next;
	}

	printed = B_TRUE;
	next:
	if (dump_opt['h'] > 1) {
	if (!printed)
	(void) printf("unrecognized record:\n");
	dump_nvlist(events[i], 2);
	}
	}
	free(buf);
	}

	/ARGSUSED/
	static void
	dump_dnode(objset_t os, uint64_t object, void data, size_t size)
	{
	}

	static uint64_t
	blkid2offset(const dnode_phys_t dnp, const blkptr_t bp,
	const zbookmark_phys_t *zb)
	{
	if (dnp == NULL) {
	ASSERT(zb->zb_level < 0);
	if (zb->zb_object == 0)
	return (zb->zb_blkid);
	return (zb->zb_blkid * BP_GET_LSIZE(bp));
	}

	ASSERT(zb->zb_level >= 0);

	return ((zb->zb_blkid <<
	(zb->zb_level * (dnp->dn_indblkshift - SPA_BLKPTRSHIFT))) *
	dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT);
	}

	static void
	snprintf_zstd_header(spa_t spa, char blkbuf, size_t buflen,
	const blkptr_t *bp)
	{
	abd_t *pabd;
	void *buf;
	zio_t *zio;
	zfs_zstdhdr_t zstd_hdr;
	int error;

	if (BP_GET_COMPRESS(bp) != ZIO_COMPRESS_ZSTD)
	return;

	if (BP_IS_HOLE(bp))
	return;

	if (BP_IS_EMBEDDED(bp)) {
	buf = malloc(SPA_MAXBLOCKSIZE);
	if (buf == NULL) {
	(void) fprintf(stderr, "out of memory\n");
	exit(1);
	}
	decode_embedded_bp_compressed(bp, buf);
	memcpy(&zstd_hdr, buf, sizeof (zstd_hdr));
	free(buf);
	zstd_hdr.c_len = BE_32(zstd_hdr.c_len);
	zstd_hdr.raw_version_level = BE_32(zstd_hdr.raw_version_level);
	(void) snprintf(blkbuf + strlen(blkbuf),
	buflen - strlen(blkbuf),
	" ZSTD:size=%u:version=%u:level=%u:EMBEDDED",
	zstd_hdr.c_len, zfs_get_hdrversion(&zstd_hdr),
	zfs_get_hdrlevel(&zstd_hdr));
	return;
	}

	pabd = abd_alloc_for_io(SPA_MAXBLOCKSIZE, B_FALSE);
	zio = zio_root(spa, NULL, NULL, 0);

	/* Decrypt but don't decompress so we can read the compression header */
	zio_nowait(zio_read(zio, spa, bp, pabd, BP_GET_PSIZE(bp), NULL, NULL,
	ZIO_PRIORITY_SYNC_READ, ZIO_FLAG_CANFAIL \| ZIO_FLAG_RAW_COMPRESS,
	NULL));
	error = zio_wait(zio);
	if (error) {
	(void) fprintf(stderr, "read failed: %d\n", error);
	return;
	}
	buf = abd_borrow_buf_copy(pabd, BP_GET_LSIZE(bp));
	memcpy(&zstd_hdr, buf, sizeof (zstd_hdr));
	zstd_hdr.c_len = BE_32(zstd_hdr.c_len);
	zstd_hdr.raw_version_level = BE_32(zstd_hdr.raw_version_level);

	(void) snprintf(blkbuf + strlen(blkbuf),
	buflen - strlen(blkbuf),
	" ZSTD:size=%u:version=%u:level=%u:NORMAL",
	zstd_hdr.c_len, zfs_get_hdrversion(&zstd_hdr),
	zfs_get_hdrlevel(&zstd_hdr));

	abd_return_buf_copy(pabd, buf, BP_GET_LSIZE(bp));
	}

	static void
	snprintf_blkptr_compact(char blkbuf, size_t buflen, const blkptr_t bp,
	boolean_t bp_freed)
	{
	const dva_t *dva = bp->blk_dva;
	int ndvas = dump_opt['d'] > 5 ? BP_GET_NDVAS(bp) : 1;
	int i;

	if (dump_opt['b'] >= 6) {
	snprintf_blkptr(blkbuf, buflen, bp);
	if (bp_freed) {
	(void) snprintf(blkbuf + strlen(blkbuf),
	buflen - strlen(blkbuf), " %s", "FREE");
	}
	return;
	}

	if (BP_IS_EMBEDDED(bp)) {
	(void) sprintf(blkbuf,
	"EMBEDDED et=%u %llxL/%llxP B=%llu",
	(int)BPE_GET_ETYPE(bp),
	(u_longlong_t)BPE_GET_LSIZE(bp),
	(u_longlong_t)BPE_GET_PSIZE(bp),
	(u_longlong_t)bp->blk_birth);
	return;
	}

	blkbuf[0] = '\0';

	for (i = 0; i < ndvas; i++)
	(void) snprintf(blkbuf + strlen(blkbuf),
	buflen - strlen(blkbuf), "%llu:%llx:%llx ",
	(u_longlong_t)DVA_GET_VDEV(&dva[i]),
	(u_longlong_t)DVA_GET_OFFSET(&dva[i]),
	(u_longlong_t)DVA_GET_ASIZE(&dva[i]));

	if (BP_IS_HOLE(bp)) {
	(void) snprintf(blkbuf + strlen(blkbuf),
	buflen - strlen(blkbuf),
	"%llxL B=%llu",
	(u_longlong_t)BP_GET_LSIZE(bp),
	(u_longlong_t)bp->blk_birth);
	} else {
	(void) snprintf(blkbuf + strlen(blkbuf),
	buflen - strlen(blkbuf),
	"%llxL/%llxP F=%llu B=%llu/%llu",
	(u_longlong_t)BP_GET_LSIZE(bp),
	(u_longlong_t)BP_GET_PSIZE(bp),
	(u_longlong_t)BP_GET_FILL(bp),
	(u_longlong_t)bp->blk_birth,
	(u_longlong_t)BP_PHYSICAL_BIRTH(bp));
	if (bp_freed)
	(void) snprintf(blkbuf + strlen(blkbuf),
	buflen - strlen(blkbuf), " %s", "FREE");
	(void) snprintf(blkbuf + strlen(blkbuf),
	buflen - strlen(blkbuf), " cksum=%llx:%llx:%llx:%llx",
	(u_longlong_t)bp->blk_cksum.zc_word[0],
	(u_longlong_t)bp->blk_cksum.zc_word[1],
	(u_longlong_t)bp->blk_cksum.zc_word[2],
	(u_longlong_t)bp->blk_cksum.zc_word[3]);
	}
	}

	static void
	print_indirect(spa_t spa, blkptr_t bp, const zbookmark_phys_t *zb,
	const dnode_phys_t *dnp)
	{
	char blkbuf[BP_SPRINTF_LEN];
	int l;

	if (!BP_IS_EMBEDDED(bp)) {
	ASSERT3U(BP_GET_TYPE(bp), ==, dnp->dn_type);
	ASSERT3U(BP_GET_LEVEL(bp), ==, zb->zb_level);
	}

	(void) printf("%16llx ", (u_longlong_t)blkid2offset(dnp, bp, zb));

	ASSERT(zb->zb_level >= 0);

	for (l = dnp->dn_nlevels - 1; l >= -1; l--) {
	if (l == zb->zb_level) {
	(void) printf("L%llx", (u_longlong_t)zb->zb_level);
	} else {
	(void) printf(" ");
	}
	}

	snprintf_blkptr_compact(blkbuf, sizeof (blkbuf), bp, B_FALSE);
	if (dump_opt['Z'] && BP_GET_COMPRESS(bp) == ZIO_COMPRESS_ZSTD)
	snprintf_zstd_header(spa, blkbuf, sizeof (blkbuf), bp);
	(void) printf("%s\n", blkbuf);
	}

	static int
	visit_indirect(spa_t spa, const dnode_phys_t dnp,
	blkptr_t bp, const zbookmark_phys_t zb)
	{
	int err = 0;

	if (bp->blk_birth == 0)
	return (0);

	print_indirect(spa, bp, zb, dnp);

	if (BP_GET_LEVEL(bp) > 0 && !BP_IS_HOLE(bp)) {
	arc_flags_t flags = ARC_FLAG_WAIT;
	int i;
	blkptr_t *cbp;
	int epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT;
	arc_buf_t *buf;
	uint64_t fill = 0;
	ASSERT(!BP_IS_REDACTED(bp));

	err = arc_read(NULL, spa, bp, arc_getbuf_func, &buf,
	ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL, &flags, zb);
	if (err)
	return (err);
	ASSERT(buf->b_data);

	/* recursively visit blocks below this */
	cbp = buf->b_data;
	for (i = 0; i < epb; i++, cbp++) {
	zbookmark_phys_t czb;

	SET_BOOKMARK(&czb, zb->zb_objset, zb->zb_object,
	zb->zb_level - 1,
	zb->zb_blkid * epb + i);
	err = visit_indirect(spa, dnp, cbp, &czb);
	if (err)
	break;
	fill += BP_GET_FILL(cbp);
	}
	if (!err)
	ASSERT3U(fill, ==, BP_GET_FILL(bp));
	arc_buf_destroy(buf, &buf);
	}

	return (err);
	}

	/ARGSUSED/
	static void
	dump_indirect(dnode_t *dn)
	{
	dnode_phys_t *dnp = dn->dn_phys;
	int j;
	zbookmark_phys_t czb;

	(void) printf("Indirect blocks:\n");

	SET_BOOKMARK(&czb, dmu_objset_id(dn->dn_objset),
	dn->dn_object, dnp->dn_nlevels - 1, 0);
	for (j = 0; j < dnp->dn_nblkptr; j++) {
	czb.zb_blkid = j;
	(void) visit_indirect(dmu_objset_spa(dn->dn_objset), dnp,
	&dnp->dn_blkptr[j], &czb);
	}

	(void) printf("\n");
	}

	/ARGSUSED/
	static void
	dump_dsl_dir(objset_t os, uint64_t object, void data, size_t size)
	{
	dsl_dir_phys_t *dd = data;
	time_t crtime;
	char nice[32];

	/* make sure nicenum has enough space */
	CTASSERT(sizeof (nice) >= NN_NUMBUF_SZ);

	if (dd == NULL)
	return;

	ASSERT3U(size, >=, sizeof (dsl_dir_phys_t));

	crtime = dd->dd_creation_time;
	(void) printf("\t\tcreation_time = %s", ctime(&crtime));
	(void) printf("\t\thead_dataset_obj = %llu\n",
	(u_longlong_t)dd->dd_head_dataset_obj);
	(void) printf("\t\tparent_dir_obj = %llu\n",
	(u_longlong_t)dd->dd_parent_obj);
	(void) printf("\t\torigin_obj = %llu\n",
	(u_longlong_t)dd->dd_origin_obj);
	(void) printf("\t\tchild_dir_zapobj = %llu\n",
	(u_longlong_t)dd->dd_child_dir_zapobj);
	zdb_nicenum(dd->dd_used_bytes, nice, sizeof (nice));
	(void) printf("\t\tused_bytes = %s\n", nice);
	zdb_nicenum(dd->dd_compressed_bytes, nice, sizeof (nice));
	(void) printf("\t\tcompressed_bytes = %s\n", nice);
	zdb_nicenum(dd->dd_uncompressed_bytes, nice, sizeof (nice));
	(void) printf("\t\tuncompressed_bytes = %s\n", nice);
	zdb_nicenum(dd->dd_quota, nice, sizeof (nice));
	(void) printf("\t\tquota = %s\n", nice);
	zdb_nicenum(dd->dd_reserved, nice, sizeof (nice));
	(void) printf("\t\treserved = %s\n", nice);
	(void) printf("\t\tprops_zapobj = %llu\n",
	(u_longlong_t)dd->dd_props_zapobj);
	(void) printf("\t\tdeleg_zapobj = %llu\n",
	(u_longlong_t)dd->dd_deleg_zapobj);
	(void) printf("\t\tflags = %llx\n",
	(u_longlong_t)dd->dd_flags);

	#define DO(which) \
	zdb_nicenum(dd->dd_used_breakdown[DD_USED_ ## which], nice, \
	sizeof (nice)); \
	(void) printf("\t\tused_breakdown[" #which "] = %s\n", nice)
	DO(HEAD);
	DO(SNAP);
	DO(CHILD);
	DO(CHILD_RSRV);
	DO(REFRSRV);
	#undef DO
	(void) printf("\t\tclones = %llu\n",
	(u_longlong_t)dd->dd_clones);
	}

	/ARGSUSED/
	static void
	dump_dsl_dataset(objset_t os, uint64_t object, void data, size_t size)
	{
	dsl_dataset_phys_t *ds = data;
	time_t crtime;
	char used[32], compressed[32], uncompressed[32], unique[32];
	char blkbuf[BP_SPRINTF_LEN];

	/* make sure nicenum has enough space */
	CTASSERT(sizeof (used) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (compressed) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (uncompressed) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (unique) >= NN_NUMBUF_SZ);

	if (ds == NULL)
	return;

	ASSERT(size == sizeof (*ds));
	crtime = ds->ds_creation_time;
	zdb_nicenum(ds->ds_referenced_bytes, used, sizeof (used));
	zdb_nicenum(ds->ds_compressed_bytes, compressed, sizeof (compressed));
	zdb_nicenum(ds->ds_uncompressed_bytes, uncompressed,
	sizeof (uncompressed));
	zdb_nicenum(ds->ds_unique_bytes, unique, sizeof (unique));
	snprintf_blkptr(blkbuf, sizeof (blkbuf), &ds->ds_bp);

	(void) printf("\t\tdir_obj = %llu\n",
	(u_longlong_t)ds->ds_dir_obj);
	(void) printf("\t\tprev_snap_obj = %llu\n",
	(u_longlong_t)ds->ds_prev_snap_obj);
	(void) printf("\t\tprev_snap_txg = %llu\n",
	(u_longlong_t)ds->ds_prev_snap_txg);
	(void) printf("\t\tnext_snap_obj = %llu\n",
	(u_longlong_t)ds->ds_next_snap_obj);
	(void) printf("\t\tsnapnames_zapobj = %llu\n",
	(u_longlong_t)ds->ds_snapnames_zapobj);
	(void) printf("\t\tnum_children = %llu\n",
	(u_longlong_t)ds->ds_num_children);
	(void) printf("\t\tuserrefs_obj = %llu\n",
	(u_longlong_t)ds->ds_userrefs_obj);
	(void) printf("\t\tcreation_time = %s", ctime(&crtime));
	(void) printf("\t\tcreation_txg = %llu\n",
	(u_longlong_t)ds->ds_creation_txg);
	(void) printf("\t\tdeadlist_obj = %llu\n",
	(u_longlong_t)ds->ds_deadlist_obj);
	(void) printf("\t\tused_bytes = %s\n", used);
	(void) printf("\t\tcompressed_bytes = %s\n", compressed);
	(void) printf("\t\tuncompressed_bytes = %s\n", uncompressed);
	(void) printf("\t\tunique = %s\n", unique);
	(void) printf("\t\tfsid_guid = %llu\n",
	(u_longlong_t)ds->ds_fsid_guid);
	(void) printf("\t\tguid = %llu\n",
	(u_longlong_t)ds->ds_guid);
	(void) printf("\t\tflags = %llx\n",
	(u_longlong_t)ds->ds_flags);
	(void) printf("\t\tnext_clones_obj = %llu\n",
	(u_longlong_t)ds->ds_next_clones_obj);
	(void) printf("\t\tprops_obj = %llu\n",
	(u_longlong_t)ds->ds_props_obj);
	(void) printf("\t\tbp = %s\n", blkbuf);
	}

	/* ARGSUSED */
	static int
	dump_bptree_cb(void arg, const blkptr_t bp, dmu_tx_t *tx)
	{
	char blkbuf[BP_SPRINTF_LEN];

	if (bp->blk_birth != 0) {
	snprintf_blkptr(blkbuf, sizeof (blkbuf), bp);
	(void) printf("\t%s\n", blkbuf);
	}
	return (0);
	}

	static void
	dump_bptree(objset_t os, uint64_t obj, const char name)
	{
	char bytes[32];
	bptree_phys_t *bt;
	dmu_buf_t *db;

	/* make sure nicenum has enough space */
	CTASSERT(sizeof (bytes) >= NN_NUMBUF_SZ);

	if (dump_opt['d'] < 3)
	return;

	VERIFY3U(0, ==, dmu_bonus_hold(os, obj, FTAG, &db));
	bt = db->db_data;
	zdb_nicenum(bt->bt_bytes, bytes, sizeof (bytes));
	(void) printf("\n %s: %llu datasets, %s\n",
	name, (unsigned long long)(bt->bt_end - bt->bt_begin), bytes);
	dmu_buf_rele(db, FTAG);

	if (dump_opt['d'] < 5)
	return;

	(void) printf("\n");

	(void) bptree_iterate(os, obj, B_FALSE, dump_bptree_cb, NULL, NULL);
	}

	/* ARGSUSED */
	static int
	dump_bpobj_cb(void arg, const blkptr_t bp, boolean_t bp_freed, dmu_tx_t *tx)
	{
	char blkbuf[BP_SPRINTF_LEN];

	ASSERT(bp->blk_birth != 0);
	snprintf_blkptr_compact(blkbuf, sizeof (blkbuf), bp, bp_freed);
	(void) printf("\t%s\n", blkbuf);
	return (0);
	}

	static void
	dump_full_bpobj(bpobj_t bpo, const char name, int indent)
	{
	char bytes[32];
	char comp[32];
	char uncomp[32];
	uint64_t i;

	/* make sure nicenum has enough space */
	CTASSERT(sizeof (bytes) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (comp) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (uncomp) >= NN_NUMBUF_SZ);

	if (dump_opt['d'] < 3)
	return;

	zdb_nicenum(bpo->bpo_phys->bpo_bytes, bytes, sizeof (bytes));
	if (bpo->bpo_havesubobj && bpo->bpo_phys->bpo_subobjs != 0) {
	zdb_nicenum(bpo->bpo_phys->bpo_comp, comp, sizeof (comp));
	zdb_nicenum(bpo->bpo_phys->bpo_uncomp, uncomp, sizeof (uncomp));
	if (bpo->bpo_havefreed) {
	(void) printf(" %*s: object %llu, %llu local "
	"blkptrs, %llu freed, %llu subobjs in object %llu, "
	"%s (%s/%s comp)\n",
	indent * 8, name,
	(u_longlong_t)bpo->bpo_object,
	(u_longlong_t)bpo->bpo_phys->bpo_num_blkptrs,
	(u_longlong_t)bpo->bpo_phys->bpo_num_freed,
	(u_longlong_t)bpo->bpo_phys->bpo_num_subobjs,
	(u_longlong_t)bpo->bpo_phys->bpo_subobjs,
	bytes, comp, uncomp);
	} else {
	(void) printf(" %*s: object %llu, %llu local "
	"blkptrs, %llu subobjs in object %llu, "
	"%s (%s/%s comp)\n",
	indent * 8, name,
	(u_longlong_t)bpo->bpo_object,
	(u_longlong_t)bpo->bpo_phys->bpo_num_blkptrs,
	(u_longlong_t)bpo->bpo_phys->bpo_num_subobjs,
	(u_longlong_t)bpo->bpo_phys->bpo_subobjs,
	bytes, comp, uncomp);
	}

	for (i = 0; i < bpo->bpo_phys->bpo_num_subobjs; i++) {
	uint64_t subobj;
	bpobj_t subbpo;
	int error;
	VERIFY0(dmu_read(bpo->bpo_os,
	bpo->bpo_phys->bpo_subobjs,
	i * sizeof (subobj), sizeof (subobj), &subobj, 0));
	error = bpobj_open(&subbpo, bpo->bpo_os, subobj);
	if (error != 0) {
	(void) printf("ERROR %u while trying to open "
	"subobj id %llu\n",
	error, (u_longlong_t)subobj);
	continue;
	}
	dump_full_bpobj(&subbpo, "subobj", indent + 1);
	bpobj_close(&subbpo);
	}
	} else {
	if (bpo->bpo_havefreed) {
	(void) printf(" %*s: object %llu, %llu blkptrs, "
	"%llu freed, %s\n",
	indent * 8, name,
	(u_longlong_t)bpo->bpo_object,
	(u_longlong_t)bpo->bpo_phys->bpo_num_blkptrs,
	(u_longlong_t)bpo->bpo_phys->bpo_num_freed,
	bytes);
	} else {
	(void) printf(" %*s: object %llu, %llu blkptrs, "
	"%s\n",
	indent * 8, name,
	(u_longlong_t)bpo->bpo_object,
	(u_longlong_t)bpo->bpo_phys->bpo_num_blkptrs,
	bytes);
	}
	}

	if (dump_opt['d'] < 5)
	return;


	if (indent == 0) {
	(void) bpobj_iterate_nofree(bpo, dump_bpobj_cb, NULL, NULL);
	(void) printf("\n");
	}
	}

	static int
	dump_bookmark(dsl_pool_t dp, char name, boolean_t print_redact,
	boolean_t print_list)
	{
	int err = 0;
	zfs_bookmark_phys_t prop;
	objset_t *mos = dp->dp_spa->spa_meta_objset;
	err = dsl_bookmark_lookup(dp, name, NULL, &prop);

	if (err != 0) {
	return (err);
	}

	(void) printf("\t#%s: ", strchr(name, '#') + 1);
	(void) printf("{guid: %llx creation_txg: %llu creation_time: "
	"%llu redaction_obj: %llu}\n", (u_longlong_t)prop.zbm_guid,
	(u_longlong_t)prop.zbm_creation_txg,
	(u_longlong_t)prop.zbm_creation_time,
	(u_longlong_t)prop.zbm_redaction_obj);

	IMPLY(print_list, print_redact);
	if (!print_redact \|\| prop.zbm_redaction_obj == 0)
	return (0);

	redaction_list_t *rl;
	VERIFY0(dsl_redaction_list_hold_obj(dp,
	prop.zbm_redaction_obj, FTAG, &rl));

	redaction_list_phys_t *rlp = rl->rl_phys;
	(void) printf("\tRedacted:\n\t\tProgress: ");
	if (rlp->rlp_last_object != UINT64_MAX \|\|
	rlp->rlp_last_blkid != UINT64_MAX) {
	(void) printf("%llu %llu (incomplete)\n",
	(u_longlong_t)rlp->rlp_last_object,
	(u_longlong_t)rlp->rlp_last_blkid);
	} else {
	(void) printf("complete\n");
	}
	(void) printf("\t\tSnapshots: [");
	for (unsigned int i = 0; i < rlp->rlp_num_snaps; i++) {
	if (i > 0)
	(void) printf(", ");
	(void) printf("%0llu",
	(u_longlong_t)rlp->rlp_snaps[i]);
	}
	(void) printf("]\n\t\tLength: %llu\n",
	(u_longlong_t)rlp->rlp_num_entries);

	if (!print_list) {
	dsl_redaction_list_rele(rl, FTAG);
	return (0);
	}

	if (rlp->rlp_num_entries == 0) {
	dsl_redaction_list_rele(rl, FTAG);
	(void) printf("\t\tRedaction List: []\n\n");
	return (0);
	}

	redact_block_phys_t *rbp_buf;
	uint64_t size;
	dmu_object_info_t doi;

	VERIFY0(dmu_object_info(mos, prop.zbm_redaction_obj, &doi));
	size = doi.doi_max_offset;
	rbp_buf = kmem_alloc(size, KM_SLEEP);

	err = dmu_read(mos, prop.zbm_redaction_obj, 0, size,
	rbp_buf, 0);
	if (err != 0) {
	dsl_redaction_list_rele(rl, FTAG);
	kmem_free(rbp_buf, size);
	return (err);
	}

	(void) printf("\t\tRedaction List: [{object: %llx, offset: "
	"%llx, blksz: %x, count: %llx}",
	(u_longlong_t)rbp_buf[0].rbp_object,
	(u_longlong_t)rbp_buf[0].rbp_blkid,
	(uint_t)(redact_block_get_size(&rbp_buf[0])),
	(u_longlong_t)redact_block_get_count(&rbp_buf[0]));

	for (size_t i = 1; i < rlp->rlp_num_entries; i++) {
	(void) printf(",\n\t\t{object: %llx, offset: %llx, "
	"blksz: %x, count: %llx}",
	(u_longlong_t)rbp_buf[i].rbp_object,
	(u_longlong_t)rbp_buf[i].rbp_blkid,
	(uint_t)(redact_block_get_size(&rbp_buf[i])),
	(u_longlong_t)redact_block_get_count(&rbp_buf[i]));
	}
	dsl_redaction_list_rele(rl, FTAG);
	kmem_free(rbp_buf, size);
	(void) printf("]\n\n");
	return (0);
	}

	static void
	dump_bookmarks(objset_t *os, int verbosity)
	{
	zap_cursor_t zc;
	zap_attribute_t attr;
	dsl_dataset_t *ds = dmu_objset_ds(os);
	dsl_pool_t *dp = spa_get_dsl(os->os_spa);
	objset_t *mos = os->os_spa->spa_meta_objset;
	if (verbosity < 4)
	return;
	dsl_pool_config_enter(dp, FTAG);

	for (zap_cursor_init(&zc, mos, ds->ds_bookmarks_obj);
	zap_cursor_retrieve(&zc, &attr) == 0;
	zap_cursor_advance(&zc)) {
	char osname[ZFS_MAX_DATASET_NAME_LEN];
	char buf[ZFS_MAX_DATASET_NAME_LEN];
	dmu_objset_name(os, osname);
	VERIFY3S(0, <=, snprintf(buf, sizeof (buf), "%s#%s", osname,
	attr.za_name));
	(void) dump_bookmark(dp, buf, verbosity >= 5, verbosity >= 6);
	}
	zap_cursor_fini(&zc);
	dsl_pool_config_exit(dp, FTAG);
	}

	static void
	bpobj_count_refd(bpobj_t *bpo)
	{
	mos_obj_refd(bpo->bpo_object);

	if (bpo->bpo_havesubobj && bpo->bpo_phys->bpo_subobjs != 0) {
	mos_obj_refd(bpo->bpo_phys->bpo_subobjs);
	for (uint64_t i = 0; i < bpo->bpo_phys->bpo_num_subobjs; i++) {
	uint64_t subobj;
	bpobj_t subbpo;
	int error;
	VERIFY0(dmu_read(bpo->bpo_os,
	bpo->bpo_phys->bpo_subobjs,
	i * sizeof (subobj), sizeof (subobj), &subobj, 0));
	error = bpobj_open(&subbpo, bpo->bpo_os, subobj);
	if (error != 0) {
	(void) printf("ERROR %u while trying to open "
	"subobj id %llu\n",
	error, (u_longlong_t)subobj);
	continue;
	}
	bpobj_count_refd(&subbpo);
	bpobj_close(&subbpo);
	}
	}
	}

	static int
	dsl_deadlist_entry_count_refd(void arg, dsl_deadlist_entry_t dle)
	{
	spa_t *spa = arg;
	uint64_t empty_bpobj = spa->spa_dsl_pool->dp_empty_bpobj;
	if (dle->dle_bpobj.bpo_object != empty_bpobj)
	bpobj_count_refd(&dle->dle_bpobj);
	return (0);
	}

	static int
	dsl_deadlist_entry_dump(void arg, dsl_deadlist_entry_t dle)
	{
	ASSERT(arg == NULL);
	if (dump_opt['d'] >= 5) {
	char buf[128];
	(void) snprintf(buf, sizeof (buf),
	"mintxg %llu -> obj %llu",
	(longlong_t)dle->dle_mintxg,
	(longlong_t)dle->dle_bpobj.bpo_object);

	dump_full_bpobj(&dle->dle_bpobj, buf, 0);
	} else {
	(void) printf("mintxg %llu -> obj %llu\n",
	(longlong_t)dle->dle_mintxg,
	(longlong_t)dle->dle_bpobj.bpo_object);
	}
	return (0);
	}

	static void
	dump_blkptr_list(dsl_deadlist_t dl, char name)
	{
	char bytes[32];
	char comp[32];
	char uncomp[32];
	char entries[32];
	spa_t *spa = dmu_objset_spa(dl->dl_os);
	uint64_t empty_bpobj = spa->spa_dsl_pool->dp_empty_bpobj;

	if (dl->dl_oldfmt) {
	if (dl->dl_bpobj.bpo_object != empty_bpobj)
	bpobj_count_refd(&dl->dl_bpobj);
	} else {
	mos_obj_refd(dl->dl_object);
	dsl_deadlist_iterate(dl, dsl_deadlist_entry_count_refd, spa);
	}

	/* make sure nicenum has enough space */
	CTASSERT(sizeof (bytes) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (comp) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (uncomp) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (entries) >= NN_NUMBUF_SZ);

	if (dump_opt['d'] < 3)
	return;

	if (dl->dl_oldfmt) {
	dump_full_bpobj(&dl->dl_bpobj, "old-format deadlist", 0);
	return;
	}

	zdb_nicenum(dl->dl_phys->dl_used, bytes, sizeof (bytes));
	zdb_nicenum(dl->dl_phys->dl_comp, comp, sizeof (comp));
	zdb_nicenum(dl->dl_phys->dl_uncomp, uncomp, sizeof (uncomp));
	zdb_nicenum(avl_numnodes(&dl->dl_tree), entries, sizeof (entries));
	(void) printf("\n %s: %s (%s/%s comp), %s entries\n",
	name, bytes, comp, uncomp, entries);

	if (dump_opt['d'] < 4)
	return;

	(void) printf("\n");

	dsl_deadlist_iterate(dl, dsl_deadlist_entry_dump, NULL);
	}

	static int
	verify_dd_livelist(objset_t *os)
	{
	uint64_t ll_used, used, ll_comp, comp, ll_uncomp, uncomp;
	dsl_pool_t *dp = spa_get_dsl(os->os_spa);
	dsl_dir_t *dd = os->os_dsl_dataset->ds_dir;

	ASSERT(!dmu_objset_is_snapshot(os));
	if (!dsl_deadlist_is_open(&dd->dd_livelist))
	return (0);

	/* Iterate through the livelist to check for duplicates */
	dsl_deadlist_iterate(&dd->dd_livelist, sublivelist_verify_lightweight,
	NULL);

	dsl_pool_config_enter(dp, FTAG);
	dsl_deadlist_space(&dd->dd_livelist, &ll_used,
	&ll_comp, &ll_uncomp);

	dsl_dataset_t *origin_ds;
	ASSERT(dsl_pool_config_held(dp));
	VERIFY0(dsl_dataset_hold_obj(dp,
	dsl_dir_phys(dd)->dd_origin_obj, FTAG, &origin_ds));
	VERIFY0(dsl_dataset_space_written(origin_ds, os->os_dsl_dataset,
	&used, &comp, &uncomp));
	dsl_dataset_rele(origin_ds, FTAG);
	dsl_pool_config_exit(dp, FTAG);
	/*
	* It's possible that the dataset's uncomp space is larger than the
	* livelist's because livelists do not track embedded block pointers
	*/
	if (used != ll_used \|\| comp != ll_comp \|\| uncomp < ll_uncomp) {
	char nice_used[32], nice_comp[32], nice_uncomp[32];
	(void) printf("Discrepancy in space accounting:\n");
	zdb_nicenum(used, nice_used, sizeof (nice_used));
	zdb_nicenum(comp, nice_comp, sizeof (nice_comp));
	zdb_nicenum(uncomp, nice_uncomp, sizeof (nice_uncomp));
	(void) printf("dir: used %s, comp %s, uncomp %s\n",
	nice_used, nice_comp, nice_uncomp);
	zdb_nicenum(ll_used, nice_used, sizeof (nice_used));
	zdb_nicenum(ll_comp, nice_comp, sizeof (nice_comp));
	zdb_nicenum(ll_uncomp, nice_uncomp, sizeof (nice_uncomp));
	(void) printf("livelist: used %s, comp %s, uncomp %s\n",
	nice_used, nice_comp, nice_uncomp);
	return (1);
	}
	return (0);
	}

	static avl_tree_t idx_tree;
	static avl_tree_t domain_tree;
	static boolean_t fuid_table_loaded;
	static objset_t *sa_os = NULL;
	static sa_attr_type_t *sa_attr_table = NULL;

	static int
	open_objset(const char path, void tag, objset_t **osp)
	{
	int err;
	uint64_t sa_attrs = 0;
	uint64_t version = 0;

	VERIFY3P(sa_os, ==, NULL);
	/*
	* We can't own an objset if it's redacted. Therefore, we do this
	* dance: hold the objset, then acquire a long hold on its dataset, then
	* release the pool (which is held as part of holding the objset).
	*/
	err = dmu_objset_hold(path, tag, osp);
	if (err != 0) {
	(void) fprintf(stderr, "failed to hold dataset '%s': %s\n",
	path, strerror(err));
	return (err);
	}
	dsl_dataset_long_hold(dmu_objset_ds(*osp), tag);
	dsl_pool_rele(dmu_objset_pool(*osp), tag);

	if (dmu_objset_type(osp) == DMU_OST_ZFS && !(osp)->os_encrypted) {
	(void) zap_lookup(*osp, MASTER_NODE_OBJ, ZPL_VERSION_STR,
	8, 1, &version);
	if (version >= ZPL_VERSION_SA) {
	(void) zap_lookup(*osp, MASTER_NODE_OBJ, ZFS_SA_ATTRS,
	8, 1, &sa_attrs);
	}
	err = sa_setup(*osp, sa_attrs, zfs_attr_table, ZPL_END,
	&sa_attr_table);
	if (err != 0) {
	(void) fprintf(stderr, "sa_setup failed: %s\n",
	strerror(err));
	dsl_dataset_long_rele(dmu_objset_ds(*osp), tag);
	dsl_dataset_rele(dmu_objset_ds(*osp), tag);
	*osp = NULL;
	}
	}
	sa_os = *osp;

	return (0);
	}

	static void
	close_objset(objset_t os, void tag)
	{
	VERIFY3P(os, ==, sa_os);
	if (os->os_sa != NULL)
	sa_tear_down(os);
	dsl_dataset_long_rele(dmu_objset_ds(os), tag);
	dsl_dataset_rele(dmu_objset_ds(os), tag);
	sa_attr_table = NULL;
	sa_os = NULL;
	}

	static void
	fuid_table_destroy(void)
	{
	if (fuid_table_loaded) {
	zfs_fuid_table_destroy(&idx_tree, &domain_tree);
	fuid_table_loaded = B_FALSE;
	}
	}

	/*
	* print uid or gid information.
	* For normal POSIX id just the id is printed in decimal format.
	* For CIFS files with FUID the fuid is printed in hex followed by
	* the domain-rid string.
	*/
	static void
	print_idstr(uint64_t id, const char *id_type)
	{
	if (FUID_INDEX(id)) {
	char *domain;

	domain = zfs_fuid_idx_domain(&idx_tree, FUID_INDEX(id));
	(void) printf("\t%s %llx [%s-%d]\n", id_type,
	(u_longlong_t)id, domain, (int)FUID_RID(id));
	} else {
	(void) printf("\t%s %llu\n", id_type, (u_longlong_t)id);
	}

	}

	static void
	dump_uidgid(objset_t *os, uint64_t uid, uint64_t gid)
	{
	uint32_t uid_idx, gid_idx;

	uid_idx = FUID_INDEX(uid);
	gid_idx = FUID_INDEX(gid);

	/* Load domain table, if not already loaded */
	if (!fuid_table_loaded && (uid_idx \|\| gid_idx)) {
	uint64_t fuid_obj;

	/* first find the fuid object. It lives in the master node */
	VERIFY(zap_lookup(os, MASTER_NODE_OBJ, ZFS_FUID_TABLES,
	8, 1, &fuid_obj) == 0);
	zfs_fuid_avl_tree_create(&idx_tree, &domain_tree);
	(void) zfs_fuid_table_load(os, fuid_obj,
	&idx_tree, &domain_tree);
	fuid_table_loaded = B_TRUE;
	}

	print_idstr(uid, "uid");
	print_idstr(gid, "gid");
	}

	static void
	dump_znode_sa_xattr(sa_handle_t *hdl)
	{
	nvlist_t *sa_xattr;
	nvpair_t *elem = NULL;
	int sa_xattr_size = 0;
	int sa_xattr_entries = 0;
	int error;
	char *sa_xattr_packed;

	error = sa_size(hdl, sa_attr_table[ZPL_DXATTR], &sa_xattr_size);
	if (error \|\| sa_xattr_size == 0)
	return;

	sa_xattr_packed = malloc(sa_xattr_size);
	if (sa_xattr_packed == NULL)
	return;

	error = sa_lookup(hdl, sa_attr_table[ZPL_DXATTR],
	sa_xattr_packed, sa_xattr_size);
	if (error) {
	free(sa_xattr_packed);
	return;
	}

	error = nvlist_unpack(sa_xattr_packed, sa_xattr_size, &sa_xattr, 0);
	if (error) {
	free(sa_xattr_packed);
	return;
	}

	while ((elem = nvlist_next_nvpair(sa_xattr, elem)) != NULL)
	sa_xattr_entries++;

	(void) printf("\tSA xattrs: %d bytes, %d entries\n\n",
	sa_xattr_size, sa_xattr_entries);
	while ((elem = nvlist_next_nvpair(sa_xattr, elem)) != NULL) {
	uchar_t *value;
	uint_t cnt, idx;

	(void) printf("\t\t%s = ", nvpair_name(elem));
	nvpair_value_byte_array(elem, &value, &cnt);
	for (idx = 0; idx < cnt; ++idx) {
	if (isprint(value[idx]))
	(void) putchar(value[idx]);
	else
	(void) printf("\\%3.3o", value[idx]);
	}
	(void) putchar('\n');
	}

	nvlist_free(sa_xattr);
	free(sa_xattr_packed);
	}

	static void
	dump_znode_symlink(sa_handle_t *hdl)
	{
	int sa_symlink_size = 0;
	char linktarget[MAXPATHLEN];
	linktarget[0] = '\0';
	int error;

	error = sa_size(hdl, sa_attr_table[ZPL_SYMLINK], &sa_symlink_size);
	if (error \|\| sa_symlink_size == 0) {
	return;
	}
	if (sa_lookup(hdl, sa_attr_table[ZPL_SYMLINK],
	&linktarget, sa_symlink_size) == 0)
	(void) printf("\ttarget %s\n", linktarget);
	}

	/ARGSUSED/
	static void
	dump_znode(objset_t os, uint64_t object, void data, size_t size)
	{
	char path[MAXPATHLEN * 2]; /* allow for xattr and failure prefix */
	sa_handle_t *hdl;
	uint64_t xattr, rdev, gen;
	uint64_t uid, gid, mode, fsize, parent, links;
	uint64_t pflags;
	uint64_t acctm[2], modtm[2], chgtm[2], crtm[2];
	time_t z_crtime, z_atime, z_mtime, z_ctime;
	sa_bulk_attr_t bulk[12];
	int idx = 0;
	int error;

	VERIFY3P(os, ==, sa_os);
	if (sa_handle_get(os, object, NULL, SA_HDL_PRIVATE, &hdl)) {
	(void) printf("Failed to get handle for SA znode\n");
	return;
	}

	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_UID], NULL, &uid, 8);
	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_GID], NULL, &gid, 8);
	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_LINKS], NULL,
	&links, 8);
	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_GEN], NULL, &gen, 8);
	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_MODE], NULL,
	&mode, 8);
	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_PARENT],
	NULL, &parent, 8);
	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_SIZE], NULL,
	&fsize, 8);
	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_ATIME], NULL,
	acctm, 16);
	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_MTIME], NULL,
	modtm, 16);
	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_CRTIME], NULL,
	crtm, 16);
	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_CTIME], NULL,
	chgtm, 16);
	SA_ADD_BULK_ATTR(bulk, idx, sa_attr_table[ZPL_FLAGS], NULL,
	&pflags, 8);

	if (sa_bulk_lookup(hdl, bulk, idx)) {
	(void) sa_handle_destroy(hdl);
	return;
	}

	z_crtime = (time_t)crtm[0];
	z_atime = (time_t)acctm[0];
	z_mtime = (time_t)modtm[0];
	z_ctime = (time_t)chgtm[0];

	if (dump_opt['d'] > 4) {
	error = zfs_obj_to_path(os, object, path, sizeof (path));
	if (error == ESTALE) {
	(void) snprintf(path, sizeof (path), "on delete queue");
	} else if (error != 0) {
	leaked_objects++;
	(void) snprintf(path, sizeof (path),
	"path not found, possibly leaked");
	}
	(void) printf("\tpath %s\n", path);
	}

	if (S_ISLNK(mode))
	dump_znode_symlink(hdl);
	dump_uidgid(os, uid, gid);
	(void) printf("\tatime %s", ctime(&z_atime));
	(void) printf("\tmtime %s", ctime(&z_mtime));
	(void) printf("\tctime %s", ctime(&z_ctime));
	(void) printf("\tcrtime %s", ctime(&z_crtime));
	(void) printf("\tgen %llu\n", (u_longlong_t)gen);
	(void) printf("\tmode %llo\n", (u_longlong_t)mode);
	(void) printf("\tsize %llu\n", (u_longlong_t)fsize);
	(void) printf("\tparent %llu\n", (u_longlong_t)parent);
	(void) printf("\tlinks %llu\n", (u_longlong_t)links);
	(void) printf("\tpflags %llx\n", (u_longlong_t)pflags);
	if (dmu_objset_projectquota_enabled(os) && (pflags & ZFS_PROJID)) {
	uint64_t projid;

	if (sa_lookup(hdl, sa_attr_table[ZPL_PROJID], &projid,
	sizeof (uint64_t)) == 0)
	(void) printf("\tprojid %llu\n", (u_longlong_t)projid);
	}
	if (sa_lookup(hdl, sa_attr_table[ZPL_XATTR], &xattr,
	sizeof (uint64_t)) == 0)
	(void) printf("\txattr %llu\n", (u_longlong_t)xattr);
	if (sa_lookup(hdl, sa_attr_table[ZPL_RDEV], &rdev,
	sizeof (uint64_t)) == 0)
	(void) printf("\trdev 0x%016llx\n", (u_longlong_t)rdev);
	dump_znode_sa_xattr(hdl);
	sa_handle_destroy(hdl);
	}

	/ARGSUSED/
	static void
	dump_acl(objset_t os, uint64_t object, void data, size_t size)
	{
	}

	/ARGSUSED/
	static void
	dump_dmu_objset(objset_t os, uint64_t object, void data, size_t size)
	{
	}

	static object_viewer_t *object_viewer[DMU_OT_NUMTYPES + 1] = {
	dump_none, /* unallocated */
	dump_zap, /* object directory */
	dump_uint64, /* object array */
	dump_none, /* packed nvlist */
	dump_packed_nvlist, /* packed nvlist size */
	dump_none, /* bpobj */
	dump_bpobj, /* bpobj header */
	dump_none, /* SPA space map header */
	dump_none, /* SPA space map */
	dump_none, /* ZIL intent log */
	dump_dnode, /* DMU dnode */
	dump_dmu_objset, /* DMU objset */
	dump_dsl_dir, /* DSL directory */
	dump_zap, /* DSL directory child map */
	dump_zap, /* DSL dataset snap map */
	dump_zap, /* DSL props */
	dump_dsl_dataset, /* DSL dataset */
	dump_znode, /* ZFS znode */
	dump_acl, /* ZFS V0 ACL */
	dump_uint8, /* ZFS plain file */
	dump_zpldir, /* ZFS directory */
	dump_zap, /* ZFS master node */
	dump_zap, /* ZFS delete queue */
	dump_uint8, /* zvol object */
	dump_zap, /* zvol prop */
	dump_uint8, /* other uint8[] */
	dump_uint64, /* other uint64[] */
	dump_zap, /* other ZAP */
	dump_zap, /* persistent error log */
	dump_uint8, /* SPA history */
	dump_history_offsets, /* SPA history offsets */
	dump_zap, /* Pool properties */
	dump_zap, /* DSL permissions */
	dump_acl, /* ZFS ACL */
	dump_uint8, /* ZFS SYSACL */
	dump_none, /* FUID nvlist */
	dump_packed_nvlist, /* FUID nvlist size */
	dump_zap, /* DSL dataset next clones */
	dump_zap, /* DSL scrub queue */
	dump_zap, /* ZFS user/group/project used */
	dump_zap, /* ZFS user/group/project quota */
	dump_zap, /* snapshot refcount tags */
	dump_ddt_zap, /* DDT ZAP object */
	dump_zap, /* DDT statistics */
	dump_znode, /* SA object */
	dump_zap, /* SA Master Node */
	dump_sa_attrs, /* SA attribute registration */
	dump_sa_layouts, /* SA attribute layouts */
	dump_zap, /* DSL scrub translations */
	dump_none, /* fake dedup BP */
	dump_zap, /* deadlist */
	dump_none, /* deadlist hdr */
	dump_zap, /* dsl clones */
	dump_bpobj_subobjs, /* bpobj subobjs */
	dump_unknown, /* Unknown type, must be last */
	};

	static boolean_t
	match_object_type(dmu_object_type_t obj_type, uint64_t flags)
	{
	boolean_t match = B_TRUE;

	switch (obj_type) {
	case DMU_OT_DIRECTORY_CONTENTS:
	if (!(flags & ZOR_FLAG_DIRECTORY))
	match = B_FALSE;
	break;
	case DMU_OT_PLAIN_FILE_CONTENTS:
	if (!(flags & ZOR_FLAG_PLAIN_FILE))
	match = B_FALSE;
	break;
	case DMU_OT_SPACE_MAP:
	if (!(flags & ZOR_FLAG_SPACE_MAP))
	match = B_FALSE;
	break;
	default:
	if (strcmp(zdb_ot_name(obj_type), "zap") == 0) {
	if (!(flags & ZOR_FLAG_ZAP))
	match = B_FALSE;
	break;
	}

	/*
	* If all bits except some of the supported flags are
	* set, the user combined the all-types flag (A) with
	* a negated flag to exclude some types (e.g. A-f to
	* show all object types except plain files).
	*/
	if ((flags \| ZOR_SUPPORTED_FLAGS) != ZOR_FLAG_ALL_TYPES)
	match = B_FALSE;

	break;
	}

	return (match);
	}

	static void
	dump_object(objset_t *os, uint64_t object, int verbosity,
	boolean_t print_header, uint64_t dnode_slots_used, uint64_t flags)
	{
	dmu_buf_t *db = NULL;
	dmu_object_info_t doi;
	dnode_t *dn;
	boolean_t dnode_held = B_FALSE;
	void *bonus = NULL;
	size_t bsize = 0;
	char iblk[32], dblk[32], lsize[32], asize[32], fill[32], dnsize[32];
	char bonus_size[32];
	char aux[50];
	int error;

	/* make sure nicenum has enough space */
	CTASSERT(sizeof (iblk) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (dblk) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (lsize) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (asize) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (bonus_size) >= NN_NUMBUF_SZ);

	if (*print_header) {
	(void) printf("\n%10s %3s %5s %5s %5s %6s %5s %6s %s\n",
	"Object", "lvl", "iblk", "dblk", "dsize", "dnsize",
	"lsize", "%full", "type");
	*print_header = 0;
	}

	if (object == 0) {
	dn = DMU_META_DNODE(os);
	dmu_object_info_from_dnode(dn, &doi);
	} else {
	/*
	* Encrypted datasets will have sensitive bonus buffers
	* encrypted. Therefore we cannot hold the bonus buffer and
	* must hold the dnode itself instead.
	*/
	error = dmu_object_info(os, object, &doi);
	if (error)
	fatal("dmu_object_info() failed, errno %u", error);

	if (os->os_encrypted &&
	DMU_OT_IS_ENCRYPTED(doi.doi_bonus_type)) {
	error = dnode_hold(os, object, FTAG, &dn);
	if (error)
	fatal("dnode_hold() failed, errno %u", error);
	dnode_held = B_TRUE;
	} else {
	error = dmu_bonus_hold(os, object, FTAG, &db);
	if (error)
	fatal("dmu_bonus_hold(%llu) failed, errno %u",
	object, error);
	bonus = db->db_data;
	bsize = db->db_size;
	dn = DB_DNODE((dmu_buf_impl_t *)db);
	}
	}

	/*
	* Default to showing all object types if no flags were specified.
	*/
	if (flags != 0 && flags != ZOR_FLAG_ALL_TYPES &&
	!match_object_type(doi.doi_type, flags))
	goto out;

	if (dnode_slots_used)
	*dnode_slots_used = doi.doi_dnodesize / DNODE_MIN_SIZE;

	zdb_nicenum(doi.doi_metadata_block_size, iblk, sizeof (iblk));
	zdb_nicenum(doi.doi_data_block_size, dblk, sizeof (dblk));
	zdb_nicenum(doi.doi_max_offset, lsize, sizeof (lsize));
	zdb_nicenum(doi.doi_physical_blocks_512 << 9, asize, sizeof (asize));
	zdb_nicenum(doi.doi_bonus_size, bonus_size, sizeof (bonus_size));
	zdb_nicenum(doi.doi_dnodesize, dnsize, sizeof (dnsize));
	(void) sprintf(fill, "%6.2f", 100.0 * doi.doi_fill_count *
	doi.doi_data_block_size / (object == 0 ? DNODES_PER_BLOCK : 1) /
	doi.doi_max_offset);

	aux[0] = '\0';

	if (doi.doi_checksum != ZIO_CHECKSUM_INHERIT \|\| verbosity >= 6) {
	(void) snprintf(aux + strlen(aux), sizeof (aux) - strlen(aux),
	" (K=%s)", ZDB_CHECKSUM_NAME(doi.doi_checksum));
	}

	if (doi.doi_compress == ZIO_COMPRESS_INHERIT &&
	ZIO_COMPRESS_HASLEVEL(os->os_compress) && verbosity >= 6) {
	const char *compname = NULL;
	if (zfs_prop_index_to_string(ZFS_PROP_COMPRESSION,
	ZIO_COMPRESS_RAW(os->os_compress, os->os_complevel),
	&compname) == 0) {
	(void) snprintf(aux + strlen(aux),
	sizeof (aux) - strlen(aux), " (Z=inherit=%s)",
	compname);
	} else {
	(void) snprintf(aux + strlen(aux),
	sizeof (aux) - strlen(aux),
	" (Z=inherit=%s-unknown)",
	ZDB_COMPRESS_NAME(os->os_compress));
	}
	} else if (doi.doi_compress == ZIO_COMPRESS_INHERIT && verbosity >= 6) {
	(void) snprintf(aux + strlen(aux), sizeof (aux) - strlen(aux),
	" (Z=inherit=%s)", ZDB_COMPRESS_NAME(os->os_compress));
	} else if (doi.doi_compress != ZIO_COMPRESS_INHERIT \|\| verbosity >= 6) {
	(void) snprintf(aux + strlen(aux), sizeof (aux) - strlen(aux),
	" (Z=%s)", ZDB_COMPRESS_NAME(doi.doi_compress));
	}

	(void) printf("%10lld %3u %5s %5s %5s %6s %5s %6s %s%s\n",
	(u_longlong_t)object, doi.doi_indirection, iblk, dblk,
	asize, dnsize, lsize, fill, zdb_ot_name(doi.doi_type), aux);

	if (doi.doi_bonus_type != DMU_OT_NONE && verbosity > 3) {
	(void) printf("%10s %3s %5s %5s %5s %5s %5s %6s %s\n",
	"", "", "", "", "", "", bonus_size, "bonus",
	zdb_ot_name(doi.doi_bonus_type));
	}

	if (verbosity >= 4) {
	(void) printf("\tdnode flags: %s%s%s%s\n",
	(dn->dn_phys->dn_flags & DNODE_FLAG_USED_BYTES) ?
	"USED_BYTES " : "",
	(dn->dn_phys->dn_flags & DNODE_FLAG_USERUSED_ACCOUNTED) ?
	"USERUSED_ACCOUNTED " : "",
	(dn->dn_phys->dn_flags & DNODE_FLAG_USEROBJUSED_ACCOUNTED) ?
	"USEROBJUSED_ACCOUNTED " : "",
	(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR) ?
	"SPILL_BLKPTR" : "");
	(void) printf("\tdnode maxblkid: %llu\n",
	(longlong_t)dn->dn_phys->dn_maxblkid);

	if (!dnode_held) {
	object_viewer[ZDB_OT_TYPE(doi.doi_bonus_type)](os,
	object, bonus, bsize);
	} else {
	(void) printf("\t\t(bonus encrypted)\n");
	}

	if (!os->os_encrypted \|\| !DMU_OT_IS_ENCRYPTED(doi.doi_type)) {
	object_viewer[ZDB_OT_TYPE(doi.doi_type)](os, object,
	NULL, 0);
	} else {
	(void) printf("\t\t(object encrypted)\n");
	}

	*print_header = B_TRUE;
	}

	if (verbosity >= 5)
	dump_indirect(dn);

	if (verbosity >= 5) {
	/*
	* Report the list of segments that comprise the object.
	*/
	uint64_t start = 0;
	uint64_t end;
	uint64_t blkfill = 1;
	int minlvl = 1;

	if (dn->dn_type == DMU_OT_DNODE) {
	minlvl = 0;
	blkfill = DNODES_PER_BLOCK;
	}

	for (;;) {
	char segsize[32];
	/* make sure nicenum has enough space */
	CTASSERT(sizeof (segsize) >= NN_NUMBUF_SZ);
	error = dnode_next_offset(dn,
	0, &start, minlvl, blkfill, 0);
	if (error)
	break;
	end = start;
	error = dnode_next_offset(dn,
	DNODE_FIND_HOLE, &end, minlvl, blkfill, 0);
	zdb_nicenum(end - start, segsize, sizeof (segsize));
	(void) printf("\t\tsegment [%016llx, %016llx)"
	" size %5s\n", (u_longlong_t)start,
	(u_longlong_t)end, segsize);
	if (error)
	break;
	start = end;
	}
	}

	out:
	if (db != NULL)
	dmu_buf_rele(db, FTAG);
	if (dnode_held)
	dnode_rele(dn, FTAG);
	}

	static void
	count_dir_mos_objects(dsl_dir_t *dd)
	{
	mos_obj_refd(dd->dd_object);
	mos_obj_refd(dsl_dir_phys(dd)->dd_child_dir_zapobj);
	mos_obj_refd(dsl_dir_phys(dd)->dd_deleg_zapobj);
	mos_obj_refd(dsl_dir_phys(dd)->dd_props_zapobj);
	mos_obj_refd(dsl_dir_phys(dd)->dd_clones);

	/*
	* The dd_crypto_obj can be referenced by multiple dsl_dir's.
	* Ignore the references after the first one.
	*/
	mos_obj_refd_multiple(dd->dd_crypto_obj);
	}

	static void
	count_ds_mos_objects(dsl_dataset_t *ds)
	{
	mos_obj_refd(ds->ds_object);
	mos_obj_refd(dsl_dataset_phys(ds)->ds_next_clones_obj);
	mos_obj_refd(dsl_dataset_phys(ds)->ds_props_obj);
	mos_obj_refd(dsl_dataset_phys(ds)->ds_userrefs_obj);
	mos_obj_refd(dsl_dataset_phys(ds)->ds_snapnames_zapobj);
	mos_obj_refd(ds->ds_bookmarks_obj);

	if (!dsl_dataset_is_snapshot(ds)) {
	count_dir_mos_objects(ds->ds_dir);
	}
	}

	static const char *objset_types[DMU_OST_NUMTYPES] = {
	"NONE", "META", "ZPL", "ZVOL", "OTHER", "ANY" };

	/*
	* Parse a string denoting a range of object IDs of the form
	* <start>[:<end>[:flags]], and store the results in zor.
	* Return 0 on success. On error, return 1 and update the msg
	* pointer to point to a descriptive error message.
	*/
	static int
	parse_object_range(char range, zopt_object_range_t zor, char **msg)
	{
	uint64_t flags = 0;
	char p, s, dup, flagstr, *tmp = NULL;
	size_t len;
	int i;
	int rc = 0;

	if (strchr(range, ':') == NULL) {
	zor->zor_obj_start = strtoull(range, &p, 0);
	if (*p != '\0') {
	*msg = "Invalid characters in object ID";
	rc = 1;
	}
	zor->zor_obj_start = ZDB_MAP_OBJECT_ID(zor->zor_obj_start);
	zor->zor_obj_end = zor->zor_obj_start;
	return (rc);
	}

	if (strchr(range, ':') == range) {
	*msg = "Invalid leading colon";
	rc = 1;
	return (rc);
	}

	len = strlen(range);
	if (range[len - 1] == ':') {
	*msg = "Invalid trailing colon";
	rc = 1;
	return (rc);
	}

	dup = strdup(range);
	s = strtok_r(dup, ":", &tmp);
	zor->zor_obj_start = strtoull(s, &p, 0);

	if (*p != '\0') {
	*msg = "Invalid characters in start object ID";
	rc = 1;
	goto out;
	}

	s = strtok_r(NULL, ":", &tmp);
	zor->zor_obj_end = strtoull(s, &p, 0);

	if (*p != '\0') {
	*msg = "Invalid characters in end object ID";
	rc = 1;
	goto out;
	}

	if (zor->zor_obj_start > zor->zor_obj_end) {
	*msg = "Start object ID may not exceed end object ID";
	rc = 1;
	goto out;
	}

	s = strtok_r(NULL, ":", &tmp);
	if (s == NULL) {
	zor->zor_flags = ZOR_FLAG_ALL_TYPES;
	goto out;
	} else if (strtok_r(NULL, ":", &tmp) != NULL) {
	*msg = "Invalid colon-delimited field after flags";
	rc = 1;
	goto out;
	}

	flagstr = s;
	for (i = 0; flagstr[i]; i++) {
	int bit;
	boolean_t negation = (flagstr[i] == '-');

	if (negation) {
	i++;
	if (flagstr[i] == '\0') {
	*msg = "Invalid trailing negation operator";
	rc = 1;
	goto out;
	}
	}
	bit = flagbits[(uchar_t)flagstr[i]];
	if (bit == 0) {
	*msg = "Invalid flag";
	rc = 1;
	goto out;
	}
	if (negation)
	flags &= ~bit;
	else
	flags \|= bit;
	}
	zor->zor_flags = flags;

	zor->zor_obj_start = ZDB_MAP_OBJECT_ID(zor->zor_obj_start);
	zor->zor_obj_end = ZDB_MAP_OBJECT_ID(zor->zor_obj_end);

	out:
	free(dup);
	return (rc);
	}

	static void
	dump_objset(objset_t *os)
	{
	dmu_objset_stats_t dds = { 0 };
	uint64_t object, object_count;
	uint64_t refdbytes, usedobjs, scratch;
	char numbuf[32];
	char blkbuf[BP_SPRINTF_LEN + 20];
	char osname[ZFS_MAX_DATASET_NAME_LEN];
	const char *type = "UNKNOWN";
	int verbosity = dump_opt['d'];
	boolean_t print_header;
	unsigned i;
	int error;
	uint64_t total_slots_used = 0;
	uint64_t max_slot_used = 0;
	uint64_t dnode_slots;
	uint64_t obj_start;
	uint64_t obj_end;
	uint64_t flags;

	/* make sure nicenum has enough space */
	CTASSERT(sizeof (numbuf) >= NN_NUMBUF_SZ);

	dsl_pool_config_enter(dmu_objset_pool(os), FTAG);
	dmu_objset_fast_stat(os, &dds);
	dsl_pool_config_exit(dmu_objset_pool(os), FTAG);

	print_header = B_TRUE;

	if (dds.dds_type < DMU_OST_NUMTYPES)
	type = objset_types[dds.dds_type];

	if (dds.dds_type == DMU_OST_META) {
	dds.dds_creation_txg = TXG_INITIAL;
	usedobjs = BP_GET_FILL(os->os_rootbp);
	refdbytes = dsl_dir_phys(os->os_spa->spa_dsl_pool->dp_mos_dir)->
	dd_used_bytes;
	} else {
	dmu_objset_space(os, &refdbytes, &scratch, &usedobjs, &scratch);
	}

	ASSERT3U(usedobjs, ==, BP_GET_FILL(os->os_rootbp));

	zdb_nicenum(refdbytes, numbuf, sizeof (numbuf));

	if (verbosity >= 4) {
	(void) snprintf(blkbuf, sizeof (blkbuf), ", rootbp ");
	(void) snprintf_blkptr(blkbuf + strlen(blkbuf),
	sizeof (blkbuf) - strlen(blkbuf), os->os_rootbp);
	} else {
	blkbuf[0] = '\0';
	}

	dmu_objset_name(os, osname);

	(void) printf("Dataset %s [%s], ID %llu, cr_txg %llu, "
	"%s, %llu objects%s%s\n",
	osname, type, (u_longlong_t)dmu_objset_id(os),
	(u_longlong_t)dds.dds_creation_txg,
	numbuf, (u_longlong_t)usedobjs, blkbuf,
	(dds.dds_inconsistent) ? " (inconsistent)" : "");

	for (i = 0; i < zopt_object_args; i++) {
	obj_start = zopt_object_ranges[i].zor_obj_start;
	obj_end = zopt_object_ranges[i].zor_obj_end;
	flags = zopt_object_ranges[i].zor_flags;

	object = obj_start;
	if (object == 0 \|\| obj_start == obj_end)
	dump_object(os, object, verbosity, &print_header, NULL,
	flags);
	else
	object--;

	while ((dmu_object_next(os, &object, B_FALSE, 0) == 0) &&
	object <= obj_end) {
	dump_object(os, object, verbosity, &print_header, NULL,
	flags);
	}
	}

	if (zopt_object_args > 0) {
	(void) printf("\n");
	return;
	}

	if (dump_opt['i'] != 0 \|\| verbosity >= 2)
	dump_intent_log(dmu_objset_zil(os));

	if (dmu_objset_ds(os) != NULL) {
	dsl_dataset_t *ds = dmu_objset_ds(os);
	dump_blkptr_list(&ds->ds_deadlist, "Deadlist");
	if (dsl_deadlist_is_open(&ds->ds_dir->dd_livelist) &&
	!dmu_objset_is_snapshot(os)) {
	dump_blkptr_list(&ds->ds_dir->dd_livelist, "Livelist");
	if (verify_dd_livelist(os) != 0)
	fatal("livelist is incorrect");
	}

	if (dsl_dataset_remap_deadlist_exists(ds)) {
	(void) printf("ds_remap_deadlist:\n");
	dump_blkptr_list(&ds->ds_remap_deadlist, "Deadlist");
	}
	count_ds_mos_objects(ds);
	}

	if (dmu_objset_ds(os) != NULL)
	dump_bookmarks(os, verbosity);

	if (verbosity < 2)
	return;

	if (BP_IS_HOLE(os->os_rootbp))
	return;

	dump_object(os, 0, verbosity, &print_header, NULL, 0);
	object_count = 0;
	if (DMU_USERUSED_DNODE(os) != NULL &&
	DMU_USERUSED_DNODE(os)->dn_type != 0) {
	dump_object(os, DMU_USERUSED_OBJECT, verbosity, &print_header,
	NULL, 0);
	dump_object(os, DMU_GROUPUSED_OBJECT, verbosity, &print_header,
	NULL, 0);
	}

	if (DMU_PROJECTUSED_DNODE(os) != NULL &&
	DMU_PROJECTUSED_DNODE(os)->dn_type != 0)
	dump_object(os, DMU_PROJECTUSED_OBJECT, verbosity,
	&print_header, NULL, 0);

	object = 0;
	while ((error = dmu_object_next(os, &object, B_FALSE, 0)) == 0) {
	dump_object(os, object, verbosity, &print_header, &dnode_slots,
	0);
	object_count++;
	total_slots_used += dnode_slots;
	max_slot_used = object + dnode_slots - 1;
	}

	(void) printf("\n");

	(void) printf(" Dnode slots:\n");
	(void) printf("\tTotal used: %10llu\n",
	(u_longlong_t)total_slots_used);
	(void) printf("\tMax used: %10llu\n",
	(u_longlong_t)max_slot_used);
	(void) printf("\tPercent empty: %10lf\n",
	(double)(max_slot_used - total_slots_used)*100 /
	(double)max_slot_used);
	(void) printf("\n");

	if (error != ESRCH) {
	(void) fprintf(stderr, "dmu_object_next() = %d\n", error);
	abort();
	}

	ASSERT3U(object_count, ==, usedobjs);

	if (leaked_objects != 0) {
	(void) printf("%d potentially leaked objects detected\n",
	leaked_objects);
	leaked_objects = 0;
	}
	}

	static void
	dump_uberblock(uberblock_t ub, const char header, const char *footer)
	{
	time_t timestamp = ub->ub_timestamp;

	(void) printf("%s", header ? header : "");
	(void) printf("\tmagic = %016llx\n", (u_longlong_t)ub->ub_magic);
	(void) printf("\tversion = %llu\n", (u_longlong_t)ub->ub_version);
	(void) printf("\ttxg = %llu\n", (u_longlong_t)ub->ub_txg);
	(void) printf("\tguid_sum = %llu\n", (u_longlong_t)ub->ub_guid_sum);
	(void) printf("\ttimestamp = %llu UTC = %s",
	(u_longlong_t)ub->ub_timestamp, asctime(localtime(&timestamp)));

	(void) printf("\tmmp_magic = %016llx\n",
	(u_longlong_t)ub->ub_mmp_magic);
	if (MMP_VALID(ub)) {
	(void) printf("\tmmp_delay = %0llu\n",
	(u_longlong_t)ub->ub_mmp_delay);
	if (MMP_SEQ_VALID(ub))
	(void) printf("\tmmp_seq = %u\n",
	(unsigned int) MMP_SEQ(ub));
	if (MMP_FAIL_INT_VALID(ub))
	(void) printf("\tmmp_fail = %u\n",
	(unsigned int) MMP_FAIL_INT(ub));
	if (MMP_INTERVAL_VALID(ub))
	(void) printf("\tmmp_write = %u\n",
	(unsigned int) MMP_INTERVAL(ub));
	/* After MMP_* to make summarize_uberblock_mmp cleaner */
	(void) printf("\tmmp_valid = %x\n",
	(unsigned int) ub->ub_mmp_config & 0xFF);
	}

	if (dump_opt['u'] >= 4) {
	char blkbuf[BP_SPRINTF_LEN];
	snprintf_blkptr(blkbuf, sizeof (blkbuf), &ub->ub_rootbp);
	(void) printf("\trootbp = %s\n", blkbuf);
	}
	(void) printf("\tcheckpoint_txg = %llu\n",
	(u_longlong_t)ub->ub_checkpoint_txg);
	(void) printf("%s", footer ? footer : "");
	}

	static void
	dump_config(spa_t *spa)
	{
	dmu_buf_t *db;
	size_t nvsize = 0;
	int error = 0;


	error = dmu_bonus_hold(spa->spa_meta_objset,
	spa->spa_config_object, FTAG, &db);

	if (error == 0) {
	nvsize = (uint64_t )db->db_data;
	dmu_buf_rele(db, FTAG);

	(void) printf("\nMOS Configuration:\n");
	dump_packed_nvlist(spa->spa_meta_objset,
	spa->spa_config_object, (void *)&nvsize, 1);
	} else {
	(void) fprintf(stderr, "dmu_bonus_hold(%llu) failed, errno %d",
	(u_longlong_t)spa->spa_config_object, error);
	}
	}

	static void
	dump_cachefile(const char *cachefile)
	{
	int fd;
	struct stat64 statbuf;
	char *buf;
	nvlist_t *config;

	if ((fd = open64(cachefile, O_RDONLY)) < 0) {
	(void) printf("cannot open '%s': %s\n", cachefile,
	strerror(errno));
	exit(1);
	}

	if (fstat64(fd, &statbuf) != 0) {
	(void) printf("failed to stat '%s': %s\n", cachefile,
	strerror(errno));
	exit(1);
	}

	if ((buf = malloc(statbuf.st_size)) == NULL) {
	(void) fprintf(stderr, "failed to allocate %llu bytes\n",
	(u_longlong_t)statbuf.st_size);
	exit(1);
	}

	if (read(fd, buf, statbuf.st_size) != statbuf.st_size) {
	(void) fprintf(stderr, "failed to read %llu bytes\n",
	(u_longlong_t)statbuf.st_size);
	exit(1);
	}

	(void) close(fd);

	if (nvlist_unpack(buf, statbuf.st_size, &config, 0) != 0) {
	(void) fprintf(stderr, "failed to unpack nvlist\n");
	exit(1);
	}

	free(buf);

	dump_nvlist(config, 0);

	nvlist_free(config);
	}

	/*
	* ZFS label nvlist stats
	*/
	typedef struct zdb_nvl_stats {
	int zns_list_count;
	int zns_leaf_count;
	size_t zns_leaf_largest;
	size_t zns_leaf_total;
	nvlist_t *zns_string;
	nvlist_t *zns_uint64;
	nvlist_t *zns_boolean;
	} zdb_nvl_stats_t;

	static void
	collect_nvlist_stats(nvlist_t nvl, zdb_nvl_stats_t stats)
	{
	nvlist_t list, *array;
	nvpair_t *nvp = NULL;
	char *name;
	uint_t i, items;

	stats->zns_list_count++;

	while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) {
	name = nvpair_name(nvp);

	switch (nvpair_type(nvp)) {
	case DATA_TYPE_STRING:
	fnvlist_add_string(stats->zns_string, name,
	fnvpair_value_string(nvp));
	break;
	case DATA_TYPE_UINT64:
	fnvlist_add_uint64(stats->zns_uint64, name,
	fnvpair_value_uint64(nvp));
	break;
	case DATA_TYPE_BOOLEAN:
	fnvlist_add_boolean(stats->zns_boolean, name);
	break;
	case DATA_TYPE_NVLIST:
	if (nvpair_value_nvlist(nvp, &list) == 0)
	collect_nvlist_stats(list, stats);
	break;
	case DATA_TYPE_NVLIST_ARRAY:
	if (nvpair_value_nvlist_array(nvp, &array, &items) != 0)
	break;

	for (i = 0; i < items; i++) {
	collect_nvlist_stats(array[i], stats);

	/* collect stats on leaf vdev */
	if (strcmp(name, "children") == 0) {
	size_t size;

	(void) nvlist_size(array[i], &size,
	NV_ENCODE_XDR);
	stats->zns_leaf_total += size;
	if (size > stats->zns_leaf_largest)
	stats->zns_leaf_largest = size;
	stats->zns_leaf_count++;
	}
	}
	break;
	default:
	(void) printf("skip type %d!\n", (int)nvpair_type(nvp));
	}
	}
	}

	static void
	dump_nvlist_stats(nvlist_t *nvl, size_t cap)
	{
	zdb_nvl_stats_t stats = { 0 };
	size_t size, sum = 0, total;
	size_t noise;

	/* requires nvlist with non-unique names for stat collection */
	VERIFY0(nvlist_alloc(&stats.zns_string, 0, 0));
	VERIFY0(nvlist_alloc(&stats.zns_uint64, 0, 0));
	VERIFY0(nvlist_alloc(&stats.zns_boolean, 0, 0));
	VERIFY0(nvlist_size(stats.zns_boolean, &noise, NV_ENCODE_XDR));

	(void) printf("\n\nZFS Label NVList Config Stats:\n");

	VERIFY0(nvlist_size(nvl, &total, NV_ENCODE_XDR));
	(void) printf(" %d bytes used, %d bytes free (using %4.1f%%)\n\n",
	(int)total, (int)(cap - total), 100.0 * total / cap);

	collect_nvlist_stats(nvl, &stats);

	VERIFY0(nvlist_size(stats.zns_uint64, &size, NV_ENCODE_XDR));
	size -= noise;
	sum += size;
	(void) printf("%12s %4d %6d bytes (%5.2f%%)\n", "integers:",
	(int)fnvlist_num_pairs(stats.zns_uint64),
	(int)size, 100.0 * size / total);

	VERIFY0(nvlist_size(stats.zns_string, &size, NV_ENCODE_XDR));
	size -= noise;
	sum += size;
	(void) printf("%12s %4d %6d bytes (%5.2f%%)\n", "strings:",
	(int)fnvlist_num_pairs(stats.zns_string),
	(int)size, 100.0 * size / total);

	VERIFY0(nvlist_size(stats.zns_boolean, &size, NV_ENCODE_XDR));
	size -= noise;
	sum += size;
	(void) printf("%12s %4d %6d bytes (%5.2f%%)\n", "booleans:",
	(int)fnvlist_num_pairs(stats.zns_boolean),
	(int)size, 100.0 * size / total);

	size = total - sum; /* treat remainder as nvlist overhead */
	(void) printf("%12s %4d %6d bytes (%5.2f%%)\n\n", "nvlists:",
	stats.zns_list_count, (int)size, 100.0 * size / total);

	if (stats.zns_leaf_count > 0) {
	size_t average = stats.zns_leaf_total / stats.zns_leaf_count;

	(void) printf("%12s %4d %6d bytes average\n", "leaf vdevs:",
	stats.zns_leaf_count, (int)average);
	(void) printf("%24d bytes largest\n",
	(int)stats.zns_leaf_largest);

	if (dump_opt['l'] >= 3 && average > 0)
	(void) printf(" space for %d additional leaf vdevs\n",
	(int)((cap - total) / average));
	}
	(void) printf("\n");

	nvlist_free(stats.zns_string);
	nvlist_free(stats.zns_uint64);
	nvlist_free(stats.zns_boolean);
	}

	typedef struct cksum_record {
	zio_cksum_t cksum;
	boolean_t labels[VDEV_LABELS];
	avl_node_t link;
	} cksum_record_t;

	static int
	cksum_record_compare(const void x1, const void x2)
	{
	const cksum_record_t l = (cksum_record_t )x1;
	const cksum_record_t r = (cksum_record_t )x2;
	int arraysize = ARRAY_SIZE(l->cksum.zc_word);
	int difference = 0;

	for (int i = 0; i < arraysize; i++) {
	difference = TREE_CMP(l->cksum.zc_word[i], r->cksum.zc_word[i]);
	if (difference)
	break;
	}

	return (difference);
	}

	static cksum_record_t *
	cksum_record_alloc(zio_cksum_t *cksum, int l)
	{
	cksum_record_t *rec;

	rec = umem_zalloc(sizeof (*rec), UMEM_NOFAIL);
	rec->cksum = *cksum;
	rec->labels[l] = B_TRUE;

	return (rec);
	}

	static cksum_record_t *
	cksum_record_lookup(avl_tree_t tree, zio_cksum_t cksum)
	{
	cksum_record_t lookup = { .cksum = *cksum };
	avl_index_t where;

	return (avl_find(tree, &lookup, &where));
	}

	static cksum_record_t *
	cksum_record_insert(avl_tree_t tree, zio_cksum_t cksum, int l)
	{
	cksum_record_t *rec;

	rec = cksum_record_lookup(tree, cksum);
	if (rec) {
	rec->labels[l] = B_TRUE;
	} else {
	rec = cksum_record_alloc(cksum, l);
	avl_add(tree, rec);
	}

	return (rec);
	}

	static int
	first_label(cksum_record_t *rec)
	{
	for (int i = 0; i < VDEV_LABELS; i++)
	if (rec->labels[i])
	return (i);

	return (-1);
	}

	static void
	print_label_numbers(char prefix, cksum_record_t rec)
	{
	printf("%s", prefix);
	for (int i = 0; i < VDEV_LABELS; i++)
	if (rec->labels[i] == B_TRUE)
	printf("%d ", i);
	printf("\n");
	}

	#define MAX_UBERBLOCK_COUNT (VDEV_UBERBLOCK_RING >> UBERBLOCK_SHIFT)

	typedef struct zdb_label {
	vdev_label_t label;
	uint64_t label_offset;
	nvlist_t *config_nv;
	cksum_record_t *config;
	cksum_record_t *uberblocks[MAX_UBERBLOCK_COUNT];
	boolean_t header_printed;
	boolean_t read_failed;
	boolean_t cksum_valid;
	} zdb_label_t;

	static void
	print_label_header(zdb_label_t *label, int l)
	{

	if (dump_opt['q'])
	return;

	if (label->header_printed == B_TRUE)
	return;

	(void) printf("------------------------------------\n");
	(void) printf("LABEL %d %s\n", l,
	label->cksum_valid ? "" : "(Bad label cksum)");
	(void) printf("------------------------------------\n");

	label->header_printed = B_TRUE;
	}

	static void
	print_l2arc_header(void)
	{
	(void) printf("------------------------------------\n");
	(void) printf("L2ARC device header\n");
	(void) printf("------------------------------------\n");
	}

	static void
	print_l2arc_log_blocks(void)
	{
	(void) printf("------------------------------------\n");
	(void) printf("L2ARC device log blocks\n");
	(void) printf("------------------------------------\n");
	}

	static void
	dump_l2arc_log_entries(uint64_t log_entries,
	l2arc_log_ent_phys_t *le, uint64_t i)
	{
	for (int j = 0; j < log_entries; j++) {
	dva_t dva = le[j].le_dva;
	(void) printf("lb[%4llu]\tle[%4d]\tDVA asize: %llu, "
	"vdev: %llu, offset: %llu\n",
	(u_longlong_t)i, j + 1,
	(u_longlong_t)DVA_GET_ASIZE(&dva),
	(u_longlong_t)DVA_GET_VDEV(&dva),
	(u_longlong_t)DVA_GET_OFFSET(&dva));
	(void) printf("\|\t\t\t\tbirth: %llu\n",
	(u_longlong_t)le[j].le_birth);
	(void) printf("\|\t\t\t\tlsize: %llu\n",
	(u_longlong_t)L2BLK_GET_LSIZE((&le[j])->le_prop));
	(void) printf("\|\t\t\t\tpsize: %llu\n",
	(u_longlong_t)L2BLK_GET_PSIZE((&le[j])->le_prop));
	(void) printf("\|\t\t\t\tcompr: %llu\n",
	(u_longlong_t)L2BLK_GET_COMPRESS((&le[j])->le_prop));
	(void) printf("\|\t\t\t\tcomplevel: %llu\n",
	(u_longlong_t)(&le[j])->le_complevel);
	(void) printf("\|\t\t\t\ttype: %llu\n",
	(u_longlong_t)L2BLK_GET_TYPE((&le[j])->le_prop));
	(void) printf("\|\t\t\t\tprotected: %llu\n",
	(u_longlong_t)L2BLK_GET_PROTECTED((&le[j])->le_prop));
	(void) printf("\|\t\t\t\tprefetch: %llu\n",
	(u_longlong_t)L2BLK_GET_PREFETCH((&le[j])->le_prop));
	(void) printf("\|\t\t\t\taddress: %llu\n",
	(u_longlong_t)le[j].le_daddr);
	(void) printf("\|\t\t\t\tARC state: %llu\n",
	(u_longlong_t)L2BLK_GET_STATE((&le[j])->le_prop));
	(void) printf("\|\n");
	}
	(void) printf("\n");
	}

	static void
	dump_l2arc_log_blkptr(l2arc_log_blkptr_t lbps)
	{
	(void) printf("\|\t\tdaddr: %llu\n", (u_longlong_t)lbps.lbp_daddr);
	(void) printf("\|\t\tpayload_asize: %llu\n",
	(u_longlong_t)lbps.lbp_payload_asize);
	(void) printf("\|\t\tpayload_start: %llu\n",
	(u_longlong_t)lbps.lbp_payload_start);
	(void) printf("\|\t\tlsize: %llu\n",
	(u_longlong_t)L2BLK_GET_LSIZE((&lbps)->lbp_prop));
	(void) printf("\|\t\tasize: %llu\n",
	(u_longlong_t)L2BLK_GET_PSIZE((&lbps)->lbp_prop));
	(void) printf("\|\t\tcompralgo: %llu\n",
	(u_longlong_t)L2BLK_GET_COMPRESS((&lbps)->lbp_prop));
	(void) printf("\|\t\tcksumalgo: %llu\n",
	(u_longlong_t)L2BLK_GET_CHECKSUM((&lbps)->lbp_prop));
	(void) printf("\|\n\n");
	}

	static void
	dump_l2arc_log_blocks(int fd, l2arc_dev_hdr_phys_t l2dhdr,
	l2arc_dev_hdr_phys_t *rebuild)
	{
	l2arc_log_blk_phys_t this_lb;
	uint64_t asize;
	l2arc_log_blkptr_t lbps[2];
	abd_t *abd;
	zio_cksum_t cksum;
	int failed = 0;
	l2arc_dev_t dev;

	if (!dump_opt['q'])
	print_l2arc_log_blocks();
	bcopy((&l2dhdr)->dh_start_lbps, lbps, sizeof (lbps));

	dev.l2ad_evict = l2dhdr.dh_evict;
	dev.l2ad_start = l2dhdr.dh_start;
	dev.l2ad_end = l2dhdr.dh_end;

	if (l2dhdr.dh_start_lbps[0].lbp_daddr == 0) {
	/* no log blocks to read */
	if (!dump_opt['q']) {
	(void) printf("No log blocks to read\n");
	(void) printf("\n");
	}
	return;
	} else {
	dev.l2ad_hand = lbps[0].lbp_daddr +
	L2BLK_GET_PSIZE((&lbps[0])->lbp_prop);
	}

	dev.l2ad_first = !!(l2dhdr.dh_flags & L2ARC_DEV_HDR_EVICT_FIRST);

	for (;;) {
	if (!l2arc_log_blkptr_valid(&dev, &lbps[0]))
	break;

	/* L2BLK_GET_PSIZE returns aligned size for log blocks */
	asize = L2BLK_GET_PSIZE((&lbps[0])->lbp_prop);
	if (pread64(fd, &this_lb, asize, lbps[0].lbp_daddr) != asize) {
	if (!dump_opt['q']) {
	(void) printf("Error while reading next log "
	"block\n\n");
	}
	break;
	}

	fletcher_4_native_varsize(&this_lb, asize, &cksum);
	if (!ZIO_CHECKSUM_EQUAL(cksum, lbps[0].lbp_cksum)) {
	failed++;
	if (!dump_opt['q']) {
	(void) printf("Invalid cksum\n");
	dump_l2arc_log_blkptr(lbps[0]);
	}
	break;
	}

	switch (L2BLK_GET_COMPRESS((&lbps[0])->lbp_prop)) {
	case ZIO_COMPRESS_OFF:
	break;
	default:
	abd = abd_alloc_for_io(asize, B_TRUE);
	abd_copy_from_buf_off(abd, &this_lb, 0, asize);
	zio_decompress_data(L2BLK_GET_COMPRESS(
	(&lbps[0])->lbp_prop), abd, &this_lb,
	asize, sizeof (this_lb), NULL);
	abd_free(abd);
	break;
	}

	if (this_lb.lb_magic == BSWAP_64(L2ARC_LOG_BLK_MAGIC))
	byteswap_uint64_array(&this_lb, sizeof (this_lb));
	if (this_lb.lb_magic != L2ARC_LOG_BLK_MAGIC) {
	if (!dump_opt['q'])
	(void) printf("Invalid log block magic\n\n");
	break;
	}

	rebuild->dh_lb_count++;
	rebuild->dh_lb_asize += asize;
	if (dump_opt['l'] > 1 && !dump_opt['q']) {
	(void) printf("lb[%4llu]\tmagic: %llu\n",
	(u_longlong_t)rebuild->dh_lb_count,
	(u_longlong_t)this_lb.lb_magic);
	dump_l2arc_log_blkptr(lbps[0]);
	}

	if (dump_opt['l'] > 2 && !dump_opt['q'])
	dump_l2arc_log_entries(l2dhdr.dh_log_entries,
	this_lb.lb_entries,
	rebuild->dh_lb_count);

	if (l2arc_range_check_overlap(lbps[1].lbp_payload_start,
	lbps[0].lbp_payload_start, dev.l2ad_evict) &&
	!dev.l2ad_first)
	break;

	lbps[0] = lbps[1];
	lbps[1] = this_lb.lb_prev_lbp;
	}

	if (!dump_opt['q']) {
	(void) printf("log_blk_count:\t %llu with valid cksum\n",
	(u_longlong_t)rebuild->dh_lb_count);
	(void) printf("\t\t %d with invalid cksum\n", failed);
	(void) printf("log_blk_asize:\t %llu\n\n",
	(u_longlong_t)rebuild->dh_lb_asize);
	}
	}

	static int
	dump_l2arc_header(int fd)
	{
	l2arc_dev_hdr_phys_t l2dhdr, rebuild;
	int error = B_FALSE;

	bzero(&l2dhdr, sizeof (l2dhdr));
	bzero(&rebuild, sizeof (rebuild));

	if (pread64(fd, &l2dhdr, sizeof (l2dhdr),
	VDEV_LABEL_START_SIZE) != sizeof (l2dhdr)) {
	error = B_TRUE;
	} else {
	if (l2dhdr.dh_magic == BSWAP_64(L2ARC_DEV_HDR_MAGIC))
	byteswap_uint64_array(&l2dhdr, sizeof (l2dhdr));

	if (l2dhdr.dh_magic != L2ARC_DEV_HDR_MAGIC)
	error = B_TRUE;
	}

	if (error) {
	(void) printf("L2ARC device header not found\n\n");
	/* Do not return an error here for backward compatibility */
	return (0);
	} else if (!dump_opt['q']) {
	print_l2arc_header();

	(void) printf(" magic: %llu\n",
	(u_longlong_t)l2dhdr.dh_magic);
	(void) printf(" version: %llu\n",
	(u_longlong_t)l2dhdr.dh_version);
	(void) printf(" pool_guid: %llu\n",
	(u_longlong_t)l2dhdr.dh_spa_guid);
	(void) printf(" flags: %llu\n",
	(u_longlong_t)l2dhdr.dh_flags);
	(void) printf(" start_lbps[0]: %llu\n",
	(u_longlong_t)
	l2dhdr.dh_start_lbps[0].lbp_daddr);
	(void) printf(" start_lbps[1]: %llu\n",
	(u_longlong_t)
	l2dhdr.dh_start_lbps[1].lbp_daddr);
	(void) printf(" log_blk_ent: %llu\n",
	(u_longlong_t)l2dhdr.dh_log_entries);
	(void) printf(" start: %llu\n",
	(u_longlong_t)l2dhdr.dh_start);
	(void) printf(" end: %llu\n",
	(u_longlong_t)l2dhdr.dh_end);
	(void) printf(" evict: %llu\n",
	(u_longlong_t)l2dhdr.dh_evict);
	(void) printf(" lb_asize_refcount: %llu\n",
	(u_longlong_t)l2dhdr.dh_lb_asize);
	(void) printf(" lb_count_refcount: %llu\n",
	(u_longlong_t)l2dhdr.dh_lb_count);
	(void) printf(" trim_action_time: %llu\n",
	(u_longlong_t)l2dhdr.dh_trim_action_time);
	(void) printf(" trim_state: %llu\n\n",
	(u_longlong_t)l2dhdr.dh_trim_state);
	}

	dump_l2arc_log_blocks(fd, l2dhdr, &rebuild);
	/*
	* The total aligned size of log blocks and the number of log blocks
	* reported in the header of the device may be less than what zdb
	* reports by dump_l2arc_log_blocks() which emulates l2arc_rebuild().
	* This happens because dump_l2arc_log_blocks() lacks the memory
	* pressure valve that l2arc_rebuild() has. Thus, if we are on a system
	* with low memory, l2arc_rebuild will exit prematurely and dh_lb_asize
	* and dh_lb_count will be lower to begin with than what exists on the
	* device. This is normal and zdb should not exit with an error. The
	* opposite case should never happen though, the values reported in the
	* header should never be higher than what dump_l2arc_log_blocks() and
	* l2arc_rebuild() report. If this happens there is a leak in the
	* accounting of log blocks.
	*/
	if (l2dhdr.dh_lb_asize > rebuild.dh_lb_asize \|\|
	l2dhdr.dh_lb_count > rebuild.dh_lb_count)
	return (1);

	return (0);
	}

	static void
	dump_config_from_label(zdb_label_t *label, size_t buflen, int l)
	{
	if (dump_opt['q'])
	return;

	if ((dump_opt['l'] < 3) && (first_label(label->config) != l))
	return;

	print_label_header(label, l);
	dump_nvlist(label->config_nv, 4);
	print_label_numbers(" labels = ", label->config);

	if (dump_opt['l'] >= 2)
	dump_nvlist_stats(label->config_nv, buflen);
	}

	#define ZDB_MAX_UB_HEADER_SIZE 32

	static void
	dump_label_uberblocks(zdb_label_t *label, uint64_t ashift, int label_num)
	{

	vdev_t vd;
	char header[ZDB_MAX_UB_HEADER_SIZE];

	vd.vdev_ashift = ashift;
	vd.vdev_top = &vd;

	for (int i = 0; i < VDEV_UBERBLOCK_COUNT(&vd); i++) {
	uint64_t uoff = VDEV_UBERBLOCK_OFFSET(&vd, i);
	uberblock_t ub = (void )((char *)&label->label + uoff);
	cksum_record_t *rec = label->uberblocks[i];

	if (rec == NULL) {
	if (dump_opt['u'] >= 2) {
	print_label_header(label, label_num);
	(void) printf(" Uberblock[%d] invalid\n", i);
	}
	continue;
	}

	if ((dump_opt['u'] < 3) && (first_label(rec) != label_num))
	continue;

	if ((dump_opt['u'] < 4) &&
	(ub->ub_mmp_magic == MMP_MAGIC) && ub->ub_mmp_delay &&
	(i >= VDEV_UBERBLOCK_COUNT(&vd) - MMP_BLOCKS_PER_LABEL))
	continue;

	print_label_header(label, label_num);
	(void) snprintf(header, ZDB_MAX_UB_HEADER_SIZE,
	" Uberblock[%d]\n", i);
	dump_uberblock(ub, header, "");
	print_label_numbers(" labels = ", rec);
	}
	}

	static char curpath[PATH_MAX];

	/*
	* Iterate through the path components, recursively passing
	* current one's obj and remaining path until we find the obj
	* for the last one.
	*/
	static int
	dump_path_impl(objset_t os, uint64_t obj, char name, uint64_t *retobj)
	{
	int err;
	boolean_t header = B_TRUE;
	uint64_t child_obj;
	char *s;
	dmu_buf_t *db;
	dmu_object_info_t doi;

	if ((s = strchr(name, '/')) != NULL)
	*s = '\0';
	err = zap_lookup(os, obj, name, 8, 1, &child_obj);

	(void) strlcat(curpath, name, sizeof (curpath));

	if (err != 0) {
	(void) fprintf(stderr, "failed to lookup %s: %s\n",
	curpath, strerror(err));
	return (err);
	}

	child_obj = ZFS_DIRENT_OBJ(child_obj);
	err = sa_buf_hold(os, child_obj, FTAG, &db);
	if (err != 0) {
	(void) fprintf(stderr,
	"failed to get SA dbuf for obj %llu: %s\n",
	(u_longlong_t)child_obj, strerror(err));
	return (EINVAL);
	}
	dmu_object_info_from_db(db, &doi);
	sa_buf_rele(db, FTAG);

	if (doi.doi_bonus_type != DMU_OT_SA &&
	doi.doi_bonus_type != DMU_OT_ZNODE) {
	(void) fprintf(stderr, "invalid bonus type %d for obj %llu\n",
	doi.doi_bonus_type, (u_longlong_t)child_obj);
	return (EINVAL);
	}

	if (dump_opt['v'] > 6) {
	(void) printf("obj=%llu %s type=%d bonustype=%d\n",
	(u_longlong_t)child_obj, curpath, doi.doi_type,
	doi.doi_bonus_type);
	}

	(void) strlcat(curpath, "/", sizeof (curpath));

	switch (doi.doi_type) {
	case DMU_OT_DIRECTORY_CONTENTS:
	if (s != NULL && *(s + 1) != '\0')
	return (dump_path_impl(os, child_obj, s + 1, retobj));
	fallthrough;
	case DMU_OT_PLAIN_FILE_CONTENTS:
	if (retobj != NULL) {
	*retobj = child_obj;
	} else {
	dump_object(os, child_obj, dump_opt['v'], &header,
	NULL, 0);
	}
	return (0);
	default:
	(void) fprintf(stderr, "object %llu has non-file/directory "
	"type %d\n", (u_longlong_t)obj, doi.doi_type);
	break;
	}

	return (EINVAL);
	}

	/*
	* Dump the blocks for the object specified by path inside the dataset.
	*/
	static int
	dump_path(char ds, char path, uint64_t *retobj)
	{
	int err;
	objset_t *os;
	uint64_t root_obj;

	err = open_objset(ds, FTAG, &os);
	if (err != 0)
	return (err);

	err = zap_lookup(os, MASTER_NODE_OBJ, ZFS_ROOT_OBJ, 8, 1, &root_obj);
	if (err != 0) {
	(void) fprintf(stderr, "can't lookup root znode: %s\n",
	strerror(err));
	close_objset(os, FTAG);
	return (EINVAL);
	}

	(void) snprintf(curpath, sizeof (curpath), "dataset=%s path=/", ds);

	err = dump_path_impl(os, root_obj, path, retobj);

	close_objset(os, FTAG);
	return (err);
	}

	static int
	zdb_copy_object(objset_t os, uint64_t srcobj, char destfile)
	{
	int err = 0;
	uint64_t size, readsize, oursize, offset;
	ssize_t writesize;
	sa_handle_t *hdl;

	(void) printf("Copying object %" PRIu64 " to file %s\n", srcobj,
	destfile);

	VERIFY3P(os, ==, sa_os);
	if ((err = sa_handle_get(os, srcobj, NULL, SA_HDL_PRIVATE, &hdl))) {
	(void) printf("Failed to get handle for SA znode\n");
	return (err);
	}
	if ((err = sa_lookup(hdl, sa_attr_table[ZPL_SIZE], &size, 8))) {
	(void) sa_handle_destroy(hdl);
	return (err);
	}
	(void) sa_handle_destroy(hdl);

	(void) printf("Object %" PRIu64 " is %" PRIu64 " bytes\n", srcobj,
	size);
	if (size == 0) {
	return (EINVAL);
	}

	int fd = open(destfile, O_WRONLY \| O_CREAT \| O_TRUNC, 0644);
	/*
	* We cap the size at 1 mebibyte here to prevent
	* allocation failures and nigh-infinite printing if the
	* object is extremely large.
	*/
	oursize = MIN(size, 1 << 20);
	offset = 0;
	char *buf = kmem_alloc(oursize, KM_NOSLEEP);
	if (buf == NULL) {
	return (ENOMEM);
	}

	while (offset < size) {
	readsize = MIN(size - offset, 1 << 20);
	err = dmu_read(os, srcobj, offset, readsize, buf, 0);
	if (err != 0) {
	(void) printf("got error %u from dmu_read\n", err);
	kmem_free(buf, oursize);
	return (err);
	}
	if (dump_opt['v'] > 3) {
	(void) printf("Read offset=%" PRIu64 " size=%" PRIu64
	" error=%d\n", offset, readsize, err);
	}

	writesize = write(fd, buf, readsize);
	if (writesize < 0) {
	err = errno;
	break;
	} else if (writesize != readsize) {
	/* Incomplete write */
	(void) fprintf(stderr, "Short write, only wrote %llu of"
	" %" PRIu64 " bytes, exiting...\n",
	(u_longlong_t)writesize, readsize);
	break;
	}

	offset += readsize;
	}

	(void) close(fd);

	if (buf != NULL)
	kmem_free(buf, oursize);

	return (err);
	}

	static boolean_t
	label_cksum_valid(vdev_label_t *label, uint64_t offset)
	{
	zio_checksum_info_t *ci = &zio_checksum_table[ZIO_CHECKSUM_LABEL];
	zio_cksum_t expected_cksum;
	zio_cksum_t actual_cksum;
	zio_cksum_t verifier;
	zio_eck_t *eck;
	int byteswap;

	void data = (char )label + offsetof(vdev_label_t, vl_vdev_phys);
	eck = (zio_eck_t )((char )(data) + VDEV_PHYS_SIZE) - 1;

	offset += offsetof(vdev_label_t, vl_vdev_phys);
	ZIO_SET_CHECKSUM(&verifier, offset, 0, 0, 0);

	byteswap = (eck->zec_magic == BSWAP_64(ZEC_MAGIC));
	if (byteswap)
	byteswap_uint64_array(&verifier, sizeof (zio_cksum_t));

	expected_cksum = eck->zec_cksum;
	eck->zec_cksum = verifier;

	abd_t *abd = abd_get_from_buf(data, VDEV_PHYS_SIZE);
	ci->ci_func[byteswap](abd, VDEV_PHYS_SIZE, NULL, &actual_cksum);
	abd_free(abd);

	if (byteswap)
	byteswap_uint64_array(&expected_cksum, sizeof (zio_cksum_t));

	if (ZIO_CHECKSUM_EQUAL(actual_cksum, expected_cksum))
	return (B_TRUE);

	return (B_FALSE);
	}

	static int
	dump_label(const char *dev)
	{
	char path[MAXPATHLEN];
	zdb_label_t labels[VDEV_LABELS];
	uint64_t psize, ashift, l2cache;
	struct stat64 statbuf;
	boolean_t config_found = B_FALSE;
	boolean_t error = B_FALSE;
	boolean_t read_l2arc_header = B_FALSE;
	avl_tree_t config_tree;
	avl_tree_t uberblock_tree;
	void node, cookie;
	int fd;

	bzero(labels, sizeof (labels));

	/*
	* Check if we were given absolute path and use it as is.
	* Otherwise if the provided vdev name doesn't point to a file,
	* try prepending expected disk paths and partition numbers.
	*/
	(void) strlcpy(path, dev, sizeof (path));
	if (dev[0] != '/' && stat64(path, &statbuf) != 0) {
	int error;

	error = zfs_resolve_shortname(dev, path, MAXPATHLEN);
	if (error == 0 && zfs_dev_is_whole_disk(path)) {
	if (zfs_append_partition(path, MAXPATHLEN) == -1)
	error = ENOENT;
	}

	if (error \|\| (stat64(path, &statbuf) != 0)) {
	(void) printf("failed to find device %s, try "
	"specifying absolute path instead\n", dev);
	return (1);
	}
	}

	if ((fd = open64(path, O_RDONLY)) < 0) {
	(void) printf("cannot open '%s': %s\n", path, strerror(errno));
	exit(1);
	}

	if (fstat64_blk(fd, &statbuf) != 0) {
	(void) printf("failed to stat '%s': %s\n", path,
	strerror(errno));
	(void) close(fd);
	exit(1);
	}

	if (S_ISBLK(statbuf.st_mode) && zfs_dev_flush(fd) != 0)
	(void) printf("failed to invalidate cache '%s' : %s\n", path,
	strerror(errno));

	avl_create(&config_tree, cksum_record_compare,
	sizeof (cksum_record_t), offsetof(cksum_record_t, link));
	avl_create(&uberblock_tree, cksum_record_compare,
	sizeof (cksum_record_t), offsetof(cksum_record_t, link));

	psize = statbuf.st_size;
	psize = P2ALIGN(psize, (uint64_t)sizeof (vdev_label_t));
	ashift = SPA_MINBLOCKSHIFT;

	/*
	* 1. Read the label from disk
	* 2. Verify label cksum
	* 3. Unpack the configuration and insert in config tree.
	* 4. Traverse all uberblocks and insert in uberblock tree.
	*/
	for (int l = 0; l < VDEV_LABELS; l++) {
	zdb_label_t *label = &labels[l];
	char *buf = label->label.vl_vdev_phys.vp_nvlist;
	size_t buflen = sizeof (label->label.vl_vdev_phys.vp_nvlist);
	nvlist_t *config;
	cksum_record_t *rec;
	zio_cksum_t cksum;
	vdev_t vd;

	label->label_offset = vdev_label_offset(psize, l, 0);

	if (pread64(fd, &label->label, sizeof (label->label),
	label->label_offset) != sizeof (label->label)) {
	if (!dump_opt['q'])
	(void) printf("failed to read label %d\n", l);
	label->read_failed = B_TRUE;
	error = B_TRUE;
	continue;
	}

	label->read_failed = B_FALSE;
	label->cksum_valid = label_cksum_valid(&label->label,
	label->label_offset);

	if (nvlist_unpack(buf, buflen, &config, 0) == 0) {
	nvlist_t *vdev_tree = NULL;
	size_t size;

	if ((nvlist_lookup_nvlist(config,
	ZPOOL_CONFIG_VDEV_TREE, &vdev_tree) != 0) \|\|
	(nvlist_lookup_uint64(vdev_tree,
	ZPOOL_CONFIG_ASHIFT, &ashift) != 0))
	ashift = SPA_MINBLOCKSHIFT;

	if (nvlist_size(config, &size, NV_ENCODE_XDR) != 0)
	size = buflen;

	/* If the device is a cache device clear the header. */
	if (!read_l2arc_header) {
	if (nvlist_lookup_uint64(config,
	ZPOOL_CONFIG_POOL_STATE, &l2cache) == 0 &&
	l2cache == POOL_STATE_L2CACHE) {
	read_l2arc_header = B_TRUE;
	}
	}

	fletcher_4_native_varsize(buf, size, &cksum);
	rec = cksum_record_insert(&config_tree, &cksum, l);

	label->config = rec;
	label->config_nv = config;
	config_found = B_TRUE;
	} else {
	error = B_TRUE;
	}

	vd.vdev_ashift = ashift;
	vd.vdev_top = &vd;

	for (int i = 0; i < VDEV_UBERBLOCK_COUNT(&vd); i++) {
	uint64_t uoff = VDEV_UBERBLOCK_OFFSET(&vd, i);
	uberblock_t ub = (void )((char *)label + uoff);

	if (uberblock_verify(ub))
	continue;

	fletcher_4_native_varsize(ub, sizeof (*ub), &cksum);
	rec = cksum_record_insert(&uberblock_tree, &cksum, l);

	label->uberblocks[i] = rec;
	}
	}

	/*
	* Dump the label and uberblocks.
	*/
	for (int l = 0; l < VDEV_LABELS; l++) {
	zdb_label_t *label = &labels[l];
	size_t buflen = sizeof (label->label.vl_vdev_phys.vp_nvlist);

	if (label->read_failed == B_TRUE)
	continue;

	if (label->config_nv) {
	dump_config_from_label(label, buflen, l);
	} else {
	if (!dump_opt['q'])
	(void) printf("failed to unpack label %d\n", l);
	}

	if (dump_opt['u'])
	dump_label_uberblocks(label, ashift, l);

	nvlist_free(label->config_nv);
	}

	/*
	* Dump the L2ARC header, if existent.
	*/
	if (read_l2arc_header)
	error \|= dump_l2arc_header(fd);

	cookie = NULL;
	while ((node = avl_destroy_nodes(&config_tree, &cookie)) != NULL)
	umem_free(node, sizeof (cksum_record_t));

	cookie = NULL;
	while ((node = avl_destroy_nodes(&uberblock_tree, &cookie)) != NULL)
	umem_free(node, sizeof (cksum_record_t));

	avl_destroy(&config_tree);
	avl_destroy(&uberblock_tree);

	(void) close(fd);

	return (config_found == B_FALSE ? 2 :
	(error == B_TRUE ? 1 : 0));
	}

	static uint64_t dataset_feature_count[SPA_FEATURES];
	static uint64_t global_feature_count[SPA_FEATURES];
	static uint64_t remap_deadlist_count = 0;

	/ARGSUSED/
	static int
	dump_one_objset(const char dsname, void arg)
	{
	int error;
	objset_t *os;
	spa_feature_t f;

	error = open_objset(dsname, FTAG, &os);
	if (error != 0)
	return (0);

	for (f = 0; f < SPA_FEATURES; f++) {
	if (!dsl_dataset_feature_is_active(dmu_objset_ds(os), f))
	continue;
	ASSERT(spa_feature_table[f].fi_flags &
	ZFEATURE_FLAG_PER_DATASET);
	dataset_feature_count[f]++;
	}

	if (dsl_dataset_remap_deadlist_exists(dmu_objset_ds(os))) {
	remap_deadlist_count++;
	}

	for (dsl_bookmark_node_t *dbn =
	avl_first(&dmu_objset_ds(os)->ds_bookmarks); dbn != NULL;
	dbn = AVL_NEXT(&dmu_objset_ds(os)->ds_bookmarks, dbn)) {
	mos_obj_refd(dbn->dbn_phys.zbm_redaction_obj);
	if (dbn->dbn_phys.zbm_redaction_obj != 0)
	global_feature_count[SPA_FEATURE_REDACTION_BOOKMARKS]++;
	if (dbn->dbn_phys.zbm_flags & ZBM_FLAG_HAS_FBN)
	global_feature_count[SPA_FEATURE_BOOKMARK_WRITTEN]++;
	}

	if (dsl_deadlist_is_open(&dmu_objset_ds(os)->ds_dir->dd_livelist) &&
	!dmu_objset_is_snapshot(os)) {
	global_feature_count[SPA_FEATURE_LIVELIST]++;
	}

	dump_objset(os);
	close_objset(os, FTAG);
	fuid_table_destroy();
	return (0);
	}

	/*
	* Block statistics.
	*/
	#define PSIZE_HISTO_SIZE (SPA_OLD_MAXBLOCKSIZE / SPA_MINBLOCKSIZE + 2)
	typedef struct zdb_blkstats {
	uint64_t zb_asize;
	uint64_t zb_lsize;
	uint64_t zb_psize;
	uint64_t zb_count;
	uint64_t zb_gangs;
	uint64_t zb_ditto_samevdev;
	uint64_t zb_ditto_same_ms;
	uint64_t zb_psize_histogram[PSIZE_HISTO_SIZE];
	} zdb_blkstats_t;

	/*
	* Extended object types to report deferred frees and dedup auto-ditto blocks.
	*/
	#define ZDB_OT_DEFERRED (DMU_OT_NUMTYPES + 0)
	#define ZDB_OT_DITTO (DMU_OT_NUMTYPES + 1)
	#define ZDB_OT_OTHER (DMU_OT_NUMTYPES + 2)
	#define ZDB_OT_TOTAL (DMU_OT_NUMTYPES + 3)

	static const char *zdb_ot_extname[] = {
	"deferred free",
	"dedup ditto",
	"other",
	"Total",
	};

	#define ZB_TOTAL DN_MAX_LEVELS
	#define SPA_MAX_FOR_16M (SPA_MAXBLOCKSHIFT+1)

	typedef struct zdb_cb {
	zdb_blkstats_t zcb_type[ZB_TOTAL + 1][ZDB_OT_TOTAL + 1];
	uint64_t zcb_removing_size;
	uint64_t zcb_checkpoint_size;
	uint64_t zcb_dedup_asize;
	uint64_t zcb_dedup_blocks;
	uint64_t zcb_psize_count[SPA_MAX_FOR_16M];
	uint64_t zcb_lsize_count[SPA_MAX_FOR_16M];
	uint64_t zcb_asize_count[SPA_MAX_FOR_16M];
	uint64_t zcb_psize_len[SPA_MAX_FOR_16M];
	uint64_t zcb_lsize_len[SPA_MAX_FOR_16M];
	uint64_t zcb_asize_len[SPA_MAX_FOR_16M];
	uint64_t zcb_psize_total;
	uint64_t zcb_lsize_total;
	uint64_t zcb_asize_total;
	uint64_t zcb_embedded_blocks[NUM_BP_EMBEDDED_TYPES];
	uint64_t zcb_embedded_histogram[NUM_BP_EMBEDDED_TYPES]
	[BPE_PAYLOAD_SIZE + 1];
	uint64_t zcb_start;
	hrtime_t zcb_lastprint;
	uint64_t zcb_totalasize;
	uint64_t zcb_errors[256];
	int zcb_readfails;
	int zcb_haderrors;
	spa_t *zcb_spa;
	uint32_t **zcb_vd_obsolete_counts;
	} zdb_cb_t;

	/* test if two DVA offsets from same vdev are within the same metaslab */
	static boolean_t
	same_metaslab(spa_t *spa, uint64_t vdev, uint64_t off1, uint64_t off2)
	{
	vdev_t *vd = vdev_lookup_top(spa, vdev);
	uint64_t ms_shift = vd->vdev_ms_shift;

	return ((off1 >> ms_shift) == (off2 >> ms_shift));
	}

	/*
	* Used to simplify reporting of the histogram data.
	*/
	typedef struct one_histo {
	char *name;
	uint64_t *count;
	uint64_t *len;
	uint64_t cumulative;
	} one_histo_t;

	/*
	* The number of separate histograms processed for psize, lsize and asize.
	*/
	#define NUM_HISTO 3

	/*
	* This routine will create a fixed column size output of three different
	* histograms showing by blocksize of 512 - 2^ SPA_MAX_FOR_16M
	* the count, length and cumulative length of the psize, lsize and
	* asize blocks.
	*
	* All three types of blocks are listed on a single line
	*
	* By default the table is printed in nicenumber format (e.g. 123K) but
	* if the '-P' parameter is specified then the full raw number (parseable)
	* is printed out.
	*/
	static void
	dump_size_histograms(zdb_cb_t *zcb)
	{
	/*
	* A temporary buffer that allows us to convert a number into
	* a string using zdb_nicenumber to allow either raw or human
	* readable numbers to be output.
	*/
	char numbuf[32];

	/*
	* Define titles which are used in the headers of the tables
	* printed by this routine.
	*/
	const char blocksize_title1[] = "block";
	const char blocksize_title2[] = "size";
	const char count_title[] = "Count";
	const char length_title[] = "Size";
	const char cumulative_title[] = "Cum.";

	/*
	* Setup the histogram arrays (psize, lsize, and asize).
	*/
	one_histo_t parm_histo[NUM_HISTO];

	parm_histo[0].name = "psize";
	parm_histo[0].count = zcb->zcb_psize_count;
	parm_histo[0].len = zcb->zcb_psize_len;
	parm_histo[0].cumulative = 0;

	parm_histo[1].name = "lsize";
	parm_histo[1].count = zcb->zcb_lsize_count;
	parm_histo[1].len = zcb->zcb_lsize_len;
	parm_histo[1].cumulative = 0;

	parm_histo[2].name = "asize";
	parm_histo[2].count = zcb->zcb_asize_count;
	parm_histo[2].len = zcb->zcb_asize_len;
	parm_histo[2].cumulative = 0;


	(void) printf("\nBlock Size Histogram\n");
	/*
	* Print the first line titles
	*/
	if (dump_opt['P'])
	(void) printf("\n%s\t", blocksize_title1);
	else
	(void) printf("\n%7s ", blocksize_title1);

	for (int j = 0; j < NUM_HISTO; j++) {
	if (dump_opt['P']) {
	if (j < NUM_HISTO - 1) {
	(void) printf("%s\t\t\t", parm_histo[j].name);
	} else {
	/* Don't print trailing spaces */
	(void) printf(" %s", parm_histo[j].name);
	}
	} else {
	if (j < NUM_HISTO - 1) {
	/* Left aligned strings in the output */
	(void) printf("%-7s ",
	parm_histo[j].name);
	} else {
	/* Don't print trailing spaces */
	(void) printf("%s", parm_histo[j].name);
	}
	}
	}
	(void) printf("\n");

	/*
	* Print the second line titles
	*/
	if (dump_opt['P']) {
	(void) printf("%s\t", blocksize_title2);
	} else {
	(void) printf("%7s ", blocksize_title2);
	}

	for (int i = 0; i < NUM_HISTO; i++) {
	if (dump_opt['P']) {
	(void) printf("%s\t%s\t%s\t",
	count_title, length_title, cumulative_title);
	} else {
	(void) printf("%7s%7s%7s",
	count_title, length_title, cumulative_title);
	}
	}
	(void) printf("\n");

	/*
	* Print the rows
	*/
	for (int i = SPA_MINBLOCKSHIFT; i < SPA_MAX_FOR_16M; i++) {

	/*
	* Print the first column showing the blocksize
	*/
	zdb_nicenum((1ULL << i), numbuf, sizeof (numbuf));

	if (dump_opt['P']) {
	printf("%s", numbuf);
	} else {
	printf("%7s:", numbuf);
	}

	/*
	* Print the remaining set of 3 columns per size:
	* for psize, lsize and asize
	*/
	for (int j = 0; j < NUM_HISTO; j++) {
	parm_histo[j].cumulative += parm_histo[j].len[i];

	zdb_nicenum(parm_histo[j].count[i],
	numbuf, sizeof (numbuf));
	if (dump_opt['P'])
	(void) printf("\t%s", numbuf);
	else
	(void) printf("%7s", numbuf);

	zdb_nicenum(parm_histo[j].len[i],
	numbuf, sizeof (numbuf));
	if (dump_opt['P'])
	(void) printf("\t%s", numbuf);
	else
	(void) printf("%7s", numbuf);

	zdb_nicenum(parm_histo[j].cumulative,
	numbuf, sizeof (numbuf));
	if (dump_opt['P'])
	(void) printf("\t%s", numbuf);
	else
	(void) printf("%7s", numbuf);
	}
	(void) printf("\n");
	}
	}

	static void
	zdb_count_block(zdb_cb_t zcb, zilog_t zilog, const blkptr_t *bp,
	dmu_object_type_t type)
	{
	uint64_t refcnt = 0;
	int i;

	ASSERT(type < ZDB_OT_TOTAL);

	if (zilog && zil_bp_tree_add(zilog, bp) != 0)
	return;

	spa_config_enter(zcb->zcb_spa, SCL_CONFIG, FTAG, RW_READER);

	for (i = 0; i < 4; i++) {
	int l = (i < 2) ? BP_GET_LEVEL(bp) : ZB_TOTAL;
	int t = (i & 1) ? type : ZDB_OT_TOTAL;
	int equal;
	zdb_blkstats_t *zb = &zcb->zcb_type[l][t];

	zb->zb_asize += BP_GET_ASIZE(bp);
	zb->zb_lsize += BP_GET_LSIZE(bp);
	zb->zb_psize += BP_GET_PSIZE(bp);
	zb->zb_count++;

	/*
	* The histogram is only big enough to record blocks up to
	* SPA_OLD_MAXBLOCKSIZE; larger blocks go into the last,
	* "other", bucket.
	*/
	unsigned idx = BP_GET_PSIZE(bp) >> SPA_MINBLOCKSHIFT;
	idx = MIN(idx, SPA_OLD_MAXBLOCKSIZE / SPA_MINBLOCKSIZE + 1);
	zb->zb_psize_histogram[idx]++;

	zb->zb_gangs += BP_COUNT_GANG(bp);

	switch (BP_GET_NDVAS(bp)) {
	case 2:
	if (DVA_GET_VDEV(&bp->blk_dva[0]) ==
	DVA_GET_VDEV(&bp->blk_dva[1])) {
	zb->zb_ditto_samevdev++;

	if (same_metaslab(zcb->zcb_spa,
	DVA_GET_VDEV(&bp->blk_dva[0]),
	DVA_GET_OFFSET(&bp->blk_dva[0]),
	DVA_GET_OFFSET(&bp->blk_dva[1])))
	zb->zb_ditto_same_ms++;
	}
	break;
	case 3:
	equal = (DVA_GET_VDEV(&bp->blk_dva[0]) ==
	DVA_GET_VDEV(&bp->blk_dva[1])) +
	(DVA_GET_VDEV(&bp->blk_dva[0]) ==
	DVA_GET_VDEV(&bp->blk_dva[2])) +
	(DVA_GET_VDEV(&bp->blk_dva[1]) ==
	DVA_GET_VDEV(&bp->blk_dva[2]));
	if (equal != 0) {
	zb->zb_ditto_samevdev++;

	if (DVA_GET_VDEV(&bp->blk_dva[0]) ==
	DVA_GET_VDEV(&bp->blk_dva[1]) &&
	same_metaslab(zcb->zcb_spa,
	DVA_GET_VDEV(&bp->blk_dva[0]),
	DVA_GET_OFFSET(&bp->blk_dva[0]),
	DVA_GET_OFFSET(&bp->blk_dva[1])))
	zb->zb_ditto_same_ms++;
	else if (DVA_GET_VDEV(&bp->blk_dva[0]) ==
	DVA_GET_VDEV(&bp->blk_dva[2]) &&
	same_metaslab(zcb->zcb_spa,
	DVA_GET_VDEV(&bp->blk_dva[0]),
	DVA_GET_OFFSET(&bp->blk_dva[0]),
	DVA_GET_OFFSET(&bp->blk_dva[2])))
	zb->zb_ditto_same_ms++;
	else if (DVA_GET_VDEV(&bp->blk_dva[1]) ==
	DVA_GET_VDEV(&bp->blk_dva[2]) &&
	same_metaslab(zcb->zcb_spa,
	DVA_GET_VDEV(&bp->blk_dva[1]),
	DVA_GET_OFFSET(&bp->blk_dva[1]),
	DVA_GET_OFFSET(&bp->blk_dva[2])))
	zb->zb_ditto_same_ms++;
	}
	break;
	}
	}

	spa_config_exit(zcb->zcb_spa, SCL_CONFIG, FTAG);

	if (BP_IS_EMBEDDED(bp)) {
	zcb->zcb_embedded_blocks[BPE_GET_ETYPE(bp)]++;
	zcb->zcb_embedded_histogram[BPE_GET_ETYPE(bp)]
	[BPE_GET_PSIZE(bp)]++;
	return;
	}
	/*
	* The binning histogram bins by powers of two up to
	* SPA_MAXBLOCKSIZE rather than creating bins for
	* every possible blocksize found in the pool.
	*/
	int bin = highbit64(BP_GET_PSIZE(bp)) - 1;

	zcb->zcb_psize_count[bin]++;
	zcb->zcb_psize_len[bin] += BP_GET_PSIZE(bp);
	zcb->zcb_psize_total += BP_GET_PSIZE(bp);

	bin = highbit64(BP_GET_LSIZE(bp)) - 1;

	zcb->zcb_lsize_count[bin]++;
	zcb->zcb_lsize_len[bin] += BP_GET_LSIZE(bp);
	zcb->zcb_lsize_total += BP_GET_LSIZE(bp);

	bin = highbit64(BP_GET_ASIZE(bp)) - 1;

	zcb->zcb_asize_count[bin]++;
	zcb->zcb_asize_len[bin] += BP_GET_ASIZE(bp);
	zcb->zcb_asize_total += BP_GET_ASIZE(bp);

	if (dump_opt['L'])
	return;

	if (BP_GET_DEDUP(bp)) {
	ddt_t *ddt;
	ddt_entry_t *dde;

	ddt = ddt_select(zcb->zcb_spa, bp);
	ddt_enter(ddt);
	dde = ddt_lookup(ddt, bp, B_FALSE);

	if (dde == NULL) {
	refcnt = 0;
	} else {
	ddt_phys_t *ddp = ddt_phys_select(dde, bp);
	ddt_phys_decref(ddp);
	refcnt = ddp->ddp_refcnt;
	if (ddt_phys_total_refcnt(dde) == 0)
	ddt_remove(ddt, dde);
	}
	ddt_exit(ddt);
	}

	VERIFY3U(zio_wait(zio_claim(NULL, zcb->zcb_spa,
	refcnt ? 0 : spa_min_claim_txg(zcb->zcb_spa),
	bp, NULL, NULL, ZIO_FLAG_CANFAIL)), ==, 0);
	}

	static void
	zdb_blkptr_done(zio_t *zio)
	{
	spa_t *spa = zio->io_spa;
	blkptr_t *bp = zio->io_bp;
	int ioerr = zio->io_error;
	zdb_cb_t *zcb = zio->io_private;
	zbookmark_phys_t *zb = &zio->io_bookmark;

	mutex_enter(&spa->spa_scrub_lock);
	spa->spa_load_verify_bytes -= BP_GET_PSIZE(bp);
	cv_broadcast(&spa->spa_scrub_io_cv);

	if (ioerr && !(zio->io_flags & ZIO_FLAG_SPECULATIVE)) {
	char blkbuf[BP_SPRINTF_LEN];

	zcb->zcb_haderrors = 1;
	zcb->zcb_errors[ioerr]++;

	if (dump_opt['b'] >= 2)
	snprintf_blkptr(blkbuf, sizeof (blkbuf), bp);
	else
	blkbuf[0] = '\0';

	(void) printf("zdb_blkptr_cb: "
	"Got error %d reading "
	"<%llu, %llu, %lld, %llx> %s -- skipping\n",
	ioerr,
	(u_longlong_t)zb->zb_objset,
	(u_longlong_t)zb->zb_object,
	(u_longlong_t)zb->zb_level,
	(u_longlong_t)zb->zb_blkid,
	blkbuf);
	}
	mutex_exit(&spa->spa_scrub_lock);

	abd_free(zio->io_abd);
	}

	static int
	zdb_blkptr_cb(spa_t spa, zilog_t zilog, const blkptr_t *bp,
	const zbookmark_phys_t zb, const dnode_phys_t dnp, void *arg)
	{
	zdb_cb_t *zcb = arg;
	dmu_object_type_t type;
	boolean_t is_metadata;

	if (zb->zb_level == ZB_DNODE_LEVEL)
	return (0);

	if (dump_opt['b'] >= 5 && bp->blk_birth > 0) {
	char blkbuf[BP_SPRINTF_LEN];
	snprintf_blkptr(blkbuf, sizeof (blkbuf), bp);
	(void) printf("objset %llu object %llu "
	"level %lld offset 0x%llx %s\n",
	(u_longlong_t)zb->zb_objset,
	(u_longlong_t)zb->zb_object,
	(longlong_t)zb->zb_level,
	(u_longlong_t)blkid2offset(dnp, bp, zb),
	blkbuf);
	}

	if (BP_IS_HOLE(bp) \|\| BP_IS_REDACTED(bp))
	return (0);

	type = BP_GET_TYPE(bp);

	zdb_count_block(zcb, zilog, bp,
	(type & DMU_OT_NEWTYPE) ? ZDB_OT_OTHER : type);

	is_metadata = (BP_GET_LEVEL(bp) != 0 \|\| DMU_OT_IS_METADATA(type));

	if (!BP_IS_EMBEDDED(bp) &&
	(dump_opt['c'] > 1 \|\| (dump_opt['c'] && is_metadata))) {
	size_t size = BP_GET_PSIZE(bp);
	abd_t *abd = abd_alloc(size, B_FALSE);
	int flags = ZIO_FLAG_CANFAIL \| ZIO_FLAG_SCRUB \| ZIO_FLAG_RAW;

	/* If it's an intent log block, failure is expected. */
	if (zb->zb_level == ZB_ZIL_LEVEL)
	flags \|= ZIO_FLAG_SPECULATIVE;

	mutex_enter(&spa->spa_scrub_lock);
	while (spa->spa_load_verify_bytes > max_inflight_bytes)
	cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock);
	spa->spa_load_verify_bytes += size;
	mutex_exit(&spa->spa_scrub_lock);

	zio_nowait(zio_read(NULL, spa, bp, abd, size,
	zdb_blkptr_done, zcb, ZIO_PRIORITY_ASYNC_READ, flags, zb));
	}

	zcb->zcb_readfails = 0;

	/* only call gethrtime() every 100 blocks */
	static int iters;
	if (++iters > 100)
	iters = 0;
	else
	return (0);

	if (dump_opt['b'] < 5 && gethrtime() > zcb->zcb_lastprint + NANOSEC) {
	uint64_t now = gethrtime();
	char buf[10];
	uint64_t bytes = zcb->zcb_type[ZB_TOTAL][ZDB_OT_TOTAL].zb_asize;
	uint64_t kb_per_sec =
	1 + bytes / (1 + ((now - zcb->zcb_start) / 1000 / 1000));
	uint64_t sec_remaining =
	(zcb->zcb_totalasize - bytes) / 1024 / kb_per_sec;

	/* make sure nicenum has enough space */
	CTASSERT(sizeof (buf) >= NN_NUMBUF_SZ);

	zfs_nicebytes(bytes, buf, sizeof (buf));
	(void) fprintf(stderr,
	"\r%5s completed (%4"PRIu64"MB/s) "
	"estimated time remaining: "
	"%"PRIu64"hr %02"PRIu64"min %02"PRIu64"sec ",
	buf, kb_per_sec / 1024,
	sec_remaining / 60 / 60,
	sec_remaining / 60 % 60,
	sec_remaining % 60);

	zcb->zcb_lastprint = now;
	}

	return (0);
	}

	static void
	zdb_leak(void *arg, uint64_t start, uint64_t size)
	{
	vdev_t *vd = arg;

	(void) printf("leaked space: vdev %llu, offset 0x%llx, size %llu\n",
	(u_longlong_t)vd->vdev_id, (u_longlong_t)start, (u_longlong_t)size);
	}

	static metaslab_ops_t zdb_metaslab_ops = {
	NULL /* alloc */
	};

	/* ARGSUSED */
	static int
	load_unflushed_svr_segs_cb(spa_t spa, space_map_entry_t sme,
	uint64_t txg, void *arg)
	{
	spa_vdev_removal_t *svr = arg;

	uint64_t offset = sme->sme_offset;
	uint64_t size = sme->sme_run;

	/* skip vdevs we don't care about */
	if (sme->sme_vdev != svr->svr_vdev_id)
	return (0);

	vdev_t *vd = vdev_lookup_top(spa, sme->sme_vdev);
	metaslab_t *ms = vd->vdev_ms[offset >> vd->vdev_ms_shift];
	ASSERT(sme->sme_type == SM_ALLOC \|\| sme->sme_type == SM_FREE);

	if (txg < metaslab_unflushed_txg(ms))
	return (0);

	if (sme->sme_type == SM_ALLOC)
	range_tree_add(svr->svr_allocd_segs, offset, size);
	else
	range_tree_remove(svr->svr_allocd_segs, offset, size);

	return (0);
	}

	/* ARGSUSED */
	static void
	claim_segment_impl_cb(uint64_t inner_offset, vdev_t *vd, uint64_t offset,
	uint64_t size, void *arg)
	{
	/*
	* This callback was called through a remap from
	* a device being removed. Therefore, the vdev that
	* this callback is applied to is a concrete
	* vdev.
	*/
	ASSERT(vdev_is_concrete(vd));

	VERIFY0(metaslab_claim_impl(vd, offset, size,
	spa_min_claim_txg(vd->vdev_spa)));
	}

	static void
	claim_segment_cb(void *arg, uint64_t offset, uint64_t size)
	{
	vdev_t *vd = arg;

	vdev_indirect_ops.vdev_op_remap(vd, offset, size,
	claim_segment_impl_cb, NULL);
	}

	/*
	* After accounting for all allocated blocks that are directly referenced,
	* we might have missed a reference to a block from a partially complete
	* (and thus unused) indirect mapping object. We perform a secondary pass
	* through the metaslabs we have already mapped and claim the destination
	* blocks.
	*/
	static void
	zdb_claim_removing(spa_t spa, zdb_cb_t zcb)
	{
	if (dump_opt['L'])
	return;

	if (spa->spa_vdev_removal == NULL)
	return;

	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);

	spa_vdev_removal_t *svr = spa->spa_vdev_removal;
	vdev_t *vd = vdev_lookup_top(spa, svr->svr_vdev_id);
	vdev_indirect_mapping_t *vim = vd->vdev_indirect_mapping;

	ASSERT0(range_tree_space(svr->svr_allocd_segs));

	range_tree_t *allocs = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0);
	for (uint64_t msi = 0; msi < vd->vdev_ms_count; msi++) {
	metaslab_t *msp = vd->vdev_ms[msi];

	ASSERT0(range_tree_space(allocs));
	if (msp->ms_sm != NULL)
	VERIFY0(space_map_load(msp->ms_sm, allocs, SM_ALLOC));
	range_tree_vacate(allocs, range_tree_add, svr->svr_allocd_segs);
	}
	range_tree_destroy(allocs);

	iterate_through_spacemap_logs(spa, load_unflushed_svr_segs_cb, svr);

	/*
	* Clear everything past what has been synced,
	* because we have not allocated mappings for
	* it yet.
	*/
	range_tree_clear(svr->svr_allocd_segs,
	vdev_indirect_mapping_max_offset(vim),
	vd->vdev_asize - vdev_indirect_mapping_max_offset(vim));

	zcb->zcb_removing_size += range_tree_space(svr->svr_allocd_segs);
	range_tree_vacate(svr->svr_allocd_segs, claim_segment_cb, vd);

	spa_config_exit(spa, SCL_CONFIG, FTAG);
	}

	/* ARGSUSED */
	static int
	increment_indirect_mapping_cb(void arg, const blkptr_t bp, boolean_t bp_freed,
	dmu_tx_t *tx)
	{
	zdb_cb_t *zcb = arg;
	spa_t *spa = zcb->zcb_spa;
	vdev_t *vd;
	const dva_t *dva = &bp->blk_dva[0];

	ASSERT(!bp_freed);
	ASSERT(!dump_opt['L']);
	ASSERT3U(BP_GET_NDVAS(bp), ==, 1);

	spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);
	vd = vdev_lookup_top(zcb->zcb_spa, DVA_GET_VDEV(dva));
	ASSERT3P(vd, !=, NULL);
	spa_config_exit(spa, SCL_VDEV, FTAG);

	ASSERT(vd->vdev_indirect_config.vic_mapping_object != 0);
	ASSERT3P(zcb->zcb_vd_obsolete_counts[vd->vdev_id], !=, NULL);

	vdev_indirect_mapping_increment_obsolete_count(
	vd->vdev_indirect_mapping,
	DVA_GET_OFFSET(dva), DVA_GET_ASIZE(dva),
	zcb->zcb_vd_obsolete_counts[vd->vdev_id]);

	return (0);
	}

	static uint32_t *
	zdb_load_obsolete_counts(vdev_t *vd)
	{
	vdev_indirect_mapping_t *vim = vd->vdev_indirect_mapping;
	spa_t *spa = vd->vdev_spa;
	spa_condensing_indirect_phys_t *scip =
	&spa->spa_condensing_indirect_phys;
	uint64_t obsolete_sm_object;
	uint32_t *counts;

	VERIFY0(vdev_obsolete_sm_object(vd, &obsolete_sm_object));
	EQUIV(obsolete_sm_object != 0, vd->vdev_obsolete_sm != NULL);
	counts = vdev_indirect_mapping_load_obsolete_counts(vim);
	if (vd->vdev_obsolete_sm != NULL) {
	vdev_indirect_mapping_load_obsolete_spacemap(vim, counts,
	vd->vdev_obsolete_sm);
	}
	if (scip->scip_vdev == vd->vdev_id &&
	scip->scip_prev_obsolete_sm_object != 0) {
	space_map_t *prev_obsolete_sm = NULL;
	VERIFY0(space_map_open(&prev_obsolete_sm, spa->spa_meta_objset,
	scip->scip_prev_obsolete_sm_object, 0, vd->vdev_asize, 0));
	vdev_indirect_mapping_load_obsolete_spacemap(vim, counts,
	prev_obsolete_sm);
	space_map_close(prev_obsolete_sm);
	}
	return (counts);
	}

	static void
	zdb_ddt_leak_init(spa_t spa, zdb_cb_t zcb)
	{
	ddt_bookmark_t ddb;
	ddt_entry_t dde;
	int error;
	int p;

	ASSERT(!dump_opt['L']);

	bzero(&ddb, sizeof (ddb));
	while ((error = ddt_walk(spa, &ddb, &dde)) == 0) {
	blkptr_t blk;
	ddt_phys_t *ddp = dde.dde_phys;

	if (ddb.ddb_class == DDT_CLASS_UNIQUE)
	return;

	ASSERT(ddt_phys_total_refcnt(&dde) > 1);

	for (p = 0; p < DDT_PHYS_TYPES; p++, ddp++) {
	if (ddp->ddp_phys_birth == 0)
	continue;
	ddt_bp_create(ddb.ddb_checksum,
	&dde.dde_key, ddp, &blk);
	if (p == DDT_PHYS_DITTO) {
	zdb_count_block(zcb, NULL, &blk, ZDB_OT_DITTO);
	} else {
	zcb->zcb_dedup_asize +=
	BP_GET_ASIZE(&blk) * (ddp->ddp_refcnt - 1);
	zcb->zcb_dedup_blocks++;
	}
	}
	ddt_t *ddt = spa->spa_ddt[ddb.ddb_checksum];
	ddt_enter(ddt);
	VERIFY(ddt_lookup(ddt, &blk, B_TRUE) != NULL);
	ddt_exit(ddt);
	}

	ASSERT(error == ENOENT);
	}

	typedef struct checkpoint_sm_exclude_entry_arg {
	vdev_t *cseea_vd;
	uint64_t cseea_checkpoint_size;
	} checkpoint_sm_exclude_entry_arg_t;

	static int
	checkpoint_sm_exclude_entry_cb(space_map_entry_t sme, void arg)
	{
	checkpoint_sm_exclude_entry_arg_t *cseea = arg;
	vdev_t *vd = cseea->cseea_vd;
	metaslab_t *ms = vd->vdev_ms[sme->sme_offset >> vd->vdev_ms_shift];
	uint64_t end = sme->sme_offset + sme->sme_run;

	ASSERT(sme->sme_type == SM_FREE);

	/*
	* Since the vdev_checkpoint_sm exists in the vdev level
	* and the ms_sm space maps exist in the metaslab level,
	* an entry in the checkpoint space map could theoretically
	* cross the boundaries of the metaslab that it belongs.
	*
	* In reality, because of the way that we populate and
	* manipulate the checkpoint's space maps currently,
	* there shouldn't be any entries that cross metaslabs.
	* Hence the assertion below.
	*
	* That said, there is no fundamental requirement that
	* the checkpoint's space map entries should not cross
	* metaslab boundaries. So if needed we could add code
	* that handles metaslab-crossing segments in the future.
	*/
	VERIFY3U(sme->sme_offset, >=, ms->ms_start);
	VERIFY3U(end, <=, ms->ms_start + ms->ms_size);

	/*
	* By removing the entry from the allocated segments we
	* also verify that the entry is there to begin with.
	*/
	mutex_enter(&ms->ms_lock);
	range_tree_remove(ms->ms_allocatable, sme->sme_offset, sme->sme_run);
	mutex_exit(&ms->ms_lock);

	cseea->cseea_checkpoint_size += sme->sme_run;
	return (0);
	}

	static void
	zdb_leak_init_vdev_exclude_checkpoint(vdev_t vd, zdb_cb_t zcb)
	{
	spa_t *spa = vd->vdev_spa;
	space_map_t *checkpoint_sm = NULL;
	uint64_t checkpoint_sm_obj;

	/*
	* If there is no vdev_top_zap, we are in a pool whose
	* version predates the pool checkpoint feature.
	*/
	if (vd->vdev_top_zap == 0)
	return;

	/*
	* If there is no reference of the vdev_checkpoint_sm in
	* the vdev_top_zap, then one of the following scenarios
	* is true:
	*
	* 1] There is no checkpoint
	* 2] There is a checkpoint, but no checkpointed blocks
	* have been freed yet
	* 3] The current vdev is indirect
	*
	* In these cases we return immediately.
	*/
	if (zap_contains(spa_meta_objset(spa), vd->vdev_top_zap,
	VDEV_TOP_ZAP_POOL_CHECKPOINT_SM) != 0)
	return;

	VERIFY0(zap_lookup(spa_meta_objset(spa), vd->vdev_top_zap,
	VDEV_TOP_ZAP_POOL_CHECKPOINT_SM, sizeof (uint64_t), 1,
	&checkpoint_sm_obj));

	checkpoint_sm_exclude_entry_arg_t cseea;
	cseea.cseea_vd = vd;
	cseea.cseea_checkpoint_size = 0;

	VERIFY0(space_map_open(&checkpoint_sm, spa_meta_objset(spa),
	checkpoint_sm_obj, 0, vd->vdev_asize, vd->vdev_ashift));

	VERIFY0(space_map_iterate(checkpoint_sm,
	space_map_length(checkpoint_sm),
	checkpoint_sm_exclude_entry_cb, &cseea));
	space_map_close(checkpoint_sm);

	zcb->zcb_checkpoint_size += cseea.cseea_checkpoint_size;
	}

	static void
	zdb_leak_init_exclude_checkpoint(spa_t spa, zdb_cb_t zcb)
	{
	ASSERT(!dump_opt['L']);

	vdev_t *rvd = spa->spa_root_vdev;
	for (uint64_t c = 0; c < rvd->vdev_children; c++) {
	ASSERT3U(c, ==, rvd->vdev_child[c]->vdev_id);
	zdb_leak_init_vdev_exclude_checkpoint(rvd->vdev_child[c], zcb);
	}
	}

	static int
	count_unflushed_space_cb(spa_t spa, space_map_entry_t sme,
	uint64_t txg, void *arg)
	{
	int64_t *ualloc_space = arg;

	uint64_t offset = sme->sme_offset;
	uint64_t vdev_id = sme->sme_vdev;

	vdev_t *vd = vdev_lookup_top(spa, vdev_id);
	if (!vdev_is_concrete(vd))
	return (0);

	metaslab_t *ms = vd->vdev_ms[offset >> vd->vdev_ms_shift];
	ASSERT(sme->sme_type == SM_ALLOC \|\| sme->sme_type == SM_FREE);

	if (txg < metaslab_unflushed_txg(ms))
	return (0);

	if (sme->sme_type == SM_ALLOC)
	*ualloc_space += sme->sme_run;
	else
	*ualloc_space -= sme->sme_run;

	return (0);
	}

	static int64_t
	get_unflushed_alloc_space(spa_t *spa)
	{
	if (dump_opt['L'])
	return (0);

	int64_t ualloc_space = 0;
	iterate_through_spacemap_logs(spa, count_unflushed_space_cb,
	&ualloc_space);
	return (ualloc_space);
	}

	static int
	load_unflushed_cb(spa_t spa, space_map_entry_t sme, uint64_t txg, void *arg)
	{
	maptype_t *uic_maptype = arg;

	uint64_t offset = sme->sme_offset;
	uint64_t size = sme->sme_run;
	uint64_t vdev_id = sme->sme_vdev;

	vdev_t *vd = vdev_lookup_top(spa, vdev_id);

	/* skip indirect vdevs */
	if (!vdev_is_concrete(vd))
	return (0);

	metaslab_t *ms = vd->vdev_ms[offset >> vd->vdev_ms_shift];

	ASSERT(sme->sme_type == SM_ALLOC \|\| sme->sme_type == SM_FREE);
	ASSERT(uic_maptype == SM_ALLOC \|\| uic_maptype == SM_FREE);

	if (txg < metaslab_unflushed_txg(ms))
	return (0);

	if (*uic_maptype == sme->sme_type)
	range_tree_add(ms->ms_allocatable, offset, size);
	else
	range_tree_remove(ms->ms_allocatable, offset, size);

	return (0);
	}

	static void
	load_unflushed_to_ms_allocatables(spa_t *spa, maptype_t maptype)
	{
	iterate_through_spacemap_logs(spa, load_unflushed_cb, &maptype);
	}

	static void
	load_concrete_ms_allocatable_trees(spa_t *spa, maptype_t maptype)
	{
	vdev_t *rvd = spa->spa_root_vdev;
	for (uint64_t i = 0; i < rvd->vdev_children; i++) {
	vdev_t *vd = rvd->vdev_child[i];

	ASSERT3U(i, ==, vd->vdev_id);

	if (vd->vdev_ops == &vdev_indirect_ops)
	continue;

	for (uint64_t m = 0; m < vd->vdev_ms_count; m++) {
	metaslab_t *msp = vd->vdev_ms[m];

	(void) fprintf(stderr,
	"\rloading concrete vdev %llu, "
	"metaslab %llu of %llu ...",
	(longlong_t)vd->vdev_id,
	(longlong_t)msp->ms_id,
	(longlong_t)vd->vdev_ms_count);

	mutex_enter(&msp->ms_lock);
	range_tree_vacate(msp->ms_allocatable, NULL, NULL);

	/*
	* We don't want to spend the CPU manipulating the
	* size-ordered tree, so clear the range_tree ops.
	*/
	msp->ms_allocatable->rt_ops = NULL;

	if (msp->ms_sm != NULL) {
	VERIFY0(space_map_load(msp->ms_sm,
	msp->ms_allocatable, maptype));
	}
	if (!msp->ms_loaded)
	msp->ms_loaded = B_TRUE;
	mutex_exit(&msp->ms_lock);
	}
	}

	load_unflushed_to_ms_allocatables(spa, maptype);
	}

	/*
	* vm_idxp is an in-out parameter which (for indirect vdevs) is the
	* index in vim_entries that has the first entry in this metaslab.
	* On return, it will be set to the first entry after this metaslab.
	*/
	static void
	load_indirect_ms_allocatable_tree(vdev_t vd, metaslab_t msp,
	uint64_t *vim_idxp)
	{
	vdev_indirect_mapping_t *vim = vd->vdev_indirect_mapping;

	mutex_enter(&msp->ms_lock);
	range_tree_vacate(msp->ms_allocatable, NULL, NULL);

	/*
	* We don't want to spend the CPU manipulating the
	* size-ordered tree, so clear the range_tree ops.
	*/
	msp->ms_allocatable->rt_ops = NULL;

	for (; *vim_idxp < vdev_indirect_mapping_num_entries(vim);
	(*vim_idxp)++) {
	vdev_indirect_mapping_entry_phys_t *vimep =
	&vim->vim_entries[*vim_idxp];
	uint64_t ent_offset = DVA_MAPPING_GET_SRC_OFFSET(vimep);
	uint64_t ent_len = DVA_GET_ASIZE(&vimep->vimep_dst);
	ASSERT3U(ent_offset, >=, msp->ms_start);
	if (ent_offset >= msp->ms_start + msp->ms_size)
	break;

	/*
	* Mappings do not cross metaslab boundaries,
	* because we create them by walking the metaslabs.
	*/
	ASSERT3U(ent_offset + ent_len, <=,
	msp->ms_start + msp->ms_size);
	range_tree_add(msp->ms_allocatable, ent_offset, ent_len);
	}

	if (!msp->ms_loaded)
	msp->ms_loaded = B_TRUE;
	mutex_exit(&msp->ms_lock);
	}

	static void
	zdb_leak_init_prepare_indirect_vdevs(spa_t spa, zdb_cb_t zcb)
	{
	ASSERT(!dump_opt['L']);

	vdev_t *rvd = spa->spa_root_vdev;
	for (uint64_t c = 0; c < rvd->vdev_children; c++) {
	vdev_t *vd = rvd->vdev_child[c];

	ASSERT3U(c, ==, vd->vdev_id);

	if (vd->vdev_ops != &vdev_indirect_ops)
	continue;

	/*
	* Note: we don't check for mapping leaks on
	* removing vdevs because their ms_allocatable's
	* are used to look for leaks in allocated space.
	*/
	zcb->zcb_vd_obsolete_counts[c] = zdb_load_obsolete_counts(vd);

	/*
	* Normally, indirect vdevs don't have any
	* metaslabs. We want to set them up for
	* zio_claim().
	*/
	vdev_metaslab_group_create(vd);
	VERIFY0(vdev_metaslab_init(vd, 0));

	vdev_indirect_mapping_t *vim __maybe_unused =
	vd->vdev_indirect_mapping;
	uint64_t vim_idx = 0;
	for (uint64_t m = 0; m < vd->vdev_ms_count; m++) {

	(void) fprintf(stderr,
	"\rloading indirect vdev %llu, "
	"metaslab %llu of %llu ...",
	(longlong_t)vd->vdev_id,
	(longlong_t)vd->vdev_ms[m]->ms_id,
	(longlong_t)vd->vdev_ms_count);

	load_indirect_ms_allocatable_tree(vd, vd->vdev_ms[m],
	&vim_idx);
	}
	ASSERT3U(vim_idx, ==, vdev_indirect_mapping_num_entries(vim));
	}
	}

	static void
	zdb_leak_init(spa_t spa, zdb_cb_t zcb)
	{
	zcb->zcb_spa = spa;

	if (dump_opt['L'])
	return;

	dsl_pool_t *dp = spa->spa_dsl_pool;
	vdev_t *rvd = spa->spa_root_vdev;

	/*
	* We are going to be changing the meaning of the metaslab's
	* ms_allocatable. Ensure that the allocator doesn't try to
	* use the tree.
	*/
	spa->spa_normal_class->mc_ops = &zdb_metaslab_ops;
	spa->spa_log_class->mc_ops = &zdb_metaslab_ops;
	spa->spa_embedded_log_class->mc_ops = &zdb_metaslab_ops;

	zcb->zcb_vd_obsolete_counts =
	umem_zalloc(rvd->vdev_children * sizeof (uint32_t *),
	UMEM_NOFAIL);

	/*
	* For leak detection, we overload the ms_allocatable trees
	* to contain allocated segments instead of free segments.
	* As a result, we can't use the normal metaslab_load/unload
	* interfaces.
	*/
	zdb_leak_init_prepare_indirect_vdevs(spa, zcb);
	load_concrete_ms_allocatable_trees(spa, SM_ALLOC);

	/*
	* On load_concrete_ms_allocatable_trees() we loaded all the
	* allocated entries from the ms_sm to the ms_allocatable for
	* each metaslab. If the pool has a checkpoint or is in the
	* middle of discarding a checkpoint, some of these blocks
	* may have been freed but their ms_sm may not have been
	* updated because they are referenced by the checkpoint. In
	* order to avoid false-positives during leak-detection, we
	* go through the vdev's checkpoint space map and exclude all
	* its entries from their relevant ms_allocatable.
	*
	* We also aggregate the space held by the checkpoint and add
	* it to zcb_checkpoint_size.
	*
	* Note that at this point we are also verifying that all the
	* entries on the checkpoint_sm are marked as allocated in
	* the ms_sm of their relevant metaslab.
	* [see comment in checkpoint_sm_exclude_entry_cb()]
	*/
	zdb_leak_init_exclude_checkpoint(spa, zcb);
	ASSERT3U(zcb->zcb_checkpoint_size, ==, spa_get_checkpoint_space(spa));

	/* for cleaner progress output */
	(void) fprintf(stderr, "\n");

	if (bpobj_is_open(&dp->dp_obsolete_bpobj)) {
	ASSERT(spa_feature_is_enabled(spa,
	SPA_FEATURE_DEVICE_REMOVAL));
	(void) bpobj_iterate_nofree(&dp->dp_obsolete_bpobj,
	increment_indirect_mapping_cb, zcb, NULL);
	}

	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	zdb_ddt_leak_init(spa, zcb);
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	}

	static boolean_t
	zdb_check_for_obsolete_leaks(vdev_t vd, zdb_cb_t zcb)
	{
	boolean_t leaks = B_FALSE;
	vdev_indirect_mapping_t *vim = vd->vdev_indirect_mapping;
	uint64_t total_leaked = 0;
	boolean_t are_precise = B_FALSE;

	ASSERT(vim != NULL);

	for (uint64_t i = 0; i < vdev_indirect_mapping_num_entries(vim); i++) {
	vdev_indirect_mapping_entry_phys_t *vimep =
	&vim->vim_entries[i];
	uint64_t obsolete_bytes = 0;
	uint64_t offset = DVA_MAPPING_GET_SRC_OFFSET(vimep);
	metaslab_t *msp = vd->vdev_ms[offset >> vd->vdev_ms_shift];

	/*
	* This is not very efficient but it's easy to
	* verify correctness.
	*/
	for (uint64_t inner_offset = 0;
	inner_offset < DVA_GET_ASIZE(&vimep->vimep_dst);
	inner_offset += 1 << vd->vdev_ashift) {
	if (range_tree_contains(msp->ms_allocatable,
	offset + inner_offset, 1 << vd->vdev_ashift)) {
	obsolete_bytes += 1 << vd->vdev_ashift;
	}
	}

	int64_t bytes_leaked = obsolete_bytes -
	zcb->zcb_vd_obsolete_counts[vd->vdev_id][i];
	ASSERT3U(DVA_GET_ASIZE(&vimep->vimep_dst), >=,
	zcb->zcb_vd_obsolete_counts[vd->vdev_id][i]);

	VERIFY0(vdev_obsolete_counts_are_precise(vd, &are_precise));
	if (bytes_leaked != 0 && (are_precise \|\| dump_opt['d'] >= 5)) {
	(void) printf("obsolete indirect mapping count "
	"mismatch on %llu:%llx:%llx : %llx bytes leaked\n",
	(u_longlong_t)vd->vdev_id,
	(u_longlong_t)DVA_MAPPING_GET_SRC_OFFSET(vimep),
	(u_longlong_t)DVA_GET_ASIZE(&vimep->vimep_dst),
	(u_longlong_t)bytes_leaked);
	}
	total_leaked += ABS(bytes_leaked);
	}

	VERIFY0(vdev_obsolete_counts_are_precise(vd, &are_precise));
	if (!are_precise && total_leaked > 0) {
	int pct_leaked = total_leaked * 100 /
	vdev_indirect_mapping_bytes_mapped(vim);
	(void) printf("cannot verify obsolete indirect mapping "
	"counts of vdev %llu because precise feature was not "
	"enabled when it was removed: %d%% (%llx bytes) of mapping"
	"unreferenced\n",
	(u_longlong_t)vd->vdev_id, pct_leaked,
	(u_longlong_t)total_leaked);
	} else if (total_leaked > 0) {
	(void) printf("obsolete indirect mapping count mismatch "
	"for vdev %llu -- %llx total bytes mismatched\n",
	(u_longlong_t)vd->vdev_id,
	(u_longlong_t)total_leaked);
	leaks \|= B_TRUE;
	}

	vdev_indirect_mapping_free_obsolete_counts(vim,
	zcb->zcb_vd_obsolete_counts[vd->vdev_id]);
	zcb->zcb_vd_obsolete_counts[vd->vdev_id] = NULL;

	return (leaks);
	}

	static boolean_t
	zdb_leak_fini(spa_t spa, zdb_cb_t zcb)
	{
	if (dump_opt['L'])
	return (B_FALSE);

	boolean_t leaks = B_FALSE;
	vdev_t *rvd = spa->spa_root_vdev;
	for (unsigned c = 0; c < rvd->vdev_children; c++) {
	vdev_t *vd = rvd->vdev_child[c];

	if (zcb->zcb_vd_obsolete_counts[c] != NULL) {
	leaks \|= zdb_check_for_obsolete_leaks(vd, zcb);
	}

	for (uint64_t m = 0; m < vd->vdev_ms_count; m++) {
	metaslab_t *msp = vd->vdev_ms[m];
	ASSERT3P(msp->ms_group, ==, (msp->ms_group->mg_class ==
	spa_embedded_log_class(spa)) ?
	vd->vdev_log_mg : vd->vdev_mg);

	/*
	* ms_allocatable has been overloaded
	* to contain allocated segments. Now that
	* we finished traversing all blocks, any
	* block that remains in the ms_allocatable
	* represents an allocated block that we
	* did not claim during the traversal.
	* Claimed blocks would have been removed
	* from the ms_allocatable. For indirect
	* vdevs, space remaining in the tree
	* represents parts of the mapping that are
	* not referenced, which is not a bug.
	*/
	if (vd->vdev_ops == &vdev_indirect_ops) {
	range_tree_vacate(msp->ms_allocatable,
	NULL, NULL);
	} else {
	range_tree_vacate(msp->ms_allocatable,
	zdb_leak, vd);
	}
	if (msp->ms_loaded) {
	msp->ms_loaded = B_FALSE;
	}
	}
	}

	umem_free(zcb->zcb_vd_obsolete_counts,
	rvd->vdev_children * sizeof (uint32_t *));
	zcb->zcb_vd_obsolete_counts = NULL;

	return (leaks);
	}

	/* ARGSUSED */
	static int
	count_block_cb(void arg, const blkptr_t bp, dmu_tx_t *tx)
	{
	zdb_cb_t *zcb = arg;

	if (dump_opt['b'] >= 5) {
	char blkbuf[BP_SPRINTF_LEN];
	snprintf_blkptr(blkbuf, sizeof (blkbuf), bp);
	(void) printf("[%s] %s\n",
	"deferred free", blkbuf);
	}
	zdb_count_block(zcb, NULL, bp, ZDB_OT_DEFERRED);
	return (0);
	}

	/*
	* Iterate over livelists which have been destroyed by the user but
	* are still present in the MOS, waiting to be freed
	*/
	static void
	iterate_deleted_livelists(spa_t spa, ll_iter_t func, void arg)
	{
	objset_t *mos = spa->spa_meta_objset;
	uint64_t zap_obj;
	int err = zap_lookup(mos, DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_DELETED_CLONES, sizeof (uint64_t), 1, &zap_obj);
	if (err == ENOENT)
	return;
	ASSERT0(err);

	zap_cursor_t zc;
	zap_attribute_t attr;
	dsl_deadlist_t ll;
	/* NULL out os prior to dsl_deadlist_open in case it's garbage */
	ll.dl_os = NULL;
	for (zap_cursor_init(&zc, mos, zap_obj);
	zap_cursor_retrieve(&zc, &attr) == 0;
	(void) zap_cursor_advance(&zc)) {
	dsl_deadlist_open(&ll, mos, attr.za_first_integer);
	func(&ll, arg);
	dsl_deadlist_close(&ll);
	}
	zap_cursor_fini(&zc);
	}

	static int
	bpobj_count_block_cb(void arg, const blkptr_t bp, boolean_t bp_freed,
	dmu_tx_t *tx)
	{
	ASSERT(!bp_freed);
	return (count_block_cb(arg, bp, tx));
	}

	static int
	livelist_entry_count_blocks_cb(void args, dsl_deadlist_entry_t dle)
	{
	zdb_cb_t *zbc = args;
	bplist_t blks;
	bplist_create(&blks);
	/* determine which blocks have been alloc'd but not freed */
	VERIFY0(dsl_process_sub_livelist(&dle->dle_bpobj, &blks, NULL, NULL));
	/* count those blocks */
	(void) bplist_iterate(&blks, count_block_cb, zbc, NULL);
	bplist_destroy(&blks);
	return (0);
	}

	static void
	livelist_count_blocks(dsl_deadlist_t ll, void arg)
	{
	dsl_deadlist_iterate(ll, livelist_entry_count_blocks_cb, arg);
	}

	/*
	* Count the blocks in the livelists that have been destroyed by the user
	* but haven't yet been freed.
	*/
	static void
	deleted_livelists_count_blocks(spa_t spa, zdb_cb_t zbc)
	{
	iterate_deleted_livelists(spa, livelist_count_blocks, zbc);
	}

	static void
	dump_livelist_cb(dsl_deadlist_t ll, void arg)
	{
	ASSERT3P(arg, ==, NULL);
	global_feature_count[SPA_FEATURE_LIVELIST]++;
	dump_blkptr_list(ll, "Deleted Livelist");
	dsl_deadlist_iterate(ll, sublivelist_verify_lightweight, NULL);
	}

	/*
	* Print out, register object references to, and increment feature counts for
	* livelists that have been destroyed by the user but haven't yet been freed.
	*/
	static void
	deleted_livelists_dump_mos(spa_t *spa)
	{
	uint64_t zap_obj;
	objset_t *mos = spa->spa_meta_objset;
	int err = zap_lookup(mos, DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_DELETED_CLONES, sizeof (uint64_t), 1, &zap_obj);
	if (err == ENOENT)
	return;
	mos_obj_refd(zap_obj);
	iterate_deleted_livelists(spa, dump_livelist_cb, NULL);
	}

	static int
	dump_block_stats(spa_t *spa)
	{
	zdb_cb_t zcb;
	zdb_blkstats_t zb, tzb;
	uint64_t norm_alloc, norm_space, total_alloc, total_found;
	int flags = TRAVERSE_PRE \| TRAVERSE_PREFETCH_METADATA \|
	TRAVERSE_NO_DECRYPT \| TRAVERSE_HARD;
	boolean_t leaks = B_FALSE;
	int e, c, err;
	bp_embedded_type_t i;

	bzero(&zcb, sizeof (zcb));
	(void) printf("\nTraversing all blocks %s%s%s%s%s...\n\n",
	(dump_opt['c'] \|\| !dump_opt['L']) ? "to verify " : "",
	(dump_opt['c'] == 1) ? "metadata " : "",
	dump_opt['c'] ? "checksums " : "",
	(dump_opt['c'] && !dump_opt['L']) ? "and verify " : "",
	!dump_opt['L'] ? "nothing leaked " : "");

	/*
	* When leak detection is enabled we load all space maps as SM_ALLOC
	* maps, then traverse the pool claiming each block we discover. If
	* the pool is perfectly consistent, the segment trees will be empty
	* when we're done. Anything left over is a leak; any block we can't
	* claim (because it's not part of any space map) is a double
	* allocation, reference to a freed block, or an unclaimed log block.
	*
	* When leak detection is disabled (-L option) we still traverse the
	* pool claiming each block we discover, but we skip opening any space
	* maps.
	*/
	bzero(&zcb, sizeof (zdb_cb_t));
	zdb_leak_init(spa, &zcb);

	/*
	* If there's a deferred-free bplist, process that first.
	*/
	(void) bpobj_iterate_nofree(&spa->spa_deferred_bpobj,
	bpobj_count_block_cb, &zcb, NULL);

	if (spa_version(spa) >= SPA_VERSION_DEADLISTS) {
	(void) bpobj_iterate_nofree(&spa->spa_dsl_pool->dp_free_bpobj,
	bpobj_count_block_cb, &zcb, NULL);
	}

	zdb_claim_removing(spa, &zcb);

	if (spa_feature_is_active(spa, SPA_FEATURE_ASYNC_DESTROY)) {
	VERIFY3U(0, ==, bptree_iterate(spa->spa_meta_objset,
	spa->spa_dsl_pool->dp_bptree_obj, B_FALSE, count_block_cb,
	&zcb, NULL));
	}

	deleted_livelists_count_blocks(spa, &zcb);

	if (dump_opt['c'] > 1)
	flags \|= TRAVERSE_PREFETCH_DATA;

	zcb.zcb_totalasize = metaslab_class_get_alloc(spa_normal_class(spa));
	zcb.zcb_totalasize += metaslab_class_get_alloc(spa_special_class(spa));
	zcb.zcb_totalasize += metaslab_class_get_alloc(spa_dedup_class(spa));
	zcb.zcb_totalasize +=
	metaslab_class_get_alloc(spa_embedded_log_class(spa));
	zcb.zcb_start = zcb.zcb_lastprint = gethrtime();
	err = traverse_pool(spa, 0, flags, zdb_blkptr_cb, &zcb);

	/*
	* If we've traversed the data blocks then we need to wait for those
	* I/Os to complete. We leverage "The Godfather" zio to wait on
	* all async I/Os to complete.
	*/
	if (dump_opt['c']) {
	for (c = 0; c < max_ncpus; c++) {
	(void) zio_wait(spa->spa_async_zio_root[c]);
	spa->spa_async_zio_root[c] = zio_root(spa, NULL, NULL,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_SPECULATIVE \|
	ZIO_FLAG_GODFATHER);
	}
	}
	ASSERT0(spa->spa_load_verify_bytes);

	/*
	* Done after zio_wait() since zcb_haderrors is modified in
	* zdb_blkptr_done()
	*/
	zcb.zcb_haderrors \|= err;

	if (zcb.zcb_haderrors) {
	(void) printf("\nError counts:\n\n");
	(void) printf("\t%5s %s\n", "errno", "count");
	for (e = 0; e < 256; e++) {
	if (zcb.zcb_errors[e] != 0) {
	(void) printf("\t%5d %llu\n",
	e, (u_longlong_t)zcb.zcb_errors[e]);
	}
	}
	}

	/*
	* Report any leaked segments.
	*/
	leaks \|= zdb_leak_fini(spa, &zcb);

	tzb = &zcb.zcb_type[ZB_TOTAL][ZDB_OT_TOTAL];

	norm_alloc = metaslab_class_get_alloc(spa_normal_class(spa));
	norm_space = metaslab_class_get_space(spa_normal_class(spa));

	total_alloc = norm_alloc +
	metaslab_class_get_alloc(spa_log_class(spa)) +
	metaslab_class_get_alloc(spa_embedded_log_class(spa)) +
	metaslab_class_get_alloc(spa_special_class(spa)) +
	metaslab_class_get_alloc(spa_dedup_class(spa)) +
	get_unflushed_alloc_space(spa);
	total_found = tzb->zb_asize - zcb.zcb_dedup_asize +
	zcb.zcb_removing_size + zcb.zcb_checkpoint_size;

	if (total_found == total_alloc && !dump_opt['L']) {
	(void) printf("\n\tNo leaks (block sum matches space"
	" maps exactly)\n");
	} else if (!dump_opt['L']) {
	(void) printf("block traversal size %llu != alloc %llu "
	"(%s %lld)\n",
	(u_longlong_t)total_found,
	(u_longlong_t)total_alloc,
	(dump_opt['L']) ? "unreachable" : "leaked",
	(longlong_t)(total_alloc - total_found));
	leaks = B_TRUE;
	}

	if (tzb->zb_count == 0)
	return (2);

	(void) printf("\n");
	(void) printf("\t%-16s %14llu\n", "bp count:",
	(u_longlong_t)tzb->zb_count);
	(void) printf("\t%-16s %14llu\n", "ganged count:",
	(longlong_t)tzb->zb_gangs);
	(void) printf("\t%-16s %14llu avg: %6llu\n", "bp logical:",
	(u_longlong_t)tzb->zb_lsize,
	(u_longlong_t)(tzb->zb_lsize / tzb->zb_count));
	(void) printf("\t%-16s %14llu avg: %6llu compression: %6.2f\n",
	"bp physical:", (u_longlong_t)tzb->zb_psize,
	(u_longlong_t)(tzb->zb_psize / tzb->zb_count),
	(double)tzb->zb_lsize / tzb->zb_psize);
	(void) printf("\t%-16s %14llu avg: %6llu compression: %6.2f\n",
	"bp allocated:", (u_longlong_t)tzb->zb_asize,
	(u_longlong_t)(tzb->zb_asize / tzb->zb_count),
	(double)tzb->zb_lsize / tzb->zb_asize);
	(void) printf("\t%-16s %14llu ref>1: %6llu deduplication: %6.2f\n",
	"bp deduped:", (u_longlong_t)zcb.zcb_dedup_asize,
	(u_longlong_t)zcb.zcb_dedup_blocks,
	(double)zcb.zcb_dedup_asize / tzb->zb_asize + 1.0);
	(void) printf("\t%-16s %14llu used: %5.2f%%\n", "Normal class:",
	(u_longlong_t)norm_alloc, 100.0 * norm_alloc / norm_space);

	if (spa_special_class(spa)->mc_allocator[0].mca_rotor != NULL) {
	uint64_t alloc = metaslab_class_get_alloc(
	spa_special_class(spa));
	uint64_t space = metaslab_class_get_space(
	spa_special_class(spa));

	(void) printf("\t%-16s %14llu used: %5.2f%%\n",
	"Special class", (u_longlong_t)alloc,
	100.0 * alloc / space);
	}

	if (spa_dedup_class(spa)->mc_allocator[0].mca_rotor != NULL) {
	uint64_t alloc = metaslab_class_get_alloc(
	spa_dedup_class(spa));
	uint64_t space = metaslab_class_get_space(
	spa_dedup_class(spa));

	(void) printf("\t%-16s %14llu used: %5.2f%%\n",
	"Dedup class", (u_longlong_t)alloc,
	100.0 * alloc / space);
	}

	if (spa_embedded_log_class(spa)->mc_allocator[0].mca_rotor != NULL) {
	uint64_t alloc = metaslab_class_get_alloc(
	spa_embedded_log_class(spa));
	uint64_t space = metaslab_class_get_space(
	spa_embedded_log_class(spa));

	(void) printf("\t%-16s %14llu used: %5.2f%%\n",
	"Embedded log class", (u_longlong_t)alloc,
	100.0 * alloc / space);
	}

	for (i = 0; i < NUM_BP_EMBEDDED_TYPES; i++) {
	if (zcb.zcb_embedded_blocks[i] == 0)
	continue;
	(void) printf("\n");
	(void) printf("\tadditional, non-pointer bps of type %u: "
	"%10llu\n",
	i, (u_longlong_t)zcb.zcb_embedded_blocks[i]);

	if (dump_opt['b'] >= 3) {
	(void) printf("\t number of (compressed) bytes: "
	"number of bps\n");
	dump_histogram(zcb.zcb_embedded_histogram[i],
	sizeof (zcb.zcb_embedded_histogram[i]) /
	sizeof (zcb.zcb_embedded_histogram[i][0]), 0);
	}
	}

	if (tzb->zb_ditto_samevdev != 0) {
	(void) printf("\tDittoed blocks on same vdev: %llu\n",
	(longlong_t)tzb->zb_ditto_samevdev);
	}
	if (tzb->zb_ditto_same_ms != 0) {
	(void) printf("\tDittoed blocks in same metaslab: %llu\n",
	(longlong_t)tzb->zb_ditto_same_ms);
	}

	for (uint64_t v = 0; v < spa->spa_root_vdev->vdev_children; v++) {
	vdev_t *vd = spa->spa_root_vdev->vdev_child[v];
	vdev_indirect_mapping_t *vim = vd->vdev_indirect_mapping;

	if (vim == NULL) {
	continue;
	}

	char mem[32];
	zdb_nicenum(vdev_indirect_mapping_num_entries(vim),
	mem, vdev_indirect_mapping_size(vim));

	(void) printf("\tindirect vdev id %llu has %llu segments "
	"(%s in memory)\n",
	(longlong_t)vd->vdev_id,
	(longlong_t)vdev_indirect_mapping_num_entries(vim), mem);
	}

	if (dump_opt['b'] >= 2) {
	int l, t, level;
	(void) printf("\nBlocks\tLSIZE\tPSIZE\tASIZE"
	"\t avg\t comp\t%%Total\tType\n");

	for (t = 0; t <= ZDB_OT_TOTAL; t++) {
	char csize[32], lsize[32], psize[32], asize[32];
	char avg[32], gang[32];
	const char *typename;

	/* make sure nicenum has enough space */
	CTASSERT(sizeof (csize) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (lsize) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (psize) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (asize) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (avg) >= NN_NUMBUF_SZ);
	CTASSERT(sizeof (gang) >= NN_NUMBUF_SZ);

	if (t < DMU_OT_NUMTYPES)
	typename = dmu_ot[t].ot_name;
	else
	typename = zdb_ot_extname[t - DMU_OT_NUMTYPES];

	if (zcb.zcb_type[ZB_TOTAL][t].zb_asize == 0) {
	(void) printf("%6s\t%5s\t%5s\t%5s"
	"\t%5s\t%5s\t%6s\t%s\n",
	"-",
	"-",
	"-",
	"-",
	"-",
	"-",
	"-",
	typename);
	continue;
	}

	for (l = ZB_TOTAL - 1; l >= -1; l--) {
	level = (l == -1 ? ZB_TOTAL : l);
	zb = &zcb.zcb_type[level][t];

	if (zb->zb_asize == 0)
	continue;

	if (dump_opt['b'] < 3 && level != ZB_TOTAL)
	continue;

	if (level == 0 && zb->zb_asize ==
	zcb.zcb_type[ZB_TOTAL][t].zb_asize)
	continue;

	zdb_nicenum(zb->zb_count, csize,
	sizeof (csize));
	zdb_nicenum(zb->zb_lsize, lsize,
	sizeof (lsize));
	zdb_nicenum(zb->zb_psize, psize,
	sizeof (psize));
	zdb_nicenum(zb->zb_asize, asize,
	sizeof (asize));
	zdb_nicenum(zb->zb_asize / zb->zb_count, avg,
	sizeof (avg));
	zdb_nicenum(zb->zb_gangs, gang, sizeof (gang));

	(void) printf("%6s\t%5s\t%5s\t%5s\t%5s"
	"\t%5.2f\t%6.2f\t",
	csize, lsize, psize, asize, avg,
	(double)zb->zb_lsize / zb->zb_psize,
	100.0 * zb->zb_asize / tzb->zb_asize);

	if (level == ZB_TOTAL)
	(void) printf("%s\n", typename);
	else
	(void) printf(" L%d %s\n",
	level, typename);

	if (dump_opt['b'] >= 3 && zb->zb_gangs > 0) {
	(void) printf("\t number of ganged "
	"blocks: %s\n", gang);
	}

	if (dump_opt['b'] >= 4) {
	(void) printf("psize "
	"(in 512-byte sectors): "
	"number of blocks\n");
	dump_histogram(zb->zb_psize_histogram,
	PSIZE_HISTO_SIZE, 0);
	}
	}
	}

	/* Output a table summarizing block sizes in the pool */
	if (dump_opt['b'] >= 2) {
	dump_size_histograms(&zcb);
	}
	}

	(void) printf("\n");

	if (leaks)
	return (2);

	if (zcb.zcb_haderrors)
	return (3);

	return (0);
	}

	typedef struct zdb_ddt_entry {
	ddt_key_t zdde_key;
	uint64_t zdde_ref_blocks;
	uint64_t zdde_ref_lsize;
	uint64_t zdde_ref_psize;
	uint64_t zdde_ref_dsize;
	avl_node_t zdde_node;
	} zdb_ddt_entry_t;

	/* ARGSUSED */
	static int
	zdb_ddt_add_cb(spa_t spa, zilog_t zilog, const blkptr_t *bp,
	const zbookmark_phys_t zb, const dnode_phys_t dnp, void *arg)
	{
	avl_tree_t *t = arg;
	avl_index_t where;
	zdb_ddt_entry_t *zdde, zdde_search;

	if (zb->zb_level == ZB_DNODE_LEVEL \|\| BP_IS_HOLE(bp) \|\|
	BP_IS_EMBEDDED(bp))
	return (0);

	if (dump_opt['S'] > 1 && zb->zb_level == ZB_ROOT_LEVEL) {
	(void) printf("traversing objset %llu, %llu objects, "
	"%lu blocks so far\n",
	(u_longlong_t)zb->zb_objset,
	(u_longlong_t)BP_GET_FILL(bp),
	avl_numnodes(t));
	}

	if (BP_IS_HOLE(bp) \|\| BP_GET_CHECKSUM(bp) == ZIO_CHECKSUM_OFF \|\|
	BP_GET_LEVEL(bp) > 0 \|\| DMU_OT_IS_METADATA(BP_GET_TYPE(bp)))
	return (0);

	ddt_key_fill(&zdde_search.zdde_key, bp);

	zdde = avl_find(t, &zdde_search, &where);

	if (zdde == NULL) {
	zdde = umem_zalloc(sizeof (*zdde), UMEM_NOFAIL);
	zdde->zdde_key = zdde_search.zdde_key;
	avl_insert(t, zdde, where);
	}

	zdde->zdde_ref_blocks += 1;
	zdde->zdde_ref_lsize += BP_GET_LSIZE(bp);
	zdde->zdde_ref_psize += BP_GET_PSIZE(bp);
	zdde->zdde_ref_dsize += bp_get_dsize_sync(spa, bp);

	return (0);
	}

	static void
	dump_simulated_ddt(spa_t *spa)
	{
	avl_tree_t t;
	void *cookie = NULL;
	zdb_ddt_entry_t *zdde;
	ddt_histogram_t ddh_total;
	ddt_stat_t dds_total;

	bzero(&ddh_total, sizeof (ddh_total));
	bzero(&dds_total, sizeof (dds_total));
	avl_create(&t, ddt_entry_compare,
	sizeof (zdb_ddt_entry_t), offsetof(zdb_ddt_entry_t, zdde_node));

	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);

	(void) traverse_pool(spa, 0, TRAVERSE_PRE \| TRAVERSE_PREFETCH_METADATA \|
	TRAVERSE_NO_DECRYPT, zdb_ddt_add_cb, &t);

	spa_config_exit(spa, SCL_CONFIG, FTAG);

	while ((zdde = avl_destroy_nodes(&t, &cookie)) != NULL) {
	ddt_stat_t dds;
	uint64_t refcnt = zdde->zdde_ref_blocks;
	ASSERT(refcnt != 0);

	dds.dds_blocks = zdde->zdde_ref_blocks / refcnt;
	dds.dds_lsize = zdde->zdde_ref_lsize / refcnt;
	dds.dds_psize = zdde->zdde_ref_psize / refcnt;
	dds.dds_dsize = zdde->zdde_ref_dsize / refcnt;

	dds.dds_ref_blocks = zdde->zdde_ref_blocks;
	dds.dds_ref_lsize = zdde->zdde_ref_lsize;
	dds.dds_ref_psize = zdde->zdde_ref_psize;
	dds.dds_ref_dsize = zdde->zdde_ref_dsize;

	ddt_stat_add(&ddh_total.ddh_stat[highbit64(refcnt) - 1],
	&dds, 0);

	umem_free(zdde, sizeof (*zdde));
	}

	avl_destroy(&t);

	ddt_histogram_stat(&dds_total, &ddh_total);

	(void) printf("Simulated DDT histogram:\n");

	zpool_dump_ddt(&dds_total, &ddh_total);

	dump_dedup_ratio(&dds_total);
	}

	static int
	verify_device_removal_feature_counts(spa_t *spa)
	{
	uint64_t dr_feature_refcount = 0;
	uint64_t oc_feature_refcount = 0;
	uint64_t indirect_vdev_count = 0;
	uint64_t precise_vdev_count = 0;
	uint64_t obsolete_counts_object_count = 0;
	uint64_t obsolete_sm_count = 0;
	uint64_t obsolete_counts_count = 0;
	uint64_t scip_count = 0;
	uint64_t obsolete_bpobj_count = 0;
	int ret = 0;

	spa_condensing_indirect_phys_t *scip =
	&spa->spa_condensing_indirect_phys;
	if (scip->scip_next_mapping_object != 0) {
	vdev_t *vd = spa->spa_root_vdev->vdev_child[scip->scip_vdev];
	ASSERT(scip->scip_prev_obsolete_sm_object != 0);
	ASSERT3P(vd->vdev_ops, ==, &vdev_indirect_ops);

	(void) printf("Condensing indirect vdev %llu: new mapping "
	"object %llu, prev obsolete sm %llu\n",
	(u_longlong_t)scip->scip_vdev,
	(u_longlong_t)scip->scip_next_mapping_object,
	(u_longlong_t)scip->scip_prev_obsolete_sm_object);
	if (scip->scip_prev_obsolete_sm_object != 0) {
	space_map_t *prev_obsolete_sm = NULL;
	VERIFY0(space_map_open(&prev_obsolete_sm,
	spa->spa_meta_objset,
	scip->scip_prev_obsolete_sm_object,
	0, vd->vdev_asize, 0));
	dump_spacemap(spa->spa_meta_objset, prev_obsolete_sm);
	(void) printf("\n");
	space_map_close(prev_obsolete_sm);
	}

	scip_count += 2;
	}

	for (uint64_t i = 0; i < spa->spa_root_vdev->vdev_children; i++) {
	vdev_t *vd = spa->spa_root_vdev->vdev_child[i];
	vdev_indirect_config_t *vic = &vd->vdev_indirect_config;

	if (vic->vic_mapping_object != 0) {
	ASSERT(vd->vdev_ops == &vdev_indirect_ops \|\|
	vd->vdev_removing);
	indirect_vdev_count++;

	if (vd->vdev_indirect_mapping->vim_havecounts) {
	obsolete_counts_count++;
	}
	}

	boolean_t are_precise;
	VERIFY0(vdev_obsolete_counts_are_precise(vd, &are_precise));
	if (are_precise) {
	ASSERT(vic->vic_mapping_object != 0);
	precise_vdev_count++;
	}

	uint64_t obsolete_sm_object;
	VERIFY0(vdev_obsolete_sm_object(vd, &obsolete_sm_object));
	if (obsolete_sm_object != 0) {
	ASSERT(vic->vic_mapping_object != 0);
	obsolete_sm_count++;
	}
	}

	(void) feature_get_refcount(spa,
	&spa_feature_table[SPA_FEATURE_DEVICE_REMOVAL],
	&dr_feature_refcount);
	(void) feature_get_refcount(spa,
	&spa_feature_table[SPA_FEATURE_OBSOLETE_COUNTS],
	&oc_feature_refcount);

	if (dr_feature_refcount != indirect_vdev_count) {
	ret = 1;
	(void) printf("Number of indirect vdevs (%llu) " \
	"does not match feature count (%llu)\n",
	(u_longlong_t)indirect_vdev_count,
	(u_longlong_t)dr_feature_refcount);
	} else {
	(void) printf("Verified device_removal feature refcount " \
	"of %llu is correct\n",
	(u_longlong_t)dr_feature_refcount);
	}

	if (zap_contains(spa_meta_objset(spa), DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_OBSOLETE_BPOBJ) == 0) {
	obsolete_bpobj_count++;
	}


	obsolete_counts_object_count = precise_vdev_count;
	obsolete_counts_object_count += obsolete_sm_count;
	obsolete_counts_object_count += obsolete_counts_count;
	obsolete_counts_object_count += scip_count;
	obsolete_counts_object_count += obsolete_bpobj_count;
	obsolete_counts_object_count += remap_deadlist_count;

	if (oc_feature_refcount != obsolete_counts_object_count) {
	ret = 1;
	(void) printf("Number of obsolete counts objects (%llu) " \
	"does not match feature count (%llu)\n",
	(u_longlong_t)obsolete_counts_object_count,
	(u_longlong_t)oc_feature_refcount);
	(void) printf("pv:%llu os:%llu oc:%llu sc:%llu "
	"ob:%llu rd:%llu\n",
	(u_longlong_t)precise_vdev_count,
	(u_longlong_t)obsolete_sm_count,
	(u_longlong_t)obsolete_counts_count,
	(u_longlong_t)scip_count,
	(u_longlong_t)obsolete_bpobj_count,
	(u_longlong_t)remap_deadlist_count);
	} else {
	(void) printf("Verified indirect_refcount feature refcount " \
	"of %llu is correct\n",
	(u_longlong_t)oc_feature_refcount);
	}
	return (ret);
	}

	static void
	zdb_set_skip_mmp(char *target)
	{
	spa_t *spa;

	/*
	* Disable the activity check to allow examination of
	* active pools.
	*/
	mutex_enter(&spa_namespace_lock);
	if ((spa = spa_lookup(target)) != NULL) {
	spa->spa_import_flags \|= ZFS_IMPORT_SKIP_MMP;
	}
	mutex_exit(&spa_namespace_lock);
	}

	#define BOGUS_SUFFIX "_CHECKPOINTED_UNIVERSE"
	/*
	* Import the checkpointed state of the pool specified by the target
	* parameter as readonly. The function also accepts a pool config
	* as an optional parameter, else it attempts to infer the config by
	* the name of the target pool.
	*
	* Note that the checkpointed state's pool name will be the name of
	* the original pool with the above suffix appended to it. In addition,
	* if the target is not a pool name (e.g. a path to a dataset) then
	* the new_path parameter is populated with the updated path to
	* reflect the fact that we are looking into the checkpointed state.
	*
	* The function returns a newly-allocated copy of the name of the
	* pool containing the checkpointed state. When this copy is no
	* longer needed it should be freed with free(3C). Same thing
	* applies to the new_path parameter if allocated.
	*/
	static char *
	import_checkpointed_state(char target, nvlist_t cfg, char **new_path)
	{
	int error = 0;
	char poolname, bogus_name = NULL;
	boolean_t freecfg = B_FALSE;

	/* If the target is not a pool, the extract the pool name */
	char *path_start = strchr(target, '/');
	if (path_start != NULL) {
	size_t poolname_len = path_start - target;
	poolname = strndup(target, poolname_len);
	} else {
	poolname = target;
	}

	if (cfg == NULL) {
	zdb_set_skip_mmp(poolname);
	error = spa_get_stats(poolname, &cfg, NULL, 0);
	if (error != 0) {
	fatal("Tried to read config of pool \"%s\" but "
	"spa_get_stats() failed with error %d\n",
	poolname, error);
	}
	freecfg = B_TRUE;
	}

	if (asprintf(&bogus_name, "%s%s", poolname, BOGUS_SUFFIX) == -1)
	return (NULL);
	fnvlist_add_string(cfg, ZPOOL_CONFIG_POOL_NAME, bogus_name);

	error = spa_import(bogus_name, cfg, NULL,
	ZFS_IMPORT_MISSING_LOG \| ZFS_IMPORT_CHECKPOINT \|
	ZFS_IMPORT_SKIP_MMP);
	if (freecfg)
	nvlist_free(cfg);
	if (error != 0) {
	fatal("Tried to import pool \"%s\" but spa_import() failed "
	"with error %d\n", bogus_name, error);
	}

	if (new_path != NULL && path_start != NULL) {
	if (asprintf(new_path, "%s%s", bogus_name, path_start) == -1) {
	if (path_start != NULL)
	free(poolname);
	return (NULL);
	}
	}

	if (target != poolname)
	free(poolname);

	return (bogus_name);
	}

	typedef struct verify_checkpoint_sm_entry_cb_arg {
	vdev_t *vcsec_vd;

	/* the following fields are only used for printing progress */
	uint64_t vcsec_entryid;
	uint64_t vcsec_num_entries;
	} verify_checkpoint_sm_entry_cb_arg_t;

	#define ENTRIES_PER_PROGRESS_UPDATE 10000

	static int
	verify_checkpoint_sm_entry_cb(space_map_entry_t sme, void arg)
	{
	verify_checkpoint_sm_entry_cb_arg_t *vcsec = arg;
	vdev_t *vd = vcsec->vcsec_vd;
	metaslab_t *ms = vd->vdev_ms[sme->sme_offset >> vd->vdev_ms_shift];
	uint64_t end = sme->sme_offset + sme->sme_run;

	ASSERT(sme->sme_type == SM_FREE);

	if ((vcsec->vcsec_entryid % ENTRIES_PER_PROGRESS_UPDATE) == 0) {
	(void) fprintf(stderr,
	"\rverifying vdev %llu, space map entry %llu of %llu ...",
	(longlong_t)vd->vdev_id,
	(longlong_t)vcsec->vcsec_entryid,
	(longlong_t)vcsec->vcsec_num_entries);
	}
	vcsec->vcsec_entryid++;

	/*
	* See comment in checkpoint_sm_exclude_entry_cb()
	*/
	VERIFY3U(sme->sme_offset, >=, ms->ms_start);
	VERIFY3U(end, <=, ms->ms_start + ms->ms_size);

	/*
	* The entries in the vdev_checkpoint_sm should be marked as
	* allocated in the checkpointed state of the pool, therefore
	* their respective ms_allocateable trees should not contain them.
	*/
	mutex_enter(&ms->ms_lock);
	range_tree_verify_not_present(ms->ms_allocatable,
	sme->sme_offset, sme->sme_run);
	mutex_exit(&ms->ms_lock);

	return (0);
	}

	/*
	* Verify that all segments in the vdev_checkpoint_sm are allocated
	* according to the checkpoint's ms_sm (i.e. are not in the checkpoint's
	* ms_allocatable).
	*
	* Do so by comparing the checkpoint space maps (vdev_checkpoint_sm) of
	* each vdev in the current state of the pool to the metaslab space maps
	* (ms_sm) of the checkpointed state of the pool.
	*
	* Note that the function changes the state of the ms_allocatable
	* trees of the current spa_t. The entries of these ms_allocatable
	* trees are cleared out and then repopulated from with the free
	* entries of their respective ms_sm space maps.
	*/
	static void
	verify_checkpoint_vdev_spacemaps(spa_t checkpoint, spa_t current)
	{
	vdev_t *ckpoint_rvd = checkpoint->spa_root_vdev;
	vdev_t *current_rvd = current->spa_root_vdev;

	load_concrete_ms_allocatable_trees(checkpoint, SM_FREE);

	for (uint64_t c = 0; c < ckpoint_rvd->vdev_children; c++) {
	vdev_t *ckpoint_vd = ckpoint_rvd->vdev_child[c];
	vdev_t *current_vd = current_rvd->vdev_child[c];

	space_map_t *checkpoint_sm = NULL;
	uint64_t checkpoint_sm_obj;

	if (ckpoint_vd->vdev_ops == &vdev_indirect_ops) {
	/*
	* Since we don't allow device removal in a pool
	* that has a checkpoint, we expect that all removed
	* vdevs were removed from the pool before the
	* checkpoint.
	*/
	ASSERT3P(current_vd->vdev_ops, ==, &vdev_indirect_ops);
	continue;
	}

	/*
	* If the checkpoint space map doesn't exist, then nothing
	* here is checkpointed so there's nothing to verify.
	*/
	if (current_vd->vdev_top_zap == 0 \|\|
	zap_contains(spa_meta_objset(current),
	current_vd->vdev_top_zap,
	VDEV_TOP_ZAP_POOL_CHECKPOINT_SM) != 0)
	continue;

	VERIFY0(zap_lookup(spa_meta_objset(current),
	current_vd->vdev_top_zap, VDEV_TOP_ZAP_POOL_CHECKPOINT_SM,
	sizeof (uint64_t), 1, &checkpoint_sm_obj));

	VERIFY0(space_map_open(&checkpoint_sm, spa_meta_objset(current),
	checkpoint_sm_obj, 0, current_vd->vdev_asize,
	current_vd->vdev_ashift));

	verify_checkpoint_sm_entry_cb_arg_t vcsec;
	vcsec.vcsec_vd = ckpoint_vd;
	vcsec.vcsec_entryid = 0;
	vcsec.vcsec_num_entries =
	space_map_length(checkpoint_sm) / sizeof (uint64_t);
	VERIFY0(space_map_iterate(checkpoint_sm,
	space_map_length(checkpoint_sm),
	verify_checkpoint_sm_entry_cb, &vcsec));
	if (dump_opt['m'] > 3)
	dump_spacemap(current->spa_meta_objset, checkpoint_sm);
	space_map_close(checkpoint_sm);
	}

	/*
	* If we've added vdevs since we took the checkpoint, ensure
	* that their checkpoint space maps are empty.
	*/
	if (ckpoint_rvd->vdev_children < current_rvd->vdev_children) {
	for (uint64_t c = ckpoint_rvd->vdev_children;
	c < current_rvd->vdev_children; c++) {
	vdev_t *current_vd = current_rvd->vdev_child[c];
	VERIFY3P(current_vd->vdev_checkpoint_sm, ==, NULL);
	}
	}

	/* for cleaner progress output */
	(void) fprintf(stderr, "\n");
	}

	/*
	* Verifies that all space that's allocated in the checkpoint is
	* still allocated in the current version, by checking that everything
	* in checkpoint's ms_allocatable (which is actually allocated, not
	* allocatable/free) is not present in current's ms_allocatable.
	*
	* Note that the function changes the state of the ms_allocatable
	* trees of both spas when called. The entries of all ms_allocatable
	* trees are cleared out and then repopulated from their respective
	* ms_sm space maps. In the checkpointed state we load the allocated
	* entries, and in the current state we load the free entries.
	*/
	static void
	verify_checkpoint_ms_spacemaps(spa_t checkpoint, spa_t current)
	{
	vdev_t *ckpoint_rvd = checkpoint->spa_root_vdev;
	vdev_t *current_rvd = current->spa_root_vdev;

	load_concrete_ms_allocatable_trees(checkpoint, SM_ALLOC);
	load_concrete_ms_allocatable_trees(current, SM_FREE);

	for (uint64_t i = 0; i < ckpoint_rvd->vdev_children; i++) {
	vdev_t *ckpoint_vd = ckpoint_rvd->vdev_child[i];
	vdev_t *current_vd = current_rvd->vdev_child[i];

	if (ckpoint_vd->vdev_ops == &vdev_indirect_ops) {
	/*
	* See comment in verify_checkpoint_vdev_spacemaps()
	*/
	ASSERT3P(current_vd->vdev_ops, ==, &vdev_indirect_ops);
	continue;
	}

	for (uint64_t m = 0; m < ckpoint_vd->vdev_ms_count; m++) {
	metaslab_t *ckpoint_msp = ckpoint_vd->vdev_ms[m];
	metaslab_t *current_msp = current_vd->vdev_ms[m];

	(void) fprintf(stderr,
	"\rverifying vdev %llu of %llu, "
	"metaslab %llu of %llu ...",
	(longlong_t)current_vd->vdev_id,
	(longlong_t)current_rvd->vdev_children,
	(longlong_t)current_vd->vdev_ms[m]->ms_id,
	(longlong_t)current_vd->vdev_ms_count);

	/*
	* We walk through the ms_allocatable trees that
	* are loaded with the allocated blocks from the
	* ms_sm spacemaps of the checkpoint. For each
	* one of these ranges we ensure that none of them
	* exists in the ms_allocatable trees of the
	* current state which are loaded with the ranges
	* that are currently free.
	*
	* This way we ensure that none of the blocks that
	* are part of the checkpoint were freed by mistake.
	*/
	range_tree_walk(ckpoint_msp->ms_allocatable,
	(range_tree_func_t *)range_tree_verify_not_present,
	current_msp->ms_allocatable);
	}
	}

	/* for cleaner progress output */
	(void) fprintf(stderr, "\n");
	}

	static void
	verify_checkpoint_blocks(spa_t *spa)
	{
	ASSERT(!dump_opt['L']);

	spa_t *checkpoint_spa;
	char *checkpoint_pool;
	int error = 0;

	/*
	* We import the checkpointed state of the pool (under a different
	* name) so we can do verification on it against the current state
	* of the pool.
	*/
	checkpoint_pool = import_checkpointed_state(spa->spa_name, NULL,
	NULL);
	ASSERT(strcmp(spa->spa_name, checkpoint_pool) != 0);

	error = spa_open(checkpoint_pool, &checkpoint_spa, FTAG);
	if (error != 0) {
	fatal("Tried to open pool \"%s\" but spa_open() failed with "
	"error %d\n", checkpoint_pool, error);
	}

	/*
	* Ensure that ranges in the checkpoint space maps of each vdev
	* are allocated according to the checkpointed state's metaslab
	* space maps.
	*/
	verify_checkpoint_vdev_spacemaps(checkpoint_spa, spa);

	/*
	* Ensure that allocated ranges in the checkpoint's metaslab
	* space maps remain allocated in the metaslab space maps of
	* the current state.
	*/
	verify_checkpoint_ms_spacemaps(checkpoint_spa, spa);

	/*
	* Once we are done, we get rid of the checkpointed state.
	*/
	spa_close(checkpoint_spa, FTAG);
	free(checkpoint_pool);
	}

	static void
	dump_leftover_checkpoint_blocks(spa_t *spa)
	{
	vdev_t *rvd = spa->spa_root_vdev;

	for (uint64_t i = 0; i < rvd->vdev_children; i++) {
	vdev_t *vd = rvd->vdev_child[i];

	space_map_t *checkpoint_sm = NULL;
	uint64_t checkpoint_sm_obj;

	if (vd->vdev_top_zap == 0)
	continue;

	if (zap_contains(spa_meta_objset(spa), vd->vdev_top_zap,
	VDEV_TOP_ZAP_POOL_CHECKPOINT_SM) != 0)
	continue;

	VERIFY0(zap_lookup(spa_meta_objset(spa), vd->vdev_top_zap,
	VDEV_TOP_ZAP_POOL_CHECKPOINT_SM,
	sizeof (uint64_t), 1, &checkpoint_sm_obj));

	VERIFY0(space_map_open(&checkpoint_sm, spa_meta_objset(spa),
	checkpoint_sm_obj, 0, vd->vdev_asize, vd->vdev_ashift));
	dump_spacemap(spa->spa_meta_objset, checkpoint_sm);
	space_map_close(checkpoint_sm);
	}
	}

	static int
	verify_checkpoint(spa_t *spa)
	{
	uberblock_t checkpoint;
	int error;

	if (!spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT))
	return (0);

	error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_ZPOOL_CHECKPOINT, sizeof (uint64_t),
	sizeof (uberblock_t) / sizeof (uint64_t), &checkpoint);

	if (error == ENOENT && !dump_opt['L']) {
	/*
	* If the feature is active but the uberblock is missing
	* then we must be in the middle of discarding the
	* checkpoint.
	*/
	(void) printf("\nPartially discarded checkpoint "
	"state found:\n");
	if (dump_opt['m'] > 3)
	dump_leftover_checkpoint_blocks(spa);
	return (0);
	} else if (error != 0) {
	(void) printf("lookup error %d when looking for "
	"checkpointed uberblock in MOS\n", error);
	return (error);
	}
	dump_uberblock(&checkpoint, "\nCheckpointed uberblock found:\n", "\n");

	if (checkpoint.ub_checkpoint_txg == 0) {
	(void) printf("\nub_checkpoint_txg not set in checkpointed "
	"uberblock\n");
	error = 3;
	}

	if (error == 0 && !dump_opt['L'])
	verify_checkpoint_blocks(spa);

	return (error);
	}

	/* ARGSUSED */
	static void
	mos_leaks_cb(void *arg, uint64_t start, uint64_t size)
	{
	for (uint64_t i = start; i < size; i++) {
	(void) printf("MOS object %llu referenced but not allocated\n",
	(u_longlong_t)i);
	}
	}

	static void
	mos_obj_refd(uint64_t obj)
	{
	if (obj != 0 && mos_refd_objs != NULL)
	range_tree_add(mos_refd_objs, obj, 1);
	}

	/*
	* Call on a MOS object that may already have been referenced.
	*/
	static void
	mos_obj_refd_multiple(uint64_t obj)
	{
	if (obj != 0 && mos_refd_objs != NULL &&
	!range_tree_contains(mos_refd_objs, obj, 1))
	range_tree_add(mos_refd_objs, obj, 1);
	}

	static void
	mos_leak_vdev_top_zap(vdev_t *vd)
	{
	uint64_t ms_flush_data_obj;
	int error = zap_lookup(spa_meta_objset(vd->vdev_spa),
	vd->vdev_top_zap, VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS,
	sizeof (ms_flush_data_obj), 1, &ms_flush_data_obj);
	if (error == ENOENT)
	return;
	ASSERT0(error);

	mos_obj_refd(ms_flush_data_obj);
	}

	static void
	mos_leak_vdev(vdev_t *vd)
	{
	mos_obj_refd(vd->vdev_dtl_object);
	mos_obj_refd(vd->vdev_ms_array);
	mos_obj_refd(vd->vdev_indirect_config.vic_births_object);
	mos_obj_refd(vd->vdev_indirect_config.vic_mapping_object);
	mos_obj_refd(vd->vdev_leaf_zap);
	if (vd->vdev_checkpoint_sm != NULL)
	mos_obj_refd(vd->vdev_checkpoint_sm->sm_object);
	if (vd->vdev_indirect_mapping != NULL) {
	mos_obj_refd(vd->vdev_indirect_mapping->
	vim_phys->vimp_counts_object);
	}
	if (vd->vdev_obsolete_sm != NULL)
	mos_obj_refd(vd->vdev_obsolete_sm->sm_object);

	for (uint64_t m = 0; m < vd->vdev_ms_count; m++) {
	metaslab_t *ms = vd->vdev_ms[m];
	mos_obj_refd(space_map_object(ms->ms_sm));
	}

	if (vd->vdev_top_zap != 0) {
	mos_obj_refd(vd->vdev_top_zap);
	mos_leak_vdev_top_zap(vd);
	}

	for (uint64_t c = 0; c < vd->vdev_children; c++) {
	mos_leak_vdev(vd->vdev_child[c]);
	}
	}

	static void
	mos_leak_log_spacemaps(spa_t *spa)
	{
	uint64_t spacemap_zap;
	int error = zap_lookup(spa_meta_objset(spa),
	DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_LOG_SPACEMAP_ZAP,
	sizeof (spacemap_zap), 1, &spacemap_zap);
	if (error == ENOENT)
	return;
	ASSERT0(error);

	mos_obj_refd(spacemap_zap);
	for (spa_log_sm_t *sls = avl_first(&spa->spa_sm_logs_by_txg);
	sls; sls = AVL_NEXT(&spa->spa_sm_logs_by_txg, sls))
	mos_obj_refd(sls->sls_sm_obj);
	}

	static int
	dump_mos_leaks(spa_t *spa)
	{
	int rv = 0;
	objset_t *mos = spa->spa_meta_objset;
	dsl_pool_t *dp = spa->spa_dsl_pool;

	/* Visit and mark all referenced objects in the MOS */

	mos_obj_refd(DMU_POOL_DIRECTORY_OBJECT);
	mos_obj_refd(spa->spa_pool_props_object);
	mos_obj_refd(spa->spa_config_object);
	mos_obj_refd(spa->spa_ddt_stat_object);
	mos_obj_refd(spa->spa_feat_desc_obj);
	mos_obj_refd(spa->spa_feat_enabled_txg_obj);
	mos_obj_refd(spa->spa_feat_for_read_obj);
	mos_obj_refd(spa->spa_feat_for_write_obj);
	mos_obj_refd(spa->spa_history);
	mos_obj_refd(spa->spa_errlog_last);
	mos_obj_refd(spa->spa_errlog_scrub);
	mos_obj_refd(spa->spa_all_vdev_zaps);
	mos_obj_refd(spa->spa_dsl_pool->dp_bptree_obj);
	mos_obj_refd(spa->spa_dsl_pool->dp_tmp_userrefs_obj);
	mos_obj_refd(spa->spa_dsl_pool->dp_scan->scn_phys.scn_queue_obj);
	bpobj_count_refd(&spa->spa_deferred_bpobj);
	mos_obj_refd(dp->dp_empty_bpobj);
	bpobj_count_refd(&dp->dp_obsolete_bpobj);
	bpobj_count_refd(&dp->dp_free_bpobj);
	mos_obj_refd(spa->spa_l2cache.sav_object);
	mos_obj_refd(spa->spa_spares.sav_object);

	if (spa->spa_syncing_log_sm != NULL)
	mos_obj_refd(spa->spa_syncing_log_sm->sm_object);
	mos_leak_log_spacemaps(spa);

	mos_obj_refd(spa->spa_condensing_indirect_phys.
	scip_next_mapping_object);
	mos_obj_refd(spa->spa_condensing_indirect_phys.
	scip_prev_obsolete_sm_object);
	if (spa->spa_condensing_indirect_phys.scip_next_mapping_object != 0) {
	vdev_indirect_mapping_t *vim =
	vdev_indirect_mapping_open(mos,
	spa->spa_condensing_indirect_phys.scip_next_mapping_object);
	mos_obj_refd(vim->vim_phys->vimp_counts_object);
	vdev_indirect_mapping_close(vim);
	}
	deleted_livelists_dump_mos(spa);

	if (dp->dp_origin_snap != NULL) {
	dsl_dataset_t *ds;

	dsl_pool_config_enter(dp, FTAG);
	VERIFY0(dsl_dataset_hold_obj(dp,
	dsl_dataset_phys(dp->dp_origin_snap)->ds_next_snap_obj,
	FTAG, &ds));
	count_ds_mos_objects(ds);
	dump_blkptr_list(&ds->ds_deadlist, "Deadlist");
	dsl_dataset_rele(ds, FTAG);
	dsl_pool_config_exit(dp, FTAG);

	count_ds_mos_objects(dp->dp_origin_snap);
	dump_blkptr_list(&dp->dp_origin_snap->ds_deadlist, "Deadlist");
	}
	count_dir_mos_objects(dp->dp_mos_dir);
	if (dp->dp_free_dir != NULL)
	count_dir_mos_objects(dp->dp_free_dir);
	if (dp->dp_leak_dir != NULL)
	count_dir_mos_objects(dp->dp_leak_dir);

	mos_leak_vdev(spa->spa_root_vdev);

	for (uint64_t class = 0; class < DDT_CLASSES; class++) {
	for (uint64_t type = 0; type < DDT_TYPES; type++) {
	for (uint64_t cksum = 0;
	cksum < ZIO_CHECKSUM_FUNCTIONS; cksum++) {
	ddt_t *ddt = spa->spa_ddt[cksum];
	mos_obj_refd(ddt->ddt_object[type][class]);
	}
	}
	}

	/*
	* Visit all allocated objects and make sure they are referenced.
	*/
	uint64_t object = 0;
	while (dmu_object_next(mos, &object, B_FALSE, 0) == 0) {
	if (range_tree_contains(mos_refd_objs, object, 1)) {
	range_tree_remove(mos_refd_objs, object, 1);
	} else {
	dmu_object_info_t doi;
	const char *name;
	dmu_object_info(mos, object, &doi);
	if (doi.doi_type & DMU_OT_NEWTYPE) {
	dmu_object_byteswap_t bswap =
	DMU_OT_BYTESWAP(doi.doi_type);
	name = dmu_ot_byteswap[bswap].ob_name;
	} else {
	name = dmu_ot[doi.doi_type].ot_name;
	}

	(void) printf("MOS object %llu (%s) leaked\n",
	(u_longlong_t)object, name);
	rv = 2;
	}
	}
	(void) range_tree_walk(mos_refd_objs, mos_leaks_cb, NULL);
	if (!range_tree_is_empty(mos_refd_objs))
	rv = 2;
	range_tree_vacate(mos_refd_objs, NULL, NULL);
	range_tree_destroy(mos_refd_objs);
	return (rv);
	}

	typedef struct log_sm_obsolete_stats_arg {
	uint64_t lsos_current_txg;

	uint64_t lsos_total_entries;
	uint64_t lsos_valid_entries;

	uint64_t lsos_sm_entries;
	uint64_t lsos_valid_sm_entries;
	} log_sm_obsolete_stats_arg_t;

	static int
	log_spacemap_obsolete_stats_cb(spa_t spa, space_map_entry_t sme,
	uint64_t txg, void *arg)
	{
	log_sm_obsolete_stats_arg_t *lsos = arg;

	uint64_t offset = sme->sme_offset;
	uint64_t vdev_id = sme->sme_vdev;

	if (lsos->lsos_current_txg == 0) {
	/* this is the first log */
	lsos->lsos_current_txg = txg;
	} else if (lsos->lsos_current_txg < txg) {
	/* we just changed log - print stats and reset */
	(void) printf("%-8llu valid entries out of %-8llu - txg %llu\n",
	(u_longlong_t)lsos->lsos_valid_sm_entries,
	(u_longlong_t)lsos->lsos_sm_entries,
	(u_longlong_t)lsos->lsos_current_txg);
	lsos->lsos_valid_sm_entries = 0;
	lsos->lsos_sm_entries = 0;
	lsos->lsos_current_txg = txg;
	}
	ASSERT3U(lsos->lsos_current_txg, ==, txg);

	lsos->lsos_sm_entries++;
	lsos->lsos_total_entries++;

	vdev_t *vd = vdev_lookup_top(spa, vdev_id);
	if (!vdev_is_concrete(vd))
	return (0);

	metaslab_t *ms = vd->vdev_ms[offset >> vd->vdev_ms_shift];
	ASSERT(sme->sme_type == SM_ALLOC \|\| sme->sme_type == SM_FREE);

	if (txg < metaslab_unflushed_txg(ms))
	return (0);
	lsos->lsos_valid_sm_entries++;
	lsos->lsos_valid_entries++;
	return (0);
	}

	static void
	dump_log_spacemap_obsolete_stats(spa_t *spa)
	{
	if (!spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP))
	return;

	log_sm_obsolete_stats_arg_t lsos;
	bzero(&lsos, sizeof (lsos));

	(void) printf("Log Space Map Obsolete Entry Statistics:\n");

	iterate_through_spacemap_logs(spa,
	log_spacemap_obsolete_stats_cb, &lsos);

	/* print stats for latest log */
	(void) printf("%-8llu valid entries out of %-8llu - txg %llu\n",
	(u_longlong_t)lsos.lsos_valid_sm_entries,
	(u_longlong_t)lsos.lsos_sm_entries,
	(u_longlong_t)lsos.lsos_current_txg);

	(void) printf("%-8llu valid entries out of %-8llu - total\n\n",
	(u_longlong_t)lsos.lsos_valid_entries,
	(u_longlong_t)lsos.lsos_total_entries);
	}

	static void
	dump_zpool(spa_t *spa)
	{
	dsl_pool_t *dp = spa_get_dsl(spa);
	int rc = 0;

	if (dump_opt['y']) {
	livelist_metaslab_validate(spa);
	}

	if (dump_opt['S']) {
	dump_simulated_ddt(spa);
	return;
	}

	if (!dump_opt['e'] && dump_opt['C'] > 1) {
	(void) printf("\nCached configuration:\n");
	dump_nvlist(spa->spa_config, 8);
	}

	if (dump_opt['C'])
	dump_config(spa);

	if (dump_opt['u'])
	dump_uberblock(&spa->spa_uberblock, "\nUberblock:\n", "\n");

	if (dump_opt['D'])
	dump_all_ddts(spa);

	if (dump_opt['d'] > 2 \|\| dump_opt['m'])
	dump_metaslabs(spa);
	if (dump_opt['M'])
	- dump_metaslab_groups(spa);
	+ dump_metaslab_groups(spa, dump_opt['M'] > 1);
	if (dump_opt['d'] > 2 \|\| dump_opt['m']) {
	dump_log_spacemaps(spa);
	dump_log_spacemap_obsolete_stats(spa);
	}

	if (dump_opt['d'] \|\| dump_opt['i']) {
	spa_feature_t f;
	mos_refd_objs = range_tree_create(NULL, RANGE_SEG64, NULL, 0,
	0);
	dump_objset(dp->dp_meta_objset);

	if (dump_opt['d'] >= 3) {
	dsl_pool_t *dp = spa->spa_dsl_pool;
	dump_full_bpobj(&spa->spa_deferred_bpobj,
	"Deferred frees", 0);
	if (spa_version(spa) >= SPA_VERSION_DEADLISTS) {
	dump_full_bpobj(&dp->dp_free_bpobj,
	"Pool snapshot frees", 0);
	}
	if (bpobj_is_open(&dp->dp_obsolete_bpobj)) {
	ASSERT(spa_feature_is_enabled(spa,
	SPA_FEATURE_DEVICE_REMOVAL));
	dump_full_bpobj(&dp->dp_obsolete_bpobj,
	"Pool obsolete blocks", 0);
	}

	if (spa_feature_is_active(spa,
	SPA_FEATURE_ASYNC_DESTROY)) {
	dump_bptree(spa->spa_meta_objset,
	dp->dp_bptree_obj,
	"Pool dataset frees");
	}
	dump_dtl(spa->spa_root_vdev, 0);
	}

	for (spa_feature_t f = 0; f < SPA_FEATURES; f++)
	global_feature_count[f] = UINT64_MAX;
	global_feature_count[SPA_FEATURE_REDACTION_BOOKMARKS] = 0;
	global_feature_count[SPA_FEATURE_BOOKMARK_WRITTEN] = 0;
	global_feature_count[SPA_FEATURE_LIVELIST] = 0;

	(void) dmu_objset_find(spa_name(spa), dump_one_objset,
	NULL, DS_FIND_SNAPSHOTS \| DS_FIND_CHILDREN);

	if (rc == 0 && !dump_opt['L'])
	rc = dump_mos_leaks(spa);

	for (f = 0; f < SPA_FEATURES; f++) {
	uint64_t refcount;

	uint64_t *arr;
	if (!(spa_feature_table[f].fi_flags &
	ZFEATURE_FLAG_PER_DATASET)) {
	if (global_feature_count[f] == UINT64_MAX)
	continue;
	if (!spa_feature_is_enabled(spa, f)) {
	ASSERT0(global_feature_count[f]);
	continue;
	}
	arr = global_feature_count;
	} else {
	if (!spa_feature_is_enabled(spa, f)) {
	ASSERT0(dataset_feature_count[f]);
	continue;
	}
	arr = dataset_feature_count;
	}
	if (feature_get_refcount(spa, &spa_feature_table[f],
	&refcount) == ENOTSUP)
	continue;
	if (arr[f] != refcount) {
	(void) printf("%s feature refcount mismatch: "
	"%lld consumers != %lld refcount\n",
	spa_feature_table[f].fi_uname,
	(longlong_t)arr[f], (longlong_t)refcount);
	rc = 2;
	} else {
	(void) printf("Verified %s feature refcount "
	"of %llu is correct\n",
	spa_feature_table[f].fi_uname,
	(longlong_t)refcount);
	}
	}

	if (rc == 0)
	rc = verify_device_removal_feature_counts(spa);
	}

	if (rc == 0 && (dump_opt['b'] \|\| dump_opt['c']))
	rc = dump_block_stats(spa);

	if (rc == 0)
	rc = verify_spacemap_refcounts(spa);

	if (dump_opt['s'])
	show_pool_stats(spa);

	if (dump_opt['h'])
	dump_history(spa);

	if (rc == 0)
	rc = verify_checkpoint(spa);

	if (rc != 0) {
	dump_debug_buffer();
	exit(rc);
	}
	}

	#define ZDB_FLAG_CHECKSUM 0x0001
	#define ZDB_FLAG_DECOMPRESS 0x0002
	#define ZDB_FLAG_BSWAP 0x0004
	#define ZDB_FLAG_GBH 0x0008
	#define ZDB_FLAG_INDIRECT 0x0010
	#define ZDB_FLAG_RAW 0x0020
	#define ZDB_FLAG_PRINT_BLKPTR 0x0040
	#define ZDB_FLAG_VERBOSE 0x0080

	static int flagbits[256];
	static char flagbitstr[16];

	static void
	zdb_print_blkptr(const blkptr_t *bp, int flags)
	{
	char blkbuf[BP_SPRINTF_LEN];

	if (flags & ZDB_FLAG_BSWAP)
	byteswap_uint64_array((void *)bp, sizeof (blkptr_t));

	snprintf_blkptr(blkbuf, sizeof (blkbuf), bp);
	(void) printf("%s\n", blkbuf);
	}

	static void
	zdb_dump_indirect(blkptr_t *bp, int nbps, int flags)
	{
	int i;

	for (i = 0; i < nbps; i++)
	zdb_print_blkptr(&bp[i], flags);
	}

	static void
	zdb_dump_gbh(void *buf, int flags)
	{
	zdb_dump_indirect((blkptr_t *)buf, SPA_GBH_NBLKPTRS, flags);
	}

	static void
	zdb_dump_block_raw(void *buf, uint64_t size, int flags)
	{
	if (flags & ZDB_FLAG_BSWAP)
	byteswap_uint64_array(buf, size);
	VERIFY(write(fileno(stdout), buf, size) == size);
	}

	static void
	zdb_dump_block(char label, void buf, uint64_t size, int flags)
	{
	uint64_t d = (uint64_t )buf;
	unsigned nwords = size / sizeof (uint64_t);
	int do_bswap = !!(flags & ZDB_FLAG_BSWAP);
	unsigned i, j;
	const char *hdr;
	char *c;


	if (do_bswap)
	hdr = " 7 6 5 4 3 2 1 0 f e d c b a 9 8";
	else
	hdr = " 0 1 2 3 4 5 6 7 8 9 a b c d e f";

	(void) printf("\n%s\n%6s %s 0123456789abcdef\n", label, "", hdr);

	#ifdef _LITTLE_ENDIAN
	/* correct the endianness */
	do_bswap = !do_bswap;
	#endif
	for (i = 0; i < nwords; i += 2) {
	(void) printf("%06llx: %016llx %016llx ",
	(u_longlong_t)(i * sizeof (uint64_t)),
	(u_longlong_t)(do_bswap ? BSWAP_64(d[i]) : d[i]),
	(u_longlong_t)(do_bswap ? BSWAP_64(d[i + 1]) : d[i + 1]));

	c = (char *)&d[i];
	for (j = 0; j < 2 * sizeof (uint64_t); j++)
	(void) printf("%c", isprint(c[j]) ? c[j] : '.');
	(void) printf("\n");
	}
	}

	/*
	* There are two acceptable formats:
	* leaf_name - For example: c1t0d0 or /tmp/ztest.0a
	* child[.child]* - For example: 0.1.1
	*
	* The second form can be used to specify arbitrary vdevs anywhere
	* in the hierarchy. For example, in a pool with a mirror of
	* RAID-Zs, you can specify either RAID-Z vdev with 0.0 or 0.1 .
	*/
	static vdev_t *
	zdb_vdev_lookup(vdev_t vdev, const char path)
	{
	char s, p, *q;
	unsigned i;

	if (vdev == NULL)
	return (NULL);

	/* First, assume the x.x.x.x format */
	i = strtoul(path, &s, 10);
	if (s == path \|\| (s && s != '.' && s != '\0'))
	goto name;
	if (i >= vdev->vdev_children)
	return (NULL);

	vdev = vdev->vdev_child[i];
	if (s && *s == '\0')
	return (vdev);
	return (zdb_vdev_lookup(vdev, s+1));

	name:
	for (i = 0; i < vdev->vdev_children; i++) {
	vdev_t *vc = vdev->vdev_child[i];

	if (vc->vdev_path == NULL) {
	vc = zdb_vdev_lookup(vc, path);
	if (vc == NULL)
	continue;
	else
	return (vc);
	}

	p = strrchr(vc->vdev_path, '/');
	p = p ? p + 1 : vc->vdev_path;
	q = &vc->vdev_path[strlen(vc->vdev_path) - 2];

	if (strcmp(vc->vdev_path, path) == 0)
	return (vc);
	if (strcmp(p, path) == 0)
	return (vc);
	if (strcmp(q, "s0") == 0 && strncmp(p, path, q - p) == 0)
	return (vc);
	}

	return (NULL);
	}

	static int
	name_from_objset_id(spa_t spa, uint64_t objset_id, char outstr)
	{
	dsl_dataset_t *ds;

	dsl_pool_config_enter(spa->spa_dsl_pool, FTAG);
	int error = dsl_dataset_hold_obj(spa->spa_dsl_pool, objset_id,
	NULL, &ds);
	if (error != 0) {
	(void) fprintf(stderr, "failed to hold objset %llu: %s\n",
	(u_longlong_t)objset_id, strerror(error));
	dsl_pool_config_exit(spa->spa_dsl_pool, FTAG);
	return (error);
	}
	dsl_dataset_name(ds, outstr);
	dsl_dataset_rele(ds, NULL);
	dsl_pool_config_exit(spa->spa_dsl_pool, FTAG);
	return (0);
	}

	static boolean_t
	zdb_parse_block_sizes(char sizes, uint64_t lsize, uint64_t *psize)
	{
	char s0, s1, *tmp = NULL;

	if (sizes == NULL)
	return (B_FALSE);

	s0 = strtok_r(sizes, "/", &tmp);
	if (s0 == NULL)
	return (B_FALSE);
	s1 = strtok_r(NULL, "/", &tmp);
	*lsize = strtoull(s0, NULL, 16);
	psize = s1 ? strtoull(s1, NULL, 16) : lsize;
	return (lsize >= psize && *psize > 0);
	}

	#define ZIO_COMPRESS_MASK(alg) (1ULL << (ZIO_COMPRESS_##alg))

	static boolean_t
	zdb_decompress_block(abd_t pabd, void buf, void *lbuf, uint64_t lsize,
	uint64_t psize, int flags)
	{
	boolean_t exceeded = B_FALSE;
	/*
	* We don't know how the data was compressed, so just try
	* every decompress function at every inflated blocksize.
	*/
	void *lbuf2 = umem_alloc(SPA_MAXBLOCKSIZE, UMEM_NOFAIL);
	int cfuncs[ZIO_COMPRESS_FUNCTIONS] = { 0 };
	int *cfuncp = cfuncs;
	uint64_t maxlsize = SPA_MAXBLOCKSIZE;
	uint64_t mask = ZIO_COMPRESS_MASK(ON) \| ZIO_COMPRESS_MASK(OFF) \|
	ZIO_COMPRESS_MASK(INHERIT) \| ZIO_COMPRESS_MASK(EMPTY) \|
	(getenv("ZDB_NO_ZLE") ? ZIO_COMPRESS_MASK(ZLE) : 0);
	*cfuncp++ = ZIO_COMPRESS_LZ4;
	*cfuncp++ = ZIO_COMPRESS_LZJB;
	mask \|= ZIO_COMPRESS_MASK(LZ4) \| ZIO_COMPRESS_MASK(LZJB);
	for (int c = 0; c < ZIO_COMPRESS_FUNCTIONS; c++)
	if (((1ULL << c) & mask) == 0)
	*cfuncp++ = c;

	/*
	* On the one hand, with SPA_MAXBLOCKSIZE at 16MB, this
	* could take a while and we should let the user know
	* we are not stuck. On the other hand, printing progress
	* info gets old after a while. User can specify 'v' flag
	* to see the progression.
	*/
	if (lsize == psize)
	lsize += SPA_MINBLOCKSIZE;
	else
	maxlsize = lsize;
	for (; lsize <= maxlsize; lsize += SPA_MINBLOCKSIZE) {
	for (cfuncp = cfuncs; *cfuncp; cfuncp++) {
	if (flags & ZDB_FLAG_VERBOSE) {
	(void) fprintf(stderr,
	"Trying %05llx -> %05llx (%s)\n",
	(u_longlong_t)psize,
	(u_longlong_t)lsize,
	zio_compress_table[*cfuncp].\
	ci_name);
	}

	/*
	* We randomize lbuf2, and decompress to both
	* lbuf and lbuf2. This way, we will know if
	* decompression fill exactly to lsize.
	*/
	VERIFY0(random_get_pseudo_bytes(lbuf2, lsize));

	if (zio_decompress_data(*cfuncp, pabd,
	lbuf, psize, lsize, NULL) == 0 &&
	zio_decompress_data(*cfuncp, pabd,
	lbuf2, psize, lsize, NULL) == 0 &&
	bcmp(lbuf, lbuf2, lsize) == 0)
	break;
	}
	if (*cfuncp != 0)
	break;
	}
	umem_free(lbuf2, SPA_MAXBLOCKSIZE);

	if (lsize > maxlsize) {
	exceeded = B_TRUE;
	}
	if (*cfuncp == ZIO_COMPRESS_ZLE) {
	printf("\nZLE decompression was selected. If you "
	"suspect the results are wrong,\ntry avoiding ZLE "
	"by setting and exporting ZDB_NO_ZLE=\"true\"\n");
	}

	return (exceeded);
	}

	/*
	* Read a block from a pool and print it out. The syntax of the
	* block descriptor is:
	*
	* pool:vdev_specifier:offset:[lsize/]psize[:flags]
	*
	* pool - The name of the pool you wish to read from
	* vdev_specifier - Which vdev (see comment for zdb_vdev_lookup)
	* offset - offset, in hex, in bytes
	* size - Amount of data to read, in hex, in bytes
	* flags - A string of characters specifying options
	* b: Decode a blkptr at given offset within block
	* c: Calculate and display checksums
	* d: Decompress data before dumping
	* e: Byteswap data before dumping
	* g: Display data as a gang block header
	* i: Display as an indirect block
	* r: Dump raw data to stdout
	* v: Verbose
	*
	*/
	static void
	zdb_read_block(char thing, spa_t spa)
	{
	blkptr_t blk, *bp = &blk;
	dva_t *dva = bp->blk_dva;
	int flags = 0;
	uint64_t offset = 0, psize = 0, lsize = 0, blkptr_offset = 0;
	zio_t *zio;
	vdev_t *vd;
	abd_t *pabd;
	void lbuf, buf;
	char s, p, dup, vdev, flagstr, sizes, *tmp = NULL;
	int i, error;
	boolean_t borrowed = B_FALSE, found = B_FALSE;

	dup = strdup(thing);
	s = strtok_r(dup, ":", &tmp);
	vdev = s ? s : "";
	s = strtok_r(NULL, ":", &tmp);
	offset = strtoull(s ? s : "", NULL, 16);
	sizes = strtok_r(NULL, ":", &tmp);
	s = strtok_r(NULL, ":", &tmp);
	flagstr = strdup(s ? s : "");

	s = NULL;
	tmp = NULL;
	if (!zdb_parse_block_sizes(sizes, &lsize, &psize))
	s = "invalid size(s)";
	if (!IS_P2ALIGNED(psize, DEV_BSIZE) \|\| !IS_P2ALIGNED(lsize, DEV_BSIZE))
	s = "size must be a multiple of sector size";
	if (!IS_P2ALIGNED(offset, DEV_BSIZE))
	s = "offset must be a multiple of sector size";
	if (s) {
	(void) printf("Invalid block specifier: %s - %s\n", thing, s);
	goto done;
	}

	for (s = strtok_r(flagstr, ":", &tmp);
	s != NULL;
	s = strtok_r(NULL, ":", &tmp)) {
	for (i = 0; i < strlen(flagstr); i++) {
	int bit = flagbits[(uchar_t)flagstr[i]];

	if (bit == 0) {
	(void) printf("***Ignoring flag: %c\n",
	(uchar_t)flagstr[i]);
	continue;
	}
	found = B_TRUE;
	flags \|= bit;

	p = &flagstr[i + 1];
	if (p != ':' && p != '\0') {
	int j = 0, nextbit = flagbits[(uchar_t)*p];
	char *end, offstr[8] = { 0 };
	if ((bit == ZDB_FLAG_PRINT_BLKPTR) &&
	(nextbit == 0)) {
	/* look ahead to isolate the offset */
	while (nextbit == 0 &&
	strchr(flagbitstr, *p) == NULL) {
	offstr[j] = *p;
	j++;
	if (i + j > strlen(flagstr))
	break;
	p++;
	nextbit = flagbits[(uchar_t)*p];
	}
	blkptr_offset = strtoull(offstr, &end,
	16);
	i += j;
	} else if (nextbit == 0) {
	(void) printf("***Ignoring flag arg:"
	" '%c'\n", (uchar_t)*p);
	}
	}
	}
	}
	if (blkptr_offset % sizeof (blkptr_t)) {
	printf("Block pointer offset 0x%llx "
	"must be divisible by 0x%x\n",
	(longlong_t)blkptr_offset, (int)sizeof (blkptr_t));
	goto done;
	}
	if (found == B_FALSE && strlen(flagstr) > 0) {
	printf("Invalid flag arg: '%s'\n", flagstr);
	goto done;
	}

	vd = zdb_vdev_lookup(spa->spa_root_vdev, vdev);
	if (vd == NULL) {
	(void) printf("***Invalid vdev: %s\n", vdev);
	free(dup);
	return;
	} else {
	if (vd->vdev_path)
	(void) fprintf(stderr, "Found vdev: %s\n",
	vd->vdev_path);
	else
	(void) fprintf(stderr, "Found vdev type: %s\n",
	vd->vdev_ops->vdev_op_type);
	}

	pabd = abd_alloc_for_io(SPA_MAXBLOCKSIZE, B_FALSE);
	lbuf = umem_alloc(SPA_MAXBLOCKSIZE, UMEM_NOFAIL);

	BP_ZERO(bp);

	DVA_SET_VDEV(&dva[0], vd->vdev_id);
	DVA_SET_OFFSET(&dva[0], offset);
	DVA_SET_GANG(&dva[0], !!(flags & ZDB_FLAG_GBH));
	DVA_SET_ASIZE(&dva[0], vdev_psize_to_asize(vd, psize));

	BP_SET_BIRTH(bp, TXG_INITIAL, TXG_INITIAL);

	BP_SET_LSIZE(bp, lsize);
	BP_SET_PSIZE(bp, psize);
	BP_SET_COMPRESS(bp, ZIO_COMPRESS_OFF);
	BP_SET_CHECKSUM(bp, ZIO_CHECKSUM_OFF);
	BP_SET_TYPE(bp, DMU_OT_NONE);
	BP_SET_LEVEL(bp, 0);
	BP_SET_DEDUP(bp, 0);
	BP_SET_BYTEORDER(bp, ZFS_HOST_BYTEORDER);

	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
	zio = zio_root(spa, NULL, NULL, 0);

	if (vd == vd->vdev_top) {
	/*
	* Treat this as a normal block read.
	*/
	zio_nowait(zio_read(zio, spa, bp, pabd, psize, NULL, NULL,
	ZIO_PRIORITY_SYNC_READ,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_RAW, NULL));
	} else {
	/*
	* Treat this as a vdev child I/O.
	*/
	zio_nowait(zio_vdev_child_io(zio, bp, vd, offset, pabd,
	psize, ZIO_TYPE_READ, ZIO_PRIORITY_SYNC_READ,
	ZIO_FLAG_DONT_CACHE \| ZIO_FLAG_DONT_PROPAGATE \|
	ZIO_FLAG_DONT_RETRY \| ZIO_FLAG_CANFAIL \| ZIO_FLAG_RAW \|
	ZIO_FLAG_OPTIONAL, NULL, NULL));
	}

	error = zio_wait(zio);
	spa_config_exit(spa, SCL_STATE, FTAG);

	if (error) {
	(void) printf("Read of %s failed, error: %d\n", thing, error);
	goto out;
	}

	uint64_t orig_lsize = lsize;
	buf = lbuf;
	if (flags & ZDB_FLAG_DECOMPRESS) {
	boolean_t failed = zdb_decompress_block(pabd, buf, lbuf,
	lsize, psize, flags);
	if (failed) {
	(void) printf("Decompress of %s failed\n", thing);
	goto out;
	}
	} else {
	buf = abd_borrow_buf_copy(pabd, lsize);
	borrowed = B_TRUE;
	}
	/*
	* Try to detect invalid block pointer. If invalid, try
	* decompressing.
	*/
	if ((flags & ZDB_FLAG_PRINT_BLKPTR \|\| flags & ZDB_FLAG_INDIRECT) &&
	!(flags & ZDB_FLAG_DECOMPRESS)) {
	const blkptr_t b = (const blkptr_t )(void *)
	((uintptr_t)buf + (uintptr_t)blkptr_offset);
	if (zfs_blkptr_verify(spa, b, B_FALSE, BLK_VERIFY_ONLY) ==
	B_FALSE) {
	abd_return_buf_copy(pabd, buf, lsize);
	borrowed = B_FALSE;
	buf = lbuf;
	boolean_t failed = zdb_decompress_block(pabd, buf,
	lbuf, lsize, psize, flags);
	b = (const blkptr_t )(void )
	((uintptr_t)buf + (uintptr_t)blkptr_offset);
	if (failed \|\| zfs_blkptr_verify(spa, b, B_FALSE,
	BLK_VERIFY_LOG) == B_FALSE) {
	printf("invalid block pointer at this DVA\n");
	goto out;
	}
	}
	}

	if (flags & ZDB_FLAG_PRINT_BLKPTR)
	zdb_print_blkptr((blkptr_t )(void )
	((uintptr_t)buf + (uintptr_t)blkptr_offset), flags);
	else if (flags & ZDB_FLAG_RAW)
	zdb_dump_block_raw(buf, lsize, flags);
	else if (flags & ZDB_FLAG_INDIRECT)
	zdb_dump_indirect((blkptr_t *)buf,
	orig_lsize / sizeof (blkptr_t), flags);
	else if (flags & ZDB_FLAG_GBH)
	zdb_dump_gbh(buf, flags);
	else
	zdb_dump_block(thing, buf, lsize, flags);

	/*
	* If :c was specified, iterate through the checksum table to
	* calculate and display each checksum for our specified
	* DVA and length.
	*/
	if ((flags & ZDB_FLAG_CHECKSUM) && !(flags & ZDB_FLAG_RAW) &&
	!(flags & ZDB_FLAG_GBH)) {
	zio_t *czio;
	(void) printf("\n");
	for (enum zio_checksum ck = ZIO_CHECKSUM_LABEL;
	ck < ZIO_CHECKSUM_FUNCTIONS; ck++) {

	if ((zio_checksum_table[ck].ci_flags &
	ZCHECKSUM_FLAG_EMBEDDED) \|\|
	ck == ZIO_CHECKSUM_NOPARITY) {
	continue;
	}
	BP_SET_CHECKSUM(bp, ck);
	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
	czio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL);
	czio->io_bp = bp;

	if (vd == vd->vdev_top) {
	zio_nowait(zio_read(czio, spa, bp, pabd, psize,
	NULL, NULL,
	ZIO_PRIORITY_SYNC_READ,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_RAW \|
	ZIO_FLAG_DONT_RETRY, NULL));
	} else {
	zio_nowait(zio_vdev_child_io(czio, bp, vd,
	offset, pabd, psize, ZIO_TYPE_READ,
	ZIO_PRIORITY_SYNC_READ,
	ZIO_FLAG_DONT_CACHE \|
	ZIO_FLAG_DONT_PROPAGATE \|
	ZIO_FLAG_DONT_RETRY \|
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_RAW \|
	ZIO_FLAG_SPECULATIVE \|
	ZIO_FLAG_OPTIONAL, NULL, NULL));
	}
	error = zio_wait(czio);
	if (error == 0 \|\| error == ECKSUM) {
	zio_t *ck_zio = zio_root(spa, NULL, NULL, 0);
	ck_zio->io_offset =
	DVA_GET_OFFSET(&bp->blk_dva[0]);
	ck_zio->io_bp = bp;
	zio_checksum_compute(ck_zio, ck, pabd, lsize);
	printf("%12s\tcksum=%llx:%llx:%llx:%llx\n",
	zio_checksum_table[ck].ci_name,
	(u_longlong_t)bp->blk_cksum.zc_word[0],
	(u_longlong_t)bp->blk_cksum.zc_word[1],
	(u_longlong_t)bp->blk_cksum.zc_word[2],
	(u_longlong_t)bp->blk_cksum.zc_word[3]);
	zio_wait(ck_zio);
	} else {
	printf("error %d reading block\n", error);
	}
	spa_config_exit(spa, SCL_STATE, FTAG);
	}
	}

	if (borrowed)
	abd_return_buf_copy(pabd, buf, lsize);

	out:
	abd_free(pabd);
	umem_free(lbuf, SPA_MAXBLOCKSIZE);
	done:
	free(flagstr);
	free(dup);
	}

	static void
	zdb_embedded_block(char *thing)
	{
	blkptr_t bp;
	unsigned long long words = (void )&bp;
	char *buf;
	int err;

	bzero(&bp, sizeof (bp));
	err = sscanf(thing, "%llx:%llx:%llx:%llx:%llx:%llx:%llx:%llx:"
	"%llx:%llx:%llx:%llx:%llx:%llx:%llx:%llx",
	words + 0, words + 1, words + 2, words + 3,
	words + 4, words + 5, words + 6, words + 7,
	words + 8, words + 9, words + 10, words + 11,
	words + 12, words + 13, words + 14, words + 15);
	if (err != 16) {
	(void) fprintf(stderr, "invalid input format\n");
	exit(1);
	}
	ASSERT3U(BPE_GET_LSIZE(&bp), <=, SPA_MAXBLOCKSIZE);
	buf = malloc(SPA_MAXBLOCKSIZE);
	if (buf == NULL) {
	(void) fprintf(stderr, "out of memory\n");
	exit(1);
	}
	err = decode_embedded_bp(&bp, buf, BPE_GET_LSIZE(&bp));
	if (err != 0) {
	(void) fprintf(stderr, "decode failed: %u\n", err);
	exit(1);
	}
	zdb_dump_block_raw(buf, BPE_GET_LSIZE(&bp), 0);
	free(buf);
	}

	int
	main(int argc, char **argv)
	{
	int c;
	struct rlimit rl = { 1024, 1024 };
	spa_t *spa = NULL;
	objset_t *os = NULL;
	int dump_all = 1;
	int verbose = 0;
	int error = 0;
	char **searchdirs = NULL;
	int nsearch = 0;
	char target, target_pool, dsname[ZFS_MAX_DATASET_NAME_LEN];
	nvlist_t *policy = NULL;
	uint64_t max_txg = UINT64_MAX;
	int64_t objset_id = -1;
	uint64_t object;
	int flags = ZFS_IMPORT_MISSING_LOG;
	int rewind = ZPOOL_NEVER_REWIND;
	char spa_config_path_env, objset_str;
	boolean_t target_is_spa = B_TRUE, dataset_lookup = B_FALSE;
	nvlist_t *cfg = NULL;

	(void) setrlimit(RLIMIT_NOFILE, &rl);
	(void) enable_extended_FILE_stdio(-1, -1);

	dprintf_setup(&argc, argv);

	/*
	* If there is an environment variable SPA_CONFIG_PATH it overrides
	* default spa_config_path setting. If -U flag is specified it will
	* override this environment variable settings once again.
	*/
	spa_config_path_env = getenv("SPA_CONFIG_PATH");
	if (spa_config_path_env != NULL)
	spa_config_path = spa_config_path_env;

	/*
	* For performance reasons, we set this tunable down. We do so before
	* the arg parsing section so that the user can override this value if
	* they choose.
	*/
	zfs_btree_verify_intensity = 3;

	while ((c = getopt(argc, argv,
	"AbcCdDeEFGhiI:klLmMo:Op:PqrRsSt:uU:vVx:XYyZ")) != -1) {
	switch (c) {
	case 'b':
	case 'c':
	case 'C':
	case 'd':
	case 'D':
	case 'E':
	case 'G':
	case 'h':
	case 'i':
	case 'l':
	case 'm':
	case 'M':
	case 'O':
	case 'r':
	case 'R':
	case 's':
	case 'S':
	case 'u':
	case 'y':
	case 'Z':
	dump_opt[c]++;
	dump_all = 0;
	break;
	case 'A':
	case 'e':
	case 'F':
	case 'k':
	case 'L':
	case 'P':
	case 'q':
	case 'X':
	dump_opt[c]++;
	break;
	case 'Y':
	zfs_reconstruct_indirect_combinations_max = INT_MAX;
	zfs_deadman_enabled = 0;
	break;
	/* NB: Sort single match options below. */
	case 'I':
	max_inflight_bytes = strtoull(optarg, NULL, 0);
	if (max_inflight_bytes == 0) {
	(void) fprintf(stderr, "maximum number "
	"of inflight bytes must be greater "
	"than 0\n");
	usage();
	}
	break;
	case 'o':
	error = set_global_var(optarg);
	if (error != 0)
	usage();
	break;
	case 'p':
	if (searchdirs == NULL) {
	searchdirs = umem_alloc(sizeof (char *),
	UMEM_NOFAIL);
	} else {
	char *tmp = umem_alloc((nsearch + 1)
	sizeof (char *), UMEM_NOFAIL);
	bcopy(searchdirs, tmp, nsearch *
	sizeof (char *));
	umem_free(searchdirs,
	nsearch * sizeof (char *));
	searchdirs = tmp;
	}
	searchdirs[nsearch++] = optarg;
	break;
	case 't':
	max_txg = strtoull(optarg, NULL, 0);
	if (max_txg < TXG_INITIAL) {
	(void) fprintf(stderr, "incorrect txg "
	"specified: %s\n", optarg);
	usage();
	}
	break;
	case 'U':
	spa_config_path = optarg;
	if (spa_config_path[0] != '/') {
	(void) fprintf(stderr,
	"cachefile must be an absolute path "
	"(i.e. start with a slash)\n");
	usage();
	}
	break;
	case 'v':
	verbose++;
	break;
	case 'V':
	flags = ZFS_IMPORT_VERBATIM;
	break;
	case 'x':
	vn_dumpdir = optarg;
	break;
	default:
	usage();
	break;
	}
	}

	if (!dump_opt['e'] && searchdirs != NULL) {
	(void) fprintf(stderr, "-p option requires use of -e\n");
	usage();
	}
	if (dump_opt['d'] \|\| dump_opt['r']) {
	/* <pool>[/<dataset \| objset id> is accepted */
	if (argv[2] && (objset_str = strchr(argv[2], '/')) != NULL &&
	objset_str++ != NULL) {
	char *endptr;
	errno = 0;
	objset_id = strtoull(objset_str, &endptr, 0);
	/* dataset 0 is the same as opening the pool */
	if (errno == 0 && endptr != objset_str &&
	objset_id != 0) {
	target_is_spa = B_FALSE;
	dataset_lookup = B_TRUE;
	} else if (objset_id != 0) {
	printf("failed to open objset %s "
	"%llu %s", objset_str,
	(u_longlong_t)objset_id,
	strerror(errno));
	exit(1);
	}
	/* normal dataset name not an objset ID */
	if (endptr == objset_str) {
	objset_id = -1;
	}
	}
	}

	#if defined(_LP64)
	/*
	* ZDB does not typically re-read blocks; therefore limit the ARC
	* to 256 MB, which can be used entirely for metadata.
	*/
	zfs_arc_min = zfs_arc_meta_min = 2ULL << SPA_MAXBLOCKSHIFT;
	zfs_arc_max = zfs_arc_meta_limit = 256 * 1024 * 1024;
	#endif

	/*
	* "zdb -c" uses checksum-verifying scrub i/os which are async reads.
	* "zdb -b" uses traversal prefetch which uses async reads.
	* For good performance, let several of them be active at once.
	*/
	zfs_vdev_async_read_max_active = 10;

	/*
	* Disable reference tracking for better performance.
	*/
	reference_tracking_enable = B_FALSE;

	/*
	* Do not fail spa_load when spa_load_verify fails. This is needed
	* to load non-idle pools.
	*/
	spa_load_verify_dryrun = B_TRUE;

	/*
	* ZDB should have ability to read spacemaps.
	*/
	spa_mode_readable_spacemaps = B_TRUE;

	kernel_init(SPA_MODE_READ);

	if (dump_all)
	verbose = MAX(verbose, 1);

	for (c = 0; c < 256; c++) {
	if (dump_all && strchr("AeEFklLOPrRSXy", c) == NULL)
	dump_opt[c] = 1;
	if (dump_opt[c])
	dump_opt[c] += verbose;
	}

	libspl_assert_ok = (dump_opt['A'] == 1) \|\| (dump_opt['A'] > 2);
	zfs_recover = (dump_opt['A'] > 1);

	argc -= optind;
	argv += optind;
	if (argc < 2 && dump_opt['R'])
	usage();

	if (dump_opt['E']) {
	if (argc != 1)
	usage();
	zdb_embedded_block(argv[0]);
	return (0);
	}

	if (argc < 1) {
	if (!dump_opt['e'] && dump_opt['C']) {
	dump_cachefile(spa_config_path);
	return (0);
	}
	usage();
	}

	if (dump_opt['l'])
	return (dump_label(argv[0]));

	if (dump_opt['O']) {
	if (argc != 2)
	usage();
	dump_opt['v'] = verbose + 3;
	return (dump_path(argv[0], argv[1], NULL));
	}
	if (dump_opt['r']) {
	if (argc != 3)
	usage();
	dump_opt['v'] = verbose;
	error = dump_path(argv[0], argv[1], &object);
	}

	if (dump_opt['X'] \|\| dump_opt['F'])
	rewind = ZPOOL_DO_REWIND \|
	(dump_opt['X'] ? ZPOOL_EXTREME_REWIND : 0);

	if (nvlist_alloc(&policy, NV_UNIQUE_NAME_TYPE, 0) != 0 \|\|
	nvlist_add_uint64(policy, ZPOOL_LOAD_REQUEST_TXG, max_txg) != 0 \|\|
	nvlist_add_uint32(policy, ZPOOL_LOAD_REWIND_POLICY, rewind) != 0)
	fatal("internal error: %s", strerror(ENOMEM));

	error = 0;
	target = argv[0];

	if (strpbrk(target, "/@") != NULL) {
	size_t targetlen;

	target_pool = strdup(target);
	*strpbrk(target_pool, "/@") = '\0';

	target_is_spa = B_FALSE;
	targetlen = strlen(target);
	if (targetlen && target[targetlen - 1] == '/')
	target[targetlen - 1] = '\0';
	} else {
	target_pool = target;
	}

	if (dump_opt['e']) {
	importargs_t args = { 0 };

	args.paths = nsearch;
	args.path = searchdirs;
	args.can_be_active = B_TRUE;

	error = zpool_find_config(NULL, target_pool, &cfg, &args,
	&libzpool_config_ops);

	if (error == 0) {

	if (nvlist_add_nvlist(cfg,
	ZPOOL_LOAD_POLICY, policy) != 0) {
	fatal("can't open '%s': %s",
	target, strerror(ENOMEM));
	}

	if (dump_opt['C'] > 1) {
	(void) printf("\nConfiguration for import:\n");
	dump_nvlist(cfg, 8);
	}

	/*
	* Disable the activity check to allow examination of
	* active pools.
	*/
	error = spa_import(target_pool, cfg, NULL,
	flags \| ZFS_IMPORT_SKIP_MMP);
	}
	}

	if (searchdirs != NULL) {
	umem_free(searchdirs, nsearch * sizeof (char *));
	searchdirs = NULL;
	}

	/*
	* import_checkpointed_state makes the assumption that the
	* target pool that we pass it is already part of the spa
	* namespace. Because of that we need to make sure to call
	* it always after the -e option has been processed, which
	* imports the pool to the namespace if it's not in the
	* cachefile.
	*/
	char *checkpoint_pool = NULL;
	char *checkpoint_target = NULL;
	if (dump_opt['k']) {
	checkpoint_pool = import_checkpointed_state(target, cfg,
	&checkpoint_target);

	if (checkpoint_target != NULL)
	target = checkpoint_target;
	}

	if (cfg != NULL) {
	nvlist_free(cfg);
	cfg = NULL;
	}

	if (target_pool != target)
	free(target_pool);

	if (error == 0) {
	if (dump_opt['k'] && (target_is_spa \|\| dump_opt['R'])) {
	ASSERT(checkpoint_pool != NULL);
	ASSERT(checkpoint_target == NULL);

	error = spa_open(checkpoint_pool, &spa, FTAG);
	if (error != 0) {
	fatal("Tried to open pool \"%s\" but "
	"spa_open() failed with error %d\n",
	checkpoint_pool, error);
	}

	} else if (target_is_spa \|\| dump_opt['R'] \|\| objset_id == 0) {
	zdb_set_skip_mmp(target);
	error = spa_open_rewind(target, &spa, FTAG, policy,
	NULL);
	if (error) {
	/*
	* If we're missing the log device then
	* try opening the pool after clearing the
	* log state.
	*/
	mutex_enter(&spa_namespace_lock);
	if ((spa = spa_lookup(target)) != NULL &&
	spa->spa_log_state == SPA_LOG_MISSING) {
	spa->spa_log_state = SPA_LOG_CLEAR;
	error = 0;
	}
	mutex_exit(&spa_namespace_lock);

	if (!error) {
	error = spa_open_rewind(target, &spa,
	FTAG, policy, NULL);
	}
	}
	} else if (strpbrk(target, "#") != NULL) {
	dsl_pool_t *dp;
	error = dsl_pool_hold(target, FTAG, &dp);
	if (error != 0) {
	fatal("can't dump '%s': %s", target,
	strerror(error));
	}
	error = dump_bookmark(dp, target, B_TRUE, verbose > 1);
	dsl_pool_rele(dp, FTAG);
	if (error != 0) {
	fatal("can't dump '%s': %s", target,
	strerror(error));
	}
	return (error);
	} else {
	zdb_set_skip_mmp(target);
	if (dataset_lookup == B_TRUE) {
	/*
	* Use the supplied id to get the name
	* for open_objset.
	*/
	error = spa_open(target, &spa, FTAG);
	if (error == 0) {
	error = name_from_objset_id(spa,
	objset_id, dsname);
	spa_close(spa, FTAG);
	if (error == 0)
	target = dsname;
	}
	}
	if (error == 0)
	error = open_objset(target, FTAG, &os);
	if (error == 0)
	spa = dmu_objset_spa(os);
	}
	}
	nvlist_free(policy);

	if (error)
	fatal("can't open '%s': %s", target, strerror(error));

	/*
	* Set the pool failure mode to panic in order to prevent the pool
	* from suspending. A suspended I/O will have no way to resume and
	* can prevent the zdb(8) command from terminating as expected.
	*/
	if (spa != NULL)
	spa->spa_failmode = ZIO_FAILURE_MODE_PANIC;

	argv++;
	argc--;
	if (dump_opt['r']) {
	error = zdb_copy_object(os, object, argv[1]);
	} else if (!dump_opt['R']) {
	flagbits['d'] = ZOR_FLAG_DIRECTORY;
	flagbits['f'] = ZOR_FLAG_PLAIN_FILE;
	flagbits['m'] = ZOR_FLAG_SPACE_MAP;
	flagbits['z'] = ZOR_FLAG_ZAP;
	flagbits['A'] = ZOR_FLAG_ALL_TYPES;

	if (argc > 0 && dump_opt['d']) {
	zopt_object_args = argc;
	zopt_object_ranges = calloc(zopt_object_args,
	sizeof (zopt_object_range_t));
	for (unsigned i = 0; i < zopt_object_args; i++) {
	int err;
	char *msg = NULL;

	err = parse_object_range(argv[i],
	&zopt_object_ranges[i], &msg);
	if (err != 0)
	fatal("Bad object or range: '%s': %s\n",
	argv[i], msg ? msg : "");
	}
	} else if (argc > 0 && dump_opt['m']) {
	zopt_metaslab_args = argc;
	zopt_metaslab = calloc(zopt_metaslab_args,
	sizeof (uint64_t));
	for (unsigned i = 0; i < zopt_metaslab_args; i++) {
	errno = 0;
	zopt_metaslab[i] = strtoull(argv[i], NULL, 0);
	if (zopt_metaslab[i] == 0 && errno != 0)
	fatal("bad number %s: %s", argv[i],
	strerror(errno));
	}
	}
	if (os != NULL) {
	dump_objset(os);
	} else if (zopt_object_args > 0 && !dump_opt['m']) {
	dump_objset(spa->spa_meta_objset);
	} else {
	dump_zpool(spa);
	}
	} else {
	flagbits['b'] = ZDB_FLAG_PRINT_BLKPTR;
	flagbits['c'] = ZDB_FLAG_CHECKSUM;
	flagbits['d'] = ZDB_FLAG_DECOMPRESS;
	flagbits['e'] = ZDB_FLAG_BSWAP;
	flagbits['g'] = ZDB_FLAG_GBH;
	flagbits['i'] = ZDB_FLAG_INDIRECT;
	flagbits['r'] = ZDB_FLAG_RAW;
	flagbits['v'] = ZDB_FLAG_VERBOSE;

	for (int i = 0; i < argc; i++)
	zdb_read_block(argv[i], spa);
	}

	if (dump_opt['k']) {
	free(checkpoint_pool);
	if (!target_is_spa)
	free(checkpoint_target);
	}

	if (os != NULL) {
	close_objset(os, FTAG);
	} else {
	spa_close(spa, FTAG);
	}

	fuid_table_destroy();

	dump_debug_buffer();

	kernel_fini();

	return (error);
	}
	diff --git a/sys/contrib/openzfs/cmd/zed/agents/fmd_api.c b/sys/contrib/openzfs/cmd/zed/agents/fmd_api.c
	index 607b387ca3a8..c9231fba2aee 100644
	--- a/sys/contrib/openzfs/cmd/zed/agents/fmd_api.c
	+++ b/sys/contrib/openzfs/cmd/zed/agents/fmd_api.c
	@@ -1,760 +1,761 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2004, 2010, Oracle and/or its affiliates. All rights reserved.
	*
	* Copyright (c) 2016, Intel Corporation.
	*/

	/*
	* This file implements the minimal FMD module API required to support the
	* fault logic modules in ZED. This support includes module registration,
	* memory allocation, module property accessors, basic case management,
	* one-shot timers and SERD engines.
	*
	* In the ZED runtime, the modules are called from a single thread so no
	* locking is required in this emulated FMD environment.
	*/

	#include <sys/types.h>
	#include <sys/fm/protocol.h>
	#include <uuid/uuid.h>
	#include <signal.h>
	#include <strings.h>
	#include <time.h>

	#include "fmd_api.h"
	#include "fmd_serd.h"

	#include "zfs_agents.h"
	#include "../zed_log.h"

	typedef struct fmd_modstat {
	fmd_stat_t ms_accepted; /* total events accepted by module */
	fmd_stat_t ms_caseopen; /* cases currently open */
	fmd_stat_t ms_casesolved; /* total cases solved by module */
	fmd_stat_t ms_caseclosed; /* total cases closed by module */
	} fmd_modstat_t;

	typedef struct fmd_module {
	const char mod_name; / basename of module (ro) */
	const fmd_hdl_info_t mod_info; / module info registered with handle */
	void mod_spec; / fmd_hdl_get/setspecific data value */
	fmd_stat_t mod_ustat; / module specific custom stats */
	uint_t mod_ustat_cnt; /* count of ustat stats */
	fmd_modstat_t mod_stats; /* fmd built-in per-module statistics */
	fmd_serd_hash_t mod_serds; /* hash of serd engs owned by module */
	char mod_vers; / a copy of module version string */
	} fmd_module_t;

	/*
	* ZED has two FMD hardwired module instances
	*/
	fmd_module_t zfs_retire_module;
	fmd_module_t zfs_diagnosis_module;

	/*
	* Enable a reasonable set of defaults for libumem debugging on DEBUG builds.
	*/

	#ifdef DEBUG
	const char *
	_umem_debug_init(void)
	{
	return ("default,verbose"); /* $UMEM_DEBUG setting */
	}

	const char *
	_umem_logging_init(void)
	{
	return ("fail,contents"); /* $UMEM_LOGGING setting */
	}
	#endif

	/*
	* Register a module with fmd and finish module initialization.
	* Returns an integer indicating whether it succeeded (zero) or
	* failed (non-zero).
	*/
	int
	fmd_hdl_register(fmd_hdl_t hdl, int version, const fmd_hdl_info_t mip)
	{
	fmd_module_t mp = (fmd_module_t )hdl;

	mp->mod_info = mip;
	mp->mod_name = mip->fmdi_desc + 4; /* drop 'ZFS ' prefix */
	mp->mod_spec = NULL;

	/* bare minimum module stats */
	(void) strcpy(mp->mod_stats.ms_accepted.fmds_name, "fmd.accepted");
	(void) strcpy(mp->mod_stats.ms_caseopen.fmds_name, "fmd.caseopen");
	(void) strcpy(mp->mod_stats.ms_casesolved.fmds_name, "fmd.casesolved");
	(void) strcpy(mp->mod_stats.ms_caseclosed.fmds_name, "fmd.caseclosed");

	fmd_serd_hash_create(&mp->mod_serds);

	fmd_hdl_debug(hdl, "register module");

	return (0);
	}

	void
	fmd_hdl_unregister(fmd_hdl_t *hdl)
	{
	fmd_module_t mp = (fmd_module_t )hdl;
	fmd_modstat_t *msp = &mp->mod_stats;
	const fmd_hdl_ops_t *ops = mp->mod_info->fmdi_ops;

	/* dump generic module stats */
	fmd_hdl_debug(hdl, "%s: %llu", msp->ms_accepted.fmds_name,
	msp->ms_accepted.fmds_value.ui64);
	if (ops->fmdo_close != NULL) {
	fmd_hdl_debug(hdl, "%s: %llu", msp->ms_caseopen.fmds_name,
	msp->ms_caseopen.fmds_value.ui64);
	fmd_hdl_debug(hdl, "%s: %llu", msp->ms_casesolved.fmds_name,
	msp->ms_casesolved.fmds_value.ui64);
	fmd_hdl_debug(hdl, "%s: %llu", msp->ms_caseclosed.fmds_name,
	msp->ms_caseclosed.fmds_value.ui64);
	}

	/* dump module specific stats */
	if (mp->mod_ustat != NULL) {
	int i;

	for (i = 0; i < mp->mod_ustat_cnt; i++) {
	fmd_hdl_debug(hdl, "%s: %llu",
	mp->mod_ustat[i].fmds_name,
	mp->mod_ustat[i].fmds_value.ui64);
	}
	}

	fmd_serd_hash_destroy(&mp->mod_serds);

	fmd_hdl_debug(hdl, "unregister module");
	}

	/*
	* fmd_hdl_setspecific() is used to associate a data pointer with
	* the specified handle for the duration of the module's lifetime.
	* This pointer can be retrieved using fmd_hdl_getspecific().
	*/
	void
	fmd_hdl_setspecific(fmd_hdl_t hdl, void spec)
	{
	fmd_module_t mp = (fmd_module_t )hdl;

	mp->mod_spec = spec;
	}

	/*
	* Return the module-specific data pointer previously associated
	* with the handle using fmd_hdl_setspecific().
	*/
	void *
	fmd_hdl_getspecific(fmd_hdl_t *hdl)
	{
	fmd_module_t mp = (fmd_module_t )hdl;

	return (mp->mod_spec);
	}

	void *
	fmd_hdl_alloc(fmd_hdl_t *hdl, size_t size, int flags)
	{
	return (umem_alloc(size, flags));
	}

	void *
	fmd_hdl_zalloc(fmd_hdl_t *hdl, size_t size, int flags)
	{
	return (umem_zalloc(size, flags));
	}

	void
	fmd_hdl_free(fmd_hdl_t hdl, void data, size_t size)
	{
	umem_free(data, size);
	}

	/*
	* Record a module debug message using the specified format.
	*/
	void
	fmd_hdl_debug(fmd_hdl_t hdl, const char format, ...)
	{
	char message[256];
	va_list vargs;
	fmd_module_t mp = (fmd_module_t )hdl;

	va_start(vargs, format);
	(void) vsnprintf(message, sizeof (message), format, vargs);
	va_end(vargs);

	/* prefix message with module name */
	zed_log_msg(LOG_INFO, "%s: %s", mp->mod_name, message);
	}

	/* Property Retrieval */

	int32_t
	fmd_prop_get_int32(fmd_hdl_t hdl, const char name)
	{
	/*
	* These can be looked up in mp->modinfo->fmdi_props
	* For now we just hard code for phase 2. In the
	* future, there can be a ZED based override.
	*/
	if (strcmp(name, "spare_on_remove") == 0)
	return (1);

	if (strcmp(name, "io_N") == 0 \|\| strcmp(name, "checksum_N") == 0)
	return (10); /* N = 10 events */

	return (0);
	}

	int64_t
	fmd_prop_get_int64(fmd_hdl_t hdl, const char name)
	{
	/*
	* These can be looked up in mp->modinfo->fmdi_props
	* For now we just hard code for phase 2. In the
	* future, there can be a ZED based override.
	*/
	if (strcmp(name, "remove_timeout") == 0)
	return (15ULL * 1000ULL * 1000ULL * 1000ULL); /* 15 sec */

	if (strcmp(name, "io_T") == 0 \|\| strcmp(name, "checksum_T") == 0)
	return (1000ULL * 1000ULL * 1000ULL * 600ULL); /* 10 min */

	return (0);
	}

	/* FMD Statistics */

	fmd_stat_t *
	fmd_stat_create(fmd_hdl_t hdl, uint_t flags, uint_t nstats, fmd_stat_t statv)
	{
	fmd_module_t mp = (fmd_module_t )hdl;

	if (flags == FMD_STAT_NOALLOC) {
	mp->mod_ustat = statv;
	mp->mod_ustat_cnt = nstats;
	}

	return (statv);
	}

	/* Case Management */

	fmd_case_t *
	fmd_case_open(fmd_hdl_t hdl, void data)
	{
	fmd_module_t mp = (fmd_module_t )hdl;
	uuid_t uuid;

	fmd_case_t *cp;

	cp = fmd_hdl_zalloc(hdl, sizeof (fmd_case_t), FMD_SLEEP);
	cp->ci_mod = hdl;
	cp->ci_state = FMD_CASE_UNSOLVED;
	cp->ci_flags = FMD_CF_DIRTY;
	cp->ci_data = data;
	cp->ci_bufptr = NULL;
	cp->ci_bufsiz = 0;

	uuid_generate(uuid);
	uuid_unparse(uuid, cp->ci_uuid);

	fmd_hdl_debug(hdl, "case opened (%s)", cp->ci_uuid);
	mp->mod_stats.ms_caseopen.fmds_value.ui64++;

	return (cp);
	}

	void
	fmd_case_solve(fmd_hdl_t hdl, fmd_case_t cp)
	{
	fmd_module_t mp = (fmd_module_t )hdl;

	/*
	* For ZED, the event was already sent from fmd_case_add_suspect()
	*/

	if (cp->ci_state >= FMD_CASE_SOLVED)
	fmd_hdl_debug(hdl, "case is already solved or closed");

	cp->ci_state = FMD_CASE_SOLVED;

	fmd_hdl_debug(hdl, "case solved (%s)", cp->ci_uuid);
	mp->mod_stats.ms_casesolved.fmds_value.ui64++;
	}

	void
	fmd_case_close(fmd_hdl_t hdl, fmd_case_t cp)
	{
	fmd_module_t mp = (fmd_module_t )hdl;
	const fmd_hdl_ops_t *ops = mp->mod_info->fmdi_ops;

	fmd_hdl_debug(hdl, "case closed (%s)", cp->ci_uuid);

	if (ops->fmdo_close != NULL)
	ops->fmdo_close(hdl, cp);

	mp->mod_stats.ms_caseopen.fmds_value.ui64--;
	mp->mod_stats.ms_caseclosed.fmds_value.ui64++;

	if (cp->ci_bufptr != NULL && cp->ci_bufsiz > 0)
	fmd_hdl_free(hdl, cp->ci_bufptr, cp->ci_bufsiz);

	fmd_hdl_free(hdl, cp, sizeof (fmd_case_t));
	}

	void
	fmd_case_uuresolved(fmd_hdl_t hdl, const char uuid)
	{
	fmd_hdl_debug(hdl, "case resolved by uuid (%s)", uuid);
	}

	int
	fmd_case_solved(fmd_hdl_t hdl, fmd_case_t cp)
	{
	return ((cp->ci_state >= FMD_CASE_SOLVED) ? FMD_B_TRUE : FMD_B_FALSE);
	}

	void
	fmd_case_add_ereport(fmd_hdl_t hdl, fmd_case_t cp, fmd_event_t *ep)
	{
	}

	static void
	zed_log_fault(nvlist_t nvl, const char uuid, const char *code)
	{
	nvlist_t *rsrc;
	char *strval;
	uint64_t guid;
	uint8_t byte;

	zed_log_msg(LOG_INFO, "\nzed_fault_event:");

	if (uuid != NULL)
	zed_log_msg(LOG_INFO, "\t%s: %s", FM_SUSPECT_UUID, uuid);
	if (nvlist_lookup_string(nvl, FM_CLASS, &strval) == 0)
	zed_log_msg(LOG_INFO, "\t%s: %s", FM_CLASS, strval);
	if (code != NULL)
	zed_log_msg(LOG_INFO, "\t%s: %s", FM_SUSPECT_DIAG_CODE, code);
	if (nvlist_lookup_uint8(nvl, FM_FAULT_CERTAINTY, &byte) == 0)
	zed_log_msg(LOG_INFO, "\t%s: %llu", FM_FAULT_CERTAINTY, byte);
	if (nvlist_lookup_nvlist(nvl, FM_FAULT_RESOURCE, &rsrc) == 0) {
	if (nvlist_lookup_string(rsrc, FM_FMRI_SCHEME, &strval) == 0)
	zed_log_msg(LOG_INFO, "\t%s: %s", FM_FMRI_SCHEME,
	strval);
	if (nvlist_lookup_uint64(rsrc, FM_FMRI_ZFS_POOL, &guid) == 0)
	zed_log_msg(LOG_INFO, "\t%s: %llu", FM_FMRI_ZFS_POOL,
	guid);
	if (nvlist_lookup_uint64(rsrc, FM_FMRI_ZFS_VDEV, &guid) == 0)
	zed_log_msg(LOG_INFO, "\t%s: %llu \n", FM_FMRI_ZFS_VDEV,
	guid);
	}
	}

	static const char *
	fmd_fault_mkcode(nvlist_t *fault)
	{
	char class, code = "-";

	/*
	* Note: message codes come from: openzfs/usr/src/cmd/fm/dicts/ZFS.po
	*/
	if (nvlist_lookup_string(fault, FM_CLASS, &class) == 0) {
	if (strcmp(class, "fault.fs.zfs.vdev.io") == 0)
	code = "ZFS-8000-FD";
	else if (strcmp(class, "fault.fs.zfs.vdev.checksum") == 0)
	code = "ZFS-8000-GH";
	else if (strcmp(class, "fault.fs.zfs.io_failure_wait") == 0)
	code = "ZFS-8000-HC";
	else if (strcmp(class, "fault.fs.zfs.io_failure_continue") == 0)
	code = "ZFS-8000-JQ";
	else if (strcmp(class, "fault.fs.zfs.log_replay") == 0)
	code = "ZFS-8000-K4";
	else if (strcmp(class, "fault.fs.zfs.pool") == 0)
	code = "ZFS-8000-CS";
	else if (strcmp(class, "fault.fs.zfs.device") == 0)
	code = "ZFS-8000-D3";

	}
	return (code);
	}

	void
	fmd_case_add_suspect(fmd_hdl_t hdl, fmd_case_t cp, nvlist_t *fault)
	{
	nvlist_t *nvl;
	const char *code = fmd_fault_mkcode(fault);
	int64_t tod[2];
	int err = 0;

	/*
	* payload derived from fmd_protocol_list()
	*/

	(void) gettimeofday(&cp->ci_tv, NULL);
	tod[0] = cp->ci_tv.tv_sec;
	tod[1] = cp->ci_tv.tv_usec;

	nvl = fmd_nvl_alloc(hdl, FMD_SLEEP);

	err \|= nvlist_add_uint8(nvl, FM_VERSION, FM_SUSPECT_VERSION);
	err \|= nvlist_add_string(nvl, FM_CLASS, FM_LIST_SUSPECT_CLASS);
	err \|= nvlist_add_string(nvl, FM_SUSPECT_UUID, cp->ci_uuid);
	err \|= nvlist_add_string(nvl, FM_SUSPECT_DIAG_CODE, code);
	err \|= nvlist_add_int64_array(nvl, FM_SUSPECT_DIAG_TIME, tod, 2);
	err \|= nvlist_add_uint32(nvl, FM_SUSPECT_FAULT_SZ, 1);
	- err \|= nvlist_add_nvlist_array(nvl, FM_SUSPECT_FAULT_LIST, &fault, 1);
	+ err \|= nvlist_add_nvlist_array(nvl, FM_SUSPECT_FAULT_LIST,
	+ (const nvlist_t **)&fault, 1);

	if (err)
	zed_log_die("failed to populate nvlist");

	zed_log_fault(fault, cp->ci_uuid, code);
	zfs_agent_post_event(FM_LIST_SUSPECT_CLASS, NULL, nvl);

	nvlist_free(nvl);
	nvlist_free(fault);
	}

	void
	fmd_case_setspecific(fmd_hdl_t hdl, fmd_case_t cp, void *data)
	{
	cp->ci_data = data;
	}

	void *
	fmd_case_getspecific(fmd_hdl_t hdl, fmd_case_t cp)
	{
	return (cp->ci_data);
	}

	void
	fmd_buf_create(fmd_hdl_t hdl, fmd_case_t cp, const char *name, size_t size)
	{
	assert(strcmp(name, "data") == 0);
	assert(cp->ci_bufptr == NULL);
	assert(size < (1024 * 1024));

	cp->ci_bufptr = fmd_hdl_alloc(hdl, size, FMD_SLEEP);
	cp->ci_bufsiz = size;
	}

	void
	fmd_buf_read(fmd_hdl_t hdl, fmd_case_t cp,
	const char name, void buf, size_t size)
	{
	assert(strcmp(name, "data") == 0);
	assert(cp->ci_bufptr != NULL);
	assert(size <= cp->ci_bufsiz);

	bcopy(cp->ci_bufptr, buf, size);
	}

	void
	fmd_buf_write(fmd_hdl_t hdl, fmd_case_t cp,
	const char name, const void buf, size_t size)
	{
	assert(strcmp(name, "data") == 0);
	assert(cp->ci_bufptr != NULL);
	assert(cp->ci_bufsiz >= size);

	bcopy(buf, cp->ci_bufptr, size);
	}

	/* SERD Engines */

	void
	fmd_serd_create(fmd_hdl_t hdl, const char name, uint_t n, hrtime_t t)
	{
	fmd_module_t mp = (fmd_module_t )hdl;

	if (fmd_serd_eng_lookup(&mp->mod_serds, name) != NULL) {
	zed_log_msg(LOG_ERR, "failed to create SERD engine '%s': "
	" name already exists", name);
	return;
	}

	(void) fmd_serd_eng_insert(&mp->mod_serds, name, n, t);
	}

	void
	fmd_serd_destroy(fmd_hdl_t hdl, const char name)
	{
	fmd_module_t mp = (fmd_module_t )hdl;

	fmd_serd_eng_delete(&mp->mod_serds, name);

	fmd_hdl_debug(hdl, "serd_destroy %s", name);
	}

	int
	fmd_serd_exists(fmd_hdl_t hdl, const char name)
	{
	fmd_module_t mp = (fmd_module_t )hdl;

	return (fmd_serd_eng_lookup(&mp->mod_serds, name) != NULL);
	}

	void
	fmd_serd_reset(fmd_hdl_t hdl, const char name)
	{
	fmd_module_t mp = (fmd_module_t )hdl;
	fmd_serd_eng_t *sgp;

	if ((sgp = fmd_serd_eng_lookup(&mp->mod_serds, name)) == NULL) {
	zed_log_msg(LOG_ERR, "serd engine '%s' does not exist", name);
	return;
	}

	fmd_serd_eng_reset(sgp);

	fmd_hdl_debug(hdl, "serd_reset %s", name);
	}

	int
	fmd_serd_record(fmd_hdl_t hdl, const char name, fmd_event_t *ep)
	{
	fmd_module_t mp = (fmd_module_t )hdl;
	fmd_serd_eng_t *sgp;
	int err;

	if ((sgp = fmd_serd_eng_lookup(&mp->mod_serds, name)) == NULL) {
	zed_log_msg(LOG_ERR, "failed to add record to SERD engine '%s'",
	name);
	return (FMD_B_FALSE);
	}
	err = fmd_serd_eng_record(sgp, ep->ev_hrt);

	return (err);
	}

	/* FMD Timers */

	static void
	_timer_notify(union sigval sv)
	{
	fmd_timer_t *ftp = sv.sival_ptr;
	fmd_hdl_t *hdl = ftp->ft_hdl;
	fmd_module_t mp = (fmd_module_t )hdl;
	const fmd_hdl_ops_t *ops = mp->mod_info->fmdi_ops;
	struct itimerspec its;

	fmd_hdl_debug(hdl, "timer fired (%p)", ftp->ft_tid);

	/* disarm the timer */
	bzero(&its, sizeof (struct itimerspec));
	timer_settime(ftp->ft_tid, 0, &its, NULL);

	/* Note that the fmdo_timeout can remove this timer */
	if (ops->fmdo_timeout != NULL)
	ops->fmdo_timeout(hdl, ftp, ftp->ft_arg);
	}

	/*
	* Install a new timer which will fire at least delta nanoseconds after the
	* current time. After the timeout has expired, the module's fmdo_timeout
	* entry point is called.
	*/
	fmd_timer_t *
	fmd_timer_install(fmd_hdl_t hdl, void arg, fmd_event_t *ep, hrtime_t delta)
	{
	struct sigevent sev;
	struct itimerspec its;
	fmd_timer_t *ftp;

	ftp = fmd_hdl_alloc(hdl, sizeof (fmd_timer_t), FMD_SLEEP);
	ftp->ft_arg = arg;
	ftp->ft_hdl = hdl;

	its.it_value.tv_sec = delta / 1000000000;
	its.it_value.tv_nsec = delta % 1000000000;
	its.it_interval.tv_sec = its.it_value.tv_sec;
	its.it_interval.tv_nsec = its.it_value.tv_nsec;

	sev.sigev_notify = SIGEV_THREAD;
	sev.sigev_notify_function = _timer_notify;
	sev.sigev_notify_attributes = NULL;
	sev.sigev_value.sival_ptr = ftp;

	timer_create(CLOCK_REALTIME, &sev, &ftp->ft_tid);
	timer_settime(ftp->ft_tid, 0, &its, NULL);

	fmd_hdl_debug(hdl, "installing timer for %d secs (%p)",
	(int)its.it_value.tv_sec, ftp->ft_tid);

	return (ftp);
	}

	void
	fmd_timer_remove(fmd_hdl_t hdl, fmd_timer_t ftp)
	{
	fmd_hdl_debug(hdl, "removing timer (%p)", ftp->ft_tid);

	timer_delete(ftp->ft_tid);

	fmd_hdl_free(hdl, ftp, sizeof (fmd_timer_t));
	}

	/* Name-Value Pair Lists */

	nvlist_t *
	fmd_nvl_create_fault(fmd_hdl_t hdl, const char class, uint8_t certainty,
	nvlist_t asru, nvlist_t fru, nvlist_t *resource)
	{
	nvlist_t *nvl;
	int err = 0;

	if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0)
	zed_log_die("failed to xalloc fault nvlist");

	err \|= nvlist_add_uint8(nvl, FM_VERSION, FM_FAULT_VERSION);
	err \|= nvlist_add_string(nvl, FM_CLASS, class);
	err \|= nvlist_add_uint8(nvl, FM_FAULT_CERTAINTY, certainty);

	if (asru != NULL)
	err \|= nvlist_add_nvlist(nvl, FM_FAULT_ASRU, asru);
	if (fru != NULL)
	err \|= nvlist_add_nvlist(nvl, FM_FAULT_FRU, fru);
	if (resource != NULL)
	err \|= nvlist_add_nvlist(nvl, FM_FAULT_RESOURCE, resource);

	if (err)
	zed_log_die("failed to populate nvlist: %s\n", strerror(err));

	return (nvl);
	}

	/*
	* sourced from fmd_string.c
	*/
	static int
	fmd_strmatch(const char s, const char p)
	{
	char c;

	if (p == NULL)
	return (0);

	if (s == NULL)
	s = ""; /* treat NULL string as the empty string */

	do {
	if ((c = *p++) == '\0')
	return (*s == '\0');

	if (c == '*') {
	while (p == '')
	p++; /* consecutive 's can be collapsed /

	if (*p == '\0')
	return (1);

	while (*s != '\0') {
	if (fmd_strmatch(s++, p) != 0)
	return (1);
	}

	return (0);
	}
	} while (c == *s++);

	return (0);
	}

	int
	fmd_nvl_class_match(fmd_hdl_t hdl, nvlist_t nvl, const char *pattern)
	{
	char *class;

	return (nvl != NULL &&
	nvlist_lookup_string(nvl, FM_CLASS, &class) == 0 &&
	fmd_strmatch(class, pattern));
	}

	nvlist_t *
	fmd_nvl_alloc(fmd_hdl_t *hdl, int flags)
	{
	nvlist_t *nvl = NULL;

	if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0)
	return (NULL);

	return (nvl);
	}


	/*
	* ZED Agent specific APIs
	*/

	fmd_hdl_t *
	fmd_module_hdl(const char *name)
	{
	if (strcmp(name, "zfs-retire") == 0)
	return ((fmd_hdl_t *)&zfs_retire_module);
	if (strcmp(name, "zfs-diagnosis") == 0)
	return ((fmd_hdl_t *)&zfs_diagnosis_module);

	return (NULL);
	}

	boolean_t
	fmd_module_initialized(fmd_hdl_t *hdl)
	{
	fmd_module_t mp = (fmd_module_t )hdl;

	return (mp->mod_info != NULL);
	}

	/*
	* fmd_module_recv is called for each event that is received by
	* the fault manager that has a class that matches one of the
	* module's subscriptions.
	*/
	void
	fmd_module_recv(fmd_hdl_t hdl, nvlist_t nvl, const char *class)
	{
	fmd_module_t mp = (fmd_module_t )hdl;
	const fmd_hdl_ops_t *ops = mp->mod_info->fmdi_ops;
	fmd_event_t faux_event = {0};
	int64_t *tv;
	uint_t n;

	/*
	* Will need to normalized this if we persistently store the case data
	*/
	if (nvlist_lookup_int64_array(nvl, FM_EREPORT_TIME, &tv, &n) == 0)
	faux_event.ev_hrt = tv[0] * NANOSEC + tv[1];
	else
	faux_event.ev_hrt = 0;

	ops->fmdo_recv(hdl, &faux_event, nvl, class);

	mp->mod_stats.ms_accepted.fmds_value.ui64++;

	/* TBD - should we initiate fm_module_gc() periodically? */
	}
	diff --git a/sys/contrib/openzfs/cmd/zed/agents/zfs_mod.c b/sys/contrib/openzfs/cmd/zed/agents/zfs_mod.c
	index 3bcdf6e1d718..b496b29ac136 100644
	--- a/sys/contrib/openzfs/cmd/zed/agents/zfs_mod.c
	+++ b/sys/contrib/openzfs/cmd/zed/agents/zfs_mod.c
	@@ -1,991 +1,991 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2012 by Delphix. All rights reserved.
	* Copyright 2014 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2016, 2017, Intel Corporation.
	* Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
	*/

	/*
	* ZFS syseventd module.
	*
	* file origin: openzfs/usr/src/cmd/syseventd/modules/zfs_mod/zfs_mod.c
	*
	* The purpose of this module is to identify when devices are added to the
	* system, and appropriately online or replace the affected vdevs.
	*
	* When a device is added to the system:
	*
	* 1. Search for any vdevs whose devid matches that of the newly added
	* device.
	*
	* 2. If no vdevs are found, then search for any vdevs whose udev path
	* matches that of the new device.
	*
	* 3. If no vdevs match by either method, then ignore the event.
	*
	* 4. Attempt to online the device with a flag to indicate that it should
	* be unspared when resilvering completes. If this succeeds, then the
	* same device was inserted and we should continue normally.
	*
	* 5. If the pool does not have the 'autoreplace' property set, attempt to
	* online the device again without the unspare flag, which will
	* generate a FMA fault.
	*
	* 6. If the pool has the 'autoreplace' property set, and the matching vdev
	* is a whole disk, then label the new disk and attempt a 'zpool
	* replace'.
	*
	* The module responds to EC_DEV_ADD events. The special ESC_ZFS_VDEV_CHECK
	* event indicates that a device failed to open during pool load, but the
	* autoreplace property was set. In this case, we deferred the associated
	* FMA fault until our module had a chance to process the autoreplace logic.
	* If the device could not be replaced, then the second online attempt will
	* trigger the FMA fault that we skipped earlier.
	*
	* On Linux udev provides a disk insert for both the disk and the partition.
	*/

	#include <ctype.h>
	#include <fcntl.h>
	#include <libnvpair.h>
	#include <libzfs.h>
	#include <libzutil.h>
	#include <limits.h>
	#include <stddef.h>
	#include <stdlib.h>
	#include <string.h>
	#include <syslog.h>
	#include <sys/list.h>
	#include <sys/sunddi.h>
	#include <sys/sysevent/eventdefs.h>
	#include <sys/sysevent/dev.h>
	#include <thread_pool.h>
	#include <pthread.h>
	#include <unistd.h>
	#include <errno.h>
	#include "zfs_agents.h"
	#include "../zed_log.h"

	#define DEV_BYID_PATH "/dev/disk/by-id/"
	#define DEV_BYPATH_PATH "/dev/disk/by-path/"
	#define DEV_BYVDEV_PATH "/dev/disk/by-vdev/"

	typedef void (zfs_process_func_t)(zpool_handle_t , nvlist_t *, boolean_t);

	libzfs_handle_t *g_zfshdl;
	list_t g_pool_list; /* list of unavailable pools at initialization */
	list_t g_device_list; /* list of disks with asynchronous label request */
	tpool_t *g_tpool;
	boolean_t g_enumeration_done;
	pthread_t g_zfs_tid; /* zfs_enum_pools() thread */

	typedef struct unavailpool {
	zpool_handle_t *uap_zhp;
	list_node_t uap_node;
	} unavailpool_t;

	typedef struct pendingdev {
	char pd_physpath[128];
	list_node_t pd_node;
	} pendingdev_t;

	static int
	zfs_toplevel_state(zpool_handle_t *zhp)
	{
	nvlist_t *nvroot;
	vdev_stat_t *vs;
	unsigned int c;

	verify(nvlist_lookup_nvlist(zpool_get_config(zhp, NULL),
	ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0);
	verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_VDEV_STATS,
	(uint64_t **)&vs, &c) == 0);
	return (vs->vs_state);
	}

	static int
	zfs_unavail_pool(zpool_handle_t zhp, void data)
	{
	zed_log_msg(LOG_INFO, "zfs_unavail_pool: examining '%s' (state %d)",
	zpool_get_name(zhp), (int)zfs_toplevel_state(zhp));

	if (zfs_toplevel_state(zhp) < VDEV_STATE_DEGRADED) {
	unavailpool_t *uap;
	uap = malloc(sizeof (unavailpool_t));
	uap->uap_zhp = zhp;
	list_insert_tail((list_t *)data, uap);
	} else {
	zpool_close(zhp);
	}
	return (0);
	}

	/*
	* Two stage replace on Linux
	* since we get disk notifications
	* we can wait for partitioned disk slice to show up!
	*
	* First stage tags the disk, initiates async partitioning, and returns
	* Second stage finds the tag and proceeds to ZFS labeling/replace
	*
	* disk-add --> label-disk + tag-disk --> partition-add --> zpool_vdev_attach
	*
	* 1. physical match with no fs, no partition
	* tag it top, partition disk
	*
	* 2. physical match again, see partition and tag
	*
	*/

	/*
	* The device associated with the given vdev (either by devid or physical path)
	* has been added to the system. If 'isdisk' is set, then we only attempt a
	* replacement if it's a whole disk. This also implies that we should label the
	* disk first.
	*
	* First, we attempt to online the device (making sure to undo any spare
	* operation when finished). If this succeeds, then we're done. If it fails,
	* and the new state is VDEV_CANT_OPEN, it indicates that the device was opened,
	* but that the label was not what we expected. If the 'autoreplace' property
	* is enabled, then we relabel the disk (if specified), and attempt a 'zpool
	* replace'. If the online is successful, but the new state is something else
	* (REMOVED or FAULTED), it indicates that we're out of sync or in some sort of
	* race, and we should avoid attempting to relabel the disk.
	*
	* Also can arrive here from a ESC_ZFS_VDEV_CHECK event
	*/
	static void
	zfs_process_add(zpool_handle_t zhp, nvlist_t vdev, boolean_t labeled)
	{
	char *path;
	vdev_state_t newstate;
	nvlist_t nvroot, newvd;
	pendingdev_t *device;
	uint64_t wholedisk = 0ULL;
	uint64_t offline = 0ULL;
	uint64_t guid = 0ULL;
	char physpath = NULL, new_devid = NULL, *enc_sysfs_path = NULL;
	char rawpath[PATH_MAX], fullpath[PATH_MAX];
	char devpath[PATH_MAX];
	int ret;
	boolean_t is_dm = B_FALSE;
	boolean_t is_sd = B_FALSE;
	uint_t c;
	vdev_stat_t *vs;

	if (nvlist_lookup_string(vdev, ZPOOL_CONFIG_PATH, &path) != 0)
	return;

	/* Skip healthy disks */
	verify(nvlist_lookup_uint64_array(vdev, ZPOOL_CONFIG_VDEV_STATS,
	(uint64_t **)&vs, &c) == 0);
	if (vs->vs_state == VDEV_STATE_HEALTHY) {
	zed_log_msg(LOG_INFO, "%s: %s is already healthy, skip it.",
	__func__, path);
	return;
	}

	(void) nvlist_lookup_string(vdev, ZPOOL_CONFIG_PHYS_PATH, &physpath);
	(void) nvlist_lookup_string(vdev, ZPOOL_CONFIG_VDEV_ENC_SYSFS_PATH,
	&enc_sysfs_path);
	(void) nvlist_lookup_uint64(vdev, ZPOOL_CONFIG_WHOLE_DISK, &wholedisk);
	(void) nvlist_lookup_uint64(vdev, ZPOOL_CONFIG_OFFLINE, &offline);
	(void) nvlist_lookup_uint64(vdev, ZPOOL_CONFIG_GUID, &guid);

	if (offline)
	return; /* don't intervene if it was taken offline */

	is_dm = zfs_dev_is_dm(path);
	zed_log_msg(LOG_INFO, "zfs_process_add: pool '%s' vdev '%s', phys '%s'"
	" wholedisk %d, %s dm (guid %llu)", zpool_get_name(zhp), path,
	physpath ? physpath : "NULL", wholedisk, is_dm ? "is" : "not",
	(long long unsigned int)guid);

	/*
	* The VDEV guid is preferred for identification (gets passed in path)
	*/
	if (guid != 0) {
	(void) snprintf(fullpath, sizeof (fullpath), "%llu",
	(long long unsigned int)guid);
	} else {
	/*
	* otherwise use path sans partition suffix for whole disks
	*/
	(void) strlcpy(fullpath, path, sizeof (fullpath));
	if (wholedisk) {
	char *spath = zfs_strip_partition(fullpath);
	if (!spath) {
	zed_log_msg(LOG_INFO, "%s: Can't alloc",
	__func__);
	return;
	}

	(void) strlcpy(fullpath, spath, sizeof (fullpath));
	free(spath);
	}
	}

	/*
	* Attempt to online the device.
	*/
	if (zpool_vdev_online(zhp, fullpath,
	ZFS_ONLINE_CHECKREMOVE \| ZFS_ONLINE_UNSPARE, &newstate) == 0 &&
	(newstate == VDEV_STATE_HEALTHY \|\|
	newstate == VDEV_STATE_DEGRADED)) {
	zed_log_msg(LOG_INFO, " zpool_vdev_online: vdev %s is %s",
	fullpath, (newstate == VDEV_STATE_HEALTHY) ?
	"HEALTHY" : "DEGRADED");
	return;
	}

	/*
	* vdev_id alias rule for using scsi_debug devices (FMA automated
	* testing)
	*/
	if (physpath != NULL && strcmp("scsidebug", physpath) == 0)
	is_sd = B_TRUE;

	/*
	* If the pool doesn't have the autoreplace property set, then use
	* vdev online to trigger a FMA fault by posting an ereport.
	*/
	if (!zpool_get_prop_int(zhp, ZPOOL_PROP_AUTOREPLACE, NULL) \|\|
	!(wholedisk \|\| is_dm) \|\| (physpath == NULL)) {
	(void) zpool_vdev_online(zhp, fullpath, ZFS_ONLINE_FORCEFAULT,
	&newstate);
	zed_log_msg(LOG_INFO, "Pool's autoreplace is not enabled or "
	"not a whole disk for '%s'", fullpath);
	return;
	}

	/*
	* Convert physical path into its current device node. Rawpath
	* needs to be /dev/disk/by-vdev for a scsi_debug device since
	* /dev/disk/by-path will not be present.
	*/
	(void) snprintf(rawpath, sizeof (rawpath), "%s%s",
	is_sd ? DEV_BYVDEV_PATH : DEV_BYPATH_PATH, physpath);

	if (realpath(rawpath, devpath) == NULL && !is_dm) {
	zed_log_msg(LOG_INFO, " realpath: %s failed (%s)",
	rawpath, strerror(errno));

	(void) zpool_vdev_online(zhp, fullpath, ZFS_ONLINE_FORCEFAULT,
	&newstate);

	zed_log_msg(LOG_INFO, " zpool_vdev_online: %s FORCEFAULT (%s)",
	fullpath, libzfs_error_description(g_zfshdl));
	return;
	}

	/* Only autoreplace bad disks */
	if ((vs->vs_state != VDEV_STATE_DEGRADED) &&
	(vs->vs_state != VDEV_STATE_FAULTED) &&
	(vs->vs_state != VDEV_STATE_CANT_OPEN)) {
	return;
	}

	nvlist_lookup_string(vdev, "new_devid", &new_devid);

	if (is_dm) {
	/* Don't label device mapper or multipath disks. */
	} else if (!labeled) {
	/*
	* we're auto-replacing a raw disk, so label it first
	*/
	char *leafname;

	/*
	* If this is a request to label a whole disk, then attempt to
	* write out the label. Before we can label the disk, we need
	* to map the physical string that was matched on to the under
	* lying device node.
	*
	* If any part of this process fails, then do a force online
	* to trigger a ZFS fault for the device (and any hot spare
	* replacement).
	*/
	leafname = strrchr(devpath, '/') + 1;

	/*
	* If this is a request to label a whole disk, then attempt to
	* write out the label.
	*/
	if (zpool_label_disk(g_zfshdl, zhp, leafname) != 0) {
	zed_log_msg(LOG_INFO, " zpool_label_disk: could not "
	"label '%s' (%s)", leafname,
	libzfs_error_description(g_zfshdl));

	(void) zpool_vdev_online(zhp, fullpath,
	ZFS_ONLINE_FORCEFAULT, &newstate);
	return;
	}

	/*
	* The disk labeling is asynchronous on Linux. Just record
	* this label request and return as there will be another
	* disk add event for the partition after the labeling is
	* completed.
	*/
	device = malloc(sizeof (pendingdev_t));
	(void) strlcpy(device->pd_physpath, physpath,
	sizeof (device->pd_physpath));
	list_insert_tail(&g_device_list, device);

	zed_log_msg(LOG_INFO, " zpool_label_disk: async '%s' (%llu)",
	leafname, (u_longlong_t)guid);

	return; /* resumes at EC_DEV_ADD.ESC_DISK for partition */

	} else /* labeled */ {
	boolean_t found = B_FALSE;
	/*
	* match up with request above to label the disk
	*/
	for (device = list_head(&g_device_list); device != NULL;
	device = list_next(&g_device_list, device)) {
	if (strcmp(physpath, device->pd_physpath) == 0) {
	list_remove(&g_device_list, device);
	free(device);
	found = B_TRUE;
	break;
	}
	zed_log_msg(LOG_INFO, "zpool_label_disk: %s != %s",
	physpath, device->pd_physpath);
	}
	if (!found) {
	/* unexpected partition slice encountered */
	zed_log_msg(LOG_INFO, "labeled disk %s unexpected here",
	fullpath);
	(void) zpool_vdev_online(zhp, fullpath,
	ZFS_ONLINE_FORCEFAULT, &newstate);
	return;
	}

	zed_log_msg(LOG_INFO, " zpool_label_disk: resume '%s' (%llu)",
	physpath, (u_longlong_t)guid);

	(void) snprintf(devpath, sizeof (devpath), "%s%s",
	DEV_BYID_PATH, new_devid);
	}

	/*
	* Construct the root vdev to pass to zpool_vdev_attach(). While adding
	* the entire vdev structure is harmless, we construct a reduced set of
	* path/physpath/wholedisk to keep it simple.
	*/
	if (nvlist_alloc(&nvroot, NV_UNIQUE_NAME, 0) != 0) {
	zed_log_msg(LOG_WARNING, "zfs_mod: nvlist_alloc out of memory");
	return;
	}
	if (nvlist_alloc(&newvd, NV_UNIQUE_NAME, 0) != 0) {
	zed_log_msg(LOG_WARNING, "zfs_mod: nvlist_alloc out of memory");
	nvlist_free(nvroot);
	return;
	}

	if (nvlist_add_string(newvd, ZPOOL_CONFIG_TYPE, VDEV_TYPE_DISK) != 0 \|\|
	nvlist_add_string(newvd, ZPOOL_CONFIG_PATH, path) != 0 \|\|
	nvlist_add_string(newvd, ZPOOL_CONFIG_DEVID, new_devid) != 0 \|\|
	(physpath != NULL && nvlist_add_string(newvd,
	ZPOOL_CONFIG_PHYS_PATH, physpath) != 0) \|\|
	(enc_sysfs_path != NULL && nvlist_add_string(newvd,
	ZPOOL_CONFIG_VDEV_ENC_SYSFS_PATH, enc_sysfs_path) != 0) \|\|
	nvlist_add_uint64(newvd, ZPOOL_CONFIG_WHOLE_DISK, wholedisk) != 0 \|\|
	nvlist_add_string(nvroot, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT) != 0 \|\|
	- nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, &newvd,
	- 1) != 0) {
	+ nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	+ (const nvlist_t **)&newvd, 1) != 0) {
	zed_log_msg(LOG_WARNING, "zfs_mod: unable to add nvlist pairs");
	nvlist_free(newvd);
	nvlist_free(nvroot);
	return;
	}

	nvlist_free(newvd);

	/*
	* Wait for udev to verify the links exist, then auto-replace
	* the leaf disk at same physical location.
	*/
	if (zpool_label_disk_wait(path, 3000) != 0) {
	zed_log_msg(LOG_WARNING, "zfs_mod: expected replacement "
	"disk %s is missing", path);
	nvlist_free(nvroot);
	return;
	}

	/*
	* Prefer sequential resilvering when supported (mirrors and dRAID),
	* otherwise fallback to a traditional healing resilver.
	*/
	ret = zpool_vdev_attach(zhp, fullpath, path, nvroot, B_TRUE, B_TRUE);
	if (ret != 0) {
	ret = zpool_vdev_attach(zhp, fullpath, path, nvroot,
	B_TRUE, B_FALSE);
	}

	zed_log_msg(LOG_INFO, " zpool_vdev_replace: %s with %s (%s)",
	fullpath, path, (ret == 0) ? "no errors" :
	libzfs_error_description(g_zfshdl));

	nvlist_free(nvroot);
	}

	/*
	* Utility functions to find a vdev matching given criteria.
	*/
	typedef struct dev_data {
	const char *dd_compare;
	const char *dd_prop;
	zfs_process_func_t dd_func;
	boolean_t dd_found;
	boolean_t dd_islabeled;
	uint64_t dd_pool_guid;
	uint64_t dd_vdev_guid;
	const char *dd_new_devid;
	} dev_data_t;

	static void
	zfs_iter_vdev(zpool_handle_t zhp, nvlist_t nvl, void *data)
	{
	dev_data_t *dp = data;
	char *path = NULL;
	uint_t c, children;
	nvlist_t **child;

	/*
	* First iterate over any children.
	*/
	if (nvlist_lookup_nvlist_array(nvl, ZPOOL_CONFIG_CHILDREN,
	&child, &children) == 0) {
	for (c = 0; c < children; c++)
	zfs_iter_vdev(zhp, child[c], data);
	}

	/*
	* Iterate over any spares and cache devices
	*/
	if (nvlist_lookup_nvlist_array(nvl, ZPOOL_CONFIG_SPARES,
	&child, &children) == 0) {
	for (c = 0; c < children; c++)
	zfs_iter_vdev(zhp, child[c], data);
	}
	if (nvlist_lookup_nvlist_array(nvl, ZPOOL_CONFIG_L2CACHE,
	&child, &children) == 0) {
	for (c = 0; c < children; c++)
	zfs_iter_vdev(zhp, child[c], data);
	}

	/* once a vdev was matched and processed there is nothing left to do */
	if (dp->dd_found)
	return;

	/*
	* Match by GUID if available otherwise fallback to devid or physical
	*/
	if (dp->dd_vdev_guid != 0) {
	uint64_t guid;

	if (nvlist_lookup_uint64(nvl, ZPOOL_CONFIG_GUID,
	&guid) != 0 \|\| guid != dp->dd_vdev_guid) {
	return;
	}
	zed_log_msg(LOG_INFO, " zfs_iter_vdev: matched on %llu", guid);
	dp->dd_found = B_TRUE;

	} else if (dp->dd_compare != NULL) {
	/*
	* NOTE: On Linux there is an event for partition, so unlike
	* illumos, substring matching is not required to accommodate
	* the partition suffix. An exact match will be present in
	* the dp->dd_compare value.
	*/
	if (nvlist_lookup_string(nvl, dp->dd_prop, &path) != 0 \|\|
	strcmp(dp->dd_compare, path) != 0)
	return;

	zed_log_msg(LOG_INFO, " zfs_iter_vdev: matched %s on %s",
	dp->dd_prop, path);
	dp->dd_found = B_TRUE;

	/* pass the new devid for use by replacing code */
	if (dp->dd_new_devid != NULL) {
	(void) nvlist_add_string(nvl, "new_devid",
	dp->dd_new_devid);
	}
	}

	(dp->dd_func)(zhp, nvl, dp->dd_islabeled);
	}

	static void
	zfs_enable_ds(void *arg)
	{
	unavailpool_t pool = (unavailpool_t )arg;

	(void) zpool_enable_datasets(pool->uap_zhp, NULL, 0);
	zpool_close(pool->uap_zhp);
	free(pool);
	}

	static int
	zfs_iter_pool(zpool_handle_t zhp, void data)
	{
	nvlist_t config, nvl;
	dev_data_t *dp = data;
	uint64_t pool_guid;
	unavailpool_t *pool;

	zed_log_msg(LOG_INFO, "zfs_iter_pool: evaluating vdevs on %s (by %s)",
	zpool_get_name(zhp), dp->dd_vdev_guid ? "GUID" : dp->dd_prop);

	/*
	* For each vdev in this pool, look for a match to apply dd_func
	*/
	if ((config = zpool_get_config(zhp, NULL)) != NULL) {
	if (dp->dd_pool_guid == 0 \|\|
	(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
	&pool_guid) == 0 && pool_guid == dp->dd_pool_guid)) {
	(void) nvlist_lookup_nvlist(config,
	ZPOOL_CONFIG_VDEV_TREE, &nvl);
	zfs_iter_vdev(zhp, nvl, data);
	}
	}

	/*
	* if this pool was originally unavailable,
	* then enable its datasets asynchronously
	*/
	if (g_enumeration_done) {
	for (pool = list_head(&g_pool_list); pool != NULL;
	pool = list_next(&g_pool_list, pool)) {

	if (strcmp(zpool_get_name(zhp),
	zpool_get_name(pool->uap_zhp)))
	continue;
	if (zfs_toplevel_state(zhp) >= VDEV_STATE_DEGRADED) {
	list_remove(&g_pool_list, pool);
	(void) tpool_dispatch(g_tpool, zfs_enable_ds,
	pool);
	break;
	}
	}
	}

	zpool_close(zhp);
	return (dp->dd_found); /* cease iteration after a match */
	}

	/*
	* Given a physical device location, iterate over all
	* (pool, vdev) pairs which correspond to that location.
	*/
	static boolean_t
	devphys_iter(const char physical, const char devid, zfs_process_func_t func,
	boolean_t is_slice)
	{
	dev_data_t data = { 0 };

	data.dd_compare = physical;
	data.dd_func = func;
	data.dd_prop = ZPOOL_CONFIG_PHYS_PATH;
	data.dd_found = B_FALSE;
	data.dd_islabeled = is_slice;
	data.dd_new_devid = devid; /* used by auto replace code */

	(void) zpool_iter(g_zfshdl, zfs_iter_pool, &data);

	return (data.dd_found);
	}

	/*
	* Given a device identifier, find any vdevs with a matching devid.
	* On Linux we can match devid directly which is always a whole disk.
	*/
	static boolean_t
	devid_iter(const char *devid, zfs_process_func_t func, boolean_t is_slice)
	{
	dev_data_t data = { 0 };

	data.dd_compare = devid;
	data.dd_func = func;
	data.dd_prop = ZPOOL_CONFIG_DEVID;
	data.dd_found = B_FALSE;
	data.dd_islabeled = is_slice;
	data.dd_new_devid = devid;

	(void) zpool_iter(g_zfshdl, zfs_iter_pool, &data);

	return (data.dd_found);
	}

	/*
	* Given a device guid, find any vdevs with a matching guid.
	*/
	static boolean_t
	guid_iter(uint64_t pool_guid, uint64_t vdev_guid, const char *devid,
	zfs_process_func_t func, boolean_t is_slice)
	{
	dev_data_t data = { 0 };

	data.dd_func = func;
	data.dd_found = B_FALSE;
	data.dd_pool_guid = pool_guid;
	data.dd_vdev_guid = vdev_guid;
	data.dd_islabeled = is_slice;
	data.dd_new_devid = devid;

	(void) zpool_iter(g_zfshdl, zfs_iter_pool, &data);

	return (data.dd_found);
	}

	/*
	* Handle a EC_DEV_ADD.ESC_DISK event.
	*
	* illumos
	* Expects: DEV_PHYS_PATH string in schema
	* Matches: vdev's ZPOOL_CONFIG_PHYS_PATH or ZPOOL_CONFIG_DEVID
	*
	* path: '/dev/dsk/c0t1d0s0' (persistent)
	* devid: 'id1,sd@SATA_____Hitachi_HDS72101______JP2940HZ3H74MC/a'
	* phys_path: '/pci@0,0/pci103c,1609@11/disk@1,0:a'
	*
	* linux
	* provides: DEV_PHYS_PATH and DEV_IDENTIFIER strings in schema
	* Matches: vdev's ZPOOL_CONFIG_PHYS_PATH or ZPOOL_CONFIG_DEVID
	*
	* path: '/dev/sdc1' (not persistent)
	* devid: 'ata-SAMSUNG_HD204UI_S2HGJD2Z805891-part1'
	* phys_path: 'pci-0000:04:00.0-sas-0x4433221106000000-lun-0'
	*/
	static int
	zfs_deliver_add(nvlist_t *nvl, boolean_t is_lofi)
	{
	char devpath = NULL, devid;
	uint64_t pool_guid = 0, vdev_guid = 0;
	boolean_t is_slice;

	/*
	* Expecting a devid string and an optional physical location and guid
	*/
	if (nvlist_lookup_string(nvl, DEV_IDENTIFIER, &devid) != 0)
	return (-1);

	(void) nvlist_lookup_string(nvl, DEV_PHYS_PATH, &devpath);
	(void) nvlist_lookup_uint64(nvl, ZFS_EV_POOL_GUID, &pool_guid);
	(void) nvlist_lookup_uint64(nvl, ZFS_EV_VDEV_GUID, &vdev_guid);

	is_slice = (nvlist_lookup_boolean(nvl, DEV_IS_PART) == 0);

	zed_log_msg(LOG_INFO, "zfs_deliver_add: adding %s (%s) (is_slice %d)",
	devid, devpath ? devpath : "NULL", is_slice);

	/*
	* Iterate over all vdevs looking for a match in the following order:
	* 1. ZPOOL_CONFIG_DEVID (identifies the unique disk)
	* 2. ZPOOL_CONFIG_PHYS_PATH (identifies disk physical location).
	* 3. ZPOOL_CONFIG_GUID (identifies unique vdev).
	*/
	if (devid_iter(devid, zfs_process_add, is_slice))
	return (0);
	if (devpath != NULL && devphys_iter(devpath, devid, zfs_process_add,
	is_slice))
	return (0);
	if (vdev_guid != 0)
	(void) guid_iter(pool_guid, vdev_guid, devid, zfs_process_add,
	is_slice);

	return (0);
	}

	/*
	* Called when we receive a VDEV_CHECK event, which indicates a device could not
	* be opened during initial pool open, but the autoreplace property was set on
	* the pool. In this case, we treat it as if it were an add event.
	*/
	static int
	zfs_deliver_check(nvlist_t *nvl)
	{
	dev_data_t data = { 0 };

	if (nvlist_lookup_uint64(nvl, ZFS_EV_POOL_GUID,
	&data.dd_pool_guid) != 0 \|\|
	nvlist_lookup_uint64(nvl, ZFS_EV_VDEV_GUID,
	&data.dd_vdev_guid) != 0 \|\|
	data.dd_vdev_guid == 0)
	return (0);

	zed_log_msg(LOG_INFO, "zfs_deliver_check: pool '%llu', vdev %llu",
	data.dd_pool_guid, data.dd_vdev_guid);

	data.dd_func = zfs_process_add;

	(void) zpool_iter(g_zfshdl, zfs_iter_pool, &data);

	return (0);
	}

	static int
	zfsdle_vdev_online(zpool_handle_t zhp, void data)
	{
	char *devname = data;
	boolean_t avail_spare, l2cache;
	nvlist_t *tgt;
	int error;

	zed_log_msg(LOG_INFO, "zfsdle_vdev_online: searching for '%s' in '%s'",
	devname, zpool_get_name(zhp));

	if ((tgt = zpool_find_vdev_by_physpath(zhp, devname,
	&avail_spare, &l2cache, NULL)) != NULL) {
	char *path, fullpath[MAXPATHLEN];
	uint64_t wholedisk;

	error = nvlist_lookup_string(tgt, ZPOOL_CONFIG_PATH, &path);
	if (error) {
	zpool_close(zhp);
	return (0);
	}

	error = nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_WHOLE_DISK,
	&wholedisk);
	if (error)
	wholedisk = 0;

	if (wholedisk) {
	path = strrchr(path, '/');
	if (path != NULL) {
	path = zfs_strip_partition(path + 1);
	if (path == NULL) {
	zpool_close(zhp);
	return (0);
	}
	} else {
	zpool_close(zhp);
	return (0);
	}

	(void) strlcpy(fullpath, path, sizeof (fullpath));
	free(path);

	/*
	* We need to reopen the pool associated with this
	* device so that the kernel can update the size of
	* the expanded device. When expanding there is no
	* need to restart the scrub from the beginning.
	*/
	boolean_t scrub_restart = B_FALSE;
	(void) zpool_reopen_one(zhp, &scrub_restart);
	} else {
	(void) strlcpy(fullpath, path, sizeof (fullpath));
	}

	if (zpool_get_prop_int(zhp, ZPOOL_PROP_AUTOEXPAND, NULL)) {
	vdev_state_t newstate;

	if (zpool_get_state(zhp) != POOL_STATE_UNAVAIL) {
	error = zpool_vdev_online(zhp, fullpath, 0,
	&newstate);
	zed_log_msg(LOG_INFO, "zfsdle_vdev_online: "
	"setting device '%s' to ONLINE state "
	"in pool '%s': %d", fullpath,
	zpool_get_name(zhp), error);
	}
	}
	zpool_close(zhp);
	return (1);
	}
	zpool_close(zhp);
	return (0);
	}

	/*
	* This function handles the ESC_DEV_DLE device change event. Use the
	* provided vdev guid when looking up a disk or partition, when the guid
	* is not present assume the entire disk is owned by ZFS and append the
	* expected -part1 partition information then lookup by physical path.
	*/
	static int
	zfs_deliver_dle(nvlist_t *nvl)
	{
	char *devname, name[MAXPATHLEN];
	uint64_t guid;

	if (nvlist_lookup_uint64(nvl, ZFS_EV_VDEV_GUID, &guid) == 0) {
	sprintf(name, "%llu", (u_longlong_t)guid);
	} else if (nvlist_lookup_string(nvl, DEV_PHYS_PATH, &devname) == 0) {
	strlcpy(name, devname, MAXPATHLEN);
	zfs_append_partition(name, MAXPATHLEN);
	} else {
	zed_log_msg(LOG_INFO, "zfs_deliver_dle: no guid or physpath");
	}

	if (zpool_iter(g_zfshdl, zfsdle_vdev_online, name) != 1) {
	zed_log_msg(LOG_INFO, "zfs_deliver_dle: device '%s' not "
	"found", name);
	return (1);
	}

	return (0);
	}

	/*
	* syseventd daemon module event handler
	*
	* Handles syseventd daemon zfs device related events:
	*
	* EC_DEV_ADD.ESC_DISK
	* EC_DEV_STATUS.ESC_DEV_DLE
	* EC_ZFS.ESC_ZFS_VDEV_CHECK
	*
	* Note: assumes only one thread active at a time (not thread safe)
	*/
	static int
	zfs_slm_deliver_event(const char class, const char subclass, nvlist_t *nvl)
	{
	int ret;
	boolean_t is_lofi = B_FALSE, is_check = B_FALSE, is_dle = B_FALSE;

	if (strcmp(class, EC_DEV_ADD) == 0) {
	/*
	* We're mainly interested in disk additions, but we also listen
	* for new loop devices, to allow for simplified testing.
	*/
	if (strcmp(subclass, ESC_DISK) == 0)
	is_lofi = B_FALSE;
	else if (strcmp(subclass, ESC_LOFI) == 0)
	is_lofi = B_TRUE;
	else
	return (0);

	is_check = B_FALSE;
	} else if (strcmp(class, EC_ZFS) == 0 &&
	strcmp(subclass, ESC_ZFS_VDEV_CHECK) == 0) {
	/*
	* This event signifies that a device failed to open
	* during pool load, but the 'autoreplace' property was
	* set, so we should pretend it's just been added.
	*/
	is_check = B_TRUE;
	} else if (strcmp(class, EC_DEV_STATUS) == 0 &&
	strcmp(subclass, ESC_DEV_DLE) == 0) {
	is_dle = B_TRUE;
	} else {
	return (0);
	}

	if (is_dle)
	ret = zfs_deliver_dle(nvl);
	else if (is_check)
	ret = zfs_deliver_check(nvl);
	else
	ret = zfs_deliver_add(nvl, is_lofi);

	return (ret);
	}

	/ARGSUSED/
	static void *
	zfs_enum_pools(void *arg)
	{
	(void) zpool_iter(g_zfshdl, zfs_unavail_pool, (void *)&g_pool_list);
	/*
	* Linux - instead of using a thread pool, each list entry
	* will spawn a thread when an unavailable pool transitions
	* to available. zfs_slm_fini will wait for these threads.
	*/
	g_enumeration_done = B_TRUE;
	return (NULL);
	}

	/*
	* called from zed daemon at startup
	*
	* sent messages from zevents or udev monitor
	*
	* For now, each agent has its own libzfs instance
	*/
	int
	zfs_slm_init()
	{
	if ((g_zfshdl = libzfs_init()) == NULL)
	return (-1);

	/*
	* collect a list of unavailable pools (asynchronously,
	* since this can take a while)
	*/
	list_create(&g_pool_list, sizeof (struct unavailpool),
	offsetof(struct unavailpool, uap_node));

	if (pthread_create(&g_zfs_tid, NULL, zfs_enum_pools, NULL) != 0) {
	list_destroy(&g_pool_list);
	libzfs_fini(g_zfshdl);
	return (-1);
	}

	pthread_setname_np(g_zfs_tid, "enum-pools");
	list_create(&g_device_list, sizeof (struct pendingdev),
	offsetof(struct pendingdev, pd_node));

	return (0);
	}

	void
	zfs_slm_fini()
	{
	unavailpool_t *pool;
	pendingdev_t *device;

	/* wait for zfs_enum_pools thread to complete */
	(void) pthread_join(g_zfs_tid, NULL);
	/* destroy the thread pool */
	if (g_tpool != NULL) {
	tpool_wait(g_tpool);
	tpool_destroy(g_tpool);
	}

	while ((pool = (list_head(&g_pool_list))) != NULL) {
	list_remove(&g_pool_list, pool);
	zpool_close(pool->uap_zhp);
	free(pool);
	}
	list_destroy(&g_pool_list);

	while ((device = (list_head(&g_device_list))) != NULL) {
	list_remove(&g_device_list, device);
	free(device);
	}
	list_destroy(&g_device_list);

	libzfs_fini(g_zfshdl);
	}

	void
	zfs_slm_event(const char class, const char subclass, nvlist_t *nvl)
	{
	zed_log_msg(LOG_INFO, "zfs_slm_event: %s.%s", class, subclass);
	(void) zfs_slm_deliver_event(class, subclass, nvl);
	}
	diff --git a/sys/contrib/openzfs/cmd/zed/agents/zfs_retire.c b/sys/contrib/openzfs/cmd/zed/agents/zfs_retire.c
	index 6c009bdc1235..5fb7b4e84c5e 100644
	--- a/sys/contrib/openzfs/cmd/zed/agents/zfs_retire.c
	+++ b/sys/contrib/openzfs/cmd/zed/agents/zfs_retire.c
	@@ -1,566 +1,566 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2006, 2010, Oracle and/or its affiliates. All rights reserved.
	*
	* Copyright (c) 2016, Intel Corporation.
	* Copyright (c) 2018, loli10K <ezomori.nozomu@gmail.com>
	*/

	/*
	* The ZFS retire agent is responsible for managing hot spares across all pools.
	* When we see a device fault or a device removal, we try to open the associated
	* pool and look for any hot spares. We iterate over any available hot spares
	* and attempt a 'zpool replace' for each one.
	*
	* For vdevs diagnosed as faulty, the agent is also responsible for proactively
	* marking the vdev FAULTY (for I/O errors) or DEGRADED (for checksum errors).
	*/

	#include <sys/fs/zfs.h>
	#include <sys/fm/protocol.h>
	#include <sys/fm/fs/zfs.h>
	#include <libzutil.h>
	#include <libzfs.h>
	#include <string.h>
	#include <libgen.h>

	#include "zfs_agents.h"
	#include "fmd_api.h"


	typedef struct zfs_retire_repaired {
	struct zfs_retire_repaired *zrr_next;
	uint64_t zrr_pool;
	uint64_t zrr_vdev;
	} zfs_retire_repaired_t;

	typedef struct zfs_retire_data {
	libzfs_handle_t *zrd_hdl;
	zfs_retire_repaired_t *zrd_repaired;
	} zfs_retire_data_t;

	static void
	zfs_retire_clear_data(fmd_hdl_t hdl, zfs_retire_data_t zdp)
	{
	zfs_retire_repaired_t *zrp;

	while ((zrp = zdp->zrd_repaired) != NULL) {
	zdp->zrd_repaired = zrp->zrr_next;
	fmd_hdl_free(hdl, zrp, sizeof (zfs_retire_repaired_t));
	}
	}

	/*
	* Find a pool with a matching GUID.
	*/
	typedef struct find_cbdata {
	uint64_t cb_guid;
	zpool_handle_t *cb_zhp;
	nvlist_t *cb_vdev;
	} find_cbdata_t;

	static int
	find_pool(zpool_handle_t zhp, void data)
	{
	find_cbdata_t *cbp = data;

	if (cbp->cb_guid ==
	zpool_get_prop_int(zhp, ZPOOL_PROP_GUID, NULL)) {
	cbp->cb_zhp = zhp;
	return (1);
	}

	zpool_close(zhp);
	return (0);
	}

	/*
	* Find a vdev within a tree with a matching GUID.
	*/
	static nvlist_t *
	find_vdev(libzfs_handle_t zhdl, nvlist_t nv, uint64_t search_guid)
	{
	uint64_t guid;
	nvlist_t **child;
	uint_t c, children;
	nvlist_t *ret;

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) == 0 &&
	guid == search_guid) {
	fmd_hdl_debug(fmd_module_hdl("zfs-retire"),
	"matched vdev %llu", guid);
	return (nv);
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0)
	return (NULL);

	for (c = 0; c < children; c++) {
	if ((ret = find_vdev(zhdl, child[c], search_guid)) != NULL)
	return (ret);
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE,
	&child, &children) != 0)
	return (NULL);

	for (c = 0; c < children; c++) {
	if ((ret = find_vdev(zhdl, child[c], search_guid)) != NULL)
	return (ret);
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES,
	&child, &children) != 0)
	return (NULL);

	for (c = 0; c < children; c++) {
	if ((ret = find_vdev(zhdl, child[c], search_guid)) != NULL)
	return (ret);
	}

	return (NULL);
	}

	/*
	* Given a (pool, vdev) GUID pair, find the matching pool and vdev.
	*/
	static zpool_handle_t *
	find_by_guid(libzfs_handle_t *zhdl, uint64_t pool_guid, uint64_t vdev_guid,
	nvlist_t **vdevp)
	{
	find_cbdata_t cb;
	zpool_handle_t *zhp;
	nvlist_t config, nvroot;

	/*
	* Find the corresponding pool and make sure the vdev still exists.
	*/
	cb.cb_guid = pool_guid;
	if (zpool_iter(zhdl, find_pool, &cb) != 1)
	return (NULL);

	zhp = cb.cb_zhp;
	config = zpool_get_config(zhp, NULL);
	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) != 0) {
	zpool_close(zhp);
	return (NULL);
	}

	if (vdev_guid != 0) {
	if ((*vdevp = find_vdev(zhdl, nvroot, vdev_guid)) == NULL) {
	zpool_close(zhp);
	return (NULL);
	}
	}

	return (zhp);
	}

	/*
	* Given a vdev, attempt to replace it with every known spare until one
	* succeeds or we run out of devices to try.
	* Return whether we were successful or not in replacing the device.
	*/
	static boolean_t
	replace_with_spare(fmd_hdl_t hdl, zpool_handle_t zhp, nvlist_t *vdev)
	{
	nvlist_t config, nvroot, *replacement;
	nvlist_t **spares;
	uint_t s, nspares;
	char *dev_name;
	zprop_source_t source;
	int ashift;

	config = zpool_get_config(zhp, NULL);
	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) != 0)
	return (B_FALSE);

	/*
	* Find out if there are any hot spares available in the pool.
	*/
	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	&spares, &nspares) != 0)
	return (B_FALSE);

	/*
	* lookup "ashift" pool property, we may need it for the replacement
	*/
	ashift = zpool_get_prop_int(zhp, ZPOOL_PROP_ASHIFT, &source);

	replacement = fmd_nvl_alloc(hdl, FMD_SLEEP);

	(void) nvlist_add_string(replacement, ZPOOL_CONFIG_TYPE,
	VDEV_TYPE_ROOT);

	dev_name = zpool_vdev_name(NULL, zhp, vdev, B_FALSE);

	/*
	* Try to replace each spare, ending when we successfully
	* replace it.
	*/
	for (s = 0; s < nspares; s++) {
	boolean_t rebuild = B_FALSE;
	char spare_name, type;

	if (nvlist_lookup_string(spares[s], ZPOOL_CONFIG_PATH,
	&spare_name) != 0)
	continue;

	/* prefer sequential resilvering for distributed spares */
	if ((nvlist_lookup_string(spares[s], ZPOOL_CONFIG_TYPE,
	&type) == 0) && strcmp(type, VDEV_TYPE_DRAID_SPARE) == 0)
	rebuild = B_TRUE;

	/* if set, add the "ashift" pool property to the spare nvlist */
	if (source != ZPROP_SRC_DEFAULT)
	(void) nvlist_add_uint64(spares[s],
	ZPOOL_CONFIG_ASHIFT, ashift);

	(void) nvlist_add_nvlist_array(replacement,
	- ZPOOL_CONFIG_CHILDREN, &spares[s], 1);
	+ ZPOOL_CONFIG_CHILDREN, (const nvlist_t **)&spares[s], 1);

	fmd_hdl_debug(hdl, "zpool_vdev_replace '%s' with spare '%s'",
	dev_name, zfs_basename(spare_name));

	if (zpool_vdev_attach(zhp, dev_name, spare_name,
	replacement, B_TRUE, rebuild) == 0) {
	free(dev_name);
	nvlist_free(replacement);
	return (B_TRUE);
	}
	}

	free(dev_name);
	nvlist_free(replacement);

	return (B_FALSE);
	}

	/*
	* Repair this vdev if we had diagnosed a 'fault.fs.zfs.device' and
	* ASRU is now usable. ZFS has found the device to be present and
	* functioning.
	*/
	/ARGSUSED/
	static void
	zfs_vdev_repair(fmd_hdl_t hdl, nvlist_t nvl)
	{
	zfs_retire_data_t *zdp = fmd_hdl_getspecific(hdl);
	zfs_retire_repaired_t *zrp;
	uint64_t pool_guid, vdev_guid;
	if (nvlist_lookup_uint64(nvl, FM_EREPORT_PAYLOAD_ZFS_POOL_GUID,
	&pool_guid) != 0 \|\| nvlist_lookup_uint64(nvl,
	FM_EREPORT_PAYLOAD_ZFS_VDEV_GUID, &vdev_guid) != 0)
	return;

	/*
	* Before checking the state of the ASRU, go through and see if we've
	* already made an attempt to repair this ASRU. This list is cleared
	* whenever we receive any kind of list event, and is designed to
	* prevent us from generating a feedback loop when we attempt repairs
	* against a faulted pool. The problem is that checking the unusable
	* state of the ASRU can involve opening the pool, which can post
	* statechange events but otherwise leave the pool in the faulted
	* state. This list allows us to detect when a statechange event is
	* due to our own request.
	*/
	for (zrp = zdp->zrd_repaired; zrp != NULL; zrp = zrp->zrr_next) {
	if (zrp->zrr_pool == pool_guid &&
	zrp->zrr_vdev == vdev_guid)
	return;
	}

	zrp = fmd_hdl_alloc(hdl, sizeof (zfs_retire_repaired_t), FMD_SLEEP);
	zrp->zrr_next = zdp->zrd_repaired;
	zrp->zrr_pool = pool_guid;
	zrp->zrr_vdev = vdev_guid;
	zdp->zrd_repaired = zrp;

	fmd_hdl_debug(hdl, "marking repaired vdev %llu on pool %llu",
	vdev_guid, pool_guid);
	}

	/ARGSUSED/
	static void
	zfs_retire_recv(fmd_hdl_t hdl, fmd_event_t ep, nvlist_t *nvl,
	const char *class)
	{
	uint64_t pool_guid, vdev_guid;
	zpool_handle_t *zhp;
	nvlist_t resource, fault;
	nvlist_t **faults;
	uint_t f, nfaults;
	zfs_retire_data_t *zdp = fmd_hdl_getspecific(hdl);
	libzfs_handle_t *zhdl = zdp->zrd_hdl;
	boolean_t fault_device, degrade_device;
	boolean_t is_repair;
	char *scheme;
	nvlist_t *vdev = NULL;
	char *uuid;
	int repair_done = 0;
	boolean_t retire;
	boolean_t is_disk;
	vdev_aux_t aux;
	uint64_t state = 0;

	fmd_hdl_debug(hdl, "zfs_retire_recv: '%s'", class);

	nvlist_lookup_uint64(nvl, FM_EREPORT_PAYLOAD_ZFS_VDEV_STATE, &state);

	/*
	* If this is a resource notifying us of device removal then simply
	* check for an available spare and continue unless the device is a
	* l2arc vdev, in which case we just offline it.
	*/
	if (strcmp(class, "resource.fs.zfs.removed") == 0 \|\|
	(strcmp(class, "resource.fs.zfs.statechange") == 0 &&
	(state == VDEV_STATE_REMOVED \|\| state == VDEV_STATE_FAULTED))) {
	char *devtype;
	char *devname;

	if (nvlist_lookup_uint64(nvl, FM_EREPORT_PAYLOAD_ZFS_POOL_GUID,
	&pool_guid) != 0 \|\|
	nvlist_lookup_uint64(nvl, FM_EREPORT_PAYLOAD_ZFS_VDEV_GUID,
	&vdev_guid) != 0)
	return;

	if ((zhp = find_by_guid(zhdl, pool_guid, vdev_guid,
	&vdev)) == NULL)
	return;

	devname = zpool_vdev_name(NULL, zhp, vdev, B_FALSE);

	/* Can't replace l2arc with a spare: offline the device */
	if (nvlist_lookup_string(nvl, FM_EREPORT_PAYLOAD_ZFS_VDEV_TYPE,
	&devtype) == 0 && strcmp(devtype, VDEV_TYPE_L2CACHE) == 0) {
	fmd_hdl_debug(hdl, "zpool_vdev_offline '%s'", devname);
	zpool_vdev_offline(zhp, devname, B_TRUE);
	} else if (!fmd_prop_get_int32(hdl, "spare_on_remove") \|\|
	replace_with_spare(hdl, zhp, vdev) == B_FALSE) {
	/* Could not handle with spare */
	fmd_hdl_debug(hdl, "no spare for '%s'", devname);
	}

	free(devname);
	zpool_close(zhp);
	return;
	}

	if (strcmp(class, FM_LIST_RESOLVED_CLASS) == 0)
	return;

	/*
	* Note: on Linux statechange events are more than just
	* healthy ones so we need to confirm the actual state value.
	*/
	if (strcmp(class, "resource.fs.zfs.statechange") == 0 &&
	state == VDEV_STATE_HEALTHY) {
	zfs_vdev_repair(hdl, nvl);
	return;
	}
	if (strcmp(class, "sysevent.fs.zfs.vdev_remove") == 0) {
	zfs_vdev_repair(hdl, nvl);
	return;
	}

	zfs_retire_clear_data(hdl, zdp);

	if (strcmp(class, FM_LIST_REPAIRED_CLASS) == 0)
	is_repair = B_TRUE;
	else
	is_repair = B_FALSE;

	/*
	* We subscribe to zfs faults as well as all repair events.
	*/
	if (nvlist_lookup_nvlist_array(nvl, FM_SUSPECT_FAULT_LIST,
	&faults, &nfaults) != 0)
	return;

	for (f = 0; f < nfaults; f++) {
	fault = faults[f];

	fault_device = B_FALSE;
	degrade_device = B_FALSE;
	is_disk = B_FALSE;

	if (nvlist_lookup_boolean_value(fault, FM_SUSPECT_RETIRE,
	&retire) == 0 && retire == 0)
	continue;

	/*
	* While we subscribe to fault.fs.zfs.*, we only take action
	* for faults targeting a specific vdev (open failure or SERD
	* failure). We also subscribe to fault.io.* events, so that
	* faulty disks will be faulted in the ZFS configuration.
	*/
	if (fmd_nvl_class_match(hdl, fault, "fault.fs.zfs.vdev.io")) {
	fault_device = B_TRUE;
	} else if (fmd_nvl_class_match(hdl, fault,
	"fault.fs.zfs.vdev.checksum")) {
	degrade_device = B_TRUE;
	} else if (fmd_nvl_class_match(hdl, fault,
	"fault.fs.zfs.device")) {
	fault_device = B_FALSE;
	} else if (fmd_nvl_class_match(hdl, fault, "fault.io.*")) {
	is_disk = B_TRUE;
	fault_device = B_TRUE;
	} else {
	continue;
	}

	if (is_disk) {
	continue;
	} else {
	/*
	* This is a ZFS fault. Lookup the resource, and
	* attempt to find the matching vdev.
	*/
	if (nvlist_lookup_nvlist(fault, FM_FAULT_RESOURCE,
	&resource) != 0 \|\|
	nvlist_lookup_string(resource, FM_FMRI_SCHEME,
	&scheme) != 0)
	continue;

	if (strcmp(scheme, FM_FMRI_SCHEME_ZFS) != 0)
	continue;

	if (nvlist_lookup_uint64(resource, FM_FMRI_ZFS_POOL,
	&pool_guid) != 0)
	continue;

	if (nvlist_lookup_uint64(resource, FM_FMRI_ZFS_VDEV,
	&vdev_guid) != 0) {
	if (is_repair)
	vdev_guid = 0;
	else
	continue;
	}

	if ((zhp = find_by_guid(zhdl, pool_guid, vdev_guid,
	&vdev)) == NULL)
	continue;

	aux = VDEV_AUX_ERR_EXCEEDED;
	}

	if (vdev_guid == 0) {
	/*
	* For pool-level repair events, clear the entire pool.
	*/
	fmd_hdl_debug(hdl, "zpool_clear of pool '%s'",
	zpool_get_name(zhp));
	(void) zpool_clear(zhp, NULL, NULL);
	zpool_close(zhp);
	continue;
	}

	/*
	* If this is a repair event, then mark the vdev as repaired and
	* continue.
	*/
	if (is_repair) {
	repair_done = 1;
	fmd_hdl_debug(hdl, "zpool_clear of pool '%s' vdev %llu",
	zpool_get_name(zhp), vdev_guid);
	(void) zpool_vdev_clear(zhp, vdev_guid);
	zpool_close(zhp);
	continue;
	}

	/*
	* Actively fault the device if needed.
	*/
	if (fault_device)
	(void) zpool_vdev_fault(zhp, vdev_guid, aux);
	if (degrade_device)
	(void) zpool_vdev_degrade(zhp, vdev_guid, aux);

	if (fault_device \|\| degrade_device)
	fmd_hdl_debug(hdl, "zpool_vdev_%s: vdev %llu on '%s'",
	fault_device ? "fault" : "degrade", vdev_guid,
	zpool_get_name(zhp));

	/*
	* Attempt to substitute a hot spare.
	*/
	(void) replace_with_spare(hdl, zhp, vdev);

	zpool_close(zhp);
	}

	if (strcmp(class, FM_LIST_REPAIRED_CLASS) == 0 && repair_done &&
	nvlist_lookup_string(nvl, FM_SUSPECT_UUID, &uuid) == 0)
	fmd_case_uuresolved(hdl, uuid);
	}

	static const fmd_hdl_ops_t fmd_ops = {
	zfs_retire_recv, /* fmdo_recv */
	NULL, /* fmdo_timeout */
	NULL, /* fmdo_close */
	NULL, /* fmdo_stats */
	NULL, /* fmdo_gc */
	};

	static const fmd_prop_t fmd_props[] = {
	{ "spare_on_remove", FMD_TYPE_BOOL, "true" },
	{ NULL, 0, NULL }
	};

	static const fmd_hdl_info_t fmd_info = {
	"ZFS Retire Agent", "1.0", &fmd_ops, fmd_props
	};

	void
	_zfs_retire_init(fmd_hdl_t *hdl)
	{
	zfs_retire_data_t *zdp;
	libzfs_handle_t *zhdl;

	if ((zhdl = libzfs_init()) == NULL)
	return;

	if (fmd_hdl_register(hdl, FMD_API_VERSION, &fmd_info) != 0) {
	libzfs_fini(zhdl);
	return;
	}

	zdp = fmd_hdl_zalloc(hdl, sizeof (zfs_retire_data_t), FMD_SLEEP);
	zdp->zrd_hdl = zhdl;

	fmd_hdl_setspecific(hdl, zdp);
	}

	void
	_zfs_retire_fini(fmd_hdl_t *hdl)
	{
	zfs_retire_data_t *zdp = fmd_hdl_getspecific(hdl);

	if (zdp != NULL) {
	zfs_retire_clear_data(hdl, zdp);
	libzfs_fini(zdp->zrd_hdl);
	fmd_hdl_free(hdl, zdp, sizeof (zfs_retire_data_t));
	}
	}
	diff --git a/sys/contrib/openzfs/cmd/zfs/zfs_main.c b/sys/contrib/openzfs/cmd/zfs/zfs_main.c
	index d05cb29c69d6..5aa2508c382f 100644
	--- a/sys/contrib/openzfs/cmd/zfs/zfs_main.c
	+++ b/sys/contrib/openzfs/cmd/zfs/zfs_main.c
	@@ -1,8793 +1,8794 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright 2012 Milan Jurik. All rights reserved.
	* Copyright (c) 2012, Joyent, Inc. All rights reserved.
	* Copyright (c) 2013 Steven Hartland. All rights reserved.
	* Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>.
	* Copyright 2016 Nexenta Systems, Inc.
	* Copyright (c) 2019 Datto Inc.
	* Copyright (c) 2019, loli10K <ezomori.nozomu@gmail.com>
	* Copyright 2019 Joyent, Inc.
	* Copyright (c) 2019, 2020 by Christian Schwarz. All rights reserved.
	*/

	#include <assert.h>
	#include <ctype.h>
	#include <sys/debug.h>
	#include <errno.h>
	#include <getopt.h>
	#include <libgen.h>
	#include <libintl.h>
	#include <libuutil.h>
	#include <libnvpair.h>
	#include <locale.h>
	#include <stddef.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <strings.h>
	#include <unistd.h>
	#include <fcntl.h>
	#include <zone.h>
	#include <grp.h>
	#include <pwd.h>
	#include <umem.h>
	#include <pthread.h>
	#include <signal.h>
	#include <sys/list.h>
	#include <sys/mkdev.h>
	#include <sys/mntent.h>
	#include <sys/mnttab.h>
	#include <sys/mount.h>
	#include <sys/stat.h>
	#include <sys/fs/zfs.h>
	#include <sys/systeminfo.h>
	#include <sys/types.h>
	#include <time.h>
	#include <sys/zfs_project.h>

	#include <libzfs.h>
	#include <libzfs_core.h>
	#include <zfs_prop.h>
	#include <zfs_deleg.h>
	#include <libzutil.h>
	#ifdef HAVE_IDMAP
	#include <aclutils.h>
	#include <directory.h>
	#endif /* HAVE_IDMAP */

	#include "zfs_iter.h"
	#include "zfs_util.h"
	#include "zfs_comutil.h"
	#include "zfs_projectutil.h"

	libzfs_handle_t *g_zfs;

	static char history_str[HIS_MAX_RECORD_LEN];
	static boolean_t log_history = B_TRUE;

	static int zfs_do_clone(int argc, char **argv);
	static int zfs_do_create(int argc, char **argv);
	static int zfs_do_destroy(int argc, char **argv);
	static int zfs_do_get(int argc, char **argv);
	static int zfs_do_inherit(int argc, char **argv);
	static int zfs_do_list(int argc, char **argv);
	static int zfs_do_mount(int argc, char **argv);
	static int zfs_do_rename(int argc, char **argv);
	static int zfs_do_rollback(int argc, char **argv);
	static int zfs_do_set(int argc, char **argv);
	static int zfs_do_upgrade(int argc, char **argv);
	static int zfs_do_snapshot(int argc, char **argv);
	static int zfs_do_unmount(int argc, char **argv);
	static int zfs_do_share(int argc, char **argv);
	static int zfs_do_unshare(int argc, char **argv);
	static int zfs_do_send(int argc, char **argv);
	static int zfs_do_receive(int argc, char **argv);
	static int zfs_do_promote(int argc, char **argv);
	static int zfs_do_userspace(int argc, char **argv);
	static int zfs_do_allow(int argc, char **argv);
	static int zfs_do_unallow(int argc, char **argv);
	static int zfs_do_hold(int argc, char **argv);
	static int zfs_do_holds(int argc, char **argv);
	static int zfs_do_release(int argc, char **argv);
	static int zfs_do_diff(int argc, char **argv);
	static int zfs_do_bookmark(int argc, char **argv);
	static int zfs_do_channel_program(int argc, char **argv);
	static int zfs_do_load_key(int argc, char **argv);
	static int zfs_do_unload_key(int argc, char **argv);
	static int zfs_do_change_key(int argc, char **argv);
	static int zfs_do_project(int argc, char **argv);
	static int zfs_do_version(int argc, char **argv);
	static int zfs_do_redact(int argc, char **argv);
	static int zfs_do_wait(int argc, char **argv);

	#ifdef __FreeBSD__
	static int zfs_do_jail(int argc, char **argv);
	static int zfs_do_unjail(int argc, char **argv);
	#endif

	/*
	* Enable a reasonable set of defaults for libumem debugging on DEBUG builds.
	*/

	#ifdef DEBUG
	const char *
	_umem_debug_init(void)
	{
	return ("default,verbose"); /* $UMEM_DEBUG setting */
	}

	const char *
	_umem_logging_init(void)
	{
	return ("fail,contents"); /* $UMEM_LOGGING setting */
	}
	#endif

	typedef enum {
	HELP_CLONE,
	HELP_CREATE,
	HELP_DESTROY,
	HELP_GET,
	HELP_INHERIT,
	HELP_UPGRADE,
	HELP_LIST,
	HELP_MOUNT,
	HELP_PROMOTE,
	HELP_RECEIVE,
	HELP_RENAME,
	HELP_ROLLBACK,
	HELP_SEND,
	HELP_SET,
	HELP_SHARE,
	HELP_SNAPSHOT,
	HELP_UNMOUNT,
	HELP_UNSHARE,
	HELP_ALLOW,
	HELP_UNALLOW,
	HELP_USERSPACE,
	HELP_GROUPSPACE,
	HELP_PROJECTSPACE,
	HELP_PROJECT,
	HELP_HOLD,
	HELP_HOLDS,
	HELP_RELEASE,
	HELP_DIFF,
	HELP_BOOKMARK,
	HELP_CHANNEL_PROGRAM,
	HELP_LOAD_KEY,
	HELP_UNLOAD_KEY,
	HELP_CHANGE_KEY,
	HELP_VERSION,
	HELP_REDACT,
	HELP_JAIL,
	HELP_UNJAIL,
	HELP_WAIT,
	} zfs_help_t;

	typedef struct zfs_command {
	const char *name;
	int (func)(int argc, char *argv);
	zfs_help_t usage;
	} zfs_command_t;

	/*
	* Master command table. Each ZFS command has a name, associated function, and
	* usage message. The usage messages need to be internationalized, so we have
	* to have a function to return the usage message based on a command index.
	*
	* These commands are organized according to how they are displayed in the usage
	* message. An empty command (one with a NULL name) indicates an empty line in
	* the generic usage message.
	*/
	static zfs_command_t command_table[] = {
	{ "version", zfs_do_version, HELP_VERSION },
	{ NULL },
	{ "create", zfs_do_create, HELP_CREATE },
	{ "destroy", zfs_do_destroy, HELP_DESTROY },
	{ NULL },
	{ "snapshot", zfs_do_snapshot, HELP_SNAPSHOT },
	{ "rollback", zfs_do_rollback, HELP_ROLLBACK },
	{ "clone", zfs_do_clone, HELP_CLONE },
	{ "promote", zfs_do_promote, HELP_PROMOTE },
	{ "rename", zfs_do_rename, HELP_RENAME },
	{ "bookmark", zfs_do_bookmark, HELP_BOOKMARK },
	{ "program", zfs_do_channel_program, HELP_CHANNEL_PROGRAM },
	{ NULL },
	{ "list", zfs_do_list, HELP_LIST },
	{ NULL },
	{ "set", zfs_do_set, HELP_SET },
	{ "get", zfs_do_get, HELP_GET },
	{ "inherit", zfs_do_inherit, HELP_INHERIT },
	{ "upgrade", zfs_do_upgrade, HELP_UPGRADE },
	{ NULL },
	{ "userspace", zfs_do_userspace, HELP_USERSPACE },
	{ "groupspace", zfs_do_userspace, HELP_GROUPSPACE },
	{ "projectspace", zfs_do_userspace, HELP_PROJECTSPACE },
	{ NULL },
	{ "project", zfs_do_project, HELP_PROJECT },
	{ NULL },
	{ "mount", zfs_do_mount, HELP_MOUNT },
	{ "unmount", zfs_do_unmount, HELP_UNMOUNT },
	{ "share", zfs_do_share, HELP_SHARE },
	{ "unshare", zfs_do_unshare, HELP_UNSHARE },
	{ NULL },
	{ "send", zfs_do_send, HELP_SEND },
	{ "receive", zfs_do_receive, HELP_RECEIVE },
	{ NULL },
	{ "allow", zfs_do_allow, HELP_ALLOW },
	{ NULL },
	{ "unallow", zfs_do_unallow, HELP_UNALLOW },
	{ NULL },
	{ "hold", zfs_do_hold, HELP_HOLD },
	{ "holds", zfs_do_holds, HELP_HOLDS },
	{ "release", zfs_do_release, HELP_RELEASE },
	{ "diff", zfs_do_diff, HELP_DIFF },
	{ "load-key", zfs_do_load_key, HELP_LOAD_KEY },
	{ "unload-key", zfs_do_unload_key, HELP_UNLOAD_KEY },
	{ "change-key", zfs_do_change_key, HELP_CHANGE_KEY },
	{ "redact", zfs_do_redact, HELP_REDACT },
	{ "wait", zfs_do_wait, HELP_WAIT },

	#ifdef __FreeBSD__
	{ "jail", zfs_do_jail, HELP_JAIL },
	{ "unjail", zfs_do_unjail, HELP_UNJAIL },
	#endif
	};

	#define NCOMMAND (sizeof (command_table) / sizeof (command_table[0]))

	zfs_command_t *current_command;

	static const char *
	get_usage(zfs_help_t idx)
	{
	switch (idx) {
	case HELP_CLONE:
	return (gettext("\tclone [-p] [-o property=value] ... "
	"<snapshot> <filesystem\|volume>\n"));
	case HELP_CREATE:
	return (gettext("\tcreate [-Pnpuv] [-o property=value] ... "
	"<filesystem>\n"
	"\tcreate [-Pnpsv] [-b blocksize] [-o property=value] ... "
	"-V <size> <volume>\n"));
	case HELP_DESTROY:
	return (gettext("\tdestroy [-fnpRrv] <filesystem\|volume>\n"
	"\tdestroy [-dnpRrv] "
	"<filesystem\|volume>@<snap>[%<snap>][,...]\n"
	"\tdestroy <filesystem\|volume>#<bookmark>\n"));
	case HELP_GET:
	return (gettext("\tget [-rHp] [-d max] "
	"[-o \"all\" \| field[,...]]\n"
	"\t [-t type[,...]] [-s source[,...]]\n"
	"\t <\"all\" \| property[,...]> "
	"[filesystem\|volume\|snapshot\|bookmark] ...\n"));
	case HELP_INHERIT:
	return (gettext("\tinherit [-rS] <property> "
	"<filesystem\|volume\|snapshot> ...\n"));
	case HELP_UPGRADE:
	return (gettext("\tupgrade [-v]\n"
	"\tupgrade [-r] [-V version] <-a \| filesystem ...>\n"));
	case HELP_LIST:
	return (gettext("\tlist [-Hp] [-r\|-d max] [-o property[,...]] "
	"[-s property]...\n\t [-S property]... [-t type[,...]] "
	"[filesystem\|volume\|snapshot] ...\n"));
	case HELP_MOUNT:
	return (gettext("\tmount\n"
	"\tmount [-flvO] [-o opts] <-a \| filesystem>\n"));
	case HELP_PROMOTE:
	return (gettext("\tpromote <clone-filesystem>\n"));
	case HELP_RECEIVE:
	return (gettext("\treceive [-vMnsFhu] "
	"[-o <property>=<value>] ... [-x <property>] ...\n"
	"\t <filesystem\|volume\|snapshot>\n"
	"\treceive [-vMnsFhu] [-o <property>=<value>] ... "
	"[-x <property>] ... \n"
	"\t [-d \| -e] <filesystem>\n"
	"\treceive -A <filesystem\|volume>\n"));
	case HELP_RENAME:
	return (gettext("\trename [-f] <filesystem\|volume\|snapshot> "
	"<filesystem\|volume\|snapshot>\n"
	"\trename -p [-f] <filesystem\|volume> <filesystem\|volume>\n"
	"\trename -u [-f] <filesystem> <filesystem>\n"
	"\trename -r <snapshot> <snapshot>\n"));
	case HELP_ROLLBACK:
	return (gettext("\trollback [-rRf] <snapshot>\n"));
	case HELP_SEND:
	return (gettext("\tsend [-DnPpRvLecwhb] [-[i\|I] snapshot] "
	"<snapshot>\n"
	"\tsend [-DnvPLecw] [-i snapshot\|bookmark] "
	"<filesystem\|volume\|snapshot>\n"
	"\tsend [-DnPpvLec] [-i bookmark\|snapshot] "
	"--redact <bookmark> <snapshot>\n"
	"\tsend [-nvPe] -t <receive_resume_token>\n"
	"\tsend [-Pnv] --saved filesystem\n"));
	case HELP_SET:
	return (gettext("\tset <property=value> ... "
	"<filesystem\|volume\|snapshot> ...\n"));
	case HELP_SHARE:
	return (gettext("\tshare [-l] <-a [nfs\|smb] \| filesystem>\n"));
	case HELP_SNAPSHOT:
	return (gettext("\tsnapshot [-r] [-o property=value] ... "
	"<filesystem\|volume>@<snap> ...\n"));
	case HELP_UNMOUNT:
	return (gettext("\tunmount [-fu] "
	"<-a \| filesystem\|mountpoint>\n"));
	case HELP_UNSHARE:
	return (gettext("\tunshare "
	"<-a [nfs\|smb] \| filesystem\|mountpoint>\n"));
	case HELP_ALLOW:
	return (gettext("\tallow <filesystem\|volume>\n"
	"\tallow [-ldug] "
	"<\"everyone\"\|user\|group>[,...] <perm\|@setname>[,...]\n"
	"\t <filesystem\|volume>\n"
	"\tallow [-ld] -e <perm\|@setname>[,...] "
	"<filesystem\|volume>\n"
	"\tallow -c <perm\|@setname>[,...] <filesystem\|volume>\n"
	"\tallow -s @setname <perm\|@setname>[,...] "
	"<filesystem\|volume>\n"));
	case HELP_UNALLOW:
	return (gettext("\tunallow [-rldug] "
	"<\"everyone\"\|user\|group>[,...]\n"
	"\t [<perm\|@setname>[,...]] <filesystem\|volume>\n"
	"\tunallow [-rld] -e [<perm\|@setname>[,...]] "
	"<filesystem\|volume>\n"
	"\tunallow [-r] -c [<perm\|@setname>[,...]] "
	"<filesystem\|volume>\n"
	"\tunallow [-r] -s @setname [<perm\|@setname>[,...]] "
	"<filesystem\|volume>\n"));
	case HELP_USERSPACE:
	return (gettext("\tuserspace [-Hinp] [-o field[,...]] "
	"[-s field] ...\n"
	"\t [-S field] ... [-t type[,...]] "
	"<filesystem\|snapshot\|path>\n"));
	case HELP_GROUPSPACE:
	return (gettext("\tgroupspace [-Hinp] [-o field[,...]] "
	"[-s field] ...\n"
	"\t [-S field] ... [-t type[,...]] "
	"<filesystem\|snapshot\|path>\n"));
	case HELP_PROJECTSPACE:
	return (gettext("\tprojectspace [-Hp] [-o field[,...]] "
	"[-s field] ... \n"
	"\t [-S field] ... <filesystem\|snapshot\|path>\n"));
	case HELP_PROJECT:
	return (gettext("\tproject [-d\|-r] <directory\|file ...>\n"
	"\tproject -c [-0] [-d\|-r] [-p id] <directory\|file ...>\n"
	"\tproject -C [-k] [-r] <directory ...>\n"
	"\tproject [-p id] [-r] [-s] <directory ...>\n"));
	case HELP_HOLD:
	return (gettext("\thold [-r] <tag> <snapshot> ...\n"));
	case HELP_HOLDS:
	return (gettext("\tholds [-rH] <snapshot> ...\n"));
	case HELP_RELEASE:
	return (gettext("\trelease [-r] <tag> <snapshot> ...\n"));
	case HELP_DIFF:
	return (gettext("\tdiff [-FHt] <snapshot> "
	"[snapshot\|filesystem]\n"));
	case HELP_BOOKMARK:
	return (gettext("\tbookmark <snapshot\|bookmark> "
	"<newbookmark>\n"));
	case HELP_CHANNEL_PROGRAM:
	return (gettext("\tprogram [-jn] [-t <instruction limit>] "
	"[-m <memory limit (b)>]\n"
	"\t <pool> <program file> [lua args...]\n"));
	case HELP_LOAD_KEY:
	return (gettext("\tload-key [-rn] [-L <keylocation>] "
	"<-a \| filesystem\|volume>\n"));
	case HELP_UNLOAD_KEY:
	return (gettext("\tunload-key [-r] "
	"<-a \| filesystem\|volume>\n"));
	case HELP_CHANGE_KEY:
	return (gettext("\tchange-key [-l] [-o keyformat=<value>]\n"
	"\t [-o keylocation=<value>] [-o pbkdf2iters=<value>]\n"
	"\t <filesystem\|volume>\n"
	"\tchange-key -i [-l] <filesystem\|volume>\n"));
	case HELP_VERSION:
	return (gettext("\tversion\n"));
	case HELP_REDACT:
	return (gettext("\tredact <snapshot> <bookmark> "
	"<redaction_snapshot> ...\n"));
	case HELP_JAIL:
	return (gettext("\tjail <jailid\|jailname> <filesystem>\n"));
	case HELP_UNJAIL:
	return (gettext("\tunjail <jailid\|jailname> <filesystem>\n"));
	case HELP_WAIT:
	return (gettext("\twait [-t <activity>] <filesystem>\n"));
	default:
	__builtin_unreachable();
	}
	}

	void
	nomem(void)
	{
	(void) fprintf(stderr, gettext("internal error: out of memory\n"));
	exit(1);
	}

	/*
	* Utility function to guarantee malloc() success.
	*/

	void *
	safe_malloc(size_t size)
	{
	void *data;

	if ((data = calloc(1, size)) == NULL)
	nomem();

	return (data);
	}

	static void *
	safe_realloc(void *data, size_t size)
	{
	void *newp;
	if ((newp = realloc(data, size)) == NULL) {
	free(data);
	nomem();
	}

	return (newp);
	}

	static char *
	safe_strdup(char *str)
	{
	char *dupstr = strdup(str);

	if (dupstr == NULL)
	nomem();

	return (dupstr);
	}

	/*
	* Callback routine that will print out information for each of
	* the properties.
	*/
	static int
	usage_prop_cb(int prop, void *cb)
	{
	FILE *fp = cb;

	(void) fprintf(fp, "\t%-15s ", zfs_prop_to_name(prop));

	if (zfs_prop_readonly(prop))
	(void) fprintf(fp, " NO ");
	else
	(void) fprintf(fp, "YES ");

	if (zfs_prop_inheritable(prop))
	(void) fprintf(fp, " YES ");
	else
	(void) fprintf(fp, " NO ");

	if (zfs_prop_values(prop) == NULL)
	(void) fprintf(fp, "-\n");
	else
	(void) fprintf(fp, "%s\n", zfs_prop_values(prop));

	return (ZPROP_CONT);
	}

	/*
	* Display usage message. If we're inside a command, display only the usage for
	* that command. Otherwise, iterate over the entire command table and display
	* a complete usage message.
	*/
	static void
	usage(boolean_t requested)
	{
	int i;
	boolean_t show_properties = B_FALSE;
	FILE *fp = requested ? stdout : stderr;

	if (current_command == NULL) {

	(void) fprintf(fp, gettext("usage: zfs command args ...\n"));
	(void) fprintf(fp,
	gettext("where 'command' is one of the following:\n\n"));

	for (i = 0; i < NCOMMAND; i++) {
	if (command_table[i].name == NULL)
	(void) fprintf(fp, "\n");
	else
	(void) fprintf(fp, "%s",
	get_usage(command_table[i].usage));
	}

	(void) fprintf(fp, gettext("\nEach dataset is of the form: "
	"pool/[dataset/]*dataset[@name]\n"));
	} else {
	(void) fprintf(fp, gettext("usage:\n"));
	(void) fprintf(fp, "%s", get_usage(current_command->usage));
	}

	if (current_command != NULL &&
	(strcmp(current_command->name, "set") == 0 \|\|
	strcmp(current_command->name, "get") == 0 \|\|
	strcmp(current_command->name, "inherit") == 0 \|\|
	strcmp(current_command->name, "list") == 0))
	show_properties = B_TRUE;

	if (show_properties) {
	(void) fprintf(fp,
	gettext("\nThe following properties are supported:\n"));

	(void) fprintf(fp, "\n\t%-14s %s %s %s\n\n",
	"PROPERTY", "EDIT", "INHERIT", "VALUES");

	/* Iterate over all properties */
	(void) zprop_iter(usage_prop_cb, fp, B_FALSE, B_TRUE,
	ZFS_TYPE_DATASET);

	(void) fprintf(fp, "\t%-15s ", "userused@...");
	(void) fprintf(fp, " NO NO <size>\n");
	(void) fprintf(fp, "\t%-15s ", "groupused@...");
	(void) fprintf(fp, " NO NO <size>\n");
	(void) fprintf(fp, "\t%-15s ", "projectused@...");
	(void) fprintf(fp, " NO NO <size>\n");
	(void) fprintf(fp, "\t%-15s ", "userobjused@...");
	(void) fprintf(fp, " NO NO <size>\n");
	(void) fprintf(fp, "\t%-15s ", "groupobjused@...");
	(void) fprintf(fp, " NO NO <size>\n");
	(void) fprintf(fp, "\t%-15s ", "projectobjused@...");
	(void) fprintf(fp, " NO NO <size>\n");
	(void) fprintf(fp, "\t%-15s ", "userquota@...");
	(void) fprintf(fp, "YES NO <size> \| none\n");
	(void) fprintf(fp, "\t%-15s ", "groupquota@...");
	(void) fprintf(fp, "YES NO <size> \| none\n");
	(void) fprintf(fp, "\t%-15s ", "projectquota@...");
	(void) fprintf(fp, "YES NO <size> \| none\n");
	(void) fprintf(fp, "\t%-15s ", "userobjquota@...");
	(void) fprintf(fp, "YES NO <size> \| none\n");
	(void) fprintf(fp, "\t%-15s ", "groupobjquota@...");
	(void) fprintf(fp, "YES NO <size> \| none\n");
	(void) fprintf(fp, "\t%-15s ", "projectobjquota@...");
	(void) fprintf(fp, "YES NO <size> \| none\n");
	(void) fprintf(fp, "\t%-15s ", "written@<snap>");
	(void) fprintf(fp, " NO NO <size>\n");
	(void) fprintf(fp, "\t%-15s ", "written#<bookmark>");
	(void) fprintf(fp, " NO NO <size>\n");

	(void) fprintf(fp, gettext("\nSizes are specified in bytes "
	"with standard units such as K, M, G, etc.\n"));
	(void) fprintf(fp, gettext("\nUser-defined properties can "
	"be specified by using a name containing a colon (:).\n"));
	(void) fprintf(fp, gettext("\nThe {user\|group\|project}"
	"[obj]{used\|quota}@ properties must be appended with\n"
	"a user\|group\|project specifier of one of these forms:\n"
	" POSIX name (eg: \"matt\")\n"
	" POSIX id (eg: \"126829\")\n"
	" SMB name@domain (eg: \"matt@sun\")\n"
	" SMB SID (eg: \"S-1-234-567-89\")\n"));
	} else {
	(void) fprintf(fp,
	gettext("\nFor the property list, run: %s\n"),
	"zfs set\|get");
	(void) fprintf(fp,
	gettext("\nFor the delegated permission list, run: %s\n"),
	"zfs allow\|unallow");
	}

	/*
	* See comments at end of main().
	*/
	if (getenv("ZFS_ABORT") != NULL) {
	(void) printf("dumping core by request\n");
	abort();
	}

	exit(requested ? 0 : 2);
	}

	/*
	* Take a property=value argument string and add it to the given nvlist.
	* Modifies the argument inplace.
	*/
	static boolean_t
	parseprop(nvlist_t props, char propname)
	{
	char *propval;

	if ((propval = strchr(propname, '=')) == NULL) {
	(void) fprintf(stderr, gettext("missing "
	"'=' for property=value argument\n"));
	return (B_FALSE);
	}
	*propval = '\0';
	propval++;
	if (nvlist_exists(props, propname)) {
	(void) fprintf(stderr, gettext("property '%s' "
	"specified multiple times\n"), propname);
	return (B_FALSE);
	}
	if (nvlist_add_string(props, propname, propval) != 0)
	nomem();
	return (B_TRUE);
	}

	/*
	* Take a property name argument and add it to the given nvlist.
	* Modifies the argument inplace.
	*/
	static boolean_t
	parsepropname(nvlist_t props, char propname)
	{
	if (strchr(propname, '=') != NULL) {
	(void) fprintf(stderr, gettext("invalid character "
	"'=' in property argument\n"));
	return (B_FALSE);
	}
	if (nvlist_exists(props, propname)) {
	(void) fprintf(stderr, gettext("property '%s' "
	"specified multiple times\n"), propname);
	return (B_FALSE);
	}
	if (nvlist_add_boolean(props, propname) != 0)
	nomem();
	return (B_TRUE);
	}

	static int
	parse_depth(char opt, int flags)
	{
	char *tmp;
	int depth;

	depth = (int)strtol(opt, &tmp, 0);
	if (*tmp) {
	(void) fprintf(stderr,
	gettext("%s is not an integer\n"), optarg);
	usage(B_FALSE);
	}
	if (depth < 0) {
	(void) fprintf(stderr,
	gettext("Depth can not be negative.\n"));
	usage(B_FALSE);
	}
	*flags \|= (ZFS_ITER_DEPTH_LIMIT\|ZFS_ITER_RECURSE);
	return (depth);
	}

	#define PROGRESS_DELAY 2 /* seconds */

	static char *pt_reverse = "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b";
	static time_t pt_begin;
	static char *pt_header = NULL;
	static boolean_t pt_shown;

	static void
	start_progress_timer(void)
	{
	pt_begin = time(NULL) + PROGRESS_DELAY;
	pt_shown = B_FALSE;
	}

	static void
	set_progress_header(char *header)
	{
	assert(pt_header == NULL);
	pt_header = safe_strdup(header);
	if (pt_shown) {
	(void) printf("%s: ", header);
	(void) fflush(stdout);
	}
	}

	static void
	update_progress(char *update)
	{
	if (!pt_shown && time(NULL) > pt_begin) {
	int len = strlen(update);

	(void) printf("%s: %s%.s", pt_header, update, len, len,
	pt_reverse);
	(void) fflush(stdout);
	pt_shown = B_TRUE;
	} else if (pt_shown) {
	int len = strlen(update);

	(void) printf("%s%.s", update, len, len, pt_reverse);
	(void) fflush(stdout);
	}
	}

	static void
	finish_progress(char *done)
	{
	if (pt_shown) {
	(void) printf("%s\n", done);
	(void) fflush(stdout);
	}
	free(pt_header);
	pt_header = NULL;
	}

	/* This function checks if the passed fd refers to /dev/null or /dev/zero */
	#ifdef __linux__
	static boolean_t
	is_dev_nullzero(int fd)
	{
	struct stat st;
	fstat(fd, &st);
	return (major(st.st_rdev) == 1 && (minor(st.st_rdev) == 3 /* null */ \|\|
	minor(st.st_rdev) == 5 /* zero */));
	}
	#endif

	static void
	note_dev_error(int err, int fd)
	{
	#ifdef __linux__
	if (err == EINVAL && is_dev_nullzero(fd)) {
	(void) fprintf(stderr,
	gettext("Error: Writing directly to /dev/{null,zero} files"
	" on certain kernels is not currently implemented.\n"
	"(As a workaround, "
	"try \"zfs send [...] \| cat > /dev/null\")\n"));
	}
	#endif
	}

	static int
	zfs_mount_and_share(libzfs_handle_t hdl, const char dataset, zfs_type_t type)
	{
	zfs_handle_t *zhp = NULL;
	int ret = 0;

	zhp = zfs_open(hdl, dataset, type);
	if (zhp == NULL)
	return (1);

	/*
	* Volumes may neither be mounted or shared. Potentially in the
	* future filesystems detected on these volumes could be mounted.
	*/
	if (zfs_get_type(zhp) == ZFS_TYPE_VOLUME) {
	zfs_close(zhp);
	return (0);
	}

	/*
	* Mount and/or share the new filesystem as appropriate. We provide a
	* verbose error message to let the user know that their filesystem was
	* in fact created, even if we failed to mount or share it.
	*
	* If the user doesn't want the dataset automatically mounted, then
	* skip the mount/share step
	*/
	if (zfs_prop_valid_for_type(ZFS_PROP_CANMOUNT, type, B_FALSE) &&
	zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT) == ZFS_CANMOUNT_ON) {
	if (zfs_mount_delegation_check()) {
	(void) fprintf(stderr, gettext("filesystem "
	"successfully created, but it may only be "
	"mounted by root\n"));
	ret = 1;
	} else if (zfs_mount(zhp, NULL, 0) != 0) {
	(void) fprintf(stderr, gettext("filesystem "
	"successfully created, but not mounted\n"));
	ret = 1;
	} else if (zfs_share(zhp) != 0) {
	(void) fprintf(stderr, gettext("filesystem "
	"successfully created, but not shared\n"));
	ret = 1;
	}
	zfs_commit_all_shares();
	}

	zfs_close(zhp);

	return (ret);
	}

	/*
	* zfs clone [-p] [-o prop=value] ... <snap> <fs \| vol>
	*
	* Given an existing dataset, create a writable copy whose initial contents
	* are the same as the source. The newly created dataset maintains a
	* dependency on the original; the original cannot be destroyed so long as
	* the clone exists.
	*
	* The '-p' flag creates all the non-existing ancestors of the target first.
	*/
	static int
	zfs_do_clone(int argc, char **argv)
	{
	zfs_handle_t *zhp = NULL;
	boolean_t parents = B_FALSE;
	nvlist_t *props;
	int ret = 0;
	int c;

	if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0)
	nomem();

	/* check options */
	while ((c = getopt(argc, argv, "o:p")) != -1) {
	switch (c) {
	case 'o':
	if (!parseprop(props, optarg)) {
	nvlist_free(props);
	return (1);
	}
	break;
	case 'p':
	parents = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	goto usage;
	}
	}

	argc -= optind;
	argv += optind;

	/* check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing source dataset "
	"argument\n"));
	goto usage;
	}
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing target dataset "
	"argument\n"));
	goto usage;
	}
	if (argc > 2) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	goto usage;
	}

	/* open the source dataset */
	if ((zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_SNAPSHOT)) == NULL) {
	nvlist_free(props);
	return (1);
	}

	if (parents && zfs_name_valid(argv[1], ZFS_TYPE_FILESYSTEM \|
	ZFS_TYPE_VOLUME)) {
	/*
	* Now create the ancestors of the target dataset. If the
	* target already exists and '-p' option was used we should not
	* complain.
	*/
	if (zfs_dataset_exists(g_zfs, argv[1], ZFS_TYPE_FILESYSTEM \|
	ZFS_TYPE_VOLUME)) {
	zfs_close(zhp);
	nvlist_free(props);
	return (0);
	}
	if (zfs_create_ancestors(g_zfs, argv[1]) != 0) {
	zfs_close(zhp);
	nvlist_free(props);
	return (1);
	}
	}

	/* pass to libzfs */
	ret = zfs_clone(zhp, argv[1], props);

	/* create the mountpoint if necessary */
	if (ret == 0) {
	if (log_history) {
	(void) zpool_log_history(g_zfs, history_str);
	log_history = B_FALSE;
	}

	ret = zfs_mount_and_share(g_zfs, argv[1], ZFS_TYPE_DATASET);
	}

	zfs_close(zhp);
	nvlist_free(props);

	return (!!ret);

	usage:
	ASSERT3P(zhp, ==, NULL);
	nvlist_free(props);
	usage(B_FALSE);
	return (-1);
	}

	/*
	* Return a default volblocksize for the pool which always uses more than
	* half of the data sectors. This primarily applies to dRAID which always
	* writes full stripe widths.
	*/
	static uint64_t
	default_volblocksize(zpool_handle_t zhp, nvlist_t props)
	{
	uint64_t volblocksize, asize = SPA_MINBLOCKSIZE;
	nvlist_t tree, *vdevs;
	uint_t nvdevs;

	nvlist_t *config = zpool_get_config(zhp, NULL);

	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &tree) != 0 \|\|
	nvlist_lookup_nvlist_array(tree, ZPOOL_CONFIG_CHILDREN,
	&vdevs, &nvdevs) != 0) {
	return (ZVOL_DEFAULT_BLOCKSIZE);
	}

	for (int i = 0; i < nvdevs; i++) {
	nvlist_t *nv = vdevs[i];
	uint64_t ashift, ndata, nparity;

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ASHIFT, &ashift) != 0)
	continue;

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DRAID_NDATA,
	&ndata) == 0) {
	/* dRAID minimum allocation width */
	asize = MAX(asize, ndata * (1ULL << ashift));
	} else if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NPARITY,
	&nparity) == 0) {
	/* raidz minimum allocation width */
	if (nparity == 1)
	asize = MAX(asize, 2 * (1ULL << ashift));
	else
	asize = MAX(asize, 4 * (1ULL << ashift));
	} else {
	/* mirror or (non-redundant) leaf vdev */
	asize = MAX(asize, 1ULL << ashift);
	}
	}

	/*
	* Calculate the target volblocksize such that more than half
	* of the asize is used. The following table is for 4k sectors.
	*
	* n asize blksz used \| n asize blksz used
	* -------------------------+---------------------------------
	* 1 4,096 8,192 100% \| 9 36,864 32,768 88%
	* 2 8,192 8,192 100% \| 10 40,960 32,768 80%
	* 3 12,288 8,192 66% \| 11 45,056 32,768 72%
	* 4 16,384 16,384 100% \| 12 49,152 32,768 66%
	* 5 20,480 16,384 80% \| 13 53,248 32,768 61%
	* 6 24,576 16,384 66% \| 14 57,344 32,768 57%
	* 7 28,672 16,384 57% \| 15 61,440 32,768 53%
	* 8 32,768 32,768 100% \| 16 65,536 65,636 100%
	*
	* This is primarily a concern for dRAID which always allocates
	* a full stripe width. For dRAID the default stripe width is
	* n=8 in which case the volblocksize is set to 32k. Ignoring
	* compression there are no unused sectors. This same reasoning
	* applies to raidz[2,3] so target 4 sectors to minimize waste.
	*/
	uint64_t tgt_volblocksize = ZVOL_DEFAULT_BLOCKSIZE;
	while (tgt_volblocksize * 2 <= asize)
	tgt_volblocksize *= 2;

	const char *prop = zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE);
	if (nvlist_lookup_uint64(props, prop, &volblocksize) == 0) {

	/* Issue a warning when a non-optimal size is requested. */
	if (volblocksize < ZVOL_DEFAULT_BLOCKSIZE) {
	(void) fprintf(stderr, gettext("Warning: "
	"volblocksize (%llu) is less than the default "
	"minimum block size (%llu).\nTo reduce wasted "
	"space a volblocksize of %llu is recommended.\n"),
	(u_longlong_t)volblocksize,
	(u_longlong_t)ZVOL_DEFAULT_BLOCKSIZE,
	(u_longlong_t)tgt_volblocksize);
	} else if (volblocksize < tgt_volblocksize) {
	(void) fprintf(stderr, gettext("Warning: "
	"volblocksize (%llu) is much less than the "
	"minimum allocation\nunit (%llu), which wastes "
	"at least %llu%% of space. To reduce wasted "
	"space,\nuse a larger volblocksize (%llu is "
	"recommended), fewer dRAID data disks\n"
	"per group, or smaller sector size (ashift).\n"),
	(u_longlong_t)volblocksize, (u_longlong_t)asize,
	(u_longlong_t)((100 * (asize - volblocksize)) /
	asize), (u_longlong_t)tgt_volblocksize);
	}
	} else {
	volblocksize = tgt_volblocksize;
	fnvlist_add_uint64(props, prop, volblocksize);
	}

	return (volblocksize);
	}

	/*
	* zfs create [-Pnpv] [-o prop=value] ... fs
	* zfs create [-Pnpsv] [-b blocksize] [-o prop=value] ... -V vol size
	*
	* Create a new dataset. This command can be used to create filesystems
	* and volumes. Snapshot creation is handled by 'zfs snapshot'.
	* For volumes, the user must specify a size to be used.
	*
	* The '-s' flag applies only to volumes, and indicates that we should not try
	* to set the reservation for this volume. By default we set a reservation
	* equal to the size for any volume. For pools with SPA_VERSION >=
	* SPA_VERSION_REFRESERVATION, we set a refreservation instead.
	*
	* The '-p' flag creates all the non-existing ancestors of the target first.
	*
	* The '-n' flag is no-op (dry run) mode. This will perform a user-space sanity
	* check of arguments and properties, but does not check for permissions,
	* available space, etc.
	*
	* The '-u' flag prevents the newly created file system from being mounted.
	*
	* The '-v' flag is for verbose output.
	*
	* The '-P' flag is used for parseable output. It implies '-v'.
	*/
	static int
	zfs_do_create(int argc, char **argv)
	{
	zfs_type_t type = ZFS_TYPE_FILESYSTEM;
	zpool_handle_t *zpool_handle = NULL;
	nvlist_t *real_props = NULL;
	uint64_t volsize = 0;
	int c;
	boolean_t noreserve = B_FALSE;
	boolean_t bflag = B_FALSE;
	boolean_t parents = B_FALSE;
	boolean_t dryrun = B_FALSE;
	boolean_t nomount = B_FALSE;
	boolean_t verbose = B_FALSE;
	boolean_t parseable = B_FALSE;
	int ret = 1;
	nvlist_t *props;
	uint64_t intval;
	char *strval;

	if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0)
	nomem();

	/* check options */
	while ((c = getopt(argc, argv, ":PV:b:nso:puv")) != -1) {
	switch (c) {
	case 'V':
	type = ZFS_TYPE_VOLUME;
	if (zfs_nicestrtonum(g_zfs, optarg, &intval) != 0) {
	(void) fprintf(stderr, gettext("bad volume "
	"size '%s': %s\n"), optarg,
	libzfs_error_description(g_zfs));
	goto error;
	}

	if (nvlist_add_uint64(props,
	zfs_prop_to_name(ZFS_PROP_VOLSIZE), intval) != 0)
	nomem();
	volsize = intval;
	break;
	case 'P':
	verbose = B_TRUE;
	parseable = B_TRUE;
	break;
	case 'p':
	parents = B_TRUE;
	break;
	case 'b':
	bflag = B_TRUE;
	if (zfs_nicestrtonum(g_zfs, optarg, &intval) != 0) {
	(void) fprintf(stderr, gettext("bad volume "
	"block size '%s': %s\n"), optarg,
	libzfs_error_description(g_zfs));
	goto error;
	}

	if (nvlist_add_uint64(props,
	zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE),
	intval) != 0)
	nomem();
	break;
	case 'n':
	dryrun = B_TRUE;
	break;
	case 'o':
	if (!parseprop(props, optarg))
	goto error;
	break;
	case 's':
	noreserve = B_TRUE;
	break;
	case 'u':
	nomount = B_TRUE;
	break;
	case 'v':
	verbose = B_TRUE;
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing size "
	"argument\n"));
	goto badusage;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	goto badusage;
	}
	}

	if ((bflag \|\| noreserve) && type != ZFS_TYPE_VOLUME) {
	(void) fprintf(stderr, gettext("'-s' and '-b' can only be "
	"used when creating a volume\n"));
	goto badusage;
	}
	if (nomount && type != ZFS_TYPE_FILESYSTEM) {
	(void) fprintf(stderr, gettext("'-u' can only be "
	"used when creating a filesystem\n"));
	goto badusage;
	}

	argc -= optind;
	argv += optind;

	/* check number of arguments */
	if (argc == 0) {
	(void) fprintf(stderr, gettext("missing %s argument\n"),
	zfs_type_to_name(type));
	goto badusage;
	}
	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	goto badusage;
	}

	if (dryrun \|\| type == ZFS_TYPE_VOLUME) {
	char msg[ZFS_MAX_DATASET_NAME_LEN * 2];
	char *p;

	if ((p = strchr(argv[0], '/')) != NULL)
	*p = '\0';
	zpool_handle = zpool_open(g_zfs, argv[0]);
	if (p != NULL)
	*p = '/';
	if (zpool_handle == NULL)
	goto error;

	(void) snprintf(msg, sizeof (msg),
	dryrun ? gettext("cannot verify '%s'") :
	gettext("cannot create '%s'"), argv[0]);
	if (props && (real_props = zfs_valid_proplist(g_zfs, type,
	props, 0, NULL, zpool_handle, B_TRUE, msg)) == NULL) {
	zpool_close(zpool_handle);
	goto error;
	}
	}

	if (type == ZFS_TYPE_VOLUME) {
	const char *prop = zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE);
	uint64_t volblocksize = default_volblocksize(zpool_handle,
	real_props);

	if (volblocksize != ZVOL_DEFAULT_BLOCKSIZE &&
	nvlist_lookup_string(props, prop, &strval) != 0) {
	if (asprintf(&strval, "%llu",
	(u_longlong_t)volblocksize) == -1)
	nomem();
	nvlist_add_string(props, prop, strval);
	free(strval);
	}

	/*
	* If volsize is not a multiple of volblocksize, round it
	* up to the nearest multiple of the volblocksize.
	*/
	if (volsize % volblocksize) {
	volsize = P2ROUNDUP_TYPED(volsize, volblocksize,
	uint64_t);

	if (nvlist_add_uint64(props,
	zfs_prop_to_name(ZFS_PROP_VOLSIZE), volsize) != 0) {
	nvlist_free(props);
	nomem();
	}
	}
	}

	if (type == ZFS_TYPE_VOLUME && !noreserve) {
	uint64_t spa_version;
	zfs_prop_t resv_prop;

	spa_version = zpool_get_prop_int(zpool_handle,
	ZPOOL_PROP_VERSION, NULL);
	if (spa_version >= SPA_VERSION_REFRESERVATION)
	resv_prop = ZFS_PROP_REFRESERVATION;
	else
	resv_prop = ZFS_PROP_RESERVATION;

	volsize = zvol_volsize_to_reservation(zpool_handle, volsize,
	real_props);

	if (nvlist_lookup_string(props, zfs_prop_to_name(resv_prop),
	&strval) != 0) {
	if (nvlist_add_uint64(props,
	zfs_prop_to_name(resv_prop), volsize) != 0) {
	nvlist_free(props);
	nomem();
	}
	}
	}
	if (zpool_handle != NULL) {
	zpool_close(zpool_handle);
	nvlist_free(real_props);
	}

	if (parents && zfs_name_valid(argv[0], type)) {
	/*
	* Now create the ancestors of target dataset. If the target
	* already exists and '-p' option was used we should not
	* complain.
	*/
	if (zfs_dataset_exists(g_zfs, argv[0], type)) {
	ret = 0;
	goto error;
	}
	if (verbose) {
	(void) printf(parseable ? "create_ancestors\t%s\n" :
	dryrun ? "would create ancestors of %s\n" :
	"create ancestors of %s\n", argv[0]);
	}
	if (!dryrun) {
	if (zfs_create_ancestors(g_zfs, argv[0]) != 0) {
	goto error;
	}
	}
	}

	if (verbose) {
	nvpair_t *nvp = NULL;
	(void) printf(parseable ? "create\t%s\n" :
	dryrun ? "would create %s\n" : "create %s\n", argv[0]);
	while ((nvp = nvlist_next_nvpair(props, nvp)) != NULL) {
	uint64_t uval;
	char *sval;

	switch (nvpair_type(nvp)) {
	case DATA_TYPE_UINT64:
	VERIFY0(nvpair_value_uint64(nvp, &uval));
	(void) printf(parseable ?
	"property\t%s\t%llu\n" : "\t%s=%llu\n",
	nvpair_name(nvp), (u_longlong_t)uval);
	break;
	case DATA_TYPE_STRING:
	VERIFY0(nvpair_value_string(nvp, &sval));
	(void) printf(parseable ?
	"property\t%s\t%s\n" : "\t%s=%s\n",
	nvpair_name(nvp), sval);
	break;
	default:
	(void) fprintf(stderr, "property '%s' "
	"has illegal type %d\n",
	nvpair_name(nvp), nvpair_type(nvp));
	abort();
	}
	}
	}
	if (dryrun) {
	ret = 0;
	goto error;
	}

	/* pass to libzfs */
	if (zfs_create(g_zfs, argv[0], type, props) != 0)
	goto error;

	if (log_history) {
	(void) zpool_log_history(g_zfs, history_str);
	log_history = B_FALSE;
	}

	if (nomount) {
	ret = 0;
	goto error;
	}

	ret = zfs_mount_and_share(g_zfs, argv[0], ZFS_TYPE_DATASET);
	error:
	nvlist_free(props);
	return (ret);
	badusage:
	nvlist_free(props);
	usage(B_FALSE);
	return (2);
	}

	/*
	* zfs destroy [-rRf] <fs, vol>
	* zfs destroy [-rRd] <snap>
	*
	* -r Recursively destroy all children
	* -R Recursively destroy all dependents, including clones
	* -f Force unmounting of any dependents
	* -d If we can't destroy now, mark for deferred destruction
	*
	* Destroys the given dataset. By default, it will unmount any filesystems,
	* and refuse to destroy a dataset that has any dependents. A dependent can
	* either be a child, or a clone of a child.
	*/
	typedef struct destroy_cbdata {
	boolean_t cb_first;
	boolean_t cb_force;
	boolean_t cb_recurse;
	boolean_t cb_error;
	boolean_t cb_doclones;
	zfs_handle_t *cb_target;
	boolean_t cb_defer_destroy;
	boolean_t cb_verbose;
	boolean_t cb_parsable;
	boolean_t cb_dryrun;
	nvlist_t *cb_nvl;
	nvlist_t *cb_batchedsnaps;

	/* first snap in contiguous run */
	char *cb_firstsnap;
	/* previous snap in contiguous run */
	char *cb_prevsnap;
	int64_t cb_snapused;
	char *cb_snapspec;
	char *cb_bookmark;
	uint64_t cb_snap_count;
	} destroy_cbdata_t;

	/*
	* Check for any dependents based on the '-r' or '-R' flags.
	*/
	static int
	destroy_check_dependent(zfs_handle_t zhp, void data)
	{
	destroy_cbdata_t *cbp = data;
	const char *tname = zfs_get_name(cbp->cb_target);
	const char *name = zfs_get_name(zhp);

	if (strncmp(tname, name, strlen(tname)) == 0 &&
	(name[strlen(tname)] == '/' \|\| name[strlen(tname)] == '@')) {
	/*
	* This is a direct descendant, not a clone somewhere else in
	* the hierarchy.
	*/
	if (cbp->cb_recurse)
	goto out;

	if (cbp->cb_first) {
	(void) fprintf(stderr, gettext("cannot destroy '%s': "
	"%s has children\n"),
	zfs_get_name(cbp->cb_target),
	zfs_type_to_name(zfs_get_type(cbp->cb_target)));
	(void) fprintf(stderr, gettext("use '-r' to destroy "
	"the following datasets:\n"));
	cbp->cb_first = B_FALSE;
	cbp->cb_error = B_TRUE;
	}

	(void) fprintf(stderr, "%s\n", zfs_get_name(zhp));
	} else {
	/*
	* This is a clone. We only want to report this if the '-r'
	* wasn't specified, or the target is a snapshot.
	*/
	if (!cbp->cb_recurse &&
	zfs_get_type(cbp->cb_target) != ZFS_TYPE_SNAPSHOT)
	goto out;

	if (cbp->cb_first) {
	(void) fprintf(stderr, gettext("cannot destroy '%s': "
	"%s has dependent clones\n"),
	zfs_get_name(cbp->cb_target),
	zfs_type_to_name(zfs_get_type(cbp->cb_target)));
	(void) fprintf(stderr, gettext("use '-R' to destroy "
	"the following datasets:\n"));
	cbp->cb_first = B_FALSE;
	cbp->cb_error = B_TRUE;
	cbp->cb_dryrun = B_TRUE;
	}

	(void) fprintf(stderr, "%s\n", zfs_get_name(zhp));
	}

	out:
	zfs_close(zhp);
	return (0);
	}

	static int
	destroy_batched(destroy_cbdata_t *cb)
	{
	int error = zfs_destroy_snaps_nvl(g_zfs,
	cb->cb_batchedsnaps, B_FALSE);
	fnvlist_free(cb->cb_batchedsnaps);
	cb->cb_batchedsnaps = fnvlist_alloc();
	return (error);
	}

	static int
	destroy_callback(zfs_handle_t zhp, void data)
	{
	destroy_cbdata_t *cb = data;
	const char *name = zfs_get_name(zhp);
	int error;

	if (cb->cb_verbose) {
	if (cb->cb_parsable) {
	(void) printf("destroy\t%s\n", name);
	} else if (cb->cb_dryrun) {
	(void) printf(gettext("would destroy %s\n"),
	name);
	} else {
	(void) printf(gettext("will destroy %s\n"),
	name);
	}
	}

	/*
	* Ignore pools (which we've already flagged as an error before getting
	* here).
	*/
	if (strchr(zfs_get_name(zhp), '/') == NULL &&
	zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM) {
	zfs_close(zhp);
	return (0);
	}
	if (cb->cb_dryrun) {
	zfs_close(zhp);
	return (0);
	}

	/*
	* We batch up all contiguous snapshots (even of different
	* filesystems) and destroy them with one ioctl. We can't
	* simply do all snap deletions and then all fs deletions,
	* because we must delete a clone before its origin.
	*/
	if (zfs_get_type(zhp) == ZFS_TYPE_SNAPSHOT) {
	cb->cb_snap_count++;
	fnvlist_add_boolean(cb->cb_batchedsnaps, name);
	if (cb->cb_snap_count % 10 == 0 && cb->cb_defer_destroy)
	error = destroy_batched(cb);
	} else {
	error = destroy_batched(cb);
	if (error != 0 \|\|
	zfs_unmount(zhp, NULL, cb->cb_force ? MS_FORCE : 0) != 0 \|\|
	zfs_destroy(zhp, cb->cb_defer_destroy) != 0) {
	zfs_close(zhp);
	/*
	* When performing a recursive destroy we ignore errors
	* so that the recursive destroy could continue
	* destroying past problem datasets
	*/
	if (cb->cb_recurse) {
	cb->cb_error = B_TRUE;
	return (0);
	}
	return (-1);
	}
	}

	zfs_close(zhp);
	return (0);
	}

	static int
	destroy_print_cb(zfs_handle_t zhp, void arg)
	{
	destroy_cbdata_t *cb = arg;
	const char *name = zfs_get_name(zhp);
	int err = 0;

	if (nvlist_exists(cb->cb_nvl, name)) {
	if (cb->cb_firstsnap == NULL)
	cb->cb_firstsnap = strdup(name);
	if (cb->cb_prevsnap != NULL)
	free(cb->cb_prevsnap);
	/* this snap continues the current range */
	cb->cb_prevsnap = strdup(name);
	if (cb->cb_firstsnap == NULL \|\| cb->cb_prevsnap == NULL)
	nomem();
	if (cb->cb_verbose) {
	if (cb->cb_parsable) {
	(void) printf("destroy\t%s\n", name);
	} else if (cb->cb_dryrun) {
	(void) printf(gettext("would destroy %s\n"),
	name);
	} else {
	(void) printf(gettext("will destroy %s\n"),
	name);
	}
	}
	} else if (cb->cb_firstsnap != NULL) {
	/* end of this range */
	uint64_t used = 0;
	err = lzc_snaprange_space(cb->cb_firstsnap,
	cb->cb_prevsnap, &used);
	cb->cb_snapused += used;
	free(cb->cb_firstsnap);
	cb->cb_firstsnap = NULL;
	free(cb->cb_prevsnap);
	cb->cb_prevsnap = NULL;
	}
	zfs_close(zhp);
	return (err);
	}

	static int
	destroy_print_snapshots(zfs_handle_t fs_zhp, destroy_cbdata_t cb)
	{
	int err;
	assert(cb->cb_firstsnap == NULL);
	assert(cb->cb_prevsnap == NULL);
	err = zfs_iter_snapshots_sorted(fs_zhp, destroy_print_cb, cb, 0, 0);
	if (cb->cb_firstsnap != NULL) {
	uint64_t used = 0;
	if (err == 0) {
	err = lzc_snaprange_space(cb->cb_firstsnap,
	cb->cb_prevsnap, &used);
	}
	cb->cb_snapused += used;
	free(cb->cb_firstsnap);
	cb->cb_firstsnap = NULL;
	free(cb->cb_prevsnap);
	cb->cb_prevsnap = NULL;
	}
	return (err);
	}

	static int
	snapshot_to_nvl_cb(zfs_handle_t zhp, void arg)
	{
	destroy_cbdata_t *cb = arg;
	int err = 0;

	/* Check for clones. */
	if (!cb->cb_doclones && !cb->cb_defer_destroy) {
	cb->cb_target = zhp;
	cb->cb_first = B_TRUE;
	err = zfs_iter_dependents(zhp, B_TRUE,
	destroy_check_dependent, cb);
	}

	if (err == 0) {
	if (nvlist_add_boolean(cb->cb_nvl, zfs_get_name(zhp)))
	nomem();
	}
	zfs_close(zhp);
	return (err);
	}

	static int
	gather_snapshots(zfs_handle_t zhp, void arg)
	{
	destroy_cbdata_t *cb = arg;
	int err = 0;

	err = zfs_iter_snapspec(zhp, cb->cb_snapspec, snapshot_to_nvl_cb, cb);
	if (err == ENOENT)
	err = 0;
	if (err != 0)
	goto out;

	if (cb->cb_verbose) {
	err = destroy_print_snapshots(zhp, cb);
	if (err != 0)
	goto out;
	}

	if (cb->cb_recurse)
	err = zfs_iter_filesystems(zhp, gather_snapshots, cb);

	out:
	zfs_close(zhp);
	return (err);
	}

	static int
	destroy_clones(destroy_cbdata_t *cb)
	{
	nvpair_t *pair;
	for (pair = nvlist_next_nvpair(cb->cb_nvl, NULL);
	pair != NULL;
	pair = nvlist_next_nvpair(cb->cb_nvl, pair)) {
	zfs_handle_t *zhp = zfs_open(g_zfs, nvpair_name(pair),
	ZFS_TYPE_SNAPSHOT);
	if (zhp != NULL) {
	boolean_t defer = cb->cb_defer_destroy;
	int err;

	/*
	* We can't defer destroy non-snapshots, so set it to
	* false while destroying the clones.
	*/
	cb->cb_defer_destroy = B_FALSE;
	err = zfs_iter_dependents(zhp, B_FALSE,
	destroy_callback, cb);
	cb->cb_defer_destroy = defer;
	zfs_close(zhp);
	if (err != 0)
	return (err);
	}
	}
	return (0);
	}

	static int
	zfs_do_destroy(int argc, char **argv)
	{
	destroy_cbdata_t cb = { 0 };
	int rv = 0;
	int err = 0;
	int c;
	zfs_handle_t *zhp = NULL;
	char at, pound;
	zfs_type_t type = ZFS_TYPE_DATASET;

	/* check options */
	while ((c = getopt(argc, argv, "vpndfrR")) != -1) {
	switch (c) {
	case 'v':
	cb.cb_verbose = B_TRUE;
	break;
	case 'p':
	cb.cb_verbose = B_TRUE;
	cb.cb_parsable = B_TRUE;
	break;
	case 'n':
	cb.cb_dryrun = B_TRUE;
	break;
	case 'd':
	cb.cb_defer_destroy = B_TRUE;
	type = ZFS_TYPE_SNAPSHOT;
	break;
	case 'f':
	cb.cb_force = B_TRUE;
	break;
	case 'r':
	cb.cb_recurse = B_TRUE;
	break;
	case 'R':
	cb.cb_recurse = B_TRUE;
	cb.cb_doclones = B_TRUE;
	break;
	case '?':
	default:
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* check number of arguments */
	if (argc == 0) {
	(void) fprintf(stderr, gettext("missing dataset argument\n"));
	usage(B_FALSE);
	}
	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	at = strchr(argv[0], '@');
	pound = strchr(argv[0], '#');
	if (at != NULL) {

	/* Build the list of snaps to destroy in cb_nvl. */
	cb.cb_nvl = fnvlist_alloc();

	*at = '\0';
	zhp = zfs_open(g_zfs, argv[0],
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME);
	if (zhp == NULL) {
	nvlist_free(cb.cb_nvl);
	return (1);
	}

	cb.cb_snapspec = at + 1;
	if (gather_snapshots(zfs_handle_dup(zhp), &cb) != 0 \|\|
	cb.cb_error) {
	rv = 1;
	goto out;
	}

	if (nvlist_empty(cb.cb_nvl)) {
	(void) fprintf(stderr, gettext("could not find any "
	"snapshots to destroy; check snapshot names.\n"));
	rv = 1;
	goto out;
	}

	if (cb.cb_verbose) {
	char buf[16];
	zfs_nicebytes(cb.cb_snapused, buf, sizeof (buf));
	if (cb.cb_parsable) {
	(void) printf("reclaim\t%llu\n",
	(u_longlong_t)cb.cb_snapused);
	} else if (cb.cb_dryrun) {
	(void) printf(gettext("would reclaim %s\n"),
	buf);
	} else {
	(void) printf(gettext("will reclaim %s\n"),
	buf);
	}
	}

	if (!cb.cb_dryrun) {
	if (cb.cb_doclones) {
	cb.cb_batchedsnaps = fnvlist_alloc();
	err = destroy_clones(&cb);
	if (err == 0) {
	err = zfs_destroy_snaps_nvl(g_zfs,
	cb.cb_batchedsnaps, B_FALSE);
	}
	if (err != 0) {
	rv = 1;
	goto out;
	}
	}
	if (err == 0) {
	err = zfs_destroy_snaps_nvl(g_zfs, cb.cb_nvl,
	cb.cb_defer_destroy);
	}
	}

	if (err != 0)
	rv = 1;
	} else if (pound != NULL) {
	int err;
	nvlist_t *nvl;

	if (cb.cb_dryrun) {
	(void) fprintf(stderr,
	"dryrun is not supported with bookmark\n");
	return (-1);
	}

	if (cb.cb_defer_destroy) {
	(void) fprintf(stderr,
	"defer destroy is not supported with bookmark\n");
	return (-1);
	}

	if (cb.cb_recurse) {
	(void) fprintf(stderr,
	"recursive is not supported with bookmark\n");
	return (-1);
	}

	/*
	* Unfortunately, zfs_bookmark() doesn't honor the
	* casesensitivity setting. However, we can't simply
	* remove this check, because lzc_destroy_bookmarks()
	* ignores non-existent bookmarks, so this is necessary
	* to get a proper error message.
	*/
	if (!zfs_bookmark_exists(argv[0])) {
	(void) fprintf(stderr, gettext("bookmark '%s' "
	"does not exist.\n"), argv[0]);
	return (1);
	}

	nvl = fnvlist_alloc();
	fnvlist_add_boolean(nvl, argv[0]);

	err = lzc_destroy_bookmarks(nvl, NULL);
	if (err != 0) {
	(void) zfs_standard_error(g_zfs, err,
	"cannot destroy bookmark");
	}

	nvlist_free(nvl);

	return (err);
	} else {
	/* Open the given dataset */
	if ((zhp = zfs_open(g_zfs, argv[0], type)) == NULL)
	return (1);

	cb.cb_target = zhp;

	/*
	* Perform an explicit check for pools before going any further.
	*/
	if (!cb.cb_recurse && strchr(zfs_get_name(zhp), '/') == NULL &&
	zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM) {
	(void) fprintf(stderr, gettext("cannot destroy '%s': "
	"operation does not apply to pools\n"),
	zfs_get_name(zhp));
	(void) fprintf(stderr, gettext("use 'zfs destroy -r "
	"%s' to destroy all datasets in the pool\n"),
	zfs_get_name(zhp));
	(void) fprintf(stderr, gettext("use 'zpool destroy %s' "
	"to destroy the pool itself\n"), zfs_get_name(zhp));
	rv = 1;
	goto out;
	}

	/*
	* Check for any dependents and/or clones.
	*/
	cb.cb_first = B_TRUE;
	if (!cb.cb_doclones &&
	zfs_iter_dependents(zhp, B_TRUE, destroy_check_dependent,
	&cb) != 0) {
	rv = 1;
	goto out;
	}

	if (cb.cb_error) {
	rv = 1;
	goto out;
	}
	cb.cb_batchedsnaps = fnvlist_alloc();
	if (zfs_iter_dependents(zhp, B_FALSE, destroy_callback,
	&cb) != 0) {
	rv = 1;
	goto out;
	}

	/*
	* Do the real thing. The callback will close the
	* handle regardless of whether it succeeds or not.
	*/
	err = destroy_callback(zhp, &cb);
	zhp = NULL;
	if (err == 0) {
	err = zfs_destroy_snaps_nvl(g_zfs,
	cb.cb_batchedsnaps, cb.cb_defer_destroy);
	}
	if (err != 0 \|\| cb.cb_error == B_TRUE)
	rv = 1;
	}

	out:
	fnvlist_free(cb.cb_batchedsnaps);
	fnvlist_free(cb.cb_nvl);
	if (zhp != NULL)
	zfs_close(zhp);
	return (rv);
	}

	static boolean_t
	is_recvd_column(zprop_get_cbdata_t *cbp)
	{
	int i;
	zfs_get_column_t col;

	for (i = 0; i < ZFS_GET_NCOLS &&
	(col = cbp->cb_columns[i]) != GET_COL_NONE; i++)
	if (col == GET_COL_RECVD)
	return (B_TRUE);
	return (B_FALSE);
	}

	/*
	* zfs get [-rHp] [-o all \| field[,field]...] [-s source[,source]...]
	* < all \| property[,property]... > < fs \| snap \| vol > ...
	*
	* -r recurse over any child datasets
	* -H scripted mode. Headers are stripped, and fields are separated
	* by tabs instead of spaces.
	* -o Set of fields to display. One of "name,property,value,
	* received,source". Default is "name,property,value,source".
	* "all" is an alias for all five.
	* -s Set of sources to allow. One of
	* "local,default,inherited,received,temporary,none". Default is
	* all six.
	* -p Display values in parsable (literal) format.
	*
	* Prints properties for the given datasets. The user can control which
	* columns to display as well as which property types to allow.
	*/

	/*
	* Invoked to display the properties for a single dataset.
	*/
	static int
	get_callback(zfs_handle_t zhp, void data)
	{
	char buf[ZFS_MAXPROPLEN];
	char rbuf[ZFS_MAXPROPLEN];
	zprop_source_t sourcetype;
	char source[ZFS_MAX_DATASET_NAME_LEN];
	zprop_get_cbdata_t *cbp = data;
	nvlist_t *user_props = zfs_get_user_props(zhp);
	zprop_list_t *pl = cbp->cb_proplist;
	nvlist_t *propval;
	char *strval;
	char *sourceval;
	boolean_t received = is_recvd_column(cbp);

	for (; pl != NULL; pl = pl->pl_next) {
	char *recvdval = NULL;
	/*
	* Skip the special fake placeholder. This will also skip over
	* the name property when 'all' is specified.
	*/
	if (pl->pl_prop == ZFS_PROP_NAME &&
	pl == cbp->cb_proplist)
	continue;

	if (pl->pl_prop != ZPROP_INVAL) {
	if (zfs_prop_get(zhp, pl->pl_prop, buf,
	sizeof (buf), &sourcetype, source,
	sizeof (source),
	cbp->cb_literal) != 0) {
	if (pl->pl_all)
	continue;
	if (!zfs_prop_valid_for_type(pl->pl_prop,
	ZFS_TYPE_DATASET, B_FALSE)) {
	(void) fprintf(stderr,
	gettext("No such property '%s'\n"),
	zfs_prop_to_name(pl->pl_prop));
	continue;
	}
	sourcetype = ZPROP_SRC_NONE;
	(void) strlcpy(buf, "-", sizeof (buf));
	}

	if (received && (zfs_prop_get_recvd(zhp,
	zfs_prop_to_name(pl->pl_prop), rbuf, sizeof (rbuf),
	cbp->cb_literal) == 0))
	recvdval = rbuf;

	zprop_print_one_property(zfs_get_name(zhp), cbp,
	zfs_prop_to_name(pl->pl_prop),
	buf, sourcetype, source, recvdval);
	} else if (zfs_prop_userquota(pl->pl_user_prop)) {
	sourcetype = ZPROP_SRC_LOCAL;

	if (zfs_prop_get_userquota(zhp, pl->pl_user_prop,
	buf, sizeof (buf), cbp->cb_literal) != 0) {
	sourcetype = ZPROP_SRC_NONE;
	(void) strlcpy(buf, "-", sizeof (buf));
	}

	zprop_print_one_property(zfs_get_name(zhp), cbp,
	pl->pl_user_prop, buf, sourcetype, source, NULL);
	} else if (zfs_prop_written(pl->pl_user_prop)) {
	sourcetype = ZPROP_SRC_LOCAL;

	if (zfs_prop_get_written(zhp, pl->pl_user_prop,
	buf, sizeof (buf), cbp->cb_literal) != 0) {
	sourcetype = ZPROP_SRC_NONE;
	(void) strlcpy(buf, "-", sizeof (buf));
	}

	zprop_print_one_property(zfs_get_name(zhp), cbp,
	pl->pl_user_prop, buf, sourcetype, source, NULL);
	} else {
	if (nvlist_lookup_nvlist(user_props,
	pl->pl_user_prop, &propval) != 0) {
	if (pl->pl_all)
	continue;
	sourcetype = ZPROP_SRC_NONE;
	strval = "-";
	} else {
	verify(nvlist_lookup_string(propval,
	ZPROP_VALUE, &strval) == 0);
	verify(nvlist_lookup_string(propval,
	ZPROP_SOURCE, &sourceval) == 0);

	if (strcmp(sourceval,
	zfs_get_name(zhp)) == 0) {
	sourcetype = ZPROP_SRC_LOCAL;
	} else if (strcmp(sourceval,
	ZPROP_SOURCE_VAL_RECVD) == 0) {
	sourcetype = ZPROP_SRC_RECEIVED;
	} else {
	sourcetype = ZPROP_SRC_INHERITED;
	(void) strlcpy(source,
	sourceval, sizeof (source));
	}
	}

	if (received && (zfs_prop_get_recvd(zhp,
	pl->pl_user_prop, rbuf, sizeof (rbuf),
	cbp->cb_literal) == 0))
	recvdval = rbuf;

	zprop_print_one_property(zfs_get_name(zhp), cbp,
	pl->pl_user_prop, strval, sourcetype,
	source, recvdval);
	}
	}

	return (0);
	}

	static int
	zfs_do_get(int argc, char **argv)
	{
	zprop_get_cbdata_t cb = { 0 };
	int i, c, flags = ZFS_ITER_ARGS_CAN_BE_PATHS;
	int types = ZFS_TYPE_DATASET \| ZFS_TYPE_BOOKMARK;
	char value, fields;
	int ret = 0;
	int limit = 0;
	zprop_list_t fake_name = { 0 };

	/*
	* Set up default columns and sources.
	*/
	cb.cb_sources = ZPROP_SRC_ALL;
	cb.cb_columns[0] = GET_COL_NAME;
	cb.cb_columns[1] = GET_COL_PROPERTY;
	cb.cb_columns[2] = GET_COL_VALUE;
	cb.cb_columns[3] = GET_COL_SOURCE;
	cb.cb_type = ZFS_TYPE_DATASET;

	/* check options */
	while ((c = getopt(argc, argv, ":d:o:s:rt:Hp")) != -1) {
	switch (c) {
	case 'p':
	cb.cb_literal = B_TRUE;
	break;
	case 'd':
	limit = parse_depth(optarg, &flags);
	break;
	case 'r':
	flags \|= ZFS_ITER_RECURSE;
	break;
	case 'H':
	cb.cb_scripted = B_TRUE;
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	usage(B_FALSE);
	break;
	case 'o':
	/*
	* Process the set of columns to display. We zero out
	* the structure to give us a blank slate.
	*/
	bzero(&cb.cb_columns, sizeof (cb.cb_columns));
	i = 0;
	while (*optarg != '\0') {
	static char *col_subopts[] =
	{ "name", "property", "value", "received",
	"source", "all", NULL };

	if (i == ZFS_GET_NCOLS) {
	(void) fprintf(stderr, gettext("too "
	"many fields given to -o "
	"option\n"));
	usage(B_FALSE);
	}

	switch (getsubopt(&optarg, col_subopts,
	&value)) {
	case 0:
	cb.cb_columns[i++] = GET_COL_NAME;
	break;
	case 1:
	cb.cb_columns[i++] = GET_COL_PROPERTY;
	break;
	case 2:
	cb.cb_columns[i++] = GET_COL_VALUE;
	break;
	case 3:
	cb.cb_columns[i++] = GET_COL_RECVD;
	flags \|= ZFS_ITER_RECVD_PROPS;
	break;
	case 4:
	cb.cb_columns[i++] = GET_COL_SOURCE;
	break;
	case 5:
	if (i > 0) {
	(void) fprintf(stderr,
	gettext("\"all\" conflicts "
	"with specific fields "
	"given to -o option\n"));
	usage(B_FALSE);
	}
	cb.cb_columns[0] = GET_COL_NAME;
	cb.cb_columns[1] = GET_COL_PROPERTY;
	cb.cb_columns[2] = GET_COL_VALUE;
	cb.cb_columns[3] = GET_COL_RECVD;
	cb.cb_columns[4] = GET_COL_SOURCE;
	flags \|= ZFS_ITER_RECVD_PROPS;
	i = ZFS_GET_NCOLS;
	break;
	default:
	(void) fprintf(stderr,
	gettext("invalid column name "
	"'%s'\n"), value);
	usage(B_FALSE);
	}
	}
	break;

	case 's':
	cb.cb_sources = 0;
	while (*optarg != '\0') {
	static char *source_subopts[] = {
	"local", "default", "inherited",
	"received", "temporary", "none",
	NULL };

	switch (getsubopt(&optarg, source_subopts,
	&value)) {
	case 0:
	cb.cb_sources \|= ZPROP_SRC_LOCAL;
	break;
	case 1:
	cb.cb_sources \|= ZPROP_SRC_DEFAULT;
	break;
	case 2:
	cb.cb_sources \|= ZPROP_SRC_INHERITED;
	break;
	case 3:
	cb.cb_sources \|= ZPROP_SRC_RECEIVED;
	break;
	case 4:
	cb.cb_sources \|= ZPROP_SRC_TEMPORARY;
	break;
	case 5:
	cb.cb_sources \|= ZPROP_SRC_NONE;
	break;
	default:
	(void) fprintf(stderr,
	gettext("invalid source "
	"'%s'\n"), value);
	usage(B_FALSE);
	}
	}
	break;

	case 't':
	types = 0;
	flags &= ~ZFS_ITER_PROP_LISTSNAPS;
	while (*optarg != '\0') {
	static char *type_subopts[] = { "filesystem",
	"volume", "snapshot", "snap", "bookmark",
	"all", NULL };

	switch (getsubopt(&optarg, type_subopts,
	&value)) {
	case 0:
	types \|= ZFS_TYPE_FILESYSTEM;
	break;
	case 1:
	types \|= ZFS_TYPE_VOLUME;
	break;
	case 2:
	case 3:
	types \|= ZFS_TYPE_SNAPSHOT;
	break;
	case 4:
	types \|= ZFS_TYPE_BOOKMARK;
	break;
	case 5:
	types = ZFS_TYPE_DATASET \|
	ZFS_TYPE_BOOKMARK;
	break;

	default:
	(void) fprintf(stderr,
	gettext("invalid type '%s'\n"),
	value);
	usage(B_FALSE);
	}
	}
	break;

	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing property "
	"argument\n"));
	usage(B_FALSE);
	}

	fields = argv[0];

	/*
	* Handle users who want to get all snapshots or bookmarks
	* of a dataset (ex. 'zfs get -t snapshot refer <dataset>').
	*/
	if ((types == ZFS_TYPE_SNAPSHOT \|\| types == ZFS_TYPE_BOOKMARK) &&
	argc > 1 && (flags & ZFS_ITER_RECURSE) == 0 && limit == 0) {
	flags \|= (ZFS_ITER_DEPTH_LIMIT \| ZFS_ITER_RECURSE);
	limit = 1;
	}

	if (zprop_get_list(g_zfs, fields, &cb.cb_proplist, ZFS_TYPE_DATASET)
	!= 0)
	usage(B_FALSE);

	argc--;
	argv++;

	/*
	* As part of zfs_expand_proplist(), we keep track of the maximum column
	* width for each property. For the 'NAME' (and 'SOURCE') columns, we
	* need to know the maximum name length. However, the user likely did
	* not specify 'name' as one of the properties to fetch, so we need to
	* make sure we always include at least this property for
	* print_get_headers() to work properly.
	*/
	if (cb.cb_proplist != NULL) {
	fake_name.pl_prop = ZFS_PROP_NAME;
	fake_name.pl_width = strlen(gettext("NAME"));
	fake_name.pl_next = cb.cb_proplist;
	cb.cb_proplist = &fake_name;
	}

	cb.cb_first = B_TRUE;

	/* run for each object */
	ret = zfs_for_each(argc, argv, flags, types, NULL,
	&cb.cb_proplist, limit, get_callback, &cb);

	if (cb.cb_proplist == &fake_name)
	zprop_free_list(fake_name.pl_next);
	else
	zprop_free_list(cb.cb_proplist);

	return (ret);
	}

	/*
	* inherit [-rS] <property> <fs\|vol> ...
	*
	* -r Recurse over all children
	* -S Revert to received value, if any
	*
	* For each dataset specified on the command line, inherit the given property
	* from its parent. Inheriting a property at the pool level will cause it to
	* use the default value. The '-r' flag will recurse over all children, and is
	* useful for setting a property on a hierarchy-wide basis, regardless of any
	* local modifications for each dataset.
	*/

	typedef struct inherit_cbdata {
	const char *cb_propname;
	boolean_t cb_received;
	} inherit_cbdata_t;

	static int
	inherit_recurse_cb(zfs_handle_t zhp, void data)
	{
	inherit_cbdata_t *cb = data;
	zfs_prop_t prop = zfs_name_to_prop(cb->cb_propname);

	/*
	* If we're doing it recursively, then ignore properties that
	* are not valid for this type of dataset.
	*/
	if (prop != ZPROP_INVAL &&
	!zfs_prop_valid_for_type(prop, zfs_get_type(zhp), B_FALSE))
	return (0);

	return (zfs_prop_inherit(zhp, cb->cb_propname, cb->cb_received) != 0);
	}

	static int
	inherit_cb(zfs_handle_t zhp, void data)
	{
	inherit_cbdata_t *cb = data;

	return (zfs_prop_inherit(zhp, cb->cb_propname, cb->cb_received) != 0);
	}

	static int
	zfs_do_inherit(int argc, char **argv)
	{
	int c;
	zfs_prop_t prop;
	inherit_cbdata_t cb = { 0 };
	char *propname;
	int ret = 0;
	int flags = 0;
	boolean_t received = B_FALSE;

	/* check options */
	while ((c = getopt(argc, argv, "rS")) != -1) {
	switch (c) {
	case 'r':
	flags \|= ZFS_ITER_RECURSE;
	break;
	case 'S':
	received = B_TRUE;
	break;
	case '?':
	default:
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing property argument\n"));
	usage(B_FALSE);
	}
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing dataset argument\n"));
	usage(B_FALSE);
	}

	propname = argv[0];
	argc--;
	argv++;

	if ((prop = zfs_name_to_prop(propname)) != ZPROP_INVAL) {
	if (zfs_prop_readonly(prop)) {
	(void) fprintf(stderr, gettext(
	"%s property is read-only\n"),
	propname);
	return (1);
	}
	if (!zfs_prop_inheritable(prop) && !received) {
	(void) fprintf(stderr, gettext("'%s' property cannot "
	"be inherited\n"), propname);
	if (prop == ZFS_PROP_QUOTA \|\|
	prop == ZFS_PROP_RESERVATION \|\|
	prop == ZFS_PROP_REFQUOTA \|\|
	prop == ZFS_PROP_REFRESERVATION) {
	(void) fprintf(stderr, gettext("use 'zfs set "
	"%s=none' to clear\n"), propname);
	(void) fprintf(stderr, gettext("use 'zfs "
	"inherit -S %s' to revert to received "
	"value\n"), propname);
	}
	return (1);
	}
	if (received && (prop == ZFS_PROP_VOLSIZE \|\|
	prop == ZFS_PROP_VERSION)) {
	(void) fprintf(stderr, gettext("'%s' property cannot "
	"be reverted to a received value\n"), propname);
	return (1);
	}
	} else if (!zfs_prop_user(propname)) {
	(void) fprintf(stderr, gettext("invalid property '%s'\n"),
	propname);
	usage(B_FALSE);
	}

	cb.cb_propname = propname;
	cb.cb_received = received;

	if (flags & ZFS_ITER_RECURSE) {
	ret = zfs_for_each(argc, argv, flags, ZFS_TYPE_DATASET,
	NULL, NULL, 0, inherit_recurse_cb, &cb);
	} else {
	ret = zfs_for_each(argc, argv, flags, ZFS_TYPE_DATASET,
	NULL, NULL, 0, inherit_cb, &cb);
	}

	return (ret);
	}

	typedef struct upgrade_cbdata {
	uint64_t cb_numupgraded;
	uint64_t cb_numsamegraded;
	uint64_t cb_numfailed;
	uint64_t cb_version;
	boolean_t cb_newer;
	boolean_t cb_foundone;
	char cb_lastfs[ZFS_MAX_DATASET_NAME_LEN];
	} upgrade_cbdata_t;

	static int
	same_pool(zfs_handle_t zhp, const char name)
	{
	int len1 = strcspn(name, "/@");
	const char *zhname = zfs_get_name(zhp);
	int len2 = strcspn(zhname, "/@");

	if (len1 != len2)
	return (B_FALSE);
	return (strncmp(name, zhname, len1) == 0);
	}

	static int
	upgrade_list_callback(zfs_handle_t zhp, void data)
	{
	upgrade_cbdata_t *cb = data;
	int version = zfs_prop_get_int(zhp, ZFS_PROP_VERSION);

	/* list if it's old/new */
	if ((!cb->cb_newer && version < ZPL_VERSION) \|\|
	(cb->cb_newer && version > ZPL_VERSION)) {
	char *str;
	if (cb->cb_newer) {
	str = gettext("The following filesystems are "
	"formatted using a newer software version and\n"
	"cannot be accessed on the current system.\n\n");
	} else {
	str = gettext("The following filesystems are "
	"out of date, and can be upgraded. After being\n"
	"upgraded, these filesystems (and any 'zfs send' "
	"streams generated from\n"
	"subsequent snapshots) will no longer be "
	"accessible by older software versions.\n\n");
	}

	if (!cb->cb_foundone) {
	(void) puts(str);
	(void) printf(gettext("VER FILESYSTEM\n"));
	(void) printf(gettext("--- ------------\n"));
	cb->cb_foundone = B_TRUE;
	}

	(void) printf("%2u %s\n", version, zfs_get_name(zhp));
	}

	return (0);
	}

	static int
	upgrade_set_callback(zfs_handle_t zhp, void data)
	{
	upgrade_cbdata_t *cb = data;
	int version = zfs_prop_get_int(zhp, ZFS_PROP_VERSION);
	int needed_spa_version;
	int spa_version;

	if (zfs_spa_version(zhp, &spa_version) < 0)
	return (-1);

	needed_spa_version = zfs_spa_version_map(cb->cb_version);

	if (needed_spa_version < 0)
	return (-1);

	if (spa_version < needed_spa_version) {
	/* can't upgrade */
	(void) printf(gettext("%s: can not be "
	"upgraded; the pool version needs to first "
	"be upgraded\nto version %d\n\n"),
	zfs_get_name(zhp), needed_spa_version);
	cb->cb_numfailed++;
	return (0);
	}

	/* upgrade */
	if (version < cb->cb_version) {
	char verstr[16];
	(void) snprintf(verstr, sizeof (verstr),
	"%llu", (u_longlong_t)cb->cb_version);
	if (cb->cb_lastfs[0] && !same_pool(zhp, cb->cb_lastfs)) {
	/*
	* If they did "zfs upgrade -a", then we could
	* be doing ioctls to different pools. We need
	* to log this history once to each pool, and bypass
	* the normal history logging that happens in main().
	*/
	(void) zpool_log_history(g_zfs, history_str);
	log_history = B_FALSE;
	}
	if (zfs_prop_set(zhp, "version", verstr) == 0)
	cb->cb_numupgraded++;
	else
	cb->cb_numfailed++;
	(void) strcpy(cb->cb_lastfs, zfs_get_name(zhp));
	} else if (version > cb->cb_version) {
	/* can't downgrade */
	(void) printf(gettext("%s: can not be downgraded; "
	"it is already at version %u\n"),
	zfs_get_name(zhp), version);
	cb->cb_numfailed++;
	} else {
	cb->cb_numsamegraded++;
	}
	return (0);
	}

	/*
	* zfs upgrade
	* zfs upgrade -v
	* zfs upgrade [-r] [-V <version>] <-a \| filesystem>
	*/
	static int
	zfs_do_upgrade(int argc, char **argv)
	{
	boolean_t all = B_FALSE;
	boolean_t showversions = B_FALSE;
	int ret = 0;
	upgrade_cbdata_t cb = { 0 };
	int c;
	int flags = ZFS_ITER_ARGS_CAN_BE_PATHS;

	/* check options */
	while ((c = getopt(argc, argv, "rvV:a")) != -1) {
	switch (c) {
	case 'r':
	flags \|= ZFS_ITER_RECURSE;
	break;
	case 'v':
	showversions = B_TRUE;
	break;
	case 'V':
	if (zfs_prop_string_to_index(ZFS_PROP_VERSION,
	optarg, &cb.cb_version) != 0) {
	(void) fprintf(stderr,
	gettext("invalid version %s\n"), optarg);
	usage(B_FALSE);
	}
	break;
	case 'a':
	all = B_TRUE;
	break;
	case '?':
	default:
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if ((!all && !argc) && ((flags & ZFS_ITER_RECURSE) \| cb.cb_version))
	usage(B_FALSE);
	if (showversions && (flags & ZFS_ITER_RECURSE \|\| all \|\|
	cb.cb_version \|\| argc))
	usage(B_FALSE);
	if ((all \|\| argc) && (showversions))
	usage(B_FALSE);
	if (all && argc)
	usage(B_FALSE);

	if (showversions) {
	/* Show info on available versions. */
	(void) printf(gettext("The following filesystem versions are "
	"supported:\n\n"));
	(void) printf(gettext("VER DESCRIPTION\n"));
	(void) printf("--- -----------------------------------------"
	"---------------\n");
	(void) printf(gettext(" 1 Initial ZFS filesystem version\n"));
	(void) printf(gettext(" 2 Enhanced directory entries\n"));
	(void) printf(gettext(" 3 Case insensitive and filesystem "
	"user identifier (FUID)\n"));
	(void) printf(gettext(" 4 userquota, groupquota "
	"properties\n"));
	(void) printf(gettext(" 5 System attributes\n"));
	(void) printf(gettext("\nFor more information on a particular "
	"version, including supported releases,\n"));
	(void) printf("see the ZFS Administration Guide.\n\n");
	ret = 0;
	} else if (argc \|\| all) {
	/* Upgrade filesystems */
	if (cb.cb_version == 0)
	cb.cb_version = ZPL_VERSION;
	ret = zfs_for_each(argc, argv, flags, ZFS_TYPE_FILESYSTEM,
	NULL, NULL, 0, upgrade_set_callback, &cb);
	(void) printf(gettext("%llu filesystems upgraded\n"),
	(u_longlong_t)cb.cb_numupgraded);
	if (cb.cb_numsamegraded) {
	(void) printf(gettext("%llu filesystems already at "
	"this version\n"),
	(u_longlong_t)cb.cb_numsamegraded);
	}
	if (cb.cb_numfailed != 0)
	ret = 1;
	} else {
	/* List old-version filesystems */
	boolean_t found;
	(void) printf(gettext("This system is currently running "
	"ZFS filesystem version %llu.\n\n"), ZPL_VERSION);

	flags \|= ZFS_ITER_RECURSE;
	ret = zfs_for_each(0, NULL, flags, ZFS_TYPE_FILESYSTEM,
	NULL, NULL, 0, upgrade_list_callback, &cb);

	found = cb.cb_foundone;
	cb.cb_foundone = B_FALSE;
	cb.cb_newer = B_TRUE;

	ret = zfs_for_each(0, NULL, flags, ZFS_TYPE_FILESYSTEM,
	NULL, NULL, 0, upgrade_list_callback, &cb);

	if (!cb.cb_foundone && !found) {
	(void) printf(gettext("All filesystems are "
	"formatted with the current version.\n"));
	}
	}

	return (ret);
	}

	/*
	* zfs userspace [-Hinp] [-o field[,...]] [-s field [-s field]...]
	* [-S field [-S field]...] [-t type[,...]]
	* filesystem \| snapshot \| path
	* zfs groupspace [-Hinp] [-o field[,...]] [-s field [-s field]...]
	* [-S field [-S field]...] [-t type[,...]]
	* filesystem \| snapshot \| path
	* zfs projectspace [-Hp] [-o field[,...]] [-s field [-s field]...]
	* [-S field [-S field]...] filesystem \| snapshot \| path
	*
	* -H Scripted mode; elide headers and separate columns by tabs.
	* -i Translate SID to POSIX ID.
	* -n Print numeric ID instead of user/group name.
	* -o Control which fields to display.
	* -p Use exact (parsable) numeric output.
	* -s Specify sort columns, descending order.
	* -S Specify sort columns, ascending order.
	* -t Control which object types to display.
	*
	* Displays space consumed by, and quotas on, each user in the specified
	* filesystem or snapshot.
	*/

	/* us_field_types, us_field_hdr and us_field_names should be kept in sync */
	enum us_field_types {
	USFIELD_TYPE,
	USFIELD_NAME,
	USFIELD_USED,
	USFIELD_QUOTA,
	USFIELD_OBJUSED,
	USFIELD_OBJQUOTA
	};
	static char *us_field_hdr[] = { "TYPE", "NAME", "USED", "QUOTA",
	"OBJUSED", "OBJQUOTA" };
	static char *us_field_names[] = { "type", "name", "used", "quota",
	"objused", "objquota" };
	#define USFIELD_LAST (sizeof (us_field_names) / sizeof (char *))

	#define USTYPE_PSX_GRP (1 << 0)
	#define USTYPE_PSX_USR (1 << 1)
	#define USTYPE_SMB_GRP (1 << 2)
	#define USTYPE_SMB_USR (1 << 3)
	#define USTYPE_PROJ (1 << 4)
	#define USTYPE_ALL \
	(USTYPE_PSX_GRP \| USTYPE_PSX_USR \| USTYPE_SMB_GRP \| USTYPE_SMB_USR \| \
	USTYPE_PROJ)

	static int us_type_bits[] = {
	USTYPE_PSX_GRP,
	USTYPE_PSX_USR,
	USTYPE_SMB_GRP,
	USTYPE_SMB_USR,
	USTYPE_ALL
	};
	static char *us_type_names[] = { "posixgroup", "posixuser", "smbgroup",
	"smbuser", "all" };

	typedef struct us_node {
	nvlist_t *usn_nvl;
	uu_avl_node_t usn_avlnode;
	uu_list_node_t usn_listnode;
	} us_node_t;

	typedef struct us_cbdata {
	nvlist_t **cb_nvlp;
	uu_avl_pool_t *cb_avl_pool;
	uu_avl_t *cb_avl;
	boolean_t cb_numname;
	boolean_t cb_nicenum;
	boolean_t cb_sid2posix;
	zfs_userquota_prop_t cb_prop;
	zfs_sort_column_t *cb_sortcol;
	size_t cb_width[USFIELD_LAST];
	} us_cbdata_t;

	static boolean_t us_populated = B_FALSE;

	typedef struct {
	zfs_sort_column_t *si_sortcol;
	boolean_t si_numname;
	} us_sort_info_t;

	static int
	us_field_index(char *field)
	{
	int i;

	for (i = 0; i < USFIELD_LAST; i++) {
	if (strcmp(field, us_field_names[i]) == 0)
	return (i);
	}

	return (-1);
	}

	static int
	us_compare(const void larg, const void rarg, void *unused)
	{
	const us_node_t *l = larg;
	const us_node_t *r = rarg;
	us_sort_info_t si = (us_sort_info_t )unused;
	zfs_sort_column_t *sortcol = si->si_sortcol;
	boolean_t numname = si->si_numname;
	nvlist_t *lnvl = l->usn_nvl;
	nvlist_t *rnvl = r->usn_nvl;
	int rc = 0;
	boolean_t lvb, rvb;

	for (; sortcol != NULL; sortcol = sortcol->sc_next) {
	char *lvstr = "";
	char *rvstr = "";
	uint32_t lv32 = 0;
	uint32_t rv32 = 0;
	uint64_t lv64 = 0;
	uint64_t rv64 = 0;
	zfs_prop_t prop = sortcol->sc_prop;
	const char *propname = NULL;
	boolean_t reverse = sortcol->sc_reverse;

	switch (prop) {
	case ZFS_PROP_TYPE:
	propname = "type";
	(void) nvlist_lookup_uint32(lnvl, propname, &lv32);
	(void) nvlist_lookup_uint32(rnvl, propname, &rv32);
	if (rv32 != lv32)
	rc = (rv32 < lv32) ? 1 : -1;
	break;
	case ZFS_PROP_NAME:
	propname = "name";
	if (numname) {
	compare_nums:
	(void) nvlist_lookup_uint64(lnvl, propname,
	&lv64);
	(void) nvlist_lookup_uint64(rnvl, propname,
	&rv64);
	if (rv64 != lv64)
	rc = (rv64 < lv64) ? 1 : -1;
	} else {
	if ((nvlist_lookup_string(lnvl, propname,
	&lvstr) == ENOENT) \|\|
	(nvlist_lookup_string(rnvl, propname,
	&rvstr) == ENOENT)) {
	goto compare_nums;
	}
	rc = strcmp(lvstr, rvstr);
	}
	break;
	case ZFS_PROP_USED:
	case ZFS_PROP_QUOTA:
	if (!us_populated)
	break;
	if (prop == ZFS_PROP_USED)
	propname = "used";
	else
	propname = "quota";
	(void) nvlist_lookup_uint64(lnvl, propname, &lv64);
	(void) nvlist_lookup_uint64(rnvl, propname, &rv64);
	if (rv64 != lv64)
	rc = (rv64 < lv64) ? 1 : -1;
	break;

	default:
	break;
	}

	if (rc != 0) {
	if (rc < 0)
	return (reverse ? 1 : -1);
	else
	return (reverse ? -1 : 1);
	}
	}

	/*
	* If entries still seem to be the same, check if they are of the same
	* type (smbentity is added only if we are doing SID to POSIX ID
	* translation where we can have duplicate type/name combinations).
	*/
	if (nvlist_lookup_boolean_value(lnvl, "smbentity", &lvb) == 0 &&
	nvlist_lookup_boolean_value(rnvl, "smbentity", &rvb) == 0 &&
	lvb != rvb)
	return (lvb < rvb ? -1 : 1);

	return (0);
	}

	static boolean_t
	zfs_prop_is_user(unsigned p)
	{
	return (p == ZFS_PROP_USERUSED \|\| p == ZFS_PROP_USERQUOTA \|\|
	p == ZFS_PROP_USEROBJUSED \|\| p == ZFS_PROP_USEROBJQUOTA);
	}

	static boolean_t
	zfs_prop_is_group(unsigned p)
	{
	return (p == ZFS_PROP_GROUPUSED \|\| p == ZFS_PROP_GROUPQUOTA \|\|
	p == ZFS_PROP_GROUPOBJUSED \|\| p == ZFS_PROP_GROUPOBJQUOTA);
	}

	static boolean_t
	zfs_prop_is_project(unsigned p)
	{
	return (p == ZFS_PROP_PROJECTUSED \|\| p == ZFS_PROP_PROJECTQUOTA \|\|
	p == ZFS_PROP_PROJECTOBJUSED \|\| p == ZFS_PROP_PROJECTOBJQUOTA);
	}

	static inline const char *
	us_type2str(unsigned field_type)
	{
	switch (field_type) {
	case USTYPE_PSX_USR:
	return ("POSIX User");
	case USTYPE_PSX_GRP:
	return ("POSIX Group");
	case USTYPE_SMB_USR:
	return ("SMB User");
	case USTYPE_SMB_GRP:
	return ("SMB Group");
	case USTYPE_PROJ:
	return ("Project");
	default:
	return ("Undefined");
	}
	}

	static int
	userspace_cb(void arg, const char domain, uid_t rid, uint64_t space)
	{
	us_cbdata_t cb = (us_cbdata_t )arg;
	zfs_userquota_prop_t prop = cb->cb_prop;
	char *name = NULL;
	char *propname;
	char sizebuf[32];
	us_node_t *node;
	uu_avl_pool_t *avl_pool = cb->cb_avl_pool;
	uu_avl_t *avl = cb->cb_avl;
	uu_avl_index_t idx;
	nvlist_t *props;
	us_node_t *n;
	zfs_sort_column_t *sortcol = cb->cb_sortcol;
	unsigned type = 0;
	const char *typestr;
	size_t namelen;
	size_t typelen;
	size_t sizelen;
	int typeidx, nameidx, sizeidx;
	us_sort_info_t sortinfo = { sortcol, cb->cb_numname };
	boolean_t smbentity = B_FALSE;

	if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0)
	nomem();
	node = safe_malloc(sizeof (us_node_t));
	uu_avl_node_init(node, &node->usn_avlnode, avl_pool);
	node->usn_nvl = props;

	if (domain != NULL && domain[0] != '\0') {
	#ifdef HAVE_IDMAP
	/* SMB */
	char sid[MAXNAMELEN + 32];
	uid_t id;
	uint64_t classes;
	int err;
	directory_error_t e;

	smbentity = B_TRUE;

	(void) snprintf(sid, sizeof (sid), "%s-%u", domain, rid);

	if (prop == ZFS_PROP_GROUPUSED \|\| prop == ZFS_PROP_GROUPQUOTA) {
	type = USTYPE_SMB_GRP;
	err = sid_to_id(sid, B_FALSE, &id);
	} else {
	type = USTYPE_SMB_USR;
	err = sid_to_id(sid, B_TRUE, &id);
	}

	if (err == 0) {
	rid = id;
	if (!cb->cb_sid2posix) {
	e = directory_name_from_sid(NULL, sid, &name,
	&classes);
	if (e != NULL)
	directory_error_free(e);
	if (name == NULL)
	name = sid;
	}
	}
	#else
	nvlist_free(props);
	free(node);

	return (-1);
	#endif /* HAVE_IDMAP */
	}

	if (cb->cb_sid2posix \|\| domain == NULL \|\| domain[0] == '\0') {
	/* POSIX or -i */
	if (zfs_prop_is_group(prop)) {
	type = USTYPE_PSX_GRP;
	if (!cb->cb_numname) {
	struct group *g;

	if ((g = getgrgid(rid)) != NULL)
	name = g->gr_name;
	}
	} else if (zfs_prop_is_user(prop)) {
	type = USTYPE_PSX_USR;
	if (!cb->cb_numname) {
	struct passwd *p;

	if ((p = getpwuid(rid)) != NULL)
	name = p->pw_name;
	}
	} else {
	type = USTYPE_PROJ;
	}
	}

	/*
	* Make sure that the type/name combination is unique when doing
	* SID to POSIX ID translation (hence changing the type from SMB to
	* POSIX).
	*/
	if (cb->cb_sid2posix &&
	nvlist_add_boolean_value(props, "smbentity", smbentity) != 0)
	nomem();

	/* Calculate/update width of TYPE field */
	typestr = us_type2str(type);
	typelen = strlen(gettext(typestr));
	typeidx = us_field_index("type");
	if (typelen > cb->cb_width[typeidx])
	cb->cb_width[typeidx] = typelen;
	if (nvlist_add_uint32(props, "type", type) != 0)
	nomem();

	/* Calculate/update width of NAME field */
	if ((cb->cb_numname && cb->cb_sid2posix) \|\| name == NULL) {
	if (nvlist_add_uint64(props, "name", rid) != 0)
	nomem();
	namelen = snprintf(NULL, 0, "%u", rid);
	} else {
	if (nvlist_add_string(props, "name", name) != 0)
	nomem();
	namelen = strlen(name);
	}
	nameidx = us_field_index("name");
	if (nameidx >= 0 && namelen > cb->cb_width[nameidx])
	cb->cb_width[nameidx] = namelen;

	/*
	* Check if this type/name combination is in the list and update it;
	* otherwise add new node to the list.
	*/
	if ((n = uu_avl_find(avl, node, &sortinfo, &idx)) == NULL) {
	uu_avl_insert(avl, node, idx);
	} else {
	nvlist_free(props);
	free(node);
	node = n;
	props = node->usn_nvl;
	}

	/* Calculate/update width of USED/QUOTA fields */
	if (cb->cb_nicenum) {
	if (prop == ZFS_PROP_USERUSED \|\| prop == ZFS_PROP_GROUPUSED \|\|
	prop == ZFS_PROP_USERQUOTA \|\| prop == ZFS_PROP_GROUPQUOTA \|\|
	prop == ZFS_PROP_PROJECTUSED \|\|
	prop == ZFS_PROP_PROJECTQUOTA) {
	zfs_nicebytes(space, sizebuf, sizeof (sizebuf));
	} else {
	zfs_nicenum(space, sizebuf, sizeof (sizebuf));
	}
	} else {
	(void) snprintf(sizebuf, sizeof (sizebuf), "%llu",
	(u_longlong_t)space);
	}
	sizelen = strlen(sizebuf);
	if (prop == ZFS_PROP_USERUSED \|\| prop == ZFS_PROP_GROUPUSED \|\|
	prop == ZFS_PROP_PROJECTUSED) {
	propname = "used";
	if (!nvlist_exists(props, "quota"))
	(void) nvlist_add_uint64(props, "quota", 0);
	} else if (prop == ZFS_PROP_USERQUOTA \|\| prop == ZFS_PROP_GROUPQUOTA \|\|
	prop == ZFS_PROP_PROJECTQUOTA) {
	propname = "quota";
	if (!nvlist_exists(props, "used"))
	(void) nvlist_add_uint64(props, "used", 0);
	} else if (prop == ZFS_PROP_USEROBJUSED \|\|
	prop == ZFS_PROP_GROUPOBJUSED \|\| prop == ZFS_PROP_PROJECTOBJUSED) {
	propname = "objused";
	if (!nvlist_exists(props, "objquota"))
	(void) nvlist_add_uint64(props, "objquota", 0);
	} else if (prop == ZFS_PROP_USEROBJQUOTA \|\|
	prop == ZFS_PROP_GROUPOBJQUOTA \|\|
	prop == ZFS_PROP_PROJECTOBJQUOTA) {
	propname = "objquota";
	if (!nvlist_exists(props, "objused"))
	(void) nvlist_add_uint64(props, "objused", 0);
	} else {
	return (-1);
	}
	sizeidx = us_field_index(propname);
	if (sizeidx >= 0 && sizelen > cb->cb_width[sizeidx])
	cb->cb_width[sizeidx] = sizelen;

	if (nvlist_add_uint64(props, propname, space) != 0)
	nomem();

	return (0);
	}

	static void
	print_us_node(boolean_t scripted, boolean_t parsable, int *fields, int types,
	size_t width, us_node_t node)
	{
	nvlist_t *nvl = node->usn_nvl;
	char valstr[MAXNAMELEN];
	boolean_t first = B_TRUE;
	int cfield = 0;
	int field;
	uint32_t ustype;

	/* Check type */
	(void) nvlist_lookup_uint32(nvl, "type", &ustype);
	if (!(ustype & types))
	return;

	while ((field = fields[cfield]) != USFIELD_LAST) {
	nvpair_t *nvp = NULL;
	data_type_t type;
	uint32_t val32;
	uint64_t val64;
	char *strval = "-";

	while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) {
	if (strcmp(nvpair_name(nvp),
	us_field_names[field]) == 0)
	break;
	}

	type = nvp == NULL ? DATA_TYPE_UNKNOWN : nvpair_type(nvp);
	switch (type) {
	case DATA_TYPE_UINT32:
	(void) nvpair_value_uint32(nvp, &val32);
	break;
	case DATA_TYPE_UINT64:
	(void) nvpair_value_uint64(nvp, &val64);
	break;
	case DATA_TYPE_STRING:
	(void) nvpair_value_string(nvp, &strval);
	break;
	case DATA_TYPE_UNKNOWN:
	break;
	default:
	(void) fprintf(stderr, "invalid data type\n");
	}

	switch (field) {
	case USFIELD_TYPE:
	if (type == DATA_TYPE_UINT32)
	strval = (char *)us_type2str(val32);
	break;
	case USFIELD_NAME:
	if (type == DATA_TYPE_UINT64) {
	(void) sprintf(valstr, "%llu",
	(u_longlong_t)val64);
	strval = valstr;
	}
	break;
	case USFIELD_USED:
	case USFIELD_QUOTA:
	if (type == DATA_TYPE_UINT64) {
	if (parsable) {
	(void) sprintf(valstr, "%llu",
	(u_longlong_t)val64);
	strval = valstr;
	} else if (field == USFIELD_QUOTA &&
	val64 == 0) {
	strval = "none";
	} else {
	zfs_nicebytes(val64, valstr,
	sizeof (valstr));
	strval = valstr;
	}
	}
	break;
	case USFIELD_OBJUSED:
	case USFIELD_OBJQUOTA:
	if (type == DATA_TYPE_UINT64) {
	if (parsable) {
	(void) sprintf(valstr, "%llu",
	(u_longlong_t)val64);
	strval = valstr;
	} else if (field == USFIELD_OBJQUOTA &&
	val64 == 0) {
	strval = "none";
	} else {
	zfs_nicenum(val64, valstr,
	sizeof (valstr));
	strval = valstr;
	}
	}
	break;
	}

	if (!first) {
	if (scripted)
	(void) printf("\t");
	else
	(void) printf(" ");
	}
	if (scripted)
	(void) printf("%s", strval);
	else if (field == USFIELD_TYPE \|\| field == USFIELD_NAME)
	(void) printf("%-*s", (int)width[field], strval);
	else
	(void) printf("%*s", (int)width[field], strval);

	first = B_FALSE;
	cfield++;
	}

	(void) printf("\n");
	}

	static void
	print_us(boolean_t scripted, boolean_t parsable, int *fields, int types,
	size_t width, boolean_t rmnode, uu_avl_t avl)
	{
	us_node_t *node;
	const char *col;
	int cfield = 0;
	int field;

	if (!scripted) {
	boolean_t first = B_TRUE;

	while ((field = fields[cfield]) != USFIELD_LAST) {
	col = gettext(us_field_hdr[field]);
	if (field == USFIELD_TYPE \|\| field == USFIELD_NAME) {
	(void) printf(first ? "%-s" : " %-s",
	(int)width[field], col);
	} else {
	(void) printf(first ? "%s" : " %s",
	(int)width[field], col);
	}
	first = B_FALSE;
	cfield++;
	}
	(void) printf("\n");
	}

	for (node = uu_avl_first(avl); node; node = uu_avl_next(avl, node)) {
	print_us_node(scripted, parsable, fields, types, width, node);
	if (rmnode)
	nvlist_free(node->usn_nvl);
	}
	}

	static int
	zfs_do_userspace(int argc, char **argv)
	{
	zfs_handle_t *zhp;
	zfs_userquota_prop_t p;
	uu_avl_pool_t *avl_pool;
	uu_avl_t *avl_tree;
	uu_avl_walk_t *walk;
	char *delim;
	char deffields[] = "type,name,used,quota,objused,objquota";
	char *ofield = NULL;
	char *tfield = NULL;
	int cfield = 0;
	int fields[256];
	int i;
	boolean_t scripted = B_FALSE;
	boolean_t prtnum = B_FALSE;
	boolean_t parsable = B_FALSE;
	boolean_t sid2posix = B_FALSE;
	int ret = 0;
	int c;
	zfs_sort_column_t *sortcol = NULL;
	int types = USTYPE_PSX_USR \| USTYPE_SMB_USR;
	us_cbdata_t cb;
	us_node_t *node;
	us_node_t *rmnode;
	uu_list_pool_t *listpool;
	uu_list_t *list;
	uu_avl_index_t idx = 0;
	uu_list_index_t idx2 = 0;

	if (argc < 2)
	usage(B_FALSE);

	if (strcmp(argv[0], "groupspace") == 0) {
	/* Toggle default group types */
	types = USTYPE_PSX_GRP \| USTYPE_SMB_GRP;
	} else if (strcmp(argv[0], "projectspace") == 0) {
	types = USTYPE_PROJ;
	prtnum = B_TRUE;
	}

	while ((c = getopt(argc, argv, "nHpo:s:S:t:i")) != -1) {
	switch (c) {
	case 'n':
	if (types == USTYPE_PROJ) {
	(void) fprintf(stderr,
	gettext("invalid option 'n'\n"));
	usage(B_FALSE);
	}
	prtnum = B_TRUE;
	break;
	case 'H':
	scripted = B_TRUE;
	break;
	case 'p':
	parsable = B_TRUE;
	break;
	case 'o':
	ofield = optarg;
	break;
	case 's':
	case 'S':
	if (zfs_add_sort_column(&sortcol, optarg,
	c == 's' ? B_FALSE : B_TRUE) != 0) {
	(void) fprintf(stderr,
	gettext("invalid field '%s'\n"), optarg);
	usage(B_FALSE);
	}
	break;
	case 't':
	if (types == USTYPE_PROJ) {
	(void) fprintf(stderr,
	gettext("invalid option 't'\n"));
	usage(B_FALSE);
	}
	tfield = optarg;
	break;
	case 'i':
	if (types == USTYPE_PROJ) {
	(void) fprintf(stderr,
	gettext("invalid option 'i'\n"));
	usage(B_FALSE);
	}
	sid2posix = B_TRUE;
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	usage(B_FALSE);
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing dataset name\n"));
	usage(B_FALSE);
	}
	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	/* Use default output fields if not specified using -o */
	if (ofield == NULL)
	ofield = deffields;
	do {
	if ((delim = strchr(ofield, ',')) != NULL)
	*delim = '\0';
	if ((fields[cfield++] = us_field_index(ofield)) == -1) {
	(void) fprintf(stderr, gettext("invalid type '%s' "
	"for -o option\n"), ofield);
	return (-1);
	}
	if (delim != NULL)
	ofield = delim + 1;
	} while (delim != NULL);
	fields[cfield] = USFIELD_LAST;

	/* Override output types (-t option) */
	if (tfield != NULL) {
	types = 0;

	do {
	boolean_t found = B_FALSE;

	if ((delim = strchr(tfield, ',')) != NULL)
	*delim = '\0';
	for (i = 0; i < sizeof (us_type_bits) / sizeof (int);
	i++) {
	if (strcmp(tfield, us_type_names[i]) == 0) {
	found = B_TRUE;
	types \|= us_type_bits[i];
	break;
	}
	}
	if (!found) {
	(void) fprintf(stderr, gettext("invalid type "
	"'%s' for -t option\n"), tfield);
	return (-1);
	}
	if (delim != NULL)
	tfield = delim + 1;
	} while (delim != NULL);
	}

	if ((zhp = zfs_path_to_zhandle(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM \|
	ZFS_TYPE_SNAPSHOT)) == NULL)
	return (1);
	if (zfs_get_underlying_type(zhp) != ZFS_TYPE_FILESYSTEM) {
	(void) fprintf(stderr, gettext("operation is only applicable "
	"to filesystems and their snapshots\n"));
	zfs_close(zhp);
	return (1);
	}

	if ((avl_pool = uu_avl_pool_create("us_avl_pool", sizeof (us_node_t),
	offsetof(us_node_t, usn_avlnode), us_compare, UU_DEFAULT)) == NULL)
	nomem();
	if ((avl_tree = uu_avl_create(avl_pool, NULL, UU_DEFAULT)) == NULL)
	nomem();

	/* Always add default sorting columns */
	(void) zfs_add_sort_column(&sortcol, "type", B_FALSE);
	(void) zfs_add_sort_column(&sortcol, "name", B_FALSE);

	cb.cb_sortcol = sortcol;
	cb.cb_numname = prtnum;
	cb.cb_nicenum = !parsable;
	cb.cb_avl_pool = avl_pool;
	cb.cb_avl = avl_tree;
	cb.cb_sid2posix = sid2posix;

	for (i = 0; i < USFIELD_LAST; i++)
	cb.cb_width[i] = strlen(gettext(us_field_hdr[i]));

	for (p = 0; p < ZFS_NUM_USERQUOTA_PROPS; p++) {
	if ((zfs_prop_is_user(p) &&
	!(types & (USTYPE_PSX_USR \| USTYPE_SMB_USR))) \|\|
	(zfs_prop_is_group(p) &&
	!(types & (USTYPE_PSX_GRP \| USTYPE_SMB_GRP))) \|\|
	(zfs_prop_is_project(p) && types != USTYPE_PROJ))
	continue;

	cb.cb_prop = p;
	if ((ret = zfs_userspace(zhp, p, userspace_cb, &cb)) != 0) {
	zfs_close(zhp);
	return (ret);
	}
	}
	zfs_close(zhp);

	/* Sort the list */
	if ((node = uu_avl_first(avl_tree)) == NULL)
	return (0);

	us_populated = B_TRUE;

	listpool = uu_list_pool_create("tmplist", sizeof (us_node_t),
	offsetof(us_node_t, usn_listnode), NULL, UU_DEFAULT);
	list = uu_list_create(listpool, NULL, UU_DEFAULT);
	uu_list_node_init(node, &node->usn_listnode, listpool);

	while (node != NULL) {
	rmnode = node;
	node = uu_avl_next(avl_tree, node);
	uu_avl_remove(avl_tree, rmnode);
	if (uu_list_find(list, rmnode, NULL, &idx2) == NULL)
	uu_list_insert(list, rmnode, idx2);
	}

	for (node = uu_list_first(list); node != NULL;
	node = uu_list_next(list, node)) {
	us_sort_info_t sortinfo = { sortcol, cb.cb_numname };

	if (uu_avl_find(avl_tree, node, &sortinfo, &idx) == NULL)
	uu_avl_insert(avl_tree, node, idx);
	}

	uu_list_destroy(list);
	uu_list_pool_destroy(listpool);

	/* Print and free node nvlist memory */
	print_us(scripted, parsable, fields, types, cb.cb_width, B_TRUE,
	cb.cb_avl);

	zfs_free_sort_columns(sortcol);

	/* Clean up the AVL tree */
	if ((walk = uu_avl_walk_start(cb.cb_avl, UU_WALK_ROBUST)) == NULL)
	nomem();

	while ((node = uu_avl_walk_next(walk)) != NULL) {
	uu_avl_remove(cb.cb_avl, node);
	free(node);
	}

	uu_avl_walk_end(walk);
	uu_avl_destroy(avl_tree);
	uu_avl_pool_destroy(avl_pool);

	return (ret);
	}

	/*
	* list [-Hp][-r\|-d max] [-o property[,...]] [-s property] ... [-S property]
	* [-t type[,...]] [filesystem\|volume\|snapshot] ...
	*
	* -H Scripted mode; elide headers and separate columns by tabs
	* -p Display values in parsable (literal) format.
	* -r Recurse over all children
	* -d Limit recursion by depth.
	* -o Control which fields to display.
	* -s Specify sort columns, descending order.
	* -S Specify sort columns, ascending order.
	* -t Control which object types to display.
	*
	* When given no arguments, list all filesystems in the system.
	* Otherwise, list the specified datasets, optionally recursing down them if
	* '-r' is specified.
	*/
	typedef struct list_cbdata {
	boolean_t cb_first;
	boolean_t cb_literal;
	boolean_t cb_scripted;
	zprop_list_t *cb_proplist;
	} list_cbdata_t;

	/*
	* Given a list of columns to display, output appropriate headers for each one.
	*/
	static void
	print_header(list_cbdata_t *cb)
	{
	zprop_list_t *pl = cb->cb_proplist;
	char headerbuf[ZFS_MAXPROPLEN];
	const char *header;
	int i;
	boolean_t first = B_TRUE;
	boolean_t right_justify;

	for (; pl != NULL; pl = pl->pl_next) {
	if (!first) {
	(void) printf(" ");
	} else {
	first = B_FALSE;
	}

	right_justify = B_FALSE;
	if (pl->pl_prop != ZPROP_INVAL) {
	header = zfs_prop_column_name(pl->pl_prop);
	right_justify = zfs_prop_align_right(pl->pl_prop);
	} else {
	for (i = 0; pl->pl_user_prop[i] != '\0'; i++)
	headerbuf[i] = toupper(pl->pl_user_prop[i]);
	headerbuf[i] = '\0';
	header = headerbuf;
	}

	if (pl->pl_next == NULL && !right_justify)
	(void) printf("%s", header);
	else if (right_justify)
	(void) printf("%*s", (int)pl->pl_width, header);
	else
	(void) printf("%-*s", (int)pl->pl_width, header);
	}

	(void) printf("\n");
	}

	/*
	* Given a dataset and a list of fields, print out all the properties according
	* to the described layout.
	*/
	static void
	print_dataset(zfs_handle_t zhp, list_cbdata_t cb)
	{
	zprop_list_t *pl = cb->cb_proplist;
	boolean_t first = B_TRUE;
	char property[ZFS_MAXPROPLEN];
	nvlist_t *userprops = zfs_get_user_props(zhp);
	nvlist_t *propval;
	char *propstr;
	boolean_t right_justify;

	for (; pl != NULL; pl = pl->pl_next) {
	if (!first) {
	if (cb->cb_scripted)
	(void) printf("\t");
	else
	(void) printf(" ");
	} else {
	first = B_FALSE;
	}

	if (pl->pl_prop == ZFS_PROP_NAME) {
	(void) strlcpy(property, zfs_get_name(zhp),
	sizeof (property));
	propstr = property;
	right_justify = zfs_prop_align_right(pl->pl_prop);
	} else if (pl->pl_prop != ZPROP_INVAL) {
	if (zfs_prop_get(zhp, pl->pl_prop, property,
	sizeof (property), NULL, NULL, 0,
	cb->cb_literal) != 0)
	propstr = "-";
	else
	propstr = property;
	right_justify = zfs_prop_align_right(pl->pl_prop);
	} else if (zfs_prop_userquota(pl->pl_user_prop)) {
	if (zfs_prop_get_userquota(zhp, pl->pl_user_prop,
	property, sizeof (property), cb->cb_literal) != 0)
	propstr = "-";
	else
	propstr = property;
	right_justify = B_TRUE;
	} else if (zfs_prop_written(pl->pl_user_prop)) {
	if (zfs_prop_get_written(zhp, pl->pl_user_prop,
	property, sizeof (property), cb->cb_literal) != 0)
	propstr = "-";
	else
	propstr = property;
	right_justify = B_TRUE;
	} else {
	if (nvlist_lookup_nvlist(userprops,
	pl->pl_user_prop, &propval) != 0)
	propstr = "-";
	else
	verify(nvlist_lookup_string(propval,
	ZPROP_VALUE, &propstr) == 0);
	right_justify = B_FALSE;
	}

	/*
	* If this is being called in scripted mode, or if this is the
	* last column and it is left-justified, don't include a width
	* format specifier.
	*/
	if (cb->cb_scripted \|\| (pl->pl_next == NULL && !right_justify))
	(void) printf("%s", propstr);
	else if (right_justify)
	(void) printf("%*s", (int)pl->pl_width, propstr);
	else
	(void) printf("%-*s", (int)pl->pl_width, propstr);
	}

	(void) printf("\n");
	}

	/*
	* Generic callback function to list a dataset or snapshot.
	*/
	static int
	list_callback(zfs_handle_t zhp, void data)
	{
	list_cbdata_t *cbp = data;

	if (cbp->cb_first) {
	if (!cbp->cb_scripted)
	print_header(cbp);
	cbp->cb_first = B_FALSE;
	}

	print_dataset(zhp, cbp);

	return (0);
	}

	static int
	zfs_do_list(int argc, char **argv)
	{
	int c;
	static char default_fields[] =
	"name,used,available,referenced,mountpoint";
	int types = ZFS_TYPE_DATASET;
	boolean_t types_specified = B_FALSE;
	char *fields = NULL;
	list_cbdata_t cb = { 0 };
	char *value;
	int limit = 0;
	int ret = 0;
	zfs_sort_column_t *sortcol = NULL;
	int flags = ZFS_ITER_PROP_LISTSNAPS \| ZFS_ITER_ARGS_CAN_BE_PATHS;

	/* check options */
	while ((c = getopt(argc, argv, "HS:d:o:prs:t:")) != -1) {
	switch (c) {
	case 'o':
	fields = optarg;
	break;
	case 'p':
	cb.cb_literal = B_TRUE;
	flags \|= ZFS_ITER_LITERAL_PROPS;
	break;
	case 'd':
	limit = parse_depth(optarg, &flags);
	break;
	case 'r':
	flags \|= ZFS_ITER_RECURSE;
	break;
	case 'H':
	cb.cb_scripted = B_TRUE;
	break;
	case 's':
	if (zfs_add_sort_column(&sortcol, optarg,
	B_FALSE) != 0) {
	(void) fprintf(stderr,
	gettext("invalid property '%s'\n"), optarg);
	usage(B_FALSE);
	}
	break;
	case 'S':
	if (zfs_add_sort_column(&sortcol, optarg,
	B_TRUE) != 0) {
	(void) fprintf(stderr,
	gettext("invalid property '%s'\n"), optarg);
	usage(B_FALSE);
	}
	break;
	case 't':
	types = 0;
	types_specified = B_TRUE;
	flags &= ~ZFS_ITER_PROP_LISTSNAPS;
	while (*optarg != '\0') {
	static char *type_subopts[] = { "filesystem",
	"volume", "snapshot", "snap", "bookmark",
	"all", NULL };

	switch (getsubopt(&optarg, type_subopts,
	&value)) {
	case 0:
	types \|= ZFS_TYPE_FILESYSTEM;
	break;
	case 1:
	types \|= ZFS_TYPE_VOLUME;
	break;
	case 2:
	case 3:
	types \|= ZFS_TYPE_SNAPSHOT;
	break;
	case 4:
	types \|= ZFS_TYPE_BOOKMARK;
	break;
	case 5:
	types = ZFS_TYPE_DATASET \|
	ZFS_TYPE_BOOKMARK;
	break;
	default:
	(void) fprintf(stderr,
	gettext("invalid type '%s'\n"),
	value);
	usage(B_FALSE);
	}
	}
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	usage(B_FALSE);
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (fields == NULL)
	fields = default_fields;

	/*
	* If we are only going to list snapshot names and sort by name,
	* then we can use faster version.
	*/
	if (strcmp(fields, "name") == 0 && zfs_sort_only_by_name(sortcol))
	flags \|= ZFS_ITER_SIMPLE;

	/*
	* If "-o space" and no types were specified, don't display snapshots.
	*/
	if (strcmp(fields, "space") == 0 && types_specified == B_FALSE)
	types &= ~ZFS_TYPE_SNAPSHOT;

	/*
	* Handle users who want to list all snapshots or bookmarks
	* of the current dataset (ex. 'zfs list -t snapshot <dataset>').
	*/
	if ((types == ZFS_TYPE_SNAPSHOT \|\| types == ZFS_TYPE_BOOKMARK) &&
	argc > 0 && (flags & ZFS_ITER_RECURSE) == 0 && limit == 0) {
	flags \|= (ZFS_ITER_DEPTH_LIMIT \| ZFS_ITER_RECURSE);
	limit = 1;
	}

	/*
	* If the user specifies '-o all', the zprop_get_list() doesn't
	* normally include the name of the dataset. For 'zfs list', we always
	* want this property to be first.
	*/
	if (zprop_get_list(g_zfs, fields, &cb.cb_proplist, ZFS_TYPE_DATASET)
	!= 0)
	usage(B_FALSE);

	cb.cb_first = B_TRUE;

	ret = zfs_for_each(argc, argv, flags, types, sortcol, &cb.cb_proplist,
	limit, list_callback, &cb);

	zprop_free_list(cb.cb_proplist);
	zfs_free_sort_columns(sortcol);

	if (ret == 0 && cb.cb_first && !cb.cb_scripted)
	(void) fprintf(stderr, gettext("no datasets available\n"));

	return (ret);
	}

	/*
	* zfs rename [-fu] <fs \| snap \| vol> <fs \| snap \| vol>
	* zfs rename [-f] -p <fs \| vol> <fs \| vol>
	* zfs rename [-u] -r <snap> <snap>
	*
	* Renames the given dataset to another of the same type.
	*
	* The '-p' flag creates all the non-existing ancestors of the target first.
	* The '-u' flag prevents file systems from being remounted during rename.
	*/
	-/* ARGSUSED */
	static int
	zfs_do_rename(int argc, char **argv)
	{
	zfs_handle_t *zhp;
	renameflags_t flags = { 0 };
	int c;
	int ret = 0;
	int types;
	boolean_t parents = B_FALSE;

	/* check options */
	while ((c = getopt(argc, argv, "pruf")) != -1) {
	switch (c) {
	case 'p':
	parents = B_TRUE;
	break;
	case 'r':
	flags.recursive = B_TRUE;
	break;
	case 'u':
	flags.nounmount = B_TRUE;
	break;
	case 'f':
	flags.forceunmount = B_TRUE;
	break;
	case '?':
	default:
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing source dataset "
	"argument\n"));
	usage(B_FALSE);
	}
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing target dataset "
	"argument\n"));
	usage(B_FALSE);
	}
	if (argc > 2) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	if (flags.recursive && parents) {
	(void) fprintf(stderr, gettext("-p and -r options are mutually "
	"exclusive\n"));
	usage(B_FALSE);
	}

	if (flags.nounmount && parents) {
	(void) fprintf(stderr, gettext("-u and -p options are mutually "
	"exclusive\n"));
	usage(B_FALSE);
	}

	if (flags.recursive && strchr(argv[0], '@') == 0) {
	(void) fprintf(stderr, gettext("source dataset for recursive "
	"rename must be a snapshot\n"));
	usage(B_FALSE);
	}

	if (flags.nounmount)
	types = ZFS_TYPE_FILESYSTEM;
	else if (parents)
	types = ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME;
	else
	types = ZFS_TYPE_DATASET;

	if ((zhp = zfs_open(g_zfs, argv[0], types)) == NULL)
	return (1);

	/* If we were asked and the name looks good, try to create ancestors. */
	if (parents && zfs_name_valid(argv[1], zfs_get_type(zhp)) &&
	zfs_create_ancestors(g_zfs, argv[1]) != 0) {
	zfs_close(zhp);
	return (1);
	}

	ret = (zfs_rename(zhp, argv[1], flags) != 0);

	zfs_close(zhp);
	return (ret);
	}

	/*
	* zfs promote <fs>
	*
	* Promotes the given clone fs to be the parent
	*/
	-/* ARGSUSED */
	static int
	zfs_do_promote(int argc, char **argv)
	{
	zfs_handle_t *zhp;
	int ret = 0;

	/* check options */
	if (argc > 1 && argv[1][0] == '-') {
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	argv[1][1]);
	usage(B_FALSE);
	}

	/* check number of arguments */
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing clone filesystem"
	" argument\n"));
	usage(B_FALSE);
	}
	if (argc > 2) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	zhp = zfs_open(g_zfs, argv[1], ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME);
	if (zhp == NULL)
	return (1);

	ret = (zfs_promote(zhp) != 0);


	zfs_close(zhp);
	return (ret);
	}

	static int
	zfs_do_redact(int argc, char **argv)
	{
	char *snap = NULL;
	char *bookname = NULL;
	char **rsnaps = NULL;
	int numrsnaps = 0;
	argv++;
	argc--;
	if (argc < 3) {
	(void) fprintf(stderr, gettext("too few arguments\n"));
	usage(B_FALSE);
	}

	snap = argv[0];
	bookname = argv[1];
	rsnaps = argv + 2;
	numrsnaps = argc - 2;

	nvlist_t *rsnapnv = fnvlist_alloc();

	for (int i = 0; i < numrsnaps; i++) {
	fnvlist_add_boolean(rsnapnv, rsnaps[i]);
	}

	int err = lzc_redact(snap, bookname, rsnapnv);
	fnvlist_free(rsnapnv);

	switch (err) {
	case 0:
	break;
	case ENOENT:
	(void) fprintf(stderr,
	gettext("provided snapshot %s does not exist\n"), snap);
	break;
	case EEXIST:
	(void) fprintf(stderr, gettext("specified redaction bookmark "
	"(%s) provided already exists\n"), bookname);
	break;
	case ENAMETOOLONG:
	(void) fprintf(stderr, gettext("provided bookmark name cannot "
	"be used, final name would be too long\n"));
	break;
	case E2BIG:
	(void) fprintf(stderr, gettext("too many redaction snapshots "
	"specified\n"));
	break;
	case EINVAL:
	if (strchr(bookname, '#') != NULL)
	(void) fprintf(stderr, gettext(
	"redaction bookmark name must not contain '#'\n"));
	else
	(void) fprintf(stderr, gettext(
	"redaction snapshot must be descendent of "
	"snapshot being redacted\n"));
	break;
	case EALREADY:
	(void) fprintf(stderr, gettext("attempted to redact redacted "
	"dataset or with respect to redacted dataset\n"));
	break;
	case ENOTSUP:
	(void) fprintf(stderr, gettext("redaction bookmarks feature "
	"not enabled\n"));
	break;
	case EXDEV:
	(void) fprintf(stderr, gettext("potentially invalid redaction "
	"snapshot; full dataset names required\n"));
	break;
	default:
	(void) fprintf(stderr, gettext("internal error: %s\n"),
	strerror(errno));
	}

	return (err);
	}

	/*
	* zfs rollback [-rRf] <snapshot>
	*
	* -r Delete any intervening snapshots before doing rollback
	* -R Delete any snapshots and their clones
	* -f ignored for backwards compatibility
	*
	* Given a filesystem, rollback to a specific snapshot, discarding any changes
	* since then and making it the active dataset. If more recent snapshots exist,
	* the command will complain unless the '-r' flag is given.
	*/
	typedef struct rollback_cbdata {
	uint64_t cb_create;
	uint8_t cb_younger_ds_printed;
	boolean_t cb_first;
	int cb_doclones;
	char *cb_target;
	int cb_error;
	boolean_t cb_recurse;
	} rollback_cbdata_t;

	static int
	rollback_check_dependent(zfs_handle_t zhp, void data)
	{
	rollback_cbdata_t *cbp = data;

	if (cbp->cb_first && cbp->cb_recurse) {
	(void) fprintf(stderr, gettext("cannot rollback to "
	"'%s': clones of previous snapshots exist\n"),
	cbp->cb_target);
	(void) fprintf(stderr, gettext("use '-R' to "
	"force deletion of the following clones and "
	"dependents:\n"));
	cbp->cb_first = 0;
	cbp->cb_error = 1;
	}

	(void) fprintf(stderr, "%s\n", zfs_get_name(zhp));

	zfs_close(zhp);
	return (0);
	}


	/*
	* Report some snapshots/bookmarks more recent than the one specified.
	* Used when '-r' is not specified. We reuse this same callback for the
	* snapshot dependents - if 'cb_dependent' is set, then this is a
	* dependent and we should report it without checking the transaction group.
	*/
	static int
	rollback_check(zfs_handle_t zhp, void data)
	{
	rollback_cbdata_t *cbp = data;
	/*
	* Max number of younger snapshots and/or bookmarks to display before
	* we stop the iteration.
	*/
	const uint8_t max_younger = 32;

	if (cbp->cb_doclones) {
	zfs_close(zhp);
	return (0);
	}

	if (zfs_prop_get_int(zhp, ZFS_PROP_CREATETXG) > cbp->cb_create) {
	if (cbp->cb_first && !cbp->cb_recurse) {
	(void) fprintf(stderr, gettext("cannot "
	"rollback to '%s': more recent snapshots "
	"or bookmarks exist\n"),
	cbp->cb_target);
	(void) fprintf(stderr, gettext("use '-r' to "
	"force deletion of the following "
	"snapshots and bookmarks:\n"));
	cbp->cb_first = 0;
	cbp->cb_error = 1;
	}

	if (cbp->cb_recurse) {
	if (zfs_iter_dependents(zhp, B_TRUE,
	rollback_check_dependent, cbp) != 0) {
	zfs_close(zhp);
	return (-1);
	}
	} else {
	(void) fprintf(stderr, "%s\n",
	zfs_get_name(zhp));
	cbp->cb_younger_ds_printed++;
	}
	}
	zfs_close(zhp);

	if (cbp->cb_younger_ds_printed == max_younger) {
	/*
	* This non-recursive rollback is going to fail due to the
	* presence of snapshots and/or bookmarks that are younger than
	* the rollback target.
	* We printed some of the offending objects, now we stop
	* zfs_iter_snapshot/bookmark iteration so we can fail fast and
	* avoid iterating over the rest of the younger objects
	*/
	(void) fprintf(stderr, gettext("Output limited to %d "
	"snapshots/bookmarks\n"), max_younger);
	return (-1);
	}
	return (0);
	}

	static int
	zfs_do_rollback(int argc, char **argv)
	{
	int ret = 0;
	int c;
	boolean_t force = B_FALSE;
	rollback_cbdata_t cb = { 0 };
	zfs_handle_t zhp, snap;
	char parentname[ZFS_MAX_DATASET_NAME_LEN];
	char *delim;
	uint64_t min_txg = 0;

	/* check options */
	while ((c = getopt(argc, argv, "rRf")) != -1) {
	switch (c) {
	case 'r':
	cb.cb_recurse = 1;
	break;
	case 'R':
	cb.cb_recurse = 1;
	cb.cb_doclones = 1;
	break;
	case 'f':
	force = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing dataset argument\n"));
	usage(B_FALSE);
	}
	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	/* open the snapshot */
	if ((snap = zfs_open(g_zfs, argv[0], ZFS_TYPE_SNAPSHOT)) == NULL)
	return (1);

	/* open the parent dataset */
	(void) strlcpy(parentname, argv[0], sizeof (parentname));
	verify((delim = strrchr(parentname, '@')) != NULL);
	*delim = '\0';
	if ((zhp = zfs_open(g_zfs, parentname, ZFS_TYPE_DATASET)) == NULL) {
	zfs_close(snap);
	return (1);
	}

	/*
	* Check for more recent snapshots and/or clones based on the presence
	* of '-r' and '-R'.
	*/
	cb.cb_target = argv[0];
	cb.cb_create = zfs_prop_get_int(snap, ZFS_PROP_CREATETXG);
	cb.cb_first = B_TRUE;
	cb.cb_error = 0;

	if (cb.cb_create > 0)
	min_txg = cb.cb_create;

	if ((ret = zfs_iter_snapshots(zhp, B_FALSE, rollback_check, &cb,
	min_txg, 0)) != 0)
	goto out;
	if ((ret = zfs_iter_bookmarks(zhp, rollback_check, &cb)) != 0)
	goto out;

	if ((ret = cb.cb_error) != 0)
	goto out;

	/*
	* Rollback parent to the given snapshot.
	*/
	ret = zfs_rollback(zhp, snap, force);

	out:
	zfs_close(snap);
	zfs_close(zhp);

	if (ret == 0)
	return (0);
	else
	return (1);
	}

	/*
	* zfs set property=value ... { fs \| snap \| vol } ...
	*
	* Sets the given properties for all datasets specified on the command line.
	*/

	static int
	set_callback(zfs_handle_t zhp, void data)
	{
	nvlist_t *props = data;

	if (zfs_prop_set_list(zhp, props) != 0) {
	switch (libzfs_errno(g_zfs)) {
	case EZFS_MOUNTFAILED:
	(void) fprintf(stderr, gettext("property may be set "
	"but unable to remount filesystem\n"));
	break;
	case EZFS_SHARENFSFAILED:
	(void) fprintf(stderr, gettext("property may be set "
	"but unable to reshare filesystem\n"));
	break;
	}
	return (1);
	}
	return (0);
	}

	static int
	zfs_do_set(int argc, char **argv)
	{
	nvlist_t *props = NULL;
	int ds_start = -1; /* argv idx of first dataset arg */
	int ret = 0;
	int i;

	/* check for options */
	if (argc > 1 && argv[1][0] == '-') {
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	argv[1][1]);
	usage(B_FALSE);
	}

	/* check number of arguments */
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing arguments\n"));
	usage(B_FALSE);
	}
	if (argc < 3) {
	if (strchr(argv[1], '=') == NULL) {
	(void) fprintf(stderr, gettext("missing property=value "
	"argument(s)\n"));
	} else {
	(void) fprintf(stderr, gettext("missing dataset "
	"name(s)\n"));
	}
	usage(B_FALSE);
	}

	/* validate argument order: prop=val args followed by dataset args */
	for (i = 1; i < argc; i++) {
	if (strchr(argv[i], '=') != NULL) {
	if (ds_start > 0) {
	/* out-of-order prop=val argument */
	(void) fprintf(stderr, gettext("invalid "
	"argument order\n"));
	usage(B_FALSE);
	}
	} else if (ds_start < 0) {
	ds_start = i;
	}
	}
	if (ds_start < 0) {
	(void) fprintf(stderr, gettext("missing dataset name(s)\n"));
	usage(B_FALSE);
	}

	/* Populate a list of property settings */
	if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0)
	nomem();
	for (i = 1; i < ds_start; i++) {
	if (!parseprop(props, argv[i])) {
	ret = -1;
	goto error;
	}
	}

	ret = zfs_for_each(argc - ds_start, argv + ds_start, 0,
	ZFS_TYPE_DATASET, NULL, NULL, 0, set_callback, props);

	error:
	nvlist_free(props);
	return (ret);
	}

	typedef struct snap_cbdata {
	nvlist_t *sd_nvl;
	boolean_t sd_recursive;
	const char *sd_snapname;
	} snap_cbdata_t;

	static int
	zfs_snapshot_cb(zfs_handle_t zhp, void arg)
	{
	snap_cbdata_t *sd = arg;
	char *name;
	int rv = 0;
	int error;

	if (sd->sd_recursive &&
	zfs_prop_get_int(zhp, ZFS_PROP_INCONSISTENT) != 0) {
	zfs_close(zhp);
	return (0);
	}

	error = asprintf(&name, "%s@%s", zfs_get_name(zhp), sd->sd_snapname);
	if (error == -1)
	nomem();
	fnvlist_add_boolean(sd->sd_nvl, name);
	free(name);

	if (sd->sd_recursive)
	rv = zfs_iter_filesystems(zhp, zfs_snapshot_cb, sd);
	zfs_close(zhp);
	return (rv);
	}

	/*
	* zfs snapshot [-r] [-o prop=value] ... <fs@snap>
	*
	* Creates a snapshot with the given name. While functionally equivalent to
	* 'zfs create', it is a separate command to differentiate intent.
	*/
	static int
	zfs_do_snapshot(int argc, char **argv)
	{
	int ret = 0;
	int c;
	nvlist_t *props;
	snap_cbdata_t sd = { 0 };
	boolean_t multiple_snaps = B_FALSE;

	if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0)
	nomem();
	if (nvlist_alloc(&sd.sd_nvl, NV_UNIQUE_NAME, 0) != 0)
	nomem();

	/* check options */
	while ((c = getopt(argc, argv, "ro:")) != -1) {
	switch (c) {
	case 'o':
	if (!parseprop(props, optarg)) {
	nvlist_free(sd.sd_nvl);
	nvlist_free(props);
	return (1);
	}
	break;
	case 'r':
	sd.sd_recursive = B_TRUE;
	multiple_snaps = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	goto usage;
	}
	}

	argc -= optind;
	argv += optind;

	/* check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing snapshot argument\n"));
	goto usage;
	}

	if (argc > 1)
	multiple_snaps = B_TRUE;
	for (; argc > 0; argc--, argv++) {
	char *atp;
	zfs_handle_t *zhp;

	atp = strchr(argv[0], '@');
	if (atp == NULL)
	goto usage;
	*atp = '\0';
	sd.sd_snapname = atp + 1;
	zhp = zfs_open(g_zfs, argv[0],
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME);
	if (zhp == NULL)
	goto usage;
	if (zfs_snapshot_cb(zhp, &sd) != 0)
	goto usage;
	}

	ret = zfs_snapshot_nvl(g_zfs, sd.sd_nvl, props);
	nvlist_free(sd.sd_nvl);
	nvlist_free(props);
	if (ret != 0 && multiple_snaps)
	(void) fprintf(stderr, gettext("no snapshots were created\n"));
	return (ret != 0);

	usage:
	nvlist_free(sd.sd_nvl);
	nvlist_free(props);
	usage(B_FALSE);
	return (-1);
	}


	/*
	* Send a backup stream to stdout.
	*/
	static int
	zfs_do_send(int argc, char **argv)
	{
	char *fromname = NULL;
	char *toname = NULL;
	char *resume_token = NULL;
	char *cp;
	zfs_handle_t *zhp;
	sendflags_t flags = { 0 };
	int c, err;
	nvlist_t *dbgnv = NULL;
	char *redactbook = NULL;

	struct option long_options[] = {
	{"replicate", no_argument, NULL, 'R'},
	{"skip-missing", no_argument, NULL, 's'},
	{"redact", required_argument, NULL, 'd'},
	{"props", no_argument, NULL, 'p'},
	{"parsable", no_argument, NULL, 'P'},
	{"dedup", no_argument, NULL, 'D'},
	{"verbose", no_argument, NULL, 'v'},
	{"dryrun", no_argument, NULL, 'n'},
	{"large-block", no_argument, NULL, 'L'},
	{"embed", no_argument, NULL, 'e'},
	{"resume", required_argument, NULL, 't'},
	{"compressed", no_argument, NULL, 'c'},
	{"raw", no_argument, NULL, 'w'},
	{"backup", no_argument, NULL, 'b'},
	{"holds", no_argument, NULL, 'h'},
	{"saved", no_argument, NULL, 'S'},
	{0, 0, 0, 0}
	};

	/* check options */
	while ((c = getopt_long(argc, argv, ":i:I:RsDpvnPLeht:cwbd:S",
	long_options, NULL)) != -1) {
	switch (c) {
	case 'i':
	if (fromname)
	usage(B_FALSE);
	fromname = optarg;
	break;
	case 'I':
	if (fromname)
	usage(B_FALSE);
	fromname = optarg;
	flags.doall = B_TRUE;
	break;
	case 'R':
	flags.replicate = B_TRUE;
	break;
	case 's':
	flags.skipmissing = B_TRUE;
	break;
	case 'd':
	redactbook = optarg;
	break;
	case 'p':
	flags.props = B_TRUE;
	break;
	case 'b':
	flags.backup = B_TRUE;
	break;
	case 'h':
	flags.holds = B_TRUE;
	break;
	case 'P':
	flags.parsable = B_TRUE;
	break;
	case 'v':
	flags.verbosity++;
	flags.progress = B_TRUE;
	break;
	case 'D':
	(void) fprintf(stderr,
	gettext("WARNING: deduplicated send is no "
	"longer supported. A regular,\n"
	"non-deduplicated stream will be generated.\n\n"));
	break;
	case 'n':
	flags.dryrun = B_TRUE;
	break;
	case 'L':
	flags.largeblock = B_TRUE;
	break;
	case 'e':
	flags.embed_data = B_TRUE;
	break;
	case 't':
	resume_token = optarg;
	break;
	case 'c':
	flags.compress = B_TRUE;
	break;
	case 'w':
	flags.raw = B_TRUE;
	flags.compress = B_TRUE;
	flags.embed_data = B_TRUE;
	flags.largeblock = B_TRUE;
	break;
	case 'S':
	flags.saved = B_TRUE;
	break;
	case ':':
	/*
	* If a parameter was not passed, optopt contains the
	* value that would normally lead us into the
	* appropriate case statement. If it's > 256, then this
	* must be a longopt and we should look at argv to get
	* the string. Otherwise it's just the character, so we
	* should use it directly.
	*/
	if (optopt <= UINT8_MAX) {
	(void) fprintf(stderr,
	gettext("missing argument for '%c' "
	"option\n"), optopt);
	} else {
	(void) fprintf(stderr,
	gettext("missing argument for '%s' "
	"option\n"), argv[optind - 1]);
	}
	usage(B_FALSE);
	break;
	case '?':
	default:
	/*
	* If an invalid flag was passed, optopt contains the
	* character if it was a short flag, or 0 if it was a
	* longopt.
	*/
	if (optopt != 0) {
	(void) fprintf(stderr,
	gettext("invalid option '%c'\n"), optopt);
	} else {
	(void) fprintf(stderr,
	gettext("invalid option '%s'\n"),
	argv[optind - 1]);

	}
	usage(B_FALSE);
	}
	}

	if (flags.parsable && flags.verbosity == 0)
	flags.verbosity = 1;

	argc -= optind;
	argv += optind;

	if (resume_token != NULL) {
	if (fromname != NULL \|\| flags.replicate \|\| flags.props \|\|
	flags.backup \|\| flags.holds \|\|
	flags.saved \|\| redactbook != NULL) {
	(void) fprintf(stderr,
	gettext("invalid flags combined with -t\n"));
	usage(B_FALSE);
	}
	if (argc > 0) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}
	} else {
	if (argc < 1) {
	(void) fprintf(stderr,
	gettext("missing snapshot argument\n"));
	usage(B_FALSE);
	}
	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}
	}

	if (flags.saved) {
	if (fromname != NULL \|\| flags.replicate \|\| flags.props \|\|
	flags.doall \|\| flags.backup \|\|
	flags.holds \|\| flags.largeblock \|\| flags.embed_data \|\|
	flags.compress \|\| flags.raw \|\| redactbook != NULL) {
	(void) fprintf(stderr, gettext("incompatible flags "
	"combined with saved send flag\n"));
	usage(B_FALSE);
	}
	if (strchr(argv[0], '@') != NULL) {
	(void) fprintf(stderr, gettext("saved send must "
	"specify the dataset with partially-received "
	"state\n"));
	usage(B_FALSE);
	}
	}

	if (flags.raw && redactbook != NULL) {
	(void) fprintf(stderr,
	gettext("Error: raw sends may not be redacted.\n"));
	return (1);
	}

	if (!flags.dryrun && isatty(STDOUT_FILENO)) {
	(void) fprintf(stderr,
	gettext("Error: Stream can not be written to a terminal.\n"
	"You must redirect standard output.\n"));
	return (1);
	}

	if (flags.saved) {
	zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_DATASET);
	if (zhp == NULL)
	return (1);

	err = zfs_send_saved(zhp, &flags, STDOUT_FILENO,
	resume_token);
	if (err != 0)
	note_dev_error(errno, STDOUT_FILENO);
	zfs_close(zhp);
	return (err != 0);
	} else if (resume_token != NULL) {
	err = zfs_send_resume(g_zfs, &flags, STDOUT_FILENO,
	resume_token);
	if (err != 0)
	note_dev_error(errno, STDOUT_FILENO);
	return (err);
	}

	if (flags.skipmissing && !flags.replicate) {
	(void) fprintf(stderr,
	gettext("skip-missing flag can only be used in "
	"conjunction with replicate\n"));
	usage(B_FALSE);
	}

	/*
	* For everything except -R and -I, use the new, cleaner code path.
	*/
	if (!(flags.replicate \|\| flags.doall)) {
	char frombuf[ZFS_MAX_DATASET_NAME_LEN];

	if (fromname != NULL && (strchr(fromname, '#') == NULL &&
	strchr(fromname, '@') == NULL)) {
	/*
	* Neither bookmark or snapshot was specified. Print a
	* warning, and assume snapshot.
	*/
	(void) fprintf(stderr, "Warning: incremental source "
	"didn't specify type, assuming snapshot. Use '@' "
	"or '#' prefix to avoid ambiguity.\n");
	(void) snprintf(frombuf, sizeof (frombuf), "@%s",
	fromname);
	fromname = frombuf;
	}
	if (fromname != NULL &&
	(fromname[0] == '#' \|\| fromname[0] == '@')) {
	/*
	* Incremental source name begins with # or @.
	* Default to same fs as target.
	*/
	char tmpbuf[ZFS_MAX_DATASET_NAME_LEN];
	(void) strlcpy(tmpbuf, fromname, sizeof (tmpbuf));
	(void) strlcpy(frombuf, argv[0], sizeof (frombuf));
	cp = strchr(frombuf, '@');
	if (cp != NULL)
	*cp = '\0';
	(void) strlcat(frombuf, tmpbuf, sizeof (frombuf));
	fromname = frombuf;
	}

	zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_DATASET);
	if (zhp == NULL)
	return (1);
	err = zfs_send_one(zhp, fromname, STDOUT_FILENO, &flags,
	redactbook);
	zfs_close(zhp);
	if (err != 0)
	note_dev_error(errno, STDOUT_FILENO);
	return (err != 0);
	}

	if (fromname != NULL && strchr(fromname, '#')) {
	(void) fprintf(stderr,
	gettext("Error: multiple snapshots cannot be "
	"sent from a bookmark.\n"));
	return (1);
	}

	if (redactbook != NULL) {
	(void) fprintf(stderr, gettext("Error: multiple snapshots "
	"cannot be sent redacted.\n"));
	return (1);
	}

	if ((cp = strchr(argv[0], '@')) == NULL) {
	(void) fprintf(stderr, gettext("Error: "
	"Unsupported flag with filesystem or bookmark.\n"));
	return (1);
	}
	*cp = '\0';
	toname = cp + 1;
	zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME);
	if (zhp == NULL)
	return (1);

	/*
	* If they specified the full path to the snapshot, chop off
	* everything except the short name of the snapshot, but special
	* case if they specify the origin.
	*/
	if (fromname && (cp = strchr(fromname, '@')) != NULL) {
	char origin[ZFS_MAX_DATASET_NAME_LEN];
	zprop_source_t src;

	(void) zfs_prop_get(zhp, ZFS_PROP_ORIGIN,
	origin, sizeof (origin), &src, NULL, 0, B_FALSE);

	if (strcmp(origin, fromname) == 0) {
	fromname = NULL;
	flags.fromorigin = B_TRUE;
	} else {
	*cp = '\0';
	if (cp != fromname && strcmp(argv[0], fromname)) {
	(void) fprintf(stderr,
	gettext("incremental source must be "
	"in same filesystem\n"));
	usage(B_FALSE);
	}
	fromname = cp + 1;
	if (strchr(fromname, '@') \|\| strchr(fromname, '/')) {
	(void) fprintf(stderr,
	gettext("invalid incremental source\n"));
	usage(B_FALSE);
	}
	}
	}

	if (flags.replicate && fromname == NULL)
	flags.doall = B_TRUE;

	err = zfs_send(zhp, fromname, toname, &flags, STDOUT_FILENO, NULL, 0,
	flags.verbosity >= 3 ? &dbgnv : NULL);

	if (flags.verbosity >= 3 && dbgnv != NULL) {
	/*
	* dump_nvlist prints to stdout, but that's been
	* redirected to a file. Make it print to stderr
	* instead.
	*/
	(void) dup2(STDERR_FILENO, STDOUT_FILENO);
	dump_nvlist(dbgnv, 0);
	nvlist_free(dbgnv);
	}
	zfs_close(zhp);
	note_dev_error(errno, STDOUT_FILENO);

	return (err != 0);
	}

	/*
	* Restore a backup stream from stdin.
	*/
	static int
	zfs_do_receive(int argc, char **argv)
	{
	int c, err = 0;
	recvflags_t flags = { 0 };
	boolean_t abort_resumable = B_FALSE;
	nvlist_t *props;

	if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0)
	nomem();

	/* check options */
	while ((c = getopt(argc, argv, ":o:x:dehMnuvFsA")) != -1) {
	switch (c) {
	case 'o':
	if (!parseprop(props, optarg)) {
	nvlist_free(props);
	usage(B_FALSE);
	}
	break;
	case 'x':
	if (!parsepropname(props, optarg)) {
	nvlist_free(props);
	usage(B_FALSE);
	}
	break;
	case 'd':
	if (flags.istail) {
	(void) fprintf(stderr, gettext("invalid option "
	"combination: -d and -e are mutually "
	"exclusive\n"));
	usage(B_FALSE);
	}
	flags.isprefix = B_TRUE;
	break;
	case 'e':
	if (flags.isprefix) {
	(void) fprintf(stderr, gettext("invalid option "
	"combination: -d and -e are mutually "
	"exclusive\n"));
	usage(B_FALSE);
	}
	flags.istail = B_TRUE;
	break;
	case 'h':
	flags.skipholds = B_TRUE;
	break;
	case 'M':
	flags.forceunmount = B_TRUE;
	break;
	case 'n':
	flags.dryrun = B_TRUE;
	break;
	case 'u':
	flags.nomount = B_TRUE;
	break;
	case 'v':
	flags.verbose = B_TRUE;
	break;
	case 's':
	flags.resumable = B_TRUE;
	break;
	case 'F':
	flags.force = B_TRUE;
	break;
	case 'A':
	abort_resumable = B_TRUE;
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	usage(B_FALSE);
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* zfs recv -e (use "tail" name) implies -d (remove dataset "head") */
	if (flags.istail)
	flags.isprefix = B_TRUE;

	/* check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing snapshot argument\n"));
	usage(B_FALSE);
	}
	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	if (abort_resumable) {
	if (flags.isprefix \|\| flags.istail \|\| flags.dryrun \|\|
	flags.resumable \|\| flags.nomount) {
	(void) fprintf(stderr, gettext("invalid option\n"));
	usage(B_FALSE);
	}

	char namebuf[ZFS_MAX_DATASET_NAME_LEN];
	(void) snprintf(namebuf, sizeof (namebuf),
	"%s/%%recv", argv[0]);

	if (zfs_dataset_exists(g_zfs, namebuf,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME)) {
	zfs_handle_t *zhp = zfs_open(g_zfs,
	namebuf, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME);
	if (zhp == NULL) {
	nvlist_free(props);
	return (1);
	}
	err = zfs_destroy(zhp, B_FALSE);
	zfs_close(zhp);
	} else {
	zfs_handle_t *zhp = zfs_open(g_zfs,
	argv[0], ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME);
	if (zhp == NULL)
	usage(B_FALSE);
	if (!zfs_prop_get_int(zhp, ZFS_PROP_INCONSISTENT) \|\|
	zfs_prop_get(zhp, ZFS_PROP_RECEIVE_RESUME_TOKEN,
	NULL, 0, NULL, NULL, 0, B_TRUE) == -1) {
	(void) fprintf(stderr,
	gettext("'%s' does not have any "
	"resumable receive state to abort\n"),
	argv[0]);
	nvlist_free(props);
	zfs_close(zhp);
	return (1);
	}
	err = zfs_destroy(zhp, B_FALSE);
	zfs_close(zhp);
	}
	nvlist_free(props);
	return (err != 0);
	}

	if (isatty(STDIN_FILENO)) {
	(void) fprintf(stderr,
	gettext("Error: Backup stream can not be read "
	"from a terminal.\n"
	"You must redirect standard input.\n"));
	nvlist_free(props);
	return (1);
	}
	err = zfs_receive(g_zfs, argv[0], props, &flags, STDIN_FILENO, NULL);
	nvlist_free(props);

	return (err != 0);
	}

	/*
	* allow/unallow stuff
	*/
	/* copied from zfs/sys/dsl_deleg.h */
	#define ZFS_DELEG_PERM_CREATE "create"
	#define ZFS_DELEG_PERM_DESTROY "destroy"
	#define ZFS_DELEG_PERM_SNAPSHOT "snapshot"
	#define ZFS_DELEG_PERM_ROLLBACK "rollback"
	#define ZFS_DELEG_PERM_CLONE "clone"
	#define ZFS_DELEG_PERM_PROMOTE "promote"
	#define ZFS_DELEG_PERM_RENAME "rename"
	#define ZFS_DELEG_PERM_MOUNT "mount"
	#define ZFS_DELEG_PERM_SHARE "share"
	#define ZFS_DELEG_PERM_SEND "send"
	#define ZFS_DELEG_PERM_RECEIVE "receive"
	#define ZFS_DELEG_PERM_ALLOW "allow"
	#define ZFS_DELEG_PERM_USERPROP "userprop"
	#define ZFS_DELEG_PERM_VSCAN "vscan" /* ??? */
	#define ZFS_DELEG_PERM_USERQUOTA "userquota"
	#define ZFS_DELEG_PERM_GROUPQUOTA "groupquota"
	#define ZFS_DELEG_PERM_USERUSED "userused"
	#define ZFS_DELEG_PERM_GROUPUSED "groupused"
	#define ZFS_DELEG_PERM_USEROBJQUOTA "userobjquota"
	#define ZFS_DELEG_PERM_GROUPOBJQUOTA "groupobjquota"
	#define ZFS_DELEG_PERM_USEROBJUSED "userobjused"
	#define ZFS_DELEG_PERM_GROUPOBJUSED "groupobjused"

	#define ZFS_DELEG_PERM_HOLD "hold"
	#define ZFS_DELEG_PERM_RELEASE "release"
	#define ZFS_DELEG_PERM_DIFF "diff"
	#define ZFS_DELEG_PERM_BOOKMARK "bookmark"
	#define ZFS_DELEG_PERM_LOAD_KEY "load-key"
	#define ZFS_DELEG_PERM_CHANGE_KEY "change-key"

	#define ZFS_DELEG_PERM_PROJECTUSED "projectused"
	#define ZFS_DELEG_PERM_PROJECTQUOTA "projectquota"
	#define ZFS_DELEG_PERM_PROJECTOBJUSED "projectobjused"
	#define ZFS_DELEG_PERM_PROJECTOBJQUOTA "projectobjquota"

	#define ZFS_NUM_DELEG_NOTES ZFS_DELEG_NOTE_NONE

	static zfs_deleg_perm_tab_t zfs_deleg_perm_tbl[] = {
	{ ZFS_DELEG_PERM_ALLOW, ZFS_DELEG_NOTE_ALLOW },
	{ ZFS_DELEG_PERM_CLONE, ZFS_DELEG_NOTE_CLONE },
	{ ZFS_DELEG_PERM_CREATE, ZFS_DELEG_NOTE_CREATE },
	{ ZFS_DELEG_PERM_DESTROY, ZFS_DELEG_NOTE_DESTROY },
	{ ZFS_DELEG_PERM_DIFF, ZFS_DELEG_NOTE_DIFF},
	{ ZFS_DELEG_PERM_HOLD, ZFS_DELEG_NOTE_HOLD },
	{ ZFS_DELEG_PERM_MOUNT, ZFS_DELEG_NOTE_MOUNT },
	{ ZFS_DELEG_PERM_PROMOTE, ZFS_DELEG_NOTE_PROMOTE },
	{ ZFS_DELEG_PERM_RECEIVE, ZFS_DELEG_NOTE_RECEIVE },
	{ ZFS_DELEG_PERM_RELEASE, ZFS_DELEG_NOTE_RELEASE },
	{ ZFS_DELEG_PERM_RENAME, ZFS_DELEG_NOTE_RENAME },
	{ ZFS_DELEG_PERM_ROLLBACK, ZFS_DELEG_NOTE_ROLLBACK },
	{ ZFS_DELEG_PERM_SEND, ZFS_DELEG_NOTE_SEND },
	{ ZFS_DELEG_PERM_SHARE, ZFS_DELEG_NOTE_SHARE },
	{ ZFS_DELEG_PERM_SNAPSHOT, ZFS_DELEG_NOTE_SNAPSHOT },
	{ ZFS_DELEG_PERM_BOOKMARK, ZFS_DELEG_NOTE_BOOKMARK },
	{ ZFS_DELEG_PERM_LOAD_KEY, ZFS_DELEG_NOTE_LOAD_KEY },
	{ ZFS_DELEG_PERM_CHANGE_KEY, ZFS_DELEG_NOTE_CHANGE_KEY },

	{ ZFS_DELEG_PERM_GROUPQUOTA, ZFS_DELEG_NOTE_GROUPQUOTA },
	{ ZFS_DELEG_PERM_GROUPUSED, ZFS_DELEG_NOTE_GROUPUSED },
	{ ZFS_DELEG_PERM_USERPROP, ZFS_DELEG_NOTE_USERPROP },
	{ ZFS_DELEG_PERM_USERQUOTA, ZFS_DELEG_NOTE_USERQUOTA },
	{ ZFS_DELEG_PERM_USERUSED, ZFS_DELEG_NOTE_USERUSED },
	{ ZFS_DELEG_PERM_USEROBJQUOTA, ZFS_DELEG_NOTE_USEROBJQUOTA },
	{ ZFS_DELEG_PERM_USEROBJUSED, ZFS_DELEG_NOTE_USEROBJUSED },
	{ ZFS_DELEG_PERM_GROUPOBJQUOTA, ZFS_DELEG_NOTE_GROUPOBJQUOTA },
	{ ZFS_DELEG_PERM_GROUPOBJUSED, ZFS_DELEG_NOTE_GROUPOBJUSED },
	{ ZFS_DELEG_PERM_PROJECTUSED, ZFS_DELEG_NOTE_PROJECTUSED },
	{ ZFS_DELEG_PERM_PROJECTQUOTA, ZFS_DELEG_NOTE_PROJECTQUOTA },
	{ ZFS_DELEG_PERM_PROJECTOBJUSED, ZFS_DELEG_NOTE_PROJECTOBJUSED },
	{ ZFS_DELEG_PERM_PROJECTOBJQUOTA, ZFS_DELEG_NOTE_PROJECTOBJQUOTA },
	{ NULL, ZFS_DELEG_NOTE_NONE }
	};

	/* permission structure */
	typedef struct deleg_perm {
	zfs_deleg_who_type_t dp_who_type;
	const char *dp_name;
	boolean_t dp_local;
	boolean_t dp_descend;
	} deleg_perm_t;

	/* */
	typedef struct deleg_perm_node {
	deleg_perm_t dpn_perm;

	uu_avl_node_t dpn_avl_node;
	} deleg_perm_node_t;

	typedef struct fs_perm fs_perm_t;

	/* permissions set */
	typedef struct who_perm {
	zfs_deleg_who_type_t who_type;
	const char who_name; / id */
	char who_ug_name[256]; /* user/group name */
	fs_perm_t who_fsperm; / uplink */

	uu_avl_t who_deleg_perm_avl; / permissions */
	} who_perm_t;

	/* */
	typedef struct who_perm_node {
	who_perm_t who_perm;
	uu_avl_node_t who_avl_node;
	} who_perm_node_t;

	typedef struct fs_perm_set fs_perm_set_t;
	/* fs permissions */
	struct fs_perm {
	const char *fsp_name;

	uu_avl_t fsp_sc_avl; / sets,create */
	uu_avl_t fsp_uge_avl; / user,group,everyone */

	fs_perm_set_t fsp_set; / uplink */
	};

	/* */
	typedef struct fs_perm_node {
	fs_perm_t fspn_fsperm;
	uu_avl_t *fspn_avl;

	uu_list_node_t fspn_list_node;
	} fs_perm_node_t;

	/* top level structure */
	struct fs_perm_set {
	uu_list_pool_t *fsps_list_pool;
	uu_list_t fsps_list; / list of fs_perms */

	uu_avl_pool_t *fsps_named_set_avl_pool;
	uu_avl_pool_t *fsps_who_perm_avl_pool;
	uu_avl_pool_t *fsps_deleg_perm_avl_pool;
	};

	static inline const char *
	deleg_perm_type(zfs_deleg_note_t note)
	{
	/* subcommands */
	switch (note) {
	/* SUBCOMMANDS */
	/* OTHER */
	case ZFS_DELEG_NOTE_GROUPQUOTA:
	case ZFS_DELEG_NOTE_GROUPUSED:
	case ZFS_DELEG_NOTE_USERPROP:
	case ZFS_DELEG_NOTE_USERQUOTA:
	case ZFS_DELEG_NOTE_USERUSED:
	case ZFS_DELEG_NOTE_USEROBJQUOTA:
	case ZFS_DELEG_NOTE_USEROBJUSED:
	case ZFS_DELEG_NOTE_GROUPOBJQUOTA:
	case ZFS_DELEG_NOTE_GROUPOBJUSED:
	case ZFS_DELEG_NOTE_PROJECTUSED:
	case ZFS_DELEG_NOTE_PROJECTQUOTA:
	case ZFS_DELEG_NOTE_PROJECTOBJUSED:
	case ZFS_DELEG_NOTE_PROJECTOBJQUOTA:
	/* other */
	return (gettext("other"));
	default:
	return (gettext("subcommand"));
	}
	}

	static int
	who_type2weight(zfs_deleg_who_type_t who_type)
	{
	int res;
	switch (who_type) {
	case ZFS_DELEG_NAMED_SET_SETS:
	case ZFS_DELEG_NAMED_SET:
	res = 0;
	break;
	case ZFS_DELEG_CREATE_SETS:
	case ZFS_DELEG_CREATE:
	res = 1;
	break;
	case ZFS_DELEG_USER_SETS:
	case ZFS_DELEG_USER:
	res = 2;
	break;
	case ZFS_DELEG_GROUP_SETS:
	case ZFS_DELEG_GROUP:
	res = 3;
	break;
	case ZFS_DELEG_EVERYONE_SETS:
	case ZFS_DELEG_EVERYONE:
	res = 4;
	break;
	default:
	res = -1;
	}

	return (res);
	}

	-/* ARGSUSED */
	static int
	who_perm_compare(const void larg, const void rarg, void *unused)
	{
	+ (void) unused;
	const who_perm_node_t *l = larg;
	const who_perm_node_t *r = rarg;
	zfs_deleg_who_type_t ltype = l->who_perm.who_type;
	zfs_deleg_who_type_t rtype = r->who_perm.who_type;
	int lweight = who_type2weight(ltype);
	int rweight = who_type2weight(rtype);
	int res = lweight - rweight;
	if (res == 0)
	res = strncmp(l->who_perm.who_name, r->who_perm.who_name,
	ZFS_MAX_DELEG_NAME-1);

	if (res == 0)
	return (0);
	if (res > 0)
	return (1);
	else
	return (-1);
	}

	-/* ARGSUSED */
	static int
	deleg_perm_compare(const void larg, const void rarg, void *unused)
	{
	+ (void) unused;
	const deleg_perm_node_t *l = larg;
	const deleg_perm_node_t *r = rarg;
	int res = strncmp(l->dpn_perm.dp_name, r->dpn_perm.dp_name,
	ZFS_MAX_DELEG_NAME-1);

	if (res == 0)
	return (0);

	if (res > 0)
	return (1);
	else
	return (-1);
	}

	static inline void
	fs_perm_set_init(fs_perm_set_t *fspset)
	{
	bzero(fspset, sizeof (fs_perm_set_t));

	if ((fspset->fsps_list_pool = uu_list_pool_create("fsps_list_pool",
	sizeof (fs_perm_node_t), offsetof(fs_perm_node_t, fspn_list_node),
	NULL, UU_DEFAULT)) == NULL)
	nomem();
	if ((fspset->fsps_list = uu_list_create(fspset->fsps_list_pool, NULL,
	UU_DEFAULT)) == NULL)
	nomem();

	if ((fspset->fsps_named_set_avl_pool = uu_avl_pool_create(
	"named_set_avl_pool", sizeof (who_perm_node_t), offsetof(
	who_perm_node_t, who_avl_node), who_perm_compare,
	UU_DEFAULT)) == NULL)
	nomem();

	if ((fspset->fsps_who_perm_avl_pool = uu_avl_pool_create(
	"who_perm_avl_pool", sizeof (who_perm_node_t), offsetof(
	who_perm_node_t, who_avl_node), who_perm_compare,
	UU_DEFAULT)) == NULL)
	nomem();

	if ((fspset->fsps_deleg_perm_avl_pool = uu_avl_pool_create(
	"deleg_perm_avl_pool", sizeof (deleg_perm_node_t), offsetof(
	deleg_perm_node_t, dpn_avl_node), deleg_perm_compare, UU_DEFAULT))
	== NULL)
	nomem();
	}

	static inline void fs_perm_fini(fs_perm_t *);
	static inline void who_perm_fini(who_perm_t *);

	static inline void
	fs_perm_set_fini(fs_perm_set_t *fspset)
	{
	fs_perm_node_t *node = uu_list_first(fspset->fsps_list);

	while (node != NULL) {
	fs_perm_node_t *next_node =
	uu_list_next(fspset->fsps_list, node);
	fs_perm_t *fsperm = &node->fspn_fsperm;
	fs_perm_fini(fsperm);
	uu_list_remove(fspset->fsps_list, node);
	free(node);
	node = next_node;
	}

	uu_avl_pool_destroy(fspset->fsps_named_set_avl_pool);
	uu_avl_pool_destroy(fspset->fsps_who_perm_avl_pool);
	uu_avl_pool_destroy(fspset->fsps_deleg_perm_avl_pool);
	}

	static inline void
	deleg_perm_init(deleg_perm_t *deleg_perm, zfs_deleg_who_type_t type,
	const char *name)
	{
	deleg_perm->dp_who_type = type;
	deleg_perm->dp_name = name;
	}

	static inline void
	who_perm_init(who_perm_t who_perm, fs_perm_t fsperm,
	zfs_deleg_who_type_t type, const char *name)
	{
	uu_avl_pool_t *pool;
	pool = fsperm->fsp_set->fsps_deleg_perm_avl_pool;

	bzero(who_perm, sizeof (who_perm_t));

	if ((who_perm->who_deleg_perm_avl = uu_avl_create(pool, NULL,
	UU_DEFAULT)) == NULL)
	nomem();

	who_perm->who_type = type;
	who_perm->who_name = name;
	who_perm->who_fsperm = fsperm;
	}

	static inline void
	who_perm_fini(who_perm_t *who_perm)
	{
	deleg_perm_node_t *node = uu_avl_first(who_perm->who_deleg_perm_avl);

	while (node != NULL) {
	deleg_perm_node_t *next_node =
	uu_avl_next(who_perm->who_deleg_perm_avl, node);

	uu_avl_remove(who_perm->who_deleg_perm_avl, node);
	free(node);
	node = next_node;
	}

	uu_avl_destroy(who_perm->who_deleg_perm_avl);
	}

	static inline void
	fs_perm_init(fs_perm_t fsperm, fs_perm_set_t fspset, const char *fsname)
	{
	uu_avl_pool_t *nset_pool = fspset->fsps_named_set_avl_pool;
	uu_avl_pool_t *who_pool = fspset->fsps_who_perm_avl_pool;

	bzero(fsperm, sizeof (fs_perm_t));

	if ((fsperm->fsp_sc_avl = uu_avl_create(nset_pool, NULL, UU_DEFAULT))
	== NULL)
	nomem();

	if ((fsperm->fsp_uge_avl = uu_avl_create(who_pool, NULL, UU_DEFAULT))
	== NULL)
	nomem();

	fsperm->fsp_set = fspset;
	fsperm->fsp_name = fsname;
	}

	static inline void
	fs_perm_fini(fs_perm_t *fsperm)
	{
	who_perm_node_t *node = uu_avl_first(fsperm->fsp_sc_avl);
	while (node != NULL) {
	who_perm_node_t *next_node = uu_avl_next(fsperm->fsp_sc_avl,
	node);
	who_perm_t *who_perm = &node->who_perm;
	who_perm_fini(who_perm);
	uu_avl_remove(fsperm->fsp_sc_avl, node);
	free(node);
	node = next_node;
	}

	node = uu_avl_first(fsperm->fsp_uge_avl);
	while (node != NULL) {
	who_perm_node_t *next_node = uu_avl_next(fsperm->fsp_uge_avl,
	node);
	who_perm_t *who_perm = &node->who_perm;
	who_perm_fini(who_perm);
	uu_avl_remove(fsperm->fsp_uge_avl, node);
	free(node);
	node = next_node;
	}

	uu_avl_destroy(fsperm->fsp_sc_avl);
	uu_avl_destroy(fsperm->fsp_uge_avl);
	}

	static void
	set_deleg_perm_node(uu_avl_t avl, deleg_perm_node_t node,
	zfs_deleg_who_type_t who_type, const char *name, char locality)
	{
	uu_avl_index_t idx = 0;

	deleg_perm_node_t *found_node = NULL;
	deleg_perm_t *deleg_perm = &node->dpn_perm;

	deleg_perm_init(deleg_perm, who_type, name);

	if ((found_node = uu_avl_find(avl, node, NULL, &idx))
	== NULL)
	uu_avl_insert(avl, node, idx);
	else {
	node = found_node;
	deleg_perm = &node->dpn_perm;
	}


	switch (locality) {
	case ZFS_DELEG_LOCAL:
	deleg_perm->dp_local = B_TRUE;
	break;
	case ZFS_DELEG_DESCENDENT:
	deleg_perm->dp_descend = B_TRUE;
	break;
	case ZFS_DELEG_NA:
	break;
	default:
	assert(B_FALSE); /* invalid locality */
	}
	}

	static inline int
	parse_who_perm(who_perm_t who_perm, nvlist_t nvl, char locality)
	{
	nvpair_t *nvp = NULL;
	fs_perm_set_t *fspset = who_perm->who_fsperm->fsp_set;
	uu_avl_t *avl = who_perm->who_deleg_perm_avl;
	zfs_deleg_who_type_t who_type = who_perm->who_type;

	while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) {
	const char *name = nvpair_name(nvp);
	data_type_t type = nvpair_type(nvp);
	uu_avl_pool_t *avl_pool = fspset->fsps_deleg_perm_avl_pool;
	deleg_perm_node_t *node =
	safe_malloc(sizeof (deleg_perm_node_t));

	VERIFY(type == DATA_TYPE_BOOLEAN);

	uu_avl_node_init(node, &node->dpn_avl_node, avl_pool);
	set_deleg_perm_node(avl, node, who_type, name, locality);
	}

	return (0);
	}

	static inline int
	parse_fs_perm(fs_perm_t fsperm, nvlist_t nvl)
	{
	nvpair_t *nvp = NULL;
	fs_perm_set_t *fspset = fsperm->fsp_set;

	while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) {
	nvlist_t *nvl2 = NULL;
	const char *name = nvpair_name(nvp);
	uu_avl_t *avl = NULL;
	uu_avl_pool_t *avl_pool = NULL;
	zfs_deleg_who_type_t perm_type = name[0];
	char perm_locality = name[1];
	const char *perm_name = name + 3;
	who_perm_t *who_perm = NULL;

	assert('$' == name[2]);

	if (nvpair_value_nvlist(nvp, &nvl2) != 0)
	return (-1);

	switch (perm_type) {
	case ZFS_DELEG_CREATE:
	case ZFS_DELEG_CREATE_SETS:
	case ZFS_DELEG_NAMED_SET:
	case ZFS_DELEG_NAMED_SET_SETS:
	avl_pool = fspset->fsps_named_set_avl_pool;
	avl = fsperm->fsp_sc_avl;
	break;
	case ZFS_DELEG_USER:
	case ZFS_DELEG_USER_SETS:
	case ZFS_DELEG_GROUP:
	case ZFS_DELEG_GROUP_SETS:
	case ZFS_DELEG_EVERYONE:
	case ZFS_DELEG_EVERYONE_SETS:
	avl_pool = fspset->fsps_who_perm_avl_pool;
	avl = fsperm->fsp_uge_avl;
	break;

	default:
	assert(!"unhandled zfs_deleg_who_type_t");
	}

	who_perm_node_t *found_node = NULL;
	who_perm_node_t *node = safe_malloc(
	sizeof (who_perm_node_t));
	who_perm = &node->who_perm;
	uu_avl_index_t idx = 0;

	uu_avl_node_init(node, &node->who_avl_node, avl_pool);
	who_perm_init(who_perm, fsperm, perm_type, perm_name);

	if ((found_node = uu_avl_find(avl, node, NULL, &idx))
	== NULL) {
	if (avl == fsperm->fsp_uge_avl) {
	uid_t rid = 0;
	struct passwd *p = NULL;
	struct group *g = NULL;
	const char *nice_name = NULL;

	switch (perm_type) {
	case ZFS_DELEG_USER_SETS:
	case ZFS_DELEG_USER:
	rid = atoi(perm_name);
	p = getpwuid(rid);
	if (p)
	nice_name = p->pw_name;
	break;
	case ZFS_DELEG_GROUP_SETS:
	case ZFS_DELEG_GROUP:
	rid = atoi(perm_name);
	g = getgrgid(rid);
	if (g)
	nice_name = g->gr_name;
	break;

	default:
	break;
	}

	if (nice_name != NULL) {
	(void) strlcpy(
	node->who_perm.who_ug_name,
	nice_name, 256);
	} else {
	/* User or group unknown */
	(void) snprintf(
	node->who_perm.who_ug_name,
	sizeof (node->who_perm.who_ug_name),
	"(unknown: %d)", rid);
	}
	}

	uu_avl_insert(avl, node, idx);
	} else {
	node = found_node;
	who_perm = &node->who_perm;
	}

	assert(who_perm != NULL);
	(void) parse_who_perm(who_perm, nvl2, perm_locality);
	}

	return (0);
	}

	static inline int
	parse_fs_perm_set(fs_perm_set_t fspset, nvlist_t nvl)
	{
	nvpair_t *nvp = NULL;
	uu_avl_index_t idx = 0;

	while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) {
	nvlist_t *nvl2 = NULL;
	const char *fsname = nvpair_name(nvp);
	data_type_t type = nvpair_type(nvp);
	fs_perm_t *fsperm = NULL;
	fs_perm_node_t *node = safe_malloc(sizeof (fs_perm_node_t));
	if (node == NULL)
	nomem();

	fsperm = &node->fspn_fsperm;

	VERIFY(DATA_TYPE_NVLIST == type);

	uu_list_node_init(node, &node->fspn_list_node,
	fspset->fsps_list_pool);

	idx = uu_list_numnodes(fspset->fsps_list);
	fs_perm_init(fsperm, fspset, fsname);

	if (nvpair_value_nvlist(nvp, &nvl2) != 0)
	return (-1);

	(void) parse_fs_perm(fsperm, nvl2);

	uu_list_insert(fspset->fsps_list, node, idx);
	}

	return (0);
	}

	static inline const char *
	deleg_perm_comment(zfs_deleg_note_t note)
	{
	const char *str = "";

	/* subcommands */
	switch (note) {
	/* SUBCOMMANDS */
	case ZFS_DELEG_NOTE_ALLOW:
	str = gettext("Must also have the permission that is being"
	"\n\t\t\t\tallowed");
	break;
	case ZFS_DELEG_NOTE_CLONE:
	str = gettext("Must also have the 'create' ability and 'mount'"
	"\n\t\t\t\tability in the origin file system");
	break;
	case ZFS_DELEG_NOTE_CREATE:
	str = gettext("Must also have the 'mount' ability");
	break;
	case ZFS_DELEG_NOTE_DESTROY:
	str = gettext("Must also have the 'mount' ability");
	break;
	case ZFS_DELEG_NOTE_DIFF:
	str = gettext("Allows lookup of paths within a dataset;"
	"\n\t\t\t\tgiven an object number. Ordinary users need this"
	"\n\t\t\t\tin order to use zfs diff");
	break;
	case ZFS_DELEG_NOTE_HOLD:
	str = gettext("Allows adding a user hold to a snapshot");
	break;
	case ZFS_DELEG_NOTE_MOUNT:
	str = gettext("Allows mount/umount of ZFS datasets");
	break;
	case ZFS_DELEG_NOTE_PROMOTE:
	str = gettext("Must also have the 'mount'\n\t\t\t\tand"
	" 'promote' ability in the origin file system");
	break;
	case ZFS_DELEG_NOTE_RECEIVE:
	str = gettext("Must also have the 'mount' and 'create'"
	" ability");
	break;
	case ZFS_DELEG_NOTE_RELEASE:
	str = gettext("Allows releasing a user hold which\n\t\t\t\t"
	"might destroy the snapshot");
	break;
	case ZFS_DELEG_NOTE_RENAME:
	str = gettext("Must also have the 'mount' and 'create'"
	"\n\t\t\t\tability in the new parent");
	break;
	case ZFS_DELEG_NOTE_ROLLBACK:
	str = gettext("");
	break;
	case ZFS_DELEG_NOTE_SEND:
	str = gettext("");
	break;
	case ZFS_DELEG_NOTE_SHARE:
	str = gettext("Allows sharing file systems over NFS or SMB"
	"\n\t\t\t\tprotocols");
	break;
	case ZFS_DELEG_NOTE_SNAPSHOT:
	str = gettext("");
	break;
	case ZFS_DELEG_NOTE_LOAD_KEY:
	str = gettext("Allows loading or unloading an encryption key");
	break;
	case ZFS_DELEG_NOTE_CHANGE_KEY:
	str = gettext("Allows changing or adding an encryption key");
	break;
	/*
	* case ZFS_DELEG_NOTE_VSCAN:
	* str = gettext("");
	* break;
	*/
	/* OTHER */
	case ZFS_DELEG_NOTE_GROUPQUOTA:
	str = gettext("Allows accessing any groupquota@... property");
	break;
	case ZFS_DELEG_NOTE_GROUPUSED:
	str = gettext("Allows reading any groupused@... property");
	break;
	case ZFS_DELEG_NOTE_USERPROP:
	str = gettext("Allows changing any user property");
	break;
	case ZFS_DELEG_NOTE_USERQUOTA:
	str = gettext("Allows accessing any userquota@... property");
	break;
	case ZFS_DELEG_NOTE_USERUSED:
	str = gettext("Allows reading any userused@... property");
	break;
	case ZFS_DELEG_NOTE_USEROBJQUOTA:
	str = gettext("Allows accessing any userobjquota@... property");
	break;
	case ZFS_DELEG_NOTE_GROUPOBJQUOTA:
	str = gettext("Allows accessing any \n\t\t\t\t"
	"groupobjquota@... property");
	break;
	case ZFS_DELEG_NOTE_GROUPOBJUSED:
	str = gettext("Allows reading any groupobjused@... property");
	break;
	case ZFS_DELEG_NOTE_USEROBJUSED:
	str = gettext("Allows reading any userobjused@... property");
	break;
	case ZFS_DELEG_NOTE_PROJECTQUOTA:
	str = gettext("Allows accessing any projectquota@... property");
	break;
	case ZFS_DELEG_NOTE_PROJECTOBJQUOTA:
	str = gettext("Allows accessing any \n\t\t\t\t"
	"projectobjquota@... property");
	break;
	case ZFS_DELEG_NOTE_PROJECTUSED:
	str = gettext("Allows reading any projectused@... property");
	break;
	case ZFS_DELEG_NOTE_PROJECTOBJUSED:
	str = gettext("Allows accessing any \n\t\t\t\t"
	"projectobjused@... property");
	break;
	/* other */
	default:
	str = "";
	}

	return (str);
	}

	struct allow_opts {
	boolean_t local;
	boolean_t descend;
	boolean_t user;
	boolean_t group;
	boolean_t everyone;
	boolean_t create;
	boolean_t set;
	boolean_t recursive; /* unallow only */
	boolean_t prt_usage;

	boolean_t prt_perms;
	char *who;
	char *perms;
	const char *dataset;
	};

	static inline int
	prop_cmp(const void a, const void b)
	{
	const char str1 = (const char **)a;
	const char str2 = (const char **)b;
	return (strcmp(str1, str2));
	}

	static void
	allow_usage(boolean_t un, boolean_t requested, const char *msg)
	{
	const char *opt_desc[] = {
	"-h", gettext("show this help message and exit"),
	"-l", gettext("set permission locally"),
	"-d", gettext("set permission for descents"),
	"-u", gettext("set permission for user"),
	"-g", gettext("set permission for group"),
	"-e", gettext("set permission for everyone"),
	"-c", gettext("set create time permission"),
	"-s", gettext("define permission set"),
	/* unallow only */
	"-r", gettext("remove permissions recursively"),
	};
	size_t unallow_size = sizeof (opt_desc) / sizeof (char *);
	size_t allow_size = unallow_size - 2;
	const char *props[ZFS_NUM_PROPS];
	int i;
	size_t count = 0;
	FILE *fp = requested ? stdout : stderr;
	zprop_desc_t *pdtbl = zfs_prop_get_table();
	const char *fmt = gettext("%-16s %-14s\t%s\n");

	(void) fprintf(fp, gettext("Usage: %s\n"), get_usage(un ? HELP_UNALLOW :
	HELP_ALLOW));
	(void) fprintf(fp, gettext("Options:\n"));
	for (i = 0; i < (un ? unallow_size : allow_size); i += 2) {
	const char *opt = opt_desc[i];
	const char *optdsc = opt_desc[i + 1];
	(void) fprintf(fp, gettext(" %-10s %s\n"), opt, optdsc);
	}

	(void) fprintf(fp, gettext("\nThe following permissions are "
	"supported:\n\n"));
	(void) fprintf(fp, fmt, gettext("NAME"), gettext("TYPE"),
	gettext("NOTES"));
	for (i = 0; i < ZFS_NUM_DELEG_NOTES; i++) {
	const char *perm_name = zfs_deleg_perm_tbl[i].z_perm;
	zfs_deleg_note_t perm_note = zfs_deleg_perm_tbl[i].z_note;
	const char *perm_type = deleg_perm_type(perm_note);
	const char *perm_comment = deleg_perm_comment(perm_note);
	(void) fprintf(fp, fmt, perm_name, perm_type, perm_comment);
	}

	for (i = 0; i < ZFS_NUM_PROPS; i++) {
	zprop_desc_t *pd = &pdtbl[i];
	if (pd->pd_visible != B_TRUE)
	continue;

	if (pd->pd_attr == PROP_READONLY)
	continue;

	props[count++] = pd->pd_name;
	}
	props[count] = NULL;

	qsort(props, count, sizeof (char *), prop_cmp);

	for (i = 0; i < count; i++)
	(void) fprintf(fp, fmt, props[i], gettext("property"), "");

	if (msg != NULL)
	(void) fprintf(fp, gettext("\nzfs: error: %s"), msg);

	exit(requested ? 0 : 2);
	}

	static inline const char *
	munge_args(int argc, char **argv, boolean_t un, size_t expected_argc,
	char **permsp)
	{
	if (un && argc == expected_argc - 1)
	*permsp = NULL;
	else if (argc == expected_argc)
	*permsp = argv[argc - 2];
	else
	allow_usage(un, B_FALSE,
	gettext("wrong number of parameters\n"));

	return (argv[argc - 1]);
	}

	static void
	parse_allow_args(int argc, char *argv, boolean_t un, struct allow_opts opts)
	{
	int uge_sum = opts->user + opts->group + opts->everyone;
	int csuge_sum = opts->create + opts->set + uge_sum;
	int ldcsuge_sum = csuge_sum + opts->local + opts->descend;
	int all_sum = un ? ldcsuge_sum + opts->recursive : ldcsuge_sum;

	if (uge_sum > 1)
	allow_usage(un, B_FALSE,
	gettext("-u, -g, and -e are mutually exclusive\n"));

	if (opts->prt_usage) {
	if (argc == 0 && all_sum == 0)
	allow_usage(un, B_TRUE, NULL);
	else
	usage(B_FALSE);
	}

	if (opts->set) {
	if (csuge_sum > 1)
	allow_usage(un, B_FALSE,
	gettext("invalid options combined with -s\n"));

	opts->dataset = munge_args(argc, argv, un, 3, &opts->perms);
	if (argv[0][0] != '@')
	allow_usage(un, B_FALSE,
	gettext("invalid set name: missing '@' prefix\n"));
	opts->who = argv[0];
	} else if (opts->create) {
	if (ldcsuge_sum > 1)
	allow_usage(un, B_FALSE,
	gettext("invalid options combined with -c\n"));
	opts->dataset = munge_args(argc, argv, un, 2, &opts->perms);
	} else if (opts->everyone) {
	if (csuge_sum > 1)
	allow_usage(un, B_FALSE,
	gettext("invalid options combined with -e\n"));
	opts->dataset = munge_args(argc, argv, un, 2, &opts->perms);
	} else if (uge_sum == 0 && argc > 0 && strcmp(argv[0], "everyone")
	== 0) {
	opts->everyone = B_TRUE;
	argc--;
	argv++;
	opts->dataset = munge_args(argc, argv, un, 2, &opts->perms);
	} else if (argc == 1 && !un) {
	opts->prt_perms = B_TRUE;
	opts->dataset = argv[argc-1];
	} else {
	opts->dataset = munge_args(argc, argv, un, 3, &opts->perms);
	opts->who = argv[0];
	}

	if (!opts->local && !opts->descend) {
	opts->local = B_TRUE;
	opts->descend = B_TRUE;
	}
	}

	static void
	store_allow_perm(zfs_deleg_who_type_t type, boolean_t local, boolean_t descend,
	const char who, char perms, nvlist_t *top_nvl)
	{
	int i;
	char ld[2] = { '\0', '\0' };
	char who_buf[MAXNAMELEN + 32];
	char base_type = '\0';
	char set_type = '\0';
	nvlist_t *base_nvl = NULL;
	nvlist_t *set_nvl = NULL;
	nvlist_t *nvl;

	if (nvlist_alloc(&base_nvl, NV_UNIQUE_NAME, 0) != 0)
	nomem();
	if (nvlist_alloc(&set_nvl, NV_UNIQUE_NAME, 0) != 0)
	nomem();

	switch (type) {
	case ZFS_DELEG_NAMED_SET_SETS:
	case ZFS_DELEG_NAMED_SET:
	set_type = ZFS_DELEG_NAMED_SET_SETS;
	base_type = ZFS_DELEG_NAMED_SET;
	ld[0] = ZFS_DELEG_NA;
	break;
	case ZFS_DELEG_CREATE_SETS:
	case ZFS_DELEG_CREATE:
	set_type = ZFS_DELEG_CREATE_SETS;
	base_type = ZFS_DELEG_CREATE;
	ld[0] = ZFS_DELEG_NA;
	break;
	case ZFS_DELEG_USER_SETS:
	case ZFS_DELEG_USER:
	set_type = ZFS_DELEG_USER_SETS;
	base_type = ZFS_DELEG_USER;
	if (local)
	ld[0] = ZFS_DELEG_LOCAL;
	if (descend)
	ld[1] = ZFS_DELEG_DESCENDENT;
	break;
	case ZFS_DELEG_GROUP_SETS:
	case ZFS_DELEG_GROUP:
	set_type = ZFS_DELEG_GROUP_SETS;
	base_type = ZFS_DELEG_GROUP;
	if (local)
	ld[0] = ZFS_DELEG_LOCAL;
	if (descend)
	ld[1] = ZFS_DELEG_DESCENDENT;
	break;
	case ZFS_DELEG_EVERYONE_SETS:
	case ZFS_DELEG_EVERYONE:
	set_type = ZFS_DELEG_EVERYONE_SETS;
	base_type = ZFS_DELEG_EVERYONE;
	if (local)
	ld[0] = ZFS_DELEG_LOCAL;
	if (descend)
	ld[1] = ZFS_DELEG_DESCENDENT;
	break;

	default:
	assert(set_type != '\0' && base_type != '\0');
	}

	if (perms != NULL) {
	char *curr = perms;
	char *end = curr + strlen(perms);

	while (curr < end) {
	char *delim = strchr(curr, ',');
	if (delim == NULL)
	delim = end;
	else
	*delim = '\0';

	if (curr[0] == '@')
	nvl = set_nvl;
	else
	nvl = base_nvl;

	(void) nvlist_add_boolean(nvl, curr);
	if (delim != end)
	*delim = ',';
	curr = delim + 1;
	}

	for (i = 0; i < 2; i++) {
	char locality = ld[i];
	if (locality == 0)
	continue;

	if (!nvlist_empty(base_nvl)) {
	if (who != NULL)
	(void) snprintf(who_buf,
	sizeof (who_buf), "%c%c$%s",
	base_type, locality, who);
	else
	(void) snprintf(who_buf,
	sizeof (who_buf), "%c%c$",
	base_type, locality);

	(void) nvlist_add_nvlist(top_nvl, who_buf,
	base_nvl);
	}


	if (!nvlist_empty(set_nvl)) {
	if (who != NULL)
	(void) snprintf(who_buf,
	sizeof (who_buf), "%c%c$%s",
	set_type, locality, who);
	else
	(void) snprintf(who_buf,
	sizeof (who_buf), "%c%c$",
	set_type, locality);

	(void) nvlist_add_nvlist(top_nvl, who_buf,
	set_nvl);
	}
	}
	} else {
	for (i = 0; i < 2; i++) {
	char locality = ld[i];
	if (locality == 0)
	continue;

	if (who != NULL)
	(void) snprintf(who_buf, sizeof (who_buf),
	"%c%c$%s", base_type, locality, who);
	else
	(void) snprintf(who_buf, sizeof (who_buf),
	"%c%c$", base_type, locality);
	(void) nvlist_add_boolean(top_nvl, who_buf);

	if (who != NULL)
	(void) snprintf(who_buf, sizeof (who_buf),
	"%c%c$%s", set_type, locality, who);
	else
	(void) snprintf(who_buf, sizeof (who_buf),
	"%c%c$", set_type, locality);
	(void) nvlist_add_boolean(top_nvl, who_buf);
	}
	}
	}

	static int
	construct_fsacl_list(boolean_t un, struct allow_opts opts, nvlist_t *nvlp)
	{
	if (nvlist_alloc(nvlp, NV_UNIQUE_NAME, 0) != 0)
	nomem();

	if (opts->set) {
	store_allow_perm(ZFS_DELEG_NAMED_SET, opts->local,
	opts->descend, opts->who, opts->perms, *nvlp);
	} else if (opts->create) {
	store_allow_perm(ZFS_DELEG_CREATE, opts->local,
	opts->descend, NULL, opts->perms, *nvlp);
	} else if (opts->everyone) {
	store_allow_perm(ZFS_DELEG_EVERYONE, opts->local,
	opts->descend, NULL, opts->perms, *nvlp);
	} else {
	char *curr = opts->who;
	char *end = curr + strlen(curr);

	while (curr < end) {
	const char *who;
	zfs_deleg_who_type_t who_type = ZFS_DELEG_WHO_UNKNOWN;
	char *endch;
	char *delim = strchr(curr, ',');
	char errbuf[256];
	char id[64];
	struct passwd *p = NULL;
	struct group *g = NULL;

	uid_t rid;
	if (delim == NULL)
	delim = end;
	else
	*delim = '\0';

	rid = (uid_t)strtol(curr, &endch, 0);
	if (opts->user) {
	who_type = ZFS_DELEG_USER;
	if (*endch != '\0')
	p = getpwnam(curr);
	else
	p = getpwuid(rid);

	if (p != NULL)
	rid = p->pw_uid;
	else if (*endch != '\0') {
	(void) snprintf(errbuf, 256, gettext(
	"invalid user %s\n"), curr);
	allow_usage(un, B_TRUE, errbuf);
	}
	} else if (opts->group) {
	who_type = ZFS_DELEG_GROUP;
	if (*endch != '\0')
	g = getgrnam(curr);
	else
	g = getgrgid(rid);

	if (g != NULL)
	rid = g->gr_gid;
	else if (*endch != '\0') {
	(void) snprintf(errbuf, 256, gettext(
	"invalid group %s\n"), curr);
	allow_usage(un, B_TRUE, errbuf);
	}
	} else {
	if (*endch != '\0') {
	p = getpwnam(curr);
	} else {
	p = getpwuid(rid);
	}

	if (p == NULL) {
	if (*endch != '\0') {
	g = getgrnam(curr);
	} else {
	g = getgrgid(rid);
	}
	}

	if (p != NULL) {
	who_type = ZFS_DELEG_USER;
	rid = p->pw_uid;
	} else if (g != NULL) {
	who_type = ZFS_DELEG_GROUP;
	rid = g->gr_gid;
	} else {
	(void) snprintf(errbuf, 256, gettext(
	"invalid user/group %s\n"), curr);
	allow_usage(un, B_TRUE, errbuf);
	}
	}

	(void) sprintf(id, "%u", rid);
	who = id;

	store_allow_perm(who_type, opts->local,
	opts->descend, who, opts->perms, *nvlp);
	curr = delim + 1;
	}
	}

	return (0);
	}

	static void
	print_set_creat_perms(uu_avl_t *who_avl)
	{
	const char *sc_title[] = {
	gettext("Permission sets:\n"),
	gettext("Create time permissions:\n"),
	NULL
	};
	who_perm_node_t *who_node = NULL;
	int prev_weight = -1;

	for (who_node = uu_avl_first(who_avl); who_node != NULL;
	who_node = uu_avl_next(who_avl, who_node)) {
	uu_avl_t *avl = who_node->who_perm.who_deleg_perm_avl;
	zfs_deleg_who_type_t who_type = who_node->who_perm.who_type;
	const char *who_name = who_node->who_perm.who_name;
	int weight = who_type2weight(who_type);
	boolean_t first = B_TRUE;
	deleg_perm_node_t *deleg_node;

	if (prev_weight != weight) {
	(void) printf("%s", sc_title[weight]);
	prev_weight = weight;
	}

	if (who_name == NULL \|\| strnlen(who_name, 1) == 0)
	(void) printf("\t");
	else
	(void) printf("\t%s ", who_name);

	for (deleg_node = uu_avl_first(avl); deleg_node != NULL;
	deleg_node = uu_avl_next(avl, deleg_node)) {
	if (first) {
	(void) printf("%s",
	deleg_node->dpn_perm.dp_name);
	first = B_FALSE;
	} else
	(void) printf(",%s",
	deleg_node->dpn_perm.dp_name);
	}

	(void) printf("\n");
	}
	}

	static void
	print_uge_deleg_perms(uu_avl_t *who_avl, boolean_t local, boolean_t descend,
	const char *title)
	{
	who_perm_node_t *who_node = NULL;
	boolean_t prt_title = B_TRUE;
	uu_avl_walk_t *walk;

	if ((walk = uu_avl_walk_start(who_avl, UU_WALK_ROBUST)) == NULL)
	nomem();

	while ((who_node = uu_avl_walk_next(walk)) != NULL) {
	const char *who_name = who_node->who_perm.who_name;
	const char *nice_who_name = who_node->who_perm.who_ug_name;
	uu_avl_t *avl = who_node->who_perm.who_deleg_perm_avl;
	zfs_deleg_who_type_t who_type = who_node->who_perm.who_type;
	char delim = ' ';
	deleg_perm_node_t *deleg_node;
	boolean_t prt_who = B_TRUE;

	for (deleg_node = uu_avl_first(avl);
	deleg_node != NULL;
	deleg_node = uu_avl_next(avl, deleg_node)) {
	if (local != deleg_node->dpn_perm.dp_local \|\|
	descend != deleg_node->dpn_perm.dp_descend)
	continue;

	if (prt_who) {
	const char *who = NULL;
	if (prt_title) {
	prt_title = B_FALSE;
	(void) printf("%s", title);
	}

	switch (who_type) {
	case ZFS_DELEG_USER_SETS:
	case ZFS_DELEG_USER:
	who = gettext("user");
	if (nice_who_name)
	who_name = nice_who_name;
	break;
	case ZFS_DELEG_GROUP_SETS:
	case ZFS_DELEG_GROUP:
	who = gettext("group");
	if (nice_who_name)
	who_name = nice_who_name;
	break;
	case ZFS_DELEG_EVERYONE_SETS:
	case ZFS_DELEG_EVERYONE:
	who = gettext("everyone");
	who_name = NULL;
	break;

	default:
	assert(who != NULL);
	}

	prt_who = B_FALSE;
	if (who_name == NULL)
	(void) printf("\t%s", who);
	else
	(void) printf("\t%s %s", who, who_name);
	}

	(void) printf("%c%s", delim,
	deleg_node->dpn_perm.dp_name);
	delim = ',';
	}

	if (!prt_who)
	(void) printf("\n");
	}

	uu_avl_walk_end(walk);
	}

	static void
	print_fs_perms(fs_perm_set_t *fspset)
	{
	fs_perm_node_t *node = NULL;
	char buf[MAXNAMELEN + 32];
	const char *dsname = buf;

	for (node = uu_list_first(fspset->fsps_list); node != NULL;
	node = uu_list_next(fspset->fsps_list, node)) {
	uu_avl_t *sc_avl = node->fspn_fsperm.fsp_sc_avl;
	uu_avl_t *uge_avl = node->fspn_fsperm.fsp_uge_avl;
	int left = 0;

	(void) snprintf(buf, sizeof (buf),
	gettext("---- Permissions on %s "),
	node->fspn_fsperm.fsp_name);
	(void) printf("%s", dsname);
	left = 70 - strlen(buf);
	while (left-- > 0)
	(void) printf("-");
	(void) printf("\n");

	print_set_creat_perms(sc_avl);
	print_uge_deleg_perms(uge_avl, B_TRUE, B_FALSE,
	gettext("Local permissions:\n"));
	print_uge_deleg_perms(uge_avl, B_FALSE, B_TRUE,
	gettext("Descendent permissions:\n"));
	print_uge_deleg_perms(uge_avl, B_TRUE, B_TRUE,
	gettext("Local+Descendent permissions:\n"));
	}
	}

	static fs_perm_set_t fs_perm_set = { NULL, NULL, NULL, NULL };

	struct deleg_perms {
	boolean_t un;
	nvlist_t *nvl;
	};

	static int
	set_deleg_perms(zfs_handle_t zhp, void data)
	{
	struct deleg_perms perms = (struct deleg_perms )data;
	zfs_type_t zfs_type = zfs_get_type(zhp);

	if (zfs_type != ZFS_TYPE_FILESYSTEM && zfs_type != ZFS_TYPE_VOLUME)
	return (0);

	return (zfs_set_fsacl(zhp, perms->un, perms->nvl));
	}

	static int
	zfs_do_allow_unallow_impl(int argc, char **argv, boolean_t un)
	{
	zfs_handle_t *zhp;
	nvlist_t *perm_nvl = NULL;
	nvlist_t *update_perm_nvl = NULL;
	int error = 1;
	int c;
	struct allow_opts opts = { 0 };

	const char *optstr = un ? "ldugecsrh" : "ldugecsh";

	/* check opts */
	while ((c = getopt(argc, argv, optstr)) != -1) {
	switch (c) {
	case 'l':
	opts.local = B_TRUE;
	break;
	case 'd':
	opts.descend = B_TRUE;
	break;
	case 'u':
	opts.user = B_TRUE;
	break;
	case 'g':
	opts.group = B_TRUE;
	break;
	case 'e':
	opts.everyone = B_TRUE;
	break;
	case 's':
	opts.set = B_TRUE;
	break;
	case 'c':
	opts.create = B_TRUE;
	break;
	case 'r':
	opts.recursive = B_TRUE;
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	usage(B_FALSE);
	break;
	case 'h':
	opts.prt_usage = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* check arguments */
	parse_allow_args(argc, argv, un, &opts);

	/* try to open the dataset */
	if ((zhp = zfs_open(g_zfs, opts.dataset, ZFS_TYPE_FILESYSTEM \|
	ZFS_TYPE_VOLUME)) == NULL) {
	(void) fprintf(stderr, "Failed to open dataset: %s\n",
	opts.dataset);
	return (-1);
	}

	if (zfs_get_fsacl(zhp, &perm_nvl) != 0)
	goto cleanup2;

	fs_perm_set_init(&fs_perm_set);
	if (parse_fs_perm_set(&fs_perm_set, perm_nvl) != 0) {
	(void) fprintf(stderr, "Failed to parse fsacl permissions\n");
	goto cleanup1;
	}

	if (opts.prt_perms)
	print_fs_perms(&fs_perm_set);
	else {
	(void) construct_fsacl_list(un, &opts, &update_perm_nvl);
	if (zfs_set_fsacl(zhp, un, update_perm_nvl) != 0)
	goto cleanup0;

	if (un && opts.recursive) {
	struct deleg_perms data = { un, update_perm_nvl };
	if (zfs_iter_filesystems(zhp, set_deleg_perms,
	&data) != 0)
	goto cleanup0;
	}
	}

	error = 0;

	cleanup0:
	nvlist_free(perm_nvl);
	nvlist_free(update_perm_nvl);
	cleanup1:
	fs_perm_set_fini(&fs_perm_set);
	cleanup2:
	zfs_close(zhp);

	return (error);
	}

	static int
	zfs_do_allow(int argc, char **argv)
	{
	return (zfs_do_allow_unallow_impl(argc, argv, B_FALSE));
	}

	static int
	zfs_do_unallow(int argc, char **argv)
	{
	return (zfs_do_allow_unallow_impl(argc, argv, B_TRUE));
	}

	static int
	zfs_do_hold_rele_impl(int argc, char **argv, boolean_t holding)
	{
	int errors = 0;
	int i;
	const char *tag;
	boolean_t recursive = B_FALSE;
	const char *opts = holding ? "rt" : "r";
	int c;

	/* check options */
	while ((c = getopt(argc, argv, opts)) != -1) {
	switch (c) {
	case 'r':
	recursive = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* check number of arguments */
	if (argc < 2)
	usage(B_FALSE);

	tag = argv[0];
	--argc;
	++argv;

	if (holding && tag[0] == '.') {
	/* tags starting with '.' are reserved for libzfs */
	(void) fprintf(stderr, gettext("tag may not start with '.'\n"));
	usage(B_FALSE);
	}

	for (i = 0; i < argc; ++i) {
	zfs_handle_t *zhp;
	char parent[ZFS_MAX_DATASET_NAME_LEN];
	const char *delim;
	char *path = argv[i];

	delim = strchr(path, '@');
	if (delim == NULL) {
	(void) fprintf(stderr,
	gettext("'%s' is not a snapshot\n"), path);
	++errors;
	continue;
	}
	(void) strncpy(parent, path, delim - path);
	parent[delim - path] = '\0';

	zhp = zfs_open(g_zfs, parent,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME);
	if (zhp == NULL) {
	++errors;
	continue;
	}
	if (holding) {
	if (zfs_hold(zhp, delim+1, tag, recursive, -1) != 0)
	++errors;
	} else {
	if (zfs_release(zhp, delim+1, tag, recursive) != 0)
	++errors;
	}
	zfs_close(zhp);
	}

	return (errors != 0);
	}

	/*
	* zfs hold [-r] [-t] <tag> <snap> ...
	*
	* -r Recursively hold
	*
	* Apply a user-hold with the given tag to the list of snapshots.
	*/
	static int
	zfs_do_hold(int argc, char **argv)
	{
	return (zfs_do_hold_rele_impl(argc, argv, B_TRUE));
	}

	/*
	* zfs release [-r] <tag> <snap> ...
	*
	* -r Recursively release
	*
	* Release a user-hold with the given tag from the list of snapshots.
	*/
	static int
	zfs_do_release(int argc, char **argv)
	{
	return (zfs_do_hold_rele_impl(argc, argv, B_FALSE));
	}

	typedef struct holds_cbdata {
	boolean_t cb_recursive;
	const char *cb_snapname;
	nvlist_t **cb_nvlp;
	size_t cb_max_namelen;
	size_t cb_max_taglen;
	} holds_cbdata_t;

	#define STRFTIME_FMT_STR "%a %b %e %H:%M %Y"
	#define DATETIME_BUF_LEN (32)
	/*
	*
	*/
	static void
	print_holds(boolean_t scripted, int nwidth, int tagwidth, nvlist_t *nvl)
	{
	int i;
	nvpair_t *nvp = NULL;
	char *hdr_cols[] = { "NAME", "TAG", "TIMESTAMP" };
	const char *col;

	if (!scripted) {
	for (i = 0; i < 3; i++) {
	col = gettext(hdr_cols[i]);
	if (i < 2)
	(void) printf("%-*s ", i ? tagwidth : nwidth,
	col);
	else
	(void) printf("%s\n", col);
	}
	}

	while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) {
	char *zname = nvpair_name(nvp);
	nvlist_t *nvl2;
	nvpair_t *nvp2 = NULL;
	(void) nvpair_value_nvlist(nvp, &nvl2);
	while ((nvp2 = nvlist_next_nvpair(nvl2, nvp2)) != NULL) {
	char tsbuf[DATETIME_BUF_LEN];
	char *tagname = nvpair_name(nvp2);
	uint64_t val = 0;
	time_t time;
	struct tm t;

	(void) nvpair_value_uint64(nvp2, &val);
	time = (time_t)val;
	(void) localtime_r(&time, &t);
	(void) strftime(tsbuf, DATETIME_BUF_LEN,
	gettext(STRFTIME_FMT_STR), &t);

	if (scripted) {
	(void) printf("%s\t%s\t%s\n", zname,
	tagname, tsbuf);
	} else {
	(void) printf("%-s %-s %s\n", nwidth,
	zname, tagwidth, tagname, tsbuf);
	}
	}
	}
	}

	/*
	* Generic callback function to list a dataset or snapshot.
	*/
	static int
	holds_callback(zfs_handle_t zhp, void data)
	{
	holds_cbdata_t *cbp = data;
	nvlist_t top_nvl = cbp->cb_nvlp;
	nvlist_t *nvl = NULL;
	nvpair_t *nvp = NULL;
	const char *zname = zfs_get_name(zhp);
	size_t znamelen = strlen(zname);

	if (cbp->cb_recursive) {
	const char *snapname;
	char *delim = strchr(zname, '@');
	if (delim == NULL)
	return (0);

	snapname = delim + 1;
	if (strcmp(cbp->cb_snapname, snapname))
	return (0);
	}

	if (zfs_get_holds(zhp, &nvl) != 0)
	return (-1);

	if (znamelen > cbp->cb_max_namelen)
	cbp->cb_max_namelen = znamelen;

	while ((nvp = nvlist_next_nvpair(nvl, nvp)) != NULL) {
	const char *tag = nvpair_name(nvp);
	size_t taglen = strlen(tag);
	if (taglen > cbp->cb_max_taglen)
	cbp->cb_max_taglen = taglen;
	}

	return (nvlist_add_nvlist(top_nvl, zname, nvl));
	}

	/*
	* zfs holds [-rH] <snap> ...
	*
	* -r Lists holds that are set on the named snapshots recursively.
	* -H Scripted mode; elide headers and separate columns by tabs.
	*/
	static int
	zfs_do_holds(int argc, char **argv)
	{
	int errors = 0;
	int c;
	int i;
	boolean_t scripted = B_FALSE;
	boolean_t recursive = B_FALSE;
	const char *opts = "rH";
	nvlist_t *nvl;

	int types = ZFS_TYPE_SNAPSHOT;
	holds_cbdata_t cb = { 0 };

	int limit = 0;
	int ret = 0;
	int flags = 0;

	/* check options */
	while ((c = getopt(argc, argv, opts)) != -1) {
	switch (c) {
	case 'r':
	recursive = B_TRUE;
	break;
	case 'H':
	scripted = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	if (recursive) {
	types \|= ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME;
	flags \|= ZFS_ITER_RECURSE;
	}

	argc -= optind;
	argv += optind;

	/* check number of arguments */
	if (argc < 1)
	usage(B_FALSE);

	if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0)
	nomem();

	for (i = 0; i < argc; ++i) {
	char *snapshot = argv[i];
	const char *delim;
	const char *snapname;

	delim = strchr(snapshot, '@');
	if (delim == NULL) {
	(void) fprintf(stderr,
	gettext("'%s' is not a snapshot\n"), snapshot);
	++errors;
	continue;
	}
	snapname = delim + 1;
	if (recursive)
	snapshot[delim - snapshot] = '\0';

	cb.cb_recursive = recursive;
	cb.cb_snapname = snapname;
	cb.cb_nvlp = &nvl;

	/*
	* 1. collect holds data, set format options
	*/
	ret = zfs_for_each(argc, argv, flags, types, NULL, NULL, limit,
	holds_callback, &cb);
	if (ret != 0)
	++errors;
	}

	/*
	* 2. print holds data
	*/
	print_holds(scripted, cb.cb_max_namelen, cb.cb_max_taglen, nvl);

	if (nvlist_empty(nvl))
	(void) fprintf(stderr, gettext("no datasets available\n"));

	nvlist_free(nvl);

	return (0 != errors);
	}

	#define CHECK_SPINNER 30
	#define SPINNER_TIME 3 /* seconds */
	#define MOUNT_TIME 1 /* seconds */

	typedef struct get_all_state {
	boolean_t ga_verbose;
	get_all_cb_t *ga_cbp;
	} get_all_state_t;

	static int
	get_one_dataset(zfs_handle_t zhp, void data)
	{
	static char *spin[] = { "-", "\\", "\|", "/" };
	static int spinval = 0;
	static int spincheck = 0;
	static time_t last_spin_time = (time_t)0;
	get_all_state_t *state = data;
	zfs_type_t type = zfs_get_type(zhp);

	if (state->ga_verbose) {
	if (--spincheck < 0) {
	time_t now = time(NULL);
	if (last_spin_time + SPINNER_TIME < now) {
	update_progress(spin[spinval++ % 4]);
	last_spin_time = now;
	}
	spincheck = CHECK_SPINNER;
	}
	}

	/*
	* Iterate over any nested datasets.
	*/
	if (zfs_iter_filesystems(zhp, get_one_dataset, data) != 0) {
	zfs_close(zhp);
	return (1);
	}

	/*
	* Skip any datasets whose type does not match.
	*/
	if ((type & ZFS_TYPE_FILESYSTEM) == 0) {
	zfs_close(zhp);
	return (0);
	}
	libzfs_add_handle(state->ga_cbp, zhp);
	assert(state->ga_cbp->cb_used <= state->ga_cbp->cb_alloc);

	return (0);
	}

	static void
	get_all_datasets(get_all_cb_t *cbp, boolean_t verbose)
	{
	get_all_state_t state = {
	.ga_verbose = verbose,
	.ga_cbp = cbp
	};

	if (verbose)
	set_progress_header(gettext("Reading ZFS config"));
	(void) zfs_iter_root(g_zfs, get_one_dataset, &state);

	if (verbose)
	finish_progress(gettext("done."));
	}

	/*
	* Generic callback for sharing or mounting filesystems. Because the code is so
	* similar, we have a common function with an extra parameter to determine which
	* mode we are using.
	*/
	typedef enum { OP_SHARE, OP_MOUNT } share_mount_op_t;

	typedef struct share_mount_state {
	share_mount_op_t sm_op;
	boolean_t sm_verbose;
	int sm_flags;
	char *sm_options;
	char sm_proto; / only valid for OP_SHARE */
	pthread_mutex_t sm_lock; /* protects the remaining fields */
	uint_t sm_total; /* number of filesystems to process */
	uint_t sm_done; /* number of filesystems processed */
	int sm_status; /* -1 if any of the share/mount operations failed */
	} share_mount_state_t;

	/*
	* Share or mount a dataset.
	*/
	static int
	share_mount_one(zfs_handle_t zhp, int op, int flags, char protocol,
	boolean_t explicit, const char *options)
	{
	char mountpoint[ZFS_MAXPROPLEN];
	char shareopts[ZFS_MAXPROPLEN];
	char smbshareopts[ZFS_MAXPROPLEN];
	const char *cmdname = op == OP_SHARE ? "share" : "mount";
	struct mnttab mnt;
	uint64_t zoned, canmount;
	boolean_t shared_nfs, shared_smb;

	assert(zfs_get_type(zhp) & ZFS_TYPE_FILESYSTEM);

	/*
	* Check to make sure we can mount/share this dataset. If we
	* are in the global zone and the filesystem is exported to a
	* local zone, or if we are in a local zone and the
	* filesystem is not exported, then it is an error.
	*/
	zoned = zfs_prop_get_int(zhp, ZFS_PROP_ZONED);

	if (zoned && getzoneid() == GLOBAL_ZONEID) {
	if (!explicit)
	return (0);

	(void) fprintf(stderr, gettext("cannot %s '%s': "
	"dataset is exported to a local zone\n"), cmdname,
	zfs_get_name(zhp));
	return (1);

	} else if (!zoned && getzoneid() != GLOBAL_ZONEID) {
	if (!explicit)
	return (0);

	(void) fprintf(stderr, gettext("cannot %s '%s': "
	"permission denied\n"), cmdname,
	zfs_get_name(zhp));
	return (1);
	}

	/*
	* Ignore any filesystems which don't apply to us. This
	* includes those with a legacy mountpoint, or those with
	* legacy share options.
	*/
	verify(zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT, mountpoint,
	sizeof (mountpoint), NULL, NULL, 0, B_FALSE) == 0);
	verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS, shareopts,
	sizeof (shareopts), NULL, NULL, 0, B_FALSE) == 0);
	verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB, smbshareopts,
	sizeof (smbshareopts), NULL, NULL, 0, B_FALSE) == 0);

	if (op == OP_SHARE && strcmp(shareopts, "off") == 0 &&
	strcmp(smbshareopts, "off") == 0) {
	if (!explicit)
	return (0);

	(void) fprintf(stderr, gettext("cannot share '%s': "
	"legacy share\n"), zfs_get_name(zhp));
	(void) fprintf(stderr, gettext("use exports(5) or "
	"smb.conf(5) to share this filesystem, or set "
	"the sharenfs or sharesmb property\n"));
	return (1);
	}

	/*
	* We cannot share or mount legacy filesystems. If the
	* shareopts is non-legacy but the mountpoint is legacy, we
	* treat it as a legacy share.
	*/
	if (strcmp(mountpoint, "legacy") == 0) {
	if (!explicit)
	return (0);

	(void) fprintf(stderr, gettext("cannot %s '%s': "
	"legacy mountpoint\n"), cmdname, zfs_get_name(zhp));
	(void) fprintf(stderr, gettext("use %s(8) to "
	"%s this filesystem\n"), cmdname, cmdname);
	return (1);
	}

	if (strcmp(mountpoint, "none") == 0) {
	if (!explicit)
	return (0);

	(void) fprintf(stderr, gettext("cannot %s '%s': no "
	"mountpoint set\n"), cmdname, zfs_get_name(zhp));
	return (1);
	}

	/*
	* canmount explicit outcome
	* on no pass through
	* on yes pass through
	* off no return 0
	* off yes display error, return 1
	* noauto no return 0
	* noauto yes pass through
	*/
	canmount = zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT);
	if (canmount == ZFS_CANMOUNT_OFF) {
	if (!explicit)
	return (0);

	(void) fprintf(stderr, gettext("cannot %s '%s': "
	"'canmount' property is set to 'off'\n"), cmdname,
	zfs_get_name(zhp));
	return (1);
	} else if (canmount == ZFS_CANMOUNT_NOAUTO && !explicit) {
	/*
	* When performing a 'zfs mount -a', we skip any mounts for
	* datasets that have 'noauto' set. Sharing a dataset with
	* 'noauto' set is only allowed if it's mounted.
	*/
	if (op == OP_MOUNT)
	return (0);
	if (op == OP_SHARE && !zfs_is_mounted(zhp, NULL)) {
	/* also purge it from existing exports */
	zfs_unshareall_bypath(zhp, mountpoint);
	return (0);
	}
	}

	/*
	* If this filesystem is encrypted and does not have
	* a loaded key, we can not mount it.
	*/
	if ((flags & MS_CRYPT) == 0 &&
	zfs_prop_get_int(zhp, ZFS_PROP_ENCRYPTION) != ZIO_CRYPT_OFF &&
	zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS) ==
	ZFS_KEYSTATUS_UNAVAILABLE) {
	if (!explicit)
	return (0);

	(void) fprintf(stderr, gettext("cannot %s '%s': "
	"encryption key not loaded\n"), cmdname, zfs_get_name(zhp));
	return (1);
	}

	/*
	* If this filesystem is inconsistent and has a receive resume
	* token, we can not mount it.
	*/
	if (zfs_prop_get_int(zhp, ZFS_PROP_INCONSISTENT) &&
	zfs_prop_get(zhp, ZFS_PROP_RECEIVE_RESUME_TOKEN,
	NULL, 0, NULL, NULL, 0, B_TRUE) == 0) {
	if (!explicit)
	return (0);

	(void) fprintf(stderr, gettext("cannot %s '%s': "
	"Contains partially-completed state from "
	"\"zfs receive -s\", which can be resumed with "
	"\"zfs send -t\"\n"),
	cmdname, zfs_get_name(zhp));
	return (1);
	}

	if (zfs_prop_get_int(zhp, ZFS_PROP_REDACTED) && !(flags & MS_FORCE)) {
	if (!explicit)
	return (0);

	(void) fprintf(stderr, gettext("cannot %s '%s': "
	"Dataset is not complete, was created by receiving "
	"a redacted zfs send stream.\n"), cmdname,
	zfs_get_name(zhp));
	return (1);
	}

	/*
	* At this point, we have verified that the mountpoint and/or
	* shareopts are appropriate for auto management. If the
	* filesystem is already mounted or shared, return (failing
	* for explicit requests); otherwise mount or share the
	* filesystem.
	*/
	switch (op) {
	case OP_SHARE:

	shared_nfs = zfs_is_shared_nfs(zhp, NULL);
	shared_smb = zfs_is_shared_smb(zhp, NULL);

	if ((shared_nfs && shared_smb) \|\|
	(shared_nfs && strcmp(shareopts, "on") == 0 &&
	strcmp(smbshareopts, "off") == 0) \|\|
	(shared_smb && strcmp(smbshareopts, "on") == 0 &&
	strcmp(shareopts, "off") == 0)) {
	if (!explicit)
	return (0);

	(void) fprintf(stderr, gettext("cannot share "
	"'%s': filesystem already shared\n"),
	zfs_get_name(zhp));
	return (1);
	}

	if (!zfs_is_mounted(zhp, NULL) &&
	zfs_mount(zhp, NULL, flags) != 0)
	return (1);

	if (protocol == NULL) {
	if (zfs_shareall(zhp) != 0)
	return (1);
	} else if (strcmp(protocol, "nfs") == 0) {
	if (zfs_share_nfs(zhp))
	return (1);
	} else if (strcmp(protocol, "smb") == 0) {
	if (zfs_share_smb(zhp))
	return (1);
	} else {
	(void) fprintf(stderr, gettext("cannot share "
	"'%s': invalid share type '%s' "
	"specified\n"),
	zfs_get_name(zhp), protocol);
	return (1);
	}

	break;

	case OP_MOUNT:
	if (options == NULL)
	mnt.mnt_mntopts = "";
	else
	mnt.mnt_mntopts = (char *)options;

	if (!hasmntopt(&mnt, MNTOPT_REMOUNT) &&
	zfs_is_mounted(zhp, NULL)) {
	if (!explicit)
	return (0);

	(void) fprintf(stderr, gettext("cannot mount "
	"'%s': filesystem already mounted\n"),
	zfs_get_name(zhp));
	return (1);
	}

	if (zfs_mount(zhp, options, flags) != 0)
	return (1);
	break;
	}

	return (0);
	}

	/*
	* Reports progress in the form "(current/total)". Not thread-safe.
	*/
	static void
	report_mount_progress(int current, int total)
	{
	static time_t last_progress_time = 0;
	time_t now = time(NULL);
	char info[32];

	/* display header if we're here for the first time */
	if (current == 1) {
	set_progress_header(gettext("Mounting ZFS filesystems"));
	} else if (current != total && last_progress_time + MOUNT_TIME >= now) {
	/* too soon to report again */
	return;
	}

	last_progress_time = now;

	(void) sprintf(info, "(%d/%d)", current, total);

	if (current == total)
	finish_progress(info);
	else
	update_progress(info);
	}

	/*
	* zfs_foreach_mountpoint() callback that mounts or shares one filesystem and
	* updates the progress meter.
	*/
	static int
	share_mount_one_cb(zfs_handle_t zhp, void arg)
	{
	share_mount_state_t *sms = arg;
	int ret;

	ret = share_mount_one(zhp, sms->sm_op, sms->sm_flags, sms->sm_proto,
	B_FALSE, sms->sm_options);

	pthread_mutex_lock(&sms->sm_lock);
	if (ret != 0)
	sms->sm_status = ret;
	sms->sm_done++;
	if (sms->sm_verbose)
	report_mount_progress(sms->sm_done, sms->sm_total);
	pthread_mutex_unlock(&sms->sm_lock);
	return (ret);
	}

	static void
	append_options(char mntopts, char newopts)
	{
	int len = strlen(mntopts);

	/* original length plus new string to append plus 1 for the comma */
	if (len + 1 + strlen(newopts) >= MNT_LINE_MAX) {
	(void) fprintf(stderr, gettext("the opts argument for "
	"'%s' option is too long (more than %d chars)\n"),
	"-o", MNT_LINE_MAX);
	usage(B_FALSE);
	}

	if (*mntopts)
	mntopts[len++] = ',';

	(void) strcpy(&mntopts[len], newopts);
	}

	static int
	share_mount(int op, int argc, char **argv)
	{
	int do_all = 0;
	boolean_t verbose = B_FALSE;
	int c, ret = 0;
	char *options = NULL;
	int flags = 0;

	/* check options */
	while ((c = getopt(argc, argv, op == OP_MOUNT ? ":alvo:Of" : "al"))
	!= -1) {
	switch (c) {
	case 'a':
	do_all = 1;
	break;
	case 'v':
	verbose = B_TRUE;
	break;
	case 'l':
	flags \|= MS_CRYPT;
	break;
	case 'o':
	if (*optarg == '\0') {
	(void) fprintf(stderr, gettext("empty mount "
	"options (-o) specified\n"));
	usage(B_FALSE);
	}

	if (options == NULL)
	options = safe_malloc(MNT_LINE_MAX + 1);

	/* option validation is done later */
	append_options(options, optarg);
	break;
	case 'O':
	flags \|= MS_OVERLAY;
	break;
	case 'f':
	flags \|= MS_FORCE;
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	usage(B_FALSE);
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* check number of arguments */
	if (do_all) {
	char *protocol = NULL;

	if (op == OP_SHARE && argc > 0) {
	if (strcmp(argv[0], "nfs") != 0 &&
	strcmp(argv[0], "smb") != 0) {
	(void) fprintf(stderr, gettext("share type "
	"must be 'nfs' or 'smb'\n"));
	usage(B_FALSE);
	}
	protocol = argv[0];
	argc--;
	argv++;
	}

	if (argc != 0) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	start_progress_timer();
	get_all_cb_t cb = { 0 };
	get_all_datasets(&cb, verbose);

	if (cb.cb_used == 0) {
	free(options);
	return (0);
	}

	share_mount_state_t share_mount_state = { 0 };
	share_mount_state.sm_op = op;
	share_mount_state.sm_verbose = verbose;
	share_mount_state.sm_flags = flags;
	share_mount_state.sm_options = options;
	share_mount_state.sm_proto = protocol;
	share_mount_state.sm_total = cb.cb_used;
	pthread_mutex_init(&share_mount_state.sm_lock, NULL);

	/*
	* libshare isn't mt-safe, so only do the operation in parallel
	* if we're mounting. Additionally, the key-loading option must
	* be serialized so that we can prompt the user for their keys
	* in a consistent manner.
	*/
	zfs_foreach_mountpoint(g_zfs, cb.cb_handles, cb.cb_used,
	share_mount_one_cb, &share_mount_state,
	op == OP_MOUNT && !(flags & MS_CRYPT));
	zfs_commit_all_shares();

	ret = share_mount_state.sm_status;

	for (int i = 0; i < cb.cb_used; i++)
	zfs_close(cb.cb_handles[i]);
	free(cb.cb_handles);
	} else if (argc == 0) {
	FILE *mnttab;
	struct mnttab entry;

	if ((op == OP_SHARE) \|\| (options != NULL)) {
	(void) fprintf(stderr, gettext("missing filesystem "
	"argument (specify -a for all)\n"));
	usage(B_FALSE);
	}

	/*
	* When mount is given no arguments, go through
	* /proc/self/mounts and display any active ZFS mounts.
	* We hide any snapshots, since they are controlled
	* automatically.
	*/

	if ((mnttab = fopen(MNTTAB, "re")) == NULL) {
	free(options);
	return (ENOENT);
	}

	while (getmntent(mnttab, &entry) == 0) {
	if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0 \|\|
	strchr(entry.mnt_special, '@') != NULL)
	continue;

	(void) printf("%-30s %s\n", entry.mnt_special,
	entry.mnt_mountp);
	}

	(void) fclose(mnttab);
	} else {
	zfs_handle_t *zhp;

	if (argc > 1) {
	(void) fprintf(stderr,
	gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	if ((zhp = zfs_open(g_zfs, argv[0],
	ZFS_TYPE_FILESYSTEM)) == NULL) {
	ret = 1;
	} else {
	ret = share_mount_one(zhp, op, flags, NULL, B_TRUE,
	options);
	zfs_commit_all_shares();
	zfs_close(zhp);
	}
	}

	free(options);
	return (ret);
	}

	/*
	* zfs mount -a [nfs]
	* zfs mount filesystem
	*
	* Mount all filesystems, or mount the given filesystem.
	*/
	static int
	zfs_do_mount(int argc, char **argv)
	{
	return (share_mount(OP_MOUNT, argc, argv));
	}

	/*
	* zfs share -a [nfs \| smb]
	* zfs share filesystem
	*
	* Share all filesystems, or share the given filesystem.
	*/
	static int
	zfs_do_share(int argc, char **argv)
	{
	return (share_mount(OP_SHARE, argc, argv));
	}

	typedef struct unshare_unmount_node {
	zfs_handle_t *un_zhp;
	char *un_mountp;
	uu_avl_node_t un_avlnode;
	} unshare_unmount_node_t;

	-/* ARGSUSED */
	static int
	unshare_unmount_compare(const void larg, const void rarg, void *unused)
	{
	+ (void) unused;
	const unshare_unmount_node_t *l = larg;
	const unshare_unmount_node_t *r = rarg;

	return (strcmp(l->un_mountp, r->un_mountp));
	}

	/*
	* Convenience routine used by zfs_do_umount() and manual_unmount(). Given an
	* absolute path, find the entry /proc/self/mounts, verify that it's a
	* ZFS filesystem, and unmount it appropriately.
	*/
	static int
	unshare_unmount_path(int op, char *path, int flags, boolean_t is_manual)
	{
	zfs_handle_t *zhp;
	int ret = 0;
	struct stat64 statbuf;
	struct extmnttab entry;
	const char *cmdname = (op == OP_SHARE) ? "unshare" : "unmount";
	ino_t path_inode;

	/*
	* Search for the given (major,minor) pair in the mount table.
	*/

	if (getextmntent(path, &entry, &statbuf) != 0) {
	if (op == OP_SHARE) {
	(void) fprintf(stderr, gettext("cannot %s '%s': not "
	"currently mounted\n"), cmdname, path);
	return (1);
	}
	(void) fprintf(stderr, gettext("warning: %s not in"
	"/proc/self/mounts\n"), path);
	if ((ret = umount2(path, flags)) != 0)
	(void) fprintf(stderr, gettext("%s: %s\n"), path,
	strerror(errno));
	return (ret != 0);
	}
	path_inode = statbuf.st_ino;

	if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0) {
	(void) fprintf(stderr, gettext("cannot %s '%s': not a ZFS "
	"filesystem\n"), cmdname, path);
	return (1);
	}

	if ((zhp = zfs_open(g_zfs, entry.mnt_special,
	ZFS_TYPE_FILESYSTEM)) == NULL)
	return (1);

	ret = 1;
	if (stat64(entry.mnt_mountp, &statbuf) != 0) {
	(void) fprintf(stderr, gettext("cannot %s '%s': %s\n"),
	cmdname, path, strerror(errno));
	goto out;
	} else if (statbuf.st_ino != path_inode) {
	(void) fprintf(stderr, gettext("cannot "
	"%s '%s': not a mountpoint\n"), cmdname, path);
	goto out;
	}

	if (op == OP_SHARE) {
	char nfs_mnt_prop[ZFS_MAXPROPLEN];
	char smbshare_prop[ZFS_MAXPROPLEN];

	verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS, nfs_mnt_prop,
	sizeof (nfs_mnt_prop), NULL, NULL, 0, B_FALSE) == 0);
	verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB, smbshare_prop,
	sizeof (smbshare_prop), NULL, NULL, 0, B_FALSE) == 0);

	if (strcmp(nfs_mnt_prop, "off") == 0 &&
	strcmp(smbshare_prop, "off") == 0) {
	(void) fprintf(stderr, gettext("cannot unshare "
	"'%s': legacy share\n"), path);
	(void) fprintf(stderr, gettext("use exportfs(8) "
	"or smbcontrol(1) to unshare this filesystem\n"));
	} else if (!zfs_is_shared(zhp)) {
	(void) fprintf(stderr, gettext("cannot unshare '%s': "
	"not currently shared\n"), path);
	} else {
	ret = zfs_unshareall_bypath(zhp, path);
	zfs_commit_all_shares();
	}
	} else {
	char mtpt_prop[ZFS_MAXPROPLEN];

	verify(zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT, mtpt_prop,
	sizeof (mtpt_prop), NULL, NULL, 0, B_FALSE) == 0);

	if (is_manual) {
	ret = zfs_unmount(zhp, NULL, flags);
	} else if (strcmp(mtpt_prop, "legacy") == 0) {
	(void) fprintf(stderr, gettext("cannot unmount "
	"'%s': legacy mountpoint\n"),
	zfs_get_name(zhp));
	(void) fprintf(stderr, gettext("use umount(8) "
	"to unmount this filesystem\n"));
	} else {
	ret = zfs_unmountall(zhp, flags);
	}
	}

	out:
	zfs_close(zhp);

	return (ret != 0);
	}

	/*
	* Generic callback for unsharing or unmounting a filesystem.
	*/
	static int
	unshare_unmount(int op, int argc, char **argv)
	{
	int do_all = 0;
	int flags = 0;
	int ret = 0;
	int c;
	zfs_handle_t *zhp;
	char nfs_mnt_prop[ZFS_MAXPROPLEN];
	char sharesmb[ZFS_MAXPROPLEN];

	/* check options */
	while ((c = getopt(argc, argv, op == OP_SHARE ? ":a" : "afu")) != -1) {
	switch (c) {
	case 'a':
	do_all = 1;
	break;
	case 'f':
	flags \|= MS_FORCE;
	break;
	case 'u':
	flags \|= MS_CRYPT;
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	usage(B_FALSE);
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (do_all) {
	/*
	* We could make use of zfs_for_each() to walk all datasets in
	* the system, but this would be very inefficient, especially
	* since we would have to linearly search /proc/self/mounts for
	* each one. Instead, do one pass through /proc/self/mounts
	* looking for zfs entries and call zfs_unmount() for each one.
	*
	* Things get a little tricky if the administrator has created
	* mountpoints beneath other ZFS filesystems. In this case, we
	* have to unmount the deepest filesystems first. To accomplish
	* this, we place all the mountpoints in an AVL tree sorted by
	* the special type (dataset name), and walk the result in
	* reverse to make sure to get any snapshots first.
	*/
	FILE *mnttab;
	struct mnttab entry;
	uu_avl_pool_t *pool;
	uu_avl_t *tree = NULL;
	unshare_unmount_node_t *node;
	uu_avl_index_t idx;
	uu_avl_walk_t *walk;
	char *protocol = NULL;

	if (op == OP_SHARE && argc > 0) {
	if (strcmp(argv[0], "nfs") != 0 &&
	strcmp(argv[0], "smb") != 0) {
	(void) fprintf(stderr, gettext("share type "
	"must be 'nfs' or 'smb'\n"));
	usage(B_FALSE);
	}
	protocol = argv[0];
	argc--;
	argv++;
	}

	if (argc != 0) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	if (((pool = uu_avl_pool_create("unmount_pool",
	sizeof (unshare_unmount_node_t),
	offsetof(unshare_unmount_node_t, un_avlnode),
	unshare_unmount_compare, UU_DEFAULT)) == NULL) \|\|
	((tree = uu_avl_create(pool, NULL, UU_DEFAULT)) == NULL))
	nomem();

	if ((mnttab = fopen(MNTTAB, "re")) == NULL)
	return (ENOENT);

	while (getmntent(mnttab, &entry) == 0) {

	/* ignore non-ZFS entries */
	if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0)
	continue;

	/* ignore snapshots */
	if (strchr(entry.mnt_special, '@') != NULL)
	continue;

	if ((zhp = zfs_open(g_zfs, entry.mnt_special,
	ZFS_TYPE_FILESYSTEM)) == NULL) {
	ret = 1;
	continue;
	}

	/*
	* Ignore datasets that are excluded/restricted by
	* parent pool name.
	*/
	if (zpool_skip_pool(zfs_get_pool_name(zhp))) {
	zfs_close(zhp);
	continue;
	}

	switch (op) {
	case OP_SHARE:
	verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS,
	nfs_mnt_prop,
	sizeof (nfs_mnt_prop),
	NULL, NULL, 0, B_FALSE) == 0);
	if (strcmp(nfs_mnt_prop, "off") != 0)
	break;
	verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB,
	nfs_mnt_prop,
	sizeof (nfs_mnt_prop),
	NULL, NULL, 0, B_FALSE) == 0);
	if (strcmp(nfs_mnt_prop, "off") == 0)
	continue;
	break;
	case OP_MOUNT:
	/* Ignore legacy mounts */
	verify(zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT,
	nfs_mnt_prop,
	sizeof (nfs_mnt_prop),
	NULL, NULL, 0, B_FALSE) == 0);
	if (strcmp(nfs_mnt_prop, "legacy") == 0)
	continue;
	/* Ignore canmount=noauto mounts */
	if (zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT) ==
	ZFS_CANMOUNT_NOAUTO)
	continue;
	break;
	default:
	break;
	}

	node = safe_malloc(sizeof (unshare_unmount_node_t));
	node->un_zhp = zhp;
	node->un_mountp = safe_strdup(entry.mnt_mountp);

	uu_avl_node_init(node, &node->un_avlnode, pool);

	if (uu_avl_find(tree, node, NULL, &idx) == NULL) {
	uu_avl_insert(tree, node, idx);
	} else {
	zfs_close(node->un_zhp);
	free(node->un_mountp);
	free(node);
	}
	}
	(void) fclose(mnttab);

	/*
	* Walk the AVL tree in reverse, unmounting each filesystem and
	* removing it from the AVL tree in the process.
	*/
	if ((walk = uu_avl_walk_start(tree,
	UU_WALK_REVERSE \| UU_WALK_ROBUST)) == NULL)
	nomem();

	while ((node = uu_avl_walk_next(walk)) != NULL) {
	const char *mntarg = NULL;

	uu_avl_remove(tree, node);
	switch (op) {
	case OP_SHARE:
	if (zfs_unshareall_bytype(node->un_zhp,
	node->un_mountp, protocol) != 0)
	ret = 1;
	break;

	case OP_MOUNT:
	if (zfs_unmount(node->un_zhp,
	mntarg, flags) != 0)
	ret = 1;
	break;
	}

	zfs_close(node->un_zhp);
	free(node->un_mountp);
	free(node);
	}

	if (op == OP_SHARE)
	zfs_commit_shares(protocol);

	uu_avl_walk_end(walk);
	uu_avl_destroy(tree);
	uu_avl_pool_destroy(pool);

	} else {
	if (argc != 1) {
	if (argc == 0)
	(void) fprintf(stderr,
	gettext("missing filesystem argument\n"));
	else
	(void) fprintf(stderr,
	gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	/*
	* We have an argument, but it may be a full path or a ZFS
	* filesystem. Pass full paths off to unmount_path() (shared by
	* manual_unmount), otherwise open the filesystem and pass to
	* zfs_unmount().
	*/
	if (argv[0][0] == '/')
	return (unshare_unmount_path(op, argv[0],
	flags, B_FALSE));

	if ((zhp = zfs_open(g_zfs, argv[0],
	ZFS_TYPE_FILESYSTEM)) == NULL)
	return (1);

	verify(zfs_prop_get(zhp, op == OP_SHARE ?
	ZFS_PROP_SHARENFS : ZFS_PROP_MOUNTPOINT,
	nfs_mnt_prop, sizeof (nfs_mnt_prop), NULL,
	NULL, 0, B_FALSE) == 0);

	switch (op) {
	case OP_SHARE:
	verify(zfs_prop_get(zhp, ZFS_PROP_SHARENFS,
	nfs_mnt_prop,
	sizeof (nfs_mnt_prop),
	NULL, NULL, 0, B_FALSE) == 0);
	verify(zfs_prop_get(zhp, ZFS_PROP_SHARESMB,
	sharesmb, sizeof (sharesmb), NULL, NULL,
	0, B_FALSE) == 0);

	if (strcmp(nfs_mnt_prop, "off") == 0 &&
	strcmp(sharesmb, "off") == 0) {
	(void) fprintf(stderr, gettext("cannot "
	"unshare '%s': legacy share\n"),
	zfs_get_name(zhp));
	(void) fprintf(stderr, gettext("use "
	"exports(5) or smb.conf(5) to unshare "
	"this filesystem\n"));
	ret = 1;
	} else if (!zfs_is_shared(zhp)) {
	(void) fprintf(stderr, gettext("cannot "
	"unshare '%s': not currently "
	"shared\n"), zfs_get_name(zhp));
	ret = 1;
	} else if (zfs_unshareall(zhp) != 0) {
	ret = 1;
	}
	break;

	case OP_MOUNT:
	if (strcmp(nfs_mnt_prop, "legacy") == 0) {
	(void) fprintf(stderr, gettext("cannot "
	"unmount '%s': legacy "
	"mountpoint\n"), zfs_get_name(zhp));
	(void) fprintf(stderr, gettext("use "
	"umount(8) to unmount this "
	"filesystem\n"));
	ret = 1;
	} else if (!zfs_is_mounted(zhp, NULL)) {
	(void) fprintf(stderr, gettext("cannot "
	"unmount '%s': not currently "
	"mounted\n"),
	zfs_get_name(zhp));
	ret = 1;
	} else if (zfs_unmountall(zhp, flags) != 0) {
	ret = 1;
	}
	break;
	}

	zfs_close(zhp);
	}

	return (ret);
	}

	/*
	* zfs unmount [-fu] -a
	* zfs unmount [-fu] filesystem
	*
	* Unmount all filesystems, or a specific ZFS filesystem.
	*/
	static int
	zfs_do_unmount(int argc, char **argv)
	{
	return (unshare_unmount(OP_MOUNT, argc, argv));
	}

	/*
	* zfs unshare -a
	* zfs unshare filesystem
	*
	* Unshare all filesystems, or a specific ZFS filesystem.
	*/
	static int
	zfs_do_unshare(int argc, char **argv)
	{
	return (unshare_unmount(OP_SHARE, argc, argv));
	}

	static int
	find_command_idx(char command, int idx)
	{
	int i;

	for (i = 0; i < NCOMMAND; i++) {
	if (command_table[i].name == NULL)
	continue;

	if (strcmp(command, command_table[i].name) == 0) {
	*idx = i;
	return (0);
	}
	}
	return (1);
	}

	static int
	zfs_do_diff(int argc, char **argv)
	{
	zfs_handle_t *zhp;
	int flags = 0;
	char *tosnap = NULL;
	char *fromsnap = NULL;
	char atp, copy;
	int err = 0;
	int c;
	struct sigaction sa;

	- while ((c = getopt(argc, argv, "FHt")) != -1) {
	+ while ((c = getopt(argc, argv, "FHth")) != -1) {
	switch (c) {
	case 'F':
	flags \|= ZFS_DIFF_CLASSIFY;
	break;
	case 'H':
	flags \|= ZFS_DIFF_PARSEABLE;
	break;
	case 't':
	flags \|= ZFS_DIFF_TIMESTAMP;
	break;
	+ case 'h':
	+ flags \|= ZFS_DIFF_NO_MANGLE;
	+ break;
	default:
	(void) fprintf(stderr,
	gettext("invalid option '%c'\n"), optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr,
	gettext("must provide at least one snapshot name\n"));
	usage(B_FALSE);
	}

	if (argc > 2) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	fromsnap = argv[0];
	tosnap = (argc == 2) ? argv[1] : NULL;

	copy = NULL;
	if (*fromsnap != '@')
	copy = strdup(fromsnap);
	else if (tosnap)
	copy = strdup(tosnap);
	if (copy == NULL)
	usage(B_FALSE);

	if ((atp = strchr(copy, '@')) != NULL)
	*atp = '\0';

	if ((zhp = zfs_open(g_zfs, copy, ZFS_TYPE_FILESYSTEM)) == NULL) {
	free(copy);
	return (1);
	}
	free(copy);

	/*
	* Ignore SIGPIPE so that the library can give us
	* information on any failure
	*/
	if (sigemptyset(&sa.sa_mask) == -1) {
	err = errno;
	goto out;
	}
	sa.sa_flags = 0;
	sa.sa_handler = SIG_IGN;
	if (sigaction(SIGPIPE, &sa, NULL) == -1) {
	err = errno;
	goto out;
	}

	err = zfs_show_diffs(zhp, STDOUT_FILENO, fromsnap, tosnap, flags);
	out:
	zfs_close(zhp);

	return (err != 0);
	}

	/*
	* zfs bookmark <fs@source>\|<fs#source> <fs#bookmark>
	*
	* Creates a bookmark with the given name from the source snapshot
	* or creates a copy of an existing source bookmark.
	*/
	static int
	zfs_do_bookmark(int argc, char **argv)
	{
	char source, bookname;
	char expbuf[ZFS_MAX_DATASET_NAME_LEN];
	int source_type;
	nvlist_t *nvl;
	int ret = 0;
	int c;

	/* check options */
	while ((c = getopt(argc, argv, "")) != -1) {
	switch (c) {
	case '?':
	(void) fprintf(stderr,
	gettext("invalid option '%c'\n"), optopt);
	goto usage;
	}
	}

	argc -= optind;
	argv += optind;

	/* check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing source argument\n"));
	goto usage;
	}
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing bookmark argument\n"));
	goto usage;
	}

	source = argv[0];
	bookname = argv[1];

	if (strchr(source, '@') == NULL && strchr(source, '#') == NULL) {
	(void) fprintf(stderr,
	gettext("invalid source name '%s': "
	"must contain a '@' or '#'\n"), source);
	goto usage;
	}
	if (strchr(bookname, '#') == NULL) {
	(void) fprintf(stderr,
	gettext("invalid bookmark name '%s': "
	"must contain a '#'\n"), bookname);
	goto usage;
	}

	/*
	* expand source or bookname to full path:
	* one of them may be specified as short name
	*/
	{
	char **expand;
	char source_short, bookname_short;
	source_short = strpbrk(source, "@#");
	bookname_short = strpbrk(bookname, "#");
	if (source_short == source &&
	bookname_short == bookname) {
	(void) fprintf(stderr, gettext(
	"either source or bookmark must be specified as "
	"full dataset paths"));
	goto usage;
	} else if (source_short != source &&
	bookname_short != bookname) {
	expand = NULL;
	} else if (source_short != source) {
	strlcpy(expbuf, source, sizeof (expbuf));
	expand = &bookname;
	} else if (bookname_short != bookname) {
	strlcpy(expbuf, bookname, sizeof (expbuf));
	expand = &source;
	} else {
	abort();
	}
	if (expand != NULL) {
	strpbrk(expbuf, "@#") = '\0'; / dataset name in buf */
	(void) strlcat(expbuf, *expand, sizeof (expbuf));
	*expand = expbuf;
	}
	}

	/* determine source type */
	switch (*strpbrk(source, "@#")) {
	case '@': source_type = ZFS_TYPE_SNAPSHOT; break;
	case '#': source_type = ZFS_TYPE_BOOKMARK; break;
	default: abort();
	}

	/* test the source exists */
	zfs_handle_t *zhp;
	zhp = zfs_open(g_zfs, source, source_type);
	if (zhp == NULL)
	goto usage;
	zfs_close(zhp);

	nvl = fnvlist_alloc();
	fnvlist_add_string(nvl, bookname, source);
	ret = lzc_bookmark(nvl, NULL);
	fnvlist_free(nvl);

	if (ret != 0) {
	const char *err_msg = NULL;
	char errbuf[1024];

	(void) snprintf(errbuf, sizeof (errbuf),
	dgettext(TEXT_DOMAIN,
	"cannot create bookmark '%s'"), bookname);

	switch (ret) {
	case EXDEV:
	err_msg = "bookmark is in a different pool";
	break;
	case ZFS_ERR_BOOKMARK_SOURCE_NOT_ANCESTOR:
	err_msg = "source is not an ancestor of the "
	"new bookmark's dataset";
	break;
	case EEXIST:
	err_msg = "bookmark exists";
	break;
	case EINVAL:
	err_msg = "invalid argument";
	break;
	case ENOTSUP:
	err_msg = "bookmark feature not enabled";
	break;
	case ENOSPC:
	err_msg = "out of space";
	break;
	case ENOENT:
	err_msg = "dataset does not exist";
	break;
	default:
	(void) zfs_standard_error(g_zfs, ret, errbuf);
	break;
	}
	if (err_msg != NULL) {
	(void) fprintf(stderr, "%s: %s\n", errbuf,
	dgettext(TEXT_DOMAIN, err_msg));
	}
	}

	return (ret != 0);

	usage:
	usage(B_FALSE);
	return (-1);
	}

	static int
	zfs_do_channel_program(int argc, char **argv)
	{
	int ret, fd, c;
	char progbuf, filename, *poolname;
	size_t progsize, progread;
	nvlist_t *outnvl = NULL;
	uint64_t instrlimit = ZCP_DEFAULT_INSTRLIMIT;
	uint64_t memlimit = ZCP_DEFAULT_MEMLIMIT;
	boolean_t sync_flag = B_TRUE, json_output = B_FALSE;
	zpool_handle_t *zhp;

	/* check options */
	while ((c = getopt(argc, argv, "nt:m:j")) != -1) {
	switch (c) {
	case 't':
	case 'm': {
	uint64_t arg;
	char *endp;

	errno = 0;
	arg = strtoull(optarg, &endp, 0);
	if (errno != 0 \|\| *endp != '\0') {
	(void) fprintf(stderr, gettext(
	"invalid argument "
	"'%s': expected integer\n"), optarg);
	goto usage;
	}

	if (c == 't') {
	instrlimit = arg;
	} else {
	ASSERT3U(c, ==, 'm');
	memlimit = arg;
	}
	break;
	}
	case 'n': {
	sync_flag = B_FALSE;
	break;
	}
	case 'j': {
	json_output = B_TRUE;
	break;
	}
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	goto usage;
	}
	}

	argc -= optind;
	argv += optind;

	if (argc < 2) {
	(void) fprintf(stderr,
	gettext("invalid number of arguments\n"));
	goto usage;
	}

	poolname = argv[0];
	filename = argv[1];
	if (strcmp(filename, "-") == 0) {
	fd = 0;
	filename = "standard input";
	} else if ((fd = open(filename, O_RDONLY)) < 0) {
	(void) fprintf(stderr, gettext("cannot open '%s': %s\n"),
	filename, strerror(errno));
	return (1);
	}

	if ((zhp = zpool_open(g_zfs, poolname)) == NULL) {
	(void) fprintf(stderr, gettext("cannot open pool '%s'\n"),
	poolname);
	if (fd != 0)
	(void) close(fd);
	return (1);
	}
	zpool_close(zhp);

	/*
	* Read in the channel program, expanding the program buffer as
	* necessary.
	*/
	progread = 0;
	progsize = 1024;
	progbuf = safe_malloc(progsize);
	do {
	ret = read(fd, progbuf + progread, progsize - progread);
	progread += ret;
	if (progread == progsize && ret > 0) {
	progsize *= 2;
	progbuf = safe_realloc(progbuf, progsize);
	}
	} while (ret > 0);

	if (fd != 0)
	(void) close(fd);
	if (ret < 0) {
	free(progbuf);
	(void) fprintf(stderr,
	gettext("cannot read '%s': %s\n"),
	filename, strerror(errno));
	return (1);
	}
	progbuf[progread] = '\0';

	/*
	* Any remaining arguments are passed as arguments to the lua script as
	* a string array:
	* {
	* "argv" -> [ "arg 1", ... "arg n" ],
	* }
	*/
	nvlist_t *argnvl = fnvlist_alloc();
	- fnvlist_add_string_array(argnvl, ZCP_ARG_CLIARGV, argv + 2, argc - 2);
	+ fnvlist_add_string_array(argnvl, ZCP_ARG_CLIARGV,
	+ (const char **)argv + 2, argc - 2);

	if (sync_flag) {
	ret = lzc_channel_program(poolname, progbuf,
	instrlimit, memlimit, argnvl, &outnvl);
	} else {
	ret = lzc_channel_program_nosync(poolname, progbuf,
	instrlimit, memlimit, argnvl, &outnvl);
	}

	if (ret != 0) {
	/*
	* On error, report the error message handed back by lua if one
	* exists. Otherwise, generate an appropriate error message,
	* falling back on strerror() for an unexpected return code.
	*/
	char *errstring = NULL;
	const char *msg = gettext("Channel program execution failed");
	uint64_t instructions = 0;
	if (outnvl != NULL && nvlist_exists(outnvl, ZCP_RET_ERROR)) {
	(void) nvlist_lookup_string(outnvl,
	ZCP_RET_ERROR, &errstring);
	if (errstring == NULL)
	errstring = strerror(ret);
	if (ret == ETIME) {
	(void) nvlist_lookup_uint64(outnvl,
	ZCP_ARG_INSTRLIMIT, &instructions);
	}
	} else {
	switch (ret) {
	case EINVAL:
	errstring =
	"Invalid instruction or memory limit.";
	break;
	case ENOMEM:
	errstring = "Return value too large.";
	break;
	case ENOSPC:
	errstring = "Memory limit exhausted.";
	break;
	case ETIME:
	errstring = "Timed out.";
	break;
	case EPERM:
	errstring = "Permission denied. Channel "
	"programs must be run as root.";
	break;
	default:
	(void) zfs_standard_error(g_zfs, ret, msg);
	}
	}
	if (errstring != NULL)
	(void) fprintf(stderr, "%s:\n%s\n", msg, errstring);

	if (ret == ETIME && instructions != 0)
	(void) fprintf(stderr,
	gettext("%llu Lua instructions\n"),
	(u_longlong_t)instructions);
	} else {
	if (json_output) {
	(void) nvlist_print_json(stdout, outnvl);
	} else if (nvlist_empty(outnvl)) {
	(void) fprintf(stdout, gettext("Channel program fully "
	"executed and did not produce output.\n"));
	} else {
	(void) fprintf(stdout, gettext("Channel program fully "
	"executed and produced output:\n"));
	dump_nvlist(outnvl, 4);
	}
	}

	free(progbuf);
	fnvlist_free(outnvl);
	fnvlist_free(argnvl);
	return (ret != 0);

	usage:
	usage(B_FALSE);
	return (-1);
	}


	typedef struct loadkey_cbdata {
	boolean_t cb_loadkey;
	boolean_t cb_recursive;
	boolean_t cb_noop;
	char *cb_keylocation;
	uint64_t cb_numfailed;
	uint64_t cb_numattempted;
	} loadkey_cbdata_t;

	static int
	load_key_callback(zfs_handle_t zhp, void data)
	{
	int ret;
	boolean_t is_encroot;
	loadkey_cbdata_t *cb = data;
	uint64_t keystatus = zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS);

	/*
	* If we are working recursively, we want to skip loading / unloading
	* keys for non-encryption roots and datasets whose keys are already
	* in the desired end-state.
	*/
	if (cb->cb_recursive) {
	ret = zfs_crypto_get_encryption_root(zhp, &is_encroot, NULL);
	if (ret != 0)
	return (ret);
	if (!is_encroot)
	return (0);

	if ((cb->cb_loadkey && keystatus == ZFS_KEYSTATUS_AVAILABLE) \|\|
	(!cb->cb_loadkey && keystatus == ZFS_KEYSTATUS_UNAVAILABLE))
	return (0);
	}

	cb->cb_numattempted++;

	if (cb->cb_loadkey)
	ret = zfs_crypto_load_key(zhp, cb->cb_noop, cb->cb_keylocation);
	else
	ret = zfs_crypto_unload_key(zhp);

	if (ret != 0) {
	cb->cb_numfailed++;
	return (ret);
	}

	return (0);
	}

	static int
	load_unload_keys(int argc, char **argv, boolean_t loadkey)
	{
	int c, ret = 0, flags = 0;
	boolean_t do_all = B_FALSE;
	loadkey_cbdata_t cb = { 0 };

	cb.cb_loadkey = loadkey;

	while ((c = getopt(argc, argv, "anrL:")) != -1) {
	/* noop and alternate keylocations only apply to zfs load-key */
	if (loadkey) {
	switch (c) {
	case 'n':
	cb.cb_noop = B_TRUE;
	continue;
	case 'L':
	cb.cb_keylocation = optarg;
	continue;
	default:
	break;
	}
	}

	switch (c) {
	case 'a':
	do_all = B_TRUE;
	cb.cb_recursive = B_TRUE;
	break;
	case 'r':
	flags \|= ZFS_ITER_RECURSE;
	cb.cb_recursive = B_TRUE;
	break;
	default:
	(void) fprintf(stderr,
	gettext("invalid option '%c'\n"), optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (!do_all && argc == 0) {
	(void) fprintf(stderr,
	gettext("Missing dataset argument or -a option\n"));
	usage(B_FALSE);
	}

	if (do_all && argc != 0) {
	(void) fprintf(stderr,
	gettext("Cannot specify dataset with -a option\n"));
	usage(B_FALSE);
	}

	if (cb.cb_recursive && cb.cb_keylocation != NULL &&
	strcmp(cb.cb_keylocation, "prompt") != 0) {
	(void) fprintf(stderr, gettext("alternate keylocation may only "
	"be 'prompt' with -r or -a\n"));
	usage(B_FALSE);
	}

	ret = zfs_for_each(argc, argv, flags,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME, NULL, NULL, 0,
	load_key_callback, &cb);

	if (cb.cb_noop \|\| (cb.cb_recursive && cb.cb_numattempted != 0)) {
	(void) printf(gettext("%llu / %llu key(s) successfully %s\n"),
	(u_longlong_t)(cb.cb_numattempted - cb.cb_numfailed),
	(u_longlong_t)cb.cb_numattempted,
	loadkey ? (cb.cb_noop ? "verified" : "loaded") :
	"unloaded");
	}

	if (cb.cb_numfailed != 0)
	ret = -1;

	return (ret);
	}

	static int
	zfs_do_load_key(int argc, char **argv)
	{
	return (load_unload_keys(argc, argv, B_TRUE));
	}


	static int
	zfs_do_unload_key(int argc, char **argv)
	{
	return (load_unload_keys(argc, argv, B_FALSE));
	}

	static int
	zfs_do_change_key(int argc, char **argv)
	{
	int c, ret;
	uint64_t keystatus;
	boolean_t loadkey = B_FALSE, inheritkey = B_FALSE;
	zfs_handle_t *zhp = NULL;
	nvlist_t *props = fnvlist_alloc();

	while ((c = getopt(argc, argv, "lio:")) != -1) {
	switch (c) {
	case 'l':
	loadkey = B_TRUE;
	break;
	case 'i':
	inheritkey = B_TRUE;
	break;
	case 'o':
	if (!parseprop(props, optarg)) {
	nvlist_free(props);
	return (1);
	}
	break;
	default:
	(void) fprintf(stderr,
	gettext("invalid option '%c'\n"), optopt);
	usage(B_FALSE);
	}
	}

	if (inheritkey && !nvlist_empty(props)) {
	(void) fprintf(stderr,
	gettext("Properties not allowed for inheriting\n"));
	usage(B_FALSE);
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("Missing dataset argument\n"));
	usage(B_FALSE);
	}

	if (argc > 1) {
	(void) fprintf(stderr, gettext("Too many arguments\n"));
	usage(B_FALSE);
	}

	zhp = zfs_open(g_zfs, argv[argc - 1],
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME);
	if (zhp == NULL)
	usage(B_FALSE);

	if (loadkey) {
	keystatus = zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS);
	if (keystatus != ZFS_KEYSTATUS_AVAILABLE) {
	ret = zfs_crypto_load_key(zhp, B_FALSE, NULL);
	if (ret != 0) {
	nvlist_free(props);
	zfs_close(zhp);
	return (-1);
	}
	}

	/* refresh the properties so the new keystatus is visible */
	zfs_refresh_properties(zhp);
	}

	ret = zfs_crypto_rewrap(zhp, props, inheritkey);
	if (ret != 0) {
	nvlist_free(props);
	zfs_close(zhp);
	return (-1);
	}

	nvlist_free(props);
	zfs_close(zhp);
	return (0);
	}

	/*
	* 1) zfs project [-d\|-r] <file\|directory ...>
	* List project ID and inherit flag of file(s) or directories.
	* -d: List the directory itself, not its children.
	* -r: List subdirectories recursively.
	*
	* 2) zfs project -C [-k] [-r] <file\|directory ...>
	* Clear project inherit flag and/or ID on the file(s) or directories.
	* -k: Keep the project ID unchanged. If not specified, the project ID
	* will be reset as zero.
	* -r: Clear on subdirectories recursively.
	*
	* 3) zfs project -c [-0] [-d\|-r] [-p id] <file\|directory ...>
	* Check project ID and inherit flag on the file(s) or directories,
	* report the outliers.
	* -0: Print file name followed by a NUL instead of newline.
	* -d: Check the directory itself, not its children.
	* -p: Specify the referenced ID for comparing with the target file(s)
	* or directories' project IDs. If not specified, the target (top)
	* directory's project ID will be used as the referenced one.
	* -r: Check subdirectories recursively.
	*
	* 4) zfs project [-p id] [-r] [-s] <file\|directory ...>
	* Set project ID and/or inherit flag on the file(s) or directories.
	* -p: Set the project ID as the given id.
	* -r: Set on subdirectories recursively. If not specify "-p" option,
	* it will use top-level directory's project ID as the given id,
	* then set both project ID and inherit flag on all descendants
	* of the top-level directory.
	* -s: Set project inherit flag.
	*/
	static int
	zfs_do_project(int argc, char **argv)
	{
	zfs_project_control_t zpc = {
	.zpc_expected_projid = ZFS_INVALID_PROJID,
	.zpc_op = ZFS_PROJECT_OP_DEFAULT,
	.zpc_dironly = B_FALSE,
	.zpc_keep_projid = B_FALSE,
	.zpc_newline = B_TRUE,
	.zpc_recursive = B_FALSE,
	.zpc_set_flag = B_FALSE,
	};
	int ret = 0, c;

	if (argc < 2)
	usage(B_FALSE);

	while ((c = getopt(argc, argv, "0Ccdkp:rs")) != -1) {
	switch (c) {
	case '0':
	zpc.zpc_newline = B_FALSE;
	break;
	case 'C':
	if (zpc.zpc_op != ZFS_PROJECT_OP_DEFAULT) {
	(void) fprintf(stderr, gettext("cannot "
	"specify '-C' '-c' '-s' together\n"));
	usage(B_FALSE);
	}

	zpc.zpc_op = ZFS_PROJECT_OP_CLEAR;
	break;
	case 'c':
	if (zpc.zpc_op != ZFS_PROJECT_OP_DEFAULT) {
	(void) fprintf(stderr, gettext("cannot "
	"specify '-C' '-c' '-s' together\n"));
	usage(B_FALSE);
	}

	zpc.zpc_op = ZFS_PROJECT_OP_CHECK;
	break;
	case 'd':
	zpc.zpc_dironly = B_TRUE;
	/* overwrite "-r" option */
	zpc.zpc_recursive = B_FALSE;
	break;
	case 'k':
	zpc.zpc_keep_projid = B_TRUE;
	break;
	case 'p': {
	char *endptr;

	errno = 0;
	zpc.zpc_expected_projid = strtoull(optarg, &endptr, 0);
	if (errno != 0 \|\| *endptr != '\0') {
	(void) fprintf(stderr,
	gettext("project ID must be less than "
	"%u\n"), UINT32_MAX);
	usage(B_FALSE);
	}
	if (zpc.zpc_expected_projid >= UINT32_MAX) {
	(void) fprintf(stderr,
	gettext("invalid project ID\n"));
	usage(B_FALSE);
	}
	break;
	}
	case 'r':
	zpc.zpc_recursive = B_TRUE;
	/* overwrite "-d" option */
	zpc.zpc_dironly = B_FALSE;
	break;
	case 's':
	if (zpc.zpc_op != ZFS_PROJECT_OP_DEFAULT) {
	(void) fprintf(stderr, gettext("cannot "
	"specify '-C' '-c' '-s' together\n"));
	usage(B_FALSE);
	}

	zpc.zpc_set_flag = B_TRUE;
	zpc.zpc_op = ZFS_PROJECT_OP_SET;
	break;
	default:
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	if (zpc.zpc_op == ZFS_PROJECT_OP_DEFAULT) {
	if (zpc.zpc_expected_projid != ZFS_INVALID_PROJID)
	zpc.zpc_op = ZFS_PROJECT_OP_SET;
	else
	zpc.zpc_op = ZFS_PROJECT_OP_LIST;
	}

	switch (zpc.zpc_op) {
	case ZFS_PROJECT_OP_LIST:
	if (zpc.zpc_keep_projid) {
	(void) fprintf(stderr,
	gettext("'-k' is only valid together with '-C'\n"));
	usage(B_FALSE);
	}
	if (!zpc.zpc_newline) {
	(void) fprintf(stderr,
	gettext("'-0' is only valid together with '-c'\n"));
	usage(B_FALSE);
	}
	break;
	case ZFS_PROJECT_OP_CHECK:
	if (zpc.zpc_keep_projid) {
	(void) fprintf(stderr,
	gettext("'-k' is only valid together with '-C'\n"));
	usage(B_FALSE);
	}
	break;
	case ZFS_PROJECT_OP_CLEAR:
	if (zpc.zpc_dironly) {
	(void) fprintf(stderr,
	gettext("'-d' is useless together with '-C'\n"));
	usage(B_FALSE);
	}
	if (!zpc.zpc_newline) {
	(void) fprintf(stderr,
	gettext("'-0' is only valid together with '-c'\n"));
	usage(B_FALSE);
	}
	if (zpc.zpc_expected_projid != ZFS_INVALID_PROJID) {
	(void) fprintf(stderr,
	gettext("'-p' is useless together with '-C'\n"));
	usage(B_FALSE);
	}
	break;
	case ZFS_PROJECT_OP_SET:
	if (zpc.zpc_dironly) {
	(void) fprintf(stderr,
	gettext("'-d' is useless for set project ID and/or "
	"inherit flag\n"));
	usage(B_FALSE);
	}
	if (zpc.zpc_keep_projid) {
	(void) fprintf(stderr,
	gettext("'-k' is only valid together with '-C'\n"));
	usage(B_FALSE);
	}
	if (!zpc.zpc_newline) {
	(void) fprintf(stderr,
	gettext("'-0' is only valid together with '-c'\n"));
	usage(B_FALSE);
	}
	break;
	default:
	ASSERT(0);
	break;
	}

	argv += optind;
	argc -= optind;
	if (argc == 0) {
	(void) fprintf(stderr,
	gettext("missing file or directory target(s)\n"));
	usage(B_FALSE);
	}

	for (int i = 0; i < argc; i++) {
	int err;

	err = zfs_project_handle(argv[i], &zpc);
	if (err && !ret)
	ret = err;
	}

	return (ret);
	}

	static int
	zfs_do_wait(int argc, char **argv)
	{
	boolean_t enabled[ZFS_WAIT_NUM_ACTIVITIES];
	int error, i;
	int c;

	/* By default, wait for all types of activity. */
	for (i = 0; i < ZFS_WAIT_NUM_ACTIVITIES; i++)
	enabled[i] = B_TRUE;

	while ((c = getopt(argc, argv, "t:")) != -1) {
	switch (c) {
	case 't':
	{
	static char *col_subopts[] = { "deleteq", NULL };
	char *value;

	/* Reset activities array */
	bzero(&enabled, sizeof (enabled));
	while (*optarg != '\0') {
	int activity = getsubopt(&optarg, col_subopts,
	&value);

	if (activity < 0) {
	(void) fprintf(stderr,
	gettext("invalid activity '%s'\n"),
	value);
	usage(B_FALSE);
	}

	enabled[activity] = B_TRUE;
	}
	break;
	}
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argv += optind;
	argc -= optind;
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing 'filesystem' "
	"argument\n"));
	usage(B_FALSE);
	}
	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	zfs_handle_t *zhp = zfs_open(g_zfs, argv[0], ZFS_TYPE_FILESYSTEM);
	if (zhp == NULL)
	return (1);

	for (;;) {
	boolean_t missing = B_FALSE;
	boolean_t any_waited = B_FALSE;

	for (int i = 0; i < ZFS_WAIT_NUM_ACTIVITIES; i++) {
	boolean_t waited;

	if (!enabled[i])
	continue;

	error = zfs_wait_status(zhp, i, &missing, &waited);
	if (error != 0 \|\| missing)
	break;

	any_waited = (any_waited \|\| waited);
	}

	if (error != 0 \|\| missing \|\| !any_waited)
	break;
	}

	zfs_close(zhp);

	return (error);
	}

	/*
	* Display version message
	*/
	static int
	zfs_do_version(int argc, char **argv)
	{
	+ (void) argc, (void) argv;
	+
	if (zfs_version_print() == -1)
	return (1);

	return (0);
	}

	int
	main(int argc, char **argv)
	{
	int ret = 0;
	int i = 0;
	char *cmdname;
	char **newargv;

	(void) setlocale(LC_ALL, "");
	(void) setlocale(LC_NUMERIC, "C");
	(void) textdomain(TEXT_DOMAIN);

	opterr = 0;

	/*
	* Make sure the user has specified some command.
	*/
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing command\n"));
	usage(B_FALSE);
	}

	cmdname = argv[1];

	/*
	* The 'umount' command is an alias for 'unmount'
	*/
	if (strcmp(cmdname, "umount") == 0)
	cmdname = "unmount";

	/*
	* The 'recv' command is an alias for 'receive'
	*/
	if (strcmp(cmdname, "recv") == 0)
	cmdname = "receive";

	/*
	* The 'snap' command is an alias for 'snapshot'
	*/
	if (strcmp(cmdname, "snap") == 0)
	cmdname = "snapshot";

	/*
	* Special case '-?'
	*/
	if ((strcmp(cmdname, "-?") == 0) \|\|
	(strcmp(cmdname, "--help") == 0))
	usage(B_TRUE);

	/*
	* Special case '-V\|--version'
	*/
	if ((strcmp(cmdname, "-V") == 0) \|\| (strcmp(cmdname, "--version") == 0))
	return (zfs_do_version(argc, argv));

	if ((g_zfs = libzfs_init()) == NULL) {
	(void) fprintf(stderr, "%s\n", libzfs_error_init(errno));
	return (1);
	}

	zfs_save_arguments(argc, argv, history_str, sizeof (history_str));

	libzfs_print_on_error(g_zfs, B_TRUE);

	/*
	* Many commands modify input strings for string parsing reasons.
	* We create a copy to protect the original argv.
	*/
	newargv = malloc((argc + 1) * sizeof (newargv[0]));
	for (i = 0; i < argc; i++)
	newargv[i] = strdup(argv[i]);
	newargv[argc] = NULL;

	/*
	* Run the appropriate command.
	*/
	libzfs_mnttab_cache(g_zfs, B_TRUE);
	if (find_command_idx(cmdname, &i) == 0) {
	current_command = &command_table[i];
	ret = command_table[i].func(argc - 1, newargv + 1);
	} else if (strchr(cmdname, '=') != NULL) {
	verify(find_command_idx("set", &i) == 0);
	current_command = &command_table[i];
	ret = command_table[i].func(argc, newargv);
	} else {
	(void) fprintf(stderr, gettext("unrecognized "
	"command '%s'\n"), cmdname);
	usage(B_FALSE);
	ret = 1;
	}

	for (i = 0; i < argc; i++)
	free(newargv[i]);
	free(newargv);

	if (ret == 0 && log_history)
	(void) zpool_log_history(g_zfs, history_str);

	libzfs_fini(g_zfs);

	/*
	* The 'ZFS_ABORT' environment variable causes us to dump core on exit
	* for the purposes of running ::findleaks.
	*/
	if (getenv("ZFS_ABORT") != NULL) {
	(void) printf("dumping core by request\n");
	abort();
	}

	return (ret);
	}

	#ifdef __FreeBSD__
	#include <sys/jail.h>
	#include <jail.h>
	/*
	* Attach/detach the given dataset to/from the given jail
	*/
	-/* ARGSUSED */
	static int
	zfs_do_jail_impl(int argc, char **argv, boolean_t attach)
	{
	zfs_handle_t *zhp;
	int jailid, ret;

	/* check number of arguments */
	if (argc < 3) {
	(void) fprintf(stderr, gettext("missing argument(s)\n"));
	usage(B_FALSE);
	}
	if (argc > 3) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	jailid = jail_getid(argv[1]);
	if (jailid < 0) {
	(void) fprintf(stderr, gettext("invalid jail id or name\n"));
	usage(B_FALSE);
	}

	zhp = zfs_open(g_zfs, argv[2], ZFS_TYPE_FILESYSTEM);
	if (zhp == NULL)
	return (1);

	ret = (zfs_jail(zhp, jailid, attach) != 0);

	zfs_close(zhp);
	return (ret);
	}

	/*
	* zfs jail jailid filesystem
	*
	* Attach the given dataset to the given jail
	*/
	-/* ARGSUSED */
	static int
	zfs_do_jail(int argc, char **argv)
	{
	return (zfs_do_jail_impl(argc, argv, B_TRUE));
	}

	/*
	* zfs unjail jailid filesystem
	*
	* Detach the given dataset from the given jail
	*/
	-/* ARGSUSED */
	static int
	zfs_do_unjail(int argc, char **argv)
	{
	return (zfs_do_jail_impl(argc, argv, B_FALSE));
	}
	#endif
	diff --git a/sys/contrib/openzfs/cmd/zpool/zpool_iter.c b/sys/contrib/openzfs/cmd/zpool/zpool_iter.c
	index abfa2b7f6b90..8cf6bab42ff1 100644
	--- a/sys/contrib/openzfs/cmd/zpool/zpool_iter.c
	+++ b/sys/contrib/openzfs/cmd/zpool/zpool_iter.c
	@@ -1,724 +1,727 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright 2007 Sun Microsystems, Inc. All rights reserved.
	* Use is subject to license terms.
	*/

	/*
	* Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>.
	*/

	#include <libintl.h>
	#include <libuutil.h>
	#include <stddef.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <strings.h>
	#include <thread_pool.h>

	#include <libzfs.h>
	#include <libzutil.h>
	#include <sys/zfs_context.h>
	#include <sys/wait.h>

	#include "zpool_util.h"

	/*
	* Private interface for iterating over pools specified on the command line.
	* Most consumers will call for_each_pool, but in order to support iostat, we
	* allow fined grained control through the zpool_list_t interface.
	*/

	typedef struct zpool_node {
	zpool_handle_t *zn_handle;
	uu_avl_node_t zn_avlnode;
	int zn_mark;
	} zpool_node_t;

	struct zpool_list {
	boolean_t zl_findall;
	boolean_t zl_literal;
	uu_avl_t *zl_avl;
	uu_avl_pool_t *zl_pool;
	zprop_list_t **zl_proplist;
	+ zfs_type_t zl_type;
	};

	/* ARGSUSED */
	static int
	zpool_compare(const void larg, const void rarg, void *unused)
	{
	zpool_handle_t l = ((zpool_node_t )larg)->zn_handle;
	zpool_handle_t r = ((zpool_node_t )rarg)->zn_handle;
	const char *lname = zpool_get_name(l);
	const char *rname = zpool_get_name(r);

	return (strcmp(lname, rname));
	}

	/*
	* Callback function for pool_list_get(). Adds the given pool to the AVL tree
	* of known pools.
	*/
	static int
	add_pool(zpool_handle_t zhp, void data)
	{
	zpool_list_t *zlp = data;
	zpool_node_t *node = safe_malloc(sizeof (zpool_node_t));
	uu_avl_index_t idx;

	node->zn_handle = zhp;
	uu_avl_node_init(node, &node->zn_avlnode, zlp->zl_pool);
	if (uu_avl_find(zlp->zl_avl, node, NULL, &idx) == NULL) {
	if (zlp->zl_proplist &&
	zpool_expand_proplist(zhp, zlp->zl_proplist,
	- zlp->zl_literal)
	- != 0) {
	+ zlp->zl_type, zlp->zl_literal) != 0) {
	zpool_close(zhp);
	free(node);
	return (-1);
	}
	uu_avl_insert(zlp->zl_avl, node, idx);
	} else {
	zpool_close(zhp);
	free(node);
	return (-1);
	}

	return (0);
	}

	/*
	* Create a list of pools based on the given arguments. If we're given no
	* arguments, then iterate over all pools in the system and add them to the AVL
	* tree. Otherwise, add only those pool explicitly specified on the command
	* line.
	*/
	zpool_list_t *
	-pool_list_get(int argc, char argv, zprop_list_t proplist,
	+pool_list_get(int argc, char argv, zprop_list_t proplist, zfs_type_t type,
	boolean_t literal, int *err)
	{
	zpool_list_t *zlp;

	zlp = safe_malloc(sizeof (zpool_list_t));

	zlp->zl_pool = uu_avl_pool_create("zfs_pool", sizeof (zpool_node_t),
	offsetof(zpool_node_t, zn_avlnode), zpool_compare, UU_DEFAULT);

	if (zlp->zl_pool == NULL)
	zpool_no_memory();

	if ((zlp->zl_avl = uu_avl_create(zlp->zl_pool, NULL,
	UU_DEFAULT)) == NULL)
	zpool_no_memory();

	zlp->zl_proplist = proplist;
	+ zlp->zl_type = type;

	zlp->zl_literal = literal;

	if (argc == 0) {
	(void) zpool_iter(g_zfs, add_pool, zlp);
	zlp->zl_findall = B_TRUE;
	} else {
	int i;

	for (i = 0; i < argc; i++) {
	zpool_handle_t *zhp;

	if ((zhp = zpool_open_canfail(g_zfs, argv[i])) !=
	NULL) {
	if (add_pool(zhp, zlp) != 0)
	*err = B_TRUE;
	} else {
	*err = B_TRUE;
	}
	}
	}

	return (zlp);
	}

	/*
	* Search for any new pools, adding them to the list. We only add pools when no
	* options were given on the command line. Otherwise, we keep the list fixed as
	* those that were explicitly specified.
	*/
	void
	pool_list_update(zpool_list_t *zlp)
	{
	if (zlp->zl_findall)
	(void) zpool_iter(g_zfs, add_pool, zlp);
	}

	/*
	* Iterate over all pools in the list, executing the callback for each
	*/
	int
	pool_list_iter(zpool_list_t *zlp, int unavail, zpool_iter_f func,
	void *data)
	{
	zpool_node_t node, next_node;
	int ret = 0;

	for (node = uu_avl_first(zlp->zl_avl); node != NULL; node = next_node) {
	next_node = uu_avl_next(zlp->zl_avl, node);
	if (zpool_get_state(node->zn_handle) != POOL_STATE_UNAVAIL \|\|
	unavail)
	ret \|= func(node->zn_handle, data);
	}

	return (ret);
	}

	/*
	* Remove the given pool from the list. When running iostat, we want to remove
	* those pools that no longer exist.
	*/
	void
	pool_list_remove(zpool_list_t zlp, zpool_handle_t zhp)
	{
	zpool_node_t search, *node;

	search.zn_handle = zhp;
	if ((node = uu_avl_find(zlp->zl_avl, &search, NULL, NULL)) != NULL) {
	uu_avl_remove(zlp->zl_avl, node);
	zpool_close(node->zn_handle);
	free(node);
	}
	}

	/*
	* Free all the handles associated with this list.
	*/
	void
	pool_list_free(zpool_list_t *zlp)
	{
	uu_avl_walk_t *walk;
	zpool_node_t *node;

	if ((walk = uu_avl_walk_start(zlp->zl_avl, UU_WALK_ROBUST)) == NULL) {
	(void) fprintf(stderr,
	gettext("internal error: out of memory"));
	exit(1);
	}

	while ((node = uu_avl_walk_next(walk)) != NULL) {
	uu_avl_remove(zlp->zl_avl, node);
	zpool_close(node->zn_handle);
	free(node);
	}

	uu_avl_walk_end(walk);
	uu_avl_destroy(zlp->zl_avl);
	uu_avl_pool_destroy(zlp->zl_pool);

	free(zlp);
	}

	/*
	* Returns the number of elements in the pool list.
	*/
	int
	pool_list_count(zpool_list_t *zlp)
	{
	return (uu_avl_numnodes(zlp->zl_avl));
	}

	/*
	* High level function which iterates over all pools given on the command line,
	* using the pool_list_* interfaces.
	*/
	int
	for_each_pool(int argc, char **argv, boolean_t unavail,
	- zprop_list_t *proplist, boolean_t literal, zpool_iter_f func, void data)
	+ zprop_list_t **proplist, zfs_type_t type, boolean_t literal,
	+ zpool_iter_f func, void *data)
	{
	zpool_list_t *list;
	int ret = 0;

	- if ((list = pool_list_get(argc, argv, proplist, literal, &ret)) == NULL)
	+ if ((list = pool_list_get(argc, argv, proplist, type, literal,
	+ &ret)) == NULL)
	return (1);

	if (pool_list_iter(list, unavail, func, data) != 0)
	ret = 1;

	pool_list_free(list);

	return (ret);
	}

	/*
	* This is the equivalent of for_each_pool() for vdevs. It iterates thorough
	* all vdevs in the pool, ignoring root vdevs and holes, calling func() on
	* each one.
	*
	* @zhp: Zpool handle
	* @func: Function to call on each vdev
	* @data: Custom data to pass to the function
	*/
	int
	for_each_vdev(zpool_handle_t zhp, pool_vdev_iter_f func, void data)
	{
	nvlist_t config, nvroot = NULL;

	if ((config = zpool_get_config(zhp, NULL)) != NULL) {
	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) == 0);
	}
	return (for_each_vdev_cb((void *) zhp, nvroot, func, data));
	}

	/*
	* Process the vcdl->vdev_cmd_data[] array to figure out all the unique column
	* names and their widths. When this function is done, vcdl->uniq_cols,
	* vcdl->uniq_cols_cnt, and vcdl->uniq_cols_width will be filled in.
	*/
	static void
	process_unique_cmd_columns(vdev_cmd_data_list_t *vcdl)
	{
	char uniq_cols = NULL, tmp = NULL;
	int *uniq_cols_width;
	vdev_cmd_data_t *data;
	int cnt = 0;
	int k;

	/* For each vdev */
	for (int i = 0; i < vcdl->count; i++) {
	data = &vcdl->data[i];
	/* For each column the vdev reported */
	for (int j = 0; j < data->cols_cnt; j++) {
	/* Is this column in our list of unique column names? */
	for (k = 0; k < cnt; k++) {
	if (strcmp(data->cols[j], uniq_cols[k]) == 0)
	break; /* yes it is */
	}
	if (k == cnt) {
	/* No entry for column, add to list */
	tmp = realloc(uniq_cols, sizeof (uniq_cols)
	(cnt + 1));
	if (tmp == NULL)
	break; /* Nothing we can do... */
	uniq_cols = tmp;
	uniq_cols[cnt] = data->cols[j];
	cnt++;
	}
	}
	}

	/*
	* We now have a list of all the unique column names. Figure out the
	* max width of each column by looking at the column name and all its
	* values.
	*/
	uniq_cols_width = safe_malloc(sizeof (uniq_cols_width) cnt);
	for (int i = 0; i < cnt; i++) {
	/* Start off with the column title's width */
	uniq_cols_width[i] = strlen(uniq_cols[i]);
	/* For each vdev */
	for (int j = 0; j < vcdl->count; j++) {
	/* For each of the vdev's values in a column */
	data = &vcdl->data[j];
	for (k = 0; k < data->cols_cnt; k++) {
	/* Does this vdev have a value for this col? */
	if (strcmp(data->cols[k], uniq_cols[i]) == 0) {
	/* Is the value width larger? */
	uniq_cols_width[i] =
	MAX(uniq_cols_width[i],
	strlen(data->lines[k]));
	}
	}
	}
	}

	vcdl->uniq_cols = uniq_cols;
	vcdl->uniq_cols_cnt = cnt;
	vcdl->uniq_cols_width = uniq_cols_width;
	}


	/*
	* Process a line of command output
	*
	* When running 'zpool iostat\|status -c' the lines of output can either be
	* in the form of:
	*
	* column_name=value
	*
	* Or just:
	*
	* value
	*
	* Process the column_name (if any) and value.
	*
	* Returns 0 if line was processed, and there are more lines can still be
	* processed.
	*
	* Returns 1 if this was the last line to process, or error.
	*/
	static int
	vdev_process_cmd_output(vdev_cmd_data_t data, char line)
	{
	char *col = NULL;
	char *val = line;
	char *equals;
	char **tmp;

	if (line == NULL)
	return (1);

	equals = strchr(line, '=');
	if (equals != NULL) {
	/*
	* We have a 'column=value' type line. Split it into the
	* column and value strings by turning the '=' into a '\0'.
	*/
	*equals = '\0';
	col = line;
	val = equals + 1;
	} else {
	val = line;
	}

	/* Do we already have a column by this name? If so, skip it. */
	if (col != NULL) {
	for (int i = 0; i < data->cols_cnt; i++) {
	if (strcmp(col, data->cols[i]) == 0)
	return (0); /* Duplicate, skip */
	}
	}

	if (val != NULL) {
	tmp = realloc(data->lines,
	(data->lines_cnt + 1) * sizeof (*data->lines));
	if (tmp == NULL)
	return (1);

	data->lines = tmp;
	data->lines[data->lines_cnt] = strdup(val);
	data->lines_cnt++;
	}

	if (col != NULL) {
	tmp = realloc(data->cols,
	(data->cols_cnt + 1) * sizeof (*data->cols));
	if (tmp == NULL)
	return (1);

	data->cols = tmp;
	data->cols[data->cols_cnt] = strdup(col);
	data->cols_cnt++;
	}

	if (val != NULL && col == NULL)
	return (1);

	return (0);
	}

	/*
	* Run the cmd and store results in *data.
	*/
	static void
	vdev_run_cmd(vdev_cmd_data_t data, char cmd)
	{
	int rc;
	char *argv[2] = {cmd, 0};
	char *env[5] = {"PATH=/bin:/sbin:/usr/bin:/usr/sbin", NULL, NULL, NULL,
	NULL};
	char **lines = NULL;
	int lines_cnt = 0;
	int i;

	/* Setup our custom environment variables */
	rc = asprintf(&env[1], "VDEV_PATH=%s",
	data->path ? data->path : "");
	if (rc == -1) {
	env[1] = NULL;
	goto out;
	}

	rc = asprintf(&env[2], "VDEV_UPATH=%s",
	data->upath ? data->upath : "");
	if (rc == -1) {
	env[2] = NULL;
	goto out;
	}

	rc = asprintf(&env[3], "VDEV_ENC_SYSFS_PATH=%s",
	data->vdev_enc_sysfs_path ?
	data->vdev_enc_sysfs_path : "");
	if (rc == -1) {
	env[3] = NULL;
	goto out;
	}

	/* Run the command */
	rc = libzfs_run_process_get_stdout_nopath(cmd, argv, env, &lines,
	&lines_cnt);
	if (rc != 0)
	goto out;

	/* Process the output we got */
	for (i = 0; i < lines_cnt; i++)
	if (vdev_process_cmd_output(data, lines[i]) != 0)
	break;

	out:
	if (lines != NULL)
	libzfs_free_str_array(lines, lines_cnt);

	/* Start with i = 1 since env[0] was statically allocated */
	for (i = 1; i < ARRAY_SIZE(env); i++)
	free(env[i]);
	}

	/*
	* Generate the search path for zpool iostat/status -c scripts.
	* The string returned must be freed.
	*/
	char *
	zpool_get_cmd_search_path(void)
	{
	const char *env;
	char *sp = NULL;

	env = getenv("ZPOOL_SCRIPTS_PATH");
	if (env != NULL)
	return (strdup(env));

	env = getenv("HOME");
	if (env != NULL) {
	if (asprintf(&sp, "%s/.zpool.d:%s",
	env, ZPOOL_SCRIPTS_DIR) != -1) {
	return (sp);
	}
	}

	if (asprintf(&sp, "%s", ZPOOL_SCRIPTS_DIR) != -1)
	return (sp);

	return (NULL);
	}

	/* Thread function run for each vdev */
	static void
	vdev_run_cmd_thread(void *cb_cmd_data)
	{
	vdev_cmd_data_t *data = cb_cmd_data;
	char cmd = NULL, cmddup, *cmdrest;

	cmddup = strdup(data->cmd);
	if (cmddup == NULL)
	return;

	cmdrest = cmddup;
	while ((cmd = strtok_r(cmdrest, ",", &cmdrest))) {
	char dir = NULL, sp, *sprest;
	char fullpath[MAXPATHLEN];

	if (strchr(cmd, '/') != NULL)
	continue;

	sp = zpool_get_cmd_search_path();
	if (sp == NULL)
	continue;

	sprest = sp;
	while ((dir = strtok_r(sprest, ":", &sprest))) {
	if (snprintf(fullpath, sizeof (fullpath),
	"%s/%s", dir, cmd) == -1)
	continue;

	if (access(fullpath, X_OK) == 0) {
	vdev_run_cmd(data, fullpath);
	break;
	}
	}
	free(sp);
	}
	free(cmddup);
	}

	/* For each vdev in the pool run a command */
	static int
	for_each_vdev_run_cb(void zhp_data, nvlist_t nv, void *cb_vcdl)
	{
	vdev_cmd_data_list_t *vcdl = cb_vcdl;
	vdev_cmd_data_t *data;
	char *path = NULL;
	char *vname = NULL;
	char *vdev_enc_sysfs_path = NULL;
	int i, match = 0;
	zpool_handle_t *zhp = zhp_data;

	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) != 0)
	return (1);

	nvlist_lookup_string(nv, ZPOOL_CONFIG_VDEV_ENC_SYSFS_PATH,
	&vdev_enc_sysfs_path);

	/* Spares show more than once if they're in use, so skip if exists */
	for (i = 0; i < vcdl->count; i++) {
	if ((strcmp(vcdl->data[i].path, path) == 0) &&
	(strcmp(vcdl->data[i].pool, zpool_get_name(zhp)) == 0)) {
	/* vdev already exists, skip it */
	return (0);
	}
	}

	/* Check for selected vdevs here, if any */
	for (i = 0; i < vcdl->vdev_names_count; i++) {
	vname = zpool_vdev_name(g_zfs, zhp, nv, vcdl->cb_name_flags);
	if (strcmp(vcdl->vdev_names[i], vname) == 0) {
	free(vname);
	match = 1;
	break; /* match */
	}
	free(vname);
	}

	/* If we selected vdevs, and this isn't one of them, then bail out */
	if (!match && vcdl->vdev_names_count)
	return (0);

	/*
	* Resize our array and add in the new element.
	*/
	if (!(vcdl->data = realloc(vcdl->data,
	sizeof (vcdl->data) (vcdl->count + 1))))
	return (ENOMEM); /* couldn't realloc */

	data = &vcdl->data[vcdl->count];

	data->pool = strdup(zpool_get_name(zhp));
	data->path = strdup(path);
	data->upath = zfs_get_underlying_path(path);
	data->cmd = vcdl->cmd;
	data->lines = data->cols = NULL;
	data->lines_cnt = data->cols_cnt = 0;
	if (vdev_enc_sysfs_path)
	data->vdev_enc_sysfs_path = strdup(vdev_enc_sysfs_path);
	else
	data->vdev_enc_sysfs_path = NULL;

	vcdl->count++;

	return (0);
	}

	/* Get the names and count of the vdevs */
	static int
	all_pools_for_each_vdev_gather_cb(zpool_handle_t zhp, void cb_vcdl)
	{
	return (for_each_vdev(zhp, for_each_vdev_run_cb, cb_vcdl));
	}

	/*
	* Now that vcdl is populated with our complete list of vdevs, spawn
	* off the commands.
	*/
	static void
	all_pools_for_each_vdev_run_vcdl(vdev_cmd_data_list_t *vcdl)
	{
	tpool_t *t;

	t = tpool_create(1, 5 * sysconf(_SC_NPROCESSORS_ONLN), 0, NULL);
	if (t == NULL)
	return;

	/* Spawn off the command for each vdev */
	for (int i = 0; i < vcdl->count; i++) {
	(void) tpool_dispatch(t, vdev_run_cmd_thread,
	(void *) &vcdl->data[i]);
	}

	/* Wait for threads to finish */
	tpool_wait(t);
	tpool_destroy(t);
	}

	/*
	* Run command 'cmd' on all vdevs in all pools in argv. Saves the first line of
	* output from the command in vcdk->data[].line for all vdevs. If you want
	* to run the command on only certain vdevs, fill in g_zfs, vdev_names,
	* vdev_names_count, and cb_name_flags. Otherwise leave them as zero.
	*
	* Returns a vdev_cmd_data_list_t that must be freed with
	* free_vdev_cmd_data_list();
	*/
	vdev_cmd_data_list_t *
	all_pools_for_each_vdev_run(int argc, char *argv, char cmd,
	libzfs_handle_t g_zfs, char *vdev_names, int vdev_names_count,
	int cb_name_flags)
	{
	vdev_cmd_data_list_t *vcdl;
	vcdl = safe_malloc(sizeof (vdev_cmd_data_list_t));
	vcdl->cmd = cmd;

	vcdl->vdev_names = vdev_names;
	vcdl->vdev_names_count = vdev_names_count;
	vcdl->cb_name_flags = cb_name_flags;
	vcdl->g_zfs = g_zfs;

	/* Gather our list of all vdevs in all pools */
	- for_each_pool(argc, argv, B_TRUE, NULL, B_FALSE,
	- all_pools_for_each_vdev_gather_cb, vcdl);
	+ for_each_pool(argc, argv, B_TRUE, NULL, ZFS_TYPE_POOL,
	+ B_FALSE, all_pools_for_each_vdev_gather_cb, vcdl);

	/* Run command on all vdevs in all pools */
	all_pools_for_each_vdev_run_vcdl(vcdl);

	/*
	* vcdl->data[] now contains all the column names and values for each
	* vdev. We need to process that into a master list of unique column
	* names, and figure out the width of each column.
	*/
	process_unique_cmd_columns(vcdl);

	return (vcdl);
	}

	/*
	* Free the vdev_cmd_data_list_t created by all_pools_for_each_vdev_run()
	*/
	void
	free_vdev_cmd_data_list(vdev_cmd_data_list_t *vcdl)
	{
	free(vcdl->uniq_cols);
	free(vcdl->uniq_cols_width);

	for (int i = 0; i < vcdl->count; i++) {
	free(vcdl->data[i].path);
	free(vcdl->data[i].pool);
	free(vcdl->data[i].upath);

	for (int j = 0; j < vcdl->data[i].lines_cnt; j++)
	free(vcdl->data[i].lines[j]);

	free(vcdl->data[i].lines);

	for (int j = 0; j < vcdl->data[i].cols_cnt; j++)
	free(vcdl->data[i].cols[j]);

	free(vcdl->data[i].cols);
	free(vcdl->data[i].vdev_enc_sysfs_path);
	}
	free(vcdl->data);
	free(vcdl);
	}
	diff --git a/sys/contrib/openzfs/cmd/zpool/zpool_main.c b/sys/contrib/openzfs/cmd/zpool/zpool_main.c
	index 3a2caa9a8101..4e2f828cb2ba 100644
	--- a/sys/contrib/openzfs/cmd/zpool/zpool_main.c
	+++ b/sys/contrib/openzfs/cmd/zpool/zpool_main.c
	@@ -1,10775 +1,11000 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2012 by Frederik Wessels. All rights reserved.
	* Copyright (c) 2012 by Cyril Plisko. All rights reserved.
	* Copyright (c) 2013 by Prasad Joshi (sTec). All rights reserved.
	* Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>.
	* Copyright (c) 2017 Datto Inc.
	* Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
	* Copyright (c) 2017, Intel Corporation.
	* Copyright (c) 2019, loli10K <ezomori.nozomu@gmail.com>
	* Copyright (c) 2021, Colm Buckley <colm@tuatha.org>
	+ * Copyright (c) 2021, Klara Inc.
	* Copyright [2021] Hewlett Packard Enterprise Development LP
	*/

	#include <assert.h>
	#include <ctype.h>
	#include <dirent.h>
	#include <errno.h>
	#include <fcntl.h>
	#include <getopt.h>
	#include <libgen.h>
	#include <libintl.h>
	#include <libuutil.h>
	#include <locale.h>
	#include <pthread.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <string.h>
	#include <strings.h>
	#include <time.h>
	#include <unistd.h>
	#include <pwd.h>
	#include <zone.h>
	#include <sys/wait.h>
	#include <zfs_prop.h>
	#include <sys/fs/zfs.h>
	#include <sys/stat.h>
	#include <sys/systeminfo.h>
	#include <sys/fm/fs/zfs.h>
	#include <sys/fm/util.h>
	#include <sys/fm/protocol.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/mount.h>
	#include <sys/sysmacros.h>

	#include <math.h>

	#include <libzfs.h>
	#include <libzutil.h>

	#include "zpool_util.h"
	#include "zfs_comutil.h"
	#include "zfeature_common.h"

	#include "statcommon.h"

	libzfs_handle_t *g_zfs;

	static int zpool_do_create(int, char **);
	static int zpool_do_destroy(int, char **);

	static int zpool_do_add(int, char **);
	static int zpool_do_remove(int, char **);
	static int zpool_do_labelclear(int, char **);

	static int zpool_do_checkpoint(int, char **);

	static int zpool_do_list(int, char **);
	static int zpool_do_iostat(int, char **);
	static int zpool_do_status(int, char **);

	static int zpool_do_online(int, char **);
	static int zpool_do_offline(int, char **);
	static int zpool_do_clear(int, char **);
	static int zpool_do_reopen(int, char **);

	static int zpool_do_reguid(int, char **);

	static int zpool_do_attach(int, char **);
	static int zpool_do_detach(int, char **);
	static int zpool_do_replace(int, char **);
	static int zpool_do_split(int, char **);

	static int zpool_do_initialize(int, char **);
	static int zpool_do_scrub(int, char **);
	static int zpool_do_resilver(int, char **);
	static int zpool_do_trim(int, char **);

	static int zpool_do_import(int, char **);
	static int zpool_do_export(int, char **);

	static int zpool_do_upgrade(int, char **);

	static int zpool_do_history(int, char **);
	static int zpool_do_events(int, char **);

	static int zpool_do_get(int, char **);
	static int zpool_do_set(int, char **);

	static int zpool_do_sync(int, char **);

	static int zpool_do_version(int, char **);

	static int zpool_do_wait(int, char **);

	static zpool_compat_status_t zpool_do_load_compat(
	const char , boolean_t );

	/*
	* These libumem hooks provide a reasonable set of defaults for the allocator's
	* debugging facilities.
	*/

	#ifdef DEBUG
	const char *
	_umem_debug_init(void)
	{
	return ("default,verbose"); /* $UMEM_DEBUG setting */
	}

	const char *
	_umem_logging_init(void)
	{
	return ("fail,contents"); /* $UMEM_LOGGING setting */
	}
	#endif

	typedef enum {
	HELP_ADD,
	HELP_ATTACH,
	HELP_CLEAR,
	HELP_CREATE,
	HELP_CHECKPOINT,
	HELP_DESTROY,
	HELP_DETACH,
	HELP_EXPORT,
	HELP_HISTORY,
	HELP_IMPORT,
	HELP_IOSTAT,
	HELP_LABELCLEAR,
	HELP_LIST,
	HELP_OFFLINE,
	HELP_ONLINE,
	HELP_REPLACE,
	HELP_REMOVE,
	HELP_INITIALIZE,
	HELP_SCRUB,
	HELP_RESILVER,
	HELP_TRIM,
	HELP_STATUS,
	HELP_UPGRADE,
	HELP_EVENTS,
	HELP_GET,
	HELP_SET,
	HELP_SPLIT,
	HELP_SYNC,
	HELP_REGUID,
	HELP_REOPEN,
	HELP_VERSION,
	HELP_WAIT
	} zpool_help_t;


	/*
	* Flags for stats to display with "zpool iostats"
	*/
	enum iostat_type {
	IOS_DEFAULT = 0,
	IOS_LATENCY = 1,
	IOS_QUEUES = 2,
	IOS_L_HISTO = 3,
	IOS_RQ_HISTO = 4,
	IOS_COUNT, /* always last element */
	};

	/* iostat_type entries as bitmasks */
	#define IOS_DEFAULT_M (1ULL << IOS_DEFAULT)
	#define IOS_LATENCY_M (1ULL << IOS_LATENCY)
	#define IOS_QUEUES_M (1ULL << IOS_QUEUES)
	#define IOS_L_HISTO_M (1ULL << IOS_L_HISTO)
	#define IOS_RQ_HISTO_M (1ULL << IOS_RQ_HISTO)

	/* Mask of all the histo bits */
	#define IOS_ANYHISTO_M (IOS_L_HISTO_M \| IOS_RQ_HISTO_M)

	/*
	* Lookup table for iostat flags to nvlist names. Basically a list
	* of all the nvlists a flag requires. Also specifies the order in
	* which data gets printed in zpool iostat.
	*/
	static const char *vsx_type_to_nvlist[IOS_COUNT][15] = {
	[IOS_L_HISTO] = {
	ZPOOL_CONFIG_VDEV_TOT_R_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_TOT_W_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_DISK_R_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_DISK_W_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_SYNC_R_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_SYNC_W_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_ASYNC_R_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_ASYNC_W_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_SCRUB_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_TRIM_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_REBUILD_LAT_HISTO,
	NULL},
	[IOS_LATENCY] = {
	ZPOOL_CONFIG_VDEV_TOT_R_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_TOT_W_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_DISK_R_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_DISK_W_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_TRIM_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_REBUILD_LAT_HISTO,
	NULL},
	[IOS_QUEUES] = {
	ZPOOL_CONFIG_VDEV_SYNC_R_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_SYNC_W_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_ASYNC_R_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_ASYNC_W_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_SCRUB_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_TRIM_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_REBUILD_ACTIVE_QUEUE,
	NULL},
	[IOS_RQ_HISTO] = {
	ZPOOL_CONFIG_VDEV_SYNC_IND_R_HISTO,
	ZPOOL_CONFIG_VDEV_SYNC_AGG_R_HISTO,
	ZPOOL_CONFIG_VDEV_SYNC_IND_W_HISTO,
	ZPOOL_CONFIG_VDEV_SYNC_AGG_W_HISTO,
	ZPOOL_CONFIG_VDEV_ASYNC_IND_R_HISTO,
	ZPOOL_CONFIG_VDEV_ASYNC_AGG_R_HISTO,
	ZPOOL_CONFIG_VDEV_ASYNC_IND_W_HISTO,
	ZPOOL_CONFIG_VDEV_ASYNC_AGG_W_HISTO,
	ZPOOL_CONFIG_VDEV_IND_SCRUB_HISTO,
	ZPOOL_CONFIG_VDEV_AGG_SCRUB_HISTO,
	ZPOOL_CONFIG_VDEV_IND_TRIM_HISTO,
	ZPOOL_CONFIG_VDEV_AGG_TRIM_HISTO,
	ZPOOL_CONFIG_VDEV_IND_REBUILD_HISTO,
	ZPOOL_CONFIG_VDEV_AGG_REBUILD_HISTO,
	NULL},
	};


	/*
	* Given a cb->cb_flags with a histogram bit set, return the iostat_type.
	* Right now, only one histo bit is ever set at one time, so we can
	* just do a highbit64(a)
	*/
	#define IOS_HISTO_IDX(a) (highbit64(a & IOS_ANYHISTO_M) - 1)

	typedef struct zpool_command {
	const char *name;
	int (func)(int, char *);
	zpool_help_t usage;
	} zpool_command_t;

	/*
	* Master command table. Each ZFS command has a name, associated function, and
	* usage message. The usage messages need to be internationalized, so we have
	* to have a function to return the usage message based on a command index.
	*
	* These commands are organized according to how they are displayed in the usage
	* message. An empty command (one with a NULL name) indicates an empty line in
	* the generic usage message.
	*/
	static zpool_command_t command_table[] = {
	{ "version", zpool_do_version, HELP_VERSION },
	{ NULL },
	{ "create", zpool_do_create, HELP_CREATE },
	{ "destroy", zpool_do_destroy, HELP_DESTROY },
	{ NULL },
	{ "add", zpool_do_add, HELP_ADD },
	{ "remove", zpool_do_remove, HELP_REMOVE },
	{ NULL },
	{ "labelclear", zpool_do_labelclear, HELP_LABELCLEAR },
	{ NULL },
	{ "checkpoint", zpool_do_checkpoint, HELP_CHECKPOINT },
	{ NULL },
	{ "list", zpool_do_list, HELP_LIST },
	{ "iostat", zpool_do_iostat, HELP_IOSTAT },
	{ "status", zpool_do_status, HELP_STATUS },
	{ NULL },
	{ "online", zpool_do_online, HELP_ONLINE },
	{ "offline", zpool_do_offline, HELP_OFFLINE },
	{ "clear", zpool_do_clear, HELP_CLEAR },
	{ "reopen", zpool_do_reopen, HELP_REOPEN },
	{ NULL },
	{ "attach", zpool_do_attach, HELP_ATTACH },
	{ "detach", zpool_do_detach, HELP_DETACH },
	{ "replace", zpool_do_replace, HELP_REPLACE },
	{ "split", zpool_do_split, HELP_SPLIT },
	{ NULL },
	{ "initialize", zpool_do_initialize, HELP_INITIALIZE },
	{ "resilver", zpool_do_resilver, HELP_RESILVER },
	{ "scrub", zpool_do_scrub, HELP_SCRUB },
	{ "trim", zpool_do_trim, HELP_TRIM },
	{ NULL },
	{ "import", zpool_do_import, HELP_IMPORT },
	{ "export", zpool_do_export, HELP_EXPORT },
	{ "upgrade", zpool_do_upgrade, HELP_UPGRADE },
	{ "reguid", zpool_do_reguid, HELP_REGUID },
	{ NULL },
	{ "history", zpool_do_history, HELP_HISTORY },
	{ "events", zpool_do_events, HELP_EVENTS },
	{ NULL },
	{ "get", zpool_do_get, HELP_GET },
	{ "set", zpool_do_set, HELP_SET },
	{ "sync", zpool_do_sync, HELP_SYNC },
	{ NULL },
	{ "wait", zpool_do_wait, HELP_WAIT },
	};

	#define NCOMMAND (ARRAY_SIZE(command_table))

	#define VDEV_ALLOC_CLASS_LOGS "logs"

	static zpool_command_t *current_command;
	+static zfs_type_t current_prop_type = (ZFS_TYPE_POOL \| ZFS_TYPE_VDEV);
	static char history_str[HIS_MAX_RECORD_LEN];
	static boolean_t log_history = B_TRUE;
	static uint_t timestamp_fmt = NODATE;

	static const char *
	get_usage(zpool_help_t idx)
	{
	switch (idx) {
	case HELP_ADD:
	return (gettext("\tadd [-fgLnP] [-o property=value] "
	"<pool> <vdev> ...\n"));
	case HELP_ATTACH:
	return (gettext("\tattach [-fsw] [-o property=value] "
	"<pool> <device> <new-device>\n"));
	case HELP_CLEAR:
	return (gettext("\tclear [-nF] <pool> [device]\n"));
	case HELP_CREATE:
	return (gettext("\tcreate [-fnd] [-o property=value] ... \n"
	"\t [-O file-system-property=value] ... \n"
	"\t [-m mountpoint] [-R root] <pool> <vdev> ...\n"));
	case HELP_CHECKPOINT:
	return (gettext("\tcheckpoint [-d [-w]] <pool> ...\n"));
	case HELP_DESTROY:
	return (gettext("\tdestroy [-f] <pool>\n"));
	case HELP_DETACH:
	return (gettext("\tdetach <pool> <device>\n"));
	case HELP_EXPORT:
	return (gettext("\texport [-af] <pool> ...\n"));
	case HELP_HISTORY:
	return (gettext("\thistory [-il] [<pool>] ...\n"));
	case HELP_IMPORT:
	return (gettext("\timport [-d dir] [-D]\n"
	"\timport [-o mntopts] [-o property=value] ... \n"
	"\t [-d dir \| -c cachefile] [-D] [-l] [-f] [-m] [-N] "
	"[-R root] [-F [-n]] -a\n"
	"\timport [-o mntopts] [-o property=value] ... \n"
	"\t [-d dir \| -c cachefile] [-D] [-l] [-f] [-m] [-N] "
	"[-R root] [-F [-n]]\n"
	"\t [--rewind-to-checkpoint] <pool \| id> [newpool]\n"));
	case HELP_IOSTAT:
	return (gettext("\tiostat [[[-c [script1,script2,...]"
	"[-lq]]\|[-rw]] [-T d \| u] [-ghHLpPvy]\n"
	"\t [[pool ...]\|[pool vdev ...]\|[vdev ...]]"
	" [[-n] interval [count]]\n"));
	case HELP_LABELCLEAR:
	return (gettext("\tlabelclear [-f] <vdev>\n"));
	case HELP_LIST:
	return (gettext("\tlist [-gHLpPv] [-o property[,...]] "
	"[-T d\|u] [pool] ... \n"
	"\t [interval [count]]\n"));
	case HELP_OFFLINE:
	return (gettext("\toffline [-f] [-t] <pool> <device> ...\n"));
	case HELP_ONLINE:
	return (gettext("\tonline [-e] <pool> <device> ...\n"));
	case HELP_REPLACE:
	return (gettext("\treplace [-fsw] [-o property=value] "
	"<pool> <device> [new-device]\n"));
	case HELP_REMOVE:
	return (gettext("\tremove [-npsw] <pool> <device> ...\n"));
	case HELP_REOPEN:
	return (gettext("\treopen [-n] <pool>\n"));
	case HELP_INITIALIZE:
	return (gettext("\tinitialize [-c \| -s] [-w] <pool> "
	"[<device> ...]\n"));
	case HELP_SCRUB:
	return (gettext("\tscrub [-s \| -p] [-w] <pool> ...\n"));
	case HELP_RESILVER:
	return (gettext("\tresilver <pool> ...\n"));
	case HELP_TRIM:
	return (gettext("\ttrim [-dw] [-r <rate>] [-c \| -s] <pool> "
	"[<device> ...]\n"));
	case HELP_STATUS:
	return (gettext("\tstatus [-c [script1,script2,...]] "
	"[-igLpPstvxD] [-T d\|u] [pool] ... \n"
	"\t [interval [count]]\n"));
	case HELP_UPGRADE:
	return (gettext("\tupgrade\n"
	"\tupgrade -v\n"
	"\tupgrade [-V version] <-a \| pool ...>\n"));
	case HELP_EVENTS:
	return (gettext("\tevents [-vHf [pool] \| -c]\n"));
	case HELP_GET:
	return (gettext("\tget [-Hp] [-o \"all\" \| field[,...]] "
	"<\"all\" \| property[,...]> <pool> ...\n"));
	case HELP_SET:
	return (gettext("\tset <property=value> <pool> \n"));
	case HELP_SPLIT:
	return (gettext("\tsplit [-gLnPl] [-R altroot] [-o mntopts]\n"
	"\t [-o property=value] <pool> <newpool> "
	"[<device> ...]\n"));
	case HELP_REGUID:
	return (gettext("\treguid <pool>\n"));
	case HELP_SYNC:
	return (gettext("\tsync [pool] ...\n"));
	case HELP_VERSION:
	return (gettext("\tversion\n"));
	case HELP_WAIT:
	return (gettext("\twait [-Hp] [-T d\|u] [-t <activity>[,...]] "
	"<pool> [interval]\n"));
	default:
	__builtin_unreachable();
	}
	}

	static void
	zpool_collect_leaves(zpool_handle_t zhp, nvlist_t nvroot, nvlist_t *res)
	{
	uint_t children = 0;
	nvlist_t **child;
	uint_t i;

	(void) nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	&child, &children);

	if (children == 0) {
	char *path = zpool_vdev_name(g_zfs, zhp, nvroot,
	VDEV_NAME_PATH);

	if (strcmp(path, VDEV_TYPE_INDIRECT) != 0 &&
	strcmp(path, VDEV_TYPE_HOLE) != 0)
	fnvlist_add_boolean(res, path);

	free(path);
	return;
	}

	for (i = 0; i < children; i++) {
	zpool_collect_leaves(zhp, child[i], res);
	}
	}

	/*
	* Callback routine that will print out a pool property value.
	*/
	static int
	-print_prop_cb(int prop, void *cb)
	+print_pool_prop_cb(int prop, void *cb)
	{
	FILE *fp = cb;

	(void) fprintf(fp, "\t%-19s ", zpool_prop_to_name(prop));

	if (zpool_prop_readonly(prop))
	(void) fprintf(fp, " NO ");
	else
	(void) fprintf(fp, " YES ");

	if (zpool_prop_values(prop) == NULL)
	(void) fprintf(fp, "-\n");
	else
	(void) fprintf(fp, "%s\n", zpool_prop_values(prop));

	return (ZPROP_CONT);
	}

	+/*
	+ * Callback routine that will print out a vdev property value.
	+ */
	+static int
	+print_vdev_prop_cb(int prop, void *cb)
	+{
	+ FILE *fp = cb;
	+
	+ (void) fprintf(fp, "\t%-19s ", vdev_prop_to_name(prop));
	+
	+ if (vdev_prop_readonly(prop))
	+ (void) fprintf(fp, " NO ");
	+ else
	+ (void) fprintf(fp, " YES ");
	+
	+ if (vdev_prop_values(prop) == NULL)
	+ (void) fprintf(fp, "-\n");
	+ else
	+ (void) fprintf(fp, "%s\n", vdev_prop_values(prop));
	+
	+ return (ZPROP_CONT);
	+}
	+
	/*
	* Display usage message. If we're inside a command, display only the usage for
	* that command. Otherwise, iterate over the entire command table and display
	* a complete usage message.
	*/
	static void
	usage(boolean_t requested)
	{
	FILE *fp = requested ? stdout : stderr;

	if (current_command == NULL) {
	int i;

	(void) fprintf(fp, gettext("usage: zpool command args ...\n"));
	(void) fprintf(fp,
	gettext("where 'command' is one of the following:\n\n"));

	for (i = 0; i < NCOMMAND; i++) {
	if (command_table[i].name == NULL)
	(void) fprintf(fp, "\n");
	else
	(void) fprintf(fp, "%s",
	get_usage(command_table[i].usage));
	}
	} else {
	(void) fprintf(fp, gettext("usage:\n"));
	(void) fprintf(fp, "%s", get_usage(current_command->usage));
	}

	if (current_command != NULL &&
	+ current_prop_type != (ZFS_TYPE_POOL \| ZFS_TYPE_VDEV) &&
	((strcmp(current_command->name, "set") == 0) \|\|
	(strcmp(current_command->name, "get") == 0) \|\|
	(strcmp(current_command->name, "list") == 0))) {

	(void) fprintf(fp,
	gettext("\nthe following properties are supported:\n"));

	(void) fprintf(fp, "\n\t%-19s %s %s\n\n",
	"PROPERTY", "EDIT", "VALUES");

	/* Iterate over all properties */
	- (void) zprop_iter(print_prop_cb, fp, B_FALSE, B_TRUE,
	- ZFS_TYPE_POOL);
	+ if (current_prop_type == ZFS_TYPE_POOL) {
	+ (void) zprop_iter(print_pool_prop_cb, fp, B_FALSE,
	+ B_TRUE, current_prop_type);

	- (void) fprintf(fp, "\t%-19s ", "feature@...");
	- (void) fprintf(fp, "YES disabled \| enabled \| active\n");
	+ (void) fprintf(fp, "\t%-19s ", "feature@...");
	+ (void) fprintf(fp, "YES "
	+ "disabled \| enabled \| active\n");

	- (void) fprintf(fp, gettext("\nThe feature@ properties must be "
	- "appended with a feature name.\nSee zpool-features(7).\n"));
	+ (void) fprintf(fp, gettext("\nThe feature@ properties "
	+ "must be appended with a feature name.\n"
	+ "See zpool-features(7).\n"));
	+ } else if (current_prop_type == ZFS_TYPE_VDEV) {
	+ (void) zprop_iter(print_vdev_prop_cb, fp, B_FALSE,
	+ B_TRUE, current_prop_type);
	+ }
	}

	/*
	* See comments at end of main().
	*/
	if (getenv("ZFS_ABORT") != NULL) {
	(void) printf("dumping core by request\n");
	abort();
	}

	exit(requested ? 0 : 2);
	}

	/*
	* zpool initialize [-c \| -s] [-w] <pool> [<vdev> ...]
	* Initialize all unused blocks in the specified vdevs, or all vdevs in the pool
	* if none specified.
	*
	* -c Cancel. Ends active initializing.
	* -s Suspend. Initializing can then be restarted with no flags.
	* -w Wait. Blocks until initializing has completed.
	*/
	int
	zpool_do_initialize(int argc, char **argv)
	{
	int c;
	char *poolname;
	zpool_handle_t *zhp;
	nvlist_t *vdevs;
	int err = 0;
	boolean_t wait = B_FALSE;

	struct option long_options[] = {
	{"cancel", no_argument, NULL, 'c'},
	{"suspend", no_argument, NULL, 's'},
	{"wait", no_argument, NULL, 'w'},
	{0, 0, 0, 0}
	};

	pool_initialize_func_t cmd_type = POOL_INITIALIZE_START;
	while ((c = getopt_long(argc, argv, "csw", long_options, NULL)) != -1) {
	switch (c) {
	case 'c':
	if (cmd_type != POOL_INITIALIZE_START &&
	cmd_type != POOL_INITIALIZE_CANCEL) {
	(void) fprintf(stderr, gettext("-c cannot be "
	"combined with other options\n"));
	usage(B_FALSE);
	}
	cmd_type = POOL_INITIALIZE_CANCEL;
	break;
	case 's':
	if (cmd_type != POOL_INITIALIZE_START &&
	cmd_type != POOL_INITIALIZE_SUSPEND) {
	(void) fprintf(stderr, gettext("-s cannot be "
	"combined with other options\n"));
	usage(B_FALSE);
	}
	cmd_type = POOL_INITIALIZE_SUSPEND;
	break;
	case 'w':
	wait = B_TRUE;
	break;
	case '?':
	if (optopt != 0) {
	(void) fprintf(stderr,
	gettext("invalid option '%c'\n"), optopt);
	} else {
	(void) fprintf(stderr,
	gettext("invalid option '%s'\n"),
	argv[optind - 1]);
	}
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name argument\n"));
	usage(B_FALSE);
	return (-1);
	}

	if (wait && (cmd_type != POOL_INITIALIZE_START)) {
	(void) fprintf(stderr, gettext("-w cannot be used with -c or "
	"-s\n"));
	usage(B_FALSE);
	}

	poolname = argv[0];
	zhp = zpool_open(g_zfs, poolname);
	if (zhp == NULL)
	return (-1);

	vdevs = fnvlist_alloc();
	if (argc == 1) {
	/* no individual leaf vdevs specified, so add them all */
	nvlist_t *config = zpool_get_config(zhp, NULL);
	nvlist_t *nvroot = fnvlist_lookup_nvlist(config,
	ZPOOL_CONFIG_VDEV_TREE);
	zpool_collect_leaves(zhp, nvroot, vdevs);
	} else {
	for (int i = 1; i < argc; i++) {
	fnvlist_add_boolean(vdevs, argv[i]);
	}
	}

	if (wait)
	err = zpool_initialize_wait(zhp, cmd_type, vdevs);
	else
	err = zpool_initialize(zhp, cmd_type, vdevs);

	fnvlist_free(vdevs);
	zpool_close(zhp);

	return (err);
	}

	/*
	* print a pool vdev config for dry runs
	*/
	static void
	print_vdev_tree(zpool_handle_t zhp, const char name, nvlist_t *nv, int indent,
	const char *match, int name_flags)
	{
	nvlist_t **child;
	uint_t c, children;
	char *vname;
	boolean_t printed = B_FALSE;

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0) {
	if (name != NULL)
	(void) printf("\t%*s%s\n", indent, "", name);
	return;
	}

	for (c = 0; c < children; c++) {
	uint64_t is_log = B_FALSE, is_hole = B_FALSE;
	char *class = "";

	(void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_HOLE,
	&is_hole);

	if (is_hole == B_TRUE) {
	continue;
	}

	(void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG,
	&is_log);
	if (is_log)
	class = VDEV_ALLOC_BIAS_LOG;
	(void) nvlist_lookup_string(child[c],
	ZPOOL_CONFIG_ALLOCATION_BIAS, &class);
	if (strcmp(match, class) != 0)
	continue;

	if (!printed && name != NULL) {
	(void) printf("\t%*s%s\n", indent, "", name);
	printed = B_TRUE;
	}
	vname = zpool_vdev_name(g_zfs, zhp, child[c], name_flags);
	print_vdev_tree(zhp, vname, child[c], indent + 2, "",
	name_flags);
	free(vname);
	}
	}

	/*
	* Print the list of l2cache devices for dry runs.
	*/
	static void
	print_cache_list(nvlist_t *nv, int indent)
	{
	nvlist_t **child;
	uint_t c, children;

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE,
	&child, &children) == 0 && children > 0) {
	(void) printf("\t%*s%s\n", indent, "", "cache");
	} else {
	return;
	}
	for (c = 0; c < children; c++) {
	char *vname;

	vname = zpool_vdev_name(g_zfs, NULL, child[c], 0);
	(void) printf("\t%*s%s\n", indent + 2, "", vname);
	free(vname);
	}
	}

	/*
	* Print the list of spares for dry runs.
	*/
	static void
	print_spare_list(nvlist_t *nv, int indent)
	{
	nvlist_t **child;
	uint_t c, children;

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES,
	&child, &children) == 0 && children > 0) {
	(void) printf("\t%*s%s\n", indent, "", "spares");
	} else {
	return;
	}
	for (c = 0; c < children; c++) {
	char *vname;

	vname = zpool_vdev_name(g_zfs, NULL, child[c], 0);
	(void) printf("\t%*s%s\n", indent + 2, "", vname);
	free(vname);
	}
	}

	static boolean_t
	prop_list_contains_feature(nvlist_t *proplist)
	{
	nvpair_t *nvp;
	for (nvp = nvlist_next_nvpair(proplist, NULL); NULL != nvp;
	nvp = nvlist_next_nvpair(proplist, nvp)) {
	if (zpool_prop_feature(nvpair_name(nvp)))
	return (B_TRUE);
	}
	return (B_FALSE);
	}

	/*
	* Add a property pair (name, string-value) into a property nvlist.
	*/
	static int
	add_prop_list(const char propname, char propval, nvlist_t **props,
	boolean_t poolprop)
	{
	zpool_prop_t prop = ZPOOL_PROP_INVAL;
	nvlist_t *proplist;
	const char *normnm;
	char *strval;

	if (*props == NULL &&
	nvlist_alloc(props, NV_UNIQUE_NAME, 0) != 0) {
	(void) fprintf(stderr,
	gettext("internal error: out of memory\n"));
	return (1);
	}

	proplist = *props;

	if (poolprop) {
	const char *vname = zpool_prop_to_name(ZPOOL_PROP_VERSION);
	const char *cname =
	zpool_prop_to_name(ZPOOL_PROP_COMPATIBILITY);

	if ((prop = zpool_name_to_prop(propname)) == ZPOOL_PROP_INVAL &&
	- !zpool_prop_feature(propname)) {
	+ (!zpool_prop_feature(propname) &&
	+ !zpool_prop_vdev(propname))) {
	(void) fprintf(stderr, gettext("property '%s' is "
	- "not a valid pool property\n"), propname);
	+ "not a valid pool or vdev property\n"), propname);
	return (2);
	}

	/*
	* feature@ properties and version should not be specified
	* at the same time.
	*/
	if ((prop == ZPOOL_PROP_INVAL && zpool_prop_feature(propname) &&
	nvlist_exists(proplist, vname)) \|\|
	(prop == ZPOOL_PROP_VERSION &&
	prop_list_contains_feature(proplist))) {
	(void) fprintf(stderr, gettext("'feature@' and "
	"'version' properties cannot be specified "
	"together\n"));
	return (2);
	}

	/*
	* if version is specified, only "legacy" compatibility
	* may be requested
	*/
	if ((prop == ZPOOL_PROP_COMPATIBILITY &&
	strcmp(propval, ZPOOL_COMPAT_LEGACY) != 0 &&
	nvlist_exists(proplist, vname)) \|\|
	(prop == ZPOOL_PROP_VERSION &&
	nvlist_exists(proplist, cname) &&
	strcmp(fnvlist_lookup_string(proplist, cname),
	ZPOOL_COMPAT_LEGACY) != 0)) {
	(void) fprintf(stderr, gettext("when 'version' is "
	"specified, the 'compatibility' feature may only "
	"be set to '" ZPOOL_COMPAT_LEGACY "'\n"));
	return (2);
	}

	- if (zpool_prop_feature(propname))
	+ if (zpool_prop_feature(propname) \|\| zpool_prop_vdev(propname))
	normnm = propname;
	else
	normnm = zpool_prop_to_name(prop);
	} else {
	zfs_prop_t fsprop = zfs_name_to_prop(propname);

	if (zfs_prop_valid_for_type(fsprop, ZFS_TYPE_FILESYSTEM,
	B_FALSE)) {
	normnm = zfs_prop_to_name(fsprop);
	} else if (zfs_prop_user(propname) \|\|
	zfs_prop_userquota(propname)) {
	normnm = propname;
	} else {
	(void) fprintf(stderr, gettext("property '%s' is "
	"not a valid filesystem property\n"), propname);
	return (2);
	}
	}

	if (nvlist_lookup_string(proplist, normnm, &strval) == 0 &&
	prop != ZPOOL_PROP_CACHEFILE) {
	(void) fprintf(stderr, gettext("property '%s' "
	"specified multiple times\n"), propname);
	return (2);
	}

	if (nvlist_add_string(proplist, normnm, propval) != 0) {
	(void) fprintf(stderr, gettext("internal "
	"error: out of memory\n"));
	return (1);
	}

	return (0);
	}

	/*
	* Set a default property pair (name, string-value) in a property nvlist
	*/
	static int
	add_prop_list_default(const char propname, char propval, nvlist_t **props,
	boolean_t poolprop)
	{
	char *pval;

	if (nvlist_lookup_string(*props, propname, &pval) == 0)
	return (0);

	return (add_prop_list(propname, propval, props, B_TRUE));
	}

	/*
	* zpool add [-fgLnP] [-o property=value] <pool> <vdev> ...
	*
	* -f Force addition of devices, even if they appear in use
	* -g Display guid for individual vdev name.
	* -L Follow links when resolving vdev path name.
	* -n Do not add the devices, but display the resulting layout if
	* they were to be added.
	* -o Set property=value.
	* -P Display full path for vdev name.
	*
	* Adds the given vdevs to 'pool'. As with create, the bulk of this work is
	* handled by make_root_vdev(), which constructs the nvlist needed to pass to
	* libzfs.
	*/
	int
	zpool_do_add(int argc, char **argv)
	{
	boolean_t force = B_FALSE;
	boolean_t dryrun = B_FALSE;
	int name_flags = 0;
	int c;
	nvlist_t *nvroot;
	char *poolname;
	int ret;
	zpool_handle_t *zhp;
	nvlist_t *config;
	nvlist_t *props = NULL;
	char *propval;

	/* check options */
	while ((c = getopt(argc, argv, "fgLno:P")) != -1) {
	switch (c) {
	case 'f':
	force = B_TRUE;
	break;
	case 'g':
	name_flags \|= VDEV_NAME_GUID;
	break;
	case 'L':
	name_flags \|= VDEV_NAME_FOLLOW_LINKS;
	break;
	case 'n':
	dryrun = B_TRUE;
	break;
	case 'o':
	if ((propval = strchr(optarg, '=')) == NULL) {
	(void) fprintf(stderr, gettext("missing "
	"'=' for -o option\n"));
	usage(B_FALSE);
	}
	*propval = '\0';
	propval++;

	if ((strcmp(optarg, ZPOOL_CONFIG_ASHIFT) != 0) \|\|
	(add_prop_list(optarg, propval, &props, B_TRUE)))
	usage(B_FALSE);
	break;
	case 'P':
	name_flags \|= VDEV_NAME_PATH;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* get pool name and check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name argument\n"));
	usage(B_FALSE);
	}
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing vdev specification\n"));
	usage(B_FALSE);
	}

	poolname = argv[0];

	argc--;
	argv++;

	if ((zhp = zpool_open(g_zfs, poolname)) == NULL)
	return (1);

	if ((config = zpool_get_config(zhp, NULL)) == NULL) {
	(void) fprintf(stderr, gettext("pool '%s' is unavailable\n"),
	poolname);
	zpool_close(zhp);
	return (1);
	}

	/* unless manually specified use "ashift" pool property (if set) */
	if (!nvlist_exists(props, ZPOOL_CONFIG_ASHIFT)) {
	int intval;
	zprop_source_t src;
	char strval[ZPOOL_MAXPROPLEN];

	intval = zpool_get_prop_int(zhp, ZPOOL_PROP_ASHIFT, &src);
	if (src != ZPROP_SRC_DEFAULT) {
	(void) sprintf(strval, "%" PRId32, intval);
	verify(add_prop_list(ZPOOL_CONFIG_ASHIFT, strval,
	&props, B_TRUE) == 0);
	}
	}

	/* pass off to make_root_vdev for processing */
	nvroot = make_root_vdev(zhp, props, force, !force, B_FALSE, dryrun,
	argc, argv);
	if (nvroot == NULL) {
	zpool_close(zhp);
	return (1);
	}

	if (dryrun) {
	nvlist_t *poolnvroot;
	nvlist_t l2child, sparechild;
	uint_t l2children, sparechildren, c;
	char *vname;
	boolean_t hadcache = B_FALSE, hadspare = B_FALSE;

	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&poolnvroot) == 0);

	(void) printf(gettext("would update '%s' to the following "
	"configuration:\n\n"), zpool_get_name(zhp));

	/* print original main pool and new tree */
	print_vdev_tree(zhp, poolname, poolnvroot, 0, "",
	name_flags \| VDEV_NAME_TYPE_ID);
	print_vdev_tree(zhp, NULL, nvroot, 0, "", name_flags);

	/* print other classes: 'dedup', 'special', and 'log' */
	if (zfs_special_devs(poolnvroot, VDEV_ALLOC_BIAS_DEDUP)) {
	print_vdev_tree(zhp, "dedup", poolnvroot, 0,
	VDEV_ALLOC_BIAS_DEDUP, name_flags);
	print_vdev_tree(zhp, NULL, nvroot, 0,
	VDEV_ALLOC_BIAS_DEDUP, name_flags);
	} else if (zfs_special_devs(nvroot, VDEV_ALLOC_BIAS_DEDUP)) {
	print_vdev_tree(zhp, "dedup", nvroot, 0,
	VDEV_ALLOC_BIAS_DEDUP, name_flags);
	}

	if (zfs_special_devs(poolnvroot, VDEV_ALLOC_BIAS_SPECIAL)) {
	print_vdev_tree(zhp, "special", poolnvroot, 0,
	VDEV_ALLOC_BIAS_SPECIAL, name_flags);
	print_vdev_tree(zhp, NULL, nvroot, 0,
	VDEV_ALLOC_BIAS_SPECIAL, name_flags);
	} else if (zfs_special_devs(nvroot, VDEV_ALLOC_BIAS_SPECIAL)) {
	print_vdev_tree(zhp, "special", nvroot, 0,
	VDEV_ALLOC_BIAS_SPECIAL, name_flags);
	}

	if (num_logs(poolnvroot) > 0) {
	print_vdev_tree(zhp, "logs", poolnvroot, 0,
	VDEV_ALLOC_BIAS_LOG, name_flags);
	print_vdev_tree(zhp, NULL, nvroot, 0,
	VDEV_ALLOC_BIAS_LOG, name_flags);
	} else if (num_logs(nvroot) > 0) {
	print_vdev_tree(zhp, "logs", nvroot, 0,
	VDEV_ALLOC_BIAS_LOG, name_flags);
	}

	/* Do the same for the caches */
	if (nvlist_lookup_nvlist_array(poolnvroot, ZPOOL_CONFIG_L2CACHE,
	&l2child, &l2children) == 0 && l2children) {
	hadcache = B_TRUE;
	(void) printf(gettext("\tcache\n"));
	for (c = 0; c < l2children; c++) {
	vname = zpool_vdev_name(g_zfs, NULL,
	l2child[c], name_flags);
	(void) printf("\t %s\n", vname);
	free(vname);
	}
	}
	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
	&l2child, &l2children) == 0 && l2children) {
	if (!hadcache)
	(void) printf(gettext("\tcache\n"));
	for (c = 0; c < l2children; c++) {
	vname = zpool_vdev_name(g_zfs, NULL,
	l2child[c], name_flags);
	(void) printf("\t %s\n", vname);
	free(vname);
	}
	}
	/* And finally the spares */
	if (nvlist_lookup_nvlist_array(poolnvroot, ZPOOL_CONFIG_SPARES,
	&sparechild, &sparechildren) == 0 && sparechildren > 0) {
	hadspare = B_TRUE;
	(void) printf(gettext("\tspares\n"));
	for (c = 0; c < sparechildren; c++) {
	vname = zpool_vdev_name(g_zfs, NULL,
	sparechild[c], name_flags);
	(void) printf("\t %s\n", vname);
	free(vname);
	}
	}
	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	&sparechild, &sparechildren) == 0 && sparechildren > 0) {
	if (!hadspare)
	(void) printf(gettext("\tspares\n"));
	for (c = 0; c < sparechildren; c++) {
	vname = zpool_vdev_name(g_zfs, NULL,
	sparechild[c], name_flags);
	(void) printf("\t %s\n", vname);
	free(vname);
	}
	}

	ret = 0;
	} else {
	ret = (zpool_add(zhp, nvroot) != 0);
	}

	nvlist_free(props);
	nvlist_free(nvroot);
	zpool_close(zhp);

	return (ret);
	}

	/*
	* zpool remove [-npsw] <pool> <vdev> ...
	*
	* Removes the given vdev from the pool.
	*/
	int
	zpool_do_remove(int argc, char **argv)
	{
	char *poolname;
	int i, ret = 0;
	zpool_handle_t *zhp = NULL;
	boolean_t stop = B_FALSE;
	int c;
	boolean_t noop = B_FALSE;
	boolean_t parsable = B_FALSE;
	boolean_t wait = B_FALSE;

	/* check options */
	while ((c = getopt(argc, argv, "npsw")) != -1) {
	switch (c) {
	case 'n':
	noop = B_TRUE;
	break;
	case 'p':
	parsable = B_TRUE;
	break;
	case 's':
	stop = B_TRUE;
	break;
	case 'w':
	wait = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* get pool name and check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name argument\n"));
	usage(B_FALSE);
	}

	poolname = argv[0];

	if ((zhp = zpool_open(g_zfs, poolname)) == NULL)
	return (1);

	if (stop && noop) {
	(void) fprintf(stderr, gettext("stop request ignored\n"));
	return (0);
	}

	if (stop) {
	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}
	if (zpool_vdev_remove_cancel(zhp) != 0)
	ret = 1;
	if (wait) {
	(void) fprintf(stderr, gettext("invalid option "
	"combination: -w cannot be used with -s\n"));
	usage(B_FALSE);
	}
	} else {
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing device\n"));
	usage(B_FALSE);
	}

	for (i = 1; i < argc; i++) {
	if (noop) {
	uint64_t size;

	if (zpool_vdev_indirect_size(zhp, argv[i],
	&size) != 0) {
	ret = 1;
	break;
	}
	if (parsable) {
	(void) printf("%s %llu\n",
	argv[i], (unsigned long long)size);
	} else {
	char valstr[32];
	zfs_nicenum(size, valstr,
	sizeof (valstr));
	(void) printf("Memory that will be "
	"used after removing %s: %s\n",
	argv[i], valstr);
	}
	} else {
	if (zpool_vdev_remove(zhp, argv[i]) != 0)
	ret = 1;
	}
	}

	if (ret == 0 && wait)
	ret = zpool_wait(zhp, ZPOOL_WAIT_REMOVE);
	}
	zpool_close(zhp);

	return (ret);
	}

	/*
	* Return 1 if a vdev is active (being used in a pool)
	* Return 0 if a vdev is inactive (offlined or faulted, or not in active pool)
	*
	* This is useful for checking if a disk in an active pool is offlined or
	* faulted.
	*/
	static int
	vdev_is_active(char *vdev_path)
	{
	int fd;
	fd = open(vdev_path, O_EXCL);
	if (fd < 0) {
	return (1); /* cant open O_EXCL - disk is active */
	}

	close(fd);
	return (0); /* disk is inactive in the pool */
	}

	/*
	* zpool labelclear [-f] <vdev>
	*
	* -f Force clearing the label for the vdevs which are members of
	* the exported or foreign pools.
	*
	* Verifies that the vdev is not active and zeros out the label information
	* on the device.
	*/
	int
	zpool_do_labelclear(int argc, char **argv)
	{
	char vdev[MAXPATHLEN];
	char *name = NULL;
	struct stat st;
	int c, fd = -1, ret = 0;
	nvlist_t *config;
	pool_state_t state;
	boolean_t inuse = B_FALSE;
	boolean_t force = B_FALSE;

	/* check options */
	while ((c = getopt(argc, argv, "f")) != -1) {
	switch (c) {
	case 'f':
	force = B_TRUE;
	break;
	default:
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* get vdev name */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing vdev name\n"));
	usage(B_FALSE);
	}
	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	/*
	* Check if we were given absolute path and use it as is.
	* Otherwise if the provided vdev name doesn't point to a file,
	* try prepending expected disk paths and partition numbers.
	*/
	(void) strlcpy(vdev, argv[0], sizeof (vdev));
	if (vdev[0] != '/' && stat(vdev, &st) != 0) {
	int error;

	error = zfs_resolve_shortname(argv[0], vdev, MAXPATHLEN);
	if (error == 0 && zfs_dev_is_whole_disk(vdev)) {
	if (zfs_append_partition(vdev, MAXPATHLEN) == -1)
	error = ENOENT;
	}

	if (error \|\| (stat(vdev, &st) != 0)) {
	(void) fprintf(stderr, gettext(
	"failed to find device %s, try specifying absolute "
	"path instead\n"), argv[0]);
	return (1);
	}
	}

	if ((fd = open(vdev, O_RDWR)) < 0) {
	(void) fprintf(stderr, gettext("failed to open %s: %s\n"),
	vdev, strerror(errno));
	return (1);
	}

	/*
	* Flush all dirty pages for the block device. This should not be
	* fatal when the device does not support BLKFLSBUF as would be the
	* case for a file vdev.
	*/
	if ((zfs_dev_flush(fd) != 0) && (errno != ENOTTY))
	(void) fprintf(stderr, gettext("failed to invalidate "
	"cache for %s: %s\n"), vdev, strerror(errno));

	if (zpool_read_label(fd, &config, NULL) != 0) {
	(void) fprintf(stderr,
	gettext("failed to read label from %s\n"), vdev);
	ret = 1;
	goto errout;
	}
	nvlist_free(config);

	ret = zpool_in_use(g_zfs, fd, &state, &name, &inuse);
	if (ret != 0) {
	(void) fprintf(stderr,
	gettext("failed to check state for %s\n"), vdev);
	ret = 1;
	goto errout;
	}

	if (!inuse)
	goto wipe_label;

	switch (state) {
	default:
	case POOL_STATE_ACTIVE:
	case POOL_STATE_SPARE:
	case POOL_STATE_L2CACHE:
	/*
	* We allow the user to call 'zpool offline -f'
	* on an offlined disk in an active pool. We can check if
	* the disk is online by calling vdev_is_active().
	*/
	if (force && !vdev_is_active(vdev))
	break;

	(void) fprintf(stderr, gettext(
	"%s is a member (%s) of pool \"%s\""),
	vdev, zpool_pool_state_to_name(state), name);

	if (force) {
	(void) fprintf(stderr, gettext(
	". Offline the disk first to clear its label."));
	}
	printf("\n");
	ret = 1;
	goto errout;

	case POOL_STATE_EXPORTED:
	if (force)
	break;
	(void) fprintf(stderr, gettext(
	"use '-f' to override the following error:\n"
	"%s is a member of exported pool \"%s\"\n"),
	vdev, name);
	ret = 1;
	goto errout;

	case POOL_STATE_POTENTIALLY_ACTIVE:
	if (force)
	break;
	(void) fprintf(stderr, gettext(
	"use '-f' to override the following error:\n"
	"%s is a member of potentially active pool \"%s\"\n"),
	vdev, name);
	ret = 1;
	goto errout;

	case POOL_STATE_DESTROYED:
	/* inuse should never be set for a destroyed pool */
	assert(0);
	break;
	}

	wipe_label:
	ret = zpool_clear_label(fd);
	if (ret != 0) {
	(void) fprintf(stderr,
	gettext("failed to clear label for %s\n"), vdev);
	}

	errout:
	free(name);
	(void) close(fd);

	return (ret);
	}

	/*
	* zpool create [-fnd] [-o property=value] ...
	* [-O file-system-property=value] ...
	* [-R root] [-m mountpoint] <pool> <dev> ...
	*
	* -f Force creation, even if devices appear in use
	* -n Do not create the pool, but display the resulting layout if it
	* were to be created.
	* -R Create a pool under an alternate root
	* -m Set default mountpoint for the root dataset. By default it's
	* '/<pool>'
	* -o Set property=value.
	* -o Set feature@feature=enabled\|disabled.
	* -d Don't automatically enable all supported pool features
	* (individual features can be enabled with -o).
	* -O Set fsproperty=value in the pool's root file system
	*
	* Creates the named pool according to the given vdev specification. The
	* bulk of the vdev processing is done in make_root_vdev() in zpool_vdev.c.
	* Once we get the nvlist back from make_root_vdev(), we either print out the
	* contents (if '-n' was specified), or pass it to libzfs to do the creation.
	*/
	int
	zpool_do_create(int argc, char **argv)
	{
	boolean_t force = B_FALSE;
	boolean_t dryrun = B_FALSE;
	boolean_t enable_pool_features = B_TRUE;

	int c;
	nvlist_t *nvroot = NULL;
	char *poolname;
	char *tname = NULL;
	int ret = 1;
	char *altroot = NULL;
	char *compat = NULL;
	char *mountpoint = NULL;
	nvlist_t *fsprops = NULL;
	nvlist_t *props = NULL;
	char *propval;

	/* check options */
	while ((c = getopt(argc, argv, ":fndR:m:o:O:t:")) != -1) {
	switch (c) {
	case 'f':
	force = B_TRUE;
	break;
	case 'n':
	dryrun = B_TRUE;
	break;
	case 'd':
	enable_pool_features = B_FALSE;
	break;
	case 'R':
	altroot = optarg;
	if (add_prop_list(zpool_prop_to_name(
	ZPOOL_PROP_ALTROOT), optarg, &props, B_TRUE))
	goto errout;
	if (add_prop_list_default(zpool_prop_to_name(
	ZPOOL_PROP_CACHEFILE), "none", &props, B_TRUE))
	goto errout;
	break;
	case 'm':
	/* Equivalent to -O mountpoint=optarg */
	mountpoint = optarg;
	break;
	case 'o':
	if ((propval = strchr(optarg, '=')) == NULL) {
	(void) fprintf(stderr, gettext("missing "
	"'=' for -o option\n"));
	goto errout;
	}
	*propval = '\0';
	propval++;

	if (add_prop_list(optarg, propval, &props, B_TRUE))
	goto errout;

	/*
	* If the user is creating a pool that doesn't support
	* feature flags, don't enable any features.
	*/
	if (zpool_name_to_prop(optarg) == ZPOOL_PROP_VERSION) {
	char *end;
	u_longlong_t ver;

	ver = strtoull(propval, &end, 10);
	if (*end == '\0' &&
	ver < SPA_VERSION_FEATURES) {
	enable_pool_features = B_FALSE;
	}
	}
	if (zpool_name_to_prop(optarg) == ZPOOL_PROP_ALTROOT)
	altroot = propval;
	if (zpool_name_to_prop(optarg) ==
	ZPOOL_PROP_COMPATIBILITY)
	compat = propval;
	break;
	case 'O':
	if ((propval = strchr(optarg, '=')) == NULL) {
	(void) fprintf(stderr, gettext("missing "
	"'=' for -O option\n"));
	goto errout;
	}
	*propval = '\0';
	propval++;

	/*
	* Mountpoints are checked and then added later.
	* Uniquely among properties, they can be specified
	* more than once, to avoid conflict with -m.
	*/
	if (0 == strcmp(optarg,
	zfs_prop_to_name(ZFS_PROP_MOUNTPOINT))) {
	mountpoint = propval;
	} else if (add_prop_list(optarg, propval, &fsprops,
	B_FALSE)) {
	goto errout;
	}
	break;
	case 't':
	/*
	* Sanity check temporary pool name.
	*/
	if (strchr(optarg, '/') != NULL) {
	(void) fprintf(stderr, gettext("cannot create "
	"'%s': invalid character '/' in temporary "
	"name\n"), optarg);
	(void) fprintf(stderr, gettext("use 'zfs "
	"create' to create a dataset\n"));
	goto errout;
	}

	if (add_prop_list(zpool_prop_to_name(
	ZPOOL_PROP_TNAME), optarg, &props, B_TRUE))
	goto errout;
	if (add_prop_list_default(zpool_prop_to_name(
	ZPOOL_PROP_CACHEFILE), "none", &props, B_TRUE))
	goto errout;
	tname = optarg;
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	goto badusage;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	goto badusage;
	}
	}

	argc -= optind;
	argv += optind;

	/* get pool name and check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name argument\n"));
	goto badusage;
	}
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing vdev specification\n"));
	goto badusage;
	}

	poolname = argv[0];

	/*
	* As a special case, check for use of '/' in the name, and direct the
	* user to use 'zfs create' instead.
	*/
	if (strchr(poolname, '/') != NULL) {
	(void) fprintf(stderr, gettext("cannot create '%s': invalid "
	"character '/' in pool name\n"), poolname);
	(void) fprintf(stderr, gettext("use 'zfs create' to "
	"create a dataset\n"));
	goto errout;
	}

	/* pass off to make_root_vdev for bulk processing */
	nvroot = make_root_vdev(NULL, props, force, !force, B_FALSE, dryrun,
	argc - 1, argv + 1);
	if (nvroot == NULL)
	goto errout;

	/* make_root_vdev() allows 0 toplevel children if there are spares */
	if (!zfs_allocatable_devs(nvroot)) {
	(void) fprintf(stderr, gettext("invalid vdev "
	"specification: at least one toplevel vdev must be "
	"specified\n"));
	goto errout;
	}

	if (altroot != NULL && altroot[0] != '/') {
	(void) fprintf(stderr, gettext("invalid alternate root '%s': "
	"must be an absolute path\n"), altroot);
	goto errout;
	}

	/*
	* Check the validity of the mountpoint and direct the user to use the
	* '-m' mountpoint option if it looks like its in use.
	*/
	if (mountpoint == NULL \|\|
	(strcmp(mountpoint, ZFS_MOUNTPOINT_LEGACY) != 0 &&
	strcmp(mountpoint, ZFS_MOUNTPOINT_NONE) != 0)) {
	char buf[MAXPATHLEN];
	DIR *dirp;

	if (mountpoint && mountpoint[0] != '/') {
	(void) fprintf(stderr, gettext("invalid mountpoint "
	"'%s': must be an absolute path, 'legacy', or "
	"'none'\n"), mountpoint);
	goto errout;
	}

	if (mountpoint == NULL) {
	if (altroot != NULL)
	(void) snprintf(buf, sizeof (buf), "%s/%s",
	altroot, poolname);
	else
	(void) snprintf(buf, sizeof (buf), "/%s",
	poolname);
	} else {
	if (altroot != NULL)
	(void) snprintf(buf, sizeof (buf), "%s%s",
	altroot, mountpoint);
	else
	(void) snprintf(buf, sizeof (buf), "%s",
	mountpoint);
	}

	if ((dirp = opendir(buf)) == NULL && errno != ENOENT) {
	(void) fprintf(stderr, gettext("mountpoint '%s' : "
	"%s\n"), buf, strerror(errno));
	(void) fprintf(stderr, gettext("use '-m' "
	"option to provide a different default\n"));
	goto errout;
	} else if (dirp) {
	int count = 0;

	while (count < 3 && readdir(dirp) != NULL)
	count++;
	(void) closedir(dirp);

	if (count > 2) {
	(void) fprintf(stderr, gettext("mountpoint "
	"'%s' exists and is not empty\n"), buf);
	(void) fprintf(stderr, gettext("use '-m' "
	"option to provide a "
	"different default\n"));
	goto errout;
	}
	}
	}

	/*
	* Now that the mountpoint's validity has been checked, ensure that
	* the property is set appropriately prior to creating the pool.
	*/
	if (mountpoint != NULL) {
	ret = add_prop_list(zfs_prop_to_name(ZFS_PROP_MOUNTPOINT),
	mountpoint, &fsprops, B_FALSE);
	if (ret != 0)
	goto errout;
	}

	ret = 1;
	if (dryrun) {
	/*
	* For a dry run invocation, print out a basic message and run
	* through all the vdevs in the list and print out in an
	* appropriate hierarchy.
	*/
	(void) printf(gettext("would create '%s' with the "
	"following layout:\n\n"), poolname);

	print_vdev_tree(NULL, poolname, nvroot, 0, "", 0);
	print_vdev_tree(NULL, "dedup", nvroot, 0,
	VDEV_ALLOC_BIAS_DEDUP, 0);
	print_vdev_tree(NULL, "special", nvroot, 0,
	VDEV_ALLOC_BIAS_SPECIAL, 0);
	print_vdev_tree(NULL, "logs", nvroot, 0,
	VDEV_ALLOC_BIAS_LOG, 0);
	print_cache_list(nvroot, 0);
	print_spare_list(nvroot, 0);

	ret = 0;
	} else {
	/*
	* Load in feature set.
	* Note: if compatibility property not given, we'll have
	* NULL, which means 'all features'.
	*/
	boolean_t requested_features[SPA_FEATURES];
	if (zpool_do_load_compat(compat, requested_features) !=
	ZPOOL_COMPATIBILITY_OK)
	goto errout;

	/*
	* props contains list of features to enable.
	* For each feature:
	* - remove it if feature@name=disabled
	* - leave it there if feature@name=enabled
	* - add it if:
	* - enable_pool_features (ie: no '-d' or '-o version')
	* - it's supported by the kernel module
	* - it's in the requested feature set
	* - warn if it's enabled but not in compat
	*/
	for (spa_feature_t i = 0; i < SPA_FEATURES; i++) {
	char propname[MAXPATHLEN];
	char *propval;
	zfeature_info_t *feat = &spa_feature_table[i];

	(void) snprintf(propname, sizeof (propname),
	"feature@%s", feat->fi_uname);

	if (!nvlist_lookup_string(props, propname, &propval)) {
	if (strcmp(propval, ZFS_FEATURE_DISABLED) == 0)
	(void) nvlist_remove_all(props,
	propname);
	if (strcmp(propval,
	ZFS_FEATURE_ENABLED) == 0 &&
	!requested_features[i])
	(void) fprintf(stderr, gettext(
	"Warning: feature \"%s\" enabled "
	"but is not in specified "
	"'compatibility' feature set.\n"),
	feat->fi_uname);
	} else if (
	enable_pool_features &&
	feat->fi_zfs_mod_supported &&
	requested_features[i]) {
	ret = add_prop_list(propname,
	ZFS_FEATURE_ENABLED, &props, B_TRUE);
	if (ret != 0)
	goto errout;
	}
	}

	ret = 1;
	if (zpool_create(g_zfs, poolname,
	nvroot, props, fsprops) == 0) {
	zfs_handle_t *pool = zfs_open(g_zfs,
	tname ? tname : poolname, ZFS_TYPE_FILESYSTEM);
	if (pool != NULL) {
	if (zfs_mount(pool, NULL, 0) == 0) {
	ret = zfs_shareall(pool);
	zfs_commit_all_shares();
	}
	zfs_close(pool);
	}
	} else if (libzfs_errno(g_zfs) == EZFS_INVALIDNAME) {
	(void) fprintf(stderr, gettext("pool name may have "
	"been omitted\n"));
	}
	}

	errout:
	nvlist_free(nvroot);
	nvlist_free(fsprops);
	nvlist_free(props);
	return (ret);
	badusage:
	nvlist_free(fsprops);
	nvlist_free(props);
	usage(B_FALSE);
	return (2);
	}

	/*
	* zpool destroy <pool>
	*
	* -f Forcefully unmount any datasets
	*
	* Destroy the given pool. Automatically unmounts any datasets in the pool.
	*/
	int
	zpool_do_destroy(int argc, char **argv)
	{
	boolean_t force = B_FALSE;
	int c;
	char *pool;
	zpool_handle_t *zhp;
	int ret;

	/* check options */
	while ((c = getopt(argc, argv, "f")) != -1) {
	switch (c) {
	case 'f':
	force = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* check arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool argument\n"));
	usage(B_FALSE);
	}
	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	pool = argv[0];

	if ((zhp = zpool_open_canfail(g_zfs, pool)) == NULL) {
	/*
	* As a special case, check for use of '/' in the name, and
	* direct the user to use 'zfs destroy' instead.
	*/
	if (strchr(pool, '/') != NULL)
	(void) fprintf(stderr, gettext("use 'zfs destroy' to "
	"destroy a dataset\n"));
	return (1);
	}

	if (zpool_disable_datasets(zhp, force) != 0) {
	(void) fprintf(stderr, gettext("could not destroy '%s': "
	"could not unmount datasets\n"), zpool_get_name(zhp));
	zpool_close(zhp);
	return (1);
	}

	/* The history must be logged as part of the export */
	log_history = B_FALSE;

	ret = (zpool_destroy(zhp, history_str) != 0);

	zpool_close(zhp);

	return (ret);
	}

	typedef struct export_cbdata {
	boolean_t force;
	boolean_t hardforce;
	} export_cbdata_t;

	/*
	* Export one pool
	*/
	static int
	zpool_export_one(zpool_handle_t zhp, void data)
	{
	export_cbdata_t *cb = data;

	if (zpool_disable_datasets(zhp, cb->force) != 0)
	return (1);

	/* The history must be logged as part of the export */
	log_history = B_FALSE;

	if (cb->hardforce) {
	if (zpool_export_force(zhp, history_str) != 0)
	return (1);
	} else if (zpool_export(zhp, cb->force, history_str) != 0) {
	return (1);
	}

	return (0);
	}

	/*
	* zpool export [-f] <pool> ...
	*
	* -a Export all pools
	* -f Forcefully unmount datasets
	*
	* Export the given pools. By default, the command will attempt to cleanly
	* unmount any active datasets within the pool. If the '-f' flag is specified,
	* then the datasets will be forcefully unmounted.
	*/
	int
	zpool_do_export(int argc, char **argv)
	{
	export_cbdata_t cb;
	boolean_t do_all = B_FALSE;
	boolean_t force = B_FALSE;
	boolean_t hardforce = B_FALSE;
	int c, ret;

	/* check options */
	while ((c = getopt(argc, argv, "afF")) != -1) {
	switch (c) {
	case 'a':
	do_all = B_TRUE;
	break;
	case 'f':
	force = B_TRUE;
	break;
	case 'F':
	hardforce = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	cb.force = force;
	cb.hardforce = hardforce;
	argc -= optind;
	argv += optind;

	if (do_all) {
	if (argc != 0) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	return (for_each_pool(argc, argv, B_TRUE, NULL,
	- B_FALSE, zpool_export_one, &cb));
	+ ZFS_TYPE_POOL, B_FALSE, zpool_export_one, &cb));
	}

	/* check arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool argument\n"));
	usage(B_FALSE);
	}

	- ret = for_each_pool(argc, argv, B_TRUE, NULL, B_FALSE, zpool_export_one,
	- &cb);
	+ ret = for_each_pool(argc, argv, B_TRUE, NULL, ZFS_TYPE_POOL,
	+ B_FALSE, zpool_export_one, &cb);

	return (ret);
	}

	/*
	* Given a vdev configuration, determine the maximum width needed for the device
	* name column.
	*/
	static int
	max_width(zpool_handle_t zhp, nvlist_t nv, int depth, int max,
	int name_flags)
	{
	char *name;
	nvlist_t **child;
	uint_t c, children;
	int ret;

	name = zpool_vdev_name(g_zfs, zhp, nv, name_flags);
	if (strlen(name) + depth > max)
	max = strlen(name) + depth;

	free(name);

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES,
	&child, &children) == 0) {
	for (c = 0; c < children; c++)
	if ((ret = max_width(zhp, child[c], depth + 2,
	max, name_flags)) > max)
	max = ret;
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE,
	&child, &children) == 0) {
	for (c = 0; c < children; c++)
	if ((ret = max_width(zhp, child[c], depth + 2,
	max, name_flags)) > max)
	max = ret;
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) == 0) {
	for (c = 0; c < children; c++)
	if ((ret = max_width(zhp, child[c], depth + 2,
	max, name_flags)) > max)
	max = ret;
	}

	return (max);
	}

	typedef struct spare_cbdata {
	uint64_t cb_guid;
	zpool_handle_t *cb_zhp;
	} spare_cbdata_t;

	static boolean_t
	find_vdev(nvlist_t *nv, uint64_t search)
	{
	uint64_t guid;
	nvlist_t **child;
	uint_t c, children;

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) == 0 &&
	search == guid)
	return (B_TRUE);

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) == 0) {
	for (c = 0; c < children; c++)
	if (find_vdev(child[c], search))
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	static int
	find_spare(zpool_handle_t zhp, void data)
	{
	spare_cbdata_t *cbp = data;
	nvlist_t config, nvroot;

	config = zpool_get_config(zhp, NULL);
	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) == 0);

	if (find_vdev(nvroot, cbp->cb_guid)) {
	cbp->cb_zhp = zhp;
	return (1);
	}

	zpool_close(zhp);
	return (0);
	}

	typedef struct status_cbdata {
	int cb_count;
	int cb_name_flags;
	int cb_namewidth;
	boolean_t cb_allpools;
	boolean_t cb_verbose;
	boolean_t cb_literal;
	boolean_t cb_explain;
	boolean_t cb_first;
	boolean_t cb_dedup_stats;
	boolean_t cb_print_status;
	boolean_t cb_print_slow_ios;
	boolean_t cb_print_vdev_init;
	boolean_t cb_print_vdev_trim;
	vdev_cmd_data_list_t *vcdl;
	} status_cbdata_t;

	/* Return 1 if string is NULL, empty, or whitespace; return 0 otherwise. */
	static int
	is_blank_str(char *str)
	{
	while (str != NULL && *str != '\0') {
	if (!isblank(*str))
	return (0);
	str++;
	}
	return (1);
	}

	/* Print command output lines for specific vdev in a specific pool */
	static void
	zpool_print_cmd(vdev_cmd_data_list_t vcdl, const char pool, char *path)
	{
	vdev_cmd_data_t *data;
	int i, j;
	char *val;

	for (i = 0; i < vcdl->count; i++) {
	if ((strcmp(vcdl->data[i].path, path) != 0) \|\|
	(strcmp(vcdl->data[i].pool, pool) != 0)) {
	/* Not the vdev we're looking for */
	continue;
	}

	data = &vcdl->data[i];
	/* Print out all the output values for this vdev */
	for (j = 0; j < vcdl->uniq_cols_cnt; j++) {
	val = NULL;
	/* Does this vdev have values for this column? */
	for (int k = 0; k < data->cols_cnt; k++) {
	if (strcmp(data->cols[k],
	vcdl->uniq_cols[j]) == 0) {
	/* yes it does, record the value */
	val = data->lines[k];
	break;
	}
	}
	/*
	* Mark empty values with dashes to make output
	* awk-able.
	*/
	if (val == NULL \|\| is_blank_str(val))
	val = "-";

	printf("%*s", vcdl->uniq_cols_width[j], val);
	if (j < vcdl->uniq_cols_cnt - 1)
	printf(" ");
	}

	/* Print out any values that aren't in a column at the end */
	for (j = data->cols_cnt; j < data->lines_cnt; j++) {
	/* Did we have any columns? If so print a spacer. */
	if (vcdl->uniq_cols_cnt > 0)
	printf(" ");

	val = data->lines[j];
	printf("%s", val ? val : "");
	}
	break;
	}
	}

	/*
	* Print vdev initialization status for leaves
	*/
	static void
	print_status_initialize(vdev_stat_t *vs, boolean_t verbose)
	{
	if (verbose) {
	if ((vs->vs_initialize_state == VDEV_INITIALIZE_ACTIVE \|\|
	vs->vs_initialize_state == VDEV_INITIALIZE_SUSPENDED \|\|
	vs->vs_initialize_state == VDEV_INITIALIZE_COMPLETE) &&
	!vs->vs_scan_removing) {
	char zbuf[1024];
	char tbuf[256];
	struct tm zaction_ts;

	time_t t = vs->vs_initialize_action_time;
	int initialize_pct = 100;
	if (vs->vs_initialize_state !=
	VDEV_INITIALIZE_COMPLETE) {
	initialize_pct = (vs->vs_initialize_bytes_done *
	100 / (vs->vs_initialize_bytes_est + 1));
	}

	(void) localtime_r(&t, &zaction_ts);
	(void) strftime(tbuf, sizeof (tbuf), "%c", &zaction_ts);

	switch (vs->vs_initialize_state) {
	case VDEV_INITIALIZE_SUSPENDED:
	(void) snprintf(zbuf, sizeof (zbuf), ", %s %s",
	gettext("suspended, started at"), tbuf);
	break;
	case VDEV_INITIALIZE_ACTIVE:
	(void) snprintf(zbuf, sizeof (zbuf), ", %s %s",
	gettext("started at"), tbuf);
	break;
	case VDEV_INITIALIZE_COMPLETE:
	(void) snprintf(zbuf, sizeof (zbuf), ", %s %s",
	gettext("completed at"), tbuf);
	break;
	}

	(void) printf(gettext(" (%d%% initialized%s)"),
	initialize_pct, zbuf);
	} else {
	(void) printf(gettext(" (uninitialized)"));
	}
	} else if (vs->vs_initialize_state == VDEV_INITIALIZE_ACTIVE) {
	(void) printf(gettext(" (initializing)"));
	}
	}

	/*
	* Print vdev TRIM status for leaves
	*/
	static void
	print_status_trim(vdev_stat_t *vs, boolean_t verbose)
	{
	if (verbose) {
	if ((vs->vs_trim_state == VDEV_TRIM_ACTIVE \|\|
	vs->vs_trim_state == VDEV_TRIM_SUSPENDED \|\|
	vs->vs_trim_state == VDEV_TRIM_COMPLETE) &&
	!vs->vs_scan_removing) {
	char zbuf[1024];
	char tbuf[256];
	struct tm zaction_ts;

	time_t t = vs->vs_trim_action_time;
	int trim_pct = 100;
	if (vs->vs_trim_state != VDEV_TRIM_COMPLETE) {
	trim_pct = (vs->vs_trim_bytes_done *
	100 / (vs->vs_trim_bytes_est + 1));
	}

	(void) localtime_r(&t, &zaction_ts);
	(void) strftime(tbuf, sizeof (tbuf), "%c", &zaction_ts);

	switch (vs->vs_trim_state) {
	case VDEV_TRIM_SUSPENDED:
	(void) snprintf(zbuf, sizeof (zbuf), ", %s %s",
	gettext("suspended, started at"), tbuf);
	break;
	case VDEV_TRIM_ACTIVE:
	(void) snprintf(zbuf, sizeof (zbuf), ", %s %s",
	gettext("started at"), tbuf);
	break;
	case VDEV_TRIM_COMPLETE:
	(void) snprintf(zbuf, sizeof (zbuf), ", %s %s",
	gettext("completed at"), tbuf);
	break;
	}

	(void) printf(gettext(" (%d%% trimmed%s)"),
	trim_pct, zbuf);
	} else if (vs->vs_trim_notsup) {
	(void) printf(gettext(" (trim unsupported)"));
	} else {
	(void) printf(gettext(" (untrimmed)"));
	}
	} else if (vs->vs_trim_state == VDEV_TRIM_ACTIVE) {
	(void) printf(gettext(" (trimming)"));
	}
	}

	/*
	* Return the color associated with a health string. This includes returning
	* NULL for no color change.
	*/
	static char *
	health_str_to_color(const char *health)
	{
	if (strcmp(health, gettext("FAULTED")) == 0 \|\|
	strcmp(health, gettext("SUSPENDED")) == 0 \|\|
	strcmp(health, gettext("UNAVAIL")) == 0) {
	return (ANSI_RED);
	}

	if (strcmp(health, gettext("OFFLINE")) == 0 \|\|
	strcmp(health, gettext("DEGRADED")) == 0 \|\|
	strcmp(health, gettext("REMOVED")) == 0) {
	return (ANSI_YELLOW);
	}

	return (NULL);
	}

	/*
	* Print out configuration state as requested by status_callback.
	*/
	static void
	print_status_config(zpool_handle_t zhp, status_cbdata_t cb, const char *name,
	nvlist_t nv, int depth, boolean_t isspare, vdev_rebuild_stat_t vrs)
	{
	nvlist_t *child, root;
	uint_t c, i, vsc, children;
	pool_scan_stat_t *ps = NULL;
	vdev_stat_t *vs;
	char rbuf[6], wbuf[6], cbuf[6];
	char *vname;
	uint64_t notpresent;
	spare_cbdata_t spare_cb;
	const char *state;
	char *type;
	char *path = NULL;
	char rcolor = NULL, wcolor = NULL, *ccolor = NULL;

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0)
	children = 0;

	verify(nvlist_lookup_uint64_array(nv, ZPOOL_CONFIG_VDEV_STATS,
	(uint64_t **)&vs, &vsc) == 0);

	verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0);

	if (strcmp(type, VDEV_TYPE_INDIRECT) == 0)
	return;

	state = zpool_state_to_name(vs->vs_state, vs->vs_aux);

	if (isspare) {
	/*
	* For hot spares, we use the terms 'INUSE' and 'AVAILABLE' for
	* online drives.
	*/
	if (vs->vs_aux == VDEV_AUX_SPARED)
	state = gettext("INUSE");
	else if (vs->vs_state == VDEV_STATE_HEALTHY)
	state = gettext("AVAIL");
	}

	printf_color(health_str_to_color(state),
	"\t%s%-s %-8s", depth, "", cb->cb_namewidth - depth,
	name, state);

	if (!isspare) {
	if (vs->vs_read_errors)
	rcolor = ANSI_RED;

	if (vs->vs_write_errors)
	wcolor = ANSI_RED;

	if (vs->vs_checksum_errors)
	ccolor = ANSI_RED;

	if (cb->cb_literal) {
	printf(" ");
	printf_color(rcolor, "%5llu",
	(u_longlong_t)vs->vs_read_errors);
	printf(" ");
	printf_color(wcolor, "%5llu",
	(u_longlong_t)vs->vs_write_errors);
	printf(" ");
	printf_color(ccolor, "%5llu",
	(u_longlong_t)vs->vs_checksum_errors);
	} else {
	zfs_nicenum(vs->vs_read_errors, rbuf, sizeof (rbuf));
	zfs_nicenum(vs->vs_write_errors, wbuf, sizeof (wbuf));
	zfs_nicenum(vs->vs_checksum_errors, cbuf,
	sizeof (cbuf));
	printf(" ");
	printf_color(rcolor, "%5s", rbuf);
	printf(" ");
	printf_color(wcolor, "%5s", wbuf);
	printf(" ");
	printf_color(ccolor, "%5s", cbuf);
	}
	if (cb->cb_print_slow_ios) {
	if (children == 0) {
	/* Only leafs vdevs have slow IOs */
	zfs_nicenum(vs->vs_slow_ios, rbuf,
	sizeof (rbuf));
	} else {
	snprintf(rbuf, sizeof (rbuf), "-");
	}

	if (cb->cb_literal)
	printf(" %5llu", (u_longlong_t)vs->vs_slow_ios);
	else
	printf(" %5s", rbuf);
	}
	}

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT,
	&notpresent) == 0) {
	verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0);
	(void) printf(" %s %s", gettext("was"), path);
	} else if (vs->vs_aux != 0) {
	(void) printf(" ");
	color_start(ANSI_RED);
	switch (vs->vs_aux) {
	case VDEV_AUX_OPEN_FAILED:
	(void) printf(gettext("cannot open"));
	break;

	case VDEV_AUX_BAD_GUID_SUM:
	(void) printf(gettext("missing device"));
	break;

	case VDEV_AUX_NO_REPLICAS:
	(void) printf(gettext("insufficient replicas"));
	break;

	case VDEV_AUX_VERSION_NEWER:
	(void) printf(gettext("newer version"));
	break;

	case VDEV_AUX_UNSUP_FEAT:
	(void) printf(gettext("unsupported feature(s)"));
	break;

	case VDEV_AUX_ASHIFT_TOO_BIG:
	(void) printf(gettext("unsupported minimum blocksize"));
	break;

	case VDEV_AUX_SPARED:
	verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID,
	&spare_cb.cb_guid) == 0);
	if (zpool_iter(g_zfs, find_spare, &spare_cb) == 1) {
	if (strcmp(zpool_get_name(spare_cb.cb_zhp),
	zpool_get_name(zhp)) == 0)
	(void) printf(gettext("currently in "
	"use"));
	else
	(void) printf(gettext("in use by "
	"pool '%s'"),
	zpool_get_name(spare_cb.cb_zhp));
	zpool_close(spare_cb.cb_zhp);
	} else {
	(void) printf(gettext("currently in use"));
	}
	break;

	case VDEV_AUX_ERR_EXCEEDED:
	(void) printf(gettext("too many errors"));
	break;

	case VDEV_AUX_IO_FAILURE:
	(void) printf(gettext("experienced I/O failures"));
	break;

	case VDEV_AUX_BAD_LOG:
	(void) printf(gettext("bad intent log"));
	break;

	case VDEV_AUX_EXTERNAL:
	(void) printf(gettext("external device fault"));
	break;

	case VDEV_AUX_SPLIT_POOL:
	(void) printf(gettext("split into new pool"));
	break;

	case VDEV_AUX_ACTIVE:
	(void) printf(gettext("currently in use"));
	break;

	case VDEV_AUX_CHILDREN_OFFLINE:
	(void) printf(gettext("all children offline"));
	break;

	case VDEV_AUX_BAD_LABEL:
	(void) printf(gettext("invalid label"));
	break;

	default:
	(void) printf(gettext("corrupted data"));
	break;
	}
	color_end();
	} else if (children == 0 && !isspare &&
	getenv("ZPOOL_STATUS_NON_NATIVE_ASHIFT_IGNORE") == NULL &&
	VDEV_STAT_VALID(vs_physical_ashift, vsc) &&
	vs->vs_configured_ashift < vs->vs_physical_ashift) {
	(void) printf(
	gettext(" block size: %dB configured, %dB native"),
	1 << vs->vs_configured_ashift, 1 << vs->vs_physical_ashift);
	}

	+ if (vs->vs_scan_removing != 0) {
	+ (void) printf(gettext(" (removing)"));
	+ } else if (vs->vs_noalloc != 0) {
	+ (void) printf(gettext(" (non-allocating)"));
	+ }
	+
	/* The root vdev has the scrub/resilver stats */
	root = fnvlist_lookup_nvlist(zpool_get_config(zhp, NULL),
	ZPOOL_CONFIG_VDEV_TREE);
	(void) nvlist_lookup_uint64_array(root, ZPOOL_CONFIG_SCAN_STATS,
	(uint64_t **)&ps, &c);

	if (ps != NULL && ps->pss_state == DSS_SCANNING && children == 0) {
	if (vs->vs_scan_processed != 0) {
	(void) printf(gettext(" (%s)"),
	(ps->pss_func == POOL_SCAN_RESILVER) ?
	"resilvering" : "repairing");
	} else if (vs->vs_resilver_deferred) {
	(void) printf(gettext(" (awaiting resilver)"));
	}
	}

	/* The top-level vdevs have the rebuild stats */
	if (vrs != NULL && vrs->vrs_state == VDEV_REBUILD_ACTIVE &&
	children == 0) {
	if (vs->vs_rebuild_processed != 0) {
	(void) printf(gettext(" (resilvering)"));
	}
	}

	if (cb->vcdl != NULL) {
	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0) {
	printf(" ");
	zpool_print_cmd(cb->vcdl, zpool_get_name(zhp), path);
	}
	}

	/* Display vdev initialization and trim status for leaves. */
	if (children == 0) {
	print_status_initialize(vs, cb->cb_print_vdev_init);
	print_status_trim(vs, cb->cb_print_vdev_trim);
	}

	(void) printf("\n");

	for (c = 0; c < children; c++) {
	uint64_t islog = B_FALSE, ishole = B_FALSE;

	/* Don't print logs or holes here */
	(void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG,
	&islog);
	(void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_HOLE,
	&ishole);
	if (islog \|\| ishole)
	continue;
	/* Only print normal classes here */
	if (nvlist_exists(child[c], ZPOOL_CONFIG_ALLOCATION_BIAS))
	continue;

	/* Provide vdev_rebuild_stats to children if available */
	if (vrs == NULL) {
	(void) nvlist_lookup_uint64_array(nv,
	ZPOOL_CONFIG_REBUILD_STATS,
	(uint64_t **)&vrs, &i);
	}

	vname = zpool_vdev_name(g_zfs, zhp, child[c],
	cb->cb_name_flags \| VDEV_NAME_TYPE_ID);
	print_status_config(zhp, cb, vname, child[c], depth + 2,
	isspare, vrs);
	free(vname);
	}
	}

	/*
	* Print the configuration of an exported pool. Iterate over all vdevs in the
	* pool, printing out the name and status for each one.
	*/
	static void
	print_import_config(status_cbdata_t cb, const char name, nvlist_t *nv,
	int depth)
	{
	nvlist_t **child;
	uint_t c, children;
	vdev_stat_t *vs;
	char type, vname;

	verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0);
	if (strcmp(type, VDEV_TYPE_MISSING) == 0 \|\|
	strcmp(type, VDEV_TYPE_HOLE) == 0)
	return;

	verify(nvlist_lookup_uint64_array(nv, ZPOOL_CONFIG_VDEV_STATS,
	(uint64_t **)&vs, &c) == 0);

	(void) printf("\t%s%-s", depth, "", cb->cb_namewidth - depth, name);
	(void) printf(" %s", zpool_state_to_name(vs->vs_state, vs->vs_aux));

	if (vs->vs_aux != 0) {
	(void) printf(" ");

	switch (vs->vs_aux) {
	case VDEV_AUX_OPEN_FAILED:
	(void) printf(gettext("cannot open"));
	break;

	case VDEV_AUX_BAD_GUID_SUM:
	(void) printf(gettext("missing device"));
	break;

	case VDEV_AUX_NO_REPLICAS:
	(void) printf(gettext("insufficient replicas"));
	break;

	case VDEV_AUX_VERSION_NEWER:
	(void) printf(gettext("newer version"));
	break;

	case VDEV_AUX_UNSUP_FEAT:
	(void) printf(gettext("unsupported feature(s)"));
	break;

	case VDEV_AUX_ERR_EXCEEDED:
	(void) printf(gettext("too many errors"));
	break;

	case VDEV_AUX_ACTIVE:
	(void) printf(gettext("currently in use"));
	break;

	case VDEV_AUX_CHILDREN_OFFLINE:
	(void) printf(gettext("all children offline"));
	break;

	case VDEV_AUX_BAD_LABEL:
	(void) printf(gettext("invalid label"));
	break;

	default:
	(void) printf(gettext("corrupted data"));
	break;
	}
	}
	(void) printf("\n");

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0)
	return;

	for (c = 0; c < children; c++) {
	uint64_t is_log = B_FALSE;

	(void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG,
	&is_log);
	if (is_log)
	continue;
	if (nvlist_exists(child[c], ZPOOL_CONFIG_ALLOCATION_BIAS))
	continue;

	vname = zpool_vdev_name(g_zfs, NULL, child[c],
	cb->cb_name_flags \| VDEV_NAME_TYPE_ID);
	print_import_config(cb, vname, child[c], depth + 2);
	free(vname);
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE,
	&child, &children) == 0) {
	(void) printf(gettext("\tcache\n"));
	for (c = 0; c < children; c++) {
	vname = zpool_vdev_name(g_zfs, NULL, child[c],
	cb->cb_name_flags);
	(void) printf("\t %s\n", vname);
	free(vname);
	}
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES,
	&child, &children) == 0) {
	(void) printf(gettext("\tspares\n"));
	for (c = 0; c < children; c++) {
	vname = zpool_vdev_name(g_zfs, NULL, child[c],
	cb->cb_name_flags);
	(void) printf("\t %s\n", vname);
	free(vname);
	}
	}
	}

	/*
	* Print specialized class vdevs.
	*
	* These are recorded as top level vdevs in the main pool child array
	* but with "is_log" set to 1 or an "alloc_bias" string. We use either
	* print_status_config() or print_import_config() to print the top level
	* class vdevs then any of their children (eg mirrored slogs) are printed
	* recursively - which works because only the top level vdev is marked.
	*/
	static void
	print_class_vdevs(zpool_handle_t zhp, status_cbdata_t cb, nvlist_t *nv,
	const char *class)
	{
	uint_t c, children;
	nvlist_t **child;
	boolean_t printed = B_FALSE;

	assert(zhp != NULL \|\| !cb->cb_verbose);

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN, &child,
	&children) != 0)
	return;

	for (c = 0; c < children; c++) {
	uint64_t is_log = B_FALSE;
	char *bias = NULL;
	char *type = NULL;

	(void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG,
	&is_log);

	if (is_log) {
	bias = VDEV_ALLOC_CLASS_LOGS;
	} else {
	(void) nvlist_lookup_string(child[c],
	ZPOOL_CONFIG_ALLOCATION_BIAS, &bias);
	(void) nvlist_lookup_string(child[c],
	ZPOOL_CONFIG_TYPE, &type);
	}

	if (bias == NULL \|\| strcmp(bias, class) != 0)
	continue;
	if (!is_log && strcmp(type, VDEV_TYPE_INDIRECT) == 0)
	continue;

	if (!printed) {
	(void) printf("\t%s\t\n", gettext(class));
	printed = B_TRUE;
	}

	char *name = zpool_vdev_name(g_zfs, zhp, child[c],
	cb->cb_name_flags \| VDEV_NAME_TYPE_ID);
	if (cb->cb_print_status)
	print_status_config(zhp, cb, name, child[c], 2,
	B_FALSE, NULL);
	else
	print_import_config(cb, name, child[c], 2);
	free(name);
	}
	}

	/*
	* Display the status for the given pool.
	*/
	static int
	show_import(nvlist_t *config, boolean_t report_error)
	{
	uint64_t pool_state;
	vdev_stat_t *vs;
	char *name;
	uint64_t guid;
	uint64_t hostid = 0;
	char *msgid;
	char *hostname = "unknown";
	nvlist_t nvroot, nvinfo;
	zpool_status_t reason;
	zpool_errata_t errata;
	const char *health;
	uint_t vsc;
	char *comment;
	status_cbdata_t cb = { 0 };

	verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME,
	&name) == 0);
	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
	&guid) == 0);
	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE,
	&pool_state) == 0);
	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) == 0);

	verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_VDEV_STATS,
	(uint64_t **)&vs, &vsc) == 0);
	health = zpool_state_to_name(vs->vs_state, vs->vs_aux);

	reason = zpool_import_status(config, &msgid, &errata);

	/*
	* If we're importing using a cachefile, then we won't report any
	* errors unless we are in the scan phase of the import.
	*/
	if (reason != ZPOOL_STATUS_OK && !report_error)
	return (reason);

	(void) printf(gettext(" pool: %s\n"), name);
	(void) printf(gettext(" id: %llu\n"), (u_longlong_t)guid);
	(void) printf(gettext(" state: %s"), health);
	if (pool_state == POOL_STATE_DESTROYED)
	(void) printf(gettext(" (DESTROYED)"));
	(void) printf("\n");

	switch (reason) {
	case ZPOOL_STATUS_MISSING_DEV_R:
	case ZPOOL_STATUS_MISSING_DEV_NR:
	case ZPOOL_STATUS_BAD_GUID_SUM:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices are "
	"missing from the system.\n"));
	break;

	case ZPOOL_STATUS_CORRUPT_LABEL_R:
	case ZPOOL_STATUS_CORRUPT_LABEL_NR:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices contains"
	" corrupted data.\n"));
	break;

	case ZPOOL_STATUS_CORRUPT_DATA:
	(void) printf(
	gettext(" status: The pool data is corrupted.\n"));
	break;

	case ZPOOL_STATUS_OFFLINE_DEV:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices "
	"are offlined.\n"));
	break;

	case ZPOOL_STATUS_CORRUPT_POOL:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool metadata is "
	"corrupted.\n"));
	break;

	case ZPOOL_STATUS_VERSION_OLDER:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool is formatted using "
	"a legacy on-disk version.\n"));
	break;

	case ZPOOL_STATUS_VERSION_NEWER:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool is formatted using "
	"an incompatible version.\n"));
	break;

	case ZPOOL_STATUS_FEAT_DISABLED:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("Some supported "
	"features are not enabled on the pool.\n\t"
	"(Note that they may be intentionally disabled "
	"if the\n\t'compatibility' property is set.)\n"));
	break;

	case ZPOOL_STATUS_COMPATIBILITY_ERR:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("Error reading or parsing "
	"the file(s) indicated by the 'compatibility'\n"
	"property.\n"));
	break;

	case ZPOOL_STATUS_INCOMPATIBLE_FEAT:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more features "
	"are enabled on the pool despite not being\n"
	"requested by the 'compatibility' property.\n"));
	break;

	case ZPOOL_STATUS_UNSUP_FEAT_READ:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool uses the following "
	"feature(s) not supported on this system:\n"));
	color_start(ANSI_YELLOW);
	zpool_print_unsup_feat(config);
	color_end();
	break;

	case ZPOOL_STATUS_UNSUP_FEAT_WRITE:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool can only be "
	"accessed in read-only mode on this system. It\n\tcannot be"
	" accessed in read-write mode because it uses the "
	"following\n\tfeature(s) not supported on this system:\n"));
	color_start(ANSI_YELLOW);
	zpool_print_unsup_feat(config);
	color_end();
	break;

	case ZPOOL_STATUS_HOSTID_ACTIVE:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool is currently "
	"imported by another system.\n"));
	break;

	case ZPOOL_STATUS_HOSTID_REQUIRED:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool has the "
	"multihost property on. It cannot\n\tbe safely imported "
	"when the system hostid is not set.\n"));
	break;

	case ZPOOL_STATUS_HOSTID_MISMATCH:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool was last accessed "
	"by another system.\n"));
	break;

	case ZPOOL_STATUS_FAULTED_DEV_R:
	case ZPOOL_STATUS_FAULTED_DEV_NR:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices are "
	"faulted.\n"));
	break;

	case ZPOOL_STATUS_BAD_LOG:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("An intent log record cannot "
	"be read.\n"));
	break;

	case ZPOOL_STATUS_RESILVERING:
	case ZPOOL_STATUS_REBUILDING:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices were "
	"being resilvered.\n"));
	break;

	case ZPOOL_STATUS_ERRATA:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("Errata #%d detected.\n"),
	errata);
	break;

	case ZPOOL_STATUS_NON_NATIVE_ASHIFT:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices are "
	"configured to use a non-native block size.\n"
	"\tExpect reduced performance.\n"));
	break;

	default:
	/*
	* No other status can be seen when importing pools.
	*/
	assert(reason == ZPOOL_STATUS_OK);
	}

	/*
	* Print out an action according to the overall state of the pool.
	*/
	if (vs->vs_state == VDEV_STATE_HEALTHY) {
	if (reason == ZPOOL_STATUS_VERSION_OLDER \|\|
	reason == ZPOOL_STATUS_FEAT_DISABLED) {
	(void) printf(gettext(" action: The pool can be "
	"imported using its name or numeric identifier, "
	"though\n\tsome features will not be available "
	"without an explicit 'zpool upgrade'.\n"));
	} else if (reason == ZPOOL_STATUS_COMPATIBILITY_ERR) {
	(void) printf(gettext(" action: The pool can be "
	"imported using its name or numeric\n\tidentifier, "
	"though the file(s) indicated by its "
	"'compatibility'\n\tproperty cannot be parsed at "
	"this time.\n"));
	} else if (reason == ZPOOL_STATUS_HOSTID_MISMATCH) {
	(void) printf(gettext(" action: The pool can be "
	"imported using its name or numeric "
	"identifier and\n\tthe '-f' flag.\n"));
	} else if (reason == ZPOOL_STATUS_ERRATA) {
	switch (errata) {
	case ZPOOL_ERRATA_NONE:
	break;

	case ZPOOL_ERRATA_ZOL_2094_SCRUB:
	(void) printf(gettext(" action: The pool can "
	"be imported using its name or numeric "
	"identifier,\n\thowever there is a compat"
	"ibility issue which should be corrected"
	"\n\tby running 'zpool scrub'\n"));
	break;

	case ZPOOL_ERRATA_ZOL_2094_ASYNC_DESTROY:
	(void) printf(gettext(" action: The pool can"
	"not be imported with this version of ZFS "
	"due to\n\tan active asynchronous destroy. "
	"Revert to an earlier version\n\tand "
	"allow the destroy to complete before "
	"updating.\n"));
	break;

	case ZPOOL_ERRATA_ZOL_6845_ENCRYPTION:
	(void) printf(gettext(" action: Existing "
	"encrypted datasets contain an on-disk "
	"incompatibility, which\n\tneeds to be "
	"corrected. Backup these datasets to new "
	"encrypted datasets\n\tand destroy the "
	"old ones.\n"));
	break;

	case ZPOOL_ERRATA_ZOL_8308_ENCRYPTION:
	(void) printf(gettext(" action: Existing "
	"encrypted snapshots and bookmarks contain "
	"an on-disk\n\tincompatibility. This may "
	"cause on-disk corruption if they are used"
	"\n\twith 'zfs recv'. To correct the "
	"issue, enable the bookmark_v2 feature.\n\t"
	"No additional action is needed if there "
	"are no encrypted snapshots or\n\t"
	"bookmarks. If preserving the encrypted "
	"snapshots and bookmarks is\n\trequired, "
	"use a non-raw send to backup and restore "
	"them. Alternately,\n\tthey may be removed"
	" to resolve the incompatibility.\n"));
	break;
	default:
	/*
	* All errata must contain an action message.
	*/
	assert(0);
	}
	} else {
	(void) printf(gettext(" action: The pool can be "
	"imported using its name or numeric "
	"identifier.\n"));
	}
	} else if (vs->vs_state == VDEV_STATE_DEGRADED) {
	(void) printf(gettext(" action: The pool can be imported "
	"despite missing or damaged devices. The\n\tfault "
	"tolerance of the pool may be compromised if imported.\n"));
	} else {
	switch (reason) {
	case ZPOOL_STATUS_VERSION_NEWER:
	(void) printf(gettext(" action: The pool cannot be "
	"imported. Access the pool on a system running "
	"newer\n\tsoftware, or recreate the pool from "
	"backup.\n"));
	break;
	case ZPOOL_STATUS_UNSUP_FEAT_READ:
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("The pool cannot be "
	"imported. Access the pool on a system that "
	"supports\n\tthe required feature(s), or recreate "
	"the pool from backup.\n"));
	break;
	case ZPOOL_STATUS_UNSUP_FEAT_WRITE:
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("The pool cannot be "
	"imported in read-write mode. Import the pool "
	"with\n"
	"\t\"-o readonly=on\", access the pool on a system "
	"that supports the\n\trequired feature(s), or "
	"recreate the pool from backup.\n"));
	break;
	case ZPOOL_STATUS_MISSING_DEV_R:
	case ZPOOL_STATUS_MISSING_DEV_NR:
	case ZPOOL_STATUS_BAD_GUID_SUM:
	(void) printf(gettext(" action: The pool cannot be "
	"imported. Attach the missing\n\tdevices and try "
	"again.\n"));
	break;
	case ZPOOL_STATUS_HOSTID_ACTIVE:
	VERIFY0(nvlist_lookup_nvlist(config,
	ZPOOL_CONFIG_LOAD_INFO, &nvinfo));

	if (nvlist_exists(nvinfo, ZPOOL_CONFIG_MMP_HOSTNAME))
	hostname = fnvlist_lookup_string(nvinfo,
	ZPOOL_CONFIG_MMP_HOSTNAME);

	if (nvlist_exists(nvinfo, ZPOOL_CONFIG_MMP_HOSTID))
	hostid = fnvlist_lookup_uint64(nvinfo,
	ZPOOL_CONFIG_MMP_HOSTID);

	(void) printf(gettext(" action: The pool must be "
	"exported from %s (hostid=%lx)\n\tbefore it "
	"can be safely imported.\n"), hostname,
	(unsigned long) hostid);
	break;
	case ZPOOL_STATUS_HOSTID_REQUIRED:
	(void) printf(gettext(" action: Set a unique system "
	"hostid with the zgenhostid(8) command.\n"));
	break;
	default:
	(void) printf(gettext(" action: The pool cannot be "
	"imported due to damaged devices or data.\n"));
	}
	}

	/* Print the comment attached to the pool. */
	if (nvlist_lookup_string(config, ZPOOL_CONFIG_COMMENT, &comment) == 0)
	(void) printf(gettext("comment: %s\n"), comment);

	/*
	* If the state is "closed" or "can't open", and the aux state
	* is "corrupt data":
	*/
	if (((vs->vs_state == VDEV_STATE_CLOSED) \|\|
	(vs->vs_state == VDEV_STATE_CANT_OPEN)) &&
	(vs->vs_aux == VDEV_AUX_CORRUPT_DATA)) {
	if (pool_state == POOL_STATE_DESTROYED)
	(void) printf(gettext("\tThe pool was destroyed, "
	"but can be imported using the '-Df' flags.\n"));
	else if (pool_state != POOL_STATE_EXPORTED)
	(void) printf(gettext("\tThe pool may be active on "
	"another system, but can be imported using\n\t"
	"the '-f' flag.\n"));
	}

	if (msgid != NULL) {
	(void) printf(gettext(
	" see: https://openzfs.github.io/openzfs-docs/msg/%s\n"),
	msgid);
	}

	(void) printf(gettext(" config:\n\n"));

	cb.cb_namewidth = max_width(NULL, nvroot, 0, strlen(name),
	VDEV_NAME_TYPE_ID);
	if (cb.cb_namewidth < 10)
	cb.cb_namewidth = 10;

	print_import_config(&cb, name, nvroot, 0);

	print_class_vdevs(NULL, &cb, nvroot, VDEV_ALLOC_BIAS_DEDUP);
	print_class_vdevs(NULL, &cb, nvroot, VDEV_ALLOC_BIAS_SPECIAL);
	print_class_vdevs(NULL, &cb, nvroot, VDEV_ALLOC_CLASS_LOGS);

	if (reason == ZPOOL_STATUS_BAD_GUID_SUM) {
	(void) printf(gettext("\n\tAdditional devices are known to "
	"be part of this pool, though their\n\texact "
	"configuration cannot be determined.\n"));
	}
	return (0);
	}

	static boolean_t
	zfs_force_import_required(nvlist_t *config)
	{
	uint64_t state;
	uint64_t hostid = 0;
	nvlist_t *nvinfo;

	state = fnvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE);
	(void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_HOSTID, &hostid);

	if (state != POOL_STATE_EXPORTED && hostid != get_system_hostid())
	return (B_TRUE);

	nvinfo = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO);
	if (nvlist_exists(nvinfo, ZPOOL_CONFIG_MMP_STATE)) {
	mmp_state_t mmp_state = fnvlist_lookup_uint64(nvinfo,
	ZPOOL_CONFIG_MMP_STATE);

	if (mmp_state != MMP_STATE_INACTIVE)
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	/*
	* Perform the import for the given configuration. This passes the heavy
	* lifting off to zpool_import_props(), and then mounts the datasets contained
	* within the pool.
	*/
	static int
	do_import(nvlist_t config, const char newname, const char *mntopts,
	nvlist_t *props, int flags)
	{
	int ret = 0;
	zpool_handle_t *zhp;
	char *name;
	uint64_t version;

	name = fnvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME);
	version = fnvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION);

	if (!SPA_VERSION_IS_SUPPORTED(version)) {
	(void) fprintf(stderr, gettext("cannot import '%s': pool "
	"is formatted using an unsupported ZFS version\n"), name);
	return (1);
	} else if (zfs_force_import_required(config) &&
	!(flags & ZFS_IMPORT_ANY_HOST)) {
	mmp_state_t mmp_state = MMP_STATE_INACTIVE;
	nvlist_t *nvinfo;

	nvinfo = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO);
	if (nvlist_exists(nvinfo, ZPOOL_CONFIG_MMP_STATE))
	mmp_state = fnvlist_lookup_uint64(nvinfo,
	ZPOOL_CONFIG_MMP_STATE);

	if (mmp_state == MMP_STATE_ACTIVE) {
	char *hostname = "<unknown>";
	uint64_t hostid = 0;

	if (nvlist_exists(nvinfo, ZPOOL_CONFIG_MMP_HOSTNAME))
	hostname = fnvlist_lookup_string(nvinfo,
	ZPOOL_CONFIG_MMP_HOSTNAME);

	if (nvlist_exists(nvinfo, ZPOOL_CONFIG_MMP_HOSTID))
	hostid = fnvlist_lookup_uint64(nvinfo,
	ZPOOL_CONFIG_MMP_HOSTID);

	(void) fprintf(stderr, gettext("cannot import '%s': "
	"pool is imported on %s (hostid: "
	"0x%lx)\nExport the pool on the other system, "
	"then run 'zpool import'.\n"),
	name, hostname, (unsigned long) hostid);
	} else if (mmp_state == MMP_STATE_NO_HOSTID) {
	(void) fprintf(stderr, gettext("Cannot import '%s': "
	"pool has the multihost property on and the\n"
	"system's hostid is not set. Set a unique hostid "
	"with the zgenhostid(8) command.\n"), name);
	} else {
	char *hostname = "<unknown>";
	uint64_t timestamp = 0;
	uint64_t hostid = 0;

	if (nvlist_exists(config, ZPOOL_CONFIG_HOSTNAME))
	hostname = fnvlist_lookup_string(config,
	ZPOOL_CONFIG_HOSTNAME);

	if (nvlist_exists(config, ZPOOL_CONFIG_TIMESTAMP))
	timestamp = fnvlist_lookup_uint64(config,
	ZPOOL_CONFIG_TIMESTAMP);

	if (nvlist_exists(config, ZPOOL_CONFIG_HOSTID))
	hostid = fnvlist_lookup_uint64(config,
	ZPOOL_CONFIG_HOSTID);

	(void) fprintf(stderr, gettext("cannot import '%s': "
	"pool was previously in use from another system.\n"
	"Last accessed by %s (hostid=%lx) at %s"
	"The pool can be imported, use 'zpool import -f' "
	"to import the pool.\n"), name, hostname,
	(unsigned long)hostid, ctime((time_t *)&timestamp));
	}

	return (1);
	}

	if (zpool_import_props(g_zfs, config, newname, props, flags) != 0)
	return (1);

	if (newname != NULL)
	name = (char *)newname;

	if ((zhp = zpool_open_canfail(g_zfs, name)) == NULL)
	return (1);

	/*
	* Loading keys is best effort. We don't want to return immediately
	* if it fails but we do want to give the error to the caller.
	*/
	if (flags & ZFS_IMPORT_LOAD_KEYS) {
	ret = zfs_crypto_attempt_load_keys(g_zfs, name);
	if (ret != 0)
	ret = 1;
	}

	if (zpool_get_state(zhp) != POOL_STATE_UNAVAIL &&
	!(flags & ZFS_IMPORT_ONLY) &&
	zpool_enable_datasets(zhp, mntopts, 0) != 0) {
	zpool_close(zhp);
	return (1);
	}

	zpool_close(zhp);
	return (ret);
	}

	static int
	import_pools(nvlist_t pools, nvlist_t props, char *mntopts, int flags,
	char orig_name, char new_name,
	boolean_t do_destroyed, boolean_t pool_specified, boolean_t do_all,
	importargs_t *import)
	{
	nvlist_t *config = NULL;
	nvlist_t *found_config = NULL;
	uint64_t pool_state;

	/*
	* At this point we have a list of import candidate configs. Even if
	* we were searching by pool name or guid, we still need to
	* post-process the list to deal with pool state and possible
	* duplicate names.
	*/
	int err = 0;
	nvpair_t *elem = NULL;
	boolean_t first = B_TRUE;
	while ((elem = nvlist_next_nvpair(pools, elem)) != NULL) {

	verify(nvpair_value_nvlist(elem, &config) == 0);

	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE,
	&pool_state) == 0);
	if (!do_destroyed && pool_state == POOL_STATE_DESTROYED)
	continue;
	if (do_destroyed && pool_state != POOL_STATE_DESTROYED)
	continue;

	verify(nvlist_add_nvlist(config, ZPOOL_LOAD_POLICY,
	import->policy) == 0);

	if (!pool_specified) {
	if (first)
	first = B_FALSE;
	else if (!do_all)
	(void) printf("\n");

	if (do_all) {
	err \|= do_import(config, NULL, mntopts,
	props, flags);
	} else {
	/*
	* If we're importing from cachefile, then
	* we don't want to report errors until we
	* are in the scan phase of the import. If
	* we get an error, then we return that error
	* to invoke the scan phase.
	*/
	if (import->cachefile && !import->scan)
	err = show_import(config, B_FALSE);
	else
	(void) show_import(config, B_TRUE);
	}
	} else if (import->poolname != NULL) {
	char *name;

	/*
	* We are searching for a pool based on name.
	*/
	verify(nvlist_lookup_string(config,
	ZPOOL_CONFIG_POOL_NAME, &name) == 0);

	if (strcmp(name, import->poolname) == 0) {
	if (found_config != NULL) {
	(void) fprintf(stderr, gettext(
	"cannot import '%s': more than "
	"one matching pool\n"),
	import->poolname);
	(void) fprintf(stderr, gettext(
	"import by numeric ID instead\n"));
	err = B_TRUE;
	}
	found_config = config;
	}
	} else {
	uint64_t guid;

	/*
	* Search for a pool by guid.
	*/
	verify(nvlist_lookup_uint64(config,
	ZPOOL_CONFIG_POOL_GUID, &guid) == 0);

	if (guid == import->guid)
	found_config = config;
	}
	}

	/*
	* If we were searching for a specific pool, verify that we found a
	* pool, and then do the import.
	*/
	if (pool_specified && err == 0) {
	if (found_config == NULL) {
	(void) fprintf(stderr, gettext("cannot import '%s': "
	"no such pool available\n"), orig_name);
	err = B_TRUE;
	} else {
	err \|= do_import(found_config, new_name,
	mntopts, props, flags);
	}
	}

	/*
	* If we were just looking for pools, report an error if none were
	* found.
	*/
	if (!pool_specified && first)
	(void) fprintf(stderr,
	gettext("no pools available to import\n"));
	return (err);
	}

	typedef struct target_exists_args {
	const char *poolname;
	uint64_t poolguid;
	} target_exists_args_t;

	static int
	name_or_guid_exists(zpool_handle_t zhp, void data)
	{
	target_exists_args_t *args = data;
	nvlist_t *config = zpool_get_config(zhp, NULL);
	int found = 0;

	if (config == NULL)
	return (0);

	if (args->poolname != NULL) {
	char *pool_name;

	verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME,
	&pool_name) == 0);
	if (strcmp(pool_name, args->poolname) == 0)
	found = 1;
	} else {
	uint64_t pool_guid;

	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
	&pool_guid) == 0);
	if (pool_guid == args->poolguid)
	found = 1;
	}
	zpool_close(zhp);

	return (found);
	}
	/*
	* zpool checkpoint <pool>
	* checkpoint --discard <pool>
	*
	* -d Discard the checkpoint from a checkpointed
	* --discard pool.
	*
	* -w Wait for discarding a checkpoint to complete.
	* --wait
	*
	* Checkpoints the specified pool, by taking a "snapshot" of its
	* current state. A pool can only have one checkpoint at a time.
	*/
	int
	zpool_do_checkpoint(int argc, char **argv)
	{
	boolean_t discard, wait;
	char *pool;
	zpool_handle_t *zhp;
	int c, err;

	struct option long_options[] = {
	{"discard", no_argument, NULL, 'd'},
	{"wait", no_argument, NULL, 'w'},
	{0, 0, 0, 0}
	};

	discard = B_FALSE;
	wait = B_FALSE;
	while ((c = getopt_long(argc, argv, ":dw", long_options, NULL)) != -1) {
	switch (c) {
	case 'd':
	discard = B_TRUE;
	break;
	case 'w':
	wait = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	if (wait && !discard) {
	(void) fprintf(stderr, gettext("--wait only valid when "
	"--discard also specified\n"));
	usage(B_FALSE);
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool argument\n"));
	usage(B_FALSE);
	}

	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	pool = argv[0];

	if ((zhp = zpool_open(g_zfs, pool)) == NULL) {
	/* As a special case, check for use of '/' in the name */
	if (strchr(pool, '/') != NULL)
	(void) fprintf(stderr, gettext("'zpool checkpoint' "
	"doesn't work on datasets. To save the state "
	"of a dataset from a specific point in time "
	"please use 'zfs snapshot'\n"));
	return (1);
	}

	if (discard) {
	err = (zpool_discard_checkpoint(zhp) != 0);
	if (err == 0 && wait)
	err = zpool_wait(zhp, ZPOOL_WAIT_CKPT_DISCARD);
	} else {
	err = (zpool_checkpoint(zhp) != 0);
	}

	zpool_close(zhp);

	return (err);
	}

	#define CHECKPOINT_OPT 1024

	/*
	* zpool import [-d dir] [-D]
	* import [-o mntopts] [-o prop=value] ... [-R root] [-D] [-l]
	* [-d dir \| -c cachefile \| -s] [-f] -a
	* import [-o mntopts] [-o prop=value] ... [-R root] [-D] [-l]
	* [-d dir \| -c cachefile \| -s] [-f] [-n] [-F] <pool \| id>
	* [newpool]
	*
	* -c Read pool information from a cachefile instead of searching
	* devices. If importing from a cachefile config fails, then
	* fallback to searching for devices only in the directories that
	* exist in the cachefile.
	*
	* -d Scan in a specific directory, other than /dev/. More than
	* one directory can be specified using multiple '-d' options.
	*
	* -D Scan for previously destroyed pools or import all or only
	* specified destroyed pools.
	*
	* -R Temporarily import the pool, with all mountpoints relative to
	* the given root. The pool will remain exported when the machine
	* is rebooted.
	*
	* -V Import even in the presence of faulted vdevs. This is an
	* intentionally undocumented option for testing purposes, and
	* treats the pool configuration as complete, leaving any bad
	* vdevs in the FAULTED state. In other words, it does verbatim
	* import.
	*
	* -f Force import, even if it appears that the pool is active.
	*
	* -F Attempt rewind if necessary.
	*
	* -n See if rewind would work, but don't actually rewind.
	*
	* -N Import the pool but don't mount datasets.
	*
	* -T Specify a starting txg to use for import. This option is
	* intentionally undocumented option for testing purposes.
	*
	* -a Import all pools found.
	*
	* -l Load encryption keys while importing.
	*
	* -o Set property=value and/or temporary mount options (without '=').
	*
	* -s Scan using the default search path, the libblkid cache will
	* not be consulted.
	*
	* --rewind-to-checkpoint
	* Import the pool and revert back to the checkpoint.
	*
	* The import command scans for pools to import, and import pools based on pool
	* name and GUID. The pool can also be renamed as part of the import process.
	*/
	int
	zpool_do_import(int argc, char **argv)
	{
	char **searchdirs = NULL;
	char env, envdup = NULL;
	int nsearch = 0;
	int c;
	int err = 0;
	nvlist_t *pools = NULL;
	boolean_t do_all = B_FALSE;
	boolean_t do_destroyed = B_FALSE;
	char *mntopts = NULL;
	uint64_t searchguid = 0;
	char *searchname = NULL;
	char *propval;
	nvlist_t *policy = NULL;
	nvlist_t *props = NULL;
	int flags = ZFS_IMPORT_NORMAL;
	uint32_t rewind_policy = ZPOOL_NO_REWIND;
	boolean_t dryrun = B_FALSE;
	boolean_t do_rewind = B_FALSE;
	boolean_t xtreme_rewind = B_FALSE;
	boolean_t do_scan = B_FALSE;
	boolean_t pool_exists = B_FALSE;
	boolean_t pool_specified = B_FALSE;
	uint64_t txg = -1ULL;
	char *cachefile = NULL;
	importargs_t idata = { 0 };
	char *endptr;

	struct option long_options[] = {
	{"rewind-to-checkpoint", no_argument, NULL, CHECKPOINT_OPT},
	{0, 0, 0, 0}
	};

	/* check options */
	while ((c = getopt_long(argc, argv, ":aCc:d:DEfFlmnNo:R:stT:VX",
	long_options, NULL)) != -1) {
	switch (c) {
	case 'a':
	do_all = B_TRUE;
	break;
	case 'c':
	cachefile = optarg;
	break;
	case 'd':
	searchdirs = safe_realloc(searchdirs,
	(nsearch + 1) * sizeof (char *));
	searchdirs[nsearch++] = optarg;
	break;
	case 'D':
	do_destroyed = B_TRUE;
	break;
	case 'f':
	flags \|= ZFS_IMPORT_ANY_HOST;
	break;
	case 'F':
	do_rewind = B_TRUE;
	break;
	case 'l':
	flags \|= ZFS_IMPORT_LOAD_KEYS;
	break;
	case 'm':
	flags \|= ZFS_IMPORT_MISSING_LOG;
	break;
	case 'n':
	dryrun = B_TRUE;
	break;
	case 'N':
	flags \|= ZFS_IMPORT_ONLY;
	break;
	case 'o':
	if ((propval = strchr(optarg, '=')) != NULL) {
	*propval = '\0';
	propval++;
	if (add_prop_list(optarg, propval,
	&props, B_TRUE))
	goto error;
	} else {
	mntopts = optarg;
	}
	break;
	case 'R':
	if (add_prop_list(zpool_prop_to_name(
	ZPOOL_PROP_ALTROOT), optarg, &props, B_TRUE))
	goto error;
	if (add_prop_list_default(zpool_prop_to_name(
	ZPOOL_PROP_CACHEFILE), "none", &props, B_TRUE))
	goto error;
	break;
	case 's':
	do_scan = B_TRUE;
	break;
	case 't':
	flags \|= ZFS_IMPORT_TEMP_NAME;
	if (add_prop_list_default(zpool_prop_to_name(
	ZPOOL_PROP_CACHEFILE), "none", &props, B_TRUE))
	goto error;
	break;

	case 'T':
	errno = 0;
	txg = strtoull(optarg, &endptr, 0);
	if (errno != 0 \|\| *endptr != '\0') {
	(void) fprintf(stderr,
	gettext("invalid txg value\n"));
	usage(B_FALSE);
	}
	rewind_policy = ZPOOL_DO_REWIND \| ZPOOL_EXTREME_REWIND;
	break;
	case 'V':
	flags \|= ZFS_IMPORT_VERBATIM;
	break;
	case 'X':
	xtreme_rewind = B_TRUE;
	break;
	case CHECKPOINT_OPT:
	flags \|= ZFS_IMPORT_CHECKPOINT;
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	usage(B_FALSE);
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (cachefile && nsearch != 0) {
	(void) fprintf(stderr, gettext("-c is incompatible with -d\n"));
	usage(B_FALSE);
	}

	if (cachefile && do_scan) {
	(void) fprintf(stderr, gettext("-c is incompatible with -s\n"));
	usage(B_FALSE);
	}

	if ((flags & ZFS_IMPORT_LOAD_KEYS) && (flags & ZFS_IMPORT_ONLY)) {
	(void) fprintf(stderr, gettext("-l is incompatible with -N\n"));
	usage(B_FALSE);
	}

	if ((flags & ZFS_IMPORT_LOAD_KEYS) && !do_all && argc == 0) {
	(void) fprintf(stderr, gettext("-l is only meaningful during "
	"an import\n"));
	usage(B_FALSE);
	}

	if ((dryrun \|\| xtreme_rewind) && !do_rewind) {
	(void) fprintf(stderr,
	gettext("-n or -X only meaningful with -F\n"));
	usage(B_FALSE);
	}
	if (dryrun)
	rewind_policy = ZPOOL_TRY_REWIND;
	else if (do_rewind)
	rewind_policy = ZPOOL_DO_REWIND;
	if (xtreme_rewind)
	rewind_policy \|= ZPOOL_EXTREME_REWIND;

	/* In the future, we can capture further policy and include it here */
	if (nvlist_alloc(&policy, NV_UNIQUE_NAME, 0) != 0 \|\|
	nvlist_add_uint64(policy, ZPOOL_LOAD_REQUEST_TXG, txg) != 0 \|\|
	nvlist_add_uint32(policy, ZPOOL_LOAD_REWIND_POLICY,
	rewind_policy) != 0)
	goto error;

	/* check argument count */
	if (do_all) {
	if (argc != 0) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}
	} else {
	if (argc > 2) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}
	}

	/*
	* Check for the effective uid. We do this explicitly here because
	* otherwise any attempt to discover pools will silently fail.
	*/
	if (argc == 0 && geteuid() != 0) {
	(void) fprintf(stderr, gettext("cannot "
	"discover pools: permission denied\n"));
	if (searchdirs != NULL)
	free(searchdirs);

	nvlist_free(props);
	nvlist_free(policy);
	return (1);
	}

	/*
	* Depending on the arguments given, we do one of the following:
	*
	* <none> Iterate through all pools and display information about
	* each one.
	*
	* -a Iterate through all pools and try to import each one.
	*
	* <id> Find the pool that corresponds to the given GUID/pool
	* name and import that one.
	*
	* -D Above options applies only to destroyed pools.
	*/
	if (argc != 0) {
	char *endptr;

	errno = 0;
	searchguid = strtoull(argv[0], &endptr, 10);
	if (errno != 0 \|\| *endptr != '\0') {
	searchname = argv[0];
	searchguid = 0;
	}
	pool_specified = B_TRUE;

	/*
	* User specified a name or guid. Ensure it's unique.
	*/
	target_exists_args_t search = {searchname, searchguid};
	pool_exists = zpool_iter(g_zfs, name_or_guid_exists, &search);
	}

	/*
	* Check the environment for the preferred search path.
	*/
	if ((searchdirs == NULL) && (env = getenv("ZPOOL_IMPORT_PATH"))) {
	char dir, tmp = NULL;

	envdup = strdup(env);

	for (dir = strtok_r(envdup, ":", &tmp);
	dir != NULL;
	dir = strtok_r(NULL, ":", &tmp)) {
	searchdirs = safe_realloc(searchdirs,
	(nsearch + 1) * sizeof (char *));
	searchdirs[nsearch++] = dir;
	}
	}

	idata.path = searchdirs;
	idata.paths = nsearch;
	idata.poolname = searchname;
	idata.guid = searchguid;
	idata.cachefile = cachefile;
	idata.scan = do_scan;
	idata.policy = policy;

	pools = zpool_search_import(g_zfs, &idata, &libzfs_config_ops);

	if (pools != NULL && pool_exists &&
	(argc == 1 \|\| strcmp(argv[0], argv[1]) == 0)) {
	(void) fprintf(stderr, gettext("cannot import '%s': "
	"a pool with that name already exists\n"),
	argv[0]);
	(void) fprintf(stderr, gettext("use the form '%s "
	"<pool \| id> <newpool>' to give it a new name\n"),
	"zpool import");
	err = 1;
	} else if (pools == NULL && pool_exists) {
	(void) fprintf(stderr, gettext("cannot import '%s': "
	"a pool with that name is already created/imported,\n"),
	argv[0]);
	(void) fprintf(stderr, gettext("and no additional pools "
	"with that name were found\n"));
	err = 1;
	} else if (pools == NULL) {
	if (argc != 0) {
	(void) fprintf(stderr, gettext("cannot import '%s': "
	"no such pool available\n"), argv[0]);
	}
	err = 1;
	}

	if (err == 1) {
	free(searchdirs);
	free(envdup);
	nvlist_free(policy);
	nvlist_free(pools);
	nvlist_free(props);
	return (1);
	}

	err = import_pools(pools, props, mntopts, flags,
	argc >= 1 ? argv[0] : NULL,
	argc >= 2 ? argv[1] : NULL,
	do_destroyed, pool_specified, do_all, &idata);

	/*
	* If we're using the cachefile and we failed to import, then
	* fallback to scanning the directory for pools that match
	* those in the cachefile.
	*/
	if (err != 0 && cachefile != NULL) {
	(void) printf(gettext("cachefile import failed, retrying\n"));

	/*
	* We use the scan flag to gather the directories that exist
	* in the cachefile. If we need to fallback to searching for
	* the pool config, we will only search devices in these
	* directories.
	*/
	idata.scan = B_TRUE;
	nvlist_free(pools);
	pools = zpool_search_import(g_zfs, &idata, &libzfs_config_ops);

	err = import_pools(pools, props, mntopts, flags,
	argc >= 1 ? argv[0] : NULL,
	argc >= 2 ? argv[1] : NULL,
	do_destroyed, pool_specified, do_all, &idata);
	}

	error:
	nvlist_free(props);
	nvlist_free(pools);
	nvlist_free(policy);
	free(searchdirs);
	free(envdup);

	return (err ? 1 : 0);
	}

	/*
	* zpool sync [-f] [pool] ...
	*
	* -f (undocumented) force uberblock (and config including zpool cache file)
	* update.
	*
	* Sync the specified pool(s).
	* Without arguments "zpool sync" will sync all pools.
	* This command initiates TXG sync(s) and will return after the TXG(s) commit.
	*
	*/
	static int
	zpool_do_sync(int argc, char **argv)
	{
	int ret;
	boolean_t force = B_FALSE;

	/* check options */
	while ((ret = getopt(argc, argv, "f")) != -1) {
	switch (ret) {
	case 'f':
	force = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* if argc == 0 we will execute zpool_sync_one on all pools */
	- ret = for_each_pool(argc, argv, B_FALSE, NULL, B_FALSE, zpool_sync_one,
	- &force);
	+ ret = for_each_pool(argc, argv, B_FALSE, NULL, ZFS_TYPE_POOL,
	+ B_FALSE, zpool_sync_one, &force);

	return (ret);
	}

	typedef struct iostat_cbdata {
	uint64_t cb_flags;
	- int cb_name_flags;
	int cb_namewidth;
	int cb_iteration;
	- char *cb_vdev_names; / Only show these vdevs */
	- unsigned int cb_vdev_names_count;
	boolean_t cb_verbose;
	boolean_t cb_literal;
	boolean_t cb_scripted;
	zpool_list_t *cb_list;
	vdev_cmd_data_list_t *vcdl;
	+ vdev_cbdata_t cb_vdevs;
	} iostat_cbdata_t;

	/* iostat labels */
	typedef struct name_and_columns {
	const char name; / Column name */
	unsigned int columns; /* Center name to this number of columns */
	} name_and_columns_t;

	#define IOSTAT_MAX_LABELS 15 /* Max number of labels on one line */

	static const name_and_columns_t iostat_top_labels[][IOSTAT_MAX_LABELS] =
	{
	[IOS_DEFAULT] = {{"capacity", 2}, {"operations", 2}, {"bandwidth", 2},
	{NULL}},
	[IOS_LATENCY] = {{"total_wait", 2}, {"disk_wait", 2}, {"syncq_wait", 2},
	{"asyncq_wait", 2}, {"scrub", 1}, {"trim", 1}, {"rebuild", 1},
	{NULL}},
	[IOS_QUEUES] = {{"syncq_read", 2}, {"syncq_write", 2},
	{"asyncq_read", 2}, {"asyncq_write", 2}, {"scrubq_read", 2},
	{"trimq_write", 2}, {"rebuildq_write", 2}, {NULL}},
	[IOS_L_HISTO] = {{"total_wait", 2}, {"disk_wait", 2}, {"syncq_wait", 2},
	{"asyncq_wait", 2}, {NULL}},
	[IOS_RQ_HISTO] = {{"sync_read", 2}, {"sync_write", 2},
	{"async_read", 2}, {"async_write", 2}, {"scrub", 2},
	{"trim", 2}, {"rebuild", 2}, {NULL}},
	};

	/* Shorthand - if "columns" field not set, default to 1 column */
	static const name_and_columns_t iostat_bottom_labels[][IOSTAT_MAX_LABELS] =
	{
	[IOS_DEFAULT] = {{"alloc"}, {"free"}, {"read"}, {"write"}, {"read"},
	{"write"}, {NULL}},
	[IOS_LATENCY] = {{"read"}, {"write"}, {"read"}, {"write"}, {"read"},
	{"write"}, {"read"}, {"write"}, {"wait"}, {"wait"}, {"wait"},
	{NULL}},
	[IOS_QUEUES] = {{"pend"}, {"activ"}, {"pend"}, {"activ"}, {"pend"},
	{"activ"}, {"pend"}, {"activ"}, {"pend"}, {"activ"},
	{"pend"}, {"activ"}, {"pend"}, {"activ"}, {NULL}},
	[IOS_L_HISTO] = {{"read"}, {"write"}, {"read"}, {"write"}, {"read"},
	{"write"}, {"read"}, {"write"}, {"scrub"}, {"trim"}, {"rebuild"},
	{NULL}},
	[IOS_RQ_HISTO] = {{"ind"}, {"agg"}, {"ind"}, {"agg"}, {"ind"}, {"agg"},
	{"ind"}, {"agg"}, {"ind"}, {"agg"}, {"ind"}, {"agg"},
	{"ind"}, {"agg"}, {NULL}},
	};

	static const char *histo_to_title[] = {
	[IOS_L_HISTO] = "latency",
	[IOS_RQ_HISTO] = "req_size",
	};

	/*
	* Return the number of labels in a null-terminated name_and_columns_t
	* array.
	*
	*/
	static unsigned int
	label_array_len(const name_and_columns_t *labels)
	{
	int i = 0;

	while (labels[i].name)
	i++;

	return (i);
	}

	/*
	* Return the number of strings in a null-terminated string array.
	* For example:
	*
	* const char foo[] = {"bar", "baz", NULL}
	*
	* returns 2
	*/
	static uint64_t
	str_array_len(const char *array[])
	{
	uint64_t i = 0;
	while (array[i])
	i++;

	return (i);
	}


	/*
	* Return a default column width for default/latency/queue columns. This does
	* not include histograms, which have their columns autosized.
	*/
	static unsigned int
	default_column_width(iostat_cbdata_t *cb, enum iostat_type type)
	{
	unsigned long column_width = 5; /* Normal niceprint */
	static unsigned long widths[] = {
	/*
	* Choose some sane default column sizes for printing the
	* raw numbers.
	*/
	[IOS_DEFAULT] = 15, /* 1PB capacity */
	[IOS_LATENCY] = 10, /* 1B ns = 10sec */
	[IOS_QUEUES] = 6, /* 1M queue entries */
	[IOS_L_HISTO] = 10, /* 1B ns = 10sec */
	[IOS_RQ_HISTO] = 6, /* 1M queue entries */
	};

	if (cb->cb_literal)
	column_width = widths[type];

	return (column_width);
	}

	/*
	* Print the column labels, i.e:
	*
	* capacity operations bandwidth
	* alloc free read write read write ...
	*
	* If force_column_width is set, use it for the column width. If not set, use
	* the default column width.
	*/
	static void
	print_iostat_labels(iostat_cbdata_t *cb, unsigned int force_column_width,
	const name_and_columns_t labels[][IOSTAT_MAX_LABELS])
	{
	int i, idx, s;
	int text_start, rw_column_width, spaces_to_end;
	uint64_t flags = cb->cb_flags;
	uint64_t f;
	unsigned int column_width = force_column_width;

	/* For each bit set in flags */
	for (f = flags; f; f &= ~(1ULL << idx)) {
	idx = lowbit64(f) - 1;
	if (!force_column_width)
	column_width = default_column_width(cb, idx);
	/* Print our top labels centered over "read write" label. */
	for (i = 0; i < label_array_len(labels[idx]); i++) {
	const char *name = labels[idx][i].name;
	/*
	* We treat labels[][].columns == 0 as shorthand
	* for one column. It makes writing out the label
	* tables more concise.
	*/
	unsigned int columns = MAX(1, labels[idx][i].columns);
	unsigned int slen = strlen(name);

	rw_column_width = (column_width * columns) +
	(2 * (columns - 1));

	text_start = (int)((rw_column_width) / columns -
	slen / columns);
	if (text_start < 0)
	text_start = 0;

	printf(" "); /* Two spaces between columns */

	/* Space from beginning of column to label */
	for (s = 0; s < text_start; s++)
	printf(" ");

	printf("%s", name);

	/* Print space after label to end of column */
	spaces_to_end = rw_column_width - text_start - slen;
	if (spaces_to_end < 0)
	spaces_to_end = 0;

	for (s = 0; s < spaces_to_end; s++)
	printf(" ");
	}
	}
	}


	/*
	* print_cmd_columns - Print custom column titles from -c
	*
	* If the user specified the "zpool status\|iostat -c" then print their custom
	* column titles in the header. For example, print_cmd_columns() would print
	* the " col1 col2" part of this:
	*
	* $ zpool iostat -vc 'echo col1=val1; echo col2=val2'
	* ...
	* capacity operations bandwidth
	* pool alloc free read write read write col1 col2
	* ---------- ----- ----- ----- ----- ----- ----- ---- ----
	* mypool 269K 1008M 0 0 107 946
	* mirror 269K 1008M 0 0 107 946
	* sdb - - 0 0 102 473 val1 val2
	* sdc - - 0 0 5 473 val1 val2
	* ---------- ----- ----- ----- ----- ----- ----- ---- ----
	*/
	static void
	print_cmd_columns(vdev_cmd_data_list_t *vcdl, int use_dashes)
	{
	int i, j;
	vdev_cmd_data_t *data = &vcdl->data[0];

	if (vcdl->count == 0 \|\| data == NULL)
	return;

	/*
	* Each vdev cmd should have the same column names unless the user did
	* something weird with their cmd. Just take the column names from the
	* first vdev and assume it works for all of them.
	*/
	for (i = 0; i < vcdl->uniq_cols_cnt; i++) {
	printf(" ");
	if (use_dashes) {
	for (j = 0; j < vcdl->uniq_cols_width[i]; j++)
	printf("-");
	} else {
	printf_color(ANSI_BOLD, "%*s", vcdl->uniq_cols_width[i],
	vcdl->uniq_cols[i]);
	}
	}
	}


	/*
	* Utility function to print out a line of dashes like:
	*
	* -------------------------------- ----- ----- ----- ----- -----
	*
	* ...or a dashed named-row line like:
	*
	* logs - - - - -
	*
	* @cb: iostat data
	*
	* @force_column_width If non-zero, use the value as the column width.
	* Otherwise use the default column widths.
	*
	* @name: Print a dashed named-row line starting
	* with @name. Otherwise, print a regular
	* dashed line.
	*/
	static void
	print_iostat_dashes(iostat_cbdata_t *cb, unsigned int force_column_width,
	const char *name)
	{
	int i;
	unsigned int namewidth;
	uint64_t flags = cb->cb_flags;
	uint64_t f;
	int idx;
	const name_and_columns_t *labels;
	const char *title;


	if (cb->cb_flags & IOS_ANYHISTO_M) {
	title = histo_to_title[IOS_HISTO_IDX(cb->cb_flags)];
	- } else if (cb->cb_vdev_names_count) {
	+ } else if (cb->cb_vdevs.cb_names_count) {
	title = "vdev";
	} else {
	title = "pool";
	}

	namewidth = MAX(MAX(strlen(title), cb->cb_namewidth),
	name ? strlen(name) : 0);


	if (name) {
	printf("%-*s", namewidth, name);
	} else {
	for (i = 0; i < namewidth; i++)
	(void) printf("-");
	}

	/* For each bit in flags */
	for (f = flags; f; f &= ~(1ULL << idx)) {
	unsigned int column_width;
	idx = lowbit64(f) - 1;
	if (force_column_width)
	column_width = force_column_width;
	else
	column_width = default_column_width(cb, idx);

	labels = iostat_bottom_labels[idx];
	for (i = 0; i < label_array_len(labels); i++) {
	if (name)
	printf(" %*s-", column_width - 1, " ");
	else
	printf(" %.*s", column_width,
	"--------------------");
	}
	}
	}


	static void
	print_iostat_separator_impl(iostat_cbdata_t *cb,
	unsigned int force_column_width)
	{
	print_iostat_dashes(cb, force_column_width, NULL);
	}

	static void
	print_iostat_separator(iostat_cbdata_t *cb)
	{
	print_iostat_separator_impl(cb, 0);
	}

	static void
	print_iostat_header_impl(iostat_cbdata_t *cb, unsigned int force_column_width,
	const char *histo_vdev_name)
	{
	unsigned int namewidth;
	const char *title;

	if (cb->cb_flags & IOS_ANYHISTO_M) {
	title = histo_to_title[IOS_HISTO_IDX(cb->cb_flags)];
	- } else if (cb->cb_vdev_names_count) {
	+ } else if (cb->cb_vdevs.cb_names_count) {
	title = "vdev";
	} else {
	title = "pool";
	}

	namewidth = MAX(MAX(strlen(title), cb->cb_namewidth),
	histo_vdev_name ? strlen(histo_vdev_name) : 0);

	if (histo_vdev_name)
	printf("%-*s", namewidth, histo_vdev_name);
	else
	printf("%*s", namewidth, "");


	print_iostat_labels(cb, force_column_width, iostat_top_labels);
	printf("\n");

	printf("%-*s", namewidth, title);

	print_iostat_labels(cb, force_column_width, iostat_bottom_labels);
	if (cb->vcdl != NULL)
	print_cmd_columns(cb->vcdl, 0);

	printf("\n");

	print_iostat_separator_impl(cb, force_column_width);

	if (cb->vcdl != NULL)
	print_cmd_columns(cb->vcdl, 1);

	printf("\n");
	}

	static void
	print_iostat_header(iostat_cbdata_t *cb)
	{
	print_iostat_header_impl(cb, 0, NULL);
	}


	/*
	* Display a single statistic.
	*/
	static void
	print_one_stat(uint64_t value, enum zfs_nicenum_format format,
	unsigned int column_size, boolean_t scripted)
	{
	char buf[64];

	zfs_nicenum_format(value, buf, sizeof (buf), format);

	if (scripted)
	printf("\t%s", buf);
	else
	printf(" %*s", column_size, buf);
	}

	/*
	* Calculate the default vdev stats
	*
	* Subtract oldvs from newvs, apply a scaling factor, and save the resulting
	* stats into calcvs.
	*/
	static void
	calc_default_iostats(vdev_stat_t oldvs, vdev_stat_t newvs,
	vdev_stat_t *calcvs)
	{
	int i;

	memcpy(calcvs, newvs, sizeof (*calcvs));
	for (i = 0; i < ARRAY_SIZE(calcvs->vs_ops); i++)
	calcvs->vs_ops[i] = (newvs->vs_ops[i] - oldvs->vs_ops[i]);

	for (i = 0; i < ARRAY_SIZE(calcvs->vs_bytes); i++)
	calcvs->vs_bytes[i] = (newvs->vs_bytes[i] - oldvs->vs_bytes[i]);
	}

	/*
	* Internal representation of the extended iostats data.
	*
	* The extended iostat stats are exported in nvlists as either uint64_t arrays
	* or single uint64_t's. We make both look like arrays to make them easier
	* to process. In order to make single uint64_t's look like arrays, we set
	* __data to the stat data, and then set *data = &__data with count = 1. Then,
	* we can just use *data and count.
	*/
	struct stat_array {
	uint64_t *data;
	uint_t count; /* Number of entries in data[] */
	uint64_t __data; /* Only used when data is a single uint64_t */
	};

	static uint64_t
	stat_histo_max(struct stat_array *nva, unsigned int len)
	{
	uint64_t max = 0;
	int i;
	for (i = 0; i < len; i++)
	max = MAX(max, array64_max(nva[i].data, nva[i].count));

	return (max);
	}

	/*
	* Helper function to lookup a uint64_t array or uint64_t value and store its
	* data as a stat_array. If the nvpair is a single uint64_t value, then we make
	* it look like a one element array to make it easier to process.
	*/
	static int
	nvpair64_to_stat_array(nvlist_t nvl, const char name,
	struct stat_array *nva)
	{
	nvpair_t *tmp;
	int ret;

	verify(nvlist_lookup_nvpair(nvl, name, &tmp) == 0);
	switch (nvpair_type(tmp)) {
	case DATA_TYPE_UINT64_ARRAY:
	ret = nvpair_value_uint64_array(tmp, &nva->data, &nva->count);
	break;
	case DATA_TYPE_UINT64:
	ret = nvpair_value_uint64(tmp, &nva->__data);
	nva->data = &nva->__data;
	nva->count = 1;
	break;
	default:
	/* Not a uint64_t */
	ret = EINVAL;
	break;
	}

	return (ret);
	}

	/*
	* Given a list of nvlist names, look up the extended stats in newnv and oldnv,
	* subtract them, and return the results in a newly allocated stat_array.
	* You must free the returned array after you are done with it with
	* free_calc_stats().
	*
	* Additionally, you can set "oldnv" to NULL if you simply want the newnv
	* values.
	*/
	static struct stat_array *
	calc_and_alloc_stats_ex(const char *names, unsigned int len, nvlist_t oldnv,
	nvlist_t *newnv)
	{
	nvlist_t oldnvx = NULL, newnvx;
	struct stat_array oldnva, newnva, *calcnva;
	int i, j;
	unsigned int alloc_size = (sizeof (struct stat_array)) * len;

	/* Extract our extended stats nvlist from the main list */
	verify(nvlist_lookup_nvlist(newnv, ZPOOL_CONFIG_VDEV_STATS_EX,
	&newnvx) == 0);
	if (oldnv) {
	verify(nvlist_lookup_nvlist(oldnv, ZPOOL_CONFIG_VDEV_STATS_EX,
	&oldnvx) == 0);
	}

	newnva = safe_malloc(alloc_size);
	oldnva = safe_malloc(alloc_size);
	calcnva = safe_malloc(alloc_size);

	for (j = 0; j < len; j++) {
	verify(nvpair64_to_stat_array(newnvx, names[j],
	&newnva[j]) == 0);
	calcnva[j].count = newnva[j].count;
	alloc_size = calcnva[j].count * sizeof (calcnva[j].data[0]);
	calcnva[j].data = safe_malloc(alloc_size);
	memcpy(calcnva[j].data, newnva[j].data, alloc_size);

	if (oldnvx) {
	verify(nvpair64_to_stat_array(oldnvx, names[j],
	&oldnva[j]) == 0);
	for (i = 0; i < oldnva[j].count; i++)
	calcnva[j].data[i] -= oldnva[j].data[i];
	}
	}
	free(newnva);
	free(oldnva);
	return (calcnva);
	}

	static void
	free_calc_stats(struct stat_array *nva, unsigned int len)
	{
	int i;
	for (i = 0; i < len; i++)
	free(nva[i].data);

	free(nva);
	}

	static void
	print_iostat_histo(struct stat_array *nva, unsigned int len,
	iostat_cbdata_t *cb, unsigned int column_width, unsigned int namewidth,
	double scale)
	{
	int i, j;
	char buf[6];
	uint64_t val;
	enum zfs_nicenum_format format;
	unsigned int buckets;
	unsigned int start_bucket;

	if (cb->cb_literal)
	format = ZFS_NICENUM_RAW;
	else
	format = ZFS_NICENUM_1024;

	/* All these histos are the same size, so just use nva[0].count */
	buckets = nva[0].count;

	if (cb->cb_flags & IOS_RQ_HISTO_M) {
	/* Start at 512 - req size should never be lower than this */
	start_bucket = 9;
	} else {
	start_bucket = 0;
	}

	for (j = start_bucket; j < buckets; j++) {
	/* Print histogram bucket label */
	if (cb->cb_flags & IOS_L_HISTO_M) {
	/* Ending range of this bucket */
	val = (1UL << (j + 1)) - 1;
	zfs_nicetime(val, buf, sizeof (buf));
	} else {
	/* Request size (starting range of bucket) */
	val = (1UL << j);
	zfs_nicenum(val, buf, sizeof (buf));
	}

	if (cb->cb_scripted)
	printf("%llu", (u_longlong_t)val);
	else
	printf("%-*s", namewidth, buf);

	/* Print the values on the line */
	for (i = 0; i < len; i++) {
	print_one_stat(nva[i].data[j] * scale, format,
	column_width, cb->cb_scripted);
	}
	printf("\n");
	}
	}

	static void
	print_solid_separator(unsigned int length)
	{
	while (length--)
	printf("-");
	printf("\n");
	}

	static void
	print_iostat_histos(iostat_cbdata_t cb, nvlist_t oldnv,
	nvlist_t newnv, double scale, const char name)
	{
	unsigned int column_width;
	unsigned int namewidth;
	unsigned int entire_width;
	enum iostat_type type;
	struct stat_array *nva;
	const char **names;
	unsigned int names_len;

	/* What type of histo are we? */
	type = IOS_HISTO_IDX(cb->cb_flags);

	/* Get NULL-terminated array of nvlist names for our histo */
	names = vsx_type_to_nvlist[type];
	names_len = str_array_len(names); /* num of names */

	nva = calc_and_alloc_stats_ex(names, names_len, oldnv, newnv);

	if (cb->cb_literal) {
	column_width = MAX(5,
	(unsigned int) log10(stat_histo_max(nva, names_len)) + 1);
	} else {
	column_width = 5;
	}

	namewidth = MAX(cb->cb_namewidth,
	strlen(histo_to_title[IOS_HISTO_IDX(cb->cb_flags)]));

	/*
	* Calculate the entire line width of what we're printing. The
	* +2 is for the two spaces between columns:
	*/
	/* read write */
	/* ----- ----- */
	/* \|___\| <---------- column_width */
	/* */
	/* \|__________\| <--- entire_width */
	/* */
	entire_width = namewidth + (column_width + 2) *
	label_array_len(iostat_bottom_labels[type]);

	if (cb->cb_scripted)
	printf("%s\n", name);
	else
	print_iostat_header_impl(cb, column_width, name);

	print_iostat_histo(nva, names_len, cb, column_width,
	namewidth, scale);

	free_calc_stats(nva, names_len);
	if (!cb->cb_scripted)
	print_solid_separator(entire_width);
	}

	/*
	* Calculate the average latency of a power-of-two latency histogram
	*/
	static uint64_t
	single_histo_average(uint64_t *histo, unsigned int buckets)
	{
	int i;
	uint64_t count = 0, total = 0;

	for (i = 0; i < buckets; i++) {
	/*
	* Our buckets are power-of-two latency ranges. Use the
	* midpoint latency of each bucket to calculate the average.
	* For example:
	*
	* Bucket Midpoint
	* 8ns-15ns: 12ns
	* 16ns-31ns: 24ns
	* ...
	*/
	if (histo[i] != 0) {
	total += histo[i] * (((1UL << i) + ((1UL << i)/2)));
	count += histo[i];
	}
	}

	/* Prevent divide by zero */
	return (count == 0 ? 0 : total / count);
	}

	static void
	print_iostat_queues(iostat_cbdata_t cb, nvlist_t oldnv,
	nvlist_t *newnv)
	{
	int i;
	uint64_t val;
	const char *names[] = {
	ZPOOL_CONFIG_VDEV_SYNC_R_PEND_QUEUE,
	ZPOOL_CONFIG_VDEV_SYNC_R_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_SYNC_W_PEND_QUEUE,
	ZPOOL_CONFIG_VDEV_SYNC_W_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_ASYNC_R_PEND_QUEUE,
	ZPOOL_CONFIG_VDEV_ASYNC_R_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_ASYNC_W_PEND_QUEUE,
	ZPOOL_CONFIG_VDEV_ASYNC_W_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_SCRUB_PEND_QUEUE,
	ZPOOL_CONFIG_VDEV_SCRUB_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_TRIM_PEND_QUEUE,
	ZPOOL_CONFIG_VDEV_TRIM_ACTIVE_QUEUE,
	ZPOOL_CONFIG_VDEV_REBUILD_PEND_QUEUE,
	ZPOOL_CONFIG_VDEV_REBUILD_ACTIVE_QUEUE,
	};

	struct stat_array *nva;

	unsigned int column_width = default_column_width(cb, IOS_QUEUES);
	enum zfs_nicenum_format format;

	nva = calc_and_alloc_stats_ex(names, ARRAY_SIZE(names), NULL, newnv);

	if (cb->cb_literal)
	format = ZFS_NICENUM_RAW;
	else
	format = ZFS_NICENUM_1024;

	for (i = 0; i < ARRAY_SIZE(names); i++) {
	val = nva[i].data[0];
	print_one_stat(val, format, column_width, cb->cb_scripted);
	}

	free_calc_stats(nva, ARRAY_SIZE(names));
	}

	static void
	print_iostat_latency(iostat_cbdata_t cb, nvlist_t oldnv,
	nvlist_t *newnv)
	{
	int i;
	uint64_t val;
	const char *names[] = {
	ZPOOL_CONFIG_VDEV_TOT_R_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_TOT_W_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_DISK_R_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_DISK_W_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_SYNC_R_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_SYNC_W_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_ASYNC_R_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_ASYNC_W_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_SCRUB_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_TRIM_LAT_HISTO,
	ZPOOL_CONFIG_VDEV_REBUILD_LAT_HISTO,
	};
	struct stat_array *nva;

	unsigned int column_width = default_column_width(cb, IOS_LATENCY);
	enum zfs_nicenum_format format;

	nva = calc_and_alloc_stats_ex(names, ARRAY_SIZE(names), oldnv, newnv);

	if (cb->cb_literal)
	format = ZFS_NICENUM_RAWTIME;
	else
	format = ZFS_NICENUM_TIME;

	/* Print our avg latencies on the line */
	for (i = 0; i < ARRAY_SIZE(names); i++) {
	/* Compute average latency for a latency histo */
	val = single_histo_average(nva[i].data, nva[i].count);
	print_one_stat(val, format, column_width, cb->cb_scripted);
	}
	free_calc_stats(nva, ARRAY_SIZE(names));
	}

	/*
	* Print default statistics (capacity/operations/bandwidth)
	*/
	static void
	print_iostat_default(vdev_stat_t vs, iostat_cbdata_t cb, double scale)
	{
	unsigned int column_width = default_column_width(cb, IOS_DEFAULT);
	enum zfs_nicenum_format format;
	char na; /* char to print for "not applicable" values */

	if (cb->cb_literal) {
	format = ZFS_NICENUM_RAW;
	na = '0';
	} else {
	format = ZFS_NICENUM_1024;
	na = '-';
	}

	/* only toplevel vdevs have capacity stats */
	if (vs->vs_space == 0) {
	if (cb->cb_scripted)
	printf("\t%c\t%c", na, na);
	else
	printf(" %c %c", column_width, na, column_width,
	na);
	} else {
	print_one_stat(vs->vs_alloc, format, column_width,
	cb->cb_scripted);
	print_one_stat(vs->vs_space - vs->vs_alloc, format,
	column_width, cb->cb_scripted);
	}

	print_one_stat((uint64_t)(vs->vs_ops[ZIO_TYPE_READ] * scale),
	format, column_width, cb->cb_scripted);
	print_one_stat((uint64_t)(vs->vs_ops[ZIO_TYPE_WRITE] * scale),
	format, column_width, cb->cb_scripted);
	print_one_stat((uint64_t)(vs->vs_bytes[ZIO_TYPE_READ] * scale),
	format, column_width, cb->cb_scripted);
	print_one_stat((uint64_t)(vs->vs_bytes[ZIO_TYPE_WRITE] * scale),
	format, column_width, cb->cb_scripted);
	}

	static const char *class_name[] = {
	VDEV_ALLOC_BIAS_DEDUP,
	VDEV_ALLOC_BIAS_SPECIAL,
	VDEV_ALLOC_CLASS_LOGS
	};

	/*
	* Print out all the statistics for the given vdev. This can either be the
	* toplevel configuration, or called recursively. If 'name' is NULL, then this
	* is a verbose output, and we don't want to display the toplevel pool stats.
	*
	* Returns the number of stat lines printed.
	*/
	static unsigned int
	print_vdev_stats(zpool_handle_t zhp, const char name, nvlist_t *oldnv,
	nvlist_t newnv, iostat_cbdata_t cb, int depth)
	{
	nvlist_t oldchild, newchild;
	uint_t c, children, oldchildren;
	vdev_stat_t oldvs, newvs, *calcvs;
	vdev_stat_t zerovs = { 0 };
	char *vname;
	int i;
	int ret = 0;
	uint64_t tdelta;
	double scale;

	if (strcmp(name, VDEV_TYPE_INDIRECT) == 0)
	return (ret);

	calcvs = safe_malloc(sizeof (*calcvs));

	if (oldnv != NULL) {
	verify(nvlist_lookup_uint64_array(oldnv,
	ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&oldvs, &c) == 0);
	} else {
	oldvs = &zerovs;
	}

	/* Do we only want to see a specific vdev? */
	- for (i = 0; i < cb->cb_vdev_names_count; i++) {
	+ for (i = 0; i < cb->cb_vdevs.cb_names_count; i++) {
	/* Yes we do. Is this the vdev? */
	- if (strcmp(name, cb->cb_vdev_names[i]) == 0) {
	+ if (strcmp(name, cb->cb_vdevs.cb_names[i]) == 0) {
	/*
	* This is our vdev. Since it is the only vdev we
	* will be displaying, make depth = 0 so that it
	* doesn't get indented.
	*/
	depth = 0;
	break;
	}
	}

	- if (cb->cb_vdev_names_count && (i == cb->cb_vdev_names_count)) {
	+ if (cb->cb_vdevs.cb_names_count && (i == cb->cb_vdevs.cb_names_count)) {
	/* Couldn't match the name */
	goto children;
	}


	verify(nvlist_lookup_uint64_array(newnv, ZPOOL_CONFIG_VDEV_STATS,
	(uint64_t **)&newvs, &c) == 0);

	/*
	* Print the vdev name unless it's is a histogram. Histograms
	* display the vdev name in the header itself.
	*/
	if (!(cb->cb_flags & IOS_ANYHISTO_M)) {
	if (cb->cb_scripted) {
	printf("%s", name);
	} else {
	if (strlen(name) + depth > cb->cb_namewidth)
	(void) printf("%*s%s", depth, "", name);
	else
	(void) printf("%s%s%s", depth, "", name,
	(int)(cb->cb_namewidth - strlen(name) -
	depth), "");
	}
	}

	/* Calculate our scaling factor */
	tdelta = newvs->vs_timestamp - oldvs->vs_timestamp;
	if ((oldvs->vs_timestamp == 0) && (cb->cb_flags & IOS_ANYHISTO_M)) {
	/*
	* If we specify printing histograms with no time interval, then
	* print the histogram numbers over the entire lifetime of the
	* vdev.
	*/
	scale = 1;
	} else {
	if (tdelta == 0)
	scale = 1.0;
	else
	scale = (double)NANOSEC / tdelta;
	}

	if (cb->cb_flags & IOS_DEFAULT_M) {
	calc_default_iostats(oldvs, newvs, calcvs);
	print_iostat_default(calcvs, cb, scale);
	}
	if (cb->cb_flags & IOS_LATENCY_M)
	print_iostat_latency(cb, oldnv, newnv);
	if (cb->cb_flags & IOS_QUEUES_M)
	print_iostat_queues(cb, oldnv, newnv);
	if (cb->cb_flags & IOS_ANYHISTO_M) {
	printf("\n");
	print_iostat_histos(cb, oldnv, newnv, scale, name);
	}

	if (cb->vcdl != NULL) {
	char *path;
	if (nvlist_lookup_string(newnv, ZPOOL_CONFIG_PATH,
	&path) == 0) {
	printf(" ");
	zpool_print_cmd(cb->vcdl, zpool_get_name(zhp), path);
	}
	}

	if (!(cb->cb_flags & IOS_ANYHISTO_M))
	printf("\n");

	ret++;

	children:

	free(calcvs);

	if (!cb->cb_verbose)
	return (ret);

	if (nvlist_lookup_nvlist_array(newnv, ZPOOL_CONFIG_CHILDREN,
	&newchild, &children) != 0)
	return (ret);

	if (oldnv) {
	if (nvlist_lookup_nvlist_array(oldnv, ZPOOL_CONFIG_CHILDREN,
	&oldchild, &oldchildren) != 0)
	return (ret);

	children = MIN(oldchildren, children);
	}

	/*
	* print normal top-level devices
	*/
	for (c = 0; c < children; c++) {
	uint64_t ishole = B_FALSE, islog = B_FALSE;

	(void) nvlist_lookup_uint64(newchild[c], ZPOOL_CONFIG_IS_HOLE,
	&ishole);

	(void) nvlist_lookup_uint64(newchild[c], ZPOOL_CONFIG_IS_LOG,
	&islog);

	if (ishole \|\| islog)
	continue;

	if (nvlist_exists(newchild[c], ZPOOL_CONFIG_ALLOCATION_BIAS))
	continue;

	vname = zpool_vdev_name(g_zfs, zhp, newchild[c],
	- cb->cb_name_flags);
	+ cb->cb_vdevs.cb_name_flags);
	ret += print_vdev_stats(zhp, vname, oldnv ? oldchild[c] : NULL,
	newchild[c], cb, depth + 2);
	free(vname);
	}

	/*
	* print all other top-level devices
	*/
	for (uint_t n = 0; n < 3; n++) {
	boolean_t printed = B_FALSE;

	for (c = 0; c < children; c++) {
	uint64_t islog = B_FALSE;
	char *bias = NULL;
	char *type = NULL;

	(void) nvlist_lookup_uint64(newchild[c],
	ZPOOL_CONFIG_IS_LOG, &islog);
	if (islog) {
	bias = VDEV_ALLOC_CLASS_LOGS;
	} else {
	(void) nvlist_lookup_string(newchild[c],
	ZPOOL_CONFIG_ALLOCATION_BIAS, &bias);
	(void) nvlist_lookup_string(newchild[c],
	ZPOOL_CONFIG_TYPE, &type);
	}
	if (bias == NULL \|\| strcmp(bias, class_name[n]) != 0)
	continue;
	if (!islog && strcmp(type, VDEV_TYPE_INDIRECT) == 0)
	continue;

	if (!printed) {
	if ((!(cb->cb_flags & IOS_ANYHISTO_M)) &&
	- !cb->cb_scripted && !cb->cb_vdev_names) {
	+ !cb->cb_scripted &&
	+ !cb->cb_vdevs.cb_names) {
	print_iostat_dashes(cb, 0,
	class_name[n]);
	}
	printf("\n");
	printed = B_TRUE;
	}

	vname = zpool_vdev_name(g_zfs, zhp, newchild[c],
	- cb->cb_name_flags);
	+ cb->cb_vdevs.cb_name_flags);
	ret += print_vdev_stats(zhp, vname, oldnv ?
	oldchild[c] : NULL, newchild[c], cb, depth + 2);
	free(vname);
	}
	}

	/*
	* Include level 2 ARC devices in iostat output
	*/
	if (nvlist_lookup_nvlist_array(newnv, ZPOOL_CONFIG_L2CACHE,
	&newchild, &children) != 0)
	return (ret);

	if (oldnv) {
	if (nvlist_lookup_nvlist_array(oldnv, ZPOOL_CONFIG_L2CACHE,
	&oldchild, &oldchildren) != 0)
	return (ret);

	children = MIN(oldchildren, children);
	}

	if (children > 0) {
	if ((!(cb->cb_flags & IOS_ANYHISTO_M)) && !cb->cb_scripted &&
	- !cb->cb_vdev_names) {
	+ !cb->cb_vdevs.cb_names) {
	print_iostat_dashes(cb, 0, "cache");
	}
	printf("\n");

	for (c = 0; c < children; c++) {
	vname = zpool_vdev_name(g_zfs, zhp, newchild[c],
	- cb->cb_name_flags);
	+ cb->cb_vdevs.cb_name_flags);
	ret += print_vdev_stats(zhp, vname, oldnv ? oldchild[c]
	: NULL, newchild[c], cb, depth + 2);
	free(vname);
	}
	}

	return (ret);
	}

	static int
	refresh_iostat(zpool_handle_t zhp, void data)
	{
	iostat_cbdata_t *cb = data;
	boolean_t missing;

	/*
	* If the pool has disappeared, remove it from the list and continue.
	*/
	if (zpool_refresh_stats(zhp, &missing) != 0)
	return (-1);

	if (missing)
	pool_list_remove(cb->cb_list, zhp);

	return (0);
	}

	/*
	* Callback to print out the iostats for the given pool.
	*/
	static int
	print_iostat(zpool_handle_t zhp, void data)
	{
	iostat_cbdata_t *cb = data;
	nvlist_t oldconfig, newconfig;
	nvlist_t oldnvroot, newnvroot;
	int ret;

	newconfig = zpool_get_config(zhp, &oldconfig);

	if (cb->cb_iteration == 1)
	oldconfig = NULL;

	verify(nvlist_lookup_nvlist(newconfig, ZPOOL_CONFIG_VDEV_TREE,
	&newnvroot) == 0);

	if (oldconfig == NULL)
	oldnvroot = NULL;
	else
	verify(nvlist_lookup_nvlist(oldconfig, ZPOOL_CONFIG_VDEV_TREE,
	&oldnvroot) == 0);

	ret = print_vdev_stats(zhp, zpool_get_name(zhp), oldnvroot, newnvroot,
	cb, 0);
	if ((ret != 0) && !(cb->cb_flags & IOS_ANYHISTO_M) &&
	- !cb->cb_scripted && cb->cb_verbose && !cb->cb_vdev_names_count) {
	+ !cb->cb_scripted && cb->cb_verbose &&
	+ !cb->cb_vdevs.cb_names_count) {
	print_iostat_separator(cb);
	if (cb->vcdl != NULL) {
	print_cmd_columns(cb->vcdl, 1);
	}
	printf("\n");
	}

	return (ret);
	}

	static int
	get_columns(void)
	{
	struct winsize ws;
	int columns = 80;
	int error;

	if (isatty(STDOUT_FILENO)) {
	error = ioctl(STDOUT_FILENO, TIOCGWINSZ, &ws);
	if (error == 0)
	columns = ws.ws_col;
	} else {
	columns = 999;
	}

	return (columns);
	}

	/*
	* Return the required length of the pool/vdev name column. The minimum
	* allowed width and output formatting flags must be provided.
	*/
	static int
	get_namewidth(zpool_handle_t *zhp, int min_width, int flags, boolean_t verbose)
	{
	nvlist_t config, nvroot;
	int width = min_width;

	if ((config = zpool_get_config(zhp, NULL)) != NULL) {
	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) == 0);
	unsigned int poolname_len = strlen(zpool_get_name(zhp));
	if (verbose == B_FALSE) {
	width = MAX(poolname_len, min_width);
	} else {
	width = MAX(poolname_len,
	max_width(zhp, nvroot, 0, min_width, flags));
	}
	}

	return (width);
	}

	/*
	* Parse the input string, get the 'interval' and 'count' value if there is one.
	*/
	static void
	get_interval_count(int argcp, char argv, float iv,
	unsigned long *cnt)
	{
	float interval = 0;
	unsigned long count = 0;
	int argc = *argcp;

	/*
	* Determine if the last argument is an integer or a pool name
	*/
	if (argc > 0 && zfs_isnumber(argv[argc - 1])) {
	char *end;

	errno = 0;
	interval = strtof(argv[argc - 1], &end);

	if (*end == '\0' && errno == 0) {
	if (interval == 0) {
	(void) fprintf(stderr, gettext(
	"interval cannot be zero\n"));
	usage(B_FALSE);
	}
	/*
	* Ignore the last parameter
	*/
	argc--;
	} else {
	/*
	* If this is not a valid number, just plow on. The
	* user will get a more informative error message later
	* on.
	*/
	interval = 0;
	}
	}

	/*
	* If the last argument is also an integer, then we have both a count
	* and an interval.
	*/
	if (argc > 0 && zfs_isnumber(argv[argc - 1])) {
	char *end;

	errno = 0;
	count = interval;
	interval = strtof(argv[argc - 1], &end);

	if (*end == '\0' && errno == 0) {
	if (interval == 0) {
	(void) fprintf(stderr, gettext(
	"interval cannot be zero\n"));
	usage(B_FALSE);
	}

	/*
	* Ignore the last parameter
	*/
	argc--;
	} else {
	interval = 0;
	}
	}

	*iv = interval;
	*cnt = count;
	*argcp = argc;
	}

	static void
	get_timestamp_arg(char c)
	{
	if (c == 'u')
	timestamp_fmt = UDATE;
	else if (c == 'd')
	timestamp_fmt = DDATE;
	else
	usage(B_FALSE);
	}

	/*
	* Return stat flags that are supported by all pools by both the module and
	* zpool iostat. "*data" should be initialized to all 0xFFs before running.
	* It will get ANDed down until only the flags that are supported on all pools
	* remain.
	*/
	static int
	get_stat_flags_cb(zpool_handle_t zhp, void data)
	{
	uint64_t *mask = data;
	nvlist_t config, nvroot, *nvx;
	uint64_t flags = 0;
	int i, j;

	config = zpool_get_config(zhp, NULL);
	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) == 0);

	/* Default stats are always supported, but for completeness.. */
	if (nvlist_exists(nvroot, ZPOOL_CONFIG_VDEV_STATS))
	flags \|= IOS_DEFAULT_M;

	/* Get our extended stats nvlist from the main list */
	if (nvlist_lookup_nvlist(nvroot, ZPOOL_CONFIG_VDEV_STATS_EX,
	&nvx) != 0) {
	/*
	* No extended stats; they're probably running an older
	* module. No big deal, we support that too.
	*/
	goto end;
	}

	/* For each extended stat, make sure all its nvpairs are supported */
	for (j = 0; j < ARRAY_SIZE(vsx_type_to_nvlist); j++) {
	if (!vsx_type_to_nvlist[j][0])
	continue;

	/* Start off by assuming the flag is supported, then check */
	flags \|= (1ULL << j);
	for (i = 0; vsx_type_to_nvlist[j][i]; i++) {
	if (!nvlist_exists(nvx, vsx_type_to_nvlist[j][i])) {
	/* flag isn't supported */
	flags = flags & ~(1ULL << j);
	break;
	}
	}
	}
	end:
	mask = mask & flags;
	return (0);
	}

	/*
	* Return a bitmask of stats that are supported on all pools by both the module
	* and zpool iostat.
	*/
	static uint64_t
	get_stat_flags(zpool_list_t *list)
	{
	uint64_t mask = -1;

	/*
	* get_stat_flags_cb() will lop off bits from "mask" until only the
	* flags that are supported on all pools remain.
	*/
	pool_list_iter(list, B_FALSE, get_stat_flags_cb, &mask);
	return (mask);
	}

	/*
	- * Return 1 if cb_data->cb_vdev_names[0] is this vdev's name, 0 otherwise.
	+ * Return 1 if cb_data->cb_names[0] is this vdev's name, 0 otherwise.
	*/
	static int
	is_vdev_cb(void zhp_data, nvlist_t nv, void *cb_data)
	{
	- iostat_cbdata_t *cb = cb_data;
	+ vdev_cbdata_t *cb = cb_data;
	char *name = NULL;
	- int ret = 0;
	+ int ret = 1; /* assume match */
	zpool_handle_t *zhp = zhp_data;

	name = zpool_vdev_name(g_zfs, zhp, nv, cb->cb_name_flags);

	- if (strcmp(name, cb->cb_vdev_names[0]) == 0)
	- ret = 1; /* match */
	+ if (strcmp(name, cb->cb_names[0])) {
	+ free(name);
	+ name = zpool_vdev_name(g_zfs, zhp, nv, VDEV_NAME_GUID);
	+ ret = (strcmp(name, cb->cb_names[0]) == 0);
	+ }
	free(name);

	return (ret);
	}

	/*
	- * Returns 1 if cb_data->cb_vdev_names[0] is a vdev name, 0 otherwise.
	+ * Returns 1 if cb_data->cb_names[0] is a vdev name, 0 otherwise.
	*/
	static int
	is_vdev(zpool_handle_t zhp, void cb_data)
	{
	return (for_each_vdev(zhp, is_vdev_cb, cb_data));
	}

	/*
	* Check if vdevs are in a pool
	*
	* Return 1 if all argv[] strings are vdev names in pool "pool_name". Otherwise
	* return 0. If pool_name is NULL, then search all pools.
	*/
	static int
	are_vdevs_in_pool(int argc, char *argv, char pool_name,
	- iostat_cbdata_t *cb)
	+ vdev_cbdata_t *cb)
	{
	char **tmp_name;
	int ret = 0;
	int i;
	int pool_count = 0;

	if ((argc == 0) \|\| !*argv)
	return (0);

	if (pool_name)
	pool_count = 1;

	- /* Temporarily hijack cb_vdev_names for a second... */
	- tmp_name = cb->cb_vdev_names;
	+ /* Temporarily hijack cb_names for a second... */
	+ tmp_name = cb->cb_names;

	/* Go though our list of prospective vdev names */
	for (i = 0; i < argc; i++) {
	- cb->cb_vdev_names = argv + i;
	+ cb->cb_names = argv + i;

	/* Is this name a vdev in our pools? */
	ret = for_each_pool(pool_count, &pool_name, B_TRUE, NULL,
	- B_FALSE, is_vdev, cb);
	+ ZFS_TYPE_POOL, B_FALSE, is_vdev, cb);
	if (!ret) {
	/* No match */
	break;
	}
	}

	- cb->cb_vdev_names = tmp_name;
	+ cb->cb_names = tmp_name;

	return (ret);
	}

	static int
	is_pool_cb(zpool_handle_t zhp, void data)
	{
	char *name = data;
	if (strcmp(name, zpool_get_name(zhp)) == 0)
	return (1);

	return (0);
	}

	/*
	* Do we have a pool named *name? If so, return 1, otherwise 0.
	*/
	static int
	is_pool(char *name)
	{
	- return (for_each_pool(0, NULL, B_TRUE, NULL, B_FALSE, is_pool_cb,
	- name));
	+ return (for_each_pool(0, NULL, B_TRUE, NULL, ZFS_TYPE_POOL, B_FALSE,
	+ is_pool_cb, name));
	}

	/* Are all our argv[] strings pool names? If so return 1, 0 otherwise. */
	static int
	are_all_pools(int argc, char **argv)
	{
	if ((argc == 0) \|\| !*argv)
	return (0);

	while (--argc >= 0)
	if (!is_pool(argv[argc]))
	return (0);

	return (1);
	}

	/*
	* Helper function to print out vdev/pool names we can't resolve. Used for an
	* error message.
	*/
	static void
	error_list_unresolved_vdevs(int argc, char *argv, char pool_name,
	- iostat_cbdata_t *cb)
	+ vdev_cbdata_t *cb)
	{
	int i;
	char *name;
	char *str;
	for (i = 0; i < argc; i++) {
	name = argv[i];

	if (is_pool(name))
	str = gettext("pool");
	else if (are_vdevs_in_pool(1, &name, pool_name, cb))
	str = gettext("vdev in this pool");
	else if (are_vdevs_in_pool(1, &name, NULL, cb))
	str = gettext("vdev in another pool");
	else
	str = gettext("unknown");

	fprintf(stderr, "\t%s (%s)\n", name, str);
	}
	}

	/*
	* Same as get_interval_count(), but with additional checks to not misinterpret
	* guids as interval/count values. Assumes VDEV_NAME_GUID is set in
	- * cb.cb_name_flags.
	+ * cb.cb_vdevs.cb_name_flags.
	*/
	static void
	get_interval_count_filter_guids(int argc, char argv, float interval,
	unsigned long count, iostat_cbdata_t cb)
	{
	char **tmpargv = argv;
	int argc_for_interval = 0;

	/* Is the last arg an interval value? Or a guid? */
	- if (argc >= 1 && !are_vdevs_in_pool(1, &argv[argc - 1], NULL, cb)) {
	+ if (argc >= 1 && !are_vdevs_in_pool(1, &argv[argc - 1], NULL,
	+ &cb->cb_vdevs)) {
	/*
	* The last arg is not a guid, so it's probably an
	* interval value.
	*/
	argc_for_interval++;

	if (*argc >= 2 &&
	- !are_vdevs_in_pool(1, &argv[*argc - 2], NULL, cb)) {
	+ !are_vdevs_in_pool(1, &argv[*argc - 2], NULL,
	+ &cb->cb_vdevs)) {
	/*
	* The 2nd to last arg is not a guid, so it's probably
	* an interval value.
	*/
	argc_for_interval++;
	}
	}

	/* Point to our list of possible intervals */
	tmpargv = &argv[*argc - argc_for_interval];

	argc = argc - argc_for_interval;
	get_interval_count(&argc_for_interval, tmpargv,
	interval, count);
	}

	/*
	* Floating point sleep(). Allows you to pass in a floating point value for
	* seconds.
	*/
	static void
	fsleep(float sec)
	{
	struct timespec req;
	req.tv_sec = floor(sec);
	req.tv_nsec = (sec - (float)req.tv_sec) * NANOSEC;
	nanosleep(&req, NULL);
	}

	/*
	* Terminal height, in rows. Returns -1 if stdout is not connected to a TTY or
	* if we were unable to determine its size.
	*/
	static int
	terminal_height(void)
	{
	struct winsize win;

	if (isatty(STDOUT_FILENO) == 0)
	return (-1);

	if (ioctl(STDOUT_FILENO, TIOCGWINSZ, &win) != -1 && win.ws_row > 0)
	return (win.ws_row);

	return (-1);
	}

	/*
	* Run one of the zpool status/iostat -c scripts with the help (-h) option and
	* print the result.
	*
	* name: Short name of the script ('iostat').
	* path: Full path to the script ('/usr/local/etc/zfs/zpool.d/iostat');
	*/
	static void
	print_zpool_script_help(char name, char path)
	{
	char *argv[] = {path, "-h", NULL};
	char **lines = NULL;
	int lines_cnt = 0;
	int rc;

	rc = libzfs_run_process_get_stdout_nopath(path, argv, NULL, &lines,
	&lines_cnt);
	if (rc != 0 \|\| lines == NULL \|\| lines_cnt <= 0) {
	if (lines != NULL)
	libzfs_free_str_array(lines, lines_cnt);
	return;
	}

	for (int i = 0; i < lines_cnt; i++)
	if (!is_blank_str(lines[i]))
	printf(" %-14s %s\n", name, lines[i]);

	libzfs_free_str_array(lines, lines_cnt);
	}

	/*
	* Go though the zpool status/iostat -c scripts in the user's path, run their
	* help option (-h), and print out the results.
	*/
	static void
	print_zpool_dir_scripts(char *dirpath)
	{
	DIR *dir;
	struct dirent *ent;
	char fullpath[MAXPATHLEN];
	struct stat dir_stat;

	if ((dir = opendir(dirpath)) != NULL) {
	/* print all the files and directories within directory */
	while ((ent = readdir(dir)) != NULL) {
	sprintf(fullpath, "%s/%s", dirpath, ent->d_name);

	/* Print the scripts */
	if (stat(fullpath, &dir_stat) == 0)
	if (dir_stat.st_mode & S_IXUSR &&
	S_ISREG(dir_stat.st_mode))
	print_zpool_script_help(ent->d_name,
	fullpath);
	}
	closedir(dir);
	}
	}

	/*
	* Print out help text for all zpool status/iostat -c scripts.
	*/
	static void
	print_zpool_script_list(char *subcommand)
	{
	char dir, sp, *tmp;

	printf(gettext("Available 'zpool %s -c' commands:\n"), subcommand);

	sp = zpool_get_cmd_search_path();
	if (sp == NULL)
	return;

	for (dir = strtok_r(sp, ":", &tmp);
	dir != NULL;
	dir = strtok_r(NULL, ":", &tmp))
	print_zpool_dir_scripts(dir);

	free(sp);
	}

	/*
	* Set the minimum pool/vdev name column width. The width must be at least 10,
	* but may be as large as the column width - 42 so it still fits on one line.
	* NOTE: 42 is the width of the default capacity/operations/bandwidth output
	*/
	static int
	get_namewidth_iostat(zpool_handle_t zhp, void data)
	{
	iostat_cbdata_t *cb = data;
	int width, available_width;

	/*
	* get_namewidth() returns the maximum width of any name in that column
	* for any pool/vdev/device line that will be output.
	*/
	- width = get_namewidth(zhp, cb->cb_namewidth, cb->cb_name_flags,
	+ width = get_namewidth(zhp, cb->cb_namewidth, cb->cb_vdevs.cb_name_flags,
	cb->cb_verbose);

	/*
	* The width we are calculating is the width of the header and also the
	* padding width for names that are less than maximum width. The stats
	* take up 42 characters, so the width available for names is:
	*/
	available_width = get_columns() - 42;

	/*
	* If the maximum width fits on a screen, then great! Make everything
	* line up by justifying all lines to the same width. If that max
	* width is larger than what's available, the name plus stats won't fit
	* on one line, and justifying to that width would cause every line to
	* wrap on the screen. We only want lines with long names to wrap.
	* Limit the padding to what won't wrap.
	*/
	if (width > available_width)
	width = available_width;

	/*
	* And regardless of whatever the screen width is (get_columns can
	* return 0 if the width is not known or less than 42 for a narrow
	* terminal) have the width be a minimum of 10.
	*/
	if (width < 10)
	width = 10;

	/* Save the calculated width */
	cb->cb_namewidth = width;

	return (0);
	}

	/*
	* zpool iostat [[-c [script1,script2,...]] [-lq]\|[-rw]] [-ghHLpPvy] [-n name]
	* [-T d\|u] [[ pool ...]\|[pool vdev ...]\|[vdev ...]]
	* [interval [count]]
	*
	* -c CMD For each vdev, run command CMD
	* -g Display guid for individual vdev name.
	* -L Follow links when resolving vdev path name.
	* -P Display full path for vdev name.
	* -v Display statistics for individual vdevs
	* -h Display help
	* -p Display values in parsable (exact) format.
	* -H Scripted mode. Don't display headers, and separate properties
	* by a single tab.
	* -l Display average latency
	* -q Display queue depths
	* -w Display latency histograms
	* -r Display request size histogram
	* -T Display a timestamp in date(1) or Unix format
	* -n Only print headers once
	*
	* This command can be tricky because we want to be able to deal with pool
	* creation/destruction as well as vdev configuration changes. The bulk of this
	* processing is handled by the pool_list_* routines in zpool_iter.c. We rely
	* on pool_list_update() to detect the addition of new pools. Configuration
	* changes are all handled within libzfs.
	*/
	int
	zpool_do_iostat(int argc, char **argv)
	{
	int c;
	int ret;
	int npools;
	float interval = 0;
	unsigned long count = 0;
	int winheight = 24;
	zpool_list_t *list;
	boolean_t verbose = B_FALSE;
	boolean_t latency = B_FALSE, l_histo = B_FALSE, rq_histo = B_FALSE;
	boolean_t queues = B_FALSE, parsable = B_FALSE, scripted = B_FALSE;
	boolean_t omit_since_boot = B_FALSE;
	boolean_t guid = B_FALSE;
	boolean_t follow_links = B_FALSE;
	boolean_t full_name = B_FALSE;
	boolean_t headers_once = B_FALSE;
	iostat_cbdata_t cb = { 0 };
	char *cmd = NULL;

	/* Used for printing error message */
	const char flag_to_arg[] = {[IOS_LATENCY] = 'l', [IOS_QUEUES] = 'q',
	[IOS_L_HISTO] = 'w', [IOS_RQ_HISTO] = 'r'};

	uint64_t unsupported_flags;

	/* check options */
	while ((c = getopt(argc, argv, "c:gLPT:vyhplqrwnH")) != -1) {
	switch (c) {
	case 'c':
	if (cmd != NULL) {
	fprintf(stderr,
	gettext("Can't set -c flag twice\n"));
	exit(1);
	}

	if (getenv("ZPOOL_SCRIPTS_ENABLED") != NULL &&
	!libzfs_envvar_is_set("ZPOOL_SCRIPTS_ENABLED")) {
	fprintf(stderr, gettext(
	"Can't run -c, disabled by "
	"ZPOOL_SCRIPTS_ENABLED.\n"));
	exit(1);
	}

	if ((getuid() <= 0 \|\| geteuid() <= 0) &&
	!libzfs_envvar_is_set("ZPOOL_SCRIPTS_AS_ROOT")) {
	fprintf(stderr, gettext(
	"Can't run -c with root privileges "
	"unless ZPOOL_SCRIPTS_AS_ROOT is set.\n"));
	exit(1);
	}
	cmd = optarg;
	verbose = B_TRUE;
	break;
	case 'g':
	guid = B_TRUE;
	break;
	case 'L':
	follow_links = B_TRUE;
	break;
	case 'P':
	full_name = B_TRUE;
	break;
	case 'T':
	get_timestamp_arg(*optarg);
	break;
	case 'v':
	verbose = B_TRUE;
	break;
	case 'p':
	parsable = B_TRUE;
	break;
	case 'l':
	latency = B_TRUE;
	break;
	case 'q':
	queues = B_TRUE;
	break;
	case 'H':
	scripted = B_TRUE;
	break;
	case 'w':
	l_histo = B_TRUE;
	break;
	case 'r':
	rq_histo = B_TRUE;
	break;
	case 'y':
	omit_since_boot = B_TRUE;
	break;
	case 'n':
	headers_once = B_TRUE;
	break;
	case 'h':
	usage(B_FALSE);
	break;
	case '?':
	if (optopt == 'c') {
	print_zpool_script_list("iostat");
	exit(0);
	} else {
	fprintf(stderr,
	gettext("invalid option '%c'\n"), optopt);
	}
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	cb.cb_literal = parsable;
	cb.cb_scripted = scripted;

	if (guid)
	- cb.cb_name_flags \|= VDEV_NAME_GUID;
	+ cb.cb_vdevs.cb_name_flags \|= VDEV_NAME_GUID;
	if (follow_links)
	- cb.cb_name_flags \|= VDEV_NAME_FOLLOW_LINKS;
	+ cb.cb_vdevs.cb_name_flags \|= VDEV_NAME_FOLLOW_LINKS;
	if (full_name)
	- cb.cb_name_flags \|= VDEV_NAME_PATH;
	+ cb.cb_vdevs.cb_name_flags \|= VDEV_NAME_PATH;
	cb.cb_iteration = 0;
	cb.cb_namewidth = 0;
	cb.cb_verbose = verbose;

	/* Get our interval and count values (if any) */
	if (guid) {
	get_interval_count_filter_guids(&argc, argv, &interval,
	&count, &cb);
	} else {
	get_interval_count(&argc, argv, &interval, &count);
	}

	if (argc == 0) {
	/* No args, so just print the defaults. */
	} else if (are_all_pools(argc, argv)) {
	/* All the args are pool names */
	- } else if (are_vdevs_in_pool(argc, argv, NULL, &cb)) {
	+ } else if (are_vdevs_in_pool(argc, argv, NULL, &cb.cb_vdevs)) {
	/* All the args are vdevs */
	- cb.cb_vdev_names = argv;
	- cb.cb_vdev_names_count = argc;
	+ cb.cb_vdevs.cb_names = argv;
	+ cb.cb_vdevs.cb_names_count = argc;
	argc = 0; /* No pools to process */
	} else if (are_all_pools(1, argv)) {
	/* The first arg is a pool name */
	- if (are_vdevs_in_pool(argc - 1, argv + 1, argv[0], &cb)) {
	+ if (are_vdevs_in_pool(argc - 1, argv + 1, argv[0],
	+ &cb.cb_vdevs)) {
	/* ...and the rest are vdev names */
	- cb.cb_vdev_names = argv + 1;
	- cb.cb_vdev_names_count = argc - 1;
	+ cb.cb_vdevs.cb_names = argv + 1;
	+ cb.cb_vdevs.cb_names_count = argc - 1;
	argc = 1; /* One pool to process */
	} else {
	fprintf(stderr, gettext("Expected either a list of "));
	fprintf(stderr, gettext("pools, or list of vdevs in"));
	fprintf(stderr, " \"%s\", ", argv[0]);
	fprintf(stderr, gettext("but got:\n"));
	error_list_unresolved_vdevs(argc - 1, argv + 1,
	- argv[0], &cb);
	+ argv[0], &cb.cb_vdevs);
	fprintf(stderr, "\n");
	usage(B_FALSE);
	return (1);
	}
	} else {
	/*
	* The args don't make sense. The first arg isn't a pool name,
	* nor are all the args vdevs.
	*/
	fprintf(stderr, gettext("Unable to parse pools/vdevs list.\n"));
	fprintf(stderr, "\n");
	return (1);
	}

	- if (cb.cb_vdev_names_count != 0) {
	+ if (cb.cb_vdevs.cb_names_count != 0) {
	/*
	* If user specified vdevs, it implies verbose.
	*/
	cb.cb_verbose = B_TRUE;
	}

	/*
	* Construct the list of all interesting pools.
	*/
	ret = 0;
	- if ((list = pool_list_get(argc, argv, NULL, parsable, &ret)) == NULL)
	+ if ((list = pool_list_get(argc, argv, NULL, ZFS_TYPE_POOL, parsable,
	+ &ret)) == NULL)
	return (1);

	if (pool_list_count(list) == 0 && argc != 0) {
	pool_list_free(list);
	return (1);
	}

	if (pool_list_count(list) == 0 && interval == 0) {
	pool_list_free(list);
	(void) fprintf(stderr, gettext("no pools available\n"));
	return (1);
	}

	if ((l_histo \|\| rq_histo) && (cmd != NULL \|\| latency \|\| queues)) {
	pool_list_free(list);
	(void) fprintf(stderr,
	gettext("[-r\|-w] isn't allowed with [-c\|-l\|-q]\n"));
	usage(B_FALSE);
	return (1);
	}

	if (l_histo && rq_histo) {
	pool_list_free(list);
	(void) fprintf(stderr,
	gettext("Only one of [-r\|-w] can be passed at a time\n"));
	usage(B_FALSE);
	return (1);
	}

	/*
	* Enter the main iostat loop.
	*/
	cb.cb_list = list;

	if (l_histo) {
	/*
	* Histograms tables look out of place when you try to display
	* them with the other stats, so make a rule that you can only
	* print histograms by themselves.
	*/
	cb.cb_flags = IOS_L_HISTO_M;
	} else if (rq_histo) {
	cb.cb_flags = IOS_RQ_HISTO_M;
	} else {
	cb.cb_flags = IOS_DEFAULT_M;
	if (latency)
	cb.cb_flags \|= IOS_LATENCY_M;
	if (queues)
	cb.cb_flags \|= IOS_QUEUES_M;
	}

	/*
	* See if the module supports all the stats we want to display.
	*/
	unsupported_flags = cb.cb_flags & ~get_stat_flags(list);
	if (unsupported_flags) {
	uint64_t f;
	int idx;
	fprintf(stderr,
	gettext("The loaded zfs module doesn't support:"));

	/* for each bit set in unsupported_flags */
	for (f = unsupported_flags; f; f &= ~(1ULL << idx)) {
	idx = lowbit64(f) - 1;
	fprintf(stderr, " -%c", flag_to_arg[idx]);
	}

	fprintf(stderr, ". Try running a newer module.\n");
	pool_list_free(list);

	return (1);
	}

	for (;;) {
	if ((npools = pool_list_count(list)) == 0)
	(void) fprintf(stderr, gettext("no pools available\n"));
	else {
	/*
	* If this is the first iteration and -y was supplied
	* we skip any printing.
	*/
	boolean_t skip = (omit_since_boot &&
	cb.cb_iteration == 0);

	/*
	* Refresh all statistics. This is done as an
	* explicit step before calculating the maximum name
	* width, so that any * configuration changes are
	* properly accounted for.
	*/
	(void) pool_list_iter(list, B_FALSE, refresh_iostat,
	&cb);

	/*
	* Iterate over all pools to determine the maximum width
	* for the pool / device name column across all pools.
	*/
	cb.cb_namewidth = 0;
	(void) pool_list_iter(list, B_FALSE,
	get_namewidth_iostat, &cb);

	if (timestamp_fmt != NODATE)
	print_timestamp(timestamp_fmt);

	if (cmd != NULL && cb.cb_verbose &&
	!(cb.cb_flags & IOS_ANYHISTO_M)) {
	cb.vcdl = all_pools_for_each_vdev_run(argc,
	- argv, cmd, g_zfs, cb.cb_vdev_names,
	- cb.cb_vdev_names_count, cb.cb_name_flags);
	+ argv, cmd, g_zfs, cb.cb_vdevs.cb_names,
	+ cb.cb_vdevs.cb_names_count,
	+ cb.cb_vdevs.cb_name_flags);
	} else {
	cb.vcdl = NULL;
	}


	/*
	* Check terminal size so we can print headers
	* even when terminal window has its height
	* changed.
	*/
	winheight = terminal_height();
	/*
	* Are we connected to TTY? If not, headers_once
	* should be true, to avoid breaking scripts.
	*/
	if (winheight < 0)
	headers_once = B_TRUE;

	/*
	* If it's the first time and we're not skipping it,
	* or either skip or verbose mode, print the header.
	*
	* The histogram code explicitly prints its header on
	* every vdev, so skip this for histograms.
	*/
	if (((++cb.cb_iteration == 1 && !skip) \|\|
	(skip != verbose) \|\|
	(!headers_once &&
	(cb.cb_iteration % winheight) == 0)) &&
	(!(cb.cb_flags & IOS_ANYHISTO_M)) &&
	!cb.cb_scripted)
	print_iostat_header(&cb);

	if (skip) {
	(void) fsleep(interval);
	continue;
	}

	pool_list_iter(list, B_FALSE, print_iostat, &cb);

	/*
	* If there's more than one pool, and we're not in
	* verbose mode (which prints a separator for us),
	* then print a separator.
	*
	* In addition, if we're printing specific vdevs then
	* we also want an ending separator.
	*/
	if (((npools > 1 && !verbose &&
	!(cb.cb_flags & IOS_ANYHISTO_M)) \|\|
	(!(cb.cb_flags & IOS_ANYHISTO_M) &&
	- cb.cb_vdev_names_count)) &&
	+ cb.cb_vdevs.cb_names_count)) &&
	!cb.cb_scripted) {
	print_iostat_separator(&cb);
	if (cb.vcdl != NULL)
	print_cmd_columns(cb.vcdl, 1);
	printf("\n");
	}

	if (cb.vcdl != NULL)
	free_vdev_cmd_data_list(cb.vcdl);

	}

	/*
	* Flush the output so that redirection to a file isn't buffered
	* indefinitely.
	*/
	(void) fflush(stdout);

	if (interval == 0)
	break;

	if (count != 0 && --count == 0)
	break;

	(void) fsleep(interval);
	}

	pool_list_free(list);

	return (ret);
	}

	typedef struct list_cbdata {
	boolean_t cb_verbose;
	int cb_name_flags;
	int cb_namewidth;
	boolean_t cb_scripted;
	zprop_list_t *cb_proplist;
	boolean_t cb_literal;
	} list_cbdata_t;


	/*
	* Given a list of columns to display, output appropriate headers for each one.
	*/
	static void
	print_header(list_cbdata_t *cb)
	{
	zprop_list_t *pl = cb->cb_proplist;
	char headerbuf[ZPOOL_MAXPROPLEN];
	const char *header;
	boolean_t first = B_TRUE;
	boolean_t right_justify;
	size_t width = 0;

	for (; pl != NULL; pl = pl->pl_next) {
	width = pl->pl_width;
	if (first && cb->cb_verbose) {
	/*
	* Reset the width to accommodate the verbose listing
	* of devices.
	*/
	width = cb->cb_namewidth;
	}

	if (!first)
	(void) printf(" ");
	else
	first = B_FALSE;

	right_justify = B_FALSE;
	if (pl->pl_prop != ZPROP_INVAL) {
	header = zpool_prop_column_name(pl->pl_prop);
	right_justify = zpool_prop_align_right(pl->pl_prop);
	} else {
	int i;

	for (i = 0; pl->pl_user_prop[i] != '\0'; i++)
	headerbuf[i] = toupper(pl->pl_user_prop[i]);
	headerbuf[i] = '\0';
	header = headerbuf;
	}

	if (pl->pl_next == NULL && !right_justify)
	(void) printf("%s", header);
	else if (right_justify)
	(void) printf("%*s", (int)width, header);
	else
	(void) printf("%-*s", (int)width, header);
	}

	(void) printf("\n");
	}

	/*
	* Given a pool and a list of properties, print out all the properties according
	* to the described layout. Used by zpool_do_list().
	*/
	static void
	print_pool(zpool_handle_t zhp, list_cbdata_t cb)
	{
	zprop_list_t *pl = cb->cb_proplist;
	boolean_t first = B_TRUE;
	char property[ZPOOL_MAXPROPLEN];
	char *propstr;
	boolean_t right_justify;
	size_t width;

	for (; pl != NULL; pl = pl->pl_next) {

	width = pl->pl_width;
	if (first && cb->cb_verbose) {
	/*
	* Reset the width to accommodate the verbose listing
	* of devices.
	*/
	width = cb->cb_namewidth;
	}

	if (!first) {
	if (cb->cb_scripted)
	(void) printf("\t");
	else
	(void) printf(" ");
	} else {
	first = B_FALSE;
	}

	right_justify = B_FALSE;
	if (pl->pl_prop != ZPROP_INVAL) {
	if (zpool_get_prop(zhp, pl->pl_prop, property,
	sizeof (property), NULL, cb->cb_literal) != 0)
	propstr = "-";
	else
	propstr = property;

	right_justify = zpool_prop_align_right(pl->pl_prop);
	} else if ((zpool_prop_feature(pl->pl_user_prop) \|\|
	zpool_prop_unsupported(pl->pl_user_prop)) &&
	zpool_prop_get_feature(zhp, pl->pl_user_prop, property,
	sizeof (property)) == 0) {
	propstr = property;
	} else {
	propstr = "-";
	}


	/*
	* If this is being called in scripted mode, or if this is the
	* last column and it is left-justified, don't include a width
	* format specifier.
	*/
	if (cb->cb_scripted \|\| (pl->pl_next == NULL && !right_justify))
	(void) printf("%s", propstr);
	else if (right_justify)
	(void) printf("%*s", (int)width, propstr);
	else
	(void) printf("%-*s", (int)width, propstr);
	}

	(void) printf("\n");
	}

	static void
	print_one_column(zpool_prop_t prop, uint64_t value, const char *str,
	boolean_t scripted, boolean_t valid, enum zfs_nicenum_format format)
	{
	char propval[64];
	boolean_t fixed;
	size_t width = zprop_width(prop, &fixed, ZFS_TYPE_POOL);

	switch (prop) {
	case ZPOOL_PROP_EXPANDSZ:
	case ZPOOL_PROP_CHECKPOINT:
	case ZPOOL_PROP_DEDUPRATIO:
	if (value == 0)
	(void) strlcpy(propval, "-", sizeof (propval));
	else
	zfs_nicenum_format(value, propval, sizeof (propval),
	format);
	break;
	case ZPOOL_PROP_FRAGMENTATION:
	if (value == ZFS_FRAG_INVALID) {
	(void) strlcpy(propval, "-", sizeof (propval));
	} else if (format == ZFS_NICENUM_RAW) {
	(void) snprintf(propval, sizeof (propval), "%llu",
	(unsigned long long)value);
	} else {
	(void) snprintf(propval, sizeof (propval), "%llu%%",
	(unsigned long long)value);
	}
	break;
	case ZPOOL_PROP_CAPACITY:
	/* capacity value is in parts-per-10,000 (aka permyriad) */
	if (format == ZFS_NICENUM_RAW)
	(void) snprintf(propval, sizeof (propval), "%llu",
	(unsigned long long)value / 100);
	else
	(void) snprintf(propval, sizeof (propval),
	value < 1000 ? "%1.2f%%" : value < 10000 ?
	"%2.1f%%" : "%3.0f%%", value / 100.0);
	break;
	case ZPOOL_PROP_HEALTH:
	width = 8;
	(void) strlcpy(propval, str, sizeof (propval));
	break;
	default:
	zfs_nicenum_format(value, propval, sizeof (propval), format);
	}

	if (!valid)
	(void) strlcpy(propval, "-", sizeof (propval));

	if (scripted)
	(void) printf("\t%s", propval);
	else
	(void) printf(" %*s", (int)width, propval);
	}

	/*
	* print static default line per vdev
	* not compatible with '-o' <proplist> option
	*/
	static void
	print_list_stats(zpool_handle_t zhp, const char name, nvlist_t *nv,
	list_cbdata_t *cb, int depth, boolean_t isspare)
	{
	nvlist_t **child;
	vdev_stat_t *vs;
	uint_t c, children;
	char *vname;
	boolean_t scripted = cb->cb_scripted;
	uint64_t islog = B_FALSE;
	char dashes = "%-s - - - - "
	"- - - - -\n";

	verify(nvlist_lookup_uint64_array(nv, ZPOOL_CONFIG_VDEV_STATS,
	(uint64_t **)&vs, &c) == 0);

	if (name != NULL) {
	boolean_t toplevel = (vs->vs_space != 0);
	uint64_t cap;
	enum zfs_nicenum_format format;
	const char *state;

	if (cb->cb_literal)
	format = ZFS_NICENUM_RAW;
	else
	format = ZFS_NICENUM_1024;

	if (strcmp(name, VDEV_TYPE_INDIRECT) == 0)
	return;

	if (scripted)
	(void) printf("\t%s", name);
	else if (strlen(name) + depth > cb->cb_namewidth)
	(void) printf("%*s%s", depth, "", name);
	else
	(void) printf("%s%s%s", depth, "", name,
	(int)(cb->cb_namewidth - strlen(name) - depth), "");

	/*
	* Print the properties for the individual vdevs. Some
	* properties are only applicable to toplevel vdevs. The
	* 'toplevel' boolean value is passed to the print_one_column()
	* to indicate that the value is valid.
	*/
	print_one_column(ZPOOL_PROP_SIZE, vs->vs_space, NULL, scripted,
	toplevel, format);
	print_one_column(ZPOOL_PROP_ALLOCATED, vs->vs_alloc, NULL,
	scripted, toplevel, format);
	print_one_column(ZPOOL_PROP_FREE, vs->vs_space - vs->vs_alloc,
	NULL, scripted, toplevel, format);
	print_one_column(ZPOOL_PROP_CHECKPOINT,
	vs->vs_checkpoint_space, NULL, scripted, toplevel, format);
	print_one_column(ZPOOL_PROP_EXPANDSZ, vs->vs_esize, NULL,
	scripted, B_TRUE, format);
	print_one_column(ZPOOL_PROP_FRAGMENTATION,
	vs->vs_fragmentation, NULL, scripted,
	(vs->vs_fragmentation != ZFS_FRAG_INVALID && toplevel),
	format);
	cap = (vs->vs_space == 0) ? 0 :
	(vs->vs_alloc * 10000 / vs->vs_space);
	print_one_column(ZPOOL_PROP_CAPACITY, cap, NULL,
	scripted, toplevel, format);
	print_one_column(ZPOOL_PROP_DEDUPRATIO, 0, NULL,
	scripted, toplevel, format);
	state = zpool_state_to_name(vs->vs_state, vs->vs_aux);
	if (isspare) {
	if (vs->vs_aux == VDEV_AUX_SPARED)
	state = "INUSE";
	else if (vs->vs_state == VDEV_STATE_HEALTHY)
	state = "AVAIL";
	}
	print_one_column(ZPOOL_PROP_HEALTH, 0, state, scripted,
	B_TRUE, format);
	(void) printf("\n");
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0)
	return;

	/* list the normal vdevs first */
	for (c = 0; c < children; c++) {
	uint64_t ishole = B_FALSE;

	if (nvlist_lookup_uint64(child[c],
	ZPOOL_CONFIG_IS_HOLE, &ishole) == 0 && ishole)
	continue;

	if (nvlist_lookup_uint64(child[c],
	ZPOOL_CONFIG_IS_LOG, &islog) == 0 && islog)
	continue;

	if (nvlist_exists(child[c], ZPOOL_CONFIG_ALLOCATION_BIAS))
	continue;

	vname = zpool_vdev_name(g_zfs, zhp, child[c],
	cb->cb_name_flags);
	print_list_stats(zhp, vname, child[c], cb, depth + 2, B_FALSE);
	free(vname);
	}

	/* list the classes: 'logs', 'dedup', and 'special' */
	for (uint_t n = 0; n < 3; n++) {
	boolean_t printed = B_FALSE;

	for (c = 0; c < children; c++) {
	char *bias = NULL;
	char *type = NULL;

	if (nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG,
	&islog) == 0 && islog) {
	bias = VDEV_ALLOC_CLASS_LOGS;
	} else {
	(void) nvlist_lookup_string(child[c],
	ZPOOL_CONFIG_ALLOCATION_BIAS, &bias);
	(void) nvlist_lookup_string(child[c],
	ZPOOL_CONFIG_TYPE, &type);
	}
	if (bias == NULL \|\| strcmp(bias, class_name[n]) != 0)
	continue;
	if (!islog && strcmp(type, VDEV_TYPE_INDIRECT) == 0)
	continue;

	if (!printed) {
	/* LINTED E_SEC_PRINTF_VAR_FMT */
	(void) printf(dashes, cb->cb_namewidth,
	class_name[n]);
	printed = B_TRUE;
	}
	vname = zpool_vdev_name(g_zfs, zhp, child[c],
	cb->cb_name_flags);
	print_list_stats(zhp, vname, child[c], cb, depth + 2,
	B_FALSE);
	free(vname);
	}
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE,
	&child, &children) == 0 && children > 0) {
	/* LINTED E_SEC_PRINTF_VAR_FMT */
	(void) printf(dashes, cb->cb_namewidth, "cache");
	for (c = 0; c < children; c++) {
	vname = zpool_vdev_name(g_zfs, zhp, child[c],
	cb->cb_name_flags);
	print_list_stats(zhp, vname, child[c], cb, depth + 2,
	B_FALSE);
	free(vname);
	}
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES, &child,
	&children) == 0 && children > 0) {
	/* LINTED E_SEC_PRINTF_VAR_FMT */
	(void) printf(dashes, cb->cb_namewidth, "spare");
	for (c = 0; c < children; c++) {
	vname = zpool_vdev_name(g_zfs, zhp, child[c],
	cb->cb_name_flags);
	print_list_stats(zhp, vname, child[c], cb, depth + 2,
	B_TRUE);
	free(vname);
	}
	}
	}

	/*
	* Generic callback function to list a pool.
	*/
	static int
	list_callback(zpool_handle_t zhp, void data)
	{
	list_cbdata_t *cbp = data;

	print_pool(zhp, cbp);

	if (cbp->cb_verbose) {
	nvlist_t config, nvroot;

	config = zpool_get_config(zhp, NULL);
	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) == 0);
	print_list_stats(zhp, NULL, nvroot, cbp, 0, B_FALSE);
	}

	return (0);
	}

	/*
	* Set the minimum pool/vdev name column width. The width must be at least 9,
	* but may be as large as needed.
	*/
	static int
	get_namewidth_list(zpool_handle_t zhp, void data)
	{
	list_cbdata_t *cb = data;
	int width;

	width = get_namewidth(zhp, cb->cb_namewidth, cb->cb_name_flags,
	cb->cb_verbose);

	if (width < 9)
	width = 9;

	cb->cb_namewidth = width;

	return (0);
	}

	/*
	* zpool list [-gHLpP] [-o prop[,prop]*] [-T d\|u] [pool] ... [interval [count]]
	*
	* -g Display guid for individual vdev name.
	* -H Scripted mode. Don't display headers, and separate properties
	* by a single tab.
	* -L Follow links when resolving vdev path name.
	* -o List of properties to display. Defaults to
	* "name,size,allocated,free,expandsize,fragmentation,capacity,"
	* "dedupratio,health,altroot"
	* -p Display values in parsable (exact) format.
	* -P Display full path for vdev name.
	* -T Display a timestamp in date(1) or Unix format
	*
	* List all pools in the system, whether or not they're healthy. Output space
	* statistics for each one, as well as health status summary.
	*/
	int
	zpool_do_list(int argc, char **argv)
	{
	int c;
	int ret = 0;
	list_cbdata_t cb = { 0 };
	static char default_props[] =
	"name,size,allocated,free,checkpoint,expandsize,fragmentation,"
	"capacity,dedupratio,health,altroot";
	char *props = default_props;
	float interval = 0;
	unsigned long count = 0;
	zpool_list_t *list;
	boolean_t first = B_TRUE;
	+ current_prop_type = ZFS_TYPE_POOL;

	/* check options */
	while ((c = getopt(argc, argv, ":gHLo:pPT:v")) != -1) {
	switch (c) {
	case 'g':
	cb.cb_name_flags \|= VDEV_NAME_GUID;
	break;
	case 'H':
	cb.cb_scripted = B_TRUE;
	break;
	case 'L':
	cb.cb_name_flags \|= VDEV_NAME_FOLLOW_LINKS;
	break;
	case 'o':
	props = optarg;
	break;
	case 'P':
	cb.cb_name_flags \|= VDEV_NAME_PATH;
	break;
	case 'p':
	cb.cb_literal = B_TRUE;
	break;
	case 'T':
	get_timestamp_arg(*optarg);
	break;
	case 'v':
	cb.cb_verbose = B_TRUE;
	cb.cb_namewidth = 8; /* 8 until precalc is avail */
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	usage(B_FALSE);
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	get_interval_count(&argc, argv, &interval, &count);

	if (zprop_get_list(g_zfs, props, &cb.cb_proplist, ZFS_TYPE_POOL) != 0)
	usage(B_FALSE);

	for (;;) {
	if ((list = pool_list_get(argc, argv, &cb.cb_proplist,
	- cb.cb_literal, &ret)) == NULL)
	+ ZFS_TYPE_POOL, cb.cb_literal, &ret)) == NULL)
	return (1);

	if (pool_list_count(list) == 0)
	break;

	cb.cb_namewidth = 0;
	(void) pool_list_iter(list, B_FALSE, get_namewidth_list, &cb);

	if (timestamp_fmt != NODATE)
	print_timestamp(timestamp_fmt);

	if (!cb.cb_scripted && (first \|\| cb.cb_verbose)) {
	print_header(&cb);
	first = B_FALSE;
	}
	ret = pool_list_iter(list, B_TRUE, list_callback, &cb);

	if (interval == 0)
	break;

	if (count != 0 && --count == 0)
	break;

	pool_list_free(list);
	(void) fsleep(interval);
	}

	if (argc == 0 && !cb.cb_scripted && pool_list_count(list) == 0) {
	(void) printf(gettext("no pools available\n"));
	ret = 0;
	}

	pool_list_free(list);
	zprop_free_list(cb.cb_proplist);
	return (ret);
	}

	static int
	zpool_do_attach_or_replace(int argc, char **argv, int replacing)
	{
	boolean_t force = B_FALSE;
	boolean_t rebuild = B_FALSE;
	boolean_t wait = B_FALSE;
	int c;
	nvlist_t *nvroot;
	char poolname, old_disk, *new_disk;
	zpool_handle_t *zhp;
	nvlist_t *props = NULL;
	char *propval;
	int ret;

	/* check options */
	while ((c = getopt(argc, argv, "fo:sw")) != -1) {
	switch (c) {
	case 'f':
	force = B_TRUE;
	break;
	case 'o':
	if ((propval = strchr(optarg, '=')) == NULL) {
	(void) fprintf(stderr, gettext("missing "
	"'=' for -o option\n"));
	usage(B_FALSE);
	}
	*propval = '\0';
	propval++;

	if ((strcmp(optarg, ZPOOL_CONFIG_ASHIFT) != 0) \|\|
	(add_prop_list(optarg, propval, &props, B_TRUE)))
	usage(B_FALSE);
	break;
	case 's':
	rebuild = B_TRUE;
	break;
	case 'w':
	wait = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* get pool name and check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name argument\n"));
	usage(B_FALSE);
	}

	poolname = argv[0];

	if (argc < 2) {
	(void) fprintf(stderr,
	gettext("missing <device> specification\n"));
	usage(B_FALSE);
	}

	old_disk = argv[1];

	if (argc < 3) {
	if (!replacing) {
	(void) fprintf(stderr,
	gettext("missing <new_device> specification\n"));
	usage(B_FALSE);
	}
	new_disk = old_disk;
	argc -= 1;
	argv += 1;
	} else {
	new_disk = argv[2];
	argc -= 2;
	argv += 2;
	}

	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	if ((zhp = zpool_open(g_zfs, poolname)) == NULL) {
	nvlist_free(props);
	return (1);
	}

	if (zpool_get_config(zhp, NULL) == NULL) {
	(void) fprintf(stderr, gettext("pool '%s' is unavailable\n"),
	poolname);
	zpool_close(zhp);
	nvlist_free(props);
	return (1);
	}

	/* unless manually specified use "ashift" pool property (if set) */
	if (!nvlist_exists(props, ZPOOL_CONFIG_ASHIFT)) {
	int intval;
	zprop_source_t src;
	char strval[ZPOOL_MAXPROPLEN];

	intval = zpool_get_prop_int(zhp, ZPOOL_PROP_ASHIFT, &src);
	if (src != ZPROP_SRC_DEFAULT) {
	(void) sprintf(strval, "%" PRId32, intval);
	verify(add_prop_list(ZPOOL_CONFIG_ASHIFT, strval,
	&props, B_TRUE) == 0);
	}
	}

	nvroot = make_root_vdev(zhp, props, force, B_FALSE, replacing, B_FALSE,
	argc, argv);
	if (nvroot == NULL) {
	zpool_close(zhp);
	nvlist_free(props);
	return (1);
	}

	ret = zpool_vdev_attach(zhp, old_disk, new_disk, nvroot, replacing,
	rebuild);

	if (ret == 0 && wait)
	ret = zpool_wait(zhp,
	replacing ? ZPOOL_WAIT_REPLACE : ZPOOL_WAIT_RESILVER);

	nvlist_free(props);
	nvlist_free(nvroot);
	zpool_close(zhp);

	return (ret);
	}

	/*
	* zpool replace [-fsw] [-o property=value] <pool> <device> <new_device>
	*
	* -f Force attach, even if <new_device> appears to be in use.
	* -s Use sequential instead of healing reconstruction for resilver.
	* -o Set property=value.
	* -w Wait for replacing to complete before returning
	*
	* Replace <device> with <new_device>.
	*/
	/* ARGSUSED */
	int
	zpool_do_replace(int argc, char **argv)
	{
	return (zpool_do_attach_or_replace(argc, argv, B_TRUE));
	}

	/*
	* zpool attach [-fsw] [-o property=value] <pool> <device> <new_device>
	*
	* -f Force attach, even if <new_device> appears to be in use.
	* -s Use sequential instead of healing reconstruction for resilver.
	* -o Set property=value.
	* -w Wait for resilvering to complete before returning
	*
	* Attach <new_device> to the mirror containing <device>. If <device> is not
	* part of a mirror, then <device> will be transformed into a mirror of
	* <device> and <new_device>. In either case, <new_device> will begin life
	* with a DTL of [0, now], and will immediately begin to resilver itself.
	*/
	int
	zpool_do_attach(int argc, char **argv)
	{
	return (zpool_do_attach_or_replace(argc, argv, B_FALSE));
	}

	/*
	* zpool detach [-f] <pool> <device>
	*
	* -f Force detach of <device>, even if DTLs argue against it
	* (not supported yet)
	*
	* Detach a device from a mirror. The operation will be refused if <device>
	* is the last device in the mirror, or if the DTLs indicate that this device
	* has the only valid copy of some data.
	*/
	/* ARGSUSED */
	int
	zpool_do_detach(int argc, char **argv)
	{
	int c;
	char poolname, path;
	zpool_handle_t *zhp;
	int ret;

	/* check options */
	while ((c = getopt(argc, argv, "")) != -1) {
	switch (c) {
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* get pool name and check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name argument\n"));
	usage(B_FALSE);
	}

	if (argc < 2) {
	(void) fprintf(stderr,
	gettext("missing <device> specification\n"));
	usage(B_FALSE);
	}

	poolname = argv[0];
	path = argv[1];

	if ((zhp = zpool_open(g_zfs, poolname)) == NULL)
	return (1);

	ret = zpool_vdev_detach(zhp, path);

	zpool_close(zhp);

	return (ret);
	}

	/*
	* zpool split [-gLnP] [-o prop=val] ...
	* [-o mntopt] ...
	* [-R altroot] <pool> <newpool> [<device> ...]
	*
	* -g Display guid for individual vdev name.
	* -L Follow links when resolving vdev path name.
	* -n Do not split the pool, but display the resulting layout if
	* it were to be split.
	* -o Set property=value, or set mount options.
	* -P Display full path for vdev name.
	* -R Mount the split-off pool under an alternate root.
	* -l Load encryption keys while importing.
	*
	* Splits the named pool and gives it the new pool name. Devices to be split
	* off may be listed, provided that no more than one device is specified
	* per top-level vdev mirror. The newly split pool is left in an exported
	* state unless -R is specified.
	*
	* Restrictions: the top-level of the pool pool must only be made up of
	* mirrors; all devices in the pool must be healthy; no device may be
	* undergoing a resilvering operation.
	*/
	int
	zpool_do_split(int argc, char **argv)
	{
	char srcpool, newpool, *propval;
	char *mntopts = NULL;
	splitflags_t flags;
	int c, ret = 0;
	boolean_t loadkeys = B_FALSE;
	zpool_handle_t *zhp;
	nvlist_t config, props = NULL;

	flags.dryrun = B_FALSE;
	flags.import = B_FALSE;
	flags.name_flags = 0;

	/* check options */
	while ((c = getopt(argc, argv, ":gLR:lno:P")) != -1) {
	switch (c) {
	case 'g':
	flags.name_flags \|= VDEV_NAME_GUID;
	break;
	case 'L':
	flags.name_flags \|= VDEV_NAME_FOLLOW_LINKS;
	break;
	case 'R':
	flags.import = B_TRUE;
	if (add_prop_list(
	zpool_prop_to_name(ZPOOL_PROP_ALTROOT), optarg,
	&props, B_TRUE) != 0) {
	nvlist_free(props);
	usage(B_FALSE);
	}
	break;
	case 'l':
	loadkeys = B_TRUE;
	break;
	case 'n':
	flags.dryrun = B_TRUE;
	break;
	case 'o':
	if ((propval = strchr(optarg, '=')) != NULL) {
	*propval = '\0';
	propval++;
	if (add_prop_list(optarg, propval,
	&props, B_TRUE) != 0) {
	nvlist_free(props);
	usage(B_FALSE);
	}
	} else {
	mntopts = optarg;
	}
	break;
	case 'P':
	flags.name_flags \|= VDEV_NAME_PATH;
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	usage(B_FALSE);
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	break;
	}
	}

	if (!flags.import && mntopts != NULL) {
	(void) fprintf(stderr, gettext("setting mntopts is only "
	"valid when importing the pool\n"));
	usage(B_FALSE);
	}

	if (!flags.import && loadkeys) {
	(void) fprintf(stderr, gettext("loading keys is only "
	"valid when importing the pool\n"));
	usage(B_FALSE);
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("Missing pool name\n"));
	usage(B_FALSE);
	}
	if (argc < 2) {
	(void) fprintf(stderr, gettext("Missing new pool name\n"));
	usage(B_FALSE);
	}

	srcpool = argv[0];
	newpool = argv[1];

	argc -= 2;
	argv += 2;

	if ((zhp = zpool_open(g_zfs, srcpool)) == NULL) {
	nvlist_free(props);
	return (1);
	}

	config = split_mirror_vdev(zhp, newpool, props, flags, argc, argv);
	if (config == NULL) {
	ret = 1;
	} else {
	if (flags.dryrun) {
	(void) printf(gettext("would create '%s' with the "
	"following layout:\n\n"), newpool);
	print_vdev_tree(NULL, newpool, config, 0, "",
	flags.name_flags);
	print_vdev_tree(NULL, "dedup", config, 0,
	VDEV_ALLOC_BIAS_DEDUP, 0);
	print_vdev_tree(NULL, "special", config, 0,
	VDEV_ALLOC_BIAS_SPECIAL, 0);
	}
	}

	zpool_close(zhp);

	if (ret != 0 \|\| flags.dryrun \|\| !flags.import) {
	nvlist_free(config);
	nvlist_free(props);
	return (ret);
	}

	/*
	* The split was successful. Now we need to open the new
	* pool and import it.
	*/
	if ((zhp = zpool_open_canfail(g_zfs, newpool)) == NULL) {
	nvlist_free(config);
	nvlist_free(props);
	return (1);
	}

	if (loadkeys) {
	ret = zfs_crypto_attempt_load_keys(g_zfs, newpool);
	if (ret != 0)
	ret = 1;
	}

	if (zpool_get_state(zhp) != POOL_STATE_UNAVAIL &&
	zpool_enable_datasets(zhp, mntopts, 0) != 0) {
	ret = 1;
	(void) fprintf(stderr, gettext("Split was successful, but "
	"the datasets could not all be mounted\n"));
	(void) fprintf(stderr, gettext("Try doing '%s' with a "
	"different altroot\n"), "zpool import");
	}
	zpool_close(zhp);
	nvlist_free(config);
	nvlist_free(props);

	return (ret);
	}



	/*
	* zpool online <pool> <device> ...
	*/
	int
	zpool_do_online(int argc, char **argv)
	{
	int c, i;
	char *poolname;
	zpool_handle_t *zhp;
	int ret = 0;
	vdev_state_t newstate;
	int flags = 0;

	/* check options */
	while ((c = getopt(argc, argv, "e")) != -1) {
	switch (c) {
	case 'e':
	flags \|= ZFS_ONLINE_EXPAND;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* get pool name and check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name\n"));
	usage(B_FALSE);
	}
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing device name\n"));
	usage(B_FALSE);
	}

	poolname = argv[0];

	if ((zhp = zpool_open(g_zfs, poolname)) == NULL)
	return (1);

	for (i = 1; i < argc; i++) {
	if (zpool_vdev_online(zhp, argv[i], flags, &newstate) == 0) {
	if (newstate != VDEV_STATE_HEALTHY) {
	(void) printf(gettext("warning: device '%s' "
	"onlined, but remains in faulted state\n"),
	argv[i]);
	if (newstate == VDEV_STATE_FAULTED)
	(void) printf(gettext("use 'zpool "
	"clear' to restore a faulted "
	"device\n"));
	else
	(void) printf(gettext("use 'zpool "
	"replace' to replace devices "
	"that are no longer present\n"));
	}
	} else {
	ret = 1;
	}
	}

	zpool_close(zhp);

	return (ret);
	}

	/*
	* zpool offline [-ft] <pool> <device> ...
	*
	* -f Force the device into a faulted state.
	*
	* -t Only take the device off-line temporarily. The offline/faulted
	* state will not be persistent across reboots.
	*/
	/* ARGSUSED */
	int
	zpool_do_offline(int argc, char **argv)
	{
	int c, i;
	char *poolname;
	zpool_handle_t *zhp;
	int ret = 0;
	boolean_t istmp = B_FALSE;
	boolean_t fault = B_FALSE;

	/* check options */
	while ((c = getopt(argc, argv, "ft")) != -1) {
	switch (c) {
	case 'f':
	fault = B_TRUE;
	break;
	case 't':
	istmp = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* get pool name and check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name\n"));
	usage(B_FALSE);
	}
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing device name\n"));
	usage(B_FALSE);
	}

	poolname = argv[0];

	if ((zhp = zpool_open(g_zfs, poolname)) == NULL)
	return (1);

	for (i = 1; i < argc; i++) {
	if (fault) {
	uint64_t guid = zpool_vdev_path_to_guid(zhp, argv[i]);
	vdev_aux_t aux;
	if (istmp == B_FALSE) {
	/* Force the fault to persist across imports */
	aux = VDEV_AUX_EXTERNAL_PERSIST;
	} else {
	aux = VDEV_AUX_EXTERNAL;
	}

	if (guid == 0 \|\| zpool_vdev_fault(zhp, guid, aux) != 0)
	ret = 1;
	} else {
	if (zpool_vdev_offline(zhp, argv[i], istmp) != 0)
	ret = 1;
	}
	}

	zpool_close(zhp);

	return (ret);
	}

	/*
	* zpool clear <pool> [device]
	*
	* Clear all errors associated with a pool or a particular device.
	*/
	int
	zpool_do_clear(int argc, char **argv)
	{
	int c;
	int ret = 0;
	boolean_t dryrun = B_FALSE;
	boolean_t do_rewind = B_FALSE;
	boolean_t xtreme_rewind = B_FALSE;
	uint32_t rewind_policy = ZPOOL_NO_REWIND;
	nvlist_t *policy = NULL;
	zpool_handle_t *zhp;
	char pool, device;

	/* check options */
	while ((c = getopt(argc, argv, "FnX")) != -1) {
	switch (c) {
	case 'F':
	do_rewind = B_TRUE;
	break;
	case 'n':
	dryrun = B_TRUE;
	break;
	case 'X':
	xtreme_rewind = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name\n"));
	usage(B_FALSE);
	}

	if (argc > 2) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	if ((dryrun \|\| xtreme_rewind) && !do_rewind) {
	(void) fprintf(stderr,
	gettext("-n or -X only meaningful with -F\n"));
	usage(B_FALSE);
	}
	if (dryrun)
	rewind_policy = ZPOOL_TRY_REWIND;
	else if (do_rewind)
	rewind_policy = ZPOOL_DO_REWIND;
	if (xtreme_rewind)
	rewind_policy \|= ZPOOL_EXTREME_REWIND;

	/* In future, further rewind policy choices can be passed along here */
	if (nvlist_alloc(&policy, NV_UNIQUE_NAME, 0) != 0 \|\|
	nvlist_add_uint32(policy, ZPOOL_LOAD_REWIND_POLICY,
	rewind_policy) != 0) {
	return (1);
	}

	pool = argv[0];
	device = argc == 2 ? argv[1] : NULL;

	if ((zhp = zpool_open_canfail(g_zfs, pool)) == NULL) {
	nvlist_free(policy);
	return (1);
	}

	if (zpool_clear(zhp, device, policy) != 0)
	ret = 1;

	zpool_close(zhp);

	nvlist_free(policy);

	return (ret);
	}

	/*
	* zpool reguid <pool>
	*/
	int
	zpool_do_reguid(int argc, char **argv)
	{
	int c;
	char *poolname;
	zpool_handle_t *zhp;
	int ret = 0;

	/* check options */
	while ((c = getopt(argc, argv, "")) != -1) {
	switch (c) {
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* get pool name and check number of arguments */
	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name\n"));
	usage(B_FALSE);
	}

	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	poolname = argv[0];
	if ((zhp = zpool_open(g_zfs, poolname)) == NULL)
	return (1);

	ret = zpool_reguid(zhp);

	zpool_close(zhp);
	return (ret);
	}


	/*
	* zpool reopen <pool>
	*
	* Reopen the pool so that the kernel can update the sizes of all vdevs.
	*/
	int
	zpool_do_reopen(int argc, char **argv)
	{
	int c;
	int ret = 0;
	boolean_t scrub_restart = B_TRUE;

	/* check options */
	while ((c = getopt(argc, argv, "n")) != -1) {
	switch (c) {
	case 'n':
	scrub_restart = B_FALSE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	/* if argc == 0 we will execute zpool_reopen_one on all pools */
	- ret = for_each_pool(argc, argv, B_TRUE, NULL, B_FALSE, zpool_reopen_one,
	- &scrub_restart);
	+ ret = for_each_pool(argc, argv, B_TRUE, NULL, ZFS_TYPE_POOL,
	+ B_FALSE, zpool_reopen_one, &scrub_restart);

	return (ret);
	}

	typedef struct scrub_cbdata {
	int cb_type;
	pool_scrub_cmd_t cb_scrub_cmd;
	} scrub_cbdata_t;

	static boolean_t
	zpool_has_checkpoint(zpool_handle_t *zhp)
	{
	nvlist_t config, nvroot;

	config = zpool_get_config(zhp, NULL);

	if (config != NULL) {
	pool_checkpoint_stat_t *pcs = NULL;
	uint_t c;

	nvroot = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE);
	(void) nvlist_lookup_uint64_array(nvroot,
	ZPOOL_CONFIG_CHECKPOINT_STATS, (uint64_t **)&pcs, &c);

	if (pcs == NULL \|\| pcs->pcs_state == CS_NONE)
	return (B_FALSE);

	assert(pcs->pcs_state == CS_CHECKPOINT_EXISTS \|\|
	pcs->pcs_state == CS_CHECKPOINT_DISCARDING);
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	static int
	scrub_callback(zpool_handle_t zhp, void data)
	{
	scrub_cbdata_t *cb = data;
	int err;

	/*
	* Ignore faulted pools.
	*/
	if (zpool_get_state(zhp) == POOL_STATE_UNAVAIL) {
	(void) fprintf(stderr, gettext("cannot scan '%s': pool is "
	"currently unavailable\n"), zpool_get_name(zhp));
	return (1);
	}

	err = zpool_scan(zhp, cb->cb_type, cb->cb_scrub_cmd);

	if (err == 0 && zpool_has_checkpoint(zhp) &&
	cb->cb_type == POOL_SCAN_SCRUB) {
	(void) printf(gettext("warning: will not scrub state that "
	"belongs to the checkpoint of pool '%s'\n"),
	zpool_get_name(zhp));
	}

	return (err != 0);
	}

	static int
	wait_callback(zpool_handle_t zhp, void data)
	{
	zpool_wait_activity_t *act = data;
	return (zpool_wait(zhp, *act));
	}

	/*
	* zpool scrub [-s \| -p] [-w] <pool> ...
	*
	* -s Stop. Stops any in-progress scrub.
	* -p Pause. Pause in-progress scrub.
	* -w Wait. Blocks until scrub has completed.
	*/
	int
	zpool_do_scrub(int argc, char **argv)
	{
	int c;
	scrub_cbdata_t cb;
	boolean_t wait = B_FALSE;
	int error;

	cb.cb_type = POOL_SCAN_SCRUB;
	cb.cb_scrub_cmd = POOL_SCRUB_NORMAL;

	/* check options */
	while ((c = getopt(argc, argv, "spw")) != -1) {
	switch (c) {
	case 's':
	cb.cb_type = POOL_SCAN_NONE;
	break;
	case 'p':
	cb.cb_scrub_cmd = POOL_SCRUB_PAUSE;
	break;
	case 'w':
	wait = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	if (cb.cb_type == POOL_SCAN_NONE &&
	cb.cb_scrub_cmd == POOL_SCRUB_PAUSE) {
	(void) fprintf(stderr, gettext("invalid option combination: "
	"-s and -p are mutually exclusive\n"));
	usage(B_FALSE);
	}

	if (wait && (cb.cb_type == POOL_SCAN_NONE \|\|
	cb.cb_scrub_cmd == POOL_SCRUB_PAUSE)) {
	(void) fprintf(stderr, gettext("invalid option combination: "
	"-w cannot be used with -p or -s\n"));
	usage(B_FALSE);
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name argument\n"));
	usage(B_FALSE);
	}

	- error = for_each_pool(argc, argv, B_TRUE, NULL, B_FALSE,
	- scrub_callback, &cb);
	+ error = for_each_pool(argc, argv, B_TRUE, NULL, ZFS_TYPE_POOL,
	+ B_FALSE, scrub_callback, &cb);

	if (wait && !error) {
	zpool_wait_activity_t act = ZPOOL_WAIT_SCRUB;
	- error = for_each_pool(argc, argv, B_TRUE, NULL, B_FALSE,
	- wait_callback, &act);
	+ error = for_each_pool(argc, argv, B_TRUE, NULL, ZFS_TYPE_POOL,
	+ B_FALSE, wait_callback, &act);
	}

	return (error);
	}

	/*
	* zpool resilver <pool> ...
	*
	* Restarts any in-progress resilver
	*/
	int
	zpool_do_resilver(int argc, char **argv)
	{
	int c;
	scrub_cbdata_t cb;

	cb.cb_type = POOL_SCAN_RESILVER;
	cb.cb_scrub_cmd = POOL_SCRUB_NORMAL;

	/* check options */
	while ((c = getopt(argc, argv, "")) != -1) {
	switch (c) {
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name argument\n"));
	usage(B_FALSE);
	}

	- return (for_each_pool(argc, argv, B_TRUE, NULL, B_FALSE,
	- scrub_callback, &cb));
	+ return (for_each_pool(argc, argv, B_TRUE, NULL, ZFS_TYPE_POOL,
	+ B_FALSE, scrub_callback, &cb));
	}

	/*
	* zpool trim [-d] [-r <rate>] [-c \| -s] <pool> [<device> ...]
	*
	* -c Cancel. Ends any in-progress trim.
	* -d Secure trim. Requires kernel and device support.
	* -r <rate> Sets the TRIM rate in bytes (per second). Supports
	* adding a multiplier suffix such as 'k' or 'm'.
	* -s Suspend. TRIM can then be restarted with no flags.
	* -w Wait. Blocks until trimming has completed.
	*/
	int
	zpool_do_trim(int argc, char **argv)
	{
	struct option long_options[] = {
	{"cancel", no_argument, NULL, 'c'},
	{"secure", no_argument, NULL, 'd'},
	{"rate", required_argument, NULL, 'r'},
	{"suspend", no_argument, NULL, 's'},
	{"wait", no_argument, NULL, 'w'},
	{0, 0, 0, 0}
	};

	pool_trim_func_t cmd_type = POOL_TRIM_START;
	uint64_t rate = 0;
	boolean_t secure = B_FALSE;
	boolean_t wait = B_FALSE;

	int c;
	while ((c = getopt_long(argc, argv, "cdr:sw", long_options, NULL))
	!= -1) {
	switch (c) {
	case 'c':
	if (cmd_type != POOL_TRIM_START &&
	cmd_type != POOL_TRIM_CANCEL) {
	(void) fprintf(stderr, gettext("-c cannot be "
	"combined with other options\n"));
	usage(B_FALSE);
	}
	cmd_type = POOL_TRIM_CANCEL;
	break;
	case 'd':
	if (cmd_type != POOL_TRIM_START) {
	(void) fprintf(stderr, gettext("-d cannot be "
	"combined with the -c or -s options\n"));
	usage(B_FALSE);
	}
	secure = B_TRUE;
	break;
	case 'r':
	if (cmd_type != POOL_TRIM_START) {
	(void) fprintf(stderr, gettext("-r cannot be "
	"combined with the -c or -s options\n"));
	usage(B_FALSE);
	}
	if (zfs_nicestrtonum(g_zfs, optarg, &rate) == -1) {
	(void) fprintf(stderr, "%s: %s\n",
	gettext("invalid value for rate"),
	libzfs_error_description(g_zfs));
	usage(B_FALSE);
	}
	break;
	case 's':
	if (cmd_type != POOL_TRIM_START &&
	cmd_type != POOL_TRIM_SUSPEND) {
	(void) fprintf(stderr, gettext("-s cannot be "
	"combined with other options\n"));
	usage(B_FALSE);
	}
	cmd_type = POOL_TRIM_SUSPEND;
	break;
	case 'w':
	wait = B_TRUE;
	break;
	case '?':
	if (optopt != 0) {
	(void) fprintf(stderr,
	gettext("invalid option '%c'\n"), optopt);
	} else {
	(void) fprintf(stderr,
	gettext("invalid option '%s'\n"),
	argv[optind - 1]);
	}
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing pool name argument\n"));
	usage(B_FALSE);
	return (-1);
	}

	if (wait && (cmd_type != POOL_TRIM_START)) {
	(void) fprintf(stderr, gettext("-w cannot be used with -c or "
	"-s\n"));
	usage(B_FALSE);
	}

	char *poolname = argv[0];
	zpool_handle_t *zhp = zpool_open(g_zfs, poolname);
	if (zhp == NULL)
	return (-1);

	trimflags_t trim_flags = {
	.secure = secure,
	.rate = rate,
	.wait = wait,
	};

	nvlist_t *vdevs = fnvlist_alloc();
	if (argc == 1) {
	/* no individual leaf vdevs specified, so add them all */
	nvlist_t *config = zpool_get_config(zhp, NULL);
	nvlist_t *nvroot = fnvlist_lookup_nvlist(config,
	ZPOOL_CONFIG_VDEV_TREE);
	zpool_collect_leaves(zhp, nvroot, vdevs);
	trim_flags.fullpool = B_TRUE;
	} else {
	trim_flags.fullpool = B_FALSE;
	for (int i = 1; i < argc; i++) {
	fnvlist_add_boolean(vdevs, argv[i]);
	}
	}

	int error = zpool_trim(zhp, cmd_type, vdevs, &trim_flags);

	fnvlist_free(vdevs);
	zpool_close(zhp);

	return (error);
	}

	/*
	* Converts a total number of seconds to a human readable string broken
	* down in to days/hours/minutes/seconds.
	*/
	static void
	secs_to_dhms(uint64_t total, char *buf)
	{
	uint64_t days = total / 60 / 60 / 24;
	uint64_t hours = (total / 60 / 60) % 24;
	uint64_t mins = (total / 60) % 60;
	uint64_t secs = (total % 60);

	if (days > 0) {
	(void) sprintf(buf, "%llu days %02llu:%02llu:%02llu",
	(u_longlong_t)days, (u_longlong_t)hours,
	(u_longlong_t)mins, (u_longlong_t)secs);
	} else {
	(void) sprintf(buf, "%02llu:%02llu:%02llu",
	(u_longlong_t)hours, (u_longlong_t)mins,
	(u_longlong_t)secs);
	}
	}

	/*
	* Print out detailed scrub status.
	*/
	static void
	print_scan_scrub_resilver_status(pool_scan_stat_t *ps)
	{
	time_t start, end, pause;
	uint64_t pass_scanned, scanned, pass_issued, issued, total;
	uint64_t elapsed, scan_rate, issue_rate;
	double fraction_done;
	char processed_buf[7], scanned_buf[7], issued_buf[7], total_buf[7];
	char srate_buf[7], irate_buf[7], time_buf[32];

	printf(" ");
	printf_color(ANSI_BOLD, gettext("scan:"));
	printf(" ");

	/* If there's never been a scan, there's not much to say. */
	if (ps == NULL \|\| ps->pss_func == POOL_SCAN_NONE \|\|
	ps->pss_func >= POOL_SCAN_FUNCS) {
	(void) printf(gettext("none requested\n"));
	return;
	}

	start = ps->pss_start_time;
	end = ps->pss_end_time;
	pause = ps->pss_pass_scrub_pause;

	zfs_nicebytes(ps->pss_processed, processed_buf, sizeof (processed_buf));

	assert(ps->pss_func == POOL_SCAN_SCRUB \|\|
	ps->pss_func == POOL_SCAN_RESILVER);

	/* Scan is finished or canceled. */
	if (ps->pss_state == DSS_FINISHED) {
	secs_to_dhms(end - start, time_buf);

	if (ps->pss_func == POOL_SCAN_SCRUB) {
	(void) printf(gettext("scrub repaired %s "
	"in %s with %llu errors on %s"), processed_buf,
	time_buf, (u_longlong_t)ps->pss_errors,
	ctime(&end));
	} else if (ps->pss_func == POOL_SCAN_RESILVER) {
	(void) printf(gettext("resilvered %s "
	"in %s with %llu errors on %s"), processed_buf,
	time_buf, (u_longlong_t)ps->pss_errors,
	ctime(&end));
	}
	return;
	} else if (ps->pss_state == DSS_CANCELED) {
	if (ps->pss_func == POOL_SCAN_SCRUB) {
	(void) printf(gettext("scrub canceled on %s"),
	ctime(&end));
	} else if (ps->pss_func == POOL_SCAN_RESILVER) {
	(void) printf(gettext("resilver canceled on %s"),
	ctime(&end));
	}
	return;
	}

	assert(ps->pss_state == DSS_SCANNING);

	/* Scan is in progress. Resilvers can't be paused. */
	if (ps->pss_func == POOL_SCAN_SCRUB) {
	if (pause == 0) {
	(void) printf(gettext("scrub in progress since %s"),
	ctime(&start));
	} else {
	(void) printf(gettext("scrub paused since %s"),
	ctime(&pause));
	(void) printf(gettext("\tscrub started on %s"),
	ctime(&start));
	}
	} else if (ps->pss_func == POOL_SCAN_RESILVER) {
	(void) printf(gettext("resilver in progress since %s"),
	ctime(&start));
	}

	scanned = ps->pss_examined;
	pass_scanned = ps->pss_pass_exam;
	issued = ps->pss_issued;
	pass_issued = ps->pss_pass_issued;
	total = ps->pss_to_examine;

	/* we are only done with a block once we have issued the IO for it */
	fraction_done = (double)issued / total;

	/* elapsed time for this pass, rounding up to 1 if it's 0 */
	elapsed = time(NULL) - ps->pss_pass_start;
	elapsed -= ps->pss_pass_scrub_spent_paused;
	elapsed = (elapsed != 0) ? elapsed : 1;

	scan_rate = pass_scanned / elapsed;
	issue_rate = pass_issued / elapsed;
	uint64_t total_secs_left = (issue_rate != 0 && total >= issued) ?
	((total - issued) / issue_rate) : UINT64_MAX;
	secs_to_dhms(total_secs_left, time_buf);

	/* format all of the numbers we will be reporting */
	zfs_nicebytes(scanned, scanned_buf, sizeof (scanned_buf));
	zfs_nicebytes(issued, issued_buf, sizeof (issued_buf));
	zfs_nicebytes(total, total_buf, sizeof (total_buf));
	zfs_nicebytes(scan_rate, srate_buf, sizeof (srate_buf));
	zfs_nicebytes(issue_rate, irate_buf, sizeof (irate_buf));

	/* do not print estimated time if we have a paused scrub */
	if (pause == 0) {
	(void) printf(gettext("\t%s scanned at %s/s, "
	"%s issued at %s/s, %s total\n"),
	scanned_buf, srate_buf, issued_buf, irate_buf, total_buf);
	} else {
	(void) printf(gettext("\t%s scanned, %s issued, %s total\n"),
	scanned_buf, issued_buf, total_buf);
	}

	if (ps->pss_func == POOL_SCAN_RESILVER) {
	(void) printf(gettext("\t%s resilvered, %.2f%% done"),
	processed_buf, 100 * fraction_done);
	} else if (ps->pss_func == POOL_SCAN_SCRUB) {
	(void) printf(gettext("\t%s repaired, %.2f%% done"),
	processed_buf, 100 * fraction_done);
	}

	if (pause == 0) {
	if (total_secs_left != UINT64_MAX &&
	issue_rate >= 10 * 1024 * 1024) {
	(void) printf(gettext(", %s to go\n"), time_buf);
	} else {
	(void) printf(gettext(", no estimated "
	"completion time\n"));
	}
	} else {
	(void) printf(gettext("\n"));
	}
	}

	static void
	print_rebuild_status_impl(vdev_rebuild_stat_t vrs, char vdev_name)
	{
	if (vrs == NULL \|\| vrs->vrs_state == VDEV_REBUILD_NONE)
	return;

	printf(" ");
	printf_color(ANSI_BOLD, gettext("scan:"));
	printf(" ");

	uint64_t bytes_scanned = vrs->vrs_bytes_scanned;
	uint64_t bytes_issued = vrs->vrs_bytes_issued;
	uint64_t bytes_rebuilt = vrs->vrs_bytes_rebuilt;
	uint64_t bytes_est = vrs->vrs_bytes_est;
	uint64_t scan_rate = (vrs->vrs_pass_bytes_scanned /
	(vrs->vrs_pass_time_ms + 1)) * 1000;
	uint64_t issue_rate = (vrs->vrs_pass_bytes_issued /
	(vrs->vrs_pass_time_ms + 1)) * 1000;
	double scan_pct = MIN((double)bytes_scanned * 100 /
	(bytes_est + 1), 100);

	/* Format all of the numbers we will be reporting */
	char bytes_scanned_buf[7], bytes_issued_buf[7];
	char bytes_rebuilt_buf[7], bytes_est_buf[7];
	char scan_rate_buf[7], issue_rate_buf[7], time_buf[32];
	zfs_nicebytes(bytes_scanned, bytes_scanned_buf,
	sizeof (bytes_scanned_buf));
	zfs_nicebytes(bytes_issued, bytes_issued_buf,
	sizeof (bytes_issued_buf));
	zfs_nicebytes(bytes_rebuilt, bytes_rebuilt_buf,
	sizeof (bytes_rebuilt_buf));
	zfs_nicebytes(bytes_est, bytes_est_buf, sizeof (bytes_est_buf));
	zfs_nicebytes(scan_rate, scan_rate_buf, sizeof (scan_rate_buf));
	zfs_nicebytes(issue_rate, issue_rate_buf, sizeof (issue_rate_buf));

	time_t start = vrs->vrs_start_time;
	time_t end = vrs->vrs_end_time;

	/* Rebuild is finished or canceled. */
	if (vrs->vrs_state == VDEV_REBUILD_COMPLETE) {
	secs_to_dhms(vrs->vrs_scan_time_ms / 1000, time_buf);
	(void) printf(gettext("resilvered (%s) %s in %s "
	"with %llu errors on %s"), vdev_name, bytes_rebuilt_buf,
	time_buf, (u_longlong_t)vrs->vrs_errors, ctime(&end));
	return;
	} else if (vrs->vrs_state == VDEV_REBUILD_CANCELED) {
	(void) printf(gettext("resilver (%s) canceled on %s"),
	vdev_name, ctime(&end));
	return;
	} else if (vrs->vrs_state == VDEV_REBUILD_ACTIVE) {
	(void) printf(gettext("resilver (%s) in progress since %s"),
	vdev_name, ctime(&start));
	}

	assert(vrs->vrs_state == VDEV_REBUILD_ACTIVE);

	secs_to_dhms(MAX((int64_t)bytes_est - (int64_t)bytes_scanned, 0) /
	MAX(scan_rate, 1), time_buf);

	(void) printf(gettext("\t%s scanned at %s/s, %s issued %s/s, "
	"%s total\n"), bytes_scanned_buf, scan_rate_buf,
	bytes_issued_buf, issue_rate_buf, bytes_est_buf);
	(void) printf(gettext("\t%s resilvered, %.2f%% done"),
	bytes_rebuilt_buf, scan_pct);

	if (vrs->vrs_state == VDEV_REBUILD_ACTIVE) {
	if (scan_rate >= 10 * 1024 * 1024) {
	(void) printf(gettext(", %s to go\n"), time_buf);
	} else {
	(void) printf(gettext(", no estimated "
	"completion time\n"));
	}
	} else {
	(void) printf(gettext("\n"));
	}
	}

	/*
	* Print rebuild status for top-level vdevs.
	*/
	static void
	print_rebuild_status(zpool_handle_t zhp, nvlist_t nvroot)
	{
	nvlist_t **child;
	uint_t children;

	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0)
	children = 0;

	for (uint_t c = 0; c < children; c++) {
	vdev_rebuild_stat_t *vrs;
	uint_t i;

	if (nvlist_lookup_uint64_array(child[c],
	ZPOOL_CONFIG_REBUILD_STATS, (uint64_t **)&vrs, &i) == 0) {
	char *name = zpool_vdev_name(g_zfs, zhp,
	child[c], VDEV_NAME_TYPE_ID);
	print_rebuild_status_impl(vrs, name);
	free(name);
	}
	}
	}

	/*
	* As we don't scrub checkpointed blocks, we want to warn the user that we
	* skipped scanning some blocks if a checkpoint exists or existed at any
	* time during the scan. If a sequential instead of healing reconstruction
	* was performed then the blocks were reconstructed. However, their checksums
	* have not been verified so we still print the warning.
	*/
	static void
	print_checkpoint_scan_warning(pool_scan_stat_t ps, pool_checkpoint_stat_t pcs)
	{
	if (ps == NULL \|\| pcs == NULL)
	return;

	if (pcs->pcs_state == CS_NONE \|\|
	pcs->pcs_state == CS_CHECKPOINT_DISCARDING)
	return;

	assert(pcs->pcs_state == CS_CHECKPOINT_EXISTS);

	if (ps->pss_state == DSS_NONE)
	return;

	if ((ps->pss_state == DSS_FINISHED \|\| ps->pss_state == DSS_CANCELED) &&
	ps->pss_end_time < pcs->pcs_start_time)
	return;

	if (ps->pss_state == DSS_FINISHED \|\| ps->pss_state == DSS_CANCELED) {
	(void) printf(gettext(" scan warning: skipped blocks "
	"that are only referenced by the checkpoint.\n"));
	} else {
	assert(ps->pss_state == DSS_SCANNING);
	(void) printf(gettext(" scan warning: skipping blocks "
	"that are only referenced by the checkpoint.\n"));
	}
	}

	/*
	* Returns B_TRUE if there is an active rebuild in progress. Otherwise,
	* B_FALSE is returned and 'rebuild_end_time' is set to the end time for
	* the last completed (or cancelled) rebuild.
	*/
	static boolean_t
	check_rebuilding(nvlist_t nvroot, uint64_t rebuild_end_time)
	{
	nvlist_t **child;
	uint_t children;
	boolean_t rebuilding = B_FALSE;
	uint64_t end_time = 0;

	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0)
	children = 0;

	for (uint_t c = 0; c < children; c++) {
	vdev_rebuild_stat_t *vrs;
	uint_t i;

	if (nvlist_lookup_uint64_array(child[c],
	ZPOOL_CONFIG_REBUILD_STATS, (uint64_t **)&vrs, &i) == 0) {

	if (vrs->vrs_end_time > end_time)
	end_time = vrs->vrs_end_time;

	if (vrs->vrs_state == VDEV_REBUILD_ACTIVE) {
	rebuilding = B_TRUE;
	end_time = 0;
	break;
	}
	}
	}

	if (rebuild_end_time != NULL)
	*rebuild_end_time = end_time;

	return (rebuilding);
	}

	/*
	* Print the scan status.
	*/
	static void
	print_scan_status(zpool_handle_t zhp, nvlist_t nvroot)
	{
	uint64_t rebuild_end_time = 0, resilver_end_time = 0;
	boolean_t have_resilver = B_FALSE, have_scrub = B_FALSE;
	boolean_t active_resilver = B_FALSE;
	pool_checkpoint_stat_t *pcs = NULL;
	pool_scan_stat_t *ps = NULL;
	uint_t c;

	if (nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_SCAN_STATS,
	(uint64_t **)&ps, &c) == 0) {
	if (ps->pss_func == POOL_SCAN_RESILVER) {
	resilver_end_time = ps->pss_end_time;
	active_resilver = (ps->pss_state == DSS_SCANNING);
	}

	have_resilver = (ps->pss_func == POOL_SCAN_RESILVER);
	have_scrub = (ps->pss_func == POOL_SCAN_SCRUB);
	}

	boolean_t active_rebuild = check_rebuilding(nvroot, &rebuild_end_time);
	boolean_t have_rebuild = (active_rebuild \|\| (rebuild_end_time > 0));

	/* Always print the scrub status when available. */
	if (have_scrub)
	print_scan_scrub_resilver_status(ps);

	/*
	* When there is an active resilver or rebuild print its status.
	* Otherwise print the status of the last resilver or rebuild.
	*/
	if (active_resilver \|\| (!active_rebuild && have_resilver &&
	resilver_end_time && resilver_end_time > rebuild_end_time)) {
	print_scan_scrub_resilver_status(ps);
	} else if (active_rebuild \|\| (!active_resilver && have_rebuild &&
	rebuild_end_time && rebuild_end_time > resilver_end_time)) {
	print_rebuild_status(zhp, nvroot);
	}

	(void) nvlist_lookup_uint64_array(nvroot,
	ZPOOL_CONFIG_CHECKPOINT_STATS, (uint64_t **)&pcs, &c);
	print_checkpoint_scan_warning(ps, pcs);
	}

	/*
	* Print out detailed removal status.
	*/
	static void
	print_removal_status(zpool_handle_t zhp, pool_removal_stat_t prs)
	{
	char copied_buf[7], examined_buf[7], total_buf[7], rate_buf[7];
	time_t start, end;
	nvlist_t config, nvroot;
	nvlist_t **child;
	uint_t children;
	char *vdev_name;

	if (prs == NULL \|\| prs->prs_state == DSS_NONE)
	return;

	/*
	* Determine name of vdev.
	*/
	config = zpool_get_config(zhp, NULL);
	nvroot = fnvlist_lookup_nvlist(config,
	ZPOOL_CONFIG_VDEV_TREE);
	verify(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	&child, &children) == 0);
	assert(prs->prs_removing_vdev < children);
	vdev_name = zpool_vdev_name(g_zfs, zhp,
	child[prs->prs_removing_vdev], B_TRUE);

	printf_color(ANSI_BOLD, gettext("remove: "));

	start = prs->prs_start_time;
	end = prs->prs_end_time;
	zfs_nicenum(prs->prs_copied, copied_buf, sizeof (copied_buf));

	/*
	* Removal is finished or canceled.
	*/
	if (prs->prs_state == DSS_FINISHED) {
	uint64_t minutes_taken = (end - start) / 60;

	(void) printf(gettext("Removal of vdev %llu copied %s "
	"in %lluh%um, completed on %s"),
	(longlong_t)prs->prs_removing_vdev,
	copied_buf,
	(u_longlong_t)(minutes_taken / 60),
	(uint_t)(minutes_taken % 60),
	ctime((time_t *)&end));
	} else if (prs->prs_state == DSS_CANCELED) {
	(void) printf(gettext("Removal of %s canceled on %s"),
	vdev_name, ctime(&end));
	} else {
	uint64_t copied, total, elapsed, mins_left, hours_left;
	double fraction_done;
	uint_t rate;

	assert(prs->prs_state == DSS_SCANNING);

	/*
	* Removal is in progress.
	*/
	(void) printf(gettext(
	"Evacuation of %s in progress since %s"),
	vdev_name, ctime(&start));

	copied = prs->prs_copied > 0 ? prs->prs_copied : 1;
	total = prs->prs_to_copy;
	fraction_done = (double)copied / total;

	/* elapsed time for this pass */
	elapsed = time(NULL) - prs->prs_start_time;
	elapsed = elapsed > 0 ? elapsed : 1;
	rate = copied / elapsed;
	rate = rate > 0 ? rate : 1;
	mins_left = ((total - copied) / rate) / 60;
	hours_left = mins_left / 60;

	zfs_nicenum(copied, examined_buf, sizeof (examined_buf));
	zfs_nicenum(total, total_buf, sizeof (total_buf));
	zfs_nicenum(rate, rate_buf, sizeof (rate_buf));

	/*
	* do not print estimated time if hours_left is more than
	* 30 days
	*/
	(void) printf(gettext(
	"\t%s copied out of %s at %s/s, %.2f%% done"),
	examined_buf, total_buf, rate_buf, 100 * fraction_done);
	if (hours_left < (30 * 24)) {
	(void) printf(gettext(", %lluh%um to go\n"),
	(u_longlong_t)hours_left, (uint_t)(mins_left % 60));
	} else {
	(void) printf(gettext(
	", (copy is slow, no estimated time)\n"));
	}
	}
	free(vdev_name);

	if (prs->prs_mapping_memory > 0) {
	char mem_buf[7];
	zfs_nicenum(prs->prs_mapping_memory, mem_buf, sizeof (mem_buf));
	(void) printf(gettext(
	"\t%s memory used for removed device mappings\n"),
	mem_buf);
	}
	}

	static void
	print_checkpoint_status(pool_checkpoint_stat_t *pcs)
	{
	time_t start;
	char space_buf[7];

	if (pcs == NULL \|\| pcs->pcs_state == CS_NONE)
	return;

	(void) printf(gettext("checkpoint: "));

	start = pcs->pcs_start_time;
	zfs_nicenum(pcs->pcs_space, space_buf, sizeof (space_buf));

	if (pcs->pcs_state == CS_CHECKPOINT_EXISTS) {
	char *date = ctime(&start);

	/*
	* ctime() adds a newline at the end of the generated
	* string, thus the weird format specifier and the
	* strlen() call used to chop it off from the output.
	*/
	(void) printf(gettext("created %.*s, consumes %s\n"),
	(int)(strlen(date) - 1), date, space_buf);
	return;
	}

	assert(pcs->pcs_state == CS_CHECKPOINT_DISCARDING);

	(void) printf(gettext("discarding, %s remaining.\n"),
	space_buf);
	}

	static void
	print_error_log(zpool_handle_t *zhp)
	{
	nvlist_t *nverrlist = NULL;
	nvpair_t *elem;
	char *pathname;
	size_t len = MAXPATHLEN * 2;

	if (zpool_get_errlog(zhp, &nverrlist) != 0)
	return;

	(void) printf("errors: Permanent errors have been "
	"detected in the following files:\n\n");

	pathname = safe_malloc(len);
	elem = NULL;
	while ((elem = nvlist_next_nvpair(nverrlist, elem)) != NULL) {
	nvlist_t *nv;
	uint64_t dsobj, obj;

	verify(nvpair_value_nvlist(elem, &nv) == 0);
	verify(nvlist_lookup_uint64(nv, ZPOOL_ERR_DATASET,
	&dsobj) == 0);
	verify(nvlist_lookup_uint64(nv, ZPOOL_ERR_OBJECT,
	&obj) == 0);
	zpool_obj_to_path(zhp, dsobj, obj, pathname, len);
	(void) printf("%7s %s\n", "", pathname);
	}
	free(pathname);
	nvlist_free(nverrlist);
	}

	static void
	print_spares(zpool_handle_t zhp, status_cbdata_t cb, nvlist_t **spares,
	uint_t nspares)
	{
	uint_t i;
	char *name;

	if (nspares == 0)
	return;

	(void) printf(gettext("\tspares\n"));

	for (i = 0; i < nspares; i++) {
	name = zpool_vdev_name(g_zfs, zhp, spares[i],
	cb->cb_name_flags);
	print_status_config(zhp, cb, name, spares[i], 2, B_TRUE, NULL);
	free(name);
	}
	}

	static void
	print_l2cache(zpool_handle_t zhp, status_cbdata_t cb, nvlist_t **l2cache,
	uint_t nl2cache)
	{
	uint_t i;
	char *name;

	if (nl2cache == 0)
	return;

	(void) printf(gettext("\tcache\n"));

	for (i = 0; i < nl2cache; i++) {
	name = zpool_vdev_name(g_zfs, zhp, l2cache[i],
	cb->cb_name_flags);
	print_status_config(zhp, cb, name, l2cache[i], 2,
	B_FALSE, NULL);
	free(name);
	}
	}

	static void
	print_dedup_stats(nvlist_t *config)
	{
	ddt_histogram_t *ddh;
	ddt_stat_t *dds;
	ddt_object_t *ddo;
	uint_t c;
	char dspace[6], mspace[6];

	/*
	* If the pool was faulted then we may not have been able to
	* obtain the config. Otherwise, if we have anything in the dedup
	* table continue processing the stats.
	*/
	if (nvlist_lookup_uint64_array(config, ZPOOL_CONFIG_DDT_OBJ_STATS,
	(uint64_t **)&ddo, &c) != 0)
	return;

	(void) printf("\n");
	(void) printf(gettext(" dedup: "));
	if (ddo->ddo_count == 0) {
	(void) printf(gettext("no DDT entries\n"));
	return;
	}

	zfs_nicebytes(ddo->ddo_dspace, dspace, sizeof (dspace));
	zfs_nicebytes(ddo->ddo_mspace, mspace, sizeof (mspace));
	(void) printf("DDT entries %llu, size %s on disk, %s in core\n",
	(u_longlong_t)ddo->ddo_count,
	dspace,
	mspace);

	verify(nvlist_lookup_uint64_array(config, ZPOOL_CONFIG_DDT_STATS,
	(uint64_t **)&dds, &c) == 0);
	verify(nvlist_lookup_uint64_array(config, ZPOOL_CONFIG_DDT_HISTOGRAM,
	(uint64_t **)&ddh, &c) == 0);
	zpool_dump_ddt(dds, ddh);
	}

	/*
	* Display a summary of pool status. Displays a summary such as:
	*
	* pool: tank
	* status: DEGRADED
	* reason: One or more devices ...
	* see: https://openzfs.github.io/openzfs-docs/msg/ZFS-xxxx-01
	* config:
	* mirror DEGRADED
	* c1t0d0 OK
	* c2t0d0 UNAVAIL
	*
	* When given the '-v' option, we print out the complete config. If the '-e'
	* option is specified, then we print out error rate information as well.
	*/
	static int
	status_callback(zpool_handle_t zhp, void data)
	{
	status_cbdata_t *cbp = data;
	nvlist_t config, nvroot;
	char *msgid;
	zpool_status_t reason;
	zpool_errata_t errata;
	const char *health;
	uint_t c;
	vdev_stat_t *vs;

	config = zpool_get_config(zhp, NULL);
	reason = zpool_get_status(zhp, &msgid, &errata);

	cbp->cb_count++;

	/*
	* If we were given 'zpool status -x', only report those pools with
	* problems.
	*/
	if (cbp->cb_explain &&
	(reason == ZPOOL_STATUS_OK \|\|
	reason == ZPOOL_STATUS_VERSION_OLDER \|\|
	reason == ZPOOL_STATUS_FEAT_DISABLED \|\|
	reason == ZPOOL_STATUS_COMPATIBILITY_ERR \|\|
	reason == ZPOOL_STATUS_INCOMPATIBLE_FEAT)) {
	if (!cbp->cb_allpools) {
	(void) printf(gettext("pool '%s' is healthy\n"),
	zpool_get_name(zhp));
	if (cbp->cb_first)
	cbp->cb_first = B_FALSE;
	}
	return (0);
	}

	if (cbp->cb_first)
	cbp->cb_first = B_FALSE;
	else
	(void) printf("\n");

	nvroot = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE);
	verify(nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_VDEV_STATS,
	(uint64_t **)&vs, &c) == 0);

	health = zpool_get_state_str(zhp);

	printf(" ");
	printf_color(ANSI_BOLD, gettext("pool:"));
	printf(" %s\n", zpool_get_name(zhp));
	printf(" ");
	printf_color(ANSI_BOLD, gettext("state: "));

	printf_color(health_str_to_color(health), "%s", health);

	printf("\n");

	switch (reason) {
	case ZPOOL_STATUS_MISSING_DEV_R:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices could "
	"not be opened. Sufficient replicas exist for\n\tthe pool "
	"to continue functioning in a degraded state.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Attach the missing device "
	"and online it using 'zpool online'.\n"));
	break;

	case ZPOOL_STATUS_MISSING_DEV_NR:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices could "
	"not be opened. There are insufficient\n\treplicas for the"
	" pool to continue functioning.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Attach the missing device "
	"and online it using 'zpool online'.\n"));
	break;

	case ZPOOL_STATUS_CORRUPT_LABEL_R:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices could "
	"not be used because the label is missing or\n\tinvalid. "
	"Sufficient replicas exist for the pool to continue\n\t"
	"functioning in a degraded state.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Replace the device using "
	"'zpool replace'.\n"));
	break;

	case ZPOOL_STATUS_CORRUPT_LABEL_NR:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices could "
	"not be used because the label is missing \n\tor invalid. "
	"There are insufficient replicas for the pool to "
	"continue\n\tfunctioning.\n"));
	zpool_explain_recover(zpool_get_handle(zhp),
	zpool_get_name(zhp), reason, config);
	break;

	case ZPOOL_STATUS_FAILING_DEV:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices has "
	"experienced an unrecoverable error. An\n\tattempt was "
	"made to correct the error. Applications are "
	"unaffected.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Determine if the "
	"device needs to be replaced, and clear the errors\n\tusing"
	" 'zpool clear' or replace the device with 'zpool "
	"replace'.\n"));
	break;

	case ZPOOL_STATUS_OFFLINE_DEV:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices has "
	"been taken offline by the administrator.\n\tSufficient "
	"replicas exist for the pool to continue functioning in "
	"a\n\tdegraded state.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Online the device "
	"using 'zpool online' or replace the device with\n\t'zpool "
	"replace'.\n"));
	break;

	case ZPOOL_STATUS_REMOVED_DEV:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices has "
	"been removed by the administrator.\n\tSufficient "
	"replicas exist for the pool to continue functioning in "
	"a\n\tdegraded state.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Online the device "
	"using zpool online' or replace the device with\n\t'zpool "
	"replace'.\n"));
	break;

	case ZPOOL_STATUS_RESILVERING:
	case ZPOOL_STATUS_REBUILDING:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices is "
	"currently being resilvered. The pool will\n\tcontinue "
	"to function, possibly in a degraded state.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Wait for the resilver to "
	"complete.\n"));
	break;

	case ZPOOL_STATUS_REBUILD_SCRUB:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices have "
	"been sequentially resilvered, scrubbing\n\tthe pool "
	"is recommended.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Use 'zpool scrub' to "
	"verify all data checksums.\n"));
	break;

	case ZPOOL_STATUS_CORRUPT_DATA:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices has "
	"experienced an error resulting in data\n\tcorruption. "
	"Applications may be affected.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Restore the file in question"
	" if possible. Otherwise restore the\n\tentire pool from "
	"backup.\n"));
	break;

	case ZPOOL_STATUS_CORRUPT_POOL:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool metadata is "
	"corrupted and the pool cannot be opened.\n"));
	zpool_explain_recover(zpool_get_handle(zhp),
	zpool_get_name(zhp), reason, config);
	break;

	case ZPOOL_STATUS_VERSION_OLDER:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool is formatted using "
	"a legacy on-disk format. The pool can\n\tstill be used, "
	"but some features are unavailable.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Upgrade the pool using "
	"'zpool upgrade'. Once this is done, the\n\tpool will no "
	"longer be accessible on software that does not support\n\t"
	"feature flags.\n"));
	break;

	case ZPOOL_STATUS_VERSION_NEWER:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool has been upgraded "
	"to a newer, incompatible on-disk version.\n\tThe pool "
	"cannot be accessed on this system.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Access the pool from a "
	"system running more recent software, or\n\trestore the "
	"pool from backup.\n"));
	break;

	case ZPOOL_STATUS_FEAT_DISABLED:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("Some supported and "
	"requested features are not enabled on the pool.\n\t"
	"The pool can still be used, but some features are "
	"unavailable.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Enable all features using "
	"'zpool upgrade'. Once this is done,\n\tthe pool may no "
	"longer be accessible by software that does not support\n\t"
	"the features. See zpool-features(7) for details.\n"));
	break;

	case ZPOOL_STATUS_COMPATIBILITY_ERR:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("This pool has a "
	"compatibility list specified, but it could not be\n\t"
	"read/parsed at this time. The pool can still be used, "
	"but this\n\tshould be investigated.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Check the value of the "
	"'compatibility' property against the\n\t"
	"appropriate file in " ZPOOL_SYSCONF_COMPAT_D " or "
	ZPOOL_DATA_COMPAT_D ".\n"));
	break;

	case ZPOOL_STATUS_INCOMPATIBLE_FEAT:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more features "
	"are enabled on the pool despite not being\n\t"
	"requested by the 'compatibility' property.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Consider setting "
	"'compatibility' to an appropriate value, or\n\t"
	"adding needed features to the relevant file in\n\t"
	ZPOOL_SYSCONF_COMPAT_D " or " ZPOOL_DATA_COMPAT_D ".\n"));
	break;

	case ZPOOL_STATUS_UNSUP_FEAT_READ:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool cannot be accessed "
	"on this system because it uses the\n\tfollowing feature(s)"
	" not supported on this system:\n"));
	zpool_print_unsup_feat(config);
	(void) printf("\n");
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Access the pool from a "
	"system that supports the required feature(s),\n\tor "
	"restore the pool from backup.\n"));
	break;

	case ZPOOL_STATUS_UNSUP_FEAT_WRITE:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool can only be "
	"accessed in read-only mode on this system. It\n\tcannot be"
	" accessed in read-write mode because it uses the "
	"following\n\tfeature(s) not supported on this system:\n"));
	zpool_print_unsup_feat(config);
	(void) printf("\n");
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("The pool cannot be accessed "
	"in read-write mode. Import the pool with\n"
	"\t\"-o readonly=on\", access the pool from a system that "
	"supports the\n\trequired feature(s), or restore the "
	"pool from backup.\n"));
	break;

	case ZPOOL_STATUS_FAULTED_DEV_R:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices are "
	"faulted in response to persistent errors.\n\tSufficient "
	"replicas exist for the pool to continue functioning "
	"in a\n\tdegraded state.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Replace the faulted device, "
	"or use 'zpool clear' to mark the device\n\trepaired.\n"));
	break;

	case ZPOOL_STATUS_FAULTED_DEV_NR:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices are "
	"faulted in response to persistent errors. There are "
	"insufficient replicas for the pool to\n\tcontinue "
	"functioning.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Destroy and re-create the "
	"pool from a backup source. Manually marking the device\n"
	"\trepaired using 'zpool clear' may allow some data "
	"to be recovered.\n"));
	break;

	case ZPOOL_STATUS_IO_FAILURE_MMP:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("The pool is suspended "
	"because multihost writes failed or were delayed;\n\t"
	"another system could import the pool undetected.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Make sure the pool's devices"
	" are connected, then reboot your system and\n\timport the "
	"pool.\n"));
	break;

	case ZPOOL_STATUS_IO_FAILURE_WAIT:
	case ZPOOL_STATUS_IO_FAILURE_CONTINUE:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("One or more devices are "
	"faulted in response to IO failures.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Make sure the affected "
	"devices are connected, then run 'zpool clear'.\n"));
	break;

	case ZPOOL_STATUS_BAD_LOG:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("An intent log record "
	"could not be read.\n"
	"\tWaiting for administrator intervention to fix the "
	"faulted pool.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Either restore the affected "
	"device(s) and run 'zpool online',\n"
	"\tor ignore the intent log records by running "
	"'zpool clear'.\n"));
	break;

	case ZPOOL_STATUS_NON_NATIVE_ASHIFT:
	(void) printf(gettext("status: One or more devices are "
	"configured to use a non-native block size.\n"
	"\tExpect reduced performance.\n"));
	(void) printf(gettext("action: Replace affected devices with "
	"devices that support the\n\tconfigured block size, or "
	"migrate data to a properly configured\n\tpool.\n"));
	break;

	case ZPOOL_STATUS_HOSTID_MISMATCH:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("Mismatch between pool hostid"
	" and system hostid on imported pool.\n\tThis pool was "
	"previously imported into a system with a different "
	"hostid,\n\tand then was verbatim imported into this "
	"system.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("Export this pool on all "
	"systems on which it is imported.\n"
	"\tThen import it to correct the mismatch.\n"));
	break;

	case ZPOOL_STATUS_ERRATA:
	printf_color(ANSI_BOLD, gettext("status: "));
	printf_color(ANSI_YELLOW, gettext("Errata #%d detected.\n"),
	errata);

	switch (errata) {
	case ZPOOL_ERRATA_NONE:
	break;

	case ZPOOL_ERRATA_ZOL_2094_SCRUB:
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("To correct the issue"
	" run 'zpool scrub'.\n"));
	break;

	case ZPOOL_ERRATA_ZOL_6845_ENCRYPTION:
	(void) printf(gettext("\tExisting encrypted datasets "
	"contain an on-disk incompatibility\n\twhich "
	"needs to be corrected.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("To correct the issue"
	" backup existing encrypted datasets to new\n\t"
	"encrypted datasets and destroy the old ones. "
	"'zfs mount -o ro' can\n\tbe used to temporarily "
	"mount existing encrypted datasets readonly.\n"));
	break;

	case ZPOOL_ERRATA_ZOL_8308_ENCRYPTION:
	(void) printf(gettext("\tExisting encrypted snapshots "
	"and bookmarks contain an on-disk\n\tincompat"
	"ibility. This may cause on-disk corruption if "
	"they are used\n\twith 'zfs recv'.\n"));
	printf_color(ANSI_BOLD, gettext("action: "));
	printf_color(ANSI_YELLOW, gettext("To correct the"
	"issue, enable the bookmark_v2 feature. No "
	"additional\n\taction is needed if there are no "
	"encrypted snapshots or bookmarks.\n\tIf preserving"
	"the encrypted snapshots and bookmarks is required,"
	" use\n\ta non-raw send to backup and restore them."
	" Alternately, they may be\n\tremoved to resolve "
	"the incompatibility.\n"));
	break;

	default:
	/*
	* All errata which allow the pool to be imported
	* must contain an action message.
	*/
	assert(0);
	}
	break;

	default:
	/*
	* The remaining errors can't actually be generated, yet.
	*/
	assert(reason == ZPOOL_STATUS_OK);
	}

	if (msgid != NULL) {
	printf(" ");
	printf_color(ANSI_BOLD, gettext("see:"));
	printf(gettext(
	" https://openzfs.github.io/openzfs-docs/msg/%s\n"),
	msgid);
	}

	if (config != NULL) {
	uint64_t nerr;
	nvlist_t spares, l2cache;
	uint_t nspares, nl2cache;
	pool_checkpoint_stat_t *pcs = NULL;
	pool_removal_stat_t *prs = NULL;

	print_scan_status(zhp, nvroot);

	(void) nvlist_lookup_uint64_array(nvroot,
	ZPOOL_CONFIG_REMOVAL_STATS, (uint64_t **)&prs, &c);
	print_removal_status(zhp, prs);

	(void) nvlist_lookup_uint64_array(nvroot,
	ZPOOL_CONFIG_CHECKPOINT_STATS, (uint64_t **)&pcs, &c);
	print_checkpoint_status(pcs);

	cbp->cb_namewidth = max_width(zhp, nvroot, 0, 0,
	cbp->cb_name_flags \| VDEV_NAME_TYPE_ID);
	if (cbp->cb_namewidth < 10)
	cbp->cb_namewidth = 10;

	color_start(ANSI_BOLD);
	(void) printf(gettext("config:\n\n"));
	(void) printf(gettext("\t%-*s %-8s %5s %5s %5s"),
	cbp->cb_namewidth, "NAME", "STATE", "READ", "WRITE",
	"CKSUM");
	color_end();

	if (cbp->cb_print_slow_ios) {
	printf_color(ANSI_BOLD, " %5s", gettext("SLOW"));
	}

	if (cbp->vcdl != NULL)
	print_cmd_columns(cbp->vcdl, 0);

	printf("\n");

	print_status_config(zhp, cbp, zpool_get_name(zhp), nvroot, 0,
	B_FALSE, NULL);

	print_class_vdevs(zhp, cbp, nvroot, VDEV_ALLOC_BIAS_DEDUP);
	print_class_vdevs(zhp, cbp, nvroot, VDEV_ALLOC_BIAS_SPECIAL);
	print_class_vdevs(zhp, cbp, nvroot, VDEV_ALLOC_CLASS_LOGS);

	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
	&l2cache, &nl2cache) == 0)
	print_l2cache(zhp, cbp, l2cache, nl2cache);

	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	&spares, &nspares) == 0)
	print_spares(zhp, cbp, spares, nspares);

	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_ERRCOUNT,
	&nerr) == 0) {
	nvlist_t *nverrlist = NULL;

	/*
	* If the approximate error count is small, get a
	* precise count by fetching the entire log and
	* uniquifying the results.
	*/
	if (nerr > 0 && nerr < 100 && !cbp->cb_verbose &&
	zpool_get_errlog(zhp, &nverrlist) == 0) {
	nvpair_t *elem;

	elem = NULL;
	nerr = 0;
	while ((elem = nvlist_next_nvpair(nverrlist,
	elem)) != NULL) {
	nerr++;
	}
	}
	nvlist_free(nverrlist);

	(void) printf("\n");

	if (nerr == 0)
	(void) printf(gettext("errors: No known data "
	"errors\n"));
	else if (!cbp->cb_verbose)
	(void) printf(gettext("errors: %llu data "
	"errors, use '-v' for a list\n"),
	(u_longlong_t)nerr);
	else
	print_error_log(zhp);
	}

	if (cbp->cb_dedup_stats)
	print_dedup_stats(config);
	} else {
	(void) printf(gettext("config: The configuration cannot be "
	"determined.\n"));
	}

	return (0);
	}

	/*
	* zpool status [-c [script1,script2,...]] [-igLpPstvx] [-T d\|u] [pool] ...
	* [interval [count]]
	*
	* -c CMD For each vdev, run command CMD
	* -i Display vdev initialization status.
	* -g Display guid for individual vdev name.
	* -L Follow links when resolving vdev path name.
	* -p Display values in parsable (exact) format.
	* -P Display full path for vdev name.
	* -s Display slow IOs column.
	* -v Display complete error logs
	* -x Display only pools with potential problems
	* -D Display dedup status (undocumented)
	* -t Display vdev TRIM status.
	* -T Display a timestamp in date(1) or Unix format
	*
	* Describes the health status of all pools or some subset.
	*/
	int
	zpool_do_status(int argc, char **argv)
	{
	int c;
	int ret;
	float interval = 0;
	unsigned long count = 0;
	status_cbdata_t cb = { 0 };
	char *cmd = NULL;

	/* check options */
	while ((c = getopt(argc, argv, "c:igLpPsvxDtT:")) != -1) {
	switch (c) {
	case 'c':
	if (cmd != NULL) {
	fprintf(stderr,
	gettext("Can't set -c flag twice\n"));
	exit(1);
	}

	if (getenv("ZPOOL_SCRIPTS_ENABLED") != NULL &&
	!libzfs_envvar_is_set("ZPOOL_SCRIPTS_ENABLED")) {
	fprintf(stderr, gettext(
	"Can't run -c, disabled by "
	"ZPOOL_SCRIPTS_ENABLED.\n"));
	exit(1);
	}

	if ((getuid() <= 0 \|\| geteuid() <= 0) &&
	!libzfs_envvar_is_set("ZPOOL_SCRIPTS_AS_ROOT")) {
	fprintf(stderr, gettext(
	"Can't run -c with root privileges "
	"unless ZPOOL_SCRIPTS_AS_ROOT is set.\n"));
	exit(1);
	}
	cmd = optarg;
	break;
	case 'i':
	cb.cb_print_vdev_init = B_TRUE;
	break;
	case 'g':
	cb.cb_name_flags \|= VDEV_NAME_GUID;
	break;
	case 'L':
	cb.cb_name_flags \|= VDEV_NAME_FOLLOW_LINKS;
	break;
	case 'p':
	cb.cb_literal = B_TRUE;
	break;
	case 'P':
	cb.cb_name_flags \|= VDEV_NAME_PATH;
	break;
	case 's':
	cb.cb_print_slow_ios = B_TRUE;
	break;
	case 'v':
	cb.cb_verbose = B_TRUE;
	break;
	case 'x':
	cb.cb_explain = B_TRUE;
	break;
	case 'D':
	cb.cb_dedup_stats = B_TRUE;
	break;
	case 't':
	cb.cb_print_vdev_trim = B_TRUE;
	break;
	case 'T':
	get_timestamp_arg(*optarg);
	break;
	case '?':
	if (optopt == 'c') {
	print_zpool_script_list("status");
	exit(0);
	} else {
	fprintf(stderr,
	gettext("invalid option '%c'\n"), optopt);
	}
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	get_interval_count(&argc, argv, &interval, &count);

	if (argc == 0)
	cb.cb_allpools = B_TRUE;

	cb.cb_first = B_TRUE;
	cb.cb_print_status = B_TRUE;

	for (;;) {
	if (timestamp_fmt != NODATE)
	print_timestamp(timestamp_fmt);

	if (cmd != NULL)
	cb.vcdl = all_pools_for_each_vdev_run(argc, argv, cmd,
	NULL, NULL, 0, 0);

	- ret = for_each_pool(argc, argv, B_TRUE, NULL, cb.cb_literal,
	- status_callback, &cb);
	+ ret = for_each_pool(argc, argv, B_TRUE, NULL, ZFS_TYPE_POOL,
	+ cb.cb_literal, status_callback, &cb);

	if (cb.vcdl != NULL)
	free_vdev_cmd_data_list(cb.vcdl);

	if (argc == 0 && cb.cb_count == 0)
	(void) fprintf(stderr, gettext("no pools available\n"));
	else if (cb.cb_explain && cb.cb_first && cb.cb_allpools)
	(void) printf(gettext("all pools are healthy\n"));

	if (ret != 0)
	return (ret);

	if (interval == 0)
	break;

	if (count != 0 && --count == 0)
	break;

	(void) fsleep(interval);
	}

	return (0);
	}

	typedef struct upgrade_cbdata {
	int cb_first;
	int cb_argc;
	uint64_t cb_version;
	char **cb_argv;
	} upgrade_cbdata_t;

	static int
	check_unsupp_fs(zfs_handle_t zhp, void unsupp_fs)
	{
	int zfs_version = (int)zfs_prop_get_int(zhp, ZFS_PROP_VERSION);
	int count = (int )unsupp_fs;

	if (zfs_version > ZPL_VERSION) {
	(void) printf(gettext("%s (v%d) is not supported by this "
	"implementation of ZFS.\n"),
	zfs_get_name(zhp), zfs_version);
	(*count)++;
	}

	zfs_iter_filesystems(zhp, check_unsupp_fs, unsupp_fs);

	zfs_close(zhp);

	return (0);
	}

	static int
	upgrade_version(zpool_handle_t *zhp, uint64_t version)
	{
	int ret;
	nvlist_t *config;
	uint64_t oldversion;
	int unsupp_fs = 0;

	config = zpool_get_config(zhp, NULL);
	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
	&oldversion) == 0);

	char compat[ZFS_MAXPROPLEN];
	if (zpool_get_prop(zhp, ZPOOL_PROP_COMPATIBILITY, compat,
	ZFS_MAXPROPLEN, NULL, B_FALSE) != 0)
	compat[0] = '\0';

	assert(SPA_VERSION_IS_SUPPORTED(oldversion));
	assert(oldversion < version);

	ret = zfs_iter_root(zpool_get_handle(zhp), check_unsupp_fs, &unsupp_fs);
	if (ret != 0)
	return (ret);

	if (unsupp_fs) {
	(void) fprintf(stderr, gettext("Upgrade not performed due "
	"to %d unsupported filesystems (max v%d).\n"),
	unsupp_fs, (int)ZPL_VERSION);
	return (1);
	}

	if (strcmp(compat, ZPOOL_COMPAT_LEGACY) == 0) {
	(void) fprintf(stderr, gettext("Upgrade not performed because "
	"'compatibility' property set to '"
	ZPOOL_COMPAT_LEGACY "'.\n"));
	return (1);
	}

	ret = zpool_upgrade(zhp, version);
	if (ret != 0)
	return (ret);

	if (version >= SPA_VERSION_FEATURES) {
	(void) printf(gettext("Successfully upgraded "
	"'%s' from version %llu to feature flags.\n"),
	zpool_get_name(zhp), (u_longlong_t)oldversion);
	} else {
	(void) printf(gettext("Successfully upgraded "
	"'%s' from version %llu to version %llu.\n"),
	zpool_get_name(zhp), (u_longlong_t)oldversion,
	(u_longlong_t)version);
	}

	return (0);
	}

	static int
	upgrade_enable_all(zpool_handle_t zhp, int countp)
	{
	int i, ret, count;
	boolean_t firstff = B_TRUE;
	nvlist_t *enabled = zpool_get_features(zhp);

	char compat[ZFS_MAXPROPLEN];
	if (zpool_get_prop(zhp, ZPOOL_PROP_COMPATIBILITY, compat,
	ZFS_MAXPROPLEN, NULL, B_FALSE) != 0)
	compat[0] = '\0';

	boolean_t requested_features[SPA_FEATURES];
	if (zpool_do_load_compat(compat, requested_features) !=
	ZPOOL_COMPATIBILITY_OK)
	return (-1);

	count = 0;
	for (i = 0; i < SPA_FEATURES; i++) {
	const char *fname = spa_feature_table[i].fi_uname;
	const char *fguid = spa_feature_table[i].fi_guid;

	if (!spa_feature_table[i].fi_zfs_mod_supported)
	continue;

	if (!nvlist_exists(enabled, fguid) && requested_features[i]) {
	char *propname;
	verify(-1 != asprintf(&propname, "feature@%s", fname));
	ret = zpool_set_prop(zhp, propname,
	ZFS_FEATURE_ENABLED);
	if (ret != 0) {
	free(propname);
	return (ret);
	}
	count++;

	if (firstff) {
	(void) printf(gettext("Enabled the "
	"following features on '%s':\n"),
	zpool_get_name(zhp));
	firstff = B_FALSE;
	}
	(void) printf(gettext(" %s\n"), fname);
	free(propname);
	}
	}

	if (countp != NULL)
	*countp = count;
	return (0);
	}

	static int
	upgrade_cb(zpool_handle_t zhp, void arg)
	{
	upgrade_cbdata_t *cbp = arg;
	nvlist_t *config;
	uint64_t version;
	boolean_t modified_pool = B_FALSE;
	int ret;

	config = zpool_get_config(zhp, NULL);
	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
	&version) == 0);

	assert(SPA_VERSION_IS_SUPPORTED(version));

	if (version < cbp->cb_version) {
	cbp->cb_first = B_FALSE;
	ret = upgrade_version(zhp, cbp->cb_version);
	if (ret != 0)
	return (ret);
	modified_pool = B_TRUE;

	/*
	* If they did "zpool upgrade -a", then we could
	* be doing ioctls to different pools. We need
	* to log this history once to each pool, and bypass
	* the normal history logging that happens in main().
	*/
	(void) zpool_log_history(g_zfs, history_str);
	log_history = B_FALSE;
	}

	if (cbp->cb_version >= SPA_VERSION_FEATURES) {
	int count;
	ret = upgrade_enable_all(zhp, &count);
	if (ret != 0)
	return (ret);

	if (count > 0) {
	cbp->cb_first = B_FALSE;
	modified_pool = B_TRUE;
	}
	}

	if (modified_pool) {
	(void) printf("\n");
	(void) after_zpool_upgrade(zhp);
	}

	return (0);
	}

	static int
	upgrade_list_older_cb(zpool_handle_t zhp, void arg)
	{
	upgrade_cbdata_t *cbp = arg;
	nvlist_t *config;
	uint64_t version;

	config = zpool_get_config(zhp, NULL);
	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
	&version) == 0);

	assert(SPA_VERSION_IS_SUPPORTED(version));

	if (version < SPA_VERSION_FEATURES) {
	if (cbp->cb_first) {
	(void) printf(gettext("The following pools are "
	"formatted with legacy version numbers and can\n"
	"be upgraded to use feature flags. After "
	"being upgraded, these pools\nwill no "
	"longer be accessible by software that does not "
	"support feature\nflags.\n\n"
	"Note that setting a pool's 'compatibility' "
	"feature to '" ZPOOL_COMPAT_LEGACY "' will\n"
	"inhibit upgrades.\n\n"));
	(void) printf(gettext("VER POOL\n"));
	(void) printf(gettext("--- ------------\n"));
	cbp->cb_first = B_FALSE;
	}

	(void) printf("%2llu %s\n", (u_longlong_t)version,
	zpool_get_name(zhp));
	}

	return (0);
	}

	static int
	upgrade_list_disabled_cb(zpool_handle_t zhp, void arg)
	{
	upgrade_cbdata_t *cbp = arg;
	nvlist_t *config;
	uint64_t version;

	config = zpool_get_config(zhp, NULL);
	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
	&version) == 0);

	if (version >= SPA_VERSION_FEATURES) {
	int i;
	boolean_t poolfirst = B_TRUE;
	nvlist_t *enabled = zpool_get_features(zhp);

	for (i = 0; i < SPA_FEATURES; i++) {
	const char *fguid = spa_feature_table[i].fi_guid;
	const char *fname = spa_feature_table[i].fi_uname;

	if (!spa_feature_table[i].fi_zfs_mod_supported)
	continue;

	if (!nvlist_exists(enabled, fguid)) {
	if (cbp->cb_first) {
	(void) printf(gettext("\nSome "
	"supported features are not "
	"enabled on the following pools. "
	"Once a\nfeature is enabled the "
	"pool may become incompatible with "
	"software\nthat does not support "
	"the feature. See "
	"zpool-features(7) for "
	"details.\n\n"
	"Note that the pool "
	"'compatibility' feature can be "
	"used to inhibit\nfeature "
	"upgrades.\n\n"));
	(void) printf(gettext("POOL "
	"FEATURE\n"));
	(void) printf(gettext("------"
	"---------\n"));
	cbp->cb_first = B_FALSE;
	}

	if (poolfirst) {
	(void) printf(gettext("%s\n"),
	zpool_get_name(zhp));
	poolfirst = B_FALSE;
	}

	(void) printf(gettext(" %s\n"), fname);
	}
	/*
	* If they did "zpool upgrade -a", then we could
	* be doing ioctls to different pools. We need
	* to log this history once to each pool, and bypass
	* the normal history logging that happens in main().
	*/
	(void) zpool_log_history(g_zfs, history_str);
	log_history = B_FALSE;
	}
	}

	return (0);
	}

	/* ARGSUSED */
	static int
	upgrade_one(zpool_handle_t zhp, void data)
	{
	boolean_t modified_pool = B_FALSE;
	upgrade_cbdata_t *cbp = data;
	uint64_t cur_version;
	int ret;

	if (strcmp("log", zpool_get_name(zhp)) == 0) {
	(void) fprintf(stderr, gettext("'log' is now a reserved word\n"
	"Pool 'log' must be renamed using export and import"
	" to upgrade.\n"));
	return (1);
	}

	cur_version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL);
	if (cur_version > cbp->cb_version) {
	(void) printf(gettext("Pool '%s' is already formatted "
	"using more current version '%llu'.\n\n"),
	zpool_get_name(zhp), (u_longlong_t)cur_version);
	return (0);
	}

	if (cbp->cb_version != SPA_VERSION && cur_version == cbp->cb_version) {
	(void) printf(gettext("Pool '%s' is already formatted "
	"using version %llu.\n\n"), zpool_get_name(zhp),
	(u_longlong_t)cbp->cb_version);
	return (0);
	}

	if (cur_version != cbp->cb_version) {
	modified_pool = B_TRUE;
	ret = upgrade_version(zhp, cbp->cb_version);
	if (ret != 0)
	return (ret);
	}

	if (cbp->cb_version >= SPA_VERSION_FEATURES) {
	int count = 0;
	ret = upgrade_enable_all(zhp, &count);
	if (ret != 0)
	return (ret);

	if (count != 0) {
	modified_pool = B_TRUE;
	} else if (cur_version == SPA_VERSION) {
	(void) printf(gettext("Pool '%s' already has all "
	"supported and requested features enabled.\n"),
	zpool_get_name(zhp));
	}
	}

	if (modified_pool) {
	(void) printf("\n");
	(void) after_zpool_upgrade(zhp);
	}

	return (0);
	}

	/*
	* zpool upgrade
	* zpool upgrade -v
	* zpool upgrade [-V version] <-a \| pool ...>
	*
	* With no arguments, display downrev'd ZFS pool available for upgrade.
	* Individual pools can be upgraded by specifying the pool, and '-a' will
	* upgrade all pools.
	*/
	int
	zpool_do_upgrade(int argc, char **argv)
	{
	int c;
	upgrade_cbdata_t cb = { 0 };
	int ret = 0;
	boolean_t showversions = B_FALSE;
	boolean_t upgradeall = B_FALSE;
	char *end;


	/* check options */
	while ((c = getopt(argc, argv, ":avV:")) != -1) {
	switch (c) {
	case 'a':
	upgradeall = B_TRUE;
	break;
	case 'v':
	showversions = B_TRUE;
	break;
	case 'V':
	cb.cb_version = strtoll(optarg, &end, 10);
	if (*end != '\0' \|\|
	!SPA_VERSION_IS_SUPPORTED(cb.cb_version)) {
	(void) fprintf(stderr,
	gettext("invalid version '%s'\n"), optarg);
	usage(B_FALSE);
	}
	break;
	case ':':
	(void) fprintf(stderr, gettext("missing argument for "
	"'%c' option\n"), optopt);
	usage(B_FALSE);
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	cb.cb_argc = argc;
	cb.cb_argv = argv;
	argc -= optind;
	argv += optind;

	if (cb.cb_version == 0) {
	cb.cb_version = SPA_VERSION;
	} else if (!upgradeall && argc == 0) {
	(void) fprintf(stderr, gettext("-V option is "
	"incompatible with other arguments\n"));
	usage(B_FALSE);
	}

	if (showversions) {
	if (upgradeall \|\| argc != 0) {
	(void) fprintf(stderr, gettext("-v option is "
	"incompatible with other arguments\n"));
	usage(B_FALSE);
	}
	} else if (upgradeall) {
	if (argc != 0) {
	(void) fprintf(stderr, gettext("-a option should not "
	"be used along with a pool name\n"));
	usage(B_FALSE);
	}
	}

	(void) printf(gettext("This system supports ZFS pool feature "
	"flags.\n\n"));
	if (showversions) {
	int i;

	(void) printf(gettext("The following features are "
	"supported:\n\n"));
	(void) printf(gettext("FEAT DESCRIPTION\n"));
	(void) printf("----------------------------------------------"
	"---------------\n");
	for (i = 0; i < SPA_FEATURES; i++) {
	zfeature_info_t *fi = &spa_feature_table[i];
	if (!fi->fi_zfs_mod_supported)
	continue;
	const char *ro =
	(fi->fi_flags & ZFEATURE_FLAG_READONLY_COMPAT) ?
	" (read-only compatible)" : "";

	(void) printf("%-37s%s\n", fi->fi_uname, ro);
	(void) printf(" %s\n", fi->fi_desc);
	}
	(void) printf("\n");

	(void) printf(gettext("The following legacy versions are also "
	"supported:\n\n"));
	(void) printf(gettext("VER DESCRIPTION\n"));
	(void) printf("--- -----------------------------------------"
	"---------------\n");
	(void) printf(gettext(" 1 Initial ZFS version\n"));
	(void) printf(gettext(" 2 Ditto blocks "
	"(replicated metadata)\n"));
	(void) printf(gettext(" 3 Hot spares and double parity "
	"RAID-Z\n"));
	(void) printf(gettext(" 4 zpool history\n"));
	(void) printf(gettext(" 5 Compression using the gzip "
	"algorithm\n"));
	(void) printf(gettext(" 6 bootfs pool property\n"));
	(void) printf(gettext(" 7 Separate intent log devices\n"));
	(void) printf(gettext(" 8 Delegated administration\n"));
	(void) printf(gettext(" 9 refquota and refreservation "
	"properties\n"));
	(void) printf(gettext(" 10 Cache devices\n"));
	(void) printf(gettext(" 11 Improved scrub performance\n"));
	(void) printf(gettext(" 12 Snapshot properties\n"));
	(void) printf(gettext(" 13 snapused property\n"));
	(void) printf(gettext(" 14 passthrough-x aclinherit\n"));
	(void) printf(gettext(" 15 user/group space accounting\n"));
	(void) printf(gettext(" 16 stmf property support\n"));
	(void) printf(gettext(" 17 Triple-parity RAID-Z\n"));
	(void) printf(gettext(" 18 Snapshot user holds\n"));
	(void) printf(gettext(" 19 Log device removal\n"));
	(void) printf(gettext(" 20 Compression using zle "
	"(zero-length encoding)\n"));
	(void) printf(gettext(" 21 Deduplication\n"));
	(void) printf(gettext(" 22 Received properties\n"));
	(void) printf(gettext(" 23 Slim ZIL\n"));
	(void) printf(gettext(" 24 System attributes\n"));
	(void) printf(gettext(" 25 Improved scrub stats\n"));
	(void) printf(gettext(" 26 Improved snapshot deletion "
	"performance\n"));
	(void) printf(gettext(" 27 Improved snapshot creation "
	"performance\n"));
	(void) printf(gettext(" 28 Multiple vdev replacements\n"));
	(void) printf(gettext("\nFor more information on a particular "
	"version, including supported releases,\n"));
	(void) printf(gettext("see the ZFS Administration Guide.\n\n"));
	} else if (argc == 0 && upgradeall) {
	cb.cb_first = B_TRUE;
	ret = zpool_iter(g_zfs, upgrade_cb, &cb);
	if (ret == 0 && cb.cb_first) {
	if (cb.cb_version == SPA_VERSION) {
	(void) printf(gettext("All pools are already "
	"formatted using feature flags.\n\n"));
	(void) printf(gettext("Every feature flags "
	"pool already has all supported and "
	"requested features enabled.\n"));
	} else {
	(void) printf(gettext("All pools are already "
	"formatted with version %llu or higher.\n"),
	(u_longlong_t)cb.cb_version);
	}
	}
	} else if (argc == 0) {
	cb.cb_first = B_TRUE;
	ret = zpool_iter(g_zfs, upgrade_list_older_cb, &cb);
	assert(ret == 0);

	if (cb.cb_first) {
	(void) printf(gettext("All pools are formatted "
	"using feature flags.\n\n"));
	} else {
	(void) printf(gettext("\nUse 'zpool upgrade -v' "
	"for a list of available legacy versions.\n"));
	}

	cb.cb_first = B_TRUE;
	ret = zpool_iter(g_zfs, upgrade_list_disabled_cb, &cb);
	assert(ret == 0);

	if (cb.cb_first) {
	(void) printf(gettext("Every feature flags pool has "
	"all supported and requested features enabled.\n"));
	} else {
	(void) printf(gettext("\n"));
	}
	} else {
	- ret = for_each_pool(argc, argv, B_FALSE, NULL, B_FALSE,
	- upgrade_one, &cb);
	+ ret = for_each_pool(argc, argv, B_FALSE, NULL, ZFS_TYPE_POOL,
	+ B_FALSE, upgrade_one, &cb);
	}

	return (ret);
	}

	typedef struct hist_cbdata {
	boolean_t first;
	boolean_t longfmt;
	boolean_t internal;
	} hist_cbdata_t;

	static void
	print_history_records(nvlist_t nvhis, hist_cbdata_t cb)
	{
	nvlist_t **records;
	uint_t numrecords;
	int i;

	verify(nvlist_lookup_nvlist_array(nvhis, ZPOOL_HIST_RECORD,
	&records, &numrecords) == 0);
	for (i = 0; i < numrecords; i++) {
	nvlist_t *rec = records[i];
	char tbuf[64] = "";

	if (nvlist_exists(rec, ZPOOL_HIST_TIME)) {
	time_t tsec;
	struct tm t;

	tsec = fnvlist_lookup_uint64(records[i],
	ZPOOL_HIST_TIME);
	(void) localtime_r(&tsec, &t);
	(void) strftime(tbuf, sizeof (tbuf), "%F.%T", &t);
	}

	if (nvlist_exists(rec, ZPOOL_HIST_ELAPSED_NS)) {
	uint64_t elapsed_ns = fnvlist_lookup_int64(records[i],
	ZPOOL_HIST_ELAPSED_NS);
	(void) snprintf(tbuf + strlen(tbuf),
	sizeof (tbuf) - strlen(tbuf),
	" (%lldms)", (long long)elapsed_ns / 1000 / 1000);
	}

	if (nvlist_exists(rec, ZPOOL_HIST_CMD)) {
	(void) printf("%s %s", tbuf,
	fnvlist_lookup_string(rec, ZPOOL_HIST_CMD));
	} else if (nvlist_exists(rec, ZPOOL_HIST_INT_EVENT)) {
	int ievent =
	fnvlist_lookup_uint64(rec, ZPOOL_HIST_INT_EVENT);
	if (!cb->internal)
	continue;
	if (ievent >= ZFS_NUM_LEGACY_HISTORY_EVENTS) {
	(void) printf("%s unrecognized record:\n",
	tbuf);
	dump_nvlist(rec, 4);
	continue;
	}
	(void) printf("%s [internal %s txg:%lld] %s", tbuf,
	zfs_history_event_names[ievent],
	(longlong_t)fnvlist_lookup_uint64(
	rec, ZPOOL_HIST_TXG),
	fnvlist_lookup_string(rec, ZPOOL_HIST_INT_STR));
	} else if (nvlist_exists(rec, ZPOOL_HIST_INT_NAME)) {
	if (!cb->internal)
	continue;
	(void) printf("%s [txg:%lld] %s", tbuf,
	(longlong_t)fnvlist_lookup_uint64(
	rec, ZPOOL_HIST_TXG),
	fnvlist_lookup_string(rec, ZPOOL_HIST_INT_NAME));
	if (nvlist_exists(rec, ZPOOL_HIST_DSNAME)) {
	(void) printf(" %s (%llu)",
	fnvlist_lookup_string(rec,
	ZPOOL_HIST_DSNAME),
	(u_longlong_t)fnvlist_lookup_uint64(rec,
	ZPOOL_HIST_DSID));
	}
	(void) printf(" %s", fnvlist_lookup_string(rec,
	ZPOOL_HIST_INT_STR));
	} else if (nvlist_exists(rec, ZPOOL_HIST_IOCTL)) {
	if (!cb->internal)
	continue;
	(void) printf("%s ioctl %s\n", tbuf,
	fnvlist_lookup_string(rec, ZPOOL_HIST_IOCTL));
	if (nvlist_exists(rec, ZPOOL_HIST_INPUT_NVL)) {
	(void) printf(" input:\n");
	dump_nvlist(fnvlist_lookup_nvlist(rec,
	ZPOOL_HIST_INPUT_NVL), 8);
	}
	if (nvlist_exists(rec, ZPOOL_HIST_OUTPUT_NVL)) {
	(void) printf(" output:\n");
	dump_nvlist(fnvlist_lookup_nvlist(rec,
	ZPOOL_HIST_OUTPUT_NVL), 8);
	}
	if (nvlist_exists(rec, ZPOOL_HIST_OUTPUT_SIZE)) {
	(void) printf(" output nvlist omitted; "
	"original size: %lldKB\n",
	(longlong_t)fnvlist_lookup_int64(rec,
	ZPOOL_HIST_OUTPUT_SIZE) / 1024);
	}
	if (nvlist_exists(rec, ZPOOL_HIST_ERRNO)) {
	(void) printf(" errno: %lld\n",
	(longlong_t)fnvlist_lookup_int64(rec,
	ZPOOL_HIST_ERRNO));
	}
	} else {
	if (!cb->internal)
	continue;
	(void) printf("%s unrecognized record:\n", tbuf);
	dump_nvlist(rec, 4);
	}

	if (!cb->longfmt) {
	(void) printf("\n");
	continue;
	}
	(void) printf(" [");
	if (nvlist_exists(rec, ZPOOL_HIST_WHO)) {
	uid_t who = fnvlist_lookup_uint64(rec, ZPOOL_HIST_WHO);
	struct passwd *pwd = getpwuid(who);
	(void) printf("user %d ", (int)who);
	if (pwd != NULL)
	(void) printf("(%s) ", pwd->pw_name);
	}
	if (nvlist_exists(rec, ZPOOL_HIST_HOST)) {
	(void) printf("on %s",
	fnvlist_lookup_string(rec, ZPOOL_HIST_HOST));
	}
	if (nvlist_exists(rec, ZPOOL_HIST_ZONE)) {
	(void) printf(":%s",
	fnvlist_lookup_string(rec, ZPOOL_HIST_ZONE));
	}

	(void) printf("]");
	(void) printf("\n");
	}
	}

	/*
	* Print out the command history for a specific pool.
	*/
	static int
	get_history_one(zpool_handle_t zhp, void data)
	{
	nvlist_t *nvhis;
	int ret;
	hist_cbdata_t cb = (hist_cbdata_t )data;
	uint64_t off = 0;
	boolean_t eof = B_FALSE;

	cb->first = B_FALSE;

	(void) printf(gettext("History for '%s':\n"), zpool_get_name(zhp));

	while (!eof) {
	if ((ret = zpool_get_history(zhp, &nvhis, &off, &eof)) != 0)
	return (ret);

	print_history_records(nvhis, cb);
	nvlist_free(nvhis);
	}
	(void) printf("\n");

	return (ret);
	}

	/*
	* zpool history <pool>
	*
	* Displays the history of commands that modified pools.
	*/
	int
	zpool_do_history(int argc, char **argv)
	{
	hist_cbdata_t cbdata = { 0 };
	int ret;
	int c;

	cbdata.first = B_TRUE;
	/* check options */
	while ((c = getopt(argc, argv, "li")) != -1) {
	switch (c) {
	case 'l':
	cbdata.longfmt = B_TRUE;
	break;
	case 'i':
	cbdata.internal = B_TRUE;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}
	argc -= optind;
	argv += optind;

	- ret = for_each_pool(argc, argv, B_FALSE, NULL, B_FALSE, get_history_one,
	- &cbdata);
	+ ret = for_each_pool(argc, argv, B_FALSE, NULL, ZFS_TYPE_POOL,
	+ B_FALSE, get_history_one, &cbdata);

	if (argc == 0 && cbdata.first == B_TRUE) {
	(void) fprintf(stderr, gettext("no pools available\n"));
	return (0);
	}

	return (ret);
	}

	typedef struct ev_opts {
	int verbose;
	int scripted;
	int follow;
	int clear;
	char poolname[ZFS_MAX_DATASET_NAME_LEN];
	} ev_opts_t;

	static void
	zpool_do_events_short(nvlist_t nvl, ev_opts_t opts)
	{
	char ctime_str[26], str[32], *ptr;
	int64_t *tv;
	uint_t n;

	verify(nvlist_lookup_int64_array(nvl, FM_EREPORT_TIME, &tv, &n) == 0);
	memset(str, ' ', 32);
	(void) ctime_r((const time_t *)&tv[0], ctime_str);
	(void) memcpy(str, ctime_str+4, 6); /* 'Jun 30' */
	(void) memcpy(str+7, ctime_str+20, 4); /* '1993' */
	(void) memcpy(str+12, ctime_str+11, 8); /* '21:49:08' */
	(void) sprintf(str+20, ".%09lld", (longlong_t)tv[1]); /* '.123456789' */
	if (opts->scripted)
	(void) printf(gettext("%s\t"), str);
	else
	(void) printf(gettext("%s "), str);

	verify(nvlist_lookup_string(nvl, FM_CLASS, &ptr) == 0);
	(void) printf(gettext("%s\n"), ptr);
	}

	static void
	zpool_do_events_nvprint(nvlist_t *nvl, int depth)
	{
	nvpair_t *nvp;

	for (nvp = nvlist_next_nvpair(nvl, NULL);
	nvp != NULL; nvp = nvlist_next_nvpair(nvl, nvp)) {

	data_type_t type = nvpair_type(nvp);
	const char *name = nvpair_name(nvp);

	boolean_t b;
	uint8_t i8;
	uint16_t i16;
	uint32_t i32;
	uint64_t i64;
	char *str;
	nvlist_t *cnv;

	printf(gettext("%*s%s = "), depth, "", name);

	switch (type) {
	case DATA_TYPE_BOOLEAN:
	printf(gettext("%s"), "1");
	break;

	case DATA_TYPE_BOOLEAN_VALUE:
	(void) nvpair_value_boolean_value(nvp, &b);
	printf(gettext("%s"), b ? "1" : "0");
	break;

	case DATA_TYPE_BYTE:
	(void) nvpair_value_byte(nvp, &i8);
	printf(gettext("0x%x"), i8);
	break;

	case DATA_TYPE_INT8:
	(void) nvpair_value_int8(nvp, (void *)&i8);
	printf(gettext("0x%x"), i8);
	break;

	case DATA_TYPE_UINT8:
	(void) nvpair_value_uint8(nvp, &i8);
	printf(gettext("0x%x"), i8);
	break;

	case DATA_TYPE_INT16:
	(void) nvpair_value_int16(nvp, (void *)&i16);
	printf(gettext("0x%x"), i16);
	break;

	case DATA_TYPE_UINT16:
	(void) nvpair_value_uint16(nvp, &i16);
	printf(gettext("0x%x"), i16);
	break;

	case DATA_TYPE_INT32:
	(void) nvpair_value_int32(nvp, (void *)&i32);
	printf(gettext("0x%x"), i32);
	break;

	case DATA_TYPE_UINT32:
	(void) nvpair_value_uint32(nvp, &i32);
	printf(gettext("0x%x"), i32);
	break;

	case DATA_TYPE_INT64:
	(void) nvpair_value_int64(nvp, (void *)&i64);
	printf(gettext("0x%llx"), (u_longlong_t)i64);
	break;

	case DATA_TYPE_UINT64:
	(void) nvpair_value_uint64(nvp, &i64);
	/*
	* translate vdev state values to readable
	* strings to aide zpool events consumers
	*/
	if (strcmp(name,
	FM_EREPORT_PAYLOAD_ZFS_VDEV_STATE) == 0 \|\|
	strcmp(name,
	FM_EREPORT_PAYLOAD_ZFS_VDEV_LASTSTATE) == 0) {
	printf(gettext("\"%s\" (0x%llx)"),
	zpool_state_to_name(i64, VDEV_AUX_NONE),
	(u_longlong_t)i64);
	} else {
	printf(gettext("0x%llx"), (u_longlong_t)i64);
	}
	break;

	case DATA_TYPE_HRTIME:
	(void) nvpair_value_hrtime(nvp, (void *)&i64);
	printf(gettext("0x%llx"), (u_longlong_t)i64);
	break;

	case DATA_TYPE_STRING:
	(void) nvpair_value_string(nvp, &str);
	printf(gettext("\"%s\""), str ? str : "<NULL>");
	break;

	case DATA_TYPE_NVLIST:
	printf(gettext("(embedded nvlist)\n"));
	(void) nvpair_value_nvlist(nvp, &cnv);
	zpool_do_events_nvprint(cnv, depth + 8);
	printf(gettext("%*s(end %s)"), depth, "", name);
	break;

	case DATA_TYPE_NVLIST_ARRAY: {
	nvlist_t **val;
	uint_t i, nelem;

	(void) nvpair_value_nvlist_array(nvp, &val, &nelem);
	printf(gettext("(%d embedded nvlists)\n"), nelem);
	for (i = 0; i < nelem; i++) {
	printf(gettext("%*s%s[%d] = %s\n"),
	depth, "", name, i, "(embedded nvlist)");
	zpool_do_events_nvprint(val[i], depth + 8);
	printf(gettext("%*s(end %s[%i])\n"),
	depth, "", name, i);
	}
	printf(gettext("%*s(end %s)\n"), depth, "", name);
	}
	break;

	case DATA_TYPE_INT8_ARRAY: {
	int8_t *val;
	uint_t i, nelem;

	(void) nvpair_value_int8_array(nvp, &val, &nelem);
	for (i = 0; i < nelem; i++)
	printf(gettext("0x%x "), val[i]);

	break;
	}

	case DATA_TYPE_UINT8_ARRAY: {
	uint8_t *val;
	uint_t i, nelem;

	(void) nvpair_value_uint8_array(nvp, &val, &nelem);
	for (i = 0; i < nelem; i++)
	printf(gettext("0x%x "), val[i]);

	break;
	}

	case DATA_TYPE_INT16_ARRAY: {
	int16_t *val;
	uint_t i, nelem;

	(void) nvpair_value_int16_array(nvp, &val, &nelem);
	for (i = 0; i < nelem; i++)
	printf(gettext("0x%x "), val[i]);

	break;
	}

	case DATA_TYPE_UINT16_ARRAY: {
	uint16_t *val;
	uint_t i, nelem;

	(void) nvpair_value_uint16_array(nvp, &val, &nelem);
	for (i = 0; i < nelem; i++)
	printf(gettext("0x%x "), val[i]);

	break;
	}

	case DATA_TYPE_INT32_ARRAY: {
	int32_t *val;
	uint_t i, nelem;

	(void) nvpair_value_int32_array(nvp, &val, &nelem);
	for (i = 0; i < nelem; i++)
	printf(gettext("0x%x "), val[i]);

	break;
	}

	case DATA_TYPE_UINT32_ARRAY: {
	uint32_t *val;
	uint_t i, nelem;

	(void) nvpair_value_uint32_array(nvp, &val, &nelem);
	for (i = 0; i < nelem; i++)
	printf(gettext("0x%x "), val[i]);

	break;
	}

	case DATA_TYPE_INT64_ARRAY: {
	int64_t *val;
	uint_t i, nelem;

	(void) nvpair_value_int64_array(nvp, &val, &nelem);
	for (i = 0; i < nelem; i++)
	printf(gettext("0x%llx "),
	(u_longlong_t)val[i]);

	break;
	}

	case DATA_TYPE_UINT64_ARRAY: {
	uint64_t *val;
	uint_t i, nelem;

	(void) nvpair_value_uint64_array(nvp, &val, &nelem);
	for (i = 0; i < nelem; i++)
	printf(gettext("0x%llx "),
	(u_longlong_t)val[i]);

	break;
	}

	case DATA_TYPE_STRING_ARRAY: {
	char **str;
	uint_t i, nelem;

	(void) nvpair_value_string_array(nvp, &str, &nelem);
	for (i = 0; i < nelem; i++)
	printf(gettext("\"%s\" "),
	str[i] ? str[i] : "<NULL>");

	break;
	}

	case DATA_TYPE_BOOLEAN_ARRAY:
	case DATA_TYPE_BYTE_ARRAY:
	case DATA_TYPE_DOUBLE:
	case DATA_TYPE_DONTCARE:
	case DATA_TYPE_UNKNOWN:
	printf(gettext("<unknown>"));
	break;
	}

	printf(gettext("\n"));
	}
	}

	static int
	zpool_do_events_next(ev_opts_t *opts)
	{
	nvlist_t *nvl;
	int zevent_fd, ret, dropped;
	char *pool;

	zevent_fd = open(ZFS_DEV, O_RDWR);
	VERIFY(zevent_fd >= 0);

	if (!opts->scripted)
	(void) printf(gettext("%-30s %s\n"), "TIME", "CLASS");

	while (1) {
	ret = zpool_events_next(g_zfs, &nvl, &dropped,
	(opts->follow ? ZEVENT_NONE : ZEVENT_NONBLOCK), zevent_fd);
	if (ret \|\| nvl == NULL)
	break;

	if (dropped > 0)
	(void) printf(gettext("dropped %d events\n"), dropped);

	if (strlen(opts->poolname) > 0 &&
	nvlist_lookup_string(nvl, FM_FMRI_ZFS_POOL, &pool) == 0 &&
	strcmp(opts->poolname, pool) != 0)
	continue;

	zpool_do_events_short(nvl, opts);

	if (opts->verbose) {
	zpool_do_events_nvprint(nvl, 8);
	printf(gettext("\n"));
	}
	(void) fflush(stdout);

	nvlist_free(nvl);
	}

	VERIFY(0 == close(zevent_fd));

	return (ret);
	}

	static int
	zpool_do_events_clear(ev_opts_t *opts)
	{
	int count, ret;

	ret = zpool_events_clear(g_zfs, &count);
	if (!ret)
	(void) printf(gettext("cleared %d events\n"), count);

	return (ret);
	}

	/*
	* zpool events [-vHf [pool] \| -c]
	*
	* Displays events logs by ZFS.
	*/
	int
	zpool_do_events(int argc, char **argv)
	{
	ev_opts_t opts = { 0 };
	int ret;
	int c;

	/* check options */
	while ((c = getopt(argc, argv, "vHfc")) != -1) {
	switch (c) {
	case 'v':
	opts.verbose = 1;
	break;
	case 'H':
	opts.scripted = 1;
	break;
	case 'f':
	opts.follow = 1;
	break;
	case 'c':
	opts.clear = 1;
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}
	argc -= optind;
	argv += optind;

	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	} else if (argc == 1) {
	(void) strlcpy(opts.poolname, argv[0], sizeof (opts.poolname));
	if (!zfs_name_valid(opts.poolname, ZFS_TYPE_POOL)) {
	(void) fprintf(stderr,
	gettext("invalid pool name '%s'\n"), opts.poolname);
	usage(B_FALSE);
	}
	}

	if ((argc == 1 \|\| opts.verbose \|\| opts.scripted \|\| opts.follow) &&
	opts.clear) {
	(void) fprintf(stderr,
	gettext("invalid options combined with -c\n"));
	usage(B_FALSE);
	}

	if (opts.clear)
	ret = zpool_do_events_clear(&opts);
	else
	ret = zpool_do_events_next(&opts);

	return (ret);
	}

	static int
	-get_callback(zpool_handle_t zhp, void data)
	+get_callback_vdev(zpool_handle_t zhp, char vdevname, void *data)
	{
	zprop_get_cbdata_t cbp = (zprop_get_cbdata_t )data;
	- char value[MAXNAMELEN];
	+ char value[ZFS_MAXPROPLEN];
	zprop_source_t srctype;
	- zprop_list_t *pl;
	-
	- for (pl = cbp->cb_proplist; pl != NULL; pl = pl->pl_next) {

	+ for (zprop_list_t *pl = cbp->cb_proplist; pl != NULL;
	+ pl = pl->pl_next) {
	+ char *prop_name;
	/*
	- * Skip the special fake placeholder. This will also skip
	+ * If the first property is pool name, it is a special
	+ * placeholder that we can skip. This will also skip
	* over the name property when 'all' is specified.
	*/
	if (pl->pl_prop == ZPOOL_PROP_NAME &&
	pl == cbp->cb_proplist)
	continue;

	- if (pl->pl_prop == ZPROP_INVAL &&
	- (zpool_prop_feature(pl->pl_user_prop) \|\|
	- zpool_prop_unsupported(pl->pl_user_prop))) {
	- srctype = ZPROP_SRC_LOCAL;
	+ if (pl->pl_prop == ZPROP_INVAL) {
	+ prop_name = pl->pl_user_prop;
	+ } else {
	+ prop_name = (char *)vdev_prop_to_name(pl->pl_prop);
	+ }
	+ if (zpool_get_vdev_prop(zhp, vdevname, pl->pl_prop,
	+ prop_name, value, sizeof (value), &srctype,
	+ cbp->cb_literal) == 0) {
	+ zprop_print_one_property(vdevname, cbp, prop_name,
	+ value, srctype, NULL, NULL);
	+ }
	+ }

	- if (zpool_prop_get_feature(zhp, pl->pl_user_prop,
	- value, sizeof (value)) == 0) {
	- zprop_print_one_property(zpool_get_name(zhp),
	- cbp, pl->pl_user_prop, value, srctype,
	- NULL, NULL);
	- }
	+ return (0);
	+}
	+
	+static int
	+get_callback_vdev_width_cb(void zhp_data, nvlist_t nv, void *data)
	+{
	+ zpool_handle_t *zhp = zhp_data;
	+ zprop_get_cbdata_t cbp = (zprop_get_cbdata_t )data;
	+ char *vdevname = zpool_vdev_name(g_zfs, zhp, nv,
	+ cbp->cb_vdevs.cb_name_flags);
	+ int ret;
	+
	+ /* Adjust the column widths for the vdev properties */
	+ ret = vdev_expand_proplist(zhp, vdevname, &cbp->cb_proplist);
	+
	+ return (ret);
	+}
	+
	+static int
	+get_callback_vdev_cb(void zhp_data, nvlist_t nv, void *data)
	+{
	+ zpool_handle_t *zhp = zhp_data;
	+ zprop_get_cbdata_t cbp = (zprop_get_cbdata_t )data;
	+ char *vdevname = zpool_vdev_name(g_zfs, zhp, nv,
	+ cbp->cb_vdevs.cb_name_flags);
	+ int ret;
	+
	+ /* Display the properties */
	+ ret = get_callback_vdev(zhp, vdevname, data);
	+
	+ return (ret);
	+}
	+
	+static int
	+get_callback(zpool_handle_t zhp, void data)
	+{
	+ zprop_get_cbdata_t cbp = (zprop_get_cbdata_t )data;
	+ char value[MAXNAMELEN];
	+ zprop_source_t srctype;
	+ zprop_list_t *pl;
	+ int vid;
	+
	+ if (cbp->cb_type == ZFS_TYPE_VDEV) {
	+ if (strcmp(cbp->cb_vdevs.cb_names[0], "all-vdevs") == 0) {
	+ for_each_vdev(zhp, get_callback_vdev_width_cb, data);
	+ for_each_vdev(zhp, get_callback_vdev_cb, data);
	} else {
	- if (zpool_get_prop(zhp, pl->pl_prop, value,
	- sizeof (value), &srctype, cbp->cb_literal) != 0)
	+ /* Adjust column widths for vdev properties */
	+ for (vid = 0; vid < cbp->cb_vdevs.cb_names_count;
	+ vid++) {
	+ vdev_expand_proplist(zhp,
	+ cbp->cb_vdevs.cb_names[vid],
	+ &cbp->cb_proplist);
	+ }
	+ /* Display the properties */
	+ for (vid = 0; vid < cbp->cb_vdevs.cb_names_count;
	+ vid++) {
	+ get_callback_vdev(zhp,
	+ cbp->cb_vdevs.cb_names[vid], data);
	+ }
	+ }
	+ } else {
	+ assert(cbp->cb_type == ZFS_TYPE_POOL);
	+ for (pl = cbp->cb_proplist; pl != NULL; pl = pl->pl_next) {
	+ /*
	+ * Skip the special fake placeholder. This will also
	+ * skip over the name property when 'all' is specified.
	+ */
	+ if (pl->pl_prop == ZPOOL_PROP_NAME &&
	+ pl == cbp->cb_proplist)
	continue;

	- zprop_print_one_property(zpool_get_name(zhp), cbp,
	- zpool_prop_to_name(pl->pl_prop), value, srctype,
	- NULL, NULL);
	+ if (pl->pl_prop == ZPROP_INVAL &&
	+ (zpool_prop_feature(pl->pl_user_prop) \|\|
	+ zpool_prop_unsupported(pl->pl_user_prop))) {
	+ srctype = ZPROP_SRC_LOCAL;
	+
	+ if (zpool_prop_get_feature(zhp,
	+ pl->pl_user_prop, value,
	+ sizeof (value)) == 0) {
	+ zprop_print_one_property(
	+ zpool_get_name(zhp), cbp,
	+ pl->pl_user_prop, value, srctype,
	+ NULL, NULL);
	+ }
	+ } else {
	+ if (zpool_get_prop(zhp, pl->pl_prop, value,
	+ sizeof (value), &srctype,
	+ cbp->cb_literal) != 0)
	+ continue;
	+
	+ zprop_print_one_property(zpool_get_name(zhp),
	+ cbp, zpool_prop_to_name(pl->pl_prop),
	+ value, srctype, NULL, NULL);
	+ }
	}
	}
	+
	return (0);
	}

	/*
	* zpool get [-Hp] [-o "all" \| field[,...]] <"all" \| property[,...]> <pool> ...
	*
	* -H Scripted mode. Don't display headers, and separate properties
	* by a single tab.
	* -o List of columns to display. Defaults to
	* "name,property,value,source".
	* -p Display values in parsable (exact) format.
	*
	* Get properties of pools in the system. Output space statistics
	* for each one as well as other attributes.
	*/
	int
	zpool_do_get(int argc, char **argv)
	{
	zprop_get_cbdata_t cb = { 0 };
	zprop_list_t fake_name = { 0 };
	int ret;
	int c, i;
	char *value;
	+ char *propstr = NULL;

	cb.cb_first = B_TRUE;

	/*
	* Set up default columns and sources.
	*/
	cb.cb_sources = ZPROP_SRC_ALL;
	cb.cb_columns[0] = GET_COL_NAME;
	cb.cb_columns[1] = GET_COL_PROPERTY;
	cb.cb_columns[2] = GET_COL_VALUE;
	cb.cb_columns[3] = GET_COL_SOURCE;
	cb.cb_type = ZFS_TYPE_POOL;
	+ cb.cb_vdevs.cb_name_flags \|= VDEV_NAME_TYPE_ID;
	+ current_prop_type = cb.cb_type;

	/* check options */
	while ((c = getopt(argc, argv, ":Hpo:")) != -1) {
	switch (c) {
	case 'p':
	cb.cb_literal = B_TRUE;
	break;
	case 'H':
	cb.cb_scripted = B_TRUE;
	break;
	case 'o':
	bzero(&cb.cb_columns, sizeof (cb.cb_columns));
	i = 0;
	while (*optarg != '\0') {
	static char *col_subopts[] =
	{ "name", "property", "value", "source",
	"all", NULL };

	if (i == ZFS_GET_NCOLS) {
	(void) fprintf(stderr, gettext("too "
	"many fields given to -o "
	"option\n"));
	usage(B_FALSE);
	}

	switch (getsubopt(&optarg, col_subopts,
	&value)) {
	case 0:
	cb.cb_columns[i++] = GET_COL_NAME;
	break;
	case 1:
	cb.cb_columns[i++] = GET_COL_PROPERTY;
	break;
	case 2:
	cb.cb_columns[i++] = GET_COL_VALUE;
	break;
	case 3:
	cb.cb_columns[i++] = GET_COL_SOURCE;
	break;
	case 4:
	if (i > 0) {
	(void) fprintf(stderr,
	gettext("\"all\" conflicts "
	"with specific fields "
	"given to -o option\n"));
	usage(B_FALSE);
	}
	cb.cb_columns[0] = GET_COL_NAME;
	cb.cb_columns[1] = GET_COL_PROPERTY;
	cb.cb_columns[2] = GET_COL_VALUE;
	cb.cb_columns[3] = GET_COL_SOURCE;
	i = ZFS_GET_NCOLS;
	break;
	default:
	(void) fprintf(stderr,
	gettext("invalid column name "
	"'%s'\n"), value);
	usage(B_FALSE);
	}
	}
	break;
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing property "
	"argument\n"));
	usage(B_FALSE);
	}

	- if (zprop_get_list(g_zfs, argv[0], &cb.cb_proplist,
	- ZFS_TYPE_POOL) != 0)
	- usage(B_FALSE);
	+ /* Properties list is needed later by zprop_get_list() */
	+ propstr = argv[0];

	argc--;
	argv++;

	+ if (argc == 0) {
	+ /* No args, so just print the defaults. */
	+ } else if (are_all_pools(argc, argv)) {
	+ /* All the args are pool names */
	+ } else if (are_all_pools(1, argv)) {
	+ /* The first arg is a pool name */
	+ if ((argc == 2 && strcmp(argv[1], "all-vdevs") == 0) \|\|
	+ are_vdevs_in_pool(argc - 1, argv + 1, argv[0],
	+ &cb.cb_vdevs)) {
	+ /* ... and the rest are vdev names */
	+ cb.cb_vdevs.cb_names = argv + 1;
	+ cb.cb_vdevs.cb_names_count = argc - 1;
	+ cb.cb_type = ZFS_TYPE_VDEV;
	+ argc = 1; /* One pool to process */
	+ } else {
	+ fprintf(stderr, gettext("Expected a list of vdevs in"
	+ " \"%s\", but got:\n"), argv[0]);
	+ error_list_unresolved_vdevs(argc - 1, argv + 1,
	+ argv[0], &cb.cb_vdevs);
	+ fprintf(stderr, "\n");
	+ usage(B_FALSE);
	+ return (1);
	+ }
	+ } else {
	+ /*
	+ * The first arg isn't a pool name,
	+ */
	+ fprintf(stderr, gettext("missing pool name.\n"));
	+ fprintf(stderr, "\n");
	+ usage(B_FALSE);
	+ return (1);
	+ }
	+
	+ if (zprop_get_list(g_zfs, propstr, &cb.cb_proplist,
	+ cb.cb_type) != 0) {
	+ /* Use correct list of valid properties (pool or vdev) */
	+ current_prop_type = cb.cb_type;
	+ usage(B_FALSE);
	+ }
	+
	if (cb.cb_proplist != NULL) {
	fake_name.pl_prop = ZPOOL_PROP_NAME;
	fake_name.pl_width = strlen(gettext("NAME"));
	fake_name.pl_next = cb.cb_proplist;
	cb.cb_proplist = &fake_name;
	}

	- ret = for_each_pool(argc, argv, B_TRUE, &cb.cb_proplist, cb.cb_literal,
	- get_callback, &cb);
	+ ret = for_each_pool(argc, argv, B_TRUE, &cb.cb_proplist, cb.cb_type,
	+ cb.cb_literal, get_callback, &cb);

	if (cb.cb_proplist == &fake_name)
	zprop_free_list(fake_name.pl_next);
	else
	zprop_free_list(cb.cb_proplist);

	return (ret);
	}

	typedef struct set_cbdata {
	char *cb_propname;
	char *cb_value;
	+ zfs_type_t cb_type;
	+ vdev_cbdata_t cb_vdevs;
	boolean_t cb_any_successful;
	} set_cbdata_t;

	static int
	-set_callback(zpool_handle_t zhp, void data)
	+set_pool_callback(zpool_handle_t zhp, set_cbdata_t cb)
	{
	int error;
	- set_cbdata_t cb = (set_cbdata_t )data;

	/* Check if we have out-of-bounds features */
	if (strcmp(cb->cb_propname, ZPOOL_CONFIG_COMPATIBILITY) == 0) {
	boolean_t features[SPA_FEATURES];
	if (zpool_do_load_compat(cb->cb_value, features) !=
	ZPOOL_COMPATIBILITY_OK)
	return (-1);

	nvlist_t *enabled = zpool_get_features(zhp);
	spa_feature_t i;
	for (i = 0; i < SPA_FEATURES; i++) {
	const char *fguid = spa_feature_table[i].fi_guid;
	if (nvlist_exists(enabled, fguid) && !features[i])
	break;
	}
	if (i < SPA_FEATURES)
	(void) fprintf(stderr, gettext("Warning: one or "
	"more features already enabled on pool '%s'\n"
	"are not present in this compatibility set.\n"),
	zpool_get_name(zhp));
	}

	/* if we're setting a feature, check it's in compatibility set */
	if (zpool_prop_feature(cb->cb_propname) &&
	strcmp(cb->cb_value, ZFS_FEATURE_ENABLED) == 0) {
	char *fname = strchr(cb->cb_propname, '@') + 1;
	spa_feature_t f;

	if (zfeature_lookup_name(fname, &f) == 0) {
	char compat[ZFS_MAXPROPLEN];
	if (zpool_get_prop(zhp, ZPOOL_PROP_COMPATIBILITY,
	compat, ZFS_MAXPROPLEN, NULL, B_FALSE) != 0)
	compat[0] = '\0';

	boolean_t features[SPA_FEATURES];
	if (zpool_do_load_compat(compat, features) !=
	ZPOOL_COMPATIBILITY_OK) {
	(void) fprintf(stderr, gettext("Error: "
	"cannot enable feature '%s' on pool '%s'\n"
	"because the pool's 'compatibility' "
	"property cannot be parsed.\n"),
	fname, zpool_get_name(zhp));
	return (-1);
	}

	if (!features[f]) {
	(void) fprintf(stderr, gettext("Error: "
	"cannot enable feature '%s' on pool '%s'\n"
	"as it is not specified in this pool's "
	"current compatibility set.\n"
	"Consider setting 'compatibility' to a "
	"less restrictive set, or to 'off'.\n"),
	fname, zpool_get_name(zhp));
	return (-1);
	}
	}
	}

	error = zpool_set_prop(zhp, cb->cb_propname, cb->cb_value);

	- if (!error)
	- cb->cb_any_successful = B_TRUE;
	+ return (error);
	+}
	+
	+static int
	+set_callback(zpool_handle_t zhp, void data)
	+{
	+ int error;
	+ set_cbdata_t cb = (set_cbdata_t )data;
	+
	+ if (cb->cb_type == ZFS_TYPE_VDEV) {
	+ error = zpool_set_vdev_prop(zhp, *cb->cb_vdevs.cb_names,
	+ cb->cb_propname, cb->cb_value);
	+ } else {
	+ assert(cb->cb_type == ZFS_TYPE_POOL);
	+ error = set_pool_callback(zhp, cb);
	+ }

	+ cb->cb_any_successful = !error;
	return (error);
	}

	int
	zpool_do_set(int argc, char **argv)
	{
	set_cbdata_t cb = { 0 };
	int error;

	+ current_prop_type = ZFS_TYPE_POOL;
	if (argc > 1 && argv[1][0] == '-') {
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	argv[1][1]);
	usage(B_FALSE);
	}

	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing property=value "
	"argument\n"));
	usage(B_FALSE);
	}

	if (argc < 3) {
	(void) fprintf(stderr, gettext("missing pool name\n"));
	usage(B_FALSE);
	}

	- if (argc > 3) {
	+ if (argc > 4) {
	(void) fprintf(stderr, gettext("too many pool names\n"));
	usage(B_FALSE);
	}

	cb.cb_propname = argv[1];
	+ cb.cb_type = ZFS_TYPE_POOL;
	+ cb.cb_vdevs.cb_name_flags \|= VDEV_NAME_TYPE_ID;
	cb.cb_value = strchr(cb.cb_propname, '=');
	if (cb.cb_value == NULL) {
	(void) fprintf(stderr, gettext("missing value in "
	"property=value argument\n"));
	usage(B_FALSE);
	}

	*(cb.cb_value) = '\0';
	cb.cb_value++;
	+ argc -= 2;
	+ argv += 2;
	+
	+ if (are_vdevs_in_pool(argc, argv, NULL, &cb.cb_vdevs)) {
	+ /* Argument is a vdev */
	+ cb.cb_vdevs.cb_names = argv;
	+ cb.cb_vdevs.cb_names_count = 1;
	+ cb.cb_type = ZFS_TYPE_VDEV;
	+ argc = 0; /* No pools to process */
	+ } else if (are_all_pools(1, argv)) {
	+ /* The first arg is a pool name */
	+ if (are_vdevs_in_pool(argc - 1, argv + 1, argv[0],
	+ &cb.cb_vdevs)) {
	+ /* 2nd argument is a vdev */
	+ cb.cb_vdevs.cb_names = argv + 1;
	+ cb.cb_vdevs.cb_names_count = 1;
	+ cb.cb_type = ZFS_TYPE_VDEV;
	+ argc = 1; /* One pool to process */
	+ } else if (argc > 1) {
	+ (void) fprintf(stderr,
	+ gettext("too many pool names\n"));
	+ usage(B_FALSE);
	+ }
	+ }

	- error = for_each_pool(argc - 2, argv + 2, B_TRUE, NULL, B_FALSE,
	- set_callback, &cb);
	+ error = for_each_pool(argc, argv, B_TRUE, NULL, ZFS_TYPE_POOL,
	+ B_FALSE, set_callback, &cb);

	return (error);
	}

	/* Add up the total number of bytes left to initialize/trim across all vdevs */
	static uint64_t
	vdev_activity_remaining(nvlist_t *nv, zpool_wait_activity_t activity)
	{
	uint64_t bytes_remaining;
	nvlist_t **child;
	uint_t c, children;
	vdev_stat_t *vs;

	assert(activity == ZPOOL_WAIT_INITIALIZE \|\|
	activity == ZPOOL_WAIT_TRIM);

	verify(nvlist_lookup_uint64_array(nv, ZPOOL_CONFIG_VDEV_STATS,
	(uint64_t **)&vs, &c) == 0);

	if (activity == ZPOOL_WAIT_INITIALIZE &&
	vs->vs_initialize_state == VDEV_INITIALIZE_ACTIVE)
	bytes_remaining = vs->vs_initialize_bytes_est -
	vs->vs_initialize_bytes_done;
	else if (activity == ZPOOL_WAIT_TRIM &&
	vs->vs_trim_state == VDEV_TRIM_ACTIVE)
	bytes_remaining = vs->vs_trim_bytes_est -
	vs->vs_trim_bytes_done;
	else
	bytes_remaining = 0;

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0)
	children = 0;

	for (c = 0; c < children; c++)
	bytes_remaining += vdev_activity_remaining(child[c], activity);

	return (bytes_remaining);
	}

	/* Add up the total number of bytes left to rebuild across top-level vdevs */
	static uint64_t
	vdev_activity_top_remaining(nvlist_t *nv)
	{
	uint64_t bytes_remaining = 0;
	nvlist_t **child;
	uint_t children;
	int error;

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0)
	children = 0;

	for (uint_t c = 0; c < children; c++) {
	vdev_rebuild_stat_t *vrs;
	uint_t i;

	error = nvlist_lookup_uint64_array(child[c],
	ZPOOL_CONFIG_REBUILD_STATS, (uint64_t **)&vrs, &i);
	if (error == 0) {
	if (vrs->vrs_state == VDEV_REBUILD_ACTIVE) {
	bytes_remaining += (vrs->vrs_bytes_est -
	vrs->vrs_bytes_rebuilt);
	}
	}
	}

	return (bytes_remaining);
	}

	/* Whether any vdevs are 'spare' or 'replacing' vdevs */
	static boolean_t
	vdev_any_spare_replacing(nvlist_t *nv)
	{
	nvlist_t **child;
	uint_t c, children;
	char *vdev_type;

	(void) nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &vdev_type);

	if (strcmp(vdev_type, VDEV_TYPE_REPLACING) == 0 \|\|
	strcmp(vdev_type, VDEV_TYPE_SPARE) == 0 \|\|
	strcmp(vdev_type, VDEV_TYPE_DRAID_SPARE) == 0) {
	return (B_TRUE);
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0)
	children = 0;

	for (c = 0; c < children; c++) {
	if (vdev_any_spare_replacing(child[c]))
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	typedef struct wait_data {
	char *wd_poolname;
	boolean_t wd_scripted;
	boolean_t wd_exact;
	boolean_t wd_headers_once;
	boolean_t wd_should_exit;
	/* Which activities to wait for */
	boolean_t wd_enabled[ZPOOL_WAIT_NUM_ACTIVITIES];
	float wd_interval;
	pthread_cond_t wd_cv;
	pthread_mutex_t wd_mutex;
	} wait_data_t;

	/*
	* Print to stdout a single line, containing one column for each activity that
	* we are waiting for specifying how many bytes of work are left for that
	* activity.
	*/
	static void
	print_wait_status_row(wait_data_t wd, zpool_handle_t zhp, int row)
	{
	nvlist_t config, nvroot;
	uint_t c;
	int i;
	pool_checkpoint_stat_t *pcs = NULL;
	pool_scan_stat_t *pss = NULL;
	pool_removal_stat_t *prs = NULL;
	char *headers[] = {"DISCARD", "FREE", "INITIALIZE", "REPLACE",
	"REMOVE", "RESILVER", "SCRUB", "TRIM"};
	int col_widths[ZPOOL_WAIT_NUM_ACTIVITIES];

	/* Calculate the width of each column */
	for (i = 0; i < ZPOOL_WAIT_NUM_ACTIVITIES; i++) {
	/*
	* Make sure we have enough space in the col for pretty-printed
	* numbers and for the column header, and then leave a couple
	* spaces between cols for readability.
	*/
	col_widths[i] = MAX(strlen(headers[i]), 6) + 2;
	}

	/* Print header if appropriate */
	int term_height = terminal_height();
	boolean_t reprint_header = (!wd->wd_headers_once && term_height > 0 &&
	row % (term_height-1) == 0);
	if (!wd->wd_scripted && (row == 0 \|\| reprint_header)) {
	for (i = 0; i < ZPOOL_WAIT_NUM_ACTIVITIES; i++) {
	if (wd->wd_enabled[i])
	(void) printf("%*s", col_widths[i], headers[i]);
	}
	(void) printf("\n");
	}

	/* Bytes of work remaining in each activity */
	int64_t bytes_rem[ZPOOL_WAIT_NUM_ACTIVITIES] = {0};

	bytes_rem[ZPOOL_WAIT_FREE] =
	zpool_get_prop_int(zhp, ZPOOL_PROP_FREEING, NULL);

	config = zpool_get_config(zhp, NULL);
	nvroot = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE);

	(void) nvlist_lookup_uint64_array(nvroot,
	ZPOOL_CONFIG_CHECKPOINT_STATS, (uint64_t **)&pcs, &c);
	if (pcs != NULL && pcs->pcs_state == CS_CHECKPOINT_DISCARDING)
	bytes_rem[ZPOOL_WAIT_CKPT_DISCARD] = pcs->pcs_space;

	(void) nvlist_lookup_uint64_array(nvroot,
	ZPOOL_CONFIG_REMOVAL_STATS, (uint64_t **)&prs, &c);
	if (prs != NULL && prs->prs_state == DSS_SCANNING)
	bytes_rem[ZPOOL_WAIT_REMOVE] = prs->prs_to_copy -
	prs->prs_copied;

	(void) nvlist_lookup_uint64_array(nvroot,
	ZPOOL_CONFIG_SCAN_STATS, (uint64_t **)&pss, &c);
	if (pss != NULL && pss->pss_state == DSS_SCANNING &&
	pss->pss_pass_scrub_pause == 0) {
	int64_t rem = pss->pss_to_examine - pss->pss_issued;
	if (pss->pss_func == POOL_SCAN_SCRUB)
	bytes_rem[ZPOOL_WAIT_SCRUB] = rem;
	else
	bytes_rem[ZPOOL_WAIT_RESILVER] = rem;
	} else if (check_rebuilding(nvroot, NULL)) {
	bytes_rem[ZPOOL_WAIT_RESILVER] =
	vdev_activity_top_remaining(nvroot);
	}

	bytes_rem[ZPOOL_WAIT_INITIALIZE] =
	vdev_activity_remaining(nvroot, ZPOOL_WAIT_INITIALIZE);
	bytes_rem[ZPOOL_WAIT_TRIM] =
	vdev_activity_remaining(nvroot, ZPOOL_WAIT_TRIM);

	/*
	* A replace finishes after resilvering finishes, so the amount of work
	* left for a replace is the same as for resilvering.
	*
	* It isn't quite correct to say that if we have any 'spare' or
	* 'replacing' vdevs and a resilver is happening, then a replace is in
	* progress, like we do here. When a hot spare is used, the faulted vdev
	* is not removed after the hot spare is resilvered, so parent 'spare'
	* vdev is not removed either. So we could have a 'spare' vdev, but be
	* resilvering for a different reason. However, we use it as a heuristic
	* because we don't have access to the DTLs, which could tell us whether
	* or not we have really finished resilvering a hot spare.
	*/
	if (vdev_any_spare_replacing(nvroot))
	bytes_rem[ZPOOL_WAIT_REPLACE] = bytes_rem[ZPOOL_WAIT_RESILVER];

	if (timestamp_fmt != NODATE)
	print_timestamp(timestamp_fmt);

	for (i = 0; i < ZPOOL_WAIT_NUM_ACTIVITIES; i++) {
	char buf[64];
	if (!wd->wd_enabled[i])
	continue;

	if (wd->wd_exact)
	(void) snprintf(buf, sizeof (buf), "%" PRIi64,
	bytes_rem[i]);
	else
	zfs_nicenum(bytes_rem[i], buf, sizeof (buf));

	if (wd->wd_scripted)
	(void) printf(i == 0 ? "%s" : "\t%s", buf);
	else
	(void) printf(" %*s", col_widths[i] - 1, buf);
	}
	(void) printf("\n");
	(void) fflush(stdout);
	}

	static void *
	wait_status_thread(void *arg)
	{
	wait_data_t wd = (wait_data_t )arg;
	zpool_handle_t *zhp;

	if ((zhp = zpool_open(g_zfs, wd->wd_poolname)) == NULL)
	return (void *)(1);

	for (int row = 0; ; row++) {
	boolean_t missing;
	struct timespec timeout;
	int ret = 0;
	(void) clock_gettime(CLOCK_REALTIME, &timeout);

	if (zpool_refresh_stats(zhp, &missing) != 0 \|\| missing \|\|
	zpool_props_refresh(zhp) != 0) {
	zpool_close(zhp);
	return (void *)(uintptr_t)(missing ? 0 : 1);
	}

	print_wait_status_row(wd, zhp, row);

	timeout.tv_sec += floor(wd->wd_interval);
	long nanos = timeout.tv_nsec +
	(wd->wd_interval - floor(wd->wd_interval)) * NANOSEC;
	if (nanos >= NANOSEC) {
	timeout.tv_sec++;
	timeout.tv_nsec = nanos - NANOSEC;
	} else {
	timeout.tv_nsec = nanos;
	}
	pthread_mutex_lock(&wd->wd_mutex);
	if (!wd->wd_should_exit)
	ret = pthread_cond_timedwait(&wd->wd_cv, &wd->wd_mutex,
	&timeout);
	pthread_mutex_unlock(&wd->wd_mutex);
	if (ret == 0) {
	break; /* signaled by main thread */
	} else if (ret != ETIMEDOUT) {
	(void) fprintf(stderr, gettext("pthread_cond_timedwait "
	"failed: %s\n"), strerror(ret));
	zpool_close(zhp);
	return (void *)(uintptr_t)(1);
	}
	}

	zpool_close(zhp);
	return (void *)(0);
	}

	int
	zpool_do_wait(int argc, char **argv)
	{
	boolean_t verbose = B_FALSE;
	int c;
	char *value;
	int i;
	unsigned long count;
	pthread_t status_thr;
	int error = 0;
	zpool_handle_t *zhp;

	wait_data_t wd;
	wd.wd_scripted = B_FALSE;
	wd.wd_exact = B_FALSE;
	wd.wd_headers_once = B_FALSE;
	wd.wd_should_exit = B_FALSE;

	pthread_mutex_init(&wd.wd_mutex, NULL);
	pthread_cond_init(&wd.wd_cv, NULL);

	/* By default, wait for all types of activity. */
	for (i = 0; i < ZPOOL_WAIT_NUM_ACTIVITIES; i++)
	wd.wd_enabled[i] = B_TRUE;

	while ((c = getopt(argc, argv, "HpT:t:")) != -1) {
	switch (c) {
	case 'H':
	wd.wd_scripted = B_TRUE;
	break;
	case 'n':
	wd.wd_headers_once = B_TRUE;
	break;
	case 'p':
	wd.wd_exact = B_TRUE;
	break;
	case 'T':
	get_timestamp_arg(*optarg);
	break;
	case 't':
	{
	static char *col_subopts[] = { "discard", "free",
	"initialize", "replace", "remove", "resilver",
	"scrub", "trim", NULL };

	/* Reset activities array */
	bzero(&wd.wd_enabled, sizeof (wd.wd_enabled));
	while (*optarg != '\0') {
	int activity = getsubopt(&optarg, col_subopts,
	&value);

	if (activity < 0) {
	(void) fprintf(stderr,
	gettext("invalid activity '%s'\n"),
	value);
	usage(B_FALSE);
	}

	wd.wd_enabled[activity] = B_TRUE;
	}
	break;
	}
	case '?':
	(void) fprintf(stderr, gettext("invalid option '%c'\n"),
	optopt);
	usage(B_FALSE);
	}
	}

	argc -= optind;
	argv += optind;

	get_interval_count(&argc, argv, &wd.wd_interval, &count);
	if (count != 0) {
	/* This subcmd only accepts an interval, not a count */
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	if (wd.wd_interval != 0)
	verbose = B_TRUE;

	if (argc < 1) {
	(void) fprintf(stderr, gettext("missing 'pool' argument\n"));
	usage(B_FALSE);
	}
	if (argc > 1) {
	(void) fprintf(stderr, gettext("too many arguments\n"));
	usage(B_FALSE);
	}

	wd.wd_poolname = argv[0];

	if ((zhp = zpool_open(g_zfs, wd.wd_poolname)) == NULL)
	return (1);

	if (verbose) {
	/*
	* We use a separate thread for printing status updates because
	* the main thread will call lzc_wait(), which blocks as long
	* as an activity is in progress, which can be a long time.
	*/
	if (pthread_create(&status_thr, NULL, wait_status_thread, &wd)
	!= 0) {
	(void) fprintf(stderr, gettext("failed to create status"
	"thread: %s\n"), strerror(errno));
	zpool_close(zhp);
	return (1);
	}
	}

	/*
	* Loop over all activities that we are supposed to wait for until none
	* of them are in progress. Note that this means we can end up waiting
	* for more activities to complete than just those that were in progress
	* when we began waiting; if an activity we are interested in begins
	* while we are waiting for another activity, we will wait for both to
	* complete before exiting.
	*/
	for (;;) {
	boolean_t missing = B_FALSE;
	boolean_t any_waited = B_FALSE;

	for (i = 0; i < ZPOOL_WAIT_NUM_ACTIVITIES; i++) {
	boolean_t waited;

	if (!wd.wd_enabled[i])
	continue;

	error = zpool_wait_status(zhp, i, &missing, &waited);
	if (error != 0 \|\| missing)
	break;

	any_waited = (any_waited \|\| waited);
	}

	if (error != 0 \|\| missing \|\| !any_waited)
	break;
	}

	zpool_close(zhp);

	if (verbose) {
	uintptr_t status;
	pthread_mutex_lock(&wd.wd_mutex);
	wd.wd_should_exit = B_TRUE;
	pthread_cond_signal(&wd.wd_cv);
	pthread_mutex_unlock(&wd.wd_mutex);
	(void) pthread_join(status_thr, (void *)&status);
	if (status != 0)
	error = status;
	}

	pthread_mutex_destroy(&wd.wd_mutex);
	pthread_cond_destroy(&wd.wd_cv);
	return (error);
	}

	static int
	find_command_idx(char command, int idx)
	{
	int i;

	for (i = 0; i < NCOMMAND; i++) {
	if (command_table[i].name == NULL)
	continue;

	if (strcmp(command, command_table[i].name) == 0) {
	*idx = i;
	return (0);
	}
	}
	return (1);
	}

	/*
	* Display version message
	*/
	static int
	zpool_do_version(int argc, char **argv)
	{
	if (zfs_version_print() == -1)
	return (1);

	return (0);
	}

	/*
	* Do zpool_load_compat() and print error message on failure
	*/
	static zpool_compat_status_t
	zpool_do_load_compat(const char compat, boolean_t list)
	{
	char report[1024];

	zpool_compat_status_t ret;

	ret = zpool_load_compat(compat, list, report, 1024);
	switch (ret) {

	case ZPOOL_COMPATIBILITY_OK:
	break;

	case ZPOOL_COMPATIBILITY_NOFILES:
	case ZPOOL_COMPATIBILITY_BADFILE:
	case ZPOOL_COMPATIBILITY_BADTOKEN:
	(void) fprintf(stderr, "Error: %s\n", report);
	break;

	case ZPOOL_COMPATIBILITY_WARNTOKEN:
	(void) fprintf(stderr, "Warning: %s\n", report);
	ret = ZPOOL_COMPATIBILITY_OK;
	break;
	}
	return (ret);
	}

	int
	main(int argc, char **argv)
	{
	int ret = 0;
	int i = 0;
	char *cmdname;
	char **newargv;

	(void) setlocale(LC_ALL, "");
	(void) setlocale(LC_NUMERIC, "C");
	(void) textdomain(TEXT_DOMAIN);
	srand(time(NULL));

	opterr = 0;

	/*
	* Make sure the user has specified some command.
	*/
	if (argc < 2) {
	(void) fprintf(stderr, gettext("missing command\n"));
	usage(B_FALSE);
	}

	cmdname = argv[1];

	/*
	* Special case '-?'
	*/
	if ((strcmp(cmdname, "-?") == 0) \|\| strcmp(cmdname, "--help") == 0)
	usage(B_TRUE);

	/*
	* Special case '-V\|--version'
	*/
	if ((strcmp(cmdname, "-V") == 0) \|\| (strcmp(cmdname, "--version") == 0))
	return (zpool_do_version(argc, argv));

	if ((g_zfs = libzfs_init()) == NULL) {
	(void) fprintf(stderr, "%s\n", libzfs_error_init(errno));
	return (1);
	}

	libzfs_print_on_error(g_zfs, B_TRUE);

	zfs_save_arguments(argc, argv, history_str, sizeof (history_str));

	/*
	* Many commands modify input strings for string parsing reasons.
	* We create a copy to protect the original argv.
	*/
	newargv = malloc((argc + 1) * sizeof (newargv[0]));
	for (i = 0; i < argc; i++)
	newargv[i] = strdup(argv[i]);
	newargv[argc] = NULL;

	/*
	* Run the appropriate command.
	*/
	if (find_command_idx(cmdname, &i) == 0) {
	current_command = &command_table[i];
	ret = command_table[i].func(argc - 1, newargv + 1);
	} else if (strchr(cmdname, '=')) {
	verify(find_command_idx("set", &i) == 0);
	current_command = &command_table[i];
	ret = command_table[i].func(argc, newargv);
	} else if (strcmp(cmdname, "freeze") == 0 && argc == 3) {
	/*
	* 'freeze' is a vile debugging abomination, so we treat
	* it as such.
	*/
	zfs_cmd_t zc = {"\0"};

	(void) strlcpy(zc.zc_name, argv[2], sizeof (zc.zc_name));
	ret = zfs_ioctl(g_zfs, ZFS_IOC_POOL_FREEZE, &zc);
	if (ret != 0) {
	(void) fprintf(stderr,
	gettext("failed to freeze pool: %d\n"), errno);
	ret = 1;
	}

	log_history = 0;
	} else {
	(void) fprintf(stderr, gettext("unrecognized "
	"command '%s'\n"), cmdname);
	usage(B_FALSE);
	ret = 1;
	}

	for (i = 0; i < argc; i++)
	free(newargv[i]);
	free(newargv);

	if (ret == 0 && log_history)
	(void) zpool_log_history(g_zfs, history_str);

	libzfs_fini(g_zfs);

	/*
	* The 'ZFS_ABORT' environment variable causes us to dump core on exit
	* for the purposes of running ::findleaks.
	*/
	if (getenv("ZFS_ABORT") != NULL) {
	(void) printf("dumping core by request\n");
	abort();
	}

	return (ret);
	}
	diff --git a/sys/contrib/openzfs/cmd/zpool/zpool_util.h b/sys/contrib/openzfs/cmd/zpool/zpool_util.h
	index 6665eaf0d44e..583f48cca82a 100644
	--- a/sys/contrib/openzfs/cmd/zpool/zpool_util.h
	+++ b/sys/contrib/openzfs/cmd/zpool/zpool_util.h
	@@ -1,140 +1,141 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	*/

	#ifndef ZPOOL_UTIL_H
	#define ZPOOL_UTIL_H

	#include <libnvpair.h>
	#include <libzfs.h>
	#include <libzutil.h>

	#ifdef __cplusplus
	extern "C" {
	#endif

	/* Path to scripts you can run with "zpool status/iostat -c" */
	#define ZPOOL_SCRIPTS_DIR SYSCONFDIR"/zfs/zpool.d"

	/*
	* Basic utility functions
	*/
	void *safe_malloc(size_t);
	void safe_realloc(void , size_t);
	void zpool_no_memory(void);
	uint_t num_logs(nvlist_t *nv);
	uint64_t array64_max(uint64_t array[], unsigned int len);
	int highbit64(uint64_t i);
	int lowbit64(uint64_t i);

	/*
	* Misc utility functions
	*/
	char *zpool_get_cmd_search_path(void);

	/*
	* Virtual device functions
	*/

	nvlist_t make_root_vdev(zpool_handle_t zhp, nvlist_t *props, int force,
	int check_rep, boolean_t replacing, boolean_t dryrun, int argc,
	char **argv);
	nvlist_t split_mirror_vdev(zpool_handle_t zhp, char *newname,
	nvlist_t props, splitflags_t flags, int argc, char *argv);

	/*
	* Pool list functions
	*/
	-int for_each_pool(int, char , boolean_t unavail, zprop_list_t ,
	+int for_each_pool(int, char , boolean_t unavail, zprop_list_t , zfs_type_t,
	boolean_t, zpool_iter_f, void *);

	/* Vdev list functions */
	int for_each_vdev(zpool_handle_t zhp, pool_vdev_iter_f func, void data);

	typedef struct zpool_list zpool_list_t;

	-zpool_list_t pool_list_get(int, char , zprop_list_t , boolean_t, int );
	+zpool_list_t pool_list_get(int, char , zprop_list_t *, zfs_type_t,
	+ boolean_t, int *);
	void pool_list_update(zpool_list_t *);
	int pool_list_iter(zpool_list_t , int unavail, zpool_iter_f, void );
	void pool_list_free(zpool_list_t *);
	int pool_list_count(zpool_list_t *);
	void pool_list_remove(zpool_list_t , zpool_handle_t );

	extern libzfs_handle_t *g_zfs;


	typedef struct vdev_cmd_data
	{
	char *lines; / Array of lines of output, minus the column name */
	int lines_cnt; /* Number of lines in the array */

	char *cols; / Array of column names */
	int cols_cnt; /* Number of column names */


	char path; / vdev path */
	char upath; / vdev underlying path */
	char pool; / Pool name */
	char cmd; / backpointer to cmd */
	char vdev_enc_sysfs_path; / enclosure sysfs path (if any) */
	} vdev_cmd_data_t;

	typedef struct vdev_cmd_data_list
	{
	char cmd; / Command to run */
	unsigned int count; /* Number of vdev_cmd_data items (vdevs) */

	/* fields used to select only certain vdevs, if requested */
	libzfs_handle_t *g_zfs;
	char **vdev_names;
	int vdev_names_count;
	int cb_name_flags;

	vdev_cmd_data_t data; / Array of vdevs */

	/* List of unique column names and widths */
	char **uniq_cols;
	int uniq_cols_cnt;
	int *uniq_cols_width;

	} vdev_cmd_data_list_t;

	vdev_cmd_data_list_t all_pools_for_each_vdev_run(int argc, char *argv,
	char cmd, libzfs_handle_t g_zfs, char **vdev_names, int vdev_names_count,
	int cb_name_flags);

	void free_vdev_cmd_data_list(vdev_cmd_data_list_t *vcdl);

	int check_device(const char *path, boolean_t force,
	boolean_t isspare, boolean_t iswholedisk);
	boolean_t check_sector_size_database(char path, int sector_size);
	void vdev_error(const char *fmt, ...) __attribute__((format(printf, 1, 2)));
	int check_file(const char *file, boolean_t force, boolean_t isspare);
	void after_zpool_upgrade(zpool_handle_t *zhp);
	int check_file_generic(const char *file, boolean_t force, boolean_t isspare);

	#ifdef __cplusplus
	}
	#endif

	#endif /* ZPOOL_UTIL_H */
	diff --git a/sys/contrib/openzfs/cmd/zpool/zpool_vdev.c b/sys/contrib/openzfs/cmd/zpool/zpool_vdev.c
	index dcc67e7e2014..0653a09faea2 100644
	--- a/sys/contrib/openzfs/cmd/zpool/zpool_vdev.c
	+++ b/sys/contrib/openzfs/cmd/zpool/zpool_vdev.c
	@@ -1,1880 +1,1880 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2013, 2018 by Delphix. All rights reserved.
	* Copyright (c) 2016, 2017 Intel Corporation.
	* Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>.
	*/

	/*
	* Functions to convert between a list of vdevs and an nvlist representing the
	* configuration. Each entry in the list can be one of:
	*
	* Device vdevs
	* disk=(path=..., devid=...)
	* file=(path=...)
	*
	* Group vdevs
	* raidz[1\|2]=(...)
	* mirror=(...)
	*
	* Hot spares
	*
	* While the underlying implementation supports it, group vdevs cannot contain
	* other group vdevs. All userland verification of devices is contained within
	* this file. If successful, the nvlist returned can be passed directly to the
	* kernel; we've done as much verification as possible in userland.
	*
	* Hot spares are a special case, and passed down as an array of disk vdevs, at
	* the same level as the root of the vdev tree.
	*
	* The only function exported by this file is 'make_root_vdev'. The
	* function performs several passes:
	*
	* 1. Construct the vdev specification. Performs syntax validation and
	* makes sure each device is valid.
	* 2. Check for devices in use. Using libblkid to make sure that no
	* devices are also in use. Some can be overridden using the 'force'
	* flag, others cannot.
	* 3. Check for replication errors if the 'force' flag is not specified.
	* validates that the replication level is consistent across the
	* entire pool.
	* 4. Call libzfs to label any whole disks with an EFI label.
	*/

	#include <assert.h>
	#include <ctype.h>
	#include <errno.h>
	#include <fcntl.h>
	#include <libintl.h>
	#include <libnvpair.h>
	#include <libzutil.h>
	#include <limits.h>
	#include <sys/spa.h>
	#include <stdio.h>
	#include <string.h>
	#include <unistd.h>
	#include "zpool_util.h"
	#include <sys/zfs_context.h>
	#include <sys/stat.h>

	/*
	* For any given vdev specification, we can have multiple errors. The
	* vdev_error() function keeps track of whether we have seen an error yet, and
	* prints out a header if its the first error we've seen.
	*/
	boolean_t error_seen;
	boolean_t is_force;

	void
	vdev_error(const char *fmt, ...)
	{
	va_list ap;

	if (!error_seen) {
	(void) fprintf(stderr, gettext("invalid vdev specification\n"));
	if (!is_force)
	(void) fprintf(stderr, gettext("use '-f' to override "
	"the following errors:\n"));
	else
	(void) fprintf(stderr, gettext("the following errors "
	"must be manually repaired:\n"));
	error_seen = B_TRUE;
	}

	va_start(ap, fmt);
	(void) vfprintf(stderr, fmt, ap);
	va_end(ap);
	}

	/*
	* Check that a file is valid. All we can do in this case is check that it's
	* not in use by another pool, and not in use by swap.
	*/
	int
	check_file_generic(const char *file, boolean_t force, boolean_t isspare)
	{
	char *name;
	int fd;
	int ret = 0;
	pool_state_t state;
	boolean_t inuse;

	if ((fd = open(file, O_RDONLY)) < 0)
	return (0);

	if (zpool_in_use(g_zfs, fd, &state, &name, &inuse) == 0 && inuse) {
	const char *desc;

	switch (state) {
	case POOL_STATE_ACTIVE:
	desc = gettext("active");
	break;

	case POOL_STATE_EXPORTED:
	desc = gettext("exported");
	break;

	case POOL_STATE_POTENTIALLY_ACTIVE:
	desc = gettext("potentially active");
	break;

	default:
	desc = gettext("unknown");
	break;
	}

	/*
	* Allow hot spares to be shared between pools.
	*/
	if (state == POOL_STATE_SPARE && isspare) {
	free(name);
	(void) close(fd);
	return (0);
	}

	if (state == POOL_STATE_ACTIVE \|\|
	state == POOL_STATE_SPARE \|\| !force) {
	switch (state) {
	case POOL_STATE_SPARE:
	vdev_error(gettext("%s is reserved as a hot "
	"spare for pool %s\n"), file, name);
	break;
	default:
	vdev_error(gettext("%s is part of %s pool "
	"'%s'\n"), file, desc, name);
	break;
	}
	ret = -1;
	}

	free(name);
	}

	(void) close(fd);
	return (ret);
	}

	/*
	* This may be a shorthand device path or it could be total gibberish.
	* Check to see if it is a known device available in zfs_vdev_paths.
	* As part of this check, see if we've been given an entire disk
	* (minus the slice number).
	*/
	static int
	is_shorthand_path(const char arg, char path, size_t path_size,
	struct stat64 statbuf, boolean_t wholedisk)
	{
	int error;

	error = zfs_resolve_shortname(arg, path, path_size);
	if (error == 0) {
	*wholedisk = zfs_dev_is_whole_disk(path);
	if (*wholedisk \|\| (stat64(path, statbuf) == 0))
	return (0);
	}

	strlcpy(path, arg, path_size);
	memset(statbuf, 0, sizeof (*statbuf));
	*wholedisk = B_FALSE;

	return (error);
	}

	/*
	* Determine if the given path is a hot spare within the given configuration.
	* If no configuration is given we rely solely on the label.
	*/
	static boolean_t
	is_spare(nvlist_t config, const char path)
	{
	int fd;
	pool_state_t state;
	char *name = NULL;
	nvlist_t *label;
	uint64_t guid, spareguid;
	nvlist_t *nvroot;
	nvlist_t **spares;
	uint_t i, nspares;
	boolean_t inuse;

	if (zpool_is_draid_spare(path))
	return (B_TRUE);

	if ((fd = open(path, O_RDONLY\|O_DIRECT)) < 0)
	return (B_FALSE);

	if (zpool_in_use(g_zfs, fd, &state, &name, &inuse) != 0 \|\|
	!inuse \|\|
	state != POOL_STATE_SPARE \|\|
	zpool_read_label(fd, &label, NULL) != 0) {
	free(name);
	(void) close(fd);
	return (B_FALSE);
	}
	free(name);
	(void) close(fd);

	if (config == NULL) {
	nvlist_free(label);
	return (B_TRUE);
	}

	verify(nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &guid) == 0);
	nvlist_free(label);

	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) == 0);
	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	&spares, &nspares) == 0) {
	for (i = 0; i < nspares; i++) {
	verify(nvlist_lookup_uint64(spares[i],
	ZPOOL_CONFIG_GUID, &spareguid) == 0);
	if (spareguid == guid)
	return (B_TRUE);
	}
	}

	return (B_FALSE);
	}

	/*
	* Create a leaf vdev. Determine if this is a file or a device. If it's a
	* device, fill in the device id to make a complete nvlist. Valid forms for a
	* leaf vdev are:
	*
	* /dev/xxx Complete disk path
	* /xxx Full path to file
	* xxx Shorthand for <zfs_vdev_paths>/xxx
	* draid* Virtual dRAID spare
	*/
	static nvlist_t *
	make_leaf_vdev(nvlist_t props, const char arg, boolean_t is_primary)
	{
	char path[MAXPATHLEN];
	struct stat64 statbuf;
	nvlist_t *vdev = NULL;
	char *type = NULL;
	boolean_t wholedisk = B_FALSE;
	uint64_t ashift = 0;
	int err;

	/*
	* Determine what type of vdev this is, and put the full path into
	* 'path'. We detect whether this is a device of file afterwards by
	* checking the st_mode of the file.
	*/
	if (arg[0] == '/') {
	/*
	* Complete device or file path. Exact type is determined by
	* examining the file descriptor afterwards. Symbolic links
	* are resolved to their real paths to determine whole disk
	* and S_ISBLK/S_ISREG type checks. However, we are careful
	* to store the given path as ZPOOL_CONFIG_PATH to ensure we
	* can leverage udev's persistent device labels.
	*/
	if (realpath(arg, path) == NULL) {
	(void) fprintf(stderr,
	gettext("cannot resolve path '%s'\n"), arg);
	return (NULL);
	}

	wholedisk = zfs_dev_is_whole_disk(path);
	if (!wholedisk && (stat64(path, &statbuf) != 0)) {
	(void) fprintf(stderr,
	gettext("cannot open '%s': %s\n"),
	path, strerror(errno));
	return (NULL);
	}

	/* After whole disk check restore original passed path */
	strlcpy(path, arg, sizeof (path));
	} else if (zpool_is_draid_spare(arg)) {
	if (!is_primary) {
	(void) fprintf(stderr,
	gettext("cannot open '%s': dRAID spares can only "
	"be used to replace primary vdevs\n"), arg);
	return (NULL);
	}

	wholedisk = B_TRUE;
	strlcpy(path, arg, sizeof (path));
	type = VDEV_TYPE_DRAID_SPARE;
	} else {
	err = is_shorthand_path(arg, path, sizeof (path),
	&statbuf, &wholedisk);
	if (err != 0) {
	/*
	* If we got ENOENT, then the user gave us
	* gibberish, so try to direct them with a
	* reasonable error message. Otherwise,
	* regurgitate strerror() since it's the best we
	* can do.
	*/
	if (err == ENOENT) {
	(void) fprintf(stderr,
	gettext("cannot open '%s': no such "
	"device in %s\n"), arg, DISK_ROOT);
	(void) fprintf(stderr,
	gettext("must be a full path or "
	"shorthand device name\n"));
	return (NULL);
	} else {
	(void) fprintf(stderr,
	gettext("cannot open '%s': %s\n"),
	path, strerror(errno));
	return (NULL);
	}
	}
	}

	if (type == NULL) {
	/*
	* Determine whether this is a device or a file.
	*/
	if (wholedisk \|\| S_ISBLK(statbuf.st_mode)) {
	type = VDEV_TYPE_DISK;
	} else if (S_ISREG(statbuf.st_mode)) {
	type = VDEV_TYPE_FILE;
	} else {
	fprintf(stderr, gettext("cannot use '%s': must "
	"be a block device or regular file\n"), path);
	return (NULL);
	}
	}

	/*
	* Finally, we have the complete device or file, and we know that it is
	* acceptable to use. Construct the nvlist to describe this vdev. All
	* vdevs have a 'path' element, and devices also have a 'devid' element.
	*/
	verify(nvlist_alloc(&vdev, NV_UNIQUE_NAME, 0) == 0);
	verify(nvlist_add_string(vdev, ZPOOL_CONFIG_PATH, path) == 0);
	verify(nvlist_add_string(vdev, ZPOOL_CONFIG_TYPE, type) == 0);

	if (strcmp(type, VDEV_TYPE_DISK) == 0)
	verify(nvlist_add_uint64(vdev, ZPOOL_CONFIG_WHOLE_DISK,
	(uint64_t)wholedisk) == 0);

	/*
	* Override defaults if custom properties are provided.
	*/
	if (props != NULL) {
	char *value = NULL;

	if (nvlist_lookup_string(props,
	zpool_prop_to_name(ZPOOL_PROP_ASHIFT), &value) == 0) {
	if (zfs_nicestrtonum(NULL, value, &ashift) != 0) {
	(void) fprintf(stderr,
	gettext("ashift must be a number.\n"));
	return (NULL);
	}
	if (ashift != 0 &&
	(ashift < ASHIFT_MIN \|\| ashift > ASHIFT_MAX)) {
	(void) fprintf(stderr,
	gettext("invalid 'ashift=%" PRIu64 "' "
	"property: only values between %" PRId32 " "
	"and %" PRId32 " are allowed.\n"),
	ashift, ASHIFT_MIN, ASHIFT_MAX);
	return (NULL);
	}
	}
	}

	/*
	* If the device is known to incorrectly report its physical sector
	* size explicitly provide the known correct value.
	*/
	if (ashift == 0) {
	int sector_size;

	if (check_sector_size_database(path, &sector_size) == B_TRUE)
	ashift = highbit64(sector_size) - 1;
	}

	if (ashift > 0)
	(void) nvlist_add_uint64(vdev, ZPOOL_CONFIG_ASHIFT, ashift);

	return (vdev);
	}

	/*
	* Go through and verify the replication level of the pool is consistent.
	* Performs the following checks:
	*
	* For the new spec, verifies that devices in mirrors and raidz are the
	* same size.
	*
	* If the current configuration already has inconsistent replication
	* levels, ignore any other potential problems in the new spec.
	*
	* Otherwise, make sure that the current spec (if there is one) and the new
	* spec have consistent replication levels.
	*
	* If there is no current spec (create), make sure new spec has at least
	* one general purpose vdev.
	*/
	typedef struct replication_level {
	char *zprl_type;
	uint64_t zprl_children;
	uint64_t zprl_parity;
	} replication_level_t;

	#define ZPOOL_FUZZ (16 * 1024 * 1024)

	/*
	* N.B. For the purposes of comparing replication levels dRAID can be
	* considered functionally equivalent to raidz.
	*/
	static boolean_t
	is_raidz_mirror(replication_level_t a, replication_level_t b,
	replication_level_t raidz, replication_level_t mirror)
	{
	if ((strcmp(a->zprl_type, "raidz") == 0 \|\|
	strcmp(a->zprl_type, "draid") == 0) &&
	strcmp(b->zprl_type, "mirror") == 0) {
	*raidz = a;
	*mirror = b;
	return (B_TRUE);
	}
	return (B_FALSE);
	}

	/*
	* Comparison for determining if dRAID and raidz where passed in either order.
	*/
	static boolean_t
	is_raidz_draid(replication_level_t a, replication_level_t b)
	{
	if ((strcmp(a->zprl_type, "raidz") == 0 \|\|
	strcmp(a->zprl_type, "draid") == 0) &&
	(strcmp(b->zprl_type, "raidz") == 0 \|\|
	strcmp(b->zprl_type, "draid") == 0)) {
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	/*
	* Given a list of toplevel vdevs, return the current replication level. If
	* the config is inconsistent, then NULL is returned. If 'fatal' is set, then
	* an error message will be displayed for each self-inconsistent vdev.
	*/
	static replication_level_t *
	get_replication(nvlist_t *nvroot, boolean_t fatal)
	{
	nvlist_t **top;
	uint_t t, toplevels;
	nvlist_t **child;
	uint_t c, children;
	nvlist_t *nv;
	char *type;
	replication_level_t lastrep = {0};
	replication_level_t rep;
	replication_level_t *ret;
	replication_level_t raidz, mirror;
	boolean_t dontreport;

	ret = safe_malloc(sizeof (replication_level_t));

	verify(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	&top, &toplevels) == 0);

	for (t = 0; t < toplevels; t++) {
	uint64_t is_log = B_FALSE;

	nv = top[t];

	/*
	* For separate logs we ignore the top level vdev replication
	* constraints.
	*/
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &is_log);
	if (is_log)
	continue;

	/* Ignore holes introduced by removing aux devices */
	verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0);
	if (strcmp(type, VDEV_TYPE_HOLE) == 0)
	continue;

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0) {
	/*
	* This is a 'file' or 'disk' vdev.
	*/
	rep.zprl_type = type;
	rep.zprl_children = 1;
	rep.zprl_parity = 0;
	} else {
	int64_t vdev_size;

	/*
	* This is a mirror or RAID-Z vdev. Go through and make
	* sure the contents are all the same (files vs. disks),
	* keeping track of the number of elements in the
	* process.
	*
	* We also check that the size of each vdev (if it can
	* be determined) is the same.
	*/
	rep.zprl_type = type;
	rep.zprl_children = 0;

	if (strcmp(type, VDEV_TYPE_RAIDZ) == 0 \|\|
	strcmp(type, VDEV_TYPE_DRAID) == 0) {
	verify(nvlist_lookup_uint64(nv,
	ZPOOL_CONFIG_NPARITY,
	&rep.zprl_parity) == 0);
	assert(rep.zprl_parity != 0);
	} else {
	rep.zprl_parity = 0;
	}

	/*
	* The 'dontreport' variable indicates that we've
	* already reported an error for this spec, so don't
	* bother doing it again.
	*/
	type = NULL;
	dontreport = 0;
	vdev_size = -1LL;
	for (c = 0; c < children; c++) {
	nvlist_t *cnv = child[c];
	char *path;
	struct stat64 statbuf;
	int64_t size = -1LL;
	char *childtype;
	int fd, err;

	rep.zprl_children++;

	verify(nvlist_lookup_string(cnv,
	ZPOOL_CONFIG_TYPE, &childtype) == 0);

	/*
	* If this is a replacing or spare vdev, then
	* get the real first child of the vdev: do this
	* in a loop because replacing and spare vdevs
	* can be nested.
	*/
	while (strcmp(childtype,
	VDEV_TYPE_REPLACING) == 0 \|\|
	strcmp(childtype, VDEV_TYPE_SPARE) == 0) {
	nvlist_t **rchild;
	uint_t rchildren;

	verify(nvlist_lookup_nvlist_array(cnv,
	ZPOOL_CONFIG_CHILDREN, &rchild,
	&rchildren) == 0);
	assert(rchildren == 2);
	cnv = rchild[0];

	verify(nvlist_lookup_string(cnv,
	ZPOOL_CONFIG_TYPE,
	&childtype) == 0);
	}

	verify(nvlist_lookup_string(cnv,
	ZPOOL_CONFIG_PATH, &path) == 0);

	/*
	* If we have a raidz/mirror that combines disks
	* with files, report it as an error.
	*/
	if (!dontreport && type != NULL &&
	strcmp(type, childtype) != 0) {
	if (ret != NULL)
	free(ret);
	ret = NULL;
	if (fatal)
	vdev_error(gettext(
	"mismatched replication "
	"level: %s contains both "
	"files and devices\n"),
	rep.zprl_type);
	else
	return (NULL);
	dontreport = B_TRUE;
	}

	/*
	* According to stat(2), the value of 'st_size'
	* is undefined for block devices and character
	* devices. But there is no effective way to
	* determine the real size in userland.
	*
	* Instead, we'll take advantage of an
	* implementation detail of spec_size(). If the
	* device is currently open, then we (should)
	* return a valid size.
	*
	* If we still don't get a valid size (indicated
	* by a size of 0 or MAXOFFSET_T), then ignore
	* this device altogether.
	*/
	if ((fd = open(path, O_RDONLY)) >= 0) {
	err = fstat64_blk(fd, &statbuf);
	(void) close(fd);
	} else {
	err = stat64(path, &statbuf);
	}

	if (err != 0 \|\|
	statbuf.st_size == 0 \|\|
	statbuf.st_size == MAXOFFSET_T)
	continue;

	size = statbuf.st_size;

	/*
	* Also make sure that devices and
	* slices have a consistent size. If
	* they differ by a significant amount
	* (~16MB) then report an error.
	*/
	if (!dontreport &&
	(vdev_size != -1LL &&
	(llabs(size - vdev_size) >
	ZPOOL_FUZZ))) {
	if (ret != NULL)
	free(ret);
	ret = NULL;
	if (fatal)
	vdev_error(gettext(
	"%s contains devices of "
	"different sizes\n"),
	rep.zprl_type);
	else
	return (NULL);
	dontreport = B_TRUE;
	}

	type = childtype;
	vdev_size = size;
	}
	}

	/*
	* At this point, we have the replication of the last toplevel
	* vdev in 'rep'. Compare it to 'lastrep' to see if it is
	* different.
	*/
	if (lastrep.zprl_type != NULL) {
	if (is_raidz_mirror(&lastrep, &rep, &raidz, &mirror) \|\|
	is_raidz_mirror(&rep, &lastrep, &raidz, &mirror)) {
	/*
	* Accepted raidz and mirror when they can
	* handle the same number of disk failures.
	*/
	if (raidz->zprl_parity !=
	mirror->zprl_children - 1) {
	if (ret != NULL)
	free(ret);
	ret = NULL;
	if (fatal)
	vdev_error(gettext(
	"mismatched replication "
	"level: "
	"%s and %s vdevs with "
	"different redundancy, "
	"%llu vs. %llu (%llu-way) "
	"are present\n"),
	raidz->zprl_type,
	mirror->zprl_type,
	(u_longlong_t)
	raidz->zprl_parity,
	(u_longlong_t)
	mirror->zprl_children - 1,
	(u_longlong_t)
	mirror->zprl_children);
	else
	return (NULL);
	}
	} else if (is_raidz_draid(&lastrep, &rep)) {
	/*
	* Accepted raidz and draid when they can
	* handle the same number of disk failures.
	*/
	if (lastrep.zprl_parity != rep.zprl_parity) {
	if (ret != NULL)
	free(ret);
	ret = NULL;
	if (fatal)
	vdev_error(gettext(
	"mismatched replication "
	"level: %s and %s vdevs "
	"with different "
	"redundancy, %llu vs. "
	"%llu are present\n"),
	lastrep.zprl_type,
	rep.zprl_type,
	(u_longlong_t)
	lastrep.zprl_parity,
	(u_longlong_t)
	rep.zprl_parity);
	else
	return (NULL);
	}
	} else if (strcmp(lastrep.zprl_type, rep.zprl_type) !=
	0) {
	if (ret != NULL)
	free(ret);
	ret = NULL;
	if (fatal)
	vdev_error(gettext(
	"mismatched replication level: "
	"both %s and %s vdevs are "
	"present\n"),
	lastrep.zprl_type, rep.zprl_type);
	else
	return (NULL);
	} else if (lastrep.zprl_parity != rep.zprl_parity) {
	if (ret)
	free(ret);
	ret = NULL;
	if (fatal)
	vdev_error(gettext(
	"mismatched replication level: "
	"both %llu and %llu device parity "
	"%s vdevs are present\n"),
	(u_longlong_t)
	lastrep.zprl_parity,
	(u_longlong_t)rep.zprl_parity,
	rep.zprl_type);
	else
	return (NULL);
	} else if (lastrep.zprl_children != rep.zprl_children) {
	if (ret)
	free(ret);
	ret = NULL;
	if (fatal)
	vdev_error(gettext(
	"mismatched replication level: "
	"both %llu-way and %llu-way %s "
	"vdevs are present\n"),
	(u_longlong_t)
	lastrep.zprl_children,
	(u_longlong_t)
	rep.zprl_children,
	rep.zprl_type);
	else
	return (NULL);
	}
	}
	lastrep = rep;
	}

	if (ret != NULL)
	*ret = rep;

	return (ret);
	}

	/*
	* Check the replication level of the vdev spec against the current pool. Calls
	* get_replication() to make sure the new spec is self-consistent. If the pool
	* has a consistent replication level, then we ignore any errors. Otherwise,
	* report any difference between the two.
	*/
	static int
	check_replication(nvlist_t config, nvlist_t newroot)
	{
	nvlist_t **child;
	uint_t children;
	replication_level_t current = NULL, new;
	replication_level_t raidz, mirror;
	int ret;

	/*
	* If we have a current pool configuration, check to see if it's
	* self-consistent. If not, simply return success.
	*/
	if (config != NULL) {
	nvlist_t *nvroot;

	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) == 0);
	if ((current = get_replication(nvroot, B_FALSE)) == NULL)
	return (0);
	}
	/*
	* for spares there may be no children, and therefore no
	* replication level to check
	*/
	if ((nvlist_lookup_nvlist_array(newroot, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0) \|\| (children == 0)) {
	free(current);
	return (0);
	}

	/*
	* If all we have is logs then there's no replication level to check.
	*/
	if (num_logs(newroot) == children) {
	free(current);
	return (0);
	}

	/*
	* Get the replication level of the new vdev spec, reporting any
	* inconsistencies found.
	*/
	if ((new = get_replication(newroot, B_TRUE)) == NULL) {
	free(current);
	return (-1);
	}

	/*
	* Check to see if the new vdev spec matches the replication level of
	* the current pool.
	*/
	ret = 0;
	if (current != NULL) {
	if (is_raidz_mirror(current, new, &raidz, &mirror) \|\|
	is_raidz_mirror(new, current, &raidz, &mirror)) {
	if (raidz->zprl_parity != mirror->zprl_children - 1) {
	vdev_error(gettext(
	"mismatched replication level: pool and "
	"new vdev with different redundancy, %s "
	"and %s vdevs, %llu vs. %llu (%llu-way)\n"),
	raidz->zprl_type,
	mirror->zprl_type,
	(u_longlong_t)raidz->zprl_parity,
	(u_longlong_t)mirror->zprl_children - 1,
	(u_longlong_t)mirror->zprl_children);
	ret = -1;
	}
	} else if (strcmp(current->zprl_type, new->zprl_type) != 0) {
	vdev_error(gettext(
	"mismatched replication level: pool uses %s "
	"and new vdev is %s\n"),
	current->zprl_type, new->zprl_type);
	ret = -1;
	} else if (current->zprl_parity != new->zprl_parity) {
	vdev_error(gettext(
	"mismatched replication level: pool uses %llu "
	"device parity and new vdev uses %llu\n"),
	(u_longlong_t)current->zprl_parity,
	(u_longlong_t)new->zprl_parity);
	ret = -1;
	} else if (current->zprl_children != new->zprl_children) {
	vdev_error(gettext(
	"mismatched replication level: pool uses %llu-way "
	"%s and new vdev uses %llu-way %s\n"),
	(u_longlong_t)current->zprl_children,
	current->zprl_type,
	(u_longlong_t)new->zprl_children,
	new->zprl_type);
	ret = -1;
	}
	}

	free(new);
	if (current != NULL)
	free(current);

	return (ret);
	}

	static int
	zero_label(char *path)
	{
	const int size = 4096;
	char buf[size];
	int err, fd;

	if ((fd = open(path, O_WRONLY\|O_EXCL)) < 0) {
	(void) fprintf(stderr, gettext("cannot open '%s': %s\n"),
	path, strerror(errno));
	return (-1);
	}

	memset(buf, 0, size);
	err = write(fd, buf, size);
	(void) fdatasync(fd);
	(void) close(fd);

	if (err == -1) {
	(void) fprintf(stderr, gettext("cannot zero first %d bytes "
	"of '%s': %s\n"), size, path, strerror(errno));
	return (-1);
	}

	if (err != size) {
	(void) fprintf(stderr, gettext("could only zero %d/%d bytes "
	"of '%s'\n"), err, size, path);
	return (-1);
	}

	return (0);
	}

	/*
	* Go through and find any whole disks in the vdev specification, labelling them
	* as appropriate. When constructing the vdev spec, we were unable to open this
	* device in order to provide a devid. Now that we have labelled the disk and
	* know that slice 0 is valid, we can construct the devid now.
	*
	* If the disk was already labeled with an EFI label, we will have gotten the
	* devid already (because we were able to open the whole disk). Otherwise, we
	* need to get the devid after we label the disk.
	*/
	static int
	make_disks(zpool_handle_t zhp, nvlist_t nv)
	{
	nvlist_t **child;
	uint_t c, children;
	char type, path;
	char devpath[MAXPATHLEN];
	char udevpath[MAXPATHLEN];
	uint64_t wholedisk;
	struct stat64 statbuf;
	int is_exclusive = 0;
	int fd;
	int ret;

	verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0);

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0) {

	if (strcmp(type, VDEV_TYPE_DISK) != 0)
	return (0);

	/*
	* We have a disk device. If this is a whole disk write
	* out the efi partition table, otherwise write zero's to
	* the first 4k of the partition. This is to ensure that
	* libblkid will not misidentify the partition due to a
	* magic value left by the previous filesystem.
	*/
	verify(!nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path));
	verify(!nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK,
	&wholedisk));

	if (!wholedisk) {
	/*
	* Update device id string for mpath nodes (Linux only)
	*/
	if (is_mpath_whole_disk(path))
	update_vdev_config_dev_strs(nv);

	if (!is_spare(NULL, path))
	(void) zero_label(path);
	return (0);
	}

	if (realpath(path, devpath) == NULL) {
	ret = errno;
	(void) fprintf(stderr,
	gettext("cannot resolve path '%s'\n"), path);
	return (ret);
	}

	/*
	* Remove any previously existing symlink from a udev path to
	* the device before labeling the disk. This ensures that
	* only newly created links are used. Otherwise there is a
	* window between when udev deletes and recreates the link
	* during which access attempts will fail with ENOENT.
	*/
	strlcpy(udevpath, path, MAXPATHLEN);
	(void) zfs_append_partition(udevpath, MAXPATHLEN);

	fd = open(devpath, O_RDWR\|O_EXCL);
	if (fd == -1) {
	if (errno == EBUSY)
	is_exclusive = 1;
	#ifdef __FreeBSD__
	if (errno == EPERM)
	is_exclusive = 1;
	#endif
	} else {
	(void) close(fd);
	}

	/*
	* If the partition exists, contains a valid spare label,
	* and is opened exclusively there is no need to partition
	* it. Hot spares have already been partitioned and are
	* held open exclusively by the kernel as a safety measure.
	*
	* If the provided path is for a /dev/disk/ device its
	* symbolic link will be removed, partition table created,
	* and then block until udev creates the new link.
	*/
	if (!is_exclusive && !is_spare(NULL, udevpath)) {
	char *devnode = strrchr(devpath, '/') + 1;

	ret = strncmp(udevpath, UDISK_ROOT, strlen(UDISK_ROOT));
	if (ret == 0) {
	ret = lstat64(udevpath, &statbuf);
	if (ret == 0 && S_ISLNK(statbuf.st_mode))
	(void) unlink(udevpath);
	}

	/*
	* When labeling a pool the raw device node name
	* is provided as it appears under /dev/.
	*/
	if (zpool_label_disk(g_zfs, zhp, devnode) == -1)
	return (-1);

	/*
	* Wait for udev to signal the device is available
	* by the provided path.
	*/
	ret = zpool_label_disk_wait(udevpath, DISK_LABEL_WAIT);
	if (ret) {
	(void) fprintf(stderr,
	gettext("missing link: %s was "
	"partitioned but %s is missing\n"),
	devnode, udevpath);
	return (ret);
	}

	ret = zero_label(udevpath);
	if (ret)
	return (ret);
	}

	/*
	* Update the path to refer to the partition. The presence of
	* the 'whole_disk' field indicates to the CLI that we should
	* chop off the partition number when displaying the device in
	* future output.
	*/
	verify(nvlist_add_string(nv, ZPOOL_CONFIG_PATH, udevpath) == 0);

	/*
	* Update device id strings for whole disks (Linux only)
	*/
	update_vdev_config_dev_strs(nv);

	return (0);
	}

	for (c = 0; c < children; c++)
	if ((ret = make_disks(zhp, child[c])) != 0)
	return (ret);

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES,
	&child, &children) == 0)
	for (c = 0; c < children; c++)
	if ((ret = make_disks(zhp, child[c])) != 0)
	return (ret);

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE,
	&child, &children) == 0)
	for (c = 0; c < children; c++)
	if ((ret = make_disks(zhp, child[c])) != 0)
	return (ret);

	return (0);
	}

	/*
	* Go through and find any devices that are in use. We rely on libdiskmgt for
	* the majority of this task.
	*/
	static boolean_t
	is_device_in_use(nvlist_t config, nvlist_t nv, boolean_t force,
	boolean_t replacing, boolean_t isspare)
	{
	nvlist_t **child;
	uint_t c, children;
	char type, path;
	int ret = 0;
	char buf[MAXPATHLEN];
	uint64_t wholedisk = B_FALSE;
	boolean_t anyinuse = B_FALSE;

	verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0);

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0) {

	verify(!nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path));
	if (strcmp(type, VDEV_TYPE_DISK) == 0)
	verify(!nvlist_lookup_uint64(nv,
	ZPOOL_CONFIG_WHOLE_DISK, &wholedisk));

	/*
	* As a generic check, we look to see if this is a replace of a
	* hot spare within the same pool. If so, we allow it
	* regardless of what libblkid or zpool_in_use() says.
	*/
	if (replacing) {
	(void) strlcpy(buf, path, sizeof (buf));
	if (wholedisk) {
	ret = zfs_append_partition(buf, sizeof (buf));
	if (ret == -1)
	return (-1);
	}

	if (is_spare(config, buf))
	return (B_FALSE);
	}

	if (strcmp(type, VDEV_TYPE_DISK) == 0)
	ret = check_device(path, force, isspare, wholedisk);

	else if (strcmp(type, VDEV_TYPE_FILE) == 0)
	ret = check_file(path, force, isspare);

	return (ret != 0);
	}

	for (c = 0; c < children; c++)
	if (is_device_in_use(config, child[c], force, replacing,
	B_FALSE))
	anyinuse = B_TRUE;

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES,
	&child, &children) == 0)
	for (c = 0; c < children; c++)
	if (is_device_in_use(config, child[c], force, replacing,
	B_TRUE))
	anyinuse = B_TRUE;

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE,
	&child, &children) == 0)
	for (c = 0; c < children; c++)
	if (is_device_in_use(config, child[c], force, replacing,
	B_FALSE))
	anyinuse = B_TRUE;

	return (anyinuse);
	}

	/*
	* Returns the parity level extracted from a raidz or draid type.
	* If the parity cannot be determined zero is returned.
	*/
	static int
	get_parity(const char *type)
	{
	long parity = 0;
	const char *p;

	if (strncmp(type, VDEV_TYPE_RAIDZ, strlen(VDEV_TYPE_RAIDZ)) == 0) {
	p = type + strlen(VDEV_TYPE_RAIDZ);

	if (*p == '\0') {
	/* when unspecified default to single parity */
	return (1);
	} else if (*p == '0') {
	/* no zero prefixes allowed */
	return (0);
	} else {
	/* 0-3, no suffixes allowed */
	char *end;
	errno = 0;
	parity = strtol(p, &end, 10);
	if (errno != 0 \|\| *end != '\0' \|\|
	parity < 1 \|\| parity > VDEV_RAIDZ_MAXPARITY) {
	return (0);
	}
	}
	} else if (strncmp(type, VDEV_TYPE_DRAID,
	strlen(VDEV_TYPE_DRAID)) == 0) {
	p = type + strlen(VDEV_TYPE_DRAID);

	if (p == '\0' \|\| p == ':') {
	/* when unspecified default to single parity */
	return (1);
	} else if (*p == '0') {
	/* no zero prefixes allowed */
	return (0);
	} else {
	/* 0-3, allowed suffixes: '\0' or ':' */
	char *end;
	errno = 0;
	parity = strtol(p, &end, 10);
	if (errno != 0 \|\|
	parity < 1 \|\| parity > VDEV_DRAID_MAXPARITY \|\|
	(end != '\0' && end != ':')) {
	return (0);
	}
	}
	}

	return ((int)parity);
	}

	/*
	* Assign the minimum and maximum number of devices allowed for
	* the specified type. On error NULL is returned, otherwise the
	* type prefix is returned (raidz, mirror, etc).
	*/
	static const char *
	is_grouping(const char type, int mindev, int *maxdev)
	{
	int nparity;

	if (strncmp(type, VDEV_TYPE_RAIDZ, strlen(VDEV_TYPE_RAIDZ)) == 0 \|\|
	strncmp(type, VDEV_TYPE_DRAID, strlen(VDEV_TYPE_DRAID)) == 0) {
	nparity = get_parity(type);
	if (nparity == 0)
	return (NULL);
	if (mindev != NULL)
	*mindev = nparity + 1;
	if (maxdev != NULL)
	*maxdev = 255;

	if (strncmp(type, VDEV_TYPE_RAIDZ,
	strlen(VDEV_TYPE_RAIDZ)) == 0) {
	return (VDEV_TYPE_RAIDZ);
	} else {
	return (VDEV_TYPE_DRAID);
	}
	}

	if (maxdev != NULL)
	*maxdev = INT_MAX;

	if (strcmp(type, "mirror") == 0) {
	if (mindev != NULL)
	*mindev = 2;
	return (VDEV_TYPE_MIRROR);
	}

	if (strcmp(type, "spare") == 0) {
	if (mindev != NULL)
	*mindev = 1;
	return (VDEV_TYPE_SPARE);
	}

	if (strcmp(type, "log") == 0) {
	if (mindev != NULL)
	*mindev = 1;
	return (VDEV_TYPE_LOG);
	}

	if (strcmp(type, VDEV_ALLOC_BIAS_SPECIAL) == 0 \|\|
	strcmp(type, VDEV_ALLOC_BIAS_DEDUP) == 0) {
	if (mindev != NULL)
	*mindev = 1;
	return (type);
	}

	if (strcmp(type, "cache") == 0) {
	if (mindev != NULL)
	*mindev = 1;
	return (VDEV_TYPE_L2CACHE);
	}

	return (NULL);
	}

	/*
	* Extract the configuration parameters encoded in the dRAID type and
	* use them to generate a dRAID configuration. The expected format is:
	*
	* draid[<parity>][:<data><d\|D>][:<children><c\|C>][:<spares><s\|S>]
	*
	* The intent is to be able to generate a good configuration when no
	* additional information is provided. The only mandatory component
	* of the 'type' is the 'draid' prefix. If a value is not provided
	* then reasonable defaults are used. The optional components may
	* appear in any order but the d/s/c suffix is required.
	*
	* Valid inputs:
	* - data: number of data devices per group (1-255)
	* - parity: number of parity blocks per group (1-3)
	* - spares: number of distributed spare (0-100)
	* - children: total number of devices (1-255)
	*
	* Examples:
	* - zpool create tank draid <devices...>
	* - zpool create tank draid2:8d:51c:2s <devices...>
	*/
	static int
	draid_config_by_type(nvlist_t nv, const char type, uint64_t children)
	{
	uint64_t nparity = 1;
	uint64_t nspares = 0;
	uint64_t ndata = UINT64_MAX;
	uint64_t ngroups = 1;
	long value;

	if (strncmp(type, VDEV_TYPE_DRAID, strlen(VDEV_TYPE_DRAID)) != 0)
	return (EINVAL);

	nparity = (uint64_t)get_parity(type);
	if (nparity == 0)
	return (EINVAL);

	char p = (char )type;
	while ((p = strchr(p, ':')) != NULL) {
	char *end;

	p = p + 1;
	errno = 0;

	if (!isdigit(p[0])) {
	(void) fprintf(stderr, gettext("invalid dRAID "
	"syntax; expected [:<number><c\|d\|s>] not '%s'\n"),
	type);
	return (EINVAL);
	}

	/* Expected non-zero value with c/d/s suffix */
	value = strtol(p, &end, 10);
	char suffix = tolower(*end);
	if (errno != 0 \|\|
	(suffix != 'c' && suffix != 'd' && suffix != 's')) {
	(void) fprintf(stderr, gettext("invalid dRAID "
	"syntax; expected [:<number><c\|d\|s>] not '%s'\n"),
	type);
	return (EINVAL);
	}

	if (suffix == 'c') {
	if ((uint64_t)value != children) {
	fprintf(stderr,
	gettext("invalid number of dRAID children; "
	"%llu required but %llu provided\n"),
	(u_longlong_t)value,
	(u_longlong_t)children);
	return (EINVAL);
	}
	} else if (suffix == 'd') {
	ndata = (uint64_t)value;
	} else if (suffix == 's') {
	nspares = (uint64_t)value;
	} else {
	verify(0); /* Unreachable */
	}
	}

	/*
	* When a specific number of data disks is not provided limit a
	* redundancy group to 8 data disks. This value was selected to
	* provide a reasonable tradeoff between capacity and performance.
	*/
	if (ndata == UINT64_MAX) {
	if (children > nspares + nparity) {
	ndata = MIN(children - nspares - nparity, 8);
	} else {
	fprintf(stderr, gettext("request number of "
	"distributed spares %llu and parity level %llu\n"
	"leaves no disks available for data\n"),
	(u_longlong_t)nspares, (u_longlong_t)nparity);
	return (EINVAL);
	}
	}

	/* Verify the maximum allowed group size is never exceeded. */
	if (ndata == 0 \|\| (ndata + nparity > children - nspares)) {
	fprintf(stderr, gettext("requested number of dRAID data "
	"disks per group %llu is too high,\nat most %llu disks "
	"are available for data\n"), (u_longlong_t)ndata,
	(u_longlong_t)(children - nspares - nparity));
	return (EINVAL);
	}

	if (nparity == 0 \|\| nparity > VDEV_DRAID_MAXPARITY) {
	fprintf(stderr,
	gettext("invalid dRAID parity level %llu; must be "
	"between 1 and %d\n"), (u_longlong_t)nparity,
	VDEV_DRAID_MAXPARITY);
	return (EINVAL);
	}

	/*
	* Verify the requested number of spares can be satisfied.
	* An arbitrary limit of 100 distributed spares is applied.
	*/
	if (nspares > 100 \|\| nspares > (children - (ndata + nparity))) {
	fprintf(stderr,
	gettext("invalid number of dRAID spares %llu; additional "
	"disks would be required\n"), (u_longlong_t)nspares);
	return (EINVAL);
	}

	/* Verify the requested number children is sufficient. */
	if (children < (ndata + nparity + nspares)) {
	fprintf(stderr, gettext("%llu disks were provided, but at "
	"least %llu disks are required for this config\n"),
	(u_longlong_t)children,
	(u_longlong_t)(ndata + nparity + nspares));
	}

	if (children > VDEV_DRAID_MAX_CHILDREN) {
	fprintf(stderr, gettext("%llu disks were provided, but "
	"dRAID only supports up to %u disks"),
	(u_longlong_t)children, VDEV_DRAID_MAX_CHILDREN);
	}

	/*
	* Calculate the minimum number of groups required to fill a slice.
	* This is the LCM of the stripe width (ndata + nparity) and the
	* number of data drives (children - nspares).
	*/
	while (ngroups * (ndata + nparity) % (children - nspares) != 0)
	ngroups++;

	/* Store the basic dRAID configuration. */
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_NPARITY, nparity);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_DRAID_NDATA, ndata);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_DRAID_NSPARES, nspares);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_DRAID_NGROUPS, ngroups);

	return (0);
	}

	/*
	* Construct a syntactically valid vdev specification,
	* and ensure that all devices and files exist and can be opened.
	* Note: we don't bother freeing anything in the error paths
	* because the program is just going to exit anyway.
	*/
	static nvlist_t *
	construct_spec(nvlist_t props, int argc, char *argv)
	{
	nvlist_t nvroot, nv, top, spares, **l2cache;
	int t, toplevels, mindev, maxdev, nspares, nlogs, nl2cache;
	const char type, fulltype;
	boolean_t is_log, is_special, is_dedup, is_spare;
	boolean_t seen_logs;

	top = NULL;
	toplevels = 0;
	spares = NULL;
	l2cache = NULL;
	nspares = 0;
	nlogs = 0;
	nl2cache = 0;
	is_log = is_special = is_dedup = is_spare = B_FALSE;
	seen_logs = B_FALSE;
	nvroot = NULL;

	while (argc > 0) {
	fulltype = argv[0];
	nv = NULL;

	/*
	* If it's a mirror, raidz, or draid the subsequent arguments
	* are its leaves -- until we encounter the next mirror,
	* raidz or draid.
	*/
	if ((type = is_grouping(fulltype, &mindev, &maxdev)) != NULL) {
	nvlist_t **child = NULL;
	int c, children = 0;

	if (strcmp(type, VDEV_TYPE_SPARE) == 0) {
	if (spares != NULL) {
	(void) fprintf(stderr,
	gettext("invalid vdev "
	"specification: 'spare' can be "
	"specified only once\n"));
	goto spec_out;
	}
	is_spare = B_TRUE;
	is_log = is_special = is_dedup = B_FALSE;
	}

	if (strcmp(type, VDEV_TYPE_LOG) == 0) {
	if (seen_logs) {
	(void) fprintf(stderr,
	gettext("invalid vdev "
	"specification: 'log' can be "
	"specified only once\n"));
	goto spec_out;
	}
	seen_logs = B_TRUE;
	is_log = B_TRUE;
	is_special = is_dedup = is_spare = B_FALSE;
	argc--;
	argv++;
	/*
	* A log is not a real grouping device.
	* We just set is_log and continue.
	*/
	continue;
	}

	if (strcmp(type, VDEV_ALLOC_BIAS_SPECIAL) == 0) {
	is_special = B_TRUE;
	is_log = is_dedup = is_spare = B_FALSE;
	argc--;
	argv++;
	continue;
	}

	if (strcmp(type, VDEV_ALLOC_BIAS_DEDUP) == 0) {
	is_dedup = B_TRUE;
	is_log = is_special = is_spare = B_FALSE;
	argc--;
	argv++;
	continue;
	}

	if (strcmp(type, VDEV_TYPE_L2CACHE) == 0) {
	if (l2cache != NULL) {
	(void) fprintf(stderr,
	gettext("invalid vdev "
	"specification: 'cache' can be "
	"specified only once\n"));
	goto spec_out;
	}
	is_log = is_special = B_FALSE;
	is_dedup = is_spare = B_FALSE;
	}

	if (is_log \|\| is_special \|\| is_dedup) {
	if (strcmp(type, VDEV_TYPE_MIRROR) != 0) {
	(void) fprintf(stderr,
	gettext("invalid vdev "
	"specification: unsupported '%s' "
	"device: %s\n"), is_log ? "log" :
	"special", type);
	goto spec_out;
	}
	nlogs++;
	}

	for (c = 1; c < argc; c++) {
	if (is_grouping(argv[c], NULL, NULL) != NULL)
	break;

	children++;
	child = realloc(child,
	children * sizeof (nvlist_t *));
	if (child == NULL)
	zpool_no_memory();
	if ((nv = make_leaf_vdev(props, argv[c],
	!(is_log \|\| is_special \|\| is_dedup \|\|
	is_spare))) == NULL) {
	for (c = 0; c < children - 1; c++)
	nvlist_free(child[c]);
	free(child);
	goto spec_out;
	}

	child[children - 1] = nv;
	}

	if (children < mindev) {
	(void) fprintf(stderr, gettext("invalid vdev "
	"specification: %s requires at least %d "
	"devices\n"), argv[0], mindev);
	for (c = 0; c < children; c++)
	nvlist_free(child[c]);
	free(child);
	goto spec_out;
	}

	if (children > maxdev) {
	(void) fprintf(stderr, gettext("invalid vdev "
	"specification: %s supports no more than "
	"%d devices\n"), argv[0], maxdev);
	for (c = 0; c < children; c++)
	nvlist_free(child[c]);
	free(child);
	goto spec_out;
	}

	argc -= c;
	argv += c;

	if (strcmp(type, VDEV_TYPE_SPARE) == 0) {
	spares = child;
	nspares = children;
	continue;
	} else if (strcmp(type, VDEV_TYPE_L2CACHE) == 0) {
	l2cache = child;
	nl2cache = children;
	continue;
	} else {
	/* create a top-level vdev with children */
	verify(nvlist_alloc(&nv, NV_UNIQUE_NAME,
	0) == 0);
	verify(nvlist_add_string(nv, ZPOOL_CONFIG_TYPE,
	type) == 0);
	verify(nvlist_add_uint64(nv,
	ZPOOL_CONFIG_IS_LOG, is_log) == 0);
	if (is_log) {
	verify(nvlist_add_string(nv,
	ZPOOL_CONFIG_ALLOCATION_BIAS,
	VDEV_ALLOC_BIAS_LOG) == 0);
	}
	if (is_special) {
	verify(nvlist_add_string(nv,
	ZPOOL_CONFIG_ALLOCATION_BIAS,
	VDEV_ALLOC_BIAS_SPECIAL) == 0);
	}
	if (is_dedup) {
	verify(nvlist_add_string(nv,
	ZPOOL_CONFIG_ALLOCATION_BIAS,
	VDEV_ALLOC_BIAS_DEDUP) == 0);
	}
	if (strcmp(type, VDEV_TYPE_RAIDZ) == 0) {
	verify(nvlist_add_uint64(nv,
	ZPOOL_CONFIG_NPARITY,
	mindev - 1) == 0);
	}
	if (strcmp(type, VDEV_TYPE_DRAID) == 0) {
	if (draid_config_by_type(nv,
	fulltype, children) != 0) {
	for (c = 0; c < children; c++)
	nvlist_free(child[c]);
	free(child);
	goto spec_out;
	}
	}
	verify(nvlist_add_nvlist_array(nv,
	- ZPOOL_CONFIG_CHILDREN, child,
	- children) == 0);
	+ ZPOOL_CONFIG_CHILDREN,
	+ (const nvlist_t **)child, children) == 0);

	for (c = 0; c < children; c++)
	nvlist_free(child[c]);
	free(child);
	}
	} else {
	/*
	* We have a device. Pass off to make_leaf_vdev() to
	* construct the appropriate nvlist describing the vdev.
	*/
	if ((nv = make_leaf_vdev(props, argv[0], !(is_log \|\|
	is_special \|\| is_dedup \|\| is_spare))) == NULL)
	goto spec_out;

	verify(nvlist_add_uint64(nv,
	ZPOOL_CONFIG_IS_LOG, is_log) == 0);
	if (is_log) {
	verify(nvlist_add_string(nv,
	ZPOOL_CONFIG_ALLOCATION_BIAS,
	VDEV_ALLOC_BIAS_LOG) == 0);
	nlogs++;
	}

	if (is_special) {
	verify(nvlist_add_string(nv,
	ZPOOL_CONFIG_ALLOCATION_BIAS,
	VDEV_ALLOC_BIAS_SPECIAL) == 0);
	}
	if (is_dedup) {
	verify(nvlist_add_string(nv,
	ZPOOL_CONFIG_ALLOCATION_BIAS,
	VDEV_ALLOC_BIAS_DEDUP) == 0);
	}
	argc--;
	argv++;
	}

	toplevels++;
	top = realloc(top, toplevels * sizeof (nvlist_t *));
	if (top == NULL)
	zpool_no_memory();
	top[toplevels - 1] = nv;
	}

	if (toplevels == 0 && nspares == 0 && nl2cache == 0) {
	(void) fprintf(stderr, gettext("invalid vdev "
	"specification: at least one toplevel vdev must be "
	"specified\n"));
	goto spec_out;
	}

	if (seen_logs && nlogs == 0) {
	(void) fprintf(stderr, gettext("invalid vdev specification: "
	"log requires at least 1 device\n"));
	goto spec_out;
	}

	/*
	* Finally, create nvroot and add all top-level vdevs to it.
	*/
	verify(nvlist_alloc(&nvroot, NV_UNIQUE_NAME, 0) == 0);
	verify(nvlist_add_string(nvroot, ZPOOL_CONFIG_TYPE,
	VDEV_TYPE_ROOT) == 0);
	verify(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	- top, toplevels) == 0);
	+ (const nvlist_t **)top, toplevels) == 0);
	if (nspares != 0)
	verify(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	- spares, nspares) == 0);
	+ (const nvlist_t **)spares, nspares) == 0);
	if (nl2cache != 0)
	verify(nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
	- l2cache, nl2cache) == 0);
	+ (const nvlist_t **)l2cache, nl2cache) == 0);

	spec_out:
	for (t = 0; t < toplevels; t++)
	nvlist_free(top[t]);
	for (t = 0; t < nspares; t++)
	nvlist_free(spares[t]);
	for (t = 0; t < nl2cache; t++)
	nvlist_free(l2cache[t]);

	free(spares);
	free(l2cache);
	free(top);

	return (nvroot);
	}

	nvlist_t *
	split_mirror_vdev(zpool_handle_t zhp, char newname, nvlist_t *props,
	splitflags_t flags, int argc, char **argv)
	{
	nvlist_t newroot = NULL, *child;
	uint_t c, children;

	if (argc > 0) {
	if ((newroot = construct_spec(props, argc, argv)) == NULL) {
	(void) fprintf(stderr, gettext("Unable to build a "
	"pool from the specified devices\n"));
	return (NULL);
	}

	if (!flags.dryrun && make_disks(zhp, newroot) != 0) {
	nvlist_free(newroot);
	return (NULL);
	}

	/* avoid any tricks in the spec */
	verify(nvlist_lookup_nvlist_array(newroot,
	ZPOOL_CONFIG_CHILDREN, &child, &children) == 0);
	for (c = 0; c < children; c++) {
	char *path;
	const char *type;
	int min, max;

	verify(nvlist_lookup_string(child[c],
	ZPOOL_CONFIG_PATH, &path) == 0);
	if ((type = is_grouping(path, &min, &max)) != NULL) {
	(void) fprintf(stderr, gettext("Cannot use "
	"'%s' as a device for splitting\n"), type);
	nvlist_free(newroot);
	return (NULL);
	}
	}
	}

	if (zpool_vdev_split(zhp, newname, &newroot, props, flags) != 0) {
	nvlist_free(newroot);
	return (NULL);
	}

	return (newroot);
	}

	static int
	num_normal_vdevs(nvlist_t *nvroot)
	{
	nvlist_t **top;
	uint_t t, toplevels, normal = 0;

	verify(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	&top, &toplevels) == 0);

	for (t = 0; t < toplevels; t++) {
	uint64_t log = B_FALSE;

	(void) nvlist_lookup_uint64(top[t], ZPOOL_CONFIG_IS_LOG, &log);
	if (log)
	continue;
	if (nvlist_exists(top[t], ZPOOL_CONFIG_ALLOCATION_BIAS))
	continue;

	normal++;
	}

	return (normal);
	}

	/*
	* Get and validate the contents of the given vdev specification. This ensures
	* that the nvlist returned is well-formed, that all the devices exist, and that
	* they are not currently in use by any other known consumer. The 'poolconfig'
	* parameter is the current configuration of the pool when adding devices
	* existing pool, and is used to perform additional checks, such as changing the
	* replication level of the pool. It can be 'NULL' to indicate that this is a
	* new pool. The 'force' flag controls whether devices should be forcefully
	* added, even if they appear in use.
	*/
	nvlist_t *
	make_root_vdev(zpool_handle_t zhp, nvlist_t props, int force, int check_rep,
	boolean_t replacing, boolean_t dryrun, int argc, char **argv)
	{
	nvlist_t *newroot;
	nvlist_t *poolconfig = NULL;
	is_force = force;

	/*
	* Construct the vdev specification. If this is successful, we know
	* that we have a valid specification, and that all devices can be
	* opened.
	*/
	if ((newroot = construct_spec(props, argc, argv)) == NULL)
	return (NULL);

	if (zhp && ((poolconfig = zpool_get_config(zhp, NULL)) == NULL)) {
	nvlist_free(newroot);
	return (NULL);
	}

	/*
	* Validate each device to make sure that it's not shared with another
	* subsystem. We do this even if 'force' is set, because there are some
	* uses (such as a dedicated dump device) that even '-f' cannot
	* override.
	*/
	if (is_device_in_use(poolconfig, newroot, force, replacing, B_FALSE)) {
	nvlist_free(newroot);
	return (NULL);
	}

	/*
	* Check the replication level of the given vdevs and report any errors
	* found. We include the existing pool spec, if any, as we need to
	* catch changes against the existing replication level.
	*/
	if (check_rep && check_replication(poolconfig, newroot) != 0) {
	nvlist_free(newroot);
	return (NULL);
	}

	/*
	* On pool create the new vdev spec must have one normal vdev.
	*/
	if (poolconfig == NULL && num_normal_vdevs(newroot) == 0) {
	vdev_error(gettext("at least one general top-level vdev must "
	"be specified\n"));
	nvlist_free(newroot);
	return (NULL);
	}

	/*
	* Run through the vdev specification and label any whole disks found.
	*/
	if (!dryrun && make_disks(zhp, newroot) != 0) {
	nvlist_free(newroot);
	return (NULL);
	}

	return (newroot);
	}
	diff --git a/sys/contrib/openzfs/cmd/ztest/ztest.c b/sys/contrib/openzfs/cmd/ztest/ztest.c
	index 5a5c381409a2..a2568962d877 100644
	--- a/sys/contrib/openzfs/cmd/ztest/ztest.c
	+++ b/sys/contrib/openzfs/cmd/ztest/ztest.c
	@@ -1,8202 +1,8204 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2018 by Delphix. All rights reserved.
	* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2013 Steven Hartland. All rights reserved.
	* Copyright (c) 2014 Integros [integros.com]
	* Copyright 2017 Joyent, Inc.
	* Copyright (c) 2017, Intel Corporation.
	*/

	/*
	* The objective of this program is to provide a DMU/ZAP/SPA stress test
	* that runs entirely in userland, is easy to use, and easy to extend.
	*
	* The overall design of the ztest program is as follows:
	*
	* (1) For each major functional area (e.g. adding vdevs to a pool,
	* creating and destroying datasets, reading and writing objects, etc)
	* we have a simple routine to test that functionality. These
	* individual routines do not have to do anything "stressful".
	*
	* (2) We turn these simple functionality tests into a stress test by
	* running them all in parallel, with as many threads as desired,
	* and spread across as many datasets, objects, and vdevs as desired.
	*
	* (3) While all this is happening, we inject faults into the pool to
	* verify that self-healing data really works.
	*
	* (4) Every time we open a dataset, we change its checksum and compression
	* functions. Thus even individual objects vary from block to block
	* in which checksum they use and whether they're compressed.
	*
	* (5) To verify that we never lose on-disk consistency after a crash,
	* we run the entire test in a child of the main process.
	* At random times, the child self-immolates with a SIGKILL.
	* This is the software equivalent of pulling the power cord.
	* The parent then runs the test again, using the existing
	* storage pool, as many times as desired. If backwards compatibility
	* testing is enabled ztest will sometimes run the "older" version
	* of ztest after a SIGKILL.
	*
	* (6) To verify that we don't have future leaks or temporal incursions,
	* many of the functional tests record the transaction group number
	* as part of their data. When reading old data, they verify that
	* the transaction group number is less than the current, open txg.
	* If you add a new test, please do this if applicable.
	*
	* (7) Threads are created with a reduced stack size, for sanity checking.
	* Therefore, it's important not to allocate huge buffers on the stack.
	*
	* When run with no arguments, ztest runs for about five minutes and
	* produces no output if successful. To get a little bit of information,
	* specify -V. To get more information, specify -VV, and so on.
	*
	* To turn this into an overnight stress test, use -T to specify run time.
	*
	* You can ask more vdevs [-v], datasets [-d], or threads [-t]
	* to increase the pool capacity, fanout, and overall stress level.
	*
	* Use the -k option to set the desired frequency of kills.
	*
	* When ztest invokes itself it passes all relevant information through a
	* temporary file which is mmap-ed in the child process. This allows shared
	* memory to survive the exec syscall. The ztest_shared_hdr_t struct is always
	* stored at offset 0 of this file and contains information on the size and
	* number of shared structures in the file. The information stored in this file
	* must remain backwards compatible with older versions of ztest so that
	* ztest can invoke them during backwards compatibility testing (-B).
	*/

	#include <sys/zfs_context.h>
	#include <sys/spa.h>
	#include <sys/dmu.h>
	#include <sys/txg.h>
	#include <sys/dbuf.h>
	#include <sys/zap.h>
	#include <sys/dmu_objset.h>
	#include <sys/poll.h>
	#include <sys/stat.h>
	#include <sys/time.h>
	#include <sys/wait.h>
	#include <sys/mman.h>
	#include <sys/resource.h>
	#include <sys/zio.h>
	#include <sys/zil.h>
	#include <sys/zil_impl.h>
	#include <sys/vdev_draid.h>
	#include <sys/vdev_impl.h>
	#include <sys/vdev_file.h>
	#include <sys/vdev_initialize.h>
	#include <sys/vdev_raidz.h>
	#include <sys/vdev_trim.h>
	#include <sys/spa_impl.h>
	#include <sys/metaslab_impl.h>
	#include <sys/dsl_prop.h>
	#include <sys/dsl_dataset.h>
	#include <sys/dsl_destroy.h>
	#include <sys/dsl_scan.h>
	#include <sys/zio_checksum.h>
	#include <sys/zfs_refcount.h>
	#include <sys/zfeature.h>
	#include <sys/dsl_userhold.h>
	#include <sys/abd.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <unistd.h>
	#include <getopt.h>
	#include <signal.h>
	#include <umem.h>
	#include <ctype.h>
	#include <math.h>
	#include <sys/fs/zfs.h>
	#include <zfs_fletcher.h>
	#include <libnvpair.h>
	#include <libzutil.h>
	#include <sys/crypto/icp.h>
	#if (__GLIBC__ && !__UCLIBC__)
	#include <execinfo.h> /* for backtrace() */
	#endif

	static int ztest_fd_data = -1;
	static int ztest_fd_rand = -1;

	typedef struct ztest_shared_hdr {
	uint64_t zh_hdr_size;
	uint64_t zh_opts_size;
	uint64_t zh_size;
	uint64_t zh_stats_size;
	uint64_t zh_stats_count;
	uint64_t zh_ds_size;
	uint64_t zh_ds_count;
	} ztest_shared_hdr_t;

	static ztest_shared_hdr_t *ztest_shared_hdr;

	enum ztest_class_state {
	ZTEST_VDEV_CLASS_OFF,
	ZTEST_VDEV_CLASS_ON,
	ZTEST_VDEV_CLASS_RND
	};

	#define ZO_GVARS_MAX_ARGLEN ((size_t)64)
	#define ZO_GVARS_MAX_COUNT ((size_t)10)

	typedef struct ztest_shared_opts {
	char zo_pool[ZFS_MAX_DATASET_NAME_LEN];
	char zo_dir[ZFS_MAX_DATASET_NAME_LEN];
	char zo_alt_ztest[MAXNAMELEN];
	char zo_alt_libpath[MAXNAMELEN];
	uint64_t zo_vdevs;
	uint64_t zo_vdevtime;
	size_t zo_vdev_size;
	int zo_ashift;
	int zo_mirrors;
	int zo_raid_children;
	int zo_raid_parity;
	char zo_raid_type[8];
	int zo_draid_data;
	int zo_draid_spares;
	int zo_datasets;
	int zo_threads;
	uint64_t zo_passtime;
	uint64_t zo_killrate;
	int zo_verbose;
	int zo_init;
	uint64_t zo_time;
	uint64_t zo_maxloops;
	uint64_t zo_metaslab_force_ganging;
	int zo_mmp_test;
	int zo_special_vdevs;
	int zo_dump_dbgmsg;
	int zo_gvars_count;
	char zo_gvars[ZO_GVARS_MAX_COUNT][ZO_GVARS_MAX_ARGLEN];
	} ztest_shared_opts_t;

	/* Default values for command line options. */
	#define DEFAULT_POOL "ztest"
	#define DEFAULT_VDEV_DIR "/tmp"
	#define DEFAULT_VDEV_COUNT 5
	#define DEFAULT_VDEV_SIZE (SPA_MINDEVSIZE * 4) /* 256m default size */
	#define DEFAULT_VDEV_SIZE_STR "256M"
	#define DEFAULT_ASHIFT SPA_MINBLOCKSHIFT
	#define DEFAULT_MIRRORS 2
	#define DEFAULT_RAID_CHILDREN 4
	#define DEFAULT_RAID_PARITY 1
	#define DEFAULT_DRAID_DATA 4
	#define DEFAULT_DRAID_SPARES 1
	#define DEFAULT_DATASETS_COUNT 7
	#define DEFAULT_THREADS 23
	#define DEFAULT_RUN_TIME 300 /* 300 seconds */
	#define DEFAULT_RUN_TIME_STR "300 sec"
	#define DEFAULT_PASS_TIME 60 /* 60 seconds */
	#define DEFAULT_PASS_TIME_STR "60 sec"
	#define DEFAULT_KILL_RATE 70 /* 70% kill rate */
	#define DEFAULT_KILLRATE_STR "70%"
	#define DEFAULT_INITS 1
	#define DEFAULT_MAX_LOOPS 50 /* 5 minutes */
	#define DEFAULT_FORCE_GANGING (64 << 10)
	#define DEFAULT_FORCE_GANGING_STR "64K"

	/* Simplifying assumption: -1 is not a valid default. */
	#define NO_DEFAULT -1

	static const ztest_shared_opts_t ztest_opts_defaults = {
	.zo_pool = DEFAULT_POOL,
	.zo_dir = DEFAULT_VDEV_DIR,
	.zo_alt_ztest = { '\0' },
	.zo_alt_libpath = { '\0' },
	.zo_vdevs = DEFAULT_VDEV_COUNT,
	.zo_ashift = DEFAULT_ASHIFT,
	.zo_mirrors = DEFAULT_MIRRORS,
	.zo_raid_children = DEFAULT_RAID_CHILDREN,
	.zo_raid_parity = DEFAULT_RAID_PARITY,
	.zo_raid_type = VDEV_TYPE_RAIDZ,
	.zo_vdev_size = DEFAULT_VDEV_SIZE,
	.zo_draid_data = DEFAULT_DRAID_DATA, /* data drives */
	.zo_draid_spares = DEFAULT_DRAID_SPARES, /* distributed spares */
	.zo_datasets = DEFAULT_DATASETS_COUNT,
	.zo_threads = DEFAULT_THREADS,
	.zo_passtime = DEFAULT_PASS_TIME,
	.zo_killrate = DEFAULT_KILL_RATE,
	.zo_verbose = 0,
	.zo_mmp_test = 0,
	.zo_init = DEFAULT_INITS,
	.zo_time = DEFAULT_RUN_TIME,
	.zo_maxloops = DEFAULT_MAX_LOOPS, /* max loops during spa_freeze() */
	.zo_metaslab_force_ganging = DEFAULT_FORCE_GANGING,
	.zo_special_vdevs = ZTEST_VDEV_CLASS_RND,
	.zo_gvars_count = 0,
	};

	extern uint64_t metaslab_force_ganging;
	extern uint64_t metaslab_df_alloc_threshold;
	extern unsigned long zfs_deadman_synctime_ms;
	extern int metaslab_preload_limit;
	extern boolean_t zfs_compressed_arc_enabled;
	extern int zfs_abd_scatter_enabled;
	extern int dmu_object_alloc_chunk_shift;
	extern boolean_t zfs_force_some_double_word_sm_entries;
	extern unsigned long zio_decompress_fail_fraction;
	extern unsigned long zfs_reconstruct_indirect_damage_fraction;


	static ztest_shared_opts_t *ztest_shared_opts;
	static ztest_shared_opts_t ztest_opts;
	static char *ztest_wkeydata = "abcdefghijklmnopqrstuvwxyz012345";

	typedef struct ztest_shared_ds {
	uint64_t zd_seq;
	} ztest_shared_ds_t;

	static ztest_shared_ds_t *ztest_shared_ds;
	#define ZTEST_GET_SHARED_DS(d) (&ztest_shared_ds[d])

	#define BT_MAGIC 0x123456789abcdefULL
	#define MAXFAULTS(zs) \
	(MAX((zs)->zs_mirrors, 1) * (ztest_opts.zo_raid_parity + 1) - 1)

	enum ztest_io_type {
	ZTEST_IO_WRITE_TAG,
	ZTEST_IO_WRITE_PATTERN,
	ZTEST_IO_WRITE_ZEROES,
	ZTEST_IO_TRUNCATE,
	ZTEST_IO_SETATTR,
	ZTEST_IO_REWRITE,
	ZTEST_IO_TYPES
	};

	typedef struct ztest_block_tag {
	uint64_t bt_magic;
	uint64_t bt_objset;
	uint64_t bt_object;
	uint64_t bt_dnodesize;
	uint64_t bt_offset;
	uint64_t bt_gen;
	uint64_t bt_txg;
	uint64_t bt_crtxg;
	} ztest_block_tag_t;

	typedef struct bufwad {
	uint64_t bw_index;
	uint64_t bw_txg;
	uint64_t bw_data;
	} bufwad_t;

	/*
	* It would be better to use a rangelock_t per object. Unfortunately
	* the rangelock_t is not a drop-in replacement for rl_t, because we
	* still need to map from object ID to rangelock_t.
	*/
	typedef enum {
	RL_READER,
	RL_WRITER,
	RL_APPEND
	} rl_type_t;

	typedef struct rll {
	void *rll_writer;
	int rll_readers;
	kmutex_t rll_lock;
	kcondvar_t rll_cv;
	} rll_t;

	typedef struct rl {
	uint64_t rl_object;
	uint64_t rl_offset;
	uint64_t rl_size;
	rll_t *rl_lock;
	} rl_t;

	#define ZTEST_RANGE_LOCKS 64
	#define ZTEST_OBJECT_LOCKS 64

	/*
	* Object descriptor. Used as a template for object lookup/create/remove.
	*/
	typedef struct ztest_od {
	uint64_t od_dir;
	uint64_t od_object;
	dmu_object_type_t od_type;
	dmu_object_type_t od_crtype;
	uint64_t od_blocksize;
	uint64_t od_crblocksize;
	uint64_t od_crdnodesize;
	uint64_t od_gen;
	uint64_t od_crgen;
	char od_name[ZFS_MAX_DATASET_NAME_LEN];
	} ztest_od_t;

	/*
	* Per-dataset state.
	*/
	typedef struct ztest_ds {
	ztest_shared_ds_t *zd_shared;
	objset_t *zd_os;
	pthread_rwlock_t zd_zilog_lock;
	zilog_t *zd_zilog;
	ztest_od_t zd_od; / debugging aid */
	char zd_name[ZFS_MAX_DATASET_NAME_LEN];
	kmutex_t zd_dirobj_lock;
	rll_t zd_object_lock[ZTEST_OBJECT_LOCKS];
	rll_t zd_range_lock[ZTEST_RANGE_LOCKS];
	} ztest_ds_t;

	/*
	* Per-iteration state.
	*/
	typedef void ztest_func_t(ztest_ds_t *zd, uint64_t id);

	typedef struct ztest_info {
	ztest_func_t zi_func; / test function */
	uint64_t zi_iters; /* iterations per execution */
	uint64_t zi_interval; / execute every <interval> seconds */
	const char zi_funcname; / name of test function */
	} ztest_info_t;

	typedef struct ztest_shared_callstate {
	uint64_t zc_count; /* per-pass count */
	uint64_t zc_time; /* per-pass time */
	uint64_t zc_next; /* next time to call this function */
	} ztest_shared_callstate_t;

	static ztest_shared_callstate_t *ztest_shared_callstate;
	#define ZTEST_GET_SHARED_CALLSTATE(c) (&ztest_shared_callstate[c])

	ztest_func_t ztest_dmu_read_write;
	ztest_func_t ztest_dmu_write_parallel;
	ztest_func_t ztest_dmu_object_alloc_free;
	ztest_func_t ztest_dmu_object_next_chunk;
	ztest_func_t ztest_dmu_commit_callbacks;
	ztest_func_t ztest_zap;
	ztest_func_t ztest_zap_parallel;
	ztest_func_t ztest_zil_commit;
	ztest_func_t ztest_zil_remount;
	ztest_func_t ztest_dmu_read_write_zcopy;
	ztest_func_t ztest_dmu_objset_create_destroy;
	ztest_func_t ztest_dmu_prealloc;
	ztest_func_t ztest_fzap;
	ztest_func_t ztest_dmu_snapshot_create_destroy;
	ztest_func_t ztest_dsl_prop_get_set;
	ztest_func_t ztest_spa_prop_get_set;
	ztest_func_t ztest_spa_create_destroy;
	ztest_func_t ztest_fault_inject;
	ztest_func_t ztest_dmu_snapshot_hold;
	ztest_func_t ztest_mmp_enable_disable;
	ztest_func_t ztest_scrub;
	ztest_func_t ztest_dsl_dataset_promote_busy;
	ztest_func_t ztest_vdev_attach_detach;
	ztest_func_t ztest_vdev_LUN_growth;
	ztest_func_t ztest_vdev_add_remove;
	ztest_func_t ztest_vdev_class_add;
	ztest_func_t ztest_vdev_aux_add_remove;
	ztest_func_t ztest_split_pool;
	ztest_func_t ztest_reguid;
	ztest_func_t ztest_spa_upgrade;
	ztest_func_t ztest_device_removal;
	ztest_func_t ztest_spa_checkpoint_create_discard;
	ztest_func_t ztest_initialize;
	ztest_func_t ztest_trim;
	ztest_func_t ztest_fletcher;
	ztest_func_t ztest_fletcher_incr;
	ztest_func_t ztest_verify_dnode_bt;

	uint64_t zopt_always = 0ULL * NANOSEC; /* all the time */
	uint64_t zopt_incessant = 1ULL * NANOSEC / 10; /* every 1/10 second */
	uint64_t zopt_often = 1ULL * NANOSEC; /* every second */
	uint64_t zopt_sometimes = 10ULL * NANOSEC; /* every 10 seconds */
	uint64_t zopt_rarely = 60ULL * NANOSEC; /* every 60 seconds */

	#define ZTI_INIT(func, iters, interval) \
	{ .zi_func = (func), \
	.zi_iters = (iters), \
	.zi_interval = (interval), \
	.zi_funcname = # func }

	ztest_info_t ztest_info[] = {
	ZTI_INIT(ztest_dmu_read_write, 1, &zopt_always),
	ZTI_INIT(ztest_dmu_write_parallel, 10, &zopt_always),
	ZTI_INIT(ztest_dmu_object_alloc_free, 1, &zopt_always),
	ZTI_INIT(ztest_dmu_object_next_chunk, 1, &zopt_sometimes),
	ZTI_INIT(ztest_dmu_commit_callbacks, 1, &zopt_always),
	ZTI_INIT(ztest_zap, 30, &zopt_always),
	ZTI_INIT(ztest_zap_parallel, 100, &zopt_always),
	ZTI_INIT(ztest_split_pool, 1, &zopt_always),
	ZTI_INIT(ztest_zil_commit, 1, &zopt_incessant),
	ZTI_INIT(ztest_zil_remount, 1, &zopt_sometimes),
	ZTI_INIT(ztest_dmu_read_write_zcopy, 1, &zopt_often),
	ZTI_INIT(ztest_dmu_objset_create_destroy, 1, &zopt_often),
	ZTI_INIT(ztest_dsl_prop_get_set, 1, &zopt_often),
	ZTI_INIT(ztest_spa_prop_get_set, 1, &zopt_sometimes),
	#if 0
	ZTI_INIT(ztest_dmu_prealloc, 1, &zopt_sometimes),
	#endif
	ZTI_INIT(ztest_fzap, 1, &zopt_sometimes),
	ZTI_INIT(ztest_dmu_snapshot_create_destroy, 1, &zopt_sometimes),
	ZTI_INIT(ztest_spa_create_destroy, 1, &zopt_sometimes),
	ZTI_INIT(ztest_fault_inject, 1, &zopt_sometimes),
	ZTI_INIT(ztest_dmu_snapshot_hold, 1, &zopt_sometimes),
	ZTI_INIT(ztest_mmp_enable_disable, 1, &zopt_sometimes),
	ZTI_INIT(ztest_reguid, 1, &zopt_rarely),
	ZTI_INIT(ztest_scrub, 1, &zopt_rarely),
	ZTI_INIT(ztest_spa_upgrade, 1, &zopt_rarely),
	ZTI_INIT(ztest_dsl_dataset_promote_busy, 1, &zopt_rarely),
	ZTI_INIT(ztest_vdev_attach_detach, 1, &zopt_sometimes),
	ZTI_INIT(ztest_vdev_LUN_growth, 1, &zopt_rarely),
	ZTI_INIT(ztest_vdev_add_remove, 1, &ztest_opts.zo_vdevtime),
	ZTI_INIT(ztest_vdev_class_add, 1, &ztest_opts.zo_vdevtime),
	ZTI_INIT(ztest_vdev_aux_add_remove, 1, &ztest_opts.zo_vdevtime),
	ZTI_INIT(ztest_device_removal, 1, &zopt_sometimes),
	ZTI_INIT(ztest_spa_checkpoint_create_discard, 1, &zopt_rarely),
	ZTI_INIT(ztest_initialize, 1, &zopt_sometimes),
	ZTI_INIT(ztest_trim, 1, &zopt_sometimes),
	ZTI_INIT(ztest_fletcher, 1, &zopt_rarely),
	ZTI_INIT(ztest_fletcher_incr, 1, &zopt_rarely),
	ZTI_INIT(ztest_verify_dnode_bt, 1, &zopt_sometimes),
	};

	#define ZTEST_FUNCS (sizeof (ztest_info) / sizeof (ztest_info_t))

	/*
	* The following struct is used to hold a list of uncalled commit callbacks.
	* The callbacks are ordered by txg number.
	*/
	typedef struct ztest_cb_list {
	kmutex_t zcl_callbacks_lock;
	list_t zcl_callbacks;
	} ztest_cb_list_t;

	/*
	* Stuff we need to share writably between parent and child.
	*/
	typedef struct ztest_shared {
	boolean_t zs_do_init;
	hrtime_t zs_proc_start;
	hrtime_t zs_proc_stop;
	hrtime_t zs_thread_start;
	hrtime_t zs_thread_stop;
	hrtime_t zs_thread_kill;
	uint64_t zs_enospc_count;
	uint64_t zs_vdev_next_leaf;
	uint64_t zs_vdev_aux;
	uint64_t zs_alloc;
	uint64_t zs_space;
	uint64_t zs_splits;
	uint64_t zs_mirrors;
	uint64_t zs_metaslab_sz;
	uint64_t zs_metaslab_df_alloc_threshold;
	uint64_t zs_guid;
	} ztest_shared_t;

	#define ID_PARALLEL -1ULL

	static char ztest_dev_template[] = "%s/%s.%llua";
	static char ztest_aux_template[] = "%s/%s.%s.%llu";
	ztest_shared_t *ztest_shared;

	static spa_t *ztest_spa = NULL;
	static ztest_ds_t *ztest_ds;

	static kmutex_t ztest_vdev_lock;
	static boolean_t ztest_device_removal_active = B_FALSE;
	static boolean_t ztest_pool_scrubbed = B_FALSE;
	static kmutex_t ztest_checkpoint_lock;

	/*
	* The ztest_name_lock protects the pool and dataset namespace used by
	* the individual tests. To modify the namespace, consumers must grab
	* this lock as writer. Grabbing the lock as reader will ensure that the
	* namespace does not change while the lock is held.
	*/
	static pthread_rwlock_t ztest_name_lock;

	static boolean_t ztest_dump_core = B_TRUE;
	static boolean_t ztest_exiting;

	/* Global commit callback list */
	static ztest_cb_list_t zcl;
	/* Commit cb delay */
	static uint64_t zc_min_txg_delay = UINT64_MAX;
	static int zc_cb_counter = 0;

	/*
	* Minimum number of commit callbacks that need to be registered for us to check
	* whether the minimum txg delay is acceptable.
	*/
	#define ZTEST_COMMIT_CB_MIN_REG 100

	/*
	* If a number of txgs equal to this threshold have been created after a commit
	* callback has been registered but not called, then we assume there is an
	* implementation bug.
	*/
	#define ZTEST_COMMIT_CB_THRESH (TXG_CONCURRENT_STATES + 1000)

	enum ztest_object {
	ZTEST_META_DNODE = 0,
	ZTEST_DIROBJ,
	ZTEST_OBJECTS
	};

	static void usage(boolean_t) __NORETURN;
	static int ztest_scrub_impl(spa_t *spa);

	/*
	* These libumem hooks provide a reasonable set of defaults for the allocator's
	* debugging facilities.
	*/
	const char *
	_umem_debug_init(void)
	{
	return ("default,verbose"); /* $UMEM_DEBUG setting */
	}

	const char *
	_umem_logging_init(void)
	{
	return ("fail,contents"); /* $UMEM_LOGGING setting */
	}

	static void
	dump_debug_buffer(void)
	{
	ssize_t ret __attribute__((unused));

	if (!ztest_opts.zo_dump_dbgmsg)
	return;

	/*
	* We use write() instead of printf() so that this function
	* is safe to call from a signal handler.
	*/
	ret = write(STDOUT_FILENO, "\n", 1);
	zfs_dbgmsg_print("ztest");
	}

	#define BACKTRACE_SZ 100

	static void sig_handler(int signo)
	{
	struct sigaction action;
	#if (__GLIBC__ && !__UCLIBC__) /* backtrace() is a GNU extension */
	int nptrs;
	void *buffer[BACKTRACE_SZ];

	nptrs = backtrace(buffer, BACKTRACE_SZ);
	backtrace_symbols_fd(buffer, nptrs, STDERR_FILENO);
	#endif
	dump_debug_buffer();

	/*
	* Restore default action and re-raise signal so SIGSEGV and
	* SIGABRT can trigger a core dump.
	*/
	action.sa_handler = SIG_DFL;
	sigemptyset(&action.sa_mask);
	action.sa_flags = 0;
	(void) sigaction(signo, &action, NULL);
	raise(signo);
	}

	#define FATAL_MSG_SZ 1024

	char *fatal_msg;

	static __attribute__((noreturn)) __attribute__((format(printf, 2, 3))) void
	fatal(int do_perror, char *message, ...)
	{
	va_list args;
	int save_errno = errno;
	char *buf;

	(void) fflush(stdout);
	buf = umem_alloc(FATAL_MSG_SZ, UMEM_NOFAIL);

	va_start(args, message);
	(void) sprintf(buf, "ztest: ");
	/* LINTED */
	(void) vsprintf(buf + strlen(buf), message, args);
	va_end(args);
	if (do_perror) {
	(void) snprintf(buf + strlen(buf), FATAL_MSG_SZ - strlen(buf),
	": %s", strerror(save_errno));
	}
	(void) fprintf(stderr, "%s\n", buf);
	fatal_msg = buf; /* to ease debugging */

	if (ztest_dump_core)
	abort();
	else
	dump_debug_buffer();

	exit(3);
	}

	static int
	str2shift(const char *buf)
	{
	const char *ends = "BKMGTPEZ";
	int i;

	if (buf[0] == '\0')
	return (0);
	for (i = 0; i < strlen(ends); i++) {
	if (toupper(buf[0]) == ends[i])
	break;
	}
	if (i == strlen(ends)) {
	(void) fprintf(stderr, "ztest: invalid bytes suffix: %s\n",
	buf);
	usage(B_FALSE);
	}
	if (buf[1] == '\0' \|\| (toupper(buf[1]) == 'B' && buf[2] == '\0')) {
	return (10*i);
	}
	(void) fprintf(stderr, "ztest: invalid bytes suffix: %s\n", buf);
	usage(B_FALSE);
	}

	static uint64_t
	nicenumtoull(const char *buf)
	{
	char *end;
	uint64_t val;

	val = strtoull(buf, &end, 0);
	if (end == buf) {
	(void) fprintf(stderr, "ztest: bad numeric value: %s\n", buf);
	usage(B_FALSE);
	} else if (end[0] == '.') {
	double fval = strtod(buf, &end);
	fval *= pow(2, str2shift(end));
	/*
	* UINT64_MAX is not exactly representable as a double.
	* The closest representation is UINT64_MAX + 1, so we
	* use a >= comparison instead of > for the bounds check.
	*/
	if (fval >= (double)UINT64_MAX) {
	(void) fprintf(stderr, "ztest: value too large: %s\n",
	buf);
	usage(B_FALSE);
	}
	val = (uint64_t)fval;
	} else {
	int shift = str2shift(end);
	if (shift >= 64 \|\| (val << shift) >> shift != val) {
	(void) fprintf(stderr, "ztest: value too large: %s\n",
	buf);
	usage(B_FALSE);
	}
	val <<= shift;
	}
	return (val);
	}

	typedef struct ztest_option {
	const char short_opt;
	const char *long_opt;
	const char *long_opt_param;
	const char *comment;
	unsigned int default_int;
	char *default_str;
	} ztest_option_t;

	/*
	* The following option_table is used for generating the usage info as well as
	* the long and short option information for calling getopt_long().
	*/
	static ztest_option_t option_table[] = {
	{ 'v', "vdevs", "INTEGER", "Number of vdevs", DEFAULT_VDEV_COUNT,
	NULL},
	{ 's', "vdev-size", "INTEGER", "Size of each vdev",
	NO_DEFAULT, DEFAULT_VDEV_SIZE_STR},
	{ 'a', "alignment-shift", "INTEGER",
	"Alignment shift; use 0 for random", DEFAULT_ASHIFT, NULL},
	{ 'm', "mirror-copies", "INTEGER", "Number of mirror copies",
	DEFAULT_MIRRORS, NULL},
	{ 'r', "raid-disks", "INTEGER", "Number of raidz/draid disks",
	DEFAULT_RAID_CHILDREN, NULL},
	{ 'R', "raid-parity", "INTEGER", "Raid parity",
	DEFAULT_RAID_PARITY, NULL},
	{ 'K', "raid-kind", "raidz\|draid\|random", "Raid kind",
	NO_DEFAULT, "random"},
	{ 'D', "draid-data", "INTEGER", "Number of draid data drives",
	DEFAULT_DRAID_DATA, NULL},
	{ 'S', "draid-spares", "INTEGER", "Number of draid spares",
	DEFAULT_DRAID_SPARES, NULL},
	{ 'd', "datasets", "INTEGER", "Number of datasets",
	DEFAULT_DATASETS_COUNT, NULL},
	{ 't', "threads", "INTEGER", "Number of ztest threads",
	DEFAULT_THREADS, NULL},
	{ 'g', "gang-block-threshold", "INTEGER",
	"Metaslab gang block threshold",
	NO_DEFAULT, DEFAULT_FORCE_GANGING_STR},
	{ 'i', "init-count", "INTEGER", "Number of times to initialize pool",
	DEFAULT_INITS, NULL},
	{ 'k', "kill-percentage", "INTEGER", "Kill percentage",
	NO_DEFAULT, DEFAULT_KILLRATE_STR},
	{ 'p', "pool-name", "STRING", "Pool name",
	NO_DEFAULT, DEFAULT_POOL},
	{ 'f', "vdev-file-directory", "PATH", "File directory for vdev files",
	NO_DEFAULT, DEFAULT_VDEV_DIR},
	{ 'M', "multi-host", NULL,
	"Multi-host; simulate pool imported on remote host",
	NO_DEFAULT, NULL},
	{ 'E', "use-existing-pool", NULL,
	"Use existing pool instead of creating new one", NO_DEFAULT, NULL},
	{ 'T', "run-time", "INTEGER", "Total run time",
	NO_DEFAULT, DEFAULT_RUN_TIME_STR},
	{ 'P', "pass-time", "INTEGER", "Time per pass",
	NO_DEFAULT, DEFAULT_PASS_TIME_STR},
	{ 'F', "freeze-loops", "INTEGER", "Max loops in spa_freeze()",
	DEFAULT_MAX_LOOPS, NULL},
	{ 'B', "alt-ztest", "PATH", "Alternate ztest path",
	NO_DEFAULT, NULL},
	{ 'C', "vdev-class-state", "on\|off\|random", "vdev class state",
	NO_DEFAULT, "random"},
	{ 'o', "option", "\"OPTION=INTEGER\"",
	"Set global variable to an unsigned 32-bit integer value",
	NO_DEFAULT, NULL},
	{ 'G', "dump-debug-msg", NULL,
	"Dump zfs_dbgmsg buffer before exiting due to an error",
	NO_DEFAULT, NULL},
	{ 'V', "verbose", NULL,
	"Verbose (use multiple times for ever more verbosity)",
	NO_DEFAULT, NULL},
	{ 'h', "help", NULL, "Show this help",
	NO_DEFAULT, NULL},
	{0, 0, 0, 0, 0, 0}
	};

	static struct option *long_opts = NULL;
	static char *short_opts = NULL;

	static void
	init_options(void)
	{
	ASSERT3P(long_opts, ==, NULL);
	ASSERT3P(short_opts, ==, NULL);

	int count = sizeof (option_table) / sizeof (option_table[0]);
	long_opts = umem_alloc(sizeof (struct option) * count, UMEM_NOFAIL);

	short_opts = umem_alloc(sizeof (char) * 2 * count, UMEM_NOFAIL);
	int short_opt_index = 0;

	for (int i = 0; i < count; i++) {
	long_opts[i].val = option_table[i].short_opt;
	long_opts[i].name = option_table[i].long_opt;
	long_opts[i].has_arg = option_table[i].long_opt_param != NULL
	? required_argument : no_argument;
	long_opts[i].flag = NULL;
	short_opts[short_opt_index++] = option_table[i].short_opt;
	if (option_table[i].long_opt_param != NULL) {
	short_opts[short_opt_index++] = ':';
	}
	}
	}

	static void
	fini_options(void)
	{
	int count = sizeof (option_table) / sizeof (option_table[0]);

	umem_free(long_opts, sizeof (struct option) * count);
	umem_free(short_opts, sizeof (char) * 2 * count);

	long_opts = NULL;
	short_opts = NULL;
	}

	static void
	usage(boolean_t requested)
	{
	char option[80];
	FILE *fp = requested ? stdout : stderr;

	(void) fprintf(fp, "Usage: %s [OPTIONS...]\n", DEFAULT_POOL);
	for (int i = 0; option_table[i].short_opt != 0; i++) {
	if (option_table[i].long_opt_param != NULL) {
	(void) sprintf(option, " -%c --%s=%s",
	option_table[i].short_opt,
	option_table[i].long_opt,
	option_table[i].long_opt_param);
	} else {
	(void) sprintf(option, " -%c --%s",
	option_table[i].short_opt,
	option_table[i].long_opt);
	}
	(void) fprintf(fp, " %-40s%s", option,
	option_table[i].comment);

	if (option_table[i].long_opt_param != NULL) {
	if (option_table[i].default_str != NULL) {
	(void) fprintf(fp, " (default: %s)",
	option_table[i].default_str);
	} else if (option_table[i].default_int != NO_DEFAULT) {
	(void) fprintf(fp, " (default: %u)",
	option_table[i].default_int);
	}
	}
	(void) fprintf(fp, "\n");
	}
	exit(requested ? 0 : 1);
	}

	static uint64_t
	ztest_random(uint64_t range)
	{
	uint64_t r;

	ASSERT3S(ztest_fd_rand, >=, 0);

	if (range == 0)
	return (0);

	if (read(ztest_fd_rand, &r, sizeof (r)) != sizeof (r))
	fatal(B_TRUE, "short read from /dev/urandom");

	return (r % range);
	}

	static void
	ztest_parse_name_value(const char input, ztest_shared_opts_t zo)
	{
	char name[32];
	char *value;
	int state = ZTEST_VDEV_CLASS_RND;

	(void) strlcpy(name, input, sizeof (name));

	value = strchr(name, '=');
	if (value == NULL) {
	(void) fprintf(stderr, "missing value in property=value "
	"'-C' argument (%s)\n", input);
	usage(B_FALSE);
	}
	*(value) = '\0';
	value++;

	if (strcmp(value, "on") == 0) {
	state = ZTEST_VDEV_CLASS_ON;
	} else if (strcmp(value, "off") == 0) {
	state = ZTEST_VDEV_CLASS_OFF;
	} else if (strcmp(value, "random") == 0) {
	state = ZTEST_VDEV_CLASS_RND;
	} else {
	(void) fprintf(stderr, "invalid property value '%s'\n", value);
	usage(B_FALSE);
	}

	if (strcmp(name, "special") == 0) {
	zo->zo_special_vdevs = state;
	} else {
	(void) fprintf(stderr, "invalid property name '%s'\n", name);
	usage(B_FALSE);
	}
	if (zo->zo_verbose >= 3)
	(void) printf("%s vdev state is '%s'\n", name, value);
	}

	static void
	process_options(int argc, char **argv)
	{
	char *path;
	ztest_shared_opts_t *zo = &ztest_opts;

	int opt;
	uint64_t value;
	char altdir[MAXNAMELEN] = { 0 };
	char raid_kind[8] = { "random" };

	bcopy(&ztest_opts_defaults, zo, sizeof (*zo));

	init_options();

	while ((opt = getopt_long(argc, argv, short_opts, long_opts,
	NULL)) != EOF) {
	value = 0;
	switch (opt) {
	case 'v':
	case 's':
	case 'a':
	case 'm':
	case 'r':
	case 'R':
	case 'D':
	case 'S':
	case 'd':
	case 't':
	case 'g':
	case 'i':
	case 'k':
	case 'T':
	case 'P':
	case 'F':
	value = nicenumtoull(optarg);
	}
	switch (opt) {
	case 'v':
	zo->zo_vdevs = value;
	break;
	case 's':
	zo->zo_vdev_size = MAX(SPA_MINDEVSIZE, value);
	break;
	case 'a':
	zo->zo_ashift = value;
	break;
	case 'm':
	zo->zo_mirrors = value;
	break;
	case 'r':
	zo->zo_raid_children = MAX(1, value);
	break;
	case 'R':
	zo->zo_raid_parity = MIN(MAX(value, 1), 3);
	break;
	case 'K':
	(void) strlcpy(raid_kind, optarg, sizeof (raid_kind));
	break;
	case 'D':
	zo->zo_draid_data = MAX(1, value);
	break;
	case 'S':
	zo->zo_draid_spares = MAX(1, value);
	break;
	case 'd':
	zo->zo_datasets = MAX(1, value);
	break;
	case 't':
	zo->zo_threads = MAX(1, value);
	break;
	case 'g':
	zo->zo_metaslab_force_ganging =
	MAX(SPA_MINBLOCKSIZE << 1, value);
	break;
	case 'i':
	zo->zo_init = value;
	break;
	case 'k':
	zo->zo_killrate = value;
	break;
	case 'p':
	(void) strlcpy(zo->zo_pool, optarg,
	sizeof (zo->zo_pool));
	break;
	case 'f':
	path = realpath(optarg, NULL);
	if (path == NULL) {
	(void) fprintf(stderr, "error: %s: %s\n",
	optarg, strerror(errno));
	usage(B_FALSE);
	} else {
	(void) strlcpy(zo->zo_dir, path,
	sizeof (zo->zo_dir));
	free(path);
	}
	break;
	case 'M':
	zo->zo_mmp_test = 1;
	break;
	case 'V':
	zo->zo_verbose++;
	break;
	case 'E':
	zo->zo_init = 0;
	break;
	case 'T':
	zo->zo_time = value;
	break;
	case 'P':
	zo->zo_passtime = MAX(1, value);
	break;
	case 'F':
	zo->zo_maxloops = MAX(1, value);
	break;
	case 'B':
	(void) strlcpy(altdir, optarg, sizeof (altdir));
	break;
	case 'C':
	ztest_parse_name_value(optarg, zo);
	break;
	case 'o':
	if (zo->zo_gvars_count >= ZO_GVARS_MAX_COUNT) {
	(void) fprintf(stderr,
	"max global var count (%zu) exceeded\n",
	ZO_GVARS_MAX_COUNT);
	usage(B_FALSE);
	}
	char *v = zo->zo_gvars[zo->zo_gvars_count];
	if (strlcpy(v, optarg, ZO_GVARS_MAX_ARGLEN) >=
	ZO_GVARS_MAX_ARGLEN) {
	(void) fprintf(stderr,
	"global var option '%s' is too long\n",
	optarg);
	usage(B_FALSE);
	}
	zo->zo_gvars_count++;
	break;
	case 'G':
	zo->zo_dump_dbgmsg = 1;
	break;
	case 'h':
	usage(B_TRUE);
	break;
	case '?':
	default:
	usage(B_FALSE);
	break;
	}
	}

	fini_options();

	/* When raid choice is 'random' add a draid pool 50% of the time */
	if (strcmp(raid_kind, "random") == 0) {
	(void) strlcpy(raid_kind, (ztest_random(2) == 0) ?
	"draid" : "raidz", sizeof (raid_kind));

	if (ztest_opts.zo_verbose >= 3)
	(void) printf("choosing RAID type '%s'\n", raid_kind);
	}

	if (strcmp(raid_kind, "draid") == 0) {
	uint64_t min_devsize;

	/* With fewer disk use 256M, otherwise 128M is OK */
	min_devsize = (ztest_opts.zo_raid_children < 16) ?
	(256ULL << 20) : (128ULL << 20);

	/* No top-level mirrors with dRAID for now */
	zo->zo_mirrors = 0;

	/* Use more appropriate defaults for dRAID */
	if (zo->zo_vdevs == ztest_opts_defaults.zo_vdevs)
	zo->zo_vdevs = 1;
	if (zo->zo_raid_children ==
	ztest_opts_defaults.zo_raid_children)
	zo->zo_raid_children = 16;
	if (zo->zo_ashift < 12)
	zo->zo_ashift = 12;
	if (zo->zo_vdev_size < min_devsize)
	zo->zo_vdev_size = min_devsize;

	if (zo->zo_draid_data + zo->zo_raid_parity >
	zo->zo_raid_children - zo->zo_draid_spares) {
	(void) fprintf(stderr, "error: too few draid "
	"children (%d) for stripe width (%d)\n",
	zo->zo_raid_children,
	zo->zo_draid_data + zo->zo_raid_parity);
	usage(B_FALSE);
	}

	(void) strlcpy(zo->zo_raid_type, VDEV_TYPE_DRAID,
	sizeof (zo->zo_raid_type));

	} else /* using raidz */ {
	ASSERT0(strcmp(raid_kind, "raidz"));

	zo->zo_raid_parity = MIN(zo->zo_raid_parity,
	zo->zo_raid_children - 1);
	}

	zo->zo_vdevtime =
	(zo->zo_vdevs > 0 ? zo->zo_time * NANOSEC / zo->zo_vdevs :
	UINT64_MAX >> 2);

	if (strlen(altdir) > 0) {
	char *cmd;
	char *realaltdir;
	char *bin;
	char *ztest;
	char *isa;
	int isalen;

	cmd = umem_alloc(MAXPATHLEN, UMEM_NOFAIL);
	realaltdir = umem_alloc(MAXPATHLEN, UMEM_NOFAIL);

	VERIFY3P(NULL, !=, realpath(getexecname(), cmd));
	if (0 != access(altdir, F_OK)) {
	ztest_dump_core = B_FALSE;
	fatal(B_TRUE, "invalid alternate ztest path: %s",
	altdir);
	}
	VERIFY3P(NULL, !=, realpath(altdir, realaltdir));

	/*
	* 'cmd' should be of the form "<anything>/usr/bin/<isa>/ztest".
	* We want to extract <isa> to determine if we should use
	* 32 or 64 bit binaries.
	*/
	bin = strstr(cmd, "/usr/bin/");
	ztest = strstr(bin, "/ztest");
	isa = bin + 9;
	isalen = ztest - isa;
	(void) snprintf(zo->zo_alt_ztest, sizeof (zo->zo_alt_ztest),
	"%s/usr/bin/%.*s/ztest", realaltdir, isalen, isa);
	(void) snprintf(zo->zo_alt_libpath, sizeof (zo->zo_alt_libpath),
	"%s/usr/lib/%.*s", realaltdir, isalen, isa);

	if (0 != access(zo->zo_alt_ztest, X_OK)) {
	ztest_dump_core = B_FALSE;
	fatal(B_TRUE, "invalid alternate ztest: %s",
	zo->zo_alt_ztest);
	} else if (0 != access(zo->zo_alt_libpath, X_OK)) {
	ztest_dump_core = B_FALSE;
	fatal(B_TRUE, "invalid alternate lib directory %s",
	zo->zo_alt_libpath);
	}

	umem_free(cmd, MAXPATHLEN);
	umem_free(realaltdir, MAXPATHLEN);
	}
	}

	static void
	ztest_kill(ztest_shared_t *zs)
	{
	zs->zs_alloc = metaslab_class_get_alloc(spa_normal_class(ztest_spa));
	zs->zs_space = metaslab_class_get_space(spa_normal_class(ztest_spa));

	/*
	* Before we kill off ztest, make sure that the config is updated.
	* See comment above spa_write_cachefile().
	*/
	mutex_enter(&spa_namespace_lock);
	spa_write_cachefile(ztest_spa, B_FALSE, B_FALSE);
	mutex_exit(&spa_namespace_lock);

	(void) kill(getpid(), SIGKILL);
	}

	/* ARGSUSED */
	static void
	ztest_record_enospc(const char *s)
	{
	ztest_shared->zs_enospc_count++;
	}

	static uint64_t
	ztest_get_ashift(void)
	{
	if (ztest_opts.zo_ashift == 0)
	return (SPA_MINBLOCKSHIFT + ztest_random(5));
	return (ztest_opts.zo_ashift);
	}

	static boolean_t
	ztest_is_draid_spare(const char *name)
	{
	uint64_t spare_id = 0, parity = 0, vdev_id = 0;

	if (sscanf(name, VDEV_TYPE_DRAID "%"PRIu64"-%"PRIu64"-%"PRIu64"",
	&parity, &vdev_id, &spare_id) == 3) {
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	static nvlist_t *
	make_vdev_file(char path, char aux, char *pool, size_t size, uint64_t ashift)
	{
	char *pathbuf;
	uint64_t vdev;
	nvlist_t *file;
	boolean_t draid_spare = B_FALSE;

	pathbuf = umem_alloc(MAXPATHLEN, UMEM_NOFAIL);

	if (ashift == 0)
	ashift = ztest_get_ashift();

	if (path == NULL) {
	path = pathbuf;

	if (aux != NULL) {
	vdev = ztest_shared->zs_vdev_aux;
	(void) snprintf(path, MAXPATHLEN,
	ztest_aux_template, ztest_opts.zo_dir,
	pool == NULL ? ztest_opts.zo_pool : pool,
	aux, vdev);
	} else {
	vdev = ztest_shared->zs_vdev_next_leaf++;
	(void) snprintf(path, MAXPATHLEN,
	ztest_dev_template, ztest_opts.zo_dir,
	pool == NULL ? ztest_opts.zo_pool : pool, vdev);
	}
	} else {
	draid_spare = ztest_is_draid_spare(path);
	}

	if (size != 0 && !draid_spare) {
	int fd = open(path, O_RDWR \| O_CREAT \| O_TRUNC, 0666);
	if (fd == -1)
	fatal(B_TRUE, "can't open %s", path);
	if (ftruncate(fd, size) != 0)
	fatal(B_TRUE, "can't ftruncate %s", path);
	(void) close(fd);
	}

	file = fnvlist_alloc();
	fnvlist_add_string(file, ZPOOL_CONFIG_TYPE,
	draid_spare ? VDEV_TYPE_DRAID_SPARE : VDEV_TYPE_FILE);
	fnvlist_add_string(file, ZPOOL_CONFIG_PATH, path);
	fnvlist_add_uint64(file, ZPOOL_CONFIG_ASHIFT, ashift);
	umem_free(pathbuf, MAXPATHLEN);

	return (file);
	}

	static nvlist_t *
	make_vdev_raid(char path, char aux, char *pool, size_t size,
	uint64_t ashift, int r)
	{
	nvlist_t raid, *child;
	int c;

	if (r < 2)
	return (make_vdev_file(path, aux, pool, size, ashift));
	child = umem_alloc(r * sizeof (nvlist_t *), UMEM_NOFAIL);

	for (c = 0; c < r; c++)
	child[c] = make_vdev_file(path, aux, pool, size, ashift);

	raid = fnvlist_alloc();
	fnvlist_add_string(raid, ZPOOL_CONFIG_TYPE,
	ztest_opts.zo_raid_type);
	fnvlist_add_uint64(raid, ZPOOL_CONFIG_NPARITY,
	ztest_opts.zo_raid_parity);
	- fnvlist_add_nvlist_array(raid, ZPOOL_CONFIG_CHILDREN, child, r);
	+ fnvlist_add_nvlist_array(raid, ZPOOL_CONFIG_CHILDREN,
	+ (const nvlist_t **)child, r);

	if (strcmp(ztest_opts.zo_raid_type, VDEV_TYPE_DRAID) == 0) {
	uint64_t ndata = ztest_opts.zo_draid_data;
	uint64_t nparity = ztest_opts.zo_raid_parity;
	uint64_t nspares = ztest_opts.zo_draid_spares;
	uint64_t children = ztest_opts.zo_raid_children;
	uint64_t ngroups = 1;

	/*
	* Calculate the minimum number of groups required to fill a
	* slice. This is the LCM of the stripe width (data + parity)
	* and the number of data drives (children - spares).
	*/
	while (ngroups * (ndata + nparity) % (children - nspares) != 0)
	ngroups++;

	/* Store the basic dRAID configuration. */
	fnvlist_add_uint64(raid, ZPOOL_CONFIG_DRAID_NDATA, ndata);
	fnvlist_add_uint64(raid, ZPOOL_CONFIG_DRAID_NSPARES, nspares);
	fnvlist_add_uint64(raid, ZPOOL_CONFIG_DRAID_NGROUPS, ngroups);
	}

	for (c = 0; c < r; c++)
	fnvlist_free(child[c]);

	umem_free(child, r * sizeof (nvlist_t *));

	return (raid);
	}

	static nvlist_t *
	make_vdev_mirror(char path, char aux, char *pool, size_t size,
	uint64_t ashift, int r, int m)
	{
	nvlist_t mirror, *child;
	int c;

	if (m < 1)
	return (make_vdev_raid(path, aux, pool, size, ashift, r));

	child = umem_alloc(m * sizeof (nvlist_t *), UMEM_NOFAIL);

	for (c = 0; c < m; c++)
	child[c] = make_vdev_raid(path, aux, pool, size, ashift, r);

	mirror = fnvlist_alloc();
	fnvlist_add_string(mirror, ZPOOL_CONFIG_TYPE, VDEV_TYPE_MIRROR);
	- fnvlist_add_nvlist_array(mirror, ZPOOL_CONFIG_CHILDREN, child, m);
	+ fnvlist_add_nvlist_array(mirror, ZPOOL_CONFIG_CHILDREN,
	+ (const nvlist_t **)child, m);

	for (c = 0; c < m; c++)
	fnvlist_free(child[c]);

	umem_free(child, m * sizeof (nvlist_t *));

	return (mirror);
	}

	static nvlist_t *
	make_vdev_root(char path, char aux, char *pool, size_t size, uint64_t ashift,
	const char *class, int r, int m, int t)
	{
	nvlist_t root, *child;
	int c;
	boolean_t log;

	ASSERT3S(t, >, 0);

	log = (class != NULL && strcmp(class, "log") == 0);

	child = umem_alloc(t * sizeof (nvlist_t *), UMEM_NOFAIL);

	for (c = 0; c < t; c++) {
	child[c] = make_vdev_mirror(path, aux, pool, size, ashift,
	r, m);
	fnvlist_add_uint64(child[c], ZPOOL_CONFIG_IS_LOG, log);

	if (class != NULL && class[0] != '\0') {
	ASSERT(m > 1 \|\| log); /* expecting a mirror */
	fnvlist_add_string(child[c],
	ZPOOL_CONFIG_ALLOCATION_BIAS, class);
	}
	}

	root = fnvlist_alloc();
	fnvlist_add_string(root, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT);
	fnvlist_add_nvlist_array(root, aux ? aux : ZPOOL_CONFIG_CHILDREN,
	- child, t);
	+ (const nvlist_t **)child, t);

	for (c = 0; c < t; c++)
	fnvlist_free(child[c]);

	umem_free(child, t * sizeof (nvlist_t *));

	return (root);
	}

	/*
	* Find a random spa version. Returns back a random spa version in the
	* range [initial_version, SPA_VERSION_FEATURES].
	*/
	static uint64_t
	ztest_random_spa_version(uint64_t initial_version)
	{
	uint64_t version = initial_version;

	if (version <= SPA_VERSION_BEFORE_FEATURES) {
	version = version +
	ztest_random(SPA_VERSION_BEFORE_FEATURES - version + 1);
	}

	if (version > SPA_VERSION_BEFORE_FEATURES)
	version = SPA_VERSION_FEATURES;

	ASSERT(SPA_VERSION_IS_SUPPORTED(version));
	return (version);
	}

	static int
	ztest_random_blocksize(void)
	{
	ASSERT3U(ztest_spa->spa_max_ashift, !=, 0);

	/*
	* Choose a block size >= the ashift.
	* If the SPA supports new MAXBLOCKSIZE, test up to 1MB blocks.
	*/
	int maxbs = SPA_OLD_MAXBLOCKSHIFT;
	if (spa_maxblocksize(ztest_spa) == SPA_MAXBLOCKSIZE)
	maxbs = 20;
	uint64_t block_shift =
	ztest_random(maxbs - ztest_spa->spa_max_ashift + 1);
	return (1 << (SPA_MINBLOCKSHIFT + block_shift));
	}

	static int
	ztest_random_dnodesize(void)
	{
	int slots;
	int max_slots = spa_maxdnodesize(ztest_spa) >> DNODE_SHIFT;

	if (max_slots == DNODE_MIN_SLOTS)
	return (DNODE_MIN_SIZE);

	/*
	* Weight the random distribution more heavily toward smaller
	* dnode sizes since that is more likely to reflect real-world
	* usage.
	*/
	ASSERT3U(max_slots, >, 4);
	switch (ztest_random(10)) {
	case 0:
	slots = 5 + ztest_random(max_slots - 4);
	break;
	case 1 ... 4:
	slots = 2 + ztest_random(3);
	break;
	default:
	slots = 1;
	break;
	}

	return (slots << DNODE_SHIFT);
	}

	static int
	ztest_random_ibshift(void)
	{
	return (DN_MIN_INDBLKSHIFT +
	ztest_random(DN_MAX_INDBLKSHIFT - DN_MIN_INDBLKSHIFT + 1));
	}

	static uint64_t
	ztest_random_vdev_top(spa_t *spa, boolean_t log_ok)
	{
	uint64_t top;
	vdev_t *rvd = spa->spa_root_vdev;
	vdev_t *tvd;

	ASSERT3U(spa_config_held(spa, SCL_ALL, RW_READER), !=, 0);

	do {
	top = ztest_random(rvd->vdev_children);
	tvd = rvd->vdev_child[top];
	} while (!vdev_is_concrete(tvd) \|\| (tvd->vdev_islog && !log_ok) \|\|
	tvd->vdev_mg == NULL \|\| tvd->vdev_mg->mg_class == NULL);

	return (top);
	}

	static uint64_t
	ztest_random_dsl_prop(zfs_prop_t prop)
	{
	uint64_t value;

	do {
	value = zfs_prop_random_value(prop, ztest_random(-1ULL));
	} while (prop == ZFS_PROP_CHECKSUM && value == ZIO_CHECKSUM_OFF);

	return (value);
	}

	static int
	ztest_dsl_prop_set_uint64(char *osname, zfs_prop_t prop, uint64_t value,
	boolean_t inherit)
	{
	const char *propname = zfs_prop_to_name(prop);
	const char *valname;
	char *setpoint;
	uint64_t curval;
	int error;

	error = dsl_prop_set_int(osname, propname,
	(inherit ? ZPROP_SRC_NONE : ZPROP_SRC_LOCAL), value);

	if (error == ENOSPC) {
	ztest_record_enospc(FTAG);
	return (error);
	}
	ASSERT0(error);

	setpoint = umem_alloc(MAXPATHLEN, UMEM_NOFAIL);
	VERIFY0(dsl_prop_get_integer(osname, propname, &curval, setpoint));

	if (ztest_opts.zo_verbose >= 6) {
	int err;

	err = zfs_prop_index_to_string(prop, curval, &valname);
	if (err)
	(void) printf("%s %s = %llu at '%s'\n", osname,
	propname, (unsigned long long)curval, setpoint);
	else
	(void) printf("%s %s = %s at '%s'\n",
	osname, propname, valname, setpoint);
	}
	umem_free(setpoint, MAXPATHLEN);

	return (error);
	}

	static int
	ztest_spa_prop_set_uint64(zpool_prop_t prop, uint64_t value)
	{
	spa_t *spa = ztest_spa;
	nvlist_t *props = NULL;
	int error;

	props = fnvlist_alloc();
	fnvlist_add_uint64(props, zpool_prop_to_name(prop), value);

	error = spa_prop_set(spa, props);

	fnvlist_free(props);

	if (error == ENOSPC) {
	ztest_record_enospc(FTAG);
	return (error);
	}
	ASSERT0(error);

	return (error);
	}

	static int
	ztest_dmu_objset_own(const char *name, dmu_objset_type_t type,
	boolean_t readonly, boolean_t decrypt, void tag, objset_t *osp)
	{
	int err;
	char *cp = NULL;
	char ddname[ZFS_MAX_DATASET_NAME_LEN];

	strcpy(ddname, name);
	cp = strchr(ddname, '@');
	if (cp != NULL)
	*cp = '\0';

	err = dmu_objset_own(name, type, readonly, decrypt, tag, osp);
	while (decrypt && err == EACCES) {
	dsl_crypto_params_t *dcp;
	nvlist_t *crypto_args = fnvlist_alloc();

	fnvlist_add_uint8_array(crypto_args, "wkeydata",
	(uint8_t *)ztest_wkeydata, WRAPPING_KEY_LEN);
	VERIFY0(dsl_crypto_params_create_nvlist(DCP_CMD_NONE, NULL,
	crypto_args, &dcp));
	err = spa_keystore_load_wkey(ddname, dcp, B_FALSE);
	/*
	* Note: if there was an error loading, the wkey was not
	* consumed, and needs to be freed.
	*/
	dsl_crypto_params_free(dcp, (err != 0));
	fnvlist_free(crypto_args);

	if (err == EINVAL) {
	/*
	* We couldn't load a key for this dataset so try
	* the parent. This loop will eventually hit the
	* encryption root since ztest only makes clones
	* as children of their origin datasets.
	*/
	cp = strrchr(ddname, '/');
	if (cp == NULL)
	return (err);

	*cp = '\0';
	err = EACCES;
	continue;
	} else if (err != 0) {
	break;
	}

	err = dmu_objset_own(name, type, readonly, decrypt, tag, osp);
	break;
	}

	return (err);
	}

	static void
	ztest_rll_init(rll_t *rll)
	{
	rll->rll_writer = NULL;
	rll->rll_readers = 0;
	mutex_init(&rll->rll_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&rll->rll_cv, NULL, CV_DEFAULT, NULL);
	}

	static void
	ztest_rll_destroy(rll_t *rll)
	{
	ASSERT3P(rll->rll_writer, ==, NULL);
	ASSERT0(rll->rll_readers);
	mutex_destroy(&rll->rll_lock);
	cv_destroy(&rll->rll_cv);
	}

	static void
	ztest_rll_lock(rll_t *rll, rl_type_t type)
	{
	mutex_enter(&rll->rll_lock);

	if (type == RL_READER) {
	while (rll->rll_writer != NULL)
	(void) cv_wait(&rll->rll_cv, &rll->rll_lock);
	rll->rll_readers++;
	} else {
	while (rll->rll_writer != NULL \|\| rll->rll_readers)
	(void) cv_wait(&rll->rll_cv, &rll->rll_lock);
	rll->rll_writer = curthread;
	}

	mutex_exit(&rll->rll_lock);
	}

	static void
	ztest_rll_unlock(rll_t *rll)
	{
	mutex_enter(&rll->rll_lock);

	if (rll->rll_writer) {
	ASSERT0(rll->rll_readers);
	rll->rll_writer = NULL;
	} else {
	ASSERT3S(rll->rll_readers, >, 0);
	ASSERT3P(rll->rll_writer, ==, NULL);
	rll->rll_readers--;
	}

	if (rll->rll_writer == NULL && rll->rll_readers == 0)
	cv_broadcast(&rll->rll_cv);

	mutex_exit(&rll->rll_lock);
	}

	static void
	ztest_object_lock(ztest_ds_t *zd, uint64_t object, rl_type_t type)
	{
	rll_t *rll = &zd->zd_object_lock[object & (ZTEST_OBJECT_LOCKS - 1)];

	ztest_rll_lock(rll, type);
	}

	static void
	ztest_object_unlock(ztest_ds_t *zd, uint64_t object)
	{
	rll_t *rll = &zd->zd_object_lock[object & (ZTEST_OBJECT_LOCKS - 1)];

	ztest_rll_unlock(rll);
	}

	static rl_t *
	ztest_range_lock(ztest_ds_t *zd, uint64_t object, uint64_t offset,
	uint64_t size, rl_type_t type)
	{
	uint64_t hash = object ^ (offset % (ZTEST_RANGE_LOCKS + 1));
	rll_t *rll = &zd->zd_range_lock[hash & (ZTEST_RANGE_LOCKS - 1)];
	rl_t *rl;

	rl = umem_alloc(sizeof (*rl), UMEM_NOFAIL);
	rl->rl_object = object;
	rl->rl_offset = offset;
	rl->rl_size = size;
	rl->rl_lock = rll;

	ztest_rll_lock(rll, type);

	return (rl);
	}

	static void
	ztest_range_unlock(rl_t *rl)
	{
	rll_t *rll = rl->rl_lock;

	ztest_rll_unlock(rll);

	umem_free(rl, sizeof (*rl));
	}

	static void
	ztest_zd_init(ztest_ds_t zd, ztest_shared_ds_t szd, objset_t *os)
	{
	zd->zd_os = os;
	zd->zd_zilog = dmu_objset_zil(os);
	zd->zd_shared = szd;
	dmu_objset_name(os, zd->zd_name);
	int l;

	if (zd->zd_shared != NULL)
	zd->zd_shared->zd_seq = 0;

	VERIFY0(pthread_rwlock_init(&zd->zd_zilog_lock, NULL));
	mutex_init(&zd->zd_dirobj_lock, NULL, MUTEX_DEFAULT, NULL);

	for (l = 0; l < ZTEST_OBJECT_LOCKS; l++)
	ztest_rll_init(&zd->zd_object_lock[l]);

	for (l = 0; l < ZTEST_RANGE_LOCKS; l++)
	ztest_rll_init(&zd->zd_range_lock[l]);
	}

	static void
	ztest_zd_fini(ztest_ds_t *zd)
	{
	int l;

	mutex_destroy(&zd->zd_dirobj_lock);
	(void) pthread_rwlock_destroy(&zd->zd_zilog_lock);

	for (l = 0; l < ZTEST_OBJECT_LOCKS; l++)
	ztest_rll_destroy(&zd->zd_object_lock[l]);

	for (l = 0; l < ZTEST_RANGE_LOCKS; l++)
	ztest_rll_destroy(&zd->zd_range_lock[l]);
	}

	#define TXG_MIGHTWAIT (ztest_random(10) == 0 ? TXG_NOWAIT : TXG_WAIT)

	static uint64_t
	ztest_tx_assign(dmu_tx_t tx, uint64_t txg_how, const char tag)
	{
	uint64_t txg;
	int error;

	/*
	* Attempt to assign tx to some transaction group.
	*/
	error = dmu_tx_assign(tx, txg_how);
	if (error) {
	if (error == ERESTART) {
	ASSERT3U(txg_how, ==, TXG_NOWAIT);
	dmu_tx_wait(tx);
	} else {
	ASSERT3U(error, ==, ENOSPC);
	ztest_record_enospc(tag);
	}
	dmu_tx_abort(tx);
	return (0);
	}
	txg = dmu_tx_get_txg(tx);
	ASSERT3U(txg, !=, 0);
	return (txg);
	}

	static void
	ztest_bt_generate(ztest_block_tag_t bt, objset_t os, uint64_t object,
	uint64_t dnodesize, uint64_t offset, uint64_t gen, uint64_t txg,
	uint64_t crtxg)
	{
	bt->bt_magic = BT_MAGIC;
	bt->bt_objset = dmu_objset_id(os);
	bt->bt_object = object;
	bt->bt_dnodesize = dnodesize;
	bt->bt_offset = offset;
	bt->bt_gen = gen;
	bt->bt_txg = txg;
	bt->bt_crtxg = crtxg;
	}

	static void
	ztest_bt_verify(ztest_block_tag_t bt, objset_t os, uint64_t object,
	uint64_t dnodesize, uint64_t offset, uint64_t gen, uint64_t txg,
	uint64_t crtxg)
	{
	ASSERT3U(bt->bt_magic, ==, BT_MAGIC);
	ASSERT3U(bt->bt_objset, ==, dmu_objset_id(os));
	ASSERT3U(bt->bt_object, ==, object);
	ASSERT3U(bt->bt_dnodesize, ==, dnodesize);
	ASSERT3U(bt->bt_offset, ==, offset);
	ASSERT3U(bt->bt_gen, <=, gen);
	ASSERT3U(bt->bt_txg, <=, txg);
	ASSERT3U(bt->bt_crtxg, ==, crtxg);
	}

	static ztest_block_tag_t *
	ztest_bt_bonus(dmu_buf_t *db)
	{
	dmu_object_info_t doi;
	ztest_block_tag_t *bt;

	dmu_object_info_from_db(db, &doi);
	ASSERT3U(doi.doi_bonus_size, <=, db->db_size);
	ASSERT3U(doi.doi_bonus_size, >=, sizeof (*bt));
	bt = (void )((char )db->db_data + doi.doi_bonus_size - sizeof (*bt));

	return (bt);
	}

	/*
	* Generate a token to fill up unused bonus buffer space. Try to make
	* it unique to the object, generation, and offset to verify that data
	* is not getting overwritten by data from other dnodes.
	*/
	#define ZTEST_BONUS_FILL_TOKEN(obj, ds, gen, offset) \
	(((ds) << 48) \| ((gen) << 32) \| ((obj) << 8) \| (offset))

	/*
	* Fill up the unused bonus buffer region before the block tag with a
	* verifiable pattern. Filling the whole bonus area with non-zero data
	* helps ensure that all dnode traversal code properly skips the
	* interior regions of large dnodes.
	*/
	static void
	ztest_fill_unused_bonus(dmu_buf_t db, void end, uint64_t obj,
	objset_t *os, uint64_t gen)
	{
	uint64_t *bonusp;

	ASSERT(IS_P2ALIGNED((char )end - (char )db->db_data, 8));

	for (bonusp = db->db_data; bonusp < (uint64_t *)end; bonusp++) {
	uint64_t token = ZTEST_BONUS_FILL_TOKEN(obj, dmu_objset_id(os),
	gen, bonusp - (uint64_t *)db->db_data);
	*bonusp = token;
	}
	}

	/*
	* Verify that the unused area of a bonus buffer is filled with the
	* expected tokens.
	*/
	static void
	ztest_verify_unused_bonus(dmu_buf_t db, void end, uint64_t obj,
	objset_t *os, uint64_t gen)
	{
	uint64_t *bonusp;

	for (bonusp = db->db_data; bonusp < (uint64_t *)end; bonusp++) {
	uint64_t token = ZTEST_BONUS_FILL_TOKEN(obj, dmu_objset_id(os),
	gen, bonusp - (uint64_t *)db->db_data);
	VERIFY3U(*bonusp, ==, token);
	}
	}

	/*
	* ZIL logging ops
	*/

	#define lrz_type lr_mode
	#define lrz_blocksize lr_uid
	#define lrz_ibshift lr_gid
	#define lrz_bonustype lr_rdev
	#define lrz_dnodesize lr_crtime[1]

	static void
	ztest_log_create(ztest_ds_t zd, dmu_tx_t tx, lr_create_t *lr)
	{
	char name = (void )(lr + 1); /* name follows lr */
	size_t namesize = strlen(name) + 1;
	itx_t *itx;

	if (zil_replaying(zd->zd_zilog, tx))
	return;

	itx = zil_itx_create(TX_CREATE, sizeof (*lr) + namesize);
	bcopy(&lr->lr_common + 1, &itx->itx_lr + 1,
	sizeof (*lr) + namesize - sizeof (lr_t));

	zil_itx_assign(zd->zd_zilog, itx, tx);
	}

	static void
	ztest_log_remove(ztest_ds_t zd, dmu_tx_t tx, lr_remove_t *lr, uint64_t object)
	{
	char name = (void )(lr + 1); /* name follows lr */
	size_t namesize = strlen(name) + 1;
	itx_t *itx;

	if (zil_replaying(zd->zd_zilog, tx))
	return;

	itx = zil_itx_create(TX_REMOVE, sizeof (*lr) + namesize);
	bcopy(&lr->lr_common + 1, &itx->itx_lr + 1,
	sizeof (*lr) + namesize - sizeof (lr_t));

	itx->itx_oid = object;
	zil_itx_assign(zd->zd_zilog, itx, tx);
	}

	static void
	ztest_log_write(ztest_ds_t zd, dmu_tx_t tx, lr_write_t *lr)
	{
	itx_t *itx;
	itx_wr_state_t write_state = ztest_random(WR_NUM_STATES);

	if (zil_replaying(zd->zd_zilog, tx))
	return;

	if (lr->lr_length > zil_max_log_data(zd->zd_zilog))
	write_state = WR_INDIRECT;

	itx = zil_itx_create(TX_WRITE,
	sizeof (*lr) + (write_state == WR_COPIED ? lr->lr_length : 0));

	if (write_state == WR_COPIED &&
	dmu_read(zd->zd_os, lr->lr_foid, lr->lr_offset, lr->lr_length,
	((lr_write_t *)&itx->itx_lr) + 1, DMU_READ_NO_PREFETCH) != 0) {
	zil_itx_destroy(itx);
	itx = zil_itx_create(TX_WRITE, sizeof (*lr));
	write_state = WR_NEED_COPY;
	}
	itx->itx_private = zd;
	itx->itx_wr_state = write_state;
	itx->itx_sync = (ztest_random(8) == 0);

	bcopy(&lr->lr_common + 1, &itx->itx_lr + 1,
	sizeof (*lr) - sizeof (lr_t));

	zil_itx_assign(zd->zd_zilog, itx, tx);
	}

	static void
	ztest_log_truncate(ztest_ds_t zd, dmu_tx_t tx, lr_truncate_t *lr)
	{
	itx_t *itx;

	if (zil_replaying(zd->zd_zilog, tx))
	return;

	itx = zil_itx_create(TX_TRUNCATE, sizeof (*lr));
	bcopy(&lr->lr_common + 1, &itx->itx_lr + 1,
	sizeof (*lr) - sizeof (lr_t));

	itx->itx_sync = B_FALSE;
	zil_itx_assign(zd->zd_zilog, itx, tx);
	}

	static void
	ztest_log_setattr(ztest_ds_t zd, dmu_tx_t tx, lr_setattr_t *lr)
	{
	itx_t *itx;

	if (zil_replaying(zd->zd_zilog, tx))
	return;

	itx = zil_itx_create(TX_SETATTR, sizeof (*lr));
	bcopy(&lr->lr_common + 1, &itx->itx_lr + 1,
	sizeof (*lr) - sizeof (lr_t));

	itx->itx_sync = B_FALSE;
	zil_itx_assign(zd->zd_zilog, itx, tx);
	}

	/*
	* ZIL replay ops
	*/
	static int
	ztest_replay_create(void arg1, void arg2, boolean_t byteswap)
	{
	ztest_ds_t *zd = arg1;
	lr_create_t *lr = arg2;
	char name = (void )(lr + 1); /* name follows lr */
	objset_t *os = zd->zd_os;
	ztest_block_tag_t *bbt;
	dmu_buf_t *db;
	dmu_tx_t *tx;
	uint64_t txg;
	int error = 0;
	int bonuslen;

	if (byteswap)
	byteswap_uint64_array(lr, sizeof (*lr));

	ASSERT3U(lr->lr_doid, ==, ZTEST_DIROBJ);
	ASSERT3S(name[0], !=, '\0');

	tx = dmu_tx_create(os);

	dmu_tx_hold_zap(tx, lr->lr_doid, B_TRUE, name);

	if (lr->lrz_type == DMU_OT_ZAP_OTHER) {
	dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, B_TRUE, NULL);
	} else {
	dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT);
	}

	txg = ztest_tx_assign(tx, TXG_WAIT, FTAG);
	if (txg == 0)
	return (ENOSPC);

	ASSERT3U(dmu_objset_zil(os)->zl_replay, ==, !!lr->lr_foid);
	bonuslen = DN_BONUS_SIZE(lr->lrz_dnodesize);

	if (lr->lrz_type == DMU_OT_ZAP_OTHER) {
	if (lr->lr_foid == 0) {
	lr->lr_foid = zap_create_dnsize(os,
	lr->lrz_type, lr->lrz_bonustype,
	bonuslen, lr->lrz_dnodesize, tx);
	} else {
	error = zap_create_claim_dnsize(os, lr->lr_foid,
	lr->lrz_type, lr->lrz_bonustype,
	bonuslen, lr->lrz_dnodesize, tx);
	}
	} else {
	if (lr->lr_foid == 0) {
	lr->lr_foid = dmu_object_alloc_dnsize(os,
	lr->lrz_type, 0, lr->lrz_bonustype,
	bonuslen, lr->lrz_dnodesize, tx);
	} else {
	error = dmu_object_claim_dnsize(os, lr->lr_foid,
	lr->lrz_type, 0, lr->lrz_bonustype,
	bonuslen, lr->lrz_dnodesize, tx);
	}
	}

	if (error) {
	ASSERT3U(error, ==, EEXIST);
	ASSERT(zd->zd_zilog->zl_replay);
	dmu_tx_commit(tx);
	return (error);
	}

	ASSERT3U(lr->lr_foid, !=, 0);

	if (lr->lrz_type != DMU_OT_ZAP_OTHER)
	VERIFY0(dmu_object_set_blocksize(os, lr->lr_foid,
	lr->lrz_blocksize, lr->lrz_ibshift, tx));

	VERIFY0(dmu_bonus_hold(os, lr->lr_foid, FTAG, &db));
	bbt = ztest_bt_bonus(db);
	dmu_buf_will_dirty(db, tx);
	ztest_bt_generate(bbt, os, lr->lr_foid, lr->lrz_dnodesize, -1ULL,
	lr->lr_gen, txg, txg);
	ztest_fill_unused_bonus(db, bbt, lr->lr_foid, os, lr->lr_gen);
	dmu_buf_rele(db, FTAG);

	VERIFY0(zap_add(os, lr->lr_doid, name, sizeof (uint64_t), 1,
	&lr->lr_foid, tx));

	(void) ztest_log_create(zd, tx, lr);

	dmu_tx_commit(tx);

	return (0);
	}

	static int
	ztest_replay_remove(void arg1, void arg2, boolean_t byteswap)
	{
	ztest_ds_t *zd = arg1;
	lr_remove_t *lr = arg2;
	char name = (void )(lr + 1); /* name follows lr */
	objset_t *os = zd->zd_os;
	dmu_object_info_t doi;
	dmu_tx_t *tx;
	uint64_t object, txg;

	if (byteswap)
	byteswap_uint64_array(lr, sizeof (*lr));

	ASSERT3U(lr->lr_doid, ==, ZTEST_DIROBJ);
	ASSERT3S(name[0], !=, '\0');

	VERIFY0(
	zap_lookup(os, lr->lr_doid, name, sizeof (object), 1, &object));
	ASSERT3U(object, !=, 0);

	ztest_object_lock(zd, object, RL_WRITER);

	VERIFY0(dmu_object_info(os, object, &doi));

	tx = dmu_tx_create(os);

	dmu_tx_hold_zap(tx, lr->lr_doid, B_FALSE, name);
	dmu_tx_hold_free(tx, object, 0, DMU_OBJECT_END);

	txg = ztest_tx_assign(tx, TXG_WAIT, FTAG);
	if (txg == 0) {
	ztest_object_unlock(zd, object);
	return (ENOSPC);
	}

	if (doi.doi_type == DMU_OT_ZAP_OTHER) {
	VERIFY0(zap_destroy(os, object, tx));
	} else {
	VERIFY0(dmu_object_free(os, object, tx));
	}

	VERIFY0(zap_remove(os, lr->lr_doid, name, tx));

	(void) ztest_log_remove(zd, tx, lr, object);

	dmu_tx_commit(tx);

	ztest_object_unlock(zd, object);

	return (0);
	}

	static int
	ztest_replay_write(void arg1, void arg2, boolean_t byteswap)
	{
	ztest_ds_t *zd = arg1;
	lr_write_t *lr = arg2;
	objset_t *os = zd->zd_os;
	void data = lr + 1; / data follows lr */
	uint64_t offset, length;
	ztest_block_tag_t *bt = data;
	ztest_block_tag_t *bbt;
	uint64_t gen, txg, lrtxg, crtxg;
	dmu_object_info_t doi;
	dmu_tx_t *tx;
	dmu_buf_t *db;
	arc_buf_t *abuf = NULL;
	rl_t *rl;

	if (byteswap)
	byteswap_uint64_array(lr, sizeof (*lr));

	offset = lr->lr_offset;
	length = lr->lr_length;

	/* If it's a dmu_sync() block, write the whole block */
	if (lr->lr_common.lrc_reclen == sizeof (lr_write_t)) {
	uint64_t blocksize = BP_GET_LSIZE(&lr->lr_blkptr);
	if (length < blocksize) {
	offset -= offset % blocksize;
	length = blocksize;
	}
	}

	if (bt->bt_magic == BSWAP_64(BT_MAGIC))
	byteswap_uint64_array(bt, sizeof (*bt));

	if (bt->bt_magic != BT_MAGIC)
	bt = NULL;

	ztest_object_lock(zd, lr->lr_foid, RL_READER);
	rl = ztest_range_lock(zd, lr->lr_foid, offset, length, RL_WRITER);

	VERIFY0(dmu_bonus_hold(os, lr->lr_foid, FTAG, &db));

	dmu_object_info_from_db(db, &doi);

	bbt = ztest_bt_bonus(db);
	ASSERT3U(bbt->bt_magic, ==, BT_MAGIC);
	gen = bbt->bt_gen;
	crtxg = bbt->bt_crtxg;
	lrtxg = lr->lr_common.lrc_txg;

	tx = dmu_tx_create(os);

	dmu_tx_hold_write(tx, lr->lr_foid, offset, length);

	if (ztest_random(8) == 0 && length == doi.doi_data_block_size &&
	P2PHASE(offset, length) == 0)
	abuf = dmu_request_arcbuf(db, length);

	txg = ztest_tx_assign(tx, TXG_WAIT, FTAG);
	if (txg == 0) {
	if (abuf != NULL)
	dmu_return_arcbuf(abuf);
	dmu_buf_rele(db, FTAG);
	ztest_range_unlock(rl);
	ztest_object_unlock(zd, lr->lr_foid);
	return (ENOSPC);
	}

	if (bt != NULL) {
	/*
	* Usually, verify the old data before writing new data --
	* but not always, because we also want to verify correct
	* behavior when the data was not recently read into cache.
	*/
	ASSERT0(offset % doi.doi_data_block_size);
	if (ztest_random(4) != 0) {
	int prefetch = ztest_random(2) ?
	DMU_READ_PREFETCH : DMU_READ_NO_PREFETCH;
	ztest_block_tag_t rbt;

	VERIFY(dmu_read(os, lr->lr_foid, offset,
	sizeof (rbt), &rbt, prefetch) == 0);
	if (rbt.bt_magic == BT_MAGIC) {
	ztest_bt_verify(&rbt, os, lr->lr_foid, 0,
	offset, gen, txg, crtxg);
	}
	}

	/*
	* Writes can appear to be newer than the bonus buffer because
	* the ztest_get_data() callback does a dmu_read() of the
	* open-context data, which may be different than the data
	* as it was when the write was generated.
	*/
	if (zd->zd_zilog->zl_replay) {
	ztest_bt_verify(bt, os, lr->lr_foid, 0, offset,
	MAX(gen, bt->bt_gen), MAX(txg, lrtxg),
	bt->bt_crtxg);
	}

	/*
	* Set the bt's gen/txg to the bonus buffer's gen/txg
	* so that all of the usual ASSERTs will work.
	*/
	ztest_bt_generate(bt, os, lr->lr_foid, 0, offset, gen, txg,
	crtxg);
	}

	if (abuf == NULL) {
	dmu_write(os, lr->lr_foid, offset, length, data, tx);
	} else {
	bcopy(data, abuf->b_data, length);
	dmu_assign_arcbuf_by_dbuf(db, offset, abuf, tx);
	}

	(void) ztest_log_write(zd, tx, lr);

	dmu_buf_rele(db, FTAG);

	dmu_tx_commit(tx);

	ztest_range_unlock(rl);
	ztest_object_unlock(zd, lr->lr_foid);

	return (0);
	}

	static int
	ztest_replay_truncate(void arg1, void arg2, boolean_t byteswap)
	{
	ztest_ds_t *zd = arg1;
	lr_truncate_t *lr = arg2;
	objset_t *os = zd->zd_os;
	dmu_tx_t *tx;
	uint64_t txg;
	rl_t *rl;

	if (byteswap)
	byteswap_uint64_array(lr, sizeof (*lr));

	ztest_object_lock(zd, lr->lr_foid, RL_READER);
	rl = ztest_range_lock(zd, lr->lr_foid, lr->lr_offset, lr->lr_length,
	RL_WRITER);

	tx = dmu_tx_create(os);

	dmu_tx_hold_free(tx, lr->lr_foid, lr->lr_offset, lr->lr_length);

	txg = ztest_tx_assign(tx, TXG_WAIT, FTAG);
	if (txg == 0) {
	ztest_range_unlock(rl);
	ztest_object_unlock(zd, lr->lr_foid);
	return (ENOSPC);
	}

	VERIFY0(dmu_free_range(os, lr->lr_foid, lr->lr_offset,
	lr->lr_length, tx));

	(void) ztest_log_truncate(zd, tx, lr);

	dmu_tx_commit(tx);

	ztest_range_unlock(rl);
	ztest_object_unlock(zd, lr->lr_foid);

	return (0);
	}

	static int
	ztest_replay_setattr(void arg1, void arg2, boolean_t byteswap)
	{
	ztest_ds_t *zd = arg1;
	lr_setattr_t *lr = arg2;
	objset_t *os = zd->zd_os;
	dmu_tx_t *tx;
	dmu_buf_t *db;
	ztest_block_tag_t *bbt;
	uint64_t txg, lrtxg, crtxg, dnodesize;

	if (byteswap)
	byteswap_uint64_array(lr, sizeof (*lr));

	ztest_object_lock(zd, lr->lr_foid, RL_WRITER);

	VERIFY0(dmu_bonus_hold(os, lr->lr_foid, FTAG, &db));

	tx = dmu_tx_create(os);
	dmu_tx_hold_bonus(tx, lr->lr_foid);

	txg = ztest_tx_assign(tx, TXG_WAIT, FTAG);
	if (txg == 0) {
	dmu_buf_rele(db, FTAG);
	ztest_object_unlock(zd, lr->lr_foid);
	return (ENOSPC);
	}

	bbt = ztest_bt_bonus(db);
	ASSERT3U(bbt->bt_magic, ==, BT_MAGIC);
	crtxg = bbt->bt_crtxg;
	lrtxg = lr->lr_common.lrc_txg;
	dnodesize = bbt->bt_dnodesize;

	if (zd->zd_zilog->zl_replay) {
	ASSERT3U(lr->lr_size, !=, 0);
	ASSERT3U(lr->lr_mode, !=, 0);
	ASSERT3U(lrtxg, !=, 0);
	} else {
	/*
	* Randomly change the size and increment the generation.
	*/
	lr->lr_size = (ztest_random(db->db_size / sizeof (bbt)) + 1)
	sizeof (*bbt);
	lr->lr_mode = bbt->bt_gen + 1;
	ASSERT0(lrtxg);
	}

	/*
	* Verify that the current bonus buffer is not newer than our txg.
	*/
	ztest_bt_verify(bbt, os, lr->lr_foid, dnodesize, -1ULL, lr->lr_mode,
	MAX(txg, lrtxg), crtxg);

	dmu_buf_will_dirty(db, tx);

	ASSERT3U(lr->lr_size, >=, sizeof (*bbt));
	ASSERT3U(lr->lr_size, <=, db->db_size);
	VERIFY0(dmu_set_bonus(db, lr->lr_size, tx));
	bbt = ztest_bt_bonus(db);

	ztest_bt_generate(bbt, os, lr->lr_foid, dnodesize, -1ULL, lr->lr_mode,
	txg, crtxg);
	ztest_fill_unused_bonus(db, bbt, lr->lr_foid, os, bbt->bt_gen);
	dmu_buf_rele(db, FTAG);

	(void) ztest_log_setattr(zd, tx, lr);

	dmu_tx_commit(tx);

	ztest_object_unlock(zd, lr->lr_foid);

	return (0);
	}

	zil_replay_func_t *ztest_replay_vector[TX_MAX_TYPE] = {
	NULL, /* 0 no such transaction type */
	ztest_replay_create, /* TX_CREATE */
	NULL, /* TX_MKDIR */
	NULL, /* TX_MKXATTR */
	NULL, /* TX_SYMLINK */
	ztest_replay_remove, /* TX_REMOVE */
	NULL, /* TX_RMDIR */
	NULL, /* TX_LINK */
	NULL, /* TX_RENAME */
	ztest_replay_write, /* TX_WRITE */
	ztest_replay_truncate, /* TX_TRUNCATE */
	ztest_replay_setattr, /* TX_SETATTR */
	NULL, /* TX_ACL */
	NULL, /* TX_CREATE_ACL */
	NULL, /* TX_CREATE_ATTR */
	NULL, /* TX_CREATE_ACL_ATTR */
	NULL, /* TX_MKDIR_ACL */
	NULL, /* TX_MKDIR_ATTR */
	NULL, /* TX_MKDIR_ACL_ATTR */
	NULL, /* TX_WRITE2 */
	};

	/*
	* ZIL get_data callbacks
	*/

	/* ARGSUSED */
	static void
	ztest_get_done(zgd_t *zgd, int error)
	{
	ztest_ds_t *zd = zgd->zgd_private;
	uint64_t object = ((rl_t *)zgd->zgd_lr)->rl_object;

	if (zgd->zgd_db)
	dmu_buf_rele(zgd->zgd_db, zgd);

	ztest_range_unlock((rl_t *)zgd->zgd_lr);
	ztest_object_unlock(zd, object);

	umem_free(zgd, sizeof (*zgd));
	}

	static int
	ztest_get_data(void arg, uint64_t arg2, lr_write_t lr, char *buf,
	struct lwb lwb, zio_t zio)
	{
	ztest_ds_t *zd = arg;
	objset_t *os = zd->zd_os;
	uint64_t object = lr->lr_foid;
	uint64_t offset = lr->lr_offset;
	uint64_t size = lr->lr_length;
	uint64_t txg = lr->lr_common.lrc_txg;
	uint64_t crtxg;
	dmu_object_info_t doi;
	dmu_buf_t *db;
	zgd_t *zgd;
	int error;

	ASSERT3P(lwb, !=, NULL);
	ASSERT3P(zio, !=, NULL);
	ASSERT3U(size, !=, 0);

	ztest_object_lock(zd, object, RL_READER);
	error = dmu_bonus_hold(os, object, FTAG, &db);
	if (error) {
	ztest_object_unlock(zd, object);
	return (error);
	}

	crtxg = ztest_bt_bonus(db)->bt_crtxg;

	if (crtxg == 0 \|\| crtxg > txg) {
	dmu_buf_rele(db, FTAG);
	ztest_object_unlock(zd, object);
	return (ENOENT);
	}

	dmu_object_info_from_db(db, &doi);
	dmu_buf_rele(db, FTAG);
	db = NULL;

	zgd = umem_zalloc(sizeof (*zgd), UMEM_NOFAIL);
	zgd->zgd_lwb = lwb;
	zgd->zgd_private = zd;

	if (buf != NULL) { /* immediate write */
	zgd->zgd_lr = (struct zfs_locked_range *)ztest_range_lock(zd,
	object, offset, size, RL_READER);

	error = dmu_read(os, object, offset, size, buf,
	DMU_READ_NO_PREFETCH);
	ASSERT0(error);
	} else {
	size = doi.doi_data_block_size;
	if (ISP2(size)) {
	offset = P2ALIGN(offset, size);
	} else {
	ASSERT3U(offset, <, size);
	offset = 0;
	}

	zgd->zgd_lr = (struct zfs_locked_range *)ztest_range_lock(zd,
	object, offset, size, RL_READER);

	error = dmu_buf_hold(os, object, offset, zgd, &db,
	DMU_READ_NO_PREFETCH);

	if (error == 0) {
	blkptr_t *bp = &lr->lr_blkptr;

	zgd->zgd_db = db;
	zgd->zgd_bp = bp;

	ASSERT3U(db->db_offset, ==, offset);
	ASSERT3U(db->db_size, ==, size);

	error = dmu_sync(zio, lr->lr_common.lrc_txg,
	ztest_get_done, zgd);

	if (error == 0)
	return (0);
	}
	}

	ztest_get_done(zgd, error);

	return (error);
	}

	static void *
	ztest_lr_alloc(size_t lrsize, char *name)
	{
	char *lr;
	size_t namesize = name ? strlen(name) + 1 : 0;

	lr = umem_zalloc(lrsize + namesize, UMEM_NOFAIL);

	if (name)
	bcopy(name, lr + lrsize, namesize);

	return (lr);
	}

	static void
	ztest_lr_free(void lr, size_t lrsize, char name)
	{
	size_t namesize = name ? strlen(name) + 1 : 0;

	umem_free(lr, lrsize + namesize);
	}

	/*
	* Lookup a bunch of objects. Returns the number of objects not found.
	*/
	static int
	ztest_lookup(ztest_ds_t zd, ztest_od_t od, int count)
	{
	int missing = 0;
	int error;
	int i;

	ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock));

	for (i = 0; i < count; i++, od++) {
	od->od_object = 0;
	error = zap_lookup(zd->zd_os, od->od_dir, od->od_name,
	sizeof (uint64_t), 1, &od->od_object);
	if (error) {
	ASSERT3S(error, ==, ENOENT);
	ASSERT0(od->od_object);
	missing++;
	} else {
	dmu_buf_t *db;
	ztest_block_tag_t *bbt;
	dmu_object_info_t doi;

	ASSERT3U(od->od_object, !=, 0);
	ASSERT0(missing); /* there should be no gaps */

	ztest_object_lock(zd, od->od_object, RL_READER);
	VERIFY0(dmu_bonus_hold(zd->zd_os, od->od_object,
	FTAG, &db));
	dmu_object_info_from_db(db, &doi);
	bbt = ztest_bt_bonus(db);
	ASSERT3U(bbt->bt_magic, ==, BT_MAGIC);
	od->od_type = doi.doi_type;
	od->od_blocksize = doi.doi_data_block_size;
	od->od_gen = bbt->bt_gen;
	dmu_buf_rele(db, FTAG);
	ztest_object_unlock(zd, od->od_object);
	}
	}

	return (missing);
	}

	static int
	ztest_create(ztest_ds_t zd, ztest_od_t od, int count)
	{
	int missing = 0;
	int i;

	ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock));

	for (i = 0; i < count; i++, od++) {
	if (missing) {
	od->od_object = 0;
	missing++;
	continue;
	}

	lr_create_t lr = ztest_lr_alloc(sizeof (lr), od->od_name);

	lr->lr_doid = od->od_dir;
	lr->lr_foid = 0; /* 0 to allocate, > 0 to claim */
	lr->lrz_type = od->od_crtype;
	lr->lrz_blocksize = od->od_crblocksize;
	lr->lrz_ibshift = ztest_random_ibshift();
	lr->lrz_bonustype = DMU_OT_UINT64_OTHER;
	lr->lrz_dnodesize = od->od_crdnodesize;
	lr->lr_gen = od->od_crgen;
	lr->lr_crtime[0] = time(NULL);

	if (ztest_replay_create(zd, lr, B_FALSE) != 0) {
	ASSERT0(missing);
	od->od_object = 0;
	missing++;
	} else {
	od->od_object = lr->lr_foid;
	od->od_type = od->od_crtype;
	od->od_blocksize = od->od_crblocksize;
	od->od_gen = od->od_crgen;
	ASSERT3U(od->od_object, !=, 0);
	}

	ztest_lr_free(lr, sizeof (*lr), od->od_name);
	}

	return (missing);
	}

	static int
	ztest_remove(ztest_ds_t zd, ztest_od_t od, int count)
	{
	int missing = 0;
	int error;
	int i;

	ASSERT(MUTEX_HELD(&zd->zd_dirobj_lock));

	od += count - 1;

	for (i = count - 1; i >= 0; i--, od--) {
	if (missing) {
	missing++;
	continue;
	}

	/*
	* No object was found.
	*/
	if (od->od_object == 0)
	continue;

	lr_remove_t lr = ztest_lr_alloc(sizeof (lr), od->od_name);

	lr->lr_doid = od->od_dir;

	if ((error = ztest_replay_remove(zd, lr, B_FALSE)) != 0) {
	ASSERT3U(error, ==, ENOSPC);
	missing++;
	} else {
	od->od_object = 0;
	}
	ztest_lr_free(lr, sizeof (*lr), od->od_name);
	}

	return (missing);
	}

	static int
	ztest_write(ztest_ds_t *zd, uint64_t object, uint64_t offset, uint64_t size,
	void *data)
	{
	lr_write_t *lr;
	int error;

	lr = ztest_lr_alloc(sizeof (*lr) + size, NULL);

	lr->lr_foid = object;
	lr->lr_offset = offset;
	lr->lr_length = size;
	lr->lr_blkoff = 0;
	BP_ZERO(&lr->lr_blkptr);

	bcopy(data, lr + 1, size);

	error = ztest_replay_write(zd, lr, B_FALSE);

	ztest_lr_free(lr, sizeof (*lr) + size, NULL);

	return (error);
	}

	static int
	ztest_truncate(ztest_ds_t *zd, uint64_t object, uint64_t offset, uint64_t size)
	{
	lr_truncate_t *lr;
	int error;

	lr = ztest_lr_alloc(sizeof (*lr), NULL);

	lr->lr_foid = object;
	lr->lr_offset = offset;
	lr->lr_length = size;

	error = ztest_replay_truncate(zd, lr, B_FALSE);

	ztest_lr_free(lr, sizeof (*lr), NULL);

	return (error);
	}

	static int
	ztest_setattr(ztest_ds_t *zd, uint64_t object)
	{
	lr_setattr_t *lr;
	int error;

	lr = ztest_lr_alloc(sizeof (*lr), NULL);

	lr->lr_foid = object;
	lr->lr_size = 0;
	lr->lr_mode = 0;

	error = ztest_replay_setattr(zd, lr, B_FALSE);

	ztest_lr_free(lr, sizeof (*lr), NULL);

	return (error);
	}

	static void
	ztest_prealloc(ztest_ds_t *zd, uint64_t object, uint64_t offset, uint64_t size)
	{
	objset_t *os = zd->zd_os;
	dmu_tx_t *tx;
	uint64_t txg;
	rl_t *rl;

	txg_wait_synced(dmu_objset_pool(os), 0);

	ztest_object_lock(zd, object, RL_READER);
	rl = ztest_range_lock(zd, object, offset, size, RL_WRITER);

	tx = dmu_tx_create(os);

	dmu_tx_hold_write(tx, object, offset, size);

	txg = ztest_tx_assign(tx, TXG_WAIT, FTAG);

	if (txg != 0) {
	dmu_prealloc(os, object, offset, size, tx);
	dmu_tx_commit(tx);
	txg_wait_synced(dmu_objset_pool(os), txg);
	} else {
	(void) dmu_free_long_range(os, object, offset, size);
	}

	ztest_range_unlock(rl);
	ztest_object_unlock(zd, object);
	}

	static void
	ztest_io(ztest_ds_t *zd, uint64_t object, uint64_t offset)
	{
	int err;
	ztest_block_tag_t wbt;
	dmu_object_info_t doi;
	enum ztest_io_type io_type;
	uint64_t blocksize;
	void *data;

	VERIFY0(dmu_object_info(zd->zd_os, object, &doi));
	blocksize = doi.doi_data_block_size;
	data = umem_alloc(blocksize, UMEM_NOFAIL);

	/*
	* Pick an i/o type at random, biased toward writing block tags.
	*/
	io_type = ztest_random(ZTEST_IO_TYPES);
	if (ztest_random(2) == 0)
	io_type = ZTEST_IO_WRITE_TAG;

	(void) pthread_rwlock_rdlock(&zd->zd_zilog_lock);

	switch (io_type) {

	case ZTEST_IO_WRITE_TAG:
	ztest_bt_generate(&wbt, zd->zd_os, object, doi.doi_dnodesize,
	offset, 0, 0, 0);
	(void) ztest_write(zd, object, offset, sizeof (wbt), &wbt);
	break;

	case ZTEST_IO_WRITE_PATTERN:
	(void) memset(data, 'a' + (object + offset) % 5, blocksize);
	if (ztest_random(2) == 0) {
	/*
	* Induce fletcher2 collisions to ensure that
	* zio_ddt_collision() detects and resolves them
	* when using fletcher2-verify for deduplication.
	*/
	((uint64_t *)data)[0] ^= 1ULL << 63;
	((uint64_t *)data)[4] ^= 1ULL << 63;
	}
	(void) ztest_write(zd, object, offset, blocksize, data);
	break;

	case ZTEST_IO_WRITE_ZEROES:
	bzero(data, blocksize);
	(void) ztest_write(zd, object, offset, blocksize, data);
	break;

	case ZTEST_IO_TRUNCATE:
	(void) ztest_truncate(zd, object, offset, blocksize);
	break;

	case ZTEST_IO_SETATTR:
	(void) ztest_setattr(zd, object);
	break;
	default:
	break;

	case ZTEST_IO_REWRITE:
	(void) pthread_rwlock_rdlock(&ztest_name_lock);
	err = ztest_dsl_prop_set_uint64(zd->zd_name,
	ZFS_PROP_CHECKSUM, spa_dedup_checksum(ztest_spa),
	B_FALSE);
	VERIFY(err == 0 \|\| err == ENOSPC);
	err = ztest_dsl_prop_set_uint64(zd->zd_name,
	ZFS_PROP_COMPRESSION,
	ztest_random_dsl_prop(ZFS_PROP_COMPRESSION),
	B_FALSE);
	VERIFY(err == 0 \|\| err == ENOSPC);
	(void) pthread_rwlock_unlock(&ztest_name_lock);

	VERIFY0(dmu_read(zd->zd_os, object, offset, blocksize, data,
	DMU_READ_NO_PREFETCH));

	(void) ztest_write(zd, object, offset, blocksize, data);
	break;
	}

	(void) pthread_rwlock_unlock(&zd->zd_zilog_lock);

	umem_free(data, blocksize);
	}

	/*
	* Initialize an object description template.
	*/
	static void
	ztest_od_init(ztest_od_t od, uint64_t id, char tag, uint64_t index,
	dmu_object_type_t type, uint64_t blocksize, uint64_t dnodesize,
	uint64_t gen)
	{
	od->od_dir = ZTEST_DIROBJ;
	od->od_object = 0;

	od->od_crtype = type;
	od->od_crblocksize = blocksize ? blocksize : ztest_random_blocksize();
	od->od_crdnodesize = dnodesize ? dnodesize : ztest_random_dnodesize();
	od->od_crgen = gen;

	od->od_type = DMU_OT_NONE;
	od->od_blocksize = 0;
	od->od_gen = 0;

	(void) snprintf(od->od_name, sizeof (od->od_name),
	"%s(%"PRId64")[%"PRIu64"]",
	tag, id, index);
	}

	/*
	* Lookup or create the objects for a test using the od template.
	* If the objects do not all exist, or if 'remove' is specified,
	* remove any existing objects and create new ones. Otherwise,
	* use the existing objects.
	*/
	static int
	ztest_object_init(ztest_ds_t zd, ztest_od_t od, size_t size, boolean_t remove)
	{
	int count = size / sizeof (*od);
	int rv = 0;

	mutex_enter(&zd->zd_dirobj_lock);
	if ((ztest_lookup(zd, od, count) != 0 \|\| remove) &&
	(ztest_remove(zd, od, count) != 0 \|\|
	ztest_create(zd, od, count) != 0))
	rv = -1;
	zd->zd_od = od;
	mutex_exit(&zd->zd_dirobj_lock);

	return (rv);
	}

	/* ARGSUSED */
	void
	ztest_zil_commit(ztest_ds_t *zd, uint64_t id)
	{
	zilog_t *zilog = zd->zd_zilog;

	(void) pthread_rwlock_rdlock(&zd->zd_zilog_lock);

	zil_commit(zilog, ztest_random(ZTEST_OBJECTS));

	/*
	* Remember the committed values in zd, which is in parent/child
	* shared memory. If we die, the next iteration of ztest_run()
	* will verify that the log really does contain this record.
	*/
	mutex_enter(&zilog->zl_lock);
	ASSERT3P(zd->zd_shared, !=, NULL);
	ASSERT3U(zd->zd_shared->zd_seq, <=, zilog->zl_commit_lr_seq);
	zd->zd_shared->zd_seq = zilog->zl_commit_lr_seq;
	mutex_exit(&zilog->zl_lock);

	(void) pthread_rwlock_unlock(&zd->zd_zilog_lock);
	}

	/*
	* This function is designed to simulate the operations that occur during a
	* mount/unmount operation. We hold the dataset across these operations in an
	* attempt to expose any implicit assumptions about ZIL management.
	*/
	/* ARGSUSED */
	void
	ztest_zil_remount(ztest_ds_t *zd, uint64_t id)
	{
	objset_t *os = zd->zd_os;

	/*
	* We hold the ztest_vdev_lock so we don't cause problems with
	* other threads that wish to remove a log device, such as
	* ztest_device_removal().
	*/
	mutex_enter(&ztest_vdev_lock);

	/*
	* We grab the zd_dirobj_lock to ensure that no other thread is
	* updating the zil (i.e. adding in-memory log records) and the
	* zd_zilog_lock to block any I/O.
	*/
	mutex_enter(&zd->zd_dirobj_lock);
	(void) pthread_rwlock_wrlock(&zd->zd_zilog_lock);

	/* zfsvfs_teardown() */
	zil_close(zd->zd_zilog);

	/* zfsvfs_setup() */
	VERIFY3P(zil_open(os, ztest_get_data), ==, zd->zd_zilog);
	zil_replay(os, zd, ztest_replay_vector);

	(void) pthread_rwlock_unlock(&zd->zd_zilog_lock);
	mutex_exit(&zd->zd_dirobj_lock);
	mutex_exit(&ztest_vdev_lock);
	}

	/*
	* Verify that we can't destroy an active pool, create an existing pool,
	* or create a pool with a bad vdev spec.
	*/
	/* ARGSUSED */
	void
	ztest_spa_create_destroy(ztest_ds_t *zd, uint64_t id)
	{
	ztest_shared_opts_t *zo = &ztest_opts;
	spa_t *spa;
	nvlist_t *nvroot;

	if (zo->zo_mmp_test)
	return;

	/*
	* Attempt to create using a bad file.
	*/
	nvroot = make_vdev_root("/dev/bogus", NULL, NULL, 0, 0, NULL, 0, 0, 1);
	VERIFY3U(ENOENT, ==,
	spa_create("ztest_bad_file", nvroot, NULL, NULL, NULL));
	fnvlist_free(nvroot);

	/*
	* Attempt to create using a bad mirror.
	*/
	nvroot = make_vdev_root("/dev/bogus", NULL, NULL, 0, 0, NULL, 0, 2, 1);
	VERIFY3U(ENOENT, ==,
	spa_create("ztest_bad_mirror", nvroot, NULL, NULL, NULL));
	fnvlist_free(nvroot);

	/*
	* Attempt to create an existing pool. It shouldn't matter
	* what's in the nvroot; we should fail with EEXIST.
	*/
	(void) pthread_rwlock_rdlock(&ztest_name_lock);
	nvroot = make_vdev_root("/dev/bogus", NULL, NULL, 0, 0, NULL, 0, 0, 1);
	VERIFY3U(EEXIST, ==,
	spa_create(zo->zo_pool, nvroot, NULL, NULL, NULL));
	fnvlist_free(nvroot);

	/*
	* We open a reference to the spa and then we try to export it
	* expecting one of the following errors:
	*
	* EBUSY
	* Because of the reference we just opened.
	*
	* ZFS_ERR_EXPORT_IN_PROGRESS
	* For the case that there is another ztest thread doing
	* an export concurrently.
	*/
	VERIFY0(spa_open(zo->zo_pool, &spa, FTAG));
	int error = spa_destroy(zo->zo_pool);
	if (error != EBUSY && error != ZFS_ERR_EXPORT_IN_PROGRESS) {
	fatal(B_FALSE, "spa_destroy(%s) returned unexpected value %d",
	spa->spa_name, error);
	}
	spa_close(spa, FTAG);

	(void) pthread_rwlock_unlock(&ztest_name_lock);
	}

	/*
	* Start and then stop the MMP threads to ensure the startup and shutdown code
	* works properly. Actual protection and property-related code tested via ZTS.
	*/
	/* ARGSUSED */
	void
	ztest_mmp_enable_disable(ztest_ds_t *zd, uint64_t id)
	{
	ztest_shared_opts_t *zo = &ztest_opts;
	spa_t *spa = ztest_spa;

	if (zo->zo_mmp_test)
	return;

	/*
	* Since enabling MMP involves setting a property, it could not be done
	* while the pool is suspended.
	*/
	if (spa_suspended(spa))
	return;

	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	mutex_enter(&spa->spa_props_lock);

	zfs_multihost_fail_intervals = 0;

	if (!spa_multihost(spa)) {
	spa->spa_multihost = B_TRUE;
	mmp_thread_start(spa);
	}

	mutex_exit(&spa->spa_props_lock);
	spa_config_exit(spa, SCL_CONFIG, FTAG);

	txg_wait_synced(spa_get_dsl(spa), 0);
	mmp_signal_all_threads();
	txg_wait_synced(spa_get_dsl(spa), 0);

	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	mutex_enter(&spa->spa_props_lock);

	if (spa_multihost(spa)) {
	mmp_thread_stop(spa);
	spa->spa_multihost = B_FALSE;
	}

	mutex_exit(&spa->spa_props_lock);
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	}

	/* ARGSUSED */
	void
	ztest_spa_upgrade(ztest_ds_t *zd, uint64_t id)
	{
	spa_t *spa;
	uint64_t initial_version = SPA_VERSION_INITIAL;
	uint64_t version, newversion;
	nvlist_t nvroot, props;
	char *name;

	if (ztest_opts.zo_mmp_test)
	return;

	/* dRAID added after feature flags, skip upgrade test. */
	if (strcmp(ztest_opts.zo_raid_type, VDEV_TYPE_DRAID) == 0)
	return;

	mutex_enter(&ztest_vdev_lock);
	name = kmem_asprintf("%s_upgrade", ztest_opts.zo_pool);

	/*
	* Clean up from previous runs.
	*/
	(void) spa_destroy(name);

	nvroot = make_vdev_root(NULL, NULL, name, ztest_opts.zo_vdev_size, 0,
	NULL, ztest_opts.zo_raid_children, ztest_opts.zo_mirrors, 1);

	/*
	* If we're configuring a RAIDZ device then make sure that the
	* initial version is capable of supporting that feature.
	*/
	switch (ztest_opts.zo_raid_parity) {
	case 0:
	case 1:
	initial_version = SPA_VERSION_INITIAL;
	break;
	case 2:
	initial_version = SPA_VERSION_RAIDZ2;
	break;
	case 3:
	initial_version = SPA_VERSION_RAIDZ3;
	break;
	}

	/*
	* Create a pool with a spa version that can be upgraded. Pick
	* a value between initial_version and SPA_VERSION_BEFORE_FEATURES.
	*/
	do {
	version = ztest_random_spa_version(initial_version);
	} while (version > SPA_VERSION_BEFORE_FEATURES);

	props = fnvlist_alloc();
	fnvlist_add_uint64(props,
	zpool_prop_to_name(ZPOOL_PROP_VERSION), version);
	VERIFY0(spa_create(name, nvroot, props, NULL, NULL));
	fnvlist_free(nvroot);
	fnvlist_free(props);

	VERIFY0(spa_open(name, &spa, FTAG));
	VERIFY3U(spa_version(spa), ==, version);
	newversion = ztest_random_spa_version(version + 1);

	if (ztest_opts.zo_verbose >= 4) {
	(void) printf("upgrading spa version from "
	"%"PRIu64" to %"PRIu64"\n",
	version, newversion);
	}

	spa_upgrade(spa, newversion);
	VERIFY3U(spa_version(spa), >, version);
	VERIFY3U(spa_version(spa), ==, fnvlist_lookup_uint64(spa->spa_config,
	zpool_prop_to_name(ZPOOL_PROP_VERSION)));
	spa_close(spa, FTAG);

	kmem_strfree(name);
	mutex_exit(&ztest_vdev_lock);
	}

	static void
	ztest_spa_checkpoint(spa_t *spa)
	{
	ASSERT(MUTEX_HELD(&ztest_checkpoint_lock));

	int error = spa_checkpoint(spa->spa_name);

	switch (error) {
	case 0:
	case ZFS_ERR_DEVRM_IN_PROGRESS:
	case ZFS_ERR_DISCARDING_CHECKPOINT:
	case ZFS_ERR_CHECKPOINT_EXISTS:
	break;
	case ENOSPC:
	ztest_record_enospc(FTAG);
	break;
	default:
	fatal(B_FALSE, "spa_checkpoint(%s) = %d", spa->spa_name, error);
	}
	}

	static void
	ztest_spa_discard_checkpoint(spa_t *spa)
	{
	ASSERT(MUTEX_HELD(&ztest_checkpoint_lock));

	int error = spa_checkpoint_discard(spa->spa_name);

	switch (error) {
	case 0:
	case ZFS_ERR_DISCARDING_CHECKPOINT:
	case ZFS_ERR_NO_CHECKPOINT:
	break;
	default:
	fatal(B_FALSE, "spa_discard_checkpoint(%s) = %d",
	spa->spa_name, error);
	}

	}

	/* ARGSUSED */
	void
	ztest_spa_checkpoint_create_discard(ztest_ds_t *zd, uint64_t id)
	{
	spa_t *spa = ztest_spa;

	mutex_enter(&ztest_checkpoint_lock);
	if (ztest_random(2) == 0) {
	ztest_spa_checkpoint(spa);
	} else {
	ztest_spa_discard_checkpoint(spa);
	}
	mutex_exit(&ztest_checkpoint_lock);
	}


	static vdev_t *
	vdev_lookup_by_path(vdev_t vd, const char path)
	{
	vdev_t *mvd;
	int c;

	if (vd->vdev_path != NULL && strcmp(path, vd->vdev_path) == 0)
	return (vd);

	for (c = 0; c < vd->vdev_children; c++)
	if ((mvd = vdev_lookup_by_path(vd->vdev_child[c], path)) !=
	NULL)
	return (mvd);

	return (NULL);
	}

	static int
	spa_num_top_vdevs(spa_t *spa)
	{
	vdev_t *rvd = spa->spa_root_vdev;
	ASSERT3U(spa_config_held(spa, SCL_VDEV, RW_READER), ==, SCL_VDEV);
	return (rvd->vdev_children);
	}

	/*
	* Verify that vdev_add() works as expected.
	*/
	/* ARGSUSED */
	void
	ztest_vdev_add_remove(ztest_ds_t *zd, uint64_t id)
	{
	ztest_shared_t *zs = ztest_shared;
	spa_t *spa = ztest_spa;
	uint64_t leaves;
	uint64_t guid;
	nvlist_t *nvroot;
	int error;

	if (ztest_opts.zo_mmp_test)
	return;

	mutex_enter(&ztest_vdev_lock);
	leaves = MAX(zs->zs_mirrors + zs->zs_splits, 1) *
	ztest_opts.zo_raid_children;

	spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);

	ztest_shared->zs_vdev_next_leaf = spa_num_top_vdevs(spa) * leaves;

	/*
	* If we have slogs then remove them 1/4 of the time.
	*/
	if (spa_has_slogs(spa) && ztest_random(4) == 0) {
	metaslab_group_t *mg;

	/*
	* find the first real slog in log allocation class
	*/
	mg = spa_log_class(spa)->mc_allocator[0].mca_rotor;
	while (!mg->mg_vd->vdev_islog)
	mg = mg->mg_next;

	guid = mg->mg_vd->vdev_guid;

	spa_config_exit(spa, SCL_VDEV, FTAG);

	/*
	* We have to grab the zs_name_lock as writer to
	* prevent a race between removing a slog (dmu_objset_find)
	* and destroying a dataset. Removing the slog will
	* grab a reference on the dataset which may cause
	* dsl_destroy_head() to fail with EBUSY thus
	* leaving the dataset in an inconsistent state.
	*/
	pthread_rwlock_wrlock(&ztest_name_lock);
	error = spa_vdev_remove(spa, guid, B_FALSE);
	pthread_rwlock_unlock(&ztest_name_lock);

	switch (error) {
	case 0:
	case EEXIST: /* Generic zil_reset() error */
	case EBUSY: /* Replay required */
	case EACCES: /* Crypto key not loaded */
	case ZFS_ERR_CHECKPOINT_EXISTS:
	case ZFS_ERR_DISCARDING_CHECKPOINT:
	break;
	default:
	fatal(B_FALSE, "spa_vdev_remove() = %d", error);
	}
	} else {
	spa_config_exit(spa, SCL_VDEV, FTAG);

	/*
	* Make 1/4 of the devices be log devices
	*/
	nvroot = make_vdev_root(NULL, NULL, NULL,
	ztest_opts.zo_vdev_size, 0, (ztest_random(4) == 0) ?
	"log" : NULL, ztest_opts.zo_raid_children, zs->zs_mirrors,
	1);

	error = spa_vdev_add(spa, nvroot);
	fnvlist_free(nvroot);

	switch (error) {
	case 0:
	break;
	case ENOSPC:
	ztest_record_enospc("spa_vdev_add");
	break;
	default:
	fatal(B_FALSE, "spa_vdev_add() = %d", error);
	}
	}

	mutex_exit(&ztest_vdev_lock);
	}

	/* ARGSUSED */
	void
	ztest_vdev_class_add(ztest_ds_t *zd, uint64_t id)
	{
	ztest_shared_t *zs = ztest_shared;
	spa_t *spa = ztest_spa;
	uint64_t leaves;
	nvlist_t *nvroot;
	const char *class = (ztest_random(2) == 0) ?
	VDEV_ALLOC_BIAS_SPECIAL : VDEV_ALLOC_BIAS_DEDUP;
	int error;

	/*
	* By default add a special vdev 50% of the time
	*/
	if ((ztest_opts.zo_special_vdevs == ZTEST_VDEV_CLASS_OFF) \|\|
	(ztest_opts.zo_special_vdevs == ZTEST_VDEV_CLASS_RND &&
	ztest_random(2) == 0)) {
	return;
	}

	mutex_enter(&ztest_vdev_lock);

	/* Only test with mirrors */
	if (zs->zs_mirrors < 2) {
	mutex_exit(&ztest_vdev_lock);
	return;
	}

	/* requires feature@allocation_classes */
	if (!spa_feature_is_enabled(spa, SPA_FEATURE_ALLOCATION_CLASSES)) {
	mutex_exit(&ztest_vdev_lock);
	return;
	}

	leaves = MAX(zs->zs_mirrors + zs->zs_splits, 1) *
	ztest_opts.zo_raid_children;

	spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);
	ztest_shared->zs_vdev_next_leaf = spa_num_top_vdevs(spa) * leaves;
	spa_config_exit(spa, SCL_VDEV, FTAG);

	nvroot = make_vdev_root(NULL, NULL, NULL, ztest_opts.zo_vdev_size, 0,
	class, ztest_opts.zo_raid_children, zs->zs_mirrors, 1);

	error = spa_vdev_add(spa, nvroot);
	fnvlist_free(nvroot);

	if (error == ENOSPC)
	ztest_record_enospc("spa_vdev_add");
	else if (error != 0)
	fatal(B_FALSE, "spa_vdev_add() = %d", error);

	/*
	* 50% of the time allow small blocks in the special class
	*/
	if (error == 0 &&
	spa_special_class(spa)->mc_groups == 1 && ztest_random(2) == 0) {
	if (ztest_opts.zo_verbose >= 3)
	(void) printf("Enabling special VDEV small blocks\n");
	(void) ztest_dsl_prop_set_uint64(zd->zd_name,
	ZFS_PROP_SPECIAL_SMALL_BLOCKS, 32768, B_FALSE);
	}

	mutex_exit(&ztest_vdev_lock);

	if (ztest_opts.zo_verbose >= 3) {
	metaslab_class_t *mc;

	if (strcmp(class, VDEV_ALLOC_BIAS_SPECIAL) == 0)
	mc = spa_special_class(spa);
	else
	mc = spa_dedup_class(spa);
	(void) printf("Added a %s mirrored vdev (of %d)\n",
	class, (int)mc->mc_groups);
	}
	}

	/*
	* Verify that adding/removing aux devices (l2arc, hot spare) works as expected.
	*/
	/* ARGSUSED */
	void
	ztest_vdev_aux_add_remove(ztest_ds_t *zd, uint64_t id)
	{
	ztest_shared_t *zs = ztest_shared;
	spa_t *spa = ztest_spa;
	vdev_t *rvd = spa->spa_root_vdev;
	spa_aux_vdev_t *sav;
	char *aux;
	char *path;
	uint64_t guid = 0;
	int error, ignore_err = 0;

	if (ztest_opts.zo_mmp_test)
	return;

	path = umem_alloc(MAXPATHLEN, UMEM_NOFAIL);

	if (ztest_random(2) == 0) {
	sav = &spa->spa_spares;
	aux = ZPOOL_CONFIG_SPARES;
	} else {
	sav = &spa->spa_l2cache;
	aux = ZPOOL_CONFIG_L2CACHE;
	}

	mutex_enter(&ztest_vdev_lock);

	spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);

	if (sav->sav_count != 0 && ztest_random(4) == 0) {
	/*
	* Pick a random device to remove.
	*/
	vdev_t *svd = sav->sav_vdevs[ztest_random(sav->sav_count)];

	/* dRAID spares cannot be removed; try anyways to see ENOTSUP */
	if (strstr(svd->vdev_path, VDEV_TYPE_DRAID) != NULL)
	ignore_err = ENOTSUP;

	guid = svd->vdev_guid;
	} else {
	/*
	* Find an unused device we can add.
	*/
	zs->zs_vdev_aux = 0;
	for (;;) {
	int c;
	(void) snprintf(path, MAXPATHLEN, ztest_aux_template,
	ztest_opts.zo_dir, ztest_opts.zo_pool, aux,
	zs->zs_vdev_aux);
	for (c = 0; c < sav->sav_count; c++)
	if (strcmp(sav->sav_vdevs[c]->vdev_path,
	path) == 0)
	break;
	if (c == sav->sav_count &&
	vdev_lookup_by_path(rvd, path) == NULL)
	break;
	zs->zs_vdev_aux++;
	}
	}

	spa_config_exit(spa, SCL_VDEV, FTAG);

	if (guid == 0) {
	/*
	* Add a new device.
	*/
	nvlist_t *nvroot = make_vdev_root(NULL, aux, NULL,
	(ztest_opts.zo_vdev_size * 5) / 4, 0, NULL, 0, 0, 1);
	error = spa_vdev_add(spa, nvroot);

	switch (error) {
	case 0:
	break;
	default:
	fatal(B_FALSE, "spa_vdev_add(%p) = %d", nvroot, error);
	}
	fnvlist_free(nvroot);
	} else {
	/*
	* Remove an existing device. Sometimes, dirty its
	* vdev state first to make sure we handle removal
	* of devices that have pending state changes.
	*/
	if (ztest_random(2) == 0)
	(void) vdev_online(spa, guid, 0, NULL);

	error = spa_vdev_remove(spa, guid, B_FALSE);

	switch (error) {
	case 0:
	case EBUSY:
	case ZFS_ERR_CHECKPOINT_EXISTS:
	case ZFS_ERR_DISCARDING_CHECKPOINT:
	break;
	default:
	if (error != ignore_err)
	fatal(B_FALSE,
	"spa_vdev_remove(%"PRIu64") = %d",
	guid, error);
	}
	}

	mutex_exit(&ztest_vdev_lock);

	umem_free(path, MAXPATHLEN);
	}

	/*
	* split a pool if it has mirror tlvdevs
	*/
	/* ARGSUSED */
	void
	ztest_split_pool(ztest_ds_t *zd, uint64_t id)
	{
	ztest_shared_t *zs = ztest_shared;
	spa_t *spa = ztest_spa;
	vdev_t *rvd = spa->spa_root_vdev;
	nvlist_t tree, child, config, split, *schild;
	uint_t c, children, schildren = 0, lastlogid = 0;
	int error = 0;

	if (ztest_opts.zo_mmp_test)
	return;

	mutex_enter(&ztest_vdev_lock);

	/* ensure we have a usable config; mirrors of raidz aren't supported */
	if (zs->zs_mirrors < 3 \|\| ztest_opts.zo_raid_children > 1) {
	mutex_exit(&ztest_vdev_lock);
	return;
	}

	/* clean up the old pool, if any */
	(void) spa_destroy("splitp");

	spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);

	/* generate a config from the existing config */
	mutex_enter(&spa->spa_props_lock);
	tree = fnvlist_lookup_nvlist(spa->spa_config, ZPOOL_CONFIG_VDEV_TREE);
	mutex_exit(&spa->spa_props_lock);

	VERIFY0(nvlist_lookup_nvlist_array(tree, ZPOOL_CONFIG_CHILDREN,
	&child, &children));

	schild = malloc(rvd->vdev_children * sizeof (nvlist_t *));
	for (c = 0; c < children; c++) {
	vdev_t *tvd = rvd->vdev_child[c];
	nvlist_t **mchild;
	uint_t mchildren;

	if (tvd->vdev_islog \|\| tvd->vdev_ops == &vdev_hole_ops) {
	schild[schildren] = fnvlist_alloc();
	fnvlist_add_string(schild[schildren],
	ZPOOL_CONFIG_TYPE, VDEV_TYPE_HOLE);
	fnvlist_add_uint64(schild[schildren],
	ZPOOL_CONFIG_IS_HOLE, 1);
	if (lastlogid == 0)
	lastlogid = schildren;
	++schildren;
	continue;
	}
	lastlogid = 0;
	VERIFY0(nvlist_lookup_nvlist_array(child[c],
	ZPOOL_CONFIG_CHILDREN, &mchild, &mchildren));
	schild[schildren++] = fnvlist_dup(mchild[0]);
	}

	/* OK, create a config that can be used to split */
	split = fnvlist_alloc();
	fnvlist_add_string(split, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT);
	- fnvlist_add_nvlist_array(split, ZPOOL_CONFIG_CHILDREN, schild,
	- lastlogid != 0 ? lastlogid : schildren);
	+ fnvlist_add_nvlist_array(split, ZPOOL_CONFIG_CHILDREN,
	+ (const nvlist_t **)schild, lastlogid != 0 ? lastlogid : schildren);

	config = fnvlist_alloc();
	fnvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, split);

	for (c = 0; c < schildren; c++)
	fnvlist_free(schild[c]);
	free(schild);
	fnvlist_free(split);

	spa_config_exit(spa, SCL_VDEV, FTAG);

	(void) pthread_rwlock_wrlock(&ztest_name_lock);
	error = spa_vdev_split_mirror(spa, "splitp", config, NULL, B_FALSE);
	(void) pthread_rwlock_unlock(&ztest_name_lock);

	fnvlist_free(config);

	if (error == 0) {
	(void) printf("successful split - results:\n");
	mutex_enter(&spa_namespace_lock);
	show_pool_stats(spa);
	show_pool_stats(spa_lookup("splitp"));
	mutex_exit(&spa_namespace_lock);
	++zs->zs_splits;
	--zs->zs_mirrors;
	}
	mutex_exit(&ztest_vdev_lock);
	}

	/*
	* Verify that we can attach and detach devices.
	*/
	/* ARGSUSED */
	void
	ztest_vdev_attach_detach(ztest_ds_t *zd, uint64_t id)
	{
	ztest_shared_t *zs = ztest_shared;
	spa_t *spa = ztest_spa;
	spa_aux_vdev_t *sav = &spa->spa_spares;
	vdev_t *rvd = spa->spa_root_vdev;
	vdev_t oldvd, newvd, *pvd;
	nvlist_t *root;
	uint64_t leaves;
	uint64_t leaf, top;
	uint64_t ashift = ztest_get_ashift();
	uint64_t oldguid, pguid;
	uint64_t oldsize, newsize;
	char oldpath, newpath;
	int replacing;
	int oldvd_has_siblings = B_FALSE;
	int newvd_is_spare = B_FALSE;
	int newvd_is_dspare = B_FALSE;
	int oldvd_is_log;
	int error, expected_error;

	if (ztest_opts.zo_mmp_test)
	return;

	oldpath = umem_alloc(MAXPATHLEN, UMEM_NOFAIL);
	newpath = umem_alloc(MAXPATHLEN, UMEM_NOFAIL);

	mutex_enter(&ztest_vdev_lock);
	leaves = MAX(zs->zs_mirrors, 1) * ztest_opts.zo_raid_children;

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);

	/*
	* If a vdev is in the process of being removed, its removal may
	* finish while we are in progress, leading to an unexpected error
	* value. Don't bother trying to attach while we are in the middle
	* of removal.
	*/
	if (ztest_device_removal_active) {
	spa_config_exit(spa, SCL_ALL, FTAG);
	goto out;
	}

	/*
	* Decide whether to do an attach or a replace.
	*/
	replacing = ztest_random(2);

	/*
	* Pick a random top-level vdev.
	*/
	top = ztest_random_vdev_top(spa, B_TRUE);

	/*
	* Pick a random leaf within it.
	*/
	leaf = ztest_random(leaves);

	/*
	* Locate this vdev.
	*/
	oldvd = rvd->vdev_child[top];

	/* pick a child from the mirror */
	if (zs->zs_mirrors >= 1) {
	ASSERT3P(oldvd->vdev_ops, ==, &vdev_mirror_ops);
	ASSERT3U(oldvd->vdev_children, >=, zs->zs_mirrors);
	oldvd = oldvd->vdev_child[leaf / ztest_opts.zo_raid_children];
	}

	/* pick a child out of the raidz group */
	if (ztest_opts.zo_raid_children > 1) {
	if (strcmp(oldvd->vdev_ops->vdev_op_type, "raidz") == 0)
	ASSERT3P(oldvd->vdev_ops, ==, &vdev_raidz_ops);
	else
	ASSERT3P(oldvd->vdev_ops, ==, &vdev_draid_ops);
	ASSERT3U(oldvd->vdev_children, ==, ztest_opts.zo_raid_children);
	oldvd = oldvd->vdev_child[leaf % ztest_opts.zo_raid_children];
	}

	/*
	* If we're already doing an attach or replace, oldvd may be a
	* mirror vdev -- in which case, pick a random child.
	*/
	while (oldvd->vdev_children != 0) {
	oldvd_has_siblings = B_TRUE;
	ASSERT3U(oldvd->vdev_children, >=, 2);
	oldvd = oldvd->vdev_child[ztest_random(oldvd->vdev_children)];
	}

	oldguid = oldvd->vdev_guid;
	oldsize = vdev_get_min_asize(oldvd);
	oldvd_is_log = oldvd->vdev_top->vdev_islog;
	(void) strcpy(oldpath, oldvd->vdev_path);
	pvd = oldvd->vdev_parent;
	pguid = pvd->vdev_guid;

	/*
	* If oldvd has siblings, then half of the time, detach it. Prior
	* to the detach the pool is scrubbed in order to prevent creating
	* unrepairable blocks as a result of the data corruption injection.
	*/
	if (oldvd_has_siblings && ztest_random(2) == 0) {
	spa_config_exit(spa, SCL_ALL, FTAG);

	error = ztest_scrub_impl(spa);
	if (error)
	goto out;

	error = spa_vdev_detach(spa, oldguid, pguid, B_FALSE);
	if (error != 0 && error != ENODEV && error != EBUSY &&
	error != ENOTSUP && error != ZFS_ERR_CHECKPOINT_EXISTS &&
	error != ZFS_ERR_DISCARDING_CHECKPOINT)
	fatal(B_FALSE, "detach (%s) returned %d",
	oldpath, error);
	goto out;
	}

	/*
	* For the new vdev, choose with equal probability between the two
	* standard paths (ending in either 'a' or 'b') or a random hot spare.
	*/
	if (sav->sav_count != 0 && ztest_random(3) == 0) {
	newvd = sav->sav_vdevs[ztest_random(sav->sav_count)];
	newvd_is_spare = B_TRUE;

	if (newvd->vdev_ops == &vdev_draid_spare_ops)
	newvd_is_dspare = B_TRUE;

	(void) strcpy(newpath, newvd->vdev_path);
	} else {
	(void) snprintf(newpath, MAXPATHLEN, ztest_dev_template,
	ztest_opts.zo_dir, ztest_opts.zo_pool,
	top * leaves + leaf);
	if (ztest_random(2) == 0)
	newpath[strlen(newpath) - 1] = 'b';
	newvd = vdev_lookup_by_path(rvd, newpath);
	}

	if (newvd) {
	/*
	* Reopen to ensure the vdev's asize field isn't stale.
	*/
	vdev_reopen(newvd);
	newsize = vdev_get_min_asize(newvd);
	} else {
	/*
	* Make newsize a little bigger or smaller than oldsize.
	* If it's smaller, the attach should fail.
	* If it's larger, and we're doing a replace,
	* we should get dynamic LUN growth when we're done.
	*/
	newsize = 10 * oldsize / (9 + ztest_random(3));
	}

	/*
	* If pvd is not a mirror or root, the attach should fail with ENOTSUP,
	* unless it's a replace; in that case any non-replacing parent is OK.
	*
	* If newvd is already part of the pool, it should fail with EBUSY.
	*
	* If newvd is too small, it should fail with EOVERFLOW.
	*
	* If newvd is a distributed spare and it's being attached to a
	* dRAID which is not its parent it should fail with EINVAL.
	*/
	if (pvd->vdev_ops != &vdev_mirror_ops &&
	pvd->vdev_ops != &vdev_root_ops && (!replacing \|\|
	pvd->vdev_ops == &vdev_replacing_ops \|\|
	pvd->vdev_ops == &vdev_spare_ops))
	expected_error = ENOTSUP;
	else if (newvd_is_spare && (!replacing \|\| oldvd_is_log))
	expected_error = ENOTSUP;
	else if (newvd == oldvd)
	expected_error = replacing ? 0 : EBUSY;
	else if (vdev_lookup_by_path(rvd, newpath) != NULL)
	expected_error = EBUSY;
	else if (!newvd_is_dspare && newsize < oldsize)
	expected_error = EOVERFLOW;
	else if (ashift > oldvd->vdev_top->vdev_ashift)
	expected_error = EDOM;
	else if (newvd_is_dspare && pvd != vdev_draid_spare_get_parent(newvd))
	expected_error = ENOTSUP;
	else
	expected_error = 0;

	spa_config_exit(spa, SCL_ALL, FTAG);

	/*
	* Build the nvlist describing newpath.
	*/
	root = make_vdev_root(newpath, NULL, NULL, newvd == NULL ? newsize : 0,
	ashift, NULL, 0, 0, 1);

	/*
	* When supported select either a healing or sequential resilver.
	*/
	boolean_t rebuilding = B_FALSE;
	if (pvd->vdev_ops == &vdev_mirror_ops \|\|
	pvd->vdev_ops == &vdev_root_ops) {
	rebuilding = !!ztest_random(2);
	}

	error = spa_vdev_attach(spa, oldguid, root, replacing, rebuilding);

	fnvlist_free(root);

	/*
	* If our parent was the replacing vdev, but the replace completed,
	* then instead of failing with ENOTSUP we may either succeed,
	* fail with ENODEV, or fail with EOVERFLOW.
	*/
	if (expected_error == ENOTSUP &&
	(error == 0 \|\| error == ENODEV \|\| error == EOVERFLOW))
	expected_error = error;

	/*
	* If someone grew the LUN, the replacement may be too small.
	*/
	if (error == EOVERFLOW \|\| error == EBUSY)
	expected_error = error;

	if (error == ZFS_ERR_CHECKPOINT_EXISTS \|\|
	error == ZFS_ERR_DISCARDING_CHECKPOINT \|\|
	error == ZFS_ERR_RESILVER_IN_PROGRESS \|\|
	error == ZFS_ERR_REBUILD_IN_PROGRESS)
	expected_error = error;

	if (error != expected_error && expected_error != EBUSY) {
	fatal(B_FALSE, "attach (%s %"PRIu64", %s %"PRIu64", %d) "
	"returned %d, expected %d",
	oldpath, oldsize, newpath,
	newsize, replacing, error, expected_error);
	}
	out:
	mutex_exit(&ztest_vdev_lock);

	umem_free(oldpath, MAXPATHLEN);
	umem_free(newpath, MAXPATHLEN);
	}

	/* ARGSUSED */
	void
	ztest_device_removal(ztest_ds_t *zd, uint64_t id)
	{
	spa_t *spa = ztest_spa;
	vdev_t *vd;
	uint64_t guid;
	int error;

	mutex_enter(&ztest_vdev_lock);

	if (ztest_device_removal_active) {
	mutex_exit(&ztest_vdev_lock);
	return;
	}

	/*
	* Remove a random top-level vdev and wait for removal to finish.
	*/
	spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);
	vd = vdev_lookup_top(spa, ztest_random_vdev_top(spa, B_FALSE));
	guid = vd->vdev_guid;
	spa_config_exit(spa, SCL_VDEV, FTAG);

	error = spa_vdev_remove(spa, guid, B_FALSE);
	if (error == 0) {
	ztest_device_removal_active = B_TRUE;
	mutex_exit(&ztest_vdev_lock);

	/*
	* spa->spa_vdev_removal is created in a sync task that
	* is initiated via dsl_sync_task_nowait(). Since the
	* task may not run before spa_vdev_remove() returns, we
	* must wait at least 1 txg to ensure that the removal
	* struct has been created.
	*/
	txg_wait_synced(spa_get_dsl(spa), 0);

	while (spa->spa_removing_phys.sr_state == DSS_SCANNING)
	txg_wait_synced(spa_get_dsl(spa), 0);
	} else {
	mutex_exit(&ztest_vdev_lock);
	return;
	}

	/*
	* The pool needs to be scrubbed after completing device removal.
	* Failure to do so may result in checksum errors due to the
	* strategy employed by ztest_fault_inject() when selecting which
	* offset are redundant and can be damaged.
	*/
	error = spa_scan(spa, POOL_SCAN_SCRUB);
	if (error == 0) {
	while (dsl_scan_scrubbing(spa_get_dsl(spa)))
	txg_wait_synced(spa_get_dsl(spa), 0);
	}

	mutex_enter(&ztest_vdev_lock);
	ztest_device_removal_active = B_FALSE;
	mutex_exit(&ztest_vdev_lock);
	}

	/*
	* Callback function which expands the physical size of the vdev.
	*/
	static vdev_t *
	grow_vdev(vdev_t vd, void arg)
	{
	spa_t *spa __maybe_unused = vd->vdev_spa;
	size_t *newsize = arg;
	size_t fsize;
	int fd;

	ASSERT3S(spa_config_held(spa, SCL_STATE, RW_READER), ==, SCL_STATE);
	ASSERT(vd->vdev_ops->vdev_op_leaf);

	if ((fd = open(vd->vdev_path, O_RDWR)) == -1)
	return (vd);

	fsize = lseek(fd, 0, SEEK_END);
	VERIFY0(ftruncate(fd, *newsize));

	if (ztest_opts.zo_verbose >= 6) {
	(void) printf("%s grew from %lu to %lu bytes\n",
	vd->vdev_path, (ulong_t)fsize, (ulong_t)*newsize);
	}
	(void) close(fd);
	return (NULL);
	}

	/*
	* Callback function which expands a given vdev by calling vdev_online().
	*/
	/* ARGSUSED */
	static vdev_t *
	online_vdev(vdev_t vd, void arg)
	{
	spa_t *spa = vd->vdev_spa;
	vdev_t *tvd = vd->vdev_top;
	uint64_t guid = vd->vdev_guid;
	uint64_t generation = spa->spa_config_generation + 1;
	vdev_state_t newstate = VDEV_STATE_UNKNOWN;
	int error;

	ASSERT3S(spa_config_held(spa, SCL_STATE, RW_READER), ==, SCL_STATE);
	ASSERT(vd->vdev_ops->vdev_op_leaf);

	/* Calling vdev_online will initialize the new metaslabs */
	spa_config_exit(spa, SCL_STATE, spa);
	error = vdev_online(spa, guid, ZFS_ONLINE_EXPAND, &newstate);
	spa_config_enter(spa, SCL_STATE, spa, RW_READER);

	/*
	* If vdev_online returned an error or the underlying vdev_open
	* failed then we abort the expand. The only way to know that
	* vdev_open fails is by checking the returned newstate.
	*/
	if (error \|\| newstate != VDEV_STATE_HEALTHY) {
	if (ztest_opts.zo_verbose >= 5) {
	(void) printf("Unable to expand vdev, state %u, "
	"error %d\n", newstate, error);
	}
	return (vd);
	}
	ASSERT3U(newstate, ==, VDEV_STATE_HEALTHY);

	/*
	* Since we dropped the lock we need to ensure that we're
	* still talking to the original vdev. It's possible this
	* vdev may have been detached/replaced while we were
	* trying to online it.
	*/
	if (generation != spa->spa_config_generation) {
	if (ztest_opts.zo_verbose >= 5) {
	(void) printf("vdev configuration has changed, "
	"guid %"PRIu64", state %"PRIu64", "
	"expected gen %"PRIu64", got gen %"PRIu64"\n",
	guid,
	tvd->vdev_state,
	generation,
	spa->spa_config_generation);
	}
	return (vd);
	}
	return (NULL);
	}

	/*
	* Traverse the vdev tree calling the supplied function.
	* We continue to walk the tree until we either have walked all
	* children or we receive a non-NULL return from the callback.
	* If a NULL callback is passed, then we just return back the first
	* leaf vdev we encounter.
	*/
	static vdev_t *
	vdev_walk_tree(vdev_t vd, vdev_t (func)(vdev_t , void ), void arg)
	{
	uint_t c;

	if (vd->vdev_ops->vdev_op_leaf) {
	if (func == NULL)
	return (vd);
	else
	return (func(vd, arg));
	}

	for (c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];
	if ((cvd = vdev_walk_tree(cvd, func, arg)) != NULL)
	return (cvd);
	}
	return (NULL);
	}

	/*
	* Verify that dynamic LUN growth works as expected.
	*/
	/* ARGSUSED */
	void
	ztest_vdev_LUN_growth(ztest_ds_t *zd, uint64_t id)
	{
	spa_t *spa = ztest_spa;
	vdev_t vd, tvd;
	metaslab_class_t *mc;
	metaslab_group_t *mg;
	size_t psize, newsize;
	uint64_t top;
	uint64_t old_class_space, new_class_space, old_ms_count, new_ms_count;

	mutex_enter(&ztest_checkpoint_lock);
	mutex_enter(&ztest_vdev_lock);
	spa_config_enter(spa, SCL_STATE, spa, RW_READER);

	/*
	* If there is a vdev removal in progress, it could complete while
	* we are running, in which case we would not be able to verify
	* that the metaslab_class space increased (because it decreases
	* when the device removal completes).
	*/
	if (ztest_device_removal_active) {
	spa_config_exit(spa, SCL_STATE, spa);
	mutex_exit(&ztest_vdev_lock);
	mutex_exit(&ztest_checkpoint_lock);
	return;
	}

	top = ztest_random_vdev_top(spa, B_TRUE);

	tvd = spa->spa_root_vdev->vdev_child[top];
	mg = tvd->vdev_mg;
	mc = mg->mg_class;
	old_ms_count = tvd->vdev_ms_count;
	old_class_space = metaslab_class_get_space(mc);

	/*
	* Determine the size of the first leaf vdev associated with
	* our top-level device.
	*/
	vd = vdev_walk_tree(tvd, NULL, NULL);
	ASSERT3P(vd, !=, NULL);
	ASSERT(vd->vdev_ops->vdev_op_leaf);

	psize = vd->vdev_psize;

	/*
	* We only try to expand the vdev if it's healthy, less than 4x its
	* original size, and it has a valid psize.
	*/
	if (tvd->vdev_state != VDEV_STATE_HEALTHY \|\|
	psize == 0 \|\| psize >= 4 * ztest_opts.zo_vdev_size) {
	spa_config_exit(spa, SCL_STATE, spa);
	mutex_exit(&ztest_vdev_lock);
	mutex_exit(&ztest_checkpoint_lock);
	return;
	}
	ASSERT3U(psize, >, 0);
	newsize = psize + MAX(psize / 8, SPA_MAXBLOCKSIZE);
	ASSERT3U(newsize, >, psize);

	if (ztest_opts.zo_verbose >= 6) {
	(void) printf("Expanding LUN %s from %lu to %lu\n",
	vd->vdev_path, (ulong_t)psize, (ulong_t)newsize);
	}

	/*
	* Growing the vdev is a two step process:
	* 1). expand the physical size (i.e. relabel)
	* 2). online the vdev to create the new metaslabs
	*/
	if (vdev_walk_tree(tvd, grow_vdev, &newsize) != NULL \|\|
	vdev_walk_tree(tvd, online_vdev, NULL) != NULL \|\|
	tvd->vdev_state != VDEV_STATE_HEALTHY) {
	if (ztest_opts.zo_verbose >= 5) {
	(void) printf("Could not expand LUN because "
	"the vdev configuration changed.\n");
	}
	spa_config_exit(spa, SCL_STATE, spa);
	mutex_exit(&ztest_vdev_lock);
	mutex_exit(&ztest_checkpoint_lock);
	return;
	}

	spa_config_exit(spa, SCL_STATE, spa);

	/*
	* Expanding the LUN will update the config asynchronously,
	* thus we must wait for the async thread to complete any
	* pending tasks before proceeding.
	*/
	for (;;) {
	boolean_t done;
	mutex_enter(&spa->spa_async_lock);
	done = (spa->spa_async_thread == NULL && !spa->spa_async_tasks);
	mutex_exit(&spa->spa_async_lock);
	if (done)
	break;
	txg_wait_synced(spa_get_dsl(spa), 0);
	(void) poll(NULL, 0, 100);
	}

	spa_config_enter(spa, SCL_STATE, spa, RW_READER);

	tvd = spa->spa_root_vdev->vdev_child[top];
	new_ms_count = tvd->vdev_ms_count;
	new_class_space = metaslab_class_get_space(mc);

	if (tvd->vdev_mg != mg \|\| mg->mg_class != mc) {
	if (ztest_opts.zo_verbose >= 5) {
	(void) printf("Could not verify LUN expansion due to "
	"intervening vdev offline or remove.\n");
	}
	spa_config_exit(spa, SCL_STATE, spa);
	mutex_exit(&ztest_vdev_lock);
	mutex_exit(&ztest_checkpoint_lock);
	return;
	}

	/*
	* Make sure we were able to grow the vdev.
	*/
	if (new_ms_count <= old_ms_count) {
	fatal(B_FALSE,
	"LUN expansion failed: ms_count %"PRIu64" < %"PRIu64"\n",
	old_ms_count, new_ms_count);
	}

	/*
	* Make sure we were able to grow the pool.
	*/
	if (new_class_space <= old_class_space) {
	fatal(B_FALSE,
	"LUN expansion failed: class_space %"PRIu64" < %"PRIu64"\n",
	old_class_space, new_class_space);
	}

	if (ztest_opts.zo_verbose >= 5) {
	char oldnumbuf[NN_NUMBUF_SZ], newnumbuf[NN_NUMBUF_SZ];

	nicenum(old_class_space, oldnumbuf, sizeof (oldnumbuf));
	nicenum(new_class_space, newnumbuf, sizeof (newnumbuf));
	(void) printf("%s grew from %s to %s\n",
	spa->spa_name, oldnumbuf, newnumbuf);
	}

	spa_config_exit(spa, SCL_STATE, spa);
	mutex_exit(&ztest_vdev_lock);
	mutex_exit(&ztest_checkpoint_lock);
	}

	/*
	* Verify that dmu_objset_{create,destroy,open,close} work as expected.
	*/
	/* ARGSUSED */
	static void
	ztest_objset_create_cb(objset_t os, void arg, cred_t cr, dmu_tx_t tx)
	{
	/*
	* Create the objects common to all ztest datasets.
	*/
	VERIFY0(zap_create_claim(os, ZTEST_DIROBJ,
	DMU_OT_ZAP_OTHER, DMU_OT_NONE, 0, tx));
	}

	static int
	ztest_dataset_create(char *dsname)
	{
	int err;
	uint64_t rand;
	dsl_crypto_params_t *dcp = NULL;

	/*
	* 50% of the time, we create encrypted datasets
	* using a random cipher suite and a hard-coded
	* wrapping key.
	*/
	rand = ztest_random(2);
	if (rand != 0) {
	nvlist_t *crypto_args = fnvlist_alloc();
	nvlist_t *props = fnvlist_alloc();

	/* slight bias towards the default cipher suite */
	rand = ztest_random(ZIO_CRYPT_FUNCTIONS);
	if (rand < ZIO_CRYPT_AES_128_CCM)
	rand = ZIO_CRYPT_ON;

	fnvlist_add_uint64(props,
	zfs_prop_to_name(ZFS_PROP_ENCRYPTION), rand);
	fnvlist_add_uint8_array(crypto_args, "wkeydata",
	(uint8_t *)ztest_wkeydata, WRAPPING_KEY_LEN);

	/*
	* These parameters aren't really used by the kernel. They
	* are simply stored so that userspace knows how to load
	* the wrapping key.
	*/
	fnvlist_add_uint64(props,
	zfs_prop_to_name(ZFS_PROP_KEYFORMAT), ZFS_KEYFORMAT_RAW);
	fnvlist_add_string(props,
	zfs_prop_to_name(ZFS_PROP_KEYLOCATION), "prompt");
	fnvlist_add_uint64(props,
	zfs_prop_to_name(ZFS_PROP_PBKDF2_SALT), 0ULL);
	fnvlist_add_uint64(props,
	zfs_prop_to_name(ZFS_PROP_PBKDF2_ITERS), 0ULL);

	VERIFY0(dsl_crypto_params_create_nvlist(DCP_CMD_NONE, props,
	crypto_args, &dcp));

	/*
	* Cycle through all available encryption implementations
	* to verify interoperability.
	*/
	VERIFY0(gcm_impl_set("cycle"));
	VERIFY0(aes_impl_set("cycle"));

	fnvlist_free(crypto_args);
	fnvlist_free(props);
	}

	err = dmu_objset_create(dsname, DMU_OST_OTHER, 0, dcp,
	ztest_objset_create_cb, NULL);
	dsl_crypto_params_free(dcp, !!err);

	rand = ztest_random(100);
	if (err \|\| rand < 80)
	return (err);

	if (ztest_opts.zo_verbose >= 5)
	(void) printf("Setting dataset %s to sync always\n", dsname);
	return (ztest_dsl_prop_set_uint64(dsname, ZFS_PROP_SYNC,
	ZFS_SYNC_ALWAYS, B_FALSE));
	}

	/* ARGSUSED */
	static int
	ztest_objset_destroy_cb(const char name, void arg)
	{
	objset_t *os;
	dmu_object_info_t doi;
	int error;

	/*
	* Verify that the dataset contains a directory object.
	*/
	VERIFY0(ztest_dmu_objset_own(name, DMU_OST_OTHER, B_TRUE,
	B_TRUE, FTAG, &os));
	error = dmu_object_info(os, ZTEST_DIROBJ, &doi);
	if (error != ENOENT) {
	/* We could have crashed in the middle of destroying it */
	ASSERT0(error);
	ASSERT3U(doi.doi_type, ==, DMU_OT_ZAP_OTHER);
	ASSERT3S(doi.doi_physical_blocks_512, >=, 0);
	}
	dmu_objset_disown(os, B_TRUE, FTAG);

	/*
	* Destroy the dataset.
	*/
	if (strchr(name, '@') != NULL) {
	VERIFY0(dsl_destroy_snapshot(name, B_TRUE));
	} else {
	error = dsl_destroy_head(name);
	if (error == ENOSPC) {
	/* There could be checkpoint or insufficient slop */
	ztest_record_enospc(FTAG);
	} else if (error != EBUSY) {
	/* There could be a hold on this dataset */
	ASSERT0(error);
	}
	}
	return (0);
	}

	static boolean_t
	ztest_snapshot_create(char *osname, uint64_t id)
	{
	char snapname[ZFS_MAX_DATASET_NAME_LEN];
	int error;

	(void) snprintf(snapname, sizeof (snapname), "%"PRIu64"", id);

	error = dmu_objset_snapshot_one(osname, snapname);
	if (error == ENOSPC) {
	ztest_record_enospc(FTAG);
	return (B_FALSE);
	}
	if (error != 0 && error != EEXIST) {
	fatal(B_FALSE, "ztest_snapshot_create(%s@%s) = %d", osname,
	snapname, error);
	}
	return (B_TRUE);
	}

	static boolean_t
	ztest_snapshot_destroy(char *osname, uint64_t id)
	{
	char snapname[ZFS_MAX_DATASET_NAME_LEN];
	int error;

	(void) snprintf(snapname, sizeof (snapname), "%s@%"PRIu64"",
	osname, id);

	error = dsl_destroy_snapshot(snapname, B_FALSE);
	if (error != 0 && error != ENOENT)
	fatal(B_FALSE, "ztest_snapshot_destroy(%s) = %d",
	snapname, error);
	return (B_TRUE);
	}

	/* ARGSUSED */
	void
	ztest_dmu_objset_create_destroy(ztest_ds_t *zd, uint64_t id)
	{
	ztest_ds_t *zdtmp;
	int iters;
	int error;
	objset_t os, os2;
	char name[ZFS_MAX_DATASET_NAME_LEN];
	zilog_t *zilog;
	int i;

	zdtmp = umem_alloc(sizeof (ztest_ds_t), UMEM_NOFAIL);

	(void) pthread_rwlock_rdlock(&ztest_name_lock);

	(void) snprintf(name, sizeof (name), "%s/temp_%"PRIu64"",
	ztest_opts.zo_pool, id);

	/*
	* If this dataset exists from a previous run, process its replay log
	* half of the time. If we don't replay it, then dsl_destroy_head()
	* (invoked from ztest_objset_destroy_cb()) should just throw it away.
	*/
	if (ztest_random(2) == 0 &&
	ztest_dmu_objset_own(name, DMU_OST_OTHER, B_FALSE,
	B_TRUE, FTAG, &os) == 0) {
	ztest_zd_init(zdtmp, NULL, os);
	zil_replay(os, zdtmp, ztest_replay_vector);
	ztest_zd_fini(zdtmp);
	dmu_objset_disown(os, B_TRUE, FTAG);
	}

	/*
	* There may be an old instance of the dataset we're about to
	* create lying around from a previous run. If so, destroy it
	* and all of its snapshots.
	*/
	(void) dmu_objset_find(name, ztest_objset_destroy_cb, NULL,
	DS_FIND_CHILDREN \| DS_FIND_SNAPSHOTS);

	/*
	* Verify that the destroyed dataset is no longer in the namespace.
	*/
	VERIFY3U(ENOENT, ==, ztest_dmu_objset_own(name, DMU_OST_OTHER, B_TRUE,
	B_TRUE, FTAG, &os));

	/*
	* Verify that we can create a new dataset.
	*/
	error = ztest_dataset_create(name);
	if (error) {
	if (error == ENOSPC) {
	ztest_record_enospc(FTAG);
	goto out;
	}
	fatal(B_FALSE, "dmu_objset_create(%s) = %d", name, error);
	}

	VERIFY0(ztest_dmu_objset_own(name, DMU_OST_OTHER, B_FALSE, B_TRUE,
	FTAG, &os));

	ztest_zd_init(zdtmp, NULL, os);

	/*
	* Open the intent log for it.
	*/
	zilog = zil_open(os, ztest_get_data);

	/*
	* Put some objects in there, do a little I/O to them,
	* and randomly take a couple of snapshots along the way.
	*/
	iters = ztest_random(5);
	for (i = 0; i < iters; i++) {
	ztest_dmu_object_alloc_free(zdtmp, id);
	if (ztest_random(iters) == 0)
	(void) ztest_snapshot_create(name, i);
	}

	/*
	* Verify that we cannot create an existing dataset.
	*/
	VERIFY3U(EEXIST, ==,
	dmu_objset_create(name, DMU_OST_OTHER, 0, NULL, NULL, NULL));

	/*
	* Verify that we can hold an objset that is also owned.
	*/
	VERIFY0(dmu_objset_hold(name, FTAG, &os2));
	dmu_objset_rele(os2, FTAG);

	/*
	* Verify that we cannot own an objset that is already owned.
	*/
	VERIFY3U(EBUSY, ==, ztest_dmu_objset_own(name, DMU_OST_OTHER,
	B_FALSE, B_TRUE, FTAG, &os2));

	zil_close(zilog);
	dmu_objset_disown(os, B_TRUE, FTAG);
	ztest_zd_fini(zdtmp);
	out:
	(void) pthread_rwlock_unlock(&ztest_name_lock);

	umem_free(zdtmp, sizeof (ztest_ds_t));
	}

	/*
	* Verify that dmu_snapshot_{create,destroy,open,close} work as expected.
	*/
	void
	ztest_dmu_snapshot_create_destroy(ztest_ds_t *zd, uint64_t id)
	{
	(void) pthread_rwlock_rdlock(&ztest_name_lock);
	(void) ztest_snapshot_destroy(zd->zd_name, id);
	(void) ztest_snapshot_create(zd->zd_name, id);
	(void) pthread_rwlock_unlock(&ztest_name_lock);
	}

	/*
	* Cleanup non-standard snapshots and clones.
	*/
	static void
	ztest_dsl_dataset_cleanup(char *osname, uint64_t id)
	{
	char *snap1name;
	char *clone1name;
	char *snap2name;
	char *clone2name;
	char *snap3name;
	int error;

	snap1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL);
	clone1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL);
	snap2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL);
	clone2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL);
	snap3name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL);

	(void) snprintf(snap1name, ZFS_MAX_DATASET_NAME_LEN, "%s@s1_%"PRIu64"",
	osname, id);
	(void) snprintf(clone1name, ZFS_MAX_DATASET_NAME_LEN, "%s/c1_%"PRIu64"",
	osname, id);
	(void) snprintf(snap2name, ZFS_MAX_DATASET_NAME_LEN, "%s@s2_%"PRIu64"",
	clone1name, id);
	(void) snprintf(clone2name, ZFS_MAX_DATASET_NAME_LEN, "%s/c2_%"PRIu64"",
	osname, id);
	(void) snprintf(snap3name, ZFS_MAX_DATASET_NAME_LEN, "%s@s3_%"PRIu64"",
	clone1name, id);

	error = dsl_destroy_head(clone2name);
	if (error && error != ENOENT)
	fatal(B_FALSE, "dsl_destroy_head(%s) = %d", clone2name, error);
	error = dsl_destroy_snapshot(snap3name, B_FALSE);
	if (error && error != ENOENT)
	fatal(B_FALSE, "dsl_destroy_snapshot(%s) = %d",
	snap3name, error);
	error = dsl_destroy_snapshot(snap2name, B_FALSE);
	if (error && error != ENOENT)
	fatal(B_FALSE, "dsl_destroy_snapshot(%s) = %d",
	snap2name, error);
	error = dsl_destroy_head(clone1name);
	if (error && error != ENOENT)
	fatal(B_FALSE, "dsl_destroy_head(%s) = %d", clone1name, error);
	error = dsl_destroy_snapshot(snap1name, B_FALSE);
	if (error && error != ENOENT)
	fatal(B_FALSE, "dsl_destroy_snapshot(%s) = %d",
	snap1name, error);

	umem_free(snap1name, ZFS_MAX_DATASET_NAME_LEN);
	umem_free(clone1name, ZFS_MAX_DATASET_NAME_LEN);
	umem_free(snap2name, ZFS_MAX_DATASET_NAME_LEN);
	umem_free(clone2name, ZFS_MAX_DATASET_NAME_LEN);
	umem_free(snap3name, ZFS_MAX_DATASET_NAME_LEN);
	}

	/*
	* Verify dsl_dataset_promote handles EBUSY
	*/
	void
	ztest_dsl_dataset_promote_busy(ztest_ds_t *zd, uint64_t id)
	{
	objset_t *os;
	char *snap1name;
	char *clone1name;
	char *snap2name;
	char *clone2name;
	char *snap3name;
	char *osname = zd->zd_name;
	int error;

	snap1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL);
	clone1name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL);
	snap2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL);
	clone2name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL);
	snap3name = umem_alloc(ZFS_MAX_DATASET_NAME_LEN, UMEM_NOFAIL);

	(void) pthread_rwlock_rdlock(&ztest_name_lock);

	ztest_dsl_dataset_cleanup(osname, id);

	(void) snprintf(snap1name, ZFS_MAX_DATASET_NAME_LEN, "%s@s1_%"PRIu64"",
	osname, id);
	(void) snprintf(clone1name, ZFS_MAX_DATASET_NAME_LEN, "%s/c1_%"PRIu64"",
	osname, id);
	(void) snprintf(snap2name, ZFS_MAX_DATASET_NAME_LEN, "%s@s2_%"PRIu64"",
	clone1name, id);
	(void) snprintf(clone2name, ZFS_MAX_DATASET_NAME_LEN, "%s/c2_%"PRIu64"",
	osname, id);
	(void) snprintf(snap3name, ZFS_MAX_DATASET_NAME_LEN, "%s@s3_%"PRIu64"",
	clone1name, id);

	error = dmu_objset_snapshot_one(osname, strchr(snap1name, '@') + 1);
	if (error && error != EEXIST) {
	if (error == ENOSPC) {
	ztest_record_enospc(FTAG);
	goto out;
	}
	fatal(B_FALSE, "dmu_take_snapshot(%s) = %d", snap1name, error);
	}

	error = dmu_objset_clone(clone1name, snap1name);
	if (error) {
	if (error == ENOSPC) {
	ztest_record_enospc(FTAG);
	goto out;
	}
	fatal(B_FALSE, "dmu_objset_create(%s) = %d", clone1name, error);
	}

	error = dmu_objset_snapshot_one(clone1name, strchr(snap2name, '@') + 1);
	if (error && error != EEXIST) {
	if (error == ENOSPC) {
	ztest_record_enospc(FTAG);
	goto out;
	}
	fatal(B_FALSE, "dmu_open_snapshot(%s) = %d", snap2name, error);
	}

	error = dmu_objset_snapshot_one(clone1name, strchr(snap3name, '@') + 1);
	if (error && error != EEXIST) {
	if (error == ENOSPC) {
	ztest_record_enospc(FTAG);
	goto out;
	}
	fatal(B_FALSE, "dmu_open_snapshot(%s) = %d", snap3name, error);
	}

	error = dmu_objset_clone(clone2name, snap3name);
	if (error) {
	if (error == ENOSPC) {
	ztest_record_enospc(FTAG);
	goto out;
	}
	fatal(B_FALSE, "dmu_objset_create(%s) = %d", clone2name, error);
	}

	error = ztest_dmu_objset_own(snap2name, DMU_OST_ANY, B_TRUE, B_TRUE,
	FTAG, &os);
	if (error)
	fatal(B_FALSE, "dmu_objset_own(%s) = %d", snap2name, error);
	error = dsl_dataset_promote(clone2name, NULL);
	if (error == ENOSPC) {
	dmu_objset_disown(os, B_TRUE, FTAG);
	ztest_record_enospc(FTAG);
	goto out;
	}
	if (error != EBUSY)
	fatal(B_FALSE, "dsl_dataset_promote(%s), %d, not EBUSY",
	clone2name, error);
	dmu_objset_disown(os, B_TRUE, FTAG);

	out:
	ztest_dsl_dataset_cleanup(osname, id);

	(void) pthread_rwlock_unlock(&ztest_name_lock);

	umem_free(snap1name, ZFS_MAX_DATASET_NAME_LEN);
	umem_free(clone1name, ZFS_MAX_DATASET_NAME_LEN);
	umem_free(snap2name, ZFS_MAX_DATASET_NAME_LEN);
	umem_free(clone2name, ZFS_MAX_DATASET_NAME_LEN);
	umem_free(snap3name, ZFS_MAX_DATASET_NAME_LEN);
	}

	#undef OD_ARRAY_SIZE
	#define OD_ARRAY_SIZE 4

	/*
	* Verify that dmu_object_{alloc,free} work as expected.
	*/
	void
	ztest_dmu_object_alloc_free(ztest_ds_t *zd, uint64_t id)
	{
	ztest_od_t *od;
	int batchsize;
	int size;
	int b;

	size = sizeof (ztest_od_t) * OD_ARRAY_SIZE;
	od = umem_alloc(size, UMEM_NOFAIL);
	batchsize = OD_ARRAY_SIZE;

	for (b = 0; b < batchsize; b++)
	ztest_od_init(od + b, id, FTAG, b, DMU_OT_UINT64_OTHER,
	0, 0, 0);

	/*
	* Destroy the previous batch of objects, create a new batch,
	* and do some I/O on the new objects.
	*/
	if (ztest_object_init(zd, od, size, B_TRUE) != 0)
	return;

	while (ztest_random(4 * batchsize) != 0)
	ztest_io(zd, od[ztest_random(batchsize)].od_object,
	ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT);

	umem_free(od, size);
	}

	/*
	* Rewind the global allocator to verify object allocation backfilling.
	*/
	void
	ztest_dmu_object_next_chunk(ztest_ds_t *zd, uint64_t id)
	{
	objset_t *os = zd->zd_os;
	int dnodes_per_chunk = 1 << dmu_object_alloc_chunk_shift;
	uint64_t object;

	/*
	* Rewind the global allocator randomly back to a lower object number
	* to force backfilling and reclamation of recently freed dnodes.
	*/
	mutex_enter(&os->os_obj_lock);
	object = ztest_random(os->os_obj_next_chunk);
	os->os_obj_next_chunk = P2ALIGN(object, dnodes_per_chunk);
	mutex_exit(&os->os_obj_lock);
	}

	#undef OD_ARRAY_SIZE
	#define OD_ARRAY_SIZE 2

	/*
	* Verify that dmu_{read,write} work as expected.
	*/
	void
	ztest_dmu_read_write(ztest_ds_t *zd, uint64_t id)
	{
	int size;
	ztest_od_t *od;

	objset_t *os = zd->zd_os;
	size = sizeof (ztest_od_t) * OD_ARRAY_SIZE;
	od = umem_alloc(size, UMEM_NOFAIL);
	dmu_tx_t *tx;
	int freeit, error;
	uint64_t i, n, s, txg;
	bufwad_t packbuf, bigbuf, pack, bigH, *bigT;
	uint64_t packobj, packoff, packsize, bigobj, bigoff, bigsize;
	uint64_t chunksize = (1000 + ztest_random(1000)) * sizeof (uint64_t);
	uint64_t regions = 997;
	uint64_t stride = 123456789ULL;
	uint64_t width = 40;
	int free_percent = 5;

	/*
	* This test uses two objects, packobj and bigobj, that are always
	* updated together (i.e. in the same tx) so that their contents are
	* in sync and can be compared. Their contents relate to each other
	* in a simple way: packobj is a dense array of 'bufwad' structures,
	* while bigobj is a sparse array of the same bufwads. Specifically,
	* for any index n, there are three bufwads that should be identical:
	*
	* packobj, at offset n * sizeof (bufwad_t)
	* bigobj, at the head of the nth chunk
	* bigobj, at the tail of the nth chunk
	*
	* The chunk size is arbitrary. It doesn't have to be a power of two,
	* and it doesn't have any relation to the object blocksize.
	* The only requirement is that it can hold at least two bufwads.
	*
	* Normally, we write the bufwad to each of these locations.
	* However, free_percent of the time we instead write zeroes to
	* packobj and perform a dmu_free_range() on bigobj. By comparing
	* bigobj to packobj, we can verify that the DMU is correctly
	* tracking which parts of an object are allocated and free,
	* and that the contents of the allocated blocks are correct.
	*/

	/*
	* Read the directory info. If it's the first time, set things up.
	*/
	ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, chunksize);
	ztest_od_init(od + 1, id, FTAG, 1, DMU_OT_UINT64_OTHER, 0, 0,
	chunksize);

	if (ztest_object_init(zd, od, size, B_FALSE) != 0) {
	umem_free(od, size);
	return;
	}

	bigobj = od[0].od_object;
	packobj = od[1].od_object;
	chunksize = od[0].od_gen;
	ASSERT3U(chunksize, ==, od[1].od_gen);

	/*
	* Prefetch a random chunk of the big object.
	* Our aim here is to get some async reads in flight
	* for blocks that we may free below; the DMU should
	* handle this race correctly.
	*/
	n = ztest_random(regions) * stride + ztest_random(width);
	s = 1 + ztest_random(2 * width - 1);
	dmu_prefetch(os, bigobj, 0, n * chunksize, s * chunksize,
	ZIO_PRIORITY_SYNC_READ);

	/*
	* Pick a random index and compute the offsets into packobj and bigobj.
	*/
	n = ztest_random(regions) * stride + ztest_random(width);
	s = 1 + ztest_random(width - 1);

	packoff = n * sizeof (bufwad_t);
	packsize = s * sizeof (bufwad_t);

	bigoff = n * chunksize;
	bigsize = s * chunksize;

	packbuf = umem_alloc(packsize, UMEM_NOFAIL);
	bigbuf = umem_alloc(bigsize, UMEM_NOFAIL);

	/*
	* free_percent of the time, free a range of bigobj rather than
	* overwriting it.
	*/
	freeit = (ztest_random(100) < free_percent);

	/*
	* Read the current contents of our objects.
	*/
	error = dmu_read(os, packobj, packoff, packsize, packbuf,
	DMU_READ_PREFETCH);
	ASSERT0(error);
	error = dmu_read(os, bigobj, bigoff, bigsize, bigbuf,
	DMU_READ_PREFETCH);
	ASSERT0(error);

	/*
	* Get a tx for the mods to both packobj and bigobj.
	*/
	tx = dmu_tx_create(os);

	dmu_tx_hold_write(tx, packobj, packoff, packsize);

	if (freeit)
	dmu_tx_hold_free(tx, bigobj, bigoff, bigsize);
	else
	dmu_tx_hold_write(tx, bigobj, bigoff, bigsize);

	/* This accounts for setting the checksum/compression. */
	dmu_tx_hold_bonus(tx, bigobj);

	txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG);
	if (txg == 0) {
	umem_free(packbuf, packsize);
	umem_free(bigbuf, bigsize);
	umem_free(od, size);
	return;
	}

	enum zio_checksum cksum;
	do {
	cksum = (enum zio_checksum)
	ztest_random_dsl_prop(ZFS_PROP_CHECKSUM);
	} while (cksum >= ZIO_CHECKSUM_LEGACY_FUNCTIONS);
	dmu_object_set_checksum(os, bigobj, cksum, tx);

	enum zio_compress comp;
	do {
	comp = (enum zio_compress)
	ztest_random_dsl_prop(ZFS_PROP_COMPRESSION);
	} while (comp >= ZIO_COMPRESS_LEGACY_FUNCTIONS);
	dmu_object_set_compress(os, bigobj, comp, tx);

	/*
	* For each index from n to n + s, verify that the existing bufwad
	* in packobj matches the bufwads at the head and tail of the
	* corresponding chunk in bigobj. Then update all three bufwads
	* with the new values we want to write out.
	*/
	for (i = 0; i < s; i++) {
	/* LINTED */
	pack = (bufwad_t )((char )packbuf + i * sizeof (bufwad_t));
	/* LINTED */
	bigH = (bufwad_t )((char )bigbuf + i * chunksize);
	/* LINTED */
	bigT = (bufwad_t )((char )bigH + chunksize) - 1;

	ASSERT3U((uintptr_t)bigH - (uintptr_t)bigbuf, <, bigsize);
	ASSERT3U((uintptr_t)bigT - (uintptr_t)bigbuf, <, bigsize);

	if (pack->bw_txg > txg)
	fatal(B_FALSE,
	"future leak: got %"PRIx64", open txg is %"PRIx64"",
	pack->bw_txg, txg);

	if (pack->bw_data != 0 && pack->bw_index != n + i)
	fatal(B_FALSE, "wrong index: "
	"got %"PRIx64", wanted %"PRIx64"+%"PRIx64"",
	pack->bw_index, n, i);

	if (bcmp(pack, bigH, sizeof (bufwad_t)) != 0)
	fatal(B_FALSE, "pack/bigH mismatch in %p/%p",
	pack, bigH);

	if (bcmp(pack, bigT, sizeof (bufwad_t)) != 0)
	fatal(B_FALSE, "pack/bigT mismatch in %p/%p",
	pack, bigT);

	if (freeit) {
	bzero(pack, sizeof (bufwad_t));
	} else {
	pack->bw_index = n + i;
	pack->bw_txg = txg;
	pack->bw_data = 1 + ztest_random(-2ULL);
	}
	bigH = pack;
	bigT = pack;
	}

	/*
	* We've verified all the old bufwads, and made new ones.
	* Now write them out.
	*/
	dmu_write(os, packobj, packoff, packsize, packbuf, tx);

	if (freeit) {
	if (ztest_opts.zo_verbose >= 7) {
	(void) printf("freeing offset %"PRIx64" size %"PRIx64""
	" txg %"PRIx64"\n",
	bigoff, bigsize, txg);
	}
	VERIFY0(dmu_free_range(os, bigobj, bigoff, bigsize, tx));
	} else {
	if (ztest_opts.zo_verbose >= 7) {
	(void) printf("writing offset %"PRIx64" size %"PRIx64""
	" txg %"PRIx64"\n",
	bigoff, bigsize, txg);
	}
	dmu_write(os, bigobj, bigoff, bigsize, bigbuf, tx);
	}

	dmu_tx_commit(tx);

	/*
	* Sanity check the stuff we just wrote.
	*/
	{
	void *packcheck = umem_alloc(packsize, UMEM_NOFAIL);
	void *bigcheck = umem_alloc(bigsize, UMEM_NOFAIL);

	VERIFY0(dmu_read(os, packobj, packoff,
	packsize, packcheck, DMU_READ_PREFETCH));
	VERIFY0(dmu_read(os, bigobj, bigoff,
	bigsize, bigcheck, DMU_READ_PREFETCH));

	ASSERT0(bcmp(packbuf, packcheck, packsize));
	ASSERT0(bcmp(bigbuf, bigcheck, bigsize));

	umem_free(packcheck, packsize);
	umem_free(bigcheck, bigsize);
	}

	umem_free(packbuf, packsize);
	umem_free(bigbuf, bigsize);
	umem_free(od, size);
	}

	static void
	compare_and_update_pbbufs(uint64_t s, bufwad_t packbuf, bufwad_t bigbuf,
	uint64_t bigsize, uint64_t n, uint64_t chunksize, uint64_t txg)
	{
	uint64_t i;
	bufwad_t *pack;
	bufwad_t *bigH;
	bufwad_t *bigT;

	/*
	* For each index from n to n + s, verify that the existing bufwad
	* in packobj matches the bufwads at the head and tail of the
	* corresponding chunk in bigobj. Then update all three bufwads
	* with the new values we want to write out.
	*/
	for (i = 0; i < s; i++) {
	/* LINTED */
	pack = (bufwad_t )((char )packbuf + i * sizeof (bufwad_t));
	/* LINTED */
	bigH = (bufwad_t )((char )bigbuf + i * chunksize);
	/* LINTED */
	bigT = (bufwad_t )((char )bigH + chunksize) - 1;

	ASSERT3U((uintptr_t)bigH - (uintptr_t)bigbuf, <, bigsize);
	ASSERT3U((uintptr_t)bigT - (uintptr_t)bigbuf, <, bigsize);

	if (pack->bw_txg > txg)
	fatal(B_FALSE,
	"future leak: got %"PRIx64", open txg is %"PRIx64"",
	pack->bw_txg, txg);

	if (pack->bw_data != 0 && pack->bw_index != n + i)
	fatal(B_FALSE, "wrong index: "
	"got %"PRIx64", wanted %"PRIx64"+%"PRIx64"",
	pack->bw_index, n, i);

	if (bcmp(pack, bigH, sizeof (bufwad_t)) != 0)
	fatal(B_FALSE, "pack/bigH mismatch in %p/%p",
	pack, bigH);

	if (bcmp(pack, bigT, sizeof (bufwad_t)) != 0)
	fatal(B_FALSE, "pack/bigT mismatch in %p/%p",
	pack, bigT);

	pack->bw_index = n + i;
	pack->bw_txg = txg;
	pack->bw_data = 1 + ztest_random(-2ULL);

	bigH = pack;
	bigT = pack;
	}
	}

	#undef OD_ARRAY_SIZE
	#define OD_ARRAY_SIZE 2

	void
	ztest_dmu_read_write_zcopy(ztest_ds_t *zd, uint64_t id)
	{
	objset_t *os = zd->zd_os;
	ztest_od_t *od;
	dmu_tx_t *tx;
	uint64_t i;
	int error;
	int size;
	uint64_t n, s, txg;
	bufwad_t packbuf, bigbuf;
	uint64_t packobj, packoff, packsize, bigobj, bigoff, bigsize;
	uint64_t blocksize = ztest_random_blocksize();
	uint64_t chunksize = blocksize;
	uint64_t regions = 997;
	uint64_t stride = 123456789ULL;
	uint64_t width = 9;
	dmu_buf_t *bonus_db;
	arc_buf_t **bigbuf_arcbufs;
	dmu_object_info_t doi;

	size = sizeof (ztest_od_t) * OD_ARRAY_SIZE;
	od = umem_alloc(size, UMEM_NOFAIL);

	/*
	* This test uses two objects, packobj and bigobj, that are always
	* updated together (i.e. in the same tx) so that their contents are
	* in sync and can be compared. Their contents relate to each other
	* in a simple way: packobj is a dense array of 'bufwad' structures,
	* while bigobj is a sparse array of the same bufwads. Specifically,
	* for any index n, there are three bufwads that should be identical:
	*
	* packobj, at offset n * sizeof (bufwad_t)
	* bigobj, at the head of the nth chunk
	* bigobj, at the tail of the nth chunk
	*
	* The chunk size is set equal to bigobj block size so that
	* dmu_assign_arcbuf_by_dbuf() can be tested for object updates.
	*/

	/*
	* Read the directory info. If it's the first time, set things up.
	*/
	ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, blocksize, 0, 0);
	ztest_od_init(od + 1, id, FTAG, 1, DMU_OT_UINT64_OTHER, 0, 0,
	chunksize);


	if (ztest_object_init(zd, od, size, B_FALSE) != 0) {
	umem_free(od, size);
	return;
	}

	bigobj = od[0].od_object;
	packobj = od[1].od_object;
	blocksize = od[0].od_blocksize;
	chunksize = blocksize;
	ASSERT3U(chunksize, ==, od[1].od_gen);

	VERIFY0(dmu_object_info(os, bigobj, &doi));
	VERIFY(ISP2(doi.doi_data_block_size));
	VERIFY3U(chunksize, ==, doi.doi_data_block_size);
	VERIFY3U(chunksize, >=, 2 * sizeof (bufwad_t));

	/*
	* Pick a random index and compute the offsets into packobj and bigobj.
	*/
	n = ztest_random(regions) * stride + ztest_random(width);
	s = 1 + ztest_random(width - 1);

	packoff = n * sizeof (bufwad_t);
	packsize = s * sizeof (bufwad_t);

	bigoff = n * chunksize;
	bigsize = s * chunksize;

	packbuf = umem_zalloc(packsize, UMEM_NOFAIL);
	bigbuf = umem_zalloc(bigsize, UMEM_NOFAIL);

	VERIFY0(dmu_bonus_hold(os, bigobj, FTAG, &bonus_db));

	bigbuf_arcbufs = umem_zalloc(2 * s * sizeof (arc_buf_t *), UMEM_NOFAIL);

	/*
	* Iteration 0 test zcopy for DB_UNCACHED dbufs.
	* Iteration 1 test zcopy to already referenced dbufs.
	* Iteration 2 test zcopy to dirty dbuf in the same txg.
	* Iteration 3 test zcopy to dbuf dirty in previous txg.
	* Iteration 4 test zcopy when dbuf is no longer dirty.
	* Iteration 5 test zcopy when it can't be done.
	* Iteration 6 one more zcopy write.
	*/
	for (i = 0; i < 7; i++) {
	uint64_t j;
	uint64_t off;

	/*
	* In iteration 5 (i == 5) use arcbufs
	* that don't match bigobj blksz to test
	* dmu_assign_arcbuf_by_dbuf() when it can't directly
	* assign an arcbuf to a dbuf.
	*/
	for (j = 0; j < s; j++) {
	if (i != 5 \|\| chunksize < (SPA_MINBLOCKSIZE * 2)) {
	bigbuf_arcbufs[j] =
	dmu_request_arcbuf(bonus_db, chunksize);
	} else {
	bigbuf_arcbufs[2 * j] =
	dmu_request_arcbuf(bonus_db, chunksize / 2);
	bigbuf_arcbufs[2 * j + 1] =
	dmu_request_arcbuf(bonus_db, chunksize / 2);
	}
	}

	/*
	* Get a tx for the mods to both packobj and bigobj.
	*/
	tx = dmu_tx_create(os);

	dmu_tx_hold_write(tx, packobj, packoff, packsize);
	dmu_tx_hold_write(tx, bigobj, bigoff, bigsize);

	txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG);
	if (txg == 0) {
	umem_free(packbuf, packsize);
	umem_free(bigbuf, bigsize);
	for (j = 0; j < s; j++) {
	if (i != 5 \|\|
	chunksize < (SPA_MINBLOCKSIZE * 2)) {
	dmu_return_arcbuf(bigbuf_arcbufs[j]);
	} else {
	dmu_return_arcbuf(
	bigbuf_arcbufs[2 * j]);
	dmu_return_arcbuf(
	bigbuf_arcbufs[2 * j + 1]);
	}
	}
	umem_free(bigbuf_arcbufs, 2 * s * sizeof (arc_buf_t *));
	umem_free(od, size);
	dmu_buf_rele(bonus_db, FTAG);
	return;
	}

	/*
	* 50% of the time don't read objects in the 1st iteration to
	* test dmu_assign_arcbuf_by_dbuf() for the case when there are
	* no existing dbufs for the specified offsets.
	*/
	if (i != 0 \|\| ztest_random(2) != 0) {
	error = dmu_read(os, packobj, packoff,
	packsize, packbuf, DMU_READ_PREFETCH);
	ASSERT0(error);
	error = dmu_read(os, bigobj, bigoff, bigsize,
	bigbuf, DMU_READ_PREFETCH);
	ASSERT0(error);
	}
	compare_and_update_pbbufs(s, packbuf, bigbuf, bigsize,
	n, chunksize, txg);

	/*
	* We've verified all the old bufwads, and made new ones.
	* Now write them out.
	*/
	dmu_write(os, packobj, packoff, packsize, packbuf, tx);
	if (ztest_opts.zo_verbose >= 7) {
	(void) printf("writing offset %"PRIx64" size %"PRIx64""
	" txg %"PRIx64"\n",
	bigoff, bigsize, txg);
	}
	for (off = bigoff, j = 0; j < s; j++, off += chunksize) {
	dmu_buf_t *dbt;
	if (i != 5 \|\| chunksize < (SPA_MINBLOCKSIZE * 2)) {
	bcopy((caddr_t)bigbuf + (off - bigoff),
	bigbuf_arcbufs[j]->b_data, chunksize);
	} else {
	bcopy((caddr_t)bigbuf + (off - bigoff),
	bigbuf_arcbufs[2 * j]->b_data,
	chunksize / 2);
	bcopy((caddr_t)bigbuf + (off - bigoff) +
	chunksize / 2,
	bigbuf_arcbufs[2 * j + 1]->b_data,
	chunksize / 2);
	}

	if (i == 1) {
	VERIFY(dmu_buf_hold(os, bigobj, off,
	FTAG, &dbt, DMU_READ_NO_PREFETCH) == 0);
	}
	if (i != 5 \|\| chunksize < (SPA_MINBLOCKSIZE * 2)) {
	VERIFY0(dmu_assign_arcbuf_by_dbuf(bonus_db,
	off, bigbuf_arcbufs[j], tx));
	} else {
	VERIFY0(dmu_assign_arcbuf_by_dbuf(bonus_db,
	off, bigbuf_arcbufs[2 * j], tx));
	VERIFY0(dmu_assign_arcbuf_by_dbuf(bonus_db,
	off + chunksize / 2,
	bigbuf_arcbufs[2 * j + 1], tx));
	}
	if (i == 1) {
	dmu_buf_rele(dbt, FTAG);
	}
	}
	dmu_tx_commit(tx);

	/*
	* Sanity check the stuff we just wrote.
	*/
	{
	void *packcheck = umem_alloc(packsize, UMEM_NOFAIL);
	void *bigcheck = umem_alloc(bigsize, UMEM_NOFAIL);

	VERIFY0(dmu_read(os, packobj, packoff,
	packsize, packcheck, DMU_READ_PREFETCH));
	VERIFY0(dmu_read(os, bigobj, bigoff,
	bigsize, bigcheck, DMU_READ_PREFETCH));

	ASSERT0(bcmp(packbuf, packcheck, packsize));
	ASSERT0(bcmp(bigbuf, bigcheck, bigsize));

	umem_free(packcheck, packsize);
	umem_free(bigcheck, bigsize);
	}
	if (i == 2) {
	txg_wait_open(dmu_objset_pool(os), 0, B_TRUE);
	} else if (i == 3) {
	txg_wait_synced(dmu_objset_pool(os), 0);
	}
	}

	dmu_buf_rele(bonus_db, FTAG);
	umem_free(packbuf, packsize);
	umem_free(bigbuf, bigsize);
	umem_free(bigbuf_arcbufs, 2 * s * sizeof (arc_buf_t *));
	umem_free(od, size);
	}

	/* ARGSUSED */
	void
	ztest_dmu_write_parallel(ztest_ds_t *zd, uint64_t id)
	{
	ztest_od_t *od;

	od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL);
	uint64_t offset = (1ULL << (ztest_random(20) + 43)) +
	(ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT);

	/*
	* Have multiple threads write to large offsets in an object
	* to verify that parallel writes to an object -- even to the
	* same blocks within the object -- doesn't cause any trouble.
	*/
	ztest_od_init(od, ID_PARALLEL, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, 0);

	if (ztest_object_init(zd, od, sizeof (ztest_od_t), B_FALSE) != 0)
	return;

	while (ztest_random(10) != 0)
	ztest_io(zd, od->od_object, offset);

	umem_free(od, sizeof (ztest_od_t));
	}

	void
	ztest_dmu_prealloc(ztest_ds_t *zd, uint64_t id)
	{
	ztest_od_t *od;
	uint64_t offset = (1ULL << (ztest_random(4) + SPA_MAXBLOCKSHIFT)) +
	(ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT);
	uint64_t count = ztest_random(20) + 1;
	uint64_t blocksize = ztest_random_blocksize();
	void *data;

	od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL);

	ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, blocksize, 0, 0);

	if (ztest_object_init(zd, od, sizeof (ztest_od_t),
	!ztest_random(2)) != 0) {
	umem_free(od, sizeof (ztest_od_t));
	return;
	}

	if (ztest_truncate(zd, od->od_object, offset, count * blocksize) != 0) {
	umem_free(od, sizeof (ztest_od_t));
	return;
	}

	ztest_prealloc(zd, od->od_object, offset, count * blocksize);

	data = umem_zalloc(blocksize, UMEM_NOFAIL);

	while (ztest_random(count) != 0) {
	uint64_t randoff = offset + (ztest_random(count) * blocksize);
	if (ztest_write(zd, od->od_object, randoff, blocksize,
	data) != 0)
	break;
	while (ztest_random(4) != 0)
	ztest_io(zd, od->od_object, randoff);
	}

	umem_free(data, blocksize);
	umem_free(od, sizeof (ztest_od_t));
	}

	/*
	* Verify that zap_{create,destroy,add,remove,update} work as expected.
	*/
	#define ZTEST_ZAP_MIN_INTS 1
	#define ZTEST_ZAP_MAX_INTS 4
	#define ZTEST_ZAP_MAX_PROPS 1000

	void
	ztest_zap(ztest_ds_t *zd, uint64_t id)
	{
	objset_t *os = zd->zd_os;
	ztest_od_t *od;
	uint64_t object;
	uint64_t txg, last_txg;
	uint64_t value[ZTEST_ZAP_MAX_INTS];
	uint64_t zl_ints, zl_intsize, prop;
	int i, ints;
	dmu_tx_t *tx;
	char propname[100], txgname[100];
	int error;
	char *hc[2] = { "s.acl.h", ".s.open.h.hyLZlg" };

	od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL);
	ztest_od_init(od, id, FTAG, 0, DMU_OT_ZAP_OTHER, 0, 0, 0);

	if (ztest_object_init(zd, od, sizeof (ztest_od_t),
	!ztest_random(2)) != 0)
	goto out;

	object = od->od_object;

	/*
	* Generate a known hash collision, and verify that
	* we can lookup and remove both entries.
	*/
	tx = dmu_tx_create(os);
	dmu_tx_hold_zap(tx, object, B_TRUE, NULL);
	txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG);
	if (txg == 0)
	goto out;
	for (i = 0; i < 2; i++) {
	value[i] = i;
	VERIFY0(zap_add(os, object, hc[i], sizeof (uint64_t),
	1, &value[i], tx));
	}
	for (i = 0; i < 2; i++) {
	VERIFY3U(EEXIST, ==, zap_add(os, object, hc[i],
	sizeof (uint64_t), 1, &value[i], tx));
	VERIFY0(
	zap_length(os, object, hc[i], &zl_intsize, &zl_ints));
	ASSERT3U(zl_intsize, ==, sizeof (uint64_t));
	ASSERT3U(zl_ints, ==, 1);
	}
	for (i = 0; i < 2; i++) {
	VERIFY0(zap_remove(os, object, hc[i], tx));
	}
	dmu_tx_commit(tx);

	/*
	* Generate a bunch of random entries.
	*/
	ints = MAX(ZTEST_ZAP_MIN_INTS, object % ZTEST_ZAP_MAX_INTS);

	prop = ztest_random(ZTEST_ZAP_MAX_PROPS);
	(void) sprintf(propname, "prop_%"PRIu64"", prop);
	(void) sprintf(txgname, "txg_%"PRIu64"", prop);
	bzero(value, sizeof (value));
	last_txg = 0;

	/*
	* If these zap entries already exist, validate their contents.
	*/
	error = zap_length(os, object, txgname, &zl_intsize, &zl_ints);
	if (error == 0) {
	ASSERT3U(zl_intsize, ==, sizeof (uint64_t));
	ASSERT3U(zl_ints, ==, 1);

	VERIFY0(zap_lookup(os, object, txgname, zl_intsize,
	zl_ints, &last_txg));

	VERIFY0(zap_length(os, object, propname, &zl_intsize,
	&zl_ints));

	ASSERT3U(zl_intsize, ==, sizeof (uint64_t));
	ASSERT3U(zl_ints, ==, ints);

	VERIFY0(zap_lookup(os, object, propname, zl_intsize,
	zl_ints, value));

	for (i = 0; i < ints; i++) {
	ASSERT3U(value[i], ==, last_txg + object + i);
	}
	} else {
	ASSERT3U(error, ==, ENOENT);
	}

	/*
	* Atomically update two entries in our zap object.
	* The first is named txg_%llu, and contains the txg
	* in which the property was last updated. The second
	* is named prop_%llu, and the nth element of its value
	* should be txg + object + n.
	*/
	tx = dmu_tx_create(os);
	dmu_tx_hold_zap(tx, object, B_TRUE, NULL);
	txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG);
	if (txg == 0)
	goto out;

	if (last_txg > txg)
	fatal(B_FALSE, "zap future leak: old %"PRIu64" new %"PRIu64"",
	last_txg, txg);

	for (i = 0; i < ints; i++)
	value[i] = txg + object + i;

	VERIFY0(zap_update(os, object, txgname, sizeof (uint64_t),
	1, &txg, tx));
	VERIFY0(zap_update(os, object, propname, sizeof (uint64_t),
	ints, value, tx));

	dmu_tx_commit(tx);

	/*
	* Remove a random pair of entries.
	*/
	prop = ztest_random(ZTEST_ZAP_MAX_PROPS);
	(void) sprintf(propname, "prop_%"PRIu64"", prop);
	(void) sprintf(txgname, "txg_%"PRIu64"", prop);

	error = zap_length(os, object, txgname, &zl_intsize, &zl_ints);

	if (error == ENOENT)
	goto out;

	ASSERT0(error);

	tx = dmu_tx_create(os);
	dmu_tx_hold_zap(tx, object, B_TRUE, NULL);
	txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG);
	if (txg == 0)
	goto out;
	VERIFY0(zap_remove(os, object, txgname, tx));
	VERIFY0(zap_remove(os, object, propname, tx));
	dmu_tx_commit(tx);
	out:
	umem_free(od, sizeof (ztest_od_t));
	}

	/*
	* Test case to test the upgrading of a microzap to fatzap.
	*/
	void
	ztest_fzap(ztest_ds_t *zd, uint64_t id)
	{
	objset_t *os = zd->zd_os;
	ztest_od_t *od;
	uint64_t object, txg, value;

	od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL);
	ztest_od_init(od, id, FTAG, 0, DMU_OT_ZAP_OTHER, 0, 0, 0);

	if (ztest_object_init(zd, od, sizeof (ztest_od_t),
	!ztest_random(2)) != 0)
	goto out;
	object = od->od_object;

	/*
	* Add entries to this ZAP and make sure it spills over
	* and gets upgraded to a fatzap. Also, since we are adding
	* 2050 entries we should see ptrtbl growth and leaf-block split.
	*/
	for (value = 0; value < 2050; value++) {
	char name[ZFS_MAX_DATASET_NAME_LEN];
	dmu_tx_t *tx;
	int error;

	(void) snprintf(name, sizeof (name), "fzap-%"PRIu64"-%"PRIu64"",
	id, value);

	tx = dmu_tx_create(os);
	dmu_tx_hold_zap(tx, object, B_TRUE, name);
	txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG);
	if (txg == 0)
	goto out;
	error = zap_add(os, object, name, sizeof (uint64_t), 1,
	&value, tx);
	ASSERT(error == 0 \|\| error == EEXIST);
	dmu_tx_commit(tx);
	}
	out:
	umem_free(od, sizeof (ztest_od_t));
	}

	/* ARGSUSED */
	void
	ztest_zap_parallel(ztest_ds_t *zd, uint64_t id)
	{
	objset_t *os = zd->zd_os;
	ztest_od_t *od;
	uint64_t txg, object, count, wsize, wc, zl_wsize, zl_wc;
	dmu_tx_t *tx;
	int i, namelen, error;
	int micro = ztest_random(2);
	char name[20], string_value[20];
	void *data;

	od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL);
	ztest_od_init(od, ID_PARALLEL, FTAG, micro, DMU_OT_ZAP_OTHER, 0, 0, 0);

	if (ztest_object_init(zd, od, sizeof (ztest_od_t), B_FALSE) != 0) {
	umem_free(od, sizeof (ztest_od_t));
	return;
	}

	object = od->od_object;

	/*
	* Generate a random name of the form 'xxx.....' where each
	* x is a random printable character and the dots are dots.
	* There are 94 such characters, and the name length goes from
	* 6 to 20, so there are 94^3 * 15 = 12,458,760 possible names.
	*/
	namelen = ztest_random(sizeof (name) - 5) + 5 + 1;

	for (i = 0; i < 3; i++)
	name[i] = '!' + ztest_random('~' - '!' + 1);
	for (; i < namelen - 1; i++)
	name[i] = '.';
	name[i] = '\0';

	if ((namelen & 1) \|\| micro) {
	wsize = sizeof (txg);
	wc = 1;
	data = &txg;
	} else {
	wsize = 1;
	wc = namelen;
	data = string_value;
	}

	count = -1ULL;
	VERIFY0(zap_count(os, object, &count));
	ASSERT3S(count, !=, -1ULL);

	/*
	* Select an operation: length, lookup, add, update, remove.
	*/
	i = ztest_random(5);

	if (i >= 2) {
	tx = dmu_tx_create(os);
	dmu_tx_hold_zap(tx, object, B_TRUE, NULL);
	txg = ztest_tx_assign(tx, TXG_MIGHTWAIT, FTAG);
	if (txg == 0) {
	umem_free(od, sizeof (ztest_od_t));
	return;
	}
	bcopy(name, string_value, namelen);
	} else {
	tx = NULL;
	txg = 0;
	bzero(string_value, namelen);
	}

	switch (i) {

	case 0:
	error = zap_length(os, object, name, &zl_wsize, &zl_wc);
	if (error == 0) {
	ASSERT3U(wsize, ==, zl_wsize);
	ASSERT3U(wc, ==, zl_wc);
	} else {
	ASSERT3U(error, ==, ENOENT);
	}
	break;

	case 1:
	error = zap_lookup(os, object, name, wsize, wc, data);
	if (error == 0) {
	if (data == string_value &&
	bcmp(name, data, namelen) != 0)
	fatal(B_FALSE, "name '%s' != val '%s' len %d",
	name, (char *)data, namelen);
	} else {
	ASSERT3U(error, ==, ENOENT);
	}
	break;

	case 2:
	error = zap_add(os, object, name, wsize, wc, data, tx);
	ASSERT(error == 0 \|\| error == EEXIST);
	break;

	case 3:
	VERIFY0(zap_update(os, object, name, wsize, wc, data, tx));
	break;

	case 4:
	error = zap_remove(os, object, name, tx);
	ASSERT(error == 0 \|\| error == ENOENT);
	break;
	}

	if (tx != NULL)
	dmu_tx_commit(tx);

	umem_free(od, sizeof (ztest_od_t));
	}

	/*
	* Commit callback data.
	*/
	typedef struct ztest_cb_data {
	list_node_t zcd_node;
	uint64_t zcd_txg;
	int zcd_expected_err;
	boolean_t zcd_added;
	boolean_t zcd_called;
	spa_t *zcd_spa;
	} ztest_cb_data_t;

	/* This is the actual commit callback function */
	static void
	ztest_commit_callback(void *arg, int error)
	{
	ztest_cb_data_t *data = arg;
	uint64_t synced_txg;

	VERIFY3P(data, !=, NULL);
	VERIFY3S(data->zcd_expected_err, ==, error);
	VERIFY(!data->zcd_called);

	synced_txg = spa_last_synced_txg(data->zcd_spa);
	if (data->zcd_txg > synced_txg)
	fatal(B_FALSE,
	"commit callback of txg %"PRIu64" called prematurely, "
	"last synced txg = %"PRIu64"\n",
	data->zcd_txg, synced_txg);

	data->zcd_called = B_TRUE;

	if (error == ECANCELED) {
	ASSERT0(data->zcd_txg);
	ASSERT(!data->zcd_added);

	/*
	* The private callback data should be destroyed here, but
	* since we are going to check the zcd_called field after
	* dmu_tx_abort(), we will destroy it there.
	*/
	return;
	}

	ASSERT(data->zcd_added);
	ASSERT3U(data->zcd_txg, !=, 0);

	(void) mutex_enter(&zcl.zcl_callbacks_lock);

	/* See if this cb was called more quickly */
	if ((synced_txg - data->zcd_txg) < zc_min_txg_delay)
	zc_min_txg_delay = synced_txg - data->zcd_txg;

	/* Remove our callback from the list */
	list_remove(&zcl.zcl_callbacks, data);

	(void) mutex_exit(&zcl.zcl_callbacks_lock);

	umem_free(data, sizeof (ztest_cb_data_t));
	}

	/* Allocate and initialize callback data structure */
	static ztest_cb_data_t *
	ztest_create_cb_data(objset_t *os, uint64_t txg)
	{
	ztest_cb_data_t *cb_data;

	cb_data = umem_zalloc(sizeof (ztest_cb_data_t), UMEM_NOFAIL);

	cb_data->zcd_txg = txg;
	cb_data->zcd_spa = dmu_objset_spa(os);
	list_link_init(&cb_data->zcd_node);

	return (cb_data);
	}

	/*
	* Commit callback test.
	*/
	void
	ztest_dmu_commit_callbacks(ztest_ds_t *zd, uint64_t id)
	{
	objset_t *os = zd->zd_os;
	ztest_od_t *od;
	dmu_tx_t *tx;
	ztest_cb_data_t cb_data[3], tmp_cb;
	uint64_t old_txg, txg;
	int i, error = 0;

	od = umem_alloc(sizeof (ztest_od_t), UMEM_NOFAIL);
	ztest_od_init(od, id, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, 0);

	if (ztest_object_init(zd, od, sizeof (ztest_od_t), B_FALSE) != 0) {
	umem_free(od, sizeof (ztest_od_t));
	return;
	}

	tx = dmu_tx_create(os);

	cb_data[0] = ztest_create_cb_data(os, 0);
	dmu_tx_callback_register(tx, ztest_commit_callback, cb_data[0]);

	dmu_tx_hold_write(tx, od->od_object, 0, sizeof (uint64_t));

	/* Every once in a while, abort the transaction on purpose */
	if (ztest_random(100) == 0)
	error = -1;

	if (!error)
	error = dmu_tx_assign(tx, TXG_NOWAIT);

	txg = error ? 0 : dmu_tx_get_txg(tx);

	cb_data[0]->zcd_txg = txg;
	cb_data[1] = ztest_create_cb_data(os, txg);
	dmu_tx_callback_register(tx, ztest_commit_callback, cb_data[1]);

	if (error) {
	/*
	* It's not a strict requirement to call the registered
	* callbacks from inside dmu_tx_abort(), but that's what
	* it's supposed to happen in the current implementation
	* so we will check for that.
	*/
	for (i = 0; i < 2; i++) {
	cb_data[i]->zcd_expected_err = ECANCELED;
	VERIFY(!cb_data[i]->zcd_called);
	}

	dmu_tx_abort(tx);

	for (i = 0; i < 2; i++) {
	VERIFY(cb_data[i]->zcd_called);
	umem_free(cb_data[i], sizeof (ztest_cb_data_t));
	}

	umem_free(od, sizeof (ztest_od_t));
	return;
	}

	cb_data[2] = ztest_create_cb_data(os, txg);
	dmu_tx_callback_register(tx, ztest_commit_callback, cb_data[2]);

	/*
	* Read existing data to make sure there isn't a future leak.
	*/
	VERIFY0(dmu_read(os, od->od_object, 0, sizeof (uint64_t),
	&old_txg, DMU_READ_PREFETCH));

	if (old_txg > txg)
	fatal(B_FALSE,
	"future leak: got %"PRIu64", open txg is %"PRIu64"",
	old_txg, txg);

	dmu_write(os, od->od_object, 0, sizeof (uint64_t), &txg, tx);

	(void) mutex_enter(&zcl.zcl_callbacks_lock);

	/*
	* Since commit callbacks don't have any ordering requirement and since
	* it is theoretically possible for a commit callback to be called
	* after an arbitrary amount of time has elapsed since its txg has been
	* synced, it is difficult to reliably determine whether a commit
	* callback hasn't been called due to high load or due to a flawed
	* implementation.
	*
	* In practice, we will assume that if after a certain number of txgs a
	* commit callback hasn't been called, then most likely there's an
	* implementation bug..
	*/
	tmp_cb = list_head(&zcl.zcl_callbacks);
	if (tmp_cb != NULL &&
	tmp_cb->zcd_txg + ZTEST_COMMIT_CB_THRESH < txg) {
	fatal(B_FALSE,
	"Commit callback threshold exceeded, "
	"oldest txg: %"PRIu64", open txg: %"PRIu64"\n",
	tmp_cb->zcd_txg, txg);
	}

	/*
	* Let's find the place to insert our callbacks.
	*
	* Even though the list is ordered by txg, it is possible for the
	* insertion point to not be the end because our txg may already be
	* quiescing at this point and other callbacks in the open txg
	* (from other objsets) may have sneaked in.
	*/
	tmp_cb = list_tail(&zcl.zcl_callbacks);
	while (tmp_cb != NULL && tmp_cb->zcd_txg > txg)
	tmp_cb = list_prev(&zcl.zcl_callbacks, tmp_cb);

	/* Add the 3 callbacks to the list */
	for (i = 0; i < 3; i++) {
	if (tmp_cb == NULL)
	list_insert_head(&zcl.zcl_callbacks, cb_data[i]);
	else
	list_insert_after(&zcl.zcl_callbacks, tmp_cb,
	cb_data[i]);

	cb_data[i]->zcd_added = B_TRUE;
	VERIFY(!cb_data[i]->zcd_called);

	tmp_cb = cb_data[i];
	}

	zc_cb_counter += 3;

	(void) mutex_exit(&zcl.zcl_callbacks_lock);

	dmu_tx_commit(tx);

	umem_free(od, sizeof (ztest_od_t));
	}

	/*
	* Visit each object in the dataset. Verify that its properties
	* are consistent what was stored in the block tag when it was created,
	* and that its unused bonus buffer space has not been overwritten.
	*/
	/* ARGSUSED */
	void
	ztest_verify_dnode_bt(ztest_ds_t *zd, uint64_t id)
	{
	objset_t *os = zd->zd_os;
	uint64_t obj;
	int err = 0;

	for (obj = 0; err == 0; err = dmu_object_next(os, &obj, FALSE, 0)) {
	ztest_block_tag_t *bt = NULL;
	dmu_object_info_t doi;
	dmu_buf_t *db;

	ztest_object_lock(zd, obj, RL_READER);
	if (dmu_bonus_hold(os, obj, FTAG, &db) != 0) {
	ztest_object_unlock(zd, obj);
	continue;
	}

	dmu_object_info_from_db(db, &doi);
	if (doi.doi_bonus_size >= sizeof (*bt))
	bt = ztest_bt_bonus(db);

	if (bt && bt->bt_magic == BT_MAGIC) {
	ztest_bt_verify(bt, os, obj, doi.doi_dnodesize,
	bt->bt_offset, bt->bt_gen, bt->bt_txg,
	bt->bt_crtxg);
	ztest_verify_unused_bonus(db, bt, obj, os, bt->bt_gen);
	}

	dmu_buf_rele(db, FTAG);
	ztest_object_unlock(zd, obj);
	}
	}

	/* ARGSUSED */
	void
	ztest_dsl_prop_get_set(ztest_ds_t *zd, uint64_t id)
	{
	zfs_prop_t proplist[] = {
	ZFS_PROP_CHECKSUM,
	ZFS_PROP_COMPRESSION,
	ZFS_PROP_COPIES,
	ZFS_PROP_DEDUP
	};
	int p;

	(void) pthread_rwlock_rdlock(&ztest_name_lock);

	for (p = 0; p < sizeof (proplist) / sizeof (proplist[0]); p++)
	(void) ztest_dsl_prop_set_uint64(zd->zd_name, proplist[p],
	ztest_random_dsl_prop(proplist[p]), (int)ztest_random(2));

	VERIFY0(ztest_dsl_prop_set_uint64(zd->zd_name, ZFS_PROP_RECORDSIZE,
	ztest_random_blocksize(), (int)ztest_random(2)));

	(void) pthread_rwlock_unlock(&ztest_name_lock);
	}

	/* ARGSUSED */
	void
	ztest_spa_prop_get_set(ztest_ds_t *zd, uint64_t id)
	{
	nvlist_t *props = NULL;

	(void) pthread_rwlock_rdlock(&ztest_name_lock);

	(void) ztest_spa_prop_set_uint64(ZPOOL_PROP_AUTOTRIM, ztest_random(2));

	VERIFY0(spa_prop_get(ztest_spa, &props));

	if (ztest_opts.zo_verbose >= 6)
	dump_nvlist(props, 4);

	fnvlist_free(props);

	(void) pthread_rwlock_unlock(&ztest_name_lock);
	}

	static int
	user_release_one(const char snapname, const char holdname)
	{
	nvlist_t snaps, holds;
	int error;

	snaps = fnvlist_alloc();
	holds = fnvlist_alloc();
	fnvlist_add_boolean(holds, holdname);
	fnvlist_add_nvlist(snaps, snapname, holds);
	fnvlist_free(holds);
	error = dsl_dataset_user_release(snaps, NULL);
	fnvlist_free(snaps);
	return (error);
	}

	/*
	* Test snapshot hold/release and deferred destroy.
	*/
	void
	ztest_dmu_snapshot_hold(ztest_ds_t *zd, uint64_t id)
	{
	int error;
	objset_t *os = zd->zd_os;
	objset_t *origin;
	char snapname[100];
	char fullname[100];
	char clonename[100];
	char tag[100];
	char osname[ZFS_MAX_DATASET_NAME_LEN];
	nvlist_t *holds;

	(void) pthread_rwlock_rdlock(&ztest_name_lock);

	dmu_objset_name(os, osname);

	(void) snprintf(snapname, sizeof (snapname), "sh1_%"PRIu64"", id);
	(void) snprintf(fullname, sizeof (fullname), "%s@%s", osname, snapname);
	(void) snprintf(clonename, sizeof (clonename), "%s/ch1_%"PRIu64"",
	osname, id);
	(void) snprintf(tag, sizeof (tag), "tag_%"PRIu64"", id);

	/*
	* Clean up from any previous run.
	*/
	error = dsl_destroy_head(clonename);
	if (error != ENOENT)
	ASSERT0(error);
	error = user_release_one(fullname, tag);
	if (error != ESRCH && error != ENOENT)
	ASSERT0(error);
	error = dsl_destroy_snapshot(fullname, B_FALSE);
	if (error != ENOENT)
	ASSERT0(error);

	/*
	* Create snapshot, clone it, mark snap for deferred destroy,
	* destroy clone, verify snap was also destroyed.
	*/
	error = dmu_objset_snapshot_one(osname, snapname);
	if (error) {
	if (error == ENOSPC) {
	ztest_record_enospc("dmu_objset_snapshot");
	goto out;
	}
	fatal(B_FALSE, "dmu_objset_snapshot(%s) = %d", fullname, error);
	}

	error = dmu_objset_clone(clonename, fullname);
	if (error) {
	if (error == ENOSPC) {
	ztest_record_enospc("dmu_objset_clone");
	goto out;
	}
	fatal(B_FALSE, "dmu_objset_clone(%s) = %d", clonename, error);
	}

	error = dsl_destroy_snapshot(fullname, B_TRUE);
	if (error) {
	fatal(B_FALSE, "dsl_destroy_snapshot(%s, B_TRUE) = %d",
	fullname, error);
	}

	error = dsl_destroy_head(clonename);
	if (error)
	fatal(B_FALSE, "dsl_destroy_head(%s) = %d", clonename, error);

	error = dmu_objset_hold(fullname, FTAG, &origin);
	if (error != ENOENT)
	fatal(B_FALSE, "dmu_objset_hold(%s) = %d", fullname, error);

	/*
	* Create snapshot, add temporary hold, verify that we can't
	* destroy a held snapshot, mark for deferred destroy,
	* release hold, verify snapshot was destroyed.
	*/
	error = dmu_objset_snapshot_one(osname, snapname);
	if (error) {
	if (error == ENOSPC) {
	ztest_record_enospc("dmu_objset_snapshot");
	goto out;
	}
	fatal(B_FALSE, "dmu_objset_snapshot(%s) = %d", fullname, error);
	}

	holds = fnvlist_alloc();
	fnvlist_add_string(holds, fullname, tag);
	error = dsl_dataset_user_hold(holds, 0, NULL);
	fnvlist_free(holds);

	if (error == ENOSPC) {
	ztest_record_enospc("dsl_dataset_user_hold");
	goto out;
	} else if (error) {
	fatal(B_FALSE, "dsl_dataset_user_hold(%s, %s) = %u",
	fullname, tag, error);
	}

	error = dsl_destroy_snapshot(fullname, B_FALSE);
	if (error != EBUSY) {
	fatal(B_FALSE, "dsl_destroy_snapshot(%s, B_FALSE) = %d",
	fullname, error);
	}

	error = dsl_destroy_snapshot(fullname, B_TRUE);
	if (error) {
	fatal(B_FALSE, "dsl_destroy_snapshot(%s, B_TRUE) = %d",
	fullname, error);
	}

	error = user_release_one(fullname, tag);
	if (error)
	fatal(B_FALSE, "user_release_one(%s, %s) = %d",
	fullname, tag, error);

	VERIFY3U(dmu_objset_hold(fullname, FTAG, &origin), ==, ENOENT);

	out:
	(void) pthread_rwlock_unlock(&ztest_name_lock);
	}

	/*
	* Inject random faults into the on-disk data.
	*/
	/* ARGSUSED */
	void
	ztest_fault_inject(ztest_ds_t *zd, uint64_t id)
	{
	ztest_shared_t *zs = ztest_shared;
	spa_t *spa = ztest_spa;
	int fd;
	uint64_t offset;
	uint64_t leaves;
	uint64_t bad = 0x1990c0ffeedecadeull;
	uint64_t top, leaf;
	char *path0;
	char *pathrand;
	size_t fsize;
	int bshift = SPA_MAXBLOCKSHIFT + 2;
	int iters = 1000;
	int maxfaults;
	int mirror_save;
	vdev_t *vd0 = NULL;
	uint64_t guid0 = 0;
	boolean_t islog = B_FALSE;

	path0 = umem_alloc(MAXPATHLEN, UMEM_NOFAIL);
	pathrand = umem_alloc(MAXPATHLEN, UMEM_NOFAIL);

	mutex_enter(&ztest_vdev_lock);

	/*
	* Device removal is in progress, fault injection must be disabled
	* until it completes and the pool is scrubbed. The fault injection
	* strategy for damaging blocks does not take in to account evacuated
	* blocks which may have already been damaged.
	*/
	if (ztest_device_removal_active) {
	mutex_exit(&ztest_vdev_lock);
	goto out;
	}

	maxfaults = MAXFAULTS(zs);
	leaves = MAX(zs->zs_mirrors, 1) * ztest_opts.zo_raid_children;
	mirror_save = zs->zs_mirrors;
	mutex_exit(&ztest_vdev_lock);

	ASSERT3U(leaves, >=, 1);

	/*
	* While ztest is running the number of leaves will not change. This
	* is critical for the fault injection logic as it determines where
	* errors can be safely injected such that they are always repairable.
	*
	* When restarting ztest a different number of leaves may be requested
	* which will shift the regions to be damaged. This is fine as long
	* as the pool has been scrubbed prior to using the new mapping.
	* Failure to do can result in non-repairable damage being injected.
	*/
	if (ztest_pool_scrubbed == B_FALSE)
	goto out;

	/*
	* Grab the name lock as reader. There are some operations
	* which don't like to have their vdevs changed while
	* they are in progress (i.e. spa_change_guid). Those
	* operations will have grabbed the name lock as writer.
	*/
	(void) pthread_rwlock_rdlock(&ztest_name_lock);

	/*
	* We need SCL_STATE here because we're going to look at vd0->vdev_tsd.
	*/
	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);

	if (ztest_random(2) == 0) {
	/*
	* Inject errors on a normal data device or slog device.
	*/
	top = ztest_random_vdev_top(spa, B_TRUE);
	leaf = ztest_random(leaves) + zs->zs_splits;

	/*
	* Generate paths to the first leaf in this top-level vdev,
	* and to the random leaf we selected. We'll induce transient
	* write failures and random online/offline activity on leaf 0,
	* and we'll write random garbage to the randomly chosen leaf.
	*/
	(void) snprintf(path0, MAXPATHLEN, ztest_dev_template,
	ztest_opts.zo_dir, ztest_opts.zo_pool,
	top * leaves + zs->zs_splits);
	(void) snprintf(pathrand, MAXPATHLEN, ztest_dev_template,
	ztest_opts.zo_dir, ztest_opts.zo_pool,
	top * leaves + leaf);

	vd0 = vdev_lookup_by_path(spa->spa_root_vdev, path0);
	if (vd0 != NULL && vd0->vdev_top->vdev_islog)
	islog = B_TRUE;

	/*
	* If the top-level vdev needs to be resilvered
	* then we only allow faults on the device that is
	* resilvering.
	*/
	if (vd0 != NULL && maxfaults != 1 &&
	(!vdev_resilver_needed(vd0->vdev_top, NULL, NULL) \|\|
	vd0->vdev_resilver_txg != 0)) {
	/*
	* Make vd0 explicitly claim to be unreadable,
	* or unwritable, or reach behind its back
	* and close the underlying fd. We can do this if
	* maxfaults == 0 because we'll fail and reexecute,
	* and we can do it if maxfaults >= 2 because we'll
	* have enough redundancy. If maxfaults == 1, the
	* combination of this with injection of random data
	* corruption below exceeds the pool's fault tolerance.
	*/
	vdev_file_t *vf = vd0->vdev_tsd;

	zfs_dbgmsg("injecting fault to vdev %llu; maxfaults=%d",
	(long long)vd0->vdev_id, (int)maxfaults);

	if (vf != NULL && ztest_random(3) == 0) {
	(void) close(vf->vf_file->f_fd);
	vf->vf_file->f_fd = -1;
	} else if (ztest_random(2) == 0) {
	vd0->vdev_cant_read = B_TRUE;
	} else {
	vd0->vdev_cant_write = B_TRUE;
	}
	guid0 = vd0->vdev_guid;
	}
	} else {
	/*
	* Inject errors on an l2cache device.
	*/
	spa_aux_vdev_t *sav = &spa->spa_l2cache;

	if (sav->sav_count == 0) {
	spa_config_exit(spa, SCL_STATE, FTAG);
	(void) pthread_rwlock_unlock(&ztest_name_lock);
	goto out;
	}
	vd0 = sav->sav_vdevs[ztest_random(sav->sav_count)];
	guid0 = vd0->vdev_guid;
	(void) strcpy(path0, vd0->vdev_path);
	(void) strcpy(pathrand, vd0->vdev_path);

	leaf = 0;
	leaves = 1;
	maxfaults = INT_MAX; /* no limit on cache devices */
	}

	spa_config_exit(spa, SCL_STATE, FTAG);
	(void) pthread_rwlock_unlock(&ztest_name_lock);

	/*
	* If we can tolerate two or more faults, or we're dealing
	* with a slog, randomly online/offline vd0.
	*/
	if ((maxfaults >= 2 \|\| islog) && guid0 != 0) {
	if (ztest_random(10) < 6) {
	int flags = (ztest_random(2) == 0 ?
	ZFS_OFFLINE_TEMPORARY : 0);

	/*
	* We have to grab the zs_name_lock as writer to
	* prevent a race between offlining a slog and
	* destroying a dataset. Offlining the slog will
	* grab a reference on the dataset which may cause
	* dsl_destroy_head() to fail with EBUSY thus
	* leaving the dataset in an inconsistent state.
	*/
	if (islog)
	(void) pthread_rwlock_wrlock(&ztest_name_lock);

	VERIFY3U(vdev_offline(spa, guid0, flags), !=, EBUSY);

	if (islog)
	(void) pthread_rwlock_unlock(&ztest_name_lock);
	} else {
	/*
	* Ideally we would like to be able to randomly
	* call vdev_[on\|off]line without holding locks
	* to force unpredictable failures but the side
	* effects of vdev_[on\|off]line prevent us from
	* doing so. We grab the ztest_vdev_lock here to
	* prevent a race between injection testing and
	* aux_vdev removal.
	*/
	mutex_enter(&ztest_vdev_lock);
	(void) vdev_online(spa, guid0, 0, NULL);
	mutex_exit(&ztest_vdev_lock);
	}
	}

	if (maxfaults == 0)
	goto out;

	/*
	* We have at least single-fault tolerance, so inject data corruption.
	*/
	fd = open(pathrand, O_RDWR);

	if (fd == -1) /* we hit a gap in the device namespace */
	goto out;

	fsize = lseek(fd, 0, SEEK_END);

	while (--iters != 0) {
	/*
	* The offset must be chosen carefully to ensure that
	* we do not inject a given logical block with errors
	* on two different leaf devices, because ZFS can not
	* tolerate that (if maxfaults==1).
	*
	* To achieve this we divide each leaf device into
	* chunks of size (# leaves * SPA_MAXBLOCKSIZE * 4).
	* Each chunk is further divided into error-injection
	* ranges (can accept errors) and clear ranges (we do
	* not inject errors in those). Each error-injection
	* range can accept errors only for a single leaf vdev.
	* Error-injection ranges are separated by clear ranges.
	*
	* For example, with 3 leaves, each chunk looks like:
	* 0 to 32M: injection range for leaf 0
	* 32M to 64M: clear range - no injection allowed
	* 64M to 96M: injection range for leaf 1
	* 96M to 128M: clear range - no injection allowed
	* 128M to 160M: injection range for leaf 2
	* 160M to 192M: clear range - no injection allowed
	*
	* Each clear range must be large enough such that a
	* single block cannot straddle it. This way a block
	* can't be a target in two different injection ranges
	* (on different leaf vdevs).
	*/
	offset = ztest_random(fsize / (leaves << bshift)) *
	(leaves << bshift) + (leaf << bshift) +
	(ztest_random(1ULL << (bshift - 1)) & -8ULL);

	/*
	* Only allow damage to the labels at one end of the vdev.
	*
	* If all labels are damaged, the device will be totally
	* inaccessible, which will result in loss of data,
	* because we also damage (parts of) the other side of
	* the mirror/raidz.
	*
	* Additionally, we will always have both an even and an
	* odd label, so that we can handle crashes in the
	* middle of vdev_config_sync().
	*/
	if ((leaf & 1) == 0 && offset < VDEV_LABEL_START_SIZE)
	continue;

	/*
	* The two end labels are stored at the "end" of the disk, but
	* the end of the disk (vdev_psize) is aligned to
	* sizeof (vdev_label_t).
	*/
	uint64_t psize = P2ALIGN(fsize, sizeof (vdev_label_t));
	if ((leaf & 1) == 1 &&
	offset + sizeof (bad) > psize - VDEV_LABEL_END_SIZE)
	continue;

	mutex_enter(&ztest_vdev_lock);
	if (mirror_save != zs->zs_mirrors) {
	mutex_exit(&ztest_vdev_lock);
	(void) close(fd);
	goto out;
	}

	if (pwrite(fd, &bad, sizeof (bad), offset) != sizeof (bad))
	fatal(B_TRUE,
	"can't inject bad word at 0x%"PRIx64" in %s",
	offset, pathrand);

	mutex_exit(&ztest_vdev_lock);

	if (ztest_opts.zo_verbose >= 7)
	(void) printf("injected bad word into %s,"
	" offset 0x%"PRIx64"\n", pathrand, offset);
	}

	(void) close(fd);
	out:
	umem_free(path0, MAXPATHLEN);
	umem_free(pathrand, MAXPATHLEN);
	}

	/*
	* By design ztest will never inject uncorrectable damage in to the pool.
	* Issue a scrub, wait for it to complete, and verify there is never any
	* persistent damage.
	*
	* Only after a full scrub has been completed is it safe to start injecting
	* data corruption. See the comment in zfs_fault_inject().
	*/
	static int
	ztest_scrub_impl(spa_t *spa)
	{
	int error = spa_scan(spa, POOL_SCAN_SCRUB);
	if (error)
	return (error);

	while (dsl_scan_scrubbing(spa_get_dsl(spa)))
	txg_wait_synced(spa_get_dsl(spa), 0);

	if (spa_get_errlog_size(spa) > 0)
	return (ECKSUM);

	ztest_pool_scrubbed = B_TRUE;

	return (0);
	}

	/*
	* Scrub the pool.
	*/
	/* ARGSUSED */
	void
	ztest_scrub(ztest_ds_t *zd, uint64_t id)
	{
	spa_t *spa = ztest_spa;
	int error;

	/*
	* Scrub in progress by device removal.
	*/
	if (ztest_device_removal_active)
	return;

	/*
	* Start a scrub, wait a moment, then force a restart.
	*/
	(void) spa_scan(spa, POOL_SCAN_SCRUB);
	(void) poll(NULL, 0, 100);

	error = ztest_scrub_impl(spa);
	if (error == EBUSY)
	error = 0;
	ASSERT0(error);
	}

	/*
	* Change the guid for the pool.
	*/
	/* ARGSUSED */
	void
	ztest_reguid(ztest_ds_t *zd, uint64_t id)
	{
	spa_t *spa = ztest_spa;
	uint64_t orig, load;
	int error;

	if (ztest_opts.zo_mmp_test)
	return;

	orig = spa_guid(spa);
	load = spa_load_guid(spa);

	(void) pthread_rwlock_wrlock(&ztest_name_lock);
	error = spa_change_guid(spa);
	(void) pthread_rwlock_unlock(&ztest_name_lock);

	if (error != 0)
	return;

	if (ztest_opts.zo_verbose >= 4) {
	(void) printf("Changed guid old %"PRIu64" -> %"PRIu64"\n",
	orig, spa_guid(spa));
	}

	VERIFY3U(orig, !=, spa_guid(spa));
	VERIFY3U(load, ==, spa_load_guid(spa));
	}

	void
	ztest_fletcher(ztest_ds_t *zd, uint64_t id)
	{
	hrtime_t end = gethrtime() + NANOSEC;

	while (gethrtime() <= end) {
	int run_count = 100;
	void *buf;
	struct abd abd_data, abd_meta;
	uint32_t size;
	int *ptr;
	int i;
	zio_cksum_t zc_ref;
	zio_cksum_t zc_ref_byteswap;

	size = ztest_random_blocksize();

	buf = umem_alloc(size, UMEM_NOFAIL);
	abd_data = abd_alloc(size, B_FALSE);
	abd_meta = abd_alloc(size, B_TRUE);

	for (i = 0, ptr = buf; i < size / sizeof (*ptr); i++, ptr++)
	*ptr = ztest_random(UINT_MAX);

	abd_copy_from_buf_off(abd_data, buf, 0, size);
	abd_copy_from_buf_off(abd_meta, buf, 0, size);

	VERIFY0(fletcher_4_impl_set("scalar"));
	fletcher_4_native(buf, size, NULL, &zc_ref);
	fletcher_4_byteswap(buf, size, NULL, &zc_ref_byteswap);

	VERIFY0(fletcher_4_impl_set("cycle"));
	while (run_count-- > 0) {
	zio_cksum_t zc;
	zio_cksum_t zc_byteswap;

	fletcher_4_byteswap(buf, size, NULL, &zc_byteswap);
	fletcher_4_native(buf, size, NULL, &zc);

	VERIFY0(bcmp(&zc, &zc_ref, sizeof (zc)));
	VERIFY0(bcmp(&zc_byteswap, &zc_ref_byteswap,
	sizeof (zc_byteswap)));

	/* Test ABD - data */
	abd_fletcher_4_byteswap(abd_data, size, NULL,
	&zc_byteswap);
	abd_fletcher_4_native(abd_data, size, NULL, &zc);

	VERIFY0(bcmp(&zc, &zc_ref, sizeof (zc)));
	VERIFY0(bcmp(&zc_byteswap, &zc_ref_byteswap,
	sizeof (zc_byteswap)));

	/* Test ABD - metadata */
	abd_fletcher_4_byteswap(abd_meta, size, NULL,
	&zc_byteswap);
	abd_fletcher_4_native(abd_meta, size, NULL, &zc);

	VERIFY0(bcmp(&zc, &zc_ref, sizeof (zc)));
	VERIFY0(bcmp(&zc_byteswap, &zc_ref_byteswap,
	sizeof (zc_byteswap)));

	}

	umem_free(buf, size);
	abd_free(abd_data);
	abd_free(abd_meta);
	}
	}

	void
	ztest_fletcher_incr(ztest_ds_t *zd, uint64_t id)
	{
	void *buf;
	size_t size;
	int *ptr;
	int i;
	zio_cksum_t zc_ref;
	zio_cksum_t zc_ref_bswap;

	hrtime_t end = gethrtime() + NANOSEC;

	while (gethrtime() <= end) {
	int run_count = 100;

	size = ztest_random_blocksize();
	buf = umem_alloc(size, UMEM_NOFAIL);

	for (i = 0, ptr = buf; i < size / sizeof (*ptr); i++, ptr++)
	*ptr = ztest_random(UINT_MAX);

	VERIFY0(fletcher_4_impl_set("scalar"));
	fletcher_4_native(buf, size, NULL, &zc_ref);
	fletcher_4_byteswap(buf, size, NULL, &zc_ref_bswap);

	VERIFY0(fletcher_4_impl_set("cycle"));

	while (run_count-- > 0) {
	zio_cksum_t zc;
	zio_cksum_t zc_bswap;
	size_t pos = 0;

	ZIO_SET_CHECKSUM(&zc, 0, 0, 0, 0);
	ZIO_SET_CHECKSUM(&zc_bswap, 0, 0, 0, 0);

	while (pos < size) {
	size_t inc = 64 * ztest_random(size / 67);
	/* sometimes add few bytes to test non-simd */
	if (ztest_random(100) < 10)
	inc += P2ALIGN(ztest_random(64),
	sizeof (uint32_t));

	if (inc > (size - pos))
	inc = size - pos;

	fletcher_4_incremental_native(buf + pos, inc,
	&zc);
	fletcher_4_incremental_byteswap(buf + pos, inc,
	&zc_bswap);

	pos += inc;
	}

	VERIFY3U(pos, ==, size);

	VERIFY(ZIO_CHECKSUM_EQUAL(zc, zc_ref));
	VERIFY(ZIO_CHECKSUM_EQUAL(zc_bswap, zc_ref_bswap));

	/*
	* verify if incremental on the whole buffer is
	* equivalent to non-incremental version
	*/
	ZIO_SET_CHECKSUM(&zc, 0, 0, 0, 0);
	ZIO_SET_CHECKSUM(&zc_bswap, 0, 0, 0, 0);

	fletcher_4_incremental_native(buf, size, &zc);
	fletcher_4_incremental_byteswap(buf, size, &zc_bswap);

	VERIFY(ZIO_CHECKSUM_EQUAL(zc, zc_ref));
	VERIFY(ZIO_CHECKSUM_EQUAL(zc_bswap, zc_ref_bswap));
	}

	umem_free(buf, size);
	}
	}

	static int
	ztest_set_global_vars(void)
	{
	for (size_t i = 0; i < ztest_opts.zo_gvars_count; i++) {
	char *kv = ztest_opts.zo_gvars[i];
	VERIFY3U(strlen(kv), <=, ZO_GVARS_MAX_ARGLEN);
	VERIFY3U(strlen(kv), >, 0);
	int err = set_global_var(kv);
	if (ztest_opts.zo_verbose > 0) {
	(void) printf("setting global var %s ... %s\n", kv,
	err ? "failed" : "ok");
	}
	if (err != 0) {
	(void) fprintf(stderr,
	"failed to set global var '%s'\n", kv);
	return (err);
	}
	}
	return (0);
	}

	static char **
	ztest_global_vars_to_zdb_args(void)
	{
	char *args = calloc(2ztest_opts.zo_gvars_count + 1, sizeof (char *));
	char **cur = args;
	for (size_t i = 0; i < ztest_opts.zo_gvars_count; i++) {
	char *kv = ztest_opts.zo_gvars[i];
	*cur = "-o";
	cur++;
	*cur = strdup(kv);
	cur++;
	}
	ASSERT3P(cur, ==, &args[2*ztest_opts.zo_gvars_count]);
	*cur = NULL;
	return (args);
	}

	/* The end of strings is indicated by a NULL element */
	static char *
	join_strings(char *strings, const char sep)
	{
	size_t totallen = 0;
	for (char *sp = strings; sp != NULL; sp++) {
	totallen += strlen(*sp);
	totallen += strlen(sep);
	}
	if (totallen > 0) {
	ASSERT(totallen >= strlen(sep));
	totallen -= strlen(sep);
	}

	size_t buflen = totallen + 1;
	char o = malloc(buflen); / trailing 0 byte */
	o[0] = '\0';
	for (char *sp = strings; sp != NULL; sp++) {
	size_t would;
	would = strlcat(o, *sp, buflen);
	VERIFY3U(would, <, buflen);
	if (*(sp+1) == NULL) {
	break;
	}
	would = strlcat(o, sep, buflen);
	VERIFY3U(would, <, buflen);
	}
	ASSERT3S(strlen(o), ==, totallen);
	return (o);
	}

	static int
	ztest_check_path(char *path)
	{
	struct stat s;
	/* return true on success */
	return (!stat(path, &s));
	}

	static void
	ztest_get_zdb_bin(char *bin, int len)
	{
	char *zdb_path;
	/*
	* Try to use ZDB_PATH and in-tree zdb path. If not successful, just
	* let popen to search through PATH.
	*/
	if ((zdb_path = getenv("ZDB_PATH"))) {
	strlcpy(bin, zdb_path, len); /* In env */
	if (!ztest_check_path(bin)) {
	ztest_dump_core = 0;
	fatal(B_TRUE, "invalid ZDB_PATH '%s'", bin);
	}
	return;
	}

	VERIFY3P(realpath(getexecname(), bin), !=, NULL);
	if (strstr(bin, "/ztest/")) {
	strstr(bin, "/ztest/")[0] = '\0'; /* In-tree */
	strcat(bin, "/zdb/zdb");
	if (ztest_check_path(bin))
	return;
	}
	strcpy(bin, "zdb");
	}

	static vdev_t *
	ztest_random_concrete_vdev_leaf(vdev_t *vd)
	{
	if (vd == NULL)
	return (NULL);

	if (vd->vdev_children == 0)
	return (vd);

	vdev_t *eligible[vd->vdev_children];
	int eligible_idx = 0, i;
	for (i = 0; i < vd->vdev_children; i++) {
	vdev_t *cvd = vd->vdev_child[i];
	if (cvd->vdev_top->vdev_removing)
	continue;
	if (cvd->vdev_children > 0 \|\|
	(vdev_is_concrete(cvd) && !cvd->vdev_detached)) {
	eligible[eligible_idx++] = cvd;
	}
	}
	VERIFY3S(eligible_idx, >, 0);

	uint64_t child_no = ztest_random(eligible_idx);
	return (ztest_random_concrete_vdev_leaf(eligible[child_no]));
	}

	/* ARGSUSED */
	void
	ztest_initialize(ztest_ds_t *zd, uint64_t id)
	{
	spa_t *spa = ztest_spa;
	int error = 0;

	mutex_enter(&ztest_vdev_lock);

	spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);

	/* Random leaf vdev */
	vdev_t *rand_vd = ztest_random_concrete_vdev_leaf(spa->spa_root_vdev);
	if (rand_vd == NULL) {
	spa_config_exit(spa, SCL_VDEV, FTAG);
	mutex_exit(&ztest_vdev_lock);
	return;
	}

	/*
	* The random vdev we've selected may change as soon as we
	* drop the spa_config_lock. We create local copies of things
	* we're interested in.
	*/
	uint64_t guid = rand_vd->vdev_guid;
	char *path = strdup(rand_vd->vdev_path);
	boolean_t active = rand_vd->vdev_initialize_thread != NULL;

	zfs_dbgmsg("vd %px, guid %llu", rand_vd, (u_longlong_t)guid);
	spa_config_exit(spa, SCL_VDEV, FTAG);

	uint64_t cmd = ztest_random(POOL_INITIALIZE_FUNCS);

	nvlist_t *vdev_guids = fnvlist_alloc();
	nvlist_t *vdev_errlist = fnvlist_alloc();
	fnvlist_add_uint64(vdev_guids, path, guid);
	error = spa_vdev_initialize(spa, vdev_guids, cmd, vdev_errlist);
	fnvlist_free(vdev_guids);
	fnvlist_free(vdev_errlist);

	switch (cmd) {
	case POOL_INITIALIZE_CANCEL:
	if (ztest_opts.zo_verbose >= 4) {
	(void) printf("Cancel initialize %s", path);
	if (!active)
	(void) printf(" failed (no initialize active)");
	(void) printf("\n");
	}
	break;
	case POOL_INITIALIZE_START:
	if (ztest_opts.zo_verbose >= 4) {
	(void) printf("Start initialize %s", path);
	if (active && error == 0)
	(void) printf(" failed (already active)");
	else if (error != 0)
	(void) printf(" failed (error %d)", error);
	(void) printf("\n");
	}
	break;
	case POOL_INITIALIZE_SUSPEND:
	if (ztest_opts.zo_verbose >= 4) {
	(void) printf("Suspend initialize %s", path);
	if (!active)
	(void) printf(" failed (no initialize active)");
	(void) printf("\n");
	}
	break;
	}
	free(path);
	mutex_exit(&ztest_vdev_lock);
	}

	/* ARGSUSED */
	void
	ztest_trim(ztest_ds_t *zd, uint64_t id)
	{
	spa_t *spa = ztest_spa;
	int error = 0;

	mutex_enter(&ztest_vdev_lock);

	spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);

	/* Random leaf vdev */
	vdev_t *rand_vd = ztest_random_concrete_vdev_leaf(spa->spa_root_vdev);
	if (rand_vd == NULL) {
	spa_config_exit(spa, SCL_VDEV, FTAG);
	mutex_exit(&ztest_vdev_lock);
	return;
	}

	/*
	* The random vdev we've selected may change as soon as we
	* drop the spa_config_lock. We create local copies of things
	* we're interested in.
	*/
	uint64_t guid = rand_vd->vdev_guid;
	char *path = strdup(rand_vd->vdev_path);
	boolean_t active = rand_vd->vdev_trim_thread != NULL;

	zfs_dbgmsg("vd %p, guid %llu", rand_vd, (u_longlong_t)guid);
	spa_config_exit(spa, SCL_VDEV, FTAG);

	uint64_t cmd = ztest_random(POOL_TRIM_FUNCS);
	uint64_t rate = 1 << ztest_random(30);
	boolean_t partial = (ztest_random(5) > 0);
	boolean_t secure = (ztest_random(5) > 0);

	nvlist_t *vdev_guids = fnvlist_alloc();
	nvlist_t *vdev_errlist = fnvlist_alloc();
	fnvlist_add_uint64(vdev_guids, path, guid);
	error = spa_vdev_trim(spa, vdev_guids, cmd, rate, partial,
	secure, vdev_errlist);
	fnvlist_free(vdev_guids);
	fnvlist_free(vdev_errlist);

	switch (cmd) {
	case POOL_TRIM_CANCEL:
	if (ztest_opts.zo_verbose >= 4) {
	(void) printf("Cancel TRIM %s", path);
	if (!active)
	(void) printf(" failed (no TRIM active)");
	(void) printf("\n");
	}
	break;
	case POOL_TRIM_START:
	if (ztest_opts.zo_verbose >= 4) {
	(void) printf("Start TRIM %s", path);
	if (active && error == 0)
	(void) printf(" failed (already active)");
	else if (error != 0)
	(void) printf(" failed (error %d)", error);
	(void) printf("\n");
	}
	break;
	case POOL_TRIM_SUSPEND:
	if (ztest_opts.zo_verbose >= 4) {
	(void) printf("Suspend TRIM %s", path);
	if (!active)
	(void) printf(" failed (no TRIM active)");
	(void) printf("\n");
	}
	break;
	}
	free(path);
	mutex_exit(&ztest_vdev_lock);
	}

	/*
	* Verify pool integrity by running zdb.
	*/
	static void
	ztest_run_zdb(char *pool)
	{
	int status;
	char *bin;
	char *zdb;
	char *zbuf;
	const int len = MAXPATHLEN + MAXNAMELEN + 20;
	FILE *fp;

	bin = umem_alloc(len, UMEM_NOFAIL);
	zdb = umem_alloc(len, UMEM_NOFAIL);
	zbuf = umem_alloc(1024, UMEM_NOFAIL);

	ztest_get_zdb_bin(bin, len);

	char **set_gvars_args = ztest_global_vars_to_zdb_args();
	char *set_gvars_args_joined = join_strings(set_gvars_args, " ");
	free(set_gvars_args);

	size_t would = snprintf(zdb, len,
	"%s -bcc%s%s -G -d -Y -e -y %s -p %s %s",
	bin,
	ztest_opts.zo_verbose >= 3 ? "s" : "",
	ztest_opts.zo_verbose >= 4 ? "v" : "",
	set_gvars_args_joined,
	ztest_opts.zo_dir,
	pool);
	ASSERT3U(would, <, len);

	free(set_gvars_args_joined);

	if (ztest_opts.zo_verbose >= 5)
	(void) printf("Executing %s\n", strstr(zdb, "zdb "));

	fp = popen(zdb, "r");

	while (fgets(zbuf, 1024, fp) != NULL)
	if (ztest_opts.zo_verbose >= 3)
	(void) printf("%s", zbuf);

	status = pclose(fp);

	if (status == 0)
	goto out;

	ztest_dump_core = 0;
	if (WIFEXITED(status))
	fatal(B_FALSE, "'%s' exit code %d", zdb, WEXITSTATUS(status));
	else
	fatal(B_FALSE, "'%s' died with signal %d",
	zdb, WTERMSIG(status));
	out:
	umem_free(bin, len);
	umem_free(zdb, len);
	umem_free(zbuf, 1024);
	}

	static void
	ztest_walk_pool_directory(char *header)
	{
	spa_t *spa = NULL;

	if (ztest_opts.zo_verbose >= 6)
	(void) printf("%s\n", header);

	mutex_enter(&spa_namespace_lock);
	while ((spa = spa_next(spa)) != NULL)
	if (ztest_opts.zo_verbose >= 6)
	(void) printf("\t%s\n", spa_name(spa));
	mutex_exit(&spa_namespace_lock);
	}

	static void
	ztest_spa_import_export(char oldname, char newname)
	{
	nvlist_t config, newconfig;
	uint64_t pool_guid;
	spa_t *spa;
	int error;

	if (ztest_opts.zo_verbose >= 4) {
	(void) printf("import/export: old = %s, new = %s\n",
	oldname, newname);
	}

	/*
	* Clean up from previous runs.
	*/
	(void) spa_destroy(newname);

	/*
	* Get the pool's configuration and guid.
	*/
	VERIFY0(spa_open(oldname, &spa, FTAG));

	/*
	* Kick off a scrub to tickle scrub/export races.
	*/
	if (ztest_random(2) == 0)
	(void) spa_scan(spa, POOL_SCAN_SCRUB);

	pool_guid = spa_guid(spa);
	spa_close(spa, FTAG);

	ztest_walk_pool_directory("pools before export");

	/*
	* Export it.
	*/
	VERIFY0(spa_export(oldname, &config, B_FALSE, B_FALSE));

	ztest_walk_pool_directory("pools after export");

	/*
	* Try to import it.
	*/
	newconfig = spa_tryimport(config);
	ASSERT3P(newconfig, !=, NULL);
	fnvlist_free(newconfig);

	/*
	* Import it under the new name.
	*/
	error = spa_import(newname, config, NULL, 0);
	if (error != 0) {
	dump_nvlist(config, 0);
	fatal(B_FALSE, "couldn't import pool %s as %s: error %u",
	oldname, newname, error);
	}

	ztest_walk_pool_directory("pools after import");

	/*
	* Try to import it again -- should fail with EEXIST.
	*/
	VERIFY3U(EEXIST, ==, spa_import(newname, config, NULL, 0));

	/*
	* Try to import it under a different name -- should fail with EEXIST.
	*/
	VERIFY3U(EEXIST, ==, spa_import(oldname, config, NULL, 0));

	/*
	* Verify that the pool is no longer visible under the old name.
	*/
	VERIFY3U(ENOENT, ==, spa_open(oldname, &spa, FTAG));

	/*
	* Verify that we can open and close the pool using the new name.
	*/
	VERIFY0(spa_open(newname, &spa, FTAG));
	ASSERT3U(pool_guid, ==, spa_guid(spa));
	spa_close(spa, FTAG);

	fnvlist_free(config);
	}

	static void
	ztest_resume(spa_t *spa)
	{
	if (spa_suspended(spa) && ztest_opts.zo_verbose >= 6)
	(void) printf("resuming from suspended state\n");
	spa_vdev_state_enter(spa, SCL_NONE);
	vdev_clear(spa, NULL);
	(void) spa_vdev_state_exit(spa, NULL, 0);
	(void) zio_resume(spa);
	}

	static void
	ztest_resume_thread(void *arg)
	{
	spa_t *spa = arg;

	while (!ztest_exiting) {
	if (spa_suspended(spa))
	ztest_resume(spa);
	(void) poll(NULL, 0, 100);

	/*
	* Periodically change the zfs_compressed_arc_enabled setting.
	*/
	if (ztest_random(10) == 0)
	zfs_compressed_arc_enabled = ztest_random(2);

	/*
	* Periodically change the zfs_abd_scatter_enabled setting.
	*/
	if (ztest_random(10) == 0)
	zfs_abd_scatter_enabled = ztest_random(2);
	}

	thread_exit();
	}

	static void
	ztest_deadman_thread(void *arg)
	{
	ztest_shared_t *zs = arg;
	spa_t *spa = ztest_spa;
	hrtime_t delay, overdue, last_run = gethrtime();

	delay = (zs->zs_thread_stop - zs->zs_thread_start) +
	MSEC2NSEC(zfs_deadman_synctime_ms);

	while (!ztest_exiting) {
	/*
	* Wait for the delay timer while checking occasionally
	* if we should stop.
	*/
	if (gethrtime() < last_run + delay) {
	(void) poll(NULL, 0, 1000);
	continue;
	}

	/*
	* If the pool is suspended then fail immediately. Otherwise,
	* check to see if the pool is making any progress. If
	* vdev_deadman() discovers that there hasn't been any recent
	* I/Os then it will end up aborting the tests.
	*/
	if (spa_suspended(spa) \|\| spa->spa_root_vdev == NULL) {
	fatal(B_FALSE,
	"aborting test after %lu seconds because "
	"pool has transitioned to a suspended state.",
	zfs_deadman_synctime_ms / 1000);
	}
	vdev_deadman(spa->spa_root_vdev, FTAG);

	/*
	* If the process doesn't complete within a grace period of
	* zfs_deadman_synctime_ms over the expected finish time,
	* then it may be hung and is terminated.
	*/
	overdue = zs->zs_proc_stop + MSEC2NSEC(zfs_deadman_synctime_ms);
	if (gethrtime() > overdue) {
	fatal(B_FALSE,
	"aborting test after %llu seconds because "
	"the process is overdue for termination.",
	(gethrtime() - zs->zs_proc_start) / NANOSEC);
	}

	(void) printf("ztest has been running for %lld seconds\n",
	(gethrtime() - zs->zs_proc_start) / NANOSEC);

	last_run = gethrtime();
	delay = MSEC2NSEC(zfs_deadman_checktime_ms);
	}

	thread_exit();
	}

	static void
	ztest_execute(int test, ztest_info_t *zi, uint64_t id)
	{
	ztest_ds_t *zd = &ztest_ds[id % ztest_opts.zo_datasets];
	ztest_shared_callstate_t *zc = ZTEST_GET_SHARED_CALLSTATE(test);
	hrtime_t functime = gethrtime();
	int i;

	for (i = 0; i < zi->zi_iters; i++)
	zi->zi_func(zd, id);

	functime = gethrtime() - functime;

	atomic_add_64(&zc->zc_count, 1);
	atomic_add_64(&zc->zc_time, functime);

	if (ztest_opts.zo_verbose >= 4)
	(void) printf("%6.2f sec in %s\n",
	(double)functime / NANOSEC, zi->zi_funcname);
	}

	static void
	ztest_thread(void *arg)
	{
	int rand;
	uint64_t id = (uintptr_t)arg;
	ztest_shared_t *zs = ztest_shared;
	uint64_t call_next;
	hrtime_t now;
	ztest_info_t *zi;
	ztest_shared_callstate_t *zc;

	while ((now = gethrtime()) < zs->zs_thread_stop) {
	/*
	* See if it's time to force a crash.
	*/
	if (now > zs->zs_thread_kill)
	ztest_kill(zs);

	/*
	* If we're getting ENOSPC with some regularity, stop.
	*/
	if (zs->zs_enospc_count > 10)
	break;

	/*
	* Pick a random function to execute.
	*/
	rand = ztest_random(ZTEST_FUNCS);
	zi = &ztest_info[rand];
	zc = ZTEST_GET_SHARED_CALLSTATE(rand);
	call_next = zc->zc_next;

	if (now >= call_next &&
	atomic_cas_64(&zc->zc_next, call_next, call_next +
	ztest_random(2 * zi->zi_interval[0] + 1)) == call_next) {
	ztest_execute(rand, zi, id);
	}
	}

	thread_exit();
	}

	static void
	ztest_dataset_name(char dsname, char pool, int d)
	{
	(void) snprintf(dsname, ZFS_MAX_DATASET_NAME_LEN, "%s/ds_%d", pool, d);
	}

	static void
	ztest_dataset_destroy(int d)
	{
	char name[ZFS_MAX_DATASET_NAME_LEN];
	int t;

	ztest_dataset_name(name, ztest_opts.zo_pool, d);

	if (ztest_opts.zo_verbose >= 3)
	(void) printf("Destroying %s to free up space\n", name);

	/*
	* Cleanup any non-standard clones and snapshots. In general,
	* ztest thread t operates on dataset (t % zopt_datasets),
	* so there may be more than one thing to clean up.
	*/
	for (t = d; t < ztest_opts.zo_threads;
	t += ztest_opts.zo_datasets)
	ztest_dsl_dataset_cleanup(name, t);

	(void) dmu_objset_find(name, ztest_objset_destroy_cb, NULL,
	DS_FIND_SNAPSHOTS \| DS_FIND_CHILDREN);
	}

	static void
	ztest_dataset_dirobj_verify(ztest_ds_t *zd)
	{
	uint64_t usedobjs, dirobjs, scratch;

	/*
	* ZTEST_DIROBJ is the object directory for the entire dataset.
	* Therefore, the number of objects in use should equal the
	* number of ZTEST_DIROBJ entries, +1 for ZTEST_DIROBJ itself.
	* If not, we have an object leak.
	*
	* Note that we can only check this in ztest_dataset_open(),
	* when the open-context and syncing-context values agree.
	* That's because zap_count() returns the open-context value,
	* while dmu_objset_space() returns the rootbp fill count.
	*/
	VERIFY0(zap_count(zd->zd_os, ZTEST_DIROBJ, &dirobjs));
	dmu_objset_space(zd->zd_os, &scratch, &scratch, &usedobjs, &scratch);
	ASSERT3U(dirobjs + 1, ==, usedobjs);
	}

	static int
	ztest_dataset_open(int d)
	{
	ztest_ds_t *zd = &ztest_ds[d];
	uint64_t committed_seq = ZTEST_GET_SHARED_DS(d)->zd_seq;
	objset_t *os;
	zilog_t *zilog;
	char name[ZFS_MAX_DATASET_NAME_LEN];
	int error;

	ztest_dataset_name(name, ztest_opts.zo_pool, d);

	(void) pthread_rwlock_rdlock(&ztest_name_lock);

	error = ztest_dataset_create(name);
	if (error == ENOSPC) {
	(void) pthread_rwlock_unlock(&ztest_name_lock);
	ztest_record_enospc(FTAG);
	return (error);
	}
	ASSERT(error == 0 \|\| error == EEXIST);

	VERIFY0(ztest_dmu_objset_own(name, DMU_OST_OTHER, B_FALSE,
	B_TRUE, zd, &os));
	(void) pthread_rwlock_unlock(&ztest_name_lock);

	ztest_zd_init(zd, ZTEST_GET_SHARED_DS(d), os);

	zilog = zd->zd_zilog;

	if (zilog->zl_header->zh_claim_lr_seq != 0 &&
	zilog->zl_header->zh_claim_lr_seq < committed_seq)
	fatal(B_FALSE, "missing log records: "
	"claimed %"PRIu64" < committed %"PRIu64"",
	zilog->zl_header->zh_claim_lr_seq, committed_seq);

	ztest_dataset_dirobj_verify(zd);

	zil_replay(os, zd, ztest_replay_vector);

	ztest_dataset_dirobj_verify(zd);

	if (ztest_opts.zo_verbose >= 6)
	(void) printf("%s replay %"PRIu64" blocks, "
	"%"PRIu64" records, seq %"PRIu64"\n",
	zd->zd_name,
	zilog->zl_parse_blk_count,
	zilog->zl_parse_lr_count,
	zilog->zl_replaying_seq);

	zilog = zil_open(os, ztest_get_data);

	if (zilog->zl_replaying_seq != 0 &&
	zilog->zl_replaying_seq < committed_seq)
	fatal(B_FALSE, "missing log records: "
	"replayed %"PRIu64" < committed %"PRIu64"",
	zilog->zl_replaying_seq, committed_seq);

	return (0);
	}

	static void
	ztest_dataset_close(int d)
	{
	ztest_ds_t *zd = &ztest_ds[d];

	zil_close(zd->zd_zilog);
	dmu_objset_disown(zd->zd_os, B_TRUE, zd);

	ztest_zd_fini(zd);
	}

	/* ARGSUSED */
	static int
	ztest_replay_zil_cb(const char name, void arg)
	{
	objset_t *os;
	ztest_ds_t *zdtmp;

	VERIFY0(ztest_dmu_objset_own(name, DMU_OST_ANY, B_TRUE,
	B_TRUE, FTAG, &os));

	zdtmp = umem_alloc(sizeof (ztest_ds_t), UMEM_NOFAIL);

	ztest_zd_init(zdtmp, NULL, os);
	zil_replay(os, zdtmp, ztest_replay_vector);
	ztest_zd_fini(zdtmp);

	if (dmu_objset_zil(os)->zl_parse_lr_count != 0 &&
	ztest_opts.zo_verbose >= 6) {
	zilog_t *zilog = dmu_objset_zil(os);

	(void) printf("%s replay %"PRIu64" blocks, "
	"%"PRIu64" records, seq %"PRIu64"\n",
	name,
	zilog->zl_parse_blk_count,
	zilog->zl_parse_lr_count,
	zilog->zl_replaying_seq);
	}

	umem_free(zdtmp, sizeof (ztest_ds_t));

	dmu_objset_disown(os, B_TRUE, FTAG);
	return (0);
	}

	static void
	ztest_freeze(void)
	{
	ztest_ds_t *zd = &ztest_ds[0];
	spa_t *spa;
	int numloops = 0;

	if (ztest_opts.zo_verbose >= 3)
	(void) printf("testing spa_freeze()...\n");

	kernel_init(SPA_MODE_READ \| SPA_MODE_WRITE);
	VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG));
	VERIFY0(ztest_dataset_open(0));
	ztest_spa = spa;

	/*
	* Force the first log block to be transactionally allocated.
	* We have to do this before we freeze the pool -- otherwise
	* the log chain won't be anchored.
	*/
	while (BP_IS_HOLE(&zd->zd_zilog->zl_header->zh_log)) {
	ztest_dmu_object_alloc_free(zd, 0);
	zil_commit(zd->zd_zilog, 0);
	}

	txg_wait_synced(spa_get_dsl(spa), 0);

	/*
	* Freeze the pool. This stops spa_sync() from doing anything,
	* so that the only way to record changes from now on is the ZIL.
	*/
	spa_freeze(spa);

	/*
	* Because it is hard to predict how much space a write will actually
	* require beforehand, we leave ourselves some fudge space to write over
	* capacity.
	*/
	uint64_t capacity = metaslab_class_get_space(spa_normal_class(spa)) / 2;

	/*
	* Run tests that generate log records but don't alter the pool config
	* or depend on DSL sync tasks (snapshots, objset create/destroy, etc).
	* We do a txg_wait_synced() after each iteration to force the txg
	* to increase well beyond the last synced value in the uberblock.
	* The ZIL should be OK with that.
	*
	* Run a random number of times less than zo_maxloops and ensure we do
	* not run out of space on the pool.
	*/
	while (ztest_random(10) != 0 &&
	numloops++ < ztest_opts.zo_maxloops &&
	metaslab_class_get_alloc(spa_normal_class(spa)) < capacity) {
	ztest_od_t od;
	ztest_od_init(&od, 0, FTAG, 0, DMU_OT_UINT64_OTHER, 0, 0, 0);
	VERIFY0(ztest_object_init(zd, &od, sizeof (od), B_FALSE));
	ztest_io(zd, od.od_object,
	ztest_random(ZTEST_RANGE_LOCKS) << SPA_MAXBLOCKSHIFT);
	txg_wait_synced(spa_get_dsl(spa), 0);
	}

	/*
	* Commit all of the changes we just generated.
	*/
	zil_commit(zd->zd_zilog, 0);
	txg_wait_synced(spa_get_dsl(spa), 0);

	/*
	* Close our dataset and close the pool.
	*/
	ztest_dataset_close(0);
	spa_close(spa, FTAG);
	kernel_fini();

	/*
	* Open and close the pool and dataset to induce log replay.
	*/
	kernel_init(SPA_MODE_READ \| SPA_MODE_WRITE);
	VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG));
	ASSERT3U(spa_freeze_txg(spa), ==, UINT64_MAX);
	VERIFY0(ztest_dataset_open(0));
	ztest_spa = spa;
	txg_wait_synced(spa_get_dsl(spa), 0);
	ztest_dataset_close(0);
	ztest_reguid(NULL, 0);

	spa_close(spa, FTAG);
	kernel_fini();
	}

	static void
	ztest_import_impl(ztest_shared_t *zs)
	{
	importargs_t args = { 0 };
	nvlist_t *cfg = NULL;
	int nsearch = 1;
	char *searchdirs[nsearch];
	int flags = ZFS_IMPORT_MISSING_LOG;

	searchdirs[0] = ztest_opts.zo_dir;
	args.paths = nsearch;
	args.path = searchdirs;
	args.can_be_active = B_FALSE;

	VERIFY0(zpool_find_config(NULL, ztest_opts.zo_pool, &cfg, &args,
	&libzpool_config_ops));
	VERIFY0(spa_import(ztest_opts.zo_pool, cfg, NULL, flags));
	fnvlist_free(cfg);
	}

	/*
	* Import a storage pool with the given name.
	*/
	static void
	ztest_import(ztest_shared_t *zs)
	{
	spa_t *spa;

	mutex_init(&ztest_vdev_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&ztest_checkpoint_lock, NULL, MUTEX_DEFAULT, NULL);
	VERIFY0(pthread_rwlock_init(&ztest_name_lock, NULL));

	kernel_init(SPA_MODE_READ \| SPA_MODE_WRITE);

	ztest_import_impl(zs);

	VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG));
	zs->zs_metaslab_sz =
	1ULL << spa->spa_root_vdev->vdev_child[0]->vdev_ms_shift;
	spa_close(spa, FTAG);

	kernel_fini();

	if (!ztest_opts.zo_mmp_test) {
	ztest_run_zdb(ztest_opts.zo_pool);
	ztest_freeze();
	ztest_run_zdb(ztest_opts.zo_pool);
	}

	(void) pthread_rwlock_destroy(&ztest_name_lock);
	mutex_destroy(&ztest_vdev_lock);
	mutex_destroy(&ztest_checkpoint_lock);
	}

	/*
	* Kick off threads to run tests on all datasets in parallel.
	*/
	static void
	ztest_run(ztest_shared_t *zs)
	{
	spa_t *spa;
	objset_t *os;
	kthread_t resume_thread, deadman_thread;
	kthread_t **run_threads;
	uint64_t object;
	int error;
	int t, d;

	ztest_exiting = B_FALSE;

	/*
	* Initialize parent/child shared state.
	*/
	mutex_init(&ztest_vdev_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&ztest_checkpoint_lock, NULL, MUTEX_DEFAULT, NULL);
	VERIFY0(pthread_rwlock_init(&ztest_name_lock, NULL));

	zs->zs_thread_start = gethrtime();
	zs->zs_thread_stop =
	zs->zs_thread_start + ztest_opts.zo_passtime * NANOSEC;
	zs->zs_thread_stop = MIN(zs->zs_thread_stop, zs->zs_proc_stop);
	zs->zs_thread_kill = zs->zs_thread_stop;
	if (ztest_random(100) < ztest_opts.zo_killrate) {
	zs->zs_thread_kill -=
	ztest_random(ztest_opts.zo_passtime * NANOSEC);
	}

	mutex_init(&zcl.zcl_callbacks_lock, NULL, MUTEX_DEFAULT, NULL);

	list_create(&zcl.zcl_callbacks, sizeof (ztest_cb_data_t),
	offsetof(ztest_cb_data_t, zcd_node));

	/*
	* Open our pool. It may need to be imported first depending on
	* what tests were running when the previous pass was terminated.
	*/
	kernel_init(SPA_MODE_READ \| SPA_MODE_WRITE);
	error = spa_open(ztest_opts.zo_pool, &spa, FTAG);
	if (error) {
	VERIFY3S(error, ==, ENOENT);
	ztest_import_impl(zs);
	VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG));
	zs->zs_metaslab_sz =
	1ULL << spa->spa_root_vdev->vdev_child[0]->vdev_ms_shift;
	}

	metaslab_preload_limit = ztest_random(20) + 1;
	ztest_spa = spa;

	VERIFY0(vdev_raidz_impl_set("cycle"));

	dmu_objset_stats_t dds;
	VERIFY0(ztest_dmu_objset_own(ztest_opts.zo_pool,
	DMU_OST_ANY, B_TRUE, B_TRUE, FTAG, &os));
	dsl_pool_config_enter(dmu_objset_pool(os), FTAG);
	dmu_objset_fast_stat(os, &dds);
	dsl_pool_config_exit(dmu_objset_pool(os), FTAG);
	zs->zs_guid = dds.dds_guid;
	dmu_objset_disown(os, B_TRUE, FTAG);

	/*
	* Create a thread to periodically resume suspended I/O.
	*/
	resume_thread = thread_create(NULL, 0, ztest_resume_thread,
	spa, 0, NULL, TS_RUN \| TS_JOINABLE, defclsyspri);

	/*
	* Create a deadman thread and set to panic if we hang.
	*/
	deadman_thread = thread_create(NULL, 0, ztest_deadman_thread,
	zs, 0, NULL, TS_RUN \| TS_JOINABLE, defclsyspri);

	spa->spa_deadman_failmode = ZIO_FAILURE_MODE_PANIC;

	/*
	* Verify that we can safely inquire about any object,
	* whether it's allocated or not. To make it interesting,
	* we probe a 5-wide window around each power of two.
	* This hits all edge cases, including zero and the max.
	*/
	for (t = 0; t < 64; t++) {
	for (d = -5; d <= 5; d++) {
	error = dmu_object_info(spa->spa_meta_objset,
	(1ULL << t) + d, NULL);
	ASSERT(error == 0 \|\| error == ENOENT \|\|
	error == EINVAL);
	}
	}

	/*
	* If we got any ENOSPC errors on the previous run, destroy something.
	*/
	if (zs->zs_enospc_count != 0) {
	int d = ztest_random(ztest_opts.zo_datasets);
	ztest_dataset_destroy(d);
	}
	zs->zs_enospc_count = 0;

	/*
	* If we were in the middle of ztest_device_removal() and were killed
	* we need to ensure the removal and scrub complete before running
	* any tests that check ztest_device_removal_active. The removal will
	* be restarted automatically when the spa is opened, but we need to
	* initiate the scrub manually if it is not already in progress. Note
	* that we always run the scrub whenever an indirect vdev exists
	* because we have no way of knowing for sure if ztest_device_removal()
	* fully completed its scrub before the pool was reimported.
	*/
	if (spa->spa_removing_phys.sr_state == DSS_SCANNING \|\|
	spa->spa_removing_phys.sr_prev_indirect_vdev != -1) {
	while (spa->spa_removing_phys.sr_state == DSS_SCANNING)
	txg_wait_synced(spa_get_dsl(spa), 0);

	error = ztest_scrub_impl(spa);
	if (error == EBUSY)
	error = 0;
	ASSERT0(error);
	}

	run_threads = umem_zalloc(ztest_opts.zo_threads * sizeof (kthread_t *),
	UMEM_NOFAIL);

	if (ztest_opts.zo_verbose >= 4)
	(void) printf("starting main threads...\n");

	/*
	* Replay all logs of all datasets in the pool. This is primarily for
	* temporary datasets which wouldn't otherwise get replayed, which
	* can trigger failures when attempting to offline a SLOG in
	* ztest_fault_inject().
	*/
	(void) dmu_objset_find(ztest_opts.zo_pool, ztest_replay_zil_cb,
	NULL, DS_FIND_CHILDREN);

	/*
	* Kick off all the tests that run in parallel.
	*/
	for (t = 0; t < ztest_opts.zo_threads; t++) {
	if (t < ztest_opts.zo_datasets && ztest_dataset_open(t) != 0) {
	umem_free(run_threads, ztest_opts.zo_threads *
	sizeof (kthread_t *));
	return;
	}

	run_threads[t] = thread_create(NULL, 0, ztest_thread,
	(void *)(uintptr_t)t, 0, NULL, TS_RUN \| TS_JOINABLE,
	defclsyspri);
	}

	/*
	* Wait for all of the tests to complete.
	*/
	for (t = 0; t < ztest_opts.zo_threads; t++)
	VERIFY0(thread_join(run_threads[t]));

	/*
	* Close all datasets. This must be done after all the threads
	* are joined so we can be sure none of the datasets are in-use
	* by any of the threads.
	*/
	for (t = 0; t < ztest_opts.zo_threads; t++) {
	if (t < ztest_opts.zo_datasets)
	ztest_dataset_close(t);
	}

	txg_wait_synced(spa_get_dsl(spa), 0);

	zs->zs_alloc = metaslab_class_get_alloc(spa_normal_class(spa));
	zs->zs_space = metaslab_class_get_space(spa_normal_class(spa));

	umem_free(run_threads, ztest_opts.zo_threads * sizeof (kthread_t *));

	/* Kill the resume and deadman threads */
	ztest_exiting = B_TRUE;
	VERIFY0(thread_join(resume_thread));
	VERIFY0(thread_join(deadman_thread));
	ztest_resume(spa);

	/*
	* Right before closing the pool, kick off a bunch of async I/O;
	* spa_close() should wait for it to complete.
	*/
	for (object = 1; object < 50; object++) {
	dmu_prefetch(spa->spa_meta_objset, object, 0, 0, 1ULL << 20,
	ZIO_PRIORITY_SYNC_READ);
	}

	/* Verify that at least one commit cb was called in a timely fashion */
	if (zc_cb_counter >= ZTEST_COMMIT_CB_MIN_REG)
	VERIFY0(zc_min_txg_delay);

	spa_close(spa, FTAG);

	/*
	* Verify that we can loop over all pools.
	*/
	mutex_enter(&spa_namespace_lock);
	for (spa = spa_next(NULL); spa != NULL; spa = spa_next(spa))
	if (ztest_opts.zo_verbose > 3)
	(void) printf("spa_next: found %s\n", spa_name(spa));
	mutex_exit(&spa_namespace_lock);

	/*
	* Verify that we can export the pool and reimport it under a
	* different name.
	*/
	if ((ztest_random(2) == 0) && !ztest_opts.zo_mmp_test) {
	char name[ZFS_MAX_DATASET_NAME_LEN];
	(void) snprintf(name, sizeof (name), "%s_import",
	ztest_opts.zo_pool);
	ztest_spa_import_export(ztest_opts.zo_pool, name);
	ztest_spa_import_export(name, ztest_opts.zo_pool);
	}

	kernel_fini();

	list_destroy(&zcl.zcl_callbacks);
	mutex_destroy(&zcl.zcl_callbacks_lock);
	(void) pthread_rwlock_destroy(&ztest_name_lock);
	mutex_destroy(&ztest_vdev_lock);
	mutex_destroy(&ztest_checkpoint_lock);
	}

	static void
	print_time(hrtime_t t, char *timebuf)
	{
	hrtime_t s = t / NANOSEC;
	hrtime_t m = s / 60;
	hrtime_t h = m / 60;
	hrtime_t d = h / 24;

	s -= m * 60;
	m -= h * 60;
	h -= d * 24;

	timebuf[0] = '\0';

	if (d)
	(void) sprintf(timebuf,
	"%llud%02lluh%02llum%02llus", d, h, m, s);
	else if (h)
	(void) sprintf(timebuf, "%lluh%02llum%02llus", h, m, s);
	else if (m)
	(void) sprintf(timebuf, "%llum%02llus", m, s);
	else
	(void) sprintf(timebuf, "%llus", s);
	}

	static nvlist_t *
	make_random_props(void)
	{
	nvlist_t *props;

	props = fnvlist_alloc();

	if (ztest_random(2) == 0)
	return (props);

	fnvlist_add_uint64(props,
	zpool_prop_to_name(ZPOOL_PROP_AUTOREPLACE), 1);

	return (props);
	}

	/*
	* Create a storage pool with the given name and initial vdev size.
	* Then test spa_freeze() functionality.
	*/
	static void
	ztest_init(ztest_shared_t *zs)
	{
	spa_t *spa;
	nvlist_t nvroot, props;
	int i;

	mutex_init(&ztest_vdev_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&ztest_checkpoint_lock, NULL, MUTEX_DEFAULT, NULL);
	VERIFY0(pthread_rwlock_init(&ztest_name_lock, NULL));

	kernel_init(SPA_MODE_READ \| SPA_MODE_WRITE);

	/*
	* Create the storage pool.
	*/
	(void) spa_destroy(ztest_opts.zo_pool);
	ztest_shared->zs_vdev_next_leaf = 0;
	zs->zs_splits = 0;
	zs->zs_mirrors = ztest_opts.zo_mirrors;
	nvroot = make_vdev_root(NULL, NULL, NULL, ztest_opts.zo_vdev_size, 0,
	NULL, ztest_opts.zo_raid_children, zs->zs_mirrors, 1);
	props = make_random_props();

	/*
	* We don't expect the pool to suspend unless maxfaults == 0,
	* in which case ztest_fault_inject() temporarily takes away
	* the only valid replica.
	*/
	fnvlist_add_uint64(props,
	zpool_prop_to_name(ZPOOL_PROP_FAILUREMODE),
	MAXFAULTS(zs) ? ZIO_FAILURE_MODE_PANIC : ZIO_FAILURE_MODE_WAIT);

	for (i = 0; i < SPA_FEATURES; i++) {
	char *buf;

	if (!spa_feature_table[i].fi_zfs_mod_supported)
	continue;

	/*
	* 75% chance of using the log space map feature. We want ztest
	* to exercise both the code paths that use the log space map
	* feature and the ones that don't.
	*/
	if (i == SPA_FEATURE_LOG_SPACEMAP && ztest_random(4) == 0)
	continue;

	VERIFY3S(-1, !=, asprintf(&buf, "feature@%s",
	spa_feature_table[i].fi_uname));
	fnvlist_add_uint64(props, buf, 0);
	free(buf);
	}

	VERIFY0(spa_create(ztest_opts.zo_pool, nvroot, props, NULL, NULL));
	fnvlist_free(nvroot);
	fnvlist_free(props);

	VERIFY0(spa_open(ztest_opts.zo_pool, &spa, FTAG));
	zs->zs_metaslab_sz =
	1ULL << spa->spa_root_vdev->vdev_child[0]->vdev_ms_shift;
	spa_close(spa, FTAG);

	kernel_fini();

	if (!ztest_opts.zo_mmp_test) {
	ztest_run_zdb(ztest_opts.zo_pool);
	ztest_freeze();
	ztest_run_zdb(ztest_opts.zo_pool);
	}

	(void) pthread_rwlock_destroy(&ztest_name_lock);
	mutex_destroy(&ztest_vdev_lock);
	mutex_destroy(&ztest_checkpoint_lock);
	}

	static void
	setup_data_fd(void)
	{
	static char ztest_name_data[] = "/tmp/ztest.data.XXXXXX";

	ztest_fd_data = mkstemp(ztest_name_data);
	ASSERT3S(ztest_fd_data, >=, 0);
	(void) unlink(ztest_name_data);
	}

	static int
	shared_data_size(ztest_shared_hdr_t *hdr)
	{
	int size;

	size = hdr->zh_hdr_size;
	size += hdr->zh_opts_size;
	size += hdr->zh_size;
	size += hdr->zh_stats_size * hdr->zh_stats_count;
	size += hdr->zh_ds_size * hdr->zh_ds_count;

	return (size);
	}

	static void
	setup_hdr(void)
	{
	int size;
	ztest_shared_hdr_t *hdr;

	hdr = (void )mmap(0, P2ROUNDUP(sizeof (hdr), getpagesize()),
	PROT_READ \| PROT_WRITE, MAP_SHARED, ztest_fd_data, 0);
	ASSERT3P(hdr, !=, MAP_FAILED);

	VERIFY0(ftruncate(ztest_fd_data, sizeof (ztest_shared_hdr_t)));

	hdr->zh_hdr_size = sizeof (ztest_shared_hdr_t);
	hdr->zh_opts_size = sizeof (ztest_shared_opts_t);
	hdr->zh_size = sizeof (ztest_shared_t);
	hdr->zh_stats_size = sizeof (ztest_shared_callstate_t);
	hdr->zh_stats_count = ZTEST_FUNCS;
	hdr->zh_ds_size = sizeof (ztest_shared_ds_t);
	hdr->zh_ds_count = ztest_opts.zo_datasets;

	size = shared_data_size(hdr);
	VERIFY0(ftruncate(ztest_fd_data, size));

	(void) munmap((caddr_t)hdr, P2ROUNDUP(sizeof (*hdr), getpagesize()));
	}

	static void
	setup_data(void)
	{
	int size, offset;
	ztest_shared_hdr_t *hdr;
	uint8_t *buf;

	hdr = (void )mmap(0, P2ROUNDUP(sizeof (hdr), getpagesize()),
	PROT_READ, MAP_SHARED, ztest_fd_data, 0);
	ASSERT3P(hdr, !=, MAP_FAILED);

	size = shared_data_size(hdr);

	(void) munmap((caddr_t)hdr, P2ROUNDUP(sizeof (*hdr), getpagesize()));
	hdr = ztest_shared_hdr = (void *)mmap(0, P2ROUNDUP(size, getpagesize()),
	PROT_READ \| PROT_WRITE, MAP_SHARED, ztest_fd_data, 0);
	ASSERT3P(hdr, !=, MAP_FAILED);
	buf = (uint8_t *)hdr;

	offset = hdr->zh_hdr_size;
	ztest_shared_opts = (void *)&buf[offset];
	offset += hdr->zh_opts_size;
	ztest_shared = (void *)&buf[offset];
	offset += hdr->zh_size;
	ztest_shared_callstate = (void *)&buf[offset];
	offset += hdr->zh_stats_size * hdr->zh_stats_count;
	ztest_shared_ds = (void *)&buf[offset];
	}

	static boolean_t
	exec_child(char cmd, char libpath, boolean_t ignorekill, int *statusp)
	{
	pid_t pid;
	int status;
	char *cmdbuf = NULL;

	pid = fork();

	if (cmd == NULL) {
	cmdbuf = umem_alloc(MAXPATHLEN, UMEM_NOFAIL);
	(void) strlcpy(cmdbuf, getexecname(), MAXPATHLEN);
	cmd = cmdbuf;
	}

	if (pid == -1)
	fatal(B_TRUE, "fork failed");

	if (pid == 0) { /* child */
	char *emptyargv[2] = { cmd, NULL };
	char fd_data_str[12];

	struct rlimit rl = { 1024, 1024 };
	(void) setrlimit(RLIMIT_NOFILE, &rl);

	(void) close(ztest_fd_rand);
	VERIFY3S(11, >=,
	snprintf(fd_data_str, 12, "%d", ztest_fd_data));
	VERIFY0(setenv("ZTEST_FD_DATA", fd_data_str, 1));

	(void) enable_extended_FILE_stdio(-1, -1);
	if (libpath != NULL)
	VERIFY0(setenv("LD_LIBRARY_PATH", libpath, 1));
	(void) execv(cmd, emptyargv);
	ztest_dump_core = B_FALSE;
	fatal(B_TRUE, "exec failed: %s", cmd);
	}

	if (cmdbuf != NULL) {
	umem_free(cmdbuf, MAXPATHLEN);
	cmd = NULL;
	}

	while (waitpid(pid, &status, 0) != pid)
	continue;
	if (statusp != NULL)
	*statusp = status;

	if (WIFEXITED(status)) {
	if (WEXITSTATUS(status) != 0) {
	(void) fprintf(stderr, "child exited with code %d\n",
	WEXITSTATUS(status));
	exit(2);
	}
	return (B_FALSE);
	} else if (WIFSIGNALED(status)) {
	if (!ignorekill \|\| WTERMSIG(status) != SIGKILL) {
	(void) fprintf(stderr, "child died with signal %d\n",
	WTERMSIG(status));
	exit(3);
	}
	return (B_TRUE);
	} else {
	(void) fprintf(stderr, "something strange happened to child\n");
	exit(4);
	}
	}

	static void
	ztest_run_init(void)
	{
	int i;

	ztest_shared_t *zs = ztest_shared;

	/*
	* Blow away any existing copy of zpool.cache
	*/
	(void) remove(spa_config_path);

	if (ztest_opts.zo_init == 0) {
	if (ztest_opts.zo_verbose >= 1)
	(void) printf("Importing pool %s\n",
	ztest_opts.zo_pool);
	ztest_import(zs);
	return;
	}

	/*
	* Create and initialize our storage pool.
	*/
	for (i = 1; i <= ztest_opts.zo_init; i++) {
	bzero(zs, sizeof (ztest_shared_t));
	if (ztest_opts.zo_verbose >= 3 &&
	ztest_opts.zo_init != 1) {
	(void) printf("ztest_init(), pass %d\n", i);
	}
	ztest_init(zs);
	}
	}

	int
	main(int argc, char **argv)
	{
	int kills = 0;
	int iters = 0;
	int older = 0;
	int newer = 0;
	ztest_shared_t *zs;
	ztest_info_t *zi;
	ztest_shared_callstate_t *zc;
	char timebuf[100];
	char numbuf[NN_NUMBUF_SZ];
	char *cmd;
	boolean_t hasalt;
	int f, err;
	char *fd_data_str = getenv("ZTEST_FD_DATA");
	struct sigaction action;

	(void) setvbuf(stdout, NULL, _IOLBF, 0);

	dprintf_setup(&argc, argv);
	zfs_deadman_synctime_ms = 300000;
	zfs_deadman_checktime_ms = 30000;
	/*
	* As two-word space map entries may not come up often (especially
	* if pool and vdev sizes are small) we want to force at least some
	* of them so the feature get tested.
	*/
	zfs_force_some_double_word_sm_entries = B_TRUE;

	/*
	* Verify that even extensively damaged split blocks with many
	* segments can be reconstructed in a reasonable amount of time
	* when reconstruction is known to be possible.
	*
	* Note: the lower this value is, the more damage we inflict, and
	* the more time ztest spends in recovering that damage. We chose
	* to induce damage 1/100th of the time so recovery is tested but
	* not so frequently that ztest doesn't get to test other code paths.
	*/
	zfs_reconstruct_indirect_damage_fraction = 100;

	action.sa_handler = sig_handler;
	sigemptyset(&action.sa_mask);
	action.sa_flags = 0;

	if (sigaction(SIGSEGV, &action, NULL) < 0) {
	(void) fprintf(stderr, "ztest: cannot catch SIGSEGV: %s.\n",
	strerror(errno));
	exit(EXIT_FAILURE);
	}

	if (sigaction(SIGABRT, &action, NULL) < 0) {
	(void) fprintf(stderr, "ztest: cannot catch SIGABRT: %s.\n",
	strerror(errno));
	exit(EXIT_FAILURE);
	}

	/*
	* Force random_get_bytes() to use /dev/urandom in order to prevent
	* ztest from needlessly depleting the system entropy pool.
	*/
	random_path = "/dev/urandom";
	ztest_fd_rand = open(random_path, O_RDONLY);
	ASSERT3S(ztest_fd_rand, >=, 0);

	if (!fd_data_str) {
	process_options(argc, argv);

	setup_data_fd();
	setup_hdr();
	setup_data();
	bcopy(&ztest_opts, ztest_shared_opts,
	sizeof (*ztest_shared_opts));
	} else {
	ztest_fd_data = atoi(fd_data_str);
	setup_data();
	bcopy(ztest_shared_opts, &ztest_opts, sizeof (ztest_opts));
	}
	ASSERT3U(ztest_opts.zo_datasets, ==, ztest_shared_hdr->zh_ds_count);

	err = ztest_set_global_vars();
	if (err != 0 && !fd_data_str) {
	/* error message done by ztest_set_global_vars */
	exit(EXIT_FAILURE);
	} else {
	/* children should not be spawned if setting gvars fails */
	VERIFY3S(err, ==, 0);
	}

	/* Override location of zpool.cache */
	VERIFY3S(asprintf((char **)&spa_config_path, "%s/zpool.cache",
	ztest_opts.zo_dir), !=, -1);

	ztest_ds = umem_alloc(ztest_opts.zo_datasets * sizeof (ztest_ds_t),
	UMEM_NOFAIL);
	zs = ztest_shared;

	if (fd_data_str) {
	metaslab_force_ganging = ztest_opts.zo_metaslab_force_ganging;
	metaslab_df_alloc_threshold =
	zs->zs_metaslab_df_alloc_threshold;

	if (zs->zs_do_init)
	ztest_run_init();
	else
	ztest_run(zs);
	exit(0);
	}

	hasalt = (strlen(ztest_opts.zo_alt_ztest) != 0);

	if (ztest_opts.zo_verbose >= 1) {
	(void) printf("%"PRIu64" vdevs, %d datasets, %d threads,"
	"%d %s disks, %"PRIu64" seconds...\n\n",
	ztest_opts.zo_vdevs,
	ztest_opts.zo_datasets,
	ztest_opts.zo_threads,
	ztest_opts.zo_raid_children,
	ztest_opts.zo_raid_type,
	ztest_opts.zo_time);
	}

	cmd = umem_alloc(MAXNAMELEN, UMEM_NOFAIL);
	(void) strlcpy(cmd, getexecname(), MAXNAMELEN);

	zs->zs_do_init = B_TRUE;
	if (strlen(ztest_opts.zo_alt_ztest) != 0) {
	if (ztest_opts.zo_verbose >= 1) {
	(void) printf("Executing older ztest for "
	"initialization: %s\n", ztest_opts.zo_alt_ztest);
	}
	VERIFY(!exec_child(ztest_opts.zo_alt_ztest,
	ztest_opts.zo_alt_libpath, B_FALSE, NULL));
	} else {
	VERIFY(!exec_child(NULL, NULL, B_FALSE, NULL));
	}
	zs->zs_do_init = B_FALSE;

	zs->zs_proc_start = gethrtime();
	zs->zs_proc_stop = zs->zs_proc_start + ztest_opts.zo_time * NANOSEC;

	for (f = 0; f < ZTEST_FUNCS; f++) {
	zi = &ztest_info[f];
	zc = ZTEST_GET_SHARED_CALLSTATE(f);
	if (zs->zs_proc_start + zi->zi_interval[0] > zs->zs_proc_stop)
	zc->zc_next = UINT64_MAX;
	else
	zc->zc_next = zs->zs_proc_start +
	ztest_random(2 * zi->zi_interval[0] + 1);
	}

	/*
	* Run the tests in a loop. These tests include fault injection
	* to verify that self-healing data works, and forced crashes
	* to verify that we never lose on-disk consistency.
	*/
	while (gethrtime() < zs->zs_proc_stop) {
	int status;
	boolean_t killed;

	/*
	* Initialize the workload counters for each function.
	*/
	for (f = 0; f < ZTEST_FUNCS; f++) {
	zc = ZTEST_GET_SHARED_CALLSTATE(f);
	zc->zc_count = 0;
	zc->zc_time = 0;
	}

	/* Set the allocation switch size */
	zs->zs_metaslab_df_alloc_threshold =
	ztest_random(zs->zs_metaslab_sz / 4) + 1;

	if (!hasalt \|\| ztest_random(2) == 0) {
	if (hasalt && ztest_opts.zo_verbose >= 1) {
	(void) printf("Executing newer ztest: %s\n",
	cmd);
	}
	newer++;
	killed = exec_child(cmd, NULL, B_TRUE, &status);
	} else {
	if (hasalt && ztest_opts.zo_verbose >= 1) {
	(void) printf("Executing older ztest: %s\n",
	ztest_opts.zo_alt_ztest);
	}
	older++;
	killed = exec_child(ztest_opts.zo_alt_ztest,
	ztest_opts.zo_alt_libpath, B_TRUE, &status);
	}

	if (killed)
	kills++;
	iters++;

	if (ztest_opts.zo_verbose >= 1) {
	hrtime_t now = gethrtime();

	now = MIN(now, zs->zs_proc_stop);
	print_time(zs->zs_proc_stop - now, timebuf);
	nicenum(zs->zs_space, numbuf, sizeof (numbuf));

	(void) printf("Pass %3d, %8s, %3"PRIu64" ENOSPC, "
	"%4.1f%% of %5s used, %3.0f%% done, %8s to go\n",
	iters,
	WIFEXITED(status) ? "Complete" : "SIGKILL",
	zs->zs_enospc_count,
	100.0 * zs->zs_alloc / zs->zs_space,
	numbuf,
	100.0 * (now - zs->zs_proc_start) /
	(ztest_opts.zo_time * NANOSEC), timebuf);
	}

	if (ztest_opts.zo_verbose >= 2) {
	(void) printf("\nWorkload summary:\n\n");
	(void) printf("%7s %9s %s\n",
	"Calls", "Time", "Function");
	(void) printf("%7s %9s %s\n",
	"-----", "----", "--------");
	for (f = 0; f < ZTEST_FUNCS; f++) {
	zi = &ztest_info[f];
	zc = ZTEST_GET_SHARED_CALLSTATE(f);
	print_time(zc->zc_time, timebuf);
	(void) printf("%7"PRIu64" %9s %s\n",
	zc->zc_count, timebuf,
	zi->zi_funcname);
	}
	(void) printf("\n");
	}

	if (!ztest_opts.zo_mmp_test)
	ztest_run_zdb(ztest_opts.zo_pool);
	}

	if (ztest_opts.zo_verbose >= 1) {
	if (hasalt) {
	(void) printf("%d runs of older ztest: %s\n", older,
	ztest_opts.zo_alt_ztest);
	(void) printf("%d runs of newer ztest: %s\n", newer,
	cmd);
	}
	(void) printf("%d killed, %d completed, %.0f%% kill rate\n",
	kills, iters - kills, (100.0 * kills) / MAX(1, iters));
	}

	umem_free(cmd, MAXNAMELEN);

	return (0);
	}
	diff --git a/sys/contrib/openzfs/config/Shellcheck.am b/sys/contrib/openzfs/config/Shellcheck.am
	index 6b805b797d12..e255e6733dd0 100644
	--- a/sys/contrib/openzfs/config/Shellcheck.am
	+++ b/sys/contrib/openzfs/config/Shellcheck.am
	@@ -1,22 +1,22 @@
	.PHONY: shellcheck
	shellcheck: $(SCRIPTS) $(SHELLCHECKSCRIPTS)
	if HAVE_SHELLCHECK
	- [ -z "$(SCRIPTS)$(SHELLCHECKSCRIPTS)" ] && exit; shellcheck $$([ -n "$(SHELLCHECK_SHELL)" ] && echo "--shell=$(SHELLCHECK_SHELL)") --exclude=SC1090,SC1091$(SHELLCHECK_IGNORE) --format=gcc $(SCRIPTS) $(SHELLCHECKSCRIPTS)
	+ [ -z "$(SCRIPTS)$(SHELLCHECKSCRIPTS)" ] && exit; shellcheck --format=gcc --exclude=SC1090,SC1091,SC2250 $$([ -n "$(SHELLCHECK_SHELL)" ] && echo "--shell=$(SHELLCHECK_SHELL)") $(SHELLCHECK_OPTS) $(SCRIPTS) $(SHELLCHECKSCRIPTS)
	else
	@[ -z "$(SCRIPTS)$(SHELLCHECKSCRIPTS)" ] && exit; echo "skipping shellcheck of" $(SCRIPTS) $(SHELLCHECKSCRIPTS) "because shellcheck is not installed"
	endif
	@set -e; for dir in $(SHELLCHECKDIRS); do $(MAKE) -C $$dir shellcheck; done


	# command -v is specified by POSIX and every shell in existence supports it
	.PHONY: checkbashisms
	checkbashisms: $(SCRIPTS) $(SHELLCHECKSCRIPTS)
	if HAVE_CHECKBASHISMS
	[ -z "$(SCRIPTS)$(SHELLCHECKSCRIPTS)" ] && exit; ! if [ -z "$(SHELLCHECK_SHELL)" ]; then \
	checkbashisms -npx $(SCRIPTS) $(SHELLCHECKSCRIPTS); else \
	for f in $(SCRIPTS) $(SHELLCHECKSCRIPTS); do echo $$f >&3; { echo '#!/bin/$(SHELLCHECK_SHELL)'; cat $$f; } \| checkbashisms -npx; done; \
	fi 3>&2 2>&1 \| grep -vFe "'command' with option other than -p" -e 'command -v' $(CHECKBASHISMS_IGNORE) >&2
	else
	@[ -z "$(SCRIPTS)$(SHELLCHECKSCRIPTS)" ] && exit; echo "skipping checkbashisms of" $(SCRIPTS) $(SHELLCHECKSCRIPTS) "because checkbashisms is not installed"
	endif
	@set -e; for dir in $(SHELLCHECKDIRS); do $(MAKE) -C $$dir checkbashisms; done
	diff --git a/sys/contrib/openzfs/config/kernel-bio.m4 b/sys/contrib/openzfs/config/kernel-bio.m4
	index aad4d31cf29b..d088d7023cb0 100644
	--- a/sys/contrib/openzfs/config/kernel-bio.m4
	+++ b/sys/contrib/openzfs/config/kernel-bio.m4
	@@ -1,431 +1,515 @@
	dnl #
	dnl # 2.6.36 API change,
	dnl # REQ_FAILFAST_{DEV\|TRANSPORT\|DRIVER}
	dnl # REQ_DISCARD
	dnl # REQ_FLUSH
	dnl #
	dnl # 4.8 - 4.9 API,
	dnl # REQ_FLUSH was renamed to REQ_PREFLUSH
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_REQ], [
	ZFS_LINUX_TEST_SRC([req_failfast_mask], [
	#include <linux/bio.h>
	],[
	int flags __attribute__ ((unused));
	flags = REQ_FAILFAST_MASK;
	])

	ZFS_LINUX_TEST_SRC([req_discard], [
	#include <linux/bio.h>
	],[
	int flags __attribute__ ((unused));
	flags = REQ_DISCARD;
	])

	ZFS_LINUX_TEST_SRC([req_flush], [
	#include <linux/bio.h>
	],[
	int flags __attribute__ ((unused));
	flags = REQ_FLUSH;
	])

	ZFS_LINUX_TEST_SRC([req_preflush], [
	#include <linux/bio.h>
	],[
	int flags __attribute__ ((unused));
	flags = REQ_PREFLUSH;
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_REQ_FAILFAST_MASK], [
	AC_MSG_CHECKING([whether REQ_FAILFAST_MASK is defined])
	ZFS_LINUX_TEST_RESULT([req_failfast_mask], [
	AC_MSG_RESULT(yes)
	],[
	ZFS_LINUX_TEST_ERROR([REQ_FAILFAST_MASK])
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_REQ_DISCARD], [
	AC_MSG_CHECKING([whether REQ_DISCARD is defined])
	ZFS_LINUX_TEST_RESULT([req_discard], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_REQ_DISCARD, 1, [REQ_DISCARD is defined])
	],[
	AC_MSG_RESULT(no)
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_REQ_FLUSH], [
	AC_MSG_CHECKING([whether REQ_FLUSH is defined])
	ZFS_LINUX_TEST_RESULT([req_flush], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_REQ_FLUSH, 1, [REQ_FLUSH is defined])
	],[
	AC_MSG_RESULT(no)
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_REQ_PREFLUSH], [
	AC_MSG_CHECKING([whether REQ_PREFLUSH is defined])
	ZFS_LINUX_TEST_RESULT([req_preflush], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_REQ_PREFLUSH, 1, [REQ_PREFLUSH is defined])
	],[
	AC_MSG_RESULT(no)
	])
	])

	dnl #
	dnl # Linux 4.8 API,
	dnl #
	dnl # The bio_op() helper was introduced as a replacement for explicitly
	dnl # checking the bio->bi_rw flags. The following checks are used to
	dnl # detect if a specific operation is supported.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO_OPS], [
	ZFS_LINUX_TEST_SRC([req_op_discard], [
	#include <linux/blk_types.h>
	],[
	int op __attribute__ ((unused)) = REQ_OP_DISCARD;
	])

	ZFS_LINUX_TEST_SRC([req_op_secure_erase], [
	#include <linux/blk_types.h>
	],[
	int op __attribute__ ((unused)) = REQ_OP_SECURE_ERASE;
	])

	ZFS_LINUX_TEST_SRC([req_op_flush], [
	#include <linux/blk_types.h>
	],[
	int op __attribute__ ((unused)) = REQ_OP_FLUSH;
	])

	ZFS_LINUX_TEST_SRC([bio_bi_opf], [
	#include <linux/bio.h>
	],[
	struct bio bio __attribute__ ((unused));
	bio.bi_opf = 0;
	])

	ZFS_LINUX_TEST_SRC([bio_set_op_attrs], [
	#include <linux/bio.h>
	],[
	struct bio *bio __attribute__ ((unused)) = NULL;
	bio_set_op_attrs(bio, 0, 0);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_REQ_OP_DISCARD], [
	AC_MSG_CHECKING([whether REQ_OP_DISCARD is defined])
	ZFS_LINUX_TEST_RESULT([req_op_discard], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_REQ_OP_DISCARD, 1, [REQ_OP_DISCARD is defined])
	],[
	AC_MSG_RESULT(no)
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_REQ_OP_SECURE_ERASE], [
	AC_MSG_CHECKING([whether REQ_OP_SECURE_ERASE is defined])
	ZFS_LINUX_TEST_RESULT([req_op_secure_erase], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_REQ_OP_SECURE_ERASE, 1,
	[REQ_OP_SECURE_ERASE is defined])
	],[
	AC_MSG_RESULT(no)
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_REQ_OP_FLUSH], [
	AC_MSG_CHECKING([whether REQ_OP_FLUSH is defined])
	ZFS_LINUX_TEST_RESULT([req_op_flush], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_REQ_OP_FLUSH, 1, [REQ_OP_FLUSH is defined])
	],[
	AC_MSG_RESULT(no)
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_BI_OPF], [
	AC_MSG_CHECKING([whether bio->bi_opf is defined])
	ZFS_LINUX_TEST_RESULT([bio_bi_opf], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BIO_BI_OPF, 1, [bio->bi_opf is defined])
	],[
	AC_MSG_RESULT(no)
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_SET_OP_ATTRS], [
	AC_MSG_CHECKING([whether bio_set_op_attrs is available])
	ZFS_LINUX_TEST_RESULT([bio_set_op_attrs], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BIO_SET_OP_ATTRS, 1,
	[bio_set_op_attrs is available])
	],[
	AC_MSG_RESULT(no)
	])
	])

	dnl #
	dnl # Linux 4.14 API,
	dnl #
	dnl # The bio_set_dev() helper macro was introduced as part of the transition
	dnl # to have struct gendisk in struct bio.
	dnl #
	dnl # Linux 5.0 API,
	dnl #
	dnl # The bio_set_dev() helper macro was updated to internally depend on
	dnl # bio_associate_blkg() symbol which is exported GPL-only.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO_SET_DEV], [
	ZFS_LINUX_TEST_SRC([bio_set_dev], [
	#include <linux/bio.h>
	#include <linux/fs.h>
	],[
	struct block_device *bdev = NULL;
	struct bio *bio = NULL;
	bio_set_dev(bio, bdev);
	], [], [ZFS_META_LICENSE])
	])

	+dnl #
	+dnl # Linux 5.16 API
	+dnl #
	+dnl # bio_set_dev is no longer a helper macro and is now an inline function,
	+dnl # meaning that the function it calls internally can no longer be overridden
	+dnl # by our code
	+dnl #
	+AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO_SET_DEV_MACRO], [
	+ ZFS_LINUX_TEST_SRC([bio_set_dev_macro], [
	+ #include <linux/bio.h>
	+ #include <linux/fs.h>
	+ ],[
	+ #ifndef bio_set_dev
	+ #error Not a macro
	+ #endif
	+ ], [], [ZFS_META_LICENSE])
	+])
	+
	AC_DEFUN([ZFS_AC_KERNEL_BIO_SET_DEV], [
	AC_MSG_CHECKING([whether bio_set_dev() is available])
	ZFS_LINUX_TEST_RESULT([bio_set_dev], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BIO_SET_DEV, 1, [bio_set_dev() is available])

	AC_MSG_CHECKING([whether bio_set_dev() is GPL-only])
	ZFS_LINUX_TEST_RESULT([bio_set_dev_license], [
	AC_MSG_RESULT(no)
	],[
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BIO_SET_DEV_GPL_ONLY, 1,
	[bio_set_dev() GPL-only])
	])
	+
	+ AC_MSG_CHECKING([whether bio_set_dev() is a macro])
	+ ZFS_LINUX_TEST_RESULT([bio_set_dev_macro], [
	+ AC_MSG_RESULT(yes)
	+ AC_DEFINE(HAVE_BIO_SET_DEV_MACRO, 1,
	+ [bio_set_dev() is a macro])
	+ ],[
	+ AC_MSG_RESULT(no)
	+ ])
	],[
	AC_MSG_RESULT(no)
	])
	])

	dnl #
	dnl # 4.3 API change
	dnl # Error argument dropped from bio_endio in favor of newly introduced
	dnl # bio->bi_error. This also replaces bio->bi_flags value BIO_UPTODATE.
	dnl # Introduced by torvalds/linux@4246a0b63bd8f56a1469b12eafeb875b1041a451
	dnl # ("block: add a bi_error field to struct bio").
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO_END_IO_T_ARGS], [
	ZFS_LINUX_TEST_SRC([bio_end_io_t_args], [
	#include <linux/bio.h>
	void wanted_end_io(struct bio *bio) { return; }
	bio_end_io_t *end_io __attribute__ ((unused)) = wanted_end_io;
	], [])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_END_IO_T_ARGS], [
	AC_MSG_CHECKING([whether bio_end_io_t wants 1 arg])
	ZFS_LINUX_TEST_RESULT([bio_end_io_t_args], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_1ARG_BIO_END_IO_T, 1,
	[bio_end_io_t wants 1 arg])
	], [
	AC_MSG_RESULT(no)
	])
	])

	dnl #
	dnl # 4.13 API change
	dnl # The bio->bi_error field was replaced with bio->bi_status which is an
	dnl # enum which describes all possible error types.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO_BI_STATUS], [
	ZFS_LINUX_TEST_SRC([bio_bi_status], [
	#include <linux/bio.h>
	], [
	struct bio bio __attribute__ ((unused));
	blk_status_t status __attribute__ ((unused)) = BLK_STS_OK;
	bio.bi_status = status;
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_BI_STATUS], [
	AC_MSG_CHECKING([whether bio->bi_status exists])
	ZFS_LINUX_TEST_RESULT([bio_bi_status], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BIO_BI_STATUS, 1, [bio->bi_status exists])
	],[
	AC_MSG_RESULT(no)
	])
	])

	dnl #
	dnl # 3.14 API change,
	dnl # Immutable biovecs. A number of fields of struct bio are moved to
	dnl # struct bvec_iter.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO_BVEC_ITER], [
	ZFS_LINUX_TEST_SRC([bio_bvec_iter], [
	#include <linux/bio.h>
	],[
	struct bio bio;
	bio.bi_iter.bi_sector = 0;
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_BVEC_ITER], [
	AC_MSG_CHECKING([whether bio has bi_iter])
	ZFS_LINUX_TEST_RESULT([bio_bvec_iter], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BIO_BVEC_ITER, 1, [bio has bi_iter])
	],[
	AC_MSG_RESULT(no)
	])
	])

	dnl #
	dnl # 4.8 API change
	dnl # The rw argument has been removed from submit_bio/submit_bio_wait.
	dnl # Callers are now expected to set bio->bi_rw instead of passing it in.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO_SUBMIT_BIO], [
	ZFS_LINUX_TEST_SRC([submit_bio], [
	#include <linux/bio.h>
	],[
	struct bio *bio = NULL;
	(void) submit_bio(bio);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_SUBMIT_BIO], [
	AC_MSG_CHECKING([whether submit_bio() wants 1 arg])
	ZFS_LINUX_TEST_RESULT([submit_bio], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_1ARG_SUBMIT_BIO, 1, [submit_bio() wants 1 arg])
	],[
	AC_MSG_RESULT(no)
	])
	])

	dnl #
	dnl # 2.6.34 API change
	dnl # current->bio_list
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO_CURRENT_BIO_LIST], [
	ZFS_LINUX_TEST_SRC([current_bio_list], [
	#include <linux/sched.h>
	], [
	current->bio_list = (struct bio_list *) NULL;
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_CURRENT_BIO_LIST], [
	AC_MSG_CHECKING([whether current->bio_list exists])
	ZFS_LINUX_TEST_RESULT([current_bio_list], [
	AC_MSG_RESULT(yes)
	],[
	ZFS_LINUX_TEST_ERROR([bio_list])
	])
	])

	dnl #
	dnl # Linux 5.5 API,
	dnl #
	dnl # The Linux 5.5 kernel updated percpu_ref_tryget() which is inlined by
	dnl # blkg_tryget() to use rcu_read_lock() instead of rcu_read_lock_sched().
	dnl # As a side effect the function was converted to GPL-only.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKG_TRYGET], [
	ZFS_LINUX_TEST_SRC([blkg_tryget], [
	#include <linux/blk-cgroup.h>
	#include <linux/bio.h>
	#include <linux/fs.h>
	],[
	struct blkcg_gq blkg __attribute__ ((unused)) = {};
	bool rc __attribute__ ((unused));
	rc = blkg_tryget(&blkg);
	], [], [ZFS_META_LICENSE])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKG_TRYGET], [
	AC_MSG_CHECKING([whether blkg_tryget() is available])
	ZFS_LINUX_TEST_RESULT([blkg_tryget], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BLKG_TRYGET, 1, [blkg_tryget() is available])

	AC_MSG_CHECKING([whether blkg_tryget() is GPL-only])
	ZFS_LINUX_TEST_RESULT([blkg_tryget_license], [
	AC_MSG_RESULT(no)
	],[
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BLKG_TRYGET_GPL_ONLY, 1,
	[blkg_tryget() GPL-only])
	])
	],[
	AC_MSG_RESULT(no)
	])
	])

	dnl #
	dnl # Linux 5.12 API,
	dnl #
	dnl # The Linux 5.12 kernel updated struct bio to create a new bi_bdev member
	dnl # and bio->bi_disk was moved to bio->bi_bdev->bd_disk
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO_BDEV_DISK], [
	ZFS_LINUX_TEST_SRC([bio_bdev_disk], [
	#include <linux/blk_types.h>
	#include <linux/blkdev.h>
	],[
	struct bio *b = NULL;
	struct gendisk *d = b->bi_bdev->bd_disk;
	blk_register_queue(d);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO_BDEV_DISK], [
	AC_MSG_CHECKING([whether bio->bi_bdev->bd_disk exists])
	ZFS_LINUX_TEST_RESULT([bio_bdev_disk], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BIO_BDEV_DISK, 1, [bio->bi_bdev->bd_disk exists])
	],[
	AC_MSG_RESULT(no)
	])
	])

	+dnl #
	+dnl # Linux 5.16 API
	+dnl #
	+dnl # The Linux 5.16 API for submit_bio changed the return type to be
	+dnl # void instead of int
	+dnl #
	+AC_DEFUN([ZFS_AC_KERNEL_SRC_BDEV_SUBMIT_BIO_RETURNS_VOID], [
	+ ZFS_LINUX_TEST_SRC([bio_bdev_submit_bio_void], [
	+ #include <linux/blkdev.h>
	+ ],[
	+ struct block_device_operations *bdev = NULL;
	+ __attribute__((unused)) void(f)(struct bio ) = bdev->submit_bio;
	+ ])
	+])
	+
	+AC_DEFUN([ZFS_AC_KERNEL_BDEV_SUBMIT_BIO_RETURNS_VOID], [
	+ AC_MSG_CHECKING(
	+ [whether block_device_operations->submit_bio() returns void])
	+ ZFS_LINUX_TEST_RESULT([bio_bdev_submit_bio_void], [
	+ AC_MSG_RESULT(yes)
	+ AC_DEFINE(HAVE_BDEV_SUBMIT_BIO_RETURNS_VOID, 1,
	+ [block_device_operations->submit_bio() returns void])
	+ ],[
	+ AC_MSG_RESULT(no)
	+ ])
	+])
	+
	+dnl #
	+dnl # Linux 5.16 API
	+dnl #
	+dnl # The Linux 5.16 API moved struct blkcg_gq into linux/blk-cgroup.h, which
	+dnl # has been around since 2015. This test looks for the presence of that
	+dnl # header, so that it can be conditionally included where it exists, but
	+dnl # still be backward compatible with kernels that pre-date its introduction.
	+dnl #
	+AC_DEFUN([ZFS_AC_KERNEL_SRC_BLK_CGROUP_HEADER], [
	+ ZFS_LINUX_TEST_SRC([blk_cgroup_header], [
	+ #include <linux/blk-cgroup.h>
	+ ], [])
	+])
	+
	+AC_DEFUN([ZFS_AC_KERNEL_BLK_CGROUP_HEADER], [
	+ AC_MSG_CHECKING([for existence of linux/blk-cgroup.h])
	+ ZFS_LINUX_TEST_RESULT([blk_cgroup_header],[
	+ AC_MSG_RESULT(yes)
	+ AC_DEFINE(HAVE_LINUX_BLK_CGROUP_HEADER, 1,
	+ [linux/blk-cgroup.h exists])
	+ ],[
	+ AC_MSG_RESULT(no)
	+ ])
	+])
	+
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BIO], [
	ZFS_AC_KERNEL_SRC_REQ
	ZFS_AC_KERNEL_SRC_BIO_OPS
	ZFS_AC_KERNEL_SRC_BIO_SET_DEV
	ZFS_AC_KERNEL_SRC_BIO_END_IO_T_ARGS
	ZFS_AC_KERNEL_SRC_BIO_BI_STATUS
	ZFS_AC_KERNEL_SRC_BIO_BVEC_ITER
	ZFS_AC_KERNEL_SRC_BIO_SUBMIT_BIO
	ZFS_AC_KERNEL_SRC_BIO_CURRENT_BIO_LIST
	ZFS_AC_KERNEL_SRC_BLKG_TRYGET
	ZFS_AC_KERNEL_SRC_BIO_BDEV_DISK
	+ ZFS_AC_KERNEL_SRC_BDEV_SUBMIT_BIO_RETURNS_VOID
	+ ZFS_AC_KERNEL_SRC_BIO_SET_DEV_MACRO
	+ ZFS_AC_KERNEL_SRC_BLK_CGROUP_HEADER
	])

	AC_DEFUN([ZFS_AC_KERNEL_BIO], [
	ZFS_AC_KERNEL_BIO_REQ_FAILFAST_MASK
	ZFS_AC_KERNEL_BIO_REQ_DISCARD
	ZFS_AC_KERNEL_BIO_REQ_FLUSH
	ZFS_AC_KERNEL_BIO_REQ_PREFLUSH

	ZFS_AC_KERNEL_BIO_REQ_OP_DISCARD
	ZFS_AC_KERNEL_BIO_REQ_OP_SECURE_ERASE
	ZFS_AC_KERNEL_BIO_REQ_OP_FLUSH
	ZFS_AC_KERNEL_BIO_BI_OPF
	ZFS_AC_KERNEL_BIO_SET_OP_ATTRS

	ZFS_AC_KERNEL_BIO_SET_DEV
	ZFS_AC_KERNEL_BIO_END_IO_T_ARGS
	ZFS_AC_KERNEL_BIO_BI_STATUS
	ZFS_AC_KERNEL_BIO_BVEC_ITER
	ZFS_AC_KERNEL_BIO_SUBMIT_BIO
	ZFS_AC_KERNEL_BIO_CURRENT_BIO_LIST
	ZFS_AC_KERNEL_BLKG_TRYGET
	ZFS_AC_KERNEL_BIO_BDEV_DISK
	+ ZFS_AC_KERNEL_BDEV_SUBMIT_BIO_RETURNS_VOID
	+ ZFS_AC_KERNEL_BLK_CGROUP_HEADER
	])
	diff --git a/sys/contrib/openzfs/config/kernel-blkdev.m4 b/sys/contrib/openzfs/config/kernel-blkdev.m4
	index 61e66421f8ec..9c60e5dd4210 100644
	--- a/sys/contrib/openzfs/config/kernel-blkdev.m4
	+++ b/sys/contrib/openzfs/config/kernel-blkdev.m4
	@@ -1,321 +1,343 @@
	dnl #
	dnl # 2.6.38 API change,
	dnl # Added blkdev_get_by_path()
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_GET_BY_PATH], [
	ZFS_LINUX_TEST_SRC([blkdev_get_by_path], [
	#include <linux/fs.h>
	#include <linux/blkdev.h>
	], [
	struct block_device *bdev __attribute__ ((unused)) = NULL;
	const char *path = "path";
	fmode_t mode = 0;
	void *holder = NULL;

	bdev = blkdev_get_by_path(path, mode, holder);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_GET_BY_PATH], [
	AC_MSG_CHECKING([whether blkdev_get_by_path() exists])
	ZFS_LINUX_TEST_RESULT([blkdev_get_by_path], [
	AC_MSG_RESULT(yes)
	], [
	ZFS_LINUX_TEST_ERROR([blkdev_get_by_path()])
	])
	])

	dnl #
	dnl # 2.6.38 API change,
	dnl # Added blkdev_put()
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_PUT], [
	ZFS_LINUX_TEST_SRC([blkdev_put], [
	#include <linux/fs.h>
	#include <linux/blkdev.h>
	], [
	struct block_device *bdev = NULL;
	fmode_t mode = 0;

	blkdev_put(bdev, mode);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_PUT], [
	AC_MSG_CHECKING([whether blkdev_put() exists])
	ZFS_LINUX_TEST_RESULT([blkdev_put], [
	AC_MSG_RESULT(yes)
	], [
	ZFS_LINUX_TEST_ERROR([blkdev_put()])
	])
	])

	dnl #
	dnl # 4.1 API, exported blkdev_reread_part() symbol, back ported to the
	dnl # 3.10.0 CentOS 7.x enterprise kernels.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_REREAD_PART], [
	ZFS_LINUX_TEST_SRC([blkdev_reread_part], [
	#include <linux/fs.h>
	#include <linux/blkdev.h>
	], [
	struct block_device *bdev = NULL;
	int error;

	error = blkdev_reread_part(bdev);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_REREAD_PART], [
	AC_MSG_CHECKING([whether blkdev_reread_part() exists])
	ZFS_LINUX_TEST_RESULT([blkdev_reread_part], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BLKDEV_REREAD_PART, 1,
	[blkdev_reread_part() exists])
	], [
	AC_MSG_RESULT(no)
	])
	])

	dnl #
	dnl # check_disk_change() was removed in 5.10
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_CHECK_DISK_CHANGE], [
	ZFS_LINUX_TEST_SRC([check_disk_change], [
	#include <linux/fs.h>
	#include <linux/blkdev.h>
	], [
	struct block_device *bdev = NULL;
	bool error;

	error = check_disk_change(bdev);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_CHECK_DISK_CHANGE], [
	AC_MSG_CHECKING([whether check_disk_change() exists])
	ZFS_LINUX_TEST_RESULT([check_disk_change], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_CHECK_DISK_CHANGE, 1,
	[check_disk_change() exists])
	], [
	AC_MSG_RESULT(no)
	])
	])

	dnl #
	dnl # 5.10 API, check_disk_change() is removed, in favor of
	dnl # bdev_check_media_change(), which doesn't force revalidation
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_CHECK_MEDIA_CHANGE], [
	ZFS_LINUX_TEST_SRC([bdev_check_media_change], [
	#include <linux/fs.h>
	#include <linux/blkdev.h>
	], [
	struct block_device *bdev = NULL;
	int error;

	error = bdev_check_media_change(bdev);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEV_CHECK_MEDIA_CHANGE], [
	AC_MSG_CHECKING([whether bdev_check_media_change() exists])
	ZFS_LINUX_TEST_RESULT([bdev_check_media_change], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BDEV_CHECK_MEDIA_CHANGE, 1,
	[bdev_check_media_change() exists])
	], [
	AC_MSG_RESULT(no)
	])
	])

	dnl #
	dnl # 2.6.22 API change
	dnl # Single argument invalidate_bdev()
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_INVALIDATE_BDEV], [
	ZFS_LINUX_TEST_SRC([invalidate_bdev], [
	#include <linux/buffer_head.h>
	#include <linux/blkdev.h>
	],[
	struct block_device *bdev = NULL;
	invalidate_bdev(bdev);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_INVALIDATE_BDEV], [
	AC_MSG_CHECKING([whether invalidate_bdev() exists])
	ZFS_LINUX_TEST_RESULT([invalidate_bdev], [
	AC_MSG_RESULT(yes)
	],[
	ZFS_LINUX_TEST_ERROR([invalidate_bdev()])
	])
	])

	dnl #
	dnl # 5.11 API, lookup_bdev() takes dev_t argument.
	dnl # 2.6.27 API, lookup_bdev() was first exported.
	dnl # 4.4.0-6.21 API, lookup_bdev() on Ubuntu takes mode argument.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_LOOKUP_BDEV], [
	ZFS_LINUX_TEST_SRC([lookup_bdev_devt], [
	#include <linux/blkdev.h>
	], [
	int error __attribute__ ((unused));
	const char path[] = "/example/path";
	dev_t dev;

	error = lookup_bdev(path, &dev);
	])

	ZFS_LINUX_TEST_SRC([lookup_bdev_1arg], [
	#include <linux/fs.h>
	#include <linux/blkdev.h>
	], [
	struct block_device *bdev __attribute__ ((unused));
	const char path[] = "/example/path";

	bdev = lookup_bdev(path);
	])

	ZFS_LINUX_TEST_SRC([lookup_bdev_mode], [
	#include <linux/fs.h>
	], [
	struct block_device *bdev __attribute__ ((unused));
	const char path[] = "/example/path";

	bdev = lookup_bdev(path, FMODE_READ);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_LOOKUP_BDEV], [
	AC_MSG_CHECKING([whether lookup_bdev() wants dev_t arg])
	ZFS_LINUX_TEST_RESULT_SYMBOL([lookup_bdev_devt],
	[lookup_bdev], [fs/block_dev.c], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_DEVT_LOOKUP_BDEV, 1,
	[lookup_bdev() wants dev_t arg])
	], [
	AC_MSG_RESULT(no)

	AC_MSG_CHECKING([whether lookup_bdev() wants 1 arg])
	ZFS_LINUX_TEST_RESULT_SYMBOL([lookup_bdev_1arg],
	[lookup_bdev], [fs/block_dev.c], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_1ARG_LOOKUP_BDEV, 1,
	[lookup_bdev() wants 1 arg])
	], [
	AC_MSG_RESULT(no)

	AC_MSG_CHECKING([whether lookup_bdev() wants mode arg])
	ZFS_LINUX_TEST_RESULT_SYMBOL([lookup_bdev_mode],
	[lookup_bdev], [fs/block_dev.c], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_MODE_LOOKUP_BDEV, 1,
	[lookup_bdev() wants mode arg])
	], [
	ZFS_LINUX_TEST_ERROR([lookup_bdev()])
	])
	])
	])
	])

	dnl #
	dnl # 2.6.30 API change
	dnl #
	dnl # The bdev_physical_block_size() interface was added to provide a way
	dnl # to determine the smallest write which can be performed without a
	dnl # read-modify-write operation.
	dnl #
	dnl # Unfortunately, this interface isn't entirely reliable because
	dnl # drives are sometimes known to misreport this value.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_PHYSICAL_BLOCK_SIZE], [
	ZFS_LINUX_TEST_SRC([bdev_physical_block_size], [
	#include <linux/blkdev.h>
	],[
	struct block_device *bdev __attribute__ ((unused)) = NULL;
	bdev_physical_block_size(bdev);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEV_PHYSICAL_BLOCK_SIZE], [
	AC_MSG_CHECKING([whether bdev_physical_block_size() is available])
	ZFS_LINUX_TEST_RESULT([bdev_physical_block_size], [
	AC_MSG_RESULT(yes)
	],[
	ZFS_LINUX_TEST_ERROR([bdev_physical_block_size()])
	])
	])

	dnl #
	dnl # 2.6.30 API change
	dnl # Added bdev_logical_block_size().
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_LOGICAL_BLOCK_SIZE], [
	ZFS_LINUX_TEST_SRC([bdev_logical_block_size], [
	#include <linux/blkdev.h>
	],[
	struct block_device *bdev __attribute__ ((unused)) = NULL;
	bdev_logical_block_size(bdev);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEV_LOGICAL_BLOCK_SIZE], [
	AC_MSG_CHECKING([whether bdev_logical_block_size() is available])
	ZFS_LINUX_TEST_RESULT([bdev_logical_block_size], [
	AC_MSG_RESULT(yes)
	],[
	ZFS_LINUX_TEST_ERROR([bdev_logical_block_size()])
	])
	])

	dnl #
	dnl # 5.11 API change
	dnl # Added bdev_whole() helper.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_WHOLE], [
	ZFS_LINUX_TEST_SRC([bdev_whole], [
	#include <linux/blkdev.h>
	],[
	struct block_device *bdev = NULL;
	bdev = bdev_whole(bdev);
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_BDEV_WHOLE], [
	AC_MSG_CHECKING([whether bdev_whole() is available])
	ZFS_LINUX_TEST_RESULT([bdev_whole], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_BDEV_WHOLE, 1, [bdev_whole() is available])
	],[
	AC_MSG_RESULT(no)
	])
	])

	+dnl #
	+dnl # 5.13 API change
	+dnl # blkdev_get_by_path() no longer handles ERESTARTSYS
	+dnl #
	+dnl # Unfortunately we're forced to rely solely on the kernel version
	+dnl # number in order to determine the expected behavior. This was an
	+dnl # internal change to blkdev_get_by_dev(), see commit a8ed1a0607.
	+dnl #
	+AC_DEFUN([ZFS_AC_KERNEL_BLKDEV_GET_ERESTARTSYS], [
	+ AC_MSG_CHECKING([whether blkdev_get_by_path() handles ERESTARTSYS])
	+ AS_VERSION_COMPARE([$LINUX_VERSION], [5.13.0], [
	+ AC_MSG_RESULT(yes)
	+ AC_DEFINE(HAVE_BLKDEV_GET_ERESTARTSYS, 1,
	+ [blkdev_get_by_path() handles ERESTARTSYS])
	+ ],[
	+ AC_MSG_RESULT(no)
	+ ],[
	+ AC_MSG_RESULT(no)
	+ ])
	+])
	+
	AC_DEFUN([ZFS_AC_KERNEL_SRC_BLKDEV], [
	ZFS_AC_KERNEL_SRC_BLKDEV_GET_BY_PATH
	ZFS_AC_KERNEL_SRC_BLKDEV_PUT
	ZFS_AC_KERNEL_SRC_BLKDEV_REREAD_PART
	ZFS_AC_KERNEL_SRC_BLKDEV_INVALIDATE_BDEV
	ZFS_AC_KERNEL_SRC_BLKDEV_LOOKUP_BDEV
	ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_LOGICAL_BLOCK_SIZE
	ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_PHYSICAL_BLOCK_SIZE
	ZFS_AC_KERNEL_SRC_BLKDEV_CHECK_DISK_CHANGE
	ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_CHECK_MEDIA_CHANGE
	ZFS_AC_KERNEL_SRC_BLKDEV_BDEV_WHOLE
	])

	AC_DEFUN([ZFS_AC_KERNEL_BLKDEV], [
	ZFS_AC_KERNEL_BLKDEV_GET_BY_PATH
	ZFS_AC_KERNEL_BLKDEV_PUT
	ZFS_AC_KERNEL_BLKDEV_REREAD_PART
	ZFS_AC_KERNEL_BLKDEV_INVALIDATE_BDEV
	ZFS_AC_KERNEL_BLKDEV_LOOKUP_BDEV
	ZFS_AC_KERNEL_BLKDEV_BDEV_LOGICAL_BLOCK_SIZE
	ZFS_AC_KERNEL_BLKDEV_BDEV_PHYSICAL_BLOCK_SIZE
	ZFS_AC_KERNEL_BLKDEV_CHECK_DISK_CHANGE
	ZFS_AC_KERNEL_BLKDEV_BDEV_CHECK_MEDIA_CHANGE
	ZFS_AC_KERNEL_BLKDEV_BDEV_WHOLE
	+ ZFS_AC_KERNEL_BLKDEV_GET_ERESTARTSYS
	])
	diff --git a/sys/contrib/openzfs/config/kernel-fpu.m4 b/sys/contrib/openzfs/config/kernel-fpu.m4
	index 4d6fe052289c..faa64f1ec46b 100644
	--- a/sys/contrib/openzfs/config/kernel-fpu.m4
	+++ b/sys/contrib/openzfs/config/kernel-fpu.m4
	@@ -1,131 +1,146 @@
	dnl #
	dnl # Handle differences in kernel FPU code.
	dnl #
	dnl # Kernel
	+dnl # 5.16: XCR code put into asm/fpu/xcr.h
	+dnl # HAVE_KERNEL_FPU_XCR_HEADER
	+dnl #
	dnl # 5.0: Wrappers have been introduced to save/restore the FPU state.
	dnl # This change was made to the 4.19.38 and 4.14.120 LTS kernels.
	dnl # HAVE_KERNEL_FPU_INTERNAL
	dnl #
	dnl # 4.2: Use __kernel_fpu_{begin,end}()
	dnl # HAVE_UNDERSCORE_KERNEL_FPU & KERNEL_EXPORTS_X86_FPU
	dnl #
	dnl # Pre-4.2: Use kernel_fpu_{begin,end}()
	dnl # HAVE_KERNEL_FPU & KERNEL_EXPORTS_X86_FPU
	dnl #
	dnl # N.B. The header check is performed before all other checks since it
	dnl # depends on HAVE_KERNEL_FPU_API_HEADER being set in confdefs.h.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_FPU_HEADER], [
	AC_MSG_CHECKING([whether fpu headers are available])
	ZFS_LINUX_TRY_COMPILE([
	#include <linux/module.h>
	#include <asm/fpu/api.h>
	],[
	],[
	AC_DEFINE(HAVE_KERNEL_FPU_API_HEADER, 1,
	[kernel has asm/fpu/api.h])
	AC_MSG_RESULT(asm/fpu/api.h)
	+ AC_MSG_CHECKING([whether fpu/xcr header is available])
	+ ZFS_LINUX_TRY_COMPILE([
	+ #include <linux/module.h>
	+ #include <asm/fpu/xcr.h>
	+ ],[
	+ ],[
	+ AC_DEFINE(HAVE_KERNEL_FPU_XCR_HEADER, 1,
	+ [kernel has asm/fpu/xcr.h])
	+ AC_MSG_RESULT(asm/fpu/xcr.h)
	+ ],[
	+ AC_MSG_RESULT(no asm/fpu/xcr.h)
	+ ])
	],[
	AC_MSG_RESULT(i387.h & xcr.h)
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_SRC_FPU], [
	ZFS_LINUX_TEST_SRC([kernel_fpu], [
	#include <linux/types.h>
	#ifdef HAVE_KERNEL_FPU_API_HEADER
	#include <asm/fpu/api.h>
	#else
	#include <asm/i387.h>
	#include <asm/xcr.h>
	#endif
	], [
	kernel_fpu_begin();
	kernel_fpu_end();
	], [], [ZFS_META_LICENSE])

	ZFS_LINUX_TEST_SRC([__kernel_fpu], [
	#include <linux/types.h>
	#ifdef HAVE_KERNEL_FPU_API_HEADER
	#include <asm/fpu/api.h>
	#else
	#include <asm/i387.h>
	#include <asm/xcr.h>
	#endif
	], [
	__kernel_fpu_begin();
	__kernel_fpu_end();
	], [], [ZFS_META_LICENSE])

	ZFS_LINUX_TEST_SRC([fpu_internal], [
	#if defined(__x86_64) \|\| defined(__x86_64__) \|\| \
	defined(__i386) \|\| defined(__i386__)
	#if !defined(__x86)
	#define __x86
	#endif
	#endif

	#if !defined(__x86)
	#error Unsupported architecture
	#endif

	#include <linux/types.h>
	#ifdef HAVE_KERNEL_FPU_API_HEADER
	#include <asm/fpu/api.h>
	#include <asm/fpu/internal.h>
	#else
	#include <asm/i387.h>
	#include <asm/xcr.h>
	#endif

	#if !defined(XSTATE_XSAVE)
	#error XSTATE_XSAVE not defined
	#endif

	#if !defined(XSTATE_XRESTORE)
	#error XSTATE_XRESTORE not defined
	#endif
	],[
	struct fpu *fpu = &current->thread.fpu;
	union fpregs_state *st = &fpu->state;
	struct fregs_state *fr __attribute__ ((unused)) = &st->fsave;
	struct fxregs_state *fxr __attribute__ ((unused)) = &st->fxsave;
	struct xregs_state *xr __attribute__ ((unused)) = &st->xsave;
	])
	])

	AC_DEFUN([ZFS_AC_KERNEL_FPU], [
	dnl #
	dnl # Legacy kernel
	dnl #
	AC_MSG_CHECKING([whether kernel fpu is available])
	ZFS_LINUX_TEST_RESULT_SYMBOL([kernel_fpu_license],
	[kernel_fpu_begin], [arch/x86/kernel/fpu/core.c], [
	AC_MSG_RESULT(kernel_fpu_*)
	AC_DEFINE(HAVE_KERNEL_FPU, 1,
	[kernel has kernel_fpu_* functions])
	AC_DEFINE(KERNEL_EXPORTS_X86_FPU, 1,
	[kernel exports FPU functions])
	],[
	dnl #
	dnl # Linux 4.2 kernel
	dnl #
	ZFS_LINUX_TEST_RESULT_SYMBOL([__kernel_fpu_license],
	[__kernel_fpu_begin],
	[arch/x86/kernel/fpu/core.c arch/x86/kernel/i387.c], [
	AC_MSG_RESULT(__kernel_fpu_*)
	AC_DEFINE(HAVE_UNDERSCORE_KERNEL_FPU, 1,
	[kernel has __kernel_fpu_* functions])
	AC_DEFINE(KERNEL_EXPORTS_X86_FPU, 1,
	[kernel exports FPU functions])
	],[
	ZFS_LINUX_TEST_RESULT([fpu_internal], [
	AC_MSG_RESULT(internal)
	AC_DEFINE(HAVE_KERNEL_FPU_INTERNAL, 1,
	[kernel fpu internal])
	],[
	AC_MSG_RESULT(unavailable)
	])
	])
	])
	])
	diff --git a/sys/contrib/openzfs/config/kernel-pagemap-folio_wait_bit.m4 b/sys/contrib/openzfs/config/kernel-pagemap-folio_wait_bit.m4
	new file mode 100644
	index 000000000000..e0aaa4a57411
	--- /dev/null
	+++ b/sys/contrib/openzfs/config/kernel-pagemap-folio_wait_bit.m4
	@@ -0,0 +1,26 @@
	+dnl #
	+dnl # Linux 5.16 no longer allows directly calling wait_on_page_bit, and
	+dnl # instead requires you to call folio-specific functions. In this case,
	+dnl # wait_on_page_bit(pg, PG_writeback) becomes
	+dnl # folio_wait_bit(pg, PG_writeback)
	+dnl #
	+AC_DEFUN([ZFS_AC_KERNEL_SRC_PAGEMAP_FOLIO_WAIT_BIT], [
	+ ZFS_LINUX_TEST_SRC([pagemap_has_folio_wait_bit], [
	+ #include <linux/pagemap.h>
	+ ],[
	+ static struct folio *f = NULL;
	+
	+ folio_wait_bit(f, PG_writeback);
	+ ])
	+])
	+
	+AC_DEFUN([ZFS_AC_KERNEL_PAGEMAP_FOLIO_WAIT_BIT], [
	+ AC_MSG_CHECKING([folio_wait_bit() exists])
	+ ZFS_LINUX_TEST_RESULT([pagemap_has_folio_wait_bit], [
	+ AC_MSG_RESULT([yes])
	+ AC_DEFINE(HAVE_PAGEMAP_FOLIO_WAIT_BIT, 1,
	+ [folio_wait_bit() exists])
	+ ],[
	+ AC_MSG_RESULT([no])
	+ ])
	+])
	diff --git a/sys/contrib/openzfs/config/kernel-vfs-iov_iter.m4 b/sys/contrib/openzfs/config/kernel-vfs-iov_iter.m4
	index bee6d0be9666..ecdda939f1cf 100644
	--- a/sys/contrib/openzfs/config/kernel-vfs-iov_iter.m4
	+++ b/sys/contrib/openzfs/config/kernel-vfs-iov_iter.m4
	@@ -1,162 +1,184 @@
	dnl #
	dnl # Check for available iov_iter functionality.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_SRC_VFS_IOV_ITER], [
	ZFS_LINUX_TEST_SRC([iov_iter_types], [
	#include <linux/fs.h>
	#include <linux/uio.h>
	],[
	int type __attribute__ ((unused)) =
	ITER_IOVEC \| ITER_KVEC \| ITER_BVEC \| ITER_PIPE;
	])

	ZFS_LINUX_TEST_SRC([iov_iter_advance], [
	#include <linux/fs.h>
	#include <linux/uio.h>
	],[
	struct iov_iter iter = { 0 };
	size_t advance = 512;

	iov_iter_advance(&iter, advance);
	])

	ZFS_LINUX_TEST_SRC([iov_iter_revert], [
	#include <linux/fs.h>
	#include <linux/uio.h>
	],[
	struct iov_iter iter = { 0 };
	size_t revert = 512;

	iov_iter_revert(&iter, revert);
	])

	ZFS_LINUX_TEST_SRC([iov_iter_fault_in_readable], [
	#include <linux/fs.h>
	#include <linux/uio.h>
	],[
	struct iov_iter iter = { 0 };
	size_t size = 512;
	int error __attribute__ ((unused));

	error = iov_iter_fault_in_readable(&iter, size);
	])

	ZFS_LINUX_TEST_SRC([iov_iter_count], [
	#include <linux/fs.h>
	#include <linux/uio.h>
	],[
	struct iov_iter iter = { 0 };
	size_t bytes __attribute__ ((unused));

	bytes = iov_iter_count(&iter);
	])

	ZFS_LINUX_TEST_SRC([copy_to_iter], [
	#include <linux/fs.h>
	#include <linux/uio.h>
	],[
	struct iov_iter iter = { 0 };
	char buf[512] = { 0 };
	size_t size = 512;
	size_t bytes __attribute__ ((unused));

	bytes = copy_to_iter((const void *)&buf, size, &iter);
	])

	ZFS_LINUX_TEST_SRC([copy_from_iter], [
	#include <linux/fs.h>
	#include <linux/uio.h>
	],[
	struct iov_iter iter = { 0 };
	char buf[512] = { 0 };
	size_t size = 512;
	size_t bytes __attribute__ ((unused));

	bytes = copy_from_iter((void *)&buf, size, &iter);
	])
	+
	+ ZFS_LINUX_TEST_SRC([iov_iter_type], [
	+ #include <linux/fs.h>
	+ #include <linux/uio.h>
	+ ],[
	+ struct iov_iter iter = { 0 };
	+ __attribute__((unused)) enum iter_type i = iov_iter_type(&iter);
	+ ])
	])

	AC_DEFUN([ZFS_AC_KERNEL_VFS_IOV_ITER], [
	enable_vfs_iov_iter="yes"

	AC_MSG_CHECKING([whether iov_iter types are available])
	ZFS_LINUX_TEST_RESULT([iov_iter_types], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_IOV_ITER_TYPES, 1,
	[iov_iter types are available])
	],[
	AC_MSG_RESULT(no)
	enable_vfs_iov_iter="no"
	])

	AC_MSG_CHECKING([whether iov_iter_advance() is available])
	ZFS_LINUX_TEST_RESULT([iov_iter_advance], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_IOV_ITER_ADVANCE, 1,
	[iov_iter_advance() is available])
	],[
	AC_MSG_RESULT(no)
	enable_vfs_iov_iter="no"
	])

	AC_MSG_CHECKING([whether iov_iter_revert() is available])
	ZFS_LINUX_TEST_RESULT([iov_iter_revert], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_IOV_ITER_REVERT, 1,
	[iov_iter_revert() is available])
	],[
	AC_MSG_RESULT(no)
	enable_vfs_iov_iter="no"
	])

	AC_MSG_CHECKING([whether iov_iter_fault_in_readable() is available])
	ZFS_LINUX_TEST_RESULT([iov_iter_fault_in_readable], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_IOV_ITER_FAULT_IN_READABLE, 1,
	[iov_iter_fault_in_readable() is available])
	],[
	AC_MSG_RESULT(no)
	enable_vfs_iov_iter="no"
	])

	AC_MSG_CHECKING([whether iov_iter_count() is available])
	ZFS_LINUX_TEST_RESULT([iov_iter_count], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_IOV_ITER_COUNT, 1,
	[iov_iter_count() is available])
	],[
	AC_MSG_RESULT(no)
	enable_vfs_iov_iter="no"
	])

	AC_MSG_CHECKING([whether copy_to_iter() is available])
	ZFS_LINUX_TEST_RESULT([copy_to_iter], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_COPY_TO_ITER, 1,
	[copy_to_iter() is available])
	],[
	AC_MSG_RESULT(no)
	enable_vfs_iov_iter="no"
	])

	AC_MSG_CHECKING([whether copy_from_iter() is available])
	ZFS_LINUX_TEST_RESULT([copy_from_iter], [
	AC_MSG_RESULT(yes)
	AC_DEFINE(HAVE_COPY_FROM_ITER, 1,
	[copy_from_iter() is available])
	],[
	AC_MSG_RESULT(no)
	enable_vfs_iov_iter="no"
	])

	+ dnl #
	+ dnl # This checks for iov_iter_type() in linux/uio.h. It is not
	+ dnl # required, however, and the module will compiled without it
	+ dnl # using direct access of the member attribute
	+ dnl #
	+ AC_MSG_CHECKING([whether iov_iter_type() is available])
	+ ZFS_LINUX_TEST_RESULT([iov_iter_type], [
	+ AC_MSG_RESULT(yes)
	+ AC_DEFINE(HAVE_IOV_ITER_TYPE, 1,
	+ [iov_iter_type() is available])
	+ ],[
	+ AC_MSG_RESULT(no)
	+ ])
	+
	dnl #
	dnl # As of the 4.9 kernel support is provided for iovecs, kvecs,
	dnl # bvecs and pipes in the iov_iter structure. As long as the
	dnl # other support interfaces are all available the iov_iter can
	dnl # be correctly used in the uio structure.
	dnl #
	AS_IF([test "x$enable_vfs_iov_iter" = "xyes"], [
	AC_DEFINE(HAVE_VFS_IOV_ITER, 1,
	[All required iov_iter interfaces are available])
	])
	])
	diff --git a/sys/contrib/openzfs/config/kernel.m4 b/sys/contrib/openzfs/config/kernel.m4
	index 0b94f3bd9cb6..bdd3caed2b3d 100644
	--- a/sys/contrib/openzfs/config/kernel.m4
	+++ b/sys/contrib/openzfs/config/kernel.m4
	@@ -1,891 +1,893 @@
	dnl #
	dnl # Default ZFS kernel configuration
	dnl #
	AC_DEFUN([ZFS_AC_CONFIG_KERNEL], [
	AM_COND_IF([BUILD_LINUX], [
	dnl # Setup the kernel build environment.
	ZFS_AC_KERNEL
	ZFS_AC_QAT

	dnl # Sanity checks for module building and CONFIG_* defines
	ZFS_AC_KERNEL_TEST_MODULE
	ZFS_AC_KERNEL_CONFIG_DEFINED

	dnl # Sequential ZFS_LINUX_TRY_COMPILE tests
	ZFS_AC_KERNEL_FPU_HEADER
	ZFS_AC_KERNEL_OBJTOOL_HEADER
	ZFS_AC_KERNEL_WAIT_QUEUE_ENTRY_T
	ZFS_AC_KERNEL_MISC_MINOR
	ZFS_AC_KERNEL_DECLARE_EVENT_CLASS

	dnl # Parallel ZFS_LINUX_TEST_SRC / ZFS_LINUX_TEST_RESULT tests
	ZFS_AC_KERNEL_TEST_SRC
	ZFS_AC_KERNEL_TEST_RESULT

	AS_IF([test "$LINUX_OBJ" != "$LINUX"], [
	KERNEL_MAKE="$KERNEL_MAKE O=$LINUX_OBJ"
	])

	AC_SUBST(KERNEL_MAKE)
	])
	])

	dnl #
	dnl # Generate and compile all of the kernel API test cases to determine
	dnl # which interfaces are available. By invoking the kernel build system
	dnl # only once the compilation can be done in parallel significantly
	dnl # speeding up the process.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_TEST_SRC], [
	ZFS_AC_KERNEL_SRC_OBJTOOL
	ZFS_AC_KERNEL_SRC_GLOBAL_PAGE_STATE
	ZFS_AC_KERNEL_SRC_ACCESS_OK_TYPE
	ZFS_AC_KERNEL_SRC_PDE_DATA
	ZFS_AC_KERNEL_SRC_FALLOCATE
	ZFS_AC_KERNEL_SRC_2ARGS_ZLIB_DEFLATE_WORKSPACESIZE
	ZFS_AC_KERNEL_SRC_RWSEM
	ZFS_AC_KERNEL_SRC_SCHED
	ZFS_AC_KERNEL_SRC_USLEEP_RANGE
	ZFS_AC_KERNEL_SRC_KMEM_CACHE
	ZFS_AC_KERNEL_SRC_KVMALLOC
	ZFS_AC_KERNEL_SRC_VMALLOC_PAGE_KERNEL
	ZFS_AC_KERNEL_SRC_WAIT
	ZFS_AC_KERNEL_SRC_INODE_TIMES
	ZFS_AC_KERNEL_SRC_INODE_LOCK
	ZFS_AC_KERNEL_SRC_GROUP_INFO_GID
	ZFS_AC_KERNEL_SRC_RW
	ZFS_AC_KERNEL_SRC_TIMER_SETUP
	ZFS_AC_KERNEL_SRC_SUPER_USER_NS
	ZFS_AC_KERNEL_SRC_PROC_OPERATIONS
	ZFS_AC_KERNEL_SRC_BLOCK_DEVICE_OPERATIONS
	ZFS_AC_KERNEL_SRC_BIO
	ZFS_AC_KERNEL_SRC_BLKDEV
	ZFS_AC_KERNEL_SRC_BLK_QUEUE
	ZFS_AC_KERNEL_SRC_REVALIDATE_DISK
	ZFS_AC_KERNEL_SRC_GET_DISK_RO
	ZFS_AC_KERNEL_SRC_GENERIC_READLINK_GLOBAL
	ZFS_AC_KERNEL_SRC_DISCARD_GRANULARITY
	ZFS_AC_KERNEL_SRC_INODE_OWNER_OR_CAPABLE
	ZFS_AC_KERNEL_SRC_XATTR
	ZFS_AC_KERNEL_SRC_ACL
	ZFS_AC_KERNEL_SRC_INODE_GETATTR
	ZFS_AC_KERNEL_SRC_INODE_SET_FLAGS
	ZFS_AC_KERNEL_SRC_INODE_SET_IVERSION
	ZFS_AC_KERNEL_SRC_SHOW_OPTIONS
	ZFS_AC_KERNEL_SRC_FILE_INODE
	ZFS_AC_KERNEL_SRC_FILE_DENTRY
	ZFS_AC_KERNEL_SRC_FSYNC
	ZFS_AC_KERNEL_SRC_AIO_FSYNC
	ZFS_AC_KERNEL_SRC_EVICT_INODE
	ZFS_AC_KERNEL_SRC_DIRTY_INODE
	ZFS_AC_KERNEL_SRC_SHRINKER
	ZFS_AC_KERNEL_SRC_MKDIR
	ZFS_AC_KERNEL_SRC_LOOKUP_FLAGS
	ZFS_AC_KERNEL_SRC_CREATE
	ZFS_AC_KERNEL_SRC_GET_LINK
	ZFS_AC_KERNEL_SRC_PUT_LINK
	ZFS_AC_KERNEL_SRC_TMPFILE
	ZFS_AC_KERNEL_SRC_AUTOMOUNT
	ZFS_AC_KERNEL_SRC_ENCODE_FH_WITH_INODE
	ZFS_AC_KERNEL_SRC_COMMIT_METADATA
	ZFS_AC_KERNEL_SRC_CLEAR_INODE
	ZFS_AC_KERNEL_SRC_SETATTR_PREPARE
	ZFS_AC_KERNEL_SRC_INSERT_INODE_LOCKED
	ZFS_AC_KERNEL_SRC_DENTRY
	ZFS_AC_KERNEL_SRC_TRUNCATE_SETSIZE
	ZFS_AC_KERNEL_SRC_SECURITY_INODE
	ZFS_AC_KERNEL_SRC_FST_MOUNT
	ZFS_AC_KERNEL_SRC_BDI
	ZFS_AC_KERNEL_SRC_SET_NLINK
	ZFS_AC_KERNEL_SRC_SGET
	ZFS_AC_KERNEL_SRC_LSEEK_EXECUTE
	ZFS_AC_KERNEL_SRC_VFS_GETATTR
	ZFS_AC_KERNEL_SRC_VFS_FSYNC_2ARGS
	ZFS_AC_KERNEL_SRC_VFS_ITERATE
	ZFS_AC_KERNEL_SRC_VFS_DIRECT_IO
	ZFS_AC_KERNEL_SRC_VFS_RW_ITERATE
	ZFS_AC_KERNEL_SRC_VFS_GENERIC_WRITE_CHECKS
	ZFS_AC_KERNEL_SRC_VFS_IOV_ITER
	ZFS_AC_KERNEL_SRC_KMAP_ATOMIC_ARGS
	ZFS_AC_KERNEL_SRC_FOLLOW_DOWN_ONE
	ZFS_AC_KERNEL_SRC_MAKE_REQUEST_FN
	ZFS_AC_KERNEL_SRC_GENERIC_IO_ACCT
	ZFS_AC_KERNEL_SRC_FPU
	ZFS_AC_KERNEL_SRC_FMODE_T
	ZFS_AC_KERNEL_SRC_KUIDGID_T
	ZFS_AC_KERNEL_SRC_KUID_HELPERS
	ZFS_AC_KERNEL_SRC_MODULE_PARAM_CALL_CONST
	ZFS_AC_KERNEL_SRC_RENAME
	ZFS_AC_KERNEL_SRC_CURRENT_TIME
	ZFS_AC_KERNEL_SRC_USERNS_CAPABILITIES
	ZFS_AC_KERNEL_SRC_IN_COMPAT_SYSCALL
	ZFS_AC_KERNEL_SRC_KTIME
	ZFS_AC_KERNEL_SRC_TOTALRAM_PAGES_FUNC
	ZFS_AC_KERNEL_SRC_TOTALHIGH_PAGES
	ZFS_AC_KERNEL_SRC_KSTRTOUL
	ZFS_AC_KERNEL_SRC_PERCPU
	ZFS_AC_KERNEL_SRC_CPU_HOTPLUG
	ZFS_AC_KERNEL_SRC_GENERIC_FILLATTR_USERNS
	ZFS_AC_KERNEL_SRC_MKNOD
	ZFS_AC_KERNEL_SRC_SYMLINK
	ZFS_AC_KERNEL_SRC_BIO_MAX_SEGS
	ZFS_AC_KERNEL_SRC_SIGNAL_STOP
	ZFS_AC_KERNEL_SRC_SIGINFO
	ZFS_AC_KERNEL_SRC_SET_SPECIAL_STATE
	ZFS_AC_KERNEL_SRC_VFS_SET_PAGE_DIRTY_NOBUFFERS
	ZFS_AC_KERNEL_SRC_STANDALONE_LINUX_STDARG
	+ ZFS_AC_KERNEL_SRC_PAGEMAP_FOLIO_WAIT_BIT

	AC_MSG_CHECKING([for available kernel interfaces])
	ZFS_LINUX_TEST_COMPILE_ALL([kabi])
	AC_MSG_RESULT([done])
	])

	dnl #
	dnl # Check results of kernel interface tests.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_TEST_RESULT], [
	ZFS_AC_KERNEL_ACCESS_OK_TYPE
	ZFS_AC_KERNEL_GLOBAL_PAGE_STATE
	ZFS_AC_KERNEL_OBJTOOL
	ZFS_AC_KERNEL_PDE_DATA
	ZFS_AC_KERNEL_FALLOCATE
	ZFS_AC_KERNEL_2ARGS_ZLIB_DEFLATE_WORKSPACESIZE
	ZFS_AC_KERNEL_RWSEM
	ZFS_AC_KERNEL_SCHED
	ZFS_AC_KERNEL_USLEEP_RANGE
	ZFS_AC_KERNEL_KMEM_CACHE
	ZFS_AC_KERNEL_KVMALLOC
	ZFS_AC_KERNEL_VMALLOC_PAGE_KERNEL
	ZFS_AC_KERNEL_WAIT
	ZFS_AC_KERNEL_INODE_TIMES
	ZFS_AC_KERNEL_INODE_LOCK
	ZFS_AC_KERNEL_GROUP_INFO_GID
	ZFS_AC_KERNEL_RW
	ZFS_AC_KERNEL_TIMER_SETUP
	ZFS_AC_KERNEL_SUPER_USER_NS
	ZFS_AC_KERNEL_PROC_OPERATIONS
	ZFS_AC_KERNEL_BLOCK_DEVICE_OPERATIONS
	ZFS_AC_KERNEL_BIO
	ZFS_AC_KERNEL_BLKDEV
	ZFS_AC_KERNEL_BLK_QUEUE
	ZFS_AC_KERNEL_REVALIDATE_DISK
	ZFS_AC_KERNEL_GET_DISK_RO
	ZFS_AC_KERNEL_GENERIC_READLINK_GLOBAL
	ZFS_AC_KERNEL_DISCARD_GRANULARITY
	ZFS_AC_KERNEL_INODE_OWNER_OR_CAPABLE
	ZFS_AC_KERNEL_XATTR
	ZFS_AC_KERNEL_ACL
	ZFS_AC_KERNEL_INODE_GETATTR
	ZFS_AC_KERNEL_INODE_SET_FLAGS
	ZFS_AC_KERNEL_INODE_SET_IVERSION
	ZFS_AC_KERNEL_SHOW_OPTIONS
	ZFS_AC_KERNEL_FILE_INODE
	ZFS_AC_KERNEL_FILE_DENTRY
	ZFS_AC_KERNEL_FSYNC
	ZFS_AC_KERNEL_AIO_FSYNC
	ZFS_AC_KERNEL_EVICT_INODE
	ZFS_AC_KERNEL_DIRTY_INODE
	ZFS_AC_KERNEL_SHRINKER
	ZFS_AC_KERNEL_MKDIR
	ZFS_AC_KERNEL_LOOKUP_FLAGS
	ZFS_AC_KERNEL_CREATE
	ZFS_AC_KERNEL_GET_LINK
	ZFS_AC_KERNEL_PUT_LINK
	ZFS_AC_KERNEL_TMPFILE
	ZFS_AC_KERNEL_AUTOMOUNT
	ZFS_AC_KERNEL_ENCODE_FH_WITH_INODE
	ZFS_AC_KERNEL_COMMIT_METADATA
	ZFS_AC_KERNEL_CLEAR_INODE
	ZFS_AC_KERNEL_SETATTR_PREPARE
	ZFS_AC_KERNEL_INSERT_INODE_LOCKED
	ZFS_AC_KERNEL_DENTRY
	ZFS_AC_KERNEL_TRUNCATE_SETSIZE
	ZFS_AC_KERNEL_SECURITY_INODE
	ZFS_AC_KERNEL_FST_MOUNT
	ZFS_AC_KERNEL_BDI
	ZFS_AC_KERNEL_SET_NLINK
	ZFS_AC_KERNEL_SGET
	ZFS_AC_KERNEL_LSEEK_EXECUTE
	ZFS_AC_KERNEL_VFS_GETATTR
	ZFS_AC_KERNEL_VFS_FSYNC_2ARGS
	ZFS_AC_KERNEL_VFS_ITERATE
	ZFS_AC_KERNEL_VFS_DIRECT_IO
	ZFS_AC_KERNEL_VFS_RW_ITERATE
	ZFS_AC_KERNEL_VFS_GENERIC_WRITE_CHECKS
	ZFS_AC_KERNEL_VFS_IOV_ITER
	ZFS_AC_KERNEL_KMAP_ATOMIC_ARGS
	ZFS_AC_KERNEL_FOLLOW_DOWN_ONE
	ZFS_AC_KERNEL_MAKE_REQUEST_FN
	ZFS_AC_KERNEL_GENERIC_IO_ACCT
	ZFS_AC_KERNEL_FPU
	ZFS_AC_KERNEL_FMODE_T
	ZFS_AC_KERNEL_KUIDGID_T
	ZFS_AC_KERNEL_KUID_HELPERS
	ZFS_AC_KERNEL_MODULE_PARAM_CALL_CONST
	ZFS_AC_KERNEL_RENAME
	ZFS_AC_KERNEL_CURRENT_TIME
	ZFS_AC_KERNEL_USERNS_CAPABILITIES
	ZFS_AC_KERNEL_IN_COMPAT_SYSCALL
	ZFS_AC_KERNEL_KTIME
	ZFS_AC_KERNEL_TOTALRAM_PAGES_FUNC
	ZFS_AC_KERNEL_TOTALHIGH_PAGES
	ZFS_AC_KERNEL_KSTRTOUL
	ZFS_AC_KERNEL_PERCPU
	ZFS_AC_KERNEL_CPU_HOTPLUG
	ZFS_AC_KERNEL_GENERIC_FILLATTR_USERNS
	ZFS_AC_KERNEL_MKNOD
	ZFS_AC_KERNEL_SYMLINK
	ZFS_AC_KERNEL_BIO_MAX_SEGS
	ZFS_AC_KERNEL_SIGNAL_STOP
	ZFS_AC_KERNEL_SIGINFO
	ZFS_AC_KERNEL_SET_SPECIAL_STATE
	ZFS_AC_KERNEL_VFS_SET_PAGE_DIRTY_NOBUFFERS
	ZFS_AC_KERNEL_STANDALONE_LINUX_STDARG
	+ ZFS_AC_KERNEL_PAGEMAP_FOLIO_WAIT_BIT
	])

	dnl #
	dnl # Detect name used for Module.symvers file in kernel
	dnl #
	AC_DEFUN([ZFS_AC_MODULE_SYMVERS], [
	modpost=$LINUX/scripts/Makefile.modpost
	AC_MSG_CHECKING([kernel file name for module symbols])
	AS_IF([test "x$enable_linux_builtin" != xyes -a -f "$modpost"], [
	AS_IF([grep -q Modules.symvers $modpost], [
	LINUX_SYMBOLS=Modules.symvers
	], [
	LINUX_SYMBOLS=Module.symvers
	])

	AS_IF([test ! -f "$LINUX_OBJ/$LINUX_SYMBOLS"], [
	AC_MSG_ERROR([
	*** Please make sure the kernel devel package for your distribution
	*** is installed. If you are building with a custom kernel, make sure
	*** the kernel is configured, built, and the '--with-linux=PATH'
	*** configure option refers to the location of the kernel source.
	])
	])
	], [
	LINUX_SYMBOLS=NONE
	])
	AC_MSG_RESULT($LINUX_SYMBOLS)
	AC_SUBST(LINUX_SYMBOLS)
	])

	dnl #
	dnl # Detect the kernel to be built against
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL], [
	AC_ARG_WITH([linux],
	AS_HELP_STRING([--with-linux=PATH],
	[Path to kernel source]),
	[kernelsrc="$withval"])

	AC_ARG_WITH(linux-obj,
	AS_HELP_STRING([--with-linux-obj=PATH],
	[Path to kernel build objects]),
	[kernelbuild="$withval"])

	AC_MSG_CHECKING([kernel source directory])
	AS_IF([test -z "$kernelsrc"], [
	AS_IF([test -e "/lib/modules/$(uname -r)/source"], [
	headersdir="/lib/modules/$(uname -r)/source"
	sourcelink=$(readlink -f "$headersdir")
	], [test -e "/lib/modules/$(uname -r)/build"], [
	headersdir="/lib/modules/$(uname -r)/build"
	sourcelink=$(readlink -f "$headersdir")
	], [
	sourcelink=$(ls -1d /usr/src/kernels/* \
	/usr/src/linux-* \
	2>/dev/null \| grep -v obj \| tail -1)
	])

	AS_IF([test -n "$sourcelink" && test -e ${sourcelink}], [
	kernelsrc=`readlink -f ${sourcelink}`
	], [
	kernelsrc="[Not found]"
	])
	], [
	AS_IF([test "$kernelsrc" = "NONE"], [
	kernsrcver=NONE
	])
	withlinux=yes
	])

	AC_MSG_RESULT([$kernelsrc])
	AS_IF([test ! -d "$kernelsrc"], [
	AC_MSG_ERROR([
	*** Please make sure the kernel devel package for your distribution
	*** is installed and then try again. If that fails, you can specify the
	*** location of the kernel source with the '--with-linux=PATH' option.])
	])

	AC_MSG_CHECKING([kernel build directory])
	AS_IF([test -z "$kernelbuild"], [
	AS_IF([test x$withlinux != xyes -a -e "/lib/modules/$(uname -r)/build"], [
	kernelbuild=`readlink -f /lib/modules/$(uname -r)/build`
	], [test -d ${kernelsrc}-obj/${target_cpu}/${target_cpu}], [
	kernelbuild=${kernelsrc}-obj/${target_cpu}/${target_cpu}
	], [test -d ${kernelsrc}-obj/${target_cpu}/default], [
	kernelbuild=${kernelsrc}-obj/${target_cpu}/default
	], [test -d `dirname ${kernelsrc}`/build-${target_cpu}], [
	kernelbuild=`dirname ${kernelsrc}`/build-${target_cpu}
	], [
	kernelbuild=${kernelsrc}
	])
	])
	AC_MSG_RESULT([$kernelbuild])

	AC_MSG_CHECKING([kernel source version])
	utsrelease1=$kernelbuild/include/linux/version.h
	utsrelease2=$kernelbuild/include/linux/utsrelease.h
	utsrelease3=$kernelbuild/include/generated/utsrelease.h
	AS_IF([test -r $utsrelease1 && fgrep -q UTS_RELEASE $utsrelease1], [
	utsrelease=$utsrelease1
	], [test -r $utsrelease2 && fgrep -q UTS_RELEASE $utsrelease2], [
	utsrelease=$utsrelease2
	], [test -r $utsrelease3 && fgrep -q UTS_RELEASE $utsrelease3], [
	utsrelease=$utsrelease3
	])

	AS_IF([test -n "$utsrelease"], [
	kernsrcver=$($AWK '/UTS_RELEASE/ { gsub(/"/, "", $[3]); print $[3] }' $utsrelease)
	AS_IF([test -z "$kernsrcver"], [
	AC_MSG_RESULT([Not found])
	AC_MSG_ERROR([
	*** Cannot determine kernel version.
	])
	])
	], [
	AC_MSG_RESULT([Not found])
	if test "x$enable_linux_builtin" != xyes; then
	AC_MSG_ERROR([
	*** Cannot find UTS_RELEASE definition.
	])
	else
	AC_MSG_ERROR([
	*** Cannot find UTS_RELEASE definition.
	*** Please run 'make prepare' inside the kernel source tree.])
	fi
	])

	AC_MSG_RESULT([$kernsrcver])

	AS_VERSION_COMPARE([$kernsrcver], [$ZFS_META_KVER_MIN], [
	AC_MSG_ERROR([
	*** Cannot build against kernel version $kernsrcver.
	*** The minimum supported kernel version is $ZFS_META_KVER_MIN.
	])
	])

	LINUX=${kernelsrc}
	LINUX_OBJ=${kernelbuild}
	LINUX_VERSION=${kernsrcver}

	AC_SUBST(LINUX)
	AC_SUBST(LINUX_OBJ)
	AC_SUBST(LINUX_VERSION)

	ZFS_AC_MODULE_SYMVERS
	])

	dnl #
	dnl # Detect the QAT module to be built against, QAT provides hardware
	dnl # acceleration for data compression:
	dnl #
	dnl # https://01.org/intel-quickassist-technology
	dnl #
	dnl # 1) Download and install QAT driver from the above link
	dnl # 2) Start QAT driver in your system:
	dnl # service qat_service start
	dnl # 3) Enable QAT in ZFS, e.g.:
	dnl # ./configure --with-qat=<qat-driver-path>/QAT1.6
	dnl # make
	dnl # 4) Set GZIP compression in ZFS dataset:
	dnl # zfs set compression = gzip <dataset>
	dnl #
	dnl # Then the data written to this ZFS pool is compressed by QAT accelerator
	dnl # automatically, and de-compressed by QAT when read from the pool.
	dnl #
	dnl # 1) Get QAT hardware statistics with:
	dnl # cat /proc/icp_dh895xcc_dev/qat
	dnl # 2) To disable QAT:
	dnl # insmod zfs.ko zfs_qat_disable=1
	dnl #
	AC_DEFUN([ZFS_AC_QAT], [
	AC_ARG_WITH([qat],
	AS_HELP_STRING([--with-qat=PATH],
	[Path to qat source]),
	AS_IF([test "$withval" = "yes"],
	AC_MSG_ERROR([--with-qat=PATH requires a PATH]),
	[qatsrc="$withval"]))

	AC_ARG_WITH([qat-obj],
	AS_HELP_STRING([--with-qat-obj=PATH],
	[Path to qat build objects]),
	[qatbuild="$withval"])

	AS_IF([test ! -z "${qatsrc}"], [
	AC_MSG_CHECKING([qat source directory])
	AC_MSG_RESULT([$qatsrc])
	QAT_SRC="${qatsrc}/quickassist"
	AS_IF([ test ! -e "$QAT_SRC/include/cpa.h"], [
	AC_MSG_ERROR([
	*** Please make sure the qat driver package is installed
	*** and specify the location of the qat source with the
	*** '--with-qat=PATH' option then try again. Failed to
	*** find cpa.h in:
	${QAT_SRC}/include])
	])
	])

	AS_IF([test ! -z "${qatsrc}"], [
	AC_MSG_CHECKING([qat build directory])
	AS_IF([test -z "$qatbuild"], [
	qatbuild="${qatsrc}/build"
	])

	AC_MSG_RESULT([$qatbuild])
	QAT_OBJ=${qatbuild}
	AS_IF([ ! test -e "$QAT_OBJ/icp_qa_al.ko" && ! test -e "$QAT_OBJ/qat_api.ko"], [
	AC_MSG_ERROR([
	*** Please make sure the qat driver is installed then try again.
	*** Failed to find icp_qa_al.ko or qat_api.ko in:
	$QAT_OBJ])
	])

	AC_SUBST(QAT_SRC)
	AC_SUBST(QAT_OBJ)

	AC_DEFINE(HAVE_QAT, 1,
	[qat is enabled and existed])
	])

	dnl #
	dnl # Detect the name used for the QAT Module.symvers file.
	dnl #
	AS_IF([test ! -z "${qatsrc}"], [
	AC_MSG_CHECKING([qat file for module symbols])
	QAT_SYMBOLS=$QAT_SRC/lookaside/access_layer/src/Module.symvers

	AS_IF([test -r $QAT_SYMBOLS], [
	AC_MSG_RESULT([$QAT_SYMBOLS])
	AC_SUBST(QAT_SYMBOLS)
	],[
	AC_MSG_ERROR([
	*** Please make sure the qat driver is installed then try again.
	*** Failed to find Module.symvers in:
	$QAT_SYMBOLS
	])
	])
	])
	])

	dnl #
	dnl # Basic toolchain sanity check.
	dnl #
	AC_DEFUN([ZFS_AC_KERNEL_TEST_MODULE], [
	AC_MSG_CHECKING([whether modules can be built])
	ZFS_LINUX_TRY_COMPILE([], [], [
	AC_MSG_RESULT([yes])
	],[
	AC_MSG_RESULT([no])
	if test "x$enable_linux_builtin" != xyes; then
	AC_MSG_ERROR([
	*** Unable to build an empty module.
	])
	else
	AC_MSG_ERROR([
	*** Unable to build an empty module.
	*** Please run 'make scripts' inside the kernel source tree.])
	fi
	])
	])

	dnl #
	dnl # ZFS_LINUX_CONFTEST_H
	dnl #
	AC_DEFUN([ZFS_LINUX_CONFTEST_H], [
	test -d build/$2 \|\| mkdir -p build/$2
	cat - <<_ACEOF >build/$2/$2.h
	$1
	_ACEOF
	])

	dnl #
	dnl # ZFS_LINUX_CONFTEST_C
	dnl #
	AC_DEFUN([ZFS_LINUX_CONFTEST_C], [
	test -d build/$2 \|\| mkdir -p build/$2
	cat confdefs.h - <<_ACEOF >build/$2/$2.c
	$1
	_ACEOF
	])

	dnl #
	dnl # ZFS_LINUX_CONFTEST_MAKEFILE
	dnl #
	dnl # $1 - test case name
	dnl # $2 - add to top-level Makefile
	dnl # $3 - additional build flags
	dnl #
	AC_DEFUN([ZFS_LINUX_CONFTEST_MAKEFILE], [
	test -d build \|\| mkdir -p build
	test -d build/$1 \|\| mkdir -p build/$1

	file=build/$1/Makefile

	dnl # Example command line to manually build source.
	cat - <<_ACEOF >$file
	# Example command line to manually build source
	# make modules -C $LINUX_OBJ $ARCH_UM M=$PWD/build/$1

	ccflags-y := -Werror $FRAME_LARGER_THAN
	_ACEOF

	dnl # Additional custom CFLAGS as requested.
	m4_ifval($3, [echo "ccflags-y += $3" >>$file], [])

	dnl # Test case source
	echo "obj-m := $1.o" >>$file

	AS_IF([test "x$2" = "xyes"], [echo "obj-m += $1/" >>build/Makefile], [])
	])

	dnl #
	dnl # ZFS_LINUX_TEST_PROGRAM(C)([PROLOGUE], [BODY])
	dnl #
	m4_define([ZFS_LINUX_TEST_PROGRAM], [
	#include <linux/module.h>
	$1

	int
	main (void)
	{
	$2
	;
	return 0;
	}

	MODULE_DESCRIPTION("conftest");
	MODULE_AUTHOR(ZFS_META_AUTHOR);
	MODULE_VERSION(ZFS_META_VERSION "-" ZFS_META_RELEASE);
	MODULE_LICENSE($3);
	])

	dnl #
	dnl # ZFS_LINUX_TEST_REMOVE
	dnl #
	dnl # Removes the specified test source and results.
	dnl #
	AC_DEFUN([ZFS_LINUX_TEST_REMOVE], [
	test -d build/$1 && rm -Rf build/$1
	test -f build/Makefile && sed '/$1/d' build/Makefile
	])

	dnl #
	dnl # ZFS_LINUX_COMPILE
	dnl #
	dnl # $1 - build dir
	dnl # $2 - test command
	dnl # $3 - pass command
	dnl # $4 - fail command
	dnl # $5 - set KBUILD_MODPOST_NOFINAL='yes'
	dnl # $6 - set KBUILD_MODPOST_WARN='yes'
	dnl #
	dnl # Used internally by ZFS_LINUX_TEST_{COMPILE,MODPOST}
	dnl #
	AC_DEFUN([ZFS_LINUX_COMPILE], [
	AC_TRY_COMMAND([
	KBUILD_MODPOST_NOFINAL="$5" KBUILD_MODPOST_WARN="$6"
	make modules -k -j$TEST_JOBS -C $LINUX_OBJ $ARCH_UM
	M=$PWD/$1 >$1/build.log 2>&1])
	AS_IF([AC_TRY_COMMAND([$2])], [$3], [$4])
	])

	dnl #
	dnl # ZFS_LINUX_TEST_COMPILE
	dnl #
	dnl # Perform a full compile excluding the final modpost phase.
	dnl #
	AC_DEFUN([ZFS_LINUX_TEST_COMPILE], [
	ZFS_LINUX_COMPILE([$2], [test -f $2/build.log], [
	mv $2/Makefile $2/Makefile.compile.$1
	mv $2/build.log $2/build.log.$1
	],[
	AC_MSG_ERROR([
	*** Unable to compile test source to determine kernel interfaces.])
	], [yes], [])
	])

	dnl #
	dnl # ZFS_LINUX_TEST_MODPOST
	dnl #
	dnl # Perform a full compile including the modpost phase. This may
	dnl # be an incremental build if the objects have already been built.
	dnl #
	AC_DEFUN([ZFS_LINUX_TEST_MODPOST], [
	ZFS_LINUX_COMPILE([$2], [test -f $2/build.log], [
	mv $2/Makefile $2/Makefile.modpost.$1
	cat $2/build.log >>build/build.log.$1
	],[
	AC_MSG_ERROR([
	*** Unable to modpost test source to determine kernel interfaces.])
	], [], [yes])
	])

	dnl #
	dnl # Perform the compilation of the test cases in two phases.
	dnl #
	dnl # Phase 1) attempt to build the object files for all of the tests
	dnl # defined by the ZFS_LINUX_TEST_SRC macro. But do not
	dnl # perform the final modpost stage.
	dnl #
	dnl # Phase 2) disable all tests which failed the initial compilation,
	dnl # then invoke the final modpost step for the remaining tests.
	dnl #
	dnl # This allows us efficiently build the test cases in parallel while
	dnl # remaining resilient to build failures which are expected when
	dnl # detecting the available kernel interfaces.
	dnl #
	dnl # The maximum allowed parallelism can be controlled by setting the
	dnl # TEST_JOBS environment variable. Otherwise, it default to $(nproc).
	dnl #
	AC_DEFUN([ZFS_LINUX_TEST_COMPILE_ALL], [
	dnl # Phase 1 - Compilation only, final linking is skipped.
	ZFS_LINUX_TEST_COMPILE([$1], [build])

	dnl #
	dnl # Phase 2 - When building external modules disable test cases
	dnl # which failed to compile and invoke modpost to verify the
	dnl # final linking.
	dnl #
	dnl # Test names suffixed with '_license' call modpost independently
	dnl # to ensure that a single incompatibility does not result in the
	dnl # modpost phase exiting early. This check is not performed on
	dnl # every symbol since the majority are compatible and doing so
	dnl # would significantly slow down this phase.
	dnl #
	dnl # When configuring for builtin (--enable-linux-builtin)
	dnl # fake the linking step artificially create the expected .ko
	dnl # files for tests which did compile. This is required for
	dnl # kernels which do not have loadable module support or have
	dnl # not yet been built.
	dnl #
	AS_IF([test "x$enable_linux_builtin" = "xno"], [
	for dir in $(awk '/^obj-m/ { print [$]3 }' \
	build/Makefile.compile.$1); do
	name=${dir%/}
	AS_IF([test -f build/$name/$name.o], [
	AS_IF([test "${name##*_}" = "license"], [
	ZFS_LINUX_TEST_MODPOST([$1],
	[build/$name])
	echo "obj-n += $dir" >>build/Makefile
	], [
	echo "obj-m += $dir" >>build/Makefile
	])
	], [
	echo "obj-n += $dir" >>build/Makefile
	])
	done

	ZFS_LINUX_TEST_MODPOST([$1], [build])
	], [
	for dir in $(awk '/^obj-m/ { print [$]3 }' \
	build/Makefile.compile.$1); do
	name=${dir%/}
	AS_IF([test -f build/$name/$name.o], [
	touch build/$name/$name.ko
	])
	done
	])
	])

	dnl #
	dnl # ZFS_LINUX_TEST_SRC
	dnl #
	dnl # $1 - name
	dnl # $2 - global
	dnl # $3 - source
	dnl # $4 - extra cflags
	dnl # $5 - check license-compatibility
	dnl #
	dnl # Check if the test source is buildable at all and then if it is
	dnl # license compatible.
	dnl #
	dnl # N.B because all of the test cases are compiled in parallel they
	dnl # must never depend on the results of previous tests. Each test
	dnl # needs to be entirely independent.
	dnl #
	AC_DEFUN([ZFS_LINUX_TEST_SRC], [
	ZFS_LINUX_CONFTEST_C([ZFS_LINUX_TEST_PROGRAM([[$2]], [[$3]],
	[["Dual BSD/GPL"]])], [$1])
	ZFS_LINUX_CONFTEST_MAKEFILE([$1], [yes], [$4])

	AS_IF([ test -n "$5" ], [
	ZFS_LINUX_CONFTEST_C([ZFS_LINUX_TEST_PROGRAM(
	[[$2]], [[$3]], [[$5]])], [$1_license])
	ZFS_LINUX_CONFTEST_MAKEFILE([$1_license], [yes], [$4])
	])
	])

	dnl #
	dnl # ZFS_LINUX_TEST_RESULT
	dnl #
	dnl # $1 - name of a test source (ZFS_LINUX_TEST_SRC)
	dnl # $2 - run on success (valid .ko generated)
	dnl # $3 - run on failure (unable to compile)
	dnl #
	AC_DEFUN([ZFS_LINUX_TEST_RESULT], [
	AS_IF([test -d build/$1], [
	AS_IF([test -f build/$1/$1.ko], [$2], [$3])
	], [
	AC_MSG_ERROR([
	*** No matching source for the "$1" test, check that
	*** both the test source and result macros refer to the same name.
	])
	])
	])

	dnl #
	dnl # ZFS_LINUX_TEST_ERROR
	dnl #
	dnl # Generic error message which can be used when none of the expected
	dnl # kernel interfaces were detected.
	dnl #
	AC_DEFUN([ZFS_LINUX_TEST_ERROR], [
	AC_MSG_ERROR([
	*** None of the expected "$1" interfaces were detected.
	*** This may be because your kernel version is newer than what is
	*** supported, or you are using a patched custom kernel with
	*** incompatible modifications.
	***
	*** ZFS Version: $ZFS_META_ALIAS
	*** Compatible Kernels: $ZFS_META_KVER_MIN - $ZFS_META_KVER_MAX
	])
	])

	dnl #
	dnl # ZFS_LINUX_TEST_RESULT_SYMBOL
	dnl #
	dnl # Like ZFS_LINUX_TEST_RESULT except ZFS_CHECK_SYMBOL_EXPORT is called to
	dnl # verify symbol exports, unless --enable-linux-builtin was provided to
	dnl # configure.
	dnl #
	AC_DEFUN([ZFS_LINUX_TEST_RESULT_SYMBOL], [
	AS_IF([ ! test -f build/$1/$1.ko], [
	$5
	], [
	AS_IF([test "x$enable_linux_builtin" != "xyes"], [
	ZFS_CHECK_SYMBOL_EXPORT([$2], [$3], [$4], [$5])
	], [
	$4
	])
	])
	])

	dnl #
	dnl # ZFS_LINUX_COMPILE_IFELSE
	dnl #
	AC_DEFUN([ZFS_LINUX_COMPILE_IFELSE], [
	ZFS_LINUX_TEST_REMOVE([conftest])

	m4_ifvaln([$1], [ZFS_LINUX_CONFTEST_C([$1], [conftest])])
	m4_ifvaln([$5], [ZFS_LINUX_CONFTEST_H([$5], [conftest])],
	[ZFS_LINUX_CONFTEST_H([], [conftest])])

	ZFS_LINUX_CONFTEST_MAKEFILE([conftest], [no],
	[m4_ifvaln([$5], [-I$PWD/build/conftest], [])])
	ZFS_LINUX_COMPILE([build/conftest], [$2], [$3], [$4], [], [])
	])

	dnl #
	dnl # ZFS_LINUX_TRY_COMPILE
	dnl #
	dnl # $1 - global
	dnl # $2 - source
	dnl # $3 - run on success (valid .ko generated)
	dnl # $4 - run on failure (unable to compile)
	dnl #
	dnl # When configuring as builtin (--enable-linux-builtin) for kernels
	dnl # without loadable module support (CONFIG_MODULES=n) only the object
	dnl # file is created. See ZFS_LINUX_TEST_COMPILE_ALL for details.
	dnl #
	AC_DEFUN([ZFS_LINUX_TRY_COMPILE], [
	AS_IF([test "x$enable_linux_builtin" = "xyes"], [
	ZFS_LINUX_COMPILE_IFELSE(
	[ZFS_LINUX_TEST_PROGRAM([[$1]], [[$2]],
	[[ZFS_META_LICENSE]])],
	[test -f build/conftest/conftest.o], [$3], [$4])
	], [
	ZFS_LINUX_COMPILE_IFELSE(
	[ZFS_LINUX_TEST_PROGRAM([[$1]], [[$2]],
	[[ZFS_META_LICENSE]])],
	[test -f build/conftest/conftest.ko], [$3], [$4])
	])
	])

	dnl #
	dnl # ZFS_CHECK_SYMBOL_EXPORT
	dnl #
	dnl # Check if a symbol is exported on not by consulting the symbols
	dnl # file, or optionally the source code.
	dnl #
	AC_DEFUN([ZFS_CHECK_SYMBOL_EXPORT], [
	grep -q -E '[[[:space:]]]$1[[[:space:]]]' \
	$LINUX_OBJ/$LINUX_SYMBOLS 2>/dev/null
	rc=$?
	if test $rc -ne 0; then
	export=0
	for file in $2; do
	grep -q -E "EXPORT_SYMBOL.*($1)" \
	"$LINUX/$file" 2>/dev/null
	rc=$?
	if test $rc -eq 0; then
	export=1
	break;
	fi
	done
	if test $export -eq 0; then :
	$4
	else :
	$3
	fi
	else :
	$3
	fi
	])

	dnl #
	dnl # ZFS_LINUX_TRY_COMPILE_SYMBOL
	dnl #
	dnl # Like ZFS_LINUX_TRY_COMPILER except ZFS_CHECK_SYMBOL_EXPORT is called
	dnl # to verify symbol exports, unless --enable-linux-builtin was provided
	dnl # to configure.
	dnl #
	AC_DEFUN([ZFS_LINUX_TRY_COMPILE_SYMBOL], [
	ZFS_LINUX_TRY_COMPILE([$1], [$2], [rc=0], [rc=1])
	if test $rc -ne 0; then :
	$6
	else
	if test "x$enable_linux_builtin" != xyes; then
	ZFS_CHECK_SYMBOL_EXPORT([$3], [$4], [rc=0], [rc=1])
	fi
	if test $rc -ne 0; then :
	$6
	else :
	$5
	fi
	fi
	])

	dnl #
	dnl # ZFS_LINUX_TRY_COMPILE_HEADER
	dnl # like ZFS_LINUX_TRY_COMPILE, except the contents conftest.h are
	dnl # provided via the fifth parameter
	dnl #
	AC_DEFUN([ZFS_LINUX_TRY_COMPILE_HEADER], [
	ZFS_LINUX_COMPILE_IFELSE(
	[ZFS_LINUX_TEST_PROGRAM([[$1]], [[$2]], [[ZFS_META_LICENSE]])],
	[test -f build/conftest/conftest.ko],
	[$3], [$4], [$5])
	])
	diff --git a/sys/contrib/openzfs/contrib/Makefile.am b/sys/contrib/openzfs/contrib/Makefile.am
	index 5ec13ece5327..f13b5f9f629c 100644
	--- a/sys/contrib/openzfs/contrib/Makefile.am
	+++ b/sys/contrib/openzfs/contrib/Makefile.am
	@@ -1,12 +1,13 @@
	include $(top_srcdir)/config/Shellcheck.am

	SUBDIRS = bash_completion.d pyzfs zcp
	if BUILD_LINUX
	SUBDIRS += bpftrace dracut initramfs
	endif
	if PAM_ZFS_ENABLED
	SUBDIRS += pam_zfs_key
	endif
	DIST_SUBDIRS = bash_completion.d bpftrace dracut initramfs pam_zfs_key pyzfs zcp

	SHELLCHECKDIRS = bash_completion.d bpftrace dracut initramfs
	+SHELLCHECK_OPTS = --enable=all
	diff --git a/sys/contrib/openzfs/contrib/bash_completion.d/Makefile.am b/sys/contrib/openzfs/contrib/bash_completion.d/Makefile.am
	index 8c8d1acebe15..f381613ed656 100644
	--- a/sys/contrib/openzfs/contrib/bash_completion.d/Makefile.am
	+++ b/sys/contrib/openzfs/contrib/bash_completion.d/Makefile.am
	@@ -1,13 +1,13 @@
	include $(top_srcdir)/config/Substfiles.am
	include $(top_srcdir)/config/Shellcheck.am

	bashcompletiondir = $(sysconfdir)/bash_completion.d

	noinst_DATA = zfs

	EXTRA_DIST += $(noinst_DATA)
	SUBSTFILES += $(noinst_DATA)

	SHELLCHECKSCRIPTS = $(noinst_DATA)
	SHELLCHECK_SHELL = bash
	-SHELLCHECK_IGNORE = ,SC2207
	+SHELLCHECK_OPTS = --enable=all
	diff --git a/sys/contrib/openzfs/contrib/bash_completion.d/zfs.in b/sys/contrib/openzfs/contrib/bash_completion.d/zfs.in
	index 41ce2f871e34..72e1092a039b 100644
	--- a/sys/contrib/openzfs/contrib/bash_completion.d/zfs.in
	+++ b/sys/contrib/openzfs/contrib/bash_completion.d/zfs.in
	@@ -1,484 +1,484 @@
	# Copyright (c) 2010-2016, Aneurin Price <aneurin.price@gmail.com>

	# Permission is hereby granted, free of charge, to any person
	# obtaining a copy of this software and associated documentation
	# files (the "Software"), to deal in the Software without
	# restriction, including without limitation the rights to use,
	# copy, modify, merge, publish, distribute, sublicense, and/or sell
	# copies of the Software, and to permit persons to whom the
	# Software is furnished to do so, subject to the following
	# conditions:

	# The above copyright notice and this permission notice shall be
	# included in all copies or substantial portions of the Software.

	# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
	# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
	# OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
	# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
	# HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
	# WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
	# FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
	# OTHER DEALINGS IN THE SOFTWARE.

	__ZFS_CMD="@sbindir@/zfs"
	__ZPOOL_CMD="@sbindir@/zpool"

	# Disable bash's built-in hostname completion, as this makes it impossible to
	# provide completions containing an @-sign, which is necessary for completing
	# snapshot names. If bash_completion is in use, this will already be disabled
	# and replaced with better completions anyway.
	shopt -u hostcomplete

	__zfs_get_commands()
	{
	$__ZFS_CMD 2>&1 \| awk '/^\t[a-z]/ {print $1}' \| cut -f1 -d '\|' \| uniq
	}

	__zfs_get_properties()
	{
	$__ZFS_CMD get 2>&1 \| awk '$2 == "YES" \|\| $2 == "NO" {print $1}'; echo all name space
	}

	__zfs_get_editable_properties()
	{
	$__ZFS_CMD get 2>&1 \| awk '$2 == "YES" {print $1"="}'
	}

	__zfs_get_inheritable_properties()
	{
	$__ZFS_CMD get 2>&1 \| awk '$3 == "YES" {print $1}'
	}

	__zfs_list_datasets()
	{
	$__ZFS_CMD list -H -o name -s name -t filesystem,volume "$@"
	}

	__zfs_list_filesystems()
	{
	$__ZFS_CMD list -H -o name -s name -t filesystem
	}

	__zfs_match_snapshot()
	{
	local base_dataset="${cur%@*}"
	if [ "$base_dataset" != "$cur" ]
	then
	$__ZFS_CMD list -H -o name -s name -t snapshot -d 1 "$base_dataset"
	else
	if [ "$cur" != "" ] && __zfs_list_datasets "$cur" &> /dev/null
	then
	- $__ZFS_CMD list -H -o name -s name -t filesystem -r "$cur" \| tail -n +2
	+ $__ZFS_CMD list -H -o name -s name -t filesystem,volume -r "$cur" \| tail -n +2
	# We output the base dataset name even though we might be
	# completing a command that can only take a snapshot, because it
	# prevents bash from considering the completion finished when it
	# ends in the bare @.
	echo "$cur"
	echo "$cur@"
	else
	local datasets
	datasets="$(__zfs_list_datasets)"
	# As above
	echo "$datasets"
	if [[ "$cur" == */ ]]
	then
	# If the current command ends with a slash, then the only way
	# it can be completed with a single tab press (ie. in this pass)
	# is if it has exactly one child, so that's the only time we
	# need to offer a suggestion with an @ appended.
	local num_children
	# This is actually off by one as zfs list includes the named
	# dataset in addition to its children
	num_children=$(__zfs_list_datasets -d 1 "${cur%/}" 2> /dev/null \| wc -l)
	if [[ $num_children != 2 ]]
	then
	return 0
	fi
	fi
	echo "$datasets" \| awk '{print $1 "@"}'
	fi
	fi
	}

	__zfs_match_snapshot_or_bookmark()
	{
	local base_dataset="${cur%[#@]*}"
	if [ "$base_dataset" != "$cur" ]
	then
	if [[ $cur == @ ]]
	then
	$__ZFS_CMD list -H -o name -s name -t snapshot -d 1 "$base_dataset"
	else
	$__ZFS_CMD list -H -o name -s name -t bookmark -d 1 "$base_dataset"
	fi
	else
	$__ZFS_CMD list -H -o name -s name -t filesystem,volume
	if [ -e "$cur" ] && $__ZFS_CMD list -H -o name -s name -t filesystem,volume "$cur" &> /dev/null
	then
	echo "$cur@"
	echo "$cur#"
	fi
	fi
	}

	__zfs_match_multiple_snapshots()
	{
	local existing_opts
	existing_opts="$(expr "$cur" : '$.*$[%,]')"
	if [ -e "$existing_opts" ]
	then
	local base_dataset="${cur%@*}"
	if [ "$base_dataset" != "$cur" ]
	then
	local cur="${cur##*,}"
	if [[ $cur =~ ^%\|%.*% ]]
	then
	# correct range syntax is start%end
	return 1
	fi
	local range_start
	range_start="$(expr "$cur" : '$.*%$')"
	# shellcheck disable=SC2016
	$__ZFS_CMD list -H -o name -s name -t snapshot -d 1 "$base_dataset" \| sed 's$.*@$'"$range_start"'$g'
	fi
	else
	__zfs_match_snapshot_or_bookmark
	fi
	}

	__zfs_list_volumes()
	{
	$__ZFS_CMD list -H -o name -s name -t volume
	}

	__zfs_argument_chosen()
	{
	local word property
	for word in $(seq $((COMP_CWORD-1)) -1 2)
	do
	local prev="${COMP_WORDS[$word]}"
	if [[ ${COMP_WORDS[$word-1]} != -[tos] ]]
	then
	if [[ "$prev" == [^,], ]] \|\| [[ "$prev" == [@:\#] ]]
	then
	return 0
	fi
	for property in "$@"
	do
	if [[ $prev == "$property"* ]]
	then
	return 0
	fi
	done
	fi
	done
	return 1
	}

	__zfs_complete_ordered_arguments()
	{
	local list1=$1
	local list2=$2
	local cur=$3
	local extra=$4
	# shellcheck disable=SC2086
	if __zfs_argument_chosen $list1
	then
	- COMPREPLY=($(compgen -W "$list2 $extra" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$list2 $extra" -- "$cur")
	else
	- COMPREPLY=($(compgen -W "$list1 $extra" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$list1 $extra" -- "$cur")
	fi
	}

	__zfs_complete_multiple_options()
	{
	local options=$1
	local cur=$2
	local existing_opts

	- COMPREPLY=($(compgen -W "$options" -- "${cur##*,}"))
	+ mapfile -t COMPREPLY < <(compgen -W "$options" -- "${cur##*,}")
	existing_opts=$(expr "$cur" : '$.*,$')
	- if [[ $existing_opts ]]
	+ if [ -n "$existing_opts" ]
	then
	COMPREPLY=( "${COMPREPLY[@]/#/${existing_opts}}" )
	fi
	}

	__zfs_complete_switch()
	{
	local options=$1
	if [[ ${cur:0:1} == - ]]
	then
	- COMPREPLY=($(compgen -W "-{$options}" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "-{$options}" -- "$cur")
	return 0
	else
	return 1
	fi
	}

	__zfs_complete_nospace()
	{
	# Google indicates that there may still be bash versions out there that
	# don't have compopt.
	if type compopt &> /dev/null
	then
	compopt -o nospace
	fi
	}

	__zfs_complete()
	{
	local cur prev cmd cmds
	COMPREPLY=()
	if type _get_comp_words_by_ref &> /dev/null
	then
	# Don't split on colon
	_get_comp_words_by_ref -n : -c cur -p prev -w COMP_WORDS -i COMP_CWORD
	else
	cur="${COMP_WORDS[COMP_CWORD]}"
	prev="${COMP_WORDS[COMP_CWORD-1]}"
	fi
	cmd="${COMP_WORDS[1]}"

	if [[ ${prev##*/} == zfs ]]
	then
	cmds=$(__zfs_get_commands)
	- COMPREPLY=($(compgen -W "$cmds -?" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$cmds -?" -- "$cur")
	return 0
	fi

	case "${cmd}" in
	bookmark)
	if __zfs_argument_chosen
	then
	- COMPREPLY=($(compgen -W "${prev%@*}# ${prev/@/#}" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "${prev%@*}# ${prev/@/#}" -- "$cur")
	else
	- COMPREPLY=($(compgen -W "$(__zfs_match_snapshot)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_match_snapshot)" -- "$cur")
	fi
	;;
	clone)
	case "${prev}" in
	-o)
	- COMPREPLY=($(compgen -W "$(__zfs_get_editable_properties)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_get_editable_properties)" -- "$cur")
	__zfs_complete_nospace
	;;
	*)
	if ! __zfs_complete_switch "o,p"
	then
	if __zfs_argument_chosen
	then
	- COMPREPLY=($(compgen -W "$(__zfs_list_datasets)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_list_datasets)" -- "$cur")
	else
	- COMPREPLY=($(compgen -W "$(__zfs_match_snapshot)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_match_snapshot)" -- "$cur")
	fi
	fi
	;;
	esac
	;;
	get)
	case "${prev}" in
	-d)
	- COMPREPLY=($(compgen -W "" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "" -- "$cur")
	;;
	-t)
	__zfs_complete_multiple_options "filesystem volume snapshot bookmark all" "$cur"
	;;
	-s)
	__zfs_complete_multiple_options "local default inherited temporary received none" "$cur"
	;;
	-o)
	__zfs_complete_multiple_options "name property value source received all" "$cur"
	;;
	*)
	if ! __zfs_complete_switch "H,r,p,d,o,t,s"
	then
	# shellcheck disable=SC2046
	if __zfs_argument_chosen $(__zfs_get_properties)
	then
	- COMPREPLY=($(compgen -W "$(__zfs_match_snapshot)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_match_snapshot)" -- "$cur")
	else
	__zfs_complete_multiple_options "$(__zfs_get_properties)" "$cur"
	fi
	fi
	;;
	esac
	;;
	inherit)
	if ! __zfs_complete_switch "r"
	then
	__zfs_complete_ordered_arguments "$(__zfs_get_inheritable_properties)" "$(__zfs_match_snapshot)" "$cur"
	fi
	;;
	list)
	case "${prev}" in
	-d)
	- COMPREPLY=($(compgen -W "" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "" -- "$cur")
	;;
	-t)
	__zfs_complete_multiple_options "filesystem volume snapshot bookmark all" "$cur"
	;;
	-o)
	__zfs_complete_multiple_options "$(__zfs_get_properties)" "$cur"
	;;
	-s\|-S)
	- COMPREPLY=($(compgen -W "$(__zfs_get_properties)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_get_properties)" -- "$cur")
	;;
	*)
	if ! __zfs_complete_switch "H,r,d,o,t,s,S"
	then
	- COMPREPLY=($(compgen -W "$(__zfs_match_snapshot)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_match_snapshot)" -- "$cur")
	fi
	;;
	esac
	;;
	promote)
	- COMPREPLY=($(compgen -W "$(__zfs_list_filesystems)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_list_filesystems)" -- "$cur")
	;;
	rollback)
	if ! __zfs_complete_switch "r,R,f"
	then
	- COMPREPLY=($(compgen -W "$(__zfs_match_snapshot)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_match_snapshot)" -- "$cur")
	fi
	;;
	send)
	if ! __zfs_complete_switch "D,n,P,p,R,v,e,L,i,I"
	then
	if __zfs_argument_chosen
	then
	- COMPREPLY=($(compgen -W "$(__zfs_match_snapshot)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_match_snapshot)" -- "$cur")
	else
	if [[ $prev == -i ]]
	then
	- COMPREPLY=($(compgen -W "$(__zfs_match_snapshot_or_bookmark)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_match_snapshot_or_bookmark)" -- "$cur")
	else
	- COMPREPLY=($(compgen -W "$(__zfs_match_snapshot)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_match_snapshot)" -- "$cur")
	fi
	fi
	fi
	;;
	snapshot)
	case "${prev}" in
	-o)
	- COMPREPLY=($(compgen -W "$(__zfs_get_editable_properties)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_get_editable_properties)" -- "$cur")
	__zfs_complete_nospace
	;;
	*)
	if ! __zfs_complete_switch "o,r"
	then
	- COMPREPLY=($(compgen -W "$(__zfs_match_snapshot)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_match_snapshot)" -- "$cur")
	__zfs_complete_nospace
	fi
	;;
	esac
	;;
	set)
	__zfs_complete_ordered_arguments "$(__zfs_get_editable_properties)" "$(__zfs_match_snapshot)" "$cur"
	__zfs_complete_nospace
	;;
	upgrade)
	case "${prev}" in
	-a\|-V\|-v)
	- COMPREPLY=($(compgen -W "" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "" -- "$cur")
	;;
	*)
	if ! __zfs_complete_switch "a,V,v,r"
	then
	- COMPREPLY=($(compgen -W "$(__zfs_list_filesystems)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_list_filesystems)" -- "$cur")
	fi
	;;
	esac
	;;
	destroy)
	if ! __zfs_complete_switch "d,f,n,p,R,r,v"
	then
	__zfs_complete_multiple_options "$(__zfs_match_multiple_snapshots)" "$cur"
	__zfs_complete_nospace
	fi
	;;
	*)
	- COMPREPLY=($(compgen -W "$(__zfs_match_snapshot)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zfs_match_snapshot)" -- "$cur")
	;;
	esac
	if type __ltrim_colon_completions &> /dev/null
	then
	__ltrim_colon_completions "$cur"
	fi
	return 0
	}

	__zpool_get_commands()
	{
	$__ZPOOL_CMD 2>&1 \| awk '/^\t[a-z]/ {print $1}' \| uniq
	}

	__zpool_get_properties()
	{
	$__ZPOOL_CMD get 2>&1 \| awk '$2 == "YES" \|\| $2 == "NO" {print $1}'; echo all
	}

	__zpool_get_editable_properties()
	{
	$__ZPOOL_CMD get 2>&1 \| awk '$2 == "YES" {print $1"="}'
	}

	__zpool_list_pools()
	{
	$__ZPOOL_CMD list -H -o name
	}

	__zpool_complete()
	{
	local cur prev cmd cmds pools
	COMPREPLY=()
	cur="${COMP_WORDS[COMP_CWORD]}"
	prev="${COMP_WORDS[COMP_CWORD-1]}"
	cmd="${COMP_WORDS[1]}"

	if [[ ${prev##*/} == zpool ]]
	then
	cmds=$(__zpool_get_commands)
	- COMPREPLY=($(compgen -W "$cmds" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$cmds" -- "$cur")
	return 0
	fi

	case "${cmd}" in
	get)
	__zfs_complete_ordered_arguments "$(__zpool_get_properties)" "$(__zpool_list_pools)" "$cur"
	return 0
	;;
	import)
	if [[ $prev == -d ]]
	then
	_filedir -d
	else
	- COMPREPLY=($(compgen -W "$(__zpool_list_pools) -d" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zpool_list_pools) -d" -- "$cur")
	fi
	return 0
	;;
	set)
	__zfs_complete_ordered_arguments "$(__zpool_get_editable_properties)" "$(__zpool_list_pools)" "$cur"
	__zfs_complete_nospace
	return 0
	;;
	add\|attach\|clear\|create\|detach\|offline\|online\|remove\|replace)
	pools="$(__zpool_list_pools)"
	# shellcheck disable=SC2086
	if __zfs_argument_chosen $pools
	then
	_filedir
	else
	- COMPREPLY=($(compgen -W "$pools" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$pools" -- "$cur")
	fi
	return 0
	;;
	*)
	- COMPREPLY=($(compgen -W "$(__zpool_list_pools)" -- "$cur"))
	+ mapfile -t COMPREPLY < <(compgen -W "$(__zpool_list_pools)" -- "$cur")
	return 0
	;;
	esac

	}

	complete -F __zfs_complete zfs
	complete -F __zpool_complete zpool
	diff --git a/sys/contrib/openzfs/contrib/bpftrace/Makefile.am b/sys/contrib/openzfs/contrib/bpftrace/Makefile.am
	index 05e4f1c507ce..87732331f8f5 100644
	--- a/sys/contrib/openzfs/contrib/bpftrace/Makefile.am
	+++ b/sys/contrib/openzfs/contrib/bpftrace/Makefile.am
	@@ -1,7 +1,8 @@
	include $(top_srcdir)/config/Shellcheck.am

	EXTRA_DIST = \
	taskqlatency.bt \
	zfs-trace.sh

	SHELLCHECKSCRIPTS = zfs-trace.sh
	+SHELLCHECK_OPTS = --enable=all
	diff --git a/sys/contrib/openzfs/contrib/dracut/02zfsexpandknowledge/Makefile.am b/sys/contrib/openzfs/contrib/dracut/02zfsexpandknowledge/Makefile.am
	index b1bbb6bd3aac..cdf06202b7fc 100644
	--- a/sys/contrib/openzfs/contrib/dracut/02zfsexpandknowledge/Makefile.am
	+++ b/sys/contrib/openzfs/contrib/dracut/02zfsexpandknowledge/Makefile.am
	@@ -1,8 +1,9 @@
	include $(top_srcdir)/config/Substfiles.am
	include $(top_srcdir)/config/Shellcheck.am

	pkgdracutdir = $(dracutdir)/modules.d/02zfsexpandknowledge
	pkgdracut_SCRIPTS = \
	module-setup.sh

	+SHELLCHECK_OPTS = --enable=all
	SUBSTFILES += $(pkgdracut_SCRIPTS)
	diff --git a/sys/contrib/openzfs/contrib/dracut/90zfs/Makefile.am b/sys/contrib/openzfs/contrib/dracut/90zfs/Makefile.am
	index 3f7050300994..d25a3d250a55 100644
	--- a/sys/contrib/openzfs/contrib/dracut/90zfs/Makefile.am
	+++ b/sys/contrib/openzfs/contrib/dracut/90zfs/Makefile.am
	@@ -1,24 +1,25 @@
	include $(top_srcdir)/config/Substfiles.am
	include $(top_srcdir)/config/Shellcheck.am

	pkgdracutdir = $(dracutdir)/modules.d/90zfs
	pkgdracut_SCRIPTS = \
	export-zfs.sh \
	module-setup.sh \
	mount-zfs.sh \
	parse-zfs.sh \
	zfs-generator.sh \
	zfs-load-key.sh \
	zfs-needshutdown.sh \
	zfs-lib.sh \
	import-opts-generator.sh

	pkgdracut_DATA = \
	zfs-env-bootfs.service \
	zfs-snapshot-bootfs.service \
	zfs-rollback-bootfs.service

	SUBSTFILES += $(pkgdracut_SCRIPTS) $(pkgdracut_DATA)
	+SHELLCHECK_OPTS = --enable=all

	# Provided by /bin/sleep, and, again, every implementation of that supports this
	CHECKBASHISMS_IGNORE = -e 'sleep only takes one integer' -e 'sleep 0.'
	diff --git a/sys/contrib/openzfs/contrib/dracut/90zfs/export-zfs.sh.in b/sys/contrib/openzfs/contrib/dracut/90zfs/export-zfs.sh.in
	index 892650383475..9ad72bd6f5cc 100755
	--- a/sys/contrib/openzfs/contrib/dracut/90zfs/export-zfs.sh.in
	+++ b/sys/contrib/openzfs/contrib/dracut/90zfs/export-zfs.sh.in
	@@ -1,30 +1,30 @@
	#!/bin/sh

	. /lib/dracut-zfs-lib.sh

	_do_zpool_export() {
	ret=0
	errs=""
	final="${1}"

	info "ZFS: Exporting ZFS storage pools..."
	errs=$(export_all -F 2>&1)
	ret=$?
	[ -z "${errs}" ] \|\| echo "${errs}" \| vwarn
	if [ "x${ret}" != "x0" ]; then
	info "ZFS: There was a problem exporting pools."
	fi

	if [ "x${final}" != "x" ]; then
	info "ZFS: pool list"
	zpool list 2>&1 \| vinfo
	fi

	- return ${ret}
	+ return "${ret}"
	}

	if command -v zpool >/dev/null; then
	_do_zpool_export "${1}"
	else
	:
	fi
	diff --git a/sys/contrib/openzfs/contrib/dracut/90zfs/import-opts-generator.sh.in b/sys/contrib/openzfs/contrib/dracut/90zfs/import-opts-generator.sh.in
	index 8bc8c9b35b73..1900676f3b3b 100755
	--- a/sys/contrib/openzfs/contrib/dracut/90zfs/import-opts-generator.sh.in
	+++ b/sys/contrib/openzfs/contrib/dracut/90zfs/import-opts-generator.sh.in
	@@ -1,5 +1,6 @@
	#!/bin/sh

	. /lib/dracut-zfs-lib.sh

	+# shellcheck disable=SC2154
	echo ZPOOL_IMPORT_OPTS="$ZPOOL_IMPORT_OPTS"
	diff --git a/sys/contrib/openzfs/contrib/dracut/90zfs/module-setup.sh.in b/sys/contrib/openzfs/contrib/dracut/90zfs/module-setup.sh.in
	index c27f905bfa8c..10f915b2ef86 100755
	--- a/sys/contrib/openzfs/contrib/dracut/90zfs/module-setup.sh.in
	+++ b/sys/contrib/openzfs/contrib/dracut/90zfs/module-setup.sh.in
	@@ -1,143 +1,143 @@
	#!/usr/bin/env bash
	# shellcheck disable=SC2154

	check() {
	# We depend on udev-rules being loaded
	[ "${1}" = "-d" ] && return 0

	# Verify the zfs tool chain
	for tool in "@sbindir@/zgenhostid" "@sbindir@/zpool" "@sbindir@/zfs" "@mounthelperdir@/mount.zfs" ; do
	test -x "$tool" \|\| return 1
	done

	return 0
	}

	depends() {
	echo udev-rules
	return 0
	}

	installkernel() {
	instmods zfs
	instmods zcommon
	instmods znvpair
	instmods zavl
	instmods zunicode
	instmods zlua
	instmods icp
	instmods spl
	instmods zlib_deflate
	instmods zlib_inflate
	}

	install() {
	inst_rules @udevruledir@/90-zfs.rules
	inst_rules @udevruledir@/69-vdev.rules
	inst_rules @udevruledir@/60-zvol.rules
	dracut_install hostid
	dracut_install grep
	dracut_install @sbindir@/zgenhostid
	dracut_install @sbindir@/zfs
	dracut_install @sbindir@/zpool
	# Workaround for https://github.com/openzfs/zfs/issues/4749 by
	# ensuring libgcc_s.so(.1) is included
	if ldd @sbindir@/zpool \| grep -qF 'libgcc_s.so'; then
	# Dracut will have already tracked and included it
	:;
	elif command -v gcc-config >/dev/null 2>&1; then
	# On systems with gcc-config (Gentoo, Funtoo, etc.):
	# Use the current profile to resolve the appropriate path
	s="$(gcc-config -c)"
	dracut_install "/usr/lib/gcc/${s%-}/${s##-}/libgcc_s.so"*
	elif [ "$(echo /usr/lib/libgcc_s.so)" != "/usr/lib/libgcc_s.so" ]; then
	# Try a simple path first
	dracut_install /usr/lib/libgcc_s.so*
	elif [ "$(echo /lib/libgcc_s.so)" != "/lib/libgcc_s.so" ]; then
	# SUSE
	dracut_install /lib/libgcc_s.so
	else
	# Fallback: Guess the path and include all matches
	dracut_install /usr/lib/gcc//libgcc_s.so
	fi
	# shellcheck disable=SC2050
	- if [ @LIBFETCH_DYNAMIC@ != 0 ]; then
	+ if [ "@LIBFETCH_DYNAMIC@" != 0 ]; then
	for d in $libdirs; do
	- [ -e "$d/"@LIBFETCH_SONAME@ ] && dracut_install "$d/"@LIBFETCH_SONAME@
	+ [ -e "$d/@LIBFETCH_SONAME@" ] && dracut_install "$d/@LIBFETCH_SONAME@"
	done
	fi
	dracut_install @mounthelperdir@/mount.zfs
	dracut_install @udevdir@/vdev_id
	dracut_install awk
	dracut_install cut
	dracut_install tr
	dracut_install head
	dracut_install @udevdir@/zvol_id
	inst_hook cmdline 95 "${moddir}/parse-zfs.sh"
	if [ -n "$systemdutildir" ] ; then
	inst_script "${moddir}/zfs-generator.sh" "$systemdutildir"/system-generators/dracut-zfs-generator
	fi
	inst_hook pre-mount 90 "${moddir}/zfs-load-key.sh"
	inst_hook mount 98 "${moddir}/mount-zfs.sh"
	inst_hook cleanup 99 "${moddir}/zfs-needshutdown.sh"
	inst_hook shutdown 20 "${moddir}/export-zfs.sh"

	inst_simple "${moddir}/zfs-lib.sh" "/lib/dracut-zfs-lib.sh"
	if [ -e @sysconfdir@/zfs/zpool.cache ]; then
	inst @sysconfdir@/zfs/zpool.cache
	type mark_hostonly >/dev/null 2>&1 && mark_hostonly @sysconfdir@/zfs/zpool.cache
	fi

	if [ -e @sysconfdir@/zfs/vdev_id.conf ]; then
	inst @sysconfdir@/zfs/vdev_id.conf
	type mark_hostonly >/dev/null 2>&1 && mark_hostonly @sysconfdir@/zfs/vdev_id.conf
	fi

	# Synchronize initramfs and system hostid
	if [ -f @sysconfdir@/hostid ]; then
	inst @sysconfdir@/hostid
	type mark_hostonly >/dev/null 2>&1 && mark_hostonly @sysconfdir@/hostid
	elif HOSTID="$(hostid 2>/dev/null)" && [ "${HOSTID}" != "00000000" ]; then
	zgenhostid -o "${initdir}@sysconfdir@/hostid" "${HOSTID}"
	type mark_hostonly >/dev/null 2>&1 && mark_hostonly @sysconfdir@/hostid
	fi

	if dracut_module_included "systemd"; then
	mkdir -p "${initdir}/$systemdsystemunitdir/zfs-import.target.wants"
	for _service in "zfs-import-scan.service" "zfs-import-cache.service" ; do
	dracut_install "@systemdunitdir@/$_service"
	if ! [ -L "${initdir}/$systemdsystemunitdir/zfs-import.target.wants/$_service" ]; then
	- ln -sf ../$_service "${initdir}/$systemdsystemunitdir/zfs-import.target.wants/$_service"
	+ ln -sf "../$_service" "${initdir}/$systemdsystemunitdir/zfs-import.target.wants/$_service"
	type mark_hostonly >/dev/null 2>&1 && mark_hostonly "@systemdunitdir@/$_service"
	fi
	done

	inst "${moddir}"/zfs-env-bootfs.service "${systemdsystemunitdir}"/zfs-env-bootfs.service
	ln -s ../zfs-env-bootfs.service "${initdir}/${systemdsystemunitdir}/zfs-import.target.wants"/zfs-env-bootfs.service
	type mark_hostonly >/dev/null 2>&1 && mark_hostonly @systemdunitdir@/zfs-env-bootfs.service

	dracut_install systemd-ask-password
	dracut_install systemd-tty-ask-password-agent

	mkdir -p "${initdir}/$systemdsystemunitdir/initrd.target.wants"
	dracut_install @systemdunitdir@/zfs-import.target
	if ! [ -L "${initdir}/$systemdsystemunitdir/initrd.target.wants"/zfs-import.target ]; then
	ln -s ../zfs-import.target "${initdir}/$systemdsystemunitdir/initrd.target.wants"/zfs-import.target
	type mark_hostonly >/dev/null 2>&1 && mark_hostonly @systemdunitdir@/zfs-import.target
	fi

	for _service in zfs-snapshot-bootfs.service zfs-rollback-bootfs.service ; do
	inst "${moddir}/$_service" "${systemdsystemunitdir}/$_service"
	if ! [ -L "${initdir}/$systemdsystemunitdir/initrd.target.wants/$_service" ]; then
	ln -s "../$_service" "${initdir}/$systemdsystemunitdir/initrd.target.wants/$_service"
	fi
	done

	# There isn't a pkg-config variable for this,
	# and dracut doesn't automatically resolve anything this'd be next to
	local systemdsystemenvironmentgeneratordir
	systemdsystemenvironmentgeneratordir="$(pkg-config --variable=prefix systemd \|\| echo "/usr")/lib/systemd/system-environment-generators"
	mkdir -p "${initdir}/${systemdsystemenvironmentgeneratordir}"
	inst "${moddir}"/import-opts-generator.sh "${systemdsystemenvironmentgeneratordir}"/zfs-import-opts.sh
	fi
	}
	diff --git a/sys/contrib/openzfs/contrib/dracut/90zfs/parse-zfs.sh.in b/sys/contrib/openzfs/contrib/dracut/90zfs/parse-zfs.sh.in
	index 0f92f5c80cce..724c5e2c6dff 100755
	--- a/sys/contrib/openzfs/contrib/dracut/90zfs/parse-zfs.sh.in
	+++ b/sys/contrib/openzfs/contrib/dracut/90zfs/parse-zfs.sh.in
	@@ -1,63 +1,66 @@
	#!/bin/sh
	# shellcheck disable=SC2034,SC2154

	. /lib/dracut-lib.sh

	# Let the command line override our host id.
	spl_hostid=$(getarg spl_hostid=)
	if [ -n "${spl_hostid}" ] ; then
	info "ZFS: Using hostid from command line: ${spl_hostid}"
	zgenhostid -f "${spl_hostid}"
	elif [ -f "/etc/hostid" ] ; then
	info "ZFS: Using hostid from /etc/hostid: $(hostid)"
	else
	warn "ZFS: No hostid found on kernel command line or /etc/hostid."
	warn "ZFS: Pools may not import correctly."
	fi

	wait_for_zfs=0
	case "${root}" in
	""\|zfs\|zfs:)
	# We'll take root unset, root=zfs, or root=zfs:
	# No root set, so we want to read the bootfs attribute. We
	# can't do that until udev settles so we'll set dummy values
	# and hope for the best later on.
	root="zfs:AUTO"
	rootok=1
	wait_for_zfs=1

	info "ZFS: Enabling autodetection of bootfs after udev settles."
	;;

	ZFS=\|zfs:\|FILESYSTEM=*)
	# root is explicit ZFS root. Parse it now. We can handle
	# a root=... param in any of the following formats:
	# root=ZFS=rpool/ROOT
	# root=zfs:rpool/ROOT
	# root=zfs:FILESYSTEM=rpool/ROOT
	# root=FILESYSTEM=rpool/ROOT
	# root=ZFS=pool+with+space/ROOT+WITH+SPACE (translates to root=ZFS=pool with space/ROOT WITH SPACE)

	# Strip down to just the pool/fs
	root="${root#zfs:}"
	root="${root#FILESYSTEM=}"
	root="zfs:${root#ZFS=}"
	# switch + with spaces because kernel cmdline does not allow us to quote parameters
	root=$(echo "$root" \| tr '+' ' ')
	rootok=1
	wait_for_zfs=1

	info "ZFS: Set ${root} as bootfs."
	;;
	+
	+ *)
	+ info "ZFS: no ZFS-on-root"
	esac

	# Make sure Dracut is happy that we have a root and will wait for ZFS
	# modules to settle before mounting.
	-if [ ${wait_for_zfs} -eq 1 ]; then
	+if [ "${wait_for_zfs}" -eq 1 ]; then
	ln -s /dev/null /dev/root 2>/dev/null
	initqueuedir="${hookdir}/initqueue/finished"
	test -d "${initqueuedir}" \|\| {
	initqueuedir="${hookdir}/initqueue-finished"
	}
	echo '[ -e /dev/zfs ]' > "${initqueuedir}/zfs.sh"
	fi
	diff --git a/sys/contrib/openzfs/contrib/dracut/90zfs/zfs-lib.sh.in b/sys/contrib/openzfs/contrib/dracut/90zfs/zfs-lib.sh.in
	index defc0bfc8e76..d7c3e96c1213 100755
	--- a/sys/contrib/openzfs/contrib/dracut/90zfs/zfs-lib.sh.in
	+++ b/sys/contrib/openzfs/contrib/dracut/90zfs/zfs-lib.sh.in
	@@ -1,207 +1,209 @@
	#!/bin/sh

	command -v getarg >/dev/null \|\| . /lib/dracut-lib.sh
	command -v getargbool >/dev/null \|\| {
	# Compatibility with older Dracut versions.
	# With apologies to the Dracut developers.
	getargbool() {
	_default="$1"; shift
	! _b=$(getarg "$@") && [ -z "$_b" ] && _b="$_default"
	if [ -n "$_b" ]; then
	[ "$_b" = "0" ] && return 1
	[ "$_b" = "no" ] && return 1
	[ "$_b" = "off" ] && return 1
	fi
	return 0
	}
	}

	OLDIFS="${IFS}"
	NEWLINE="
	"
	TAB=" "

	ZPOOL_IMPORT_OPTS=""
	if getargbool 0 zfs_force -y zfs.force -y zfsforce ; then
	warn "ZFS: Will force-import pools if necessary."
	ZPOOL_IMPORT_OPTS="${ZPOOL_IMPORT_OPTS} -f"
	fi

	# find_bootfs
	# returns the first dataset with the bootfs attribute.
	find_bootfs() {
	IFS="${NEWLINE}"
	for dataset in $(zpool list -H -o bootfs); do
	case "${dataset}" in
	"" \| "-")
	continue
	;;
	"no pools available")
	IFS="${OLDIFS}"
	return 1
	;;
	*)
	IFS="${OLDIFS}"
	echo "${dataset}"
	return 0
	;;
	esac
	done

	IFS="${OLDIFS}"
	return 1
	}

	# import_pool POOL
	# imports the given zfs pool if it isn't imported already.
	import_pool() {
	pool="${1}"

	if ! zpool list -H "${pool}" > /dev/null 2>&1; then
	info "ZFS: Importing pool ${pool}..."
	# shellcheck disable=SC2086
	if ! zpool import -N ${ZPOOL_IMPORT_OPTS} "${pool}" ; then
	warn "ZFS: Unable to import pool ${pool}"
	return 1
	fi
	fi

	return 0
	}

	_mount_dataset_cb() {
	+ # shellcheck disable=SC2154
	mount -o zfsutil -t zfs "${1}" "${NEWROOT}${2}"
	}

	# mount_dataset DATASET
	# mounts the given zfs dataset.
	mount_dataset() {
	dataset="${1}"
	mountpoint="$(zfs get -H -o value mountpoint "${dataset}")"
	ret=0

	# We need zfsutil for non-legacy mounts and not for legacy mounts.
	if [ "${mountpoint}" = "legacy" ] ; then
	mount -t zfs "${dataset}" "${NEWROOT}" \|\| ret=$?
	else
	mount -o zfsutil -t zfs "${dataset}" "${NEWROOT}" \|\| ret=$?

	if [ "$ret" = "0" ]; then
	for_relevant_root_children "${dataset}" _mount_dataset_cb \|\| ret=$?
	fi
	fi

	- return ${ret}
	+ return "${ret}"
	}

	# for_relevant_root_children DATASET EXEC
	# Runs "EXEC dataset mountpoint" for all children of DATASET that are needed for system bringup
	# Used by zfs-generator.sh and friends, too!
	for_relevant_root_children() {
	dataset="${1}"
	exec="${2}"

	zfs list -t filesystem -Ho name,mountpoint,canmount -r "${dataset}" \|
	(
	_ret=0
	while IFS="${TAB}" read -r dataset mountpoint canmount; do
	[ "$canmount" != "on" ] && continue

	case "$mountpoint" in
	/etc\|/bin\|/lib\|/lib??\|/libx32\|/usr)
	# If these aren't mounted we may not be able to get to the real init at all, or pollute the dataset holding the rootfs
	"${exec}" "${dataset}" "${mountpoint}" \|\| _ret=$?
	;;
	*)
	# Up to the real init to remount everything else it might need
	;;
	esac
	done
	- exit ${_ret}
	+ exit "${_ret}"
	)
	}

	# export_all OPTS
	# exports all imported zfs pools.
	export_all() {
	ret=0

	IFS="${NEWLINE}"
	for pool in $(zpool list -H -o name) ; do
	if zpool list -H "${pool}" > /dev/null 2>&1; then
	zpool export "${pool}" "$@" \|\| ret=$?
	fi
	done
	IFS="${OLDIFS}"

	- return ${ret}
	+ return "${ret}"
	}

	# ask_for_password
	#
	# Wraps around plymouth ask-for-password and adds fallback to tty password ask
	# if plymouth is not present.
	#
	# --cmd command
	# Command to execute. Required.
	# --prompt prompt
	# Password prompt. Note that function already adds ':' at the end.
	# Recommended.
	# --tries n
	# How many times repeat command on its failure. Default is 3.
	# --ply-[cmd\|prompt\|tries]
	# Command/prompt/tries specific for plymouth password ask only.
	# --tty-[cmd\|prompt\|tries]
	# Command/prompt/tries specific for tty password ask only.
	# --tty-echo-off
	# Turn off input echo before tty command is executed and turn on after.
	# It's useful when password is read from stdin.
	ask_for_password() {
	ply_tries=3
	tty_tries=3
	while [ "$#" -gt 0 ]; do
	case "$1" in
	--cmd) ply_cmd="$2"; tty_cmd="$2"; shift;;
	--ply-cmd) ply_cmd="$2"; shift;;
	--tty-cmd) tty_cmd="$2"; shift;;
	--prompt) ply_prompt="$2"; tty_prompt="$2"; shift;;
	--ply-prompt) ply_prompt="$2"; shift;;
	--tty-prompt) tty_prompt="$2"; shift;;
	--tries) ply_tries="$2"; tty_tries="$2"; shift;;
	--ply-tries) ply_tries="$2"; shift;;
	--tty-tries) tty_tries="$2"; shift;;
	--tty-echo-off) tty_echo_off=yes;;
	+ *) echo "ask_for_password(): wrong opt '$1'" >&2;;
	esac
	shift
	done

	{ flock -s 9;
	# Prompt for password with plymouth, if installed and running.
	if plymouth --ping 2>/dev/null; then
	plymouth ask-for-password \
	--prompt "$ply_prompt" --number-of-tries="$ply_tries" \| \
	eval "$ply_cmd"
	ret=$?
	else
	if [ "$tty_echo_off" = yes ]; then
	stty_orig="$(stty -g)"
	stty -echo
	fi

	i=1
	while [ "$i" -le "$tty_tries" ]; do
	[ -n "$tty_prompt" ] && \
	printf "%s [%i/%i]:" "$tty_prompt" "$i" "$tty_tries" >&2
	eval "$tty_cmd" && ret=0 && break
	ret=$?
	i=$((i+1))
	[ -n "$tty_prompt" ] && printf '\n' >&2
	done
	unset i
	[ "$tty_echo_off" = yes ] && stty "$stty_orig"
	fi
	} 9>/.console_lock

	- [ $ret -ne 0 ] && echo "Wrong password" >&2
	- return $ret
	+ [ "$ret" -ne 0 ] && echo "Wrong password" >&2
	+ return "$ret"
	}
	diff --git a/sys/contrib/openzfs/contrib/initramfs/Makefile.am b/sys/contrib/openzfs/contrib/initramfs/Makefile.am
	index 931ceb1316a5..76e66216c40b 100644
	--- a/sys/contrib/openzfs/contrib/initramfs/Makefile.am
	+++ b/sys/contrib/openzfs/contrib/initramfs/Makefile.am
	@@ -1,12 +1,13 @@
	include $(top_srcdir)/config/Shellcheck.am

	initrddir = /usr/share/initramfs-tools

	dist_initrd_SCRIPTS = \
	zfsunlock

	SUBDIRS = conf.d conf-hooks.d hooks scripts
	SHELLCHECKDIRS = hooks scripts
	+SHELLCHECK_OPTS = --enable=all

	EXTRA_DIST = \
	README.initramfs.markdown
	diff --git a/sys/contrib/openzfs/contrib/initramfs/hooks/Makefile.am b/sys/contrib/openzfs/contrib/initramfs/hooks/Makefile.am
	index 0cd1aafcd359..7fb7a75285e1 100644
	--- a/sys/contrib/openzfs/contrib/initramfs/hooks/Makefile.am
	+++ b/sys/contrib/openzfs/contrib/initramfs/hooks/Makefile.am
	@@ -1,10 +1,11 @@
	include $(top_srcdir)/config/Substfiles.am
	include $(top_srcdir)/config/Shellcheck.am

	hooksdir = /usr/share/initramfs-tools/hooks

	hooks_SCRIPTS = \
	zfs \
	zfsunlock

	+SHELLCHECK_OPTS = --enable=all
	SUBSTFILES += $(hooks_SCRIPTS)
	diff --git a/sys/contrib/openzfs/contrib/initramfs/hooks/zfs.in b/sys/contrib/openzfs/contrib/initramfs/hooks/zfs.in
	index 9d5c397cf224..176a1568ec77 100755
	--- a/sys/contrib/openzfs/contrib/initramfs/hooks/zfs.in
	+++ b/sys/contrib/openzfs/contrib/initramfs/hooks/zfs.in
	@@ -1,56 +1,56 @@
	#!/bin/sh
	#
	# Add OpenZFS filesystem capabilities to an initrd, usually for a native ZFS root.
	#

	if [ "$1" = "prereqs" ]; then
	echo "udev"
	exit
	fi

	. /usr/share/initramfs-tools/hook-functions

	for req in "@sbindir@/zpool" "@sbindir@/zfs" "@mounthelperdir@/mount.zfs"; do
	copy_exec "$req" \|\| {
	echo "$req not available!" >&2
	exit 2
	}
	done

	copy_exec "@udevdir@/vdev_id"
	copy_exec "@udevdir@/zvol_id"
	if command -v systemd-ask-password > /dev/null; then
	copy_exec "$(command -v systemd-ask-password)"
	fi

	# We use pthreads, but i-t from buster doesn't automatically
	# copy this indirect dependency: this can be removed when buster finally dies.
	find /lib/ -type f -name "libgcc_s.so.[1-9]" \| while read -r libgcc; do
	copy_exec "$libgcc"
	done

	# shellcheck disable=SC2050
	-if [ @LIBFETCH_DYNAMIC@ != 0 ]; then
	- find /lib/ -name @LIBFETCH_SONAME@ \| while read -r libfetch; do
	+if [ "@LIBFETCH_DYNAMIC@" != 0 ]; then
	+ find /lib/ -name "@LIBFETCH_SONAME@" \| while read -r libfetch; do
	copy_exec "$libfetch"
	done
	fi

	copy_file config "/etc/hostid"
	copy_file cache "@sysconfdir@/zfs/zpool.cache"
	copy_file config "@initconfdir@/zfs"
	copy_file config "@sysconfdir@/zfs/zfs-functions"
	copy_file config "@sysconfdir@/zfs/vdev_id.conf"
	copy_file rule "@udevruledir@/60-zvol.rules"
	copy_file rule "@udevruledir@/69-vdev.rules"

	manual_add_modules zfs

	if [ -f "/etc/hostname" ]; then
	copy_file config "/etc/hostname"
	else
	hostname="$(mktemp -t hostname.XXXXXXXXXX)"
	hostname > "$hostname"
	copy_file config "$hostname" "/etc/hostname"
	rm -f "$hostname"
	fi
	diff --git a/sys/contrib/openzfs/contrib/initramfs/scripts/Makefile.am b/sys/contrib/openzfs/contrib/initramfs/scripts/Makefile.am
	index 444a5f374bfe..470fd792b13a 100644
	--- a/sys/contrib/openzfs/contrib/initramfs/scripts/Makefile.am
	+++ b/sys/contrib/openzfs/contrib/initramfs/scripts/Makefile.am
	@@ -1,11 +1,12 @@
	include $(top_srcdir)/config/Shellcheck.am

	scriptsdir = /usr/share/initramfs-tools/scripts

	dist_scripts_SCRIPTS = \
	zfs

	SUBDIRS = local-top

	SHELLCHECKDIRS = $(SUBDIRS)
	SHELLCHECK_SHELL = sh
	+SHELLCHECK_OPTS = --enable=all
	diff --git a/sys/contrib/openzfs/contrib/initramfs/scripts/local-top/Makefile.am b/sys/contrib/openzfs/contrib/initramfs/scripts/local-top/Makefile.am
	index 897f9b2e2124..a4bb907dd184 100644
	--- a/sys/contrib/openzfs/contrib/initramfs/scripts/local-top/Makefile.am
	+++ b/sys/contrib/openzfs/contrib/initramfs/scripts/local-top/Makefile.am
	@@ -1,6 +1,8 @@
	include $(top_srcdir)/config/Shellcheck.am

	localtopdir = /usr/share/initramfs-tools/scripts/local-top

	dist_localtop_SCRIPTS = \
	zfs
	+
	+SHELLCHECK_OPTS = --enable=all
	diff --git a/sys/contrib/openzfs/contrib/initramfs/scripts/zfs b/sys/contrib/openzfs/contrib/initramfs/scripts/zfs
	index ccb42ce103b9..fcb712cff272 100644
	--- a/sys/contrib/openzfs/contrib/initramfs/scripts/zfs
	+++ b/sys/contrib/openzfs/contrib/initramfs/scripts/zfs
	@@ -1,996 +1,997 @@
	# ZFS boot stub for initramfs-tools.
	#
	# In the initramfs environment, the /init script sources this stub to
	# override the default functions in the /scripts/local script.
	#
	# Enable this by passing boot=zfs on the kernel command line.
	#
	# $quiet, $root, $rpool, $bootfs come from the cmdline:
	# shellcheck disable=SC2154

	# Source the common functions
	. /etc/zfs/zfs-functions

	# Start interactive shell.
	# Use debian's panic() if defined, because it allows to prevent shell access
	# by setting panic in cmdline (e.g. panic=0 or panic=15).
	# See "4.5 Disable root prompt on the initramfs" of Securing Debian Manual:
	# https://www.debian.org/doc/manuals/securing-debian-howto/ch4.en.html
	shell() {
	if command -v panic > /dev/null 2>&1; then
	panic
	else
	/bin/sh
	fi
	}

	# This runs any scripts that should run before we start importing
	# pools and mounting any filesystems.
	pre_mountroot()
	{
	if command -v run_scripts > /dev/null 2>&1
	then
	if [ -f "/scripts/local-top" ] \|\| [ -d "/scripts/local-top" ]
	then
	[ "$quiet" != "y" ] && \
	zfs_log_begin_msg "Running /scripts/local-top"
	run_scripts /scripts/local-top
	[ "$quiet" != "y" ] && zfs_log_end_msg
	fi

	if [ -f "/scripts/local-premount" ] \|\| [ -d "/scripts/local-premount" ]
	then
	[ "$quiet" != "y" ] && \
	zfs_log_begin_msg "Running /scripts/local-premount"
	run_scripts /scripts/local-premount
	[ "$quiet" != "y" ] && zfs_log_end_msg
	fi
	fi
	}

	# If plymouth is available, hide the splash image.
	disable_plymouth()
	{
	if [ -x /bin/plymouth ] && /bin/plymouth --ping
	then
	/bin/plymouth hide-splash >/dev/null 2>&1
	fi
	}

	# Get a ZFS filesystem property value.
	get_fs_value()
	{
	fs="$1"
	value=$2

	"${ZFS}" get -H -ovalue "$value" "$fs" 2> /dev/null
	}

	# Find the 'bootfs' property on pool $1.
	# If the property does not contain '/', then ignore this
	# pool by exporting it again.
	find_rootfs()
	{
	pool="$1"

	# If 'POOL_IMPORTED' isn't set, no pool imported and therefore
	# we won't be able to find a root fs.
	[ -z "${POOL_IMPORTED}" ] && return 1

	# If it's already specified, just keep it mounted and exit
	# User (kernel command line) must be correct.
	if [ -n "${ZFS_BOOTFS}" ] && [ "${ZFS_BOOTFS}" != "zfs:AUTO" ]; then
	return 0
	fi

	# Not set, try to find it in the 'bootfs' property of the pool.
	# NOTE: zpool does not support 'get -H -ovalue bootfs'...
	ZFS_BOOTFS=$("${ZPOOL}" list -H -obootfs "$pool")

	# Make sure it's not '-' and that it starts with /.
	if [ "${ZFS_BOOTFS}" != "-" ] && \
	get_fs_value "${ZFS_BOOTFS}" mountpoint \| grep -q '^/$'
	then
	# Keep it mounted
	POOL_IMPORTED=1
	return 0
	fi

	# Not boot fs here, export it and later try again..
	"${ZPOOL}" export "$pool"
	POOL_IMPORTED=
	ZFS_BOOTFS=
	return 1
	}

	# Support function to get a list of all pools, separated with ';'
	find_pools()
	{
	pools=$("$@" 2> /dev/null \| \
	sed -Ee '/pool:\|^[a-zA-Z0-9]/!d' -e 's@.*: @@' \| \
	tr '\n' ';')

	echo "${pools%%;}" # Return without the last ';'.
	}

	# Get a list of all available pools
	get_pools()
	{
	if [ -n "${ZFS_POOL_IMPORT}" ]; then
	echo "$ZFS_POOL_IMPORT"
	return 0
	fi

	# Get the base list of available pools.
	available_pools=$(find_pools "$ZPOOL" import)

	# Just in case - seen it happen (that a pool isn't visible/found
	# with a simple "zpool import" but only when using the "-d"
	# option or setting ZPOOL_IMPORT_PATH).
	if [ -d "/dev/disk/by-id" ]
	then
	npools=$(find_pools "$ZPOOL" import -d /dev/disk/by-id)
	if [ -n "$npools" ]
	then
	# Because we have found extra pool(s) here, which wasn't
	# found 'normally', we need to force USE_DISK_BY_ID to
	# make sure we're able to actually import it/them later.
	USE_DISK_BY_ID='yes'

	if [ -n "$available_pools" ]
	then
	# Filter out duplicates (pools found with the simple
	# "zpool import" but which is also found with the
	# "zpool import -d ...").
	npools=$(echo "$npools" \| sed "s,$available_pools,,")

	# Add the list to the existing list of
	# available pools
	available_pools="$available_pools;$npools"
	else
	available_pools="$npools"
	fi
	fi
	fi

	# Filter out any exceptions...
	if [ -n "$ZFS_POOL_EXCEPTIONS" ]
	then
	found=""
	apools=""
	OLD_IFS="$IFS" ; IFS=";"

	for pool in $available_pools
	do
	for exception in $ZFS_POOL_EXCEPTIONS
	do
	[ "$pool" = "$exception" ] && continue 2
	found="$pool"
	done

	if [ -n "$found" ]
	then
	if [ -n "$apools" ]
	then
	apools="$apools;$pool"
	else
	apools="$pool"
	fi
	fi
	done

	IFS="$OLD_IFS"
	available_pools="$apools"
	fi

	# Return list of available pools.
	echo "$available_pools"
	}

	# Import given pool $1
	import_pool()
	{
	pool="$1"

	# Verify that the pool isn't already imported
	# Make as sure as we can to not require '-f' to import.
	"${ZPOOL}" get name,guid -o value -H 2>/dev/null \| grep -Fxq "$pool" && return 0

	# For backwards compatibility, make sure that ZPOOL_IMPORT_PATH is set
	# to something we can use later with the real import(s). We want to
	# make sure we find all by* dirs, BUT by-vdev should be first (if it
	# exists).
	if [ -n "$USE_DISK_BY_ID" ] && [ -z "$ZPOOL_IMPORT_PATH" ]
	then
	dirs="$(for dir in /dev/disk/by-*
	do
	# Ignore by-vdev here - we want it first!
	echo "$dir" \| grep -q /by-vdev && continue
	[ ! -d "$dir" ] && continue

	printf "%s" "$dir:"
	done \| sed 's,:$,,g')"

	if [ -d "/dev/disk/by-vdev" ]
	then
	# Add by-vdev at the beginning.
	ZPOOL_IMPORT_PATH="/dev/disk/by-vdev:"
	fi

	# ... and /dev at the very end, just for good measure.
	ZPOOL_IMPORT_PATH="$ZPOOL_IMPORT_PATH$dirs:/dev"
	fi

	# Needs to be exported for "zpool" to catch it.
	[ -n "$ZPOOL_IMPORT_PATH" ] && export ZPOOL_IMPORT_PATH


	[ "$quiet" != "y" ] && zfs_log_begin_msg \
	"Importing pool '${pool}' using defaults"

	ZFS_CMD="${ZPOOL} import -N ${ZPOOL_FORCE} ${ZPOOL_IMPORT_OPTS}"
	ZFS_STDERR="$($ZFS_CMD "$pool" 2>&1)"
	ZFS_ERROR="$?"
	if [ "${ZFS_ERROR}" != 0 ]
	then
	[ "$quiet" != "y" ] && zfs_log_failure_msg "${ZFS_ERROR}"

	if [ -f "${ZPOOL_CACHE}" ]
	then
	[ "$quiet" != "y" ] && zfs_log_begin_msg \
	"Importing pool '${pool}' using cachefile."

	ZFS_CMD="${ZPOOL} import -c ${ZPOOL_CACHE} -N ${ZPOOL_FORCE} ${ZPOOL_IMPORT_OPTS}"
	ZFS_STDERR="$($ZFS_CMD "$pool" 2>&1)"
	ZFS_ERROR="$?"
	fi

	if [ "${ZFS_ERROR}" != 0 ]
	then
	[ "$quiet" != "y" ] && zfs_log_failure_msg "${ZFS_ERROR}"

	disable_plymouth
	echo ""
	echo "Command: ${ZFS_CMD} '$pool'"
	echo "Message: $ZFS_STDERR"
	echo "Error: $ZFS_ERROR"
	echo ""
	echo "Failed to import pool '$pool'."
	echo "Manually import the pool and exit."
	shell
	fi
	fi

	[ "$quiet" != "y" ] && zfs_log_end_msg

	POOL_IMPORTED=1
	return 0
	}

	# Load ZFS modules
	# Loading a module in a initrd require a slightly different approach,
	# with more logging etc.
	load_module_initrd()
	{
	[ -n "$ROOTDELAY" ] && ZFS_INITRD_PRE_MOUNTROOT_SLEEP="$ROOTDELAY"

	if [ "$ZFS_INITRD_PRE_MOUNTROOT_SLEEP" -gt 0 ] 2>/dev/null
	then
	if [ "$quiet" != "y" ]; then
	zfs_log_begin_msg "Sleeping for" \
	"$ZFS_INITRD_PRE_MOUNTROOT_SLEEP seconds..."
	fi
	sleep "$ZFS_INITRD_PRE_MOUNTROOT_SLEEP"
	[ "$quiet" != "y" ] && zfs_log_end_msg
	fi

	# Wait for all of the /dev/{hd,sd}[a-z] device nodes to appear.
	if command -v wait_for_udev > /dev/null 2>&1 ; then
	wait_for_udev 10
	elif command -v wait_for_dev > /dev/null 2>&1 ; then
	wait_for_dev
	fi

	# zpool import refuse to import without a valid /proc/self/mounts
	[ ! -f /proc/self/mounts ] && mount proc /proc

	# Load the module
	load_module "zfs" \|\| return 1

	if [ "$ZFS_INITRD_POST_MODPROBE_SLEEP" -gt 0 ] 2>/dev/null
	then
	if [ "$quiet" != "y" ]; then
	zfs_log_begin_msg "Sleeping for" \
	"$ZFS_INITRD_POST_MODPROBE_SLEEP seconds..."
	fi
	sleep "$ZFS_INITRD_POST_MODPROBE_SLEEP"
	[ "$quiet" != "y" ] && zfs_log_end_msg
	fi

	return 0
	}

	# Mount a given filesystem
	mount_fs()
	{
	fs="$1"

	# Check that the filesystem exists
	"${ZFS}" list -oname -tfilesystem -H "${fs}" > /dev/null 2>&1 \|\| return 1

	# Skip filesystems with canmount=off. The root fs should not have
	# canmount=off, but ignore it for backwards compatibility just in case.
	if [ "$fs" != "${ZFS_BOOTFS}" ]
	then
	canmount=$(get_fs_value "$fs" canmount)
	[ "$canmount" = "off" ] && return 0
	fi

	# Need the _original_ datasets mountpoint!
	mountpoint=$(get_fs_value "$fs" mountpoint)
	ZFS_CMD="mount -o zfsutil -t zfs"
	if [ "$mountpoint" = "legacy" ] \|\| [ "$mountpoint" = "none" ]; then
	# Can't use the mountpoint property. Might be one of our
	# clones. Check the 'org.zol:mountpoint' property set in
	# clone_snap() if that's usable.
	mountpoint=$(get_fs_value "$fs" org.zol:mountpoint)
	if [ "$mountpoint" = "legacy" ] \|\|
	[ "$mountpoint" = "none" ] \|\|
	[ "$mountpoint" = "-" ]
	then
	if [ "$fs" != "${ZFS_BOOTFS}" ]; then
	# We don't have a proper mountpoint and this
	# isn't the root fs.
	return 0
	else
	# Last hail-mary: Hope 'rootmnt' is set!
	mountpoint=""
	fi
	fi

	# If it's not a legacy filesystem, it can only be a
	# native one...
	if [ "$mountpoint" = "legacy" ]; then
	ZFS_CMD="mount -t zfs"
	fi
	fi

	# Possibly decrypt a filesystem using native encryption.
	decrypt_fs "$fs"

	[ "$quiet" != "y" ] && \
	zfs_log_begin_msg "Mounting '${fs}' on '${rootmnt}/${mountpoint}'"
	[ -n "${ZFS_DEBUG}" ] && \
	zfs_log_begin_msg "CMD: '$ZFS_CMD ${fs} ${rootmnt}/${mountpoint}'"

	ZFS_STDERR=$(${ZFS_CMD} "${fs}" "${rootmnt}/${mountpoint}" 2>&1)
	ZFS_ERROR=$?
	if [ "${ZFS_ERROR}" != 0 ]
	then
	[ "$quiet" != "y" ] && zfs_log_failure_msg "${ZFS_ERROR}"

	disable_plymouth
	echo ""
	echo "Command: ${ZFS_CMD} ${fs} ${rootmnt}/${mountpoint}"
	echo "Message: $ZFS_STDERR"
	echo "Error: $ZFS_ERROR"
	echo ""
	echo "Failed to mount ${fs} on ${rootmnt}/${mountpoint}."
	echo "Manually mount the filesystem and exit."
	shell
	else
	[ "$quiet" != "y" ] && zfs_log_end_msg
	fi

	return 0
	}

	# Unlock a ZFS native encrypted filesystem.
	decrypt_fs()
	{
	fs="$1"

	# If pool encryption is active and the zfs command understands '-o encryption'
	if [ "$(zpool list -H -o feature@encryption "${fs%%/*}")" = 'active' ]; then

	# Determine dataset that holds key for root dataset
	ENCRYPTIONROOT="$(get_fs_value "${fs}" encryptionroot)"
	KEYLOCATION="$(get_fs_value "${ENCRYPTIONROOT}" keylocation)"

	echo "${ENCRYPTIONROOT}" > /run/zfs_fs_name

	# If root dataset is encrypted...
	if ! [ "${ENCRYPTIONROOT}" = "-" ]; then
	KEYSTATUS="$(get_fs_value "${ENCRYPTIONROOT}" keystatus)"
	# Continue only if the key needs to be loaded
	[ "$KEYSTATUS" = "unavailable" ] \|\| return 0

	# Do not prompt if key is stored noninteractively,
	if ! [ "${KEYLOCATION}" = "prompt" ]; then
	$ZFS load-key "${ENCRYPTIONROOT}"

	# Prompt with plymouth, if active
	elif /bin/plymouth --ping 2>/dev/null; then
	echo "plymouth" > /run/zfs_console_askpwd_cmd
	for _ in 1 2 3; do
	plymouth ask-for-password --prompt "Encrypted ZFS password for ${ENCRYPTIONROOT}" \| \
	$ZFS load-key "${ENCRYPTIONROOT}" && break
	done

	# Prompt with systemd, if active
	elif [ -e /run/systemd/system ]; then
	echo "systemd-ask-password" > /run/zfs_console_askpwd_cmd
	for _ in 1 2 3; do
	systemd-ask-password "Encrypted ZFS password for ${ENCRYPTIONROOT}" --no-tty \| \
	$ZFS load-key "${ENCRYPTIONROOT}" && break
	done

	# Prompt with ZFS tty, otherwise
	else
	# Temporarily setting "printk" to "7" allows the prompt to appear even when the "quiet" kernel option has been used
	echo "load-key" > /run/zfs_console_askpwd_cmd
	read -r storeprintk _ < /proc/sys/kernel/printk
	echo 7 > /proc/sys/kernel/printk
	$ZFS load-key "${ENCRYPTIONROOT}"
	echo "$storeprintk" > /proc/sys/kernel/printk
	fi
	fi
	fi

	return 0
	}

	# Destroy a given filesystem.
	destroy_fs()
	{
	fs="$1"

	[ "$quiet" != "y" ] && \
	zfs_log_begin_msg "Destroying '$fs'"

	ZFS_CMD="${ZFS} destroy $fs"
	ZFS_STDERR="$(${ZFS_CMD} 2>&1)"
	ZFS_ERROR="$?"
	if [ "${ZFS_ERROR}" != 0 ]
	then
	[ "$quiet" != "y" ] && zfs_log_failure_msg "${ZFS_ERROR}"

	disable_plymouth
	echo ""
	echo "Command: $ZFS_CMD"
	echo "Message: $ZFS_STDERR"
	echo "Error: $ZFS_ERROR"
	echo ""
	echo "Failed to destroy '$fs'. Please make sure that '$fs' is not available."
	echo "Hint: Try: zfs destroy -Rfn $fs"
	echo "If this dryrun looks good, then remove the 'n' from '-Rfn' and try again."
	shell
	else
	[ "$quiet" != "y" ] && zfs_log_end_msg
	fi

	return 0
	}

	# Clone snapshot $1 to destination filesystem $2
	# Set 'canmount=noauto' and 'mountpoint=none' so that we get to keep
	# manual control over it's mounting (i.e., make sure it's not automatically
	# mounted with a 'zfs mount -a' in the init/systemd scripts).
	clone_snap()
	{
	snap="$1"
	destfs="$2"
	mountpoint="$3"

	[ "$quiet" != "y" ] && zfs_log_begin_msg "Cloning '$snap' to '$destfs'"

	# Clone the snapshot into a dataset we can boot from
	# + We don't want this filesystem to be automatically mounted, we
	# want control over this here and nowhere else.
	# + We don't need any mountpoint set for the same reason.
	# We use the 'org.zol:mountpoint' property to remember the mountpoint.
	ZFS_CMD="${ZFS} clone -o canmount=noauto -o mountpoint=none"
	ZFS_CMD="${ZFS_CMD} -o org.zol:mountpoint=${mountpoint}"
	ZFS_CMD="${ZFS_CMD} $snap $destfs"
	ZFS_STDERR="$(${ZFS_CMD} 2>&1)"
	ZFS_ERROR="$?"
	if [ "${ZFS_ERROR}" != 0 ]
	then
	[ "$quiet" != "y" ] && zfs_log_failure_msg "${ZFS_ERROR}"

	disable_plymouth
	echo ""
	echo "Command: $ZFS_CMD"
	echo "Message: $ZFS_STDERR"
	echo "Error: $ZFS_ERROR"
	echo ""
	echo "Failed to clone snapshot."
	echo "Make sure that the any problems are corrected and then make sure"
	echo "that the dataset '$destfs' exists and is bootable."
	shell
	else
	[ "$quiet" != "y" ] && zfs_log_end_msg
	fi

	return 0
	}

	# Rollback a given snapshot.
	rollback_snap()
	{
	snap="$1"

	[ "$quiet" != "y" ] && zfs_log_begin_msg "Rollback $snap"

	ZFS_CMD="${ZFS} rollback -Rf $snap"
	ZFS_STDERR="$(${ZFS_CMD} 2>&1)"
	ZFS_ERROR="$?"
	if [ "${ZFS_ERROR}" != 0 ]
	then
	[ "$quiet" != "y" ] && zfs_log_failure_msg "${ZFS_ERROR}"

	disable_plymouth
	echo ""
	echo "Command: $ZFS_CMD"
	echo "Message: $ZFS_STDERR"
	echo "Error: $ZFS_ERROR"
	echo ""
	echo "Failed to rollback snapshot."
	shell
	else
	[ "$quiet" != "y" ] && zfs_log_end_msg
	fi

	return 0
	}

	# Get a list of snapshots, give them as a numbered list
	# to the user to choose from.
	ask_user_snap()
	{
	fs="$1"

	# We need to temporarily disable debugging. Set 'debug' so we
	# remember to enabled it again.
	if [ -n "${ZFS_DEBUG}" ]; then
	unset ZFS_DEBUG
	set +x
	debug=1
	fi

	# Because we need the resulting snapshot, which is sent on
	# stdout to the caller, we use stderr for our questions.
	echo "What snapshot do you want to boot from?" > /dev/stderr
	# shellcheck disable=SC2046
	IFS="
	" set -- $("${ZFS}" list -H -oname -tsnapshot -r "${fs}")

	i=1
	for snap in "$@"; do
	echo " $i: $snap"
	i=$((i + 1))
	done > /dev/stderr

	# expr instead of test here because [ a -lt 0 ] errors out,
	# but expr falls back to lexicographical, which works out right
	snapnr=0
	while expr "$snapnr" "<" 1 > /dev/null \|\|
	expr "$snapnr" ">" "$#" > /dev/null
	do
	printf "%s" "Snap nr [1-$#]? " > /dev/stderr
	read -r snapnr
	done

	# Re-enable debugging.
	if [ -n "${debug}" ]; then
	ZFS_DEBUG=1
	set -x
	fi

	eval echo '$'"$snapnr"
	}

	setup_snapshot_booting()
	{
	snap="$1"
	retval=0

	# Make sure that the snapshot specified actually exists.
	- if [ ! "$(get_fs_value "${snap}" type)" ]
	+ if [ -z "$(get_fs_value "${snap}" type)" ]
	then
	# Snapshot does not exist (...@<null> ?)
	# ask the user for a snapshot to use.
	snap="$(ask_user_snap "${snap%%@*}")"
	fi

	# Separate the full snapshot ('$snap') into it's filesystem and
	# snapshot names. Would have been nice with a split() function..
	rootfs="${snap%%@*}"
	snapname="${snap##*@}"
	ZFS_BOOTFS="${rootfs}_${snapname}"

	if ! grep -qiE '(^\|[^\\](\\\\)* )(rollback)=(on\|yes\|1)( \|$)' /proc/cmdline
	then
	# If the destination dataset for the clone
	# already exists, destroy it. Recursively
	- if [ "$(get_fs_value "${rootfs}_${snapname}" type)" ]; then
	+ if [ -n "$(get_fs_value "${rootfs}_${snapname}" type)" ]
	+ then
	filesystems=$("${ZFS}" list -oname -tfilesystem -H \
	-r -Sname "${ZFS_BOOTFS}")
	for fs in $filesystems; do
	destroy_fs "${fs}"
	done
	fi
	fi

	# Get all snapshots, recursively (might need to clone /usr, /var etc
	# as well).
	for s in $("${ZFS}" list -H -oname -tsnapshot -r "${rootfs}" \| \
	grep "${snapname}")
	do
	if grep -qiE '(^\|[^\\](\\\\)* )(rollback)=(on\|yes\|1)( \|$)' /proc/cmdline
	then
	# Rollback snapshot
	rollback_snap "$s" \|\| retval=$((retval + 1))
	ZFS_BOOTFS="${rootfs}"
	else
	# Setup a destination filesystem name.
	# Ex: Called with 'rpool/ROOT/debian@snap2'
	# rpool/ROOT/debian@snap2 => rpool/ROOT/debian_snap2
	# rpool/ROOT/debian/boot@snap2 => rpool/ROOT/debian_snap2/boot
	# rpool/ROOT/debian/usr@snap2 => rpool/ROOT/debian_snap2/usr
	# rpool/ROOT/debian/var@snap2 => rpool/ROOT/debian_snap2/var
	subfs="${s##$rootfs}"
	subfs="${subfs%%@$snapname}"

	destfs="${rootfs}_${snapname}" # base fs.
	[ -n "$subfs" ] && destfs="${destfs}$subfs" # + sub fs.

	# Get the mountpoint of the filesystem, to be used
	# with clone_snap(). If legacy or none, then use
	# the sub fs value.
	mountpoint=$(get_fs_value "${s%%@*}" mountpoint)
	if [ "$mountpoint" = "legacy" ] \|\| \
	[ "$mountpoint" = "none" ]
	then
	if [ -n "${subfs}" ]; then
	mountpoint="${subfs}"
	else
	mountpoint="/"
	fi
	fi

	# Clone the snapshot into its own
	# filesystem
	clone_snap "$s" "${destfs}" "${mountpoint}" \|\| \
	retval=$((retval + 1))
	fi
	done

	# If we haven't return yet, we have a problem...
	return "${retval}"
	}

	# ================================================================

	# This is the main function.
	mountroot()
	{
	# ----------------------------------------------------------------
	# I N I T I A L S E T U P

	# ------------
	# Run the pre-mount scripts from /scripts/local-top.
	pre_mountroot

	# ------------
	# Source the default setup variables.
	[ -r '/etc/default/zfs' ] && . /etc/default/zfs

	# ------------
	# Support debug option
	if grep -qiE '(^\|[^\\](\\\\)* )(zfs_debug\|zfs\.debug\|zfsdebug)=(on\|yes\|1)( \|$)' /proc/cmdline
	then
	ZFS_DEBUG=1
	mkdir /var/log
	#exec 2> /var/log/boot.debug
	set -x
	fi

	# ------------
	# Load ZFS module etc.
	if ! load_module_initrd; then
	disable_plymouth
	echo ""
	echo "Failed to load ZFS modules."
	echo "Manually load the modules and exit."
	shell
	fi

	# ------------
	# Look for the cache file (if any).
	[ -f "${ZPOOL_CACHE}" ] \|\| unset ZPOOL_CACHE
	[ -s "${ZPOOL_CACHE}" ] \|\| unset ZPOOL_CACHE

	# ------------
	# Compatibility: 'ROOT' is for Debian GNU/Linux (etc),
	# 'root' is for Redhat/Fedora (etc),
	# 'REAL_ROOT' is for Gentoo
	if [ -z "$ROOT" ]
	then
	[ -n "$root" ] && ROOT=${root}

	[ -n "$REAL_ROOT" ] && ROOT=${REAL_ROOT}
	fi

	# ------------
	# Where to mount the root fs in the initrd - set outside this script
	# Compatibility: 'rootmnt' is for Debian GNU/Linux (etc),
	# 'NEWROOT' is for RedHat/Fedora (etc),
	# 'NEW_ROOT' is for Gentoo
	if [ -z "$rootmnt" ]
	then
	[ -n "$NEWROOT" ] && rootmnt=${NEWROOT}

	[ -n "$NEW_ROOT" ] && rootmnt=${NEW_ROOT}
	fi

	# ------------
	# No longer set in the defaults file, but it could have been set in
	# get_pools() in some circumstances. If it's something, but not 'yes',
	# it's no good to us.
	[ -n "$USE_DISK_BY_ID" ] && [ "$USE_DISK_BY_ID" != 'yes' ] && \
	unset USE_DISK_BY_ID

	# ----------------------------------------------------------------
	# P A R S E C O M M A N D L I N E O P T I O N S

	# This part is the really ugly part - there's so many options and permutations
	# 'out there', and if we should make this the 'primary' source for ZFS initrd
	# scripting, we need/should support them all.
	#
	# Supports the following kernel command line argument combinations
	# (in this order - first match win):
	#
	# rpool=<pool> (tries to finds bootfs automatically)
	# bootfs=<pool>/<dataset> (uses this for rpool - first part)
	# rpool=<pool> bootfs=<pool>/<dataset>
	# -B zfs-bootfs=<pool>/<fs> (uses this for rpool - first part)
	# rpool=rpool (default if none of the above is used)
	# root=<pool>/<dataset> (uses this for rpool - first part)
	# root=ZFS=<pool>/<dataset> (uses this for rpool - first part, without 'ZFS=')
	# root=zfs:AUTO (tries to detect both pool and rootfs)
	# root=zfs:<pool>/<dataset> (uses this for rpool - first part, without 'zfs:')
	#
	# Option <dataset> could also be <snapshot>
	# Option <pool> could also be <guid>

	# ------------
	# Support force option
	# In addition, setting one of zfs_force, zfs.force or zfsforce to
	# 'yes', 'on' or '1' will make sure we force import the pool.
	# This should (almost) never be needed, but it's here for
	# completeness.
	ZPOOL_FORCE=""
	if grep -qiE '(^\|[^\\](\\\\)* )(zfs_force\|zfs\.force\|zfsforce)=(on\|yes\|1)( \|$)' /proc/cmdline
	then
	ZPOOL_FORCE="-f"
	fi

	# ------------
	# Look for 'rpool' and 'bootfs' parameter
	[ -n "$rpool" ] && ZFS_RPOOL="${rpool#rpool=}"
	[ -n "$bootfs" ] && ZFS_BOOTFS="${bootfs#bootfs=}"

	# ------------
	# If we have 'ROOT' (see above), but not 'ZFS_BOOTFS', then use
	# 'ROOT'
	[ -n "$ROOT" ] && [ -z "${ZFS_BOOTFS}" ] && ZFS_BOOTFS="$ROOT"

	# ------------
	# Check for the `-B zfs-bootfs=%s/%u,...` kind of parameter.
	# NOTE: Only use the pool name and dataset. The rest is not
	# supported by OpenZFS (whatever it's for).
	if [ -z "$ZFS_RPOOL" ]
	then
	# The ${zfs-bootfs} variable is set at the kernel command
	# line, usually by GRUB, but it cannot be referenced here
	# directly because bourne variable names cannot contain a
	# hyphen.
	#
	# Reassign the variable by dumping the environment and
	# stripping the zfs-bootfs= prefix. Let the shell handle
	# quoting through the eval command:
	# shellcheck disable=SC2046
	eval ZFS_RPOOL=$(set \| sed -n -e 's,^zfs-bootfs=,,p')
	fi

	# ------------
	# No root fs or pool specified - do auto detect.
	if [ -z "$ZFS_RPOOL" ] && [ -z "${ZFS_BOOTFS}" ]
	then
	# Do auto detect. Do this by 'cheating' - set 'root=zfs:AUTO'
	# which will be caught later
	ROOT='zfs:AUTO'
	fi

	# ----------------------------------------------------------------
	# F I N D A N D I M P O R T C O R R E C T P O O L

	# ------------
	if [ "$ROOT" = "zfs:AUTO" ]
	then
	# Try to detect both pool and root fs.

	# If we got here, that means we don't have a hint so as to
	# the root dataset, but with root=zfs:AUTO on cmdline,
	# this says "zfs:AUTO" here and interferes with checks later
	ZFS_BOOTFS=

	[ "$quiet" != "y" ] && \
	zfs_log_begin_msg "Attempting to import additional pools."

	# Get a list of pools available for import
	if [ -n "$ZFS_RPOOL" ]
	then
	# We've specified a pool - check only that
	POOLS=$ZFS_RPOOL
	else
	POOLS=$(get_pools)
	fi

	OLD_IFS="$IFS" ; IFS=";"
	for pool in $POOLS
	do
	[ -z "$pool" ] && continue

	IFS="$OLD_IFS" import_pool "$pool"
	IFS="$OLD_IFS" find_rootfs "$pool" && break
	done
	IFS="$OLD_IFS"

	- [ "$quiet" != "y" ] && zfs_log_end_msg $ZFS_ERROR
	+ [ "$quiet" != "y" ] && zfs_log_end_msg "$ZFS_ERROR"
	else
	# No auto - use value from the command line option.

	# Strip 'zfs:' and 'ZFS='.
	ZFS_BOOTFS="${ROOT#*[:=]}"

	# Strip everything after the first slash.
	ZFS_RPOOL="${ZFS_BOOTFS%%/*}"
	fi

	# Import the pool (if not already done so in the AUTO check above).
	if [ -n "$ZFS_RPOOL" ] && [ -z "${POOL_IMPORTED}" ]
	then
	[ "$quiet" != "y" ] && \
	zfs_log_begin_msg "Importing ZFS root pool '$ZFS_RPOOL'"

	import_pool "${ZFS_RPOOL}"
	find_rootfs "${ZFS_RPOOL}"

	[ "$quiet" != "y" ] && zfs_log_end_msg
	fi

	if [ -z "${POOL_IMPORTED}" ]
	then
	# No pool imported, this is serious!
	disable_plymouth
	echo ""
	echo "Command: $ZFS_CMD"
	echo "Message: $ZFS_STDERR"
	echo "Error: $ZFS_ERROR"
	echo ""
	echo "No pool imported. Manually import the root pool"
	echo "at the command prompt and then exit."
	echo "Hint: Try: zpool import -N ${ZFS_RPOOL}"
	shell
	fi

	# In case the pool was specified as guid, resolve guid to name
	pool="$("${ZPOOL}" get name,guid -o name,value -H \| \
	awk -v pool="${ZFS_RPOOL}" '$2 == pool { print $1 }')"
	if [ -n "$pool" ]; then
	# If $ZFS_BOOTFS contains guid, replace the guid portion with $pool
	ZFS_BOOTFS=$(echo "$ZFS_BOOTFS" \| \
	sed -e "s/$("${ZPOOL}" get guid -o value "$pool" -H)/$pool/g")
	ZFS_RPOOL="${pool}"
	fi


	# ----------------------------------------------------------------
	# P R E P A R E R O O T F I L E S Y S T E M

	if [ -n "${ZFS_BOOTFS}" ]
	then
	# Booting from a snapshot?
	# Will overwrite the ZFS_BOOTFS variable like so:
	# rpool/ROOT/debian@snap2 => rpool/ROOT/debian_snap2
	echo "${ZFS_BOOTFS}" \| grep -q '@' && \
	setup_snapshot_booting "${ZFS_BOOTFS}"
	fi

	if [ -z "${ZFS_BOOTFS}" ]
	then
	# Still nothing! Let the user sort this out.
	disable_plymouth
	echo ""
	echo "Error: Unknown root filesystem - no 'bootfs' pool property and"
	echo " not specified on the kernel command line."
	echo ""
	echo "Manually mount the root filesystem on $rootmnt and then exit."
	echo "Hint: Try: mount -o zfsutil -t zfs ${ZFS_RPOOL-rpool}/ROOT/system $rootmnt"
	shell
	fi

	# ----------------------------------------------------------------
	# M O U N T F I L E S Y S T E M S

	# * Ideally, the root filesystem would be mounted like this:
	#
	# zpool import -R "$rootmnt" -N "$ZFS_RPOOL"
	# zfs mount -o mountpoint=/ "${ZFS_BOOTFS}"
	#
	# but the MOUNTPOINT prefix is preserved on descendent filesystem
	# after the pivot into the regular root, which later breaks things
	# like `zfs mount -a` and the /proc/self/mounts refresh.
	#
	# * Mount additional filesystems required
	# Such as /usr, /var, /usr/local etc.
	# NOTE: Mounted in the order specified in the
	# ZFS_INITRD_ADDITIONAL_DATASETS variable so take care!

	# Go through the complete list (recursively) of all filesystems below
	# the real root dataset
	filesystems="$("${ZFS}" list -oname -tfilesystem -H -r "${ZFS_BOOTFS}")"
	OLD_IFS="$IFS" ; IFS="
	"
	for fs in $filesystems; do
	IFS="$OLD_IFS" mount_fs "$fs"
	done
	IFS="$OLD_IFS"
	for fs in $ZFS_INITRD_ADDITIONAL_DATASETS; do
	mount_fs "$fs"
	done

	touch /run/zfs_unlock_complete
	if [ -e /run/zfs_unlock_complete_notify ]; then
	read -r < /run/zfs_unlock_complete_notify
	fi

	# ------------
	# Debugging information
	if [ -n "${ZFS_DEBUG}" ]
	then
	#exec 2>&1-

	echo "DEBUG: imported pools:"
	"${ZPOOL}" list -H
	echo

	echo "DEBUG: mounted ZFS filesystems:"
	mount \| grep zfs
	echo

	echo "=> waiting for ENTER before continuing because of 'zfsdebug=1'. "
	printf "%s" " 'c' for shell, 'r' for reboot, 'ENTER' to continue. "
	read -r b

	[ "$b" = "c" ] && /bin/sh
	[ "$b" = "r" ] && reboot -f

	set +x
	fi

	# ------------
	# Run local bottom script
	if command -v run_scripts > /dev/null 2>&1
	then
	if [ -f "/scripts/local-bottom" ] \|\| [ -d "/scripts/local-bottom" ]
	then
	[ "$quiet" != "y" ] && \
	zfs_log_begin_msg "Running /scripts/local-bottom"
	run_scripts /scripts/local-bottom
	[ "$quiet" != "y" ] && zfs_log_end_msg
	fi
	fi
	}
	diff --git a/sys/contrib/openzfs/contrib/pyzfs/libzfs_core/_constants.py b/sys/contrib/openzfs/contrib/pyzfs/libzfs_core/_constants.py
	index 2dfed224c29d..3273652f758a 100644
	--- a/sys/contrib/openzfs/contrib/pyzfs/libzfs_core/_constants.py
	+++ b/sys/contrib/openzfs/contrib/pyzfs/libzfs_core/_constants.py
	@@ -1,114 +1,116 @@
	#
	# Copyright 2015 ClusterHQ
	#
	# Licensed under the Apache License, Version 2.0 (the "License");
	# you may not use this file except in compliance with the License.
	# You may obtain a copy of the License at
	#
	# http://www.apache.org/licenses/LICENSE-2.0
	#
	# Unless required by applicable law or agreed to in writing, software
	# distributed under the License is distributed on an "AS IS" BASIS,
	# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
	# See the License for the specific language governing permissions and
	# limitations under the License.
	#

	"""
	Important `libzfs_core` constants.
	"""

	from __future__ import absolute_import, division, print_function
	import errno
	import sys


	# Compat for platform-specific errnos
	if sys.platform.startswith('freebsd'):
	ECHRNG = errno.ENXIO
	ECKSUM = 97 # EINTEGRITY
	ETIME = errno.ETIMEDOUT
	else:
	ECHRNG = errno.ECHRNG
	ECKSUM = errno.EBADE
	ETIME = errno.ETIME


	# https://stackoverflow.com/a/1695250
	def enum_with_offset(offset, sequential, named):
	enums = dict(((b, a + offset) for a, b in enumerate(sequential)), **named)
	return type('Enum', (), enums)


	def enum(sequential, *named):
	return enum_with_offset(0, sequential, named)


	#: Maximum length of any ZFS name.
	MAXNAMELEN = 255
	#: Default channel program limits
	ZCP_DEFAULT_INSTRLIMIT = 10 * 1000 * 1000
	ZCP_DEFAULT_MEMLIMIT = 10 * 1024 * 1024
	#: Encryption wrapping key length
	WRAPPING_KEY_LEN = 32
	#: Encryption key location enum
	zfs_key_location = enum(
	'ZFS_KEYLOCATION_NONE',
	'ZFS_KEYLOCATION_PROMPT',
	'ZFS_KEYLOCATION_URI'
	)
	#: Encryption key format enum
	zfs_keyformat = enum(
	'ZFS_KEYFORMAT_NONE',
	'ZFS_KEYFORMAT_RAW',
	'ZFS_KEYFORMAT_HEX',
	'ZFS_KEYFORMAT_PASSPHRASE'
	)
	# Encryption algorithms enum
	zio_encrypt = enum(
	'ZIO_CRYPT_INHERIT',
	'ZIO_CRYPT_ON',
	'ZIO_CRYPT_OFF',
	'ZIO_CRYPT_AES_128_CCM',
	'ZIO_CRYPT_AES_192_CCM',
	'ZIO_CRYPT_AES_256_CCM',
	'ZIO_CRYPT_AES_128_GCM',
	'ZIO_CRYPT_AES_192_GCM',
	'ZIO_CRYPT_AES_256_GCM'
	)
	# ZFS-specific error codes
	zfs_errno = enum_with_offset(1024, [
	'ZFS_ERR_CHECKPOINT_EXISTS',
	'ZFS_ERR_DISCARDING_CHECKPOINT',
	'ZFS_ERR_NO_CHECKPOINT',
	'ZFS_ERR_DEVRM_IN_PROGRESS',
	'ZFS_ERR_VDEV_TOO_BIG',
	'ZFS_ERR_IOC_CMD_UNAVAIL',
	'ZFS_ERR_IOC_ARG_UNAVAIL',
	'ZFS_ERR_IOC_ARG_REQUIRED',
	'ZFS_ERR_IOC_ARG_BADTYPE',
	'ZFS_ERR_WRONG_PARENT',
	'ZFS_ERR_FROM_IVSET_GUID_MISSING',
	'ZFS_ERR_FROM_IVSET_GUID_MISMATCH',
	'ZFS_ERR_SPILL_BLOCK_FLAG_MISSING',
	'ZFS_ERR_UNKNOWN_SEND_STREAM_FEATURE',
	'ZFS_ERR_EXPORT_IN_PROGRESS',
	'ZFS_ERR_BOOKMARK_SOURCE_NOT_ANCESTOR',
	'ZFS_ERR_STREAM_TRUNCATED',
	'ZFS_ERR_STREAM_LARGE_BLOCK_MISMATCH',
	'ZFS_ERR_RESILVER_IN_PROGRESS',
	'ZFS_ERR_REBUILD_IN_PROGRESS',
	'ZFS_ERR_BADPROP',
	+ 'ZFS_ERR_VDEV_NOTSUP',
	],
	{}
	)
	# compat before we used the enum helper for these values
	ZFS_ERR_CHECKPOINT_EXISTS = zfs_errno.ZFS_ERR_CHECKPOINT_EXISTS
	assert(ZFS_ERR_CHECKPOINT_EXISTS == 1024)
	ZFS_ERR_DISCARDING_CHECKPOINT = zfs_errno.ZFS_ERR_DISCARDING_CHECKPOINT
	ZFS_ERR_NO_CHECKPOINT = zfs_errno.ZFS_ERR_NO_CHECKPOINT
	ZFS_ERR_DEVRM_IN_PROGRESS = zfs_errno.ZFS_ERR_DEVRM_IN_PROGRESS
	ZFS_ERR_VDEV_TOO_BIG = zfs_errno.ZFS_ERR_VDEV_TOO_BIG
	ZFS_ERR_WRONG_PARENT = zfs_errno.ZFS_ERR_WRONG_PARENT
	+ZFS_ERR_VDEV_NOTSUP = zfs_errno.ZFS_ERR_VDEV_NOTSUP

	# vim: softtabstop=4 tabstop=4 expandtab shiftwidth=4
	diff --git a/sys/contrib/openzfs/contrib/pyzfs/libzfs_core/bindings/libnvpair.py b/sys/contrib/openzfs/contrib/pyzfs/libzfs_core/bindings/libnvpair.py
	index 3cd72d4908de..54a805907237 100644
	--- a/sys/contrib/openzfs/contrib/pyzfs/libzfs_core/bindings/libnvpair.py
	+++ b/sys/contrib/openzfs/contrib/pyzfs/libzfs_core/bindings/libnvpair.py
	@@ -1,134 +1,144 @@
	#
	# Copyright 2015 ClusterHQ
	#
	# Licensed under the Apache License, Version 2.0 (the "License");
	# you may not use this file except in compliance with the License.
	# You may obtain a copy of the License at
	#
	# http://www.apache.org/licenses/LICENSE-2.0
	#
	# Unless required by applicable law or agreed to in writing, software
	# distributed under the License is distributed on an "AS IS" BASIS,
	# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
	# See the License for the specific language governing permissions and
	# limitations under the License.
	#

	"""
	Python bindings for ``libnvpair``.
	"""
	from __future__ import absolute_import, division, print_function

	CDEF = """
	typedef ... nvlist_t;
	typedef ... nvpair_t;


	typedef enum {
	DATA_TYPE_UNKNOWN = 0,
	DATA_TYPE_BOOLEAN,
	DATA_TYPE_BYTE,
	DATA_TYPE_INT16,
	DATA_TYPE_UINT16,
	DATA_TYPE_INT32,
	DATA_TYPE_UINT32,
	DATA_TYPE_INT64,
	DATA_TYPE_UINT64,
	DATA_TYPE_STRING,
	DATA_TYPE_BYTE_ARRAY,
	DATA_TYPE_INT16_ARRAY,
	DATA_TYPE_UINT16_ARRAY,
	DATA_TYPE_INT32_ARRAY,
	DATA_TYPE_UINT32_ARRAY,
	DATA_TYPE_INT64_ARRAY,
	DATA_TYPE_UINT64_ARRAY,
	DATA_TYPE_STRING_ARRAY,
	DATA_TYPE_HRTIME,
	DATA_TYPE_NVLIST,
	DATA_TYPE_NVLIST_ARRAY,
	DATA_TYPE_BOOLEAN_VALUE,
	DATA_TYPE_INT8,
	DATA_TYPE_UINT8,
	DATA_TYPE_BOOLEAN_ARRAY,
	DATA_TYPE_INT8_ARRAY,
	DATA_TYPE_UINT8_ARRAY
	} data_type_t;
	typedef enum { B_FALSE, B_TRUE } boolean_t;

	typedef unsigned char uchar_t;
	typedef unsigned int uint_t;

	int nvlist_alloc(nvlist_t **, uint_t, int);
	void nvlist_free(nvlist_t *);

	int nvlist_unpack(char , size_t, nvlist_t *, int);

	void dump_nvlist(nvlist_t *, int);
	int nvlist_dup(nvlist_t , nvlist_t *, int);

	int nvlist_add_boolean(nvlist_t , const char );
	int nvlist_add_boolean_value(nvlist_t , const char , boolean_t);
	int nvlist_add_byte(nvlist_t , const char , uchar_t);
	int nvlist_add_int8(nvlist_t , const char , int8_t);
	int nvlist_add_uint8(nvlist_t , const char , uint8_t);
	int nvlist_add_int16(nvlist_t , const char , int16_t);
	int nvlist_add_uint16(nvlist_t , const char , uint16_t);
	int nvlist_add_int32(nvlist_t , const char , int32_t);
	int nvlist_add_uint32(nvlist_t , const char , uint32_t);
	int nvlist_add_int64(nvlist_t , const char , int64_t);
	int nvlist_add_uint64(nvlist_t , const char , uint64_t);
	int nvlist_add_string(nvlist_t , const char , const char *);
	int nvlist_add_nvlist(nvlist_t , const char , nvlist_t *);
	- int nvlist_add_boolean_array(nvlist_t , const char , boolean_t *,
	- uint_t);
	- int nvlist_add_byte_array(nvlist_t , const char , uchar_t *, uint_t);
	- int nvlist_add_int8_array(nvlist_t , const char , int8_t *, uint_t);
	- int nvlist_add_uint8_array(nvlist_t , const char , uint8_t *, uint_t);
	- int nvlist_add_int16_array(nvlist_t , const char , int16_t *, uint_t);
	- int nvlist_add_uint16_array(nvlist_t , const char , uint16_t *, uint_t);
	- int nvlist_add_int32_array(nvlist_t , const char , int32_t *, uint_t);
	- int nvlist_add_uint32_array(nvlist_t , const char , uint32_t *, uint_t);
	- int nvlist_add_int64_array(nvlist_t , const char , int64_t *, uint_t);
	- int nvlist_add_uint64_array(nvlist_t , const char , uint64_t *, uint_t);
	- int nvlist_add_string_array(nvlist_t , const char , char const ,
	- uint_t);
	- int nvlist_add_nvlist_array(nvlist_t , const char , nvlist_t **, uint_t);
	+ int nvlist_add_boolean_array(nvlist_t , const char ,
	+ const boolean_t *, uint_t);
	+ int nvlist_add_byte_array(nvlist_t , const char ,
	+ const uchar_t *, uint_t);
	+ int nvlist_add_int8_array(nvlist_t , const char ,
	+ const int8_t *, uint_t);
	+ int nvlist_add_uint8_array(nvlist_t , const char ,
	+ const uint8_t *, uint_t);
	+ int nvlist_add_int16_array(nvlist_t , const char ,
	+ const int16_t *, uint_t);
	+ int nvlist_add_uint16_array(nvlist_t , const char ,
	+ const uint16_t *, uint_t);
	+ int nvlist_add_int32_array(nvlist_t , const char ,
	+ const int32_t *, uint_t);
	+ int nvlist_add_uint32_array(nvlist_t , const char ,
	+ const uint32_t *, uint_t);
	+ int nvlist_add_int64_array(nvlist_t , const char ,
	+ const int64_t *, uint_t);
	+ int nvlist_add_uint64_array(nvlist_t , const char ,
	+ const uint64_t *, uint_t);
	+ int nvlist_add_string_array(nvlist_t , const char ,
	+ const char * const *, uint_t);
	+ int nvlist_add_nvlist_array(nvlist_t , const char ,
	+ const nvlist_t * const *, uint_t);

	nvpair_t nvlist_next_nvpair(nvlist_t , nvpair_t *);
	nvpair_t nvlist_prev_nvpair(nvlist_t , nvpair_t *);
	char nvpair_name(nvpair_t );
	data_type_t nvpair_type(nvpair_t *);
	int nvpair_type_is_array(nvpair_t *);
	int nvpair_value_boolean_value(nvpair_t , boolean_t );
	int nvpair_value_byte(nvpair_t , uchar_t );
	int nvpair_value_int8(nvpair_t , int8_t );
	int nvpair_value_uint8(nvpair_t , uint8_t );
	int nvpair_value_int16(nvpair_t , int16_t );
	int nvpair_value_uint16(nvpair_t , uint16_t );
	int nvpair_value_int32(nvpair_t , int32_t );
	int nvpair_value_uint32(nvpair_t , uint32_t );
	int nvpair_value_int64(nvpair_t , int64_t );
	int nvpair_value_uint64(nvpair_t , uint64_t );
	int nvpair_value_string(nvpair_t , char *);
	int nvpair_value_nvlist(nvpair_t , nvlist_t *);
	int nvpair_value_boolean_array(nvpair_t , boolean_t , uint_t );
	int nvpair_value_byte_array(nvpair_t , uchar_t , uint_t );
	int nvpair_value_int8_array(nvpair_t , int8_t , uint_t );
	int nvpair_value_uint8_array(nvpair_t , uint8_t , uint_t );
	int nvpair_value_int16_array(nvpair_t , int16_t , uint_t );
	int nvpair_value_uint16_array(nvpair_t , uint16_t , uint_t );
	int nvpair_value_int32_array(nvpair_t , int32_t , uint_t );
	int nvpair_value_uint32_array(nvpair_t , uint32_t , uint_t );
	int nvpair_value_int64_array(nvpair_t , int64_t , uint_t );
	int nvpair_value_uint64_array(nvpair_t , uint64_t , uint_t );
	int nvpair_value_string_array(nvpair_t , char *, uint_t );
	int nvpair_value_nvlist_array(nvpair_t , nvlist_t *, uint_t );
	"""

	SOURCE = """
	#include <libzfs/sys/nvpair.h>
	"""

	LIBRARY = "nvpair"

	# vim: softtabstop=4 tabstop=4 expandtab shiftwidth=4
	diff --git a/sys/contrib/openzfs/etc/default/Makefile.am b/sys/contrib/openzfs/etc/default/Makefile.am
	index b88eb549493b..0f7c96698d45 100644
	--- a/sys/contrib/openzfs/etc/default/Makefile.am
	+++ b/sys/contrib/openzfs/etc/default/Makefile.am
	@@ -1,9 +1,8 @@
	include $(top_srcdir)/config/Substfiles.am
	include $(top_srcdir)/config/Shellcheck.am

	initconf_SCRIPTS = zfs

	SUBSTFILES += $(initconf_SCRIPTS)

	SHELLCHECK_SHELL = sh
	-SHELLCHECK_IGNORE = ,SC2034
	diff --git a/sys/contrib/openzfs/etc/default/zfs.in b/sys/contrib/openzfs/etc/default/zfs.in
	index 3b6e5486dd33..fb4a1b65e348 100644
	--- a/sys/contrib/openzfs/etc/default/zfs.in
	+++ b/sys/contrib/openzfs/etc/default/zfs.in
	@@ -1,103 +1,110 @@
	-# ZoL userland configuration.
	+# OpenZFS userland configuration.

	# NOTE: This file is intended for sysv init and initramfs.
	# Changing some of these settings may not make any difference on
	# systemd-based setup, e.g. setting ZFS_MOUNT=no will not prevent systemd
	# from launching zfs-mount.service during boot.
	# See: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=901436

	# To enable a boolean setting, set it to yes, on, true, or 1.
	# Anything else will be interpreted as unset.
	+# shellcheck disable=SC2034
	+
	+# Run `zfs load-key` during system start?
	+ZFS_LOAD_KEY='yes'
	+
	+# Run `zfs unload-key` during system stop?
	+ZFS_UNLOAD_KEY='no'

	# Run `zfs mount -a` during system start?
	ZFS_MOUNT='yes'

	# Run `zfs unmount -a` during system stop?
	ZFS_UNMOUNT='yes'

	# Run `zfs share -a` during system start?
	# nb: The shareiscsi, sharenfs, and sharesmb dataset properties.
	ZFS_SHARE='yes'

	# Run `zfs unshare -a` during system stop?
	ZFS_UNSHARE='yes'

	# By default, a verbatim import of all pools is performed at boot based on the
	# contents of the default zpool cache file. The contents of the cache are
	# managed automatically by the 'zpool import' and 'zpool export' commands.
	#
	# By setting this to 'yes', the system will instead search all devices for
	# pools and attempt to import them all at boot, even those that have been
	# exported. Under this mode, the search path can be controlled by the
	# ZPOOL_IMPORT_PATH variable and a list of pools that should not be imported
	# can be listed in the ZFS_POOL_EXCEPTIONS variable.
	#
	# Note that importing all visible pools may include pools that you don't
	# expect, such as those on removable devices and SANs, and those pools may
	# proceed to mount themselves in places you do not want them to. The results
	# can be unpredictable and possibly dangerous. Only enable this option if you
	# understand this risk and have complete physical control over your system and
	# SAN to prevent the insertion of malicious pools.
	ZPOOL_IMPORT_ALL_VISIBLE='no'

	# Specify specific path(s) to look for device nodes and/or links for the
	# pool import(s). See zpool(8) for more information about this variable.
	# It supersedes the old USE_DISK_BY_ID which indicated that it would only
	# try '/dev/disk/by-id'.
	# The old variable will still work in the code, but is deprecated.
	#ZPOOL_IMPORT_PATH="/dev/disk/by-vdev:/dev/disk/by-id"

	# List of pools that should NOT be imported at boot
	# when ZPOOL_IMPORT_ALL_VISIBLE is 'yes'.
	# This is a space separated list.
	#ZFS_POOL_EXCEPTIONS="test2"

	# Should the datasets be mounted verbosely?
	# A mount counter will be used when mounting if set to 'yes'.
	VERBOSE_MOUNT='no'

	# Should we allow overlay mounts?
	# This is standard in Linux, but not ZFS which comes from Solaris where this
	# is not allowed).
	DO_OVERLAY_MOUNTS='no'

	# Any additional option to the 'zfs import' commandline?
	# Include '-o' for each option wanted.
	# You don't need to put '-f' in here, unless you want it ALL the time.
	# Using the option 'zfsforce=1' on the grub/kernel command line will
	# do the same, but on a case-to-case basis.
	ZPOOL_IMPORT_OPTS=""

	# Full path to the ZFS cache file?
	# See "cachefile" in zpool(8).
	# The default is "@sysconfdir@/zfs/zpool.cache".
	#ZPOOL_CACHE="@sysconfdir@/zfs/zpool.cache"
	#
	# Setting ZPOOL_CACHE to an empty string ('') AND setting ZPOOL_IMPORT_OPTS to
	# "-c @sysconfdir@/zfs/zpool.cache" will _enforce_ the use of a cache file.
	# This is needed in some cases (extreme amounts of VDEVs, multipath etc).
	# Generally, the use of a cache file is usually not recommended on Linux
	# because it sometimes is more trouble than it's worth (laptops with external
	# devices or when/if device nodes changes names).
	#ZPOOL_IMPORT_OPTS="-c @sysconfdir@/zfs/zpool.cache"
	#ZPOOL_CACHE=""

	# Any additional option to the 'zfs mount' command line?
	# Include '-o' for each option wanted.
	MOUNT_EXTRA_OPTIONS=""

	# Build kernel modules with the --enable-debug switch?
	# Only applicable for Debian GNU/Linux {dkms,initramfs}.
	ZFS_DKMS_ENABLE_DEBUG='no'

	# Build kernel modules with the --enable-debuginfo switch?
	# Only applicable for Debian GNU/Linux {dkms,initramfs}.
	ZFS_DKMS_ENABLE_DEBUGINFO='no'

	# Keep debugging symbols in kernel modules?
	# Only applicable for Debian GNU/Linux {dkms,initramfs}.
	ZFS_DKMS_DISABLE_STRIP='no'

	# Optional arguments for the ZFS Event Daemon (ZED).
	# See zed(8) for more information on available options.
	#ZED_ARGS="-M"
	diff --git a/sys/contrib/openzfs/etc/init.d/.gitignore b/sys/contrib/openzfs/etc/init.d/.gitignore
	index 43a673d55343..b3402f831806 100644
	--- a/sys/contrib/openzfs/etc/init.d/.gitignore
	+++ b/sys/contrib/openzfs/etc/init.d/.gitignore
	@@ -1,5 +1,6 @@
	zfs-import
	+zfs-load-key
	zfs-mount
	zfs-share
	zfs-zed
	zfs
	diff --git a/sys/contrib/openzfs/etc/init.d/Makefile.am b/sys/contrib/openzfs/etc/init.d/Makefile.am
	index f93af1fd77e0..658623fda4aa 100644
	--- a/sys/contrib/openzfs/etc/init.d/Makefile.am
	+++ b/sys/contrib/openzfs/etc/init.d/Makefile.am
	@@ -1,10 +1,10 @@
	include $(top_srcdir)/config/Substfiles.am
	include $(top_srcdir)/config/Shellcheck.am

	EXTRA_DIST += README.md

	-init_SCRIPTS = zfs-import zfs-mount zfs-share zfs-zed
	+init_SCRIPTS = zfs-import zfs-load-key zfs-mount zfs-share zfs-zed

	SUBSTFILES += $(init_SCRIPTS)

	SHELLCHECK_SHELL = dash # local variables
	diff --git a/sys/contrib/openzfs/etc/init.d/README.md b/sys/contrib/openzfs/etc/init.d/README.md
	index c14b01937db2..f417b24c5923 100644
	--- a/sys/contrib/openzfs/etc/init.d/README.md
	+++ b/sys/contrib/openzfs/etc/init.d/README.md
	@@ -1,72 +1,75 @@
	DESCRIPTION
	These script were written with the primary intention of being portable and
	usable on as many systems as possible.

	This is, in practice, usually not possible. But the intention is there.
	And it is a good one.

	They have been tested successfully on:

	* Debian GNU/Linux Wheezy
	* Debian GNU/Linux Jessie
	* Ubuntu Trusty
	* CentOS 6.0
	* CentOS 6.6
	* Gentoo

	SUPPORT
	If you find that they don't work for your platform, please report this
	at the OpenZFS issue tracker at https://github.com/openzfs/zfs/issues.

	Please include:

	* Distribution name
	* Distribution version
	* Where to find an install CD image
	* Architecture

	If you have code to share that fixes the problem, that is much better.
	But please remember to try your best keep portability in mind. If you
	suspect that what you're writing/modifying won't work on anything else
	than your distribution, please make sure to put that code in appropriate
	if/else/fi code.

	It currently MUST be bash (or fully compatible) for this to work.

	If you're making your own distribution and you want the scripts to
	work on that, the biggest problem you'll (probably) have is the part
	at the beginning of the "zfs-functions" file which sets up the
	logging output.

	INSTALLING INIT SCRIPT LINKS
	To setup the init script links in /etc/rc?.d manually on a Debian GNU/Linux
	(or derived) system, run the following commands (the order is important!):

	- update-rc.d zfs-import start 07 S . stop 07 0 1 6 .
	- update-rc.d zfs-mount start 02 2 3 4 5 . stop 06 0 1 6 .
	- update-rc.d zfs-zed start 07 2 3 4 5 . stop 08 0 1 6 .
	- update-rc.d zfs-share start 27 2 3 4 5 . stop 05 0 1 6 .
	+ update-rc.d zfs-import start 07 S . stop 07 0 1 6 .
	+ update-rc.d zfs-load-key start 02 2 3 4 5 . stop 06 0 1 6 .
	+ update-rc.d zfs-mount start 02 2 3 4 5 . stop 06 0 1 6 .
	+ update-rc.d zfs-zed start 07 2 3 4 5 . stop 08 0 1 6 .
	+ update-rc.d zfs-share start 27 2 3 4 5 . stop 05 0 1 6 .

	To do the same on RedHat, Fedora and/or CentOS:

	chkconfig zfs-import
	+ chkconfig zfs-load-key
	chkconfig zfs-mount
	chkconfig zfs-zed
	chkconfig zfs-share

	On Gentoo:

	rc-update add zfs-import boot
	+ rc-update add zfs-load-key boot
	rc-update add zfs-mount boot
	rc-update add zfs-zed default
	rc-update add zfs-share default

	The idea here is to make sure all of the ZFS filesystems, including possibly
	separate datasets like /var, are mounted before anything else is started.

	Then, ZED, which depends on /var, can be started. It will consume and act
	on events that occurred before it started. ZED may also play a role in
	sharing filesystems in the future, so it is important to start before the
	'share' service.

	Finally, we share filesystems configured with the share\* property.
	diff --git a/sys/contrib/openzfs/etc/init.d/zfs-load-key.in b/sys/contrib/openzfs/etc/init.d/zfs-load-key.in
	new file mode 100755
	index 000000000000..2f8deffdc809
	--- /dev/null
	+++ b/sys/contrib/openzfs/etc/init.d/zfs-load-key.in
	@@ -0,0 +1,131 @@
	+#!@DEFAULT_INIT_SHELL@
	+#
	+# zfs-load-key This script will load/unload the zfs filesystems keys.
	+#
	+# chkconfig: 2345 06 99
	+# description: This script will load or unload the zfs filesystems keys during
	+# system boot/shutdown. Only filesystems with key path set
	+# in keylocation property. See the zfs(8) man page for details.
	+# probe: true
	+#
	+### BEGIN INIT INFO
	+# Provides: zfs-load-key
	+# Required-Start: $local_fs zfs-import
	+# Required-Stop: $local_fs zfs-import
	+# Default-Start: 2 3 4 5
	+# Default-Stop: 0 1 6
	+# X-Start-Before: zfs-mount
	+# X-Stop-After: zfs-zed
	+# Short-Description: Load ZFS keys for filesystems and volumes
	+# Description: Run the `zfs load-key` or `zfs unload-key` commands.
	+### END INIT INFO
	+#
	+# Released under the 2-clause BSD license.
	+#
	+# This script is based on debian/zfsutils.zfs.init from the
	+# Debian GNU/kFreeBSD zfsutils 8.1-3 package, written by Aurelien Jarno.
	+
	+# Source the common init script
	+. @sysconfdir@/zfs/zfs-functions
	+
	+# ----------------------------------------------------
	+
	+do_depend()
	+{
	+ # bootmisc will log to /var which may be a different zfs than root.
	+ before bootmisc logger zfs-mount
	+
	+ after zfs-import sysfs
	+ keyword -lxc -openvz -prefix -vserver
	+}
	+
	+# Load keys for all datasets/filesystems
	+do_load_keys()
	+{
	+ zfs_log_begin_msg "Load ZFS filesystem(s) keys"
	+
	+ "$ZFS" list -Ho name,encryptionroot,keystatus,keylocation \|
	+ while IFS=" " read -r name encryptionroot keystatus keylocation; do
	+ if [ "$encryptionroot" != "-" ] &&
	+ [ "$name" = "$encryptionroot" ] &&
	+ [ "$keystatus" = "unavailable" ] &&
	+ [ "$keylocation" != "prompt" ] &&
	+ [ "$keylocation" != "none" ]
	+ then
	+ zfs_action "Load key for $encryptionroot" \
	+ "$ZFS" load-key "$encryptionroot"
	+ fi
	+ done
	+
	+ zfs_log_end_msg 0
	+
	+ return 0
	+}
	+
	+# Unload keys for all datasets/filesystems
	+do_unload_keys()
	+{
	+ zfs_log_begin_msg "Unload ZFS filesystem(s) key"
	+
	+ "$ZFS" list -Ho name,encryptionroot,keystatus \| sed '1!G;h;$!d' \|
	+ while IFS=" " read -r name encryptionroot keystatus; do
	+ if [ "$encryptionroot" != "-" ] &&
	+ [ "$name" = "$encryptionroot" ] &&
	+ [ "$keystatus" = "available" ]
	+ then
	+ zfs_action "Unload key for $encryptionroot" \
	+ "$ZFS" unload-key "$encryptionroot"
	+ fi
	+ done
	+
	+ zfs_log_end_msg 0
	+
	+ return 0
	+}
	+
	+do_start()
	+{
	+ check_boolean "$ZFS_LOAD_KEY" \|\| exit 0
	+
	+ check_module_loaded "zfs" \|\| exit 0
	+
	+ do_load_keys
	+}
	+
	+do_stop()
	+{
	+ check_boolean "$ZFS_UNLOAD_KEY" \|\| exit 0
	+
	+ check_module_loaded "zfs" \|\| exit 0
	+
	+ do_unload_keys
	+}
	+
	+# ----------------------------------------------------
	+
	+if [ ! -e /sbin/openrc-run ]
	+then
	+ case "$1" in
	+ start)
	+ do_start
	+ ;;
	+ stop)
	+ do_stop
	+ ;;
	+ force-reload\|condrestart\|reload\|restart\|status)
	+ # no-op
	+ ;;
	+ *)
	+ [ -n "$1" ] && echo "Error: Unknown command $1."
	+ echo "Usage: $0 {start\|stop}"
	+ exit 3
	+ ;;
	+ esac
	+
	+ exit $?
	+else
	+ # Create wrapper functions since Gentoo don't use the case part.
	+ depend() { do_depend; }
	+ start() { do_start; }
	+ stop() { do_stop; }
	+fi
	diff --git a/sys/contrib/openzfs/etc/systemd/system-generators/zfs-mount-generator.c b/sys/contrib/openzfs/etc/systemd/system-generators/zfs-mount-generator.c
	index b806339deb2f..f4c6c26a0b34 100644
	--- a/sys/contrib/openzfs/etc/systemd/system-generators/zfs-mount-generator.c
	+++ b/sys/contrib/openzfs/etc/systemd/system-generators/zfs-mount-generator.c
	@@ -1,1083 +1,1000 @@
	/*
	* Copyright (c) 2017 Antonio Russo <antonio.e.russo@gmail.com>
	* Copyright (c) 2020 InsanePrawn <insane.prawny@gmail.com>
	*
	* Permission is hereby granted, free of charge, to any person obtaining
	* a copy of this software and associated documentation files (the
	* "Software"), to deal in the Software without restriction, including
	* without limitation the rights to use, copy, modify, merge, publish,
	* distribute, sublicense, and/or sell copies of the Software, and to
	* permit persons to whom the Software is furnished to do so, subject to
	* the following conditions:
	*
	* The above copyright notice and this permission notice shall be
	* included in all copies or substantial portions of the Software.
	*
	* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
	* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
	* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
	* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
	* LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
	* OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
	* WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
	*/


	#include <sys/resource.h>
	#include <sys/types.h>
	#include <sys/time.h>
	#include <sys/stat.h>
	-#include <sys/wait.h>
	-#include <sys/mman.h>
	-#include <semaphore.h>
	#include <stdbool.h>
	#include <unistd.h>
	#include <fcntl.h>
	#include <stdio.h>
	#include <time.h>
	#include <regex.h>
	#include <search.h>
	#include <dirent.h>
	#include <string.h>
	#include <stdlib.h>
	#include <limits.h>
	#include <errno.h>
	#include <libzfs.h>

	-#define STRCMP ((int()(const void , const void *))&strcmp)
	-#define PID_T_CMP ((int()(const void , const void *))&pid_t_cmp)
	-
	-static int
	-pid_t_cmp(const pid_t lhs, const pid_t rhs)
	-{
	- /*
	- * This is always valid, quoth sys_types.h(7posix):
	- * > blksize_t, pid_t, and ssize_t shall be signed integer types.
	- */
	- return (lhs - rhs);
	-}
	+/*
	+ * For debugging only.
	+ *
	+ * Free statics with trivial life-times,
	+ * but saved line filenames are replaced with a static string.
	+ */
	+#define FREE_STATICS false

	-#define EXIT_ENOMEM() \
	- do { \
	- fprintf(stderr, PROGNAME "[%d]: " \
	- "not enough memory (L%d)!\n", getpid(), __LINE__); \
	- _exit(1); \
	- } while (0)
	+#define nitems(arr) (sizeof (arr) / sizeof (*arr))
	+#define STRCMP ((int()(const void , const void *))&strcmp)


	#define PROGNAME "zfs-mount-generator"
	#define FSLIST SYSCONFDIR "/zfs/zfs-list.cache"
	#define ZFS SBINDIR "/zfs"

	#define OUTPUT_HEADER \
	"# Automatically generated by " PROGNAME "\n" \
	"\n"

	/*
	* Starts like the one in libzfs_util.c but also matches "//"
	* and captures until the end, since we actually use it for path extraxion
	*/
	#define URI_REGEX_S "^\$[A-Za-z][A-Za-z0-9+.\\-]\$:\\/\\/\$.\$$"
	static regex_t uri_regex;

	-static char *argv0;
	-
	static const char *destdir = "/tmp";
	static int destdir_fd = -1;

	static void known_pools = NULL; / tsearch() of C strings */
	-static struct {
	- sem_t noauto_not_on_sem;
	-
	- sem_t noauto_names_sem;
	- size_t noauto_names_len;
	- size_t noauto_names_max;
	- char noauto_names[][NAME_MAX];
	-} *noauto_files;
	+static void noauto_files = NULL; / tsearch() of C strings */


	static char *
	systemd_escape(const char input, const char prepend, const char *append)
	{
	size_t len = strlen(input);
	size_t applen = strlen(append);
	size_t prelen = strlen(prepend);
	char ret = malloc(4 len + prelen + applen + 1);
	- if (!ret)
	- EXIT_ENOMEM();
	+ if (!ret) {
	+ fprintf(stderr, PROGNAME "[%d]: "
	+ "out of memory to escape \"%s%s%s\"!\n",
	+ getpid(), prepend, input, append);
	+ return (NULL);
	+ }

	memcpy(ret, prepend, prelen);
	char *out = ret + prelen;

	const char *cur = input;
	if (*cur == '.') {
	memcpy(out, "\\x2e", 4);
	out += 4;
	++cur;
	}
	for (; *cur; ++cur) {
	if (*cur == '/')
	*(out++) = '-';
	else if (strchr(
	"0123456789"
	"abcdefghijklmnopqrstuvwxyz"
	"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
	":_.", *cur))
	(out++) = cur;
	else {
	sprintf(out, "\\x%02x", (int)*cur);
	out += 4;
	}
	}

	memcpy(out, append, applen + 1);
	return (ret);
	}

	static void
	simplify_path(char *path)
	{
	char *out = path;
	for (char cur = path; cur; ++cur) {
	if (*cur == '/') {
	while (*(cur + 1) == '/')
	++cur;
	*(out++) = '/';
	} else
	(out++) = cur;
	}

	*(out++) = '\0';
	}

	static bool
	strendswith(const char what, const char suff)
	{
	size_t what_l = strlen(what);
	size_t suff_l = strlen(suff);

	return ((what_l >= suff_l) &&
	(strcmp(what + what_l - suff_l, suff) == 0));
	}

	/* Assumes already-simplified path, doesn't modify input */
	static char *
	systemd_escape_path(char input, const char prepend, const char *append)
	{
	if (strcmp(input, "/") == 0) {
	char *ret;
	- if (asprintf(&ret, "%s-%s", prepend, append) == -1)
	- EXIT_ENOMEM();
	+ if (asprintf(&ret, "%s-%s", prepend, append) == -1) {
	+ fprintf(stderr, PROGNAME "[%d]: "
	+ "out of memory to escape \"%s%s%s\"!\n",
	+ getpid(), prepend, input, append);
	+ ret = NULL;
	+ }
	return (ret);
	} else {
	/*
	* path_is_normalized() (flattened for absolute paths here),
	* required for proper escaping
	*/
	if (strstr(input, "/./") \|\| strstr(input, "/../") \|\|
	strendswith(input, "/.") \|\| strendswith(input, "/.."))
	return (NULL);


	if (input[0] == '/')
	++input;

	char *back = &input[strlen(input) - 1];
	bool deslash = *back == '/';
	if (deslash)
	*back = '\0';

	char *ret = systemd_escape(input, prepend, append);

	if (deslash)
	*back = '/';
	return (ret);
	}
	}

	static FILE *
	fopenat(int dirfd, const char *pathname, int flags,
	const char *stream_mode, mode_t mode)
	{
	int fd = openat(dirfd, pathname, flags, mode);
	if (fd < 0)
	return (NULL);

	return (fdopen(fd, stream_mode));
	}

	static int
	line_worker(char line, const char cachefile)
	{
	+ int ret = 0;
	+ void *tofree_all[8];
	+ void **tofree = tofree_all;
	+
	char *toktmp;
	/* BEGIN CSTYLED */
	const char *dataset = strtok_r(line, "\t", &toktmp);
	char *p_mountpoint = strtok_r(NULL, "\t", &toktmp);
	const char *p_canmount = strtok_r(NULL, "\t", &toktmp);
	const char *p_atime = strtok_r(NULL, "\t", &toktmp);
	const char *p_relatime = strtok_r(NULL, "\t", &toktmp);
	const char *p_devices = strtok_r(NULL, "\t", &toktmp);
	const char *p_exec = strtok_r(NULL, "\t", &toktmp);
	const char *p_readonly = strtok_r(NULL, "\t", &toktmp);
	const char *p_setuid = strtok_r(NULL, "\t", &toktmp);
	const char *p_nbmand = strtok_r(NULL, "\t", &toktmp);
	const char *p_encroot = strtok_r(NULL, "\t", &toktmp) ?: "-";
	char *p_keyloc = strtok_r(NULL, "\t", &toktmp) ?: strdupa("none");
	const char *p_systemd_requires = strtok_r(NULL, "\t", &toktmp) ?: "-";
	const char *p_systemd_requiresmountsfor = strtok_r(NULL, "\t", &toktmp) ?: "-";
	const char *p_systemd_before = strtok_r(NULL, "\t", &toktmp) ?: "-";
	const char *p_systemd_after = strtok_r(NULL, "\t", &toktmp) ?: "-";
	char *p_systemd_wantedby = strtok_r(NULL, "\t", &toktmp) ?: strdupa("-");
	char *p_systemd_requiredby = strtok_r(NULL, "\t", &toktmp) ?: strdupa("-");
	const char *p_systemd_nofail = strtok_r(NULL, "\t", &toktmp) ?: "-";
	const char *p_systemd_ignore = strtok_r(NULL, "\t", &toktmp) ?: "-";
	/* END CSTYLED */

	const char *pool = dataset;
	if ((toktmp = strchr(pool, '/')) != NULL)
	pool = strndupa(pool, toktmp - pool);

	if (p_nbmand == NULL) {
	fprintf(stderr, PROGNAME "[%d]: %s: not enough tokens!\n",
	getpid(), dataset);
	- return (1);
	+ goto err;
	}

	- strncpy(argv0, dataset, strlen(argv0));
	-
	/* Minimal pre-requisites to mount a ZFS dataset */
	const char *after = "zfs-import.target";
	const char *wants = "zfs-import.target";
	const char *bindsto = NULL;
	char *wantedby = NULL;
	char *requiredby = NULL;
	bool noauto = false;
	bool wantedby_append = true;

	/*
	* zfs-import.target is not needed if the pool is already imported.
	* This avoids a dependency loop on root-on-ZFS systems:
	* systemd-random-seed.service After (via RequiresMountsFor)
	* var-lib.mount After
	* zfs-import.target After
	* zfs-import-{cache,scan}.service After
	* cryptsetup.service After
	* systemd-random-seed.service
	*/
	if (tfind(pool, &known_pools, STRCMP)) {
	after = "";
	wants = "";
	}

	if (strcmp(p_systemd_after, "-") == 0)
	p_systemd_after = NULL;
	if (strcmp(p_systemd_before, "-") == 0)
	p_systemd_before = NULL;
	if (strcmp(p_systemd_requires, "-") == 0)
	p_systemd_requires = NULL;
	if (strcmp(p_systemd_requiresmountsfor, "-") == 0)
	p_systemd_requiresmountsfor = NULL;


	if (strcmp(p_encroot, "-") != 0) {
	- char *keyloadunit =
	+ char keyloadunit = (tofree++) =
	systemd_escape(p_encroot, "zfs-load-key@", ".service");
	+ if (keyloadunit == NULL)
	+ goto err;

	if (strcmp(dataset, p_encroot) == 0) {
	const char *keymountdep = NULL;
	bool is_prompt = false;
	+ bool need_network = false;

	regmatch_t uri_matches[3];
	if (regexec(&uri_regex, p_keyloc,
	- sizeof (uri_matches) / sizeof (*uri_matches),
	- uri_matches, 0) == 0) {
	+ nitems(uri_matches), uri_matches, 0) == 0) {
	+ p_keyloc[uri_matches[1].rm_eo] = '\0';
	p_keyloc[uri_matches[2].rm_eo] = '\0';
	+ const char *scheme =
	+ &p_keyloc[uri_matches[1].rm_so];
	const char *path =
	&p_keyloc[uri_matches[2].rm_so];

	- /*
	- * Assumes all URI keylocations need
	- * the mount for their path;
	- * http://, for example, wouldn't
	- * (but it'd need network-online.target et al.)
	- */
	- keymountdep = path;
	+ if (strcmp(scheme, "https") == 0 \|\|
	+ strcmp(scheme, "http") == 0)
	+ need_network = true;
	+ else
	+ keymountdep = path;
	} else {
	if (strcmp(p_keyloc, "prompt") != 0)
	fprintf(stderr, PROGNAME "[%d]: %s: "
	"unknown non-URI keylocation=%s\n",
	getpid(), dataset, p_keyloc);

	is_prompt = true;
	}


	/* Generate the key-load .service unit */
	FILE *keyloadunit_f = fopenat(destdir_fd, keyloadunit,
	O_WRONLY \| O_CREAT \| O_TRUNC \| O_CLOEXEC, "w",
	0644);
	if (!keyloadunit_f) {
	fprintf(stderr, PROGNAME "[%d]: %s: "
	"couldn't open %s under %s: %s\n",
	getpid(), dataset, keyloadunit, destdir,
	strerror(errno));
	- return (1);
	+ goto err;
	}

	fprintf(keyloadunit_f,
	OUTPUT_HEADER
	"[Unit]\n"
	"Description=Load ZFS key for %s\n"
	"SourcePath=" FSLIST "/%s\n"
	"Documentation=man:zfs-mount-generator(8)\n"
	"DefaultDependencies=no\n"
	"Wants=%s\n"
	"After=%s\n",
	dataset, cachefile, wants, after);

	+ if (need_network)
	+ fprintf(keyloadunit_f,
	+ "Wants=network-online.target\n"
	+ "After=network-online.target\n");
	+
	if (p_systemd_requires)
	fprintf(keyloadunit_f,
	"Requires=%s\n", p_systemd_requires);

	- if (p_systemd_requiresmountsfor \|\| keymountdep) {
	- fprintf(keyloadunit_f, "RequiresMountsFor=");
	- if (p_systemd_requiresmountsfor)
	- fprintf(keyloadunit_f,
	- "%s ", p_systemd_requiresmountsfor);
	- if (keymountdep)
	- fprintf(keyloadunit_f,
	- "'%s'", keymountdep);
	- fprintf(keyloadunit_f, "\n");
	- }
	+ if (p_systemd_requiresmountsfor)
	+ fprintf(keyloadunit_f,
	+ "RequiresMountsFor=%s\n",
	+ p_systemd_requiresmountsfor);
	+ if (keymountdep)
	+ fprintf(keyloadunit_f,
	+ "RequiresMountsFor='%s'\n", keymountdep);

	/* BEGIN CSTYLED */
	fprintf(keyloadunit_f,
	"\n"
	"[Service]\n"
	"Type=oneshot\n"
	"RemainAfterExit=yes\n"
	"# This avoids a dependency loop involving systemd-journald.socket if this\n"
	"# dataset is a parent of the root filesystem.\n"
	"StandardOutput=null\n"
	"StandardError=null\n"
	"ExecStart=/bin/sh -euc '"
	"[ \"$$(" ZFS " get -H -o value keystatus \"%s\")\" = \"unavailable\" ] \|\| exit 0;",
	dataset);
	if (is_prompt)
	fprintf(keyloadunit_f,
	"for i in 1 2 3; do "
	"systemd-ask-password --id=\"zfs:%s\" \"Enter passphrase for %s:\" \|"
	"" ZFS " load-key \"%s\" && exit 0;"
	"done;"
	"exit 1",
	dataset, dataset, dataset);
	else
	fprintf(keyloadunit_f,
	"exec " ZFS " load-key \"%s\"",
	dataset);

	fprintf(keyloadunit_f,
	"'\n"
	"ExecStop=/bin/sh -euc '"
	"[ \"$$(" ZFS " get -H -o value keystatus \"%s\")\" = \"available\" ] \|\| exit 0;"
	"exec " ZFS " unload-key \"%s\""
	"'\n",
	dataset, dataset);
	/* END CSTYLED */

	(void) fclose(keyloadunit_f);
	}

	/* Update dependencies for the mount file to want this */
	bindsto = keyloadunit;
	if (after[0] == '\0')
	after = keyloadunit;
	else if (asprintf(&toktmp, "%s %s", after, keyloadunit) != -1)
	- after = toktmp;
	- else
	- EXIT_ENOMEM();
	+ after = *(tofree++) = toktmp;
	+ else {
	+ fprintf(stderr, PROGNAME "[%d]: %s: "
	+ "out of memory to generate after=\"%s %s\"!\n",
	+ getpid(), dataset, after, keyloadunit);
	+ goto err;
	+ }
	}


	/* Skip generation of the mount unit if org.openzfs.systemd:ignore=on */
	if (strcmp(p_systemd_ignore, "-") == 0 \|\|
	strcmp(p_systemd_ignore, "off") == 0) {
	/* ok */
	} else if (strcmp(p_systemd_ignore, "on") == 0)
	- return (0);
	+ goto end;
	else {
	fprintf(stderr, PROGNAME "[%d]: %s: "
	"invalid org.openzfs.systemd:ignore=%s\n",
	getpid(), dataset, p_systemd_ignore);
	- return (1);
	+ goto err;
	}

	/* Check for canmount */
	if (strcmp(p_canmount, "on") == 0) {
	/* ok */
	} else if (strcmp(p_canmount, "noauto") == 0)
	noauto = true;
	else if (strcmp(p_canmount, "off") == 0)
	- return (0);
	+ goto end;
	else {
	fprintf(stderr, PROGNAME "[%d]: %s: invalid canmount=%s\n",
	getpid(), dataset, p_canmount);
	- return (1);
	+ goto err;
	}

	/* Check for legacy and blank mountpoints */
	if (strcmp(p_mountpoint, "legacy") == 0 \|\|
	strcmp(p_mountpoint, "none") == 0)
	- return (0);
	+ goto end;
	else if (p_mountpoint[0] != '/') {
	fprintf(stderr, PROGNAME "[%d]: %s: invalid mountpoint=%s\n",
	getpid(), dataset, p_mountpoint);
	- return (1);
	+ goto err;
	}

	/* Escape the mountpoint per systemd policy */
	simplify_path(p_mountpoint);
	const char *mountfile = systemd_escape_path(p_mountpoint, "", ".mount");
	if (mountfile == NULL) {
	fprintf(stderr,
	PROGNAME "[%d]: %s: abnormal simplified mountpoint: %s\n",
	getpid(), dataset, p_mountpoint);
	- return (1);
	+ goto err;
	}


	/*
	* Parse options, cf. lib/libzfs/libzfs_mount.c:zfs_add_options
	*
	* The longest string achievable here is
	* ",atime,strictatime,nodev,noexec,rw,nosuid,nomand".
	*/
	char opts[64] = "";

	/* atime */
	if (strcmp(p_atime, "on") == 0) {
	/* relatime */
	if (strcmp(p_relatime, "on") == 0)
	strcat(opts, ",atime,relatime");
	else if (strcmp(p_relatime, "off") == 0)
	strcat(opts, ",atime,strictatime");
	else
	fprintf(stderr,
	PROGNAME "[%d]: %s: invalid relatime=%s\n",
	getpid(), dataset, p_relatime);
	} else if (strcmp(p_atime, "off") == 0) {
	strcat(opts, ",noatime");
	} else
	fprintf(stderr, PROGNAME "[%d]: %s: invalid atime=%s\n",
	getpid(), dataset, p_atime);

	/* devices */
	if (strcmp(p_devices, "on") == 0)
	strcat(opts, ",dev");
	else if (strcmp(p_devices, "off") == 0)
	strcat(opts, ",nodev");
	else
	fprintf(stderr, PROGNAME "[%d]: %s: invalid devices=%s\n",
	getpid(), dataset, p_devices);

	/* exec */
	if (strcmp(p_exec, "on") == 0)
	strcat(opts, ",exec");
	else if (strcmp(p_exec, "off") == 0)
	strcat(opts, ",noexec");
	else
	fprintf(stderr, PROGNAME "[%d]: %s: invalid exec=%s\n",
	getpid(), dataset, p_exec);

	/* readonly */
	if (strcmp(p_readonly, "on") == 0)
	strcat(opts, ",ro");
	else if (strcmp(p_readonly, "off") == 0)
	strcat(opts, ",rw");
	else
	fprintf(stderr, PROGNAME "[%d]: %s: invalid readonly=%s\n",
	getpid(), dataset, p_readonly);

	/* setuid */
	if (strcmp(p_setuid, "on") == 0)
	strcat(opts, ",suid");
	else if (strcmp(p_setuid, "off") == 0)
	strcat(opts, ",nosuid");
	else
	fprintf(stderr, PROGNAME "[%d]: %s: invalid setuid=%s\n",
	getpid(), dataset, p_setuid);

	/* nbmand */
	if (strcmp(p_nbmand, "on") == 0)
	strcat(opts, ",mand");
	else if (strcmp(p_nbmand, "off") == 0)
	strcat(opts, ",nomand");
	else
	fprintf(stderr, PROGNAME "[%d]: %s: invalid nbmand=%s\n",
	getpid(), dataset, p_setuid);

	if (strcmp(p_systemd_wantedby, "-") != 0) {
	noauto = true;

	if (strcmp(p_systemd_wantedby, "none") != 0)
	wantedby = p_systemd_wantedby;
	}

	if (strcmp(p_systemd_requiredby, "-") != 0) {
	noauto = true;

	if (strcmp(p_systemd_requiredby, "none") != 0)
	requiredby = p_systemd_requiredby;
	}

	/*
	* For datasets with canmount=on, a dependency is created for
	* local-fs.target by default. To avoid regressions, this dependency
	* is reduced to "wants" rather than "requires" when nofail!=off.
	* THIS MAY CHANGE
	* noauto=on disables this behavior completely.
	*/
	if (!noauto) {
	if (strcmp(p_systemd_nofail, "off") == 0)
	requiredby = strdupa("local-fs.target");
	else {
	wantedby = strdupa("local-fs.target");
	wantedby_append = strcmp(p_systemd_nofail, "on") != 0;
	}
	}

	/*
	* Handle existing files:
	* 1. We never overwrite existing files, although we may delete
	* files if we're sure they were created by us. (see 5.)
	* 2. We handle files differently based on canmount.
	* Units with canmount=on always have precedence over noauto.
	- * This is enforced by the noauto_not_on_sem semaphore,
	- * which is only unlocked when the last canmount=on process exits.
	+ * This is enforced by processing these units before all others.
	* It is important to use p_canmount and not noauto here,
	* since we categorise by canmount while other properties,
	* e.g. org.openzfs.systemd:wanted-by, also modify noauto.
	* 3. If no unit file exists for a noauto dataset, we create one.
	* Additionally, we use noauto_files to track the unit file names
	* (which are the systemd-escaped mountpoints) of all (exclusively)
	* noauto datasets that had a file created.
	- * 4. If the file to be created is found in the tracking array,
	+ * 4. If the file to be created is found in the tracking tree,
	* we do NOT create it.
	* 5. If a file exists for a noauto dataset,
	* we check whether the file name is in the array.
	* If it is, we have multiple noauto datasets for the same
	* mountpoint. In such cases, we remove the file for safety.
	* We leave the file name in the tracking array to avoid
	* further noauto datasets creating a file for this path again.
	*/

	- {
	- sem_t *our_sem = (strcmp(p_canmount, "on") == 0) ?
	- &noauto_files->noauto_names_sem :
	- &noauto_files->noauto_not_on_sem;
	- while (sem_wait(our_sem) == -1 && errno == EINTR)
	- ;
	- }
	-
	struct stat stbuf;
	bool already_exists = fstatat(destdir_fd, mountfile, &stbuf, 0) == 0;
	+ bool is_known = tfind(mountfile, &noauto_files, STRCMP) != NULL;

	- bool is_known = false;
	- for (size_t i = 0; i < noauto_files->noauto_names_len; ++i) {
	- if (strncmp(
	- noauto_files->noauto_names[i], mountfile, NAME_MAX) == 0) {
	- is_known = true;
	- break;
	- }
	- }
	-
	+ (tofree++) = (void )mountfile;
	if (already_exists) {
	if (is_known) {
	- /* If it's in $noauto_files, we must be noauto too */
	+ /* If it's in noauto_files, we must be noauto too */

	/* See 5 */
	errno = 0;
	(void) unlinkat(destdir_fd, mountfile, 0);

	/* See 2 */
	fprintf(stderr, PROGNAME "[%d]: %s: "
	"removing duplicate noauto unit %s%s%s\n",
	getpid(), dataset, mountfile,
	errno ? "" : " failed: ",
	errno ? "" : strerror(errno));
	} else {
	/* Don't log for canmount=noauto */
	if (strcmp(p_canmount, "on") == 0)
	fprintf(stderr, PROGNAME "[%d]: %s: "
	"%s already exists. Skipping.\n",
	getpid(), dataset, mountfile);
	}

	/* File exists: skip current dataset */
	- if (strcmp(p_canmount, "on") == 0)
	- sem_post(&noauto_files->noauto_names_sem);
	- return (0);
	+ goto end;
	} else {
	if (is_known) {
	/* See 4 */
	- if (strcmp(p_canmount, "on") == 0)
	- sem_post(&noauto_files->noauto_names_sem);
	- return (0);
	+ goto end;
	} else if (strcmp(p_canmount, "noauto") == 0) {
	- if (noauto_files->noauto_names_len ==
	- noauto_files->noauto_names_max)
	+ if (tsearch(mountfile, &noauto_files, STRCMP) == NULL)
	fprintf(stderr, PROGNAME "[%d]: %s: "
	- "noauto dataset limit (%zu) reached! "
	- "Not tracking %s. Please report this to "
	- "https://github.com/openzfs/zfs\n",
	- getpid(), dataset,
	- noauto_files->noauto_names_max, mountfile);
	- else {
	- strncpy(noauto_files->noauto_names[
	- noauto_files->noauto_names_len],
	- mountfile, NAME_MAX);
	- ++noauto_files->noauto_names_len;
	- }
	+ "out of memory for noauto datasets! "
	+ "Not tracking %s.\n",
	+ getpid(), dataset, mountfile);
	+ else
	+ /* mountfile escaped to noauto_files */
	+ *(--tofree) = NULL;
	}
	}


	FILE *mountfile_f = fopenat(destdir_fd, mountfile,
	O_WRONLY \| O_CREAT \| O_TRUNC \| O_CLOEXEC, "w", 0644);
	- if (strcmp(p_canmount, "on") == 0)
	- sem_post(&noauto_files->noauto_names_sem);
	if (!mountfile_f) {
	fprintf(stderr,
	PROGNAME "[%d]: %s: couldn't open %s under %s: %s\n",
	getpid(), dataset, mountfile, destdir, strerror(errno));
	- return (1);
	+ goto err;
	}

	fprintf(mountfile_f,
	OUTPUT_HEADER
	"[Unit]\n"
	"SourcePath=" FSLIST "/%s\n"
	"Documentation=man:zfs-mount-generator(8)\n"
	"\n"
	"Before=",
	cachefile);

	if (p_systemd_before)
	fprintf(mountfile_f, "%s ", p_systemd_before);
	fprintf(mountfile_f, "zfs-mount.service"); /* Ensures we don't race */
	if (requiredby)
	fprintf(mountfile_f, " %s", requiredby);
	if (wantedby && wantedby_append)
	fprintf(mountfile_f, " %s", wantedby);

	fprintf(mountfile_f,
	"\n"
	"After=");
	if (p_systemd_after)
	fprintf(mountfile_f, "%s ", p_systemd_after);
	fprintf(mountfile_f, "%s\n", after);

	fprintf(mountfile_f, "Wants=%s\n", wants);

	if (bindsto)
	fprintf(mountfile_f, "BindsTo=%s\n", bindsto);
	if (p_systemd_requires)
	fprintf(mountfile_f, "Requires=%s\n", p_systemd_requires);
	if (p_systemd_requiresmountsfor)
	fprintf(mountfile_f,
	"RequiresMountsFor=%s\n", p_systemd_requiresmountsfor);

	fprintf(mountfile_f,
	"\n"
	"[Mount]\n"
	"Where=%s\n"
	"What=%s\n"
	"Type=zfs\n"
	"Options=defaults%s,zfsutil\n",
	p_mountpoint, dataset, opts);

	(void) fclose(mountfile_f);

	if (!requiredby && !wantedby)
	- return (0);
	+ goto end;

	/* Finally, create the appropriate dependencies */
	char *linktgt;
	- if (asprintf(&linktgt, "../%s", mountfile) == -1)
	- EXIT_ENOMEM();
	+ if (asprintf(&linktgt, "../%s", mountfile) == -1) {
	+ fprintf(stderr, PROGNAME "[%d]: %s: "
	+ "out of memory for dependents of %s!\n",
	+ getpid(), dataset, mountfile);
	+ goto err;
	+ }
	+ *(tofree++) = linktgt;

	char *dependencies[][2] = {
	{"wants", wantedby},
	{"requires", requiredby},
	{}
	};
	for (__typeof__(&dependencies) dep = &dependencies; **dep; ++dep) {
	if (!(*dep)[1])
	continue;

	for (char reqby = strtok_r((dep)[1], " ", &toktmp);
	reqby;
	reqby = strtok_r(NULL, " ", &toktmp)) {
	char *depdir;
	- if (asprintf(&depdir, "%s.%s", reqby, (*dep)[0]) == -1)
	- EXIT_ENOMEM();
	+ if (asprintf(
	+ &depdir, "%s.%s", reqby, (*dep)[0]) == -1) {
	+ fprintf(stderr, PROGNAME "[%d]: %s: "
	+ "out of memory for dependent dir name "
	+ "\"%s.%s\"!\n",
	+ getpid(), dataset, reqby, (*dep)[0]);
	+ continue;
	+ }

	(void) mkdirat(destdir_fd, depdir, 0755);
	int depdir_fd = openat(destdir_fd, depdir,
	O_PATH \| O_DIRECTORY \| O_CLOEXEC);
	if (depdir_fd < 0) {
	fprintf(stderr, PROGNAME "[%d]: %s: "
	"couldn't open %s under %s: %s\n",
	getpid(), dataset, depdir, destdir,
	strerror(errno));
	free(depdir);
	continue;
	}

	if (symlinkat(linktgt, depdir_fd, mountfile) == -1)
	fprintf(stderr, PROGNAME "[%d]: %s: "
	"couldn't symlink at "
	"%s under %s under %s: %s\n",
	getpid(), dataset, mountfile,
	depdir, destdir, strerror(errno));

	(void) close(depdir_fd);
	free(depdir);
	}
	}

	- return (0);
	+end:
	+ if (tofree >= tofree_all + nitems(tofree_all)) {
	+ /*
	+ * This won't happen as-is:
	+ * we've got 8 slots and allocate 4 things at most.
	+ */
	+ fprintf(stderr,
	+ PROGNAME "[%d]: %s: need to free %zu > %zu!\n",
	+ getpid(), dataset, tofree - tofree_all, nitems(tofree_all));
	+ ret = tofree - tofree_all;
	+ }
	+
	+ while (tofree-- != tofree_all)
	+ free(*tofree);
	+ return (ret);
	+err:
	+ ret = 1;
	+ goto end;
	}


	static int
	pool_enumerator(zpool_handle_t pool, void data __attribute__((unused)))
	{
	int ret = 0;

	/*
	* Pools are guaranteed-unique by the kernel,
	* no risk of leaking dupes here
	*/
	char *name = strdup(zpool_get_name(pool));
	if (!name \|\| !tsearch(name, &known_pools, STRCMP)) {
	free(name);
	ret = ENOMEM;
	}

	zpool_close(pool);
	return (ret);
	}

	int
	main(int argc, char **argv)
	{
	struct timespec time_init = {};
	clock_gettime(CLOCK_MONOTONIC_RAW, &time_init);

	{
	int kmfd = open("/dev/kmsg", O_WRONLY \| O_CLOEXEC);
	if (kmfd >= 0) {
	(void) dup2(kmfd, STDERR_FILENO);
	(void) close(kmfd);
	+
	+ setlinebuf(stderr);
	}
	}

	- uint8_t debug = 0;
	-
	- argv0 = argv[0];
	switch (argc) {
	case 1:
	/* Use default */
	break;
	case 2:
	case 4:
	destdir = argv[1];
	break;
	default:
	fprintf(stderr,
	PROGNAME "[%d]: wrong argument count: %d\n",
	getpid(), argc - 1);
	_exit(1);
	}

	{
	destdir_fd = open(destdir, O_PATH \| O_DIRECTORY \| O_CLOEXEC);
	if (destdir_fd < 0) {
	fprintf(stderr, PROGNAME "[%d]: "
	"can't open destination directory %s: %s\n",
	getpid(), destdir, strerror(errno));
	_exit(1);
	}
	}

	DIR *fslist_dir = opendir(FSLIST);
	if (!fslist_dir) {
	if (errno != ENOENT)
	fprintf(stderr,
	PROGNAME "[%d]: couldn't open " FSLIST ": %s\n",
	getpid(), strerror(errno));
	_exit(0);
	}

	{
	libzfs_handle_t *libzfs = libzfs_init();
	if (libzfs) {
	if (zpool_iter(libzfs, pool_enumerator, NULL) != 0)
	fprintf(stderr, PROGNAME "[%d]: "
	"error listing pools, ignoring\n",
	getpid());
	libzfs_fini(libzfs);
	} else
	fprintf(stderr, PROGNAME "[%d]: "
	"couldn't start libzfs, ignoring\n",
	getpid());
	}

	{
	int regerr = regcomp(&uri_regex, URI_REGEX_S, 0);
	if (regerr != 0) {
	fprintf(stderr,
	PROGNAME "[%d]: invalid regex: %d\n",
	getpid(), regerr);
	_exit(1);
	}
	}

	- {
	- /*
	- * We could just get a gigabyte here and Not Care,
	- * but if vm.overcommit_memory=2, then MAP_NORESERVE is ignored
	- * and we'd try (and likely fail) to rip it out of swap
	- */
	- noauto_files = mmap(NULL, 4 * 1024 * 1024,
	- PROT_READ \| PROT_WRITE,
	- MAP_SHARED \| MAP_ANONYMOUS \| MAP_NORESERVE, -1, 0);
	- if (noauto_files == MAP_FAILED) {
	- fprintf(stderr,
	- PROGNAME "[%d]: couldn't allocate IPC region: %s\n",
	- getpid(), strerror(errno));
	- _exit(1);
	- }
	-
	- sem_init(&noauto_files->noauto_not_on_sem, true, 0);
	- sem_init(&noauto_files->noauto_names_sem, true, 1);
	- noauto_files->noauto_names_len = 0;
	- /* Works out to 16447ish, well enough */
	- noauto_files->noauto_names_max =
	- (4 * 1024 * 1024 - sizeof (*noauto_files)) / NAME_MAX;
	- }
	-
	+ bool debug = false;
	char *line = NULL;
	size_t linelen = 0;
	- struct timespec time_start = {};
	{
	const char *dbgenv = getenv("ZFS_DEBUG");
	if (dbgenv)
	debug = atoi(dbgenv);
	else {
	FILE *cmdline = fopen("/proc/cmdline", "re");
	if (cmdline != NULL) {
	if (getline(&line, &linelen, cmdline) >= 0)
	- debug = strstr(line, "debug") ? 2 : 0;
	+ debug = strstr(line, "debug");
	(void) fclose(cmdline);
	}
	}

	if (debug && !isatty(STDOUT_FILENO))
	dup2(STDERR_FILENO, STDOUT_FILENO);
	}

	- size_t forked_canmount_on = 0;
	- size_t forked_canmount_not_on = 0;
	- size_t canmount_on_pids_len = 128;
	- pid_t *canmount_on_pids =
	- malloc(canmount_on_pids_len * sizeof (*canmount_on_pids));
	- if (canmount_on_pids == NULL)
	- canmount_on_pids_len = 0;
	-
	+ struct timespec time_start = {};
	if (debug)
	clock_gettime(CLOCK_MONOTONIC_RAW, &time_start);

	- ssize_t read;
	- pid_t pid;
	+ struct line {
	+ char *line;
	+ const char *fname;
	+ struct line *next;
	+ } *lines_canmount_not_on = NULL;
	+
	+ int ret = 0;
	struct dirent *cachent;
	while ((cachent = readdir(fslist_dir)) != NULL) {
	if (strcmp(cachent->d_name, ".") == 0 \|\|
	strcmp(cachent->d_name, "..") == 0)
	continue;

	FILE *cachefile = fopenat(dirfd(fslist_dir), cachent->d_name,
	O_RDONLY \| O_CLOEXEC, "r", 0);
	if (!cachefile) {
	fprintf(stderr, PROGNAME "[%d]: "
	"couldn't open %s under " FSLIST ": %s\n",
	getpid(), cachent->d_name, strerror(errno));
	continue;
	}

	+ const char *filename = FREE_STATICS ? "(elided)" : NULL;
	+
	+ ssize_t read;
	while ((read = getline(&line, &linelen, cachefile)) >= 0) {
	line[read - 1] = '\0'; /* newline */

	- switch (pid = fork()) {
	- case -1:
	- fprintf(stderr,
	- PROGNAME "[%d]: couldn't fork for %s: %s\n",
	- getpid(), line, strerror(errno));
	- break;
	- case 0: /* child */
	- _exit(line_worker(line, cachent->d_name));
	- default: { /* parent */
	- char *tmp;
	- char *dset = strtok_r(line, "\t", &tmp);
	- strtok_r(NULL, "\t", &tmp);
	- char *canmount = strtok_r(NULL, "\t", &tmp);
	- bool canmount_on =
	- canmount && strncmp(canmount, "on", 2) == 0;
	-
	- if (debug >= 2)
	- printf(PROGNAME ": forked %d, "
	- "canmount_on=%d, dataset=%s\n",
	- (int)pid, canmount_on, dset);
	-
	- if (canmount_on &&
	- forked_canmount_on ==
	- canmount_on_pids_len) {
	- size_t new_len =
	- (canmount_on_pids_len ?: 16) * 2;
	- void *new_pidlist =
	- realloc(canmount_on_pids,
	- new_len *
	- sizeof (*canmount_on_pids));
	- if (!new_pidlist) {
	- fprintf(stderr,
	- PROGNAME "[%d]: "
	- "out of memory! "
	- "Mount ordering may be "
	- "affected.\n", getpid());
	- continue;
	- }
	-
	- canmount_on_pids = new_pidlist;
	- canmount_on_pids_len = new_len;
	- }
	+ char *canmount = line;
	+ canmount += strcspn(canmount, "\t");
	+ canmount += strspn(canmount, "\t");
	+ canmount += strcspn(canmount, "\t");
	+ canmount += strspn(canmount, "\t");
	+ bool canmount_on = strncmp(canmount, "on", 2) == 0;

	- if (canmount_on) {
	- canmount_on_pids[forked_canmount_on] =
	- pid;
	- ++forked_canmount_on;
	- } else
	- ++forked_canmount_not_on;
	- break;
	- }
	+ if (canmount_on)
	+ ret \|= line_worker(line, cachent->d_name);
	+ else {
	+ if (filename == NULL)
	+ filename =
	+ strdup(cachent->d_name) ?: "(?)";
	+
	+ struct line l = calloc(1, sizeof (l));
	+ char *nl = strdup(line);
	+ if (l == NULL \|\| nl == NULL) {
	+ fprintf(stderr, PROGNAME "[%d]: "
	+ "out of memory for \"%s\" in %s\n",
	+ getpid(), line, cachent->d_name);
	+ free(l);
	+ free(nl);
	+ continue;
	+ }
	+ l->line = nl;
	+ l->fname = filename;
	+ l->next = lines_canmount_not_on;
	+ lines_canmount_not_on = l;
	}
	}

	- (void) fclose(cachefile);
	+ fclose(cachefile);
	}
	free(line);

	- if (forked_canmount_on == 0) {
	- /* No canmount=on processes to finish, so don't deadlock here */
	- for (size_t i = 0; i < forked_canmount_not_on; ++i)
	- sem_post(&noauto_files->noauto_not_on_sem);
	- } else {
	- /* Likely a no-op, since we got these from a narrow fork loop */
	- qsort(canmount_on_pids, forked_canmount_on,
	- sizeof (*canmount_on_pids), PID_T_CMP);
	- }
	+ while (lines_canmount_not_on) {
	+ struct line *l = lines_canmount_not_on;
	+ lines_canmount_not_on = l->next;

	- int status, ret = 0;
	- struct rusage usage;
	- size_t forked_canmount_on_max = forked_canmount_on;
	- while ((pid = wait4(-1, &status, 0, &usage)) != -1) {
	- ret \|= WEXITSTATUS(status) \| WTERMSIG(status);
	-
	- if (forked_canmount_on != 0) {
	- if (bsearch(&pid, canmount_on_pids,
	- forked_canmount_on_max, sizeof (*canmount_on_pids),
	- PID_T_CMP))
	- --forked_canmount_on;
	-
	- if (forked_canmount_on == 0) {
	- /*
	- * All canmount=on processes have finished,
	- * let all the lower-priority ones finish now
	- */
	- for (size_t i = 0;
	- i < forked_canmount_not_on; ++i)
	- sem_post(
	- &noauto_files->noauto_not_on_sem);
	- }
	+ ret \|= line_worker(l->line, l->fname);
	+ if (FREE_STATICS) {
	+ free(l->line);
	+ free(l);
	}
	-
	- if (debug >= 2)
	- printf(PROGNAME ": %d done, user=%llu.%06us, "
	- "system=%llu.%06us, maxrss=%ldB, ex=0x%x\n",
	- (int)pid,
	- (unsigned long long) usage.ru_utime.tv_sec,
	- (unsigned int) usage.ru_utime.tv_usec,
	- (unsigned long long) usage.ru_stime.tv_sec,
	- (unsigned int) usage.ru_stime.tv_usec,
	- usage.ru_maxrss * 1024, status);
	}

	if (debug) {
	struct timespec time_end = {};
	clock_gettime(CLOCK_MONOTONIC_RAW, &time_end);

	+ struct rusage usage;
	getrusage(RUSAGE_SELF, &usage);
	printf(
	"\n"
	- PROGNAME ": self : "
	- "user=%llu.%06us, system=%llu.%06us, maxrss=%ldB\n",
	- (unsigned long long) usage.ru_utime.tv_sec,
	- (unsigned int) usage.ru_utime.tv_usec,
	- (unsigned long long) usage.ru_stime.tv_sec,
	- (unsigned int) usage.ru_stime.tv_usec,
	- usage.ru_maxrss * 1024);
	-
	- getrusage(RUSAGE_CHILDREN, &usage);
	- printf(PROGNAME ": children: "
	+ PROGNAME ": "
	"user=%llu.%06us, system=%llu.%06us, maxrss=%ldB\n",
	(unsigned long long) usage.ru_utime.tv_sec,
	(unsigned int) usage.ru_utime.tv_usec,
	(unsigned long long) usage.ru_stime.tv_sec,
	(unsigned int) usage.ru_stime.tv_usec,
	usage.ru_maxrss * 1024);

	if (time_start.tv_nsec > time_end.tv_nsec) {
	time_end.tv_nsec =
	1000000000 + time_end.tv_nsec - time_start.tv_nsec;
	time_end.tv_sec -= 1;
	} else
	time_end.tv_nsec -= time_start.tv_nsec;
	time_end.tv_sec -= time_start.tv_sec;

	if (time_init.tv_nsec > time_start.tv_nsec) {
	time_start.tv_nsec =
	1000000000 + time_start.tv_nsec - time_init.tv_nsec;
	time_start.tv_sec -= 1;
	} else
	time_start.tv_nsec -= time_init.tv_nsec;
	time_start.tv_sec -= time_init.tv_sec;

	time_init.tv_nsec = time_start.tv_nsec + time_end.tv_nsec;
	time_init.tv_sec =
	time_start.tv_sec + time_end.tv_sec +
	time_init.tv_nsec / 1000000000;
	time_init.tv_nsec %= 1000000000;

	- printf(PROGNAME ": wall : "
	+ printf(PROGNAME ": "
	"total=%llu.%09llus = "
	"init=%llu.%09llus + real=%llu.%09llus\n",
	(unsigned long long) time_init.tv_sec,
	(unsigned long long) time_init.tv_nsec,
	(unsigned long long) time_start.tv_sec,
	(unsigned long long) time_start.tv_nsec,
	(unsigned long long) time_end.tv_sec,
	(unsigned long long) time_end.tv_nsec);
	+
	+ fflush(stdout);
	}

	+ if (FREE_STATICS) {
	+ closedir(fslist_dir);
	+ tdestroy(noauto_files, free);
	+ tdestroy(known_pools, free);
	+ regfree(&uri_regex);
	+ }
	_exit(ret);
	}
	diff --git a/sys/contrib/openzfs/etc/zfs/zfs-functions.in b/sys/contrib/openzfs/etc/zfs/zfs-functions.in
	index a27d4c6cb073..30441dc35d4b 100644
	--- a/sys/contrib/openzfs/etc/zfs/zfs-functions.in
	+++ b/sys/contrib/openzfs/etc/zfs/zfs-functions.in
	@@ -1,429 +1,432 @@
	-# This is a script with common functions etc used by zfs-import, zfs-mount,
	-# zfs-share and zfs-zed.
	+# This is a script with common functions etc used by zfs-import, zfs-load-key,
	+# zfs-mount, zfs-share and zfs-zed.
	#
	# It is _NOT_ to be called independently
	#
	# Released under the 2-clause BSD license.
	#
	# This script is based on debian/zfsutils.zfs.init from the
	# Debian GNU/kFreeBSD zfsutils 8.1-3 package, written by Aurelien Jarno.

	PATH=/sbin:/bin:/usr/bin:/usr/sbin

	# Source function library
	if [ -f /etc/rc.d/init.d/functions ]; then
	# RedHat and derivatives
	. /etc/rc.d/init.d/functions
	elif [ -L /etc/init.d/functions.sh ]; then
	# Gentoo
	. /etc/init.d/functions.sh
	elif [ -f /lib/lsb/init-functions ]; then
	# LSB, Debian, and derivatives
	. /lib/lsb/init-functions
	fi

	# Of course the functions we need are called differently
	# on different distributions - it would be way too easy
	# otherwise!!
	if type log_failure_msg > /dev/null 2>&1 ; then
	# LSB functions - fall through
	zfs_log_begin_msg() { log_begin_msg "$1"; }
	zfs_log_end_msg() { log_end_msg "$1"; }
	zfs_log_failure_msg() { log_failure_msg "$1"; }
	zfs_log_progress_msg() { log_progress_msg "$1"; }
	elif type success > /dev/null 2>&1 ; then
	# Fedora/RedHat functions
	zfs_set_ifs() {
	# For some reason, the init function library have a problem
	# with a changed IFS, so this function goes around that.
	local tIFS="$1"
	if [ -n "$tIFS" ]
	then
	TMP_IFS="$IFS"
	IFS="$tIFS"
	fi
	}

	zfs_log_begin_msg() { printf "%s" "$1 "; }
	zfs_log_end_msg() {
	zfs_set_ifs "$OLD_IFS"
	if [ "$1" -eq 0 ]; then
	success
	else
	failure
	fi
	echo
	zfs_set_ifs "$TMP_IFS"
	}
	zfs_log_failure_msg() {
	zfs_set_ifs "$OLD_IFS"
	failure
	echo
	zfs_set_ifs "$TMP_IFS"
	}
	zfs_log_progress_msg() { printf "%s" "$""$1"; }
	elif type einfo > /dev/null 2>&1 ; then
	# Gentoo functions
	zfs_log_begin_msg() { ebegin "$1"; }
	zfs_log_end_msg() { eend "$1"; }
	zfs_log_failure_msg() { eend "$1"; }
	# zfs_log_progress_msg() { printf "%s" "$1"; }
	zfs_log_progress_msg() { :; }
	else
	# Unknown - simple substitutes.
	zfs_log_begin_msg() { printf "%s" "$1"; }
	zfs_log_end_msg() {
	ret=$1
	if [ "$ret" -ge 1 ]; then
	echo " failed!"
	else
	echo " success"
	fi
	return "$ret"
	}
	zfs_log_failure_msg() { echo "$1"; }
	zfs_log_progress_msg() { printf "%s" "$1"; }
	fi

	# Paths to what we need
	ZFS="@sbindir@/zfs"
	ZED="@sbindir@/zed"
	ZPOOL="@sbindir@/zpool"
	ZPOOL_CACHE="@sysconfdir@/zfs/zpool.cache"

	# Sensible defaults
	+ZFS_LOAD_KEY='yes'
	+ZFS_UNLOAD_KEY='no'
	ZFS_MOUNT='yes'
	ZFS_UNMOUNT='yes'
	ZFS_SHARE='yes'
	ZFS_UNSHARE='yes'

	# Source zfs configuration, overriding the defaults
	if [ -f @initconfdir@/zfs ]; then
	. @initconfdir@/zfs
	fi

	# ----------------------------------------------------

	-export ZFS ZED ZPOOL ZPOOL_CACHE ZFS_MOUNT ZFS_UNMOUNT ZFS_SHARE ZFS_UNSHARE
	+export ZFS ZED ZPOOL ZPOOL_CACHE ZFS_LOAD_KEY ZFS_UNLOAD_KEY ZFS_MOUNT ZFS_UNMOUNT \
	+ ZFS_SHARE ZFS_UNSHARE

	zfs_action()
	{
	local MSG="$1"; shift
	local CMD="$*"
	local ret

	zfs_log_begin_msg "$MSG "
	$CMD
	ret=$?
	if [ "$ret" -eq 0 ]; then
	zfs_log_end_msg $ret
	else
	zfs_log_failure_msg $ret
	fi

	return $ret
	}

	# Returns
	# 0 if daemon has been started
	# 1 if daemon was already running
	# 2 if daemon could not be started
	# 3 if unsupported
	#
	zfs_daemon_start()
	{
	local PIDFILE="$1"; shift
	local DAEMON_BIN="$1"; shift

	if type start-stop-daemon > /dev/null 2>&1 ; then
	# LSB functions
	start-stop-daemon --start --quiet --pidfile "$PIDFILE" \
	--exec "$DAEMON_BIN" --test > /dev/null \|\| return 1

	# shellcheck disable=SC2086
	start-stop-daemon --start --quiet --exec "$DAEMON_BIN" -- \
	"$@" \|\| return 2

	# On Debian, there's a 'sendsigs' script that will
	# kill basically everything quite early and zed is stopped
	# much later than that. We don't want zed to be among them,
	# so add the zed pid to list of pids to ignore.
	if [ -f "$PIDFILE" ] && [ -d /run/sendsigs.omit.d ]
	then
	ln -sf "$PIDFILE" /run/sendsigs.omit.d/zed
	fi
	elif type daemon > /dev/null 2>&1 ; then
	# Fedora/RedHat functions
	# shellcheck disable=SC2086
	daemon --pidfile "$PIDFILE" "$DAEMON_BIN" "$@"
	return $?
	else
	# Unsupported
	return 3
	fi

	return 0
	}

	# Returns
	# 0 if daemon has been stopped
	# 1 if daemon was already stopped
	# 2 if daemon could not be stopped
	# 3 if unsupported
	#
	zfs_daemon_stop()
	{
	local PIDFILE="$1"
	local DAEMON_BIN="$2"
	local DAEMON_NAME="$3"

	if type start-stop-daemon > /dev/null 2>&1 ; then
	# LSB functions
	start-stop-daemon --stop --quiet --retry=TERM/30/KILL/5 \
	--pidfile "$PIDFILE" --name "$DAEMON_NAME"
	ret="$?"
	[ "$ret" = 0 ] && rm -f "$PIDFILE"

	return "$ret"
	elif type killproc > /dev/null 2>&1 ; then
	# Fedora/RedHat functions
	killproc -p "$PIDFILE" "$DAEMON_NAME"
	ret="$?"
	[ "$ret" = 0 ] && rm -f "$PIDFILE"

	return "$ret"
	else
	# Unsupported
	return 3
	fi

	return 0
	}

	# Returns status
	zfs_daemon_status()
	{
	local PIDFILE="$1"
	local DAEMON_BIN="$2"
	local DAEMON_NAME="$3"

	if type status_of_proc > /dev/null 2>&1 ; then
	# LSB functions
	status_of_proc "$DAEMON_NAME" "$DAEMON_BIN"
	return $?
	elif type status > /dev/null 2>&1 ; then
	# Fedora/RedHat functions
	status -p "$PIDFILE" "$DAEMON_NAME"
	return $?
	else
	# Unsupported
	return 3
	fi

	return 0
	}

	zfs_daemon_reload()
	{
	local PIDFILE="$1"
	local DAEMON_NAME="$2"

	if type start-stop-daemon > /dev/null 2>&1 ; then
	# LSB functions
	start-stop-daemon --stop --signal 1 --quiet \
	--pidfile "$PIDFILE" --name "$DAEMON_NAME"
	return $?
	elif type killproc > /dev/null 2>&1 ; then
	# Fedora/RedHat functions
	killproc -p "$PIDFILE" "$DAEMON_NAME" -HUP
	return $?
	else
	# Unsupported
	return 3
	fi

	return 0
	}

	zfs_installed()
	{
	if [ ! -x "$ZPOOL" ]; then
	return 1
	else
	# Test if it works (will catch missing/broken libs etc)
	"$ZPOOL" -? > /dev/null 2>&1
	return $?
	fi

	if [ ! -x "$ZFS" ]; then
	return 2
	else
	# Test if it works (will catch missing/broken libs etc)
	"$ZFS" -? > /dev/null 2>&1
	return $?
	fi

	return 0
	}

	# Trigger udev and wait for it to settle.
	udev_trigger()
	{
	if [ -x /sbin/udevadm ]; then
	/sbin/udevadm trigger --action=change --subsystem-match=block
	/sbin/udevadm settle
	elif [ -x /sbin/udevsettle ]; then
	/sbin/udevtrigger
	/sbin/udevsettle
	fi
	}

	# Do a lot of checks to make sure it's 'safe' to continue with the import.
	checksystem()
	{
	if grep -qiE '(^\|[^\\](\\\\)* )zfs=(off\|no\|0)( \|$)' /proc/cmdline;
	then
	# Called with zfs=(off\|no\|0) - bail because we don't
	# want anything import, mounted or shared.
	# HOWEVER, only do this if we're called at the boot up
	# (from init), not if we're running interactively (as in
	# from the shell - we know what we're doing).
	# shellcheck disable=SC2154
	[ -n "$init" ] && exit 3
	fi

	# Check if ZFS is installed.
	zfs_installed \|\| return 5

	# Just make sure that /dev/zfs is created.
	udev_trigger

	return 0
	}

	get_root_pool()
	{
	# shellcheck disable=SC2046
	set -- $(mount \| grep ' on / ')
	[ "$5" = "zfs" ] && echo "${1%%/*}"
	}

	# Check if a variable is 'yes' (any case) or '1'
	# Returns TRUE if set.
	check_boolean()
	{
	local var="$1"

	echo "$var" \| grep -Eiq "^yes$\|^on$\|^true$\|^1$" && return 0 \|\| return 1
	}

	check_module_loaded()
	{
	module="$1"

	[ -r "/sys/module/${module}/version" ] && return 0 \|\| return 1
	}

	load_module()
	{
	module="$1"

	# Load the zfs module stack
	if ! check_module_loaded "$module"; then
	if ! /sbin/modprobe "$module"; then
	return 5
	fi
	fi
	return 0
	}

	# first parameter is a regular expression that filters mtab
	read_mtab()
	{
	local match="$1"
	local fs mntpnt fstype opts rest

	# Unset all MTAB_* variables
	# shellcheck disable=SC2046
	unset $(env \| sed -e '/^MTAB_/!d' -e 's,=.*,,')

	while read -r fs mntpnt fstype opts rest; do
	if echo "$fs $mntpnt $fstype $opts" \| grep -qE "$match"; then
	# * Fix problems (!?) in the mounts file. It will record
	# 'rpool 1' as 'rpool\0401' instead of 'rpool\00401'
	# which seems to be the correct (at least as far as
	# 'printf' is concerned).
	# * We need to use the external echo, because the
	# internal one would interpret the backslash code
	# (incorrectly), giving us a instead.
	mntpnt=$(/bin/echo "$mntpnt" \| sed 's,\\0,\\00,g')
	fs=$(/bin/echo "$fs" \| sed 's,\\0,\\00,')

	# Remove 'unwanted' characters.
	mntpnt=$(printf '%b' "$mntpnt" \| tr -d '/. -')
	fs=$(printf '%b' "$fs")

	# Set the variable.
	eval export "MTAB_$mntpnt=\"$fs\""
	fi
	done < /proc/self/mounts
	}

	in_mtab()
	{
	local mntpnt="$1"
	# Remove 'unwanted' characters.
	mntpnt=$(printf '%b' "$mntpnt" \| tr -d '/. -')
	local var

	var="$(eval echo "MTAB_$mntpnt")"
	[ "$(eval echo "$""$var")" != "" ]
	return "$?"
	}

	# first parameter is a regular expression that filters fstab
	read_fstab()
	{
	local match="$1"
	local i var

	# Unset all FSTAB_* variables
	# shellcheck disable=SC2046
	unset $(env \| sed -e '/^FSTAB_/!d' -e 's,=.*,,')

	i=0
	while read -r fs mntpnt fstype opts; do
	echo "$fs" \| grep -qE '^#\|^$' && continue
	echo "$mntpnt" \| grep -qE '^none\|^swap' && continue
	echo "$fstype" \| grep -qE '^swap' && continue

	if echo "$fs $mntpnt $fstype $opts" \| grep -qE "$match"; then
	eval export "FSTAB_dev_$i=$fs"
	fs=$(printf '%b' "$fs" \| tr '/' '_')
	eval export "FSTAB_$i=$mntpnt"

	i=$((i + 1))
	fi
	done < /etc/fstab
	}

	in_fstab()
	{
	local var

	var="$(eval echo "FSTAB_$1")"
	[ "${var}" != "" ]
	return $?
	}

	is_mounted()
	{
	local mntpt="$1"
	local mp

	while read -r _ mp _; do
	[ "$mp" = "$mntpt" ] && return 0
	done < /proc/self/mounts

	return 1
	}
	diff --git a/sys/contrib/openzfs/include/libzfs.h b/sys/contrib/openzfs/include/libzfs.h
	index c0883a983678..e135ae2ee066 100644
	--- a/sys/contrib/openzfs/include/libzfs.h
	+++ b/sys/contrib/openzfs/include/libzfs.h
	@@ -1,970 +1,999 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright Joyent, Inc.
	* Copyright (c) 2013 Steven Hartland. All rights reserved.
	* Copyright (c) 2016, Intel Corporation.
	* Copyright 2016 Nexenta Systems, Inc.
	* Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
	* Copyright (c) 2019 Datto Inc.
	* Copyright (c) 2021, Colm Buckley <colm@tuatha.org>
	*/

	#ifndef _LIBZFS_H
	#define _LIBZFS_H extern __attribute__((visibility("default")))

	#include <assert.h>
	#include <libnvpair.h>
	#include <sys/mnttab.h>
	#include <sys/param.h>
	#include <sys/types.h>
	#include <sys/varargs.h>
	#include <sys/fs/zfs.h>
	#include <sys/avl.h>
	#include <ucred.h>
	#include <libzfs_core.h>

	#ifdef __cplusplus
	extern "C" {
	#endif

	/*
	* Miscellaneous ZFS constants
	*/
	#define ZFS_MAXPROPLEN MAXPATHLEN
	#define ZPOOL_MAXPROPLEN MAXPATHLEN

	/*
	* libzfs errors
	*/
	typedef enum zfs_error {
	EZFS_SUCCESS = 0, /* no error -- success */
	EZFS_NOMEM = 2000, /* out of memory */
	EZFS_BADPROP, /* invalid property value */
	EZFS_PROPREADONLY, /* cannot set readonly property */
	EZFS_PROPTYPE, /* property does not apply to dataset type */
	EZFS_PROPNONINHERIT, /* property is not inheritable */
	EZFS_PROPSPACE, /* bad quota or reservation */
	EZFS_BADTYPE, /* dataset is not of appropriate type */
	EZFS_BUSY, /* pool or dataset is busy */
	EZFS_EXISTS, /* pool or dataset already exists */
	EZFS_NOENT, /* no such pool or dataset */
	EZFS_BADSTREAM, /* bad backup stream */
	EZFS_DSREADONLY, /* dataset is readonly */
	EZFS_VOLTOOBIG, /* volume is too large for 32-bit system */
	EZFS_INVALIDNAME, /* invalid dataset name */
	EZFS_BADRESTORE, /* unable to restore to destination */
	EZFS_BADBACKUP, /* backup failed */
	EZFS_BADTARGET, /* bad attach/detach/replace target */
	EZFS_NODEVICE, /* no such device in pool */
	EZFS_BADDEV, /* invalid device to add */
	EZFS_NOREPLICAS, /* no valid replicas */
	EZFS_RESILVERING, /* resilvering (healing reconstruction) */
	EZFS_BADVERSION, /* unsupported version */
	EZFS_POOLUNAVAIL, /* pool is currently unavailable */
	EZFS_DEVOVERFLOW, /* too many devices in one vdev */
	EZFS_BADPATH, /* must be an absolute path */
	EZFS_CROSSTARGET, /* rename or clone across pool or dataset */
	EZFS_ZONED, /* used improperly in local zone */
	EZFS_MOUNTFAILED, /* failed to mount dataset */
	EZFS_UMOUNTFAILED, /* failed to unmount dataset */
	EZFS_UNSHARENFSFAILED, /* failed to unshare over nfs */
	EZFS_SHARENFSFAILED, /* failed to share over nfs */
	EZFS_PERM, /* permission denied */
	EZFS_NOSPC, /* out of space */
	EZFS_FAULT, /* bad address */
	EZFS_IO, /* I/O error */
	EZFS_INTR, /* signal received */
	EZFS_ISSPARE, /* device is a hot spare */
	EZFS_INVALCONFIG, /* invalid vdev configuration */
	EZFS_RECURSIVE, /* recursive dependency */
	EZFS_NOHISTORY, /* no history object */
	EZFS_POOLPROPS, /* couldn't retrieve pool props */
	EZFS_POOL_NOTSUP, /* ops not supported for this type of pool */
	EZFS_POOL_INVALARG, /* invalid argument for this pool operation */
	EZFS_NAMETOOLONG, /* dataset name is too long */
	EZFS_OPENFAILED, /* open of device failed */
	EZFS_NOCAP, /* couldn't get capacity */
	EZFS_LABELFAILED, /* write of label failed */
	EZFS_BADWHO, /* invalid permission who */
	EZFS_BADPERM, /* invalid permission */
	EZFS_BADPERMSET, /* invalid permission set name */
	EZFS_NODELEGATION, /* delegated administration is disabled */
	EZFS_UNSHARESMBFAILED, /* failed to unshare over smb */
	EZFS_SHARESMBFAILED, /* failed to share over smb */
	EZFS_BADCACHE, /* bad cache file */
	EZFS_ISL2CACHE, /* device is for the level 2 ARC */
	EZFS_VDEVNOTSUP, /* unsupported vdev type */
	EZFS_NOTSUP, /* ops not supported on this dataset */
	EZFS_ACTIVE_SPARE, /* pool has active shared spare devices */
	EZFS_UNPLAYED_LOGS, /* log device has unplayed logs */
	EZFS_REFTAG_RELE, /* snapshot release: tag not found */
	EZFS_REFTAG_HOLD, /* snapshot hold: tag already exists */
	EZFS_TAGTOOLONG, /* snapshot hold/rele: tag too long */
	EZFS_PIPEFAILED, /* pipe create failed */
	EZFS_THREADCREATEFAILED, /* thread create failed */
	EZFS_POSTSPLIT_ONLINE, /* onlining a disk after splitting it */
	EZFS_SCRUBBING, /* currently scrubbing */
	EZFS_NO_SCRUB, /* no active scrub */
	EZFS_DIFF, /* general failure of zfs diff */
	EZFS_DIFFDATA, /* bad zfs diff data */
	EZFS_POOLREADONLY, /* pool is in read-only mode */
	EZFS_SCRUB_PAUSED, /* scrub currently paused */
	EZFS_ACTIVE_POOL, /* pool is imported on a different system */
	EZFS_CRYPTOFAILED, /* failed to setup encryption */
	EZFS_NO_PENDING, /* cannot cancel, no operation is pending */
	EZFS_CHECKPOINT_EXISTS, /* checkpoint exists */
	EZFS_DISCARDING_CHECKPOINT, /* currently discarding a checkpoint */
	EZFS_NO_CHECKPOINT, /* pool has no checkpoint */
	EZFS_DEVRM_IN_PROGRESS, /* a device is currently being removed */
	EZFS_VDEV_TOO_BIG, /* a device is too big to be used */
	EZFS_IOC_NOTSUPPORTED, /* operation not supported by zfs module */
	EZFS_TOOMANY, /* argument list too long */
	EZFS_INITIALIZING, /* currently initializing */
	EZFS_NO_INITIALIZE, /* no active initialize */
	EZFS_WRONG_PARENT, /* invalid parent dataset (e.g ZVOL) */
	EZFS_TRIMMING, /* currently trimming */
	EZFS_NO_TRIM, /* no active trim */
	EZFS_TRIM_NOTSUP, /* device does not support trim */
	EZFS_NO_RESILVER_DEFER, /* pool doesn't support resilver_defer */
	EZFS_EXPORT_IN_PROGRESS, /* currently exporting the pool */
	EZFS_REBUILDING, /* resilvering (sequential reconstrution) */
	+ EZFS_VDEV_NOTSUP, /* ops not supported for this type of vdev */
	EZFS_UNKNOWN
	} zfs_error_t;

	/*
	* The following data structures are all part
	* of the zfs_allow_t data structure which is
	* used for printing 'allow' permissions.
	* It is a linked list of zfs_allow_t's which
	* then contain avl tree's for user/group/sets/...
	* and each one of the entries in those trees have
	* avl tree's for the permissions they belong to and
	* whether they are local,descendent or local+descendent
	* permissions. The AVL trees are used primarily for
	* sorting purposes, but also so that we can quickly find
	* a given user and or permission.
	*/
	typedef struct zfs_perm_node {
	avl_node_t z_node;
	char z_pname[MAXPATHLEN];
	} zfs_perm_node_t;

	typedef struct zfs_allow_node {
	avl_node_t z_node;
	char z_key[MAXPATHLEN]; /* name, such as joe */
	avl_tree_t z_localdescend; /* local+descendent perms */
	avl_tree_t z_local; /* local permissions */
	avl_tree_t z_descend; /* descendent permissions */
	} zfs_allow_node_t;

	typedef struct zfs_allow {
	struct zfs_allow *z_next;
	char z_setpoint[MAXPATHLEN];
	avl_tree_t z_sets;
	avl_tree_t z_crperms;
	avl_tree_t z_user;
	avl_tree_t z_group;
	avl_tree_t z_everyone;
	} zfs_allow_t;

	/*
	* Basic handle types
	*/
	typedef struct zfs_handle zfs_handle_t;
	typedef struct zpool_handle zpool_handle_t;
	typedef struct libzfs_handle libzfs_handle_t;

	_LIBZFS_H int zpool_wait(zpool_handle_t *, zpool_wait_activity_t);
	_LIBZFS_H int zpool_wait_status(zpool_handle_t *, zpool_wait_activity_t,
	boolean_t , boolean_t );

	/*
	* Library initialization
	*/
	_LIBZFS_H libzfs_handle_t *libzfs_init(void);
	_LIBZFS_H void libzfs_fini(libzfs_handle_t *);

	_LIBZFS_H libzfs_handle_t zpool_get_handle(zpool_handle_t );
	_LIBZFS_H libzfs_handle_t zfs_get_handle(zfs_handle_t );

	_LIBZFS_H void libzfs_print_on_error(libzfs_handle_t *, boolean_t);

	_LIBZFS_H void zfs_save_arguments(int argc, char *, char , int);
	_LIBZFS_H int zpool_log_history(libzfs_handle_t , const char );

	_LIBZFS_H int libzfs_errno(libzfs_handle_t *);
	_LIBZFS_H const char *libzfs_error_init(int);
	_LIBZFS_H const char libzfs_error_action(libzfs_handle_t );
	_LIBZFS_H const char libzfs_error_description(libzfs_handle_t );
	_LIBZFS_H int zfs_standard_error(libzfs_handle_t , int, const char );
	_LIBZFS_H void libzfs_mnttab_init(libzfs_handle_t *);
	_LIBZFS_H void libzfs_mnttab_fini(libzfs_handle_t *);
	_LIBZFS_H void libzfs_mnttab_cache(libzfs_handle_t *, boolean_t);
	_LIBZFS_H int libzfs_mnttab_find(libzfs_handle_t , const char ,
	struct mnttab *);
	_LIBZFS_H void libzfs_mnttab_add(libzfs_handle_t , const char ,
	const char , const char );
	_LIBZFS_H void libzfs_mnttab_remove(libzfs_handle_t , const char );

	/*
	* Basic handle functions
	*/
	_LIBZFS_H zpool_handle_t zpool_open(libzfs_handle_t , const char *);
	_LIBZFS_H zpool_handle_t zpool_open_canfail(libzfs_handle_t , const char *);
	_LIBZFS_H void zpool_close(zpool_handle_t *);
	_LIBZFS_H const char zpool_get_name(zpool_handle_t );
	_LIBZFS_H int zpool_get_state(zpool_handle_t *);
	_LIBZFS_H const char *zpool_state_to_name(vdev_state_t, vdev_aux_t);
	_LIBZFS_H const char *zpool_pool_state_to_name(pool_state_t);
	_LIBZFS_H void zpool_free_handles(libzfs_handle_t *);

	/*
	* Iterate over all active pools in the system.
	*/
	typedef int (zpool_iter_f)(zpool_handle_t , void *);
	_LIBZFS_H int zpool_iter(libzfs_handle_t , zpool_iter_f, void );
	_LIBZFS_H boolean_t zpool_skip_pool(const char *);

	/*
	* Functions to create and destroy pools
	*/
	_LIBZFS_H int zpool_create(libzfs_handle_t , const char , nvlist_t *,
	nvlist_t , nvlist_t );
	_LIBZFS_H int zpool_destroy(zpool_handle_t , const char );
	_LIBZFS_H int zpool_add(zpool_handle_t , nvlist_t );

	typedef struct splitflags {
	/* do not split, but return the config that would be split off */
	int dryrun : 1;

	/* after splitting, import the pool */
	int import : 1;
	int name_flags;
	} splitflags_t;

	typedef struct trimflags {
	/* requested vdevs are for the entire pool */
	boolean_t fullpool;

	/* request a secure trim, requires support from device */
	boolean_t secure;

	/* after starting trim, block until trim completes */
	boolean_t wait;

	/* trim at the requested rate in bytes/second */
	uint64_t rate;
	} trimflags_t;

	/*
	* Functions to manipulate pool and vdev state
	*/
	_LIBZFS_H int zpool_scan(zpool_handle_t *, pool_scan_func_t, pool_scrub_cmd_t);
	_LIBZFS_H int zpool_initialize(zpool_handle_t *, pool_initialize_func_t,
	nvlist_t *);
	_LIBZFS_H int zpool_initialize_wait(zpool_handle_t *, pool_initialize_func_t,
	nvlist_t *);
	_LIBZFS_H int zpool_trim(zpool_handle_t , pool_trim_func_t, nvlist_t ,
	trimflags_t *);

	_LIBZFS_H int zpool_clear(zpool_handle_t , const char , nvlist_t *);
	_LIBZFS_H int zpool_reguid(zpool_handle_t *);
	_LIBZFS_H int zpool_reopen_one(zpool_handle_t , void );

	_LIBZFS_H int zpool_sync_one(zpool_handle_t , void );

	_LIBZFS_H int zpool_vdev_online(zpool_handle_t , const char , int,
	vdev_state_t *);
	_LIBZFS_H int zpool_vdev_offline(zpool_handle_t , const char , boolean_t);
	_LIBZFS_H int zpool_vdev_attach(zpool_handle_t , const char ,
	const char , nvlist_t , int, boolean_t);
	_LIBZFS_H int zpool_vdev_detach(zpool_handle_t , const char );
	_LIBZFS_H int zpool_vdev_remove(zpool_handle_t , const char );
	_LIBZFS_H int zpool_vdev_remove_cancel(zpool_handle_t *);
	_LIBZFS_H int zpool_vdev_indirect_size(zpool_handle_t , const char ,
	uint64_t *);
	_LIBZFS_H int zpool_vdev_split(zpool_handle_t , char , nvlist_t **,
	nvlist_t *, splitflags_t);

	_LIBZFS_H int zpool_vdev_fault(zpool_handle_t *, uint64_t, vdev_aux_t);
	_LIBZFS_H int zpool_vdev_degrade(zpool_handle_t *, uint64_t, vdev_aux_t);
	_LIBZFS_H int zpool_vdev_clear(zpool_handle_t *, uint64_t);

	_LIBZFS_H nvlist_t zpool_find_vdev(zpool_handle_t , const char , boolean_t ,
	boolean_t , boolean_t );
	_LIBZFS_H nvlist_t zpool_find_vdev_by_physpath(zpool_handle_t , const char *,
	boolean_t , boolean_t , boolean_t *);
	_LIBZFS_H int zpool_label_disk(libzfs_handle_t , zpool_handle_t ,
	const char *);
	_LIBZFS_H uint64_t zpool_vdev_path_to_guid(zpool_handle_t *zhp,
	const char *path);

	_LIBZFS_H const char zpool_get_state_str(zpool_handle_t );

	/*
	* Functions to manage pool properties
	*/
	_LIBZFS_H int zpool_set_prop(zpool_handle_t , const char , const char *);
	_LIBZFS_H int zpool_get_prop(zpool_handle_t , zpool_prop_t, char ,
	size_t proplen, zprop_source_t *, boolean_t literal);
	_LIBZFS_H uint64_t zpool_get_prop_int(zpool_handle_t *, zpool_prop_t,
	zprop_source_t *);
	_LIBZFS_H int zpool_props_refresh(zpool_handle_t *);

	_LIBZFS_H const char *zpool_prop_to_name(zpool_prop_t);
	_LIBZFS_H const char *zpool_prop_values(zpool_prop_t);

	+/*
	+ * Functions to manage vdev properties
	+ */
	+_LIBZFS_H int zpool_get_vdev_prop_value(nvlist_t , vdev_prop_t, char , char *,
	+ size_t, zprop_source_t *, boolean_t);
	+_LIBZFS_H int zpool_get_vdev_prop(zpool_handle_t , const char , vdev_prop_t,
	+ char , char , size_t, zprop_source_t *, boolean_t);
	+_LIBZFS_H int zpool_get_all_vdev_props(zpool_handle_t , const char ,
	+ nvlist_t **);
	+_LIBZFS_H int zpool_set_vdev_prop(zpool_handle_t , const char , const char *,
	+ const char *);
	+
	+_LIBZFS_H const char *vdev_prop_to_name(vdev_prop_t);
	+_LIBZFS_H const char *vdev_prop_values(vdev_prop_t);
	+_LIBZFS_H boolean_t vdev_prop_user(const char *name);
	+_LIBZFS_H const char *vdev_prop_column_name(vdev_prop_t);
	+_LIBZFS_H boolean_t vdev_prop_align_right(vdev_prop_t);
	+
	/*
	* Pool health statistics.
	*/
	typedef enum {
	/*
	* The following correspond to faults as defined in the (fault.fs.zfs.*)
	* event namespace. Each is associated with a corresponding message ID.
	* This must be kept in sync with the zfs_msgid_table in
	* lib/libzfs/libzfs_status.c.
	*/
	ZPOOL_STATUS_CORRUPT_CACHE, /* corrupt /kernel/drv/zpool.cache */
	ZPOOL_STATUS_MISSING_DEV_R, /* missing device with replicas */
	ZPOOL_STATUS_MISSING_DEV_NR, /* missing device with no replicas */
	ZPOOL_STATUS_CORRUPT_LABEL_R, /* bad device label with replicas */
	ZPOOL_STATUS_CORRUPT_LABEL_NR, /* bad device label with no replicas */
	ZPOOL_STATUS_BAD_GUID_SUM, /* sum of device guids didn't match */
	ZPOOL_STATUS_CORRUPT_POOL, /* pool metadata is corrupted */
	ZPOOL_STATUS_CORRUPT_DATA, /* data errors in user (meta)data */
	ZPOOL_STATUS_FAILING_DEV, /* device experiencing errors */
	ZPOOL_STATUS_VERSION_NEWER, /* newer on-disk version */
	ZPOOL_STATUS_HOSTID_MISMATCH, /* last accessed by another system */
	ZPOOL_STATUS_HOSTID_ACTIVE, /* currently active on another system */
	ZPOOL_STATUS_HOSTID_REQUIRED, /* multihost=on and hostid=0 */
	ZPOOL_STATUS_IO_FAILURE_WAIT, /* failed I/O, failmode 'wait' */
	ZPOOL_STATUS_IO_FAILURE_CONTINUE, /* failed I/O, failmode 'continue' */
	ZPOOL_STATUS_IO_FAILURE_MMP, /* failed MMP, failmode not 'panic' */
	ZPOOL_STATUS_BAD_LOG, /* cannot read log chain(s) */
	ZPOOL_STATUS_ERRATA, /* informational errata available */

	/*
	* If the pool has unsupported features but can still be opened in
	* read-only mode, its status is ZPOOL_STATUS_UNSUP_FEAT_WRITE. If the
	* pool has unsupported features but cannot be opened at all, its
	* status is ZPOOL_STATUS_UNSUP_FEAT_READ.
	*/
	ZPOOL_STATUS_UNSUP_FEAT_READ, /* unsupported features for read */
	ZPOOL_STATUS_UNSUP_FEAT_WRITE, /* unsupported features for write */

	/*
	* These faults have no corresponding message ID. At the time we are
	* checking the status, the original reason for the FMA fault (I/O or
	* checksum errors) has been lost.
	*/
	ZPOOL_STATUS_FAULTED_DEV_R, /* faulted device with replicas */
	ZPOOL_STATUS_FAULTED_DEV_NR, /* faulted device with no replicas */

	/*
	* The following are not faults per se, but still an error possibly
	* requiring administrative attention. There is no corresponding
	* message ID.
	*/
	ZPOOL_STATUS_VERSION_OLDER, /* older legacy on-disk version */
	ZPOOL_STATUS_FEAT_DISABLED, /* supported features are disabled */
	ZPOOL_STATUS_RESILVERING, /* device being resilvered */
	ZPOOL_STATUS_OFFLINE_DEV, /* device offline */
	ZPOOL_STATUS_REMOVED_DEV, /* removed device */
	ZPOOL_STATUS_REBUILDING, /* device being rebuilt */
	ZPOOL_STATUS_REBUILD_SCRUB, /* recommend scrubbing the pool */
	ZPOOL_STATUS_NON_NATIVE_ASHIFT, /* (e.g. 512e dev with ashift of 9) */
	ZPOOL_STATUS_COMPATIBILITY_ERR, /* bad 'compatibility' property */
	ZPOOL_STATUS_INCOMPATIBLE_FEAT, /* feature set outside compatibility */

	/*
	* Finally, the following indicates a healthy pool.
	*/
	ZPOOL_STATUS_OK
	} zpool_status_t;

	_LIBZFS_H zpool_status_t zpool_get_status(zpool_handle_t , char *,
	zpool_errata_t *);
	_LIBZFS_H zpool_status_t zpool_import_status(nvlist_t , char *,
	zpool_errata_t *);

	/*
	* Statistics and configuration functions.
	*/
	_LIBZFS_H nvlist_t zpool_get_config(zpool_handle_t , nvlist_t **);
	_LIBZFS_H nvlist_t zpool_get_features(zpool_handle_t );
	_LIBZFS_H int zpool_refresh_stats(zpool_handle_t , boolean_t );
	_LIBZFS_H int zpool_get_errlog(zpool_handle_t , nvlist_t *);

	/*
	* Import and export functions
	*/
	_LIBZFS_H int zpool_export(zpool_handle_t , boolean_t, const char );
	_LIBZFS_H int zpool_export_force(zpool_handle_t , const char );
	_LIBZFS_H int zpool_import(libzfs_handle_t , nvlist_t , const char *,
	char *altroot);
	_LIBZFS_H int zpool_import_props(libzfs_handle_t , nvlist_t , const char *,
	nvlist_t *, int);
	_LIBZFS_H void zpool_print_unsup_feat(nvlist_t *config);

	/*
	* Miscellaneous pool functions
	*/
	struct zfs_cmd;

	_LIBZFS_H const char *zfs_history_event_names[];

	typedef enum {
	VDEV_NAME_PATH = 1 << 0,
	VDEV_NAME_GUID = 1 << 1,
	VDEV_NAME_FOLLOW_LINKS = 1 << 2,
	VDEV_NAME_TYPE_ID = 1 << 3,
	} vdev_name_t;

	_LIBZFS_H char zpool_vdev_name(libzfs_handle_t , zpool_handle_t , nvlist_t ,
	int name_flags);
	_LIBZFS_H int zpool_upgrade(zpool_handle_t *, uint64_t);
	_LIBZFS_H int zpool_get_history(zpool_handle_t , nvlist_t , uint64_t ,
	boolean_t *);
	_LIBZFS_H int zpool_events_next(libzfs_handle_t , nvlist_t , int , unsigned,
	int);
	_LIBZFS_H int zpool_events_clear(libzfs_handle_t , int );
	_LIBZFS_H int zpool_events_seek(libzfs_handle_t *, uint64_t, int);
	_LIBZFS_H void zpool_obj_to_path_ds(zpool_handle_t *, uint64_t, uint64_t,
	char *, size_t);
	_LIBZFS_H void zpool_obj_to_path(zpool_handle_t , uint64_t, uint64_t, char ,
	size_t);
	_LIBZFS_H int zfs_ioctl(libzfs_handle_t , int, struct zfs_cmd );
	_LIBZFS_H int zpool_get_physpath(zpool_handle_t , char , size_t);
	_LIBZFS_H void zpool_explain_recover(libzfs_handle_t , const char , int,
	nvlist_t *);
	_LIBZFS_H int zpool_checkpoint(zpool_handle_t *);
	_LIBZFS_H int zpool_discard_checkpoint(zpool_handle_t *);
	_LIBZFS_H boolean_t zpool_is_draid_spare(const char *);

	/*
	* Basic handle manipulations. These functions do not create or destroy the
	* underlying datasets, only the references to them.
	*/
	_LIBZFS_H zfs_handle_t zfs_open(libzfs_handle_t , const char *, int);
	_LIBZFS_H zfs_handle_t zfs_handle_dup(zfs_handle_t );
	_LIBZFS_H void zfs_close(zfs_handle_t *);
	_LIBZFS_H zfs_type_t zfs_get_type(const zfs_handle_t *);
	_LIBZFS_H zfs_type_t zfs_get_underlying_type(const zfs_handle_t *);
	_LIBZFS_H const char zfs_get_name(const zfs_handle_t );
	_LIBZFS_H zpool_handle_t zfs_get_pool_handle(const zfs_handle_t );
	_LIBZFS_H const char zfs_get_pool_name(const zfs_handle_t );

	/*
	* Property management functions. Some functions are shared with the kernel,
	* and are found in sys/fs/zfs.h.
	*/

	/*
	* zfs dataset property management
	*/
	_LIBZFS_H const char *zfs_prop_default_string(zfs_prop_t);
	_LIBZFS_H uint64_t zfs_prop_default_numeric(zfs_prop_t);
	_LIBZFS_H const char *zfs_prop_column_name(zfs_prop_t);
	_LIBZFS_H boolean_t zfs_prop_align_right(zfs_prop_t);

	_LIBZFS_H nvlist_t zfs_valid_proplist(libzfs_handle_t , zfs_type_t,
	nvlist_t , uint64_t, zfs_handle_t , zpool_handle_t *, boolean_t,
	const char *);

	_LIBZFS_H const char *zfs_prop_to_name(zfs_prop_t);
	_LIBZFS_H int zfs_prop_set(zfs_handle_t , const char , const char *);
	_LIBZFS_H int zfs_prop_set_list(zfs_handle_t , nvlist_t );
	_LIBZFS_H int zfs_prop_get(zfs_handle_t , zfs_prop_t, char , size_t,
	zprop_source_t , char , size_t, boolean_t);
	_LIBZFS_H int zfs_prop_get_recvd(zfs_handle_t , const char , char *, size_t,
	boolean_t);
	_LIBZFS_H int zfs_prop_get_numeric(zfs_handle_t , zfs_prop_t, uint64_t ,
	zprop_source_t , char , size_t);
	_LIBZFS_H int zfs_prop_get_userquota_int(zfs_handle_t *zhp,
	const char propname, uint64_t propvalue);
	_LIBZFS_H int zfs_prop_get_userquota(zfs_handle_t zhp, const char propname,
	char *propbuf, int proplen, boolean_t literal);
	_LIBZFS_H int zfs_prop_get_written_int(zfs_handle_t zhp, const char propname,
	uint64_t *propvalue);
	_LIBZFS_H int zfs_prop_get_written(zfs_handle_t zhp, const char propname,
	char *propbuf, int proplen, boolean_t literal);
	_LIBZFS_H int zfs_prop_get_feature(zfs_handle_t zhp, const char propname,
	char *buf, size_t len);
	_LIBZFS_H uint64_t getprop_uint64(zfs_handle_t , zfs_prop_t, char *);
	_LIBZFS_H uint64_t zfs_prop_get_int(zfs_handle_t *, zfs_prop_t);
	_LIBZFS_H int zfs_prop_inherit(zfs_handle_t , const char , boolean_t);
	_LIBZFS_H const char *zfs_prop_values(zfs_prop_t);
	_LIBZFS_H int zfs_prop_is_string(zfs_prop_t prop);
	_LIBZFS_H nvlist_t zfs_get_all_props(zfs_handle_t );
	_LIBZFS_H nvlist_t zfs_get_user_props(zfs_handle_t );
	_LIBZFS_H nvlist_t zfs_get_recvd_props(zfs_handle_t );
	_LIBZFS_H nvlist_t zfs_get_clones_nvl(zfs_handle_t );

	_LIBZFS_H int zfs_wait_status(zfs_handle_t *, zfs_wait_activity_t,
	boolean_t , boolean_t );

	/*
	* zfs encryption management
	*/
	_LIBZFS_H int zfs_crypto_get_encryption_root(zfs_handle_t , boolean_t ,
	char *);
	_LIBZFS_H int zfs_crypto_create(libzfs_handle_t , char , nvlist_t *,
	nvlist_t , boolean_t stdin_available, uint8_t , uint_t );
	_LIBZFS_H int zfs_crypto_clone_check(libzfs_handle_t , zfs_handle_t , char *,
	nvlist_t *);
	_LIBZFS_H int zfs_crypto_attempt_load_keys(libzfs_handle_t , char );
	_LIBZFS_H int zfs_crypto_load_key(zfs_handle_t , boolean_t, char );
	_LIBZFS_H int zfs_crypto_unload_key(zfs_handle_t *);
	_LIBZFS_H int zfs_crypto_rewrap(zfs_handle_t , nvlist_t , boolean_t);

	typedef struct zprop_list {
	int pl_prop;
	char *pl_user_prop;
	struct zprop_list *pl_next;
	boolean_t pl_all;
	size_t pl_width;
	size_t pl_recvd_width;
	boolean_t pl_fixed;
	} zprop_list_t;

	_LIBZFS_H int zfs_expand_proplist(zfs_handle_t , zprop_list_t *, boolean_t,
	boolean_t);
	_LIBZFS_H void zfs_prune_proplist(zfs_handle_t , uint8_t );
	+_LIBZFS_H int vdev_expand_proplist(zpool_handle_t , const char ,
	+ zprop_list_t **);

	#define ZFS_MOUNTPOINT_NONE "none"
	#define ZFS_MOUNTPOINT_LEGACY "legacy"

	#define ZFS_FEATURE_DISABLED "disabled"
	#define ZFS_FEATURE_ENABLED "enabled"
	#define ZFS_FEATURE_ACTIVE "active"

	#define ZFS_UNSUPPORTED_INACTIVE "inactive"
	#define ZFS_UNSUPPORTED_READONLY "readonly"

	/*
	* zpool property management
	*/
	_LIBZFS_H int zpool_expand_proplist(zpool_handle_t , zprop_list_t *,
	- boolean_t);
	+ zfs_type_t, boolean_t);
	_LIBZFS_H int zpool_prop_get_feature(zpool_handle_t , const char , char *,
	size_t);
	_LIBZFS_H const char *zpool_prop_default_string(zpool_prop_t);
	_LIBZFS_H uint64_t zpool_prop_default_numeric(zpool_prop_t);
	_LIBZFS_H const char *zpool_prop_column_name(zpool_prop_t);
	_LIBZFS_H boolean_t zpool_prop_align_right(zpool_prop_t);

	/*
	* Functions shared by zfs and zpool property management.
	*/
	_LIBZFS_H int zprop_iter(zprop_func func, void *cb, boolean_t show_all,
	boolean_t ordered, zfs_type_t type);
	_LIBZFS_H int zprop_get_list(libzfs_handle_t , char , zprop_list_t **,
	zfs_type_t);
	_LIBZFS_H void zprop_free_list(zprop_list_t *);

	#define ZFS_GET_NCOLS 5

	typedef enum {
	GET_COL_NONE,
	GET_COL_NAME,
	GET_COL_PROPERTY,
	GET_COL_VALUE,
	GET_COL_RECVD,
	GET_COL_SOURCE
	} zfs_get_column_t;

	/*
	* Functions for printing zfs or zpool properties
	*/
	+typedef struct vdev_cbdata {
	+ int cb_name_flags;
	+ char **cb_names;
	+ unsigned int cb_names_count;
	+} vdev_cbdata_t;
	+
	typedef struct zprop_get_cbdata {
	int cb_sources;
	zfs_get_column_t cb_columns[ZFS_GET_NCOLS];
	int cb_colwidths[ZFS_GET_NCOLS + 1];
	boolean_t cb_scripted;
	boolean_t cb_literal;
	boolean_t cb_first;
	zprop_list_t *cb_proplist;
	zfs_type_t cb_type;
	+ vdev_cbdata_t cb_vdevs;
	} zprop_get_cbdata_t;

	_LIBZFS_H void zprop_print_one_property(const char , zprop_get_cbdata_t ,
	const char , const char , zprop_source_t, const char *,
	const char *);

	/*
	* Iterator functions.
	*/
	typedef int (zfs_iter_f)(zfs_handle_t , void *);
	_LIBZFS_H int zfs_iter_root(libzfs_handle_t , zfs_iter_f, void );
	_LIBZFS_H int zfs_iter_children(zfs_handle_t , zfs_iter_f, void );
	_LIBZFS_H int zfs_iter_dependents(zfs_handle_t *, boolean_t, zfs_iter_f,
	void *);
	_LIBZFS_H int zfs_iter_filesystems(zfs_handle_t , zfs_iter_f, void );
	_LIBZFS_H int zfs_iter_snapshots(zfs_handle_t , boolean_t, zfs_iter_f, void ,
	uint64_t, uint64_t);
	_LIBZFS_H int zfs_iter_snapshots_sorted(zfs_handle_t , zfs_iter_f, void ,
	uint64_t, uint64_t);
	_LIBZFS_H int zfs_iter_snapspec(zfs_handle_t , const char , zfs_iter_f,
	void *);
	_LIBZFS_H int zfs_iter_bookmarks(zfs_handle_t , zfs_iter_f, void );
	_LIBZFS_H int zfs_iter_mounted(zfs_handle_t , zfs_iter_f, void );

	typedef struct get_all_cb {
	zfs_handle_t **cb_handles;
	size_t cb_alloc;
	size_t cb_used;
	} get_all_cb_t;

	_LIBZFS_H void zfs_foreach_mountpoint(libzfs_handle_t , zfs_handle_t *,
	size_t, zfs_iter_f, void *, boolean_t);
	_LIBZFS_H void libzfs_add_handle(get_all_cb_t , zfs_handle_t );

	/*
	* Functions to create and destroy datasets.
	*/
	_LIBZFS_H int zfs_create(libzfs_handle_t , const char , zfs_type_t,
	nvlist_t *);
	_LIBZFS_H int zfs_create_ancestors(libzfs_handle_t , const char );
	_LIBZFS_H int zfs_destroy(zfs_handle_t *, boolean_t);
	_LIBZFS_H int zfs_destroy_snaps(zfs_handle_t , char , boolean_t);
	_LIBZFS_H int zfs_destroy_snaps_nvl(libzfs_handle_t , nvlist_t , boolean_t);
	_LIBZFS_H int zfs_destroy_snaps_nvl_os(libzfs_handle_t , nvlist_t );
	_LIBZFS_H int zfs_clone(zfs_handle_t , const char , nvlist_t *);
	_LIBZFS_H int zfs_snapshot(libzfs_handle_t , const char , boolean_t,
	nvlist_t *);
	_LIBZFS_H int zfs_snapshot_nvl(libzfs_handle_t hdl, nvlist_t snaps,
	nvlist_t *props);
	_LIBZFS_H int zfs_rollback(zfs_handle_t , zfs_handle_t , boolean_t);

	typedef struct renameflags {
	/* recursive rename */
	int recursive : 1;

	/* don't unmount file systems */
	int nounmount : 1;

	/* force unmount file systems */
	int forceunmount : 1;
	} renameflags_t;

	_LIBZFS_H int zfs_rename(zfs_handle_t , const char , renameflags_t);

	typedef struct sendflags {
	/* Amount of extra information to print. */
	int verbosity;

	/* recursive send (ie, -R) */
	boolean_t replicate;

	/* for recursive send, skip sending missing snapshots */
	boolean_t skipmissing;

	/* for incrementals, do all intermediate snapshots */
	boolean_t doall;

	/* if dataset is a clone, do incremental from its origin */
	boolean_t fromorigin;

	/* field no longer used, maintained for backwards compatibility */
	boolean_t pad;

	/* send properties (ie, -p) */
	boolean_t props;

	/* do not send (no-op, ie. -n) */
	boolean_t dryrun;

	/* parsable verbose output (ie. -P) */
	boolean_t parsable;

	/* show progress (ie. -v) */
	boolean_t progress;

	/* large blocks (>128K) are permitted */
	boolean_t largeblock;

	/* WRITE_EMBEDDED records of type DATA are permitted */
	boolean_t embed_data;

	/* compressed WRITE records are permitted */
	boolean_t compress;

	/* raw encrypted records are permitted */
	boolean_t raw;

	/* only send received properties (ie. -b) */
	boolean_t backup;

	/* include snapshot holds in send stream */
	boolean_t holds;

	/* stream represents a partially received dataset */
	boolean_t saved;
	} sendflags_t;

	typedef boolean_t (snapfilter_cb_t)(zfs_handle_t , void );

	_LIBZFS_H int zfs_send(zfs_handle_t , const char , const char *,
	sendflags_t , int, snapfilter_cb_t, void , nvlist_t **);
	_LIBZFS_H int zfs_send_one(zfs_handle_t , const char , int, sendflags_t *,
	const char *);
	_LIBZFS_H int zfs_send_progress(zfs_handle_t , int, uint64_t , uint64_t *);
	_LIBZFS_H int zfs_send_resume(libzfs_handle_t , sendflags_t , int outfd,
	const char *);
	_LIBZFS_H int zfs_send_saved(zfs_handle_t , sendflags_t , int, const char *);
	_LIBZFS_H nvlist_t zfs_send_resume_token_to_nvlist(libzfs_handle_t hdl,
	const char *token);

	_LIBZFS_H int zfs_promote(zfs_handle_t *);
	_LIBZFS_H int zfs_hold(zfs_handle_t , const char , const char *,
	boolean_t, int);
	_LIBZFS_H int zfs_hold_nvl(zfs_handle_t , int, nvlist_t );
	_LIBZFS_H int zfs_release(zfs_handle_t , const char , const char *,
	boolean_t);
	_LIBZFS_H int zfs_get_holds(zfs_handle_t , nvlist_t *);
	_LIBZFS_H uint64_t zvol_volsize_to_reservation(zpool_handle_t *, uint64_t,
	nvlist_t *);

	typedef int (zfs_userspace_cb_t)(void arg, const char *domain,
	uid_t rid, uint64_t space);

	_LIBZFS_H int zfs_userspace(zfs_handle_t *, zfs_userquota_prop_t,
	zfs_userspace_cb_t, void *);

	_LIBZFS_H int zfs_get_fsacl(zfs_handle_t , nvlist_t *);
	_LIBZFS_H int zfs_set_fsacl(zfs_handle_t , boolean_t, nvlist_t );

	typedef struct recvflags {
	/* print informational messages (ie, -v was specified) */
	boolean_t verbose;

	/* the destination is a prefix, not the exact fs (ie, -d) */
	boolean_t isprefix;

	/*
	* Only the tail of the sent snapshot path is appended to the
	* destination to determine the received snapshot name (ie, -e).
	*/
	boolean_t istail;

	/* do not actually do the recv, just check if it would work (ie, -n) */
	boolean_t dryrun;

	/* rollback/destroy filesystems as necessary (eg, -F) */
	boolean_t force;

	/* set "canmount=off" on all modified filesystems */
	boolean_t canmountoff;

	/*
	* Mark the file systems as "resumable" and do not destroy them if the
	* receive is interrupted
	*/
	boolean_t resumable;

	/* byteswap flag is used internally; callers need not specify */
	boolean_t byteswap;

	/* do not mount file systems as they are extracted (private) */
	boolean_t nomount;

	/* Was holds flag set in the compound header? */
	boolean_t holds;

	/* skip receive of snapshot holds */
	boolean_t skipholds;

	/* mount the filesystem unless nomount is specified */
	boolean_t domount;

	/* force unmount while recv snapshot (private) */
	boolean_t forceunmount;
	} recvflags_t;

	_LIBZFS_H int zfs_receive(libzfs_handle_t , const char , nvlist_t *,
	recvflags_t , int, avl_tree_t );

	typedef enum diff_flags {
	- ZFS_DIFF_PARSEABLE = 0x1,
	- ZFS_DIFF_TIMESTAMP = 0x2,
	- ZFS_DIFF_CLASSIFY = 0x4
	+ ZFS_DIFF_PARSEABLE = 1 << 0,
	+ ZFS_DIFF_TIMESTAMP = 1 << 1,
	+ ZFS_DIFF_CLASSIFY = 1 << 2,
	+ ZFS_DIFF_NO_MANGLE = 1 << 3
	} diff_flags_t;

	_LIBZFS_H int zfs_show_diffs(zfs_handle_t , int, const char , const char *,
	int);

	/*
	* Miscellaneous functions.
	*/
	_LIBZFS_H const char *zfs_type_to_name(zfs_type_t);
	_LIBZFS_H void zfs_refresh_properties(zfs_handle_t *);
	_LIBZFS_H int zfs_name_valid(const char *, zfs_type_t);
	_LIBZFS_H zfs_handle_t zfs_path_to_zhandle(libzfs_handle_t , const char *,
	zfs_type_t);
	_LIBZFS_H int zfs_parent_name(zfs_handle_t , char , size_t);
	_LIBZFS_H boolean_t zfs_dataset_exists(libzfs_handle_t , const char ,
	zfs_type_t);
	_LIBZFS_H int zfs_spa_version(zfs_handle_t , int );
	_LIBZFS_H boolean_t zfs_bookmark_exists(const char *path);

	/*
	* Mount support functions.
	*/
	_LIBZFS_H boolean_t is_mounted(libzfs_handle_t , const char special, char **);
	_LIBZFS_H boolean_t zfs_is_mounted(zfs_handle_t , char *);
	_LIBZFS_H int zfs_mount(zfs_handle_t , const char , int);
	_LIBZFS_H int zfs_mount_at(zfs_handle_t , const char , int, const char *);
	_LIBZFS_H int zfs_unmount(zfs_handle_t , const char , int);
	_LIBZFS_H int zfs_unmountall(zfs_handle_t *, int);
	_LIBZFS_H int zfs_mount_delegation_check(void);

	#if defined(__linux__) \|\| defined(__APPLE__)
	_LIBZFS_H int zfs_parse_mount_options(char mntopts, unsigned long mntflags,
	unsigned long zfsflags, int sloppy, char badopt, char *mtabopt);
	_LIBZFS_H void zfs_adjust_mount_options(zfs_handle_t zhp, const char mntpoint,
	char mntopts, char mtabopt);
	#endif

	/*
	* Share support functions.
	*/
	_LIBZFS_H boolean_t zfs_is_shared(zfs_handle_t *);
	_LIBZFS_H int zfs_share(zfs_handle_t *);
	_LIBZFS_H int zfs_unshare(zfs_handle_t *);

	/*
	* Protocol-specific share support functions.
	*/
	_LIBZFS_H boolean_t zfs_is_shared_nfs(zfs_handle_t , char *);
	_LIBZFS_H boolean_t zfs_is_shared_smb(zfs_handle_t , char *);
	_LIBZFS_H int zfs_share_nfs(zfs_handle_t *);
	_LIBZFS_H int zfs_share_smb(zfs_handle_t *);
	_LIBZFS_H int zfs_shareall(zfs_handle_t *);
	_LIBZFS_H int zfs_unshare_nfs(zfs_handle_t , const char );
	_LIBZFS_H int zfs_unshare_smb(zfs_handle_t , const char );
	_LIBZFS_H int zfs_unshareall_nfs(zfs_handle_t *);
	_LIBZFS_H int zfs_unshareall_smb(zfs_handle_t *);
	_LIBZFS_H int zfs_unshareall_bypath(zfs_handle_t , const char );
	_LIBZFS_H int zfs_unshareall_bytype(zfs_handle_t , const char , const char *);
	_LIBZFS_H int zfs_unshareall(zfs_handle_t *);
	_LIBZFS_H int zfs_deleg_share_nfs(libzfs_handle_t , char , char , char ,
	void , void , int, zfs_share_op_t);
	_LIBZFS_H void zfs_commit_nfs_shares(void);
	_LIBZFS_H void zfs_commit_smb_shares(void);
	_LIBZFS_H void zfs_commit_all_shares(void);
	_LIBZFS_H void zfs_commit_shares(const char *);

	_LIBZFS_H int zfs_nicestrtonum(libzfs_handle_t , const char , uint64_t *);

	/*
	- * Utility functions to run an external process.
	+ * Utility functions to run an _LIBZFS_Hal process.
	*/
	#define STDOUT_VERBOSE 0x01
	#define STDERR_VERBOSE 0x02
	#define NO_DEFAULT_PATH 0x04 /* Don't use $PATH to lookup the command */

	_LIBZFS_H int libzfs_run_process(const char , char *, int);
	_LIBZFS_H int libzfs_run_process_get_stdout(const char , char [], char *[],
	char *[], int );
	_LIBZFS_H int libzfs_run_process_get_stdout_nopath(const char , char [],
	char [], char [], int );

	_LIBZFS_H void libzfs_free_str_array(char **, int);

	_LIBZFS_H int libzfs_envvar_is_set(char *);

	/*
	* Utility functions for zfs version
	*/
	_LIBZFS_H void zfs_version_userland(char *, int);
	_LIBZFS_H int zfs_version_kernel(char *, int);
	_LIBZFS_H int zfs_version_print(void);

	/*
	* Given a device or file, determine if it is part of a pool.
	*/
	_LIBZFS_H int zpool_in_use(libzfs_handle_t , int, pool_state_t , char **,
	boolean_t *);

	/*
	* Label manipulation.
	*/
	_LIBZFS_H int zpool_clear_label(int);
	_LIBZFS_H int zpool_set_bootenv(zpool_handle_t , const nvlist_t );
	_LIBZFS_H int zpool_get_bootenv(zpool_handle_t , nvlist_t *);

	/*
	* Management interfaces for SMB ACL files
	*/

	_LIBZFS_H int zfs_smb_acl_add(libzfs_handle_t , char , char , char );
	_LIBZFS_H int zfs_smb_acl_remove(libzfs_handle_t , char , char , char );
	_LIBZFS_H int zfs_smb_acl_purge(libzfs_handle_t , char , char *);
	_LIBZFS_H int zfs_smb_acl_rename(libzfs_handle_t , char , char , char ,
	char *);

	/*
	* Enable and disable datasets within a pool by mounting/unmounting and
	* sharing/unsharing them.
	*/
	_LIBZFS_H int zpool_enable_datasets(zpool_handle_t , const char , int);
	_LIBZFS_H int zpool_disable_datasets(zpool_handle_t *, boolean_t);
	_LIBZFS_H void zpool_disable_datasets_os(zpool_handle_t *, boolean_t);
	_LIBZFS_H void zpool_disable_volume_os(const char *);

	/*
	* Parse a features file for -o compatibility
	*/
	typedef enum {
	ZPOOL_COMPATIBILITY_OK,
	ZPOOL_COMPATIBILITY_WARNTOKEN,
	ZPOOL_COMPATIBILITY_BADTOKEN,
	ZPOOL_COMPATIBILITY_BADFILE,
	ZPOOL_COMPATIBILITY_NOFILES
	} zpool_compat_status_t;

	_LIBZFS_H zpool_compat_status_t zpool_load_compat(const char *,
	boolean_t , char , size_t);

	#ifdef __FreeBSD__

	/*
	* Attach/detach the given filesystem to/from the given jail.
	*/
	_LIBZFS_H int zfs_jail(zfs_handle_t *zhp, int jailid, int attach);

	/*
	* Set loader options for next boot.
	*/
	_LIBZFS_H int zpool_nextboot(libzfs_handle_t *, uint64_t, uint64_t,
	const char *);

	#endif /* __FreeBSD__ */

	#ifdef __cplusplus
	}
	#endif

	#endif /* _LIBZFS_H */
	diff --git a/sys/contrib/openzfs/include/libzfs_core.h b/sys/contrib/openzfs/include/libzfs_core.h
	index 9020d70db397..7acc03fc71bb 100644
	--- a/sys/contrib/openzfs/include/libzfs_core.h
	+++ b/sys/contrib/openzfs/include/libzfs_core.h
	@@ -1,153 +1,157 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2012, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2017 Datto Inc.
	* Copyright 2017 RackTop Systems.
	* Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
	*/

	#ifndef _LIBZFS_CORE_H
	#define _LIBZFS_CORE_H extern __attribute__((visibility("default")))

	#include <libnvpair.h>
	#include <sys/param.h>
	#include <sys/types.h>
	#include <sys/fs/zfs.h>

	#ifdef __cplusplus
	extern "C" {
	#endif

	_LIBZFS_CORE_H int libzfs_core_init(void);
	_LIBZFS_CORE_H void libzfs_core_fini(void);

	struct zfs_cmd;
	_LIBZFS_CORE_H int lzc_ioctl_fd(int, unsigned long, struct zfs_cmd *);

	/*
	* NB: this type should be kept binary-compatible with dmu_objset_type_t.
	*/
	enum lzc_dataset_type {
	LZC_DATSET_TYPE_ZFS = 2,
	LZC_DATSET_TYPE_ZVOL
	};

	_LIBZFS_CORE_H int lzc_snapshot(nvlist_t , nvlist_t , nvlist_t **);
	_LIBZFS_CORE_H int lzc_create(const char , enum lzc_dataset_type, nvlist_t ,
	uint8_t *, uint_t);
	_LIBZFS_CORE_H int lzc_clone(const char , const char , nvlist_t *);
	_LIBZFS_CORE_H int lzc_promote(const char , char , int);
	_LIBZFS_CORE_H int lzc_destroy_snaps(nvlist_t , boolean_t, nvlist_t *);
	_LIBZFS_CORE_H int lzc_bookmark(nvlist_t , nvlist_t *);
	_LIBZFS_CORE_H int lzc_get_bookmarks(const char , nvlist_t , nvlist_t **);
	_LIBZFS_CORE_H int lzc_get_bookmark_props(const char , nvlist_t *);
	_LIBZFS_CORE_H int lzc_destroy_bookmarks(nvlist_t , nvlist_t *);
	_LIBZFS_CORE_H int lzc_load_key(const char , boolean_t, uint8_t , uint_t);
	_LIBZFS_CORE_H int lzc_unload_key(const char *);
	_LIBZFS_CORE_H int lzc_change_key(const char , uint64_t, nvlist_t , uint8_t *,
	uint_t);
	_LIBZFS_CORE_H int lzc_initialize(const char *, pool_initialize_func_t,
	nvlist_t , nvlist_t *);
	_LIBZFS_CORE_H int lzc_trim(const char *, pool_trim_func_t, uint64_t, boolean_t,
	nvlist_t , nvlist_t *);
	_LIBZFS_CORE_H int lzc_redact(const char , const char , nvlist_t *);

	_LIBZFS_CORE_H int lzc_snaprange_space(const char , const char , uint64_t *);

	_LIBZFS_CORE_H int lzc_hold(nvlist_t , int, nvlist_t *);
	_LIBZFS_CORE_H int lzc_release(nvlist_t , nvlist_t *);
	_LIBZFS_CORE_H int lzc_get_holds(const char , nvlist_t *);

	enum lzc_send_flags {
	LZC_SEND_FLAG_EMBED_DATA = 1 << 0,
	LZC_SEND_FLAG_LARGE_BLOCK = 1 << 1,
	LZC_SEND_FLAG_COMPRESS = 1 << 2,
	LZC_SEND_FLAG_RAW = 1 << 3,
	LZC_SEND_FLAG_SAVED = 1 << 4,
	};

	_LIBZFS_CORE_H int lzc_send(const char , const char , int,
	enum lzc_send_flags);
	_LIBZFS_CORE_H int lzc_send_resume(const char , const char , int,
	enum lzc_send_flags, uint64_t, uint64_t);
	_LIBZFS_CORE_H int lzc_send_space(const char , const char ,
	enum lzc_send_flags, uint64_t *);

	struct dmu_replay_record;

	_LIBZFS_CORE_H int lzc_send_redacted(const char , const char , int,
	enum lzc_send_flags, const char *);
	_LIBZFS_CORE_H int lzc_send_resume_redacted(const char , const char , int,
	enum lzc_send_flags, uint64_t, uint64_t, const char *);
	_LIBZFS_CORE_H int lzc_receive(const char , nvlist_t , const char *,
	boolean_t, boolean_t, int);
	_LIBZFS_CORE_H int lzc_receive_resumable(const char , nvlist_t , const char *,
	boolean_t, boolean_t, int);
	_LIBZFS_CORE_H int lzc_receive_with_header(const char , nvlist_t ,
	const char *, boolean_t, boolean_t, boolean_t, int,
	const struct dmu_replay_record *);
	_LIBZFS_CORE_H int lzc_receive_one(const char , nvlist_t , const char *,
	boolean_t, boolean_t, boolean_t, int, const struct dmu_replay_record *, int,
	uint64_t , uint64_t , uint64_t , nvlist_t *);
	_LIBZFS_CORE_H int lzc_receive_with_cmdprops(const char , nvlist_t ,
	nvlist_t , uint8_t , uint_t, const char *, boolean_t, boolean_t,
	boolean_t, int, const struct dmu_replay_record , int, uint64_t ,
	uint64_t , uint64_t , nvlist_t **);
	_LIBZFS_CORE_H int lzc_send_space(const char , const char ,
	enum lzc_send_flags, uint64_t *);
	_LIBZFS_CORE_H int lzc_send_space_resume_redacted(const char , const char ,
	enum lzc_send_flags, uint64_t, uint64_t, uint64_t, const char *,
	int, uint64_t *);
	_LIBZFS_CORE_H uint64_t lzc_send_progress(int);

	_LIBZFS_CORE_H boolean_t lzc_exists(const char *);

	_LIBZFS_CORE_H int lzc_rollback(const char , char , int);
	_LIBZFS_CORE_H int lzc_rollback_to(const char , const char );

	_LIBZFS_CORE_H int lzc_rename(const char , const char );
	_LIBZFS_CORE_H int lzc_destroy(const char *);

	_LIBZFS_CORE_H int lzc_channel_program(const char , const char , uint64_t,
	uint64_t, nvlist_t , nvlist_t *);
	_LIBZFS_CORE_H int lzc_channel_program_nosync(const char , const char ,
	uint64_t, uint64_t, nvlist_t , nvlist_t *);

	_LIBZFS_CORE_H int lzc_sync(const char , nvlist_t , nvlist_t **);
	_LIBZFS_CORE_H int lzc_reopen(const char *, boolean_t);

	_LIBZFS_CORE_H int lzc_pool_checkpoint(const char *);
	_LIBZFS_CORE_H int lzc_pool_checkpoint_discard(const char *);

	_LIBZFS_CORE_H int lzc_wait(const char , zpool_wait_activity_t, boolean_t );
	_LIBZFS_CORE_H int lzc_wait_tag(const char *, zpool_wait_activity_t, uint64_t,
	boolean_t *);
	_LIBZFS_CORE_H int lzc_wait_fs(const char , zfs_wait_activity_t, boolean_t );

	_LIBZFS_CORE_H int lzc_set_bootenv(const char , const nvlist_t );
	_LIBZFS_CORE_H int lzc_get_bootenv(const char , nvlist_t *);
	+
	+_LIBZFS_CORE_H int lzc_get_vdev_prop(const char , nvlist_t , nvlist_t **);
	+_LIBZFS_CORE_H int lzc_set_vdev_prop(const char , nvlist_t , nvlist_t **);
	+
	#ifdef __cplusplus
	}
	#endif

	#endif /* _LIBZFS_CORE_H */
	diff --git a/sys/contrib/openzfs/include/os/linux/kernel/linux/simd_x86.h b/sys/contrib/openzfs/include/os/linux/kernel/linux/simd_x86.h
	index cdd3286d2147..f2ae0fcbc21a 100644
	--- a/sys/contrib/openzfs/include/os/linux/kernel/linux/simd_x86.h
	+++ b/sys/contrib/openzfs/include/os/linux/kernel/linux/simd_x86.h
	@@ -1,646 +1,649 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (C) 2016 Gvozden Neskovic <neskovic@compeng.uni-frankfurt.de>.
	*/

	/*
	* USER API:
	*
	* Kernel fpu methods:
	* kfpu_allowed()
	* kfpu_begin()
	* kfpu_end()
	* kfpu_init()
	* kfpu_fini()
	*
	* SIMD support:
	*
	* Following functions should be called to determine whether CPU feature
	* is supported. All functions are usable in kernel and user space.
	* If a SIMD algorithm is using more than one instruction set
	* all relevant feature test functions should be called.
	*
	* Supported features:
	* zfs_sse_available()
	* zfs_sse2_available()
	* zfs_sse3_available()
	* zfs_ssse3_available()
	* zfs_sse4_1_available()
	* zfs_sse4_2_available()
	*
	* zfs_avx_available()
	* zfs_avx2_available()
	*
	* zfs_bmi1_available()
	* zfs_bmi2_available()
	*
	* zfs_avx512f_available()
	* zfs_avx512cd_available()
	* zfs_avx512er_available()
	* zfs_avx512pf_available()
	* zfs_avx512bw_available()
	* zfs_avx512dq_available()
	* zfs_avx512vl_available()
	* zfs_avx512ifma_available()
	* zfs_avx512vbmi_available()
	*
	* NOTE(AVX-512VL): If using AVX-512 instructions with 128Bit registers
	* also add zfs_avx512vl_available() to feature check.
	*/

	#ifndef _LINUX_SIMD_X86_H
	#define _LINUX_SIMD_X86_H

	/* only for __x86 */
	#if defined(__x86)

	#include <sys/types.h>
	#include <asm/cpufeature.h>

	/*
	* Disable the WARN_ON_FPU() macro to prevent additional dependencies
	* when providing the kfpu_* functions. Relevant warnings are included
	* as appropriate and are unconditionally enabled.
	*/
	#if defined(CONFIG_X86_DEBUG_FPU) && !defined(KERNEL_EXPORTS_X86_FPU)
	#undef CONFIG_X86_DEBUG_FPU
	#endif

	#if defined(HAVE_KERNEL_FPU_API_HEADER)
	#include <asm/fpu/api.h>
	#include <asm/fpu/internal.h>
	+#if defined(HAVE_KERNEL_FPU_XCR_HEADER)
	+#include <asm/fpu/xcr.h>
	+#endif
	#else
	#include <asm/i387.h>
	#include <asm/xcr.h>
	#endif

	/*
	* The following cases are for kernels which export either the
	* kernel_fpu_* or __kernel_fpu_* functions.
	*/
	#if defined(KERNEL_EXPORTS_X86_FPU)

	#define kfpu_allowed() 1
	#define kfpu_init() 0
	#define kfpu_fini() ((void) 0)

	#if defined(HAVE_UNDERSCORE_KERNEL_FPU)
	#define kfpu_begin() \
	{ \
	preempt_disable(); \
	__kernel_fpu_begin(); \
	}
	#define kfpu_end() \
	{ \
	__kernel_fpu_end(); \
	preempt_enable(); \
	}

	#elif defined(HAVE_KERNEL_FPU)
	#define kfpu_begin() kernel_fpu_begin()
	#define kfpu_end() kernel_fpu_end()

	#else
	/*
	* This case is unreachable. When KERNEL_EXPORTS_X86_FPU is defined then
	* either HAVE_UNDERSCORE_KERNEL_FPU or HAVE_KERNEL_FPU must be defined.
	*/
	#error "Unreachable kernel configuration"
	#endif

	#else /* defined(KERNEL_EXPORTS_X86_FPU) */

	/*
	* When the kernel_fpu_* symbols are unavailable then provide our own
	* versions which allow the FPU to be safely used.
	*/
	#if defined(HAVE_KERNEL_FPU_INTERNAL)

	#include <linux/mm.h>

	extern union fpregs_state **zfs_kfpu_fpregs;

	/*
	* Initialize per-cpu variables to store FPU state.
	*/
	static inline void
	kfpu_fini(void)
	{
	int cpu;

	for_each_possible_cpu(cpu) {
	if (zfs_kfpu_fpregs[cpu] != NULL) {
	free_pages((unsigned long)zfs_kfpu_fpregs[cpu],
	get_order(sizeof (union fpregs_state)));
	}
	}

	kfree(zfs_kfpu_fpregs);
	}

	static inline int
	kfpu_init(void)
	{
	zfs_kfpu_fpregs = kzalloc(num_possible_cpus() *
	sizeof (union fpregs_state *), GFP_KERNEL);
	if (zfs_kfpu_fpregs == NULL)
	return (-ENOMEM);

	/*
	* The fxsave and xsave operations require 16-/64-byte alignment of
	* the target memory. Since kmalloc() provides no alignment
	* guarantee instead use alloc_pages_node().
	*/
	unsigned int order = get_order(sizeof (union fpregs_state));
	int cpu;

	for_each_possible_cpu(cpu) {
	struct page *page = alloc_pages_node(cpu_to_node(cpu),
	GFP_KERNEL \| __GFP_ZERO, order);
	if (page == NULL) {
	kfpu_fini();
	return (-ENOMEM);
	}

	zfs_kfpu_fpregs[cpu] = page_address(page);
	}

	return (0);
	}

	#define kfpu_allowed() 1
	#define ex_handler_fprestore ex_handler_default

	/*
	* FPU save and restore instructions.
	*/
	#define __asm __asm__ __volatile__
	#define kfpu_fxsave(addr) __asm("fxsave %0" : "=m" (*(addr)))
	#define kfpu_fxsaveq(addr) __asm("fxsaveq %0" : "=m" (*(addr)))
	#define kfpu_fnsave(addr) __asm("fnsave %0; fwait" : "=m" (*(addr)))
	#define kfpu_fxrstor(addr) __asm("fxrstor %0" : : "m" (*(addr)))
	#define kfpu_fxrstorq(addr) __asm("fxrstorq %0" : : "m" (*(addr)))
	#define kfpu_frstor(addr) __asm("frstor %0" : : "m" (*(addr)))
	#define kfpu_fxsr_clean(rval) __asm("fnclex; emms; fildl %P[addr]" \
	: : [addr] "m" (rval));

	static inline void
	kfpu_save_xsave(struct xregs_state *addr, uint64_t mask)
	{
	uint32_t low, hi;
	int err;

	low = mask;
	hi = mask >> 32;
	XSTATE_XSAVE(addr, low, hi, err);
	WARN_ON_ONCE(err);
	}

	static inline void
	kfpu_save_fxsr(struct fxregs_state *addr)
	{
	if (IS_ENABLED(CONFIG_X86_32))
	kfpu_fxsave(addr);
	else
	kfpu_fxsaveq(addr);
	}

	static inline void
	kfpu_save_fsave(struct fregs_state *addr)
	{
	kfpu_fnsave(addr);
	}

	static inline void
	kfpu_begin(void)
	{
	/*
	* Preemption and interrupts must be disabled for the critical
	* region where the FPU state is being modified.
	*/
	preempt_disable();
	local_irq_disable();

	/*
	* The current FPU registers need to be preserved by kfpu_begin()
	* and restored by kfpu_end(). They are stored in a dedicated
	* per-cpu variable, not in the task struct, this allows any user
	* FPU state to be correctly preserved and restored.
	*/
	union fpregs_state *state = zfs_kfpu_fpregs[smp_processor_id()];

	if (static_cpu_has(X86_FEATURE_XSAVE)) {
	kfpu_save_xsave(&state->xsave, ~0);
	} else if (static_cpu_has(X86_FEATURE_FXSR)) {
	kfpu_save_fxsr(&state->fxsave);
	} else {
	kfpu_save_fsave(&state->fsave);
	}
	}

	static inline void
	kfpu_restore_xsave(struct xregs_state *addr, uint64_t mask)
	{
	uint32_t low, hi;

	low = mask;
	hi = mask >> 32;
	XSTATE_XRESTORE(addr, low, hi);
	}

	static inline void
	kfpu_restore_fxsr(struct fxregs_state *addr)
	{
	/*
	* On AuthenticAMD K7 and K8 processors the fxrstor instruction only
	* restores the _x87 FOP, FIP, and FDP registers when an exception
	* is pending. Clean the _x87 state to force the restore.
	*/
	if (unlikely(static_cpu_has_bug(X86_BUG_FXSAVE_LEAK)))
	kfpu_fxsr_clean(addr);

	if (IS_ENABLED(CONFIG_X86_32)) {
	kfpu_fxrstor(addr);
	} else {
	kfpu_fxrstorq(addr);
	}
	}

	static inline void
	kfpu_restore_fsave(struct fregs_state *addr)
	{
	kfpu_frstor(addr);
	}

	static inline void
	kfpu_end(void)
	{
	union fpregs_state *state = zfs_kfpu_fpregs[smp_processor_id()];

	if (static_cpu_has(X86_FEATURE_XSAVE)) {
	kfpu_restore_xsave(&state->xsave, ~0);
	} else if (static_cpu_has(X86_FEATURE_FXSR)) {
	kfpu_restore_fxsr(&state->fxsave);
	} else {
	kfpu_restore_fsave(&state->fsave);
	}

	local_irq_enable();
	preempt_enable();
	}

	#else

	/*
	* FPU support is unavailable.
	*/
	#define kfpu_allowed() 0
	#define kfpu_begin() do {} while (0)
	#define kfpu_end() do {} while (0)
	#define kfpu_init() 0
	#define kfpu_fini() ((void) 0)

	#endif /* defined(HAVE_KERNEL_FPU_INTERNAL) */
	#endif /* defined(KERNEL_EXPORTS_X86_FPU) */

	/*
	* Linux kernel provides an interface for CPU feature testing.
	*/

	/*
	* Detect register set support
	*/
	static inline boolean_t
	__simd_state_enabled(const uint64_t state)
	{
	boolean_t has_osxsave;
	uint64_t xcr0;

	#if defined(X86_FEATURE_OSXSAVE)
	has_osxsave = !!boot_cpu_has(X86_FEATURE_OSXSAVE);
	#else
	has_osxsave = B_FALSE;
	#endif
	if (!has_osxsave)
	return (B_FALSE);

	xcr0 = xgetbv(0);
	return ((xcr0 & state) == state);
	}

	#define _XSTATE_SSE_AVX (0x2 \| 0x4)
	#define _XSTATE_AVX512 (0xE0 \| _XSTATE_SSE_AVX)

	#define __ymm_enabled() __simd_state_enabled(_XSTATE_SSE_AVX)
	#define __zmm_enabled() __simd_state_enabled(_XSTATE_AVX512)

	/*
	* Check if SSE instruction set is available
	*/
	static inline boolean_t
	zfs_sse_available(void)
	{
	return (!!boot_cpu_has(X86_FEATURE_XMM));
	}

	/*
	* Check if SSE2 instruction set is available
	*/
	static inline boolean_t
	zfs_sse2_available(void)
	{
	return (!!boot_cpu_has(X86_FEATURE_XMM2));
	}

	/*
	* Check if SSE3 instruction set is available
	*/
	static inline boolean_t
	zfs_sse3_available(void)
	{
	return (!!boot_cpu_has(X86_FEATURE_XMM3));
	}

	/*
	* Check if SSSE3 instruction set is available
	*/
	static inline boolean_t
	zfs_ssse3_available(void)
	{
	return (!!boot_cpu_has(X86_FEATURE_SSSE3));
	}

	/*
	* Check if SSE4.1 instruction set is available
	*/
	static inline boolean_t
	zfs_sse4_1_available(void)
	{
	return (!!boot_cpu_has(X86_FEATURE_XMM4_1));
	}

	/*
	* Check if SSE4.2 instruction set is available
	*/
	static inline boolean_t
	zfs_sse4_2_available(void)
	{
	return (!!boot_cpu_has(X86_FEATURE_XMM4_2));
	}

	/*
	* Check if AVX instruction set is available
	*/
	static inline boolean_t
	zfs_avx_available(void)
	{
	return (boot_cpu_has(X86_FEATURE_AVX) && __ymm_enabled());
	}

	/*
	* Check if AVX2 instruction set is available
	*/
	static inline boolean_t
	zfs_avx2_available(void)
	{
	return (boot_cpu_has(X86_FEATURE_AVX2) && __ymm_enabled());
	}

	/*
	* Check if BMI1 instruction set is available
	*/
	static inline boolean_t
	zfs_bmi1_available(void)
	{
	#if defined(X86_FEATURE_BMI1)
	return (!!boot_cpu_has(X86_FEATURE_BMI1));
	#else
	return (B_FALSE);
	#endif
	}

	/*
	* Check if BMI2 instruction set is available
	*/
	static inline boolean_t
	zfs_bmi2_available(void)
	{
	#if defined(X86_FEATURE_BMI2)
	return (!!boot_cpu_has(X86_FEATURE_BMI2));
	#else
	return (B_FALSE);
	#endif
	}

	/*
	* Check if AES instruction set is available
	*/
	static inline boolean_t
	zfs_aes_available(void)
	{
	#if defined(X86_FEATURE_AES)
	return (!!boot_cpu_has(X86_FEATURE_AES));
	#else
	return (B_FALSE);
	#endif
	}

	/*
	* Check if PCLMULQDQ instruction set is available
	*/
	static inline boolean_t
	zfs_pclmulqdq_available(void)
	{
	#if defined(X86_FEATURE_PCLMULQDQ)
	return (!!boot_cpu_has(X86_FEATURE_PCLMULQDQ));
	#else
	return (B_FALSE);
	#endif
	}

	/*
	* Check if MOVBE instruction is available
	*/
	static inline boolean_t
	zfs_movbe_available(void)
	{
	#if defined(X86_FEATURE_MOVBE)
	return (!!boot_cpu_has(X86_FEATURE_MOVBE));
	#else
	return (B_FALSE);
	#endif
	}

	/*
	* AVX-512 family of instruction sets:
	*
	* AVX512F Foundation
	* AVX512CD Conflict Detection Instructions
	* AVX512ER Exponential and Reciprocal Instructions
	* AVX512PF Prefetch Instructions
	*
	* AVX512BW Byte and Word Instructions
	* AVX512DQ Double-word and Quadword Instructions
	* AVX512VL Vector Length Extensions
	*
	* AVX512IFMA Integer Fused Multiply Add (Not supported by kernel 4.4)
	* AVX512VBMI Vector Byte Manipulation Instructions
	*/

	/*
	* Check if AVX512F instruction set is available
	*/
	static inline boolean_t
	zfs_avx512f_available(void)
	{
	boolean_t has_avx512 = B_FALSE;

	#if defined(X86_FEATURE_AVX512F)
	has_avx512 = !!boot_cpu_has(X86_FEATURE_AVX512F);
	#endif
	return (has_avx512 && __zmm_enabled());
	}

	/*
	* Check if AVX512CD instruction set is available
	*/
	static inline boolean_t
	zfs_avx512cd_available(void)
	{
	boolean_t has_avx512 = B_FALSE;

	#if defined(X86_FEATURE_AVX512CD)
	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
	boot_cpu_has(X86_FEATURE_AVX512CD);
	#endif
	return (has_avx512 && __zmm_enabled());
	}

	/*
	* Check if AVX512ER instruction set is available
	*/
	static inline boolean_t
	zfs_avx512er_available(void)
	{
	boolean_t has_avx512 = B_FALSE;

	#if defined(X86_FEATURE_AVX512ER)
	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
	boot_cpu_has(X86_FEATURE_AVX512ER);
	#endif
	return (has_avx512 && __zmm_enabled());
	}

	/*
	* Check if AVX512PF instruction set is available
	*/
	static inline boolean_t
	zfs_avx512pf_available(void)
	{
	boolean_t has_avx512 = B_FALSE;

	#if defined(X86_FEATURE_AVX512PF)
	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
	boot_cpu_has(X86_FEATURE_AVX512PF);
	#endif
	return (has_avx512 && __zmm_enabled());
	}

	/*
	* Check if AVX512BW instruction set is available
	*/
	static inline boolean_t
	zfs_avx512bw_available(void)
	{
	boolean_t has_avx512 = B_FALSE;

	#if defined(X86_FEATURE_AVX512BW)
	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
	boot_cpu_has(X86_FEATURE_AVX512BW);
	#endif

	return (has_avx512 && __zmm_enabled());
	}

	/*
	* Check if AVX512DQ instruction set is available
	*/
	static inline boolean_t
	zfs_avx512dq_available(void)
	{
	boolean_t has_avx512 = B_FALSE;

	#if defined(X86_FEATURE_AVX512DQ)
	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
	boot_cpu_has(X86_FEATURE_AVX512DQ);
	#endif
	return (has_avx512 && __zmm_enabled());
	}

	/*
	* Check if AVX512VL instruction set is available
	*/
	static inline boolean_t
	zfs_avx512vl_available(void)
	{
	boolean_t has_avx512 = B_FALSE;

	#if defined(X86_FEATURE_AVX512VL)
	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
	boot_cpu_has(X86_FEATURE_AVX512VL);
	#endif
	return (has_avx512 && __zmm_enabled());
	}

	/*
	* Check if AVX512IFMA instruction set is available
	*/
	static inline boolean_t
	zfs_avx512ifma_available(void)
	{
	boolean_t has_avx512 = B_FALSE;

	#if defined(X86_FEATURE_AVX512IFMA)
	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
	boot_cpu_has(X86_FEATURE_AVX512IFMA);
	#endif
	return (has_avx512 && __zmm_enabled());
	}

	/*
	* Check if AVX512VBMI instruction set is available
	*/
	static inline boolean_t
	zfs_avx512vbmi_available(void)
	{
	boolean_t has_avx512 = B_FALSE;

	#if defined(X86_FEATURE_AVX512VBMI)
	has_avx512 = boot_cpu_has(X86_FEATURE_AVX512F) &&
	boot_cpu_has(X86_FEATURE_AVX512VBMI);
	#endif
	return (has_avx512 && __zmm_enabled());
	}

	#endif /* defined(__x86) */

	#endif /* _LINUX_SIMD_X86_H */
	diff --git a/sys/contrib/openzfs/include/sys/fs/zfs.h b/sys/contrib/openzfs/include/sys/fs/zfs.h
	index 2af11fc7196d..287b3beae91e 100644
	--- a/sys/contrib/openzfs/include/sys/fs/zfs.h
	+++ b/sys/contrib/openzfs/include/sys/fs/zfs.h
	@@ -1,1669 +1,1757 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2013, 2017 Joyent, Inc. All rights reserved.
	* Copyright (c) 2014 Integros [integros.com]
	* Copyright (c) 2017, Intel Corporation.
	* Copyright (c) 2019 Datto Inc.
	* Portions Copyright 2010 Robert Milkowski
	* Copyright (c) 2021, Colm Buckley <colm@tuatha.org>
	*/

	#ifndef _SYS_FS_ZFS_H
	#define _SYS_FS_ZFS_H extern __attribute__((visibility("default")))

	#include <sys/time.h>
	#include <sys/zio_priority.h>

	#ifdef __cplusplus
	extern "C" {
	#endif

	/*
	* Types and constants shared between userland and the kernel.
	*/

	/*
	* Each dataset can be one of the following types. These constants can be
	* combined into masks that can be passed to various functions.
	*/
	typedef enum {
	ZFS_TYPE_FILESYSTEM = (1 << 0),
	ZFS_TYPE_SNAPSHOT = (1 << 1),
	ZFS_TYPE_VOLUME = (1 << 2),
	ZFS_TYPE_POOL = (1 << 3),
	- ZFS_TYPE_BOOKMARK = (1 << 4)
	+ ZFS_TYPE_BOOKMARK = (1 << 4),
	+ ZFS_TYPE_VDEV = (1 << 5),
	} zfs_type_t;

	/*
	* NB: lzc_dataset_type should be updated whenever a new objset type is added,
	* if it represents a real type of a dataset that can be created from userland.
	*/
	typedef enum dmu_objset_type {
	DMU_OST_NONE,
	DMU_OST_META,
	DMU_OST_ZFS,
	DMU_OST_ZVOL,
	DMU_OST_OTHER, /* For testing only! */
	DMU_OST_ANY, /* Be careful! */
	DMU_OST_NUMTYPES
	} dmu_objset_type_t;

	#define ZFS_TYPE_DATASET \
	(ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME \| ZFS_TYPE_SNAPSHOT)

	/*
	* All of these include the terminating NUL byte.
	*/
	#define ZAP_MAXNAMELEN 256
	#define ZAP_MAXVALUELEN (1024 * 8)
	#define ZAP_OLDMAXVALUELEN 1024
	#define ZFS_MAX_DATASET_NAME_LEN 256

	/*
	* Dataset properties are identified by these constants and must be added to
	* the end of this list to ensure that external consumers are not affected
	* by the change. If you make any changes to this list, be sure to update
	* the property table in module/zcommon/zfs_prop.c.
	*/
	typedef enum {
	ZPROP_CONT = -2,
	ZPROP_INVAL = -1,
	ZFS_PROP_TYPE = 0,
	ZFS_PROP_CREATION,
	ZFS_PROP_USED,
	ZFS_PROP_AVAILABLE,
	ZFS_PROP_REFERENCED,
	ZFS_PROP_COMPRESSRATIO,
	ZFS_PROP_MOUNTED,
	ZFS_PROP_ORIGIN,
	ZFS_PROP_QUOTA,
	ZFS_PROP_RESERVATION,
	ZFS_PROP_VOLSIZE,
	ZFS_PROP_VOLBLOCKSIZE,
	ZFS_PROP_RECORDSIZE,
	ZFS_PROP_MOUNTPOINT,
	ZFS_PROP_SHARENFS,
	ZFS_PROP_CHECKSUM,
	ZFS_PROP_COMPRESSION,
	ZFS_PROP_ATIME,
	ZFS_PROP_DEVICES,
	ZFS_PROP_EXEC,
	ZFS_PROP_SETUID,
	ZFS_PROP_READONLY,
	ZFS_PROP_ZONED,
	ZFS_PROP_SNAPDIR,
	ZFS_PROP_ACLMODE,
	ZFS_PROP_ACLINHERIT,
	ZFS_PROP_CREATETXG,
	ZFS_PROP_NAME, /* not exposed to the user */
	ZFS_PROP_CANMOUNT,
	ZFS_PROP_ISCSIOPTIONS, /* not exposed to the user */
	ZFS_PROP_XATTR,
	ZFS_PROP_NUMCLONES, /* not exposed to the user */
	ZFS_PROP_COPIES,
	ZFS_PROP_VERSION,
	ZFS_PROP_UTF8ONLY,
	ZFS_PROP_NORMALIZE,
	ZFS_PROP_CASE,
	ZFS_PROP_VSCAN,
	ZFS_PROP_NBMAND,
	ZFS_PROP_SHARESMB,
	ZFS_PROP_REFQUOTA,
	ZFS_PROP_REFRESERVATION,
	ZFS_PROP_GUID,
	ZFS_PROP_PRIMARYCACHE,
	ZFS_PROP_SECONDARYCACHE,
	ZFS_PROP_USEDSNAP,
	ZFS_PROP_USEDDS,
	ZFS_PROP_USEDCHILD,
	ZFS_PROP_USEDREFRESERV,
	ZFS_PROP_USERACCOUNTING, /* not exposed to the user */
	ZFS_PROP_STMF_SHAREINFO, /* not exposed to the user */
	ZFS_PROP_DEFER_DESTROY,
	ZFS_PROP_USERREFS,
	ZFS_PROP_LOGBIAS,
	ZFS_PROP_UNIQUE, /* not exposed to the user */
	ZFS_PROP_OBJSETID,
	ZFS_PROP_DEDUP,
	ZFS_PROP_MLSLABEL,
	ZFS_PROP_SYNC,
	ZFS_PROP_DNODESIZE,
	ZFS_PROP_REFRATIO,
	ZFS_PROP_WRITTEN,
	ZFS_PROP_CLONES,
	ZFS_PROP_LOGICALUSED,
	ZFS_PROP_LOGICALREFERENCED,
	ZFS_PROP_INCONSISTENT, /* not exposed to the user */
	ZFS_PROP_VOLMODE,
	ZFS_PROP_FILESYSTEM_LIMIT,
	ZFS_PROP_SNAPSHOT_LIMIT,
	ZFS_PROP_FILESYSTEM_COUNT,
	ZFS_PROP_SNAPSHOT_COUNT,
	ZFS_PROP_SNAPDEV,
	ZFS_PROP_ACLTYPE,
	ZFS_PROP_SELINUX_CONTEXT,
	ZFS_PROP_SELINUX_FSCONTEXT,
	ZFS_PROP_SELINUX_DEFCONTEXT,
	ZFS_PROP_SELINUX_ROOTCONTEXT,
	ZFS_PROP_RELATIME,
	ZFS_PROP_REDUNDANT_METADATA,
	ZFS_PROP_OVERLAY,
	ZFS_PROP_PREV_SNAP,
	ZFS_PROP_RECEIVE_RESUME_TOKEN,
	ZFS_PROP_ENCRYPTION,
	ZFS_PROP_KEYLOCATION,
	ZFS_PROP_KEYFORMAT,
	ZFS_PROP_PBKDF2_SALT,
	ZFS_PROP_PBKDF2_ITERS,
	ZFS_PROP_ENCRYPTION_ROOT,
	ZFS_PROP_KEY_GUID,
	ZFS_PROP_KEYSTATUS,
	ZFS_PROP_REMAPTXG, /* obsolete - no longer used */
	ZFS_PROP_SPECIAL_SMALL_BLOCKS,
	ZFS_PROP_IVSET_GUID, /* not exposed to the user */
	ZFS_PROP_REDACTED,
	ZFS_PROP_REDACT_SNAPS,
	ZFS_NUM_PROPS
	} zfs_prop_t;

	typedef enum {
	ZFS_PROP_USERUSED,
	ZFS_PROP_USERQUOTA,
	ZFS_PROP_GROUPUSED,
	ZFS_PROP_GROUPQUOTA,
	ZFS_PROP_USEROBJUSED,
	ZFS_PROP_USEROBJQUOTA,
	ZFS_PROP_GROUPOBJUSED,
	ZFS_PROP_GROUPOBJQUOTA,
	ZFS_PROP_PROJECTUSED,
	ZFS_PROP_PROJECTQUOTA,
	ZFS_PROP_PROJECTOBJUSED,
	ZFS_PROP_PROJECTOBJQUOTA,
	ZFS_NUM_USERQUOTA_PROPS
	} zfs_userquota_prop_t;

	_SYS_FS_ZFS_H const char *zfs_userquota_prop_prefixes[ZFS_NUM_USERQUOTA_PROPS];

	/*
	* Pool properties are identified by these constants and must be added to the
	* end of this list to ensure that external consumers are not affected
	* by the change. Properties must be registered in zfs_prop_init().
	*/
	typedef enum {
	ZPOOL_PROP_INVAL = -1,
	ZPOOL_PROP_NAME,
	ZPOOL_PROP_SIZE,
	ZPOOL_PROP_CAPACITY,
	ZPOOL_PROP_ALTROOT,
	ZPOOL_PROP_HEALTH,
	ZPOOL_PROP_GUID,
	ZPOOL_PROP_VERSION,
	ZPOOL_PROP_BOOTFS,
	ZPOOL_PROP_DELEGATION,
	ZPOOL_PROP_AUTOREPLACE,
	ZPOOL_PROP_CACHEFILE,
	ZPOOL_PROP_FAILUREMODE,
	ZPOOL_PROP_LISTSNAPS,
	ZPOOL_PROP_AUTOEXPAND,
	ZPOOL_PROP_DEDUPDITTO,
	ZPOOL_PROP_DEDUPRATIO,
	ZPOOL_PROP_FREE,
	ZPOOL_PROP_ALLOCATED,
	ZPOOL_PROP_READONLY,
	ZPOOL_PROP_ASHIFT,
	ZPOOL_PROP_COMMENT,
	ZPOOL_PROP_EXPANDSZ,
	ZPOOL_PROP_FREEING,
	ZPOOL_PROP_FRAGMENTATION,
	ZPOOL_PROP_LEAKED,
	ZPOOL_PROP_MAXBLOCKSIZE,
	ZPOOL_PROP_TNAME,
	ZPOOL_PROP_MAXDNODESIZE,
	ZPOOL_PROP_MULTIHOST,
	ZPOOL_PROP_CHECKPOINT,
	ZPOOL_PROP_LOAD_GUID,
	ZPOOL_PROP_AUTOTRIM,
	ZPOOL_PROP_COMPATIBILITY,
	ZPOOL_NUM_PROPS
	} zpool_prop_t;

	/* Small enough to not hog a whole line of printout in zpool(8). */
	#define ZPROP_MAX_COMMENT 32
	+#define ZPROP_BOOLEAN_NA 2

	#define ZPROP_VALUE "value"
	#define ZPROP_SOURCE "source"

	typedef enum {
	ZPROP_SRC_NONE = 0x1,
	ZPROP_SRC_DEFAULT = 0x2,
	ZPROP_SRC_TEMPORARY = 0x4,
	ZPROP_SRC_LOCAL = 0x8,
	ZPROP_SRC_INHERITED = 0x10,
	ZPROP_SRC_RECEIVED = 0x20
	} zprop_source_t;

	#define ZPROP_SRC_ALL 0x3f

	#define ZPROP_SOURCE_VAL_RECVD "$recvd"
	#define ZPROP_N_MORE_ERRORS "N_MORE_ERRORS"

	/*
	* Dataset flag implemented as a special entry in the props zap object
	* indicating that the dataset has received properties on or after
	* SPA_VERSION_RECVD_PROPS. The first such receive blows away local properties
	* just as it did in earlier versions, and thereafter, local properties are
	* preserved.
	*/
	#define ZPROP_HAS_RECVD "$hasrecvd"

	typedef enum {
	ZPROP_ERR_NOCLEAR = 0x1, /* failure to clear existing props */
	ZPROP_ERR_NORESTORE = 0x2 /* failure to restore props on error */
	} zprop_errflags_t;

	typedef int (zprop_func)(int, void );

	/*
	* Properties to be set on the root file system of a new pool
	* are stuffed into their own nvlist, which is then included in
	* the properties nvlist with the pool properties.
	*/
	#define ZPOOL_ROOTFS_PROPS "root-props-nvl"

	/*
	* Length of 'written@' and 'written#'
	*/
	#define ZFS_WRITTEN_PROP_PREFIX_LEN 8

	+/*
	+ * VDEV properties are identified by these constants and must be added to the
	+ * end of this list to ensure that external consumers are not affected
	+ * by the change. If you make any changes to this list, be sure to update
	+ * the property table in usr/src/common/zfs/zpool_prop.c.
	+ */
	+typedef enum {
	+ VDEV_PROP_INVAL = -1,
	+#define VDEV_PROP_USER VDEV_PROP_INVAL
	+ VDEV_PROP_NAME,
	+ VDEV_PROP_CAPACITY,
	+ VDEV_PROP_STATE,
	+ VDEV_PROP_GUID,
	+ VDEV_PROP_ASIZE,
	+ VDEV_PROP_PSIZE,
	+ VDEV_PROP_ASHIFT,
	+ VDEV_PROP_SIZE,
	+ VDEV_PROP_FREE,
	+ VDEV_PROP_ALLOCATED,
	+ VDEV_PROP_COMMENT,
	+ VDEV_PROP_EXPANDSZ,
	+ VDEV_PROP_FRAGMENTATION,
	+ VDEV_PROP_BOOTSIZE,
	+ VDEV_PROP_PARITY,
	+ VDEV_PROP_PATH,
	+ VDEV_PROP_DEVID,
	+ VDEV_PROP_PHYS_PATH,
	+ VDEV_PROP_ENC_PATH,
	+ VDEV_PROP_FRU,
	+ VDEV_PROP_PARENT,
	+ VDEV_PROP_CHILDREN,
	+ VDEV_PROP_NUMCHILDREN,
	+ VDEV_PROP_READ_ERRORS,
	+ VDEV_PROP_WRITE_ERRORS,
	+ VDEV_PROP_CHECKSUM_ERRORS,
	+ VDEV_PROP_INITIALIZE_ERRORS,
	+ VDEV_PROP_OPS_NULL,
	+ VDEV_PROP_OPS_READ,
	+ VDEV_PROP_OPS_WRITE,
	+ VDEV_PROP_OPS_FREE,
	+ VDEV_PROP_OPS_CLAIM,
	+ VDEV_PROP_OPS_TRIM,
	+ VDEV_PROP_BYTES_NULL,
	+ VDEV_PROP_BYTES_READ,
	+ VDEV_PROP_BYTES_WRITE,
	+ VDEV_PROP_BYTES_FREE,
	+ VDEV_PROP_BYTES_CLAIM,
	+ VDEV_PROP_BYTES_TRIM,
	+ VDEV_PROP_REMOVING,
	+ VDEV_PROP_ALLOCATING,
	+ VDEV_NUM_PROPS
	+} vdev_prop_t;
	+
	/*
	* Dataset property functions shared between libzfs and kernel.
	*/
	_SYS_FS_ZFS_H const char *zfs_prop_default_string(zfs_prop_t);
	_SYS_FS_ZFS_H uint64_t zfs_prop_default_numeric(zfs_prop_t);
	_SYS_FS_ZFS_H boolean_t zfs_prop_readonly(zfs_prop_t);
	_SYS_FS_ZFS_H boolean_t zfs_prop_visible(zfs_prop_t prop);
	_SYS_FS_ZFS_H boolean_t zfs_prop_inheritable(zfs_prop_t);
	_SYS_FS_ZFS_H boolean_t zfs_prop_setonce(zfs_prop_t);
	_SYS_FS_ZFS_H boolean_t zfs_prop_encryption_key_param(zfs_prop_t);
	_SYS_FS_ZFS_H boolean_t zfs_prop_valid_keylocation(const char *, boolean_t);
	_SYS_FS_ZFS_H const char *zfs_prop_to_name(zfs_prop_t);
	_SYS_FS_ZFS_H zfs_prop_t zfs_name_to_prop(const char *);
	_SYS_FS_ZFS_H boolean_t zfs_prop_user(const char *);
	_SYS_FS_ZFS_H boolean_t zfs_prop_userquota(const char *);
	_SYS_FS_ZFS_H boolean_t zfs_prop_written(const char *);
	_SYS_FS_ZFS_H int zfs_prop_index_to_string(zfs_prop_t, uint64_t, const char **);
	_SYS_FS_ZFS_H int zfs_prop_string_to_index(zfs_prop_t, const char *,
	uint64_t *);
	_SYS_FS_ZFS_H uint64_t zfs_prop_random_value(zfs_prop_t, uint64_t seed);
	_SYS_FS_ZFS_H boolean_t zfs_prop_valid_for_type(int, zfs_type_t, boolean_t);

	/*
	* Pool property functions shared between libzfs and kernel.
	*/
	_SYS_FS_ZFS_H zpool_prop_t zpool_name_to_prop(const char *);
	_SYS_FS_ZFS_H const char *zpool_prop_to_name(zpool_prop_t);
	_SYS_FS_ZFS_H const char *zpool_prop_default_string(zpool_prop_t);
	_SYS_FS_ZFS_H uint64_t zpool_prop_default_numeric(zpool_prop_t);
	_SYS_FS_ZFS_H boolean_t zpool_prop_readonly(zpool_prop_t);
	_SYS_FS_ZFS_H boolean_t zpool_prop_setonce(zpool_prop_t);
	_SYS_FS_ZFS_H boolean_t zpool_prop_feature(const char *);
	_SYS_FS_ZFS_H boolean_t zpool_prop_unsupported(const char *);
	_SYS_FS_ZFS_H int zpool_prop_index_to_string(zpool_prop_t, uint64_t,
	const char **);
	_SYS_FS_ZFS_H int zpool_prop_string_to_index(zpool_prop_t, const char *,
	uint64_t *);
	_SYS_FS_ZFS_H uint64_t zpool_prop_random_value(zpool_prop_t, uint64_t seed);

	+/*
	+ * VDEV property functions shared between libzfs and kernel.
	+ */
	+_SYS_FS_ZFS_H vdev_prop_t vdev_name_to_prop(const char *);
	+_SYS_FS_ZFS_H boolean_t vdev_prop_user(const char *name);
	+_SYS_FS_ZFS_H const char *vdev_prop_to_name(vdev_prop_t);
	+_SYS_FS_ZFS_H const char *vdev_prop_default_string(vdev_prop_t);
	+_SYS_FS_ZFS_H uint64_t vdev_prop_default_numeric(vdev_prop_t);
	+_SYS_FS_ZFS_H boolean_t vdev_prop_readonly(vdev_prop_t prop);
	+_SYS_FS_ZFS_H int vdev_prop_index_to_string(vdev_prop_t, uint64_t,
	+ const char **);
	+_SYS_FS_ZFS_H int vdev_prop_string_to_index(vdev_prop_t, const char *,
	+ uint64_t *);
	+_SYS_FS_ZFS_H boolean_t zpool_prop_vdev(const char *name);
	+_SYS_FS_ZFS_H uint64_t vdev_prop_random_value(vdev_prop_t prop, uint64_t seed);
	+
	/*
	* Definitions for the Delegation.
	*/
	typedef enum {
	ZFS_DELEG_WHO_UNKNOWN = 0,
	ZFS_DELEG_USER = 'u',
	ZFS_DELEG_USER_SETS = 'U',
	ZFS_DELEG_GROUP = 'g',
	ZFS_DELEG_GROUP_SETS = 'G',
	ZFS_DELEG_EVERYONE = 'e',
	ZFS_DELEG_EVERYONE_SETS = 'E',
	ZFS_DELEG_CREATE = 'c',
	ZFS_DELEG_CREATE_SETS = 'C',
	ZFS_DELEG_NAMED_SET = 's',
	ZFS_DELEG_NAMED_SET_SETS = 'S'
	} zfs_deleg_who_type_t;

	typedef enum {
	ZFS_DELEG_NONE = 0,
	ZFS_DELEG_PERM_LOCAL = 1,
	ZFS_DELEG_PERM_DESCENDENT = 2,
	ZFS_DELEG_PERM_LOCALDESCENDENT = 3,
	ZFS_DELEG_PERM_CREATE = 4
	} zfs_deleg_inherit_t;

	#define ZFS_DELEG_PERM_UID "uid"
	#define ZFS_DELEG_PERM_GID "gid"
	#define ZFS_DELEG_PERM_GROUPS "groups"

	#define ZFS_MLSLABEL_DEFAULT "none"

	#define ZFS_SMB_ACL_SRC "src"
	#define ZFS_SMB_ACL_TARGET "target"

	typedef enum {
	ZFS_CANMOUNT_OFF = 0,
	ZFS_CANMOUNT_ON = 1,
	ZFS_CANMOUNT_NOAUTO = 2
	} zfs_canmount_type_t;

	typedef enum {
	ZFS_LOGBIAS_LATENCY = 0,
	ZFS_LOGBIAS_THROUGHPUT = 1
	} zfs_logbias_op_t;

	typedef enum zfs_share_op {
	ZFS_SHARE_NFS = 0,
	ZFS_UNSHARE_NFS = 1,
	ZFS_SHARE_SMB = 2,
	ZFS_UNSHARE_SMB = 3
	} zfs_share_op_t;

	typedef enum zfs_smb_acl_op {
	ZFS_SMB_ACL_ADD,
	ZFS_SMB_ACL_REMOVE,
	ZFS_SMB_ACL_RENAME,
	ZFS_SMB_ACL_PURGE
	} zfs_smb_acl_op_t;

	typedef enum zfs_cache_type {
	ZFS_CACHE_NONE = 0,
	ZFS_CACHE_METADATA = 1,
	ZFS_CACHE_ALL = 2
	} zfs_cache_type_t;

	typedef enum {
	ZFS_SYNC_STANDARD = 0,
	ZFS_SYNC_ALWAYS = 1,
	ZFS_SYNC_DISABLED = 2
	} zfs_sync_type_t;

	typedef enum {
	ZFS_XATTR_OFF = 0,
	ZFS_XATTR_DIR = 1,
	ZFS_XATTR_SA = 2
	} zfs_xattr_type_t;

	typedef enum {
	ZFS_DNSIZE_LEGACY = 0,
	ZFS_DNSIZE_AUTO = 1,
	ZFS_DNSIZE_1K = 1024,
	ZFS_DNSIZE_2K = 2048,
	ZFS_DNSIZE_4K = 4096,
	ZFS_DNSIZE_8K = 8192,
	ZFS_DNSIZE_16K = 16384
	} zfs_dnsize_type_t;

	typedef enum {
	ZFS_REDUNDANT_METADATA_ALL,
	ZFS_REDUNDANT_METADATA_MOST
	} zfs_redundant_metadata_type_t;

	typedef enum {
	ZFS_VOLMODE_DEFAULT = 0,
	ZFS_VOLMODE_GEOM = 1,
	ZFS_VOLMODE_DEV = 2,
	ZFS_VOLMODE_NONE = 3
	} zfs_volmode_t;

	typedef enum zfs_keystatus {
	ZFS_KEYSTATUS_NONE = 0,
	ZFS_KEYSTATUS_UNAVAILABLE,
	ZFS_KEYSTATUS_AVAILABLE,
	} zfs_keystatus_t;

	typedef enum zfs_keyformat {
	ZFS_KEYFORMAT_NONE = 0,
	ZFS_KEYFORMAT_RAW,
	ZFS_KEYFORMAT_HEX,
	ZFS_KEYFORMAT_PASSPHRASE,
	ZFS_KEYFORMAT_FORMATS
	} zfs_keyformat_t;

	typedef enum zfs_key_location {
	ZFS_KEYLOCATION_NONE = 0,
	ZFS_KEYLOCATION_PROMPT,
	ZFS_KEYLOCATION_URI,
	ZFS_KEYLOCATION_LOCATIONS
	} zfs_keylocation_t;

	#define DEFAULT_PBKDF2_ITERATIONS 350000
	#define MIN_PBKDF2_ITERATIONS 100000

	/*
	* On-disk version number.
	*/
	#define SPA_VERSION_1 1ULL
	#define SPA_VERSION_2 2ULL
	#define SPA_VERSION_3 3ULL
	#define SPA_VERSION_4 4ULL
	#define SPA_VERSION_5 5ULL
	#define SPA_VERSION_6 6ULL
	#define SPA_VERSION_7 7ULL
	#define SPA_VERSION_8 8ULL
	#define SPA_VERSION_9 9ULL
	#define SPA_VERSION_10 10ULL
	#define SPA_VERSION_11 11ULL
	#define SPA_VERSION_12 12ULL
	#define SPA_VERSION_13 13ULL
	#define SPA_VERSION_14 14ULL
	#define SPA_VERSION_15 15ULL
	#define SPA_VERSION_16 16ULL
	#define SPA_VERSION_17 17ULL
	#define SPA_VERSION_18 18ULL
	#define SPA_VERSION_19 19ULL
	#define SPA_VERSION_20 20ULL
	#define SPA_VERSION_21 21ULL
	#define SPA_VERSION_22 22ULL
	#define SPA_VERSION_23 23ULL
	#define SPA_VERSION_24 24ULL
	#define SPA_VERSION_25 25ULL
	#define SPA_VERSION_26 26ULL
	#define SPA_VERSION_27 27ULL
	#define SPA_VERSION_28 28ULL
	#define SPA_VERSION_5000 5000ULL

	/*
	* The incrementing pool version number has been replaced by pool feature
	* flags. For more details, see zfeature.c.
	*/
	#define SPA_VERSION SPA_VERSION_5000
	#define SPA_VERSION_STRING "5000"

	/*
	* Symbolic names for the changes that caused a SPA_VERSION switch.
	* Used in the code when checking for presence or absence of a feature.
	* Feel free to define multiple symbolic names for each version if there
	* were multiple changes to on-disk structures during that version.
	*
	* NOTE: When checking the current SPA_VERSION in your code, be sure
	* to use spa_version() since it reports the version of the
	* last synced uberblock. Checking the in-flight version can
	* be dangerous in some cases.
	*/
	#define SPA_VERSION_INITIAL SPA_VERSION_1
	#define SPA_VERSION_DITTO_BLOCKS SPA_VERSION_2
	#define SPA_VERSION_SPARES SPA_VERSION_3
	#define SPA_VERSION_RAIDZ2 SPA_VERSION_3
	#define SPA_VERSION_BPOBJ_ACCOUNT SPA_VERSION_3
	#define SPA_VERSION_RAIDZ_DEFLATE SPA_VERSION_3
	#define SPA_VERSION_DNODE_BYTES SPA_VERSION_3
	#define SPA_VERSION_ZPOOL_HISTORY SPA_VERSION_4
	#define SPA_VERSION_GZIP_COMPRESSION SPA_VERSION_5
	#define SPA_VERSION_BOOTFS SPA_VERSION_6
	#define SPA_VERSION_SLOGS SPA_VERSION_7
	#define SPA_VERSION_DELEGATED_PERMS SPA_VERSION_8
	#define SPA_VERSION_FUID SPA_VERSION_9
	#define SPA_VERSION_REFRESERVATION SPA_VERSION_9
	#define SPA_VERSION_REFQUOTA SPA_VERSION_9
	#define SPA_VERSION_UNIQUE_ACCURATE SPA_VERSION_9
	#define SPA_VERSION_L2CACHE SPA_VERSION_10
	#define SPA_VERSION_NEXT_CLONES SPA_VERSION_11
	#define SPA_VERSION_ORIGIN SPA_VERSION_11
	#define SPA_VERSION_DSL_SCRUB SPA_VERSION_11
	#define SPA_VERSION_SNAP_PROPS SPA_VERSION_12
	#define SPA_VERSION_USED_BREAKDOWN SPA_VERSION_13
	#define SPA_VERSION_PASSTHROUGH_X SPA_VERSION_14
	#define SPA_VERSION_USERSPACE SPA_VERSION_15
	#define SPA_VERSION_STMF_PROP SPA_VERSION_16
	#define SPA_VERSION_RAIDZ3 SPA_VERSION_17
	#define SPA_VERSION_USERREFS SPA_VERSION_18
	#define SPA_VERSION_HOLES SPA_VERSION_19
	#define SPA_VERSION_ZLE_COMPRESSION SPA_VERSION_20
	#define SPA_VERSION_DEDUP SPA_VERSION_21
	#define SPA_VERSION_RECVD_PROPS SPA_VERSION_22
	#define SPA_VERSION_SLIM_ZIL SPA_VERSION_23
	#define SPA_VERSION_SA SPA_VERSION_24
	#define SPA_VERSION_SCAN SPA_VERSION_25
	#define SPA_VERSION_DIR_CLONES SPA_VERSION_26
	#define SPA_VERSION_DEADLISTS SPA_VERSION_26
	#define SPA_VERSION_FAST_SNAP SPA_VERSION_27
	#define SPA_VERSION_MULTI_REPLACE SPA_VERSION_28
	#define SPA_VERSION_BEFORE_FEATURES SPA_VERSION_28
	#define SPA_VERSION_FEATURES SPA_VERSION_5000

	#define SPA_VERSION_IS_SUPPORTED(v) \
	(((v) >= SPA_VERSION_INITIAL && (v) <= SPA_VERSION_BEFORE_FEATURES) \|\| \
	((v) >= SPA_VERSION_FEATURES && (v) <= SPA_VERSION))

	/*
	* ZPL version - rev'd whenever an incompatible on-disk format change
	* occurs. This is independent of SPA/DMU/ZAP versioning. You must
	* also update the version_table[] and help message in zfs_prop.c.
	*/
	#define ZPL_VERSION_1 1ULL
	#define ZPL_VERSION_2 2ULL
	#define ZPL_VERSION_3 3ULL
	#define ZPL_VERSION_4 4ULL
	#define ZPL_VERSION_5 5ULL
	#define ZPL_VERSION ZPL_VERSION_5
	#define ZPL_VERSION_STRING "5"

	#define ZPL_VERSION_INITIAL ZPL_VERSION_1
	#define ZPL_VERSION_DIRENT_TYPE ZPL_VERSION_2
	#define ZPL_VERSION_FUID ZPL_VERSION_3
	#define ZPL_VERSION_NORMALIZATION ZPL_VERSION_3
	#define ZPL_VERSION_SYSATTR ZPL_VERSION_3
	#define ZPL_VERSION_USERSPACE ZPL_VERSION_4
	#define ZPL_VERSION_SA ZPL_VERSION_5

	/* Persistent L2ARC version */
	#define L2ARC_PERSISTENT_VERSION_1 1ULL
	#define L2ARC_PERSISTENT_VERSION L2ARC_PERSISTENT_VERSION_1
	#define L2ARC_PERSISTENT_VERSION_STRING "1"

	/* Rewind policy information */
	#define ZPOOL_NO_REWIND 1 /* No policy - default behavior */
	#define ZPOOL_NEVER_REWIND 2 /* Do not search for best txg or rewind */
	#define ZPOOL_TRY_REWIND 4 /* Search for best txg, but do not rewind */
	#define ZPOOL_DO_REWIND 8 /* Rewind to best txg w/in deferred frees */
	#define ZPOOL_EXTREME_REWIND 16 /* Allow extreme measures to find best txg */
	#define ZPOOL_REWIND_MASK 28 /* All the possible rewind bits */
	#define ZPOOL_REWIND_POLICIES 31 /* All the possible policy bits */

	typedef struct zpool_load_policy {
	uint32_t zlp_rewind; /* rewind policy requested */
	uint64_t zlp_maxmeta; /* max acceptable meta-data errors */
	uint64_t zlp_maxdata; /* max acceptable data errors */
	uint64_t zlp_txg; /* specific txg to load */
	} zpool_load_policy_t;

	/*
	* The following are configuration names used in the nvlist describing a pool's
	* configuration. New on-disk names should be prefixed with "<reversed-DNS>:"
	* (e.g. "org.openzfs:") to avoid conflicting names being developed
	* independently.
	*/
	#define ZPOOL_CONFIG_VERSION "version"
	#define ZPOOL_CONFIG_POOL_NAME "name"
	#define ZPOOL_CONFIG_POOL_STATE "state"
	#define ZPOOL_CONFIG_POOL_TXG "txg"
	#define ZPOOL_CONFIG_POOL_GUID "pool_guid"
	#define ZPOOL_CONFIG_CREATE_TXG "create_txg"
	#define ZPOOL_CONFIG_TOP_GUID "top_guid"
	#define ZPOOL_CONFIG_VDEV_TREE "vdev_tree"
	#define ZPOOL_CONFIG_TYPE "type"
	#define ZPOOL_CONFIG_CHILDREN "children"
	#define ZPOOL_CONFIG_ID "id"
	#define ZPOOL_CONFIG_GUID "guid"
	#define ZPOOL_CONFIG_INDIRECT_OBJECT "com.delphix:indirect_object"
	#define ZPOOL_CONFIG_INDIRECT_BIRTHS "com.delphix:indirect_births"
	#define ZPOOL_CONFIG_PREV_INDIRECT_VDEV "com.delphix:prev_indirect_vdev"
	#define ZPOOL_CONFIG_PATH "path"
	#define ZPOOL_CONFIG_DEVID "devid"
	#define ZPOOL_CONFIG_SPARE_ID "spareid"
	#define ZPOOL_CONFIG_METASLAB_ARRAY "metaslab_array"
	#define ZPOOL_CONFIG_METASLAB_SHIFT "metaslab_shift"
	#define ZPOOL_CONFIG_ASHIFT "ashift"
	#define ZPOOL_CONFIG_ASIZE "asize"
	#define ZPOOL_CONFIG_DTL "DTL"
	#define ZPOOL_CONFIG_SCAN_STATS "scan_stats" /* not stored on disk */
	#define ZPOOL_CONFIG_REMOVAL_STATS "removal_stats" /* not stored on disk */
	#define ZPOOL_CONFIG_CHECKPOINT_STATS "checkpoint_stats" /* not on disk */
	#define ZPOOL_CONFIG_VDEV_STATS "vdev_stats" /* not stored on disk */
	#define ZPOOL_CONFIG_INDIRECT_SIZE "indirect_size" /* not stored on disk */

	/* container nvlist of extended stats */
	#define ZPOOL_CONFIG_VDEV_STATS_EX "vdev_stats_ex"

	/* Active queue read/write stats */
	#define ZPOOL_CONFIG_VDEV_SYNC_R_ACTIVE_QUEUE "vdev_sync_r_active_queue"
	#define ZPOOL_CONFIG_VDEV_SYNC_W_ACTIVE_QUEUE "vdev_sync_w_active_queue"
	#define ZPOOL_CONFIG_VDEV_ASYNC_R_ACTIVE_QUEUE "vdev_async_r_active_queue"
	#define ZPOOL_CONFIG_VDEV_ASYNC_W_ACTIVE_QUEUE "vdev_async_w_active_queue"
	#define ZPOOL_CONFIG_VDEV_SCRUB_ACTIVE_QUEUE "vdev_async_scrub_active_queue"
	#define ZPOOL_CONFIG_VDEV_TRIM_ACTIVE_QUEUE "vdev_async_trim_active_queue"
	#define ZPOOL_CONFIG_VDEV_REBUILD_ACTIVE_QUEUE "vdev_rebuild_active_queue"

	/* Queue sizes */
	#define ZPOOL_CONFIG_VDEV_SYNC_R_PEND_QUEUE "vdev_sync_r_pend_queue"
	#define ZPOOL_CONFIG_VDEV_SYNC_W_PEND_QUEUE "vdev_sync_w_pend_queue"
	#define ZPOOL_CONFIG_VDEV_ASYNC_R_PEND_QUEUE "vdev_async_r_pend_queue"
	#define ZPOOL_CONFIG_VDEV_ASYNC_W_PEND_QUEUE "vdev_async_w_pend_queue"
	#define ZPOOL_CONFIG_VDEV_SCRUB_PEND_QUEUE "vdev_async_scrub_pend_queue"
	#define ZPOOL_CONFIG_VDEV_TRIM_PEND_QUEUE "vdev_async_trim_pend_queue"
	#define ZPOOL_CONFIG_VDEV_REBUILD_PEND_QUEUE "vdev_rebuild_pend_queue"

	/* Latency read/write histogram stats */
	#define ZPOOL_CONFIG_VDEV_TOT_R_LAT_HISTO "vdev_tot_r_lat_histo"
	#define ZPOOL_CONFIG_VDEV_TOT_W_LAT_HISTO "vdev_tot_w_lat_histo"
	#define ZPOOL_CONFIG_VDEV_DISK_R_LAT_HISTO "vdev_disk_r_lat_histo"
	#define ZPOOL_CONFIG_VDEV_DISK_W_LAT_HISTO "vdev_disk_w_lat_histo"
	#define ZPOOL_CONFIG_VDEV_SYNC_R_LAT_HISTO "vdev_sync_r_lat_histo"
	#define ZPOOL_CONFIG_VDEV_SYNC_W_LAT_HISTO "vdev_sync_w_lat_histo"
	#define ZPOOL_CONFIG_VDEV_ASYNC_R_LAT_HISTO "vdev_async_r_lat_histo"
	#define ZPOOL_CONFIG_VDEV_ASYNC_W_LAT_HISTO "vdev_async_w_lat_histo"
	#define ZPOOL_CONFIG_VDEV_SCRUB_LAT_HISTO "vdev_scrub_histo"
	#define ZPOOL_CONFIG_VDEV_TRIM_LAT_HISTO "vdev_trim_histo"
	#define ZPOOL_CONFIG_VDEV_REBUILD_LAT_HISTO "vdev_rebuild_histo"

	/* Request size histograms */
	#define ZPOOL_CONFIG_VDEV_SYNC_IND_R_HISTO "vdev_sync_ind_r_histo"
	#define ZPOOL_CONFIG_VDEV_SYNC_IND_W_HISTO "vdev_sync_ind_w_histo"
	#define ZPOOL_CONFIG_VDEV_ASYNC_IND_R_HISTO "vdev_async_ind_r_histo"
	#define ZPOOL_CONFIG_VDEV_ASYNC_IND_W_HISTO "vdev_async_ind_w_histo"
	#define ZPOOL_CONFIG_VDEV_IND_SCRUB_HISTO "vdev_ind_scrub_histo"
	#define ZPOOL_CONFIG_VDEV_IND_TRIM_HISTO "vdev_ind_trim_histo"
	#define ZPOOL_CONFIG_VDEV_IND_REBUILD_HISTO "vdev_ind_rebuild_histo"
	#define ZPOOL_CONFIG_VDEV_SYNC_AGG_R_HISTO "vdev_sync_agg_r_histo"
	#define ZPOOL_CONFIG_VDEV_SYNC_AGG_W_HISTO "vdev_sync_agg_w_histo"
	#define ZPOOL_CONFIG_VDEV_ASYNC_AGG_R_HISTO "vdev_async_agg_r_histo"
	#define ZPOOL_CONFIG_VDEV_ASYNC_AGG_W_HISTO "vdev_async_agg_w_histo"
	#define ZPOOL_CONFIG_VDEV_AGG_SCRUB_HISTO "vdev_agg_scrub_histo"
	#define ZPOOL_CONFIG_VDEV_AGG_TRIM_HISTO "vdev_agg_trim_histo"
	#define ZPOOL_CONFIG_VDEV_AGG_REBUILD_HISTO "vdev_agg_rebuild_histo"

	/* Number of slow IOs */
	#define ZPOOL_CONFIG_VDEV_SLOW_IOS "vdev_slow_ios"

	/* vdev enclosure sysfs path */
	#define ZPOOL_CONFIG_VDEV_ENC_SYSFS_PATH "vdev_enc_sysfs_path"

	#define ZPOOL_CONFIG_WHOLE_DISK "whole_disk"
	#define ZPOOL_CONFIG_ERRCOUNT "error_count"
	#define ZPOOL_CONFIG_NOT_PRESENT "not_present"
	#define ZPOOL_CONFIG_SPARES "spares"
	#define ZPOOL_CONFIG_IS_SPARE "is_spare"
	#define ZPOOL_CONFIG_NPARITY "nparity"
	#define ZPOOL_CONFIG_HOSTID "hostid"
	#define ZPOOL_CONFIG_HOSTNAME "hostname"
	#define ZPOOL_CONFIG_LOADED_TIME "initial_load_time"
	#define ZPOOL_CONFIG_UNSPARE "unspare"
	#define ZPOOL_CONFIG_PHYS_PATH "phys_path"
	#define ZPOOL_CONFIG_IS_LOG "is_log"
	#define ZPOOL_CONFIG_L2CACHE "l2cache"
	#define ZPOOL_CONFIG_HOLE_ARRAY "hole_array"
	#define ZPOOL_CONFIG_VDEV_CHILDREN "vdev_children"
	#define ZPOOL_CONFIG_IS_HOLE "is_hole"
	#define ZPOOL_CONFIG_DDT_HISTOGRAM "ddt_histogram"
	#define ZPOOL_CONFIG_DDT_OBJ_STATS "ddt_object_stats"
	#define ZPOOL_CONFIG_DDT_STATS "ddt_stats"
	#define ZPOOL_CONFIG_SPLIT "splitcfg"
	#define ZPOOL_CONFIG_ORIG_GUID "orig_guid"
	#define ZPOOL_CONFIG_SPLIT_GUID "split_guid"
	#define ZPOOL_CONFIG_SPLIT_LIST "guid_list"
	+#define ZPOOL_CONFIG_NONALLOCATING "non_allocating"
	#define ZPOOL_CONFIG_REMOVING "removing"
	#define ZPOOL_CONFIG_RESILVER_TXG "resilver_txg"
	#define ZPOOL_CONFIG_REBUILD_TXG "rebuild_txg"
	#define ZPOOL_CONFIG_COMMENT "comment"
	#define ZPOOL_CONFIG_SUSPENDED "suspended" /* not stored on disk */
	#define ZPOOL_CONFIG_SUSPENDED_REASON "suspended_reason" /* not stored */
	#define ZPOOL_CONFIG_TIMESTAMP "timestamp" /* not stored on disk */
	#define ZPOOL_CONFIG_BOOTFS "bootfs" /* not stored on disk */
	#define ZPOOL_CONFIG_MISSING_DEVICES "missing_vdevs" /* not stored on disk */
	#define ZPOOL_CONFIG_LOAD_INFO "load_info" /* not stored on disk */
	#define ZPOOL_CONFIG_REWIND_INFO "rewind_info" /* not stored on disk */
	#define ZPOOL_CONFIG_UNSUP_FEAT "unsup_feat" /* not stored on disk */
	#define ZPOOL_CONFIG_ENABLED_FEAT "enabled_feat" /* not stored on disk */
	#define ZPOOL_CONFIG_CAN_RDONLY "can_rdonly" /* not stored on disk */
	#define ZPOOL_CONFIG_FEATURES_FOR_READ "features_for_read"
	#define ZPOOL_CONFIG_FEATURE_STATS "feature_stats" /* not stored on disk */
	#define ZPOOL_CONFIG_ERRATA "errata" /* not stored on disk */
	#define ZPOOL_CONFIG_VDEV_TOP_ZAP "com.delphix:vdev_zap_top"
	#define ZPOOL_CONFIG_VDEV_LEAF_ZAP "com.delphix:vdev_zap_leaf"
	#define ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS "com.delphix:has_per_vdev_zaps"
	#define ZPOOL_CONFIG_RESILVER_DEFER "com.datto:resilver_defer"
	#define ZPOOL_CONFIG_CACHEFILE "cachefile" /* not stored on disk */
	#define ZPOOL_CONFIG_MMP_STATE "mmp_state" /* not stored on disk */
	#define ZPOOL_CONFIG_MMP_TXG "mmp_txg" /* not stored on disk */
	#define ZPOOL_CONFIG_MMP_SEQ "mmp_seq" /* not stored on disk */
	#define ZPOOL_CONFIG_MMP_HOSTNAME "mmp_hostname" /* not stored on disk */
	#define ZPOOL_CONFIG_MMP_HOSTID "mmp_hostid" /* not stored on disk */
	#define ZPOOL_CONFIG_ALLOCATION_BIAS "alloc_bias" /* not stored on disk */
	#define ZPOOL_CONFIG_EXPANSION_TIME "expansion_time" /* not stored */
	#define ZPOOL_CONFIG_REBUILD_STATS "org.openzfs:rebuild_stats"
	#define ZPOOL_CONFIG_COMPATIBILITY "compatibility"

	/*
	* The persistent vdev state is stored as separate values rather than a single
	* 'vdev_state' entry. This is because a device can be in multiple states, such
	* as offline and degraded.
	*/
	#define ZPOOL_CONFIG_OFFLINE "offline"
	#define ZPOOL_CONFIG_FAULTED "faulted"
	#define ZPOOL_CONFIG_DEGRADED "degraded"
	#define ZPOOL_CONFIG_REMOVED "removed"
	#define ZPOOL_CONFIG_FRU "fru"
	#define ZPOOL_CONFIG_AUX_STATE "aux_state"

	/* Pool load policy parameters */
	#define ZPOOL_LOAD_POLICY "load-policy"
	#define ZPOOL_LOAD_REWIND_POLICY "load-rewind-policy"
	#define ZPOOL_LOAD_REQUEST_TXG "load-request-txg"
	#define ZPOOL_LOAD_META_THRESH "load-meta-thresh"
	#define ZPOOL_LOAD_DATA_THRESH "load-data-thresh"

	/* Rewind data discovered */
	#define ZPOOL_CONFIG_LOAD_TIME "rewind_txg_ts"
	#define ZPOOL_CONFIG_LOAD_DATA_ERRORS "verify_data_errors"
	#define ZPOOL_CONFIG_REWIND_TIME "seconds_of_rewind"

	/* dRAID configuration */
	#define ZPOOL_CONFIG_DRAID_NDATA "draid_ndata"
	#define ZPOOL_CONFIG_DRAID_NSPARES "draid_nspares"
	#define ZPOOL_CONFIG_DRAID_NGROUPS "draid_ngroups"

	#define VDEV_TYPE_ROOT "root"
	#define VDEV_TYPE_MIRROR "mirror"
	#define VDEV_TYPE_REPLACING "replacing"
	#define VDEV_TYPE_RAIDZ "raidz"
	#define VDEV_TYPE_DRAID "draid"
	#define VDEV_TYPE_DRAID_SPARE "dspare"
	#define VDEV_TYPE_DISK "disk"
	#define VDEV_TYPE_FILE "file"
	#define VDEV_TYPE_MISSING "missing"
	#define VDEV_TYPE_HOLE "hole"
	#define VDEV_TYPE_SPARE "spare"
	#define VDEV_TYPE_LOG "log"
	#define VDEV_TYPE_L2CACHE "l2cache"
	#define VDEV_TYPE_INDIRECT "indirect"

	#define VDEV_RAIDZ_MAXPARITY 3

	#define VDEV_DRAID_MAXPARITY 3
	#define VDEV_DRAID_MIN_CHILDREN 2
	#define VDEV_DRAID_MAX_CHILDREN UINT8_MAX

	/* VDEV_TOP_ZAP_* are used in top-level vdev ZAP objects. */
	#define VDEV_TOP_ZAP_INDIRECT_OBSOLETE_SM \
	"com.delphix:indirect_obsolete_sm"
	#define VDEV_TOP_ZAP_OBSOLETE_COUNTS_ARE_PRECISE \
	"com.delphix:obsolete_counts_are_precise"
	#define VDEV_TOP_ZAP_POOL_CHECKPOINT_SM \
	"com.delphix:pool_checkpoint_sm"
	#define VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS \
	"com.delphix:ms_unflushed_phys_txgs"

	#define VDEV_TOP_ZAP_VDEV_REBUILD_PHYS \
	"org.openzfs:vdev_rebuild"

	#define VDEV_TOP_ZAP_ALLOCATION_BIAS \
	"org.zfsonlinux:allocation_bias"

	/* vdev metaslab allocation bias */
	#define VDEV_ALLOC_BIAS_LOG "log"
	#define VDEV_ALLOC_BIAS_SPECIAL "special"
	#define VDEV_ALLOC_BIAS_DEDUP "dedup"

	/* vdev initialize state */
	#define VDEV_LEAF_ZAP_INITIALIZE_LAST_OFFSET \
	"com.delphix:next_offset_to_initialize"
	#define VDEV_LEAF_ZAP_INITIALIZE_STATE \
	"com.delphix:vdev_initialize_state"
	#define VDEV_LEAF_ZAP_INITIALIZE_ACTION_TIME \
	"com.delphix:vdev_initialize_action_time"

	/* vdev TRIM state */
	#define VDEV_LEAF_ZAP_TRIM_LAST_OFFSET \
	"org.zfsonlinux:next_offset_to_trim"
	#define VDEV_LEAF_ZAP_TRIM_STATE \
	"org.zfsonlinux:vdev_trim_state"
	#define VDEV_LEAF_ZAP_TRIM_ACTION_TIME \
	"org.zfsonlinux:vdev_trim_action_time"
	#define VDEV_LEAF_ZAP_TRIM_RATE \
	"org.zfsonlinux:vdev_trim_rate"
	#define VDEV_LEAF_ZAP_TRIM_PARTIAL \
	"org.zfsonlinux:vdev_trim_partial"
	#define VDEV_LEAF_ZAP_TRIM_SECURE \
	"org.zfsonlinux:vdev_trim_secure"

	/*
	* This is needed in userland to report the minimum necessary device size.
	*/
	#define SPA_MINDEVSIZE (64ULL << 20)

	/*
	* Set if the fragmentation has not yet been calculated. This can happen
	* because the space maps have not been upgraded or the histogram feature
	* is not enabled.
	*/
	#define ZFS_FRAG_INVALID UINT64_MAX

	/*
	* The location of the pool configuration repository, shared between kernel and
	* userland.
	*/
	#define ZPOOL_CACHE_BOOT "/boot/zfs/zpool.cache"
	#define ZPOOL_CACHE "/etc/zfs/zpool.cache"
	/*
	* Settings for zpool compatibility features files
	*/
	#define ZPOOL_SYSCONF_COMPAT_D SYSCONFDIR "/zfs/compatibility.d"
	#define ZPOOL_DATA_COMPAT_D PKGDATADIR "/compatibility.d"
	#define ZPOOL_COMPAT_MAXSIZE 16384

	/*
	* Hard-wired compatibility settings
	*/
	#define ZPOOL_COMPAT_LEGACY "legacy"
	#define ZPOOL_COMPAT_OFF "off"

	/*
	* vdev states are ordered from least to most healthy.
	* A vdev that's CANT_OPEN or below is considered unusable.
	*/
	typedef enum vdev_state {
	VDEV_STATE_UNKNOWN = 0, /* Uninitialized vdev */
	VDEV_STATE_CLOSED, /* Not currently open */
	VDEV_STATE_OFFLINE, /* Not allowed to open */
	VDEV_STATE_REMOVED, /* Explicitly removed from system */
	VDEV_STATE_CANT_OPEN, /* Tried to open, but failed */
	VDEV_STATE_FAULTED, /* External request to fault device */
	VDEV_STATE_DEGRADED, /* Replicated vdev with unhealthy kids */
	VDEV_STATE_HEALTHY /* Presumed good */
	} vdev_state_t;

	#define VDEV_STATE_ONLINE VDEV_STATE_HEALTHY

	/*
	* vdev aux states. When a vdev is in the CANT_OPEN state, the aux field
	* of the vdev stats structure uses these constants to distinguish why.
	*/
	typedef enum vdev_aux {
	VDEV_AUX_NONE, /* no error */
	VDEV_AUX_OPEN_FAILED, /* ldi_open_() or vn_open() failed /
	VDEV_AUX_CORRUPT_DATA, /* bad label or disk contents */
	VDEV_AUX_NO_REPLICAS, /* insufficient number of replicas */
	VDEV_AUX_BAD_GUID_SUM, /* vdev guid sum doesn't match */
	VDEV_AUX_TOO_SMALL, /* vdev size is too small */
	VDEV_AUX_BAD_LABEL, /* the label is OK but invalid */
	VDEV_AUX_VERSION_NEWER, /* on-disk version is too new */
	VDEV_AUX_VERSION_OLDER, /* on-disk version is too old */
	VDEV_AUX_UNSUP_FEAT, /* unsupported features */
	VDEV_AUX_SPARED, /* hot spare used in another pool */
	VDEV_AUX_ERR_EXCEEDED, /* too many errors */
	VDEV_AUX_IO_FAILURE, /* experienced I/O failure */
	VDEV_AUX_BAD_LOG, /* cannot read log chain(s) */
	VDEV_AUX_EXTERNAL, /* external diagnosis or forced fault */
	VDEV_AUX_SPLIT_POOL, /* vdev was split off into another pool */
	VDEV_AUX_BAD_ASHIFT, /* vdev ashift is invalid */
	VDEV_AUX_EXTERNAL_PERSIST, /* persistent forced fault */
	VDEV_AUX_ACTIVE, /* vdev active on a different host */
	VDEV_AUX_CHILDREN_OFFLINE, /* all children are offline */
	VDEV_AUX_ASHIFT_TOO_BIG, /* vdev's min block size is too large */
	} vdev_aux_t;

	/*
	* pool state. The following states are written to disk as part of the normal
	* SPA lifecycle: ACTIVE, EXPORTED, DESTROYED, SPARE, L2CACHE. The remaining
	* states are software abstractions used at various levels to communicate
	* pool state.
	*/
	typedef enum pool_state {
	POOL_STATE_ACTIVE = 0, /* In active use */
	POOL_STATE_EXPORTED, /* Explicitly exported */
	POOL_STATE_DESTROYED, /* Explicitly destroyed */
	POOL_STATE_SPARE, /* Reserved for hot spare use */
	POOL_STATE_L2CACHE, /* Level 2 ARC device */
	POOL_STATE_UNINITIALIZED, /* Internal spa_t state */
	POOL_STATE_UNAVAIL, /* Internal libzfs state */
	POOL_STATE_POTENTIALLY_ACTIVE /* Internal libzfs state */
	} pool_state_t;

	/*
	* mmp state. The following states provide additional detail describing
	* why a pool couldn't be safely imported.
	*/
	typedef enum mmp_state {
	MMP_STATE_ACTIVE = 0, /* In active use */
	MMP_STATE_INACTIVE, /* Inactive and safe to import */
	MMP_STATE_NO_HOSTID /* System hostid is not set */
	} mmp_state_t;

	/*
	* Scan Functions.
	*/
	typedef enum pool_scan_func {
	POOL_SCAN_NONE,
	POOL_SCAN_SCRUB,
	POOL_SCAN_RESILVER,
	POOL_SCAN_FUNCS
	} pool_scan_func_t;

	/*
	* Used to control scrub pause and resume.
	*/
	typedef enum pool_scrub_cmd {
	POOL_SCRUB_NORMAL = 0,
	POOL_SCRUB_PAUSE,
	POOL_SCRUB_FLAGS_END
	} pool_scrub_cmd_t;

	typedef enum {
	CS_NONE,
	CS_CHECKPOINT_EXISTS,
	CS_CHECKPOINT_DISCARDING,
	CS_NUM_STATES
	} checkpoint_state_t;

	typedef struct pool_checkpoint_stat {
	uint64_t pcs_state; /* checkpoint_state_t */
	uint64_t pcs_start_time; /* time checkpoint/discard started */
	uint64_t pcs_space; /* checkpointed space */
	} pool_checkpoint_stat_t;

	/*
	* ZIO types. Needed to interpret vdev statistics below.
	*/
	typedef enum zio_type {
	ZIO_TYPE_NULL = 0,
	ZIO_TYPE_READ,
	ZIO_TYPE_WRITE,
	ZIO_TYPE_FREE,
	ZIO_TYPE_CLAIM,
	ZIO_TYPE_IOCTL,
	ZIO_TYPE_TRIM,
	ZIO_TYPES
	} zio_type_t;

	/*
	* Pool statistics. Note: all fields should be 64-bit because this
	* is passed between kernel and userland as an nvlist uint64 array.
	*/
	typedef struct pool_scan_stat {
	/* values stored on disk */
	uint64_t pss_func; /* pool_scan_func_t */
	uint64_t pss_state; /* dsl_scan_state_t */
	uint64_t pss_start_time; /* scan start time */
	uint64_t pss_end_time; /* scan end time */
	uint64_t pss_to_examine; /* total bytes to scan */
	uint64_t pss_examined; /* total bytes located by scanner */
	uint64_t pss_to_process; /* total bytes to process */
	uint64_t pss_processed; /* total processed bytes */
	uint64_t pss_errors; /* scan errors */

	/* values not stored on disk */
	uint64_t pss_pass_exam; /* examined bytes per scan pass */
	uint64_t pss_pass_start; /* start time of a scan pass */
	uint64_t pss_pass_scrub_pause; /* pause time of a scrub pass */
	/* cumulative time scrub spent paused, needed for rate calculation */
	uint64_t pss_pass_scrub_spent_paused;
	uint64_t pss_pass_issued; /* issued bytes per scan pass */
	uint64_t pss_issued; /* total bytes checked by scanner */
	} pool_scan_stat_t;

	typedef struct pool_removal_stat {
	uint64_t prs_state; /* dsl_scan_state_t */
	uint64_t prs_removing_vdev;
	uint64_t prs_start_time;
	uint64_t prs_end_time;
	uint64_t prs_to_copy; /* bytes that need to be copied */
	uint64_t prs_copied; /* bytes copied so far */
	/*
	* bytes of memory used for indirect mappings.
	* This includes all removed vdevs.
	*/
	uint64_t prs_mapping_memory;
	} pool_removal_stat_t;

	typedef enum dsl_scan_state {
	DSS_NONE,
	DSS_SCANNING,
	DSS_FINISHED,
	DSS_CANCELED,
	DSS_NUM_STATES
	} dsl_scan_state_t;

	typedef struct vdev_rebuild_stat {
	uint64_t vrs_state; /* vdev_rebuild_state_t */
	uint64_t vrs_start_time; /* time_t */
	uint64_t vrs_end_time; /* time_t */
	uint64_t vrs_scan_time_ms; /* total run time (millisecs) */
	uint64_t vrs_bytes_scanned; /* allocated bytes scanned */
	uint64_t vrs_bytes_issued; /* read bytes issued */
	uint64_t vrs_bytes_rebuilt; /* rebuilt bytes */
	uint64_t vrs_bytes_est; /* total bytes to scan */
	uint64_t vrs_errors; /* scanning errors */
	uint64_t vrs_pass_time_ms; /* pass run time (millisecs) */
	uint64_t vrs_pass_bytes_scanned; /* bytes scanned since start/resume */
	uint64_t vrs_pass_bytes_issued; /* bytes rebuilt since start/resume */
	} vdev_rebuild_stat_t;

	/*
	* Errata described by https://openzfs.github.io/openzfs-docs/msg/ZFS-8000-ER.
	* The ordering of this enum must be maintained to ensure the errata identifiers
	* map to the correct documentation. New errata may only be appended to the
	* list and must contain corresponding documentation at the above link.
	*/
	typedef enum zpool_errata {
	ZPOOL_ERRATA_NONE,
	ZPOOL_ERRATA_ZOL_2094_SCRUB,
	ZPOOL_ERRATA_ZOL_2094_ASYNC_DESTROY,
	ZPOOL_ERRATA_ZOL_6845_ENCRYPTION,
	ZPOOL_ERRATA_ZOL_8308_ENCRYPTION,
	} zpool_errata_t;

	/*
	* Vdev statistics. Note: all fields should be 64-bit because this
	* is passed between kernel and user land as an nvlist uint64 array.
	*
	* The vs_ops[] and vs_bytes[] arrays must always be an array size of 6 in
	* order to keep subsequent members at their known fixed offsets. When
	* adding a new field it must be added to the end the structure.
	*/
	#define VS_ZIO_TYPES 6

	typedef struct vdev_stat {
	hrtime_t vs_timestamp; /* time since vdev load */
	uint64_t vs_state; /* vdev state */
	uint64_t vs_aux; /* see vdev_aux_t */
	uint64_t vs_alloc; /* space allocated */
	uint64_t vs_space; /* total capacity */
	uint64_t vs_dspace; /* deflated capacity */
	uint64_t vs_rsize; /* replaceable dev size */
	uint64_t vs_esize; /* expandable dev size */
	uint64_t vs_ops[VS_ZIO_TYPES]; /* operation count */
	uint64_t vs_bytes[VS_ZIO_TYPES]; /* bytes read/written */
	uint64_t vs_read_errors; /* read errors */
	uint64_t vs_write_errors; /* write errors */
	uint64_t vs_checksum_errors; /* checksum errors */
	uint64_t vs_initialize_errors; /* initializing errors */
	uint64_t vs_self_healed; /* self-healed bytes */
	uint64_t vs_scan_removing; /* removing? */
	uint64_t vs_scan_processed; /* scan processed bytes */
	uint64_t vs_fragmentation; /* device fragmentation */
	uint64_t vs_initialize_bytes_done; /* bytes initialized */
	uint64_t vs_initialize_bytes_est; /* total bytes to initialize */
	uint64_t vs_initialize_state; /* vdev_initializing_state_t */
	uint64_t vs_initialize_action_time; /* time_t */
	uint64_t vs_checkpoint_space; /* checkpoint-consumed space */
	uint64_t vs_resilver_deferred; /* resilver deferred */
	uint64_t vs_slow_ios; /* slow IOs */
	uint64_t vs_trim_errors; /* trimming errors */
	uint64_t vs_trim_notsup; /* supported by device */
	uint64_t vs_trim_bytes_done; /* bytes trimmed */
	uint64_t vs_trim_bytes_est; /* total bytes to trim */
	uint64_t vs_trim_state; /* vdev_trim_state_t */
	uint64_t vs_trim_action_time; /* time_t */
	uint64_t vs_rebuild_processed; /* bytes rebuilt */
	uint64_t vs_configured_ashift; /* TLV vdev_ashift */
	uint64_t vs_logical_ashift; /* vdev_logical_ashift */
	uint64_t vs_physical_ashift; /* vdev_physical_ashift */
	+ uint64_t vs_noalloc; /* allocations halted? */
	} vdev_stat_t;

	/* BEGIN CSTYLED */
	#define VDEV_STAT_VALID(field, uint64_t_field_count) \
	((uint64_t_field_count * sizeof (uint64_t)) >= \
	(offsetof(vdev_stat_t, field) + sizeof (((vdev_stat_t *)NULL)->field)))
	/* END CSTYLED */

	/*
	* Extended stats
	*
	* These are stats which aren't included in the original iostat output. For
	* convenience, they are grouped together in vdev_stat_ex, although each stat
	* is individually exported as an nvlist.
	*/
	typedef struct vdev_stat_ex {
	/* Number of ZIOs issued to disk and waiting to finish */
	uint64_t vsx_active_queue[ZIO_PRIORITY_NUM_QUEUEABLE];

	/* Number of ZIOs pending to be issued to disk */
	uint64_t vsx_pend_queue[ZIO_PRIORITY_NUM_QUEUEABLE];

	/*
	* Below are the histograms for various latencies. Buckets are in
	* units of nanoseconds.
	*/

	/*
	* 2^37 nanoseconds = 134s. Timeouts will probably start kicking in
	* before this.
	*/
	#define VDEV_L_HISTO_BUCKETS 37 /* Latency histo buckets */
	#define VDEV_RQ_HISTO_BUCKETS 25 /* Request size histo buckets */

	/* Amount of time in ZIO queue (ns) */
	uint64_t vsx_queue_histo[ZIO_PRIORITY_NUM_QUEUEABLE]
	[VDEV_L_HISTO_BUCKETS];

	/* Total ZIO latency (ns). Includes queuing and disk access time */
	uint64_t vsx_total_histo[ZIO_TYPES][VDEV_L_HISTO_BUCKETS];

	/* Amount of time to read/write the disk (ns) */
	uint64_t vsx_disk_histo[ZIO_TYPES][VDEV_L_HISTO_BUCKETS];

	/* "lookup the bucket for a value" histogram macros */
	#define HISTO(val, buckets) (val != 0 ? MIN(highbit64(val) - 1, \
	buckets - 1) : 0)
	#define L_HISTO(a) HISTO(a, VDEV_L_HISTO_BUCKETS)
	#define RQ_HISTO(a) HISTO(a, VDEV_RQ_HISTO_BUCKETS)

	/* Physical IO histogram */
	uint64_t vsx_ind_histo[ZIO_PRIORITY_NUM_QUEUEABLE]
	[VDEV_RQ_HISTO_BUCKETS];

	/* Delegated (aggregated) physical IO histogram */
	uint64_t vsx_agg_histo[ZIO_PRIORITY_NUM_QUEUEABLE]
	[VDEV_RQ_HISTO_BUCKETS];

	} vdev_stat_ex_t;

	/*
	* Initialize functions.
	*/
	typedef enum pool_initialize_func {
	POOL_INITIALIZE_START,
	POOL_INITIALIZE_CANCEL,
	POOL_INITIALIZE_SUSPEND,
	POOL_INITIALIZE_FUNCS
	} pool_initialize_func_t;

	/*
	* TRIM functions.
	*/
	typedef enum pool_trim_func {
	POOL_TRIM_START,
	POOL_TRIM_CANCEL,
	POOL_TRIM_SUSPEND,
	POOL_TRIM_FUNCS
	} pool_trim_func_t;

	/*
	* DDT statistics. Note: all fields should be 64-bit because this
	* is passed between kernel and userland as an nvlist uint64 array.
	*/
	typedef struct ddt_object {
	uint64_t ddo_count; /* number of elements in ddt */
	uint64_t ddo_dspace; /* size of ddt on disk */
	uint64_t ddo_mspace; /* size of ddt in-core */
	} ddt_object_t;

	typedef struct ddt_stat {
	uint64_t dds_blocks; /* blocks */
	uint64_t dds_lsize; /* logical size */
	uint64_t dds_psize; /* physical size */
	uint64_t dds_dsize; /* deflated allocated size */
	uint64_t dds_ref_blocks; /* referenced blocks */
	uint64_t dds_ref_lsize; /* referenced lsize * refcnt */
	uint64_t dds_ref_psize; /* referenced psize * refcnt */
	uint64_t dds_ref_dsize; /* referenced dsize * refcnt */
	} ddt_stat_t;

	typedef struct ddt_histogram {
	ddt_stat_t ddh_stat[64]; /* power-of-two histogram buckets */
	} ddt_histogram_t;

	#define ZVOL_DRIVER "zvol"
	#define ZFS_DRIVER "zfs"
	#define ZFS_DEV "/dev/zfs"

	#define ZFS_SUPER_MAGIC 0x2fc12fc1

	/* general zvol path */
	#define ZVOL_DIR "/dev/zvol/"

	#define ZVOL_MAJOR 230
	#define ZVOL_MINOR_BITS 4
	#define ZVOL_MINOR_MASK ((1U << ZVOL_MINOR_BITS) - 1)
	#define ZVOL_MINORS (1 << 4)
	#define ZVOL_DEV_NAME "zd"

	#define ZVOL_PROP_NAME "name"
	#define ZVOL_DEFAULT_BLOCKSIZE 16384

	typedef enum {
	VDEV_INITIALIZE_NONE,
	VDEV_INITIALIZE_ACTIVE,
	VDEV_INITIALIZE_CANCELED,
	VDEV_INITIALIZE_SUSPENDED,
	VDEV_INITIALIZE_COMPLETE
	} vdev_initializing_state_t;

	typedef enum {
	VDEV_TRIM_NONE,
	VDEV_TRIM_ACTIVE,
	VDEV_TRIM_CANCELED,
	VDEV_TRIM_SUSPENDED,
	VDEV_TRIM_COMPLETE,
	} vdev_trim_state_t;

	typedef enum {
	VDEV_REBUILD_NONE,
	VDEV_REBUILD_ACTIVE,
	VDEV_REBUILD_CANCELED,
	VDEV_REBUILD_COMPLETE,
	} vdev_rebuild_state_t;

	/*
	* nvlist name constants. Facilitate restricting snapshot iteration range for
	* the "list next snapshot" ioctl
	*/
	#define SNAP_ITER_MIN_TXG "snap_iter_min_txg"
	#define SNAP_ITER_MAX_TXG "snap_iter_max_txg"

	/*
	* /dev/zfs ioctl numbers.
	*
	* These numbers cannot change over time. New ioctl numbers must be appended.
	*/
	typedef enum zfs_ioc {
	/*
	* Core features - 81/128 numbers reserved.
	*/
	#ifdef __FreeBSD__
	ZFS_IOC_FIRST = 0,
	#else
	ZFS_IOC_FIRST = ('Z' << 8),
	#endif
	ZFS_IOC = ZFS_IOC_FIRST,
	ZFS_IOC_POOL_CREATE = ZFS_IOC_FIRST, /* 0x5a00 */
	ZFS_IOC_POOL_DESTROY, /* 0x5a01 */
	ZFS_IOC_POOL_IMPORT, /* 0x5a02 */
	ZFS_IOC_POOL_EXPORT, /* 0x5a03 */
	ZFS_IOC_POOL_CONFIGS, /* 0x5a04 */
	ZFS_IOC_POOL_STATS, /* 0x5a05 */
	ZFS_IOC_POOL_TRYIMPORT, /* 0x5a06 */
	ZFS_IOC_POOL_SCAN, /* 0x5a07 */
	ZFS_IOC_POOL_FREEZE, /* 0x5a08 */
	ZFS_IOC_POOL_UPGRADE, /* 0x5a09 */
	ZFS_IOC_POOL_GET_HISTORY, /* 0x5a0a */
	ZFS_IOC_VDEV_ADD, /* 0x5a0b */
	ZFS_IOC_VDEV_REMOVE, /* 0x5a0c */
	ZFS_IOC_VDEV_SET_STATE, /* 0x5a0d */
	ZFS_IOC_VDEV_ATTACH, /* 0x5a0e */
	ZFS_IOC_VDEV_DETACH, /* 0x5a0f */
	ZFS_IOC_VDEV_SETPATH, /* 0x5a10 */
	ZFS_IOC_VDEV_SETFRU, /* 0x5a11 */
	ZFS_IOC_OBJSET_STATS, /* 0x5a12 */
	ZFS_IOC_OBJSET_ZPLPROPS, /* 0x5a13 */
	ZFS_IOC_DATASET_LIST_NEXT, /* 0x5a14 */
	ZFS_IOC_SNAPSHOT_LIST_NEXT, /* 0x5a15 */
	ZFS_IOC_SET_PROP, /* 0x5a16 */
	ZFS_IOC_CREATE, /* 0x5a17 */
	ZFS_IOC_DESTROY, /* 0x5a18 */
	ZFS_IOC_ROLLBACK, /* 0x5a19 */
	ZFS_IOC_RENAME, /* 0x5a1a */
	ZFS_IOC_RECV, /* 0x5a1b */
	ZFS_IOC_SEND, /* 0x5a1c */
	ZFS_IOC_INJECT_FAULT, /* 0x5a1d */
	ZFS_IOC_CLEAR_FAULT, /* 0x5a1e */
	ZFS_IOC_INJECT_LIST_NEXT, /* 0x5a1f */
	ZFS_IOC_ERROR_LOG, /* 0x5a20 */
	ZFS_IOC_CLEAR, /* 0x5a21 */
	ZFS_IOC_PROMOTE, /* 0x5a22 */
	ZFS_IOC_SNAPSHOT, /* 0x5a23 */
	ZFS_IOC_DSOBJ_TO_DSNAME, /* 0x5a24 */
	ZFS_IOC_OBJ_TO_PATH, /* 0x5a25 */
	ZFS_IOC_POOL_SET_PROPS, /* 0x5a26 */
	ZFS_IOC_POOL_GET_PROPS, /* 0x5a27 */
	ZFS_IOC_SET_FSACL, /* 0x5a28 */
	ZFS_IOC_GET_FSACL, /* 0x5a29 */
	ZFS_IOC_SHARE, /* 0x5a2a */
	ZFS_IOC_INHERIT_PROP, /* 0x5a2b */
	ZFS_IOC_SMB_ACL, /* 0x5a2c */
	ZFS_IOC_USERSPACE_ONE, /* 0x5a2d */
	ZFS_IOC_USERSPACE_MANY, /* 0x5a2e */
	ZFS_IOC_USERSPACE_UPGRADE, /* 0x5a2f */
	ZFS_IOC_HOLD, /* 0x5a30 */
	ZFS_IOC_RELEASE, /* 0x5a31 */
	ZFS_IOC_GET_HOLDS, /* 0x5a32 */
	ZFS_IOC_OBJSET_RECVD_PROPS, /* 0x5a33 */
	ZFS_IOC_VDEV_SPLIT, /* 0x5a34 */
	ZFS_IOC_NEXT_OBJ, /* 0x5a35 */
	ZFS_IOC_DIFF, /* 0x5a36 */
	ZFS_IOC_TMP_SNAPSHOT, /* 0x5a37 */
	ZFS_IOC_OBJ_TO_STATS, /* 0x5a38 */
	ZFS_IOC_SPACE_WRITTEN, /* 0x5a39 */
	ZFS_IOC_SPACE_SNAPS, /* 0x5a3a */
	ZFS_IOC_DESTROY_SNAPS, /* 0x5a3b */
	ZFS_IOC_POOL_REGUID, /* 0x5a3c */
	ZFS_IOC_POOL_REOPEN, /* 0x5a3d */
	ZFS_IOC_SEND_PROGRESS, /* 0x5a3e */
	ZFS_IOC_LOG_HISTORY, /* 0x5a3f */
	ZFS_IOC_SEND_NEW, /* 0x5a40 */
	ZFS_IOC_SEND_SPACE, /* 0x5a41 */
	ZFS_IOC_CLONE, /* 0x5a42 */
	ZFS_IOC_BOOKMARK, /* 0x5a43 */
	ZFS_IOC_GET_BOOKMARKS, /* 0x5a44 */
	ZFS_IOC_DESTROY_BOOKMARKS, /* 0x5a45 */
	ZFS_IOC_RECV_NEW, /* 0x5a46 */
	ZFS_IOC_POOL_SYNC, /* 0x5a47 */
	ZFS_IOC_CHANNEL_PROGRAM, /* 0x5a48 */
	ZFS_IOC_LOAD_KEY, /* 0x5a49 */
	ZFS_IOC_UNLOAD_KEY, /* 0x5a4a */
	ZFS_IOC_CHANGE_KEY, /* 0x5a4b */
	ZFS_IOC_REMAP, /* 0x5a4c */
	ZFS_IOC_POOL_CHECKPOINT, /* 0x5a4d */
	ZFS_IOC_POOL_DISCARD_CHECKPOINT, /* 0x5a4e */
	ZFS_IOC_POOL_INITIALIZE, /* 0x5a4f */
	ZFS_IOC_POOL_TRIM, /* 0x5a50 */
	ZFS_IOC_REDACT, /* 0x5a51 */
	ZFS_IOC_GET_BOOKMARK_PROPS, /* 0x5a52 */
	ZFS_IOC_WAIT, /* 0x5a53 */
	ZFS_IOC_WAIT_FS, /* 0x5a54 */
	+ ZFS_IOC_VDEV_GET_PROPS, /* 0x5a55 */
	+ ZFS_IOC_VDEV_SET_PROPS, /* 0x5a56 */

	/*
	* Per-platform (Optional) - 8/128 numbers reserved.
	*/
	ZFS_IOC_PLATFORM = ZFS_IOC_FIRST + 0x80,
	ZFS_IOC_EVENTS_NEXT, /* 0x81 (Linux) */
	ZFS_IOC_EVENTS_CLEAR, /* 0x82 (Linux) */
	ZFS_IOC_EVENTS_SEEK, /* 0x83 (Linux) */
	ZFS_IOC_NEXTBOOT, /* 0x84 (FreeBSD) */
	ZFS_IOC_JAIL, /* 0x85 (FreeBSD) */
	ZFS_IOC_UNJAIL, /* 0x86 (FreeBSD) */
	ZFS_IOC_SET_BOOTENV, /* 0x87 */
	ZFS_IOC_GET_BOOTENV, /* 0x88 */
	ZFS_IOC_LAST
	} zfs_ioc_t;

	/*
	* zvol ioctl to get dataset name
	*/
	#define BLKZNAME _IOR(0x12, 125, char[ZFS_MAX_DATASET_NAME_LEN])

	/*
	* ZFS-specific error codes used for returning descriptive errors
	* to the userland through zfs ioctls.
	*
	* The enum implicitly includes all the error codes from errno.h.
	* New code should use and extend this enum for errors that are
	* not described precisely by generic errno codes.
	*
	* These numbers should not change over time. New entries should be appended.
	*
	* (Keep in sync with contrib/pyzfs/libzfs_core/_constants.py)
	*/
	typedef enum {
	ZFS_ERR_CHECKPOINT_EXISTS = 1024,
	ZFS_ERR_DISCARDING_CHECKPOINT,
	ZFS_ERR_NO_CHECKPOINT,
	ZFS_ERR_DEVRM_IN_PROGRESS,
	ZFS_ERR_VDEV_TOO_BIG,
	ZFS_ERR_IOC_CMD_UNAVAIL,
	ZFS_ERR_IOC_ARG_UNAVAIL,
	ZFS_ERR_IOC_ARG_REQUIRED,
	ZFS_ERR_IOC_ARG_BADTYPE,
	ZFS_ERR_WRONG_PARENT,
	ZFS_ERR_FROM_IVSET_GUID_MISSING,
	ZFS_ERR_FROM_IVSET_GUID_MISMATCH,
	ZFS_ERR_SPILL_BLOCK_FLAG_MISSING,
	ZFS_ERR_UNKNOWN_SEND_STREAM_FEATURE,
	ZFS_ERR_EXPORT_IN_PROGRESS,
	ZFS_ERR_BOOKMARK_SOURCE_NOT_ANCESTOR,
	ZFS_ERR_STREAM_TRUNCATED,
	ZFS_ERR_STREAM_LARGE_BLOCK_MISMATCH,
	ZFS_ERR_RESILVER_IN_PROGRESS,
	ZFS_ERR_REBUILD_IN_PROGRESS,
	ZFS_ERR_BADPROP,
	+ ZFS_ERR_VDEV_NOTSUP,
	} zfs_errno_t;

	/*
	* Internal SPA load state. Used by FMA diagnosis engine.
	*/
	typedef enum {
	SPA_LOAD_NONE, /* no load in progress */
	SPA_LOAD_OPEN, /* normal open */
	SPA_LOAD_IMPORT, /* import in progress */
	SPA_LOAD_TRYIMPORT, /* tryimport in progress */
	SPA_LOAD_RECOVER, /* recovery requested */
	SPA_LOAD_ERROR, /* load failed */
	SPA_LOAD_CREATE /* creation in progress */
	} spa_load_state_t;

	typedef enum {
	ZPOOL_WAIT_CKPT_DISCARD,
	ZPOOL_WAIT_FREE,
	ZPOOL_WAIT_INITIALIZE,
	ZPOOL_WAIT_REPLACE,
	ZPOOL_WAIT_REMOVE,
	ZPOOL_WAIT_RESILVER,
	ZPOOL_WAIT_SCRUB,
	ZPOOL_WAIT_TRIM,
	ZPOOL_WAIT_NUM_ACTIVITIES
	} zpool_wait_activity_t;

	typedef enum {
	ZFS_WAIT_DELETEQ,
	ZFS_WAIT_NUM_ACTIVITIES
	} zfs_wait_activity_t;

	/*
	* Bookmark name values.
	*/
	#define ZPOOL_ERR_LIST "error list"
	#define ZPOOL_ERR_DATASET "dataset"
	#define ZPOOL_ERR_OBJECT "object"

	#define HIS_MAX_RECORD_LEN (MAXPATHLEN + MAXPATHLEN + 1)

	/*
	* The following are names used in the nvlist describing
	* the pool's history log.
	*/
	#define ZPOOL_HIST_RECORD "history record"
	#define ZPOOL_HIST_TIME "history time"
	#define ZPOOL_HIST_CMD "history command"
	#define ZPOOL_HIST_WHO "history who"
	#define ZPOOL_HIST_ZONE "history zone"
	#define ZPOOL_HIST_HOST "history hostname"
	#define ZPOOL_HIST_TXG "history txg"
	#define ZPOOL_HIST_INT_EVENT "history internal event"
	#define ZPOOL_HIST_INT_STR "history internal str"
	#define ZPOOL_HIST_INT_NAME "internal_name"
	#define ZPOOL_HIST_IOCTL "ioctl"
	#define ZPOOL_HIST_INPUT_NVL "in_nvl"
	#define ZPOOL_HIST_OUTPUT_NVL "out_nvl"
	#define ZPOOL_HIST_OUTPUT_SIZE "out_size"
	#define ZPOOL_HIST_DSNAME "dsname"
	#define ZPOOL_HIST_DSID "dsid"
	#define ZPOOL_HIST_ERRNO "errno"
	#define ZPOOL_HIST_ELAPSED_NS "elapsed_ns"

	/*
	* Special nvlist name that will not have its args recorded in the pool's
	* history log.
	*/
	#define ZPOOL_HIDDEN_ARGS "hidden_args"

	/*
	* The following are names used when invoking ZFS_IOC_POOL_INITIALIZE.
	*/
	#define ZPOOL_INITIALIZE_COMMAND "initialize_command"
	#define ZPOOL_INITIALIZE_VDEVS "initialize_vdevs"

	/*
	* The following are names used when invoking ZFS_IOC_POOL_TRIM.
	*/
	#define ZPOOL_TRIM_COMMAND "trim_command"
	#define ZPOOL_TRIM_VDEVS "trim_vdevs"
	#define ZPOOL_TRIM_RATE "trim_rate"
	#define ZPOOL_TRIM_SECURE "trim_secure"

	/*
	* The following are names used when invoking ZFS_IOC_POOL_WAIT.
	*/
	#define ZPOOL_WAIT_ACTIVITY "wait_activity"
	#define ZPOOL_WAIT_TAG "wait_tag"
	#define ZPOOL_WAIT_WAITED "wait_waited"

	+/*
	+ * The following are names used when invoking ZFS_IOC_VDEV_GET_PROP.
	+ */
	+#define ZPOOL_VDEV_PROPS_GET_VDEV "vdevprops_get_vdev"
	+#define ZPOOL_VDEV_PROPS_GET_PROPS "vdevprops_get_props"
	+
	+/*
	+ * The following are names used when invoking ZFS_IOC_VDEV_SET_PROP.
	+ */
	+#define ZPOOL_VDEV_PROPS_SET_VDEV "vdevprops_set_vdev"
	+#define ZPOOL_VDEV_PROPS_SET_PROPS "vdevprops_set_props"
	+
	/*
	* The following are names used when invoking ZFS_IOC_WAIT_FS.
	*/
	#define ZFS_WAIT_ACTIVITY "wait_activity"
	#define ZFS_WAIT_WAITED "wait_waited"

	/*
	* Flags for ZFS_IOC_VDEV_SET_STATE
	*/
	#define ZFS_ONLINE_CHECKREMOVE 0x1
	#define ZFS_ONLINE_UNSPARE 0x2
	#define ZFS_ONLINE_FORCEFAULT 0x4
	#define ZFS_ONLINE_EXPAND 0x8
	#define ZFS_OFFLINE_TEMPORARY 0x1

	/*
	* Flags for ZFS_IOC_POOL_IMPORT
	*/
	#define ZFS_IMPORT_NORMAL 0x0
	#define ZFS_IMPORT_VERBATIM 0x1
	#define ZFS_IMPORT_ANY_HOST 0x2
	#define ZFS_IMPORT_MISSING_LOG 0x4
	#define ZFS_IMPORT_ONLY 0x8
	#define ZFS_IMPORT_TEMP_NAME 0x10
	#define ZFS_IMPORT_SKIP_MMP 0x20
	#define ZFS_IMPORT_LOAD_KEYS 0x40
	#define ZFS_IMPORT_CHECKPOINT 0x80

	/*
	* Channel program argument/return nvlist keys and defaults.
	*/
	#define ZCP_ARG_PROGRAM "program"
	#define ZCP_ARG_ARGLIST "arg"
	#define ZCP_ARG_SYNC "sync"
	#define ZCP_ARG_INSTRLIMIT "instrlimit"
	#define ZCP_ARG_MEMLIMIT "memlimit"

	#define ZCP_ARG_CLIARGV "argv"

	#define ZCP_RET_ERROR "error"
	#define ZCP_RET_RETURN "return"

	#define ZCP_DEFAULT_INSTRLIMIT (10 * 1000 * 1000)
	#define ZCP_MAX_INSTRLIMIT (10 * ZCP_DEFAULT_INSTRLIMIT)
	#define ZCP_DEFAULT_MEMLIMIT (10 * 1024 * 1024)
	#define ZCP_MAX_MEMLIMIT (10 * ZCP_DEFAULT_MEMLIMIT)

	/*
	* Sysevent payload members. ZFS will generate the following sysevents with the
	* given payloads:
	*
	* ESC_ZFS_RESILVER_START
	* ESC_ZFS_RESILVER_FINISH
	*
	* ZFS_EV_POOL_NAME DATA_TYPE_STRING
	* ZFS_EV_POOL_GUID DATA_TYPE_UINT64
	* ZFS_EV_RESILVER_TYPE DATA_TYPE_STRING
	*
	* ESC_ZFS_POOL_DESTROY
	* ESC_ZFS_POOL_REGUID
	*
	* ZFS_EV_POOL_NAME DATA_TYPE_STRING
	* ZFS_EV_POOL_GUID DATA_TYPE_UINT64
	*
	* ESC_ZFS_VDEV_REMOVE
	* ESC_ZFS_VDEV_CLEAR
	* ESC_ZFS_VDEV_CHECK
	*
	* ZFS_EV_POOL_NAME DATA_TYPE_STRING
	* ZFS_EV_POOL_GUID DATA_TYPE_UINT64
	* ZFS_EV_VDEV_PATH DATA_TYPE_STRING (optional)
	* ZFS_EV_VDEV_GUID DATA_TYPE_UINT64
	*
	* ESC_ZFS_HISTORY_EVENT
	*
	* ZFS_EV_POOL_NAME DATA_TYPE_STRING
	* ZFS_EV_POOL_GUID DATA_TYPE_UINT64
	* ZFS_EV_HIST_TIME DATA_TYPE_UINT64 (optional)
	* ZFS_EV_HIST_CMD DATA_TYPE_STRING (optional)
	* ZFS_EV_HIST_WHO DATA_TYPE_UINT64 (optional)
	* ZFS_EV_HIST_ZONE DATA_TYPE_STRING (optional)
	* ZFS_EV_HIST_HOST DATA_TYPE_STRING (optional)
	* ZFS_EV_HIST_TXG DATA_TYPE_UINT64 (optional)
	* ZFS_EV_HIST_INT_EVENT DATA_TYPE_UINT64 (optional)
	* ZFS_EV_HIST_INT_STR DATA_TYPE_STRING (optional)
	* ZFS_EV_HIST_INT_NAME DATA_TYPE_STRING (optional)
	* ZFS_EV_HIST_IOCTL DATA_TYPE_STRING (optional)
	* ZFS_EV_HIST_DSNAME DATA_TYPE_STRING (optional)
	* ZFS_EV_HIST_DSID DATA_TYPE_UINT64 (optional)
	*
	* The ZFS_EV_HIST_* members will correspond to the ZPOOL_HIST_* members in the
	* history log nvlist. The keynames will be free of any spaces or other
	* characters that could be potentially unexpected to consumers of the
	* sysevents.
	*/
	#define ZFS_EV_POOL_NAME "pool_name"
	#define ZFS_EV_POOL_GUID "pool_guid"
	#define ZFS_EV_VDEV_PATH "vdev_path"
	#define ZFS_EV_VDEV_GUID "vdev_guid"
	#define ZFS_EV_HIST_TIME "history_time"
	#define ZFS_EV_HIST_CMD "history_command"
	#define ZFS_EV_HIST_WHO "history_who"
	#define ZFS_EV_HIST_ZONE "history_zone"
	#define ZFS_EV_HIST_HOST "history_hostname"
	#define ZFS_EV_HIST_TXG "history_txg"
	#define ZFS_EV_HIST_INT_EVENT "history_internal_event"
	#define ZFS_EV_HIST_INT_STR "history_internal_str"
	#define ZFS_EV_HIST_INT_NAME "history_internal_name"
	#define ZFS_EV_HIST_IOCTL "history_ioctl"
	#define ZFS_EV_HIST_DSNAME "history_dsname"
	#define ZFS_EV_HIST_DSID "history_dsid"
	#define ZFS_EV_RESILVER_TYPE "resilver_type"


	/*
	* We currently support block sizes from 512 bytes to 16MB.
	* The benefits of larger blocks, and thus larger IO, need to be weighed
	* against the cost of COWing a giant block to modify one byte, and the
	* large latency of reading or writing a large block.
	*
	* Note that although blocks up to 16MB are supported, the recordsize
	* property can not be set larger than zfs_max_recordsize (default 1MB).
	* See the comment near zfs_max_recordsize in dsl_dataset.c for details.
	*
	* Note that although the LSIZE field of the blkptr_t can store sizes up
	* to 32MB, the dnode's dn_datablkszsec can only store sizes up to
	* 32MB - 512 bytes. Therefore, we limit SPA_MAXBLOCKSIZE to 16MB.
	*/
	#define SPA_MINBLOCKSHIFT 9
	#define SPA_OLD_MAXBLOCKSHIFT 17
	#define SPA_MAXBLOCKSHIFT 24
	#define SPA_MINBLOCKSIZE (1ULL << SPA_MINBLOCKSHIFT)
	#define SPA_OLD_MAXBLOCKSIZE (1ULL << SPA_OLD_MAXBLOCKSHIFT)
	#define SPA_MAXBLOCKSIZE (1ULL << SPA_MAXBLOCKSHIFT)


	/* supported encryption algorithms */
	enum zio_encrypt {
	ZIO_CRYPT_INHERIT = 0,
	ZIO_CRYPT_ON,
	ZIO_CRYPT_OFF,
	ZIO_CRYPT_AES_128_CCM,
	ZIO_CRYPT_AES_192_CCM,
	ZIO_CRYPT_AES_256_CCM,
	ZIO_CRYPT_AES_128_GCM,
	ZIO_CRYPT_AES_192_GCM,
	ZIO_CRYPT_AES_256_GCM,
	ZIO_CRYPT_FUNCTIONS
	};

	#define ZIO_CRYPT_ON_VALUE ZIO_CRYPT_AES_256_GCM
	#define ZIO_CRYPT_DEFAULT ZIO_CRYPT_OFF


	#ifdef __cplusplus
	}
	#endif

	#endif /* _SYS_FS_ZFS_H */
	diff --git a/sys/contrib/openzfs/include/sys/nvpair.h b/sys/contrib/openzfs/include/sys/nvpair.h
	index 76d383a3c681..340b0d79c0a2 100644
	--- a/sys/contrib/openzfs/include/sys/nvpair.h
	+++ b/sys/contrib/openzfs/include/sys/nvpair.h
	@@ -1,407 +1,416 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2000, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2012, 2018 by Delphix. All rights reserved.
	*/

	#ifndef _SYS_NVPAIR_H
	#define _SYS_NVPAIR_H extern __attribute__((visibility("default")))

	#include <sys/types.h>
	#include <sys/time.h>
	#include <sys/errno.h>

	#ifdef __cplusplus
	extern "C" {
	#endif

	typedef enum {
	DATA_TYPE_DONTCARE = -1,
	DATA_TYPE_UNKNOWN = 0,
	DATA_TYPE_BOOLEAN,
	DATA_TYPE_BYTE,
	DATA_TYPE_INT16,
	DATA_TYPE_UINT16,
	DATA_TYPE_INT32,
	DATA_TYPE_UINT32,
	DATA_TYPE_INT64,
	DATA_TYPE_UINT64,
	DATA_TYPE_STRING,
	DATA_TYPE_BYTE_ARRAY,
	DATA_TYPE_INT16_ARRAY,
	DATA_TYPE_UINT16_ARRAY,
	DATA_TYPE_INT32_ARRAY,
	DATA_TYPE_UINT32_ARRAY,
	DATA_TYPE_INT64_ARRAY,
	DATA_TYPE_UINT64_ARRAY,
	DATA_TYPE_STRING_ARRAY,
	DATA_TYPE_HRTIME,
	DATA_TYPE_NVLIST,
	DATA_TYPE_NVLIST_ARRAY,
	DATA_TYPE_BOOLEAN_VALUE,
	DATA_TYPE_INT8,
	DATA_TYPE_UINT8,
	DATA_TYPE_BOOLEAN_ARRAY,
	DATA_TYPE_INT8_ARRAY,
	#if !defined(_KERNEL) && !defined(_STANDALONE)
	DATA_TYPE_UINT8_ARRAY,
	DATA_TYPE_DOUBLE
	#else
	DATA_TYPE_UINT8_ARRAY
	#endif
	} data_type_t;

	typedef struct nvpair {
	int32_t nvp_size; /* size of this nvpair */
	int16_t nvp_name_sz; /* length of name string */
	int16_t nvp_reserve; /* not used */
	int32_t nvp_value_elem; /* number of elements for array types */
	data_type_t nvp_type; /* type of value */
	/* name string */
	/* aligned ptr array for string arrays */
	/* aligned array of data for value */
	} nvpair_t;

	/* nvlist header */
	typedef struct nvlist {
	int32_t nvl_version;
	uint32_t nvl_nvflag; /* persistent flags */
	uint64_t nvl_priv; /* ptr to private data if not packed */
	uint32_t nvl_flag;
	int32_t nvl_pad; /* currently not used, for alignment */
	} nvlist_t;

	/* nvp implementation version */
	#define NV_VERSION 0

	/* nvlist pack encoding */
	#define NV_ENCODE_NATIVE 0
	#define NV_ENCODE_XDR 1

	/* nvlist persistent unique name flags, stored in nvl_nvflags */
	#define NV_UNIQUE_NAME 0x1
	#define NV_UNIQUE_NAME_TYPE 0x2

	/* nvlist lookup pairs related flags */
	#define NV_FLAG_NOENTOK 0x1

	/* convenience macros */
	#define NV_ALIGN(x) (((ulong_t)(x) + 7ul) & ~7ul)
	#define NV_ALIGN4(x) (((x) + 3) & ~3)

	#define NVP_SIZE(nvp) ((nvp)->nvp_size)
	#define NVP_NAME(nvp) ((char *)(nvp) + sizeof (nvpair_t))
	#define NVP_TYPE(nvp) ((nvp)->nvp_type)
	#define NVP_NELEM(nvp) ((nvp)->nvp_value_elem)
	#define NVP_VALUE(nvp) ((char *)(nvp) + NV_ALIGN(sizeof (nvpair_t) \
	+ (nvp)->nvp_name_sz))

	#define NVL_VERSION(nvl) ((nvl)->nvl_version)
	#define NVL_SIZE(nvl) ((nvl)->nvl_size)
	#define NVL_FLAG(nvl) ((nvl)->nvl_flag)

	/* NV allocator framework */
	typedef struct nv_alloc_ops nv_alloc_ops_t;

	typedef struct nv_alloc {
	const nv_alloc_ops_t *nva_ops;
	void *nva_arg;
	} nv_alloc_t;

	struct nv_alloc_ops {
	int (nv_ao_init)(nv_alloc_t , va_list);
	void (nv_ao_fini)(nv_alloc_t );
	void (nv_ao_alloc)(nv_alloc_t *, size_t);
	void (nv_ao_free)(nv_alloc_t , void *, size_t);
	void (nv_ao_reset)(nv_alloc_t );
	};

	_SYS_NVPAIR_H const nv_alloc_ops_t *nv_fixed_ops;
	_SYS_NVPAIR_H nv_alloc_t *nv_alloc_nosleep;

	#if defined(_KERNEL)
	_SYS_NVPAIR_H nv_alloc_t *nv_alloc_sleep;
	_SYS_NVPAIR_H nv_alloc_t *nv_alloc_pushpage;
	#endif

	_SYS_NVPAIR_H int nv_alloc_init(nv_alloc_t , const nv_alloc_ops_t ,
	/* args */ ...);
	_SYS_NVPAIR_H void nv_alloc_reset(nv_alloc_t *);
	_SYS_NVPAIR_H void nv_alloc_fini(nv_alloc_t *);

	/* list management */
	_SYS_NVPAIR_H int nvlist_alloc(nvlist_t **, uint_t, int);
	_SYS_NVPAIR_H void nvlist_free(nvlist_t *);
	_SYS_NVPAIR_H int nvlist_size(nvlist_t , size_t , int);
	_SYS_NVPAIR_H int nvlist_pack(nvlist_t , char , size_t , int, int);
	_SYS_NVPAIR_H int nvlist_unpack(char , size_t, nvlist_t *, int);
	-_SYS_NVPAIR_H int nvlist_dup(nvlist_t , nvlist_t *, int);
	+_SYS_NVPAIR_H int nvlist_dup(const nvlist_t , nvlist_t *, int);
	_SYS_NVPAIR_H int nvlist_merge(nvlist_t , nvlist_t , int);

	_SYS_NVPAIR_H uint_t nvlist_nvflag(nvlist_t *);

	_SYS_NVPAIR_H int nvlist_xalloc(nvlist_t *, uint_t, nv_alloc_t );
	_SYS_NVPAIR_H int nvlist_xpack(nvlist_t , char , size_t , int,
	nv_alloc_t *);
	_SYS_NVPAIR_H int nvlist_xunpack(char , size_t, nvlist_t , nv_alloc_t );
	-_SYS_NVPAIR_H int nvlist_xdup(nvlist_t , nvlist_t , nv_alloc_t );
	+_SYS_NVPAIR_H int nvlist_xdup(const nvlist_t , nvlist_t , nv_alloc_t );
	_SYS_NVPAIR_H nv_alloc_t nvlist_lookup_nv_alloc(nvlist_t );

	_SYS_NVPAIR_H int nvlist_add_nvpair(nvlist_t , nvpair_t );
	_SYS_NVPAIR_H int nvlist_add_boolean(nvlist_t , const char );
	_SYS_NVPAIR_H int nvlist_add_boolean_value(nvlist_t , const char , boolean_t);
	_SYS_NVPAIR_H int nvlist_add_byte(nvlist_t , const char , uchar_t);
	_SYS_NVPAIR_H int nvlist_add_int8(nvlist_t , const char , int8_t);
	_SYS_NVPAIR_H int nvlist_add_uint8(nvlist_t , const char , uint8_t);
	_SYS_NVPAIR_H int nvlist_add_int16(nvlist_t , const char , int16_t);
	_SYS_NVPAIR_H int nvlist_add_uint16(nvlist_t , const char , uint16_t);
	_SYS_NVPAIR_H int nvlist_add_int32(nvlist_t , const char , int32_t);
	_SYS_NVPAIR_H int nvlist_add_uint32(nvlist_t , const char , uint32_t);
	_SYS_NVPAIR_H int nvlist_add_int64(nvlist_t , const char , int64_t);
	_SYS_NVPAIR_H int nvlist_add_uint64(nvlist_t , const char , uint64_t);
	_SYS_NVPAIR_H int nvlist_add_string(nvlist_t , const char , const char *);
	-_SYS_NVPAIR_H int nvlist_add_nvlist(nvlist_t , const char , nvlist_t *);
	+_SYS_NVPAIR_H int nvlist_add_nvlist(nvlist_t , const char , const nvlist_t *);
	_SYS_NVPAIR_H int nvlist_add_boolean_array(nvlist_t , const char ,
	- boolean_t *, uint_t);
	-_SYS_NVPAIR_H int nvlist_add_byte_array(nvlist_t , const char , uchar_t *,
	- uint_t);
	-_SYS_NVPAIR_H int nvlist_add_int8_array(nvlist_t , const char , int8_t *,
	- uint_t);
	-_SYS_NVPAIR_H int nvlist_add_uint8_array(nvlist_t , const char , uint8_t *,
	- uint_t);
	-_SYS_NVPAIR_H int nvlist_add_int16_array(nvlist_t , const char , int16_t *,
	- uint_t);
	-_SYS_NVPAIR_H int nvlist_add_uint16_array(nvlist_t , const char , uint16_t *,
	- uint_t);
	-_SYS_NVPAIR_H int nvlist_add_int32_array(nvlist_t , const char , int32_t *,
	- uint_t);
	-_SYS_NVPAIR_H int nvlist_add_uint32_array(nvlist_t , const char , uint32_t *,
	- uint_t);
	-_SYS_NVPAIR_H int nvlist_add_int64_array(nvlist_t , const char , int64_t *,
	- uint_t);
	-_SYS_NVPAIR_H int nvlist_add_uint64_array(nvlist_t , const char , uint64_t *,
	- uint_t);
	+ const boolean_t *, uint_t);
	+_SYS_NVPAIR_H int nvlist_add_byte_array(nvlist_t , const char ,
	+ const uchar_t *, uint_t);
	+_SYS_NVPAIR_H int nvlist_add_int8_array(nvlist_t , const char ,
	+ const int8_t *, uint_t);
	+_SYS_NVPAIR_H int nvlist_add_uint8_array(nvlist_t , const char ,
	+ const uint8_t *, uint_t);
	+_SYS_NVPAIR_H int nvlist_add_int16_array(nvlist_t , const char ,
	+ const int16_t *, uint_t);
	+_SYS_NVPAIR_H int nvlist_add_uint16_array(nvlist_t , const char ,
	+ const uint16_t *, uint_t);
	+_SYS_NVPAIR_H int nvlist_add_int32_array(nvlist_t , const char ,
	+ const int32_t *, uint_t);
	+_SYS_NVPAIR_H int nvlist_add_uint32_array(nvlist_t , const char ,
	+ const uint32_t *, uint_t);
	+_SYS_NVPAIR_H int nvlist_add_int64_array(nvlist_t , const char ,
	+ const int64_t *, uint_t);
	+_SYS_NVPAIR_H int nvlist_add_uint64_array(nvlist_t , const char ,
	+ const uint64_t *, uint_t);
	_SYS_NVPAIR_H int nvlist_add_string_array(nvlist_t , const char ,
	- char * const *, uint_t);
	+ const char * const *, uint_t);
	_SYS_NVPAIR_H int nvlist_add_nvlist_array(nvlist_t , const char ,
	- nvlist_t **, uint_t);
	+ const nvlist_t * const *, uint_t);
	_SYS_NVPAIR_H int nvlist_add_hrtime(nvlist_t , const char , hrtime_t);
	#if !defined(_KERNEL) && !defined(_STANDALONE)
	_SYS_NVPAIR_H int nvlist_add_double(nvlist_t , const char , double);
	#endif

	_SYS_NVPAIR_H int nvlist_remove(nvlist_t , const char , data_type_t);
	_SYS_NVPAIR_H int nvlist_remove_all(nvlist_t , const char );
	_SYS_NVPAIR_H int nvlist_remove_nvpair(nvlist_t , nvpair_t );

	-_SYS_NVPAIR_H int nvlist_lookup_boolean(nvlist_t , const char );
	-_SYS_NVPAIR_H int nvlist_lookup_boolean_value(nvlist_t , const char ,
	+_SYS_NVPAIR_H int nvlist_lookup_boolean(const nvlist_t , const char );
	+_SYS_NVPAIR_H int nvlist_lookup_boolean_value(const nvlist_t , const char ,
	boolean_t *);
	-_SYS_NVPAIR_H int nvlist_lookup_byte(nvlist_t , const char , uchar_t *);
	-_SYS_NVPAIR_H int nvlist_lookup_int8(nvlist_t , const char , int8_t *);
	-_SYS_NVPAIR_H int nvlist_lookup_uint8(nvlist_t , const char , uint8_t *);
	-_SYS_NVPAIR_H int nvlist_lookup_int16(nvlist_t , const char , int16_t *);
	-_SYS_NVPAIR_H int nvlist_lookup_uint16(nvlist_t , const char , uint16_t *);
	-_SYS_NVPAIR_H int nvlist_lookup_int32(nvlist_t , const char , int32_t *);
	-_SYS_NVPAIR_H int nvlist_lookup_uint32(nvlist_t , const char , uint32_t *);
	-_SYS_NVPAIR_H int nvlist_lookup_int64(nvlist_t , const char , int64_t *);
	-_SYS_NVPAIR_H int nvlist_lookup_uint64(nvlist_t , const char , uint64_t *);
	+_SYS_NVPAIR_H int nvlist_lookup_byte(const nvlist_t , const char , uchar_t *);
	+_SYS_NVPAIR_H int nvlist_lookup_int8(const nvlist_t , const char , int8_t *);
	+_SYS_NVPAIR_H int nvlist_lookup_uint8(const nvlist_t , const char ,
	+ uint8_t *);
	+_SYS_NVPAIR_H int nvlist_lookup_int16(const nvlist_t , const char ,
	+ int16_t *);
	+_SYS_NVPAIR_H int nvlist_lookup_uint16(const nvlist_t , const char ,
	+ uint16_t *);
	+_SYS_NVPAIR_H int nvlist_lookup_int32(const nvlist_t , const char ,
	+ int32_t *);
	+_SYS_NVPAIR_H int nvlist_lookup_uint32(const nvlist_t , const char ,
	+ uint32_t *);
	+_SYS_NVPAIR_H int nvlist_lookup_int64(const nvlist_t , const char ,
	+ int64_t *);
	+_SYS_NVPAIR_H int nvlist_lookup_uint64(const nvlist_t , const char ,
	+ uint64_t *);
	_SYS_NVPAIR_H int nvlist_lookup_string(nvlist_t , const char , char **);
	_SYS_NVPAIR_H int nvlist_lookup_nvlist(nvlist_t , const char , nvlist_t **);
	_SYS_NVPAIR_H int nvlist_lookup_boolean_array(nvlist_t , const char ,
	boolean_t *, uint_t );
	_SYS_NVPAIR_H int nvlist_lookup_byte_array(nvlist_t , const char , uchar_t **,
	uint_t *);
	_SYS_NVPAIR_H int nvlist_lookup_int8_array(nvlist_t , const char , int8_t **,
	uint_t *);
	_SYS_NVPAIR_H int nvlist_lookup_uint8_array(nvlist_t , const char ,
	uint8_t *, uint_t );
	_SYS_NVPAIR_H int nvlist_lookup_int16_array(nvlist_t , const char ,
	int16_t *, uint_t );
	_SYS_NVPAIR_H int nvlist_lookup_uint16_array(nvlist_t , const char ,
	uint16_t *, uint_t );
	_SYS_NVPAIR_H int nvlist_lookup_int32_array(nvlist_t , const char ,
	int32_t *, uint_t );
	_SYS_NVPAIR_H int nvlist_lookup_uint32_array(nvlist_t , const char ,
	uint32_t *, uint_t );
	_SYS_NVPAIR_H int nvlist_lookup_int64_array(nvlist_t , const char ,
	int64_t *, uint_t );
	_SYS_NVPAIR_H int nvlist_lookup_uint64_array(nvlist_t , const char ,
	uint64_t *, uint_t );
	_SYS_NVPAIR_H int nvlist_lookup_string_array(nvlist_t , const char ,
	char **, uint_t );
	_SYS_NVPAIR_H int nvlist_lookup_nvlist_array(nvlist_t , const char ,
	nvlist_t **, uint_t );
	_SYS_NVPAIR_H int nvlist_lookup_hrtime(nvlist_t , const char , hrtime_t *);
	_SYS_NVPAIR_H int nvlist_lookup_pairs(nvlist_t *, int, ...);
	#if !defined(_KERNEL) && !defined(_STANDALONE)
	-_SYS_NVPAIR_H int nvlist_lookup_double(nvlist_t , const char , double *);
	+_SYS_NVPAIR_H int nvlist_lookup_double(const nvlist_t , const char ,
	+ double *);
	#endif

	_SYS_NVPAIR_H int nvlist_lookup_nvpair(nvlist_t , const char , nvpair_t **);
	_SYS_NVPAIR_H int nvlist_lookup_nvpair_embedded_index(nvlist_t , const char ,
	nvpair_t *, int , char **);
	-_SYS_NVPAIR_H boolean_t nvlist_exists(nvlist_t , const char );
	-_SYS_NVPAIR_H boolean_t nvlist_empty(nvlist_t *);
	+_SYS_NVPAIR_H boolean_t nvlist_exists(const nvlist_t , const char );
	+_SYS_NVPAIR_H boolean_t nvlist_empty(const nvlist_t *);

	/* processing nvpair */
	-_SYS_NVPAIR_H nvpair_t nvlist_next_nvpair(nvlist_t , nvpair_t *);
	-_SYS_NVPAIR_H nvpair_t nvlist_prev_nvpair(nvlist_t , nvpair_t *);
	-_SYS_NVPAIR_H char nvpair_name(nvpair_t );
	-_SYS_NVPAIR_H data_type_t nvpair_type(nvpair_t *);
	-_SYS_NVPAIR_H int nvpair_type_is_array(nvpair_t *);
	-_SYS_NVPAIR_H int nvpair_value_boolean_value(nvpair_t , boolean_t );
	-_SYS_NVPAIR_H int nvpair_value_byte(nvpair_t , uchar_t );
	-_SYS_NVPAIR_H int nvpair_value_int8(nvpair_t , int8_t );
	-_SYS_NVPAIR_H int nvpair_value_uint8(nvpair_t , uint8_t );
	-_SYS_NVPAIR_H int nvpair_value_int16(nvpair_t , int16_t );
	-_SYS_NVPAIR_H int nvpair_value_uint16(nvpair_t , uint16_t );
	-_SYS_NVPAIR_H int nvpair_value_int32(nvpair_t , int32_t );
	-_SYS_NVPAIR_H int nvpair_value_uint32(nvpair_t , uint32_t );
	-_SYS_NVPAIR_H int nvpair_value_int64(nvpair_t , int64_t );
	-_SYS_NVPAIR_H int nvpair_value_uint64(nvpair_t , uint64_t );
	+_SYS_NVPAIR_H nvpair_t nvlist_next_nvpair(nvlist_t , const nvpair_t *);
	+_SYS_NVPAIR_H nvpair_t nvlist_prev_nvpair(nvlist_t , const nvpair_t *);
	+_SYS_NVPAIR_H char nvpair_name(const nvpair_t );
	+_SYS_NVPAIR_H data_type_t nvpair_type(const nvpair_t *);
	+_SYS_NVPAIR_H int nvpair_type_is_array(const nvpair_t *);
	+_SYS_NVPAIR_H int nvpair_value_boolean_value(const nvpair_t , boolean_t );
	+_SYS_NVPAIR_H int nvpair_value_byte(const nvpair_t , uchar_t );
	+_SYS_NVPAIR_H int nvpair_value_int8(const nvpair_t , int8_t );
	+_SYS_NVPAIR_H int nvpair_value_uint8(const nvpair_t , uint8_t );
	+_SYS_NVPAIR_H int nvpair_value_int16(const nvpair_t , int16_t );
	+_SYS_NVPAIR_H int nvpair_value_uint16(const nvpair_t , uint16_t );
	+_SYS_NVPAIR_H int nvpair_value_int32(const nvpair_t , int32_t );
	+_SYS_NVPAIR_H int nvpair_value_uint32(const nvpair_t , uint32_t );
	+_SYS_NVPAIR_H int nvpair_value_int64(const nvpair_t , int64_t );
	+_SYS_NVPAIR_H int nvpair_value_uint64(const nvpair_t , uint64_t );
	_SYS_NVPAIR_H int nvpair_value_string(nvpair_t , char *);
	_SYS_NVPAIR_H int nvpair_value_nvlist(nvpair_t , nvlist_t *);
	_SYS_NVPAIR_H int nvpair_value_boolean_array(nvpair_t , boolean_t *,
	uint_t *);
	_SYS_NVPAIR_H int nvpair_value_byte_array(nvpair_t , uchar_t , uint_t );
	_SYS_NVPAIR_H int nvpair_value_int8_array(nvpair_t , int8_t , uint_t );
	_SYS_NVPAIR_H int nvpair_value_uint8_array(nvpair_t , uint8_t , uint_t );
	_SYS_NVPAIR_H int nvpair_value_int16_array(nvpair_t , int16_t , uint_t );
	_SYS_NVPAIR_H int nvpair_value_uint16_array(nvpair_t , uint16_t , uint_t );
	_SYS_NVPAIR_H int nvpair_value_int32_array(nvpair_t , int32_t , uint_t );
	_SYS_NVPAIR_H int nvpair_value_uint32_array(nvpair_t , uint32_t , uint_t );
	_SYS_NVPAIR_H int nvpair_value_int64_array(nvpair_t , int64_t , uint_t );
	_SYS_NVPAIR_H int nvpair_value_uint64_array(nvpair_t , uint64_t , uint_t );
	_SYS_NVPAIR_H int nvpair_value_string_array(nvpair_t , char *, uint_t );
	_SYS_NVPAIR_H int nvpair_value_nvlist_array(nvpair_t , nvlist_t *, uint_t );
	_SYS_NVPAIR_H int nvpair_value_hrtime(nvpair_t , hrtime_t );
	#if !defined(_KERNEL) && !defined(_STANDALONE)
	-_SYS_NVPAIR_H int nvpair_value_double(nvpair_t , double );
	+_SYS_NVPAIR_H int nvpair_value_double(const nvpair_t , double );
	#endif

	_SYS_NVPAIR_H nvlist_t *fnvlist_alloc(void);
	_SYS_NVPAIR_H void fnvlist_free(nvlist_t *);
	_SYS_NVPAIR_H size_t fnvlist_size(nvlist_t *);
	_SYS_NVPAIR_H char fnvlist_pack(nvlist_t , size_t *);
	_SYS_NVPAIR_H void fnvlist_pack_free(char *, size_t);
	_SYS_NVPAIR_H nvlist_t fnvlist_unpack(char , size_t);
	-_SYS_NVPAIR_H nvlist_t fnvlist_dup(nvlist_t );
	+_SYS_NVPAIR_H nvlist_t fnvlist_dup(const nvlist_t );
	_SYS_NVPAIR_H void fnvlist_merge(nvlist_t , nvlist_t );
	_SYS_NVPAIR_H size_t fnvlist_num_pairs(nvlist_t *);

	_SYS_NVPAIR_H void fnvlist_add_boolean(nvlist_t , const char );
	_SYS_NVPAIR_H void fnvlist_add_boolean_value(nvlist_t , const char ,
	boolean_t);
	_SYS_NVPAIR_H void fnvlist_add_byte(nvlist_t , const char , uchar_t);
	_SYS_NVPAIR_H void fnvlist_add_int8(nvlist_t , const char , int8_t);
	_SYS_NVPAIR_H void fnvlist_add_uint8(nvlist_t , const char , uint8_t);
	_SYS_NVPAIR_H void fnvlist_add_int16(nvlist_t , const char , int16_t);
	_SYS_NVPAIR_H void fnvlist_add_uint16(nvlist_t , const char , uint16_t);
	_SYS_NVPAIR_H void fnvlist_add_int32(nvlist_t , const char , int32_t);
	_SYS_NVPAIR_H void fnvlist_add_uint32(nvlist_t , const char , uint32_t);
	_SYS_NVPAIR_H void fnvlist_add_int64(nvlist_t , const char , int64_t);
	_SYS_NVPAIR_H void fnvlist_add_uint64(nvlist_t , const char , uint64_t);
	_SYS_NVPAIR_H void fnvlist_add_string(nvlist_t , const char , const char *);
	_SYS_NVPAIR_H void fnvlist_add_nvlist(nvlist_t , const char , nvlist_t *);
	_SYS_NVPAIR_H void fnvlist_add_nvpair(nvlist_t , nvpair_t );
	_SYS_NVPAIR_H void fnvlist_add_boolean_array(nvlist_t , const char ,
	- boolean_t *, uint_t);
	-_SYS_NVPAIR_H void fnvlist_add_byte_array(nvlist_t , const char , uchar_t *,
	- uint_t);
	-_SYS_NVPAIR_H void fnvlist_add_int8_array(nvlist_t , const char , int8_t *,
	- uint_t);
	-_SYS_NVPAIR_H void fnvlist_add_uint8_array(nvlist_t , const char , uint8_t *,
	- uint_t);
	-_SYS_NVPAIR_H void fnvlist_add_int16_array(nvlist_t , const char , int16_t *,
	- uint_t);
	+ const boolean_t *, uint_t);
	+_SYS_NVPAIR_H void fnvlist_add_byte_array(nvlist_t , const char ,
	+ const uchar_t *, uint_t);
	+_SYS_NVPAIR_H void fnvlist_add_int8_array(nvlist_t , const char ,
	+ const int8_t *, uint_t);
	+_SYS_NVPAIR_H void fnvlist_add_uint8_array(nvlist_t , const char ,
	+ const uint8_t *, uint_t);
	+_SYS_NVPAIR_H void fnvlist_add_int16_array(nvlist_t , const char ,
	+ const int16_t *, uint_t);
	_SYS_NVPAIR_H void fnvlist_add_uint16_array(nvlist_t , const char ,
	- uint16_t *, uint_t);
	-_SYS_NVPAIR_H void fnvlist_add_int32_array(nvlist_t , const char , int32_t *,
	- uint_t);
	+ const uint16_t *, uint_t);
	+_SYS_NVPAIR_H void fnvlist_add_int32_array(nvlist_t , const char ,
	+ const int32_t *, uint_t);
	_SYS_NVPAIR_H void fnvlist_add_uint32_array(nvlist_t , const char ,
	- uint32_t *, uint_t);
	-_SYS_NVPAIR_H void fnvlist_add_int64_array(nvlist_t , const char , int64_t *,
	- uint_t);
	+ const uint32_t *, uint_t);
	+_SYS_NVPAIR_H void fnvlist_add_int64_array(nvlist_t , const char ,
	+ const int64_t *, uint_t);
	_SYS_NVPAIR_H void fnvlist_add_uint64_array(nvlist_t , const char ,
	- uint64_t *, uint_t);
	+ const uint64_t *, uint_t);
	_SYS_NVPAIR_H void fnvlist_add_string_array(nvlist_t , const char ,
	- char * const *, uint_t);
	+ const char * const *, uint_t);
	_SYS_NVPAIR_H void fnvlist_add_nvlist_array(nvlist_t , const char ,
	- nvlist_t **, uint_t);
	+ const nvlist_t * const *, uint_t);

	_SYS_NVPAIR_H void fnvlist_remove(nvlist_t , const char );
	_SYS_NVPAIR_H void fnvlist_remove_nvpair(nvlist_t , nvpair_t );

	_SYS_NVPAIR_H nvpair_t fnvlist_lookup_nvpair(nvlist_t , const char *);
	-_SYS_NVPAIR_H boolean_t fnvlist_lookup_boolean(nvlist_t , const char );
	-_SYS_NVPAIR_H boolean_t fnvlist_lookup_boolean_value(nvlist_t , const char );
	-_SYS_NVPAIR_H uchar_t fnvlist_lookup_byte(nvlist_t , const char );
	-_SYS_NVPAIR_H int8_t fnvlist_lookup_int8(nvlist_t , const char );
	-_SYS_NVPAIR_H int16_t fnvlist_lookup_int16(nvlist_t , const char );
	-_SYS_NVPAIR_H int32_t fnvlist_lookup_int32(nvlist_t , const char );
	-_SYS_NVPAIR_H int64_t fnvlist_lookup_int64(nvlist_t , const char );
	-_SYS_NVPAIR_H uint8_t fnvlist_lookup_uint8(nvlist_t , const char );
	-_SYS_NVPAIR_H uint16_t fnvlist_lookup_uint16(nvlist_t , const char );
	-_SYS_NVPAIR_H uint32_t fnvlist_lookup_uint32(nvlist_t , const char );
	-_SYS_NVPAIR_H uint64_t fnvlist_lookup_uint64(nvlist_t , const char );
	+_SYS_NVPAIR_H boolean_t fnvlist_lookup_boolean(const nvlist_t , const char );
	+_SYS_NVPAIR_H boolean_t fnvlist_lookup_boolean_value(const nvlist_t *,
	+ const char *);
	+_SYS_NVPAIR_H uchar_t fnvlist_lookup_byte(const nvlist_t , const char );
	+_SYS_NVPAIR_H int8_t fnvlist_lookup_int8(const nvlist_t , const char );
	+_SYS_NVPAIR_H int16_t fnvlist_lookup_int16(const nvlist_t , const char );
	+_SYS_NVPAIR_H int32_t fnvlist_lookup_int32(const nvlist_t , const char );
	+_SYS_NVPAIR_H int64_t fnvlist_lookup_int64(const nvlist_t , const char );
	+_SYS_NVPAIR_H uint8_t fnvlist_lookup_uint8(const nvlist_t , const char );
	+_SYS_NVPAIR_H uint16_t fnvlist_lookup_uint16(const nvlist_t , const char );
	+_SYS_NVPAIR_H uint32_t fnvlist_lookup_uint32(const nvlist_t , const char );
	+_SYS_NVPAIR_H uint64_t fnvlist_lookup_uint64(const nvlist_t , const char );
	_SYS_NVPAIR_H char fnvlist_lookup_string(nvlist_t , const char *);
	_SYS_NVPAIR_H nvlist_t fnvlist_lookup_nvlist(nvlist_t , const char *);
	_SYS_NVPAIR_H boolean_t fnvlist_lookup_boolean_array(nvlist_t , const char *,
	uint_t *);
	_SYS_NVPAIR_H uchar_t fnvlist_lookup_byte_array(nvlist_t , const char *,
	uint_t *);
	_SYS_NVPAIR_H int8_t fnvlist_lookup_int8_array(nvlist_t , const char *,
	uint_t *);
	_SYS_NVPAIR_H uint8_t fnvlist_lookup_uint8_array(nvlist_t , const char *,
	uint_t *);
	_SYS_NVPAIR_H int16_t fnvlist_lookup_int16_array(nvlist_t , const char *,
	uint_t *);
	_SYS_NVPAIR_H uint16_t fnvlist_lookup_uint16_array(nvlist_t , const char *,
	uint_t *);
	_SYS_NVPAIR_H int32_t fnvlist_lookup_int32_array(nvlist_t , const char *,
	uint_t *);
	_SYS_NVPAIR_H uint32_t fnvlist_lookup_uint32_array(nvlist_t , const char *,
	uint_t *);
	_SYS_NVPAIR_H int64_t fnvlist_lookup_int64_array(nvlist_t , const char *,
	uint_t *);
	_SYS_NVPAIR_H uint64_t fnvlist_lookup_uint64_array(nvlist_t , const char *,
	uint_t *);

	-_SYS_NVPAIR_H boolean_t fnvpair_value_boolean_value(nvpair_t *nvp);
	-_SYS_NVPAIR_H uchar_t fnvpair_value_byte(nvpair_t *nvp);
	-_SYS_NVPAIR_H int8_t fnvpair_value_int8(nvpair_t *nvp);
	-_SYS_NVPAIR_H int16_t fnvpair_value_int16(nvpair_t *nvp);
	-_SYS_NVPAIR_H int32_t fnvpair_value_int32(nvpair_t *nvp);
	-_SYS_NVPAIR_H int64_t fnvpair_value_int64(nvpair_t *nvp);
	-_SYS_NVPAIR_H uint8_t fnvpair_value_uint8(nvpair_t *nvp);
	-_SYS_NVPAIR_H uint16_t fnvpair_value_uint16(nvpair_t *nvp);
	-_SYS_NVPAIR_H uint32_t fnvpair_value_uint32(nvpair_t *nvp);
	-_SYS_NVPAIR_H uint64_t fnvpair_value_uint64(nvpair_t *nvp);
	+_SYS_NVPAIR_H boolean_t fnvpair_value_boolean_value(const nvpair_t *nvp);
	+_SYS_NVPAIR_H uchar_t fnvpair_value_byte(const nvpair_t *nvp);
	+_SYS_NVPAIR_H int8_t fnvpair_value_int8(const nvpair_t *nvp);
	+_SYS_NVPAIR_H int16_t fnvpair_value_int16(const nvpair_t *nvp);
	+_SYS_NVPAIR_H int32_t fnvpair_value_int32(const nvpair_t *nvp);
	+_SYS_NVPAIR_H int64_t fnvpair_value_int64(const nvpair_t *nvp);
	+_SYS_NVPAIR_H uint8_t fnvpair_value_uint8(const nvpair_t *nvp);
	+_SYS_NVPAIR_H uint16_t fnvpair_value_uint16(const nvpair_t *nvp);
	+_SYS_NVPAIR_H uint32_t fnvpair_value_uint32(const nvpair_t *nvp);
	+_SYS_NVPAIR_H uint64_t fnvpair_value_uint64(const nvpair_t *nvp);
	_SYS_NVPAIR_H char fnvpair_value_string(nvpair_t nvp);
	_SYS_NVPAIR_H nvlist_t fnvpair_value_nvlist(nvpair_t nvp);

	#ifdef __cplusplus
	}
	#endif

	#endif /* _SYS_NVPAIR_H */
	diff --git a/sys/contrib/openzfs/include/sys/nvpair_impl.h b/sys/contrib/openzfs/include/sys/nvpair_impl.h
	index c9874b3e4db7..809e5c454712 100644
	--- a/sys/contrib/openzfs/include/sys/nvpair_impl.h
	+++ b/sys/contrib/openzfs/include/sys/nvpair_impl.h
	@@ -1,90 +1,90 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License, Version 1.0 only
	* (the "License"). You may not use this file except in compliance
	* with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright 2004 Sun Microsystems, Inc. All rights reserved.
	* Use is subject to license terms.
	*/

	/*
	* Copyright (c) 2017 by Delphix. All rights reserved.
	*/

	#ifndef _NVPAIR_IMPL_H
	#define _NVPAIR_IMPL_H

	#ifdef __cplusplus
	extern "C" {
	#endif

	#include <sys/nvpair.h>

	/*
	* The structures here provided for information and debugging purposes only
	* may be changed in the future.
	*/

	/*
	* implementation linked list for pre-packed data
	*/
	typedef struct i_nvp i_nvp_t;

	struct i_nvp {
	union {
	/* ensure alignment */
	uint64_t _nvi_align;

	struct {
	/* pointer to next nvpair */
	i_nvp_t *_nvi_next;

	/* pointer to prev nvpair */
	i_nvp_t *_nvi_prev;

	/* next pair in table bucket */
	i_nvp_t *_nvi_hashtable_next;
	} _nvi;
	} _nvi_un;

	/* nvpair */
	nvpair_t nvi_nvp;
	};
	#define nvi_next _nvi_un._nvi._nvi_next
	#define nvi_prev _nvi_un._nvi._nvi_prev
	#define nvi_hashtable_next _nvi_un._nvi._nvi_hashtable_next

	typedef struct {
	i_nvp_t nvp_list; / linked list of nvpairs */
	i_nvp_t nvp_last; / last nvpair */
	- i_nvp_t nvp_curr; / current walker nvpair */
	+ const i_nvp_t nvp_curr; / current walker nvpair */
	nv_alloc_t nvp_nva; / pluggable allocator */
	uint32_t nvp_stat; /* internal state */

	i_nvp_t *nvp_hashtable; / table of entries used for lookup */
	uint32_t nvp_nbuckets; /* # of buckets in hash table */
	uint32_t nvp_nentries; /* # of entries in hash table */
	} nvpriv_t;

	#ifdef __cplusplus
	}
	#endif

	#endif /* _NVPAIR_IMPL_H */
	diff --git a/sys/contrib/openzfs/include/sys/spa.h b/sys/contrib/openzfs/include/sys/spa.h
	index a55dbd66d8d7..2e365eabe21d 100644
	--- a/sys/contrib/openzfs/include/sys/spa.h
	+++ b/sys/contrib/openzfs/include/sys/spa.h
	@@ -1,1214 +1,1215 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2021 by Delphix. All rights reserved.
	* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
	* Copyright 2013 Saso Kiselkov. All rights reserved.
	* Copyright (c) 2014 Integros [integros.com]
	* Copyright 2017 Joyent, Inc.
	* Copyright (c) 2017, 2019, Datto Inc. All rights reserved.
	* Copyright (c) 2017, Intel Corporation.
	* Copyright (c) 2019, Allan Jude
	* Copyright (c) 2019, Klara Inc.
	*/

	#ifndef _SYS_SPA_H
	#define _SYS_SPA_H

	#include <sys/avl.h>
	#include <sys/zfs_context.h>
	#include <sys/kstat.h>
	#include <sys/nvpair.h>
	#include <sys/sysmacros.h>
	#include <sys/types.h>
	#include <sys/fs/zfs.h>
	#include <sys/spa_checksum.h>
	#include <sys/dmu.h>
	#include <sys/space_map.h>
	#include <sys/bitops.h>

	#ifdef __cplusplus
	extern "C" {
	#endif

	/*
	* Forward references that lots of things need.
	*/
	typedef struct spa spa_t;
	typedef struct vdev vdev_t;
	typedef struct metaslab metaslab_t;
	typedef struct metaslab_group metaslab_group_t;
	typedef struct metaslab_class metaslab_class_t;
	typedef struct zio zio_t;
	typedef struct zilog zilog_t;
	typedef struct spa_aux_vdev spa_aux_vdev_t;
	typedef struct ddt ddt_t;
	typedef struct ddt_entry ddt_entry_t;
	typedef struct zbookmark_phys zbookmark_phys_t;

	struct bpobj;
	struct bplist;
	struct dsl_pool;
	struct dsl_dataset;
	struct dsl_crypto_params;

	/*
	* Alignment Shift (ashift) is an immutable, internal top-level vdev property
	* which can only be set at vdev creation time. Physical writes are always done
	* according to it, which makes 2^ashift the smallest possible IO on a vdev.
	*
	* We currently allow values ranging from 512 bytes (2^9 = 512) to 64 KiB
	* (2^16 = 65,536).
	*/
	#define ASHIFT_MIN 9
	#define ASHIFT_MAX 16

	/*
	* Size of block to hold the configuration data (a packed nvlist)
	*/
	#define SPA_CONFIG_BLOCKSIZE (1ULL << 14)

	/*
	* The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB.
	* The ASIZE encoding should be at least 64 times larger (6 more bits)
	* to support up to 4-way RAID-Z mirror mode with worst-case gang block
	* overhead, three DVAs per bp, plus one more bit in case we do anything
	* else that expands the ASIZE.
	*/
	#define SPA_LSIZEBITS 16 /* LSIZE up to 32M (2^16 * 512) */
	#define SPA_PSIZEBITS 16 /* PSIZE up to 32M (2^16 * 512) */
	#define SPA_ASIZEBITS 24 /* ASIZE up to 64 times larger */

	#define SPA_COMPRESSBITS 7
	#define SPA_VDEVBITS 24
	#define SPA_COMPRESSMASK ((1U << SPA_COMPRESSBITS) - 1)

	/*
	* All SPA data is represented by 128-bit data virtual addresses (DVAs).
	* The members of the dva_t should be considered opaque outside the SPA.
	*/
	typedef struct dva {
	uint64_t dva_word[2];
	} dva_t;


	/*
	* Some checksums/hashes need a 256-bit initialization salt. This salt is kept
	* secret and is suitable for use in MAC algorithms as the key.
	*/
	typedef struct zio_cksum_salt {
	uint8_t zcs_bytes[32];
	} zio_cksum_salt_t;

	/*
	* Each block is described by its DVAs, time of birth, checksum, etc.
	* The word-by-word, bit-by-bit layout of the blkptr is as follows:
	*
	* 64 56 48 40 32 24 16 8 0
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 0 \| pad \| vdev1 \| GRID \| ASIZE \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 1 \|G\| offset1 \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 2 \| pad \| vdev2 \| GRID \| ASIZE \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 3 \|G\| offset2 \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 4 \| pad \| vdev3 \| GRID \| ASIZE \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 5 \|G\| offset3 \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 6 \|BDX\|lvl\| type \| cksum \|E\| comp\| PSIZE \| LSIZE \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 7 \| padding \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 8 \| padding \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 9 \| physical birth txg \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* a \| logical birth txg \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* b \| fill count \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* c \| checksum[0] \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* d \| checksum[1] \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* e \| checksum[2] \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* f \| checksum[3] \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	*
	* Legend:
	*
	* vdev virtual device ID
	* offset offset into virtual device
	* LSIZE logical size
	* PSIZE physical size (after compression)
	* ASIZE allocated size (including RAID-Z parity and gang block headers)
	* GRID RAID-Z layout information (reserved for future use)
	* cksum checksum function
	* comp compression function
	* G gang block indicator
	* B byteorder (endianness)
	* D dedup
	* X encryption
	* E blkptr_t contains embedded data (see below)
	* lvl level of indirection
	* type DMU object type
	* phys birth txg when dva[0] was written; zero if same as logical birth txg
	* note that typically all the dva's would be written in this
	* txg, but they could be different if they were moved by
	* device removal.
	* log. birth transaction group in which the block was logically born
	* fill count number of non-zero blocks under this bp
	* checksum[4] 256-bit checksum of the data this bp describes
	*/

	/*
	* The blkptr_t's of encrypted blocks also need to store the encryption
	* parameters so that the block can be decrypted. This layout is as follows:
	*
	* 64 56 48 40 32 24 16 8 0
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 0 \| vdev1 \| GRID \| ASIZE \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 1 \|G\| offset1 \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 2 \| vdev2 \| GRID \| ASIZE \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 3 \|G\| offset2 \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 4 \| salt \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 5 \| IV1 \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 6 \|BDX\|lvl\| type \| cksum \|E\| comp\| PSIZE \| LSIZE \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 7 \| padding \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 8 \| padding \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 9 \| physical birth txg \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* a \| logical birth txg \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* b \| IV2 \| fill count \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* c \| checksum[0] \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* d \| checksum[1] \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* e \| MAC[0] \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* f \| MAC[1] \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	*
	* Legend:
	*
	* salt Salt for generating encryption keys
	* IV1 First 64 bits of encryption IV
	* X Block requires encryption handling (set to 1)
	* E blkptr_t contains embedded data (set to 0, see below)
	* fill count number of non-zero blocks under this bp (truncated to 32 bits)
	* IV2 Last 32 bits of encryption IV
	* checksum[2] 128-bit checksum of the data this bp describes
	* MAC[2] 128-bit message authentication code for this data
	*
	* The X bit being set indicates that this block is one of 3 types. If this is
	* a level 0 block with an encrypted object type, the block is encrypted
	* (see BP_IS_ENCRYPTED()). If this is a level 0 block with an unencrypted
	* object type, this block is authenticated with an HMAC (see
	* BP_IS_AUTHENTICATED()). Otherwise (if level > 0), this bp will use the MAC
	* words to store a checksum-of-MACs from the level below (see
	* BP_HAS_INDIRECT_MAC_CKSUM()). For convenience in the code, BP_IS_PROTECTED()
	* refers to both encrypted and authenticated blocks and BP_USES_CRYPT()
	* refers to any of these 3 kinds of blocks.
	*
	* The additional encryption parameters are the salt, IV, and MAC which are
	* explained in greater detail in the block comment at the top of zio_crypt.c.
	* The MAC occupies half of the checksum space since it serves a very similar
	* purpose: to prevent data corruption on disk. The only functional difference
	* is that the checksum is used to detect on-disk corruption whether or not the
	* encryption key is loaded and the MAC provides additional protection against
	* malicious disk tampering. We use the 3rd DVA to store the salt and first
	* 64 bits of the IV. As a result encrypted blocks can only have 2 copies
	* maximum instead of the normal 3. The last 32 bits of the IV are stored in
	* the upper bits of what is usually the fill count. Note that only blocks at
	* level 0 or -2 are ever encrypted, which allows us to guarantee that these
	* 32 bits are not trampled over by other code (see zio_crypt.c for details).
	* The salt and IV are not used for authenticated bps or bps with an indirect
	* MAC checksum, so these blocks can utilize all 3 DVAs and the full 64 bits
	* for the fill count.
	*/

	/*
	* "Embedded" blkptr_t's don't actually point to a block, instead they
	* have a data payload embedded in the blkptr_t itself. See the comment
	* in blkptr.c for more details.
	*
	* The blkptr_t is laid out as follows:
	*
	* 64 56 48 40 32 24 16 8 0
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 0 \| payload \|
	* 1 \| payload \|
	* 2 \| payload \|
	* 3 \| payload \|
	* 4 \| payload \|
	* 5 \| payload \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 6 \|BDX\|lvl\| type \| etype \|E\| comp\| PSIZE\| LSIZE \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* 7 \| payload \|
	* 8 \| payload \|
	* 9 \| payload \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* a \| logical birth txg \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	* b \| payload \|
	* c \| payload \|
	* d \| payload \|
	* e \| payload \|
	* f \| payload \|
	* +-------+-------+-------+-------+-------+-------+-------+-------+
	*
	* Legend:
	*
	* payload contains the embedded data
	* B (byteorder) byteorder (endianness)
	* D (dedup) padding (set to zero)
	* X encryption (set to zero)
	* E (embedded) set to one
	* lvl indirection level
	* type DMU object type
	* etype how to interpret embedded data (BP_EMBEDDED_TYPE_*)
	* comp compression function of payload
	* PSIZE size of payload after compression, in bytes
	* LSIZE logical size of payload, in bytes
	* note that 25 bits is enough to store the largest
	* "normal" BP's LSIZE (2^16 * 2^9) in bytes
	* log. birth transaction group in which the block was logically born
	*
	* Note that LSIZE and PSIZE are stored in bytes, whereas for non-embedded
	* bp's they are stored in units of SPA_MINBLOCKSHIFT.
	* Generally, the generic BP_GET_*() macros can be used on embedded BP's.
	* The B, D, X, lvl, type, and comp fields are stored the same as with normal
	* BP's so the BP_SET_* macros can be used with them. etype, PSIZE, LSIZE must
	* be set with the BPE_SET_* macros. BP_SET_EMBEDDED() should be called before
	* other macros, as they assert that they are only used on BP's of the correct
	* "embedded-ness". Encrypted blkptr_t's cannot be embedded because they use
	* the payload space for encryption parameters (see the comment above on
	* how encryption parameters are stored).
	*/

	#define BPE_GET_ETYPE(bp) \
	(ASSERT(BP_IS_EMBEDDED(bp)), \
	BF64_GET((bp)->blk_prop, 40, 8))
	#define BPE_SET_ETYPE(bp, t) do { \
	ASSERT(BP_IS_EMBEDDED(bp)); \
	BF64_SET((bp)->blk_prop, 40, 8, t); \
	} while (0)

	#define BPE_GET_LSIZE(bp) \
	(ASSERT(BP_IS_EMBEDDED(bp)), \
	BF64_GET_SB((bp)->blk_prop, 0, 25, 0, 1))
	#define BPE_SET_LSIZE(bp, x) do { \
	ASSERT(BP_IS_EMBEDDED(bp)); \
	BF64_SET_SB((bp)->blk_prop, 0, 25, 0, 1, x); \
	} while (0)

	#define BPE_GET_PSIZE(bp) \
	(ASSERT(BP_IS_EMBEDDED(bp)), \
	BF64_GET_SB((bp)->blk_prop, 25, 7, 0, 1))
	#define BPE_SET_PSIZE(bp, x) do { \
	ASSERT(BP_IS_EMBEDDED(bp)); \
	BF64_SET_SB((bp)->blk_prop, 25, 7, 0, 1, x); \
	} while (0)

	typedef enum bp_embedded_type {
	BP_EMBEDDED_TYPE_DATA,
	BP_EMBEDDED_TYPE_RESERVED, /* Reserved for Delphix byteswap feature. */
	BP_EMBEDDED_TYPE_REDACTED,
	NUM_BP_EMBEDDED_TYPES
	} bp_embedded_type_t;

	#define BPE_NUM_WORDS 14
	#define BPE_PAYLOAD_SIZE (BPE_NUM_WORDS * sizeof (uint64_t))
	#define BPE_IS_PAYLOADWORD(bp, wp) \
	((wp) != &(bp)->blk_prop && (wp) != &(bp)->blk_birth)

	#define SPA_BLKPTRSHIFT 7 /* blkptr_t is 128 bytes */
	#define SPA_DVAS_PER_BP 3 /* Number of DVAs in a bp */
	#define SPA_SYNC_MIN_VDEVS 3 /* min vdevs to update during sync */

	/*
	* A block is a hole when it has either 1) never been written to, or
	* 2) is zero-filled. In both cases, ZFS can return all zeroes for all reads
	* without physically allocating disk space. Holes are represented in the
	* blkptr_t structure by zeroed blk_dva. Correct checking for holes is
	* done through the BP_IS_HOLE macro. For holes, the logical size, level,
	* DMU object type, and birth times are all also stored for holes that
	* were written to at some point (i.e. were punched after having been filled).
	*/
	typedef struct blkptr {
	dva_t blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */
	uint64_t blk_prop; /* size, compression, type, etc */
	uint64_t blk_pad[2]; /* Extra space for the future */
	uint64_t blk_phys_birth; /* txg when block was allocated */
	uint64_t blk_birth; /* transaction group at birth */
	uint64_t blk_fill; /* fill count */
	zio_cksum_t blk_cksum; /* 256-bit checksum */
	} blkptr_t;

	/*
	* Macros to get and set fields in a bp or DVA.
	*/

	/*
	* Note, for gang blocks, DVA_GET_ASIZE() is the total space allocated for
	* this gang DVA including its children BP's. The space allocated at this
	* DVA's vdev/offset is vdev_gang_header_asize(vdev).
	*/
	#define DVA_GET_ASIZE(dva) \
	BF64_GET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, SPA_MINBLOCKSHIFT, 0)
	#define DVA_SET_ASIZE(dva, x) \
	BF64_SET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, \
	SPA_MINBLOCKSHIFT, 0, x)

	#define DVA_GET_GRID(dva) BF64_GET((dva)->dva_word[0], 24, 8)
	#define DVA_SET_GRID(dva, x) BF64_SET((dva)->dva_word[0], 24, 8, x)

	#define DVA_GET_VDEV(dva) BF64_GET((dva)->dva_word[0], 32, SPA_VDEVBITS)
	#define DVA_SET_VDEV(dva, x) \
	BF64_SET((dva)->dva_word[0], 32, SPA_VDEVBITS, x)

	#define DVA_GET_OFFSET(dva) \
	BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0)
	#define DVA_SET_OFFSET(dva, x) \
	BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x)

	#define DVA_GET_GANG(dva) BF64_GET((dva)->dva_word[1], 63, 1)
	#define DVA_SET_GANG(dva, x) BF64_SET((dva)->dva_word[1], 63, 1, x)

	#define BP_GET_LSIZE(bp) \
	(BP_IS_EMBEDDED(bp) ? \
	(BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA ? BPE_GET_LSIZE(bp) : 0): \
	BF64_GET_SB((bp)->blk_prop, 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1))
	#define BP_SET_LSIZE(bp, x) do { \
	ASSERT(!BP_IS_EMBEDDED(bp)); \
	BF64_SET_SB((bp)->blk_prop, \
	0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \
	} while (0)

	#define BP_GET_PSIZE(bp) \
	(BP_IS_EMBEDDED(bp) ? 0 : \
	BF64_GET_SB((bp)->blk_prop, 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1))
	#define BP_SET_PSIZE(bp, x) do { \
	ASSERT(!BP_IS_EMBEDDED(bp)); \
	BF64_SET_SB((bp)->blk_prop, \
	16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \
	} while (0)

	#define BP_GET_COMPRESS(bp) \
	BF64_GET((bp)->blk_prop, 32, SPA_COMPRESSBITS)
	#define BP_SET_COMPRESS(bp, x) \
	BF64_SET((bp)->blk_prop, 32, SPA_COMPRESSBITS, x)

	#define BP_IS_EMBEDDED(bp) BF64_GET((bp)->blk_prop, 39, 1)
	#define BP_SET_EMBEDDED(bp, x) BF64_SET((bp)->blk_prop, 39, 1, x)

	#define BP_GET_CHECKSUM(bp) \
	(BP_IS_EMBEDDED(bp) ? ZIO_CHECKSUM_OFF : \
	BF64_GET((bp)->blk_prop, 40, 8))
	#define BP_SET_CHECKSUM(bp, x) do { \
	ASSERT(!BP_IS_EMBEDDED(bp)); \
	BF64_SET((bp)->blk_prop, 40, 8, x); \
	} while (0)

	#define BP_GET_TYPE(bp) BF64_GET((bp)->blk_prop, 48, 8)
	#define BP_SET_TYPE(bp, x) BF64_SET((bp)->blk_prop, 48, 8, x)

	#define BP_GET_LEVEL(bp) BF64_GET((bp)->blk_prop, 56, 5)
	#define BP_SET_LEVEL(bp, x) BF64_SET((bp)->blk_prop, 56, 5, x)

	/* encrypted, authenticated, and MAC cksum bps use the same bit */
	#define BP_USES_CRYPT(bp) BF64_GET((bp)->blk_prop, 61, 1)
	#define BP_SET_CRYPT(bp, x) BF64_SET((bp)->blk_prop, 61, 1, x)

	#define BP_IS_ENCRYPTED(bp) \
	(BP_USES_CRYPT(bp) && \
	BP_GET_LEVEL(bp) <= 0 && \
	DMU_OT_IS_ENCRYPTED(BP_GET_TYPE(bp)))

	#define BP_IS_AUTHENTICATED(bp) \
	(BP_USES_CRYPT(bp) && \
	BP_GET_LEVEL(bp) <= 0 && \
	!DMU_OT_IS_ENCRYPTED(BP_GET_TYPE(bp)))

	#define BP_HAS_INDIRECT_MAC_CKSUM(bp) \
	(BP_USES_CRYPT(bp) && BP_GET_LEVEL(bp) > 0)

	#define BP_IS_PROTECTED(bp) \
	(BP_IS_ENCRYPTED(bp) \|\| BP_IS_AUTHENTICATED(bp))

	#define BP_GET_DEDUP(bp) BF64_GET((bp)->blk_prop, 62, 1)
	#define BP_SET_DEDUP(bp, x) BF64_SET((bp)->blk_prop, 62, 1, x)

	#define BP_GET_BYTEORDER(bp) BF64_GET((bp)->blk_prop, 63, 1)
	#define BP_SET_BYTEORDER(bp, x) BF64_SET((bp)->blk_prop, 63, 1, x)

	#define BP_GET_FREE(bp) BF64_GET((bp)->blk_fill, 0, 1)
	#define BP_SET_FREE(bp, x) BF64_SET((bp)->blk_fill, 0, 1, x)

	#define BP_PHYSICAL_BIRTH(bp) \
	(BP_IS_EMBEDDED(bp) ? 0 : \
	(bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth)

	#define BP_SET_BIRTH(bp, logical, physical) \
	{ \
	ASSERT(!BP_IS_EMBEDDED(bp)); \
	(bp)->blk_birth = (logical); \
	(bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \
	}

	#define BP_GET_FILL(bp) \
	((BP_IS_ENCRYPTED(bp)) ? BF64_GET((bp)->blk_fill, 0, 32) : \
	((BP_IS_EMBEDDED(bp)) ? 1 : (bp)->blk_fill))

	#define BP_SET_FILL(bp, fill) \
	{ \
	if (BP_IS_ENCRYPTED(bp)) \
	BF64_SET((bp)->blk_fill, 0, 32, fill); \
	else \
	(bp)->blk_fill = fill; \
	}

	#define BP_GET_IV2(bp) \
	(ASSERT(BP_IS_ENCRYPTED(bp)), \
	BF64_GET((bp)->blk_fill, 32, 32))
	#define BP_SET_IV2(bp, iv2) \
	{ \
	ASSERT(BP_IS_ENCRYPTED(bp)); \
	BF64_SET((bp)->blk_fill, 32, 32, iv2); \
	}

	#define BP_IS_METADATA(bp) \
	(BP_GET_LEVEL(bp) > 0 \|\| DMU_OT_IS_METADATA(BP_GET_TYPE(bp)))

	#define BP_GET_ASIZE(bp) \
	(BP_IS_EMBEDDED(bp) ? 0 : \
	DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
	DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
	(DVA_GET_ASIZE(&(bp)->blk_dva[2]) * !BP_IS_ENCRYPTED(bp)))

	#define BP_GET_UCSIZE(bp) \
	(BP_IS_METADATA(bp) ? BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp))

	#define BP_GET_NDVAS(bp) \
	(BP_IS_EMBEDDED(bp) ? 0 : \
	!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
	!!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
	(!!DVA_GET_ASIZE(&(bp)->blk_dva[2]) * !BP_IS_ENCRYPTED(bp)))

	#define BP_COUNT_GANG(bp) \
	(BP_IS_EMBEDDED(bp) ? 0 : \
	(DVA_GET_GANG(&(bp)->blk_dva[0]) + \
	DVA_GET_GANG(&(bp)->blk_dva[1]) + \
	(DVA_GET_GANG(&(bp)->blk_dva[2]) * !BP_IS_ENCRYPTED(bp))))

	#define DVA_EQUAL(dva1, dva2) \
	((dva1)->dva_word[1] == (dva2)->dva_word[1] && \
	(dva1)->dva_word[0] == (dva2)->dva_word[0])

	#define BP_EQUAL(bp1, bp2) \
	(BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) && \
	(bp1)->blk_birth == (bp2)->blk_birth && \
	DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) && \
	DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) && \
	DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2]))


	#define DVA_IS_VALID(dva) (DVA_GET_ASIZE(dva) != 0)

	#define BP_IDENTITY(bp) (ASSERT(!BP_IS_EMBEDDED(bp)), &(bp)->blk_dva[0])
	#define BP_IS_GANG(bp) \
	(BP_IS_EMBEDDED(bp) ? B_FALSE : DVA_GET_GANG(BP_IDENTITY(bp)))
	#define DVA_IS_EMPTY(dva) ((dva)->dva_word[0] == 0ULL && \
	(dva)->dva_word[1] == 0ULL)
	#define BP_IS_HOLE(bp) \
	(!BP_IS_EMBEDDED(bp) && DVA_IS_EMPTY(BP_IDENTITY(bp)))

	#define BP_SET_REDACTED(bp) \
	{ \
	BP_SET_EMBEDDED(bp, B_TRUE); \
	BPE_SET_ETYPE(bp, BP_EMBEDDED_TYPE_REDACTED); \
	}
	#define BP_IS_REDACTED(bp) \
	(BP_IS_EMBEDDED(bp) && BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_REDACTED)

	/* BP_IS_RAIDZ(bp) assumes no block compression */
	#define BP_IS_RAIDZ(bp) (DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \
	BP_GET_PSIZE(bp))

	#define BP_ZERO(bp) \
	{ \
	(bp)->blk_dva[0].dva_word[0] = 0; \
	(bp)->blk_dva[0].dva_word[1] = 0; \
	(bp)->blk_dva[1].dva_word[0] = 0; \
	(bp)->blk_dva[1].dva_word[1] = 0; \
	(bp)->blk_dva[2].dva_word[0] = 0; \
	(bp)->blk_dva[2].dva_word[1] = 0; \
	(bp)->blk_prop = 0; \
	(bp)->blk_pad[0] = 0; \
	(bp)->blk_pad[1] = 0; \
	(bp)->blk_phys_birth = 0; \
	(bp)->blk_birth = 0; \
	(bp)->blk_fill = 0; \
	ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0); \
	}

	#ifdef _ZFS_BIG_ENDIAN
	#define ZFS_HOST_BYTEORDER (0ULL)
	#else
	#define ZFS_HOST_BYTEORDER (1ULL)
	#endif

	#define BP_SHOULD_BYTESWAP(bp) (BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER)

	#define BP_SPRINTF_LEN 400

	/*
	* This macro allows code sharing between zfs, libzpool, and mdb.
	* 'func' is either snprintf() or mdb_snprintf().
	* 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line.
	*/

	#define SNPRINTF_BLKPTR(func, ws, buf, size, bp, type, checksum, compress) \
	{ \
	static const char *copyname[] = \
	{ "zero", "single", "double", "triple" }; \
	int len = 0; \
	int copies = 0; \
	const char *crypt_type; \
	if (bp != NULL) { \
	if (BP_IS_ENCRYPTED(bp)) { \
	crypt_type = "encrypted"; \
	/* LINTED E_SUSPICIOUS_COMPARISON */ \
	} else if (BP_IS_AUTHENTICATED(bp)) { \
	crypt_type = "authenticated"; \
	} else if (BP_HAS_INDIRECT_MAC_CKSUM(bp)) { \
	crypt_type = "indirect-MAC"; \
	} else { \
	crypt_type = "unencrypted"; \
	} \
	} \
	if (bp == NULL) { \
	len += func(buf + len, size - len, "<NULL>"); \
	} else if (BP_IS_HOLE(bp)) { \
	len += func(buf + len, size - len, \
	"HOLE [L%llu %s] " \
	"size=%llxL birth=%lluL", \
	(u_longlong_t)BP_GET_LEVEL(bp), \
	type, \
	(u_longlong_t)BP_GET_LSIZE(bp), \
	(u_longlong_t)bp->blk_birth); \
	} else if (BP_IS_EMBEDDED(bp)) { \
	len = func(buf + len, size - len, \
	"EMBEDDED [L%llu %s] et=%u %s " \
	"size=%llxL/%llxP birth=%lluL", \
	(u_longlong_t)BP_GET_LEVEL(bp), \
	type, \
	(int)BPE_GET_ETYPE(bp), \
	compress, \
	(u_longlong_t)BPE_GET_LSIZE(bp), \
	(u_longlong_t)BPE_GET_PSIZE(bp), \
	(u_longlong_t)bp->blk_birth); \
	} else if (BP_IS_REDACTED(bp)) { \
	len += func(buf + len, size - len, \
	"REDACTED [L%llu %s] size=%llxL birth=%lluL", \
	(u_longlong_t)BP_GET_LEVEL(bp), \
	type, \
	(u_longlong_t)BP_GET_LSIZE(bp), \
	(u_longlong_t)bp->blk_birth); \
	} else { \
	for (int d = 0; d < BP_GET_NDVAS(bp); d++) { \
	const dva_t *dva = &bp->blk_dva[d]; \
	if (DVA_IS_VALID(dva)) \
	copies++; \
	len += func(buf + len, size - len, \
	"DVA[%d]=<%llu:%llx:%llx>%c", d, \
	(u_longlong_t)DVA_GET_VDEV(dva), \
	(u_longlong_t)DVA_GET_OFFSET(dva), \
	(u_longlong_t)DVA_GET_ASIZE(dva), \
	ws); \
	} \
	if (BP_IS_ENCRYPTED(bp)) { \
	len += func(buf + len, size - len, \
	"salt=%llx iv=%llx:%llx%c", \
	(u_longlong_t)bp->blk_dva[2].dva_word[0], \
	(u_longlong_t)bp->blk_dva[2].dva_word[1], \
	(u_longlong_t)BP_GET_IV2(bp), \
	ws); \
	} \
	if (BP_IS_GANG(bp) && \
	DVA_GET_ASIZE(&bp->blk_dva[2]) <= \
	DVA_GET_ASIZE(&bp->blk_dva[1]) / 2) \
	copies--; \
	len += func(buf + len, size - len, \
	"[L%llu %s] %s %s %s %s %s %s %s%c" \
	"size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c" \
	"cksum=%llx:%llx:%llx:%llx", \
	(u_longlong_t)BP_GET_LEVEL(bp), \
	type, \
	checksum, \
	compress, \
	crypt_type, \
	BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE", \
	BP_IS_GANG(bp) ? "gang" : "contiguous", \
	BP_GET_DEDUP(bp) ? "dedup" : "unique", \
	copyname[copies], \
	ws, \
	(u_longlong_t)BP_GET_LSIZE(bp), \
	(u_longlong_t)BP_GET_PSIZE(bp), \
	(u_longlong_t)bp->blk_birth, \
	(u_longlong_t)BP_PHYSICAL_BIRTH(bp), \
	(u_longlong_t)BP_GET_FILL(bp), \
	ws, \
	(u_longlong_t)bp->blk_cksum.zc_word[0], \
	(u_longlong_t)bp->blk_cksum.zc_word[1], \
	(u_longlong_t)bp->blk_cksum.zc_word[2], \
	(u_longlong_t)bp->blk_cksum.zc_word[3]); \
	} \
	ASSERT(len < size); \
	}

	#define BP_GET_BUFC_TYPE(bp) \
	(BP_IS_METADATA(bp) ? ARC_BUFC_METADATA : ARC_BUFC_DATA)

	typedef enum spa_import_type {
	SPA_IMPORT_EXISTING,
	SPA_IMPORT_ASSEMBLE
	} spa_import_type_t;

	typedef enum spa_mode {
	SPA_MODE_UNINIT = 0,
	SPA_MODE_READ = 1,
	SPA_MODE_WRITE = 2,
	} spa_mode_t;

	/*
	* Send TRIM commands in-line during normal pool operation while deleting.
	* OFF: no
	* ON: yes
	* NB: IN_FREEBSD_BASE is defined within the FreeBSD sources.
	*/
	typedef enum {
	SPA_AUTOTRIM_OFF = 0, /* default */
	SPA_AUTOTRIM_ON,
	#ifdef IN_FREEBSD_BASE
	SPA_AUTOTRIM_DEFAULT = SPA_AUTOTRIM_ON,
	#else
	SPA_AUTOTRIM_DEFAULT = SPA_AUTOTRIM_OFF,
	#endif
	} spa_autotrim_t;

	/*
	* Reason TRIM command was issued, used internally for accounting purposes.
	*/
	typedef enum trim_type {
	TRIM_TYPE_MANUAL = 0,
	TRIM_TYPE_AUTO = 1,
	TRIM_TYPE_SIMPLE = 2
	} trim_type_t;

	/* state manipulation functions */
	extern int spa_open(const char pool, spa_t , void tag);
	extern int spa_open_rewind(const char pool, spa_t , void tag,
	nvlist_t policy, nvlist_t *config);
	extern int spa_get_stats(const char pool, nvlist_t config, char altroot,
	size_t buflen);
	extern int spa_create(const char pool, nvlist_t nvroot, nvlist_t *props,
	nvlist_t zplprops, struct dsl_crypto_params dcp);
	extern int spa_import(char pool, nvlist_t config, nvlist_t *props,
	uint64_t flags);
	extern nvlist_t spa_tryimport(nvlist_t tryconfig);
	extern int spa_destroy(const char *pool);
	extern int spa_checkpoint(const char *pool);
	extern int spa_checkpoint_discard(const char *pool);
	extern int spa_export(const char pool, nvlist_t *oldconfig, boolean_t force,
	boolean_t hardforce);
	extern int spa_reset(const char *pool);
	extern void spa_async_request(spa_t *spa, int flag);
	extern void spa_async_unrequest(spa_t *spa, int flag);
	extern void spa_async_suspend(spa_t *spa);
	extern void spa_async_resume(spa_t *spa);
	extern int spa_async_tasks(spa_t *spa);
	extern spa_t spa_inject_addref(char pool);
	extern void spa_inject_delref(spa_t *spa);
	extern void spa_scan_stat_init(spa_t *spa);
	extern int spa_scan_get_stats(spa_t spa, pool_scan_stat_t ps);
	extern int bpobj_enqueue_alloc_cb(void arg, const blkptr_t bp, dmu_tx_t *tx);
	extern int bpobj_enqueue_free_cb(void arg, const blkptr_t bp, dmu_tx_t *tx);

	#define SPA_ASYNC_CONFIG_UPDATE 0x01
	#define SPA_ASYNC_REMOVE 0x02
	#define SPA_ASYNC_PROBE 0x04
	#define SPA_ASYNC_RESILVER_DONE 0x08
	#define SPA_ASYNC_RESILVER 0x10
	#define SPA_ASYNC_AUTOEXPAND 0x20
	#define SPA_ASYNC_REMOVE_DONE 0x40
	#define SPA_ASYNC_REMOVE_STOP 0x80
	#define SPA_ASYNC_INITIALIZE_RESTART 0x100
	#define SPA_ASYNC_TRIM_RESTART 0x200
	#define SPA_ASYNC_AUTOTRIM_RESTART 0x400
	#define SPA_ASYNC_L2CACHE_REBUILD 0x800
	#define SPA_ASYNC_L2CACHE_TRIM 0x1000
	#define SPA_ASYNC_REBUILD_DONE 0x2000

	/* device manipulation */
	extern int spa_vdev_add(spa_t spa, nvlist_t nvroot);
	extern int spa_vdev_attach(spa_t spa, uint64_t guid, nvlist_t nvroot,
	int replacing, int rebuild);
	extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid,
	int replace_done);
	-extern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare);
	+extern int spa_vdev_alloc(spa_t *spa, uint64_t guid);
	+extern int spa_vdev_noalloc(spa_t *spa, uint64_t guid);
	extern boolean_t spa_vdev_remove_active(spa_t *spa);
	extern int spa_vdev_initialize(spa_t spa, nvlist_t nv, uint64_t cmd_type,
	nvlist_t *vdev_errlist);
	extern int spa_vdev_trim(spa_t spa, nvlist_t nv, uint64_t cmd_type,
	uint64_t rate, boolean_t partial, boolean_t secure, nvlist_t *vdev_errlist);
	extern int spa_vdev_setpath(spa_t spa, uint64_t guid, const char newpath);
	extern int spa_vdev_setfru(spa_t spa, uint64_t guid, const char newfru);
	extern int spa_vdev_split_mirror(spa_t spa, char newname, nvlist_t *config,
	nvlist_t *props, boolean_t exp);

	/* spare state (which is global across all pools) */
	extern void spa_spare_add(vdev_t *vd);
	extern void spa_spare_remove(vdev_t *vd);
	extern boolean_t spa_spare_exists(uint64_t guid, uint64_t pool, int refcnt);
	extern void spa_spare_activate(vdev_t *vd);

	/* L2ARC state (which is global across all pools) */
	extern void spa_l2cache_add(vdev_t *vd);
	extern void spa_l2cache_remove(vdev_t *vd);
	extern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool);
	extern void spa_l2cache_activate(vdev_t *vd);
	extern void spa_l2cache_drop(spa_t *spa);

	/* scanning */
	extern int spa_scan(spa_t *spa, pool_scan_func_t func);
	extern int spa_scan_stop(spa_t *spa);
	extern int spa_scrub_pause_resume(spa_t *spa, pool_scrub_cmd_t flag);

	/* spa syncing */
	extern void spa_sync(spa_t spa, uint64_t txg); / only for DMU use */
	extern void spa_sync_allpools(void);

	extern int zfs_sync_pass_deferred_free;

	/* spa namespace global mutex */
	extern kmutex_t spa_namespace_lock;

	/*
	* SPA configuration functions in spa_config.c
	*/

	#define SPA_CONFIG_UPDATE_POOL 0
	#define SPA_CONFIG_UPDATE_VDEVS 1

	extern void spa_write_cachefile(spa_t *, boolean_t, boolean_t);
	extern void spa_config_load(void);
	extern nvlist_t spa_all_configs(uint64_t );
	extern void spa_config_set(spa_t spa, nvlist_t config);
	extern nvlist_t spa_config_generate(spa_t spa, vdev_t *vd, uint64_t txg,
	int getstats);
	extern void spa_config_update(spa_t *spa, int what);
	extern int spa_config_parse(spa_t spa, vdev_t vdp, nvlist_t nv,
	vdev_t *parent, uint_t id, int atype);


	/*
	* Miscellaneous SPA routines in spa_misc.c
	*/

	/* Namespace manipulation */
	extern spa_t spa_lookup(const char name);
	extern spa_t spa_add(const char name, nvlist_t config, const char altroot);
	extern void spa_remove(spa_t *spa);
	extern spa_t spa_next(spa_t prev);

	/* Refcount functions */
	extern void spa_open_ref(spa_t spa, void tag);
	extern void spa_close(spa_t spa, void tag);
	extern void spa_async_close(spa_t spa, void tag);
	extern boolean_t spa_refcount_zero(spa_t *spa);

	#define SCL_NONE 0x00
	#define SCL_CONFIG 0x01
	#define SCL_STATE 0x02
	#define SCL_L2ARC 0x04 /* hack until L2ARC 2.0 */
	#define SCL_ALLOC 0x08
	#define SCL_ZIO 0x10
	#define SCL_FREE 0x20
	#define SCL_VDEV 0x40
	#define SCL_LOCKS 7
	#define SCL_ALL ((1 << SCL_LOCKS) - 1)
	#define SCL_STATE_ALL (SCL_STATE \| SCL_L2ARC \| SCL_ZIO)

	/* Historical pool statistics */
	typedef struct spa_history_kstat {
	kmutex_t lock;
	uint64_t count;
	uint64_t size;
	kstat_t *kstat;
	void *priv;
	list_t list;
	} spa_history_kstat_t;

	typedef struct spa_history_list {
	uint64_t size;
	procfs_list_t procfs_list;
	} spa_history_list_t;

	typedef struct spa_stats {
	spa_history_list_t read_history;
	spa_history_list_t txg_history;
	spa_history_kstat_t tx_assign_histogram;
	spa_history_list_t mmp_history;
	spa_history_kstat_t state; /* pool state */
	spa_history_kstat_t iostats;
	} spa_stats_t;

	typedef enum txg_state {
	TXG_STATE_BIRTH = 0,
	TXG_STATE_OPEN = 1,
	TXG_STATE_QUIESCED = 2,
	TXG_STATE_WAIT_FOR_SYNC = 3,
	TXG_STATE_SYNCED = 4,
	TXG_STATE_COMMITTED = 5,
	} txg_state_t;

	typedef struct txg_stat {
	vdev_stat_t vs1;
	vdev_stat_t vs2;
	uint64_t txg;
	uint64_t ndirty;
	} txg_stat_t;

	/* Assorted pool IO kstats */
	typedef struct spa_iostats {
	kstat_named_t trim_extents_written;
	kstat_named_t trim_bytes_written;
	kstat_named_t trim_extents_skipped;
	kstat_named_t trim_bytes_skipped;
	kstat_named_t trim_extents_failed;
	kstat_named_t trim_bytes_failed;
	kstat_named_t autotrim_extents_written;
	kstat_named_t autotrim_bytes_written;
	kstat_named_t autotrim_extents_skipped;
	kstat_named_t autotrim_bytes_skipped;
	kstat_named_t autotrim_extents_failed;
	kstat_named_t autotrim_bytes_failed;
	kstat_named_t simple_trim_extents_written;
	kstat_named_t simple_trim_bytes_written;
	kstat_named_t simple_trim_extents_skipped;
	kstat_named_t simple_trim_bytes_skipped;
	kstat_named_t simple_trim_extents_failed;
	kstat_named_t simple_trim_bytes_failed;
	} spa_iostats_t;

	extern void spa_stats_init(spa_t *spa);
	extern void spa_stats_destroy(spa_t *spa);
	extern void spa_read_history_add(spa_t spa, const zbookmark_phys_t zb,
	uint32_t aflags);
	extern void spa_txg_history_add(spa_t *spa, uint64_t txg, hrtime_t birth_time);
	extern int spa_txg_history_set(spa_t *spa, uint64_t txg,
	txg_state_t completed_state, hrtime_t completed_time);
	extern txg_stat_t spa_txg_history_init_io(spa_t , uint64_t,
	struct dsl_pool *);
	extern void spa_txg_history_fini_io(spa_t , txg_stat_t );
	extern void spa_tx_assign_add_nsecs(spa_t *spa, uint64_t nsecs);
	extern int spa_mmp_history_set_skip(spa_t *spa, uint64_t mmp_kstat_id);
	extern int spa_mmp_history_set(spa_t *spa, uint64_t mmp_kstat_id, int io_error,
	hrtime_t duration);
	extern void spa_mmp_history_add(spa_t *spa, uint64_t txg, uint64_t timestamp,
	uint64_t mmp_delay, vdev_t *vd, int label, uint64_t mmp_kstat_id,
	int error);
	extern void spa_iostats_trim_add(spa_t *spa, trim_type_t type,
	uint64_t extents_written, uint64_t bytes_written,
	uint64_t extents_skipped, uint64_t bytes_skipped,
	uint64_t extents_failed, uint64_t bytes_failed);
	extern void spa_import_progress_add(spa_t *spa);
	extern void spa_import_progress_remove(uint64_t spa_guid);
	extern int spa_import_progress_set_mmp_check(uint64_t pool_guid,
	uint64_t mmp_sec_remaining);
	extern int spa_import_progress_set_max_txg(uint64_t pool_guid,
	uint64_t max_txg);
	extern int spa_import_progress_set_state(uint64_t pool_guid,
	spa_load_state_t spa_load_state);

	/* Pool configuration locks */
	extern int spa_config_tryenter(spa_t spa, int locks, void tag, krw_t rw);
	extern void spa_config_enter(spa_t spa, int locks, const void tag, krw_t rw);
	extern void spa_config_exit(spa_t spa, int locks, const void tag);
	extern int spa_config_held(spa_t *spa, int locks, krw_t rw);

	/* Pool vdev add/remove lock */
	extern uint64_t spa_vdev_enter(spa_t *spa);
	extern uint64_t spa_vdev_detach_enter(spa_t *spa, uint64_t guid);
	extern uint64_t spa_vdev_config_enter(spa_t *spa);
	extern void spa_vdev_config_exit(spa_t spa, vdev_t vd, uint64_t txg,
	int error, char *tag);
	extern int spa_vdev_exit(spa_t spa, vdev_t vd, uint64_t txg, int error);

	/* Pool vdev state change lock */
	extern void spa_vdev_state_enter(spa_t *spa, int oplock);
	extern int spa_vdev_state_exit(spa_t spa, vdev_t vd, int error);

	/* Log state */
	typedef enum spa_log_state {
	SPA_LOG_UNKNOWN = 0, /* unknown log state */
	SPA_LOG_MISSING, /* missing log(s) */
	SPA_LOG_CLEAR, /* clear the log(s) */
	SPA_LOG_GOOD, /* log(s) are good */
	} spa_log_state_t;

	extern spa_log_state_t spa_get_log_state(spa_t *spa);
	extern void spa_set_log_state(spa_t *spa, spa_log_state_t state);
	extern int spa_reset_logs(spa_t *spa);

	/* Log claim callback */
	extern void spa_claim_notify(zio_t *zio);
	extern void spa_deadman(void *);

	/* Accessor functions */
	extern boolean_t spa_shutting_down(spa_t *spa);
	extern struct dsl_pool spa_get_dsl(spa_t spa);
	extern boolean_t spa_is_initializing(spa_t *spa);
	extern boolean_t spa_indirect_vdevs_loaded(spa_t *spa);
	extern blkptr_t spa_get_rootblkptr(spa_t spa);
	extern void spa_set_rootblkptr(spa_t spa, const blkptr_t bp);
	extern void spa_altroot(spa_t , char , size_t);
	extern int spa_sync_pass(spa_t *spa);
	extern char spa_name(spa_t spa);
	extern uint64_t spa_guid(spa_t *spa);
	extern uint64_t spa_load_guid(spa_t *spa);
	extern uint64_t spa_last_synced_txg(spa_t *spa);
	extern uint64_t spa_first_txg(spa_t *spa);
	extern uint64_t spa_syncing_txg(spa_t *spa);
	extern uint64_t spa_final_dirty_txg(spa_t *spa);
	extern uint64_t spa_version(spa_t *spa);
	extern pool_state_t spa_state(spa_t *spa);
	extern spa_load_state_t spa_load_state(spa_t *spa);
	extern uint64_t spa_freeze_txg(spa_t *spa);
	extern uint64_t spa_get_worst_case_asize(spa_t *spa, uint64_t lsize);
	extern uint64_t spa_get_dspace(spa_t *spa);
	extern uint64_t spa_get_checkpoint_space(spa_t *spa);
	extern uint64_t spa_get_slop_space(spa_t *spa);
	extern void spa_update_dspace(spa_t *spa);
	extern uint64_t spa_version(spa_t *spa);
	extern boolean_t spa_deflate(spa_t *spa);
	extern metaslab_class_t spa_normal_class(spa_t spa);
	extern metaslab_class_t spa_log_class(spa_t spa);
	extern metaslab_class_t spa_embedded_log_class(spa_t spa);
	extern metaslab_class_t spa_special_class(spa_t spa);
	extern metaslab_class_t spa_dedup_class(spa_t spa);
	extern metaslab_class_t spa_preferred_class(spa_t spa, uint64_t size,
	dmu_object_type_t objtype, uint_t level, uint_t special_smallblk);

	extern void spa_evicting_os_register(spa_t , objset_t os);
	extern void spa_evicting_os_deregister(spa_t , objset_t os);
	extern void spa_evicting_os_wait(spa_t *spa);
	extern int spa_max_replication(spa_t *spa);
	extern int spa_prev_software_version(spa_t *spa);
	extern uint64_t spa_get_failmode(spa_t *spa);
	extern uint64_t spa_get_deadman_failmode(spa_t *spa);
	extern void spa_set_deadman_failmode(spa_t spa, const char failmode);
	extern boolean_t spa_suspended(spa_t *spa);
	extern uint64_t spa_bootfs(spa_t *spa);
	extern uint64_t spa_delegation(spa_t *spa);
	extern objset_t spa_meta_objset(spa_t spa);
	extern space_map_t spa_syncing_log_sm(spa_t spa);
	extern uint64_t spa_deadman_synctime(spa_t *spa);
	extern uint64_t spa_deadman_ziotime(spa_t *spa);
	extern uint64_t spa_dirty_data(spa_t *spa);
	extern spa_autotrim_t spa_get_autotrim(spa_t *spa);

	/* Miscellaneous support routines */
	extern void spa_load_failed(spa_t spa, const char fmt, ...)
	__attribute__((format(printf, 2, 3)));
	extern void spa_load_note(spa_t spa, const char fmt, ...)
	__attribute__((format(printf, 2, 3)));
	extern void spa_activate_mos_feature(spa_t spa, const char feature,
	dmu_tx_t *tx);
	extern void spa_deactivate_mos_feature(spa_t spa, const char feature);
	extern spa_t *spa_by_guid(uint64_t pool_guid, uint64_t device_guid);
	extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid);
	extern char spa_strdup(const char );
	extern void spa_strfree(char *);
	extern uint64_t spa_generate_guid(spa_t *spa);
	extern void snprintf_blkptr(char buf, size_t buflen, const blkptr_t bp);
	extern void spa_freeze(spa_t *spa);
	extern int spa_change_guid(spa_t *spa);
	extern void spa_upgrade(spa_t *spa, uint64_t version);
	extern void spa_evict_all(void);
	extern vdev_t spa_lookup_by_guid(spa_t spa, uint64_t guid,
	boolean_t l2cache);
	extern boolean_t spa_has_l2cache(spa_t *, uint64_t guid);
	extern boolean_t spa_has_spare(spa_t *, uint64_t guid);
	extern uint64_t dva_get_dsize_sync(spa_t spa, const dva_t dva);
	extern uint64_t bp_get_dsize_sync(spa_t spa, const blkptr_t bp);
	extern uint64_t bp_get_dsize(spa_t spa, const blkptr_t bp);
	extern boolean_t spa_has_slogs(spa_t *spa);
	extern boolean_t spa_is_root(spa_t *spa);
	extern boolean_t spa_writeable(spa_t *spa);
	extern boolean_t spa_has_pending_synctask(spa_t *spa);
	extern int spa_maxblocksize(spa_t *spa);
	extern int spa_maxdnodesize(spa_t *spa);
	extern boolean_t spa_has_checkpoint(spa_t *spa);
	extern boolean_t spa_importing_readonly_checkpoint(spa_t *spa);
	extern boolean_t spa_suspend_async_destroy(spa_t *spa);
	extern uint64_t spa_min_claim_txg(spa_t *spa);
	extern boolean_t zfs_dva_valid(spa_t spa, const dva_t dva,
	const blkptr_t *bp);
	typedef void (*spa_remap_cb_t)(uint64_t vdev, uint64_t offset, uint64_t size,
	void *arg);
	extern boolean_t spa_remap_blkptr(spa_t spa, blkptr_t bp,
	spa_remap_cb_t callback, void *arg);
	extern uint64_t spa_get_last_removal_txg(spa_t *spa);
	extern boolean_t spa_trust_config(spa_t *spa);
	extern uint64_t spa_missing_tvds_allowed(spa_t *spa);
	extern void spa_set_missing_tvds(spa_t *spa, uint64_t missing);
	extern boolean_t spa_top_vdevs_spacemap_addressable(spa_t *spa);
	extern uint64_t spa_total_metaslabs(spa_t *spa);
	extern boolean_t spa_multihost(spa_t *spa);
	extern uint32_t spa_get_hostid(spa_t *spa);
	extern void spa_activate_allocation_classes(spa_t , dmu_tx_t );
	extern boolean_t spa_livelist_delete_check(spa_t *spa);

	extern spa_mode_t spa_mode(spa_t *spa);
	extern uint64_t zfs_strtonum(const char str, char *nptr);

	extern char *spa_his_ievent_table[];

	extern void spa_history_create_obj(spa_t spa, dmu_tx_t tx);
	extern int spa_history_get(spa_t spa, uint64_t offset, uint64_t *len_read,
	char *his_buf);
	extern int spa_history_log(spa_t spa, const char his_buf);
	extern int spa_history_log_nvl(spa_t spa, nvlist_t nvl);
	extern void spa_history_log_version(spa_t spa, const char operation,
	dmu_tx_t *tx);
	extern void spa_history_log_internal(spa_t spa, const char operation,
	dmu_tx_t tx, const char fmt, ...) __printflike(4, 5);
	extern void spa_history_log_internal_ds(struct dsl_dataset ds, const char op,
	dmu_tx_t tx, const char fmt, ...) __printflike(4, 5);
	extern void spa_history_log_internal_dd(dsl_dir_t dd, const char operation,
	dmu_tx_t tx, const char fmt, ...) __printflike(4, 5);

	extern const char spa_state_to_name(spa_t spa);

	/* error handling */
	struct zbookmark_phys;
	extern void spa_log_error(spa_t spa, const zbookmark_phys_t zb);
	extern int zfs_ereport_post(const char clazz, spa_t spa, vdev_t *vd,
	const zbookmark_phys_t zb, zio_t zio, uint64_t state);
	extern boolean_t zfs_ereport_is_valid(const char clazz, spa_t spa, vdev_t *vd,
	zio_t *zio);
	extern void zfs_ereport_taskq_fini(void);
	extern void zfs_ereport_clear(spa_t spa, vdev_t vd);
	extern nvlist_t zfs_event_create(spa_t spa, vdev_t vd, const char type,
	const char name, nvlist_t aux);
	extern void zfs_post_remove(spa_t spa, vdev_t vd);
	extern void zfs_post_state_change(spa_t spa, vdev_t vd, uint64_t laststate);
	extern void zfs_post_autoreplace(spa_t spa, vdev_t vd);
	extern uint64_t spa_get_errlog_size(spa_t *spa);
	extern int spa_get_errlog(spa_t spa, void uaddr, size_t *count);
	extern void spa_errlog_rotate(spa_t *spa);
	extern void spa_errlog_drain(spa_t *spa);
	extern void spa_errlog_sync(spa_t *spa, uint64_t txg);
	extern void spa_get_errlists(spa_t spa, avl_tree_t last, avl_tree_t *scrub);

	/* vdev cache */
	extern void vdev_cache_stat_init(void);
	extern void vdev_cache_stat_fini(void);

	/* vdev mirror */
	extern void vdev_mirror_stat_init(void);
	extern void vdev_mirror_stat_fini(void);

	/* Initialization and termination */
	extern void spa_init(spa_mode_t mode);
	extern void spa_fini(void);
	extern void spa_boot_init(void);

	/* properties */
	extern int spa_prop_set(spa_t spa, nvlist_t nvp);
	extern int spa_prop_get(spa_t spa, nvlist_t *nvp);
	extern void spa_prop_clear_bootfs(spa_t spa, uint64_t obj, dmu_tx_t tx);
	extern void spa_configfile_set(spa_t , nvlist_t , boolean_t);

	/* asynchronous event notification */
	extern void spa_event_notify(spa_t spa, vdev_t vdev, nvlist_t *hist_nvl,
	const char *name);
	extern void zfs_ereport_zvol_post(const char subclass, const char name,
	const char device_name, const char raw_name);

	/* waiting for pool activities to complete */
	extern int spa_wait(const char *pool, zpool_wait_activity_t activity,
	boolean_t *waited);
	extern int spa_wait_tag(const char *name, zpool_wait_activity_t activity,
	uint64_t tag, boolean_t *waited);
	extern void spa_notify_waiters(spa_t *spa);
	extern void spa_wake_waiters(spa_t *spa);

	/* module param call functions */
	int param_set_deadman_ziotime(ZFS_MODULE_PARAM_ARGS);
	int param_set_deadman_synctime(ZFS_MODULE_PARAM_ARGS);
	int param_set_slop_shift(ZFS_MODULE_PARAM_ARGS);
	int param_set_deadman_failmode(ZFS_MODULE_PARAM_ARGS);

	#ifdef ZFS_DEBUG
	#define dprintf_bp(bp, fmt, ...) do { \
	if (zfs_flags & ZFS_DEBUG_DPRINTF) { \
	char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_SLEEP); \
	snprintf_blkptr(__blkbuf, BP_SPRINTF_LEN, (bp)); \
	dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf); \
	kmem_free(__blkbuf, BP_SPRINTF_LEN); \
	} \
	} while (0)
	#else
	#define dprintf_bp(bp, fmt, ...)
	#endif

	extern spa_mode_t spa_mode_global;
	extern int zfs_deadman_enabled;
	extern unsigned long zfs_deadman_synctime_ms;
	extern unsigned long zfs_deadman_ziotime_ms;
	extern unsigned long zfs_deadman_checktime_ms;

	#ifdef __cplusplus
	}
	#endif

	#endif /* _SYS_SPA_H */
	diff --git a/sys/contrib/openzfs/include/sys/spa_impl.h b/sys/contrib/openzfs/include/sys/spa_impl.h
	index 9714bbce9c9d..eee4783fe3f8 100644
	--- a/sys/contrib/openzfs/include/sys/spa_impl.h
	+++ b/sys/contrib/openzfs/include/sys/spa_impl.h
	@@ -1,466 +1,467 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2019 by Delphix. All rights reserved.
	* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
	* Copyright 2013 Saso Kiselkov. All rights reserved.
	* Copyright (c) 2016 Actifio, Inc. All rights reserved.
	* Copyright (c) 2017 Datto Inc.
	* Copyright (c) 2017, Intel Corporation.
	*/

	#ifndef _SYS_SPA_IMPL_H
	#define _SYS_SPA_IMPL_H

	#include <sys/spa.h>
	#include <sys/spa_checkpoint.h>
	#include <sys/spa_log_spacemap.h>
	#include <sys/vdev.h>
	#include <sys/vdev_rebuild.h>
	#include <sys/vdev_removal.h>
	#include <sys/metaslab.h>
	#include <sys/dmu.h>
	#include <sys/dsl_pool.h>
	#include <sys/uberblock_impl.h>
	#include <sys/zfs_context.h>
	#include <sys/avl.h>
	#include <sys/zfs_refcount.h>
	#include <sys/bplist.h>
	#include <sys/bpobj.h>
	#include <sys/dsl_crypt.h>
	#include <sys/zfeature.h>
	#include <sys/zthr.h>
	#include <sys/dsl_deadlist.h>
	#include <zfeature_common.h>

	#ifdef __cplusplus
	extern "C" {
	#endif

	typedef struct spa_alloc {
	kmutex_t spaa_lock;
	avl_tree_t spaa_tree;
	} ____cacheline_aligned spa_alloc_t;

	typedef struct spa_error_entry {
	zbookmark_phys_t se_bookmark;
	char *se_name;
	avl_node_t se_avl;
	} spa_error_entry_t;

	typedef struct spa_history_phys {
	uint64_t sh_pool_create_len; /* ending offset of zpool create */
	uint64_t sh_phys_max_off; /* physical EOF */
	uint64_t sh_bof; /* logical BOF */
	uint64_t sh_eof; /* logical EOF */
	uint64_t sh_records_lost; /* num of records overwritten */
	} spa_history_phys_t;

	/*
	* All members must be uint64_t, for byteswap purposes.
	*/
	typedef struct spa_removing_phys {
	uint64_t sr_state; /* dsl_scan_state_t */

	/*
	* The vdev ID that we most recently attempted to remove,
	* or -1 if no removal has been attempted.
	*/
	uint64_t sr_removing_vdev;

	/*
	* The vdev ID that we most recently successfully removed,
	* or -1 if no devices have been removed.
	*/
	uint64_t sr_prev_indirect_vdev;

	uint64_t sr_start_time;
	uint64_t sr_end_time;

	/*
	* Note that we can not use the space map's or indirect mapping's
	* accounting as a substitute for these values, because we need to
	* count frees of not-yet-copied data as though it did the copy.
	* Otherwise, we could get into a situation where copied > to_copy,
	* or we complete before copied == to_copy.
	*/
	uint64_t sr_to_copy; /* bytes that need to be copied */
	uint64_t sr_copied; /* bytes that have been copied or freed */
	} spa_removing_phys_t;

	/*
	* This struct is stored as an entry in the DMU_POOL_DIRECTORY_OBJECT
	* (with key DMU_POOL_CONDENSING_INDIRECT). It is present if a condense
	* of an indirect vdev's mapping object is in progress.
	*/
	typedef struct spa_condensing_indirect_phys {
	/*
	* The vdev ID of the indirect vdev whose indirect mapping is
	* being condensed.
	*/
	uint64_t scip_vdev;

	/*
	* The vdev's old obsolete spacemap. This spacemap's contents are
	* being integrated into the new mapping.
	*/
	uint64_t scip_prev_obsolete_sm_object;

	/*
	* The new mapping object that is being created.
	*/
	uint64_t scip_next_mapping_object;
	} spa_condensing_indirect_phys_t;

	struct spa_aux_vdev {
	uint64_t sav_object; /* MOS object for device list */
	nvlist_t sav_config; / cached device config */
	vdev_t *sav_vdevs; / devices */
	int sav_count; /* number devices */
	boolean_t sav_sync; /* sync the device list */
	nvlist_t *sav_pending; / pending device additions */
	uint_t sav_npending; /* # pending devices */
	};

	typedef struct spa_config_lock {
	kmutex_t scl_lock;
	kthread_t *scl_writer;
	int scl_write_wanted;
	int scl_count;
	kcondvar_t scl_cv;
	} ____cacheline_aligned spa_config_lock_t;

	typedef struct spa_config_dirent {
	list_node_t scd_link;
	char *scd_path;
	} spa_config_dirent_t;

	typedef enum zio_taskq_type {
	ZIO_TASKQ_ISSUE = 0,
	ZIO_TASKQ_ISSUE_HIGH,
	ZIO_TASKQ_INTERRUPT,
	ZIO_TASKQ_INTERRUPT_HIGH,
	ZIO_TASKQ_TYPES
	} zio_taskq_type_t;

	/*
	* State machine for the zpool-poolname process. The states transitions
	* are done as follows:
	*
	* From To Routine
	* PROC_NONE -> PROC_CREATED spa_activate()
	* PROC_CREATED -> PROC_ACTIVE spa_thread()
	* PROC_ACTIVE -> PROC_DEACTIVATE spa_deactivate()
	* PROC_DEACTIVATE -> PROC_GONE spa_thread()
	* PROC_GONE -> PROC_NONE spa_deactivate()
	*/
	typedef enum spa_proc_state {
	SPA_PROC_NONE, /* spa_proc = &p0, no process created */
	SPA_PROC_CREATED, /* spa_activate() has proc, is waiting */
	SPA_PROC_ACTIVE, /* taskqs created, spa_proc set */
	SPA_PROC_DEACTIVATE, /* spa_deactivate() requests process exit */
	SPA_PROC_GONE /* spa_thread() is exiting, spa_proc = &p0 */
	} spa_proc_state_t;

	typedef struct spa_taskqs {
	uint_t stqs_count;
	taskq_t **stqs_taskq;
	} spa_taskqs_t;

	typedef enum spa_all_vdev_zap_action {
	AVZ_ACTION_NONE = 0,
	AVZ_ACTION_DESTROY, /* Destroy all per-vdev ZAPs and the AVZ. */
	AVZ_ACTION_REBUILD, /* Populate the new AVZ, see spa_avz_rebuild */
	AVZ_ACTION_INITIALIZE
	} spa_avz_action_t;

	typedef enum spa_config_source {
	SPA_CONFIG_SRC_NONE = 0,
	SPA_CONFIG_SRC_SCAN, /* scan of path (default: /dev/dsk) */
	SPA_CONFIG_SRC_CACHEFILE, /* any cachefile */
	SPA_CONFIG_SRC_TRYIMPORT, /* returned from call to tryimport */
	SPA_CONFIG_SRC_SPLIT, /* new pool in a pool split */
	SPA_CONFIG_SRC_MOS /* MOS, but not always from right txg */
	} spa_config_source_t;

	struct spa {
	/*
	* Fields protected by spa_namespace_lock.
	*/
	char spa_name[ZFS_MAX_DATASET_NAME_LEN]; /* pool name */
	char spa_comment; / comment */
	avl_node_t spa_avl; /* node in spa_namespace_avl */
	nvlist_t spa_config; / last synced config */
	nvlist_t spa_config_syncing; / currently syncing config */
	nvlist_t spa_config_splitting; / config for splitting */
	nvlist_t spa_load_info; / info and errors from load */
	uint64_t spa_config_txg; /* txg of last config change */
	int spa_sync_pass; /* iterate-to-convergence */
	pool_state_t spa_state; /* pool state */
	int spa_inject_ref; /* injection references */
	uint8_t spa_sync_on; /* sync threads are running */
	spa_load_state_t spa_load_state; /* current load operation */
	boolean_t spa_indirect_vdevs_loaded; /* mappings loaded? */
	boolean_t spa_trust_config; /* do we trust vdev tree? */
	boolean_t spa_is_splitting; /* in the middle of a split? */
	spa_config_source_t spa_config_source; /* where config comes from? */
	uint64_t spa_import_flags; /* import specific flags */
	spa_taskqs_t spa_zio_taskq[ZIO_TYPES][ZIO_TASKQ_TYPES];
	dsl_pool_t *spa_dsl_pool;
	boolean_t spa_is_initializing; /* true while opening pool */
	boolean_t spa_is_exporting; /* true while exporting pool */
	metaslab_class_t spa_normal_class; / normal data class */
	metaslab_class_t spa_log_class; / intent log data class */
	metaslab_class_t spa_embedded_log_class; / log on normal vdevs */
	metaslab_class_t spa_special_class; / special allocation class */
	metaslab_class_t spa_dedup_class; / dedup allocation class */
	uint64_t spa_first_txg; /* first txg after spa_open() */
	uint64_t spa_final_txg; /* txg of export/destroy */
	uint64_t spa_freeze_txg; /* freeze pool at this txg */
	uint64_t spa_load_max_txg; /* best initial ub_txg */
	uint64_t spa_claim_max_txg; /* highest claimed birth txg */
	inode_timespec_t spa_loaded_ts; /* 1st successful open time */
	objset_t spa_meta_objset; / copy of dp->dp_meta_objset */
	kmutex_t spa_evicting_os_lock; /* Evicting objset list lock */
	list_t spa_evicting_os_list; /* Objsets being evicted. */
	kcondvar_t spa_evicting_os_cv; /* Objset Eviction Completion */
	txg_list_t spa_vdev_txg_list; /* per-txg dirty vdev list */
	vdev_t spa_root_vdev; / top-level vdev container */
	uint64_t spa_min_ashift; /* of vdevs in normal class */
	uint64_t spa_max_ashift; /* of vdevs in normal class */
	uint64_t spa_min_alloc; /* of vdevs in normal class */
	uint64_t spa_config_guid; /* config pool guid */
	uint64_t spa_load_guid; /* spa_load initialized guid */
	uint64_t spa_last_synced_guid; /* last synced guid */
	list_t spa_config_dirty_list; /* vdevs with dirty config */
	list_t spa_state_dirty_list; /* vdevs with dirty state */
	/*
	* spa_allocs is an array, whose lengths is stored in spa_alloc_count.
	* There is one tree and one lock for each allocator, to help improve
	* allocation performance in write-heavy workloads.
	*/
	spa_alloc_t *spa_allocs;
	int spa_alloc_count;

	spa_aux_vdev_t spa_spares; /* hot spares */
	spa_aux_vdev_t spa_l2cache; /* L2ARC cache devices */
	nvlist_t spa_label_features; / Features for reading MOS */
	uint64_t spa_config_object; /* MOS object for pool config */
	uint64_t spa_config_generation; /* config generation number */
	uint64_t spa_syncing_txg; /* txg currently syncing */
	bpobj_t spa_deferred_bpobj; /* deferred-free bplist */
	bplist_t spa_free_bplist[TXG_SIZE]; /* bplist of stuff to free */
	zio_cksum_salt_t spa_cksum_salt; /* secret salt for cksum */
	/* checksum context templates */
	kmutex_t spa_cksum_tmpls_lock;
	void *spa_cksum_tmpls[ZIO_CHECKSUM_FUNCTIONS];
	uberblock_t spa_ubsync; /* last synced uberblock */
	uberblock_t spa_uberblock; /* current uberblock */
	boolean_t spa_extreme_rewind; /* rewind past deferred frees */
	kmutex_t spa_scrub_lock; /* resilver/scrub lock */
	uint64_t spa_scrub_inflight; /* in-flight scrub bytes */

	/* in-flight verification bytes */
	uint64_t spa_load_verify_bytes;
	kcondvar_t spa_scrub_io_cv; /* scrub I/O completion */
	uint8_t spa_scrub_active; /* active or suspended? */
	uint8_t spa_scrub_type; /* type of scrub we're doing */
	uint8_t spa_scrub_finished; /* indicator to rotate logs */
	uint8_t spa_scrub_started; /* started since last boot */
	uint8_t spa_scrub_reopen; /* scrub doing vdev_reopen */
	uint64_t spa_scan_pass_start; /* start time per pass/reboot */
	uint64_t spa_scan_pass_scrub_pause; /* scrub pause time */
	uint64_t spa_scan_pass_scrub_spent_paused; /* total paused */
	uint64_t spa_scan_pass_exam; /* examined bytes per pass */
	uint64_t spa_scan_pass_issued; /* issued bytes per pass */

	/*
	* We are in the middle of a resilver, and another resilver
	* is needed once this one completes. This is set iff any
	* vdev_resilver_deferred is set.
	*/
	boolean_t spa_resilver_deferred;
	kmutex_t spa_async_lock; /* protect async state */
	kthread_t spa_async_thread; / thread doing async task */
	int spa_async_suspended; /* async tasks suspended */
	kcondvar_t spa_async_cv; /* wait for thread_exit() */
	uint16_t spa_async_tasks; /* async task mask */
	uint64_t spa_missing_tvds; /* unopenable tvds on load */
	uint64_t spa_missing_tvds_allowed; /* allow loading spa? */

	+ uint64_t spa_nonallocating_dspace;
	spa_removing_phys_t spa_removing_phys;
	spa_vdev_removal_t *spa_vdev_removal;

	spa_condensing_indirect_phys_t spa_condensing_indirect_phys;
	spa_condensing_indirect_t *spa_condensing_indirect;
	zthr_t spa_condense_zthr; / zthr doing condense. */

	uint64_t spa_checkpoint_txg; /* the txg of the checkpoint */
	spa_checkpoint_info_t spa_checkpoint_info; /* checkpoint accounting */
	zthr_t *spa_checkpoint_discard_zthr;

	space_map_t spa_syncing_log_sm; / current log space map */
	avl_tree_t spa_sm_logs_by_txg;
	kmutex_t spa_flushed_ms_lock; /* for metaslabs_by_flushed */
	avl_tree_t spa_metaslabs_by_flushed;
	spa_unflushed_stats_t spa_unflushed_stats;
	list_t spa_log_summary;
	uint64_t spa_log_flushall_txg;

	zthr_t spa_livelist_delete_zthr; / deleting livelists */
	zthr_t spa_livelist_condense_zthr; / condensing livelists */
	uint64_t spa_livelists_to_delete; /* set of livelists to free */
	livelist_condense_entry_t spa_to_condense; /* next to condense */

	char spa_root; / alternate root directory */
	uint64_t spa_ena; /* spa-wide ereport ENA */
	int spa_last_open_failed; /* error if last open failed */
	uint64_t spa_last_ubsync_txg; /* "best" uberblock txg */
	uint64_t spa_last_ubsync_txg_ts; /* timestamp from that ub */
	uint64_t spa_load_txg; /* ub txg that loaded */
	uint64_t spa_load_txg_ts; /* timestamp from that ub */
	uint64_t spa_load_meta_errors; /* verify metadata err count */
	uint64_t spa_load_data_errors; /* verify data err count */
	uint64_t spa_verify_min_txg; /* start txg of verify scrub */
	kmutex_t spa_errlog_lock; /* error log lock */
	uint64_t spa_errlog_last; /* last error log object */
	uint64_t spa_errlog_scrub; /* scrub error log object */
	kmutex_t spa_errlist_lock; /* error list/ereport lock */
	avl_tree_t spa_errlist_last; /* last error list */
	avl_tree_t spa_errlist_scrub; /* scrub error list */
	uint64_t spa_deflate; /* should we deflate? */
	uint64_t spa_history; /* history object */
	kmutex_t spa_history_lock; /* history lock */
	vdev_t spa_pending_vdev; / pending vdev additions */
	kmutex_t spa_props_lock; /* property lock */
	uint64_t spa_pool_props_object; /* object for properties */
	uint64_t spa_bootfs; /* default boot filesystem */
	uint64_t spa_failmode; /* failure mode for the pool */
	uint64_t spa_deadman_failmode; /* failure mode for deadman */
	uint64_t spa_delegation; /* delegation on/off */
	list_t spa_config_list; /* previous cache file(s) */
	/* per-CPU array of root of async I/O: */
	zio_t **spa_async_zio_root;
	zio_t spa_suspend_zio_root; / root of all suspended I/O */
	zio_t spa_txg_zio[TXG_SIZE]; / spa_sync() waits for this */
	kmutex_t spa_suspend_lock; /* protects suspend_zio_root */
	kcondvar_t spa_suspend_cv; /* notification of resume */
	zio_suspend_reason_t spa_suspended; /* pool is suspended */
	uint8_t spa_claiming; /* pool is doing zil_claim() */
	boolean_t spa_is_root; /* pool is root */
	int spa_minref; /* num refs when first opened */
	spa_mode_t spa_mode; /* SPA_MODE_{READ\|WRITE} */
	boolean_t spa_read_spacemaps; /* spacemaps available if ro */
	spa_log_state_t spa_log_state; /* log state */
	uint64_t spa_autoexpand; /* lun expansion on/off */
	ddt_t spa_ddt[ZIO_CHECKSUM_FUNCTIONS]; / in-core DDTs */
	uint64_t spa_ddt_stat_object; /* DDT statistics */
	uint64_t spa_dedup_dspace; /* Cache get_dedup_dspace() */
	uint64_t spa_dedup_checksum; /* default dedup checksum */
	uint64_t spa_dspace; /* dspace in normal class */
	kmutex_t spa_vdev_top_lock; /* dueling offline/remove */
	kmutex_t spa_proc_lock; /* protects spa_proc* */
	kcondvar_t spa_proc_cv; /* spa_proc_state transitions */
	spa_proc_state_t spa_proc_state; /* see definition */
	proc_t spa_proc; / "zpool-poolname" process */
	uintptr_t spa_did; /* if procp != p0, did of t1 */
	boolean_t spa_autoreplace; /* autoreplace set in open */
	int spa_vdev_locks; /* locks grabbed */
	uint64_t spa_creation_version; /* version at pool creation */
	uint64_t spa_prev_software_version; /* See ub_software_version */
	uint64_t spa_feat_for_write_obj; /* required to write to pool */
	uint64_t spa_feat_for_read_obj; /* required to read from pool */
	uint64_t spa_feat_desc_obj; /* Feature descriptions */
	uint64_t spa_feat_enabled_txg_obj; /* Feature enabled txg */
	kmutex_t spa_feat_stats_lock; /* protects spa_feat_stats */
	nvlist_t spa_feat_stats; / Cache of enabled features */
	/* cache feature refcounts */
	uint64_t spa_feat_refcount_cache[SPA_FEATURES];
	taskqid_t spa_deadman_tqid; /* Task id */
	uint64_t spa_deadman_calls; /* number of deadman calls */
	hrtime_t spa_sync_starttime; /* starting time of spa_sync */
	uint64_t spa_deadman_synctime; /* deadman sync expiration */
	uint64_t spa_deadman_ziotime; /* deadman zio expiration */
	uint64_t spa_all_vdev_zaps; /* ZAP of per-vd ZAP obj #s */
	spa_avz_action_t spa_avz_action; /* destroy/rebuild AVZ? */
	uint64_t spa_autotrim; /* automatic background trim? */
	uint64_t spa_errata; /* errata issues detected */
	spa_stats_t spa_stats; /* assorted spa statistics */
	spa_keystore_t spa_keystore; /* loaded crypto keys */

	/* arc_memory_throttle() parameters during low memory condition */
	uint64_t spa_lowmem_page_load; /* memory load during txg */
	uint64_t spa_lowmem_last_txg; /* txg window start */

	hrtime_t spa_ccw_fail_time; /* Conf cache write fail time */
	taskq_t spa_zvol_taskq; / Taskq for minor management */
	taskq_t spa_prefetch_taskq; / Taskq for prefetch threads */
	uint64_t spa_multihost; /* multihost aware (mmp) */
	mmp_thread_t spa_mmp; /* multihost mmp thread */
	list_t spa_leaf_list; /* list of leaf vdevs */
	uint64_t spa_leaf_list_gen; /* track leaf_list changes */
	uint32_t spa_hostid; /* cached system hostid */

	/* synchronization for threads in spa_wait */
	kmutex_t spa_activities_lock;
	kcondvar_t spa_activities_cv;
	kcondvar_t spa_waiters_cv;
	int spa_waiters; /* number of waiting threads */
	boolean_t spa_waiters_cancel; /* waiters should return */

	char spa_compatibility; / compatibility file(s) */

	/*
	* spa_refcount & spa_config_lock must be the last elements
	* because zfs_refcount_t changes size based on compilation options.
	* In order for the MDB module to function correctly, the other
	* fields must remain in the same location.
	*/
	spa_config_lock_t spa_config_lock[SCL_LOCKS]; /* config changes */
	zfs_refcount_t spa_refcount; /* number of opens */

	taskq_t spa_upgrade_taskq; / taskq for upgrade jobs */
	};

	extern char *spa_config_path;
	extern char *zfs_deadman_failmode;
	extern int spa_slop_shift;
	extern void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
	task_func_t func, void arg, uint_t flags, taskq_ent_t *ent);
	extern void spa_taskq_dispatch_sync(spa_t *, zio_type_t t, zio_taskq_type_t q,
	task_func_t func, void arg, uint_t flags);
	extern void spa_load_spares(spa_t *spa);
	extern void spa_load_l2cache(spa_t *spa);
	extern sysevent_t spa_event_create(spa_t spa, vdev_t vd, nvlist_t hist_nvl,
	const char *name);
	extern void spa_event_post(sysevent_t *ev);
	extern int param_set_deadman_failmode_common(const char *val);
	extern void spa_set_deadman_synctime(hrtime_t ns);
	extern void spa_set_deadman_ziotime(hrtime_t ns);
	extern const char *spa_history_zone(void);

	#ifdef __cplusplus
	}
	#endif

	#endif /* _SYS_SPA_IMPL_H */
	diff --git a/sys/contrib/openzfs/include/sys/vdev.h b/sys/contrib/openzfs/include/sys/vdev.h
	index 0a81713a44d0..4e507d081957 100644
	--- a/sys/contrib/openzfs/include/sys/vdev.h
	+++ b/sys/contrib/openzfs/include/sys/vdev.h
	@@ -1,226 +1,229 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2017, Intel Corporation.
	* Copyright (c) 2019, Datto Inc. All rights reserved.
	*/

	#ifndef _SYS_VDEV_H
	#define _SYS_VDEV_H

	#include <sys/spa.h>
	#include <sys/zio.h>
	#include <sys/dmu.h>
	#include <sys/space_map.h>
	#include <sys/metaslab.h>
	#include <sys/fs/zfs.h>

	#ifdef __cplusplus
	extern "C" {
	#endif

	typedef enum vdev_dtl_type {
	DTL_MISSING, /* 0% replication: no copies of the data */
	DTL_PARTIAL, /* less than 100% replication: some copies missing */
	DTL_SCRUB, /* unable to fully repair during scrub/resilver */
	DTL_OUTAGE, /* temporarily missing (used to attempt detach) */
	DTL_TYPES
	} vdev_dtl_type_t;

	extern int zfs_nocacheflush;

	typedef boolean_t vdev_open_children_func_t(vdev_t *vd);

	extern void vdev_dbgmsg(vdev_t vd, const char fmt, ...)
	__attribute__((format(printf, 2, 3)));
	extern void vdev_dbgmsg_print_tree(vdev_t *, int);
	extern int vdev_open(vdev_t *);
	extern void vdev_open_children(vdev_t *);
	extern void vdev_open_children_subset(vdev_t , vdev_open_children_func_t );
	extern int vdev_validate(vdev_t *);
	extern int vdev_copy_path_strict(vdev_t , vdev_t );
	extern void vdev_copy_path_relaxed(vdev_t , vdev_t );
	extern void vdev_close(vdev_t *);
	extern int vdev_create(vdev_t *, uint64_t txg, boolean_t isreplace);
	extern void vdev_reopen(vdev_t *);
	extern int vdev_validate_aux(vdev_t *vd);
	extern zio_t vdev_probe(vdev_t vd, zio_t *pio);
	extern boolean_t vdev_is_concrete(vdev_t *vd);
	extern boolean_t vdev_is_bootable(vdev_t *vd);
	extern vdev_t vdev_lookup_top(spa_t spa, uint64_t vdev);
	extern vdev_t vdev_lookup_by_guid(vdev_t vd, uint64_t guid);
	extern int vdev_count_leaves(spa_t *spa);
	extern void vdev_dtl_dirty(vdev_t *vd, vdev_dtl_type_t d,
	uint64_t txg, uint64_t size);
	extern boolean_t vdev_dtl_contains(vdev_t *vd, vdev_dtl_type_t d,
	uint64_t txg, uint64_t size);
	extern boolean_t vdev_dtl_empty(vdev_t *vd, vdev_dtl_type_t d);
	extern boolean_t vdev_default_need_resilver(vdev_t vd, const dva_t dva,
	size_t psize, uint64_t phys_birth);
	extern boolean_t vdev_dtl_need_resilver(vdev_t vd, const dva_t dva,
	size_t psize, uint64_t phys_birth);
	extern void vdev_dtl_reassess(vdev_t *vd, uint64_t txg, uint64_t scrub_txg,
	boolean_t scrub_done, boolean_t rebuild_done);
	extern boolean_t vdev_dtl_required(vdev_t *vd);
	extern boolean_t vdev_resilver_needed(vdev_t *vd,
	uint64_t minp, uint64_t maxp);
	extern void vdev_destroy_unlink_zap(vdev_t *vd, uint64_t zapobj,
	dmu_tx_t *tx);
	extern uint64_t vdev_create_link_zap(vdev_t vd, dmu_tx_t tx);
	extern void vdev_construct_zaps(vdev_t vd, dmu_tx_t tx);
	extern void vdev_destroy_spacemaps(vdev_t vd, dmu_tx_t tx);
	extern void vdev_indirect_mark_obsolete(vdev_t *vd, uint64_t offset,
	uint64_t size);
	extern void spa_vdev_indirect_mark_obsolete(spa_t *spa, uint64_t vdev,
	uint64_t offset, uint64_t size, dmu_tx_t *tx);
	extern boolean_t vdev_replace_in_progress(vdev_t *vdev);

	extern void vdev_hold(vdev_t *);
	extern void vdev_rele(vdev_t *);

	extern int vdev_metaslab_init(vdev_t *vd, uint64_t txg);
	extern void vdev_metaslab_fini(vdev_t *vd);
	extern void vdev_metaslab_set_size(vdev_t *);
	extern void vdev_expand(vdev_t *vd, uint64_t txg);
	extern void vdev_split(vdev_t *vd);
	extern void vdev_deadman(vdev_t vd, char tag);

	typedef void vdev_xlate_func_t(void arg, range_seg64_t physical_rs);

	extern boolean_t vdev_xlate_is_empty(range_seg64_t *rs);
	extern void vdev_xlate(vdev_t vd, const range_seg64_t logical_rs,
	range_seg64_t physical_rs, range_seg64_t remain_rs);
	extern void vdev_xlate_walk(vdev_t vd, const range_seg64_t logical_rs,
	vdev_xlate_func_t func, void arg);

	extern void vdev_get_stats_ex(vdev_t vd, vdev_stat_t vs, vdev_stat_ex_t *vsx);

	extern metaslab_group_t vdev_get_mg(vdev_t vd, metaslab_class_t *mc);

	extern void vdev_get_stats(vdev_t vd, vdev_stat_t vs);
	extern void vdev_clear_stats(vdev_t *vd);
	extern void vdev_stat_update(zio_t *zio, uint64_t psize);
	extern void vdev_scan_stat_init(vdev_t *vd);
	extern void vdev_propagate_state(vdev_t *vd);
	extern void vdev_set_state(vdev_t *vd, boolean_t isopen, vdev_state_t state,
	vdev_aux_t aux);
	extern boolean_t vdev_children_are_offline(vdev_t *vd);

	extern void vdev_space_update(vdev_t *vd,
	int64_t alloc_delta, int64_t defer_delta, int64_t space_delta);

	extern int64_t vdev_deflated_space(vdev_t *vd, int64_t space);

	extern uint64_t vdev_psize_to_asize(vdev_t *vd, uint64_t psize);

	/*
	* Return the amount of space allocated for a gang block header.
	*/
	static inline uint64_t
	vdev_gang_header_asize(vdev_t *vd)
	{
	return (vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE));
	}

	extern int vdev_fault(spa_t *spa, uint64_t guid, vdev_aux_t aux);
	extern int vdev_degrade(spa_t *spa, uint64_t guid, vdev_aux_t aux);
	extern int vdev_online(spa_t *spa, uint64_t guid, uint64_t flags,
	vdev_state_t *);
	extern int vdev_offline(spa_t *spa, uint64_t guid, uint64_t flags);
	extern void vdev_clear(spa_t spa, vdev_t vd);

	extern boolean_t vdev_is_dead(vdev_t *vd);
	extern boolean_t vdev_readable(vdev_t *vd);
	extern boolean_t vdev_writeable(vdev_t *vd);
	extern boolean_t vdev_allocatable(vdev_t *vd);
	extern boolean_t vdev_accessible(vdev_t vd, zio_t zio);
	extern boolean_t vdev_is_spacemap_addressable(vdev_t *vd);

	extern void vdev_cache_init(vdev_t *vd);
	extern void vdev_cache_fini(vdev_t *vd);
	extern boolean_t vdev_cache_read(zio_t *zio);
	extern void vdev_cache_write(zio_t *zio);
	extern void vdev_cache_purge(vdev_t *vd);

	extern void vdev_queue_init(vdev_t *vd);
	extern void vdev_queue_fini(vdev_t *vd);
	extern zio_t vdev_queue_io(zio_t zio);
	extern void vdev_queue_io_done(zio_t *zio);
	extern void vdev_queue_change_io_priority(zio_t *zio, zio_priority_t priority);

	extern int vdev_queue_length(vdev_t *vd);
	extern uint64_t vdev_queue_last_offset(vdev_t *vd);

	extern void vdev_config_dirty(vdev_t *vd);
	extern void vdev_config_clean(vdev_t *vd);
	extern int vdev_config_sync(vdev_t **svd, int svdcount, uint64_t txg);

	extern void vdev_state_dirty(vdev_t *vd);
	extern void vdev_state_clean(vdev_t *vd);

	extern void vdev_defer_resilver(vdev_t *vd);
	extern boolean_t vdev_clear_resilver_deferred(vdev_t vd, dmu_tx_t tx);

	typedef enum vdev_config_flag {
	VDEV_CONFIG_SPARE = 1 << 0,
	VDEV_CONFIG_L2CACHE = 1 << 1,
	VDEV_CONFIG_REMOVING = 1 << 2,
	VDEV_CONFIG_MOS = 1 << 3,
	VDEV_CONFIG_MISSING = 1 << 4
	} vdev_config_flag_t;

	extern void vdev_top_config_generate(spa_t spa, nvlist_t config);
	extern nvlist_t vdev_config_generate(spa_t spa, vdev_t *vd,
	boolean_t getstats, vdev_config_flag_t flags);

	/*
	* Label routines
	*/
	struct uberblock;
	extern uint64_t vdev_label_offset(uint64_t psize, int l, uint64_t offset);
	extern int vdev_label_number(uint64_t psise, uint64_t offset);
	extern nvlist_t vdev_label_read_config(vdev_t vd, uint64_t txg);
	extern void vdev_uberblock_load(vdev_t , struct uberblock , nvlist_t **);
	extern void vdev_config_generate_stats(vdev_t vd, nvlist_t nv);
	extern void vdev_label_write(zio_t zio, vdev_t vd, int l, abd_t *buf, uint64_t
	offset, uint64_t size, zio_done_func_t done, void priv, int flags);
	extern int vdev_label_read_bootenv(vdev_t , nvlist_t );
	extern int vdev_label_write_bootenv(vdev_t , nvlist_t );

	typedef enum {
	VDEV_LABEL_CREATE, /* create/add a new device */
	VDEV_LABEL_REPLACE, /* replace an existing device */
	VDEV_LABEL_SPARE, /* add a new hot spare */
	VDEV_LABEL_REMOVE, /* remove an existing device */
	VDEV_LABEL_L2CACHE, /* add an L2ARC cache device */
	VDEV_LABEL_SPLIT /* generating new label for split-off dev */
	} vdev_labeltype_t;

	extern int vdev_label_init(vdev_t *vd, uint64_t txg, vdev_labeltype_t reason);

	+extern int vdev_prop_set(vdev_t vd, nvlist_t innvl, nvlist_t *outnvl);
	+extern int vdev_prop_get(vdev_t vd, nvlist_t nvprops, nvlist_t *outnvl);
	+
	#ifdef __cplusplus
	}
	#endif

	#endif /* _SYS_VDEV_H */
	diff --git a/sys/contrib/openzfs/include/sys/vdev_impl.h b/sys/contrib/openzfs/include/sys/vdev_impl.h
	index 3cfde40a77fe..86959725a513 100644
	--- a/sys/contrib/openzfs/include/sys/vdev_impl.h
	+++ b/sys/contrib/openzfs/include/sys/vdev_impl.h
	@@ -1,658 +1,659 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2017, Intel Corporation.
	*/

	#ifndef _SYS_VDEV_IMPL_H
	#define _SYS_VDEV_IMPL_H

	#include <sys/avl.h>
	#include <sys/bpobj.h>
	#include <sys/dmu.h>
	#include <sys/metaslab.h>
	#include <sys/nvpair.h>
	#include <sys/space_map.h>
	#include <sys/vdev.h>
	#include <sys/dkio.h>
	#include <sys/uberblock_impl.h>
	#include <sys/vdev_indirect_mapping.h>
	#include <sys/vdev_indirect_births.h>
	#include <sys/vdev_rebuild.h>
	#include <sys/vdev_removal.h>
	#include <sys/zfs_ratelimit.h>

	#ifdef __cplusplus
	extern "C" {
	#endif

	/*
	* Virtual device descriptors.
	*
	* All storage pool operations go through the virtual device framework,
	* which provides data replication and I/O scheduling.
	*/

	/*
	* Forward declarations that lots of things need.
	*/
	typedef struct vdev_queue vdev_queue_t;
	typedef struct vdev_cache vdev_cache_t;
	typedef struct vdev_cache_entry vdev_cache_entry_t;
	struct abd;

	extern int zfs_vdev_queue_depth_pct;
	extern int zfs_vdev_def_queue_depth;
	extern uint32_t zfs_vdev_async_write_max_active;

	/*
	* Virtual device operations
	*/
	typedef int vdev_init_func_t(spa_t spa, nvlist_t nv, void **tsd);
	typedef void vdev_fini_func_t(vdev_t *vd);
	typedef int vdev_open_func_t(vdev_t vd, uint64_t size, uint64_t *max_size,
	uint64_t ashift, uint64_t pshift);
	typedef void vdev_close_func_t(vdev_t *vd);
	typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize);
	typedef uint64_t vdev_min_asize_func_t(vdev_t *vd);
	typedef uint64_t vdev_min_alloc_func_t(vdev_t *vd);
	typedef void vdev_io_start_func_t(zio_t *zio);
	typedef void vdev_io_done_func_t(zio_t *zio);
	typedef void vdev_state_change_func_t(vdev_t *vd, int, int);
	typedef boolean_t vdev_need_resilver_func_t(vdev_t vd, const dva_t dva,
	size_t psize, uint64_t phys_birth);
	typedef void vdev_hold_func_t(vdev_t *vd);
	typedef void vdev_rele_func_t(vdev_t *vd);

	typedef void vdev_remap_cb_t(uint64_t inner_offset, vdev_t *vd,
	uint64_t offset, uint64_t size, void *arg);
	typedef void vdev_remap_func_t(vdev_t *vd, uint64_t offset, uint64_t size,
	vdev_remap_cb_t callback, void *arg);
	/*
	* Given a target vdev, translates the logical range "in" to the physical
	* range "res"
	*/
	typedef void vdev_xlation_func_t(vdev_t cvd, const range_seg64_t logical,
	range_seg64_t physical, range_seg64_t remain);
	typedef uint64_t vdev_rebuild_asize_func_t(vdev_t *vd, uint64_t start,
	uint64_t size, uint64_t max_segment);
	typedef void vdev_metaslab_init_func_t(vdev_t vd, uint64_t startp,
	uint64_t *sizep);
	typedef void vdev_config_generate_func_t(vdev_t vd, nvlist_t nv);
	typedef uint64_t vdev_nparity_func_t(vdev_t *vd);
	typedef uint64_t vdev_ndisks_func_t(vdev_t *vd);

	typedef const struct vdev_ops {
	vdev_init_func_t *vdev_op_init;
	vdev_fini_func_t *vdev_op_fini;
	vdev_open_func_t *vdev_op_open;
	vdev_close_func_t *vdev_op_close;
	vdev_asize_func_t *vdev_op_asize;
	vdev_min_asize_func_t *vdev_op_min_asize;
	vdev_min_alloc_func_t *vdev_op_min_alloc;
	vdev_io_start_func_t *vdev_op_io_start;
	vdev_io_done_func_t *vdev_op_io_done;
	vdev_state_change_func_t *vdev_op_state_change;
	vdev_need_resilver_func_t *vdev_op_need_resilver;
	vdev_hold_func_t *vdev_op_hold;
	vdev_rele_func_t *vdev_op_rele;
	vdev_remap_func_t *vdev_op_remap;
	vdev_xlation_func_t *vdev_op_xlate;
	vdev_rebuild_asize_func_t *vdev_op_rebuild_asize;
	vdev_metaslab_init_func_t *vdev_op_metaslab_init;
	vdev_config_generate_func_t *vdev_op_config_generate;
	vdev_nparity_func_t *vdev_op_nparity;
	vdev_ndisks_func_t *vdev_op_ndisks;
	char vdev_op_type[16];
	boolean_t vdev_op_leaf;
	} vdev_ops_t;

	/*
	* Virtual device properties
	*/
	struct vdev_cache_entry {
	struct abd *ve_abd;
	uint64_t ve_offset;
	clock_t ve_lastused;
	avl_node_t ve_offset_node;
	avl_node_t ve_lastused_node;
	uint32_t ve_hits;
	uint16_t ve_missed_update;
	zio_t *ve_fill_io;
	};

	struct vdev_cache {
	avl_tree_t vc_offset_tree;
	avl_tree_t vc_lastused_tree;
	kmutex_t vc_lock;
	};

	typedef struct vdev_queue_class {
	uint32_t vqc_active;

	/*
	* Sorted by offset or timestamp, depending on if the queue is
	* LBA-ordered vs FIFO.
	*/
	avl_tree_t vqc_queued_tree;
	} vdev_queue_class_t;

	struct vdev_queue {
	vdev_t *vq_vdev;
	vdev_queue_class_t vq_class[ZIO_PRIORITY_NUM_QUEUEABLE];
	avl_tree_t vq_active_tree;
	avl_tree_t vq_read_offset_tree;
	avl_tree_t vq_write_offset_tree;
	avl_tree_t vq_trim_offset_tree;
	uint64_t vq_last_offset;
	zio_priority_t vq_last_prio; /* Last sent I/O priority. */
	uint32_t vq_ia_active; /* Active interactive I/Os. */
	uint32_t vq_nia_credit; /* Non-interactive I/Os credit. */
	hrtime_t vq_io_complete_ts; /* time last i/o completed */
	hrtime_t vq_io_delta_ts;
	zio_t vq_io_search; /* used as local for stack reduction */
	kmutex_t vq_lock;
	};

	typedef enum vdev_alloc_bias {
	VDEV_BIAS_NONE,
	VDEV_BIAS_LOG, /* dedicated to ZIL data (SLOG) */
	VDEV_BIAS_SPECIAL, /* dedicated to ddt, metadata, and small blks */
	VDEV_BIAS_DEDUP /* dedicated to dedup metadata */
	} vdev_alloc_bias_t;


	/*
	* On-disk indirect vdev state.
	*
	* An indirect vdev is described exclusively in the MOS config of a pool.
	* The config for an indirect vdev includes several fields, which are
	* accessed in memory by a vdev_indirect_config_t.
	*/
	typedef struct vdev_indirect_config {
	/*
	* Object (in MOS) which contains the indirect mapping. This object
	* contains an array of vdev_indirect_mapping_entry_phys_t ordered by
	* vimep_src. The bonus buffer for this object is a
	* vdev_indirect_mapping_phys_t. This object is allocated when a vdev
	* removal is initiated.
	*
	* Note that this object can be empty if none of the data on the vdev
	* has been copied yet.
	*/
	uint64_t vic_mapping_object;

	/*
	* Object (in MOS) which contains the birth times for the mapping
	* entries. This object contains an array of
	* vdev_indirect_birth_entry_phys_t sorted by vibe_offset. The bonus
	* buffer for this object is a vdev_indirect_birth_phys_t. This object
	* is allocated when a vdev removal is initiated.
	*
	* Note that this object can be empty if none of the vdev has yet been
	* copied.
	*/
	uint64_t vic_births_object;

	/*
	* This is the vdev ID which was removed previous to this vdev, or
	* UINT64_MAX if there are no previously removed vdevs.
	*/
	uint64_t vic_prev_indirect_vdev;
	} vdev_indirect_config_t;

	/*
	* Virtual device descriptor
	*/
	struct vdev {
	/*
	* Common to all vdev types.
	*/
	uint64_t vdev_id; /* child number in vdev parent */
	uint64_t vdev_guid; /* unique ID for this vdev */
	uint64_t vdev_guid_sum; /* self guid + all child guids */
	uint64_t vdev_orig_guid; /* orig. guid prior to remove */
	uint64_t vdev_asize; /* allocatable device capacity */
	uint64_t vdev_min_asize; /* min acceptable asize */
	uint64_t vdev_max_asize; /* max acceptable asize */
	uint64_t vdev_ashift; /* block alignment shift */

	/*
	* Logical block alignment shift
	*
	* The smallest sized/aligned I/O supported by the device.
	*/
	uint64_t vdev_logical_ashift;
	/*
	* Physical block alignment shift
	*
	* The device supports logical I/Os with vdev_logical_ashift
	* size/alignment, but optimum performance will be achieved by
	* aligning/sizing requests to vdev_physical_ashift. Smaller
	* requests may be inflated or incur device level read-modify-write
	* operations.
	*
	* May be 0 to indicate no preference (i.e. use vdev_logical_ashift).
	*/
	uint64_t vdev_physical_ashift;
	uint64_t vdev_state; /* see VDEV_STATE_* #defines */
	uint64_t vdev_prevstate; /* used when reopening a vdev */
	vdev_ops_t vdev_ops; / vdev operations */
	spa_t vdev_spa; / spa for this vdev */
	void vdev_tsd; / type-specific data */
	vdev_t vdev_top; / top-level vdev */
	vdev_t vdev_parent; / parent vdev */
	vdev_t *vdev_child; / array of children */
	uint64_t vdev_children; /* number of children */
	vdev_stat_t vdev_stat; /* virtual device statistics */
	vdev_stat_ex_t vdev_stat_ex; /* extended statistics */
	boolean_t vdev_expanding; /* expand the vdev? */
	boolean_t vdev_reopening; /* reopen in progress? */
	boolean_t vdev_nonrot; /* true if solid state */
	int vdev_load_error; /* error on last load */
	int vdev_open_error; /* error on last open */
	int vdev_validate_error; /* error on last validate */
	kthread_t vdev_open_thread; / thread opening children */
	kthread_t vdev_validate_thread; / thread validating children */
	uint64_t vdev_crtxg; /* txg when top-level was added */

	/*
	* Top-level vdev state.
	*/
	uint64_t vdev_ms_array; /* metaslab array object */
	uint64_t vdev_ms_shift; /* metaslab size shift */
	uint64_t vdev_ms_count; /* number of metaslabs */
	metaslab_group_t vdev_mg; / metaslab group */
	metaslab_group_t vdev_log_mg; / embedded slog metaslab group */
	metaslab_t *vdev_ms; / metaslab array */
	uint64_t vdev_pending_fastwrite; /* allocated fastwrites */
	txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */
	txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */
	txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */
	boolean_t vdev_remove_wanted; /* async remove wanted? */
	boolean_t vdev_probe_wanted; /* async probe wanted? */
	list_node_t vdev_config_dirty_node; /* config dirty list */
	list_node_t vdev_state_dirty_node; /* state dirty list */
	uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */
	uint64_t vdev_islog; /* is an intent log device */
	+ uint64_t vdev_noalloc; /* device is passivated? */
	uint64_t vdev_removing; /* device is being removed? */
	boolean_t vdev_ishole; /* is a hole in the namespace */
	uint64_t vdev_top_zap;
	vdev_alloc_bias_t vdev_alloc_bias; /* metaslab allocation bias */

	/* pool checkpoint related */
	space_map_t vdev_checkpoint_sm; / contains reserved blocks */

	/* Initialize related */
	boolean_t vdev_initialize_exit_wanted;
	vdev_initializing_state_t vdev_initialize_state;
	list_node_t vdev_initialize_node;
	kthread_t *vdev_initialize_thread;
	/* Protects vdev_initialize_thread and vdev_initialize_state. */
	kmutex_t vdev_initialize_lock;
	kcondvar_t vdev_initialize_cv;
	uint64_t vdev_initialize_offset[TXG_SIZE];
	uint64_t vdev_initialize_last_offset;
	range_tree_t vdev_initialize_tree; / valid while initializing */
	uint64_t vdev_initialize_bytes_est;
	uint64_t vdev_initialize_bytes_done;
	uint64_t vdev_initialize_action_time; /* start and end time */

	/* TRIM related */
	boolean_t vdev_trim_exit_wanted;
	boolean_t vdev_autotrim_exit_wanted;
	vdev_trim_state_t vdev_trim_state;
	list_node_t vdev_trim_node;
	kmutex_t vdev_autotrim_lock;
	kcondvar_t vdev_autotrim_cv;
	kthread_t *vdev_autotrim_thread;
	/* Protects vdev_trim_thread and vdev_trim_state. */
	kmutex_t vdev_trim_lock;
	kcondvar_t vdev_trim_cv;
	kthread_t *vdev_trim_thread;
	uint64_t vdev_trim_offset[TXG_SIZE];
	uint64_t vdev_trim_last_offset;
	uint64_t vdev_trim_bytes_est;
	uint64_t vdev_trim_bytes_done;
	uint64_t vdev_trim_rate; /* requested rate (bytes/sec) */
	uint64_t vdev_trim_partial; /* requested partial TRIM */
	uint64_t vdev_trim_secure; /* requested secure TRIM */
	uint64_t vdev_trim_action_time; /* start and end time */

	/* Rebuild related */
	boolean_t vdev_rebuilding;
	boolean_t vdev_rebuild_exit_wanted;
	boolean_t vdev_rebuild_cancel_wanted;
	boolean_t vdev_rebuild_reset_wanted;
	kmutex_t vdev_rebuild_lock;
	kcondvar_t vdev_rebuild_cv;
	kthread_t *vdev_rebuild_thread;
	vdev_rebuild_t vdev_rebuild_config;

	/* For limiting outstanding I/Os (initialize, TRIM) */
	kmutex_t vdev_initialize_io_lock;
	kcondvar_t vdev_initialize_io_cv;
	uint64_t vdev_initialize_inflight;
	kmutex_t vdev_trim_io_lock;
	kcondvar_t vdev_trim_io_cv;
	uint64_t vdev_trim_inflight[3];

	/*
	* Values stored in the config for an indirect or removing vdev.
	*/
	vdev_indirect_config_t vdev_indirect_config;

	/*
	* The vdev_indirect_rwlock protects the vdev_indirect_mapping
	* pointer from changing on indirect vdevs (when it is condensed).
	* Note that removing (not yet indirect) vdevs have different
	* access patterns (the mapping is not accessed from open context,
	* e.g. from zio_read) and locking strategy (e.g. svr_lock).
	*/
	krwlock_t vdev_indirect_rwlock;
	vdev_indirect_mapping_t *vdev_indirect_mapping;
	vdev_indirect_births_t *vdev_indirect_births;

	/*
	* In memory data structures used to manage the obsolete sm, for
	* indirect or removing vdevs.
	*
	* The vdev_obsolete_segments is the in-core record of the segments
	* that are no longer referenced anywhere in the pool (due to
	* being freed or remapped and not referenced by any snapshots).
	* During a sync, segments are added to vdev_obsolete_segments
	* via vdev_indirect_mark_obsolete(); at the end of each sync
	* pass, this is appended to vdev_obsolete_sm via
	* vdev_indirect_sync_obsolete(). The vdev_obsolete_lock
	* protects against concurrent modifications of vdev_obsolete_segments
	* from multiple zio threads.
	*/
	kmutex_t vdev_obsolete_lock;
	range_tree_t *vdev_obsolete_segments;
	space_map_t *vdev_obsolete_sm;

	/*
	* Protects the vdev_scan_io_queue field itself as well as the
	* structure's contents (when present).
	*/
	kmutex_t vdev_scan_io_queue_lock;
	struct dsl_scan_io_queue *vdev_scan_io_queue;

	/*
	* Leaf vdev state.
	*/
	range_tree_t vdev_dtl[DTL_TYPES]; / dirty time logs */
	space_map_t vdev_dtl_sm; / dirty time log space map */
	txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */
	uint64_t vdev_dtl_object; /* DTL object */
	uint64_t vdev_psize; /* physical device capacity */
	uint64_t vdev_wholedisk; /* true if this is a whole disk */
	uint64_t vdev_offline; /* persistent offline state */
	uint64_t vdev_faulted; /* persistent faulted state */
	uint64_t vdev_degraded; /* persistent degraded state */
	uint64_t vdev_removed; /* persistent removed state */
	uint64_t vdev_resilver_txg; /* persistent resilvering state */
	uint64_t vdev_rebuild_txg; /* persistent rebuilding state */
	char vdev_path; / vdev path (if any) */
	char vdev_devid; / vdev devid (if any) */
	char vdev_physpath; / vdev device path (if any) */
	char vdev_enc_sysfs_path; / enclosure sysfs path */
	char vdev_fru; / physical FRU location */
	uint64_t vdev_not_present; /* not present during import */
	uint64_t vdev_unspare; /* unspare when resilvering done */
	boolean_t vdev_nowritecache; /* true if flushwritecache failed */
	boolean_t vdev_has_trim; /* TRIM is supported */
	boolean_t vdev_has_securetrim; /* secure TRIM is supported */
	boolean_t vdev_checkremove; /* temporary online test */
	boolean_t vdev_forcefault; /* force online fault */
	boolean_t vdev_splitting; /* split or repair in progress */
	boolean_t vdev_delayed_close; /* delayed device close? */
	boolean_t vdev_tmpoffline; /* device taken offline temporarily? */
	boolean_t vdev_detached; /* device detached? */
	boolean_t vdev_cant_read; /* vdev is failing all reads */
	boolean_t vdev_cant_write; /* vdev is failing all writes */
	boolean_t vdev_isspare; /* was a hot spare */
	boolean_t vdev_isl2cache; /* was a l2cache device */
	boolean_t vdev_copy_uberblocks; /* post expand copy uberblocks */
	boolean_t vdev_resilver_deferred; /* resilver deferred */
	vdev_queue_t vdev_queue; /* I/O deadline schedule queue */
	vdev_cache_t vdev_cache; /* physical block cache */
	spa_aux_vdev_t vdev_aux; / for l2cache and spares vdevs */
	zio_t vdev_probe_zio; / root of current probe */
	vdev_aux_t vdev_label_aux; /* on-disk aux state */
	uint64_t vdev_leaf_zap;
	hrtime_t vdev_mmp_pending; /* 0 if write finished */
	uint64_t vdev_mmp_kstat_id; /* to find kstat entry */
	uint64_t vdev_expansion_time; /* vdev's last expansion time */
	list_node_t vdev_leaf_node; /* leaf vdev list */

	/*
	* For DTrace to work in userland (libzpool) context, these fields must
	* remain at the end of the structure. DTrace will use the kernel's
	* CTF definition for 'struct vdev', and since the size of a kmutex_t is
	* larger in userland, the offsets for the rest of the fields would be
	* incorrect.
	*/
	kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */
	kmutex_t vdev_stat_lock; /* vdev_stat */
	kmutex_t vdev_probe_lock; /* protects vdev_probe_zio */

	/*
	* We rate limit ZIO delay, deadman, and checksum events, since they
	* can flood ZED with tons of events when a drive is acting up.
	*/
	zfs_ratelimit_t vdev_delay_rl;
	zfs_ratelimit_t vdev_deadman_rl;
	zfs_ratelimit_t vdev_checksum_rl;
	};

	#define VDEV_PAD_SIZE (8 << 10)
	/* 2 padding areas (vl_pad1 and vl_be) to skip */
	#define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2
	#define VDEV_PHYS_SIZE (112 << 10)
	#define VDEV_UBERBLOCK_RING (128 << 10)

	/*
	* MMP blocks occupy the last MMP_BLOCKS_PER_LABEL slots in the uberblock
	* ring when MMP is enabled.
	*/
	#define MMP_BLOCKS_PER_LABEL 1

	/* The largest uberblock we support is 8k. */
	#define MAX_UBERBLOCK_SHIFT (13)
	#define VDEV_UBERBLOCK_SHIFT(vd) \
	MIN(MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT), \
	MAX_UBERBLOCK_SHIFT)
	#define VDEV_UBERBLOCK_COUNT(vd) \
	(VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd))
	#define VDEV_UBERBLOCK_OFFSET(vd, n) \
	offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)])
	#define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd))

	typedef struct vdev_phys {
	char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)];
	zio_eck_t vp_zbt;
	} vdev_phys_t;

	typedef enum vbe_vers {
	/*
	* The bootenv file is stored as ascii text in the envblock.
	* It is used by the GRUB bootloader used on Linux to store the
	* contents of the grubenv file. The file is stored as raw ASCII,
	* and is protected by an embedded checksum. By default, GRUB will
	* check if the boot filesystem supports storing the environment data
	* in a special location, and if so, will invoke filesystem specific
	* logic to retrieve it. This can be overridden by a variable, should
	* the user so desire.
	*/
	VB_RAW = 0,

	/*
	* The bootenv file is converted to an nvlist and then packed into the
	* envblock.
	*/
	VB_NVLIST = 1
	} vbe_vers_t;

	typedef struct vdev_boot_envblock {
	uint64_t vbe_version;
	char vbe_bootenv[VDEV_PAD_SIZE - sizeof (uint64_t) -
	sizeof (zio_eck_t)];
	zio_eck_t vbe_zbt;
	} vdev_boot_envblock_t;

	CTASSERT_GLOBAL(sizeof (vdev_boot_envblock_t) == VDEV_PAD_SIZE);

	typedef struct vdev_label {
	char vl_pad1[VDEV_PAD_SIZE]; /* 8K */
	vdev_boot_envblock_t vl_be; /* 8K */
	vdev_phys_t vl_vdev_phys; /* 112K */
	char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */
	} vdev_label_t; /* 256K total */

	/*
	* vdev_dirty() flags
	*/
	#define VDD_METASLAB 0x01
	#define VDD_DTL 0x02

	/* Offset of embedded boot loader region on each label */
	#define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t))
	/*
	* Size of embedded boot loader region on each label.
	* The total size of the first two labels plus the boot area is 4MB.
	*/
	#define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */

	/*
	* Size of label regions at the start and end of each leaf device.
	*/
	#define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
	#define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t))
	#define VDEV_LABELS 4
	#define VDEV_BEST_LABEL VDEV_LABELS
	#define VDEV_OFFSET_IS_LABEL(vd, off) \
	(((off) < VDEV_LABEL_START_SIZE) \|\| \
	((off) >= ((vd)->vdev_psize - VDEV_LABEL_END_SIZE)))

	#define VDEV_ALLOC_LOAD 0
	#define VDEV_ALLOC_ADD 1
	#define VDEV_ALLOC_SPARE 2
	#define VDEV_ALLOC_L2CACHE 3
	#define VDEV_ALLOC_ROOTPOOL 4
	#define VDEV_ALLOC_SPLIT 5
	#define VDEV_ALLOC_ATTACH 6

	/*
	* Allocate or free a vdev
	*/
	extern vdev_t vdev_alloc_common(spa_t spa, uint_t id, uint64_t guid,
	vdev_ops_t *ops);
	extern int vdev_alloc(spa_t spa, vdev_t vdp, nvlist_t config,
	vdev_t *parent, uint_t id, int alloctype);
	extern void vdev_free(vdev_t *vd);

	/*
	* Add or remove children and parents
	*/
	extern void vdev_add_child(vdev_t pvd, vdev_t cvd);
	extern void vdev_remove_child(vdev_t pvd, vdev_t cvd);
	extern void vdev_compact_children(vdev_t *pvd);
	extern vdev_t vdev_add_parent(vdev_t cvd, vdev_ops_t *ops);
	extern void vdev_remove_parent(vdev_t *cvd);

	/*
	* vdev sync load and sync
	*/
	extern boolean_t vdev_log_state_valid(vdev_t *vd);
	extern int vdev_load(vdev_t *vd);
	extern int vdev_dtl_load(vdev_t *vd);
	extern void vdev_sync(vdev_t *vd, uint64_t txg);
	extern void vdev_sync_done(vdev_t *vd, uint64_t txg);
	extern void vdev_dirty(vdev_t vd, int flags, void arg, uint64_t txg);
	extern void vdev_dirty_leaves(vdev_t *vd, int flags, uint64_t txg);

	/*
	* Available vdev types.
	*/
	extern vdev_ops_t vdev_root_ops;
	extern vdev_ops_t vdev_mirror_ops;
	extern vdev_ops_t vdev_replacing_ops;
	extern vdev_ops_t vdev_raidz_ops;
	extern vdev_ops_t vdev_draid_ops;
	extern vdev_ops_t vdev_draid_spare_ops;
	extern vdev_ops_t vdev_disk_ops;
	extern vdev_ops_t vdev_file_ops;
	extern vdev_ops_t vdev_missing_ops;
	extern vdev_ops_t vdev_hole_ops;
	extern vdev_ops_t vdev_spare_ops;
	extern vdev_ops_t vdev_indirect_ops;

	/*
	* Common size functions
	*/
	extern void vdev_default_xlate(vdev_t vd, const range_seg64_t logical_rs,
	range_seg64_t physical_rs, range_seg64_t remain_rs);
	extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize);
	extern uint64_t vdev_default_min_asize(vdev_t *vd);
	extern uint64_t vdev_get_min_asize(vdev_t *vd);
	extern void vdev_set_min_asize(vdev_t *vd);
	extern uint64_t vdev_get_min_alloc(vdev_t *vd);
	extern uint64_t vdev_get_nparity(vdev_t *vd);
	extern uint64_t vdev_get_ndisks(vdev_t *vd);

	/*
	* Global variables
	*/
	extern int zfs_vdev_standard_sm_blksz;
	/* zdb uses this tunable, so it must be declared here to make lint happy. */
	extern int zfs_vdev_cache_size;

	/*
	* Functions from vdev_indirect.c
	*/
	extern void vdev_indirect_sync_obsolete(vdev_t vd, dmu_tx_t tx);
	extern boolean_t vdev_indirect_should_condense(vdev_t *vd);
	extern void spa_condense_indirect_start_sync(vdev_t vd, dmu_tx_t tx);
	extern int vdev_obsolete_sm_object(vdev_t vd, uint64_t sm_obj);
	extern int vdev_obsolete_counts_are_precise(vdev_t vd, boolean_t are_precise);

	/*
	* Other miscellaneous functions
	*/
	int vdev_checkpoint_sm_object(vdev_t vd, uint64_t sm_obj);
	void vdev_metaslab_group_create(vdev_t *vd);

	/*
	* Vdev ashift optimization tunables
	*/
	extern uint64_t zfs_vdev_min_auto_ashift;
	extern uint64_t zfs_vdev_max_auto_ashift;
	int param_set_min_auto_ashift(ZFS_MODULE_PARAM_ARGS);
	int param_set_max_auto_ashift(ZFS_MODULE_PARAM_ARGS);

	#ifdef __cplusplus
	}
	#endif

	#endif /* _SYS_VDEV_IMPL_H */
	diff --git a/sys/contrib/openzfs/include/sys/zfs_sysfs.h b/sys/contrib/openzfs/include/sys/zfs_sysfs.h
	index 14ba61fc4b04..912ef234f8c5 100644
	--- a/sys/contrib/openzfs/include/sys/zfs_sysfs.h
	+++ b/sys/contrib/openzfs/include/sys/zfs_sysfs.h
	@@ -1,50 +1,51 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2018 by Delphix. All rights reserved.
	*/

	#ifndef _SYS_ZFS_SYSFS_H
	#define _SYS_ZFS_SYSFS_H extern __attribute__((visibility("default")))

	#ifdef _KERNEL

	void zfs_sysfs_init(void);
	void zfs_sysfs_fini(void);

	#else

	#define zfs_sysfs_init()
	#define zfs_sysfs_fini()

	_SYS_ZFS_SYSFS_H boolean_t zfs_mod_supported(const char , const char );
	#endif

	#define ZFS_SYSFS_POOL_PROPERTIES "properties.pool"
	+#define ZFS_SYSFS_VDEV_PROPERTIES "properties.vdev"
	#define ZFS_SYSFS_DATASET_PROPERTIES "properties.dataset"
	#define ZFS_SYSFS_KERNEL_FEATURES "features.kernel"
	#define ZFS_SYSFS_POOL_FEATURES "features.pool"

	#define ZFS_SYSFS_DIR "/sys/module/zfs"
	/* Alternate location when ZFS is built as part of the kernel (rare) */
	#define ZFS_SYSFS_ALT_DIR "/sys/fs/zfs"

	#endif /* _SYS_ZFS_SYSFS_H */
	diff --git a/sys/contrib/openzfs/include/zfs_prop.h b/sys/contrib/openzfs/include/zfs_prop.h
	index 91b5032e7058..8014c757aaff 100644
	--- a/sys/contrib/openzfs/include/zfs_prop.h
	+++ b/sys/contrib/openzfs/include/zfs_prop.h
	@@ -1,135 +1,144 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright 2010 Sun Microsystems, Inc. All rights reserved.
	* Use is subject to license terms.
	*/

	#ifndef _ZFS_PROP_H
	#define _ZFS_PROP_H extern __attribute__((visibility("default")))

	#include <sys/fs/zfs.h>
	#include <sys/types.h>

	#ifdef __cplusplus
	extern "C" {
	#endif

	/*
	* For index types (e.g. compression and checksum), we want the numeric value
	* in the kernel, but the string value in userland.
	*/
	typedef enum {
	PROP_TYPE_NUMBER, /* numeric value */
	PROP_TYPE_STRING, /* string value */
	PROP_TYPE_INDEX /* numeric value indexed by string */
	} zprop_type_t;

	typedef enum {
	PROP_DEFAULT,
	PROP_READONLY,
	PROP_INHERIT,
	/*
	* ONETIME properties are a sort of conglomeration of READONLY
	* and INHERIT. They can be set only during object creation,
	* after that they are READONLY. If not explicitly set during
	* creation, they can be inherited. ONETIME_DEFAULT properties
	* work the same way, but they will default instead of
	* inheriting a value.
	*/
	PROP_ONETIME,
	PROP_ONETIME_DEFAULT
	} zprop_attr_t;

	typedef struct zfs_index {
	const char *pi_name;
	uint64_t pi_value;
	} zprop_index_t;

	typedef struct {
	const char pd_name; / human-readable property name */
	int pd_propnum; /* property number */
	zprop_type_t pd_proptype; /* string, boolean, index, number */
	const char pd_strdefault; / default for strings */
	uint64_t pd_numdefault; /* for boolean / index / number */
	zprop_attr_t pd_attr; /* default, readonly, inherit */
	int pd_types; /* bitfield of valid dataset types */
	/* fs \| vol \| snap; or pool */
	const char pd_values; / string telling acceptable values */
	const char pd_colname; / column header for "zfs list" */
	boolean_t pd_rightalign; /* column alignment for "zfs list" */
	boolean_t pd_visible; /* do we list this property with the */
	/* "zfs get" help message */
	boolean_t pd_zfs_mod_supported; /* supported by running zfs module */
	const zprop_index_t pd_table; / for index properties, a table */
	/* defining the possible values */
	size_t pd_table_size; /* number of entries in pd_table[] */
	} zprop_desc_t;

	/*
	* zfs dataset property functions
	*/
	_ZFS_PROP_H void zfs_prop_init(void);
	_ZFS_PROP_H zprop_type_t zfs_prop_get_type(zfs_prop_t);
	_ZFS_PROP_H boolean_t zfs_prop_delegatable(zfs_prop_t prop);
	_ZFS_PROP_H zprop_desc_t *zfs_prop_get_table(void);

	/*
	* zpool property functions
	*/
	_ZFS_PROP_H void zpool_prop_init(void);
	_ZFS_PROP_H zprop_type_t zpool_prop_get_type(zpool_prop_t);
	_ZFS_PROP_H zprop_desc_t *zpool_prop_get_table(void);

	+/*
	+ * vdev property functions
	+ */
	+_ZFS_PROP_H void vdev_prop_init(void);
	+_ZFS_PROP_H zprop_type_t vdev_prop_get_type(vdev_prop_t prop);
	+_ZFS_PROP_H zprop_desc_t *vdev_prop_get_table(void);
	+
	/*
	* Common routines to initialize property tables
	*/
	_ZFS_PROP_H void zprop_register_impl(int, const char *, zprop_type_t, uint64_t,
	const char , zprop_attr_t, int, const char , const char *,
	boolean_t, boolean_t, const zprop_index_t *);
	_ZFS_PROP_H void zprop_register_string(int, const char , const char ,
	zprop_attr_t attr, int, const char , const char );
	_ZFS_PROP_H void zprop_register_number(int, const char *, uint64_t,
	zprop_attr_t, int, const char , const char );
	_ZFS_PROP_H void zprop_register_index(int, const char *, uint64_t, zprop_attr_t,
	int, const char , const char , const zprop_index_t *);
	_ZFS_PROP_H void zprop_register_hidden(int, const char *, zprop_type_t,
	zprop_attr_t, int, const char *);

	/*
	* Common routines for zfs and zpool property management
	*/
	_ZFS_PROP_H int zprop_iter_common(zprop_func, void *, boolean_t, boolean_t,
	zfs_type_t);
	_ZFS_PROP_H int zprop_name_to_prop(const char *, zfs_type_t);
	_ZFS_PROP_H int zprop_string_to_index(int, const char , uint64_t ,
	zfs_type_t);
	-_ZFS_PROP_H int zprop_index_to_string(int, uint64_t, const char **, zfs_type_t);
	+_ZFS_PROP_H int zprop_index_to_string(int, uint64_t, const char **,
	+ zfs_type_t);
	_ZFS_PROP_H uint64_t zprop_random_value(int, uint64_t, zfs_type_t);
	_ZFS_PROP_H const char *zprop_values(int, zfs_type_t);
	_ZFS_PROP_H size_t zprop_width(int, boolean_t *, zfs_type_t);
	_ZFS_PROP_H boolean_t zprop_valid_for_type(int, zfs_type_t, boolean_t);
	+_ZFS_PROP_H int zprop_valid_char(char c);

	#ifdef __cplusplus
	}
	#endif

	#endif /* _ZFS_PROP_H */
	diff --git a/sys/contrib/openzfs/lib/libnvpair/libnvpair.abi b/sys/contrib/openzfs/lib/libnvpair/libnvpair.abi
	index 9c27d178dc05..4a001c3dc3a6 100644
	--- a/sys/contrib/openzfs/lib/libnvpair/libnvpair.abi
	+++ b/sys/contrib/openzfs/lib/libnvpair/libnvpair.abi
	@@ -1,2778 +1,2805 @@
	<abi-corpus version='2.0' architecture='elf-amd-x86_64' soname='libnvpair.so.3'>
	<elf-needed>
	<dependency name='libc.so.6'/>
	</elf-needed>
	<elf-function-symbols>
	+ <elf-symbol name='_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='dump_nvlist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_boolean' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_boolean_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_boolean_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_byte' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_byte_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_int16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_int16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_int32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_int32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_int64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_int64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_int8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_int8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_nvlist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_nvlist_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_nvpair' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_string_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_uint16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_uint16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_uint32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_uint32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_uint64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_uint64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_uint8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_add_uint8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_alloc' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_dup' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_free' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_boolean' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_boolean_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_boolean_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_byte' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_byte_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_int16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_int16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_int32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_int32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_int64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_int64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_int8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_int8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_nvlist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_nvpair' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_uint16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_uint16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_uint32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_uint32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_uint64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_uint64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_uint8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_lookup_uint8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_merge' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_num_pairs' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_pack' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_pack_free' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_remove_nvpair' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_size' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvlist_unpack' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_boolean_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_byte' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_int16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_int32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_int64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_int8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_nvlist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_uint16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_uint32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_uint64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fnvpair_value_uint8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libspl_assertf' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nv_alloc_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nv_alloc_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nv_alloc_reset' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_boolean' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_boolean_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_boolean_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_byte' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_byte_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_double' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_hrtime' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_int16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_int16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_int32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_int32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_int64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_int64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_int8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_int8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_nvlist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_nvlist_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_nvpair' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_string_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_uint16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_uint16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_uint32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_uint32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_uint64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_uint64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_uint8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_add_uint8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_alloc' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_dup' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_empty' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_exists' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_free' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_boolean' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_boolean_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_boolean_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_byte' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_byte_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_double' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_hrtime' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_int16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_int16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_int32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_int32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_int64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_int64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_int8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_int8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_nv_alloc' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_nvlist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_nvlist_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_nvpair' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_nvpair_embedded_index' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_pairs' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_string_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_uint16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_uint16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_uint32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_uint32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_uint64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_uint64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_uint8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_lookup_uint8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_merge' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_next_nvpair' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_nvflag' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_pack' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prev_nvpair' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_print' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_print_json' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prt' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctl_alloc' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctl_dofmt' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctl_doindent' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctl_free' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctl_getdest' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctl_setdest' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctl_setfmt' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctl_setindent' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_boolean' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_boolean_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_boolean_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_byte' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_byte_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_double' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_hrtime' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_int16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_int16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_int32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_int32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_int64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_int64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_int8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_int8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_nvlist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_nvlist_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_string_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_uint16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_uint16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_uint32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_uint32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_uint64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_uint64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_uint8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_prtctlop_uint8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_remove_all' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_remove_nvpair' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_size' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_unpack' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_xalloc' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_xdup' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_xpack' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvlist_xunpack' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_type' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_type_is_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_boolean_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_boolean_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_byte' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_byte_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_double' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_hrtime' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_int16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_int16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_int32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_int32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_int64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_int64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_int8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_int8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_match' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_match_regex' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_nvlist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_nvlist_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_string_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_uint16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_uint16_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_uint32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_uint32_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_uint64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_uint64_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_uint8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nvpair_value_uint8_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	</elf-function-symbols>
	<elf-variable-symbols>
	<elf-symbol name='libspl_assert_ok' size='4' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nv_alloc_nosleep' size='8' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='nv_fixed_ops' size='8' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	</elf-variable-symbols>
	<abi-instr address-size='64' path='../../module/nvpair/fnvpair.c' language='LANG_C99'>
	<function-decl name='fnvlist_alloc' mangled-name='fnvlist_alloc' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_alloc'>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='fnvlist_free' mangled-name='fnvlist_free' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_free'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_size' mangled-name='fnvlist_size' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_size'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<return type-id='b59d7dce'/>
	</function-decl>
	<function-decl name='fnvlist_pack' mangled-name='fnvlist_pack' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_pack'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='78c01427' name='sizep'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='fnvlist_pack_free' mangled-name='fnvlist_pack_free' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_pack_free'>
	<parameter type-id='26a90f95' name='pack'/>
	<parameter type-id='b59d7dce' name='size'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_unpack' mangled-name='fnvlist_unpack' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_unpack'>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='b59d7dce' name='buflen'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='fnvlist_dup' mangled-name='fnvlist_dup' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_dup'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='fnvlist_merge' mangled-name='fnvlist_merge' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_merge'>
	<parameter type-id='5ce45b60' name='dst'/>
	<parameter type-id='5ce45b60' name='src'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_num_pairs' mangled-name='fnvlist_num_pairs' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_num_pairs'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<return type-id='b59d7dce'/>
	</function-decl>
	<function-decl name='fnvlist_add_boolean' mangled-name='fnvlist_add_boolean' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_boolean'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_boolean_value' mangled-name='fnvlist_add_boolean_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_boolean_value'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='c19b74c3' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_byte' mangled-name='fnvlist_add_byte' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_byte'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='d8bf0010' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_int8' mangled-name='fnvlist_add_int8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_int8'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='ee31ee44' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_uint8' mangled-name='fnvlist_add_uint8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_uint8'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='b96825af' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_int16' mangled-name='fnvlist_add_int16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_int16'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='23bd8cb5' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_uint16' mangled-name='fnvlist_add_uint16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_uint16'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='149c6638' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_int32' mangled-name='fnvlist_add_int32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_int32'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='3ff5601b' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_uint32' mangled-name='fnvlist_add_uint32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_uint32'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='8f92235e' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_int64' mangled-name='fnvlist_add_int64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_int64'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='9da381c4' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_uint64' mangled-name='fnvlist_add_uint64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_uint64'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='9c313c2d' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_string' mangled-name='fnvlist_add_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_string'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='80f4b756' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_nvlist' mangled-name='fnvlist_add_nvlist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_nvlist'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='5ce45b60' name='val'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_nvpair' mangled-name='fnvlist_add_nvpair' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_nvpair'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='3fa542f0' name='pair'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_boolean_array' mangled-name='fnvlist_add_boolean_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_boolean_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='37e3bd22' name='val'/>
	+ <parameter type-id='c5f6c15b' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_byte_array' mangled-name='fnvlist_add_byte_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_byte_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='45b65157' name='val'/>
	+ <parameter type-id='d1db479e' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_int8_array' mangled-name='fnvlist_add_int8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_int8_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='256d5229' name='val'/>
	+ <parameter type-id='a06445da' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_uint8_array' mangled-name='fnvlist_add_uint8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_uint8_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='ae3e8ca6' name='val'/>
	+ <parameter type-id='9f7200cf' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_int16_array' mangled-name='fnvlist_add_int16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_int16_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='f76f73d0' name='val'/>
	+ <parameter type-id='a3eb883d' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_uint16_array' mangled-name='fnvlist_add_uint16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_uint16_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='8a121f49' name='val'/>
	+ <parameter type-id='1b7d11c6' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_int32_array' mangled-name='fnvlist_add_int32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_int32_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='4aafb922' name='val'/>
	+ <parameter type-id='1f526493' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_uint32_array' mangled-name='fnvlist_add_uint32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_uint32_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='90421557' name='val'/>
	+ <parameter type-id='a6798dcc' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_int64_array' mangled-name='fnvlist_add_int64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_int64_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='cb785ebf' name='val'/>
	+ <parameter type-id='505bed1a' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_uint64_array' mangled-name='fnvlist_add_uint64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_uint64_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='5d6479ae' name='val'/>
	+ <parameter type-id='713a56f5' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_string_array' mangled-name='fnvlist_add_string_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_string_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='f319fae0' name='val'/>
	+ <parameter type-id='13956559' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_add_nvlist_array' mangled-name='fnvlist_add_nvlist_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_add_nvlist_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='857bb57e' name='val'/>
	+ <parameter type-id='3bbfee2e' name='val'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_remove' mangled-name='fnvlist_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_remove'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_remove_nvpair' mangled-name='fnvlist_remove_nvpair' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_remove_nvpair'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='3fa542f0' name='pair'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_nvpair' mangled-name='fnvlist_lookup_nvpair' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_nvpair'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='3fa542f0'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_boolean' mangled-name='fnvlist_lookup_boolean' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_boolean'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_boolean_value' mangled-name='fnvlist_lookup_boolean_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_boolean_value'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_byte' mangled-name='fnvlist_lookup_byte' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_byte'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='d8bf0010'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_int8' mangled-name='fnvlist_lookup_int8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_int8'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='ee31ee44'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_int16' mangled-name='fnvlist_lookup_int16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_int16'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='23bd8cb5'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_int32' mangled-name='fnvlist_lookup_int32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_int32'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='3ff5601b'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_int64' mangled-name='fnvlist_lookup_int64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_int64'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='9da381c4'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_uint8' mangled-name='fnvlist_lookup_uint8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_uint8'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_uint16' mangled-name='fnvlist_lookup_uint16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_uint16'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_uint32' mangled-name='fnvlist_lookup_uint32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_uint32'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_uint64' mangled-name='fnvlist_lookup_uint64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_uint64'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_string' mangled-name='fnvlist_lookup_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_string'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_nvlist' mangled-name='fnvlist_lookup_nvlist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_nvlist'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_boolean_array' mangled-name='fnvlist_lookup_boolean_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_boolean_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='37e3bd22'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_byte_array' mangled-name='fnvlist_lookup_byte_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_byte_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='45b65157'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_int8_array' mangled-name='fnvlist_lookup_int8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_int8_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='256d5229'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_uint8_array' mangled-name='fnvlist_lookup_uint8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_uint8_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='ae3e8ca6'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_int16_array' mangled-name='fnvlist_lookup_int16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_int16_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='f76f73d0'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_uint16_array' mangled-name='fnvlist_lookup_uint16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_uint16_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='8a121f49'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_int32_array' mangled-name='fnvlist_lookup_int32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_int32_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='4aafb922'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_uint32_array' mangled-name='fnvlist_lookup_uint32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_uint32_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='90421557'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_int64_array' mangled-name='fnvlist_lookup_int64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_int64_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='cb785ebf'/>
	</function-decl>
	<function-decl name='fnvlist_lookup_uint64_array' mangled-name='fnvlist_lookup_uint64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvlist_lookup_uint64_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='5d6479ae'/>
	</function-decl>
	<function-decl name='fnvpair_value_boolean_value' mangled-name='fnvpair_value_boolean_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_boolean_value'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='fnvpair_value_byte' mangled-name='fnvpair_value_byte' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_byte'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='d8bf0010'/>
	</function-decl>
	<function-decl name='fnvpair_value_int8' mangled-name='fnvpair_value_int8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_int8'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='ee31ee44'/>
	</function-decl>
	<function-decl name='fnvpair_value_int16' mangled-name='fnvpair_value_int16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_int16'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='23bd8cb5'/>
	</function-decl>
	<function-decl name='fnvpair_value_int32' mangled-name='fnvpair_value_int32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_int32'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='3ff5601b'/>
	</function-decl>
	<function-decl name='fnvpair_value_int64' mangled-name='fnvpair_value_int64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_int64'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='9da381c4'/>
	</function-decl>
	<function-decl name='fnvpair_value_uint8' mangled-name='fnvpair_value_uint8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_uint8'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='fnvpair_value_uint16' mangled-name='fnvpair_value_uint16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_uint16'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='fnvpair_value_uint32' mangled-name='fnvpair_value_uint32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_uint32'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='fnvpair_value_uint64' mangled-name='fnvpair_value_uint64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_uint64'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='fnvpair_value_string' mangled-name='fnvpair_value_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_string'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='fnvpair_value_nvlist' mangled-name='fnvpair_value_nvlist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fnvpair_value_nvlist'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/nvpair/nvpair.c' language='LANG_C99'>
	<pointer-type-def type-id='37e3bd22' size-in-bits='64' id='03829398'/>
	- <qualified-type-def type-id='26a90f95' const='yes' id='57de658a'/>
	- <pointer-type-def type-id='57de658a' size-in-bits='64' id='f319fae0'/>
	<pointer-type-def type-id='9b23c9ad' size-in-bits='64' id='c0563f85'/>
	+ <qualified-type-def type-id='c19b74c3' const='yes' id='12373e33'/>
	+ <pointer-type-def type-id='12373e33' size-in-bits='64' id='c5f6c15b'/>
	+ <qualified-type-def type-id='80f4b756' const='yes' id='b99c00c9'/>
	+ <pointer-type-def type-id='b99c00c9' size-in-bits='64' id='13956559'/>
	+ <qualified-type-def type-id='23bd8cb5' const='yes' id='75f7b0c5'/>
	+ <pointer-type-def type-id='75f7b0c5' size-in-bits='64' id='a3eb883d'/>
	+ <qualified-type-def type-id='3ff5601b' const='yes' id='922df12b'/>
	+ <pointer-type-def type-id='922df12b' size-in-bits='64' id='1f526493'/>
	+ <qualified-type-def type-id='9da381c4' const='yes' id='f07b7694'/>
	+ <pointer-type-def type-id='f07b7694' size-in-bits='64' id='505bed1a'/>
	+ <qualified-type-def type-id='ee31ee44' const='yes' id='721c32d4'/>
	+ <pointer-type-def type-id='721c32d4' size-in-bits='64' id='a06445da'/>
	+ <qualified-type-def type-id='8e8d4be3' const='yes' id='693c3853'/>
	+ <pointer-type-def type-id='693c3853' size-in-bits='64' id='22cce67b'/>
	+ <qualified-type-def type-id='22cce67b' const='yes' id='d2816df0'/>
	+ <pointer-type-def type-id='d2816df0' size-in-bits='64' id='3bbfee2e'/>
	+ <qualified-type-def type-id='57928edf' const='yes' id='642ee20f'/>
	+ <pointer-type-def type-id='642ee20f' size-in-bits='64' id='dace003f'/>
	+ <qualified-type-def type-id='d8bf0010' const='yes' id='a9125480'/>
	+ <pointer-type-def type-id='a9125480' size-in-bits='64' id='d1db479e'/>
	+ <qualified-type-def type-id='149c6638' const='yes' id='b01a5ac8'/>
	+ <pointer-type-def type-id='b01a5ac8' size-in-bits='64' id='1b7d11c6'/>
	+ <qualified-type-def type-id='8f92235e' const='yes' id='b9930aae'/>
	+ <pointer-type-def type-id='b9930aae' size-in-bits='64' id='a6798dcc'/>
	+ <qualified-type-def type-id='9c313c2d' const='yes' id='c3b7ba7d'/>
	+ <pointer-type-def type-id='c3b7ba7d' size-in-bits='64' id='713a56f5'/>
	+ <qualified-type-def type-id='b96825af' const='yes' id='2b61797f'/>
	+ <pointer-type-def type-id='2b61797f' size-in-bits='64' id='9f7200cf'/>
	<pointer-type-def type-id='a0eb0f08' size-in-bits='64' id='7408d286'/>
	<pointer-type-def type-id='cebdd548' size-in-bits='64' id='e379e62d'/>
	<pointer-type-def type-id='95e97e5e' size-in-bits='64' id='7292109c'/>
	<pointer-type-def type-id='f76f73d0' size-in-bits='64' id='7e73928e'/>
	<pointer-type-def type-id='4aafb922' size-in-bits='64' id='9aa04798'/>
	<pointer-type-def type-id='cb785ebf' size-in-bits='64' id='e37ce48f'/>
	<pointer-type-def type-id='256d5229' size-in-bits='64' id='ee181ab9'/>
	<pointer-type-def type-id='857bb57e' size-in-bits='64' id='75be733c'/>
	<pointer-type-def type-id='3fa542f0' size-in-bits='64' id='0b283d2e'/>
	<pointer-type-def type-id='b59d7dce' size-in-bits='64' id='78c01427'/>
	<pointer-type-def type-id='45b65157' size-in-bits='64' id='3b0247c7'/>
	<pointer-type-def type-id='8a121f49' size-in-bits='64' id='bd8768d9'/>
	<pointer-type-def type-id='90421557' size-in-bits='64' id='9507d3c7'/>
	<pointer-type-def type-id='5d6479ae' size-in-bits='64' id='892b4acc'/>
	<pointer-type-def type-id='ae3e8ca6' size-in-bits='64' id='d8774064'/>
	<pointer-type-def type-id='3502e3ff' size-in-bits='64' id='4dd26a40'/>
	<function-decl name='nv_alloc_init' mangled-name='nv_alloc_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nv_alloc_init'>
	<parameter type-id='11871392' name='nva'/>
	<parameter type-id='ee1d4944' name='nvo'/>
	<parameter is-variadic='yes'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nv_alloc_reset' mangled-name='nv_alloc_reset' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nv_alloc_reset'>
	<parameter type-id='11871392' name='nva'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nv_alloc_fini' mangled-name='nv_alloc_fini' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nv_alloc_fini'>
	<parameter type-id='11871392' name='nva'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_lookup_nv_alloc' mangled-name='nvlist_lookup_nv_alloc' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_nv_alloc'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<return type-id='11871392'/>
	</function-decl>
	<function-decl name='nvlist_nvflag' mangled-name='nvlist_nvflag' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_nvflag'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<return type-id='3502e3ff'/>
	</function-decl>
	<function-decl name='nvlist_alloc' mangled-name='nvlist_alloc' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_alloc'>
	<parameter type-id='857bb57e' name='nvlp'/>
	<parameter type-id='3502e3ff' name='nvflag'/>
	<parameter type-id='95e97e5e' name='kmflag'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_xalloc' mangled-name='nvlist_xalloc' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_xalloc'>
	<parameter type-id='857bb57e' name='nvlp'/>
	<parameter type-id='3502e3ff' name='nvflag'/>
	<parameter type-id='11871392' name='nva'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_free' mangled-name='nvlist_free' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_free'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_dup' mangled-name='nvlist_dup' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_dup'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='857bb57e' name='nvlp'/>
	<parameter type-id='95e97e5e' name='kmflag'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_xdup' mangled-name='nvlist_xdup' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_xdup'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='857bb57e' name='nvlp'/>
	<parameter type-id='11871392' name='nva'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_remove_all' mangled-name='nvlist_remove_all' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_remove_all'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_remove' mangled-name='nvlist_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_remove'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='8d0687d2' name='type'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_remove_nvpair' mangled-name='nvlist_remove_nvpair' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_remove_nvpair'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='3fa542f0' name='nvp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_boolean' mangled-name='nvlist_add_boolean' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_boolean'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_boolean_value' mangled-name='nvlist_add_boolean_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_boolean_value'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='c19b74c3' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_byte' mangled-name='nvlist_add_byte' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_byte'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='d8bf0010' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_int8' mangled-name='nvlist_add_int8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_int8'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='ee31ee44' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_uint8' mangled-name='nvlist_add_uint8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_uint8'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='b96825af' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_int16' mangled-name='nvlist_add_int16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_int16'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='23bd8cb5' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_uint16' mangled-name='nvlist_add_uint16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_uint16'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='149c6638' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_int32' mangled-name='nvlist_add_int32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_int32'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='3ff5601b' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_uint32' mangled-name='nvlist_add_uint32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_uint32'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='8f92235e' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_int64' mangled-name='nvlist_add_int64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_int64'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='9da381c4' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_uint64' mangled-name='nvlist_add_uint64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_uint64'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='9c313c2d' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_double' mangled-name='nvlist_add_double' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_double'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='a0eb0f08' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_string' mangled-name='nvlist_add_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_string'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='80f4b756' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_boolean_array' mangled-name='nvlist_add_boolean_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_boolean_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='37e3bd22' name='a'/>
	+ <parameter type-id='c5f6c15b' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_byte_array' mangled-name='nvlist_add_byte_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_byte_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='45b65157' name='a'/>
	+ <parameter type-id='d1db479e' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_int8_array' mangled-name='nvlist_add_int8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_int8_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='256d5229' name='a'/>
	+ <parameter type-id='a06445da' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_uint8_array' mangled-name='nvlist_add_uint8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_uint8_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='ae3e8ca6' name='a'/>
	+ <parameter type-id='9f7200cf' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_int16_array' mangled-name='nvlist_add_int16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_int16_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='f76f73d0' name='a'/>
	+ <parameter type-id='a3eb883d' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_uint16_array' mangled-name='nvlist_add_uint16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_uint16_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='8a121f49' name='a'/>
	+ <parameter type-id='1b7d11c6' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_int32_array' mangled-name='nvlist_add_int32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_int32_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='4aafb922' name='a'/>
	+ <parameter type-id='1f526493' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_uint32_array' mangled-name='nvlist_add_uint32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_uint32_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='90421557' name='a'/>
	+ <parameter type-id='a6798dcc' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_int64_array' mangled-name='nvlist_add_int64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_int64_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='cb785ebf' name='a'/>
	+ <parameter type-id='505bed1a' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_uint64_array' mangled-name='nvlist_add_uint64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_uint64_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='5d6479ae' name='a'/>
	+ <parameter type-id='713a56f5' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_string_array' mangled-name='nvlist_add_string_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_string_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='f319fae0' name='a'/>
	+ <parameter type-id='13956559' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_hrtime' mangled-name='nvlist_add_hrtime' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_hrtime'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='cebdd548' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_nvlist' mangled-name='nvlist_add_nvlist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_nvlist'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='5ce45b60' name='val'/>
	+ <parameter type-id='22cce67b' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_nvlist_array' mangled-name='nvlist_add_nvlist_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_nvlist_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	- <parameter type-id='857bb57e' name='a'/>
	+ <parameter type-id='3bbfee2e' name='a'/>
	<parameter type-id='3502e3ff' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_next_nvpair' mangled-name='nvlist_next_nvpair' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_next_nvpair'>
	<parameter type-id='5ce45b60' name='nvl'/>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='3fa542f0'/>
	</function-decl>
	<function-decl name='nvlist_prev_nvpair' mangled-name='nvlist_prev_nvpair' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prev_nvpair'>
	<parameter type-id='5ce45b60' name='nvl'/>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='3fa542f0'/>
	</function-decl>
	<function-decl name='nvlist_empty' mangled-name='nvlist_empty' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_empty'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='nvpair_name' mangled-name='nvpair_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_name'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='nvpair_type' mangled-name='nvpair_type' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_type'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='8d0687d2'/>
	</function-decl>
	<function-decl name='nvpair_type_is_array' mangled-name='nvpair_type_is_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_type_is_array'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_boolean' mangled-name='nvlist_lookup_boolean' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_boolean'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_boolean_value' mangled-name='nvlist_lookup_boolean_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_boolean_value'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='37e3bd22' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_byte' mangled-name='nvlist_lookup_byte' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_byte'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='45b65157' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_int8' mangled-name='nvlist_lookup_int8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_int8'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='256d5229' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_uint8' mangled-name='nvlist_lookup_uint8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_uint8'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='ae3e8ca6' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_int16' mangled-name='nvlist_lookup_int16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_int16'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='f76f73d0' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_uint16' mangled-name='nvlist_lookup_uint16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_uint16'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='8a121f49' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_int32' mangled-name='nvlist_lookup_int32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_int32'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='4aafb922' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_uint32' mangled-name='nvlist_lookup_uint32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_uint32'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='90421557' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_int64' mangled-name='nvlist_lookup_int64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_int64'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='cb785ebf' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_uint64' mangled-name='nvlist_lookup_uint64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_uint64'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='5d6479ae' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_double' mangled-name='nvlist_lookup_double' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_double'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='7408d286' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_string' mangled-name='nvlist_lookup_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_string'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='9b23c9ad' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_nvlist' mangled-name='nvlist_lookup_nvlist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_nvlist'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='857bb57e' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_boolean_array' mangled-name='nvlist_lookup_boolean_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_boolean_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='03829398' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_byte_array' mangled-name='nvlist_lookup_byte_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_byte_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='3b0247c7' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_int8_array' mangled-name='nvlist_lookup_int8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_int8_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='ee181ab9' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_uint8_array' mangled-name='nvlist_lookup_uint8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_uint8_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='d8774064' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_int16_array' mangled-name='nvlist_lookup_int16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_int16_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='7e73928e' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_uint16_array' mangled-name='nvlist_lookup_uint16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_uint16_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='bd8768d9' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_int32_array' mangled-name='nvlist_lookup_int32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_int32_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='9aa04798' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_uint32_array' mangled-name='nvlist_lookup_uint32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_uint32_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='9507d3c7' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_int64_array' mangled-name='nvlist_lookup_int64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_int64_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='e37ce48f' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_uint64_array' mangled-name='nvlist_lookup_uint64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_uint64_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='892b4acc' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_string_array' mangled-name='nvlist_lookup_string_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_string_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='c0563f85' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_nvlist_array' mangled-name='nvlist_lookup_nvlist_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_nvlist_array'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='75be733c' name='a'/>
	<parameter type-id='4dd26a40' name='n'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_hrtime' mangled-name='nvlist_lookup_hrtime' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_hrtime'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='e379e62d' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_pairs' mangled-name='nvlist_lookup_pairs' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_pairs'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='95e97e5e' name='flag'/>
	<parameter is-variadic='yes'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_nvpair' mangled-name='nvlist_lookup_nvpair' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_nvpair'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='0b283d2e' name='ret'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_lookup_nvpair_embedded_index' mangled-name='nvlist_lookup_nvpair_embedded_index' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_lookup_nvpair_embedded_index'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='0b283d2e' name='ret'/>
	<parameter type-id='7292109c' name='ip'/>
	<parameter type-id='9b23c9ad' name='ep'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_exists' mangled-name='nvlist_exists' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_exists'>
	- <parameter type-id='5ce45b60' name='nvl'/>
	+ <parameter type-id='22cce67b' name='nvl'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='nvpair_value_boolean_value' mangled-name='nvpair_value_boolean_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_boolean_value'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<parameter type-id='37e3bd22' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_byte' mangled-name='nvpair_value_byte' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_byte'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<parameter type-id='45b65157' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_int8' mangled-name='nvpair_value_int8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_int8'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<parameter type-id='256d5229' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_uint8' mangled-name='nvpair_value_uint8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_uint8'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<parameter type-id='ae3e8ca6' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_int16' mangled-name='nvpair_value_int16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_int16'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<parameter type-id='f76f73d0' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_uint16' mangled-name='nvpair_value_uint16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_uint16'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<parameter type-id='8a121f49' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_int32' mangled-name='nvpair_value_int32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_int32'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<parameter type-id='4aafb922' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_uint32' mangled-name='nvpair_value_uint32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_uint32'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<parameter type-id='90421557' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_int64' mangled-name='nvpair_value_int64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_int64'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<parameter type-id='cb785ebf' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_uint64' mangled-name='nvpair_value_uint64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_uint64'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<parameter type-id='5d6479ae' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_double' mangled-name='nvpair_value_double' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_double'>
	- <parameter type-id='3fa542f0' name='nvp'/>
	+ <parameter type-id='dace003f' name='nvp'/>
	<parameter type-id='7408d286' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_string' mangled-name='nvpair_value_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_string'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='9b23c9ad' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_nvlist' mangled-name='nvpair_value_nvlist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_nvlist'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='857bb57e' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_boolean_array' mangled-name='nvpair_value_boolean_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_boolean_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='03829398' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_byte_array' mangled-name='nvpair_value_byte_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_byte_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='3b0247c7' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_int8_array' mangled-name='nvpair_value_int8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_int8_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='ee181ab9' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_uint8_array' mangled-name='nvpair_value_uint8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_uint8_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='d8774064' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_int16_array' mangled-name='nvpair_value_int16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_int16_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='7e73928e' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_uint16_array' mangled-name='nvpair_value_uint16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_uint16_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='bd8768d9' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_int32_array' mangled-name='nvpair_value_int32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_int32_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='9aa04798' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_uint32_array' mangled-name='nvpair_value_uint32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_uint32_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='9507d3c7' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_int64_array' mangled-name='nvpair_value_int64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_int64_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='e37ce48f' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_uint64_array' mangled-name='nvpair_value_uint64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_uint64_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='892b4acc' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_string_array' mangled-name='nvpair_value_string_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_string_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='c0563f85' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_nvlist_array' mangled-name='nvpair_value_nvlist_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_nvlist_array'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='75be733c' name='val'/>
	<parameter type-id='4dd26a40' name='nelem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_hrtime' mangled-name='nvpair_value_hrtime' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_hrtime'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='e379e62d' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_add_nvpair' mangled-name='nvlist_add_nvpair' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_add_nvpair'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='3fa542f0' name='nvp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_merge' mangled-name='nvlist_merge' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_merge'>
	<parameter type-id='5ce45b60' name='dst'/>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='95e97e5e' name='flag'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_size' mangled-name='nvlist_size' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_size'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='78c01427' name='size'/>
	<parameter type-id='95e97e5e' name='encoding'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_pack' mangled-name='nvlist_pack' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_pack'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='9b23c9ad' name='bufp'/>
	<parameter type-id='78c01427' name='buflen'/>
	<parameter type-id='95e97e5e' name='encoding'/>
	<parameter type-id='95e97e5e' name='kmflag'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_xpack' mangled-name='nvlist_xpack' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_xpack'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='9b23c9ad' name='bufp'/>
	<parameter type-id='78c01427' name='buflen'/>
	<parameter type-id='95e97e5e' name='encoding'/>
	<parameter type-id='11871392' name='nva'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_unpack' mangled-name='nvlist_unpack' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_unpack'>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='b59d7dce' name='buflen'/>
	<parameter type-id='857bb57e' name='nvlp'/>
	<parameter type-id='95e97e5e' name='kmflag'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvlist_xunpack' mangled-name='nvlist_xunpack' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_xunpack'>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='b59d7dce' name='buflen'/>
	<parameter type-id='857bb57e' name='nvlp'/>
	<parameter type-id='11871392' name='nva'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/nvpair/nvpair_alloc_fixed.c' language='LANG_C99'>
	<var-decl name='nv_fixed_ops' type-id='ee1d4944' mangled-name='nv_fixed_ops' visibility='default' elf-symbol-id='nv_fixed_ops'/>
	</abi-instr>
	<abi-instr address-size='64' path='libnvpair.c' language='LANG_C99'>
	<type-decl name='char' size-in-bits='8' id='a84c031d'/>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='8' id='89feb1ec'>
	<subrange length='1' type-id='7359adad' id='52f813b4'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='160' id='664ac0b7'>
	<subrange length='20' type-id='7359adad' id='fdca39cf'/>
	</array-type-def>
	- <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
	- <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
	- <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
	- <class-decl name='re_dfa_t' is-struct='yes' visibility='default' is-declaration-only='yes' id='b48d2441'/>
	<type-decl name='double' size-in-bits='64' id='a0eb0f08'/>
	<type-decl name='int' size-in-bits='32' id='95e97e5e'/>
	<type-decl name='long int' size-in-bits='64' id='bd54fe1a'/>
	<type-decl name='long long int' size-in-bits='64' id='1eb56b1e'/>
	<type-decl name='short int' size-in-bits='16' id='a2185560'/>
	<type-decl name='signed char' size-in-bits='8' id='28577a57'/>
	<type-decl name='unnamed-enum-underlying-type-32' is-anonymous='yes' size-in-bits='32' alignment-in-bits='32' id='9cac1fee'/>
	<type-decl name='unsigned char' size-in-bits='8' id='002ac4a6'/>
	<type-decl name='unsigned int' size-in-bits='32' id='f0981eeb'/>
	<type-decl name='unsigned long int' size-in-bits='64' id='7359adad'/>
	<type-decl name='unsigned short int' size-in-bits='16' id='8efea9e5'/>
	<type-decl name='variadic parameter type' id='2c1145c5'/>
	- <type-decl name='void' id='48b5725f'/>
	<typedef-decl name='nvlist_prtctl_t' type-id='196db161' id='b0c1ff8d'/>
	<enum-decl name='nvlist_indent_mode' id='628aafab'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='NVLIST_INDENT_ABS' value='0'/>
	<enumerator name='NVLIST_INDENT_TABBED' value='1'/>
	</enum-decl>
	<enum-decl name='nvlist_prtctl_fmt' id='c8dcc53a'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='NVLIST_FMT_MEMBER_NAME' value='0'/>
	<enumerator name='NVLIST_FMT_MEMBER_POSTAMBLE' value='1'/>
	<enumerator name='NVLIST_FMT_BTWN_ARRAY' value='2'/>
	</enum-decl>
	<enum-decl name='data_type_t' naming-typedef-id='8d0687d2' id='aeeae136'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='DATA_TYPE_DONTCARE' value='-1'/>
	<enumerator name='DATA_TYPE_UNKNOWN' value='0'/>
	<enumerator name='DATA_TYPE_BOOLEAN' value='1'/>
	<enumerator name='DATA_TYPE_BYTE' value='2'/>
	<enumerator name='DATA_TYPE_INT16' value='3'/>
	<enumerator name='DATA_TYPE_UINT16' value='4'/>
	<enumerator name='DATA_TYPE_INT32' value='5'/>
	<enumerator name='DATA_TYPE_UINT32' value='6'/>
	<enumerator name='DATA_TYPE_INT64' value='7'/>
	<enumerator name='DATA_TYPE_UINT64' value='8'/>
	<enumerator name='DATA_TYPE_STRING' value='9'/>
	<enumerator name='DATA_TYPE_BYTE_ARRAY' value='10'/>
	<enumerator name='DATA_TYPE_INT16_ARRAY' value='11'/>
	<enumerator name='DATA_TYPE_UINT16_ARRAY' value='12'/>
	<enumerator name='DATA_TYPE_INT32_ARRAY' value='13'/>
	<enumerator name='DATA_TYPE_UINT32_ARRAY' value='14'/>
	<enumerator name='DATA_TYPE_INT64_ARRAY' value='15'/>
	<enumerator name='DATA_TYPE_UINT64_ARRAY' value='16'/>
	<enumerator name='DATA_TYPE_STRING_ARRAY' value='17'/>
	<enumerator name='DATA_TYPE_HRTIME' value='18'/>
	<enumerator name='DATA_TYPE_NVLIST' value='19'/>
	<enumerator name='DATA_TYPE_NVLIST_ARRAY' value='20'/>
	<enumerator name='DATA_TYPE_BOOLEAN_VALUE' value='21'/>
	<enumerator name='DATA_TYPE_INT8' value='22'/>
	<enumerator name='DATA_TYPE_UINT8' value='23'/>
	<enumerator name='DATA_TYPE_BOOLEAN_ARRAY' value='24'/>
	<enumerator name='DATA_TYPE_INT8_ARRAY' value='25'/>
	<enumerator name='DATA_TYPE_UINT8_ARRAY' value='26'/>
	<enumerator name='DATA_TYPE_DOUBLE' value='27'/>
	</enum-decl>
	<typedef-decl name='data_type_t' type-id='aeeae136' id='8d0687d2'/>
	<class-decl name='nvpair' size-in-bits='128' is-struct='yes' visibility='default' id='1c34e459'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='nvp_size' type-id='3ff5601b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='nvp_name_sz' type-id='23bd8cb5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='48'>
	<var-decl name='nvp_reserve' type-id='23bd8cb5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='nvp_value_elem' type-id='3ff5601b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='96'>
	<var-decl name='nvp_type' type-id='8d0687d2' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='nvpair_t' type-id='1c34e459' id='57928edf'/>
	<class-decl name='nvlist' size-in-bits='192' is-struct='yes' visibility='default' id='ac266fd9'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='nvl_version' type-id='3ff5601b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='nvl_nvflag' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='nvl_priv' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='nvl_flag' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='160'>
	<var-decl name='nvl_pad' type-id='3ff5601b' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='nvlist_t' type-id='ac266fd9' id='8e8d4be3'/>
	<enum-decl name='boolean_t' naming-typedef-id='c19b74c3' id='f58c8277'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='B_FALSE' value='0'/>
	<enumerator name='B_TRUE' value='1'/>
	</enum-decl>
	<typedef-decl name='boolean_t' type-id='f58c8277' id='c19b74c3'/>
	<typedef-decl name='uchar_t' type-id='002ac4a6' id='d8bf0010'/>
	<typedef-decl name='uint_t' type-id='f0981eeb' id='3502e3ff'/>
	<typedef-decl name='hrtime_t' type-id='1eb56b1e' id='cebdd548'/>
	<class-decl name='nvlist_printops' size-in-bits='3456' is-struct='yes' visibility='default' id='ebc6735b'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='print_boolean' type-id='e7f43f72' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='print_boolean_value' type-id='e7f43f73' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='print_byte' type-id='e7f43f74' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='print_int8' type-id='e7f43f75' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='print_uint8' type-id='e7f43f76' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='640'>
	<var-decl name='print_int16' type-id='e7f43f77' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='768'>
	<var-decl name='print_uint16' type-id='e7f43f78' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='896'>
	<var-decl name='print_int32' type-id='e7f43f79' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1024'>
	<var-decl name='print_uint32' type-id='e7f43f7a' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1152'>
	<var-decl name='print_int64' type-id='e7f43f7b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1280'>
	<var-decl name='print_uint64' type-id='e7f43f7c' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1408'>
	<var-decl name='print_double' type-id='e7f43f7d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1536'>
	<var-decl name='print_string' type-id='e7f43f7e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1664'>
	<var-decl name='print_hrtime' type-id='e7f43f7f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1792'>
	<var-decl name='print_nvlist' type-id='e7f43f80' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1920'>
	<var-decl name='print_boolean_array' type-id='e7f43f81' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2048'>
	<var-decl name='print_byte_array' type-id='e7f43f82' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2176'>
	<var-decl name='print_int8_array' type-id='e7f43f83' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2304'>
	<var-decl name='print_uint8_array' type-id='e7f43f84' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2432'>
	<var-decl name='print_int16_array' type-id='e7f43f85' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2560'>
	<var-decl name='print_uint16_array' type-id='e7f43f86' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2688'>
	<var-decl name='print_int32_array' type-id='e7f43f87' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2816'>
	<var-decl name='print_uint32_array' type-id='e7f43f88' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2944'>
	<var-decl name='print_int64_array' type-id='e7f43f89' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='3072'>
	<var-decl name='print_uint64_array' type-id='e7f43f8a' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='3200'>
	<var-decl name='print_string_array' type-id='e7f43f8b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='3328'>
	<var-decl name='print_nvlist_array' type-id='e7f43f8c' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f72'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='6d994334' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__1' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f73'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='6a2f50c1' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__2' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f74'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='8a1fb33a' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__3' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f75'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='506696a8' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__4' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f76'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='39b623f9' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__5' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f77'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='ea6be4eb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__6' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f78'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='f10f1e84' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__7' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f79'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='1708018d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__8' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f7a'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='90174072' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__9' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f7b'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='d2af7f32' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__10' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f7c'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='0b22f759' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__11' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f7d'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='3be4d568' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__12' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f7e'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='c0d0f877' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__13' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f7f'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='e1c54c3c' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__14' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f80'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='19ea27ae' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__15' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f81'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='7ef0e988' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__16' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f82'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='7391ed39' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__17' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f83'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='42257af5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__18' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f84'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='330cc0d0' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__19' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f85'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='506ab59a' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__20' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f86'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='ed6a3a3d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__21' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f87'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='750cc41c' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__22' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f88'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='292cdbcf' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__23' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f89'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='aaea91b5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__24' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f8a'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='7e85a9b6' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__25' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f8b'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='de20bf07' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='__anonymous_struct__26' size-in-bits='128' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f8c'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='op' type-id='2835af80' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='nvlist_prtctl' size-in-bits='576' is-struct='yes' visibility='default' id='d2e8bad9'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='nvprt_fp' type-id='822cd80b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='nvprt_indent_mode' type-id='628aafab' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='96'>
	<var-decl name='nvprt_indent' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='nvprt_indentinc' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='nvprt_nmfmt' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='nvprt_eomfmt' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='nvprt_btwnarrfmt' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='nvprt_btwnarrfmt_nl' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='nvprt_dfltops' type-id='7be54adb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='nvprt_custops' type-id='7be54adb' visibility='default'/>
	</data-member>
	</class-decl>
	- <typedef-decl name='__re_long_size_t' type-id='7359adad' id='ba516949'/>
	<typedef-decl name='reg_syntax_t' type-id='7359adad' id='1b72c3b3'/>
	<class-decl name='re_pattern_buffer' size-in-bits='512' is-struct='yes' visibility='default' id='19fc9a8c'>
	<data-member access='public' layout-offset-in-bits='0'>
	- <var-decl name='buffer' type-id='33976309' visibility='default'/>
	+ <var-decl name='buffer' type-id='cf536864' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	- <var-decl name='allocated' type-id='ba516949' visibility='default'/>
	+ <var-decl name='allocated' type-id='7359adad' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	- <var-decl name='used' type-id='ba516949' visibility='default'/>
	+ <var-decl name='used' type-id='7359adad' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='syntax' type-id='1b72c3b3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='fastmap' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='translate' type-id='cf536864' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='re_nsub' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='can_be_null' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='449'>
	<var-decl name='regs_allocated' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='451'>
	<var-decl name='fastmap_accurate' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='452'>
	<var-decl name='no_sub' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='453'>
	<var-decl name='not_bol' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='454'>
	<var-decl name='not_eol' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='455'>
	<var-decl name='newline_anchor' type-id='f0981eeb' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='regex_t' type-id='19fc9a8c' id='aca3bac8'/>
	- <typedef-decl name='int8_t' type-id='2171a512' id='ee31ee44'/>
	- <typedef-decl name='int16_t' type-id='03896e23' id='23bd8cb5'/>
	- <typedef-decl name='int32_t' type-id='33f57a65' id='3ff5601b'/>
	- <typedef-decl name='int64_t' type-id='0c9942d2' id='9da381c4'/>
	- <typedef-decl name='uint8_t' type-id='c51d6389' id='b96825af'/>
	- <typedef-decl name='uint16_t' type-id='253c2d2a' id='149c6638'/>
	- <typedef-decl name='uint32_t' type-id='62f1140c' id='8f92235e'/>
	- <typedef-decl name='uint64_t' type-id='8910171f' id='9c313c2d'/>
	- <typedef-decl name='__int8_t' type-id='28577a57' id='2171a512'/>
	- <typedef-decl name='__uint8_t' type-id='002ac4a6' id='c51d6389'/>
	- <typedef-decl name='__int16_t' type-id='a2185560' id='03896e23'/>
	- <typedef-decl name='__uint16_t' type-id='8efea9e5' id='253c2d2a'/>
	- <typedef-decl name='__int32_t' type-id='95e97e5e' id='33f57a65'/>
	- <typedef-decl name='__uint32_t' type-id='f0981eeb' id='62f1140c'/>
	- <typedef-decl name='__int64_t' type-id='bd54fe1a' id='0c9942d2'/>
	- <typedef-decl name='__uint64_t' type-id='7359adad' id='8910171f'/>
	- <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
	- <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
	- <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
	<typedef-decl name='_IO_lock_t' type-id='48b5725f' id='bb4788fa'/>
	+ <class-decl name='_IO_marker' size-in-bits='192' is-struct='yes' visibility='default' id='010ae0b9'>
	+ <data-member access='public' layout-offset-in-bits='0'>
	+ <var-decl name='_next' type-id='e4c6fa61' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='64'>
	+ <var-decl name='_sbuf' type-id='dca988a5' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='128'>
	+ <var-decl name='_pos' type-id='95e97e5e' visibility='default'/>
	+ </data-member>
	+ </class-decl>
	<class-decl name='_IO_FILE' size-in-bits='1728' is-struct='yes' visibility='default' id='ec1ed955'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='_flags' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='_IO_read_ptr' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='_IO_read_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='_IO_read_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='_IO_write_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='_IO_write_ptr' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='_IO_write_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='_IO_buf_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='_IO_buf_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='576'>
	<var-decl name='_IO_save_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='640'>
	<var-decl name='_IO_backup_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='704'>
	<var-decl name='_IO_save_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='768'>
	<var-decl name='_markers' type-id='e4c6fa61' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='832'>
	<var-decl name='_chain' type-id='dca988a5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='896'>
	<var-decl name='_fileno' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='928'>
	<var-decl name='_flags2' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='960'>
	<var-decl name='_old_offset' type-id='79989e9c' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1024'>
	<var-decl name='_cur_column' type-id='8efea9e5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1040'>
	<var-decl name='_vtable_offset' type-id='28577a57' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1048'>
	<var-decl name='_shortbuf' type-id='89feb1ec' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1088'>
	<var-decl name='_lock' type-id='cecf4ea7' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1152'>
	<var-decl name='_offset' type-id='724e4de6' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1216'>
	- <var-decl name='_codecvt' type-id='570f8c59' visibility='default'/>
	+ <var-decl name='__pad1' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1280'>
	- <var-decl name='_wide_data' type-id='c65a1f29' visibility='default'/>
	+ <var-decl name='__pad2' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1344'>
	- <var-decl name='_freeres_list' type-id='dca988a5' visibility='default'/>
	+ <var-decl name='__pad3' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1408'>
	- <var-decl name='_freeres_buf' type-id='eaa32e2f' visibility='default'/>
	+ <var-decl name='__pad4' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1472'>
	<var-decl name='__pad5' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1536'>
	<var-decl name='_mode' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1568'>
	<var-decl name='_unused2' type-id='664ac0b7' visibility='default'/>
	</data-member>
	</class-decl>
	+ <typedef-decl name='int8_t' type-id='2171a512' id='ee31ee44'/>
	+ <typedef-decl name='int16_t' type-id='03896e23' id='23bd8cb5'/>
	+ <typedef-decl name='int32_t' type-id='33f57a65' id='3ff5601b'/>
	+ <typedef-decl name='int64_t' type-id='0c9942d2' id='9da381c4'/>
	+ <typedef-decl name='uint8_t' type-id='c51d6389' id='b96825af'/>
	+ <typedef-decl name='uint16_t' type-id='253c2d2a' id='149c6638'/>
	+ <typedef-decl name='uint32_t' type-id='62f1140c' id='8f92235e'/>
	+ <typedef-decl name='uint64_t' type-id='8910171f' id='9c313c2d'/>
	+ <typedef-decl name='__int8_t' type-id='28577a57' id='2171a512'/>
	+ <typedef-decl name='__uint8_t' type-id='002ac4a6' id='c51d6389'/>
	+ <typedef-decl name='__int16_t' type-id='a2185560' id='03896e23'/>
	+ <typedef-decl name='__uint16_t' type-id='8efea9e5' id='253c2d2a'/>
	+ <typedef-decl name='__int32_t' type-id='95e97e5e' id='33f57a65'/>
	+ <typedef-decl name='__uint32_t' type-id='f0981eeb' id='62f1140c'/>
	+ <typedef-decl name='__int64_t' type-id='bd54fe1a' id='0c9942d2'/>
	+ <typedef-decl name='__uint64_t' type-id='7359adad' id='8910171f'/>
	+ <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
	+ <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
	+ <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
	<typedef-decl name='size_t' type-id='7359adad' id='b59d7dce'/>
	<pointer-type-def type-id='aa12d1ba' size-in-bits='64' id='822cd80b'/>
	<pointer-type-def type-id='ec1ed955' size-in-bits='64' id='dca988a5'/>
	- <pointer-type-def type-id='a4036571' size-in-bits='64' id='570f8c59'/>
	<pointer-type-def type-id='bb4788fa' size-in-bits='64' id='cecf4ea7'/>
	<pointer-type-def type-id='010ae0b9' size-in-bits='64' id='e4c6fa61'/>
	- <pointer-type-def type-id='79bd3751' size-in-bits='64' id='c65a1f29'/>
	<pointer-type-def type-id='c19b74c3' size-in-bits='64' id='37e3bd22'/>
	<pointer-type-def type-id='a84c031d' size-in-bits='64' id='26a90f95'/>
	<pointer-type-def type-id='26a90f95' size-in-bits='64' id='9b23c9ad'/>
	<qualified-type-def type-id='a84c031d' const='yes' id='9b45d938'/>
	<pointer-type-def type-id='9b45d938' size-in-bits='64' id='80f4b756'/>
	<pointer-type-def type-id='9f88f76e' size-in-bits='64' id='7ef0e988'/>
	<pointer-type-def type-id='c5bb1a2b' size-in-bits='64' id='c0d0f877'/>
	<pointer-type-def type-id='573fea1b' size-in-bits='64' id='de20bf07'/>
	<pointer-type-def type-id='70284cc6' size-in-bits='64' id='3be4d568'/>
	<pointer-type-def type-id='700c3bca' size-in-bits='64' id='6d994334'/>
	<pointer-type-def type-id='18ac1860' size-in-bits='64' id='506ab59a'/>
	<pointer-type-def type-id='328fee42' size-in-bits='64' id='750cc41c'/>
	<pointer-type-def type-id='7ba5cd31' size-in-bits='64' id='aaea91b5'/>
	<pointer-type-def type-id='a86d8029' size-in-bits='64' id='42257af5'/>
	<pointer-type-def type-id='0b4eb914' size-in-bits='64' id='19ea27ae'/>
	<pointer-type-def type-id='c6c8144e' size-in-bits='64' id='2835af80'/>
	<pointer-type-def type-id='20f7b475' size-in-bits='64' id='6a2f50c1'/>
	<pointer-type-def type-id='102ee17a' size-in-bits='64' id='e1c54c3c'/>
	<pointer-type-def type-id='49b69c77' size-in-bits='64' id='ea6be4eb'/>
	<pointer-type-def type-id='cb5d50f1' size-in-bits='64' id='1708018d'/>
	<pointer-type-def type-id='880d56b8' size-in-bits='64' id='d2af7f32'/>
	<pointer-type-def type-id='a739bfc6' size-in-bits='64' id='506696a8'/>
	<pointer-type-def type-id='234f35e8' size-in-bits='64' id='8a1fb33a'/>
	<pointer-type-def type-id='41f7168a' size-in-bits='64' id='f10f1e84'/>
	<pointer-type-def type-id='e8d6e508' size-in-bits='64' id='90174072'/>
	<pointer-type-def type-id='f3daafe5' size-in-bits='64' id='0b22f759'/>
	<pointer-type-def type-id='17ab04ad' size-in-bits='64' id='39b623f9'/>
	<pointer-type-def type-id='256cdd75' size-in-bits='64' id='7391ed39'/>
	<pointer-type-def type-id='cc10a041' size-in-bits='64' id='ed6a3a3d'/>
	<pointer-type-def type-id='9fd269d3' size-in-bits='64' id='292cdbcf'/>
	<pointer-type-def type-id='3bd73b0c' size-in-bits='64' id='7e85a9b6'/>
	<pointer-type-def type-id='0d445e26' size-in-bits='64' id='330cc0d0'/>
	<pointer-type-def type-id='e4b89f30' size-in-bits='64' id='ed8aa8ba'/>
	<pointer-type-def type-id='be7f4941' size-in-bits='64' id='2809de35'/>
	<pointer-type-def type-id='fe5ae69d' size-in-bits='64' id='90d5edb9'/>
	<pointer-type-def type-id='2783af3c' size-in-bits='64' id='e44553b6'/>
	<pointer-type-def type-id='33c6e3d8' size-in-bits='64' id='1263777a'/>
	<pointer-type-def type-id='dadb9eca' size-in-bits='64' id='cbda43ac'/>
	<pointer-type-def type-id='55b9e070' size-in-bits='64' id='b3fae562'/>
	<pointer-type-def type-id='8e63c78b' size-in-bits='64' id='8b41e457'/>
	<pointer-type-def type-id='c542ed33' size-in-bits='64' id='f9668a57'/>
	<pointer-type-def type-id='5dea179a' size-in-bits='64' id='001d8764'/>
	<pointer-type-def type-id='b6f659a0' size-in-bits='64' id='44f188f2'/>
	<pointer-type-def type-id='2765bd17' size-in-bits='64' id='976f721b'/>
	<pointer-type-def type-id='9e073b5c' size-in-bits='64' id='ee62ad8e'/>
	<pointer-type-def type-id='2c785071' size-in-bits='64' id='957d9f35'/>
	<pointer-type-def type-id='aad19bf7' size-in-bits='64' id='4db8acf3'/>
	<pointer-type-def type-id='0660e71a' size-in-bits='64' id='0ca7b13c'/>
	<pointer-type-def type-id='250287b8' size-in-bits='64' id='a91bad5a'/>
	<pointer-type-def type-id='e7344862' size-in-bits='64' id='519bf35c'/>
	<pointer-type-def type-id='32b6d968' size-in-bits='64' id='92988dea'/>
	<pointer-type-def type-id='5c975642' size-in-bits='64' id='7f8ee7e4'/>
	<pointer-type-def type-id='0155b993' size-in-bits='64' id='2c8c4457'/>
	<pointer-type-def type-id='6e8b02cb' size-in-bits='64' id='eb944897'/>
	<pointer-type-def type-id='d434b7d7' size-in-bits='64' id='108e6453'/>
	<pointer-type-def type-id='c645e10f' size-in-bits='64' id='5cbe16ab'/>
	<pointer-type-def type-id='de41f295' size-in-bits='64' id='d94cdfa1'/>
	<pointer-type-def type-id='b2fbf64a' size-in-bits='64' id='470a7fd4'/>
	<pointer-type-def type-id='cc22d314' size-in-bits='64' id='eddda806'/>
	<pointer-type-def type-id='23bd8cb5' size-in-bits='64' id='f76f73d0'/>
	<pointer-type-def type-id='3ff5601b' size-in-bits='64' id='4aafb922'/>
	<pointer-type-def type-id='9da381c4' size-in-bits='64' id='cb785ebf'/>
	<pointer-type-def type-id='ee31ee44' size-in-bits='64' id='256d5229'/>
	<pointer-type-def type-id='ebc6735b' size-in-bits='64' id='7be54adb'/>
	<pointer-type-def type-id='d2e8bad9' size-in-bits='64' id='196db161'/>
	<pointer-type-def type-id='8e8d4be3' size-in-bits='64' id='5ce45b60'/>
	<pointer-type-def type-id='5ce45b60' size-in-bits='64' id='857bb57e'/>
	<pointer-type-def type-id='57928edf' size-in-bits='64' id='3fa542f0'/>
	- <pointer-type-def type-id='b48d2441' size-in-bits='64' id='33976309'/>
	<pointer-type-def type-id='aca3bac8' size-in-bits='64' id='d33f11cb'/>
	<pointer-type-def type-id='d8bf0010' size-in-bits='64' id='45b65157'/>
	<pointer-type-def type-id='149c6638' size-in-bits='64' id='8a121f49'/>
	<pointer-type-def type-id='8f92235e' size-in-bits='64' id='90421557'/>
	<pointer-type-def type-id='9c313c2d' size-in-bits='64' id='5d6479ae'/>
	<pointer-type-def type-id='b96825af' size-in-bits='64' id='ae3e8ca6'/>
	<pointer-type-def type-id='002ac4a6' size-in-bits='64' id='cf536864'/>
	<pointer-type-def type-id='48b5725f' size-in-bits='64' id='eaa32e2f'/>
	- <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
	- <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
	- <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
	- <class-decl name='re_dfa_t' is-struct='yes' visibility='default' is-declaration-only='yes' id='b48d2441'/>
	<function-decl name='nvlist_prtctl_setdest' mangled-name='nvlist_prtctl_setdest' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctl_setdest'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='822cd80b' name='fp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctl_getdest' mangled-name='nvlist_prtctl_getdest' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctl_getdest'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<return type-id='822cd80b'/>
	</function-decl>
	<function-decl name='nvlist_prtctl_setindent' mangled-name='nvlist_prtctl_setindent' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctl_setindent'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='628aafab' name='mode'/>
	<parameter type-id='95e97e5e' name='start'/>
	<parameter type-id='95e97e5e' name='inc'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctl_doindent' mangled-name='nvlist_prtctl_doindent' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctl_doindent'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='95e97e5e' name='onemore'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctl_setfmt' mangled-name='nvlist_prtctl_setfmt' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctl_setfmt'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='c8dcc53a' name='which'/>
	<parameter type-id='80f4b756' name='fmt'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctl_dofmt' mangled-name='nvlist_prtctl_dofmt' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctl_dofmt'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='c8dcc53a' name='which'/>
	<parameter is-variadic='yes'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_boolean' mangled-name='nvlist_prtctlop_boolean' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_boolean'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='1263777a' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_boolean_value' mangled-name='nvlist_prtctlop_boolean_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_boolean_value'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='976f721b' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_byte' mangled-name='nvlist_prtctlop_byte' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_byte'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='519bf35c' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_int8' mangled-name='nvlist_prtctlop_int8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_int8'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='a91bad5a' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_uint8' mangled-name='nvlist_prtctlop_uint8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_uint8'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='eb944897' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_int16' mangled-name='nvlist_prtctlop_int16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_int16'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='957d9f35' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_uint16' mangled-name='nvlist_prtctlop_uint16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_uint16'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='92988dea' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_int32' mangled-name='nvlist_prtctlop_int32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_int32'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='4db8acf3' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_uint32' mangled-name='nvlist_prtctlop_uint32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_uint32'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='7f8ee7e4' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_int64' mangled-name='nvlist_prtctlop_int64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_int64'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='0ca7b13c' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_uint64' mangled-name='nvlist_prtctlop_uint64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_uint64'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='2c8c4457' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_double' mangled-name='nvlist_prtctlop_double' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_double'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='e44553b6' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_string' mangled-name='nvlist_prtctlop_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_string'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='2809de35' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_hrtime' mangled-name='nvlist_prtctlop_hrtime' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_hrtime'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='ee62ad8e' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_nvlist' mangled-name='nvlist_prtctlop_nvlist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_nvlist'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='001d8764' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_boolean_array' mangled-name='nvlist_prtctlop_boolean_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_boolean_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='ed8aa8ba' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_byte_array' mangled-name='nvlist_prtctlop_byte_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_byte_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='108e6453' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_int8_array' mangled-name='nvlist_prtctlop_int8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_int8_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='f9668a57' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_uint8_array' mangled-name='nvlist_prtctlop_uint8_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_uint8_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='eddda806' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_int16_array' mangled-name='nvlist_prtctlop_int16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_int16_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='cbda43ac' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_uint16_array' mangled-name='nvlist_prtctlop_uint16_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_uint16_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='5cbe16ab' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_int32_array' mangled-name='nvlist_prtctlop_int32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_int32_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='b3fae562' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_uint32_array' mangled-name='nvlist_prtctlop_uint32_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_uint32_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='d94cdfa1' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_int64_array' mangled-name='nvlist_prtctlop_int64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_int64_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='8b41e457' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_uint64_array' mangled-name='nvlist_prtctlop_uint64_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_uint64_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='470a7fd4' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_string_array' mangled-name='nvlist_prtctlop_string_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_string_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='90d5edb9' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctlop_nvlist_array' mangled-name='nvlist_prtctlop_nvlist_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctlop_nvlist_array'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<parameter type-id='44f188f2' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prtctl_alloc' mangled-name='nvlist_prtctl_alloc' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctl_alloc'>
	<return type-id='b0c1ff8d'/>
	</function-decl>
	<function-decl name='nvlist_prtctl_free' mangled-name='nvlist_prtctl_free' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prtctl_free'>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_print' mangled-name='nvlist_print' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_print'>
	<parameter type-id='822cd80b' name='fp'/>
	<parameter type-id='5ce45b60' name='nvl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvlist_prt' mangled-name='nvlist_prt' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_prt'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='b0c1ff8d' name='pctl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='dump_nvlist' mangled-name='dump_nvlist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='dump_nvlist'>
	<parameter type-id='5ce45b60' name='list'/>
	<parameter type-id='95e97e5e' name='indent'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='nvpair_value_match_regex' mangled-name='nvpair_value_match_regex' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_match_regex'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='95e97e5e' name='ai'/>
	<parameter type-id='26a90f95' name='value'/>
	<parameter type-id='d33f11cb' name='value_regex'/>
	<parameter type-id='9b23c9ad' name='ep'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='nvpair_value_match' mangled-name='nvpair_value_match' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvpair_value_match'>
	<parameter type-id='3fa542f0' name='nvp'/>
	<parameter type-id='95e97e5e' name='ai'/>
	<parameter type-id='26a90f95' name='value'/>
	<parameter type-id='9b23c9ad' name='ep'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-type size-in-bits='64' id='9f88f76e'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='37e3bd22'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='c5bb1a2b'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='26a90f95'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='573fea1b'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='9b23c9ad'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='70284cc6'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='a0eb0f08'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='700c3bca'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='95e97e5e'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='18ac1860'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='f76f73d0'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='328fee42'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='4aafb922'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='7ba5cd31'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='cb785ebf'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='a86d8029'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='256d5229'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='0b4eb914'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='5ce45b60'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='c6c8144e'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='857bb57e'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='20f7b475'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='c19b74c3'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='102ee17a'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='cebdd548'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='49b69c77'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='23bd8cb5'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='cb5d50f1'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='3ff5601b'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='880d56b8'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='9da381c4'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='a739bfc6'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='ee31ee44'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='234f35e8'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='d8bf0010'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='41f7168a'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='149c6638'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='e8d6e508'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='8f92235e'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='f3daafe5'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='9c313c2d'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='17ab04ad'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='b96825af'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='256cdd75'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='45b65157'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='cc10a041'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='8a121f49'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='9fd269d3'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='90421557'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='3bd73b0c'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='5d6479ae'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='0d445e26'>
	<parameter type-id='196db161'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='ae3e8ca6'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='e4b89f30'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='37e3bd22'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='be7f4941'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='26a90f95'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='fe5ae69d'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='9b23c9ad'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='2783af3c'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='a0eb0f08'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='33c6e3d8'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='95e97e5e'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='dadb9eca'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='f76f73d0'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='55b9e070'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='4aafb922'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='8e63c78b'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='cb785ebf'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='c542ed33'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='256d5229'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='5dea179a'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='5ce45b60'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='b6f659a0'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='857bb57e'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='2765bd17'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='c19b74c3'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='9e073b5c'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='cebdd548'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='2c785071'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='23bd8cb5'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='aad19bf7'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='3ff5601b'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='0660e71a'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='9da381c4'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='250287b8'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='ee31ee44'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='e7344862'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='d8bf0010'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='32b6d968'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='149c6638'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='5c975642'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='8f92235e'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='0155b993'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='9c313c2d'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='6e8b02cb'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='b96825af'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='d434b7d7'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='45b65157'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='c645e10f'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='8a121f49'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='de41f295'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='90421557'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='b2fbf64a'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='5d6479ae'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='cc22d314'>
	<parameter type-id='b0c1ff8d'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='ae3e8ca6'/>
	<parameter type-id='3502e3ff'/>
	<return type-id='95e97e5e'/>
	</function-type>
	+ <type-decl name='void' id='48b5725f'/>
	</abi-instr>
	<abi-instr address-size='64' path='libnvpair_json.c' language='LANG_C99'>
	<function-decl name='nvlist_print_json' mangled-name='nvlist_print_json' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='nvlist_print_json'>
	<parameter type-id='822cd80b' name='fp'/>
	<parameter type-id='5ce45b60' name='nvl'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='nvpair_alloc_system.c' language='LANG_C99'>
	<class-decl name='__va_list_tag' size-in-bits='192' is-struct='yes' visibility='default' id='d5027220'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='gp_offset' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='fp_offset' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='overflow_arg_area' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='reg_save_area' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='nv_alloc_ops_t' type-id='8f6cc4f4' id='03e8ffd6'/>
	<class-decl name='nv_alloc' size-in-bits='128' is-struct='yes' visibility='default' id='98213087'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='nva_ops' type-id='ee1d4944' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='nva_arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='nv_alloc_t' type-id='98213087' id='cca08635'/>
	<class-decl name='nv_alloc_ops' size-in-bits='320' is-struct='yes' visibility='default' id='8f6cc4f4'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='nv_ao_init' type-id='76da8447' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='nv_ao_fini' type-id='fe356f6f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='nv_ao_alloc' type-id='9ff7f508' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='nv_ao_free' type-id='520da3f4' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='nv_ao_reset' type-id='fe356f6f' visibility='default'/>
	</data-member>
	</class-decl>
	<pointer-type-def type-id='d5027220' size-in-bits='64' id='b7f2d5e6'/>
	<qualified-type-def type-id='03e8ffd6' const='yes' id='aca16c06'/>
	<pointer-type-def type-id='aca16c06' size-in-bits='64' id='ee1d4944'/>
	<pointer-type-def type-id='e9ff7293' size-in-bits='64' id='76da8447'/>
	<pointer-type-def type-id='cca08635' size-in-bits='64' id='11871392'/>
	<pointer-type-def type-id='51a21b4b' size-in-bits='64' id='fe356f6f'/>
	<pointer-type-def type-id='1169c032' size-in-bits='64' id='520da3f4'/>
	<pointer-type-def type-id='9fff962e' size-in-bits='64' id='9ff7f508'/>
	<var-decl name='nv_alloc_nosleep' type-id='11871392' mangled-name='nv_alloc_nosleep' visibility='default' elf-symbol-id='nv_alloc_nosleep'/>
	<function-type size-in-bits='64' id='e9ff7293'>
	<parameter type-id='11871392'/>
	<parameter type-id='b7f2d5e6'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='51a21b4b'>
	<parameter type-id='11871392'/>
	<return type-id='48b5725f'/>
	</function-type>
	<function-type size-in-bits='64' id='1169c032'>
	<parameter type-id='11871392'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='b59d7dce'/>
	<return type-id='48b5725f'/>
	</function-type>
	<function-type size-in-bits='64' id='9fff962e'>
	<parameter type-id='11871392'/>
	<parameter type-id='b59d7dce'/>
	<return type-id='eaa32e2f'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='assert.c' language='LANG_C99'>
	<var-decl name='libspl_assert_ok' type-id='95e97e5e' mangled-name='libspl_assert_ok' visibility='default' elf-symbol-id='libspl_assert_ok'/>
	<function-decl name='libspl_assertf' mangled-name='libspl_assertf' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libspl_assertf'>
	<parameter type-id='80f4b756' name='file'/>
	<parameter type-id='80f4b756' name='func'/>
	<parameter type-id='95e97e5e' name='line'/>
	<parameter type-id='80f4b756' name='format'/>
	<parameter is-variadic='yes'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	</abi-corpus>
	diff --git a/sys/contrib/openzfs/lib/libspl/os/linux/getmntany.c b/sys/contrib/openzfs/lib/libspl/os/linux/getmntany.c
	index d458b28ad309..3713ff38e17f 100644
	--- a/sys/contrib/openzfs/lib/libspl/os/linux/getmntany.c
	+++ b/sys/contrib/openzfs/lib/libspl/os/linux/getmntany.c
	@@ -1,162 +1,162 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License, Version 1.0 only
	* (the "License"). You may not use this file except in compliance
	* with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright 2005 Sun Microsystems, Inc. All rights reserved.
	* Copyright 2006 Ricardo Correia. All rights reserved.
	* Use is subject to license terms.
	*/

	/* Copyright (c) 1988 AT&T */
	/* All Rights Reserved */

	#include <stdio.h>
	#include <string.h>
	#include <mntent.h>
	#include <sys/errno.h>
	#include <sys/mnttab.h>

	#include <sys/types.h>
	#include <sys/sysmacros.h>
	#include <sys/stat.h>
	#include <unistd.h>

	#define BUFSIZE (MNT_LINE_MAX + 2)

	static __thread char buf[BUFSIZE];

	#define DIFF(xx) ( \
	(mrefp->xx != NULL) && \
	(mgetp->xx == NULL \|\| strcmp(mrefp->xx, mgetp->xx) != 0))

	int
	getmntany(FILE fp, struct mnttab mgetp, struct mnttab *mrefp)
	{
	int ret;

	while (
	((ret = _sol_getmntent(fp, mgetp)) == 0) && (
	DIFF(mnt_special) \|\| DIFF(mnt_mountp) \|\|
	DIFF(mnt_fstype) \|\| DIFF(mnt_mntopts))) { }

	return (ret);
	}

	int
	_sol_getmntent(FILE fp, struct mnttab mgetp)
	{
	struct mntent mntbuf;
	struct mntent *ret;

	ret = getmntent_r(fp, &mntbuf, buf, BUFSIZE);

	if (ret != NULL) {
	mgetp->mnt_special = mntbuf.mnt_fsname;
	mgetp->mnt_mountp = mntbuf.mnt_dir;
	mgetp->mnt_fstype = mntbuf.mnt_type;
	mgetp->mnt_mntopts = mntbuf.mnt_opts;
	return (0);
	}

	if (feof(fp))
	return (-1);

	return (MNT_TOOLONG);
	}

	static int
	-getextmntent_impl(FILE fp, struct extmnttab mp, int len)
	+getextmntent_impl(FILE fp, struct extmnttab mp)
	{
	int ret;
	struct stat64 st;

	ret = _sol_getmntent(fp, (struct mnttab *)mp);
	if (ret == 0) {
	if (stat64(mp->mnt_mountp, &st) != 0) {
	mp->mnt_major = 0;
	mp->mnt_minor = 0;
	return (ret);
	}
	mp->mnt_major = major(st.st_dev);
	mp->mnt_minor = minor(st.st_dev);
	}

	return (ret);
	}

	int
	getextmntent(const char path, struct extmnttab entry, struct stat64 *statbuf)
	{
	struct stat64 st;
	FILE *fp;
	int match;

	if (strlen(path) >= MAXPATHLEN) {
	(void) fprintf(stderr, "invalid object; pathname too long\n");
	return (-1);
	}

	/*
	* Search for the path in /proc/self/mounts. Rather than looking for the
	* specific path, which can be fooled by non-standard paths (i.e. ".."
	* or "//"), we stat() the path and search for the corresponding
	* (major,minor) device pair.
	*/
	if (stat64(path, statbuf) != 0) {
	(void) fprintf(stderr, "cannot open '%s': %s\n",
	path, strerror(errno));
	return (-1);
	}


	if ((fp = fopen(MNTTAB, "re")) == NULL) {
	(void) fprintf(stderr, "cannot open %s\n", MNTTAB);
	return (-1);
	}

	/*
	* Search for the given (major,minor) pair in the mount table.
	*/

	match = 0;
	- while (getextmntent_impl(fp, entry, sizeof (*entry)) == 0) {
	+ while (getextmntent_impl(fp, entry) == 0) {
	if (makedev(entry->mnt_major, entry->mnt_minor) ==
	statbuf->st_dev) {
	match = 1;
	break;
	}
	}
	(void) fclose(fp);

	if (!match) {
	(void) fprintf(stderr, "cannot find mountpoint for '%s'\n",
	path);
	return (-1);
	}

	if (stat64(entry->mnt_mountp, &st) != 0) {
	entry->mnt_major = 0;
	entry->mnt_minor = 0;
	return (-1);
	}

	return (0);
	}
	diff --git a/sys/contrib/openzfs/lib/libuutil/libuutil.abi b/sys/contrib/openzfs/lib/libuutil/libuutil.abi
	index c7ab5672eace..48575ebc6a9f 100644
	--- a/sys/contrib/openzfs/lib/libuutil/libuutil.abi
	+++ b/sys/contrib/openzfs/lib/libuutil/libuutil.abi
	@@ -1,1841 +1,1846 @@
	<abi-corpus version='2.0' architecture='elf-amd-x86_64' soname='libuutil.so.3'>
	<elf-needed>
	<dependency name='libpthread.so.0'/>
	<dependency name='libc.so.6'/>
	<dependency name='ld-linux-x86-64.so.2'/>
	</elf-needed>
	<elf-function-symbols>
	+ <elf-symbol name='_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='_sol_getmntent' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_char' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_char_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_int' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_int_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_long' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_long_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_ptr_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_short' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_short_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_clear_long_excl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_set_long_excl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_char' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_char_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_int' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_int_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_long' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_long_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_ptr_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_short' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_short_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_add' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_destroy_nodes' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_find' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_first' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_insert' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_insert_here' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_is_empty' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_last' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_nearest' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_numnodes' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_swap' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_update' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_update_gt' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_update_lt' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_walk' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='get_system_hostid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getexecname' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getextmntent' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getmntany' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getzoneid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libspl_assertf' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_head' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_after' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_before' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_head' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_is_empty' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_link_active' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_link_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_link_replace' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_move_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_next' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_prev' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_remove_head' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_remove_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_consumer' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_enter' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_exit' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_producer' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='mkdirp' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='print_timestamp' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='spl_pagesize' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='strlcat' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='strlcpy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_alt_exit' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_find' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_first' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_insert' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_last' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_lockup' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_nearest_next' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_nearest_prev' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_next' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_node_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_node_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_numnodes' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_pool_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_pool_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_prev' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_release' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_teardown' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_walk' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_walk_end' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_walk_next' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_avl_walk_start' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_check_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_die' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_error' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_exit_fatal' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_exit_ok' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_exit_usage' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_free' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_getpname' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_find' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_first' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_insert' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_insert_after' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_insert_before' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_last' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_lockup' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_nearest_next' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_nearest_prev' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_next' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_node_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_node_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_numnodes' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_pool_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_pool_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_prev' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_release' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_teardown' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_walk' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_walk_end' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_walk_next' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_list_walk_start' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_memdup' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_msprintf' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_panic' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_set_error' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_setpname' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_strbw' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_strcaseeq' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_strdup' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_streq' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_strerror' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_strndup' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_vdie' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_vwarn' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_vxdie' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_warn' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_xdie' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_zalloc' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	</elf-function-symbols>
	<elf-variable-symbols>
	<elf-symbol name='libspl_assert_ok' size='4' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_exit_fatal_value' size='4' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_exit_ok_value' size='4' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='uu_exit_usage_value' size='4' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	</elf-variable-symbols>
	<abi-instr address-size='64' path='../../module/avl/avl.c' language='LANG_C99'>
	<typedef-decl name='avl_tree_t' type-id='b351119f' id='f20fbd51'/>
	<typedef-decl name='avl_index_t' type-id='e475ab95' id='fba6cb51'/>
	<pointer-type-def type-id='fba6cb51' size-in-bits='64' id='32adbf30'/>
	<pointer-type-def type-id='f20fbd51' size-in-bits='64' id='a3681dea'/>
	<function-decl name='avl_walk' mangled-name='avl_walk' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_walk'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='oldnode'/>
	<parameter type-id='95e97e5e' name='left'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='avl_first' mangled-name='avl_first' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_first'>
	<parameter type-id='a3681dea' name='tree'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='avl_last' mangled-name='avl_last' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_last'>
	<parameter type-id='a3681dea' name='tree'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='avl_nearest' mangled-name='avl_nearest' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_nearest'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='fba6cb51' name='where'/>
	<parameter type-id='95e97e5e' name='direction'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='avl_find' mangled-name='avl_find' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_find'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='value'/>
	<parameter type-id='32adbf30' name='where'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='avl_insert' mangled-name='avl_insert' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_insert'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='new_data'/>
	<parameter type-id='fba6cb51' name='where'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_insert_here' mangled-name='avl_insert_here' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_insert_here'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='new_data'/>
	<parameter type-id='eaa32e2f' name='here'/>
	<parameter type-id='95e97e5e' name='direction'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_add' mangled-name='avl_add' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_add'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='new_node'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_remove' mangled-name='avl_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_remove'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_update_lt' mangled-name='avl_update_lt' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_update_lt'>
	<parameter type-id='a3681dea' name='t'/>
	<parameter type-id='eaa32e2f' name='obj'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='avl_update_gt' mangled-name='avl_update_gt' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_update_gt'>
	<parameter type-id='a3681dea' name='t'/>
	<parameter type-id='eaa32e2f' name='obj'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='avl_update' mangled-name='avl_update' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_update'>
	<parameter type-id='a3681dea' name='t'/>
	<parameter type-id='eaa32e2f' name='obj'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='avl_swap' mangled-name='avl_swap' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_swap'>
	<parameter type-id='a3681dea' name='tree1'/>
	<parameter type-id='a3681dea' name='tree2'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_create' mangled-name='avl_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_create'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='585e1de9' name='compar'/>
	<parameter type-id='b59d7dce' name='size'/>
	<parameter type-id='b59d7dce' name='offset'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_destroy' mangled-name='avl_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_destroy'>
	<parameter type-id='a3681dea' name='tree'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_numnodes' mangled-name='avl_numnodes' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_numnodes'>
	<parameter type-id='a3681dea' name='tree'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='avl_is_empty' mangled-name='avl_is_empty' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_is_empty'>
	<parameter type-id='a3681dea' name='tree'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='avl_destroy_nodes' mangled-name='avl_destroy_nodes' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_destroy_nodes'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='63e171df' name='cookie'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-type size-in-bits='64' id='96ee24a5'>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='eaa32e2f'/>
	<return type-id='95e97e5e'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='assert.c' language='LANG_C99'>
	<var-decl name='libspl_assert_ok' type-id='95e97e5e' mangled-name='libspl_assert_ok' visibility='default' elf-symbol-id='libspl_assert_ok'/>
	<function-decl name='libspl_assertf' mangled-name='libspl_assertf' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libspl_assertf'>
	<parameter type-id='80f4b756' name='file'/>
	<parameter type-id='80f4b756' name='func'/>
	<parameter type-id='95e97e5e' name='line'/>
	<parameter type-id='80f4b756' name='format'/>
	<parameter is-variadic='yes'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='atomic.c' language='LANG_C99'>
	<type-decl name='unsigned short int' size-in-bits='16' id='8efea9e5'/>
	<typedef-decl name='int16_t' type-id='03896e23' id='23bd8cb5'/>
	<typedef-decl name='int32_t' type-id='33f57a65' id='3ff5601b'/>
	<typedef-decl name='uint16_t' type-id='253c2d2a' id='149c6638'/>
	<typedef-decl name='__int16_t' type-id='a2185560' id='03896e23'/>
	<typedef-decl name='__uint16_t' type-id='8efea9e5' id='253c2d2a'/>
	<typedef-decl name='__int32_t' type-id='95e97e5e' id='33f57a65'/>
	<typedef-decl name='__ssize_t' type-id='bd54fe1a' id='41060289'/>
	<typedef-decl name='ssize_t' type-id='41060289' id='79a0948f'/>
	<qualified-type-def type-id='149c6638' volatile='yes' id='5120c5f7'/>
	<pointer-type-def type-id='5120c5f7' size-in-bits='64' id='93977ae7'/>
	<qualified-type-def type-id='8f92235e' volatile='yes' id='430e0681'/>
	<pointer-type-def type-id='430e0681' size-in-bits='64' id='3a147f31'/>
	<qualified-type-def type-id='b96825af' volatile='yes' id='84ff7d66'/>
	<pointer-type-def type-id='84ff7d66' size-in-bits='64' id='aa323ea4'/>
	<qualified-type-def type-id='ee1f298e' volatile='yes' id='6f7e09cb'/>
	<pointer-type-def type-id='6f7e09cb' size-in-bits='64' id='64698d33'/>
	<qualified-type-def type-id='48b5725f' volatile='yes' id='b0b3cbf9'/>
	<pointer-type-def type-id='b0b3cbf9' size-in-bits='64' id='fe09dd29'/>
	<function-decl name='atomic_inc_8' mangled-name='atomic_inc_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_16' mangled-name='atomic_inc_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_16'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_32' mangled-name='atomic_inc_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_32'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_ulong' mangled-name='atomic_inc_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_8' mangled-name='atomic_dec_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_16' mangled-name='atomic_dec_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_16'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_32' mangled-name='atomic_dec_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_32'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_ulong' mangled-name='atomic_dec_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_8' mangled-name='atomic_add_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_16' mangled-name='atomic_add_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_32' mangled-name='atomic_add_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_long' mangled-name='atomic_add_long' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_long'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_ptr' mangled-name='atomic_add_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_8' mangled-name='atomic_sub_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_16' mangled-name='atomic_sub_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_32' mangled-name='atomic_sub_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_long' mangled-name='atomic_sub_long' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_long'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_ptr' mangled-name='atomic_sub_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_8' mangled-name='atomic_or_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_16' mangled-name='atomic_or_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_32' mangled-name='atomic_or_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_ulong' mangled-name='atomic_or_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_8' mangled-name='atomic_and_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_16' mangled-name='atomic_and_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_32' mangled-name='atomic_and_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_ulong' mangled-name='atomic_and_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_8_nv' mangled-name='atomic_inc_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_inc_16_nv' mangled-name='atomic_inc_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_inc_32_nv' mangled-name='atomic_inc_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_inc_ulong_nv' mangled-name='atomic_inc_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_dec_8_nv' mangled-name='atomic_dec_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_dec_16_nv' mangled-name='atomic_dec_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_dec_32_nv' mangled-name='atomic_dec_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_dec_ulong_nv' mangled-name='atomic_dec_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_add_8_nv' mangled-name='atomic_add_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_add_16_nv' mangled-name='atomic_add_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_add_32_nv' mangled-name='atomic_add_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_add_long_nv' mangled-name='atomic_add_long_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_long_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_add_ptr_nv' mangled-name='atomic_add_ptr_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_ptr_nv'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_sub_8_nv' mangled-name='atomic_sub_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_sub_16_nv' mangled-name='atomic_sub_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_sub_32_nv' mangled-name='atomic_sub_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_sub_long_nv' mangled-name='atomic_sub_long_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_long_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_sub_ptr_nv' mangled-name='atomic_sub_ptr_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_ptr_nv'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_or_8_nv' mangled-name='atomic_or_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_or_16_nv' mangled-name='atomic_or_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_or_32_nv' mangled-name='atomic_or_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_or_ulong_nv' mangled-name='atomic_or_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_and_8_nv' mangled-name='atomic_and_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_and_16_nv' mangled-name='atomic_and_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_and_32_nv' mangled-name='atomic_and_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_and_ulong_nv' mangled-name='atomic_and_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_cas_8' mangled-name='atomic_cas_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='exp'/>
	<parameter type-id='b96825af' name='des'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_cas_16' mangled-name='atomic_cas_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='exp'/>
	<parameter type-id='149c6638' name='des'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_cas_32' mangled-name='atomic_cas_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='exp'/>
	<parameter type-id='8f92235e' name='des'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_cas_ulong' mangled-name='atomic_cas_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='exp'/>
	<parameter type-id='ee1f298e' name='des'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_cas_ptr' mangled-name='atomic_cas_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='eaa32e2f' name='exp'/>
	<parameter type-id='eaa32e2f' name='des'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_swap_8' mangled-name='atomic_swap_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_swap_16' mangled-name='atomic_swap_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_swap_32' mangled-name='atomic_swap_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_swap_ulong' mangled-name='atomic_swap_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_swap_ptr' mangled-name='atomic_swap_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='eaa32e2f' name='bits'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_set_long_excl' mangled-name='atomic_set_long_excl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_set_long_excl'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='3502e3ff' name='value'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='atomic_clear_long_excl' mangled-name='atomic_clear_long_excl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_clear_long_excl'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='3502e3ff' name='value'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='membar_enter' mangled-name='membar_enter' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='membar_enter'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='membar_producer' mangled-name='membar_producer' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='membar_producer'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='membar_consumer' mangled-name='membar_consumer' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='membar_consumer'>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='getexecname.c' language='LANG_C99'>
	<function-decl name='getexecname' mangled-name='getexecname' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getexecname'>
	<return type-id='80f4b756'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='list.c' language='LANG_C99'>
	<typedef-decl name='list_node_t' type-id='b0b5e45e' id='b21843b2'/>
	<typedef-decl name='list_t' type-id='e824dae9' id='0899125f'/>
	<class-decl name='list_node' size-in-bits='128' is-struct='yes' visibility='default' id='b0b5e45e'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='next' type-id='b03eadb4' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='prev' type-id='b03eadb4' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='list' size-in-bits='256' is-struct='yes' visibility='default' id='e824dae9'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='list_size' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='list_offset' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='list_head' type-id='b0b5e45e' visibility='default'/>
	</data-member>
	</class-decl>
	<pointer-type-def type-id='b0b5e45e' size-in-bits='64' id='b03eadb4'/>
	<pointer-type-def type-id='b21843b2' size-in-bits='64' id='ccc38265'/>
	<pointer-type-def type-id='0899125f' size-in-bits='64' id='352ec160'/>
	<function-decl name='list_create' mangled-name='list_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_create'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='b59d7dce' name='size'/>
	<parameter type-id='b59d7dce' name='offset'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_destroy' mangled-name='list_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_destroy'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_after' mangled-name='list_insert_after' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_after'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<parameter type-id='eaa32e2f' name='nobject'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_before' mangled-name='list_insert_before' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_before'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<parameter type-id='eaa32e2f' name='nobject'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_head' mangled-name='list_insert_head' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_head'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_tail' mangled-name='list_insert_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_tail'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_remove' mangled-name='list_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_remove'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_remove_head' mangled-name='list_remove_head' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_remove_head'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_remove_tail' mangled-name='list_remove_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_remove_tail'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_head' mangled-name='list_head' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_head'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_tail' mangled-name='list_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_tail'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_next' mangled-name='list_next' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_next'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_prev' mangled-name='list_prev' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_prev'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_move_tail' mangled-name='list_move_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_move_tail'>
	<parameter type-id='352ec160' name='dst'/>
	<parameter type-id='352ec160' name='src'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_link_replace' mangled-name='list_link_replace' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_link_replace'>
	<parameter type-id='ccc38265' name='lold'/>
	<parameter type-id='ccc38265' name='lnew'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_link_init' mangled-name='list_link_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_link_init'>
	<parameter type-id='ccc38265' name='ln'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_link_active' mangled-name='list_link_active' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_link_active'>
	<parameter type-id='ccc38265' name='ln'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='list_is_empty' mangled-name='list_is_empty' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_is_empty'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='mkdirp.c' language='LANG_C99'>
	<typedef-decl name='__mode_t' type-id='f0981eeb' id='e1c52942'/>
	<typedef-decl name='mode_t' type-id='e1c52942' id='d50d396c'/>
	<function-decl name='mkdirp' mangled-name='mkdirp' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='mkdirp'>
	<parameter type-id='80f4b756' name='d'/>
	<parameter type-id='d50d396c' name='mode'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/gethostid.c' language='LANG_C99'>
	<function-decl name='get_system_hostid' mangled-name='get_system_hostid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='get_system_hostid'>
	<return type-id='7359adad'/>
	</function-decl>
	<type-decl name='unsigned long int' size-in-bits='64' id='7359adad'/>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/getmntany.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='03085adc' size-in-bits='192' id='083f8d58'>
	<subrange length='3' type-id='7359adad' id='56f209d2'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='8' id='89feb1ec'>
	<subrange length='1' type-id='7359adad' id='52f813b4'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='160' id='664ac0b7'>
	<subrange length='20' type-id='7359adad' id='fdca39cf'/>
	</array-type-def>
	- <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
	- <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
	- <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
	<class-decl name='mnttab' size-in-bits='256' is-struct='yes' visibility='default' id='1b055409'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='mnt_special' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='mnt_mountp' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='mnt_fstype' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='mnt_mntopts' type-id='26a90f95' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='extmnttab' size-in-bits='320' is-struct='yes' visibility='default' id='0c544dc0'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='mnt_special' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='mnt_mountp' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='mnt_fstype' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='mnt_mntopts' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='mnt_major' type-id='3502e3ff' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='288'>
	<var-decl name='mnt_minor' type-id='3502e3ff' visibility='default'/>
	</data-member>
	</class-decl>
	- <class-decl name='stat64' size-in-bits='1152' is-struct='yes' visibility='default' id='0bbec9cd'>
	+ <typedef-decl name='_IO_lock_t' type-id='48b5725f' id='bb4788fa'/>
	+ <class-decl name='_IO_marker' size-in-bits='192' is-struct='yes' visibility='default' id='010ae0b9'>
	<data-member access='public' layout-offset-in-bits='0'>
	- <var-decl name='st_dev' type-id='35ed8932' visibility='default'/>
	+ <var-decl name='_next' type-id='e4c6fa61' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	- <var-decl name='st_ino' type-id='71288a47' visibility='default'/>
	+ <var-decl name='_sbuf' type-id='dca988a5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	- <var-decl name='st_nlink' type-id='80f0b9df' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='192'>
	- <var-decl name='st_mode' type-id='e1c52942' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='224'>
	- <var-decl name='st_uid' type-id='cc5fcceb' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='256'>
	- <var-decl name='st_gid' type-id='d94ec6d9' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='288'>
	- <var-decl name='__pad0' type-id='95e97e5e' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='320'>
	- <var-decl name='st_rdev' type-id='35ed8932' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='384'>
	- <var-decl name='st_size' type-id='79989e9c' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='448'>
	- <var-decl name='st_blksize' type-id='d3f10a7f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='512'>
	- <var-decl name='st_blocks' type-id='4e711bf1' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='576'>
	- <var-decl name='st_atim' type-id='a9c79a1f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='704'>
	- <var-decl name='st_mtim' type-id='a9c79a1f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='832'>
	- <var-decl name='st_ctim' type-id='a9c79a1f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='960'>
	- <var-decl name='__glibc_reserved' type-id='083f8d58' visibility='default'/>
	+ <var-decl name='_pos' type-id='95e97e5e' visibility='default'/>
	</data-member>
	</class-decl>
	- <typedef-decl name='__dev_t' type-id='7359adad' id='35ed8932'/>
	- <typedef-decl name='__uid_t' type-id='f0981eeb' id='cc5fcceb'/>
	- <typedef-decl name='__gid_t' type-id='f0981eeb' id='d94ec6d9'/>
	- <typedef-decl name='__ino64_t' type-id='7359adad' id='71288a47'/>
	- <typedef-decl name='__nlink_t' type-id='7359adad' id='80f0b9df'/>
	- <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
	- <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
	- <typedef-decl name='__time_t' type-id='bd54fe1a' id='65eda9c0'/>
	- <typedef-decl name='__blksize_t' type-id='bd54fe1a' id='d3f10a7f'/>
	- <typedef-decl name='__blkcnt64_t' type-id='bd54fe1a' id='4e711bf1'/>
	- <typedef-decl name='__syscall_slong_t' type-id='bd54fe1a' id='03085adc'/>
	- <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
	- <typedef-decl name='_IO_lock_t' type-id='48b5725f' id='bb4788fa'/>
	<class-decl name='_IO_FILE' size-in-bits='1728' is-struct='yes' visibility='default' id='ec1ed955'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='_flags' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='_IO_read_ptr' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='_IO_read_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='_IO_read_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='_IO_write_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='_IO_write_ptr' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='_IO_write_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='_IO_buf_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='_IO_buf_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='576'>
	<var-decl name='_IO_save_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='640'>
	<var-decl name='_IO_backup_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='704'>
	<var-decl name='_IO_save_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='768'>
	<var-decl name='_markers' type-id='e4c6fa61' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='832'>
	<var-decl name='_chain' type-id='dca988a5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='896'>
	<var-decl name='_fileno' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='928'>
	<var-decl name='_flags2' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='960'>
	<var-decl name='_old_offset' type-id='79989e9c' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1024'>
	<var-decl name='_cur_column' type-id='8efea9e5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1040'>
	<var-decl name='_vtable_offset' type-id='28577a57' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1048'>
	<var-decl name='_shortbuf' type-id='89feb1ec' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1088'>
	<var-decl name='_lock' type-id='cecf4ea7' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1152'>
	<var-decl name='_offset' type-id='724e4de6' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1216'>
	- <var-decl name='_codecvt' type-id='570f8c59' visibility='default'/>
	+ <var-decl name='__pad1' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1280'>
	- <var-decl name='_wide_data' type-id='c65a1f29' visibility='default'/>
	+ <var-decl name='__pad2' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1344'>
	- <var-decl name='_freeres_list' type-id='dca988a5' visibility='default'/>
	+ <var-decl name='__pad3' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1408'>
	- <var-decl name='_freeres_buf' type-id='eaa32e2f' visibility='default'/>
	+ <var-decl name='__pad4' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1472'>
	<var-decl name='__pad5' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1536'>
	<var-decl name='_mode' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1568'>
	<var-decl name='_unused2' type-id='664ac0b7' visibility='default'/>
	</data-member>
	</class-decl>
	+ <class-decl name='stat64' size-in-bits='1152' is-struct='yes' visibility='default' id='0bbec9cd'>
	+ <data-member access='public' layout-offset-in-bits='0'>
	+ <var-decl name='st_dev' type-id='35ed8932' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='64'>
	+ <var-decl name='st_ino' type-id='71288a47' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='128'>
	+ <var-decl name='st_nlink' type-id='80f0b9df' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='192'>
	+ <var-decl name='st_mode' type-id='e1c52942' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='224'>
	+ <var-decl name='st_uid' type-id='cc5fcceb' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='256'>
	+ <var-decl name='st_gid' type-id='d94ec6d9' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='288'>
	+ <var-decl name='__pad0' type-id='95e97e5e' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='320'>
	+ <var-decl name='st_rdev' type-id='35ed8932' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='384'>
	+ <var-decl name='st_size' type-id='79989e9c' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='448'>
	+ <var-decl name='st_blksize' type-id='d3f10a7f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='512'>
	+ <var-decl name='st_blocks' type-id='4e711bf1' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='576'>
	+ <var-decl name='st_atim' type-id='a9c79a1f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='704'>
	+ <var-decl name='st_mtim' type-id='a9c79a1f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='832'>
	+ <var-decl name='st_ctim' type-id='a9c79a1f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='960'>
	+ <var-decl name='__glibc_reserved' type-id='083f8d58' visibility='default'/>
	+ </data-member>
	+ </class-decl>
	+ <typedef-decl name='__dev_t' type-id='7359adad' id='35ed8932'/>
	+ <typedef-decl name='__uid_t' type-id='f0981eeb' id='cc5fcceb'/>
	+ <typedef-decl name='__gid_t' type-id='f0981eeb' id='d94ec6d9'/>
	+ <typedef-decl name='__ino64_t' type-id='7359adad' id='71288a47'/>
	+ <typedef-decl name='__nlink_t' type-id='7359adad' id='80f0b9df'/>
	+ <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
	+ <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
	+ <typedef-decl name='__time_t' type-id='bd54fe1a' id='65eda9c0'/>
	+ <typedef-decl name='__blksize_t' type-id='bd54fe1a' id='d3f10a7f'/>
	+ <typedef-decl name='__blkcnt64_t' type-id='bd54fe1a' id='4e711bf1'/>
	+ <typedef-decl name='__syscall_slong_t' type-id='bd54fe1a' id='03085adc'/>
	+ <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
	<class-decl name='timespec' size-in-bits='128' is-struct='yes' visibility='default' id='a9c79a1f'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='tv_sec' type-id='65eda9c0' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='tv_nsec' type-id='03085adc' visibility='default'/>
	</data-member>
	</class-decl>
	<pointer-type-def type-id='aa12d1ba' size-in-bits='64' id='822cd80b'/>
	<pointer-type-def type-id='ec1ed955' size-in-bits='64' id='dca988a5'/>
	- <pointer-type-def type-id='a4036571' size-in-bits='64' id='570f8c59'/>
	<pointer-type-def type-id='bb4788fa' size-in-bits='64' id='cecf4ea7'/>
	<pointer-type-def type-id='010ae0b9' size-in-bits='64' id='e4c6fa61'/>
	- <pointer-type-def type-id='79bd3751' size-in-bits='64' id='c65a1f29'/>
	<pointer-type-def type-id='0c544dc0' size-in-bits='64' id='394fc496'/>
	<pointer-type-def type-id='1b055409' size-in-bits='64' id='9d424d31'/>
	<pointer-type-def type-id='0bbec9cd' size-in-bits='64' id='62f7a03d'/>
	- <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
	- <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
	- <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
	<function-decl name='getmntany' mangled-name='getmntany' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getmntany'>
	<parameter type-id='822cd80b' name='fp'/>
	<parameter type-id='9d424d31' name='mgetp'/>
	<parameter type-id='9d424d31' name='mrefp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='_sol_getmntent' mangled-name='_sol_getmntent' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='_sol_getmntent'>
	<parameter type-id='822cd80b' name='fp'/>
	<parameter type-id='9d424d31' name='mgetp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='getextmntent' mangled-name='getextmntent' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getextmntent'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='394fc496' name='entry'/>
	<parameter type-id='62f7a03d' name='statbuf'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/zone.c' language='LANG_C99'>
	<typedef-decl name='zoneid_t' type-id='95e97e5e' id='4da03624'/>
	<function-decl name='getzoneid' mangled-name='getzoneid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getzoneid'>
	<return type-id='4da03624'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='page.c' language='LANG_C99'>
	<function-decl name='spl_pagesize' mangled-name='spl_pagesize' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='spl_pagesize'>
	<return type-id='b59d7dce'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='strlcat.c' language='LANG_C99'>
	<function-decl name='strlcat' mangled-name='strlcat' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='strlcat'>
	<parameter type-id='26a90f95' name='dst'/>
	<parameter type-id='80f4b756' name='src'/>
	<parameter type-id='b59d7dce' name='dstsize'/>
	<return type-id='b59d7dce'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='strlcpy.c' language='LANG_C99'>
	<function-decl name='strlcpy' mangled-name='strlcpy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='strlcpy'>
	<parameter type-id='26a90f95' name='dst'/>
	<parameter type-id='80f4b756' name='src'/>
	<parameter type-id='b59d7dce' name='len'/>
	<return type-id='b59d7dce'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='timestamp.c' language='LANG_C99'>
	<function-decl name='print_timestamp' mangled-name='print_timestamp' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='print_timestamp'>
	<parameter type-id='3502e3ff' name='timestamp_fmt'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='uu_alloc.c' language='LANG_C99'>
	<type-decl name='char' size-in-bits='8' id='a84c031d'/>
	<type-decl name='unsigned long int' size-in-bits='64' id='7359adad'/>
	<type-decl name='variadic parameter type' id='2c1145c5'/>
	- <type-decl name='void' id='48b5725f'/>
	<typedef-decl name='size_t' type-id='7359adad' id='b59d7dce'/>
	<pointer-type-def type-id='a84c031d' size-in-bits='64' id='26a90f95'/>
	<qualified-type-def type-id='a84c031d' const='yes' id='9b45d938'/>
	<pointer-type-def type-id='9b45d938' size-in-bits='64' id='80f4b756'/>
	<pointer-type-def type-id='48b5725f' size-in-bits='64' id='eaa32e2f'/>
	<function-decl name='uu_zalloc' mangled-name='uu_zalloc' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_zalloc'>
	<parameter type-id='b59d7dce' name='n'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_free' mangled-name='uu_free' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_free'>
	<parameter type-id='eaa32e2f' name='p'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_strdup' mangled-name='uu_strdup' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_strdup'>
	<parameter type-id='80f4b756' name='str'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='uu_strndup' mangled-name='uu_strndup' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_strndup'>
	<parameter type-id='80f4b756' name='s'/>
	<parameter type-id='b59d7dce' name='n'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='uu_memdup' mangled-name='uu_memdup' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_memdup'>
	<parameter type-id='eaa32e2f' name='buf'/>
	<parameter type-id='b59d7dce' name='sz'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_msprintf' mangled-name='uu_msprintf' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_msprintf'>
	<parameter type-id='80f4b756' name='format'/>
	<parameter is-variadic='yes'/>
	<return type-id='26a90f95'/>
	</function-decl>
	+ <type-decl name='void' id='48b5725f'/>
	</abi-instr>
	<abi-instr address-size='64' path='uu_avl.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='bf311473' size-in-bits='128' id='f0f65199'>
	<subrange length='2' type-id='7359adad' id='52efc4ef'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='320' id='36c46961'>
	<subrange length='40' type-id='7359adad' id='8f80b239'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='512' id='59daf3ef'>
	<subrange length='64' type-id='7359adad' id='b10be967'/>
	</array-type-def>
	<type-decl name='int' size-in-bits='32' id='95e97e5e'/>
	<type-decl name='long int' size-in-bits='64' id='bd54fe1a'/>
	<type-decl name='short int' size-in-bits='16' id='a2185560'/>
	<type-decl name='signed char' size-in-bits='8' id='28577a57'/>
	<array-type-def dimensions='1' type-id='e475ab95' size-in-bits='192' id='0ce65a8b'>
	<subrange length='3' type-id='7359adad' id='56f209d2'/>
	</array-type-def>
	<type-decl name='unsigned char' size-in-bits='8' id='002ac4a6'/>
	<type-decl name='unsigned int' size-in-bits='32' id='f0981eeb'/>
	<typedef-decl name='uu_compare_fn_t' type-id='add6e811' id='40f93560'/>
	<typedef-decl name='uu_walk_fn_t' type-id='96ee24a5' id='9d1aa0dc'/>
	<typedef-decl name='uu_avl_pool_t' type-id='12a530a8' id='7f84e390'/>
	<typedef-decl name='uu_avl_t' type-id='4af029d1' id='bb7f0973'/>
	<class-decl name='uu_avl_node' size-in-bits='192' is-struct='yes' visibility='default' id='f65f4326'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='uan_opaque' type-id='0ce65a8b' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='uu_avl_node_t' type-id='f65f4326' id='73a65116'/>
	<typedef-decl name='uu_avl_walk_t' type-id='e70a39e3' id='edd8457b'/>
	<typedef-decl name='uu_avl_index_t' type-id='e475ab95' id='5d7f5fc8'/>
	<class-decl name='uu_avl_walk' size-in-bits='320' is-struct='yes' visibility='default' id='e70a39e3'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='uaw_next' type-id='5842d146' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='uaw_prev' type-id='5842d146' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='uaw_avl' type-id='a5c21a38' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='uaw_next_result' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='uaw_dir' type-id='ee31ee44' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='264'>
	<var-decl name='uaw_robust' type-id='b96825af' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='uu_avl' size-in-bits='960' is-struct='yes' visibility='default' id='4af029d1'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='ua_next_enc' type-id='e475ab95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='ua_prev_enc' type-id='e475ab95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='ua_pool' type-id='de82c773' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='ua_parent_enc' type-id='e475ab95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='ua_debug' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='264'>
	<var-decl name='ua_index' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='ua_tree' type-id='b351119f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='640'>
	<var-decl name='ua_null_walk' type-id='edd8457b' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='uu_avl_pool' size-in-bits='2176' is-struct='yes' visibility='default' id='12a530a8'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='uap_next' type-id='de82c773' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='uap_prev' type-id='de82c773' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='uap_name' type-id='59daf3ef' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='640'>
	<var-decl name='uap_nodeoffset' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='704'>
	<var-decl name='uap_objsize' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='768'>
	<var-decl name='uap_cmp' type-id='d502b39f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='832'>
	<var-decl name='uap_debug' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='840'>
	<var-decl name='uap_last_index' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='896'>
	<var-decl name='uap_lock' type-id='7a6844eb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1216'>
	<var-decl name='uap_null_avl' type-id='bb7f0973' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='avl_node' size-in-bits='192' is-struct='yes' visibility='default' id='428b67b3'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='avl_child' type-id='f0f65199' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='avl_pcb' type-id='e475ab95' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='avl_tree' size-in-bits='320' is-struct='yes' visibility='default' id='b351119f'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='avl_root' type-id='bf311473' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='avl_compar' type-id='585e1de9' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='avl_offset' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='avl_numnodes' type-id='ee1f298e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='avl_pad' type-id='b59d7dce' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='ulong_t' type-id='7359adad' id='ee1f298e'/>
	<typedef-decl name='uintptr_t' type-id='7359adad' id='e475ab95'/>
	<union-decl name='pthread_mutex_t' size-in-bits='320' naming-typedef-id='7a6844eb' visibility='default' id='70681f9b'>
	<data-member access='public'>
	<var-decl name='__data' type-id='4c734837' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='__size' type-id='36c46961' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='__align' type-id='bd54fe1a' visibility='default'/>
	</data-member>
	</union-decl>
	<typedef-decl name='pthread_mutex_t' type-id='70681f9b' id='7a6844eb'/>
	<typedef-decl name='int8_t' type-id='2171a512' id='ee31ee44'/>
	<typedef-decl name='uint8_t' type-id='c51d6389' id='b96825af'/>
	<typedef-decl name='uint32_t' type-id='62f1140c' id='8f92235e'/>
	+ <class-decl name='__pthread_internal_list' size-in-bits='128' is-struct='yes' visibility='default' id='0e01899c'>
	+ <data-member access='public' layout-offset-in-bits='0'>
	+ <var-decl name='__prev' type-id='4d98cd5a' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='64'>
	+ <var-decl name='__next' type-id='4d98cd5a' visibility='default'/>
	+ </data-member>
	+ </class-decl>
	+ <typedef-decl name='__pthread_list_t' type-id='0e01899c' id='518fb49c'/>
	<class-decl name='__pthread_mutex_s' size-in-bits='320' is-struct='yes' visibility='default' id='4c734837'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='__lock' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='__count' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='__owner' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='96'>
	<var-decl name='__nusers' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='__kind' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='160'>
	<var-decl name='__spins' type-id='a2185560' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='176'>
	<var-decl name='__elision' type-id='a2185560' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='__list' type-id='518fb49c' visibility='default'/>
	</data-member>
	</class-decl>
	- <class-decl name='__pthread_internal_list' size-in-bits='128' is-struct='yes' visibility='default' id='0e01899c'>
	- <data-member access='public' layout-offset-in-bits='0'>
	- <var-decl name='__prev' type-id='4d98cd5a' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='64'>
	- <var-decl name='__next' type-id='4d98cd5a' visibility='default'/>
	- </data-member>
	- </class-decl>
	- <typedef-decl name='__pthread_list_t' type-id='0e01899c' id='518fb49c'/>
	<typedef-decl name='__int8_t' type-id='28577a57' id='2171a512'/>
	<typedef-decl name='__uint8_t' type-id='002ac4a6' id='c51d6389'/>
	<typedef-decl name='__uint32_t' type-id='f0981eeb' id='62f1140c'/>
	<pointer-type-def type-id='0e01899c' size-in-bits='64' id='4d98cd5a'/>
	<pointer-type-def type-id='428b67b3' size-in-bits='64' id='bf311473'/>
	<pointer-type-def type-id='96ee24a5' size-in-bits='64' id='585e1de9'/>
	<pointer-type-def type-id='5d7f5fc8' size-in-bits='64' id='813a2225'/>
	<pointer-type-def type-id='73a65116' size-in-bits='64' id='2dc35b9d'/>
	<pointer-type-def type-id='7f84e390' size-in-bits='64' id='de82c773'/>
	<pointer-type-def type-id='bb7f0973' size-in-bits='64' id='a5c21a38'/>
	<pointer-type-def type-id='edd8457b' size-in-bits='64' id='5842d146'/>
	<pointer-type-def type-id='40f93560' size-in-bits='64' id='d502b39f'/>
	<pointer-type-def type-id='9d1aa0dc' size-in-bits='64' id='30a42b6d'/>
	<pointer-type-def type-id='eaa32e2f' size-in-bits='64' id='63e171df'/>
	<function-decl name='uu_avl_pool_create' mangled-name='uu_avl_pool_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_pool_create'>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='b59d7dce' name='objsize'/>
	<parameter type-id='b59d7dce' name='nodeoffset'/>
	<parameter type-id='d502b39f' name='compare_func'/>
	<parameter type-id='8f92235e' name='flags'/>
	<return type-id='de82c773'/>
	</function-decl>
	<function-decl name='uu_avl_pool_destroy' mangled-name='uu_avl_pool_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_pool_destroy'>
	<parameter type-id='de82c773' name='pp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_avl_node_init' mangled-name='uu_avl_node_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_node_init'>
	<parameter type-id='eaa32e2f' name='base'/>
	<parameter type-id='2dc35b9d' name='np'/>
	<parameter type-id='de82c773' name='pp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_avl_node_fini' mangled-name='uu_avl_node_fini' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_node_fini'>
	<parameter type-id='eaa32e2f' name='base'/>
	<parameter type-id='2dc35b9d' name='np'/>
	<parameter type-id='de82c773' name='pp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_avl_create' mangled-name='uu_avl_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_create'>
	<parameter type-id='de82c773' name='pp'/>
	<parameter type-id='eaa32e2f' name='parent'/>
	<parameter type-id='8f92235e' name='flags'/>
	<return type-id='a5c21a38'/>
	</function-decl>
	<function-decl name='uu_avl_destroy' mangled-name='uu_avl_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_destroy'>
	<parameter type-id='a5c21a38' name='ap'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_avl_numnodes' mangled-name='uu_avl_numnodes' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_numnodes'>
	<parameter type-id='a5c21a38' name='ap'/>
	<return type-id='b59d7dce'/>
	</function-decl>
	<function-decl name='uu_avl_first' mangled-name='uu_avl_first' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_first'>
	<parameter type-id='a5c21a38' name='ap'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_avl_last' mangled-name='uu_avl_last' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_last'>
	<parameter type-id='a5c21a38' name='ap'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_avl_next' mangled-name='uu_avl_next' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_next'>
	<parameter type-id='a5c21a38' name='ap'/>
	<parameter type-id='eaa32e2f' name='node'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_avl_prev' mangled-name='uu_avl_prev' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_prev'>
	<parameter type-id='a5c21a38' name='ap'/>
	<parameter type-id='eaa32e2f' name='node'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_avl_walk_start' mangled-name='uu_avl_walk_start' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_walk_start'>
	<parameter type-id='a5c21a38' name='ap'/>
	<parameter type-id='8f92235e' name='flags'/>
	<return type-id='5842d146'/>
	</function-decl>
	<function-decl name='uu_avl_walk_next' mangled-name='uu_avl_walk_next' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_walk_next'>
	<parameter type-id='5842d146' name='wp'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_avl_walk_end' mangled-name='uu_avl_walk_end' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_walk_end'>
	<parameter type-id='5842d146' name='wp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_avl_walk' mangled-name='uu_avl_walk' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_walk'>
	<parameter type-id='a5c21a38' name='ap'/>
	<parameter type-id='30a42b6d' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<parameter type-id='8f92235e' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='uu_avl_remove' mangled-name='uu_avl_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_remove'>
	<parameter type-id='a5c21a38' name='ap'/>
	<parameter type-id='eaa32e2f' name='elem'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_avl_teardown' mangled-name='uu_avl_teardown' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_teardown'>
	<parameter type-id='a5c21a38' name='ap'/>
	<parameter type-id='63e171df' name='cookie'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_avl_find' mangled-name='uu_avl_find' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_find'>
	<parameter type-id='a5c21a38' name='ap'/>
	<parameter type-id='eaa32e2f' name='elem'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<parameter type-id='813a2225' name='out'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_avl_insert' mangled-name='uu_avl_insert' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_insert'>
	<parameter type-id='a5c21a38' name='ap'/>
	<parameter type-id='eaa32e2f' name='elem'/>
	<parameter type-id='5d7f5fc8' name='idx'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_avl_nearest_next' mangled-name='uu_avl_nearest_next' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_nearest_next'>
	<parameter type-id='a5c21a38' name='ap'/>
	<parameter type-id='5d7f5fc8' name='idx'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_avl_nearest_prev' mangled-name='uu_avl_nearest_prev' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_nearest_prev'>
	<parameter type-id='a5c21a38' name='ap'/>
	<parameter type-id='5d7f5fc8' name='idx'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_avl_lockup' mangled-name='uu_avl_lockup' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_lockup'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_avl_release' mangled-name='uu_avl_release' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_avl_release'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-type size-in-bits='64' id='add6e811'>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='eaa32e2f'/>
	<return type-id='95e97e5e'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='uu_ident.c' language='LANG_C99'>
	<typedef-decl name='uint_t' type-id='f0981eeb' id='3502e3ff'/>
	<function-decl name='uu_check_name' mangled-name='uu_check_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_check_name'>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='3502e3ff' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='uu_list.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='e475ab95' size-in-bits='128' id='d0e9cdae'>
	<subrange length='2' type-id='7359adad' id='52efc4ef'/>
	</array-type-def>
	<typedef-decl name='uu_list_pool_t' type-id='55168cab' id='38a2549d'/>
	<typedef-decl name='uu_list_t' type-id='1d04bdf0' id='82e88484'/>
	<class-decl name='uu_list_node' size-in-bits='128' is-struct='yes' visibility='default' id='f8f3cec5'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='uln_opaque' type-id='d0e9cdae' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='uu_list_node_t' type-id='f8f3cec5' id='c4dc472f'/>
	<typedef-decl name='uu_list_walk_t' type-id='b80e3208' id='9fed32d2'/>
	<typedef-decl name='uu_list_index_t' type-id='e475ab95' id='f0dd35ff'/>
	<class-decl name='uu_list_node_impl' size-in-bits='128' is-struct='yes' visibility='default' id='700a795c'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='uln_next' type-id='5af1298a' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='uln_prev' type-id='5af1298a' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='uu_list_node_impl_t' type-id='700a795c' id='8e5864b0'/>
	<class-decl name='uu_list_walk' size-in-bits='320' is-struct='yes' visibility='default' id='b80e3208'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='ulw_next' type-id='4d848103' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='ulw_prev' type-id='4d848103' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='ulw_list' type-id='0c0b229b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='ulw_dir' type-id='ee31ee44' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='200'>
	<var-decl name='ulw_robust' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='ulw_next_result' type-id='a085247f' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='uu_list' size-in-bits='896' is-struct='yes' visibility='default' id='1d04bdf0'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='ul_next_enc' type-id='e475ab95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='ul_prev_enc' type-id='e475ab95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='ul_pool' type-id='0941e04e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='ul_parent_enc' type-id='e475ab95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='ul_offset' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='ul_numnodes' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='ul_debug' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='392'>
	<var-decl name='ul_sorted' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='400'>
	<var-decl name='ul_index' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='ul_null_node' type-id='8e5864b0' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='576'>
	<var-decl name='ul_null_walk' type-id='9fed32d2' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='uu_list_pool' size-in-bits='2112' is-struct='yes' visibility='default' id='55168cab'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='ulp_next' type-id='0941e04e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='ulp_prev' type-id='0941e04e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='ulp_name' type-id='59daf3ef' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='640'>
	<var-decl name='ulp_nodeoffset' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='704'>
	<var-decl name='ulp_objsize' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='768'>
	<var-decl name='ulp_cmp' type-id='d502b39f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='832'>
	<var-decl name='ulp_debug' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='840'>
	<var-decl name='ulp_last_index' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='896'>
	<var-decl name='ulp_lock' type-id='7a6844eb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1216'>
	<var-decl name='ulp_null_list' type-id='82e88484' visibility='default'/>
	</data-member>
	</class-decl>
	<pointer-type-def type-id='f0dd35ff' size-in-bits='64' id='ecbc0046'/>
	<pointer-type-def type-id='700a795c' size-in-bits='64' id='5af1298a'/>
	<pointer-type-def type-id='8e5864b0' size-in-bits='64' id='a085247f'/>
	<pointer-type-def type-id='c4dc472f' size-in-bits='64' id='dbe143f4'/>
	<pointer-type-def type-id='38a2549d' size-in-bits='64' id='0941e04e'/>
	<pointer-type-def type-id='82e88484' size-in-bits='64' id='0c0b229b'/>
	<pointer-type-def type-id='9fed32d2' size-in-bits='64' id='4d848103'/>
	<function-decl name='uu_list_pool_create' mangled-name='uu_list_pool_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_pool_create'>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='b59d7dce' name='objsize'/>
	<parameter type-id='b59d7dce' name='nodeoffset'/>
	<parameter type-id='d502b39f' name='compare_func'/>
	<parameter type-id='8f92235e' name='flags'/>
	<return type-id='0941e04e'/>
	</function-decl>
	<function-decl name='uu_list_pool_destroy' mangled-name='uu_list_pool_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_pool_destroy'>
	<parameter type-id='0941e04e' name='pp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_list_node_init' mangled-name='uu_list_node_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_node_init'>
	<parameter type-id='eaa32e2f' name='base'/>
	<parameter type-id='dbe143f4' name='np_arg'/>
	<parameter type-id='0941e04e' name='pp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_list_node_fini' mangled-name='uu_list_node_fini' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_node_fini'>
	<parameter type-id='eaa32e2f' name='base'/>
	<parameter type-id='dbe143f4' name='np_arg'/>
	<parameter type-id='0941e04e' name='pp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_list_create' mangled-name='uu_list_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_create'>
	<parameter type-id='0941e04e' name='pp'/>
	<parameter type-id='eaa32e2f' name='parent'/>
	<parameter type-id='8f92235e' name='flags'/>
	<return type-id='0c0b229b'/>
	</function-decl>
	<function-decl name='uu_list_destroy' mangled-name='uu_list_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_destroy'>
	<parameter type-id='0c0b229b' name='lp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_list_insert' mangled-name='uu_list_insert' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_insert'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='eaa32e2f' name='elem'/>
	<parameter type-id='f0dd35ff' name='idx'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_list_find' mangled-name='uu_list_find' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_find'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='eaa32e2f' name='elem'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<parameter type-id='ecbc0046' name='out'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_list_nearest_next' mangled-name='uu_list_nearest_next' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_nearest_next'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='f0dd35ff' name='idx'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_list_nearest_prev' mangled-name='uu_list_nearest_prev' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_nearest_prev'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='f0dd35ff' name='idx'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_list_walk_start' mangled-name='uu_list_walk_start' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_walk_start'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='8f92235e' name='flags'/>
	<return type-id='4d848103'/>
	</function-decl>
	<function-decl name='uu_list_walk_next' mangled-name='uu_list_walk_next' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_walk_next'>
	<parameter type-id='4d848103' name='wp'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_list_walk_end' mangled-name='uu_list_walk_end' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_walk_end'>
	<parameter type-id='4d848103' name='wp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_list_walk' mangled-name='uu_list_walk' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_walk'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='30a42b6d' name='func'/>
	<parameter type-id='eaa32e2f' name='private'/>
	<parameter type-id='8f92235e' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='uu_list_remove' mangled-name='uu_list_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_remove'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='eaa32e2f' name='elem'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_list_teardown' mangled-name='uu_list_teardown' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_teardown'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='63e171df' name='cookie'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_list_insert_before' mangled-name='uu_list_insert_before' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_insert_before'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='eaa32e2f' name='target'/>
	<parameter type-id='eaa32e2f' name='elem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='uu_list_insert_after' mangled-name='uu_list_insert_after' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_insert_after'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='eaa32e2f' name='target'/>
	<parameter type-id='eaa32e2f' name='elem'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='uu_list_numnodes' mangled-name='uu_list_numnodes' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_numnodes'>
	<parameter type-id='0c0b229b' name='lp'/>
	<return type-id='b59d7dce'/>
	</function-decl>
	<function-decl name='uu_list_first' mangled-name='uu_list_first' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_first'>
	<parameter type-id='0c0b229b' name='lp'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_list_last' mangled-name='uu_list_last' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_last'>
	<parameter type-id='0c0b229b' name='lp'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_list_next' mangled-name='uu_list_next' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_next'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='eaa32e2f' name='elem'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_list_prev' mangled-name='uu_list_prev' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_prev'>
	<parameter type-id='0c0b229b' name='lp'/>
	<parameter type-id='eaa32e2f' name='elem'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='uu_list_lockup' mangled-name='uu_list_lockup' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_lockup'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_list_release' mangled-name='uu_list_release' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_list_release'>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='uu_misc.c' language='LANG_C99'>
	<function-decl name='uu_set_error' mangled-name='uu_set_error' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_set_error'>
	<parameter type-id='3502e3ff' name='code'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_error' mangled-name='uu_error' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_error'>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='uu_strerror' mangled-name='uu_strerror' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_strerror'>
	<parameter type-id='8f92235e' name='code'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='uu_panic' mangled-name='uu_panic' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_panic'>
	<parameter type-id='80f4b756' name='format'/>
	<parameter is-variadic='yes'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='uu_pname.c' language='LANG_C99'>
	<class-decl name='__va_list_tag' size-in-bits='192' is-struct='yes' visibility='default' id='d5027220'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='gp_offset' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='fp_offset' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='overflow_arg_area' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='reg_save_area' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<pointer-type-def type-id='d5027220' size-in-bits='64' id='b7f2d5e6'/>
	<pointer-type-def type-id='95e97e5e' size-in-bits='64' id='7292109c'/>
	<var-decl name='uu_exit_ok_value' type-id='95e97e5e' mangled-name='uu_exit_ok_value' visibility='default' elf-symbol-id='uu_exit_ok_value'/>
	<var-decl name='uu_exit_fatal_value' type-id='95e97e5e' mangled-name='uu_exit_fatal_value' visibility='default' elf-symbol-id='uu_exit_fatal_value'/>
	<var-decl name='uu_exit_usage_value' type-id='95e97e5e' mangled-name='uu_exit_usage_value' visibility='default' elf-symbol-id='uu_exit_usage_value'/>
	<function-decl name='uu_exit_ok' mangled-name='uu_exit_ok' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_exit_ok'>
	<return type-id='7292109c'/>
	</function-decl>
	<function-decl name='uu_exit_fatal' mangled-name='uu_exit_fatal' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_exit_fatal'>
	<return type-id='7292109c'/>
	</function-decl>
	<function-decl name='uu_exit_usage' mangled-name='uu_exit_usage' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_exit_usage'>
	<return type-id='7292109c'/>
	</function-decl>
	<function-decl name='uu_alt_exit' mangled-name='uu_alt_exit' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_alt_exit'>
	<parameter type-id='95e97e5e' name='profile'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_vwarn' mangled-name='uu_vwarn' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_vwarn'>
	<parameter type-id='80f4b756' name='format'/>
	<parameter type-id='b7f2d5e6' name='alist'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_warn' mangled-name='uu_warn' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_warn'>
	<parameter type-id='80f4b756' name='format'/>
	<parameter is-variadic='yes'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_vdie' mangled-name='uu_vdie' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_vdie'>
	<parameter type-id='80f4b756' name='format'/>
	<parameter type-id='b7f2d5e6' name='alist'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_die' mangled-name='uu_die' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_die'>
	<parameter type-id='80f4b756' name='format'/>
	<parameter is-variadic='yes'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_vxdie' mangled-name='uu_vxdie' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_vxdie'>
	<parameter type-id='95e97e5e' name='status'/>
	<parameter type-id='80f4b756' name='format'/>
	<parameter type-id='b7f2d5e6' name='alist'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_xdie' mangled-name='uu_xdie' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_xdie'>
	<parameter type-id='95e97e5e' name='status'/>
	<parameter type-id='80f4b756' name='format'/>
	<parameter is-variadic='yes'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='uu_setpname' mangled-name='uu_setpname' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_setpname'>
	<parameter type-id='26a90f95' name='arg0'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='uu_getpname' mangled-name='uu_getpname' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_getpname'>
	<return type-id='80f4b756'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='uu_string.c' language='LANG_C99'>
	<type-decl name='unnamed-enum-underlying-type-32' is-anonymous='yes' size-in-bits='32' alignment-in-bits='32' id='9cac1fee'/>
	<enum-decl name='boolean_t' naming-typedef-id='c19b74c3' id='f58c8277'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='B_FALSE' value='0'/>
	<enumerator name='B_TRUE' value='1'/>
	</enum-decl>
	<typedef-decl name='boolean_t' type-id='f58c8277' id='c19b74c3'/>
	<function-decl name='uu_streq' mangled-name='uu_streq' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_streq'>
	<parameter type-id='80f4b756' name='a'/>
	<parameter type-id='80f4b756' name='b'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='uu_strcaseeq' mangled-name='uu_strcaseeq' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_strcaseeq'>
	<parameter type-id='80f4b756' name='a'/>
	<parameter type-id='80f4b756' name='b'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='uu_strbw' mangled-name='uu_strbw' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='uu_strbw'>
	<parameter type-id='80f4b756' name='a'/>
	<parameter type-id='80f4b756' name='b'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	</abi-instr>
	</abi-corpus>
	diff --git a/sys/contrib/openzfs/lib/libzfs/libzfs.abi b/sys/contrib/openzfs/lib/libzfs/libzfs.abi
	index ab6d27e913b8..9bd1fd0db0e9 100644
	--- a/sys/contrib/openzfs/lib/libzfs/libzfs.abi
	+++ b/sys/contrib/openzfs/lib/libzfs/libzfs.abi
	@@ -1,5637 +1,5837 @@
	<abi-corpus version='2.0' architecture='elf-amd-x86_64' soname='libzfs.so.4'>
	<elf-needed>
	<dependency name='libzfs_core.so.3'/>
	<dependency name='libnvpair.so.3'/>
	<dependency name='libuuid.so.1'/>
	<dependency name='librt.so.1'/>
	<dependency name='libblkid.so.1'/>
	<dependency name='libudev.so.1'/>
	<dependency name='libuutil.so.3'/>
	<dependency name='libm.so.6'/>
	<dependency name='libcrypto.so.1.1'/>
	<dependency name='libz.so.1'/>
	+ <dependency name='libdl.so.2'/>
	<dependency name='libpthread.so.0'/>
	<dependency name='libc.so.6'/>
	<dependency name='ld-linux-x86-64.so.2'/>
	</elf-needed>
	<elf-function-symbols>
	+ <elf-symbol name='_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='_sol_getmntent' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_char' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_char_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_int' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_int_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_long' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_long_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_ptr_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_short' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_short_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_clear_long_excl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_set_long_excl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_char' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_char_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_int' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_int_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_long' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_long_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_ptr_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_short' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_short_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_add' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_destroy_nodes' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_find' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_first' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_insert' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_insert_here' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_is_empty' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_last' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_nearest' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_numnodes' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_swap' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_update' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_update_gt' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_update_lt' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='avl_walk' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='bookmark_namecheck' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='cityhash4' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='color_end' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='color_start' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='dataset_namecheck' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='dataset_nestcheck' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='efi_alloc_and_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='efi_alloc_and_read' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='efi_err_check' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='efi_free' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='efi_rescan' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='efi_type' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='efi_use_whole_disk' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='efi_write' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='entity_namecheck' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_2_byteswap' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_2_incremental_byteswap' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_2_incremental_native' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_2_native' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_byteswap' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_impl_set' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_incremental_byteswap' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_incremental_native' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_native' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_native_varsize' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='get_dataset_depth' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='get_system_hostid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getexecname' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getextmntent' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getmntany' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getprop_uint64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getzoneid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='is_mounted' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='is_mpath_whole_disk' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libspl_assertf' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_add_handle' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_envvar_is_set' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_errno' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_error_action' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_error_description' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_error_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_free_str_array' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_mnttab_add' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_mnttab_cache' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_mnttab_find' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_mnttab_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_mnttab_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_mnttab_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_print_on_error' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_run_process' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_run_process_get_stdout' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_run_process_get_stdout_nopath' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_head' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_after' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_before' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_head' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_is_empty' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_link_active' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_link_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_link_replace' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_move_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_next' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_prev' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_remove_head' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_remove_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_consumer' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_enter' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_exit' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_producer' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='mkdirp' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='mountpoint_namecheck' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='permset_namecheck' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='pool_namecheck' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='print_timestamp' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='printf_color' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='sa_commit_shares' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='sa_disable_share' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='sa_enable_share' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='sa_errorstr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='sa_is_shared' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='sa_validate_shareopts' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='snapshot_namecheck' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='spl_pagesize' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='strlcat' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='strlcpy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='tpool_abandon' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='tpool_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='tpool_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='tpool_dispatch' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='tpool_member' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='tpool_resume' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='tpool_suspend' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='tpool_suspended' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='tpool_wait' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='update_vdev_config_dev_strs' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_expand_proplist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_name_to_prop' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_align_right' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_column_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_default_numeric' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_default_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_get_table' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_get_type' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_index_to_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_random_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_readonly' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_string_to_index' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_to_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_user' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='vdev_prop_values' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfeature_depends_on' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfeature_is_supported' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfeature_is_valid_guid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfeature_lookup_guid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfeature_lookup_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_adjust_mount_options' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_allocatable_devs' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_append_partition' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_basename' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_bookmark_exists' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_clone' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_close' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_commit_all_shares' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_commit_nfs_shares' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_commit_shares' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_commit_smb_shares' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_component_namecheck' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_create_ancestors' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_crypto_attempt_load_keys' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_crypto_clone_check' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_crypto_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_crypto_get_encryption_root' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_crypto_load_key' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_crypto_rewrap' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_crypto_unload_key' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_dataset_exists' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_dataset_name_hidden' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_deleg_canonicalize_perm' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_deleg_verify_nvlist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_deleg_whokey' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_destroy_snaps' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_destroy_snaps_nvl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_destroy_snaps_nvl_os' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_dev_flush' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_dev_is_dm' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_dev_is_whole_disk' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_device_get_devid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_device_get_physical' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_dirnamelen' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_expand_proplist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_foreach_mountpoint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_all_props' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_clones_nvl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_enclosure_sysfs_path' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_fsacl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_handle' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_holds' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_pool_handle' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_pool_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_recvd_props' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_type' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_underlying_path' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_underlying_type' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_get_user_props' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_handle_dup' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_hold' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_hold_nvl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_ioctl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_is_mounted' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_is_shared' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_is_shared_nfs' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_is_shared_smb' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_isnumber' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_iter_bookmarks' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_iter_children' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_iter_dependents' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_iter_filesystems' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_iter_mounted' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_iter_root' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_iter_snapshots' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_iter_snapshots_sorted' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_iter_snapspec' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_mod_supported' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_mount' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_mount_at' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_mount_delegation_check' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_name_to_prop' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_name_valid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_nicebytes' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_nicenum' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_nicenum_format' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_niceraw' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_nicestrtonum' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_nicetime' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_open' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_parent_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_parse_mount_options' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_path_to_zhandle' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_promote' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_align_right' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_column_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_default_numeric' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_default_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_delegatable' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_encryption_key_param' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_get' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_get_int' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_get_numeric' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_get_recvd' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_get_table' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_get_type' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_get_userquota' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_get_userquota_int' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_get_written' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_get_written_int' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_index_to_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_inherit' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_inheritable' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_is_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_random_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_readonly' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_set' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_set_list' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_setonce' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_string_to_index' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_to_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_user' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_userquota' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_valid_for_type' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_valid_keylocation' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_values' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_visible' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prop_written' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_prune_proplist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_receive' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_refresh_properties' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_release' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_rename' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_resolve_shortname' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_rollback' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_save_arguments' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_send' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_send_one' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_send_progress' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_send_resume' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_send_resume_token_to_nvlist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_send_saved' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_set_fsacl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_share' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_share_nfs' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_share_smb' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_shareall' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_show_diffs' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_smb_acl_add' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_smb_acl_purge' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_smb_acl_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_smb_acl_rename' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_snapshot' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_snapshot_nvl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_spa_version' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_spa_version_map' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_special_devs' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_standard_error' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_strcmp_pathname' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_strip_partition' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_strip_path' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_type_to_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_unmount' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_unmountall' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_unshare' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_unshare_nfs' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_unshare_smb' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_unshareall' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_unshareall_bypath' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_unshareall_bytype' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_unshareall_nfs' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_unshareall_smb' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_userspace' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_valid_proplist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_version_kernel' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_version_print' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_version_userland' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_wait_status' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_zpl_version_map' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_add' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_checkpoint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_clear' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_clear_label' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_close' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_default_search_paths' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_disable_datasets' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_disable_datasets_os' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_disable_volume_os' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_discard_checkpoint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_dump_ddt' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_enable_datasets' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_events_clear' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_events_next' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_events_seek' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_expand_proplist' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_explain_recover' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_export' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_export_force' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_feature_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_find_config' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_find_vdev' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_find_vdev_by_physpath' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_free_handles' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='zpool_get_all_vdev_props' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_bootenv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_config' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_errlog' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_features' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_handle' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_history' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_load_policy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_physpath' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_prop' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_prop_int' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_state' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_state_str' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_get_status' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='zpool_get_vdev_prop' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='zpool_get_vdev_prop_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_history_unpack' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_import' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_import_props' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_import_status' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_in_use' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_initialize' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_initialize_wait' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_is_draid_spare' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_iter' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_label_disk' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_label_disk_wait' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_load_compat' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_log_history' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_name_to_prop' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_obj_to_path' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_obj_to_path_ds' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_open' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_open_canfail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_pool_state_to_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_print_unsup_feat' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_align_right' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_column_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_default_numeric' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_default_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_feature' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_get_feature' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_get_table' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_get_type' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_index_to_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_random_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_readonly' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_setonce' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_string_to_index' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_to_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_unsupported' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_prop_values' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='zpool_prop_vdev' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_props_refresh' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_read_label' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_refresh_stats' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_reguid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_reopen_one' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_scan' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_search_import' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_set_bootenv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_set_prop' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='zpool_set_vdev_prop' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_skip_pool' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_state_to_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_sync_one' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_trim' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_upgrade' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_attach' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_clear' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_degrade' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_detach' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_fault' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_indirect_size' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_name' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_offline' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_online' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_path_to_guid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_remove_cancel' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_vdev_split' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_wait' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zpool_wait_status' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_free_list' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_get_list' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_index_to_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_iter' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_iter_common' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_name_to_prop' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_print_one_property' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_random_value' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_register_hidden' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_register_impl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_register_index' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_register_number' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_register_string' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_string_to_index' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='zprop_valid_char' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_valid_for_type' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_values' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zprop_width' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zvol_volsize_to_reservation' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	</elf-function-symbols>
	<elf-variable-symbols>
	<elf-symbol name='efi_debug' size='4' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_abd_ops' size='24' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_avx2_ops' size='64' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_avx512bw_ops' size='64' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_avx512f_ops' size='64' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_sse2_ops' size='64' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_ssse3_ops' size='64' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_superscalar4_ops' size='64' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='fletcher_4_superscalar_ops' size='64' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libspl_assert_ok' size='4' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_config_ops' size='16' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='spa_feature_table' size='1904' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfeature_checks_disable' size='4' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_deleg_perm_tab' size='512' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_history_event_names' size='328' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_max_dataset_nesting' size='4' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='zfs_userquota_prop_prefixes' size='96' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	</elf-variable-symbols>
	<abi-instr address-size='64' path='../../module/avl/avl.c' language='LANG_C99'>
	<typedef-decl name='avl_index_t' type-id='e475ab95' id='fba6cb51'/>
	<pointer-type-def type-id='fba6cb51' size-in-bits='64' id='32adbf30'/>
	<pointer-type-def type-id='eaa32e2f' size-in-bits='64' id='63e171df'/>
	<function-decl name='avl_walk' mangled-name='avl_walk' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_walk'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='oldnode'/>
	<parameter type-id='95e97e5e' name='left'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='avl_first' mangled-name='avl_first' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_first'>
	<parameter type-id='a3681dea' name='tree'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='avl_last' mangled-name='avl_last' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_last'>
	<parameter type-id='a3681dea' name='tree'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='avl_nearest' mangled-name='avl_nearest' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_nearest'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='fba6cb51' name='where'/>
	<parameter type-id='95e97e5e' name='direction'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='avl_find' mangled-name='avl_find' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_find'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='value'/>
	<parameter type-id='32adbf30' name='where'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='avl_insert' mangled-name='avl_insert' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_insert'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='new_data'/>
	<parameter type-id='fba6cb51' name='where'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_insert_here' mangled-name='avl_insert_here' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_insert_here'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='new_data'/>
	<parameter type-id='eaa32e2f' name='here'/>
	<parameter type-id='95e97e5e' name='direction'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_add' mangled-name='avl_add' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_add'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='new_node'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_remove' mangled-name='avl_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_remove'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_update_lt' mangled-name='avl_update_lt' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_update_lt'>
	<parameter type-id='a3681dea' name='t'/>
	<parameter type-id='eaa32e2f' name='obj'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='avl_update_gt' mangled-name='avl_update_gt' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_update_gt'>
	<parameter type-id='a3681dea' name='t'/>
	<parameter type-id='eaa32e2f' name='obj'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='avl_update' mangled-name='avl_update' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_update'>
	<parameter type-id='a3681dea' name='t'/>
	<parameter type-id='eaa32e2f' name='obj'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='avl_swap' mangled-name='avl_swap' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_swap'>
	<parameter type-id='a3681dea' name='tree1'/>
	<parameter type-id='a3681dea' name='tree2'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_create' mangled-name='avl_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_create'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='585e1de9' name='compar'/>
	<parameter type-id='b59d7dce' name='size'/>
	<parameter type-id='b59d7dce' name='offset'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_destroy' mangled-name='avl_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_destroy'>
	<parameter type-id='a3681dea' name='tree'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='avl_numnodes' mangled-name='avl_numnodes' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_numnodes'>
	<parameter type-id='a3681dea' name='tree'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='avl_is_empty' mangled-name='avl_is_empty' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_is_empty'>
	<parameter type-id='a3681dea' name='tree'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='avl_destroy_nodes' mangled-name='avl_destroy_nodes' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='avl_destroy_nodes'>
	<parameter type-id='a3681dea' name='tree'/>
	<parameter type-id='63e171df' name='cookie'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-type size-in-bits='64' id='96ee24a5'>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='eaa32e2f'/>
	<return type-id='95e97e5e'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='rdwr_efi.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='288' id='16e6f2c6'>
	<subrange length='36' type-id='7359adad' id='ae666bde'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a65ae39c' size-in-bits='960' id='fa198beb'>
	<subrange length='1' type-id='7359adad' id='52f813b4'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='b96825af' size-in-bits='48' id='0f562bd0'>
	<subrange length='6' type-id='7359adad' id='52fa524b'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='3502e3ff' size-in-bits='384' id='dba89ba3'>
	<subrange length='12' type-id='7359adad' id='84827bdc'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='3502e3ff' size-in-bits='256' id='01d84ed4'>
	<subrange length='8' type-id='7359adad' id='56e0c0b1'/>
	</array-type-def>
	<class-decl name='dk_part' size-in-bits='960' is-struct='yes' visibility='default' id='a65ae39c'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='p_start' type-id='804dc465' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='p_size' type-id='804dc465' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='p_guid' type-id='214f32ea' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='p_tag' type-id='d908a348' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='272'>
	<var-decl name='p_flag' type-id='d908a348' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='288'>
	<var-decl name='p_name' type-id='16e6f2c6' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='576'>
	<var-decl name='p_uguid' type-id='214f32ea' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='704'>
	<var-decl name='p_resv' type-id='01d84ed4' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='dk_gpt' size-in-bits='1920' is-struct='yes' visibility='default' id='dd4a2e5a'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='efi_version' type-id='3502e3ff' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='efi_nparts' type-id='3502e3ff' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='efi_part_size' type-id='3502e3ff' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='96'>
	<var-decl name='efi_lbasize' type-id='3502e3ff' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='efi_last_lba' type-id='804dc465' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='efi_first_u_lba' type-id='804dc465' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='efi_last_u_lba' type-id='804dc465' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='efi_disk_uguid' type-id='214f32ea' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='efi_flags' type-id='3502e3ff' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='480'>
	<var-decl name='efi_reserved1' type-id='3502e3ff' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='efi_altern_lba' type-id='804dc465' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='576'>
	<var-decl name='efi_reserved' type-id='dba89ba3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='960'>
	<var-decl name='efi_parts' type-id='fa198beb' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='uuid' size-in-bits='128' is-struct='yes' visibility='default' id='214f32ea'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='time_low' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='time_mid' type-id='149c6638' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='48'>
	<var-decl name='time_hi_and_version' type-id='149c6638' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='clock_seq_hi_and_reserved' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='72'>
	<var-decl name='clock_seq_low' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='80'>
	<var-decl name='node_addr' type-id='0f562bd0' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='ushort_t' type-id='8efea9e5' id='d908a348'/>
	<pointer-type-def type-id='dd4a2e5a' size-in-bits='64' id='0d8119a8'/>
	<pointer-type-def type-id='0d8119a8' size-in-bits='64' id='c43b27a6'/>
	<var-decl name='efi_debug' type-id='95e97e5e' mangled-name='efi_debug' visibility='default' elf-symbol-id='efi_debug'/>
	<function-decl name='efi_alloc_and_init' mangled-name='efi_alloc_and_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='efi_alloc_and_init'>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='8f92235e' name='nparts'/>
	<parameter type-id='c43b27a6' name='vtoc'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='efi_alloc_and_read' mangled-name='efi_alloc_and_read' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='efi_alloc_and_read'>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='c43b27a6' name='vtoc'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='efi_rescan' mangled-name='efi_rescan' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='efi_rescan'>
	<parameter type-id='95e97e5e' name='fd'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='efi_use_whole_disk' mangled-name='efi_use_whole_disk' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='efi_use_whole_disk'>
	<parameter type-id='95e97e5e' name='fd'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='efi_write' mangled-name='efi_write' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='efi_write'>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='0d8119a8' name='vtoc'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='efi_free' mangled-name='efi_free' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='efi_free'>
	<parameter type-id='0d8119a8' name='ptr'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='efi_type' mangled-name='efi_type' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='efi_type'>
	<parameter type-id='95e97e5e' name='fd'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='efi_err_check' mangled-name='efi_err_check' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='efi_err_check'>
	<parameter type-id='0d8119a8' name='vtoc'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<type-decl name='unsigned long int' size-in-bits='64' id='7359adad'/>
	<type-decl name='unsigned short int' size-in-bits='16' id='8efea9e5'/>
	</abi-instr>
	<abi-instr address-size='64' path='libshare.c' language='LANG_C99'>
	<function-decl name='sa_enable_share' mangled-name='sa_enable_share' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='sa_enable_share'>
	<parameter type-id='80f4b756' name='zfsname'/>
	<parameter type-id='80f4b756' name='mountpoint'/>
	<parameter type-id='80f4b756' name='shareopts'/>
	<parameter type-id='26a90f95' name='protocol'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='sa_disable_share' mangled-name='sa_disable_share' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='sa_disable_share'>
	<parameter type-id='80f4b756' name='mountpoint'/>
	<parameter type-id='26a90f95' name='protocol'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='sa_is_shared' mangled-name='sa_is_shared' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='sa_is_shared'>
	<parameter type-id='80f4b756' name='mountpoint'/>
	<parameter type-id='26a90f95' name='protocol'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='sa_commit_shares' mangled-name='sa_commit_shares' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='sa_commit_shares'>
	<parameter type-id='80f4b756' name='protocol'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='sa_errorstr' mangled-name='sa_errorstr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='sa_errorstr'>
	<parameter type-id='95e97e5e' name='err'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='sa_validate_shareopts' mangled-name='sa_validate_shareopts' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='sa_validate_shareopts'>
	<parameter type-id='26a90f95' name='options'/>
	<parameter type-id='26a90f95' name='proto'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/smb.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='2040' id='11641789'>
	<subrange length='255' type-id='7359adad' id='36e7f891'/>
	</array-type-def>
	<class-decl name='smb_share_s' size-in-bits='36992' is-struct='yes' visibility='default' id='a75bc907'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='name' type-id='11641789' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2040'>
	<var-decl name='path' type-id='d16c6df4' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='34808'>
	<var-decl name='comment' type-id='11641789' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='36864'>
	<var-decl name='guest_ok' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='36928'>
	<var-decl name='next' type-id='05ed1c5f' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='smb_share_t' type-id='a75bc907' id='2d05afd9'/>
	<pointer-type-def type-id='a75bc907' size-in-bits='64' id='05ed1c5f'/>
	<pointer-type-def type-id='2d05afd9' size-in-bits='64' id='a3e5c654'/>
	<var-decl name='smb_shares' type-id='a3e5c654' visibility='default'/>
	</abi-instr>
	<abi-instr address-size='64' path='assert.c' language='LANG_C99'>
	<var-decl name='libspl_assert_ok' type-id='95e97e5e' mangled-name='libspl_assert_ok' visibility='default' elf-symbol-id='libspl_assert_ok'/>
	<function-decl name='libspl_assertf' mangled-name='libspl_assertf' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libspl_assertf'>
	<parameter type-id='80f4b756' name='file'/>
	<parameter type-id='80f4b756' name='func'/>
	<parameter type-id='95e97e5e' name='line'/>
	<parameter type-id='80f4b756' name='format'/>
	<parameter is-variadic='yes'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='atomic.c' language='LANG_C99'>
	<type-decl name='signed char' size-in-bits='8' id='28577a57'/>
	<type-decl name='unsigned short int' size-in-bits='16' id='8efea9e5'/>
	<typedef-decl name='int8_t' type-id='2171a512' id='ee31ee44'/>
	<typedef-decl name='int16_t' type-id='03896e23' id='23bd8cb5'/>
	<typedef-decl name='uint16_t' type-id='253c2d2a' id='149c6638'/>
	<typedef-decl name='__int8_t' type-id='28577a57' id='2171a512'/>
	<typedef-decl name='__int16_t' type-id='a2185560' id='03896e23'/>
	<typedef-decl name='__uint16_t' type-id='8efea9e5' id='253c2d2a'/>
	<qualified-type-def type-id='149c6638' volatile='yes' id='5120c5f7'/>
	<pointer-type-def type-id='5120c5f7' size-in-bits='64' id='93977ae7'/>
	<qualified-type-def type-id='8f92235e' volatile='yes' id='430e0681'/>
	<pointer-type-def type-id='430e0681' size-in-bits='64' id='3a147f31'/>
	<qualified-type-def type-id='b96825af' volatile='yes' id='84ff7d66'/>
	<pointer-type-def type-id='84ff7d66' size-in-bits='64' id='aa323ea4'/>
	<qualified-type-def type-id='ee1f298e' volatile='yes' id='6f7e09cb'/>
	<pointer-type-def type-id='6f7e09cb' size-in-bits='64' id='64698d33'/>
	<qualified-type-def type-id='48b5725f' volatile='yes' id='b0b3cbf9'/>
	<pointer-type-def type-id='b0b3cbf9' size-in-bits='64' id='fe09dd29'/>
	<function-decl name='atomic_inc_8' mangled-name='atomic_inc_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_16' mangled-name='atomic_inc_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_16'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_32' mangled-name='atomic_inc_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_32'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_ulong' mangled-name='atomic_inc_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_8' mangled-name='atomic_dec_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_16' mangled-name='atomic_dec_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_16'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_32' mangled-name='atomic_dec_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_32'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_ulong' mangled-name='atomic_dec_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_8' mangled-name='atomic_add_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_16' mangled-name='atomic_add_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_32' mangled-name='atomic_add_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_long' mangled-name='atomic_add_long' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_long'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_ptr' mangled-name='atomic_add_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_8' mangled-name='atomic_sub_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_16' mangled-name='atomic_sub_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_32' mangled-name='atomic_sub_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_long' mangled-name='atomic_sub_long' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_long'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_ptr' mangled-name='atomic_sub_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_8' mangled-name='atomic_or_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_16' mangled-name='atomic_or_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_32' mangled-name='atomic_or_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_ulong' mangled-name='atomic_or_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_8' mangled-name='atomic_and_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_16' mangled-name='atomic_and_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_32' mangled-name='atomic_and_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_ulong' mangled-name='atomic_and_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_8_nv' mangled-name='atomic_inc_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_inc_16_nv' mangled-name='atomic_inc_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_inc_32_nv' mangled-name='atomic_inc_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_inc_ulong_nv' mangled-name='atomic_inc_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_dec_8_nv' mangled-name='atomic_dec_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_dec_16_nv' mangled-name='atomic_dec_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_dec_32_nv' mangled-name='atomic_dec_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_dec_ulong_nv' mangled-name='atomic_dec_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_add_8_nv' mangled-name='atomic_add_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_add_16_nv' mangled-name='atomic_add_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_add_32_nv' mangled-name='atomic_add_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_add_long_nv' mangled-name='atomic_add_long_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_long_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_add_ptr_nv' mangled-name='atomic_add_ptr_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_ptr_nv'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_sub_8_nv' mangled-name='atomic_sub_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_sub_16_nv' mangled-name='atomic_sub_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_sub_32_nv' mangled-name='atomic_sub_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_sub_long_nv' mangled-name='atomic_sub_long_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_long_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_sub_ptr_nv' mangled-name='atomic_sub_ptr_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_ptr_nv'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_or_8_nv' mangled-name='atomic_or_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_or_16_nv' mangled-name='atomic_or_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_or_32_nv' mangled-name='atomic_or_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_or_ulong_nv' mangled-name='atomic_or_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_and_8_nv' mangled-name='atomic_and_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_and_16_nv' mangled-name='atomic_and_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_and_32_nv' mangled-name='atomic_and_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_and_ulong_nv' mangled-name='atomic_and_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_cas_8' mangled-name='atomic_cas_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='exp'/>
	<parameter type-id='b96825af' name='des'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_cas_16' mangled-name='atomic_cas_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='exp'/>
	<parameter type-id='149c6638' name='des'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_cas_32' mangled-name='atomic_cas_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='exp'/>
	<parameter type-id='8f92235e' name='des'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_cas_ulong' mangled-name='atomic_cas_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='exp'/>
	<parameter type-id='ee1f298e' name='des'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_cas_ptr' mangled-name='atomic_cas_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='eaa32e2f' name='exp'/>
	<parameter type-id='eaa32e2f' name='des'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_swap_8' mangled-name='atomic_swap_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_swap_16' mangled-name='atomic_swap_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_swap_32' mangled-name='atomic_swap_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_swap_ulong' mangled-name='atomic_swap_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_swap_ptr' mangled-name='atomic_swap_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='eaa32e2f' name='bits'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_set_long_excl' mangled-name='atomic_set_long_excl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_set_long_excl'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='3502e3ff' name='value'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='atomic_clear_long_excl' mangled-name='atomic_clear_long_excl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_clear_long_excl'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='3502e3ff' name='value'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='membar_enter' mangled-name='membar_enter' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='membar_enter'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='membar_producer' mangled-name='membar_producer' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='membar_producer'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='membar_consumer' mangled-name='membar_consumer' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='membar_consumer'>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='getexecname.c' language='LANG_C99'>
	<function-decl name='getexecname' mangled-name='getexecname' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getexecname'>
	<return type-id='80f4b756'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='list.c' language='LANG_C99'>
	<typedef-decl name='list_node_t' type-id='b0b5e45e' id='b21843b2'/>
	<typedef-decl name='list_t' type-id='e824dae9' id='0899125f'/>
	<class-decl name='list_node' size-in-bits='128' is-struct='yes' visibility='default' id='b0b5e45e'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='next' type-id='b03eadb4' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='prev' type-id='b03eadb4' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='list' size-in-bits='256' is-struct='yes' visibility='default' id='e824dae9'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='list_size' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='list_offset' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='list_head' type-id='b0b5e45e' visibility='default'/>
	</data-member>
	</class-decl>
	<pointer-type-def type-id='b0b5e45e' size-in-bits='64' id='b03eadb4'/>
	<pointer-type-def type-id='b21843b2' size-in-bits='64' id='ccc38265'/>
	<pointer-type-def type-id='0899125f' size-in-bits='64' id='352ec160'/>
	<function-decl name='list_create' mangled-name='list_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_create'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='b59d7dce' name='size'/>
	<parameter type-id='b59d7dce' name='offset'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_destroy' mangled-name='list_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_destroy'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_after' mangled-name='list_insert_after' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_after'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<parameter type-id='eaa32e2f' name='nobject'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_before' mangled-name='list_insert_before' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_before'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<parameter type-id='eaa32e2f' name='nobject'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_head' mangled-name='list_insert_head' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_head'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_tail' mangled-name='list_insert_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_tail'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_remove' mangled-name='list_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_remove'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_remove_head' mangled-name='list_remove_head' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_remove_head'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_remove_tail' mangled-name='list_remove_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_remove_tail'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_head' mangled-name='list_head' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_head'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_tail' mangled-name='list_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_tail'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_next' mangled-name='list_next' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_next'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_prev' mangled-name='list_prev' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_prev'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_move_tail' mangled-name='list_move_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_move_tail'>
	<parameter type-id='352ec160' name='dst'/>
	<parameter type-id='352ec160' name='src'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_link_replace' mangled-name='list_link_replace' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_link_replace'>
	<parameter type-id='ccc38265' name='lold'/>
	<parameter type-id='ccc38265' name='lnew'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_link_init' mangled-name='list_link_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_link_init'>
	<parameter type-id='ccc38265' name='ln'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_link_active' mangled-name='list_link_active' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_link_active'>
	<parameter type-id='ccc38265' name='ln'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='list_is_empty' mangled-name='list_is_empty' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_is_empty'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='mkdirp.c' language='LANG_C99'>
	<typedef-decl name='mode_t' type-id='e1c52942' id='d50d396c'/>
	<function-decl name='mkdirp' mangled-name='mkdirp' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='mkdirp'>
	<parameter type-id='80f4b756' name='d'/>
	<parameter type-id='d50d396c' name='mode'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/gethostid.c' language='LANG_C99'>
	<function-decl name='get_system_hostid' mangled-name='get_system_hostid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='get_system_hostid'>
	<return type-id='7359adad'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/getmntany.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='03085adc' size-in-bits='192' id='083f8d58'>
	<subrange length='3' type-id='7359adad' id='56f209d2'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='8' id='89feb1ec'>
	<subrange length='1' type-id='7359adad' id='52f813b4'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='160' id='664ac0b7'>
	<subrange length='20' type-id='7359adad' id='fdca39cf'/>
	</array-type-def>
	- <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
	- <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
	- <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
	<class-decl name='extmnttab' size-in-bits='320' is-struct='yes' visibility='default' id='0c544dc0'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='mnt_special' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='mnt_mountp' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='mnt_fstype' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='mnt_mntopts' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='mnt_major' type-id='3502e3ff' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='288'>
	<var-decl name='mnt_minor' type-id='3502e3ff' visibility='default'/>
	</data-member>
	</class-decl>
	- <class-decl name='stat64' size-in-bits='1152' is-struct='yes' visibility='default' id='0bbec9cd'>
	+ <typedef-decl name='_IO_lock_t' type-id='48b5725f' id='bb4788fa'/>
	+ <class-decl name='_IO_marker' size-in-bits='192' is-struct='yes' visibility='default' id='010ae0b9'>
	<data-member access='public' layout-offset-in-bits='0'>
	- <var-decl name='st_dev' type-id='35ed8932' visibility='default'/>
	+ <var-decl name='_next' type-id='e4c6fa61' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	- <var-decl name='st_ino' type-id='71288a47' visibility='default'/>
	+ <var-decl name='_sbuf' type-id='dca988a5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	- <var-decl name='st_nlink' type-id='80f0b9df' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='192'>
	- <var-decl name='st_mode' type-id='e1c52942' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='224'>
	- <var-decl name='st_uid' type-id='cc5fcceb' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='256'>
	- <var-decl name='st_gid' type-id='d94ec6d9' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='288'>
	- <var-decl name='__pad0' type-id='95e97e5e' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='320'>
	- <var-decl name='st_rdev' type-id='35ed8932' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='384'>
	- <var-decl name='st_size' type-id='79989e9c' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='448'>
	- <var-decl name='st_blksize' type-id='d3f10a7f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='512'>
	- <var-decl name='st_blocks' type-id='4e711bf1' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='576'>
	- <var-decl name='st_atim' type-id='a9c79a1f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='704'>
	- <var-decl name='st_mtim' type-id='a9c79a1f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='832'>
	- <var-decl name='st_ctim' type-id='a9c79a1f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='960'>
	- <var-decl name='__glibc_reserved' type-id='083f8d58' visibility='default'/>
	+ <var-decl name='_pos' type-id='95e97e5e' visibility='default'/>
	</data-member>
	</class-decl>
	- <typedef-decl name='__dev_t' type-id='7359adad' id='35ed8932'/>
	- <typedef-decl name='__gid_t' type-id='f0981eeb' id='d94ec6d9'/>
	- <typedef-decl name='__ino64_t' type-id='7359adad' id='71288a47'/>
	- <typedef-decl name='__mode_t' type-id='f0981eeb' id='e1c52942'/>
	- <typedef-decl name='__nlink_t' type-id='7359adad' id='80f0b9df'/>
	- <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
	- <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
	- <typedef-decl name='__time_t' type-id='bd54fe1a' id='65eda9c0'/>
	- <typedef-decl name='__blksize_t' type-id='bd54fe1a' id='d3f10a7f'/>
	- <typedef-decl name='__blkcnt64_t' type-id='bd54fe1a' id='4e711bf1'/>
	- <typedef-decl name='__syscall_slong_t' type-id='bd54fe1a' id='03085adc'/>
	- <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
	- <typedef-decl name='_IO_lock_t' type-id='48b5725f' id='bb4788fa'/>
	<class-decl name='_IO_FILE' size-in-bits='1728' is-struct='yes' visibility='default' id='ec1ed955'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='_flags' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='_IO_read_ptr' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='_IO_read_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='_IO_read_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='_IO_write_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='_IO_write_ptr' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='_IO_write_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='_IO_buf_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='_IO_buf_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='576'>
	<var-decl name='_IO_save_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='640'>
	<var-decl name='_IO_backup_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='704'>
	<var-decl name='_IO_save_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='768'>
	<var-decl name='_markers' type-id='e4c6fa61' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='832'>
	<var-decl name='_chain' type-id='dca988a5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='896'>
	<var-decl name='_fileno' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='928'>
	<var-decl name='_flags2' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='960'>
	<var-decl name='_old_offset' type-id='79989e9c' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1024'>
	<var-decl name='_cur_column' type-id='8efea9e5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1040'>
	<var-decl name='_vtable_offset' type-id='28577a57' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1048'>
	<var-decl name='_shortbuf' type-id='89feb1ec' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1088'>
	<var-decl name='_lock' type-id='cecf4ea7' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1152'>
	<var-decl name='_offset' type-id='724e4de6' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1216'>
	- <var-decl name='_codecvt' type-id='570f8c59' visibility='default'/>
	+ <var-decl name='__pad1' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1280'>
	- <var-decl name='_wide_data' type-id='c65a1f29' visibility='default'/>
	+ <var-decl name='__pad2' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1344'>
	- <var-decl name='_freeres_list' type-id='dca988a5' visibility='default'/>
	+ <var-decl name='__pad3' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1408'>
	- <var-decl name='_freeres_buf' type-id='eaa32e2f' visibility='default'/>
	+ <var-decl name='__pad4' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1472'>
	<var-decl name='__pad5' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1536'>
	<var-decl name='_mode' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1568'>
	<var-decl name='_unused2' type-id='664ac0b7' visibility='default'/>
	</data-member>
	</class-decl>
	+ <class-decl name='stat64' size-in-bits='1152' is-struct='yes' visibility='default' id='0bbec9cd'>
	+ <data-member access='public' layout-offset-in-bits='0'>
	+ <var-decl name='st_dev' type-id='35ed8932' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='64'>
	+ <var-decl name='st_ino' type-id='71288a47' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='128'>
	+ <var-decl name='st_nlink' type-id='80f0b9df' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='192'>
	+ <var-decl name='st_mode' type-id='e1c52942' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='224'>
	+ <var-decl name='st_uid' type-id='cc5fcceb' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='256'>
	+ <var-decl name='st_gid' type-id='d94ec6d9' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='288'>
	+ <var-decl name='__pad0' type-id='95e97e5e' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='320'>
	+ <var-decl name='st_rdev' type-id='35ed8932' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='384'>
	+ <var-decl name='st_size' type-id='79989e9c' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='448'>
	+ <var-decl name='st_blksize' type-id='d3f10a7f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='512'>
	+ <var-decl name='st_blocks' type-id='4e711bf1' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='576'>
	+ <var-decl name='st_atim' type-id='a9c79a1f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='704'>
	+ <var-decl name='st_mtim' type-id='a9c79a1f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='832'>
	+ <var-decl name='st_ctim' type-id='a9c79a1f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='960'>
	+ <var-decl name='__glibc_reserved' type-id='083f8d58' visibility='default'/>
	+ </data-member>
	+ </class-decl>
	+ <typedef-decl name='__dev_t' type-id='7359adad' id='35ed8932'/>
	+ <typedef-decl name='__gid_t' type-id='f0981eeb' id='d94ec6d9'/>
	+ <typedef-decl name='__ino64_t' type-id='7359adad' id='71288a47'/>
	+ <typedef-decl name='__mode_t' type-id='f0981eeb' id='e1c52942'/>
	+ <typedef-decl name='__nlink_t' type-id='7359adad' id='80f0b9df'/>
	+ <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
	+ <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
	+ <typedef-decl name='__time_t' type-id='bd54fe1a' id='65eda9c0'/>
	+ <typedef-decl name='__blksize_t' type-id='bd54fe1a' id='d3f10a7f'/>
	+ <typedef-decl name='__blkcnt64_t' type-id='bd54fe1a' id='4e711bf1'/>
	+ <typedef-decl name='__syscall_slong_t' type-id='bd54fe1a' id='03085adc'/>
	+ <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
	<class-decl name='timespec' size-in-bits='128' is-struct='yes' visibility='default' id='a9c79a1f'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='tv_sec' type-id='65eda9c0' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='tv_nsec' type-id='03085adc' visibility='default'/>
	</data-member>
	</class-decl>
	<pointer-type-def type-id='aa12d1ba' size-in-bits='64' id='822cd80b'/>
	<pointer-type-def type-id='ec1ed955' size-in-bits='64' id='dca988a5'/>
	- <pointer-type-def type-id='a4036571' size-in-bits='64' id='570f8c59'/>
	<pointer-type-def type-id='bb4788fa' size-in-bits='64' id='cecf4ea7'/>
	<pointer-type-def type-id='010ae0b9' size-in-bits='64' id='e4c6fa61'/>
	- <pointer-type-def type-id='79bd3751' size-in-bits='64' id='c65a1f29'/>
	<pointer-type-def type-id='0c544dc0' size-in-bits='64' id='394fc496'/>
	<pointer-type-def type-id='0bbec9cd' size-in-bits='64' id='62f7a03d'/>
	- <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
	- <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
	- <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
	<function-decl name='getmntany' mangled-name='getmntany' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getmntany'>
	<parameter type-id='822cd80b' name='fp'/>
	<parameter type-id='9d424d31' name='mgetp'/>
	<parameter type-id='9d424d31' name='mrefp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='_sol_getmntent' mangled-name='_sol_getmntent' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='_sol_getmntent'>
	<parameter type-id='822cd80b' name='fp'/>
	<parameter type-id='9d424d31' name='mgetp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='getextmntent' mangled-name='getextmntent' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getextmntent'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='394fc496' name='entry'/>
	<parameter type-id='62f7a03d' name='statbuf'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/zone.c' language='LANG_C99'>
	<typedef-decl name='zoneid_t' type-id='95e97e5e' id='4da03624'/>
	<function-decl name='getzoneid' mangled-name='getzoneid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getzoneid'>
	<return type-id='4da03624'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='page.c' language='LANG_C99'>
	<function-decl name='spl_pagesize' mangled-name='spl_pagesize' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='spl_pagesize'>
	<return type-id='b59d7dce'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='strlcat.c' language='LANG_C99'>
	<function-decl name='strlcat' mangled-name='strlcat' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='strlcat'>
	<parameter type-id='26a90f95' name='dst'/>
	<parameter type-id='80f4b756' name='src'/>
	<parameter type-id='b59d7dce' name='dstsize'/>
	<return type-id='b59d7dce'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='strlcpy.c' language='LANG_C99'>
	<function-decl name='strlcpy' mangled-name='strlcpy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='strlcpy'>
	<parameter type-id='26a90f95' name='dst'/>
	<parameter type-id='80f4b756' name='src'/>
	<parameter type-id='b59d7dce' name='len'/>
	<return type-id='b59d7dce'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='timestamp.c' language='LANG_C99'>
	<function-decl name='print_timestamp' mangled-name='print_timestamp' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='print_timestamp'>
	<parameter type-id='3502e3ff' name='timestamp_fmt'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='thread_pool.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='384' id='36d7f119'>
	<subrange length='48' type-id='7359adad' id='8f6d2a81'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='448' id='6093ff7c'>
	<subrange length='56' type-id='7359adad' id='f8137894'/>
	</array-type-def>
	<type-decl name='long long unsigned int' size-in-bits='64' id='3a47d82b'/>
	<array-type-def dimensions='1' type-id='f0981eeb' size-in-bits='64' id='0d532ec1'>
	<subrange length='2' type-id='7359adad' id='52efc4ef'/>
	</array-type-def>
	<typedef-decl name='tpool_t' type-id='88d1b7f9' id='b1bbf10d'/>
	<typedef-decl name='tpool_job_t' type-id='3b8579e5' id='66a0afc9'/>
	<class-decl name='tpool_job' size-in-bits='192' is-struct='yes' visibility='default' id='3b8579e5'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='tpj_next' type-id='f32b30e4' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='tpj_func' type-id='b7f9d8e6' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='tpj_arg' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='tpool_active_t' type-id='c8d086f4' id='6fcda10e'/>
	<class-decl name='tpool_active' size-in-bits='128' is-struct='yes' visibility='default' id='c8d086f4'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='tpa_next' type-id='ad33e5e7' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='tpa_tid' type-id='4051f5e7' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='tpool' size-in-bits='2496' is-struct='yes' visibility='default' id='88d1b7f9'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='tp_forw' type-id='9cf59a50' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='tp_back' type-id='9cf59a50' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='tp_mutex' type-id='7a6844eb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='tp_busycv' type-id='62fab762' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='832'>
	<var-decl name='tp_workcv' type-id='62fab762' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1216'>
	<var-decl name='tp_waitcv' type-id='62fab762' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1600'>
	<var-decl name='tp_active' type-id='ad33e5e7' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1664'>
	<var-decl name='tp_head' type-id='f32b30e4' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1728'>
	<var-decl name='tp_tail' type-id='f32b30e4' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1792'>
	<var-decl name='tp_attr' type-id='7d8569fd' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2240'>
	<var-decl name='tp_flags' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2272'>
	<var-decl name='tp_linger' type-id='3502e3ff' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2304'>
	<var-decl name='tp_njobs' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2336'>
	<var-decl name='tp_minimum' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2368'>
	<var-decl name='tp_maximum' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2400'>
	<var-decl name='tp_current' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2432'>
	<var-decl name='tp_idle' type-id='95e97e5e' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='pthread_t' type-id='7359adad' id='4051f5e7'/>
	<union-decl name='pthread_attr_t' size-in-bits='448' visibility='default' id='b63afacd'>
	<data-member access='public'>
	<var-decl name='__size' type-id='6093ff7c' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='__align' type-id='bd54fe1a' visibility='default'/>
	</data-member>
	</union-decl>
	<typedef-decl name='pthread_attr_t' type-id='b63afacd' id='7d8569fd'/>
	<union-decl name='pthread_cond_t' size-in-bits='384' naming-typedef-id='62fab762' visibility='default' id='cbb12c12'>
	<data-member access='public'>
	<var-decl name='__data' type-id='c987b47c' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='__size' type-id='36d7f119' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='__align' type-id='1eb56b1e' visibility='default'/>
	</data-member>
	</union-decl>
	<typedef-decl name='pthread_cond_t' type-id='cbb12c12' id='62fab762'/>
	<class-decl name='__pthread_cond_s' size-in-bits='384' is-struct='yes' visibility='default' id='c987b47c'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='' type-id='ac5ab595' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='' type-id='ac5ab596' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='__g_refs' type-id='0d532ec1' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='__g_size' type-id='0d532ec1' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='__g1_orig_size' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='288'>
	<var-decl name='__wrefs' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='__g_signals' type-id='0d532ec1' visibility='default'/>
	</data-member>
	</class-decl>
	<union-decl name='__anonymous_union__1' size-in-bits='64' is-anonymous='yes' visibility='default' id='ac5ab595'>
	<data-member access='public'>
	<var-decl name='__wseq' type-id='3a47d82b' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='__wseq32' type-id='e7f43f72' visibility='default'/>
	</data-member>
	</union-decl>
	<class-decl name='__anonymous_struct__' size-in-bits='64' is-struct='yes' is-anonymous='yes' visibility='default' id='e7f43f72'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='__low' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='__high' type-id='f0981eeb' visibility='default'/>
	</data-member>
	</class-decl>
	<union-decl name='__anonymous_union__2' size-in-bits='64' is-anonymous='yes' visibility='default' id='ac5ab596'>
	<data-member access='public'>
	<var-decl name='__g1_start' type-id='3a47d82b' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='__g1_start32' type-id='e7f43f72' visibility='default'/>
	</data-member>
	</union-decl>
	<pointer-type-def type-id='7d8569fd' size-in-bits='64' id='7347a39e'/>
	<pointer-type-def type-id='6fcda10e' size-in-bits='64' id='ad33e5e7'/>
	<pointer-type-def type-id='66a0afc9' size-in-bits='64' id='f32b30e4'/>
	<pointer-type-def type-id='b1bbf10d' size-in-bits='64' id='9cf59a50'/>
	<pointer-type-def type-id='c5c76c9c' size-in-bits='64' id='b7f9d8e6'/>
	<function-decl name='tpool_create' mangled-name='tpool_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='tpool_create'>
	<parameter type-id='3502e3ff' name='min_threads'/>
	<parameter type-id='3502e3ff' name='max_threads'/>
	<parameter type-id='3502e3ff' name='linger'/>
	<parameter type-id='7347a39e' name='attr'/>
	<return type-id='9cf59a50'/>
	</function-decl>
	<function-decl name='tpool_dispatch' mangled-name='tpool_dispatch' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='tpool_dispatch'>
	<parameter type-id='9cf59a50' name='tpool'/>
	<parameter type-id='b7f9d8e6' name='func'/>
	<parameter type-id='eaa32e2f' name='arg'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='tpool_destroy' mangled-name='tpool_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='tpool_destroy'>
	<parameter type-id='9cf59a50' name='tpool'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='tpool_abandon' mangled-name='tpool_abandon' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='tpool_abandon'>
	<parameter type-id='9cf59a50' name='tpool'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='tpool_wait' mangled-name='tpool_wait' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='tpool_wait'>
	<parameter type-id='9cf59a50' name='tpool'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='tpool_suspend' mangled-name='tpool_suspend' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='tpool_suspend'>
	<parameter type-id='9cf59a50' name='tpool'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='tpool_suspended' mangled-name='tpool_suspended' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='tpool_suspended'>
	<parameter type-id='9cf59a50' name='tpool'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='tpool_resume' mangled-name='tpool_resume' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='tpool_resume'>
	<parameter type-id='9cf59a50' name='tpool'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='tpool_member' mangled-name='tpool_member' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='tpool_member'>
	<parameter type-id='9cf59a50' name='tpool'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-type size-in-bits='64' id='c5c76c9c'>
	<parameter type-id='eaa32e2f'/>
	<return type-id='48b5725f'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/cityhash.c' language='LANG_C99'>
	<function-decl name='cityhash4' mangled-name='cityhash4' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='cityhash4'>
	<parameter type-id='9c313c2d' name='w1'/>
	<parameter type-id='9c313c2d' name='w2'/>
	<parameter type-id='9c313c2d' name='w3'/>
	<parameter type-id='9c313c2d' name='w4'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zfeature_common.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='83f29ca2' size-in-bits='15232' id='d96379d0'>
	<subrange length='34' type-id='7359adad' id='6a6a7e00'/>
	</array-type-def>
	<enum-decl name='spa_feature' id='33ecb627'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='SPA_FEATURE_NONE' value='-1'/>
	<enumerator name='SPA_FEATURE_ASYNC_DESTROY' value='0'/>
	<enumerator name='SPA_FEATURE_EMPTY_BPOBJ' value='1'/>
	<enumerator name='SPA_FEATURE_LZ4_COMPRESS' value='2'/>
	<enumerator name='SPA_FEATURE_MULTI_VDEV_CRASH_DUMP' value='3'/>
	<enumerator name='SPA_FEATURE_SPACEMAP_HISTOGRAM' value='4'/>
	<enumerator name='SPA_FEATURE_ENABLED_TXG' value='5'/>
	<enumerator name='SPA_FEATURE_HOLE_BIRTH' value='6'/>
	<enumerator name='SPA_FEATURE_EXTENSIBLE_DATASET' value='7'/>
	<enumerator name='SPA_FEATURE_EMBEDDED_DATA' value='8'/>
	<enumerator name='SPA_FEATURE_BOOKMARKS' value='9'/>
	<enumerator name='SPA_FEATURE_FS_SS_LIMIT' value='10'/>
	<enumerator name='SPA_FEATURE_LARGE_BLOCKS' value='11'/>
	<enumerator name='SPA_FEATURE_LARGE_DNODE' value='12'/>
	<enumerator name='SPA_FEATURE_SHA512' value='13'/>
	<enumerator name='SPA_FEATURE_SKEIN' value='14'/>
	<enumerator name='SPA_FEATURE_EDONR' value='15'/>
	<enumerator name='SPA_FEATURE_USEROBJ_ACCOUNTING' value='16'/>
	<enumerator name='SPA_FEATURE_ENCRYPTION' value='17'/>
	<enumerator name='SPA_FEATURE_PROJECT_QUOTA' value='18'/>
	<enumerator name='SPA_FEATURE_DEVICE_REMOVAL' value='19'/>
	<enumerator name='SPA_FEATURE_OBSOLETE_COUNTS' value='20'/>
	<enumerator name='SPA_FEATURE_POOL_CHECKPOINT' value='21'/>
	<enumerator name='SPA_FEATURE_SPACEMAP_V2' value='22'/>
	<enumerator name='SPA_FEATURE_ALLOCATION_CLASSES' value='23'/>
	<enumerator name='SPA_FEATURE_RESILVER_DEFER' value='24'/>
	<enumerator name='SPA_FEATURE_BOOKMARK_V2' value='25'/>
	<enumerator name='SPA_FEATURE_REDACTION_BOOKMARKS' value='26'/>
	<enumerator name='SPA_FEATURE_REDACTED_DATASETS' value='27'/>
	<enumerator name='SPA_FEATURE_BOOKMARK_WRITTEN' value='28'/>
	<enumerator name='SPA_FEATURE_LOG_SPACEMAP' value='29'/>
	<enumerator name='SPA_FEATURE_LIVELIST' value='30'/>
	<enumerator name='SPA_FEATURE_DEVICE_REBUILD' value='31'/>
	<enumerator name='SPA_FEATURE_ZSTD_COMPRESS' value='32'/>
	<enumerator name='SPA_FEATURE_DRAID' value='33'/>
	<enumerator name='SPA_FEATURES' value='34'/>
	</enum-decl>
	<typedef-decl name='spa_feature_t' type-id='33ecb627' id='d6618c78'/>
	<enum-decl name='zfeature_flags' id='6db816a4'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZFEATURE_FLAG_READONLY_COMPAT' value='1'/>
	<enumerator name='ZFEATURE_FLAG_MOS' value='2'/>
	<enumerator name='ZFEATURE_FLAG_ACTIVATE_ON_ENABLE' value='4'/>
	<enumerator name='ZFEATURE_FLAG_PER_DATASET' value='8'/>
	</enum-decl>
	<typedef-decl name='zfeature_flags_t' type-id='6db816a4' id='fc329033'/>
	<enum-decl name='zfeature_type' id='c4fa2355'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZFEATURE_TYPE_BOOLEAN' value='0'/>
	<enumerator name='ZFEATURE_TYPE_UINT64_ARRAY' value='1'/>
	<enumerator name='ZFEATURE_NUM_TYPES' value='2'/>
	</enum-decl>
	<typedef-decl name='zfeature_type_t' type-id='c4fa2355' id='732d2bb2'/>
	<class-decl name='zfeature_info' size-in-bits='448' is-struct='yes' visibility='default' id='1178d146'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='fi_feature' type-id='d6618c78' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='fi_uname' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='fi_guid' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='fi_desc' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='fi_flags' type-id='fc329033' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='288'>
	<var-decl name='fi_zfs_mod_supported' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='fi_type' type-id='732d2bb2' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='fi_depends' type-id='1acff326' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfeature_info_t' type-id='1178d146' id='83f29ca2'/>
	<qualified-type-def type-id='d6618c78' const='yes' id='81a65028'/>
	<pointer-type-def type-id='81a65028' size-in-bits='64' id='1acff326'/>
	<pointer-type-def type-id='d6618c78' size-in-bits='64' id='a8425263'/>
	<var-decl name='spa_feature_table' type-id='d96379d0' mangled-name='spa_feature_table' visibility='default' elf-symbol-id='spa_feature_table'/>
	<var-decl name='zfeature_checks_disable' type-id='c19b74c3' mangled-name='zfeature_checks_disable' visibility='default' elf-symbol-id='zfeature_checks_disable'/>
	<function-decl name='zfeature_is_valid_guid' mangled-name='zfeature_is_valid_guid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfeature_is_valid_guid'>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfeature_is_supported' mangled-name='zfeature_is_supported' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfeature_is_supported'>
	<parameter type-id='80f4b756' name='guid'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfeature_lookup_guid' mangled-name='zfeature_lookup_guid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfeature_lookup_guid'>
	<parameter type-id='80f4b756' name='guid'/>
	<parameter type-id='a8425263' name='res'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfeature_lookup_name' mangled-name='zfeature_lookup_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfeature_lookup_name'>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='a8425263' name='res'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfeature_depends_on' mangled-name='zfeature_depends_on' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfeature_depends_on'>
	<parameter type-id='d6618c78' name='fid'/>
	<parameter type-id='d6618c78' name='check'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_mod_supported' mangled-name='zfs_mod_supported' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_mod_supported'>
	<parameter type-id='80f4b756' name='scope'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zpool_feature_init' mangled-name='zpool_feature_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_feature_init'>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zfs_comutil.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='80f4b756' size-in-bits='2624' id='ef31fedf'>
	<subrange length='41' type-id='7359adad' id='cb834f44'/>
	</array-type-def>
	<class-decl name='zpool_load_policy' size-in-bits='256' is-struct='yes' visibility='default' id='2f65b36f'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='zlp_rewind' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='zlp_maxmeta' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='zlp_maxdata' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='zlp_txg' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zpool_load_policy_t' type-id='2f65b36f' id='d11b7617'/>
	<pointer-type-def type-id='d11b7617' size-in-bits='64' id='23432aaa'/>
	<var-decl name='zfs_history_event_names' type-id='ef31fedf' mangled-name='zfs_history_event_names' visibility='default' elf-symbol-id='zfs_history_event_names'/>
	<function-decl name='zfs_allocatable_devs' mangled-name='zfs_allocatable_devs' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_allocatable_devs'>
	<parameter type-id='5ce45b60' name='nv'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_special_devs' mangled-name='zfs_special_devs' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_special_devs'>
	<parameter type-id='5ce45b60' name='nv'/>
	<parameter type-id='26a90f95' name='type'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zpool_get_load_policy' mangled-name='zpool_get_load_policy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_load_policy'>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='23432aaa' name='zlpp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_zpl_version_map' mangled-name='zfs_zpl_version_map' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_zpl_version_map'>
	<parameter type-id='95e97e5e' name='spa_version'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_spa_version_map' mangled-name='zfs_spa_version_map' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_spa_version_map'>
	<parameter type-id='95e97e5e' name='zpl_version'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_dataset_name_hidden' mangled-name='zfs_dataset_name_hidden' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_dataset_name_hidden'>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zfs_deleg.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='f3f851ad' size-in-bits='infinite' id='bc4e5d90'>
	<subrange length='infinite' id='031f2035'/>
	</array-type-def>
	<enum-decl name='zfs_deleg_who_type_t' naming-typedef-id='36d4bd5a' id='b5fa5816'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZFS_DELEG_WHO_UNKNOWN' value='0'/>
	<enumerator name='ZFS_DELEG_USER' value='117'/>
	<enumerator name='ZFS_DELEG_USER_SETS' value='85'/>
	<enumerator name='ZFS_DELEG_GROUP' value='103'/>
	<enumerator name='ZFS_DELEG_GROUP_SETS' value='71'/>
	<enumerator name='ZFS_DELEG_EVERYONE' value='101'/>
	<enumerator name='ZFS_DELEG_EVERYONE_SETS' value='69'/>
	<enumerator name='ZFS_DELEG_CREATE' value='99'/>
	<enumerator name='ZFS_DELEG_CREATE_SETS' value='67'/>
	<enumerator name='ZFS_DELEG_NAMED_SET' value='115'/>
	<enumerator name='ZFS_DELEG_NAMED_SET_SETS' value='83'/>
	</enum-decl>
	<typedef-decl name='zfs_deleg_who_type_t' type-id='b5fa5816' id='36d4bd5a'/>
	<enum-decl name='zfs_deleg_note_t' naming-typedef-id='4613c173' id='729d4547'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZFS_DELEG_NOTE_CREATE' value='0'/>
	<enumerator name='ZFS_DELEG_NOTE_DESTROY' value='1'/>
	<enumerator name='ZFS_DELEG_NOTE_SNAPSHOT' value='2'/>
	<enumerator name='ZFS_DELEG_NOTE_ROLLBACK' value='3'/>
	<enumerator name='ZFS_DELEG_NOTE_CLONE' value='4'/>
	<enumerator name='ZFS_DELEG_NOTE_PROMOTE' value='5'/>
	<enumerator name='ZFS_DELEG_NOTE_RENAME' value='6'/>
	<enumerator name='ZFS_DELEG_NOTE_SEND' value='7'/>
	<enumerator name='ZFS_DELEG_NOTE_RECEIVE' value='8'/>
	<enumerator name='ZFS_DELEG_NOTE_ALLOW' value='9'/>
	<enumerator name='ZFS_DELEG_NOTE_USERPROP' value='10'/>
	<enumerator name='ZFS_DELEG_NOTE_MOUNT' value='11'/>
	<enumerator name='ZFS_DELEG_NOTE_SHARE' value='12'/>
	<enumerator name='ZFS_DELEG_NOTE_USERQUOTA' value='13'/>
	<enumerator name='ZFS_DELEG_NOTE_GROUPQUOTA' value='14'/>
	<enumerator name='ZFS_DELEG_NOTE_USERUSED' value='15'/>
	<enumerator name='ZFS_DELEG_NOTE_GROUPUSED' value='16'/>
	<enumerator name='ZFS_DELEG_NOTE_USEROBJQUOTA' value='17'/>
	<enumerator name='ZFS_DELEG_NOTE_GROUPOBJQUOTA' value='18'/>
	<enumerator name='ZFS_DELEG_NOTE_USEROBJUSED' value='19'/>
	<enumerator name='ZFS_DELEG_NOTE_GROUPOBJUSED' value='20'/>
	<enumerator name='ZFS_DELEG_NOTE_HOLD' value='21'/>
	<enumerator name='ZFS_DELEG_NOTE_RELEASE' value='22'/>
	<enumerator name='ZFS_DELEG_NOTE_DIFF' value='23'/>
	<enumerator name='ZFS_DELEG_NOTE_BOOKMARK' value='24'/>
	<enumerator name='ZFS_DELEG_NOTE_LOAD_KEY' value='25'/>
	<enumerator name='ZFS_DELEG_NOTE_CHANGE_KEY' value='26'/>
	<enumerator name='ZFS_DELEG_NOTE_PROJECTUSED' value='27'/>
	<enumerator name='ZFS_DELEG_NOTE_PROJECTQUOTA' value='28'/>
	<enumerator name='ZFS_DELEG_NOTE_PROJECTOBJUSED' value='29'/>
	<enumerator name='ZFS_DELEG_NOTE_PROJECTOBJQUOTA' value='30'/>
	<enumerator name='ZFS_DELEG_NOTE_NONE' value='31'/>
	</enum-decl>
	<typedef-decl name='zfs_deleg_note_t' type-id='729d4547' id='4613c173'/>
	<class-decl name='zfs_deleg_perm_tab' size-in-bits='128' is-struct='yes' visibility='default' id='5aa05c1f'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='z_perm' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='z_note' type-id='4613c173' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfs_deleg_perm_tab_t' type-id='5aa05c1f' id='f3f851ad'/>
	<var-decl name='zfs_deleg_perm_tab' type-id='bc4e5d90' mangled-name='zfs_deleg_perm_tab' visibility='default' elf-symbol-id='zfs_deleg_perm_tab'/>
	<function-decl name='zfs_deleg_canonicalize_perm' mangled-name='zfs_deleg_canonicalize_perm' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_deleg_canonicalize_perm'>
	<parameter type-id='80f4b756' name='perm'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zfs_deleg_verify_nvlist' mangled-name='zfs_deleg_verify_nvlist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_deleg_verify_nvlist'>
	<parameter type-id='5ce45b60' name='nvp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_deleg_whokey' mangled-name='zfs_deleg_whokey' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_deleg_whokey'>
	<parameter type-id='26a90f95' name='attr'/>
	<parameter type-id='36d4bd5a' name='type'/>
	<parameter type-id='a84c031d' name='inheritchr'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zfs_fletcher.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='9c313c2d' size-in-bits='256' id='85c64d26'>
	<subrange length='4' type-id='7359adad' id='16fe7105'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='9c313c2d' size-in-bits='512' id='c5d13f42'>
	<subrange length='8' type-id='7359adad' id='56e0c0b1'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='90dbb6d6' size-in-bits='2048' id='16582e69'>
	<subrange length='4' type-id='7359adad' id='16fe7105'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='8240361c' size-in-bits='1024' id='481f90b1'>
	<subrange length='4' type-id='7359adad' id='16fe7105'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='7c1ab40c' size-in-bits='512' id='cbd91ec1'>
	<subrange length='4' type-id='7359adad' id='16fe7105'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='6d059eaa' size-in-bits='1024' id='729b6ebb'>
	<subrange length='4' type-id='7359adad' id='16fe7105'/>
	</array-type-def>
	<class-decl name='zio_cksum' size-in-bits='256' is-struct='yes' visibility='default' id='1d53e28b'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='zc_word' type-id='85c64d26' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zio_cksum_t' type-id='1d53e28b' id='39730d0b'/>
	<enum-decl name='zio_byteorder_t' naming-typedef-id='595a65ec' id='fc861be0'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZIO_CHECKSUM_NATIVE' value='0'/>
	<enumerator name='ZIO_CHECKSUM_BYTESWAP' value='1'/>
	</enum-decl>
	<typedef-decl name='zio_byteorder_t' type-id='fc861be0' id='595a65ec'/>
	<class-decl name='zio_abd_checksum_data' size-in-bits='256' is-struct='yes' visibility='default' id='4bf4b004'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='acd_byteorder' type-id='595a65ec' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='acd_ctx' type-id='0f7df99e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='acd_zcp' type-id='c24fc2ee' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='acd_private' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zio_abd_checksum_data_t' type-id='4bf4b004' id='74e39470'/>
	<typedef-decl name='zio_abd_checksum_init_t' type-id='a5444274' id='029a8ebe'/>
	<typedef-decl name='zio_abd_checksum_fini_t' type-id='a5444274' id='d6fd5c6c'/>
	<typedef-decl name='zio_abd_checksum_iter_t' type-id='f4a1892e' id='cefa0f4a'/>
	<class-decl name='zio_abd_checksum_func' size-in-bits='192' is-struct='yes' visibility='default' id='aa14691a'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='acf_init' type-id='0bcca125' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='acf_fini' type-id='bfe36153' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='acf_iter' type-id='1e276399' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zio_abd_checksum_func_t' type-id='3f8e8d11' id='c2eb138a'/>
	<class-decl name='zfs_fletcher_superscalar' size-in-bits='256' is-struct='yes' visibility='default' id='28efb250'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='v' type-id='85c64d26' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfs_fletcher_superscalar_t' type-id='28efb250' id='6d059eaa'/>
	<class-decl name='zfs_fletcher_sse' size-in-bits='128' is-struct='yes' visibility='default' id='acd4019a'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='v' type-id='c1c22e6c' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfs_fletcher_sse_t' type-id='acd4019a' id='7c1ab40c'/>
	<class-decl name='zfs_fletcher_avx' size-in-bits='256' is-struct='yes' visibility='default' id='8c208dfa'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='v' type-id='85c64d26' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfs_fletcher_avx_t' type-id='8c208dfa' id='8240361c'/>
	<class-decl name='zfs_fletcher_avx512' size-in-bits='512' is-struct='yes' visibility='default' id='c6d0c382'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='v' type-id='c5d13f42' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfs_fletcher_avx512_t' type-id='c6d0c382' id='90dbb6d6'/>
	<union-decl name='fletcher_4_ctx' size-in-bits='2048' visibility='default' id='1f951ade'>
	<data-member access='public'>
	<var-decl name='scalar' type-id='39730d0b' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='superscalar' type-id='729b6ebb' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='sse' type-id='cbd91ec1' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='avx' type-id='481f90b1' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='avx512' type-id='16582e69' visibility='default'/>
	</data-member>
	</union-decl>
	<typedef-decl name='fletcher_4_ctx_t' type-id='1f951ade' id='4b675395'/>
	<qualified-type-def type-id='aa14691a' const='yes' id='3f8e8d11'/>
	<pointer-type-def type-id='4b675395' size-in-bits='64' id='0f7df99e'/>
	<pointer-type-def type-id='74e39470' size-in-bits='64' id='eefe7427'/>
	<pointer-type-def type-id='d6fd5c6c' size-in-bits='64' id='bfe36153'/>
	<pointer-type-def type-id='029a8ebe' size-in-bits='64' id='0bcca125'/>
	<pointer-type-def type-id='cefa0f4a' size-in-bits='64' id='1e276399'/>
	<pointer-type-def type-id='39730d0b' size-in-bits='64' id='c24fc2ee'/>
	<var-decl name='fletcher_4_abd_ops' type-id='c2eb138a' mangled-name='fletcher_4_abd_ops' visibility='default' elf-symbol-id='fletcher_4_abd_ops'/>
	<function-decl name='fletcher_init' mangled-name='fletcher_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_init'>
	<parameter type-id='c24fc2ee' name='zcp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fletcher_2_incremental_native' mangled-name='fletcher_2_incremental_native' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_2_incremental_native'>
	<parameter type-id='eaa32e2f' name='buf'/>
	<parameter type-id='b59d7dce' name='size'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='fletcher_2_native' mangled-name='fletcher_2_native' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_2_native'>
	<parameter type-id='eaa32e2f' name='buf'/>
	<parameter type-id='9c313c2d' name='size'/>
	<parameter type-id='eaa32e2f' name='ctx_template'/>
	<parameter type-id='c24fc2ee' name='zcp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fletcher_2_incremental_byteswap' mangled-name='fletcher_2_incremental_byteswap' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_2_incremental_byteswap'>
	<parameter type-id='eaa32e2f' name='buf'/>
	<parameter type-id='b59d7dce' name='size'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='fletcher_2_byteswap' mangled-name='fletcher_2_byteswap' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_2_byteswap'>
	<parameter type-id='eaa32e2f' name='buf'/>
	<parameter type-id='9c313c2d' name='size'/>
	<parameter type-id='eaa32e2f' name='ctx_template'/>
	<parameter type-id='c24fc2ee' name='zcp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fletcher_4_impl_set' mangled-name='fletcher_4_impl_set' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_4_impl_set'>
	<parameter type-id='80f4b756' name='val'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='fletcher_4_native' mangled-name='fletcher_4_native' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_4_native'>
	<parameter type-id='eaa32e2f' name='buf'/>
	<parameter type-id='9c313c2d' name='size'/>
	<parameter type-id='eaa32e2f' name='ctx_template'/>
	<parameter type-id='c24fc2ee' name='zcp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fletcher_4_native_varsize' mangled-name='fletcher_4_native_varsize' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_4_native_varsize'>
	<parameter type-id='eaa32e2f' name='buf'/>
	<parameter type-id='9c313c2d' name='size'/>
	<parameter type-id='c24fc2ee' name='zcp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fletcher_4_byteswap' mangled-name='fletcher_4_byteswap' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_4_byteswap'>
	<parameter type-id='eaa32e2f' name='buf'/>
	<parameter type-id='9c313c2d' name='size'/>
	<parameter type-id='eaa32e2f' name='ctx_template'/>
	<parameter type-id='c24fc2ee' name='zcp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fletcher_4_incremental_native' mangled-name='fletcher_4_incremental_native' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_4_incremental_native'>
	<parameter type-id='eaa32e2f' name='buf'/>
	<parameter type-id='b59d7dce' name='size'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='fletcher_4_incremental_byteswap' mangled-name='fletcher_4_incremental_byteswap' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_4_incremental_byteswap'>
	<parameter type-id='eaa32e2f' name='buf'/>
	<parameter type-id='b59d7dce' name='size'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='fletcher_4_init' mangled-name='fletcher_4_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_4_init'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='fletcher_4_fini' mangled-name='fletcher_4_fini' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='fletcher_4_fini'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-type size-in-bits='64' id='f4a1892e'>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='b59d7dce'/>
	<parameter type-id='eaa32e2f'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='a5444274'>
	<parameter type-id='eefe7427'/>
	<return type-id='48b5725f'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zfs_fletcher_avx512.c' language='LANG_C99'>
	<typedef-decl name='fletcher_4_init_f' type-id='173aa527' id='b9ae1656'/>
	<typedef-decl name='fletcher_4_fini_f' type-id='0ad5b8a8' id='c4c1f4fc'/>
	<typedef-decl name='fletcher_4_compute_f' type-id='38147eff' id='ad1dc4cb'/>
	<class-decl name='fletcher_4_func' size-in-bits='512' is-struct='yes' visibility='default' id='57f479a0'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='init_native' type-id='b9ae1656' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='fini_native' type-id='c4c1f4fc' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='compute_native' type-id='ad1dc4cb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='init_byteswap' type-id='b9ae1656' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='fini_byteswap' type-id='c4c1f4fc' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='compute_byteswap' type-id='ad1dc4cb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='valid' type-id='297d38bc' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='name' type-id='80f4b756' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='fletcher_4_ops_t' type-id='57f479a0' id='eba91718'/>
	<qualified-type-def type-id='eba91718' const='yes' id='9eeabdc8'/>
	<pointer-type-def type-id='e9e61702' size-in-bits='64' id='297d38bc'/>
	<pointer-type-def type-id='fe40251b' size-in-bits='64' id='173aa527'/>
	<pointer-type-def type-id='17fb1f83' size-in-bits='64' id='38147eff'/>
	<pointer-type-def type-id='fb39e25e' size-in-bits='64' id='0ad5b8a8'/>
	<var-decl name='fletcher_4_avx512f_ops' type-id='9eeabdc8' mangled-name='fletcher_4_avx512f_ops' visibility='default' elf-symbol-id='fletcher_4_avx512f_ops'/>
	<var-decl name='fletcher_4_avx512bw_ops' type-id='9eeabdc8' mangled-name='fletcher_4_avx512bw_ops' visibility='default' elf-symbol-id='fletcher_4_avx512bw_ops'/>
	<function-type size-in-bits='64' id='e9e61702'>
	<return type-id='c19b74c3'/>
	</function-type>
	<function-type size-in-bits='64' id='fe40251b'>
	<parameter type-id='0f7df99e'/>
	<return type-id='48b5725f'/>
	</function-type>
	<function-type size-in-bits='64' id='17fb1f83'>
	<parameter type-id='0f7df99e'/>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='9c313c2d'/>
	<return type-id='48b5725f'/>
	</function-type>
	<function-type size-in-bits='64' id='fb39e25e'>
	<parameter type-id='0f7df99e'/>
	<parameter type-id='c24fc2ee'/>
	<return type-id='48b5725f'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zfs_fletcher_intel.c' language='LANG_C99'>
	<var-decl name='fletcher_4_avx2_ops' type-id='9eeabdc8' mangled-name='fletcher_4_avx2_ops' visibility='default' elf-symbol-id='fletcher_4_avx2_ops'/>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zfs_fletcher_sse.c' language='LANG_C99'>
	<var-decl name='fletcher_4_sse2_ops' type-id='9eeabdc8' mangled-name='fletcher_4_sse2_ops' visibility='default' elf-symbol-id='fletcher_4_sse2_ops'/>
	<var-decl name='fletcher_4_ssse3_ops' type-id='9eeabdc8' mangled-name='fletcher_4_ssse3_ops' visibility='default' elf-symbol-id='fletcher_4_ssse3_ops'/>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zfs_fletcher_superscalar.c' language='LANG_C99'>
	<var-decl name='fletcher_4_superscalar_ops' type-id='9eeabdc8' mangled-name='fletcher_4_superscalar_ops' visibility='default' elf-symbol-id='fletcher_4_superscalar_ops'/>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zfs_fletcher_superscalar4.c' language='LANG_C99'>
	<var-decl name='fletcher_4_superscalar4_ops' type-id='9eeabdc8' mangled-name='fletcher_4_superscalar4_ops' visibility='default' elf-symbol-id='fletcher_4_superscalar4_ops'/>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zfs_namecheck.c' language='LANG_C99'>
	<enum-decl name='namecheck_err_t' naming-typedef-id='8e0af06e' id='f43bbcda'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='NAME_ERR_LEADING_SLASH' value='0'/>
	<enumerator name='NAME_ERR_EMPTY_COMPONENT' value='1'/>
	<enumerator name='NAME_ERR_TRAILING_SLASH' value='2'/>
	<enumerator name='NAME_ERR_INVALCHAR' value='3'/>
	<enumerator name='NAME_ERR_MULTIPLE_DELIMITERS' value='4'/>
	<enumerator name='NAME_ERR_NOLETTER' value='5'/>
	<enumerator name='NAME_ERR_RESERVED' value='6'/>
	<enumerator name='NAME_ERR_DISKLIKE' value='7'/>
	<enumerator name='NAME_ERR_TOOLONG' value='8'/>
	<enumerator name='NAME_ERR_SELF_REF' value='9'/>
	<enumerator name='NAME_ERR_PARENT_REF' value='10'/>
	<enumerator name='NAME_ERR_NO_AT' value='11'/>
	<enumerator name='NAME_ERR_NO_POUND' value='12'/>
	</enum-decl>
	<typedef-decl name='namecheck_err_t' type-id='f43bbcda' id='8e0af06e'/>
	<pointer-type-def type-id='8e0af06e' size-in-bits='64' id='053457bd'/>
	<var-decl name='zfs_max_dataset_nesting' type-id='95e97e5e' mangled-name='zfs_max_dataset_nesting' visibility='default' elf-symbol-id='zfs_max_dataset_nesting'/>
	<function-decl name='get_dataset_depth' mangled-name='get_dataset_depth' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='get_dataset_depth'>
	<parameter type-id='80f4b756' name='path'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_component_namecheck' mangled-name='zfs_component_namecheck' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_component_namecheck'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='053457bd' name='why'/>
	<parameter type-id='26a90f95' name='what'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='permset_namecheck' mangled-name='permset_namecheck' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='permset_namecheck'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='053457bd' name='why'/>
	<parameter type-id='26a90f95' name='what'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='dataset_nestcheck' mangled-name='dataset_nestcheck' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='dataset_nestcheck'>
	<parameter type-id='80f4b756' name='path'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='entity_namecheck' mangled-name='entity_namecheck' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='entity_namecheck'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='053457bd' name='why'/>
	<parameter type-id='26a90f95' name='what'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='dataset_namecheck' mangled-name='dataset_namecheck' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='dataset_namecheck'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='053457bd' name='why'/>
	<parameter type-id='26a90f95' name='what'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='bookmark_namecheck' mangled-name='bookmark_namecheck' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='bookmark_namecheck'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='053457bd' name='why'/>
	<parameter type-id='26a90f95' name='what'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='snapshot_namecheck' mangled-name='snapshot_namecheck' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='snapshot_namecheck'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='053457bd' name='why'/>
	<parameter type-id='26a90f95' name='what'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='mountpoint_namecheck' mangled-name='mountpoint_namecheck' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='mountpoint_namecheck'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='053457bd' name='why'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='pool_namecheck' mangled-name='pool_namecheck' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='pool_namecheck'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='053457bd' name='why'/>
	<parameter type-id='26a90f95' name='what'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zfs_prop.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='80f4b756' size-in-bits='768' id='35e4b367'>
	<subrange length='12' type-id='7359adad' id='84827bdc'/>
	</array-type-def>
	<enum-decl name='zprop_type_t' naming-typedef-id='31429eff' id='87676253'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='PROP_TYPE_NUMBER' value='0'/>
	<enumerator name='PROP_TYPE_STRING' value='1'/>
	<enumerator name='PROP_TYPE_INDEX' value='2'/>
	</enum-decl>
	<typedef-decl name='zprop_type_t' type-id='87676253' id='31429eff'/>
	<enum-decl name='zprop_attr_t' naming-typedef-id='999701cc' id='77d05200'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='PROP_DEFAULT' value='0'/>
	<enumerator name='PROP_READONLY' value='1'/>
	<enumerator name='PROP_INHERIT' value='2'/>
	<enumerator name='PROP_ONETIME' value='3'/>
	<enumerator name='PROP_ONETIME_DEFAULT' value='4'/>
	</enum-decl>
	<typedef-decl name='zprop_attr_t' type-id='77d05200' id='999701cc'/>
	<class-decl name='zfs_index' size-in-bits='128' is-struct='yes' visibility='default' id='87957af9'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='pi_name' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='pi_value' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zprop_index_t' type-id='87957af9' id='64636ce3'/>
	<class-decl name='zprop_desc_t' size-in-bits='704' is-struct='yes' naming-typedef-id='ffa52b96' visibility='default' id='bbff5e4b'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='pd_name' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='pd_propnum' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='96'>
	<var-decl name='pd_proptype' type-id='31429eff' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='pd_strdefault' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='pd_numdefault' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='pd_attr' type-id='999701cc' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='288'>
	<var-decl name='pd_types' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='pd_values' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='pd_colname' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='pd_rightalign' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='480'>
	<var-decl name='pd_visible' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='pd_zfs_mod_supported' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='576'>
	<var-decl name='pd_table' type-id='c8bc397b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='640'>
	<var-decl name='pd_table_size' type-id='b59d7dce' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zprop_desc_t' type-id='bbff5e4b' id='ffa52b96'/>
	<pointer-type-def type-id='80f4b756' size-in-bits='64' id='7d3cd834'/>
	<qualified-type-def type-id='64636ce3' const='yes' id='072f7953'/>
	<pointer-type-def type-id='072f7953' size-in-bits='64' id='c8bc397b'/>
	<pointer-type-def type-id='ffa52b96' size-in-bits='64' id='76c8174b'/>
	<var-decl name='zfs_userquota_prop_prefixes' type-id='35e4b367' mangled-name='zfs_userquota_prop_prefixes' visibility='default' elf-symbol-id='zfs_userquota_prop_prefixes'/>
	<function-decl name='zfs_prop_get_table' mangled-name='zfs_prop_get_table' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_get_table'>
	<return type-id='76c8174b'/>
	</function-decl>
	<function-decl name='zfs_prop_init' mangled-name='zfs_prop_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_init'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_prop_delegatable' mangled-name='zfs_prop_delegatable' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_delegatable'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_name_to_prop' mangled-name='zfs_name_to_prop' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_name_to_prop'>
	<parameter type-id='80f4b756' name='propname'/>
	<return type-id='58603c44'/>
	</function-decl>
	<function-decl name='zfs_prop_user' mangled-name='zfs_prop_user' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_user'>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_prop_userquota' mangled-name='zfs_prop_userquota' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_userquota'>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_prop_written' mangled-name='zfs_prop_written' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_written'>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_prop_string_to_index' mangled-name='zfs_prop_string_to_index' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_string_to_index'>
	<parameter type-id='58603c44' name='prop'/>
	<parameter type-id='80f4b756' name='string'/>
	<parameter type-id='5d6479ae' name='index'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_prop_index_to_string' mangled-name='zfs_prop_index_to_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_index_to_string'>
	<parameter type-id='58603c44' name='prop'/>
	<parameter type-id='9c313c2d' name='index'/>
	<parameter type-id='7d3cd834' name='string'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_prop_random_value' mangled-name='zfs_prop_random_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_random_value'>
	<parameter type-id='58603c44' name='prop'/>
	<parameter type-id='9c313c2d' name='seed'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='zfs_prop_valid_for_type' mangled-name='zfs_prop_valid_for_type' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_valid_for_type'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='2e45de5d' name='types'/>
	<parameter type-id='c19b74c3' name='headcheck'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_prop_get_type' mangled-name='zfs_prop_get_type' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_get_type'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='31429eff'/>
	</function-decl>
	<function-decl name='zfs_prop_readonly' mangled-name='zfs_prop_readonly' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_readonly'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_prop_visible' mangled-name='zfs_prop_visible' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_visible'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_prop_setonce' mangled-name='zfs_prop_setonce' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_setonce'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_prop_default_string' mangled-name='zfs_prop_default_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_default_string'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zfs_prop_default_numeric' mangled-name='zfs_prop_default_numeric' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_default_numeric'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='zfs_prop_to_name' mangled-name='zfs_prop_to_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_to_name'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zfs_prop_inheritable' mangled-name='zfs_prop_inheritable' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_inheritable'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_prop_encryption_key_param' mangled-name='zfs_prop_encryption_key_param' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_encryption_key_param'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_prop_valid_keylocation' mangled-name='zfs_prop_valid_keylocation' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_valid_keylocation'>
	<parameter type-id='80f4b756' name='str'/>
	<parameter type-id='c19b74c3' name='encrypted'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_prop_values' mangled-name='zfs_prop_values' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_values'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zfs_prop_is_string' mangled-name='zfs_prop_is_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_is_string'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_prop_column_name' mangled-name='zfs_prop_column_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_column_name'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zfs_prop_align_right' mangled-name='zfs_prop_align_right' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_align_right'>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zpool_prop.c' language='LANG_C99'>
	<function-decl name='zpool_prop_get_table' mangled-name='zpool_prop_get_table' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_get_table'>
	<return type-id='76c8174b'/>
	</function-decl>
	<function-decl name='zpool_prop_init' mangled-name='zpool_prop_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_init'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zpool_name_to_prop' mangled-name='zpool_name_to_prop' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_name_to_prop'>
	<parameter type-id='80f4b756' name='propname'/>
	<return type-id='5d0c23fb'/>
	</function-decl>
	<function-decl name='zpool_prop_to_name' mangled-name='zpool_prop_to_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_to_name'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zpool_prop_get_type' mangled-name='zpool_prop_get_type' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_get_type'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<return type-id='31429eff'/>
	</function-decl>
	<function-decl name='zpool_prop_readonly' mangled-name='zpool_prop_readonly' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_readonly'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zpool_prop_setonce' mangled-name='zpool_prop_setonce' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_setonce'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zpool_prop_default_string' mangled-name='zpool_prop_default_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_default_string'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zpool_prop_default_numeric' mangled-name='zpool_prop_default_numeric' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_default_numeric'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='zpool_prop_feature' mangled-name='zpool_prop_feature' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_feature'>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zpool_prop_unsupported' mangled-name='zpool_prop_unsupported' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_unsupported'>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zpool_prop_string_to_index' mangled-name='zpool_prop_string_to_index' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_string_to_index'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<parameter type-id='80f4b756' name='string'/>
	<parameter type-id='5d6479ae' name='index'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_prop_index_to_string' mangled-name='zpool_prop_index_to_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_index_to_string'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<parameter type-id='9c313c2d' name='index'/>
	<parameter type-id='7d3cd834' name='string'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_prop_random_value' mangled-name='zpool_prop_random_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_random_value'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<parameter type-id='9c313c2d' name='seed'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='zpool_prop_values' mangled-name='zpool_prop_values' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_values'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zpool_prop_column_name' mangled-name='zpool_prop_column_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_column_name'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zpool_prop_align_right' mangled-name='zpool_prop_align_right' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_align_right'>
	<parameter type-id='5d0c23fb' name='prop'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	+ <function-decl name='vdev_prop_get_table' mangled-name='vdev_prop_get_table' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_get_table'>
	+ <return type-id='76c8174b'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_init' mangled-name='vdev_prop_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_init'>
	+ <return type-id='48b5725f'/>
	+ </function-decl>
	+ <function-decl name='vdev_name_to_prop' mangled-name='vdev_name_to_prop' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_name_to_prop'>
	+ <parameter type-id='80f4b756' name='propname'/>
	+ <return type-id='5aa5c90c'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_user' mangled-name='vdev_prop_user' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_user'>
	+ <parameter type-id='80f4b756' name='name'/>
	+ <return type-id='c19b74c3'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_to_name' mangled-name='vdev_prop_to_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_to_name'>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <return type-id='80f4b756'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_get_type' mangled-name='vdev_prop_get_type' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_get_type'>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <return type-id='31429eff'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_readonly' mangled-name='vdev_prop_readonly' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_readonly'>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <return type-id='c19b74c3'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_default_string' mangled-name='vdev_prop_default_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_default_string'>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <return type-id='80f4b756'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_default_numeric' mangled-name='vdev_prop_default_numeric' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_default_numeric'>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <return type-id='9c313c2d'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_string_to_index' mangled-name='vdev_prop_string_to_index' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_string_to_index'>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <parameter type-id='80f4b756' name='string'/>
	+ <parameter type-id='5d6479ae' name='index'/>
	+ <return type-id='95e97e5e'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_index_to_string' mangled-name='vdev_prop_index_to_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_index_to_string'>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <parameter type-id='9c313c2d' name='index'/>
	+ <parameter type-id='7d3cd834' name='string'/>
	+ <return type-id='95e97e5e'/>
	+ </function-decl>
	+ <function-decl name='zpool_prop_vdev' mangled-name='zpool_prop_vdev' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_vdev'>
	+ <parameter type-id='80f4b756' name='name'/>
	+ <return type-id='c19b74c3'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_random_value' mangled-name='vdev_prop_random_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_random_value'>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <parameter type-id='9c313c2d' name='seed'/>
	+ <return type-id='9c313c2d'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_values' mangled-name='vdev_prop_values' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_values'>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <return type-id='80f4b756'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_column_name' mangled-name='vdev_prop_column_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_column_name'>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <return type-id='80f4b756'/>
	+ </function-decl>
	+ <function-decl name='vdev_prop_align_right' mangled-name='vdev_prop_align_right' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_prop_align_right'>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <return type-id='c19b74c3'/>
	+ </function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='../../module/zcommon/zprop_common.c' language='LANG_C99'>
	<function-decl name='zprop_register_impl' mangled-name='zprop_register_impl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_register_impl'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='31429eff' name='type'/>
	<parameter type-id='9c313c2d' name='numdefault'/>
	<parameter type-id='80f4b756' name='strdefault'/>
	<parameter type-id='999701cc' name='attr'/>
	<parameter type-id='95e97e5e' name='objset_types'/>
	<parameter type-id='80f4b756' name='values'/>
	<parameter type-id='80f4b756' name='colname'/>
	<parameter type-id='c19b74c3' name='rightalign'/>
	<parameter type-id='c19b74c3' name='visible'/>
	<parameter type-id='c8bc397b' name='idx_tbl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zprop_register_string' mangled-name='zprop_register_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_register_string'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='80f4b756' name='def'/>
	<parameter type-id='999701cc' name='attr'/>
	<parameter type-id='95e97e5e' name='objset_types'/>
	<parameter type-id='80f4b756' name='values'/>
	<parameter type-id='80f4b756' name='colname'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zprop_register_number' mangled-name='zprop_register_number' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_register_number'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='9c313c2d' name='def'/>
	<parameter type-id='999701cc' name='attr'/>
	<parameter type-id='95e97e5e' name='objset_types'/>
	<parameter type-id='80f4b756' name='values'/>
	<parameter type-id='80f4b756' name='colname'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zprop_register_index' mangled-name='zprop_register_index' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_register_index'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='9c313c2d' name='def'/>
	<parameter type-id='999701cc' name='attr'/>
	<parameter type-id='95e97e5e' name='objset_types'/>
	<parameter type-id='80f4b756' name='values'/>
	<parameter type-id='80f4b756' name='colname'/>
	<parameter type-id='c8bc397b' name='idx_tbl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zprop_register_hidden' mangled-name='zprop_register_hidden' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_register_hidden'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='31429eff' name='type'/>
	<parameter type-id='999701cc' name='attr'/>
	<parameter type-id='95e97e5e' name='objset_types'/>
	<parameter type-id='80f4b756' name='colname'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zprop_iter_common' mangled-name='zprop_iter_common' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_iter_common'>
	<parameter type-id='1ec3747a' name='func'/>
	<parameter type-id='eaa32e2f' name='cb'/>
	<parameter type-id='c19b74c3' name='show_all'/>
	<parameter type-id='c19b74c3' name='ordered'/>
	<parameter type-id='2e45de5d' name='type'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zprop_name_to_prop' mangled-name='zprop_name_to_prop' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_name_to_prop'>
	<parameter type-id='80f4b756' name='propname'/>
	<parameter type-id='2e45de5d' name='type'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zprop_string_to_index' mangled-name='zprop_string_to_index' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_string_to_index'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='80f4b756' name='string'/>
	<parameter type-id='5d6479ae' name='index'/>
	<parameter type-id='2e45de5d' name='type'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zprop_index_to_string' mangled-name='zprop_index_to_string' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_index_to_string'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='9c313c2d' name='index'/>
	<parameter type-id='7d3cd834' name='string'/>
	<parameter type-id='2e45de5d' name='type'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zprop_random_value' mangled-name='zprop_random_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_random_value'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='9c313c2d' name='seed'/>
	<parameter type-id='2e45de5d' name='type'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='zprop_values' mangled-name='zprop_values' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_values'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='2e45de5d' name='type'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zprop_valid_for_type' mangled-name='zprop_valid_for_type' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_valid_for_type'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='2e45de5d' name='type'/>
	<parameter type-id='c19b74c3' name='headcheck'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	+ <function-decl name='zprop_valid_char' mangled-name='zprop_valid_char' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_valid_char'>
	+ <parameter type-id='a84c031d' name='c'/>
	+ <return type-id='95e97e5e'/>
	+ </function-decl>
	<function-decl name='zprop_width' mangled-name='zprop_width' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_width'>
	<parameter type-id='95e97e5e' name='prop'/>
	<parameter type-id='37e3bd22' name='fixed'/>
	<parameter type-id='2e45de5d' name='type'/>
	<return type-id='b59d7dce'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_changelist.c' language='LANG_C99'>
	<type-decl name='void' id='48b5725f'/>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_config.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='bf311473' size-in-bits='128' id='f0f65199'>
	<subrange length='2' type-id='7359adad' id='52efc4ef'/>
	</array-type-def>
	<type-decl name='char' size-in-bits='8' id='a84c031d'/>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='8192' id='b54ce520'>
	<subrange length='1024' type-id='7359adad' id='c60446f8'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='2048' id='d1617432'>
	<subrange length='256' type-id='7359adad' id='36e5b9fa'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='320' id='36c46961'>
	<subrange length='40' type-id='7359adad' id='8f80b239'/>
	</array-type-def>
	- <class-decl name='re_dfa_t' is-struct='yes' visibility='default' is-declaration-only='yes' id='b48d2441'/>
	- <class-decl name='uu_avl' is-struct='yes' visibility='default' is-declaration-only='yes' id='4af029d1'/>
	- <class-decl name='uu_avl_pool' is-struct='yes' visibility='default' is-declaration-only='yes' id='12a530a8'/>
	<type-decl name='int' size-in-bits='32' id='95e97e5e'/>
	<type-decl name='long int' size-in-bits='64' id='bd54fe1a'/>
	<type-decl name='long long int' size-in-bits='64' id='1eb56b1e'/>
	<type-decl name='short int' size-in-bits='16' id='a2185560'/>
	<type-decl name='unnamed-enum-underlying-type-32' is-anonymous='yes' size-in-bits='32' alignment-in-bits='32' id='9cac1fee'/>
	<type-decl name='unsigned char' size-in-bits='8' id='002ac4a6'/>
	<type-decl name='unsigned int' size-in-bits='32' id='f0981eeb'/>
	<type-decl name='unsigned long int' size-in-bits='64' id='7359adad'/>
	<typedef-decl name='uu_avl_pool_t' type-id='12a530a8' id='7f84e390'/>
	<typedef-decl name='uu_avl_t' type-id='4af029d1' id='bb7f0973'/>
	<typedef-decl name='zfs_handle_t' type-id='f6ee4445' id='775509eb'/>
	<typedef-decl name='zpool_handle_t' type-id='67002a8a' id='b1efc708'/>
	<typedef-decl name='libzfs_handle_t' type-id='c8a9d9d8' id='95942d0c'/>
	<typedef-decl name='zpool_iter_f' type-id='3aebb66f' id='fa476e62'/>
	<typedef-decl name='zfs_iter_f' type-id='5571cde4' id='d8e49ab9'/>
	<typedef-decl name='avl_tree_t' type-id='b351119f' id='f20fbd51'/>
	<class-decl name='avl_node' size-in-bits='192' is-struct='yes' visibility='default' id='428b67b3'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='avl_child' type-id='f0f65199' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='avl_pcb' type-id='e475ab95' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='avl_tree' size-in-bits='320' is-struct='yes' visibility='default' id='b351119f'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='avl_root' type-id='bf311473' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='avl_compar' type-id='585e1de9' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='avl_offset' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='avl_numnodes' type-id='ee1f298e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='avl_pad' type-id='b59d7dce' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='dmu_objset_stats' size-in-bits='2304' is-struct='yes' visibility='default' id='098f0221'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='dds_num_clones' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='dds_creation_txg' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='dds_guid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='dds_type' type-id='230f1e16' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='224'>
	<var-decl name='dds_is_snapshot' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='232'>
	<var-decl name='dds_inconsistent' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='240'>
	<var-decl name='dds_redacted' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='248'>
	<var-decl name='dds_origin' type-id='d1617432' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='dmu_objset_stats_t' type-id='098f0221' id='b2c14f17'/>
	<enum-decl name='zfs_type_t' naming-typedef-id='2e45de5d' id='5d8f7321'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZFS_TYPE_FILESYSTEM' value='1'/>
	<enumerator name='ZFS_TYPE_SNAPSHOT' value='2'/>
	<enumerator name='ZFS_TYPE_VOLUME' value='4'/>
	<enumerator name='ZFS_TYPE_POOL' value='8'/>
	<enumerator name='ZFS_TYPE_BOOKMARK' value='16'/>
	+ <enumerator name='ZFS_TYPE_VDEV' value='32'/>
	</enum-decl>
	<typedef-decl name='zfs_type_t' type-id='5d8f7321' id='2e45de5d'/>
	<enum-decl name='dmu_objset_type' id='6b1b19f9'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='DMU_OST_NONE' value='0'/>
	<enumerator name='DMU_OST_META' value='1'/>
	<enumerator name='DMU_OST_ZFS' value='2'/>
	<enumerator name='DMU_OST_ZVOL' value='3'/>
	<enumerator name='DMU_OST_OTHER' value='4'/>
	<enumerator name='DMU_OST_ANY' value='5'/>
	<enumerator name='DMU_OST_NUMTYPES' value='6'/>
	</enum-decl>
	<typedef-decl name='dmu_objset_type_t' type-id='6b1b19f9' id='230f1e16'/>
	<class-decl name='nvlist' size-in-bits='192' is-struct='yes' visibility='default' id='ac266fd9'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='nvl_version' type-id='3ff5601b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='nvl_nvflag' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='nvl_priv' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='nvl_flag' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='160'>
	<var-decl name='nvl_pad' type-id='3ff5601b' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='nvlist_t' type-id='ac266fd9' id='8e8d4be3'/>
	<enum-decl name='boolean_t' naming-typedef-id='c19b74c3' id='f58c8277'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='B_FALSE' value='0'/>
	<enumerator name='B_TRUE' value='1'/>
	</enum-decl>
	<typedef-decl name='boolean_t' type-id='f58c8277' id='c19b74c3'/>
	<typedef-decl name='ulong_t' type-id='7359adad' id='ee1f298e'/>
	<typedef-decl name='longlong_t' type-id='1eb56b1e' id='9b3ff54f'/>
	<typedef-decl name='diskaddr_t' type-id='9b3ff54f' id='804dc465'/>
	<class-decl name='libzfs_handle' size-in-bits='18240' is-struct='yes' visibility='default' id='c8a9d9d8'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='libzfs_error' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='libzfs_fd' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='libzfs_pool_handles' type-id='4c81de99' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='libzfs_ns_avlpool' type-id='de82c773' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='libzfs_ns_avl' type-id='a5c21a38' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='libzfs_ns_gen' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='libzfs_desc_active' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='352'>
	<var-decl name='libzfs_action' type-id='b54ce520' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='8544'>
	<var-decl name='libzfs_desc' type-id='b54ce520' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='16736'>
	<var-decl name='libzfs_printerr' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='16768'>
	<var-decl name='libzfs_mnttab_enable' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='16832'>
	<var-decl name='libzfs_mnttab_cache_lock' type-id='7a6844eb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='17152'>
	<var-decl name='libzfs_mnttab_cache' type-id='f20fbd51' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='17472'>
	<var-decl name='libzfs_pool_iter' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='17504'>
	<var-decl name='libzfs_prop_debug' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='17536'>
	<var-decl name='libzfs_urire' type-id='aca3bac8' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='18048'>
	<var-decl name='libzfs_max_nvlist' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='18112'>
	<var-decl name='libfetch' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='18176'>
	<var-decl name='libfetch_load_error' type-id='26a90f95' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='zfs_handle' size-in-bits='4928' is-struct='yes' visibility='default' id='f6ee4445'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='zfs_hdl' type-id='b0382bb3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='zpool_hdl' type-id='4c81de99' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='zfs_name' type-id='d1617432' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2176'>
	<var-decl name='zfs_type' type-id='2e45de5d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2208'>
	<var-decl name='zfs_head_type' type-id='2e45de5d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2240'>
	<var-decl name='zfs_dmustats' type-id='b2c14f17' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='4544'>
	<var-decl name='zfs_props' type-id='5ce45b60' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='4608'>
	<var-decl name='zfs_user_props' type-id='5ce45b60' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='4672'>
	<var-decl name='zfs_recvd_props' type-id='5ce45b60' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='4736'>
	<var-decl name='zfs_mntcheck' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='4800'>
	<var-decl name='zfs_mntopts' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='4864'>
	<var-decl name='zfs_props_table' type-id='ae3e8ca6' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='zpool_handle' size-in-bits='2560' is-struct='yes' visibility='default' id='67002a8a'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='zpool_hdl' type-id='b0382bb3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='zpool_next' type-id='4c81de99' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='zpool_name' type-id='d1617432' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2176'>
	<var-decl name='zpool_state' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2240'>
	<var-decl name='zpool_config_size' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2304'>
	<var-decl name='zpool_config' type-id='5ce45b60' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2368'>
	<var-decl name='zpool_old_config' type-id='5ce45b60' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2432'>
	<var-decl name='zpool_props' type-id='5ce45b60' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2496'>
	<var-decl name='zpool_start_block' type-id='804dc465' visibility='default'/>
	</data-member>
	</class-decl>
	- <typedef-decl name='__re_long_size_t' type-id='7359adad' id='ba516949'/>
	<typedef-decl name='reg_syntax_t' type-id='7359adad' id='1b72c3b3'/>
	<class-decl name='re_pattern_buffer' size-in-bits='512' is-struct='yes' visibility='default' id='19fc9a8c'>
	<data-member access='public' layout-offset-in-bits='0'>
	- <var-decl name='buffer' type-id='33976309' visibility='default'/>
	+ <var-decl name='buffer' type-id='cf536864' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	- <var-decl name='allocated' type-id='ba516949' visibility='default'/>
	+ <var-decl name='allocated' type-id='7359adad' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	- <var-decl name='used' type-id='ba516949' visibility='default'/>
	+ <var-decl name='used' type-id='7359adad' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='syntax' type-id='1b72c3b3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='fastmap' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='translate' type-id='cf536864' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='re_nsub' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='can_be_null' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='449'>
	<var-decl name='regs_allocated' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='451'>
	<var-decl name='fastmap_accurate' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='452'>
	<var-decl name='no_sub' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='453'>
	<var-decl name='not_bol' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='454'>
	<var-decl name='not_eol' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='455'>
	<var-decl name='newline_anchor' type-id='f0981eeb' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='regex_t' type-id='19fc9a8c' id='aca3bac8'/>
	<typedef-decl name='uintptr_t' type-id='7359adad' id='e475ab95'/>
	<union-decl name='pthread_mutex_t' size-in-bits='320' naming-typedef-id='7a6844eb' visibility='default' id='70681f9b'>
	<data-member access='public'>
	<var-decl name='__data' type-id='4c734837' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='__size' type-id='36c46961' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='__align' type-id='bd54fe1a' visibility='default'/>
	</data-member>
	</union-decl>
	<typedef-decl name='pthread_mutex_t' type-id='70681f9b' id='7a6844eb'/>
	<typedef-decl name='int32_t' type-id='33f57a65' id='3ff5601b'/>
	<typedef-decl name='uint8_t' type-id='c51d6389' id='b96825af'/>
	<typedef-decl name='uint32_t' type-id='62f1140c' id='8f92235e'/>
	<typedef-decl name='uint64_t' type-id='8910171f' id='9c313c2d'/>
	+ <class-decl name='__pthread_internal_list' size-in-bits='128' is-struct='yes' visibility='default' id='0e01899c'>
	+ <data-member access='public' layout-offset-in-bits='0'>
	+ <var-decl name='__prev' type-id='4d98cd5a' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='64'>
	+ <var-decl name='__next' type-id='4d98cd5a' visibility='default'/>
	+ </data-member>
	+ </class-decl>
	+ <typedef-decl name='__pthread_list_t' type-id='0e01899c' id='518fb49c'/>
	<class-decl name='__pthread_mutex_s' size-in-bits='320' is-struct='yes' visibility='default' id='4c734837'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='__lock' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='__count' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='__owner' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='96'>
	<var-decl name='__nusers' type-id='f0981eeb' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='__kind' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='160'>
	<var-decl name='__spins' type-id='a2185560' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='176'>
	<var-decl name='__elision' type-id='a2185560' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='__list' type-id='518fb49c' visibility='default'/>
	</data-member>
	</class-decl>
	- <class-decl name='__pthread_internal_list' size-in-bits='128' is-struct='yes' visibility='default' id='0e01899c'>
	- <data-member access='public' layout-offset-in-bits='0'>
	- <var-decl name='__prev' type-id='4d98cd5a' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='64'>
	- <var-decl name='__next' type-id='4d98cd5a' visibility='default'/>
	- </data-member>
	- </class-decl>
	- <typedef-decl name='__pthread_list_t' type-id='0e01899c' id='518fb49c'/>
	<typedef-decl name='__uint8_t' type-id='002ac4a6' id='c51d6389'/>
	<typedef-decl name='__int32_t' type-id='95e97e5e' id='33f57a65'/>
	<typedef-decl name='__uint32_t' type-id='f0981eeb' id='62f1140c'/>
	<typedef-decl name='__uint64_t' type-id='7359adad' id='8910171f'/>
	<typedef-decl name='size_t' type-id='7359adad' id='b59d7dce'/>
	<pointer-type-def type-id='0e01899c' size-in-bits='64' id='4d98cd5a'/>
	<pointer-type-def type-id='428b67b3' size-in-bits='64' id='bf311473'/>
	<pointer-type-def type-id='c19b74c3' size-in-bits='64' id='37e3bd22'/>
	<pointer-type-def type-id='a84c031d' size-in-bits='64' id='26a90f95'/>
	<qualified-type-def type-id='a84c031d' const='yes' id='9b45d938'/>
	<pointer-type-def type-id='9b45d938' size-in-bits='64' id='80f4b756'/>
	<pointer-type-def type-id='96ee24a5' size-in-bits='64' id='585e1de9'/>
	<pointer-type-def type-id='cb9628fa' size-in-bits='64' id='5571cde4'/>
	<pointer-type-def type-id='2bce87e3' size-in-bits='64' id='3aebb66f'/>
	<pointer-type-def type-id='95942d0c' size-in-bits='64' id='b0382bb3'/>
	<pointer-type-def type-id='8e8d4be3' size-in-bits='64' id='5ce45b60'/>
	<pointer-type-def type-id='5ce45b60' size-in-bits='64' id='857bb57e'/>
	- <pointer-type-def type-id='b48d2441' size-in-bits='64' id='33976309'/>
	<pointer-type-def type-id='b96825af' size-in-bits='64' id='ae3e8ca6'/>
	<pointer-type-def type-id='002ac4a6' size-in-bits='64' id='cf536864'/>
	<pointer-type-def type-id='7f84e390' size-in-bits='64' id='de82c773'/>
	<pointer-type-def type-id='bb7f0973' size-in-bits='64' id='a5c21a38'/>
	<pointer-type-def type-id='48b5725f' size-in-bits='64' id='eaa32e2f'/>
	<pointer-type-def type-id='775509eb' size-in-bits='64' id='9200a744'/>
	<pointer-type-def type-id='b1efc708' size-in-bits='64' id='4c81de99'/>
	- <class-decl name='re_dfa_t' is-struct='yes' visibility='default' is-declaration-only='yes' id='b48d2441'/>
	<class-decl name='uu_avl' is-struct='yes' visibility='default' is-declaration-only='yes' id='4af029d1'/>
	<class-decl name='uu_avl_pool' is-struct='yes' visibility='default' is-declaration-only='yes' id='12a530a8'/>
	<function-decl name='zpool_get_config' mangled-name='zpool_get_config' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_config'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='857bb57e' name='oldconfig'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='zpool_get_features' mangled-name='zpool_get_features' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_features'>
	<parameter type-id='4c81de99' name='zhp'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='zpool_refresh_stats' mangled-name='zpool_refresh_stats' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_refresh_stats'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='37e3bd22' name='missing'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_skip_pool' mangled-name='zpool_skip_pool' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_skip_pool'>
	<parameter type-id='80f4b756' name='poolname'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zpool_iter' mangled-name='zpool_iter' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_iter'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='fa476e62' name='func'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_iter_root' mangled-name='zfs_iter_root' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_iter_root'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='d8e49ab9' name='func'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-type size-in-bits='64' id='cb9628fa'>
	<parameter type-id='9200a744'/>
	<parameter type-id='eaa32e2f'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='2bce87e3'>
	<parameter type-id='4c81de99'/>
	<parameter type-id='eaa32e2f'/>
	<return type-id='95e97e5e'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_crypto.c' language='LANG_C99'>
	<typedef-decl name='uint_t' type-id='f0981eeb' id='3502e3ff'/>
	<pointer-type-def type-id='ae3e8ca6' size-in-bits='64' id='d8774064'/>
	<pointer-type-def type-id='3502e3ff' size-in-bits='64' id='4dd26a40'/>
	<function-decl name='zfs_crypto_get_encryption_root' mangled-name='zfs_crypto_get_encryption_root' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_crypto_get_encryption_root'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='37e3bd22' name='is_encroot'/>
	<parameter type-id='26a90f95' name='buf'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_crypto_create' mangled-name='zfs_crypto_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_crypto_create'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='26a90f95' name='parent_name'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='5ce45b60' name='pool_props'/>
	<parameter type-id='c19b74c3' name='stdin_available'/>
	<parameter type-id='d8774064' name='wkeydata_out'/>
	<parameter type-id='4dd26a40' name='wkeylen_out'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_crypto_clone_check' mangled-name='zfs_crypto_clone_check' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_crypto_clone_check'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='9200a744' name='origin_zhp'/>
	<parameter type-id='26a90f95' name='parent_name'/>
	<parameter type-id='5ce45b60' name='props'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_crypto_attempt_load_keys' mangled-name='zfs_crypto_attempt_load_keys' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_crypto_attempt_load_keys'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='26a90f95' name='fsname'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_crypto_load_key' mangled-name='zfs_crypto_load_key' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_crypto_load_key'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='c19b74c3' name='noop'/>
	<parameter type-id='26a90f95' name='alt_keylocation'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_crypto_unload_key' mangled-name='zfs_crypto_unload_key' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_crypto_unload_key'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_crypto_rewrap' mangled-name='zfs_crypto_rewrap' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_crypto_rewrap'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='5ce45b60' name='raw_props'/>
	<parameter type-id='c19b74c3' name='inheritkey'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_dataset.c' language='LANG_C99'>
	<class-decl name='zprop_list' size-in-bits='448' is-struct='yes' visibility='default' id='bd9b4291'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='pl_prop' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='pl_user_prop' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='pl_next' type-id='9f1a1109' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='pl_all' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='pl_width' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='pl_recvd_width' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='pl_fixed' type-id='c19b74c3' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zprop_list_t' type-id='bd9b4291' id='bdb8ac4f'/>
	<class-decl name='renameflags' size-in-bits='32' is-struct='yes' visibility='default' id='7aee5792'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='recursive' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1'>
	<var-decl name='nounmount' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2'>
	<var-decl name='forceunmount' type-id='95e97e5e' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='renameflags_t' type-id='7aee5792' id='067170c2'/>
	<typedef-decl name='zfs_userspace_cb_t' type-id='ca64ff60' id='16c5f410'/>
	<enum-decl name='zfs_prop_t' naming-typedef-id='58603c44' id='4b000d60'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZPROP_CONT' value='-2'/>
	<enumerator name='ZPROP_INVAL' value='-1'/>
	<enumerator name='ZFS_PROP_TYPE' value='0'/>
	<enumerator name='ZFS_PROP_CREATION' value='1'/>
	<enumerator name='ZFS_PROP_USED' value='2'/>
	<enumerator name='ZFS_PROP_AVAILABLE' value='3'/>
	<enumerator name='ZFS_PROP_REFERENCED' value='4'/>
	<enumerator name='ZFS_PROP_COMPRESSRATIO' value='5'/>
	<enumerator name='ZFS_PROP_MOUNTED' value='6'/>
	<enumerator name='ZFS_PROP_ORIGIN' value='7'/>
	<enumerator name='ZFS_PROP_QUOTA' value='8'/>
	<enumerator name='ZFS_PROP_RESERVATION' value='9'/>
	<enumerator name='ZFS_PROP_VOLSIZE' value='10'/>
	<enumerator name='ZFS_PROP_VOLBLOCKSIZE' value='11'/>
	<enumerator name='ZFS_PROP_RECORDSIZE' value='12'/>
	<enumerator name='ZFS_PROP_MOUNTPOINT' value='13'/>
	<enumerator name='ZFS_PROP_SHARENFS' value='14'/>
	<enumerator name='ZFS_PROP_CHECKSUM' value='15'/>
	<enumerator name='ZFS_PROP_COMPRESSION' value='16'/>
	<enumerator name='ZFS_PROP_ATIME' value='17'/>
	<enumerator name='ZFS_PROP_DEVICES' value='18'/>
	<enumerator name='ZFS_PROP_EXEC' value='19'/>
	<enumerator name='ZFS_PROP_SETUID' value='20'/>
	<enumerator name='ZFS_PROP_READONLY' value='21'/>
	<enumerator name='ZFS_PROP_ZONED' value='22'/>
	<enumerator name='ZFS_PROP_SNAPDIR' value='23'/>
	<enumerator name='ZFS_PROP_ACLMODE' value='24'/>
	<enumerator name='ZFS_PROP_ACLINHERIT' value='25'/>
	<enumerator name='ZFS_PROP_CREATETXG' value='26'/>
	<enumerator name='ZFS_PROP_NAME' value='27'/>
	<enumerator name='ZFS_PROP_CANMOUNT' value='28'/>
	<enumerator name='ZFS_PROP_ISCSIOPTIONS' value='29'/>
	<enumerator name='ZFS_PROP_XATTR' value='30'/>
	<enumerator name='ZFS_PROP_NUMCLONES' value='31'/>
	<enumerator name='ZFS_PROP_COPIES' value='32'/>
	<enumerator name='ZFS_PROP_VERSION' value='33'/>
	<enumerator name='ZFS_PROP_UTF8ONLY' value='34'/>
	<enumerator name='ZFS_PROP_NORMALIZE' value='35'/>
	<enumerator name='ZFS_PROP_CASE' value='36'/>
	<enumerator name='ZFS_PROP_VSCAN' value='37'/>
	<enumerator name='ZFS_PROP_NBMAND' value='38'/>
	<enumerator name='ZFS_PROP_SHARESMB' value='39'/>
	<enumerator name='ZFS_PROP_REFQUOTA' value='40'/>
	<enumerator name='ZFS_PROP_REFRESERVATION' value='41'/>
	<enumerator name='ZFS_PROP_GUID' value='42'/>
	<enumerator name='ZFS_PROP_PRIMARYCACHE' value='43'/>
	<enumerator name='ZFS_PROP_SECONDARYCACHE' value='44'/>
	<enumerator name='ZFS_PROP_USEDSNAP' value='45'/>
	<enumerator name='ZFS_PROP_USEDDS' value='46'/>
	<enumerator name='ZFS_PROP_USEDCHILD' value='47'/>
	<enumerator name='ZFS_PROP_USEDREFRESERV' value='48'/>
	<enumerator name='ZFS_PROP_USERACCOUNTING' value='49'/>
	<enumerator name='ZFS_PROP_STMF_SHAREINFO' value='50'/>
	<enumerator name='ZFS_PROP_DEFER_DESTROY' value='51'/>
	<enumerator name='ZFS_PROP_USERREFS' value='52'/>
	<enumerator name='ZFS_PROP_LOGBIAS' value='53'/>
	<enumerator name='ZFS_PROP_UNIQUE' value='54'/>
	<enumerator name='ZFS_PROP_OBJSETID' value='55'/>
	<enumerator name='ZFS_PROP_DEDUP' value='56'/>
	<enumerator name='ZFS_PROP_MLSLABEL' value='57'/>
	<enumerator name='ZFS_PROP_SYNC' value='58'/>
	<enumerator name='ZFS_PROP_DNODESIZE' value='59'/>
	<enumerator name='ZFS_PROP_REFRATIO' value='60'/>
	<enumerator name='ZFS_PROP_WRITTEN' value='61'/>
	<enumerator name='ZFS_PROP_CLONES' value='62'/>
	<enumerator name='ZFS_PROP_LOGICALUSED' value='63'/>
	<enumerator name='ZFS_PROP_LOGICALREFERENCED' value='64'/>
	<enumerator name='ZFS_PROP_INCONSISTENT' value='65'/>
	<enumerator name='ZFS_PROP_VOLMODE' value='66'/>
	<enumerator name='ZFS_PROP_FILESYSTEM_LIMIT' value='67'/>
	<enumerator name='ZFS_PROP_SNAPSHOT_LIMIT' value='68'/>
	<enumerator name='ZFS_PROP_FILESYSTEM_COUNT' value='69'/>
	<enumerator name='ZFS_PROP_SNAPSHOT_COUNT' value='70'/>
	<enumerator name='ZFS_PROP_SNAPDEV' value='71'/>
	<enumerator name='ZFS_PROP_ACLTYPE' value='72'/>
	<enumerator name='ZFS_PROP_SELINUX_CONTEXT' value='73'/>
	<enumerator name='ZFS_PROP_SELINUX_FSCONTEXT' value='74'/>
	<enumerator name='ZFS_PROP_SELINUX_DEFCONTEXT' value='75'/>
	<enumerator name='ZFS_PROP_SELINUX_ROOTCONTEXT' value='76'/>
	<enumerator name='ZFS_PROP_RELATIME' value='77'/>
	<enumerator name='ZFS_PROP_REDUNDANT_METADATA' value='78'/>
	<enumerator name='ZFS_PROP_OVERLAY' value='79'/>
	<enumerator name='ZFS_PROP_PREV_SNAP' value='80'/>
	<enumerator name='ZFS_PROP_RECEIVE_RESUME_TOKEN' value='81'/>
	<enumerator name='ZFS_PROP_ENCRYPTION' value='82'/>
	<enumerator name='ZFS_PROP_KEYLOCATION' value='83'/>
	<enumerator name='ZFS_PROP_KEYFORMAT' value='84'/>
	<enumerator name='ZFS_PROP_PBKDF2_SALT' value='85'/>
	<enumerator name='ZFS_PROP_PBKDF2_ITERS' value='86'/>
	<enumerator name='ZFS_PROP_ENCRYPTION_ROOT' value='87'/>
	<enumerator name='ZFS_PROP_KEY_GUID' value='88'/>
	<enumerator name='ZFS_PROP_KEYSTATUS' value='89'/>
	<enumerator name='ZFS_PROP_REMAPTXG' value='90'/>
	<enumerator name='ZFS_PROP_SPECIAL_SMALL_BLOCKS' value='91'/>
	<enumerator name='ZFS_PROP_IVSET_GUID' value='92'/>
	<enumerator name='ZFS_PROP_REDACTED' value='93'/>
	<enumerator name='ZFS_PROP_REDACT_SNAPS' value='94'/>
	<enumerator name='ZFS_NUM_PROPS' value='95'/>
	</enum-decl>
	<typedef-decl name='zfs_prop_t' type-id='4b000d60' id='58603c44'/>
	<enum-decl name='zfs_userquota_prop_t' naming-typedef-id='279fde6a' id='5258d2f6'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZFS_PROP_USERUSED' value='0'/>
	<enumerator name='ZFS_PROP_USERQUOTA' value='1'/>
	<enumerator name='ZFS_PROP_GROUPUSED' value='2'/>
	<enumerator name='ZFS_PROP_GROUPQUOTA' value='3'/>
	<enumerator name='ZFS_PROP_USEROBJUSED' value='4'/>
	<enumerator name='ZFS_PROP_USEROBJQUOTA' value='5'/>
	<enumerator name='ZFS_PROP_GROUPOBJUSED' value='6'/>
	<enumerator name='ZFS_PROP_GROUPOBJQUOTA' value='7'/>
	<enumerator name='ZFS_PROP_PROJECTUSED' value='8'/>
	<enumerator name='ZFS_PROP_PROJECTQUOTA' value='9'/>
	<enumerator name='ZFS_PROP_PROJECTOBJUSED' value='10'/>
	<enumerator name='ZFS_PROP_PROJECTOBJQUOTA' value='11'/>
	<enumerator name='ZFS_NUM_USERQUOTA_PROPS' value='12'/>
	</enum-decl>
	<typedef-decl name='zfs_userquota_prop_t' type-id='5258d2f6' id='279fde6a'/>
	<enum-decl name='zprop_source_t' naming-typedef-id='a2256d42' id='5903f80e'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZPROP_SRC_NONE' value='1'/>
	<enumerator name='ZPROP_SRC_DEFAULT' value='2'/>
	<enumerator name='ZPROP_SRC_TEMPORARY' value='4'/>
	<enumerator name='ZPROP_SRC_LOCAL' value='8'/>
	<enumerator name='ZPROP_SRC_INHERITED' value='16'/>
	<enumerator name='ZPROP_SRC_RECEIVED' value='32'/>
	</enum-decl>
	<typedef-decl name='zprop_source_t' type-id='5903f80e' id='a2256d42'/>
	<enum-decl name='zfs_wait_activity_t' naming-typedef-id='3024501a' id='527d5dc6'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZFS_WAIT_DELETEQ' value='0'/>
	<enumerator name='ZFS_WAIT_NUM_ACTIVITIES' value='1'/>
	</enum-decl>
	<typedef-decl name='zfs_wait_activity_t' type-id='527d5dc6' id='3024501a'/>
	<class-decl name='mnttab' size-in-bits='256' is-struct='yes' visibility='default' id='1b055409'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='mnt_special' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='mnt_mountp' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='mnt_fstype' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='mnt_mntopts' type-id='26a90f95' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='__uid_t' type-id='f0981eeb' id='cc5fcceb'/>
	<typedef-decl name='uid_t' type-id='cc5fcceb' id='354978ed'/>
	<pointer-type-def type-id='26a90f95' size-in-bits='64' id='9b23c9ad'/>
	<qualified-type-def type-id='775509eb' const='yes' id='5eadf2db'/>
	<pointer-type-def type-id='5eadf2db' size-in-bits='64' id='fcd57163'/>
	<pointer-type-def type-id='7e291ce6' size-in-bits='64' id='ca64ff60'/>
	<pointer-type-def type-id='95e97e5e' size-in-bits='64' id='7292109c'/>
	<pointer-type-def type-id='1b055409' size-in-bits='64' id='9d424d31'/>
	<pointer-type-def type-id='9c313c2d' size-in-bits='64' id='5d6479ae'/>
	<pointer-type-def type-id='bd9b4291' size-in-bits='64' id='9f1a1109'/>
	<pointer-type-def type-id='bdb8ac4f' size-in-bits='64' id='3a9b2288'/>
	<pointer-type-def type-id='3a9b2288' size-in-bits='64' id='e4378506'/>
	<pointer-type-def type-id='a2256d42' size-in-bits='64' id='debc6aa3'/>
	<function-decl name='zfs_type_to_name' mangled-name='zfs_type_to_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_type_to_name'>
	<parameter type-id='2e45de5d' name='type'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zfs_name_valid' mangled-name='zfs_name_valid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_name_valid'>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='2e45de5d' name='type'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_free_handles' mangled-name='zpool_free_handles' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_free_handles'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_refresh_properties' mangled-name='zfs_refresh_properties' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_refresh_properties'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_handle_dup' mangled-name='zfs_handle_dup' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_handle_dup'>
	<parameter type-id='9200a744' name='zhp_orig'/>
	<return type-id='9200a744'/>
	</function-decl>
	<function-decl name='zfs_bookmark_exists' mangled-name='zfs_bookmark_exists' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_bookmark_exists'>
	<parameter type-id='80f4b756' name='path'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_open' mangled-name='zfs_open' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_open'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='95e97e5e' name='types'/>
	<return type-id='9200a744'/>
	</function-decl>
	<function-decl name='zfs_close' mangled-name='zfs_close' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_close'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='libzfs_mnttab_init' mangled-name='libzfs_mnttab_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_mnttab_init'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='libzfs_mnttab_fini' mangled-name='libzfs_mnttab_fini' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_mnttab_fini'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='libzfs_mnttab_cache' mangled-name='libzfs_mnttab_cache' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_mnttab_cache'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='c19b74c3' name='enable'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='libzfs_mnttab_find' mangled-name='libzfs_mnttab_find' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_mnttab_find'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='fsname'/>
	<parameter type-id='9d424d31' name='entry'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='libzfs_mnttab_add' mangled-name='libzfs_mnttab_add' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_mnttab_add'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='special'/>
	<parameter type-id='80f4b756' name='mountp'/>
	<parameter type-id='80f4b756' name='mntopts'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='libzfs_mnttab_remove' mangled-name='libzfs_mnttab_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_mnttab_remove'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='fsname'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_spa_version' mangled-name='zfs_spa_version' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_spa_version'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='7292109c' name='spa_version'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_valid_proplist' mangled-name='zfs_valid_proplist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_valid_proplist'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='2e45de5d' name='type'/>
	<parameter type-id='5ce45b60' name='nvl'/>
	<parameter type-id='9c313c2d' name='zoned'/>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='4c81de99' name='zpool_hdl'/>
	<parameter type-id='c19b74c3' name='key_params_ok'/>
	<parameter type-id='80f4b756' name='errbuf'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='zfs_prop_set' mangled-name='zfs_prop_set' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_set'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='propname'/>
	<parameter type-id='80f4b756' name='propval'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_prop_set_list' mangled-name='zfs_prop_set_list' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_set_list'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='5ce45b60' name='props'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_prop_inherit' mangled-name='zfs_prop_inherit' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_inherit'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='propname'/>
	<parameter type-id='c19b74c3' name='received'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='getprop_uint64' mangled-name='getprop_uint64' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getprop_uint64'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='58603c44' name='prop'/>
	<parameter type-id='9b23c9ad' name='source'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='zfs_prop_get_recvd' mangled-name='zfs_prop_get_recvd' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_get_recvd'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='propname'/>
	<parameter type-id='26a90f95' name='propbuf'/>
	<parameter type-id='b59d7dce' name='proplen'/>
	<parameter type-id='c19b74c3' name='literal'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_get_clones_nvl' mangled-name='zfs_get_clones_nvl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_clones_nvl'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='zfs_prop_get' mangled-name='zfs_prop_get' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_get'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='58603c44' name='prop'/>
	<parameter type-id='26a90f95' name='propbuf'/>
	<parameter type-id='b59d7dce' name='proplen'/>
	<parameter type-id='debc6aa3' name='src'/>
	<parameter type-id='26a90f95' name='statbuf'/>
	<parameter type-id='b59d7dce' name='statlen'/>
	<parameter type-id='c19b74c3' name='literal'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_prop_get_int' mangled-name='zfs_prop_get_int' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_get_int'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='58603c44' name='prop'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='zfs_prop_get_numeric' mangled-name='zfs_prop_get_numeric' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_get_numeric'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='58603c44' name='prop'/>
	<parameter type-id='5d6479ae' name='value'/>
	<parameter type-id='debc6aa3' name='src'/>
	<parameter type-id='26a90f95' name='statbuf'/>
	<parameter type-id='b59d7dce' name='statlen'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_prop_get_userquota_int' mangled-name='zfs_prop_get_userquota_int' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_get_userquota_int'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='propname'/>
	<parameter type-id='5d6479ae' name='propvalue'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_prop_get_userquota' mangled-name='zfs_prop_get_userquota' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_get_userquota'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='propname'/>
	<parameter type-id='26a90f95' name='propbuf'/>
	<parameter type-id='95e97e5e' name='proplen'/>
	<parameter type-id='c19b74c3' name='literal'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_prop_get_written_int' mangled-name='zfs_prop_get_written_int' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_get_written_int'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='propname'/>
	<parameter type-id='5d6479ae' name='propvalue'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_prop_get_written' mangled-name='zfs_prop_get_written' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prop_get_written'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='propname'/>
	<parameter type-id='26a90f95' name='propbuf'/>
	<parameter type-id='95e97e5e' name='proplen'/>
	<parameter type-id='c19b74c3' name='literal'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_get_name' mangled-name='zfs_get_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_name'>
	<parameter type-id='fcd57163' name='zhp'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zfs_get_pool_name' mangled-name='zfs_get_pool_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_pool_name'>
	<parameter type-id='fcd57163' name='zhp'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zfs_get_type' mangled-name='zfs_get_type' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_type'>
	<parameter type-id='fcd57163' name='zhp'/>
	<return type-id='2e45de5d'/>
	</function-decl>
	<function-decl name='zfs_get_underlying_type' mangled-name='zfs_get_underlying_type' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_underlying_type'>
	<parameter type-id='fcd57163' name='zhp'/>
	<return type-id='2e45de5d'/>
	</function-decl>
	<function-decl name='zfs_parent_name' mangled-name='zfs_parent_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_parent_name'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='b59d7dce' name='buflen'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_dataset_exists' mangled-name='zfs_dataset_exists' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_dataset_exists'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='2e45de5d' name='types'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_create_ancestors' mangled-name='zfs_create_ancestors' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_create_ancestors'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='path'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_create' mangled-name='zfs_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_create'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='2e45de5d' name='type'/>
	<parameter type-id='5ce45b60' name='props'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_destroy' mangled-name='zfs_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_destroy'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='c19b74c3' name='defer'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_destroy_snaps' mangled-name='zfs_destroy_snaps' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_destroy_snaps'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='26a90f95' name='snapname'/>
	<parameter type-id='c19b74c3' name='defer'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_destroy_snaps_nvl' mangled-name='zfs_destroy_snaps_nvl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_destroy_snaps_nvl'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='5ce45b60' name='snaps'/>
	<parameter type-id='c19b74c3' name='defer'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_clone' mangled-name='zfs_clone' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_clone'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='target'/>
	<parameter type-id='5ce45b60' name='props'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_promote' mangled-name='zfs_promote' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_promote'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_snapshot_nvl' mangled-name='zfs_snapshot_nvl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_snapshot_nvl'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='5ce45b60' name='snaps'/>
	<parameter type-id='5ce45b60' name='props'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_snapshot' mangled-name='zfs_snapshot' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_snapshot'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='c19b74c3' name='recursive'/>
	<parameter type-id='5ce45b60' name='props'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_rollback' mangled-name='zfs_rollback' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_rollback'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='9200a744' name='snap'/>
	<parameter type-id='c19b74c3' name='force'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_rename' mangled-name='zfs_rename' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_rename'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='target'/>
	<parameter type-id='067170c2' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_get_all_props' mangled-name='zfs_get_all_props' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_all_props'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='zfs_get_recvd_props' mangled-name='zfs_get_recvd_props' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_recvd_props'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='zfs_get_user_props' mangled-name='zfs_get_user_props' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_user_props'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='zfs_expand_proplist' mangled-name='zfs_expand_proplist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_expand_proplist'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='e4378506' name='plp'/>
	<parameter type-id='c19b74c3' name='received'/>
	<parameter type-id='c19b74c3' name='literal'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_prune_proplist' mangled-name='zfs_prune_proplist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_prune_proplist'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='ae3e8ca6' name='props'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_smb_acl_add' mangled-name='zfs_smb_acl_add' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_smb_acl_add'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='26a90f95' name='dataset'/>
	<parameter type-id='26a90f95' name='path'/>
	<parameter type-id='26a90f95' name='resource'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_smb_acl_remove' mangled-name='zfs_smb_acl_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_smb_acl_remove'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='26a90f95' name='dataset'/>
	<parameter type-id='26a90f95' name='path'/>
	<parameter type-id='26a90f95' name='resource'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_smb_acl_purge' mangled-name='zfs_smb_acl_purge' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_smb_acl_purge'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='26a90f95' name='dataset'/>
	<parameter type-id='26a90f95' name='path'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_smb_acl_rename' mangled-name='zfs_smb_acl_rename' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_smb_acl_rename'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='26a90f95' name='dataset'/>
	<parameter type-id='26a90f95' name='path'/>
	<parameter type-id='26a90f95' name='oldname'/>
	<parameter type-id='26a90f95' name='newname'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_userspace' mangled-name='zfs_userspace' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_userspace'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='279fde6a' name='type'/>
	<parameter type-id='16c5f410' name='func'/>
	<parameter type-id='eaa32e2f' name='arg'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_hold' mangled-name='zfs_hold' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_hold'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='80f4b756' name='tag'/>
	<parameter type-id='c19b74c3' name='recursive'/>
	<parameter type-id='95e97e5e' name='cleanup_fd'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_hold_nvl' mangled-name='zfs_hold_nvl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_hold_nvl'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='95e97e5e' name='cleanup_fd'/>
	<parameter type-id='5ce45b60' name='holds'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_release' mangled-name='zfs_release' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_release'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='80f4b756' name='tag'/>
	<parameter type-id='c19b74c3' name='recursive'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_get_fsacl' mangled-name='zfs_get_fsacl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_fsacl'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='857bb57e' name='nvl'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_set_fsacl' mangled-name='zfs_set_fsacl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_set_fsacl'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='c19b74c3' name='un'/>
	<parameter type-id='5ce45b60' name='nvl'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_get_holds' mangled-name='zfs_get_holds' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_holds'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='857bb57e' name='nvl'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zvol_volsize_to_reservation' mangled-name='zvol_volsize_to_reservation' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zvol_volsize_to_reservation'>
	<parameter type-id='4c81de99' name='zph'/>
	<parameter type-id='9c313c2d' name='volsize'/>
	<parameter type-id='5ce45b60' name='props'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='zfs_wait_status' mangled-name='zfs_wait_status' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_wait_status'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='3024501a' name='activity'/>
	<parameter type-id='37e3bd22' name='missing'/>
	<parameter type-id='37e3bd22' name='waited'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-type size-in-bits='64' id='7e291ce6'>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='354978ed'/>
	<parameter type-id='9c313c2d'/>
	<return type-id='95e97e5e'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_diff.c' language='LANG_C99'>
	<function-decl name='zfs_show_diffs' mangled-name='zfs_show_diffs' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_show_diffs'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='95e97e5e' name='outfd'/>
	<parameter type-id='80f4b756' name='fromsnap'/>
	<parameter type-id='80f4b756' name='tosnap'/>
	<parameter type-id='95e97e5e' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_import.c' language='LANG_C99'>
	<typedef-decl name='refresh_config_func_t' type-id='29f040d2' id='b7c58eaa'/>
	<typedef-decl name='pool_active_func_t' type-id='baa42fef' id='de5d1d8f'/>
	<class-decl name='pool_config_ops' size-in-bits='128' is-struct='yes' visibility='default' id='8b092c69'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='pco_refresh_config' type-id='e7c00489' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='pco_pool_active' type-id='9eadf5e0' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='pool_config_ops_t' type-id='1a21babe' id='b1e62775'/>
	<enum-decl name='pool_state' id='4871ac24'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='POOL_STATE_ACTIVE' value='0'/>
	<enumerator name='POOL_STATE_EXPORTED' value='1'/>
	<enumerator name='POOL_STATE_DESTROYED' value='2'/>
	<enumerator name='POOL_STATE_SPARE' value='3'/>
	<enumerator name='POOL_STATE_L2CACHE' value='4'/>
	<enumerator name='POOL_STATE_UNINITIALIZED' value='5'/>
	<enumerator name='POOL_STATE_UNAVAIL' value='6'/>
	<enumerator name='POOL_STATE_POTENTIALLY_ACTIVE' value='7'/>
	</enum-decl>
	<typedef-decl name='pool_state_t' type-id='4871ac24' id='084a08a3'/>
	<qualified-type-def type-id='8b092c69' const='yes' id='1a21babe'/>
	<pointer-type-def type-id='de5d1d8f' size-in-bits='64' id='9eadf5e0'/>
	<pointer-type-def type-id='084a08a3' size-in-bits='64' id='b9ea57b8'/>
	<pointer-type-def type-id='b7c58eaa' size-in-bits='64' id='e7c00489'/>
	<var-decl name='libzfs_config_ops' type-id='b1e62775' mangled-name='libzfs_config_ops' visibility='default' elf-symbol-id='libzfs_config_ops'/>
	<function-decl name='zpool_clear_label' mangled-name='zpool_clear_label' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_clear_label'>
	<parameter type-id='95e97e5e' name='fd'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_in_use' mangled-name='zpool_in_use' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_in_use'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='b9ea57b8' name='state'/>
	<parameter type-id='9b23c9ad' name='namestr'/>
	<parameter type-id='37e3bd22' name='inuse'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-type size-in-bits='64' id='baa42fef'>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='80f4b756'/>
	<parameter type-id='9c313c2d'/>
	<parameter type-id='37e3bd22'/>
	<return type-id='95e97e5e'/>
	</function-type>
	<function-type size-in-bits='64' id='29f040d2'>
	<parameter type-id='eaa32e2f'/>
	<parameter type-id='5ce45b60'/>
	<return type-id='5ce45b60'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_iter.c' language='LANG_C99'>
	<function-decl name='zfs_iter_filesystems' mangled-name='zfs_iter_filesystems' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_iter_filesystems'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='d8e49ab9' name='func'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_iter_snapshots' mangled-name='zfs_iter_snapshots' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_iter_snapshots'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='c19b74c3' name='simple'/>
	<parameter type-id='d8e49ab9' name='func'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<parameter type-id='9c313c2d' name='min_txg'/>
	<parameter type-id='9c313c2d' name='max_txg'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_iter_bookmarks' mangled-name='zfs_iter_bookmarks' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_iter_bookmarks'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='d8e49ab9' name='func'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_iter_snapshots_sorted' mangled-name='zfs_iter_snapshots_sorted' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_iter_snapshots_sorted'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='d8e49ab9' name='callback'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<parameter type-id='9c313c2d' name='min_txg'/>
	<parameter type-id='9c313c2d' name='max_txg'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_iter_snapspec' mangled-name='zfs_iter_snapspec' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_iter_snapspec'>
	<parameter type-id='9200a744' name='fs_zhp'/>
	<parameter type-id='80f4b756' name='spec_orig'/>
	<parameter type-id='d8e49ab9' name='func'/>
	<parameter type-id='eaa32e2f' name='arg'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_iter_children' mangled-name='zfs_iter_children' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_iter_children'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='d8e49ab9' name='func'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_iter_dependents' mangled-name='zfs_iter_dependents' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_iter_dependents'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='c19b74c3' name='allowrecursion'/>
	<parameter type-id='d8e49ab9' name='func'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_iter_mounted' mangled-name='zfs_iter_mounted' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_iter_mounted'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='d8e49ab9' name='func'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_mount.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='f1bd64e2' size-in-bits='384' id='b2c36c9f'>
	<subrange length='2' type-id='7359adad' id='52efc4ef'/>
	</array-type-def>
	<class-decl name='get_all_cb' size-in-bits='192' is-struct='yes' visibility='default' id='803dac95'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='cb_handles' type-id='4507922a' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='cb_alloc' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='cb_used' type-id='b59d7dce' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='get_all_cb_t' type-id='803dac95' id='9b293607'/>
	<class-decl name='proto_table_t' size-in-bits='192' is-struct='yes' naming-typedef-id='f1bd64e2' visibility='default' id='f4c8e1ed'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='p_prop' type-id='58603c44' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='p_name' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='p_share_err' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='160'>
	<var-decl name='p_unshare_err' type-id='95e97e5e' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='proto_table_t' type-id='f4c8e1ed' id='f1bd64e2'/>
	<pointer-type-def type-id='9b293607' size-in-bits='64' id='77bf1784'/>
	<pointer-type-def type-id='9200a744' size-in-bits='64' id='4507922a'/>
	<var-decl name='proto_table' type-id='b2c36c9f' visibility='default'/>
	<function-decl name='is_mounted' mangled-name='is_mounted' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='is_mounted'>
	<parameter type-id='b0382bb3' name='zfs_hdl'/>
	<parameter type-id='80f4b756' name='special'/>
	<parameter type-id='9b23c9ad' name='where'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_is_mounted' mangled-name='zfs_is_mounted' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_is_mounted'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='9b23c9ad' name='where'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_mount' mangled-name='zfs_mount' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_mount'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='options'/>
	<parameter type-id='95e97e5e' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_mount_at' mangled-name='zfs_mount_at' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_mount_at'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='options'/>
	<parameter type-id='95e97e5e' name='flags'/>
	<parameter type-id='80f4b756' name='mountpoint'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_unmount' mangled-name='zfs_unmount' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_unmount'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='mountpoint'/>
	<parameter type-id='95e97e5e' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_unmountall' mangled-name='zfs_unmountall' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_unmountall'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='95e97e5e' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_is_shared' mangled-name='zfs_is_shared' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_is_shared'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_share' mangled-name='zfs_share' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_share'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_unshare' mangled-name='zfs_unshare' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_unshare'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_is_shared_nfs' mangled-name='zfs_is_shared_nfs' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_is_shared_nfs'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='9b23c9ad' name='where'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_is_shared_smb' mangled-name='zfs_is_shared_smb' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_is_shared_smb'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='9b23c9ad' name='where'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_commit_nfs_shares' mangled-name='zfs_commit_nfs_shares' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_commit_nfs_shares'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_commit_smb_shares' mangled-name='zfs_commit_smb_shares' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_commit_smb_shares'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_commit_all_shares' mangled-name='zfs_commit_all_shares' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_commit_all_shares'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_commit_shares' mangled-name='zfs_commit_shares' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_commit_shares'>
	<parameter type-id='80f4b756' name='proto'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_share_nfs' mangled-name='zfs_share_nfs' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_share_nfs'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_share_smb' mangled-name='zfs_share_smb' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_share_smb'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	+ <function-decl name='zfs_shareall' mangled-name='zfs_shareall' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_shareall'>
	+ <parameter type-id='9200a744' name='zhp'/>
	+ <return type-id='95e97e5e'/>
	+ </function-decl>
	<function-decl name='zfs_unshare_nfs' mangled-name='zfs_unshare_nfs' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_unshare_nfs'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='mountpoint'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_unshare_smb' mangled-name='zfs_unshare_smb' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_unshare_smb'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='mountpoint'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_unshareall_nfs' mangled-name='zfs_unshareall_nfs' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_unshareall_nfs'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_unshareall_smb' mangled-name='zfs_unshareall_smb' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_unshareall_smb'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_unshareall' mangled-name='zfs_unshareall' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_unshareall'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_unshareall_bypath' mangled-name='zfs_unshareall_bypath' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_unshareall_bypath'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='mountpoint'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_unshareall_bytype' mangled-name='zfs_unshareall_bytype' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_unshareall_bytype'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='mountpoint'/>
	<parameter type-id='80f4b756' name='proto'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='libzfs_add_handle' mangled-name='libzfs_add_handle' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_add_handle'>
	<parameter type-id='77bf1784' name='cbp'/>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_foreach_mountpoint' mangled-name='zfs_foreach_mountpoint' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_foreach_mountpoint'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='4507922a' name='handles'/>
	<parameter type-id='b59d7dce' name='num_handles'/>
	<parameter type-id='d8e49ab9' name='func'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<parameter type-id='c19b74c3' name='parallel'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zpool_enable_datasets' mangled-name='zpool_enable_datasets' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_enable_datasets'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='mntopts'/>
	<parameter type-id='95e97e5e' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_disable_datasets' mangled-name='zpool_disable_datasets' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_disable_datasets'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='c19b74c3' name='force'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_pool.c' language='LANG_C99'>
	<class-decl name='splitflags' size-in-bits='64' is-struct='yes' visibility='default' id='dc01bf52'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='dryrun' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1'>
	<var-decl name='import' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='name_flags' type-id='95e97e5e' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='splitflags_t' type-id='dc01bf52' id='325c1e34'/>
	<class-decl name='trimflags' size-in-bits='192' is-struct='yes' visibility='default' id='8ef58008'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='fullpool' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='secure' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='wait' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='rate' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='trimflags_t' type-id='8ef58008' id='a093cbb8'/>
	<enum-decl name='zpool_compat_status_t' naming-typedef-id='901b78d1' id='20676925'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZPOOL_COMPATIBILITY_OK' value='0'/>
	<enumerator name='ZPOOL_COMPATIBILITY_WARNTOKEN' value='1'/>
	<enumerator name='ZPOOL_COMPATIBILITY_BADTOKEN' value='2'/>
	<enumerator name='ZPOOL_COMPATIBILITY_BADFILE' value='3'/>
	<enumerator name='ZPOOL_COMPATIBILITY_NOFILES' value='4'/>
	</enum-decl>
	<typedef-decl name='zpool_compat_status_t' type-id='20676925' id='901b78d1'/>
	<enum-decl name='zpool_prop_t' naming-typedef-id='5d0c23fb' id='af1ba157'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZPOOL_PROP_INVAL' value='-1'/>
	<enumerator name='ZPOOL_PROP_NAME' value='0'/>
	<enumerator name='ZPOOL_PROP_SIZE' value='1'/>
	<enumerator name='ZPOOL_PROP_CAPACITY' value='2'/>
	<enumerator name='ZPOOL_PROP_ALTROOT' value='3'/>
	<enumerator name='ZPOOL_PROP_HEALTH' value='4'/>
	<enumerator name='ZPOOL_PROP_GUID' value='5'/>
	<enumerator name='ZPOOL_PROP_VERSION' value='6'/>
	<enumerator name='ZPOOL_PROP_BOOTFS' value='7'/>
	<enumerator name='ZPOOL_PROP_DELEGATION' value='8'/>
	<enumerator name='ZPOOL_PROP_AUTOREPLACE' value='9'/>
	<enumerator name='ZPOOL_PROP_CACHEFILE' value='10'/>
	<enumerator name='ZPOOL_PROP_FAILUREMODE' value='11'/>
	<enumerator name='ZPOOL_PROP_LISTSNAPS' value='12'/>
	<enumerator name='ZPOOL_PROP_AUTOEXPAND' value='13'/>
	<enumerator name='ZPOOL_PROP_DEDUPDITTO' value='14'/>
	<enumerator name='ZPOOL_PROP_DEDUPRATIO' value='15'/>
	<enumerator name='ZPOOL_PROP_FREE' value='16'/>
	<enumerator name='ZPOOL_PROP_ALLOCATED' value='17'/>
	<enumerator name='ZPOOL_PROP_READONLY' value='18'/>
	<enumerator name='ZPOOL_PROP_ASHIFT' value='19'/>
	<enumerator name='ZPOOL_PROP_COMMENT' value='20'/>
	<enumerator name='ZPOOL_PROP_EXPANDSZ' value='21'/>
	<enumerator name='ZPOOL_PROP_FREEING' value='22'/>
	<enumerator name='ZPOOL_PROP_FRAGMENTATION' value='23'/>
	<enumerator name='ZPOOL_PROP_LEAKED' value='24'/>
	<enumerator name='ZPOOL_PROP_MAXBLOCKSIZE' value='25'/>
	<enumerator name='ZPOOL_PROP_TNAME' value='26'/>
	<enumerator name='ZPOOL_PROP_MAXDNODESIZE' value='27'/>
	<enumerator name='ZPOOL_PROP_MULTIHOST' value='28'/>
	<enumerator name='ZPOOL_PROP_CHECKPOINT' value='29'/>
	<enumerator name='ZPOOL_PROP_LOAD_GUID' value='30'/>
	<enumerator name='ZPOOL_PROP_AUTOTRIM' value='31'/>
	<enumerator name='ZPOOL_PROP_COMPATIBILITY' value='32'/>
	<enumerator name='ZPOOL_NUM_PROPS' value='33'/>
	</enum-decl>
	<typedef-decl name='zpool_prop_t' type-id='af1ba157' id='5d0c23fb'/>
	+ <enum-decl name='vdev_prop_t' naming-typedef-id='5aa5c90c' id='1573bec8'>
	+ <underlying-type type-id='9cac1fee'/>
	+ <enumerator name='VDEV_PROP_INVAL' value='-1'/>
	+ <enumerator name='VDEV_PROP_NAME' value='0'/>
	+ <enumerator name='VDEV_PROP_CAPACITY' value='1'/>
	+ <enumerator name='VDEV_PROP_STATE' value='2'/>
	+ <enumerator name='VDEV_PROP_GUID' value='3'/>
	+ <enumerator name='VDEV_PROP_ASIZE' value='4'/>
	+ <enumerator name='VDEV_PROP_PSIZE' value='5'/>
	+ <enumerator name='VDEV_PROP_ASHIFT' value='6'/>
	+ <enumerator name='VDEV_PROP_SIZE' value='7'/>
	+ <enumerator name='VDEV_PROP_FREE' value='8'/>
	+ <enumerator name='VDEV_PROP_ALLOCATED' value='9'/>
	+ <enumerator name='VDEV_PROP_COMMENT' value='10'/>
	+ <enumerator name='VDEV_PROP_EXPANDSZ' value='11'/>
	+ <enumerator name='VDEV_PROP_FRAGMENTATION' value='12'/>
	+ <enumerator name='VDEV_PROP_BOOTSIZE' value='13'/>
	+ <enumerator name='VDEV_PROP_PARITY' value='14'/>
	+ <enumerator name='VDEV_PROP_PATH' value='15'/>
	+ <enumerator name='VDEV_PROP_DEVID' value='16'/>
	+ <enumerator name='VDEV_PROP_PHYS_PATH' value='17'/>
	+ <enumerator name='VDEV_PROP_ENC_PATH' value='18'/>
	+ <enumerator name='VDEV_PROP_FRU' value='19'/>
	+ <enumerator name='VDEV_PROP_PARENT' value='20'/>
	+ <enumerator name='VDEV_PROP_CHILDREN' value='21'/>
	+ <enumerator name='VDEV_PROP_NUMCHILDREN' value='22'/>
	+ <enumerator name='VDEV_PROP_READ_ERRORS' value='23'/>
	+ <enumerator name='VDEV_PROP_WRITE_ERRORS' value='24'/>
	+ <enumerator name='VDEV_PROP_CHECKSUM_ERRORS' value='25'/>
	+ <enumerator name='VDEV_PROP_INITIALIZE_ERRORS' value='26'/>
	+ <enumerator name='VDEV_PROP_OPS_NULL' value='27'/>
	+ <enumerator name='VDEV_PROP_OPS_READ' value='28'/>
	+ <enumerator name='VDEV_PROP_OPS_WRITE' value='29'/>
	+ <enumerator name='VDEV_PROP_OPS_FREE' value='30'/>
	+ <enumerator name='VDEV_PROP_OPS_CLAIM' value='31'/>
	+ <enumerator name='VDEV_PROP_OPS_TRIM' value='32'/>
	+ <enumerator name='VDEV_PROP_BYTES_NULL' value='33'/>
	+ <enumerator name='VDEV_PROP_BYTES_READ' value='34'/>
	+ <enumerator name='VDEV_PROP_BYTES_WRITE' value='35'/>
	+ <enumerator name='VDEV_PROP_BYTES_FREE' value='36'/>
	+ <enumerator name='VDEV_PROP_BYTES_CLAIM' value='37'/>
	+ <enumerator name='VDEV_PROP_BYTES_TRIM' value='38'/>
	+ <enumerator name='VDEV_PROP_REMOVING' value='39'/>
	+ <enumerator name='VDEV_PROP_ALLOCATING' value='40'/>
	+ <enumerator name='VDEV_NUM_PROPS' value='41'/>
	+ </enum-decl>
	+ <typedef-decl name='vdev_prop_t' type-id='1573bec8' id='5aa5c90c'/>
	<enum-decl name='vdev_state' id='21566197'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='VDEV_STATE_UNKNOWN' value='0'/>
	<enumerator name='VDEV_STATE_CLOSED' value='1'/>
	<enumerator name='VDEV_STATE_OFFLINE' value='2'/>
	<enumerator name='VDEV_STATE_REMOVED' value='3'/>
	<enumerator name='VDEV_STATE_CANT_OPEN' value='4'/>
	<enumerator name='VDEV_STATE_FAULTED' value='5'/>
	<enumerator name='VDEV_STATE_DEGRADED' value='6'/>
	<enumerator name='VDEV_STATE_HEALTHY' value='7'/>
	</enum-decl>
	<typedef-decl name='vdev_state_t' type-id='21566197' id='35acf840'/>
	<enum-decl name='vdev_aux' id='7f5bcca4'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='VDEV_AUX_NONE' value='0'/>
	<enumerator name='VDEV_AUX_OPEN_FAILED' value='1'/>
	<enumerator name='VDEV_AUX_CORRUPT_DATA' value='2'/>
	<enumerator name='VDEV_AUX_NO_REPLICAS' value='3'/>
	<enumerator name='VDEV_AUX_BAD_GUID_SUM' value='4'/>
	<enumerator name='VDEV_AUX_TOO_SMALL' value='5'/>
	<enumerator name='VDEV_AUX_BAD_LABEL' value='6'/>
	<enumerator name='VDEV_AUX_VERSION_NEWER' value='7'/>
	<enumerator name='VDEV_AUX_VERSION_OLDER' value='8'/>
	<enumerator name='VDEV_AUX_UNSUP_FEAT' value='9'/>
	<enumerator name='VDEV_AUX_SPARED' value='10'/>
	<enumerator name='VDEV_AUX_ERR_EXCEEDED' value='11'/>
	<enumerator name='VDEV_AUX_IO_FAILURE' value='12'/>
	<enumerator name='VDEV_AUX_BAD_LOG' value='13'/>
	<enumerator name='VDEV_AUX_EXTERNAL' value='14'/>
	<enumerator name='VDEV_AUX_SPLIT_POOL' value='15'/>
	<enumerator name='VDEV_AUX_BAD_ASHIFT' value='16'/>
	<enumerator name='VDEV_AUX_EXTERNAL_PERSIST' value='17'/>
	<enumerator name='VDEV_AUX_ACTIVE' value='18'/>
	<enumerator name='VDEV_AUX_CHILDREN_OFFLINE' value='19'/>
	<enumerator name='VDEV_AUX_ASHIFT_TOO_BIG' value='20'/>
	</enum-decl>
	<typedef-decl name='vdev_aux_t' type-id='7f5bcca4' id='9d774e0b'/>
	<enum-decl name='pool_scan_func' id='1b092565'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='POOL_SCAN_NONE' value='0'/>
	<enumerator name='POOL_SCAN_SCRUB' value='1'/>
	<enumerator name='POOL_SCAN_RESILVER' value='2'/>
	<enumerator name='POOL_SCAN_FUNCS' value='3'/>
	</enum-decl>
	<typedef-decl name='pool_scan_func_t' type-id='1b092565' id='7313fbe2'/>
	<enum-decl name='pool_scrub_cmd' id='a1474cbd'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='POOL_SCRUB_NORMAL' value='0'/>
	<enumerator name='POOL_SCRUB_PAUSE' value='1'/>
	<enumerator name='POOL_SCRUB_FLAGS_END' value='2'/>
	</enum-decl>
	<typedef-decl name='pool_scrub_cmd_t' type-id='a1474cbd' id='b51cf3c2'/>
	<enum-decl name='pool_initialize_func' id='5c246ad4'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='POOL_INITIALIZE_START' value='0'/>
	<enumerator name='POOL_INITIALIZE_CANCEL' value='1'/>
	<enumerator name='POOL_INITIALIZE_SUSPEND' value='2'/>
	<enumerator name='POOL_INITIALIZE_FUNCS' value='3'/>
	</enum-decl>
	<typedef-decl name='pool_initialize_func_t' type-id='5c246ad4' id='7063e1ab'/>
	<enum-decl name='pool_trim_func' id='54ed608a'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='POOL_TRIM_START' value='0'/>
	<enumerator name='POOL_TRIM_CANCEL' value='1'/>
	<enumerator name='POOL_TRIM_SUSPEND' value='2'/>
	<enumerator name='POOL_TRIM_FUNCS' value='3'/>
	</enum-decl>
	<typedef-decl name='pool_trim_func_t' type-id='54ed608a' id='b1146b8d'/>
	<enum-decl name='zpool_wait_activity_t' naming-typedef-id='73446457' id='849338e3'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZPOOL_WAIT_CKPT_DISCARD' value='0'/>
	<enumerator name='ZPOOL_WAIT_FREE' value='1'/>
	<enumerator name='ZPOOL_WAIT_INITIALIZE' value='2'/>
	<enumerator name='ZPOOL_WAIT_REPLACE' value='3'/>
	<enumerator name='ZPOOL_WAIT_REMOVE' value='4'/>
	<enumerator name='ZPOOL_WAIT_RESILVER' value='5'/>
	<enumerator name='ZPOOL_WAIT_SCRUB' value='6'/>
	<enumerator name='ZPOOL_WAIT_TRIM' value='7'/>
	<enumerator name='ZPOOL_WAIT_NUM_ACTIVITIES' value='8'/>
	</enum-decl>
	<typedef-decl name='zpool_wait_activity_t' type-id='849338e3' id='73446457'/>
	<qualified-type-def type-id='8e8d4be3' const='yes' id='693c3853'/>
	<pointer-type-def type-id='693c3853' size-in-bits='64' id='22cce67b'/>
	<pointer-type-def type-id='a093cbb8' size-in-bits='64' id='b13f38c3'/>
	<pointer-type-def type-id='35acf840' size-in-bits='64' id='17f3480d'/>
	<function-decl name='zpool_props_refresh' mangled-name='zpool_props_refresh' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_props_refresh'>
	<parameter type-id='4c81de99' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_get_prop_int' mangled-name='zpool_get_prop_int' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_prop_int'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='5d0c23fb' name='prop'/>
	<parameter type-id='debc6aa3' name='src'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='zpool_state_to_name' mangled-name='zpool_state_to_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_state_to_name'>
	<parameter type-id='35acf840' name='state'/>
	<parameter type-id='9d774e0b' name='aux'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zpool_pool_state_to_name' mangled-name='zpool_pool_state_to_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_pool_state_to_name'>
	<parameter type-id='084a08a3' name='state'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zpool_get_state_str' mangled-name='zpool_get_state_str' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_state_str'>
	<parameter type-id='4c81de99' name='zhp'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zpool_get_prop' mangled-name='zpool_get_prop' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_prop'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='5d0c23fb' name='prop'/>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='b59d7dce' name='len'/>
	<parameter type-id='debc6aa3' name='srctype'/>
	<parameter type-id='c19b74c3' name='literal'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_set_prop' mangled-name='zpool_set_prop' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_set_prop'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='propname'/>
	<parameter type-id='80f4b756' name='propval'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_expand_proplist' mangled-name='zpool_expand_proplist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_expand_proplist'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='e4378506' name='plp'/>
	+ <parameter type-id='2e45de5d' name='type'/>
	<parameter type-id='c19b74c3' name='literal'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	+ <function-decl name='vdev_expand_proplist' mangled-name='vdev_expand_proplist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='vdev_expand_proplist'>
	+ <parameter type-id='4c81de99' name='zhp'/>
	+ <parameter type-id='80f4b756' name='vdevname'/>
	+ <parameter type-id='e4378506' name='plp'/>
	+ <return type-id='95e97e5e'/>
	+ </function-decl>
	<function-decl name='zpool_prop_get_feature' mangled-name='zpool_prop_get_feature' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_prop_get_feature'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='propname'/>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='b59d7dce' name='len'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_open_canfail' mangled-name='zpool_open_canfail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_open_canfail'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='pool'/>
	<return type-id='4c81de99'/>
	</function-decl>
	<function-decl name='zpool_open' mangled-name='zpool_open' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_open'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='pool'/>
	<return type-id='4c81de99'/>
	</function-decl>
	<function-decl name='zpool_close' mangled-name='zpool_close' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_close'>
	<parameter type-id='4c81de99' name='zhp'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zpool_get_name' mangled-name='zpool_get_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_name'>
	<parameter type-id='4c81de99' name='zhp'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zpool_get_state' mangled-name='zpool_get_state' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_state'>
	<parameter type-id='4c81de99' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_is_draid_spare' mangled-name='zpool_is_draid_spare' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_is_draid_spare'>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zpool_create' mangled-name='zpool_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_create'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='5ce45b60' name='nvroot'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='5ce45b60' name='fsprops'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_destroy' mangled-name='zpool_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_destroy'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='log_str'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_checkpoint' mangled-name='zpool_checkpoint' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_checkpoint'>
	<parameter type-id='4c81de99' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_discard_checkpoint' mangled-name='zpool_discard_checkpoint' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_discard_checkpoint'>
	<parameter type-id='4c81de99' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_add' mangled-name='zpool_add' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_add'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='5ce45b60' name='nvroot'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_export' mangled-name='zpool_export' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_export'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='c19b74c3' name='force'/>
	<parameter type-id='80f4b756' name='log_str'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_export_force' mangled-name='zpool_export_force' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_export_force'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='log_str'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_explain_recover' mangled-name='zpool_explain_recover' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_explain_recover'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='95e97e5e' name='reason'/>
	<parameter type-id='5ce45b60' name='config'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zpool_import' mangled-name='zpool_import' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_import'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='5ce45b60' name='config'/>
	<parameter type-id='80f4b756' name='newname'/>
	<parameter type-id='26a90f95' name='altroot'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_print_unsup_feat' mangled-name='zpool_print_unsup_feat' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_print_unsup_feat'>
	<parameter type-id='5ce45b60' name='config'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zpool_import_props' mangled-name='zpool_import_props' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_import_props'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='5ce45b60' name='config'/>
	<parameter type-id='80f4b756' name='newname'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='95e97e5e' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_initialize' mangled-name='zpool_initialize' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_initialize'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='7063e1ab' name='cmd_type'/>
	<parameter type-id='5ce45b60' name='vds'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_initialize_wait' mangled-name='zpool_initialize_wait' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_initialize_wait'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='7063e1ab' name='cmd_type'/>
	<parameter type-id='5ce45b60' name='vds'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_trim' mangled-name='zpool_trim' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_trim'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='b1146b8d' name='cmd_type'/>
	<parameter type-id='5ce45b60' name='vds'/>
	<parameter type-id='b13f38c3' name='trim_flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_scan' mangled-name='zpool_scan' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_scan'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='7313fbe2' name='func'/>
	<parameter type-id='b51cf3c2' name='cmd'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_find_vdev_by_physpath' mangled-name='zpool_find_vdev_by_physpath' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_find_vdev_by_physpath'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='ppath'/>
	<parameter type-id='37e3bd22' name='avail_spare'/>
	<parameter type-id='37e3bd22' name='l2cache'/>
	<parameter type-id='37e3bd22' name='log'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='zpool_find_vdev' mangled-name='zpool_find_vdev' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_find_vdev'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='37e3bd22' name='avail_spare'/>
	<parameter type-id='37e3bd22' name='l2cache'/>
	<parameter type-id='37e3bd22' name='log'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='zpool_get_physpath' mangled-name='zpool_get_physpath' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_physpath'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='26a90f95' name='physpath'/>
	<parameter type-id='b59d7dce' name='phypath_size'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_path_to_guid' mangled-name='zpool_vdev_path_to_guid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_path_to_guid'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='path'/>
	<return type-id='9c313c2d'/>
	</function-decl>
	<function-decl name='zpool_vdev_online' mangled-name='zpool_vdev_online' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_online'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='95e97e5e' name='flags'/>
	<parameter type-id='17f3480d' name='newstate'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_offline' mangled-name='zpool_vdev_offline' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_offline'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='c19b74c3' name='istmp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_fault' mangled-name='zpool_vdev_fault' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_fault'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='9c313c2d' name='guid'/>
	<parameter type-id='9d774e0b' name='aux'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_degrade' mangled-name='zpool_vdev_degrade' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_degrade'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='9c313c2d' name='guid'/>
	<parameter type-id='9d774e0b' name='aux'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_attach' mangled-name='zpool_vdev_attach' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_attach'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='old_disk'/>
	<parameter type-id='80f4b756' name='new_disk'/>
	<parameter type-id='5ce45b60' name='nvroot'/>
	<parameter type-id='95e97e5e' name='replacing'/>
	<parameter type-id='c19b74c3' name='rebuild'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_detach' mangled-name='zpool_vdev_detach' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_detach'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='path'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_split' mangled-name='zpool_vdev_split' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_split'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='26a90f95' name='newname'/>
	<parameter type-id='857bb57e' name='newroot'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='325c1e34' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_remove' mangled-name='zpool_vdev_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_remove'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='path'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_remove_cancel' mangled-name='zpool_vdev_remove_cancel' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_remove_cancel'>
	<parameter type-id='4c81de99' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_indirect_size' mangled-name='zpool_vdev_indirect_size' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_indirect_size'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='5d6479ae' name='sizep'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_clear' mangled-name='zpool_clear' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_clear'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='5ce45b60' name='rewindnvl'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_clear' mangled-name='zpool_vdev_clear' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_clear'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='9c313c2d' name='guid'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_reguid' mangled-name='zpool_reguid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_reguid'>
	<parameter type-id='4c81de99' name='zhp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_reopen_one' mangled-name='zpool_reopen_one' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_reopen_one'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_sync_one' mangled-name='zpool_sync_one' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_sync_one'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='eaa32e2f' name='data'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_vdev_name' mangled-name='zpool_vdev_name' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_vdev_name'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='5ce45b60' name='nv'/>
	<parameter type-id='95e97e5e' name='name_flags'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='zpool_get_errlog' mangled-name='zpool_get_errlog' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_errlog'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='857bb57e' name='nverrlistp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_upgrade' mangled-name='zpool_upgrade' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_upgrade'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='9c313c2d' name='new_version'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_save_arguments' mangled-name='zfs_save_arguments' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_save_arguments'>
	<parameter type-id='95e97e5e' name='argc'/>
	<parameter type-id='9b23c9ad' name='argv'/>
	<parameter type-id='26a90f95' name='string'/>
	<parameter type-id='95e97e5e' name='len'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zpool_log_history' mangled-name='zpool_log_history' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_log_history'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='message'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_get_history' mangled-name='zpool_get_history' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_history'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='857bb57e' name='nvhisp'/>
	<parameter type-id='5d6479ae' name='off'/>
	<parameter type-id='37e3bd22' name='eof'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_events_next' mangled-name='zpool_events_next' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_events_next'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='857bb57e' name='nvp'/>
	<parameter type-id='7292109c' name='dropped'/>
	<parameter type-id='f0981eeb' name='flags'/>
	<parameter type-id='95e97e5e' name='zevent_fd'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_events_clear' mangled-name='zpool_events_clear' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_events_clear'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='7292109c' name='count'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_events_seek' mangled-name='zpool_events_seek' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_events_seek'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='9c313c2d' name='eid'/>
	<parameter type-id='95e97e5e' name='zevent_fd'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_obj_to_path' mangled-name='zpool_obj_to_path' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_obj_to_path'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='9c313c2d' name='dsobj'/>
	<parameter type-id='9c313c2d' name='obj'/>
	<parameter type-id='26a90f95' name='pathname'/>
	<parameter type-id='b59d7dce' name='len'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zpool_obj_to_path_ds' mangled-name='zpool_obj_to_path_ds' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_obj_to_path_ds'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='9c313c2d' name='dsobj'/>
	<parameter type-id='9c313c2d' name='obj'/>
	<parameter type-id='26a90f95' name='pathname'/>
	<parameter type-id='b59d7dce' name='len'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zpool_wait' mangled-name='zpool_wait' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_wait'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='73446457' name='activity'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_wait_status' mangled-name='zpool_wait_status' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_wait_status'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='73446457' name='activity'/>
	<parameter type-id='37e3bd22' name='missing'/>
	<parameter type-id='37e3bd22' name='waited'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_set_bootenv' mangled-name='zpool_set_bootenv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_set_bootenv'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='22cce67b' name='envmap'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_get_bootenv' mangled-name='zpool_get_bootenv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_bootenv'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='857bb57e' name='nvlp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_load_compat' mangled-name='zpool_load_compat' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_load_compat'>
	<parameter type-id='80f4b756' name='compat'/>
	<parameter type-id='37e3bd22' name='features'/>
	<parameter type-id='26a90f95' name='report'/>
	<parameter type-id='b59d7dce' name='rlen'/>
	<return type-id='901b78d1'/>
	</function-decl>
	+ <function-decl name='zpool_get_vdev_prop_value' mangled-name='zpool_get_vdev_prop_value' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_vdev_prop_value'>
	+ <parameter type-id='5ce45b60' name='nvprop'/>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <parameter type-id='26a90f95' name='prop_name'/>
	+ <parameter type-id='26a90f95' name='buf'/>
	+ <parameter type-id='b59d7dce' name='len'/>
	+ <parameter type-id='debc6aa3' name='srctype'/>
	+ <parameter type-id='c19b74c3' name='literal'/>
	+ <return type-id='95e97e5e'/>
	+ </function-decl>
	+ <function-decl name='zpool_get_vdev_prop' mangled-name='zpool_get_vdev_prop' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_vdev_prop'>
	+ <parameter type-id='4c81de99' name='zhp'/>
	+ <parameter type-id='80f4b756' name='vdevname'/>
	+ <parameter type-id='5aa5c90c' name='prop'/>
	+ <parameter type-id='26a90f95' name='prop_name'/>
	+ <parameter type-id='26a90f95' name='buf'/>
	+ <parameter type-id='b59d7dce' name='len'/>
	+ <parameter type-id='debc6aa3' name='srctype'/>
	+ <parameter type-id='c19b74c3' name='literal'/>
	+ <return type-id='95e97e5e'/>
	+ </function-decl>
	+ <function-decl name='zpool_get_all_vdev_props' mangled-name='zpool_get_all_vdev_props' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_all_vdev_props'>
	+ <parameter type-id='4c81de99' name='zhp'/>
	+ <parameter type-id='80f4b756' name='vdevname'/>
	+ <parameter type-id='857bb57e' name='outnvl'/>
	+ <return type-id='95e97e5e'/>
	+ </function-decl>
	+ <function-decl name='zpool_set_vdev_prop' mangled-name='zpool_set_vdev_prop' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_set_vdev_prop'>
	+ <parameter type-id='4c81de99' name='zhp'/>
	+ <parameter type-id='80f4b756' name='vdevname'/>
	+ <parameter type-id='80f4b756' name='propname'/>
	+ <parameter type-id='80f4b756' name='propval'/>
	+ <return type-id='95e97e5e'/>
	+ </function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_sendrecv.c' language='LANG_C99'>
	<class-decl name='sendflags' size-in-bits='544' is-struct='yes' visibility='default' id='f6aa15be'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='verbosity' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='replicate' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='skipmissing' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='96'>
	<var-decl name='doall' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='fromorigin' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='160'>
	<var-decl name='pad' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='props' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='224'>
	<var-decl name='dryrun' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='parsable' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='288'>
	<var-decl name='progress' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='largeblock' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='352'>
	<var-decl name='embed_data' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='compress' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='416'>
	<var-decl name='raw' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='backup' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='480'>
	<var-decl name='holds' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='saved' type-id='c19b74c3' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='sendflags_t' type-id='f6aa15be' id='945467e6'/>
	<typedef-decl name='snapfilter_cb_t' type-id='d2a5e211' id='3d3ffb69'/>
	<class-decl name='recvflags' size-in-bits='416' is-struct='yes' visibility='default' id='34a384dc'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='verbose' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='isprefix' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='istail' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='96'>
	<var-decl name='dryrun' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='force' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='160'>
	<var-decl name='canmountoff' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='resumable' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='224'>
	<var-decl name='byteswap' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='nomount' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='288'>
	<var-decl name='holds' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='skipholds' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='352'>
	<var-decl name='domount' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='forceunmount' type-id='c19b74c3' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='recvflags_t' type-id='34a384dc' id='9e59d1d4'/>
	<pointer-type-def type-id='f20fbd51' size-in-bits='64' id='a3681dea'/>
	<pointer-type-def type-id='9e59d1d4' size-in-bits='64' id='4ea84b4f'/>
	<pointer-type-def type-id='945467e6' size-in-bits='64' id='8def7735'/>
	<pointer-type-def type-id='3d3ffb69' size-in-bits='64' id='72a26210'/>
	<function-decl name='zfs_send_progress' mangled-name='zfs_send_progress' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_send_progress'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='5d6479ae' name='bytes_written'/>
	<parameter type-id='5d6479ae' name='blocks_visited'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_send_resume_token_to_nvlist' mangled-name='zfs_send_resume_token_to_nvlist' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_send_resume_token_to_nvlist'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='token'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='zfs_send_resume' mangled-name='zfs_send_resume' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_send_resume'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='8def7735' name='flags'/>
	<parameter type-id='95e97e5e' name='outfd'/>
	<parameter type-id='80f4b756' name='resume_token'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_send_saved' mangled-name='zfs_send_saved' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_send_saved'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='8def7735' name='flags'/>
	<parameter type-id='95e97e5e' name='outfd'/>
	<parameter type-id='80f4b756' name='resume_token'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_send' mangled-name='zfs_send' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_send'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='fromsnap'/>
	<parameter type-id='80f4b756' name='tosnap'/>
	<parameter type-id='8def7735' name='flags'/>
	<parameter type-id='95e97e5e' name='outfd'/>
	<parameter type-id='72a26210' name='filter_func'/>
	<parameter type-id='eaa32e2f' name='cb_arg'/>
	<parameter type-id='857bb57e' name='debugnvp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_send_one' mangled-name='zfs_send_one' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_send_one'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='from'/>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='8def7735' name='flags'/>
	<parameter type-id='80f4b756' name='redactbook'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_receive' mangled-name='zfs_receive' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_receive'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='tosnap'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='4ea84b4f' name='flags'/>
	<parameter type-id='95e97e5e' name='infd'/>
	<parameter type-id='a3681dea' name='stream_avl'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-type size-in-bits='64' id='d2a5e211'>
	<parameter type-id='9200a744'/>
	<parameter type-id='eaa32e2f'/>
	<return type-id='c19b74c3'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_status.c' language='LANG_C99'>
	<enum-decl name='zpool_status_t' naming-typedef-id='d3dd6294' id='5e770b40'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZPOOL_STATUS_CORRUPT_CACHE' value='0'/>
	<enumerator name='ZPOOL_STATUS_MISSING_DEV_R' value='1'/>
	<enumerator name='ZPOOL_STATUS_MISSING_DEV_NR' value='2'/>
	<enumerator name='ZPOOL_STATUS_CORRUPT_LABEL_R' value='3'/>
	<enumerator name='ZPOOL_STATUS_CORRUPT_LABEL_NR' value='4'/>
	<enumerator name='ZPOOL_STATUS_BAD_GUID_SUM' value='5'/>
	<enumerator name='ZPOOL_STATUS_CORRUPT_POOL' value='6'/>
	<enumerator name='ZPOOL_STATUS_CORRUPT_DATA' value='7'/>
	<enumerator name='ZPOOL_STATUS_FAILING_DEV' value='8'/>
	<enumerator name='ZPOOL_STATUS_VERSION_NEWER' value='9'/>
	<enumerator name='ZPOOL_STATUS_HOSTID_MISMATCH' value='10'/>
	<enumerator name='ZPOOL_STATUS_HOSTID_ACTIVE' value='11'/>
	<enumerator name='ZPOOL_STATUS_HOSTID_REQUIRED' value='12'/>
	<enumerator name='ZPOOL_STATUS_IO_FAILURE_WAIT' value='13'/>
	<enumerator name='ZPOOL_STATUS_IO_FAILURE_CONTINUE' value='14'/>
	<enumerator name='ZPOOL_STATUS_IO_FAILURE_MMP' value='15'/>
	<enumerator name='ZPOOL_STATUS_BAD_LOG' value='16'/>
	<enumerator name='ZPOOL_STATUS_ERRATA' value='17'/>
	<enumerator name='ZPOOL_STATUS_UNSUP_FEAT_READ' value='18'/>
	<enumerator name='ZPOOL_STATUS_UNSUP_FEAT_WRITE' value='19'/>
	<enumerator name='ZPOOL_STATUS_FAULTED_DEV_R' value='20'/>
	<enumerator name='ZPOOL_STATUS_FAULTED_DEV_NR' value='21'/>
	<enumerator name='ZPOOL_STATUS_VERSION_OLDER' value='22'/>
	<enumerator name='ZPOOL_STATUS_FEAT_DISABLED' value='23'/>
	<enumerator name='ZPOOL_STATUS_RESILVERING' value='24'/>
	<enumerator name='ZPOOL_STATUS_OFFLINE_DEV' value='25'/>
	<enumerator name='ZPOOL_STATUS_REMOVED_DEV' value='26'/>
	<enumerator name='ZPOOL_STATUS_REBUILDING' value='27'/>
	<enumerator name='ZPOOL_STATUS_REBUILD_SCRUB' value='28'/>
	<enumerator name='ZPOOL_STATUS_NON_NATIVE_ASHIFT' value='29'/>
	<enumerator name='ZPOOL_STATUS_COMPATIBILITY_ERR' value='30'/>
	<enumerator name='ZPOOL_STATUS_INCOMPATIBLE_FEAT' value='31'/>
	<enumerator name='ZPOOL_STATUS_OK' value='32'/>
	</enum-decl>
	<typedef-decl name='zpool_status_t' type-id='5e770b40' id='d3dd6294'/>
	<enum-decl name='zpool_errata' id='d9abbf54'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZPOOL_ERRATA_NONE' value='0'/>
	<enumerator name='ZPOOL_ERRATA_ZOL_2094_SCRUB' value='1'/>
	<enumerator name='ZPOOL_ERRATA_ZOL_2094_ASYNC_DESTROY' value='2'/>
	<enumerator name='ZPOOL_ERRATA_ZOL_6845_ENCRYPTION' value='3'/>
	<enumerator name='ZPOOL_ERRATA_ZOL_8308_ENCRYPTION' value='4'/>
	</enum-decl>
	<typedef-decl name='zpool_errata_t' type-id='d9abbf54' id='688c495b'/>
	<pointer-type-def type-id='688c495b' size-in-bits='64' id='cec6f2e4'/>
	<function-decl name='zpool_get_status' mangled-name='zpool_get_status' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_status'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='9b23c9ad' name='msgid'/>
	<parameter type-id='cec6f2e4' name='errata'/>
	<return type-id='d3dd6294'/>
	</function-decl>
	<function-decl name='zpool_import_status' mangled-name='zpool_import_status' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_import_status'>
	<parameter type-id='5ce45b60' name='config'/>
	<parameter type-id='9b23c9ad' name='msgid'/>
	<parameter type-id='cec6f2e4' name='errata'/>
	<return type-id='d3dd6294'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_util.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='95e97e5e' size-in-bits='192' id='e41bdf22'>
	<subrange length='6' type-id='7359adad' id='52fa524b'/>
	</array-type-def>
	<type-decl name='variadic parameter type' id='2c1145c5'/>
	<array-type-def dimensions='1' type-id='19cefcee' size-in-bits='160' alignment-in-bits='32' id='3fcf57d2'>
	<subrange length='5' type-id='7359adad' id='53010e10'/>
	</array-type-def>
	<enum-decl name='zfs_get_column_t' naming-typedef-id='19cefcee' id='223bdcaa'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='GET_COL_NONE' value='0'/>
	<enumerator name='GET_COL_NAME' value='1'/>
	<enumerator name='GET_COL_PROPERTY' value='2'/>
	<enumerator name='GET_COL_VALUE' value='3'/>
	<enumerator name='GET_COL_RECVD' value='4'/>
	<enumerator name='GET_COL_SOURCE' value='5'/>
	</enum-decl>
	<typedef-decl name='zfs_get_column_t' type-id='223bdcaa' id='19cefcee'/>
	- <class-decl name='zprop_get_cbdata' size-in-bits='640' is-struct='yes' visibility='default' id='f3d3c319'>
	+ <class-decl name='vdev_cbdata' size-in-bits='192' is-struct='yes' visibility='default' id='b8006be8'>
	+ <data-member access='public' layout-offset-in-bits='0'>
	+ <var-decl name='cb_name_flags' type-id='95e97e5e' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='64'>
	+ <var-decl name='cb_names' type-id='9b23c9ad' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='128'>
	+ <var-decl name='cb_names_count' type-id='f0981eeb' visibility='default'/>
	+ </data-member>
	+ </class-decl>
	+ <typedef-decl name='vdev_cbdata_t' type-id='b8006be8' id='a9679c94'/>
	+ <class-decl name='zprop_get_cbdata' size-in-bits='832' is-struct='yes' visibility='default' id='f3d3c319'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='cb_sources' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='cb_columns' type-id='3fcf57d2' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='cb_colwidths' type-id='e41bdf22' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='cb_scripted' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='416'>
	<var-decl name='cb_literal' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='cb_first' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='cb_proplist' type-id='3a9b2288' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='576'>
	<var-decl name='cb_type' type-id='2e45de5d' visibility='default'/>
	</data-member>
	+ <data-member access='public' layout-offset-in-bits='640'>
	+ <var-decl name='cb_vdevs' type-id='a9679c94' visibility='default'/>
	+ </data-member>
	</class-decl>
	<typedef-decl name='zprop_get_cbdata_t' type-id='f3d3c319' id='f3d87113'/>
	<typedef-decl name='zprop_func' type-id='2e711a2a' id='1ec3747a'/>
	<pointer-type-def type-id='9b23c9ad' size-in-bits='64' id='c0563f85'/>
	<pointer-type-def type-id='c70fa2e8' size-in-bits='64' id='2e711a2a'/>
	<pointer-type-def type-id='f3d87113' size-in-bits='64' id='0d2a0670'/>
	<function-decl name='libzfs_errno' mangled-name='libzfs_errno' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_errno'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='libzfs_error_action' mangled-name='libzfs_error_action' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_error_action'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='libzfs_error_description' mangled-name='libzfs_error_description' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_error_description'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zfs_standard_error' mangled-name='zfs_standard_error' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_standard_error'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='95e97e5e' name='error'/>
	<parameter type-id='80f4b756' name='msg'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='libzfs_print_on_error' mangled-name='libzfs_print_on_error' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_print_on_error'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='c19b74c3' name='printerr'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='libzfs_run_process' mangled-name='libzfs_run_process' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_run_process'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='9b23c9ad' name='argv'/>
	<parameter type-id='95e97e5e' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='libzfs_run_process_get_stdout' mangled-name='libzfs_run_process_get_stdout' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_run_process_get_stdout'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='9b23c9ad' name='argv'/>
	<parameter type-id='9b23c9ad' name='env'/>
	<parameter type-id='c0563f85' name='lines'/>
	<parameter type-id='7292109c' name='lines_cnt'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='libzfs_run_process_get_stdout_nopath' mangled-name='libzfs_run_process_get_stdout_nopath' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_run_process_get_stdout_nopath'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='9b23c9ad' name='argv'/>
	<parameter type-id='9b23c9ad' name='env'/>
	<parameter type-id='c0563f85' name='lines'/>
	<parameter type-id='7292109c' name='lines_cnt'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='libzfs_free_str_array' mangled-name='libzfs_free_str_array' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_free_str_array'>
	<parameter type-id='9b23c9ad' name='strs'/>
	<parameter type-id='95e97e5e' name='count'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='libzfs_envvar_is_set' mangled-name='libzfs_envvar_is_set' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_envvar_is_set'>
	<parameter type-id='26a90f95' name='envvar'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='libzfs_init' mangled-name='libzfs_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_init'>
	<return type-id='b0382bb3'/>
	</function-decl>
	<function-decl name='libzfs_fini' mangled-name='libzfs_fini' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_fini'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zpool_get_handle' mangled-name='zpool_get_handle' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_get_handle'>
	<parameter type-id='4c81de99' name='zhp'/>
	<return type-id='b0382bb3'/>
	</function-decl>
	<function-decl name='zfs_get_handle' mangled-name='zfs_get_handle' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_handle'>
	<parameter type-id='9200a744' name='zhp'/>
	<return type-id='b0382bb3'/>
	</function-decl>
	<function-decl name='zfs_get_pool_handle' mangled-name='zfs_get_pool_handle' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_pool_handle'>
	<parameter type-id='fcd57163' name='zhp'/>
	<return type-id='4c81de99'/>
	</function-decl>
	<function-decl name='zfs_path_to_zhandle' mangled-name='zfs_path_to_zhandle' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_path_to_zhandle'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='2e45de5d' name='argtype'/>
	<return type-id='9200a744'/>
	</function-decl>
	<function-decl name='zprop_print_one_property' mangled-name='zprop_print_one_property' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_print_one_property'>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='0d2a0670' name='cbp'/>
	<parameter type-id='80f4b756' name='propname'/>
	<parameter type-id='80f4b756' name='value'/>
	<parameter type-id='a2256d42' name='sourcetype'/>
	<parameter type-id='80f4b756' name='source'/>
	<parameter type-id='80f4b756' name='recvd_value'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_nicestrtonum' mangled-name='zfs_nicestrtonum' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_nicestrtonum'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='80f4b756' name='value'/>
	<parameter type-id='5d6479ae' name='num'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zprop_get_list' mangled-name='zprop_get_list' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_get_list'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='26a90f95' name='props'/>
	<parameter type-id='e4378506' name='listp'/>
	<parameter type-id='2e45de5d' name='type'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zprop_free_list' mangled-name='zprop_free_list' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_free_list'>
	<parameter type-id='3a9b2288' name='pl'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zprop_iter' mangled-name='zprop_iter' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zprop_iter'>
	<parameter type-id='1ec3747a' name='func'/>
	<parameter type-id='eaa32e2f' name='cb'/>
	<parameter type-id='c19b74c3' name='show_all'/>
	<parameter type-id='c19b74c3' name='ordered'/>
	<parameter type-id='2e45de5d' name='type'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_version_userland' mangled-name='zfs_version_userland' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_version_userland'>
	<parameter type-id='26a90f95' name='version'/>
	<parameter type-id='95e97e5e' name='len'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_version_print' mangled-name='zfs_version_print' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_version_print'>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='color_start' mangled-name='color_start' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='color_start'>
	<parameter type-id='26a90f95' name='color'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='color_end' mangled-name='color_end' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='color_end'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='printf_color' mangled-name='printf_color' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='printf_color'>
	<parameter type-id='26a90f95' name='color'/>
	<parameter type-id='26a90f95' name='format'/>
	<parameter is-variadic='yes'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-type size-in-bits='64' id='c70fa2e8'>
	<parameter type-id='95e97e5e'/>
	<parameter type-id='eaa32e2f'/>
	<return type-id='95e97e5e'/>
	</function-type>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/libzfs_mount_os.c' language='LANG_C99'>
	<pointer-type-def type-id='7359adad' size-in-bits='64' id='1d2c2b85'/>
	<function-decl name='zfs_parse_mount_options' mangled-name='zfs_parse_mount_options' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_parse_mount_options'>
	<parameter type-id='26a90f95' name='mntopts'/>
	<parameter type-id='1d2c2b85' name='mntflags'/>
	<parameter type-id='1d2c2b85' name='zfsflags'/>
	<parameter type-id='95e97e5e' name='sloppy'/>
	<parameter type-id='26a90f95' name='badopt'/>
	<parameter type-id='26a90f95' name='mtabopt'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_adjust_mount_options' mangled-name='zfs_adjust_mount_options' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_adjust_mount_options'>
	<parameter type-id='9200a744' name='zhp'/>
	<parameter type-id='80f4b756' name='mntpoint'/>
	<parameter type-id='26a90f95' name='mntopts'/>
	<parameter type-id='26a90f95' name='mtabopt'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_mount_delegation_check' mangled-name='zfs_mount_delegation_check' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_mount_delegation_check'>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_disable_datasets_os' mangled-name='zpool_disable_datasets_os' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_disable_datasets_os'>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='c19b74c3' name='force'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zpool_disable_volume_os' mangled-name='zpool_disable_volume_os' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_disable_volume_os'>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/libzfs_pool_os.c' language='LANG_C99'>
	<function-decl name='zpool_label_disk' mangled-name='zpool_label_disk' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_label_disk'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='4c81de99' name='zhp'/>
	<parameter type-id='80f4b756' name='name'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/libzfs_util_os.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='32768' id='d16c6df4'>
	<subrange length='4096' type-id='7359adad' id='bc1b5ddc'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='65536' id='163f6aa5'>
	<subrange length='8192' type-id='7359adad' id='c88f397d'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='9c313c2d' size-in-bits='128' id='c1c22e6c'>
	<subrange length='2' type-id='7359adad' id='52efc4ef'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='b96825af' size-in-bits='24' id='d3490169'>
	<subrange length='3' type-id='7359adad' id='56f209d2'/>
	</array-type-def>
	<class-decl name='drr_begin' size-in-bits='2432' is-struct='yes' visibility='default' id='09fcdc01'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_magic' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_versioninfo' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='drr_creation_time' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='drr_type' type-id='230f1e16' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='224'>
	<var-decl name='drr_flags' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='drr_fromguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='drr_toname' type-id='d1617432' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='zinject_record' size-in-bits='2816' is-struct='yes' visibility='default' id='3216f820'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='zi_objset' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='zi_object' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='zi_start' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='zi_end' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='zi_guid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='zi_level' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='352'>
	<var-decl name='zi_error' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='zi_type' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='zi_freq' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='480'>
	<var-decl name='zi_failfast' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='zi_func' type-id='d1617432' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2560'>
	<var-decl name='zi_iotype' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2592'>
	<var-decl name='zi_duration' type-id='3ff5601b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2624'>
	<var-decl name='zi_timer' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2688'>
	<var-decl name='zi_nlanes' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2752'>
	<var-decl name='zi_cmd' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2784'>
	<var-decl name='zi_dvas' type-id='8f92235e' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zinject_record_t' type-id='3216f820' id='a4301ca6'/>
	<class-decl name='zfs_share' size-in-bits='256' is-struct='yes' visibility='default' id='feb6f2da'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='z_exportdata' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='z_sharedata' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='z_sharetype' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='z_sharemax' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfs_share_t' type-id='feb6f2da' id='ee5cec36'/>
	<class-decl name='zfs_cmd' size-in-bits='109952' is-struct='yes' visibility='default' id='3522cd69'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='zc_name' type-id='d16c6df4' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32768'>
	<var-decl name='zc_nvlist_src' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32832'>
	<var-decl name='zc_nvlist_src_size' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32896'>
	<var-decl name='zc_nvlist_dst' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32960'>
	<var-decl name='zc_nvlist_dst_size' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='33024'>
	<var-decl name='zc_nvlist_dst_filled' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='33056'>
	<var-decl name='zc_pad2' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='33088'>
	<var-decl name='zc_history' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='33152'>
	<var-decl name='zc_value' type-id='163f6aa5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='98688'>
	<var-decl name='zc_string' type-id='d1617432' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='100736'>
	<var-decl name='zc_guid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='100800'>
	<var-decl name='zc_nvlist_conf' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='100864'>
	<var-decl name='zc_nvlist_conf_size' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='100928'>
	<var-decl name='zc_cookie' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='100992'>
	<var-decl name='zc_objset_type' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101056'>
	<var-decl name='zc_perm_action' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101120'>
	<var-decl name='zc_history_len' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101184'>
	<var-decl name='zc_history_offset' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101248'>
	<var-decl name='zc_obj' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101312'>
	<var-decl name='zc_iflags' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101376'>
	<var-decl name='zc_share' type-id='ee5cec36' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101632'>
	<var-decl name='zc_objset_stats' type-id='b2c14f17' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='103936'>
	<var-decl name='zc_begin_record' type-id='09fcdc01' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='106368'>
	<var-decl name='zc_inject_record' type-id='a4301ca6' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109184'>
	<var-decl name='zc_defer_destroy' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109216'>
	<var-decl name='zc_flags' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109248'>
	<var-decl name='zc_action_handle' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109312'>
	<var-decl name='zc_cleanup_fd' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109344'>
	<var-decl name='zc_simple' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109352'>
	<var-decl name='zc_pad' type-id='d3490169' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109376'>
	<var-decl name='zc_sendobj' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109440'>
	<var-decl name='zc_fromobj' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109504'>
	<var-decl name='zc_createtxg' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109568'>
	<var-decl name='zc_stat' type-id='0371a9c7' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109888'>
	<var-decl name='zc_zoneid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfs_cmd_t' type-id='3522cd69' id='a5559cdd'/>
	<class-decl name='zfs_stat' size-in-bits='320' is-struct='yes' visibility='default' id='6417f0b9'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='zs_gen' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='zs_mode' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='zs_links' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='zs_ctime' type-id='c1c22e6c' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfs_stat_t' type-id='6417f0b9' id='0371a9c7'/>
	<pointer-type-def type-id='a5559cdd' size-in-bits='64' id='e4ec4540'/>
	<function-decl name='zfs_ioctl' mangled-name='zfs_ioctl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_ioctl'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='95e97e5e' name='request'/>
	<parameter type-id='e4ec4540' name='zc'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='libzfs_error_init' mangled-name='libzfs_error_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_error_init'>
	<parameter type-id='95e97e5e' name='error'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zfs_destroy_snaps_nvl_os' mangled-name='zfs_destroy_snaps_nvl_os' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_destroy_snaps_nvl_os'>
	<parameter type-id='b0382bb3' name='hdl'/>
	<parameter type-id='5ce45b60' name='snaps'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_version_kernel' mangled-name='zfs_version_kernel' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_version_kernel'>
	<parameter type-id='26a90f95' name='version'/>
	<parameter type-id='95e97e5e' name='len'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/zutil_device_path_os.c' language='LANG_C99'>
	<function-decl name='zfs_append_partition' mangled-name='zfs_append_partition' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_append_partition'>
	<parameter type-id='26a90f95' name='path'/>
	<parameter type-id='b59d7dce' name='max_len'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_strip_partition' mangled-name='zfs_strip_partition' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_strip_partition'>
	<parameter type-id='26a90f95' name='path'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='zfs_strip_path' mangled-name='zfs_strip_path' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_strip_path'>
	<parameter type-id='26a90f95' name='path'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='zfs_get_enclosure_sysfs_path' mangled-name='zfs_get_enclosure_sysfs_path' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_enclosure_sysfs_path'>
	<parameter type-id='80f4b756' name='dev_name'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='zfs_dev_is_dm' mangled-name='zfs_dev_is_dm' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_dev_is_dm'>
	<parameter type-id='80f4b756' name='dev_name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_dev_is_whole_disk' mangled-name='zfs_dev_is_whole_disk' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_dev_is_whole_disk'>
	<parameter type-id='80f4b756' name='dev_name'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_get_underlying_path' mangled-name='zfs_get_underlying_path' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_get_underlying_path'>
	<parameter type-id='80f4b756' name='dev_name'/>
	<return type-id='26a90f95'/>
	</function-decl>
	<function-decl name='is_mpath_whole_disk' mangled-name='is_mpath_whole_disk' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='is_mpath_whole_disk'>
	<parameter type-id='80f4b756' name='path'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/zutil_import_os.c' language='LANG_C99'>
	- <class-decl name='udev_device' is-struct='yes' visibility='default' is-declaration-only='yes' id='640b33ca'/>
	<qualified-type-def type-id='80f4b756' const='yes' id='b99c00c9'/>
	<pointer-type-def type-id='b99c00c9' size-in-bits='64' id='13956559'/>
	<pointer-type-def type-id='b59d7dce' size-in-bits='64' id='78c01427'/>
	<pointer-type-def type-id='640b33ca' size-in-bits='64' id='b32bae08'/>
	<class-decl name='udev_device' is-struct='yes' visibility='default' is-declaration-only='yes' id='640b33ca'/>
	<function-decl name='zfs_dev_flush' mangled-name='zfs_dev_flush' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_dev_flush'>
	<parameter type-id='95e97e5e' name='fd'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_default_search_paths' mangled-name='zpool_default_search_paths' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_default_search_paths'>
	<parameter type-id='78c01427' name='count'/>
	<return type-id='13956559'/>
	</function-decl>
	<function-decl name='zfs_device_get_devid' mangled-name='zfs_device_get_devid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_device_get_devid'>
	<parameter type-id='b32bae08' name='dev'/>
	<parameter type-id='26a90f95' name='bufptr'/>
	<parameter type-id='b59d7dce' name='buflen'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_device_get_physical' mangled-name='zfs_device_get_physical' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_device_get_physical'>
	<parameter type-id='b32bae08' name='dev'/>
	<parameter type-id='26a90f95' name='bufptr'/>
	<parameter type-id='b59d7dce' name='buflen'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_label_disk_wait' mangled-name='zpool_label_disk_wait' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_label_disk_wait'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='95e97e5e' name='timeout_ms'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='update_vdev_config_dev_strs' mangled-name='update_vdev_config_dev_strs' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='update_vdev_config_dev_strs'>
	<parameter type-id='5ce45b60' name='nv'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='zutil_device_path.c' language='LANG_C99'>
	<typedef-decl name='ssize_t' type-id='41060289' id='79a0948f'/>
	<typedef-decl name='__ssize_t' type-id='bd54fe1a' id='41060289'/>
	<function-decl name='zfs_basename' mangled-name='zfs_basename' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_basename'>
	<parameter type-id='80f4b756' name='path'/>
	<return type-id='80f4b756'/>
	</function-decl>
	<function-decl name='zfs_dirnamelen' mangled-name='zfs_dirnamelen' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_dirnamelen'>
	<parameter type-id='80f4b756' name='path'/>
	<return type-id='79a0948f'/>
	</function-decl>
	<function-decl name='zfs_resolve_shortname' mangled-name='zfs_resolve_shortname' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_resolve_shortname'>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='26a90f95' name='path'/>
	<parameter type-id='b59d7dce' name='len'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zfs_strcmp_pathname' mangled-name='zfs_strcmp_pathname' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_strcmp_pathname'>
	<parameter type-id='80f4b756' name='name'/>
	<parameter type-id='80f4b756' name='cmp'/>
	<parameter type-id='95e97e5e' name='wholedisk'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='zutil_import.c' language='LANG_C99'>
	<class-decl name='importargs' size-in-bits='448' is-struct='yes' visibility='default' id='7ac83801'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='path' type-id='9b23c9ad' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='paths' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='poolname' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='guid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='cachefile' type-id='80f4b756' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='can_be_active' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='352'>
	<var-decl name='scan' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='policy' type-id='5ce45b60' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='importargs_t' type-id='7ac83801' id='7a842a6b'/>
	<pointer-type-def type-id='7a842a6b' size-in-bits='64' id='07ee4a58'/>
	<pointer-type-def type-id='b1e62775' size-in-bits='64' id='f095e320'/>
	<function-decl name='zpool_read_label' mangled-name='zpool_read_label' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_read_label'>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='857bb57e' name='config'/>
	<parameter type-id='7292109c' name='num_labels'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='zpool_search_import' mangled-name='zpool_search_import' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_search_import'>
	<parameter type-id='eaa32e2f' name='hdl'/>
	<parameter type-id='07ee4a58' name='import'/>
	<parameter type-id='f095e320' name='pco'/>
	<return type-id='5ce45b60'/>
	</function-decl>
	<function-decl name='zpool_find_config' mangled-name='zpool_find_config' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_find_config'>
	<parameter type-id='eaa32e2f' name='hdl'/>
	<parameter type-id='80f4b756' name='target'/>
	<parameter type-id='857bb57e' name='configp'/>
	<parameter type-id='07ee4a58' name='args'/>
	<parameter type-id='f095e320' name='pco'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='zutil_nicenum.c' language='LANG_C99'>
	<enum-decl name='zfs_nicenum_format' id='29cf1969'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZFS_NICENUM_1024' value='0'/>
	<enumerator name='ZFS_NICENUM_BYTES' value='1'/>
	<enumerator name='ZFS_NICENUM_TIME' value='2'/>
	<enumerator name='ZFS_NICENUM_RAW' value='3'/>
	<enumerator name='ZFS_NICENUM_RAWTIME' value='4'/>
	</enum-decl>
	<function-decl name='zfs_isnumber' mangled-name='zfs_isnumber' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_isnumber'>
	<parameter type-id='80f4b756' name='str'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='zfs_nicenum_format' mangled-name='zfs_nicenum_format' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_nicenum_format'>
	<parameter type-id='9c313c2d' name='num'/>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='b59d7dce' name='buflen'/>
	<parameter type-id='29cf1969' name='format'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_nicenum' mangled-name='zfs_nicenum' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_nicenum'>
	<parameter type-id='9c313c2d' name='num'/>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='b59d7dce' name='buflen'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_nicetime' mangled-name='zfs_nicetime' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_nicetime'>
	<parameter type-id='9c313c2d' name='num'/>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='b59d7dce' name='buflen'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_niceraw' mangled-name='zfs_niceraw' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_niceraw'>
	<parameter type-id='9c313c2d' name='num'/>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='b59d7dce' name='buflen'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zfs_nicebytes' mangled-name='zfs_nicebytes' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zfs_nicebytes'>
	<parameter type-id='9c313c2d' name='num'/>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='b59d7dce' name='buflen'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='zutil_pool.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='853fd5dc' size-in-bits='32768' id='b505fc2f'>
	<subrange length='64' type-id='7359adad' id='b10be967'/>
	</array-type-def>
	<class-decl name='ddt_stat' size-in-bits='512' is-struct='yes' visibility='default' id='65242dfe'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='dds_blocks' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='dds_lsize' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='dds_psize' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='dds_dsize' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='dds_ref_blocks' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='dds_ref_lsize' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='dds_ref_psize' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='dds_ref_dsize' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='ddt_stat_t' type-id='65242dfe' id='853fd5dc'/>
	<class-decl name='ddt_histogram' size-in-bits='32768' is-struct='yes' visibility='default' id='bc2b3086'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='ddh_stat' type-id='b505fc2f' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='ddt_histogram_t' type-id='bc2b3086' id='2d7fe832'/>
	<qualified-type-def type-id='2d7fe832' const='yes' id='ec92d602'/>
	<pointer-type-def type-id='ec92d602' size-in-bits='64' id='932720f8'/>
	<qualified-type-def type-id='853fd5dc' const='yes' id='764c298c'/>
	<pointer-type-def type-id='764c298c' size-in-bits='64' id='dfe59052'/>
	<pointer-type-def type-id='857bb57e' size-in-bits='64' id='75be733c'/>
	<function-decl name='zpool_dump_ddt' mangled-name='zpool_dump_ddt' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_dump_ddt'>
	<parameter type-id='dfe59052' name='dds_total'/>
	<parameter type-id='932720f8' name='ddh'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='zpool_history_unpack' mangled-name='zpool_history_unpack' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='zpool_history_unpack'>
	<parameter type-id='26a90f95' name='buf'/>
	<parameter type-id='9c313c2d' name='bytes_read'/>
	<parameter type-id='5d6479ae' name='leftover'/>
	<parameter type-id='75be733c' name='records'/>
	<parameter type-id='4dd26a40' name='numrecords'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	</abi-corpus>
	diff --git a/sys/contrib/openzfs/lib/libzfs/libzfs_crypto.c b/sys/contrib/openzfs/lib/libzfs/libzfs_crypto.c
	index 644dd26859f1..cfe04a79273d 100644
	--- a/sys/contrib/openzfs/lib/libzfs/libzfs_crypto.c
	+++ b/sys/contrib/openzfs/lib/libzfs/libzfs_crypto.c
	@@ -1,1805 +1,1812 @@
	/*
	* CDDL HEADER START
	*
	* This file and its contents are supplied under the terms of the
	* Common Development and Distribution License ("CDDL"), version 1.0.
	* You may only use this file in accordance with the terms of version
	* 1.0 of the CDDL.
	*
	* A full copy of the text of the CDDL should have accompanied this
	* source. A copy of the CDDL is also available via the Internet at
	* http://www.illumos.org/license/CDDL.
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2017, Datto, Inc. All rights reserved.
	* Copyright 2020 Joyent, Inc.
	*/

	#include <sys/zfs_context.h>
	#include <sys/fs/zfs.h>
	#include <sys/dsl_crypt.h>
	#include <libintl.h>
	#include <termios.h>
	#include <signal.h>
	#include <errno.h>
	#include <openssl/evp.h>
	#if LIBFETCH_DYNAMIC
	#include <dlfcn.h>
	#endif
	#if LIBFETCH_IS_FETCH
	#include <sys/param.h>
	#include <stdio.h>
	#include <fetch.h>
	#elif LIBFETCH_IS_LIBCURL
	#include <curl/curl.h>
	#endif
	#include <libzfs.h>
	#include "libzfs_impl.h"
	#include "zfeature_common.h"

	/*
	* User keys are used to decrypt the master encryption keys of a dataset. This
	* indirection allows a user to change his / her access key without having to
	* re-encrypt the entire dataset. User keys can be provided in one of several
	* ways. Raw keys are simply given to the kernel as is. Similarly, hex keys
	* are converted to binary and passed into the kernel. Password based keys are
	* a bit more complicated. Passwords alone do not provide suitable entropy for
	* encryption and may be too short or too long to be used. In order to derive
	* a more appropriate key we use a PBKDF2 function. This function is designed
	* to take a (relatively) long time to calculate in order to discourage
	* attackers from guessing from a list of common passwords. PBKDF2 requires
	* 2 additional parameters. The first is the number of iterations to run, which
	* will ultimately determine how long it takes to derive the resulting key from
	* the password. The second parameter is a salt that is randomly generated for
	* each dataset. The salt is used to "tweak" PBKDF2 such that a group of
	* attackers cannot reasonably generate a table of commonly known passwords to
	* their output keys and expect it work for all past and future PBKDF2 users.
	* We store the salt as a hidden property of the dataset (although it is
	* technically ok if the salt is known to the attacker).
	*/

	#define MIN_PASSPHRASE_LEN 8
	#define MAX_PASSPHRASE_LEN 512
	#define MAX_KEY_PROMPT_ATTEMPTS 3

	static int caught_interrupt;

	static int get_key_material_file(libzfs_handle_t , const char , const char *,
	zfs_keyformat_t, boolean_t, uint8_t *, size_t );
	static int get_key_material_https(libzfs_handle_t , const char , const char *,
	zfs_keyformat_t, boolean_t, uint8_t *, size_t );

	static zfs_uri_handler_t uri_handlers[] = {
	{ "file", get_key_material_file },
	{ "https", get_key_material_https },
	{ "http", get_key_material_https },
	{ NULL, NULL }
	};

	static int
	pkcs11_get_urandom(uint8_t *buf, size_t bytes)
	{
	int rand;
	ssize_t bytes_read = 0;

	rand = open("/dev/urandom", O_RDONLY \| O_CLOEXEC);

	if (rand < 0)
	return (rand);

	while (bytes_read < bytes) {
	ssize_t rc = read(rand, buf + bytes_read, bytes - bytes_read);
	if (rc < 0)
	break;
	bytes_read += rc;
	}

	(void) close(rand);

	return (bytes_read);
	}

	static int
	zfs_prop_parse_keylocation(libzfs_handle_t restrict hdl, const char str,
	zfs_keylocation_t restrict locp, char *restrict schemep)
	{
	*locp = ZFS_KEYLOCATION_NONE;
	*schemep = NULL;

	if (strcmp("prompt", str) == 0) {
	*locp = ZFS_KEYLOCATION_PROMPT;
	return (0);
	}

	regmatch_t pmatch[2];

	if (regexec(&hdl->libzfs_urire, str, ARRAY_SIZE(pmatch),
	pmatch, 0) == 0) {
	size_t scheme_len;

	if (pmatch[1].rm_so == -1) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Invalid URI"));
	return (EINVAL);
	}

	scheme_len = pmatch[1].rm_eo - pmatch[1].rm_so;

	*schemep = calloc(1, scheme_len + 1);
	if (*schemep == NULL) {
	int ret = errno;

	errno = 0;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Invalid URI"));
	return (ret);
	}

	(void) memcpy(*schemep, str + pmatch[1].rm_so, scheme_len);
	*locp = ZFS_KEYLOCATION_URI;
	return (0);
	}

	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Invalid keylocation"));
	return (EINVAL);
	}

	static int
	hex_key_to_raw(char hex, int hexlen, uint8_t out)
	{
	int ret, i;
	unsigned int c;

	for (i = 0; i < hexlen; i += 2) {
	if (!isxdigit(hex[i]) \|\| !isxdigit(hex[i + 1])) {
	ret = EINVAL;
	goto error;
	}

	ret = sscanf(&hex[i], "%02x", &c);
	if (ret != 1) {
	ret = EINVAL;
	goto error;
	}

	out[i / 2] = c;
	}

	return (0);

	error:
	return (ret);
	}


	static void
	catch_signal(int sig)
	{
	caught_interrupt = sig;
	}

	static const char *
	get_format_prompt_string(zfs_keyformat_t format)
	{
	switch (format) {
	case ZFS_KEYFORMAT_RAW:
	return ("raw key");
	case ZFS_KEYFORMAT_HEX:
	return ("hex key");
	case ZFS_KEYFORMAT_PASSPHRASE:
	return ("passphrase");
	default:
	/* shouldn't happen */
	return (NULL);
	}
	}

	/* do basic validation of the key material */
	static int
	validate_key(libzfs_handle_t *hdl, zfs_keyformat_t keyformat,
	- const char *key, size_t keylen)
	+ const char *key, size_t keylen, boolean_t do_verify)
	{
	switch (keyformat) {
	case ZFS_KEYFORMAT_RAW:
	/* verify the key length is correct */
	if (keylen < WRAPPING_KEY_LEN) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Raw key too short (expected %u)."),
	WRAPPING_KEY_LEN);
	return (EINVAL);
	}

	if (keylen > WRAPPING_KEY_LEN) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Raw key too long (expected %u)."),
	WRAPPING_KEY_LEN);
	return (EINVAL);
	}
	break;
	case ZFS_KEYFORMAT_HEX:
	/* verify the key length is correct */
	if (keylen < WRAPPING_KEY_LEN * 2) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Hex key too short (expected %u)."),
	WRAPPING_KEY_LEN * 2);
	return (EINVAL);
	}

	if (keylen > WRAPPING_KEY_LEN * 2) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Hex key too long (expected %u)."),
	WRAPPING_KEY_LEN * 2);
	return (EINVAL);
	}

	/* check for invalid hex digits */
	for (size_t i = 0; i < WRAPPING_KEY_LEN * 2; i++) {
	if (!isxdigit(key[i])) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Invalid hex character detected."));
	return (EINVAL);
	}
	}
	break;
	case ZFS_KEYFORMAT_PASSPHRASE:
	- /* verify the length is within bounds */
	+ /*
	+ * Verify the length is within bounds when setting a new key,
	+ * but not when loading an existing key.
	+ */
	+ if (!do_verify)
	+ break;
	if (keylen > MAX_PASSPHRASE_LEN) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Passphrase too long (max %u)."),
	MAX_PASSPHRASE_LEN);
	return (EINVAL);
	}

	if (keylen < MIN_PASSPHRASE_LEN) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Passphrase too short (min %u)."),
	MIN_PASSPHRASE_LEN);
	return (EINVAL);
	}
	break;
	default:
	/* can't happen, checked above */
	break;
	}

	return (0);
	}

	static int
	libzfs_getpassphrase(zfs_keyformat_t keyformat, boolean_t is_reenter,
	boolean_t new_key, const char *fsname,
	char *restrict res, size_t restrict reslen)
	{
	FILE *f = stdin;
	size_t buflen = 0;
	ssize_t bytes;
	int ret = 0;
	struct termios old_term, new_term;
	struct sigaction act, osigint, osigtstp;

	*res = NULL;
	*reslen = 0;

	/*
	* handle SIGINT and ignore SIGSTP. This is necessary to
	* restore the state of the terminal.
	*/
	caught_interrupt = 0;
	act.sa_flags = 0;
	(void) sigemptyset(&act.sa_mask);
	act.sa_handler = catch_signal;

	(void) sigaction(SIGINT, &act, &osigint);
	act.sa_handler = SIG_IGN;
	(void) sigaction(SIGTSTP, &act, &osigtstp);

	(void) printf("%s %s%s",
	is_reenter ? "Re-enter" : "Enter",
	new_key ? "new " : "",
	get_format_prompt_string(keyformat));
	if (fsname != NULL)
	(void) printf(" for '%s'", fsname);
	(void) fputc(':', stdout);
	(void) fflush(stdout);

	/* disable the terminal echo for key input */
	(void) tcgetattr(fileno(f), &old_term);

	new_term = old_term;
	new_term.c_lflag &= ~(ECHO \| ECHOE \| ECHOK \| ECHONL);

	ret = tcsetattr(fileno(f), TCSAFLUSH, &new_term);
	if (ret != 0) {
	ret = errno;
	errno = 0;
	goto out;
	}

	bytes = getline(res, &buflen, f);
	if (bytes < 0) {
	ret = errno;
	errno = 0;
	goto out;
	}

	/* trim the ending newline if it exists */
	if (bytes > 0 && (*res)[bytes - 1] == '\n') {
	(*res)[bytes - 1] = '\0';
	bytes--;
	}

	*reslen = bytes;

	out:
	/* reset the terminal */
	(void) tcsetattr(fileno(f), TCSAFLUSH, &old_term);
	(void) sigaction(SIGINT, &osigint, NULL);
	(void) sigaction(SIGTSTP, &osigtstp, NULL);

	/* if we caught a signal, re-throw it now */
	if (caught_interrupt != 0)
	(void) kill(getpid(), caught_interrupt);

	/* print the newline that was not echo'd */
	(void) printf("\n");

	return (ret);
	}

	static int
	get_key_interactive(libzfs_handle_t restrict hdl, const char fsname,
	zfs_keyformat_t keyformat, boolean_t confirm_key, boolean_t newkey,
	uint8_t *restrict outbuf, size_t restrict len_out)
	{
	char buf = NULL, buf2 = NULL;
	size_t buflen = 0, buf2len = 0;
	int ret = 0;

	ASSERT(isatty(fileno(stdin)));

	/* raw keys cannot be entered on the terminal */
	if (keyformat == ZFS_KEYFORMAT_RAW) {
	ret = EINVAL;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Cannot enter raw keys on the terminal"));
	goto out;
	}

	/* prompt for the key */
	if ((ret = libzfs_getpassphrase(keyformat, B_FALSE, newkey, fsname,
	&buf, &buflen)) != 0) {
	free(buf);
	buf = NULL;
	buflen = 0;
	goto out;
	}

	if (!confirm_key)
	goto out;

	- if ((ret = validate_key(hdl, keyformat, buf, buflen)) != 0) {
	+ if ((ret = validate_key(hdl, keyformat, buf, buflen, confirm_key)) !=
	+ 0) {
	free(buf);
	return (ret);
	}

	ret = libzfs_getpassphrase(keyformat, B_TRUE, newkey, fsname, &buf2,
	&buf2len);
	if (ret != 0) {
	free(buf);
	free(buf2);
	buf = buf2 = NULL;
	buflen = buf2len = 0;
	goto out;
	}

	if (buflen != buf2len \|\| strcmp(buf, buf2) != 0) {
	free(buf);
	buf = NULL;
	buflen = 0;

	ret = EINVAL;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Provided keys do not match."));
	}

	free(buf2);

	out:
	outbuf = (uint8_t )buf;
	*len_out = buflen;
	return (ret);
	}

	static int
	get_key_material_raw(FILE *fd, zfs_keyformat_t keyformat,
	uint8_t *buf, size_t len_out)
	{
	int ret = 0;
	size_t buflen = 0;

	*len_out = 0;

	/* read the key material */
	if (keyformat != ZFS_KEYFORMAT_RAW) {
	ssize_t bytes;

	bytes = getline((char **)buf, &buflen, fd);
	if (bytes < 0) {
	ret = errno;
	errno = 0;
	goto out;
	}

	/* trim the ending newline if it exists */
	if (bytes > 0 && (*buf)[bytes - 1] == '\n') {
	(*buf)[bytes - 1] = '\0';
	bytes--;
	}

	*len_out = bytes;
	} else {
	size_t n;

	/*
	* Raw keys may have newline characters in them and so can't
	* use getline(). Here we attempt to read 33 bytes so that we
	* can properly check the key length (the file should only have
	* 32 bytes).
	*/
	buf = malloc((WRAPPING_KEY_LEN + 1) sizeof (uint8_t));
	if (*buf == NULL) {
	ret = ENOMEM;
	goto out;
	}

	n = fread(*buf, 1, WRAPPING_KEY_LEN + 1, fd);
	if (n == 0 \|\| ferror(fd)) {
	/* size errors are handled by the calling function */
	free(*buf);
	*buf = NULL;
	ret = errno;
	errno = 0;
	goto out;
	}

	*len_out = n;
	}
	out:
	return (ret);
	}

	static int
	get_key_material_file(libzfs_handle_t hdl, const char uri,
	const char *fsname, zfs_keyformat_t keyformat, boolean_t newkey,
	uint8_t *restrict buf, size_t restrict len_out)
	{
	FILE *f = NULL;
	int ret = 0;

	if (strlen(uri) < 7)
	return (EINVAL);

	if ((f = fopen(uri + 7, "re")) == NULL) {
	ret = errno;
	errno = 0;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Failed to open key material file: %s"), strerror(ret));
	return (ret);
	}

	ret = get_key_material_raw(f, keyformat, buf, len_out);

	(void) fclose(f);

	return (ret);
	}

	static int
	get_key_material_https(libzfs_handle_t hdl, const char uri,
	const char *fsname, zfs_keyformat_t keyformat, boolean_t newkey,
	uint8_t *restrict buf, size_t restrict len_out)
	{
	int ret = 0;
	FILE *key = NULL;
	boolean_t is_http = strncmp(uri, "http:", strlen("http:")) == 0;

	if (strlen(uri) < (is_http ? 7 : 8)) {
	ret = EINVAL;
	goto end;
	}

	#if LIBFETCH_DYNAMIC
	#define LOAD_FUNCTION(func) \
	__typeof__(func) *func = dlsym(hdl->libfetch, #func);

	if (hdl->libfetch == NULL)
	hdl->libfetch = dlopen(LIBFETCH_SONAME, RTLD_LAZY);

	if (hdl->libfetch == NULL) {
	hdl->libfetch = (void *)-1;
	char *err = dlerror();
	if (err)
	hdl->libfetch_load_error = strdup(err);
	}

	if (hdl->libfetch == (void *)-1) {
	ret = ENOSYS;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Couldn't load %s: %s"),
	LIBFETCH_SONAME, hdl->libfetch_load_error ?: "(?)");
	goto end;
	}

	boolean_t ok;
	#if LIBFETCH_IS_FETCH
	LOAD_FUNCTION(fetchGetURL);
	char *fetchLastErrString = dlsym(hdl->libfetch, "fetchLastErrString");

	ok = fetchGetURL && fetchLastErrString;
	#elif LIBFETCH_IS_LIBCURL
	LOAD_FUNCTION(curl_easy_init);
	LOAD_FUNCTION(curl_easy_setopt);
	LOAD_FUNCTION(curl_easy_perform);
	LOAD_FUNCTION(curl_easy_cleanup);
	LOAD_FUNCTION(curl_easy_strerror);
	LOAD_FUNCTION(curl_easy_getinfo);

	ok = curl_easy_init && curl_easy_setopt && curl_easy_perform &&
	curl_easy_cleanup && curl_easy_strerror && curl_easy_getinfo;
	#endif
	if (!ok) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"keylocation=%s back-end %s missing symbols."),
	is_http ? "http://" : "https://", LIBFETCH_SONAME);
	ret = ENOSYS;
	goto end;
	}
	#endif

	#if LIBFETCH_IS_FETCH
	key = fetchGetURL(uri, "");
	if (key == NULL) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Couldn't GET %s: %s"),
	uri, fetchLastErrString);
	ret = ENETDOWN;
	}
	#elif LIBFETCH_IS_LIBCURL
	CURL *curl = curl_easy_init();
	if (curl == NULL) {
	ret = ENOTSUP;
	goto end;
	}

	int kfd = -1;
	#ifdef O_TMPFILE
	kfd = open(getenv("TMPDIR") ?: "/tmp",
	O_RDWR \| O_TMPFILE \| O_EXCL \| O_CLOEXEC, 0600);
	if (kfd != -1)
	goto kfdok;
	#endif

	char *path;
	if (asprintf(&path,
	"%s/libzfs-XXXXXXXX.https", getenv("TMPDIR") ?: "/tmp") == -1) {
	ret = ENOMEM;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "%s"),
	strerror(ret));
	goto end;
	}

	kfd = mkostemps(path, strlen(".https"), O_CLOEXEC);
	if (kfd == -1) {
	ret = errno;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Couldn't create temporary file %s: %s"),
	path, strerror(ret));
	free(path);
	goto end;
	}
	(void) unlink(path);
	free(path);

	kfdok:
	if ((key = fdopen(kfd, "r+")) == NULL) {
	ret = errno;
	free(path);
	(void) close(kfd);
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Couldn't reopen temporary file: %s"), strerror(ret));
	goto end;
	}

	char errbuf[CURL_ERROR_SIZE] = "";
	char cainfo = getenv("SSL_CA_CERT_FILE"); / matches fetch(3) */
	char capath = getenv("SSL_CA_CERT_PATH"); / matches fetch(3) */
	char clcert = getenv("SSL_CLIENT_CERT_FILE"); / matches fetch(3) */
	char clkey = getenv("SSL_CLIENT_KEY_FILE"); / matches fetch(3) */
	(void) curl_easy_setopt(curl, CURLOPT_URL, uri);
	(void) curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1L);
	(void) curl_easy_setopt(curl, CURLOPT_TIMEOUT_MS, 30000L);
	(void) curl_easy_setopt(curl, CURLOPT_WRITEDATA, key);
	(void) curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errbuf);
	if (cainfo != NULL)
	(void) curl_easy_setopt(curl, CURLOPT_CAINFO, cainfo);
	if (capath != NULL)
	(void) curl_easy_setopt(curl, CURLOPT_CAPATH, capath);
	if (clcert != NULL)
	(void) curl_easy_setopt(curl, CURLOPT_SSLCERT, clcert);
	if (clkey != NULL)
	(void) curl_easy_setopt(curl, CURLOPT_SSLKEY, clkey);

	CURLcode res = curl_easy_perform(curl);

	if (res != CURLE_OK) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Failed to connect to %s: %s"),
	uri, strlen(errbuf) ? errbuf : curl_easy_strerror(res));
	ret = ENETDOWN;
	} else {
	long resp = 200;
	(void) curl_easy_getinfo(curl, CURLINFO_RESPONSE_CODE, &resp);

	if (resp < 200 \|\| resp >= 300) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Couldn't GET %s: %ld"),
	uri, resp);
	ret = ENOENT;
	} else
	rewind(key);
	}

	curl_easy_cleanup(curl);
	#else
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"No keylocation=%s back-end."), is_http ? "http://" : "https://");
	ret = ENOSYS;
	#endif

	end:
	if (ret == 0)
	ret = get_key_material_raw(key, keyformat, buf, len_out);

	if (key != NULL)
	fclose(key);

	return (ret);
	}

	/*
	* Attempts to fetch key material, no matter where it might live. The key
	* material is allocated and returned in km_out. *can_retry_out will be set
	* to B_TRUE if the user is providing the key material interactively, allowing
	* for re-entry attempts.
	*/
	static int
	get_key_material(libzfs_handle_t *hdl, boolean_t do_verify, boolean_t newkey,
	zfs_keyformat_t keyformat, char keylocation, const char fsname,
	uint8_t *km_out, size_t kmlen_out, boolean_t *can_retry_out)
	{
	int ret;
	zfs_keylocation_t keyloc = ZFS_KEYLOCATION_NONE;
	uint8_t *km = NULL;
	size_t kmlen = 0;
	char *uri_scheme = NULL;
	zfs_uri_handler_t *handler = NULL;
	boolean_t can_retry = B_FALSE;

	/* verify and parse the keylocation */
	ret = zfs_prop_parse_keylocation(hdl, keylocation, &keyloc,
	&uri_scheme);
	if (ret != 0)
	goto error;

	/* open the appropriate file descriptor */
	switch (keyloc) {
	case ZFS_KEYLOCATION_PROMPT:
	if (isatty(fileno(stdin))) {
	can_retry = keyformat != ZFS_KEYFORMAT_RAW;
	ret = get_key_interactive(hdl, fsname, keyformat,
	do_verify, newkey, &km, &kmlen);
	} else {
	/* fetch the key material into the buffer */
	ret = get_key_material_raw(stdin, keyformat, &km,
	&kmlen);
	}

	if (ret != 0)
	goto error;

	break;
	case ZFS_KEYLOCATION_URI:
	ret = ENOTSUP;

	for (handler = uri_handlers; handler->zuh_scheme != NULL;
	handler++) {
	if (strcmp(handler->zuh_scheme, uri_scheme) != 0)
	continue;

	if ((ret = handler->zuh_handler(hdl, keylocation,
	fsname, keyformat, newkey, &km, &kmlen)) != 0)
	goto error;

	break;
	}

	if (ret == ENOTSUP) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"URI scheme is not supported"));
	goto error;
	}

	break;
	default:
	ret = EINVAL;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Invalid keylocation."));
	goto error;
	}

	- if ((ret = validate_key(hdl, keyformat, (const char *)km, kmlen)) != 0)
	+ if ((ret = validate_key(hdl, keyformat, (const char *)km, kmlen,
	+ do_verify)) != 0)
	goto error;

	*km_out = km;
	*kmlen_out = kmlen;
	if (can_retry_out != NULL)
	*can_retry_out = can_retry;

	free(uri_scheme);
	return (0);

	error:
	free(km);

	*km_out = NULL;
	*kmlen_out = 0;

	if (can_retry_out != NULL)
	*can_retry_out = can_retry;

	free(uri_scheme);
	return (ret);
	}

	static int
	derive_key(libzfs_handle_t *hdl, zfs_keyformat_t format, uint64_t iters,
	uint8_t *key_material, size_t key_material_len, uint64_t salt,
	uint8_t **key_out)
	{
	int ret;
	uint8_t *key;

	*key_out = NULL;

	key = zfs_alloc(hdl, WRAPPING_KEY_LEN);
	if (!key)
	return (ENOMEM);

	switch (format) {
	case ZFS_KEYFORMAT_RAW:
	bcopy(key_material, key, WRAPPING_KEY_LEN);
	break;
	case ZFS_KEYFORMAT_HEX:
	ret = hex_key_to_raw((char *)key_material,
	WRAPPING_KEY_LEN * 2, key);
	if (ret != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Invalid hex key provided."));
	goto error;
	}
	break;
	case ZFS_KEYFORMAT_PASSPHRASE:
	salt = LE_64(salt);

	ret = PKCS5_PBKDF2_HMAC_SHA1((char *)key_material,
	strlen((char )key_material), ((uint8_t )&salt),
	sizeof (uint64_t), iters, WRAPPING_KEY_LEN, key);
	if (ret != 1) {
	ret = EIO;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Failed to generate key from passphrase."));
	goto error;
	}
	break;
	default:
	ret = EINVAL;
	goto error;
	}

	*key_out = key;
	return (0);

	error:
	free(key);

	*key_out = NULL;
	return (ret);
	}

	static boolean_t
	encryption_feature_is_enabled(zpool_handle_t *zph)
	{
	nvlist_t *features;
	uint64_t feat_refcount;

	/* check that features can be enabled */
	if (zpool_get_prop_int(zph, ZPOOL_PROP_VERSION, NULL)
	< SPA_VERSION_FEATURES)
	return (B_FALSE);

	/* check for crypto feature */
	features = zpool_get_features(zph);
	if (!features \|\| nvlist_lookup_uint64(features,
	spa_feature_table[SPA_FEATURE_ENCRYPTION].fi_guid,
	&feat_refcount) != 0)
	return (B_FALSE);

	return (B_TRUE);
	}

	static int
	populate_create_encryption_params_nvlists(libzfs_handle_t *hdl,
	zfs_handle_t *zhp, boolean_t newkey, zfs_keyformat_t keyformat,
	char keylocation, nvlist_t props, uint8_t *wkeydata, uint_t wkeylen)
	{
	int ret;
	uint64_t iters = 0, salt = 0;
	uint8_t *key_material = NULL;
	size_t key_material_len = 0;
	uint8_t *key_data = NULL;
	const char *fsname = (zhp) ? zfs_get_name(zhp) : NULL;

	/* get key material from keyformat and keylocation */
	ret = get_key_material(hdl, B_TRUE, newkey, keyformat, keylocation,
	fsname, &key_material, &key_material_len, NULL);
	if (ret != 0)
	goto error;

	/* passphrase formats require a salt and pbkdf2 iters property */
	if (keyformat == ZFS_KEYFORMAT_PASSPHRASE) {
	/* always generate a new salt */
	ret = pkcs11_get_urandom((uint8_t *)&salt, sizeof (uint64_t));
	if (ret != sizeof (uint64_t)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Failed to generate salt."));
	goto error;
	}

	ret = nvlist_add_uint64(props,
	zfs_prop_to_name(ZFS_PROP_PBKDF2_SALT), salt);
	if (ret != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Failed to add salt to properties."));
	goto error;
	}

	/*
	* If not otherwise specified, use the default number of
	* pbkdf2 iterations. If specified, we have already checked
	* that the given value is greater than MIN_PBKDF2_ITERATIONS
	* during zfs_valid_proplist().
	*/
	ret = nvlist_lookup_uint64(props,
	zfs_prop_to_name(ZFS_PROP_PBKDF2_ITERS), &iters);
	if (ret == ENOENT) {
	iters = DEFAULT_PBKDF2_ITERATIONS;
	ret = nvlist_add_uint64(props,
	zfs_prop_to_name(ZFS_PROP_PBKDF2_ITERS), iters);
	if (ret != 0)
	goto error;
	} else if (ret != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Failed to get pbkdf2 iterations."));
	goto error;
	}
	} else {
	/* check that pbkdf2iters was not specified by the user */
	ret = nvlist_lookup_uint64(props,
	zfs_prop_to_name(ZFS_PROP_PBKDF2_ITERS), &iters);
	if (ret == 0) {
	ret = EINVAL;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Cannot specify pbkdf2iters with a non-passphrase "
	"keyformat."));
	goto error;
	}
	}

	/* derive a key from the key material */
	ret = derive_key(hdl, keyformat, iters, key_material, key_material_len,
	salt, &key_data);
	if (ret != 0)
	goto error;

	free(key_material);

	*wkeydata = key_data;
	*wkeylen = WRAPPING_KEY_LEN;
	return (0);

	error:
	if (key_material != NULL)
	free(key_material);
	if (key_data != NULL)
	free(key_data);

	*wkeydata = NULL;
	*wkeylen = 0;
	return (ret);
	}

	static boolean_t
	proplist_has_encryption_props(nvlist_t *props)
	{
	int ret;
	uint64_t intval;
	char *strval;

	ret = nvlist_lookup_uint64(props,
	zfs_prop_to_name(ZFS_PROP_ENCRYPTION), &intval);
	if (ret == 0 && intval != ZIO_CRYPT_OFF)
	return (B_TRUE);

	ret = nvlist_lookup_string(props,
	zfs_prop_to_name(ZFS_PROP_KEYLOCATION), &strval);
	if (ret == 0 && strcmp(strval, "none") != 0)
	return (B_TRUE);

	ret = nvlist_lookup_uint64(props,
	zfs_prop_to_name(ZFS_PROP_KEYFORMAT), &intval);
	if (ret == 0)
	return (B_TRUE);

	ret = nvlist_lookup_uint64(props,
	zfs_prop_to_name(ZFS_PROP_PBKDF2_ITERS), &intval);
	if (ret == 0)
	return (B_TRUE);

	return (B_FALSE);
	}

	int
	zfs_crypto_get_encryption_root(zfs_handle_t zhp, boolean_t is_encroot,
	char *buf)
	{
	int ret;
	char prop_encroot[MAXNAMELEN];

	/* if the dataset isn't encrypted, just return */
	if (zfs_prop_get_int(zhp, ZFS_PROP_ENCRYPTION) == ZIO_CRYPT_OFF) {
	*is_encroot = B_FALSE;
	if (buf != NULL)
	buf[0] = '\0';
	return (0);
	}

	ret = zfs_prop_get(zhp, ZFS_PROP_ENCRYPTION_ROOT, prop_encroot,
	sizeof (prop_encroot), NULL, NULL, 0, B_TRUE);
	if (ret != 0) {
	*is_encroot = B_FALSE;
	if (buf != NULL)
	buf[0] = '\0';
	return (ret);
	}

	*is_encroot = strcmp(prop_encroot, zfs_get_name(zhp)) == 0;
	if (buf != NULL)
	strcpy(buf, prop_encroot);

	return (0);
	}

	int
	zfs_crypto_create(libzfs_handle_t hdl, char parent_name, nvlist_t *props,
	nvlist_t pool_props, boolean_t stdin_available, uint8_t *wkeydata_out,
	uint_t *wkeylen_out)
	{
	int ret;
	char errbuf[1024];
	uint64_t crypt = ZIO_CRYPT_INHERIT, pcrypt = ZIO_CRYPT_INHERIT;
	uint64_t keyformat = ZFS_KEYFORMAT_NONE;
	char *keylocation = NULL;
	zfs_handle_t *pzhp = NULL;
	uint8_t *wkeydata = NULL;
	uint_t wkeylen = 0;
	boolean_t local_crypt = B_TRUE;

	(void) snprintf(errbuf, sizeof (errbuf),
	dgettext(TEXT_DOMAIN, "Encryption create error"));

	/* lookup crypt from props */
	ret = nvlist_lookup_uint64(props,
	zfs_prop_to_name(ZFS_PROP_ENCRYPTION), &crypt);
	if (ret != 0)
	local_crypt = B_FALSE;

	/* lookup key location and format from props */
	(void) nvlist_lookup_uint64(props,
	zfs_prop_to_name(ZFS_PROP_KEYFORMAT), &keyformat);
	(void) nvlist_lookup_string(props,
	zfs_prop_to_name(ZFS_PROP_KEYLOCATION), &keylocation);

	if (parent_name != NULL) {
	/* get a reference to parent dataset */
	pzhp = make_dataset_handle(hdl, parent_name);
	if (pzhp == NULL) {
	ret = ENOENT;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Failed to lookup parent."));
	goto out;
	}

	/* Lookup parent's crypt */
	pcrypt = zfs_prop_get_int(pzhp, ZFS_PROP_ENCRYPTION);

	/* Params require the encryption feature */
	if (!encryption_feature_is_enabled(pzhp->zpool_hdl)) {
	if (proplist_has_encryption_props(props)) {
	ret = EINVAL;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Encryption feature not enabled."));
	goto out;
	}

	ret = 0;
	goto out;
	}
	} else {
	/*
	* special case for root dataset where encryption feature
	* feature won't be on disk yet
	*/
	if (!nvlist_exists(pool_props, "feature@encryption")) {
	if (proplist_has_encryption_props(props)) {
	ret = EINVAL;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Encryption feature not enabled."));
	goto out;
	}

	ret = 0;
	goto out;
	}

	pcrypt = ZIO_CRYPT_OFF;
	}

	/* Get the inherited encryption property if we don't have it locally */
	if (!local_crypt)
	crypt = pcrypt;

	/*
	* At this point crypt should be the actual encryption value. If
	* encryption is off just verify that no encryption properties have
	* been specified and return.
	*/
	if (crypt == ZIO_CRYPT_OFF) {
	if (proplist_has_encryption_props(props)) {
	ret = EINVAL;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Encryption must be turned on to set encryption "
	"properties."));
	goto out;
	}

	ret = 0;
	goto out;
	}

	/*
	* If we have a parent crypt it is valid to specify encryption alone.
	* This will result in a child that is encrypted with the chosen
	* encryption suite that will also inherit the parent's key. If
	* the parent is not encrypted we need an encryption suite provided.
	*/
	if (pcrypt == ZIO_CRYPT_OFF && keylocation == NULL &&
	keyformat == ZFS_KEYFORMAT_NONE) {
	ret = EINVAL;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Keyformat required for new encryption root."));
	goto out;
	}

	/*
	* Specifying a keylocation implies this will be a new encryption root.
	* Check that a keyformat is also specified.
	*/
	if (keylocation != NULL && keyformat == ZFS_KEYFORMAT_NONE) {
	ret = EINVAL;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Keyformat required for new encryption root."));
	goto out;
	}

	/* default to prompt if no keylocation is specified */
	if (keyformat != ZFS_KEYFORMAT_NONE && keylocation == NULL) {
	keylocation = "prompt";
	ret = nvlist_add_string(props,
	zfs_prop_to_name(ZFS_PROP_KEYLOCATION), keylocation);
	if (ret != 0)
	goto out;
	}

	/*
	* If a local key is provided, this dataset will be a new
	* encryption root. Populate the encryption params.
	*/
	if (keylocation != NULL) {
	/*
	* 'zfs recv -o keylocation=prompt' won't work because stdin
	* is being used by the send stream, so we disallow it.
	*/
	if (!stdin_available && strcmp(keylocation, "prompt") == 0) {
	ret = EINVAL;
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Cannot use "
	"'prompt' keylocation because stdin is in use."));
	goto out;
	}

	ret = populate_create_encryption_params_nvlists(hdl, NULL,
	B_TRUE, keyformat, keylocation, props, &wkeydata,
	&wkeylen);
	if (ret != 0)
	goto out;
	}

	if (pzhp != NULL)
	zfs_close(pzhp);

	*wkeydata_out = wkeydata;
	*wkeylen_out = wkeylen;
	return (0);

	out:
	if (pzhp != NULL)
	zfs_close(pzhp);
	if (wkeydata != NULL)
	free(wkeydata);

	*wkeydata_out = NULL;
	*wkeylen_out = 0;
	return (ret);
	}

	int
	zfs_crypto_clone_check(libzfs_handle_t hdl, zfs_handle_t origin_zhp,
	char parent_name, nvlist_t props)
	{
	char errbuf[1024];

	(void) snprintf(errbuf, sizeof (errbuf),
	dgettext(TEXT_DOMAIN, "Encryption clone error"));

	/*
	* No encryption properties should be specified. They will all be
	* inherited from the origin dataset.
	*/
	if (nvlist_exists(props, zfs_prop_to_name(ZFS_PROP_KEYFORMAT)) \|\|
	nvlist_exists(props, zfs_prop_to_name(ZFS_PROP_KEYLOCATION)) \|\|
	nvlist_exists(props, zfs_prop_to_name(ZFS_PROP_ENCRYPTION)) \|\|
	nvlist_exists(props, zfs_prop_to_name(ZFS_PROP_PBKDF2_ITERS))) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Encryption properties must inherit from origin dataset."));
	return (EINVAL);
	}

	return (0);
	}

	typedef struct loadkeys_cbdata {
	uint64_t cb_numfailed;
	uint64_t cb_numattempted;
	} loadkey_cbdata_t;

	static int
	load_keys_cb(zfs_handle_t zhp, void arg)
	{
	int ret;
	boolean_t is_encroot;
	loadkey_cbdata_t *cb = arg;
	uint64_t keystatus = zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS);

	/* only attempt to load keys for encryption roots */
	ret = zfs_crypto_get_encryption_root(zhp, &is_encroot, NULL);
	if (ret != 0 \|\| !is_encroot)
	goto out;

	/* don't attempt to load already loaded keys */
	if (keystatus == ZFS_KEYSTATUS_AVAILABLE)
	goto out;

	/* Attempt to load the key. Record status in cb. */
	cb->cb_numattempted++;

	ret = zfs_crypto_load_key(zhp, B_FALSE, NULL);
	if (ret)
	cb->cb_numfailed++;

	out:
	(void) zfs_iter_filesystems(zhp, load_keys_cb, cb);
	zfs_close(zhp);

	/* always return 0, since this function is best effort */
	return (0);
	}

	/*
	* This function is best effort. It attempts to load all the keys for the given
	* filesystem and all of its children.
	*/
	int
	zfs_crypto_attempt_load_keys(libzfs_handle_t hdl, char fsname)
	{
	int ret;
	zfs_handle_t *zhp = NULL;
	loadkey_cbdata_t cb = { 0 };

	zhp = zfs_open(hdl, fsname, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME);
	if (zhp == NULL) {
	ret = ENOENT;
	goto error;
	}

	ret = load_keys_cb(zfs_handle_dup(zhp), &cb);
	if (ret)
	goto error;

	(void) printf(gettext("%llu / %llu keys successfully loaded\n"),
	(u_longlong_t)(cb.cb_numattempted - cb.cb_numfailed),
	(u_longlong_t)cb.cb_numattempted);

	if (cb.cb_numfailed != 0) {
	ret = -1;
	goto error;
	}

	zfs_close(zhp);
	return (0);

	error:
	if (zhp != NULL)
	zfs_close(zhp);
	return (ret);
	}

	int
	zfs_crypto_load_key(zfs_handle_t zhp, boolean_t noop, char alt_keylocation)
	{
	int ret, attempts = 0;
	char errbuf[1024];
	uint64_t keystatus, iters = 0, salt = 0;
	uint64_t keyformat = ZFS_KEYFORMAT_NONE;
	char prop_keylocation[MAXNAMELEN];
	char prop_encroot[MAXNAMELEN];
	char *keylocation = NULL;
	uint8_t key_material = NULL, key_data = NULL;
	size_t key_material_len;
	boolean_t is_encroot, can_retry = B_FALSE, correctible = B_FALSE;

	(void) snprintf(errbuf, sizeof (errbuf),
	dgettext(TEXT_DOMAIN, "Key load error"));

	/* check that encryption is enabled for the pool */
	if (!encryption_feature_is_enabled(zhp->zpool_hdl)) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Encryption feature not enabled."));
	ret = EINVAL;
	goto error;
	}

	/* Fetch the keyformat. Check that the dataset is encrypted. */
	keyformat = zfs_prop_get_int(zhp, ZFS_PROP_KEYFORMAT);
	if (keyformat == ZFS_KEYFORMAT_NONE) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"'%s' is not encrypted."), zfs_get_name(zhp));
	ret = EINVAL;
	goto error;
	}

	/*
	* Fetch the key location. Check that we are working with an
	* encryption root.
	*/
	ret = zfs_crypto_get_encryption_root(zhp, &is_encroot, prop_encroot);
	if (ret != 0) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Failed to get encryption root for '%s'."),
	zfs_get_name(zhp));
	goto error;
	} else if (!is_encroot) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Keys must be loaded for encryption root of '%s' (%s)."),
	zfs_get_name(zhp), prop_encroot);
	ret = EINVAL;
	goto error;
	}

	/*
	* if the caller has elected to override the keylocation property
	* use that instead
	*/
	if (alt_keylocation != NULL) {
	keylocation = alt_keylocation;
	} else {
	ret = zfs_prop_get(zhp, ZFS_PROP_KEYLOCATION, prop_keylocation,
	sizeof (prop_keylocation), NULL, NULL, 0, B_TRUE);
	if (ret != 0) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Failed to get keylocation for '%s'."),
	zfs_get_name(zhp));
	goto error;
	}

	keylocation = prop_keylocation;
	}

	/* check that the key is unloaded unless this is a noop */
	if (!noop) {
	keystatus = zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS);
	if (keystatus == ZFS_KEYSTATUS_AVAILABLE) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Key already loaded for '%s'."), zfs_get_name(zhp));
	ret = EEXIST;
	goto error;
	}
	}

	/* passphrase formats require a salt and pbkdf2_iters property */
	if (keyformat == ZFS_KEYFORMAT_PASSPHRASE) {
	salt = zfs_prop_get_int(zhp, ZFS_PROP_PBKDF2_SALT);
	iters = zfs_prop_get_int(zhp, ZFS_PROP_PBKDF2_ITERS);
	}

	try_again:
	/* fetching and deriving the key are correctable errors. set the flag */
	correctible = B_TRUE;

	/* get key material from key format and location */
	ret = get_key_material(zhp->zfs_hdl, B_FALSE, B_FALSE, keyformat,
	keylocation, zfs_get_name(zhp), &key_material, &key_material_len,
	&can_retry);
	if (ret != 0)
	goto error;

	/* derive a key from the key material */
	ret = derive_key(zhp->zfs_hdl, keyformat, iters, key_material,
	key_material_len, salt, &key_data);
	if (ret != 0)
	goto error;

	correctible = B_FALSE;

	/* pass the wrapping key and noop flag to the ioctl */
	ret = lzc_load_key(zhp->zfs_name, noop, key_data, WRAPPING_KEY_LEN);
	if (ret != 0) {
	switch (ret) {
	case EPERM:
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Permission denied."));
	break;
	case EINVAL:
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Invalid parameters provided for dataset %s."),
	zfs_get_name(zhp));
	break;
	case EEXIST:
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Key already loaded for '%s'."), zfs_get_name(zhp));
	break;
	case EBUSY:
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"'%s' is busy."), zfs_get_name(zhp));
	break;
	case EACCES:
	correctible = B_TRUE;
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Incorrect key provided for '%s'."),
	zfs_get_name(zhp));
	break;
	}
	goto error;
	}

	free(key_material);
	free(key_data);

	return (0);

	error:
	zfs_error(zhp->zfs_hdl, EZFS_CRYPTOFAILED, errbuf);
	if (key_material != NULL) {
	free(key_material);
	key_material = NULL;
	}
	if (key_data != NULL) {
	free(key_data);
	key_data = NULL;
	}

	/*
	* Here we decide if it is ok to allow the user to retry entering their
	* key. The can_retry flag will be set if the user is entering their
	* key from an interactive prompt. The correctable flag will only be
	* set if an error that occurred could be corrected by retrying. Both
	* flags are needed to allow the user to attempt key entry again
	*/
	attempts++;
	if (can_retry && correctible && attempts < MAX_KEY_PROMPT_ATTEMPTS)
	goto try_again;

	return (ret);
	}

	int
	zfs_crypto_unload_key(zfs_handle_t *zhp)
	{
	int ret;
	char errbuf[1024];
	char prop_encroot[MAXNAMELEN];
	uint64_t keystatus, keyformat;
	boolean_t is_encroot;

	(void) snprintf(errbuf, sizeof (errbuf),
	dgettext(TEXT_DOMAIN, "Key unload error"));

	/* check that encryption is enabled for the pool */
	if (!encryption_feature_is_enabled(zhp->zpool_hdl)) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Encryption feature not enabled."));
	ret = EINVAL;
	goto error;
	}

	/* Fetch the keyformat. Check that the dataset is encrypted. */
	keyformat = zfs_prop_get_int(zhp, ZFS_PROP_KEYFORMAT);
	if (keyformat == ZFS_KEYFORMAT_NONE) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"'%s' is not encrypted."), zfs_get_name(zhp));
	ret = EINVAL;
	goto error;
	}

	/*
	* Fetch the key location. Check that we are working with an
	* encryption root.
	*/
	ret = zfs_crypto_get_encryption_root(zhp, &is_encroot, prop_encroot);
	if (ret != 0) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Failed to get encryption root for '%s'."),
	zfs_get_name(zhp));
	goto error;
	} else if (!is_encroot) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Keys must be unloaded for encryption root of '%s' (%s)."),
	zfs_get_name(zhp), prop_encroot);
	ret = EINVAL;
	goto error;
	}

	/* check that the key is loaded */
	keystatus = zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS);
	if (keystatus == ZFS_KEYSTATUS_UNAVAILABLE) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Key already unloaded for '%s'."), zfs_get_name(zhp));
	ret = EACCES;
	goto error;
	}

	/* call the ioctl */
	ret = lzc_unload_key(zhp->zfs_name);

	if (ret != 0) {
	switch (ret) {
	case EPERM:
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Permission denied."));
	break;
	case EACCES:
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Key already unloaded for '%s'."),
	zfs_get_name(zhp));
	break;
	case EBUSY:
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"'%s' is busy."), zfs_get_name(zhp));
	break;
	}
	zfs_error(zhp->zfs_hdl, EZFS_CRYPTOFAILED, errbuf);
	}

	return (ret);

	error:
	zfs_error(zhp->zfs_hdl, EZFS_CRYPTOFAILED, errbuf);
	return (ret);
	}

	static int
	zfs_crypto_verify_rewrap_nvlist(zfs_handle_t zhp, nvlist_t props,
	nvlist_t *props_out, char errbuf)
	{
	int ret;
	nvpair_t *elem = NULL;
	zfs_prop_t prop;
	nvlist_t *new_props = NULL;

	new_props = fnvlist_alloc();

	/*
	* loop through all provided properties, we should only have
	* keyformat, keylocation and pbkdf2iters. The actual validation of
	* values is done by zfs_valid_proplist().
	*/
	while ((elem = nvlist_next_nvpair(props, elem)) != NULL) {
	const char *propname = nvpair_name(elem);
	prop = zfs_name_to_prop(propname);

	switch (prop) {
	case ZFS_PROP_PBKDF2_ITERS:
	case ZFS_PROP_KEYFORMAT:
	case ZFS_PROP_KEYLOCATION:
	break;
	default:
	ret = EINVAL;
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Only keyformat, keylocation and pbkdf2iters may "
	"be set with this command."));
	goto error;
	}
	}

	new_props = zfs_valid_proplist(zhp->zfs_hdl, zhp->zfs_type, props,
	zfs_prop_get_int(zhp, ZFS_PROP_ZONED), NULL, zhp->zpool_hdl,
	B_TRUE, errbuf);
	if (new_props == NULL) {
	ret = EINVAL;
	goto error;
	}

	*props_out = new_props;
	return (0);

	error:
	nvlist_free(new_props);
	*props_out = NULL;
	return (ret);
	}

	int
	zfs_crypto_rewrap(zfs_handle_t zhp, nvlist_t raw_props, boolean_t inheritkey)
	{
	int ret;
	char errbuf[1024];
	boolean_t is_encroot;
	nvlist_t *props = NULL;
	uint8_t *wkeydata = NULL;
	uint_t wkeylen = 0;
	dcp_cmd_t cmd = (inheritkey) ? DCP_CMD_INHERIT : DCP_CMD_NEW_KEY;
	uint64_t crypt, pcrypt, keystatus, pkeystatus;
	uint64_t keyformat = ZFS_KEYFORMAT_NONE;
	zfs_handle_t *pzhp = NULL;
	char *keylocation = NULL;
	char origin_name[MAXNAMELEN];
	char prop_keylocation[MAXNAMELEN];
	char parent_name[ZFS_MAX_DATASET_NAME_LEN];

	(void) snprintf(errbuf, sizeof (errbuf),
	dgettext(TEXT_DOMAIN, "Key change error"));

	/* check that encryption is enabled for the pool */
	if (!encryption_feature_is_enabled(zhp->zpool_hdl)) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Encryption feature not enabled."));
	ret = EINVAL;
	goto error;
	}

	/* get crypt from dataset */
	crypt = zfs_prop_get_int(zhp, ZFS_PROP_ENCRYPTION);
	if (crypt == ZIO_CRYPT_OFF) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Dataset not encrypted."));
	ret = EINVAL;
	goto error;
	}

	/* get the encryption root of the dataset */
	ret = zfs_crypto_get_encryption_root(zhp, &is_encroot, NULL);
	if (ret != 0) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Failed to get encryption root for '%s'."),
	zfs_get_name(zhp));
	goto error;
	}

	/* Clones use their origin's key and cannot rewrap it */
	ret = zfs_prop_get(zhp, ZFS_PROP_ORIGIN, origin_name,
	sizeof (origin_name), NULL, NULL, 0, B_TRUE);
	if (ret == 0 && strcmp(origin_name, "") != 0) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Keys cannot be changed on clones."));
	ret = EINVAL;
	goto error;
	}

	/*
	* If the user wants to use the inheritkey variant of this function
	* we don't need to collect any crypto arguments.
	*/
	if (!inheritkey) {
	/* validate the provided properties */
	ret = zfs_crypto_verify_rewrap_nvlist(zhp, raw_props, &props,
	errbuf);
	if (ret != 0)
	goto error;

	/*
	* Load keyformat and keylocation from the nvlist. Fetch from
	* the dataset properties if not specified.
	*/
	(void) nvlist_lookup_uint64(props,
	zfs_prop_to_name(ZFS_PROP_KEYFORMAT), &keyformat);
	(void) nvlist_lookup_string(props,
	zfs_prop_to_name(ZFS_PROP_KEYLOCATION), &keylocation);

	if (is_encroot) {
	/*
	* If this is already an encryption root, just keep
	* any properties not set by the user.
	*/
	if (keyformat == ZFS_KEYFORMAT_NONE) {
	keyformat = zfs_prop_get_int(zhp,
	ZFS_PROP_KEYFORMAT);
	ret = nvlist_add_uint64(props,
	zfs_prop_to_name(ZFS_PROP_KEYFORMAT),
	keyformat);
	if (ret != 0) {
	zfs_error_aux(zhp->zfs_hdl,
	dgettext(TEXT_DOMAIN, "Failed to "
	"get existing keyformat "
	"property."));
	goto error;
	}
	}

	if (keylocation == NULL) {
	ret = zfs_prop_get(zhp, ZFS_PROP_KEYLOCATION,
	prop_keylocation, sizeof (prop_keylocation),
	NULL, NULL, 0, B_TRUE);
	if (ret != 0) {
	zfs_error_aux(zhp->zfs_hdl,
	dgettext(TEXT_DOMAIN, "Failed to "
	"get existing keylocation "
	"property."));
	goto error;
	}

	keylocation = prop_keylocation;
	}
	} else {
	/* need a new key for non-encryption roots */
	if (keyformat == ZFS_KEYFORMAT_NONE) {
	ret = EINVAL;
	zfs_error_aux(zhp->zfs_hdl,
	dgettext(TEXT_DOMAIN, "Keyformat required "
	"for new encryption root."));
	goto error;
	}

	/* default to prompt if no keylocation is specified */
	if (keylocation == NULL) {
	keylocation = "prompt";
	ret = nvlist_add_string(props,
	zfs_prop_to_name(ZFS_PROP_KEYLOCATION),
	keylocation);
	if (ret != 0)
	goto error;
	}
	}

	/* fetch the new wrapping key and associated properties */
	ret = populate_create_encryption_params_nvlists(zhp->zfs_hdl,
	zhp, B_TRUE, keyformat, keylocation, props, &wkeydata,
	&wkeylen);
	if (ret != 0)
	goto error;
	} else {
	/* check that zhp is an encryption root */
	if (!is_encroot) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Key inheritting can only be performed on "
	"encryption roots."));
	ret = EINVAL;
	goto error;
	}

	/* get the parent's name */
	ret = zfs_parent_name(zhp, parent_name, sizeof (parent_name));
	if (ret != 0) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Root dataset cannot inherit key."));
	ret = EINVAL;
	goto error;
	}

	/* get a handle to the parent */
	pzhp = make_dataset_handle(zhp->zfs_hdl, parent_name);
	if (pzhp == NULL) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Failed to lookup parent."));
	ret = ENOENT;
	goto error;
	}

	/* parent must be encrypted */
	pcrypt = zfs_prop_get_int(pzhp, ZFS_PROP_ENCRYPTION);
	if (pcrypt == ZIO_CRYPT_OFF) {
	zfs_error_aux(pzhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Parent must be encrypted."));
	ret = EINVAL;
	goto error;
	}

	/* check that the parent's key is loaded */
	pkeystatus = zfs_prop_get_int(pzhp, ZFS_PROP_KEYSTATUS);
	if (pkeystatus == ZFS_KEYSTATUS_UNAVAILABLE) {
	zfs_error_aux(pzhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Parent key must be loaded."));
	ret = EACCES;
	goto error;
	}
	}

	/* check that the key is loaded */
	keystatus = zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS);
	if (keystatus == ZFS_KEYSTATUS_UNAVAILABLE) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Key must be loaded."));
	ret = EACCES;
	goto error;
	}

	/* call the ioctl */
	ret = lzc_change_key(zhp->zfs_name, cmd, props, wkeydata, wkeylen);
	if (ret != 0) {
	switch (ret) {
	case EPERM:
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Permission denied."));
	break;
	case EINVAL:
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Invalid properties for key change."));
	break;
	case EACCES:
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"Key is not currently loaded."));
	break;
	}
	zfs_error(zhp->zfs_hdl, EZFS_CRYPTOFAILED, errbuf);
	}

	if (pzhp != NULL)
	zfs_close(pzhp);
	if (props != NULL)
	nvlist_free(props);
	if (wkeydata != NULL)
	free(wkeydata);

	return (ret);

	error:
	if (pzhp != NULL)
	zfs_close(pzhp);
	if (props != NULL)
	nvlist_free(props);
	if (wkeydata != NULL)
	free(wkeydata);

	zfs_error(zhp->zfs_hdl, EZFS_CRYPTOFAILED, errbuf);
	return (ret);
	}
	diff --git a/sys/contrib/openzfs/lib/libzfs/libzfs_diff.c b/sys/contrib/openzfs/lib/libzfs/libzfs_diff.c
	index d46e23a2fc0e..8884ac536a03 100644
	--- a/sys/contrib/openzfs/lib/libzfs/libzfs_diff.c
	+++ b/sys/contrib/openzfs/lib/libzfs/libzfs_diff.c
	@@ -1,809 +1,792 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright 2015 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2015, 2018 by Delphix. All rights reserved.
	* Copyright 2016 Joyent, Inc.
	* Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>
	*/

	/*
	* zfs diff support
	*/
	#include <ctype.h>
	#include <errno.h>
	#include <libintl.h>
	#include <string.h>
	#include <sys/types.h>
	#include <sys/stat.h>
	#include <fcntl.h>
	#include <stddef.h>
	#include <unistd.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <stropts.h>
	#include <pthread.h>
	#include <sys/zfs_ioctl.h>
	#include <libzfs.h>
	#include "libzfs_impl.h"

	#define ZDIFF_SNAPDIR "/.zfs/snapshot/"
	#define ZDIFF_PREFIX "zfs-diff-%d"

	#define ZDIFF_ADDED '+'
	-#define ZDIFF_MODIFIED 'M'
	+#define ZDIFF_MODIFIED "M"
	#define ZDIFF_REMOVED '-'
	-#define ZDIFF_RENAMED 'R'
	+#define ZDIFF_RENAMED "R"


	/*
	* Given a {dsname, object id}, get the object path
	*/
	static int
	get_stats_for_obj(differ_info_t di, const char dsname, uint64_t obj,
	char pn, int maxlen, zfs_stat_t sb)
	{
	zfs_cmd_t zc = {"\0"};
	int error;

	(void) strlcpy(zc.zc_name, dsname, sizeof (zc.zc_name));
	zc.zc_obj = obj;

	errno = 0;
	error = zfs_ioctl(di->zhp->zfs_hdl, ZFS_IOC_OBJ_TO_STATS, &zc);
	di->zerr = errno;

	/* we can get stats even if we failed to get a path */
	(void) memcpy(sb, &zc.zc_stat, sizeof (zfs_stat_t));
	if (error == 0) {
	ASSERT(di->zerr == 0);
	(void) strlcpy(pn, zc.zc_value, maxlen);
	return (0);
	}

	if (di->zerr == ESTALE) {
	(void) snprintf(pn, maxlen, "(on_delete_queue)");
	return (0);
	} else if (di->zerr == EPERM) {
	(void) snprintf(di->errbuf, sizeof (di->errbuf),
	dgettext(TEXT_DOMAIN,
	"The sys_config privilege or diff delegated permission "
	"is needed\nto discover path names"));
	return (-1);
	} else if (di->zerr == EACCES) {
	(void) snprintf(di->errbuf, sizeof (di->errbuf),
	dgettext(TEXT_DOMAIN,
	"Key must be loaded to discover path names"));
	return (-1);
	} else {
	(void) snprintf(di->errbuf, sizeof (di->errbuf),
	dgettext(TEXT_DOMAIN,
	"Unable to determine path or stats for "
	"object %lld in %s"), (longlong_t)obj, dsname);
	return (-1);
	}
	}

	/*
	* stream_bytes
	*
	* Prints a file name out a character at a time. If the character is
	* not in the range of what we consider "printable" ASCII, display it
	* as an escaped 4-digit octal value. ASCII values less than a space
	* are all control characters and we declare the upper end as the
	* DELete character. This also is the last 7-bit ASCII character.
	* We choose to treat all 8-bit ASCII as not printable for this
	* application.
	*/
	static void
	stream_bytes(FILE fp, const char string)
	{
	char c;

	while ((c = *string++) != '\0') {
	if (c > ' ' && c != '\\' && c < '\177') {
	- (void) fprintf(fp, "%c", c);
	+ (void) fputc(c, fp);
	} else {
	- (void) fprintf(fp, "\\%04o", (uint8_t)c);
	+ (void) fprintf(fp, "\\%04hho", (uint8_t)c);
	}
	}
	}

	-static void
	-print_what(FILE *fp, mode_t what)
	+static char
	+get_what(mode_t what)
	{
	- char symbol;
	-
	switch (what & S_IFMT) {
	case S_IFBLK:
	- symbol = 'B';
	- break;
	+ return ('B');
	case S_IFCHR:
	- symbol = 'C';
	- break;
	+ return ('C');
	case S_IFDIR:
	- symbol = '/';
	- break;
	+ return ('/');
	#ifdef S_IFDOOR
	case S_IFDOOR:
	- symbol = '>';
	- break;
	+ return ('>');
	#endif
	case S_IFIFO:
	- symbol = '\|';
	- break;
	+ return ('\|');
	case S_IFLNK:
	- symbol = '@';
	- break;
	+ return ('@');
	#ifdef S_IFPORT
	case S_IFPORT:
	- symbol = 'P';
	- break;
	+ return ('P');
	#endif
	case S_IFSOCK:
	- symbol = '=';
	- break;
	+ return ('=');
	case S_IFREG:
	- symbol = 'F';
	- break;
	+ return ('F');
	default:
	- symbol = '?';
	- break;
	+ return ('?');
	}
	- (void) fprintf(fp, "%c", symbol);
	}

	static void
	print_cmn(FILE fp, differ_info_t di, const char *file)
	{
	- stream_bytes(fp, di->dsmnt);
	- stream_bytes(fp, file);
	+ if (!di->no_mangle) {
	+ stream_bytes(fp, di->dsmnt);
	+ stream_bytes(fp, file);
	+ } else {
	+ (void) fputs(di->dsmnt, fp);
	+ (void) fputs(file, fp);
	+ }
	}

	static void
	print_rename(FILE fp, differ_info_t di, const char old, const char new,
	zfs_stat_t *isb)
	{
	if (di->timestamped)
	(void) fprintf(fp, "%10lld.%09lld\t",
	(longlong_t)isb->zs_ctime[0],
	(longlong_t)isb->zs_ctime[1]);
	- (void) fprintf(fp, "%c\t", ZDIFF_RENAMED);
	- if (di->classify) {
	- print_what(fp, isb->zs_mode);
	- (void) fprintf(fp, "\t");
	- }
	+ (void) fputs(ZDIFF_RENAMED "\t", fp);
	+ if (di->classify)
	+ (void) fprintf(fp, "%c\t", get_what(isb->zs_mode));
	print_cmn(fp, di, old);
	- if (di->scripted)
	- (void) fprintf(fp, "\t");
	- else
	- (void) fprintf(fp, " -> ");
	+ (void) fputs(di->scripted ? "\t" : " -> ", fp);
	print_cmn(fp, di, new);
	- (void) fprintf(fp, "\n");
	+ (void) fputc('\n', fp);
	}

	static void
	print_link_change(FILE fp, differ_info_t di, int delta, const char *file,
	zfs_stat_t *isb)
	{
	if (di->timestamped)
	(void) fprintf(fp, "%10lld.%09lld\t",
	(longlong_t)isb->zs_ctime[0],
	(longlong_t)isb->zs_ctime[1]);
	- (void) fprintf(fp, "%c\t", ZDIFF_MODIFIED);
	- if (di->classify) {
	- print_what(fp, isb->zs_mode);
	- (void) fprintf(fp, "\t");
	- }
	+ (void) fputs(ZDIFF_MODIFIED "\t", fp);
	+ if (di->classify)
	+ (void) fprintf(fp, "%c\t", get_what(isb->zs_mode));
	print_cmn(fp, di, file);
	- (void) fprintf(fp, "\t(%+d)", delta);
	- (void) fprintf(fp, "\n");
	+ (void) fprintf(fp, "\t(%+d)\n", delta);
	}

	static void
	print_file(FILE fp, differ_info_t di, char type, const char *file,
	zfs_stat_t *isb)
	{
	if (di->timestamped)
	(void) fprintf(fp, "%10lld.%09lld\t",
	(longlong_t)isb->zs_ctime[0],
	(longlong_t)isb->zs_ctime[1]);
	(void) fprintf(fp, "%c\t", type);
	- if (di->classify) {
	- print_what(fp, isb->zs_mode);
	- (void) fprintf(fp, "\t");
	- }
	+ if (di->classify)
	+ (void) fprintf(fp, "%c\t", get_what(isb->zs_mode));
	print_cmn(fp, di, file);
	- (void) fprintf(fp, "\n");
	+ (void) fputc('\n', fp);
	}

	static int
	write_inuse_diffs_one(FILE fp, differ_info_t di, uint64_t dobj)
	{
	struct zfs_stat fsb, tsb;
	mode_t fmode, tmode;
	char fobjname[MAXPATHLEN], tobjname[MAXPATHLEN];
	boolean_t already_logged = B_FALSE;
	int fobjerr, tobjerr;
	int change;

	if (dobj == di->shares)
	return (0);

	/*
	* Check the from and to snapshots for info on the object. If
	* we get ENOENT, then the object just didn't exist in that
	* snapshot. If we get ENOTSUP, then we tried to get
	* info on a non-ZPL object, which we don't care about anyway.
	* For any other error we print a warning which includes the
	* errno and continue.
	*/

	fobjerr = get_stats_for_obj(di, di->fromsnap, dobj, fobjname,
	MAXPATHLEN, &fsb);
	if (fobjerr && di->zerr != ENOTSUP && di->zerr != ENOENT) {
	zfs_error_aux(di->zhp->zfs_hdl, "%s", strerror(di->zerr));
	zfs_error(di->zhp->zfs_hdl, di->zerr, di->errbuf);
	/*
	* Let's not print an error for the same object more than
	* once if it happens in both snapshots
	*/
	already_logged = B_TRUE;
	}

	tobjerr = get_stats_for_obj(di, di->tosnap, dobj, tobjname,
	MAXPATHLEN, &tsb);

	if (tobjerr && di->zerr != ENOTSUP && di->zerr != ENOENT) {
	if (!already_logged) {
	zfs_error_aux(di->zhp->zfs_hdl,
	"%s", strerror(di->zerr));
	zfs_error(di->zhp->zfs_hdl, di->zerr, di->errbuf);
	}
	}
	/*
	* Unallocated object sharing the same meta dnode block
	*/
	if (fobjerr && tobjerr) {
	di->zerr = 0;
	return (0);
	}

	di->zerr = 0; /* negate get_stats_for_obj() from side that failed */
	fmode = fsb.zs_mode & S_IFMT;
	tmode = tsb.zs_mode & S_IFMT;
	if (fmode == S_IFDIR \|\| tmode == S_IFDIR \|\| fsb.zs_links == 0 \|\|
	tsb.zs_links == 0)
	change = 0;
	else
	change = tsb.zs_links - fsb.zs_links;

	if (fobjerr) {
	if (change) {
	print_link_change(fp, di, change, tobjname, &tsb);
	return (0);
	}
	print_file(fp, di, ZDIFF_ADDED, tobjname, &tsb);
	return (0);
	} else if (tobjerr) {
	if (change) {
	print_link_change(fp, di, change, fobjname, &fsb);
	return (0);
	}
	print_file(fp, di, ZDIFF_REMOVED, fobjname, &fsb);
	return (0);
	}

	if (fmode != tmode && fsb.zs_gen == tsb.zs_gen)
	tsb.zs_gen++; /* Force a generational difference */

	/* Simple modification or no change */
	if (fsb.zs_gen == tsb.zs_gen) {
	/* No apparent changes. Could we assert !this? */
	if (fsb.zs_ctime[0] == tsb.zs_ctime[0] &&
	fsb.zs_ctime[1] == tsb.zs_ctime[1])
	return (0);
	if (change) {
	print_link_change(fp, di, change,
	change > 0 ? fobjname : tobjname, &tsb);
	} else if (strcmp(fobjname, tobjname) == 0) {
	- print_file(fp, di, ZDIFF_MODIFIED, fobjname, &tsb);
	+ print_file(fp, di, *ZDIFF_MODIFIED, fobjname, &tsb);
	} else {
	print_rename(fp, di, fobjname, tobjname, &tsb);
	}
	return (0);
	} else {
	/* file re-created or object re-used */
	print_file(fp, di, ZDIFF_REMOVED, fobjname, &fsb);
	print_file(fp, di, ZDIFF_ADDED, tobjname, &tsb);
	return (0);
	}
	}

	static int
	write_inuse_diffs(FILE fp, differ_info_t di, dmu_diff_record_t *dr)
	{
	uint64_t o;
	int err;

	for (o = dr->ddr_first; o <= dr->ddr_last; o++) {
	if ((err = write_inuse_diffs_one(fp, di, o)) != 0)
	return (err);
	}
	return (0);
	}

	static int
	describe_free(FILE fp, differ_info_t di, uint64_t object, char *namebuf,
	int maxlen)
	{
	struct zfs_stat sb;

	(void) get_stats_for_obj(di, di->fromsnap, object, namebuf,
	maxlen, &sb);

	/* Don't print if in the delete queue on from side */
	if (di->zerr == ESTALE \|\| di->zerr == ENOENT) {
	di->zerr = 0;
	return (0);
	}

	print_file(fp, di, ZDIFF_REMOVED, namebuf, &sb);
	return (0);
	}

	static int
	write_free_diffs(FILE fp, differ_info_t di, dmu_diff_record_t *dr)
	{
	zfs_cmd_t zc = {"\0"};
	libzfs_handle_t *lhdl = di->zhp->zfs_hdl;
	char fobjname[MAXPATHLEN];

	(void) strlcpy(zc.zc_name, di->fromsnap, sizeof (zc.zc_name));
	zc.zc_obj = dr->ddr_first - 1;

	ASSERT(di->zerr == 0);

	while (zc.zc_obj < dr->ddr_last) {
	int err;

	err = zfs_ioctl(lhdl, ZFS_IOC_NEXT_OBJ, &zc);
	if (err == 0) {
	if (zc.zc_obj == di->shares) {
	zc.zc_obj++;
	continue;
	}
	if (zc.zc_obj > dr->ddr_last) {
	break;
	}
	err = describe_free(fp, di, zc.zc_obj, fobjname,
	MAXPATHLEN);
	} else if (errno == ESRCH) {
	break;
	} else {
	(void) snprintf(di->errbuf, sizeof (di->errbuf),
	dgettext(TEXT_DOMAIN,
	"next allocated object (> %lld) find failure"),
	(longlong_t)zc.zc_obj);
	di->zerr = errno;
	break;
	}
	}
	if (di->zerr)
	return (-1);
	return (0);
	}

	static void *
	differ(void *arg)
	{
	differ_info_t *di = arg;
	dmu_diff_record_t dr;
	FILE *ofp;
	int err = 0;

	if ((ofp = fdopen(di->outputfd, "w")) == NULL) {
	di->zerr = errno;
	strlcpy(di->errbuf, strerror(errno), sizeof (di->errbuf));
	(void) close(di->datafd);
	return ((void *)-1);
	}

	for (;;) {
	char cp = (char )&dr;
	int len = sizeof (dr);
	int rv;

	do {
	rv = read(di->datafd, cp, len);
	cp += rv;
	len -= rv;
	} while (len > 0 && rv > 0);

	if (rv < 0 \|\| (rv == 0 && len != sizeof (dr))) {
	di->zerr = EPIPE;
	break;
	} else if (rv == 0) {
	/* end of file at a natural breaking point */
	break;
	}

	switch (dr.ddr_type) {
	case DDR_FREE:
	err = write_free_diffs(ofp, di, &dr);
	break;
	case DDR_INUSE:
	err = write_inuse_diffs(ofp, di, &dr);
	break;
	default:
	di->zerr = EPIPE;
	break;
	}

	if (err \|\| di->zerr)
	break;
	}

	(void) fclose(ofp);
	(void) close(di->datafd);
	if (err)
	return ((void *)-1);
	if (di->zerr) {
	ASSERT(di->zerr == EPIPE);
	(void) snprintf(di->errbuf, sizeof (di->errbuf),
	dgettext(TEXT_DOMAIN,
	"Internal error: bad data from diff IOCTL"));
	return ((void *)-1);
	}
	return ((void *)0);
	}

	static int
	make_temp_snapshot(differ_info_t *di)
	{
	libzfs_handle_t *hdl = di->zhp->zfs_hdl;
	zfs_cmd_t zc = {"\0"};

	(void) snprintf(zc.zc_value, sizeof (zc.zc_value),
	ZDIFF_PREFIX, getpid());
	(void) strlcpy(zc.zc_name, di->ds, sizeof (zc.zc_name));
	zc.zc_cleanup_fd = di->cleanupfd;

	if (zfs_ioctl(hdl, ZFS_IOC_TMP_SNAPSHOT, &zc) != 0) {
	int err = errno;
	if (err == EPERM) {
	(void) snprintf(di->errbuf, sizeof (di->errbuf),
	dgettext(TEXT_DOMAIN, "The diff delegated "
	"permission is needed in order\nto create a "
	"just-in-time snapshot for diffing\n"));
	return (zfs_error(hdl, EZFS_DIFF, di->errbuf));
	} else {
	(void) snprintf(di->errbuf, sizeof (di->errbuf),
	dgettext(TEXT_DOMAIN, "Cannot create just-in-time "
	"snapshot of '%s'"), zc.zc_name);
	return (zfs_standard_error(hdl, err, di->errbuf));
	}
	}

	di->tmpsnap = zfs_strdup(hdl, zc.zc_value);
	di->tosnap = zfs_asprintf(hdl, "%s@%s", di->ds, di->tmpsnap);
	return (0);
	}

	static void
	teardown_differ_info(differ_info_t *di)
	{
	free(di->ds);
	free(di->dsmnt);
	free(di->fromsnap);
	free(di->frommnt);
	free(di->tosnap);
	free(di->tmpsnap);
	free(di->tomnt);
	(void) close(di->cleanupfd);
	}

	static int
	get_snapshot_names(differ_info_t di, const char fromsnap,
	const char *tosnap)
	{
	libzfs_handle_t *hdl = di->zhp->zfs_hdl;
	char *atptrf = NULL;
	char *atptrt = NULL;
	int fdslen, fsnlen;
	int tdslen, tsnlen;

	/*
	* Can accept
	* fdslen fsnlen tdslen tsnlen
	* dataset@snap1
	* 0. dataset@snap1 dataset@snap2 >0 >1 >0 >1
	* 1. dataset@snap1 @snap2 >0 >1 ==0 >1
	* 2. dataset@snap1 dataset >0 >1 >0 ==0
	* 3. @snap1 dataset@snap2 ==0 >1 >0 >1
	* 4. @snap1 dataset ==0 >1 >0 ==0
	*/
	if (tosnap == NULL) {
	/* only a from snapshot given, must be valid */
	(void) snprintf(di->errbuf, sizeof (di->errbuf),
	dgettext(TEXT_DOMAIN,
	"Badly formed snapshot name %s"), fromsnap);

	if (!zfs_validate_name(hdl, fromsnap, ZFS_TYPE_SNAPSHOT,
	B_FALSE)) {
	return (zfs_error(hdl, EZFS_INVALIDNAME,
	di->errbuf));
	}

	atptrf = strchr(fromsnap, '@');
	ASSERT(atptrf != NULL);
	fdslen = atptrf - fromsnap;

	di->fromsnap = zfs_strdup(hdl, fromsnap);
	di->ds = zfs_strdup(hdl, fromsnap);
	di->ds[fdslen] = '\0';

	/* the to snap will be a just-in-time snap of the head */
	return (make_temp_snapshot(di));
	}

	(void) snprintf(di->errbuf, sizeof (di->errbuf),
	dgettext(TEXT_DOMAIN,
	"Unable to determine which snapshots to compare"));

	atptrf = strchr(fromsnap, '@');
	atptrt = strchr(tosnap, '@');
	fdslen = atptrf ? atptrf - fromsnap : strlen(fromsnap);
	tdslen = atptrt ? atptrt - tosnap : strlen(tosnap);
	fsnlen = strlen(fromsnap) - fdslen; /* includes @ sign */
	tsnlen = strlen(tosnap) - tdslen; /* includes @ sign */

	if (fsnlen <= 1 \|\| tsnlen == 1 \|\| (fdslen == 0 && tdslen == 0)) {
	return (zfs_error(hdl, EZFS_INVALIDNAME, di->errbuf));
	} else if ((fdslen > 0 && tdslen > 0) &&
	((tdslen != fdslen \|\| strncmp(fromsnap, tosnap, fdslen) != 0))) {
	/*
	* not the same dataset name, might be okay if
	* tosnap is a clone of a fromsnap descendant.
	*/
	char origin[ZFS_MAX_DATASET_NAME_LEN];
	zprop_source_t src;
	zfs_handle_t *zhp;

	di->ds = zfs_alloc(di->zhp->zfs_hdl, tdslen + 1);
	(void) strncpy(di->ds, tosnap, tdslen);
	di->ds[tdslen] = '\0';

	zhp = zfs_open(hdl, di->ds, ZFS_TYPE_FILESYSTEM);
	while (zhp != NULL) {
	if (zfs_prop_get(zhp, ZFS_PROP_ORIGIN, origin,
	sizeof (origin), &src, NULL, 0, B_FALSE) != 0) {
	(void) zfs_close(zhp);
	zhp = NULL;
	break;
	}
	if (strncmp(origin, fromsnap, fsnlen) == 0)
	break;

	(void) zfs_close(zhp);
	zhp = zfs_open(hdl, origin, ZFS_TYPE_FILESYSTEM);
	}

	if (zhp == NULL) {
	(void) snprintf(di->errbuf, sizeof (di->errbuf),
	dgettext(TEXT_DOMAIN,
	"Not an earlier snapshot from the same fs"));
	return (zfs_error(hdl, EZFS_INVALIDNAME, di->errbuf));
	} else {
	(void) zfs_close(zhp);
	}

	di->isclone = B_TRUE;
	di->fromsnap = zfs_strdup(hdl, fromsnap);
	if (tsnlen) {
	di->tosnap = zfs_strdup(hdl, tosnap);
	} else {
	return (make_temp_snapshot(di));
	}
	} else {
	int dslen = fdslen ? fdslen : tdslen;

	di->ds = zfs_alloc(hdl, dslen + 1);
	(void) strncpy(di->ds, fdslen ? fromsnap : tosnap, dslen);
	di->ds[dslen] = '\0';

	di->fromsnap = zfs_asprintf(hdl, "%s%s", di->ds, atptrf);
	if (tsnlen) {
	di->tosnap = zfs_asprintf(hdl, "%s%s", di->ds, atptrt);
	} else {
	return (make_temp_snapshot(di));
	}
	}
	return (0);
	}

	static int
	get_mountpoint(differ_info_t di, char dsnm, char **mntpt)
	{
	boolean_t mounted;

	mounted = is_mounted(di->zhp->zfs_hdl, dsnm, mntpt);
	if (mounted == B_FALSE) {
	(void) snprintf(di->errbuf, sizeof (di->errbuf),
	dgettext(TEXT_DOMAIN,
	"Cannot diff an unmounted snapshot"));
	return (zfs_error(di->zhp->zfs_hdl, EZFS_BADTYPE, di->errbuf));
	}

	/* Avoid a double slash at the beginning of root-mounted datasets */
	if (*mntpt == '/' && (*mntpt + 1) == '\0')
	**mntpt = '\0';
	return (0);
	}

	static int
	get_mountpoints(differ_info_t *di)
	{
	char *strptr;
	char *frommntpt;

	/*
	* first get the mountpoint for the parent dataset
	*/
	if (get_mountpoint(di, di->ds, &di->dsmnt) != 0)
	return (-1);

	strptr = strchr(di->tosnap, '@');
	ASSERT3P(strptr, !=, NULL);
	di->tomnt = zfs_asprintf(di->zhp->zfs_hdl, "%s%s%s", di->dsmnt,
	ZDIFF_SNAPDIR, ++strptr);

	strptr = strchr(di->fromsnap, '@');
	ASSERT3P(strptr, !=, NULL);

	frommntpt = di->dsmnt;
	if (di->isclone) {
	char *mntpt;
	int err;

	*strptr = '\0';
	err = get_mountpoint(di, di->fromsnap, &mntpt);
	*strptr = '@';
	if (err != 0)
	return (-1);
	frommntpt = mntpt;
	}

	di->frommnt = zfs_asprintf(di->zhp->zfs_hdl, "%s%s%s", frommntpt,
	ZDIFF_SNAPDIR, ++strptr);

	if (di->isclone)
	free(frommntpt);

	return (0);
	}

	static int
	setup_differ_info(zfs_handle_t zhp, const char fromsnap,
	const char tosnap, differ_info_t di)
	{
	di->zhp = zhp;

	di->cleanupfd = open(ZFS_DEV, O_RDWR \| O_CLOEXEC);
	VERIFY(di->cleanupfd >= 0);

	if (get_snapshot_names(di, fromsnap, tosnap) != 0)
	return (-1);

	if (get_mountpoints(di) != 0)
	return (-1);

	if (find_shares_object(di) != 0)
	return (-1);

	return (0);
	}

	int
	zfs_show_diffs(zfs_handle_t zhp, int outfd, const char fromsnap,
	const char *tosnap, int flags)
	{
	zfs_cmd_t zc = {"\0"};
	char errbuf[1024];
	differ_info_t di = { 0 };
	pthread_t tid;
	int pipefd[2];
	int iocerr;

	(void) snprintf(errbuf, sizeof (errbuf),
	dgettext(TEXT_DOMAIN, "zfs diff failed"));

	if (setup_differ_info(zhp, fromsnap, tosnap, &di)) {
	teardown_differ_info(&di);
	return (-1);
	}

	if (pipe2(pipefd, O_CLOEXEC)) {
	zfs_error_aux(zhp->zfs_hdl, "%s", strerror(errno));
	teardown_differ_info(&di);
	return (zfs_error(zhp->zfs_hdl, EZFS_PIPEFAILED, errbuf));
	}

	di.scripted = (flags & ZFS_DIFF_PARSEABLE);
	di.classify = (flags & ZFS_DIFF_CLASSIFY);
	di.timestamped = (flags & ZFS_DIFF_TIMESTAMP);
	+ di.no_mangle = (flags & ZFS_DIFF_NO_MANGLE);

	di.outputfd = outfd;
	di.datafd = pipefd[0];

	if (pthread_create(&tid, NULL, differ, &di)) {
	zfs_error_aux(zhp->zfs_hdl, "%s", strerror(errno));
	(void) close(pipefd[0]);
	(void) close(pipefd[1]);
	teardown_differ_info(&di);
	return (zfs_error(zhp->zfs_hdl,
	EZFS_THREADCREATEFAILED, errbuf));
	}

	/* do the ioctl() */
	(void) strlcpy(zc.zc_value, di.fromsnap, strlen(di.fromsnap) + 1);
	(void) strlcpy(zc.zc_name, di.tosnap, strlen(di.tosnap) + 1);
	zc.zc_cookie = pipefd[1];

	iocerr = zfs_ioctl(zhp->zfs_hdl, ZFS_IOC_DIFF, &zc);
	if (iocerr != 0) {
	(void) snprintf(errbuf, sizeof (errbuf),
	dgettext(TEXT_DOMAIN, "Unable to obtain diffs"));
	if (errno == EPERM) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"\n The sys_mount privilege or diff delegated "
	"permission is needed\n to execute the "
	"diff ioctl"));
	} else if (errno == EXDEV) {
	zfs_error_aux(zhp->zfs_hdl, dgettext(TEXT_DOMAIN,
	"\n Not an earlier snapshot from the same fs"));
	} else if (errno != EPIPE \|\| di.zerr == 0) {
	zfs_error_aux(zhp->zfs_hdl, "%s", strerror(errno));
	}
	(void) close(pipefd[1]);
	(void) pthread_cancel(tid);
	(void) pthread_join(tid, NULL);
	teardown_differ_info(&di);
	if (di.zerr != 0 && di.zerr != EPIPE) {
	zfs_error_aux(zhp->zfs_hdl, "%s", strerror(di.zerr));
	return (zfs_error(zhp->zfs_hdl, EZFS_DIFF, di.errbuf));
	} else {
	return (zfs_error(zhp->zfs_hdl, EZFS_DIFFDATA, errbuf));
	}
	}

	(void) close(pipefd[1]);
	(void) pthread_join(tid, NULL);

	if (di.zerr != 0) {
	zfs_error_aux(zhp->zfs_hdl, "%s", strerror(di.zerr));
	return (zfs_error(zhp->zfs_hdl, EZFS_DIFF, di.errbuf));
	}
	teardown_differ_info(&di);
	return (0);
	}
	diff --git a/sys/contrib/openzfs/lib/libzfs/libzfs_impl.h b/sys/contrib/openzfs/lib/libzfs/libzfs_impl.h
	index b1cf4f825f8a..33b3feed43e6 100644
	--- a/sys/contrib/openzfs/lib/libzfs/libzfs_impl.h
	+++ b/sys/contrib/openzfs/lib/libzfs/libzfs_impl.h
	@@ -1,267 +1,268 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2018 Datto Inc.
	* Copyright 2020 Joyent, Inc.
	*/

	#ifndef _LIBZFS_IMPL_H
	#define _LIBZFS_IMPL_H

	#include <sys/fs/zfs.h>
	#include <sys/nvpair.h>
	#include <sys/dmu.h>
	#include <sys/zfs_ioctl.h>
	#include <regex.h>

	#include <libuutil.h>
	#include <libzfs.h>
	#include <libshare.h>
	#include <libzfs_core.h>

	#ifdef __cplusplus
	extern "C" {
	#endif

	struct libzfs_handle {
	int libzfs_error;
	int libzfs_fd;
	zpool_handle_t *libzfs_pool_handles;
	uu_avl_pool_t *libzfs_ns_avlpool;
	uu_avl_t *libzfs_ns_avl;
	uint64_t libzfs_ns_gen;
	int libzfs_desc_active;
	char libzfs_action[1024];
	char libzfs_desc[1024];
	int libzfs_printerr;
	boolean_t libzfs_mnttab_enable;
	/*
	* We need a lock to handle the case where parallel mount
	* threads are populating the mnttab cache simultaneously. The
	* lock only protects the integrity of the avl tree, and does
	* not protect the contents of the mnttab entries themselves.
	*/
	pthread_mutex_t libzfs_mnttab_cache_lock;
	avl_tree_t libzfs_mnttab_cache;
	int libzfs_pool_iter;
	boolean_t libzfs_prop_debug;
	regex_t libzfs_urire;
	uint64_t libzfs_max_nvlist;
	void *libfetch;
	char *libfetch_load_error;
	};

	struct zfs_handle {
	libzfs_handle_t *zfs_hdl;
	zpool_handle_t *zpool_hdl;
	char zfs_name[ZFS_MAX_DATASET_NAME_LEN];
	zfs_type_t zfs_type; /* type including snapshot */
	zfs_type_t zfs_head_type; /* type excluding snapshot */
	dmu_objset_stats_t zfs_dmustats;
	nvlist_t *zfs_props;
	nvlist_t *zfs_user_props;
	nvlist_t *zfs_recvd_props;
	boolean_t zfs_mntcheck;
	char *zfs_mntopts;
	uint8_t *zfs_props_table;
	};

	/*
	* This is different from checking zfs_type, because it will also catch
	* snapshots of volumes.
	*/
	#define ZFS_IS_VOLUME(zhp) ((zhp)->zfs_head_type == ZFS_TYPE_VOLUME)

	struct zpool_handle {
	libzfs_handle_t *zpool_hdl;
	zpool_handle_t *zpool_next;
	char zpool_name[ZFS_MAX_DATASET_NAME_LEN];
	int zpool_state;
	size_t zpool_config_size;
	nvlist_t *zpool_config;
	nvlist_t *zpool_old_config;
	nvlist_t *zpool_props;
	diskaddr_t zpool_start_block;
	};

	typedef enum {
	PROTO_NFS = 0,
	PROTO_SMB = 1,
	PROTO_END = 2
	} zfs_share_proto_t;

	/*
	* The following can be used as a bitmask and any new values
	* added must preserve that capability.
	*/
	typedef enum {
	SHARED_NOT_SHARED = 0x0,
	SHARED_NFS = 0x2,
	SHARED_SMB = 0x4
	} zfs_share_type_t;

	typedef int (zfs_uri_handler_fn_t)(struct libzfs_handle , const char *,
	const char , zfs_keyformat_t, boolean_t, uint8_t , size_t );

	typedef struct zfs_uri_handler {
	const char *zuh_scheme;
	zfs_uri_handler_fn_t zuh_handler;
	} zfs_uri_handler_t;

	#define CONFIG_BUF_MINSIZE 262144

	extern int zfs_error(libzfs_handle_t , int, const char );
	extern int zfs_error_fmt(libzfs_handle_t , int, const char , ...)
	__attribute__((format(printf, 3, 4)));
	extern void zfs_error_aux(libzfs_handle_t , const char , ...)
	__attribute__((format(printf, 2, 3)));
	extern void zfs_alloc(libzfs_handle_t , size_t);
	extern void zfs_realloc(libzfs_handle_t , void *, size_t, size_t);
	extern char zfs_asprintf(libzfs_handle_t , const char *, ...)
	__attribute__((format(printf, 2, 3)));
	extern char zfs_strdup(libzfs_handle_t , const char *);
	extern int no_memory(libzfs_handle_t *);

	extern int zfs_standard_error_fmt(libzfs_handle_t , int, const char , ...)
	__attribute__((format(printf, 3, 4)));
	extern void zfs_setprop_error(libzfs_handle_t , zfs_prop_t, int, char );
	extern int zpool_standard_error(libzfs_handle_t , int, const char );
	extern int zpool_standard_error_fmt(libzfs_handle_t , int, const char , ...)
	__attribute__((format(printf, 3, 4)));

	extern zfs_handle_t make_dataset_handle_zc(libzfs_handle_t , zfs_cmd_t *);
	extern zfs_handle_t make_dataset_simple_handle_zc(zfs_handle_t , zfs_cmd_t *);

	extern int zprop_parse_value(libzfs_handle_t , nvpair_t , int, zfs_type_t,
	nvlist_t , char , uint64_t , const char *);
	extern int zprop_expand_list(libzfs_handle_t hdl, zprop_list_t *plp,
	zfs_type_t type);

	/*
	* Use this changelist_gather() flag to force attempting mounts
	* on each change node regardless of whether or not it is currently
	* mounted.
	*/
	#define CL_GATHER_MOUNT_ALWAYS 1
	/*
	* changelist_gather() flag to force it to iterate on mounted datasets only
	*/
	#define CL_GATHER_ITER_MOUNTED 2
	/*
	* Use this changelist_gather() flag to prevent unmounting of file systems.
	*/
	#define CL_GATHER_DONT_UNMOUNT 4

	typedef struct prop_changelist prop_changelist_t;

	extern int zcmd_alloc_dst_nvlist(libzfs_handle_t , zfs_cmd_t , size_t);
	extern int zcmd_write_src_nvlist(libzfs_handle_t , zfs_cmd_t , nvlist_t *);
	extern int zcmd_write_conf_nvlist(libzfs_handle_t , zfs_cmd_t , nvlist_t *);
	extern int zcmd_expand_dst_nvlist(libzfs_handle_t , zfs_cmd_t );
	extern int zcmd_read_dst_nvlist(libzfs_handle_t , zfs_cmd_t , nvlist_t **);
	extern void zcmd_free_nvlists(zfs_cmd_t *);

	extern int changelist_prefix(prop_changelist_t *);
	extern int changelist_postfix(prop_changelist_t *);
	extern void changelist_rename(prop_changelist_t , const char , const char *);
	extern void changelist_remove(prop_changelist_t , const char );
	extern void changelist_free(prop_changelist_t *);
	extern prop_changelist_t changelist_gather(zfs_handle_t , zfs_prop_t, int,
	int);
	extern int changelist_unshare(prop_changelist_t , zfs_share_proto_t );
	extern int changelist_haszonedchild(prop_changelist_t *);

	extern void remove_mountpoint(zfs_handle_t *);
	extern int create_parents(libzfs_handle_t , char , int);

	extern zfs_handle_t make_dataset_handle(libzfs_handle_t , const char *);
	extern zfs_handle_t make_bookmark_handle(zfs_handle_t , const char *,
	nvlist_t *props);

	extern int zpool_open_silent(libzfs_handle_t , const char ,
	zpool_handle_t **);

	extern boolean_t zpool_name_valid(libzfs_handle_t , boolean_t, const char );

	extern int zfs_validate_name(libzfs_handle_t hdl, const char path, int type,
	boolean_t modifying);

	extern void namespace_clear(libzfs_handle_t *);

	extern int zfs_parse_options(char *, zfs_share_proto_t);

	typedef struct {
	zfs_prop_t p_prop;
	char *p_name;
	int p_share_err;
	int p_unshare_err;
	} proto_table_t;

	typedef struct differ_info {
	zfs_handle_t *zhp;
	char *fromsnap;
	char *frommnt;
	char *tosnap;
	char *tomnt;
	char *ds;
	char *dsmnt;
	char *tmpsnap;
	char errbuf[1024];
	boolean_t isclone;
	boolean_t scripted;
	boolean_t classify;
	boolean_t timestamped;
	+ boolean_t no_mangle;
	uint64_t shares;
	int zerr;
	int cleanupfd;
	int outputfd;
	int datafd;
	} differ_info_t;

	extern proto_table_t proto_table[PROTO_END];

	extern int do_mount(zfs_handle_t zhp, const char mntpt, char *opts,
	int flags);
	extern int do_unmount(zfs_handle_t zhp, const char mntpt, int flags);
	extern int zfs_mount_delegation_check(void);
	extern int zfs_share_proto(zfs_handle_t zhp, zfs_share_proto_t proto);
	extern int zfs_unshare_proto(zfs_handle_t , const char , zfs_share_proto_t *);
	extern int unshare_one(libzfs_handle_t hdl, const char name,
	const char *mountpoint, zfs_share_proto_t proto);
	extern boolean_t zfs_is_mountable(zfs_handle_t zhp, char buf, size_t buflen,
	zprop_source_t *source, int flags);
	extern zfs_share_type_t is_shared(const char *mountpoint,
	zfs_share_proto_t proto);
	extern int libzfs_load_module(void);
	extern int zpool_relabel_disk(libzfs_handle_t hdl, const char path,
	const char *msg);
	extern int find_shares_object(differ_info_t *di);
	extern void libzfs_set_pipe_max(int infd);
	extern void zfs_commit_proto(zfs_share_proto_t *);

	#ifdef __cplusplus
	}
	#endif

	#endif /* _LIBZFS_IMPL_H */
	diff --git a/sys/contrib/openzfs/lib/libzfs/libzfs_mount.c b/sys/contrib/openzfs/lib/libzfs/libzfs_mount.c
	index b0279d8fbc3f..e97de8f20262 100644
	--- a/sys/contrib/openzfs/lib/libzfs/libzfs_mount.c
	+++ b/sys/contrib/openzfs/lib/libzfs/libzfs_mount.c
	@@ -1,1655 +1,1659 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright 2015 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2014, 2021 by Delphix. All rights reserved.
	* Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>
	* Copyright 2017 RackTop Systems.
	* Copyright (c) 2018 Datto Inc.
	* Copyright 2018 OmniOS Community Edition (OmniOSce) Association.
	*/

	/*
	* Routines to manage ZFS mounts. We separate all the nasty routines that have
	* to deal with the OS. The following functions are the main entry points --
	* they are used by mount and unmount and when changing a filesystem's
	* mountpoint.
	*
	* zfs_is_mounted()
	* zfs_mount()
	* zfs_mount_at()
	* zfs_unmount()
	* zfs_unmountall()
	*
	* This file also contains the functions used to manage sharing filesystems via
	* NFS and iSCSI:
	*
	* zfs_is_shared()
	* zfs_share()
	* zfs_unshare()
	*
	* zfs_is_shared_nfs()
	* zfs_is_shared_smb()
	* zfs_share_proto()
	* zfs_shareall();
	* zfs_unshare_nfs()
	* zfs_unshare_smb()
	* zfs_unshareall_nfs()
	* zfs_unshareall_smb()
	* zfs_unshareall()
	* zfs_unshareall_bypath()
	*
	* The following functions are available for pool consumers, and will
	* mount/unmount and share/unshare all datasets within pool:
	*
	* zpool_enable_datasets()
	* zpool_disable_datasets()
	*/

	#include <dirent.h>
	#include <dlfcn.h>
	#include <errno.h>
	#include <fcntl.h>
	#include <libgen.h>
	#include <libintl.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <strings.h>
	#include <unistd.h>
	#include <zone.h>
	#include <sys/mntent.h>
	#include <sys/mount.h>
	#include <sys/stat.h>
	#include <sys/vfs.h>
	#include <sys/dsl_crypt.h>

	#include <libzfs.h>

	#include "libzfs_impl.h"
	#include <thread_pool.h>

	#include <libshare.h>
	#include <sys/systeminfo.h>
	#define MAXISALEN 257 /* based on sysinfo(2) man page */

	static int mount_tp_nthr = 512; /* tpool threads for multi-threaded mounting */

	static void zfs_mount_task(void *);
	static zfs_share_type_t zfs_is_shared_proto(zfs_handle_t , char *,
	zfs_share_proto_t);

	/*
	* The share protocols table must be in the same order as the zfs_share_proto_t
	* enum in libzfs_impl.h
	*/
	proto_table_t proto_table[PROTO_END] = {
	{ZFS_PROP_SHARENFS, "nfs", EZFS_SHARENFSFAILED, EZFS_UNSHARENFSFAILED},
	{ZFS_PROP_SHARESMB, "smb", EZFS_SHARESMBFAILED, EZFS_UNSHARESMBFAILED},
	};

	static zfs_share_proto_t nfs_only[] = {
	PROTO_NFS,
	PROTO_END
	};

	static zfs_share_proto_t smb_only[] = {
	PROTO_SMB,
	PROTO_END
	};
	static zfs_share_proto_t share_all_proto[] = {
	PROTO_NFS,
	PROTO_SMB,
	PROTO_END
	};



	static boolean_t
	dir_is_empty_stat(const char *dirname)
	{
	struct stat st;

	/*
	* We only want to return false if the given path is a non empty
	* directory, all other errors are handled elsewhere.
	*/
	if (stat(dirname, &st) < 0 \|\| !S_ISDIR(st.st_mode)) {
	return (B_TRUE);
	}

	/*
	* An empty directory will still have two entries in it, one
	* entry for each of "." and "..".
	*/
	if (st.st_size > 2) {
	return (B_FALSE);
	}

	return (B_TRUE);
	}

	static boolean_t
	dir_is_empty_readdir(const char *dirname)
	{
	DIR *dirp;
	struct dirent64 *dp;
	int dirfd;

	if ((dirfd = openat(AT_FDCWD, dirname,
	O_RDONLY \| O_NDELAY \| O_LARGEFILE \| O_CLOEXEC, 0)) < 0) {
	return (B_TRUE);
	}

	if ((dirp = fdopendir(dirfd)) == NULL) {
	(void) close(dirfd);
	return (B_TRUE);
	}

	while ((dp = readdir64(dirp)) != NULL) {

	if (strcmp(dp->d_name, ".") == 0 \|\|
	strcmp(dp->d_name, "..") == 0)
	continue;

	(void) closedir(dirp);
	return (B_FALSE);
	}

	(void) closedir(dirp);
	return (B_TRUE);
	}

	/*
	* Returns true if the specified directory is empty. If we can't open the
	* directory at all, return true so that the mount can fail with a more
	* informative error message.
	*/
	static boolean_t
	dir_is_empty(const char *dirname)
	{
	struct statfs64 st;

	/*
	* If the statvfs call fails or the filesystem is not a ZFS
	* filesystem, fall back to the slow path which uses readdir.
	*/
	if ((statfs64(dirname, &st) != 0) \|\|
	(st.f_type != ZFS_SUPER_MAGIC)) {
	return (dir_is_empty_readdir(dirname));
	}

	/*
	* At this point, we know the provided path is on a ZFS
	* filesystem, so we can use stat instead of readdir to
	* determine if the directory is empty or not. We try to avoid
	* using readdir because that requires opening "dirname"; this
	* open file descriptor can potentially end up in a child
	* process if there's a concurrent fork, thus preventing the
	* zfs_mount() from otherwise succeeding (the open file
	* descriptor inherited by the child process will cause the
	* parent's mount to fail with EBUSY). The performance
	* implications of replacing the open, read, and close with a
	* single stat is nice; but is not the main motivation for the
	* added complexity.
	*/
	return (dir_is_empty_stat(dirname));
	}

	/*
	* Checks to see if the mount is active. If the filesystem is mounted, we fill
	* in 'where' with the current mountpoint, and return 1. Otherwise, we return
	* 0.
	*/
	boolean_t
	is_mounted(libzfs_handle_t zfs_hdl, const char special, char **where)
	{
	struct mnttab entry;

	if (libzfs_mnttab_find(zfs_hdl, special, &entry) != 0)
	return (B_FALSE);

	if (where != NULL)
	*where = zfs_strdup(zfs_hdl, entry.mnt_mountp);

	return (B_TRUE);
	}

	boolean_t
	zfs_is_mounted(zfs_handle_t zhp, char *where)
	{
	return (is_mounted(zhp->zfs_hdl, zfs_get_name(zhp), where));
	}

	/*
	* Checks any higher order concerns about whether the given dataset is
	* mountable, false otherwise. zfs_is_mountable_internal specifically assumes
	* that the caller has verified the sanity of mounting the dataset at
	* mountpoint to the extent the caller wants.
	*/
	static boolean_t
	zfs_is_mountable_internal(zfs_handle_t zhp, const char mountpoint)
	{

	if (zfs_prop_get_int(zhp, ZFS_PROP_ZONED) &&
	getzoneid() == GLOBAL_ZONEID)
	return (B_FALSE);

	return (B_TRUE);
	}

	/*
	* Returns true if the given dataset is mountable, false otherwise. Returns the
	* mountpoint in 'buf'.
	*/
	boolean_t
	zfs_is_mountable(zfs_handle_t zhp, char buf, size_t buflen,
	zprop_source_t *source, int flags)
	{
	char sourceloc[MAXNAMELEN];
	zprop_source_t sourcetype;

	if (!zfs_prop_valid_for_type(ZFS_PROP_MOUNTPOINT, zhp->zfs_type,
	B_FALSE))
	return (B_FALSE);

	verify(zfs_prop_get(zhp, ZFS_PROP_MOUNTPOINT, buf, buflen,
	&sourcetype, sourceloc, sizeof (sourceloc), B_FALSE) == 0);

	if (strcmp(buf, ZFS_MOUNTPOINT_NONE) == 0 \|\|
	strcmp(buf, ZFS_MOUNTPOINT_LEGACY) == 0)
	return (B_FALSE);

	if (zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT) == ZFS_CANMOUNT_OFF)
	return (B_FALSE);

	if (!zfs_is_mountable_internal(zhp, buf))
	return (B_FALSE);

	if (zfs_prop_get_int(zhp, ZFS_PROP_REDACTED) && !(flags & MS_FORCE))
	return (B_FALSE);

	if (source)
	*source = sourcetype;

	return (B_TRUE);
	}

	/*
	* The filesystem is mounted by invoking the system mount utility rather
	* than by the system call mount(2). This ensures that the /etc/mtab
	* file is correctly locked for the update. Performing our own locking
	* and /etc/mtab update requires making an unsafe assumption about how
	* the mount utility performs its locking. Unfortunately, this also means
	* in the case of a mount failure we do not have the exact errno. We must
	* make due with return value from the mount process.
	*
	* In the long term a shared library called libmount is under development
	* which provides a common API to address the locking and errno issues.
	* Once the standard mount utility has been updated to use this library
	* we can add an autoconf check to conditionally use it.
	*
	* http://www.kernel.org/pub/linux/utils/util-linux/libmount-docs/index.html
	*/

	static int
	zfs_add_option(zfs_handle_t zhp, char options, int len,
	zfs_prop_t prop, char on, char off)
	{
	char *source;
	uint64_t value;

	/* Skip adding duplicate default options */
	if ((strstr(options, on) != NULL) \|\| (strstr(options, off) != NULL))
	return (0);

	/*
	* zfs_prop_get_int() is not used to ensure our mount options
	* are not influenced by the current /proc/self/mounts contents.
	*/
	value = getprop_uint64(zhp, prop, &source);

	(void) strlcat(options, ",", len);
	(void) strlcat(options, value ? on : off, len);

	return (0);
	}

	static int
	zfs_add_options(zfs_handle_t zhp, char options, int len)
	{
	int error = 0;

	error = zfs_add_option(zhp, options, len,
	ZFS_PROP_ATIME, MNTOPT_ATIME, MNTOPT_NOATIME);
	/*
	* don't add relatime/strictatime when atime=off, otherwise strictatime
	* will force atime=on
	*/
	if (strstr(options, MNTOPT_NOATIME) == NULL) {
	error = zfs_add_option(zhp, options, len,
	ZFS_PROP_RELATIME, MNTOPT_RELATIME, MNTOPT_STRICTATIME);
	}
	error = error ? error : zfs_add_option(zhp, options, len,
	ZFS_PROP_DEVICES, MNTOPT_DEVICES, MNTOPT_NODEVICES);
	error = error ? error : zfs_add_option(zhp, options, len,
	ZFS_PROP_EXEC, MNTOPT_EXEC, MNTOPT_NOEXEC);
	error = error ? error : zfs_add_option(zhp, options, len,
	ZFS_PROP_READONLY, MNTOPT_RO, MNTOPT_RW);
	error = error ? error : zfs_add_option(zhp, options, len,
	ZFS_PROP_SETUID, MNTOPT_SETUID, MNTOPT_NOSETUID);
	error = error ? error : zfs_add_option(zhp, options, len,
	ZFS_PROP_NBMAND, MNTOPT_NBMAND, MNTOPT_NONBMAND);

	return (error);
	}

	int
	zfs_mount(zfs_handle_t zhp, const char options, int flags)
	{
	char mountpoint[ZFS_MAXPROPLEN];

	if (!zfs_is_mountable(zhp, mountpoint, sizeof (mountpoint), NULL,
	flags))
	return (0);

	return (zfs_mount_at(zhp, options, flags, mountpoint));
	}

	/*
	* Mount the given filesystem.
	*/
	int
	zfs_mount_at(zfs_handle_t zhp, const char options, int flags,
	const char *mountpoint)
	{
	struct stat buf;
	char mntopts[MNT_LINE_MAX];
	char overlay[ZFS_MAXPROPLEN];
	char prop_encroot[MAXNAMELEN];
	boolean_t is_encroot;
	zfs_handle_t *encroot_hp = zhp;
	libzfs_handle_t *hdl = zhp->zfs_hdl;
	uint64_t keystatus;
	int remount = 0, rc;

	if (options == NULL) {
	(void) strlcpy(mntopts, MNTOPT_DEFAULTS, sizeof (mntopts));
	} else {
	(void) strlcpy(mntopts, options, sizeof (mntopts));
	}

	if (strstr(mntopts, MNTOPT_REMOUNT) != NULL)
	remount = 1;

	/* Potentially duplicates some checks if invoked by zfs_mount(). */
	if (!zfs_is_mountable_internal(zhp, mountpoint))
	return (0);

	/*
	* If the pool is imported read-only then all mounts must be read-only
	*/
	if (zpool_get_prop_int(zhp->zpool_hdl, ZPOOL_PROP_READONLY, NULL))
	(void) strlcat(mntopts, "," MNTOPT_RO, sizeof (mntopts));

	/*
	* Append default mount options which apply to the mount point.
	* This is done because under Linux (unlike Solaris) multiple mount
	* points may reference a single super block. This means that just
	* given a super block there is no back reference to update the per
	* mount point options.
	*/
	rc = zfs_add_options(zhp, mntopts, sizeof (mntopts));
	if (rc) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"default options unavailable"));
	return (zfs_error_fmt(hdl, EZFS_MOUNTFAILED,
	dgettext(TEXT_DOMAIN, "cannot mount '%s'"),
	mountpoint));
	}

	/*
	* If the filesystem is encrypted the key must be loaded in order to
	* mount. If the key isn't loaded, the MS_CRYPT flag decides whether
	* or not we attempt to load the keys. Note: we must call
	* zfs_refresh_properties() here since some callers of this function
	* (most notably zpool_enable_datasets()) may implicitly load our key
	* by loading the parent's key first.
	*/
	if (zfs_prop_get_int(zhp, ZFS_PROP_ENCRYPTION) != ZIO_CRYPT_OFF) {
	zfs_refresh_properties(zhp);
	keystatus = zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS);

	/*
	* If the key is unavailable and MS_CRYPT is set give the
	* user a chance to enter the key. Otherwise just fail
	* immediately.
	*/
	if (keystatus == ZFS_KEYSTATUS_UNAVAILABLE) {
	if (flags & MS_CRYPT) {
	rc = zfs_crypto_get_encryption_root(zhp,
	&is_encroot, prop_encroot);
	if (rc) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Failed to get encryption root for "
	"'%s'."), zfs_get_name(zhp));
	return (rc);
	}

	if (!is_encroot) {
	encroot_hp = zfs_open(hdl, prop_encroot,
	ZFS_TYPE_DATASET);
	if (encroot_hp == NULL)
	return (hdl->libzfs_error);
	}

	rc = zfs_crypto_load_key(encroot_hp,
	B_FALSE, NULL);

	if (!is_encroot)
	zfs_close(encroot_hp);
	if (rc)
	return (rc);
	} else {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"encryption key not loaded"));
	return (zfs_error_fmt(hdl, EZFS_MOUNTFAILED,
	dgettext(TEXT_DOMAIN, "cannot mount '%s'"),
	mountpoint));
	}
	}

	}

	/*
	* Append zfsutil option so the mount helper allow the mount
	*/
	strlcat(mntopts, "," MNTOPT_ZFSUTIL, sizeof (mntopts));

	/* Create the directory if it doesn't already exist */
	if (lstat(mountpoint, &buf) != 0) {
	if (mkdirp(mountpoint, 0755) != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"failed to create mountpoint: %s"),
	strerror(errno));
	return (zfs_error_fmt(hdl, EZFS_MOUNTFAILED,
	dgettext(TEXT_DOMAIN, "cannot mount '%s'"),
	mountpoint));
	}
	}

	/*
	* Overlay mounts are enabled by default but may be disabled
	* via the 'overlay' property. The -O flag remains for compatibility.
	*/
	if (!(flags & MS_OVERLAY)) {
	if (zfs_prop_get(zhp, ZFS_PROP_OVERLAY, overlay,
	sizeof (overlay), NULL, NULL, 0, B_FALSE) == 0) {
	if (strcmp(overlay, "on") == 0) {
	flags \|= MS_OVERLAY;
	}
	}
	}

	/*
	* Determine if the mountpoint is empty. If so, refuse to perform the
	* mount. We don't perform this check if 'remount' is
	* specified or if overlay option (-O) is given
	*/
	if ((flags & MS_OVERLAY) == 0 && !remount &&
	!dir_is_empty(mountpoint)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"directory is not empty"));
	return (zfs_error_fmt(hdl, EZFS_MOUNTFAILED,
	dgettext(TEXT_DOMAIN, "cannot mount '%s'"), mountpoint));
	}

	/* perform the mount */
	rc = do_mount(zhp, mountpoint, mntopts, flags);
	if (rc) {
	/*
	* Generic errors are nasty, but there are just way too many
	* from mount(), and they're well-understood. We pick a few
	* common ones to improve upon.
	*/
	if (rc == EBUSY) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"mountpoint or dataset is busy"));
	} else if (rc == EPERM) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Insufficient privileges"));
	} else if (rc == ENOTSUP) {
	int spa_version;

	VERIFY(zfs_spa_version(zhp, &spa_version) == 0);
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Can't mount a version %llu "
	"file system on a version %d pool. Pool must be"
	" upgraded to mount this file system."),
	(u_longlong_t)zfs_prop_get_int(zhp,
	ZFS_PROP_VERSION), spa_version);
	} else {
	zfs_error_aux(hdl, "%s", strerror(rc));
	}
	return (zfs_error_fmt(hdl, EZFS_MOUNTFAILED,
	dgettext(TEXT_DOMAIN, "cannot mount '%s'"),
	zhp->zfs_name));
	}

	/* remove the mounted entry before re-adding on remount */
	if (remount)
	libzfs_mnttab_remove(hdl, zhp->zfs_name);

	/* add the mounted entry into our cache */
	libzfs_mnttab_add(hdl, zfs_get_name(zhp), mountpoint, mntopts);
	return (0);
	}

	/*
	* Unmount a single filesystem.
	*/
	static int
	unmount_one(zfs_handle_t zhp, const char mountpoint, int flags)
	{
	int error;

	error = do_unmount(zhp, mountpoint, flags);
	if (error != 0) {
	int libzfs_err;

	switch (error) {
	case EBUSY:
	libzfs_err = EZFS_BUSY;
	break;
	case EIO:
	libzfs_err = EZFS_IO;
	break;
	case ENOENT:
	libzfs_err = EZFS_NOENT;
	break;
	case ENOMEM:
	libzfs_err = EZFS_NOMEM;
	break;
	case EPERM:
	libzfs_err = EZFS_PERM;
	break;
	default:
	libzfs_err = EZFS_UMOUNTFAILED;
	}
	- return (zfs_error_fmt(zhp->zfs_hdl, libzfs_err,
	- dgettext(TEXT_DOMAIN, "cannot unmount '%s'"),
	- mountpoint));
	+ if (zhp) {
	+ return (zfs_error_fmt(zhp->zfs_hdl, libzfs_err,
	+ dgettext(TEXT_DOMAIN, "cannot unmount '%s'"),
	+ mountpoint));
	+ } else {
	+ return (-1);
	+ }
	}

	return (0);
	}

	/*
	* Unmount the given filesystem.
	*/
	int
	zfs_unmount(zfs_handle_t zhp, const char mountpoint, int flags)
	{
	libzfs_handle_t *hdl = zhp->zfs_hdl;
	struct mnttab entry;
	char *mntpt = NULL;
	boolean_t encroot, unmounted = B_FALSE;

	/* check to see if we need to unmount the filesystem */
	if (mountpoint != NULL \|\| ((zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM) &&
	libzfs_mnttab_find(hdl, zhp->zfs_name, &entry) == 0)) {
	/*
	* mountpoint may have come from a call to
	* getmnt/getmntany if it isn't NULL. If it is NULL,
	* we know it comes from libzfs_mnttab_find which can
	* then get freed later. We strdup it to play it safe.
	*/
	if (mountpoint == NULL)
	mntpt = zfs_strdup(hdl, entry.mnt_mountp);
	else
	mntpt = zfs_strdup(hdl, mountpoint);

	/*
	* Unshare and unmount the filesystem
	*/
	if (zfs_unshare_proto(zhp, mntpt, share_all_proto) != 0) {
	free(mntpt);
	return (-1);
	}
	zfs_commit_all_shares();

	if (unmount_one(zhp, mntpt, flags) != 0) {
	free(mntpt);
	(void) zfs_shareall(zhp);
	zfs_commit_all_shares();
	return (-1);
	}

	libzfs_mnttab_remove(hdl, zhp->zfs_name);
	free(mntpt);
	unmounted = B_TRUE;
	}

	/*
	* If the MS_CRYPT flag is provided we must ensure we attempt to
	* unload the dataset's key regardless of whether we did any work
	* to unmount it. We only do this for encryption roots.
	*/
	if ((flags & MS_CRYPT) != 0 &&
	zfs_prop_get_int(zhp, ZFS_PROP_ENCRYPTION) != ZIO_CRYPT_OFF) {
	zfs_refresh_properties(zhp);

	if (zfs_crypto_get_encryption_root(zhp, &encroot, NULL) != 0 &&
	unmounted) {
	(void) zfs_mount(zhp, NULL, 0);
	return (-1);
	}

	if (encroot && zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS) ==
	ZFS_KEYSTATUS_AVAILABLE &&
	zfs_crypto_unload_key(zhp) != 0) {
	(void) zfs_mount(zhp, NULL, 0);
	return (-1);
	}
	}

	zpool_disable_volume_os(zhp->zfs_name);

	return (0);
	}

	/*
	* Unmount this filesystem and any children inheriting the mountpoint property.
	* To do this, just act like we're changing the mountpoint property, but don't
	* remount the filesystems afterwards.
	*/
	int
	zfs_unmountall(zfs_handle_t *zhp, int flags)
	{
	prop_changelist_t *clp;
	int ret;

	clp = changelist_gather(zhp, ZFS_PROP_MOUNTPOINT,
	CL_GATHER_ITER_MOUNTED, flags);
	if (clp == NULL)
	return (-1);

	ret = changelist_prefix(clp);
	changelist_free(clp);

	return (ret);
	}

	boolean_t
	zfs_is_shared(zfs_handle_t *zhp)
	{
	zfs_share_type_t rc = 0;
	zfs_share_proto_t *curr_proto;

	if (ZFS_IS_VOLUME(zhp))
	return (B_FALSE);

	for (curr_proto = share_all_proto; *curr_proto != PROTO_END;
	curr_proto++)
	rc \|= zfs_is_shared_proto(zhp, NULL, *curr_proto);

	return (rc ? B_TRUE : B_FALSE);
	}

	/*
	* Unshare a filesystem by mountpoint.
	*/
	int
	unshare_one(libzfs_handle_t hdl, const char name, const char *mountpoint,
	zfs_share_proto_t proto)
	{
	int err;

	err = sa_disable_share(mountpoint, proto_table[proto].p_name);
	if (err != SA_OK) {
	return (zfs_error_fmt(hdl, proto_table[proto].p_unshare_err,
	dgettext(TEXT_DOMAIN, "cannot unshare '%s': %s"),
	name, sa_errorstr(err)));
	}
	return (0);
	}

	/*
	* Query libshare for the given mountpoint and protocol, returning
	* a zfs_share_type_t value.
	*/
	zfs_share_type_t
	is_shared(const char *mountpoint, zfs_share_proto_t proto)
	{
	if (sa_is_shared(mountpoint, proto_table[proto].p_name)) {
	switch (proto) {
	case PROTO_NFS:
	return (SHARED_NFS);
	case PROTO_SMB:
	return (SHARED_SMB);
	default:
	return (SHARED_NOT_SHARED);
	}
	}
	return (SHARED_NOT_SHARED);
	}

	/*
	* Share the given filesystem according to the options in the specified
	* protocol specific properties (sharenfs, sharesmb). We rely
	* on "libshare" to do the dirty work for us.
	*/
	int
	zfs_share_proto(zfs_handle_t zhp, zfs_share_proto_t proto)
	{
	char mountpoint[ZFS_MAXPROPLEN];
	char shareopts[ZFS_MAXPROPLEN];
	char sourcestr[ZFS_MAXPROPLEN];
	zfs_share_proto_t *curr_proto;
	zprop_source_t sourcetype;
	int err = 0;

	if (!zfs_is_mountable(zhp, mountpoint, sizeof (mountpoint), NULL, 0))
	return (0);

	for (curr_proto = proto; *curr_proto != PROTO_END; curr_proto++) {
	/*
	* Return success if there are no share options.
	*/
	if (zfs_prop_get(zhp, proto_table[*curr_proto].p_prop,
	shareopts, sizeof (shareopts), &sourcetype, sourcestr,
	ZFS_MAXPROPLEN, B_FALSE) != 0 \|\|
	strcmp(shareopts, "off") == 0)
	continue;

	/*
	* If the 'zoned' property is set, then zfs_is_mountable()
	* will have already bailed out if we are in the global zone.
	* But local zones cannot be NFS servers, so we ignore it for
	* local zones as well.
	*/
	if (zfs_prop_get_int(zhp, ZFS_PROP_ZONED))
	continue;

	err = sa_enable_share(zfs_get_name(zhp), mountpoint, shareopts,
	proto_table[*curr_proto].p_name);
	if (err != SA_OK) {
	return (zfs_error_fmt(zhp->zfs_hdl,
	proto_table[*curr_proto].p_share_err,
	dgettext(TEXT_DOMAIN, "cannot share '%s: %s'"),
	zfs_get_name(zhp), sa_errorstr(err)));
	}

	}
	return (0);
	}

	int
	zfs_share(zfs_handle_t *zhp)
	{
	assert(!ZFS_IS_VOLUME(zhp));
	return (zfs_share_proto(zhp, share_all_proto));
	}

	int
	zfs_unshare(zfs_handle_t *zhp)
	{
	assert(!ZFS_IS_VOLUME(zhp));
	return (zfs_unshareall(zhp));
	}

	/*
	* Check to see if the filesystem is currently shared.
	*/
	static zfs_share_type_t
	zfs_is_shared_proto(zfs_handle_t zhp, char *where, zfs_share_proto_t proto)
	{
	char *mountpoint;
	zfs_share_type_t rc;

	if (!zfs_is_mounted(zhp, &mountpoint))
	return (SHARED_NOT_SHARED);

	if ((rc = is_shared(mountpoint, proto))
	!= SHARED_NOT_SHARED) {
	if (where != NULL)
	*where = mountpoint;
	else
	free(mountpoint);
	return (rc);
	} else {
	free(mountpoint);
	return (SHARED_NOT_SHARED);
	}
	}

	boolean_t
	zfs_is_shared_nfs(zfs_handle_t zhp, char *where)
	{
	return (zfs_is_shared_proto(zhp, where,
	PROTO_NFS) != SHARED_NOT_SHARED);
	}

	boolean_t
	zfs_is_shared_smb(zfs_handle_t zhp, char *where)
	{
	return (zfs_is_shared_proto(zhp, where,
	PROTO_SMB) != SHARED_NOT_SHARED);
	}

	/*
	* zfs_parse_options(options, proto)
	*
	* Call the legacy parse interface to get the protocol specific
	* options using the NULL arg to indicate that this is a "parse" only.
	*/
	int
	zfs_parse_options(char *options, zfs_share_proto_t proto)
	{
	return (sa_validate_shareopts(options, proto_table[proto].p_name));
	}

	void
	zfs_commit_proto(zfs_share_proto_t *proto)
	{
	zfs_share_proto_t *curr_proto;
	for (curr_proto = proto; *curr_proto != PROTO_END; curr_proto++) {
	sa_commit_shares(proto_table[*curr_proto].p_name);
	}
	}

	void
	zfs_commit_nfs_shares(void)
	{
	zfs_commit_proto(nfs_only);
	}

	void
	zfs_commit_smb_shares(void)
	{
	zfs_commit_proto(smb_only);
	}

	void
	zfs_commit_all_shares(void)
	{
	zfs_commit_proto(share_all_proto);
	}

	void
	zfs_commit_shares(const char *proto)
	{
	if (proto == NULL)
	zfs_commit_proto(share_all_proto);
	else if (strcmp(proto, "nfs") == 0)
	zfs_commit_proto(nfs_only);
	else if (strcmp(proto, "smb") == 0)
	zfs_commit_proto(smb_only);
	}

	int
	zfs_share_nfs(zfs_handle_t *zhp)
	{
	return (zfs_share_proto(zhp, nfs_only));
	}

	int
	zfs_share_smb(zfs_handle_t *zhp)
	{
	return (zfs_share_proto(zhp, smb_only));
	}

	int
	zfs_shareall(zfs_handle_t *zhp)
	{
	return (zfs_share_proto(zhp, share_all_proto));
	}

	/*
	* Unshare the given filesystem.
	*/
	int
	zfs_unshare_proto(zfs_handle_t zhp, const char mountpoint,
	zfs_share_proto_t *proto)
	{
	libzfs_handle_t *hdl = zhp->zfs_hdl;
	struct mnttab entry;
	char *mntpt = NULL;

	/* check to see if need to unmount the filesystem */
	if (mountpoint != NULL)
	mntpt = zfs_strdup(hdl, mountpoint);

	if (mountpoint != NULL \|\| ((zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM) &&
	libzfs_mnttab_find(hdl, zfs_get_name(zhp), &entry) == 0)) {
	zfs_share_proto_t *curr_proto;

	if (mountpoint == NULL)
	mntpt = zfs_strdup(zhp->zfs_hdl, entry.mnt_mountp);

	for (curr_proto = proto; *curr_proto != PROTO_END;
	curr_proto++) {

	if (is_shared(mntpt, *curr_proto)) {
	if (unshare_one(hdl, zhp->zfs_name,
	mntpt, *curr_proto) != 0) {
	if (mntpt != NULL)
	free(mntpt);
	return (-1);
	}
	}
	}
	}
	if (mntpt != NULL)
	free(mntpt);

	return (0);
	}

	int
	zfs_unshare_nfs(zfs_handle_t zhp, const char mountpoint)
	{
	return (zfs_unshare_proto(zhp, mountpoint, nfs_only));
	}

	int
	zfs_unshare_smb(zfs_handle_t zhp, const char mountpoint)
	{
	return (zfs_unshare_proto(zhp, mountpoint, smb_only));
	}

	/*
	* Same as zfs_unmountall(), but for NFS and SMB unshares.
	*/
	static int
	zfs_unshareall_proto(zfs_handle_t zhp, zfs_share_proto_t proto)
	{
	prop_changelist_t *clp;
	int ret;

	clp = changelist_gather(zhp, ZFS_PROP_SHARENFS, 0, 0);
	if (clp == NULL)
	return (-1);

	ret = changelist_unshare(clp, proto);
	changelist_free(clp);

	return (ret);
	}

	int
	zfs_unshareall_nfs(zfs_handle_t *zhp)
	{
	return (zfs_unshareall_proto(zhp, nfs_only));
	}

	int
	zfs_unshareall_smb(zfs_handle_t *zhp)
	{
	return (zfs_unshareall_proto(zhp, smb_only));
	}

	int
	zfs_unshareall(zfs_handle_t *zhp)
	{
	return (zfs_unshareall_proto(zhp, share_all_proto));
	}

	int
	zfs_unshareall_bypath(zfs_handle_t zhp, const char mountpoint)
	{
	return (zfs_unshare_proto(zhp, mountpoint, share_all_proto));
	}

	int
	zfs_unshareall_bytype(zfs_handle_t zhp, const char mountpoint,
	const char *proto)
	{
	if (proto == NULL)
	return (zfs_unshare_proto(zhp, mountpoint, share_all_proto));
	if (strcmp(proto, "nfs") == 0)
	return (zfs_unshare_proto(zhp, mountpoint, nfs_only));
	else if (strcmp(proto, "smb") == 0)
	return (zfs_unshare_proto(zhp, mountpoint, smb_only));
	else
	return (1);
	}

	/*
	* Remove the mountpoint associated with the current dataset, if necessary.
	* We only remove the underlying directory if:
	*
	* - The mountpoint is not 'none' or 'legacy'
	* - The mountpoint is non-empty
	* - The mountpoint is the default or inherited
	* - The 'zoned' property is set, or we're in a local zone
	*
	* Any other directories we leave alone.
	*/
	void
	remove_mountpoint(zfs_handle_t *zhp)
	{
	char mountpoint[ZFS_MAXPROPLEN];
	zprop_source_t source;

	if (!zfs_is_mountable(zhp, mountpoint, sizeof (mountpoint),
	&source, 0))
	return;

	if (source == ZPROP_SRC_DEFAULT \|\|
	source == ZPROP_SRC_INHERITED) {
	/*
	* Try to remove the directory, silently ignoring any errors.
	* The filesystem may have since been removed or moved around,
	* and this error isn't really useful to the administrator in
	* any way.
	*/
	(void) rmdir(mountpoint);
	}
	}

	/*
	* Add the given zfs handle to the cb_handles array, dynamically reallocating
	* the array if it is out of space.
	*/
	void
	libzfs_add_handle(get_all_cb_t cbp, zfs_handle_t zhp)
	{
	if (cbp->cb_alloc == cbp->cb_used) {
	size_t newsz;
	zfs_handle_t **newhandles;

	newsz = cbp->cb_alloc != 0 ? cbp->cb_alloc * 2 : 64;
	newhandles = zfs_realloc(zhp->zfs_hdl,
	cbp->cb_handles, cbp->cb_alloc * sizeof (zfs_handle_t *),
	newsz * sizeof (zfs_handle_t *));
	cbp->cb_handles = newhandles;
	cbp->cb_alloc = newsz;
	}
	cbp->cb_handles[cbp->cb_used++] = zhp;
	}

	/*
	* Recursive helper function used during file system enumeration
	*/
	static int
	zfs_iter_cb(zfs_handle_t zhp, void data)
	{
	get_all_cb_t *cbp = data;

	if (!(zfs_get_type(zhp) & ZFS_TYPE_FILESYSTEM)) {
	zfs_close(zhp);
	return (0);
	}

	if (zfs_prop_get_int(zhp, ZFS_PROP_CANMOUNT) == ZFS_CANMOUNT_NOAUTO) {
	zfs_close(zhp);
	return (0);
	}

	if (zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS) ==
	ZFS_KEYSTATUS_UNAVAILABLE) {
	zfs_close(zhp);
	return (0);
	}

	/*
	* If this filesystem is inconsistent and has a receive resume
	* token, we can not mount it.
	*/
	if (zfs_prop_get_int(zhp, ZFS_PROP_INCONSISTENT) &&
	zfs_prop_get(zhp, ZFS_PROP_RECEIVE_RESUME_TOKEN,
	NULL, 0, NULL, NULL, 0, B_TRUE) == 0) {
	zfs_close(zhp);
	return (0);
	}

	libzfs_add_handle(cbp, zhp);
	if (zfs_iter_filesystems(zhp, zfs_iter_cb, cbp) != 0) {
	zfs_close(zhp);
	return (-1);
	}
	return (0);
	}

	/*
	* Sort comparator that compares two mountpoint paths. We sort these paths so
	* that subdirectories immediately follow their parents. This means that we
	* effectively treat the '/' character as the lowest value non-nul char.
	* Since filesystems from non-global zones can have the same mountpoint
	* as other filesystems, the comparator sorts global zone filesystems to
	* the top of the list. This means that the global zone will traverse the
	* filesystem list in the correct order and can stop when it sees the
	* first zoned filesystem. In a non-global zone, only the delegated
	* filesystems are seen.
	*
	* An example sorted list using this comparator would look like:
	*
	* /foo
	* /foo/bar
	* /foo/bar/baz
	* /foo/baz
	* /foo.bar
	* /foo (NGZ1)
	* /foo (NGZ2)
	*
	* The mounting code depends on this ordering to deterministically iterate
	* over filesystems in order to spawn parallel mount tasks.
	*/
	static int
	mountpoint_cmp(const void arga, const void argb)
	{
	zfs_handle_t const zap = arga;
	zfs_handle_t za = zap;
	zfs_handle_t const zbp = argb;
	zfs_handle_t zb = zbp;
	char mounta[MAXPATHLEN];
	char mountb[MAXPATHLEN];
	const char *a = mounta;
	const char *b = mountb;
	boolean_t gota, gotb;
	uint64_t zoneda, zonedb;

	zoneda = zfs_prop_get_int(za, ZFS_PROP_ZONED);
	zonedb = zfs_prop_get_int(zb, ZFS_PROP_ZONED);
	if (zoneda && !zonedb)
	return (1);
	if (!zoneda && zonedb)
	return (-1);

	gota = (zfs_get_type(za) == ZFS_TYPE_FILESYSTEM);
	if (gota) {
	verify(zfs_prop_get(za, ZFS_PROP_MOUNTPOINT, mounta,
	sizeof (mounta), NULL, NULL, 0, B_FALSE) == 0);
	}
	gotb = (zfs_get_type(zb) == ZFS_TYPE_FILESYSTEM);
	if (gotb) {
	verify(zfs_prop_get(zb, ZFS_PROP_MOUNTPOINT, mountb,
	sizeof (mountb), NULL, NULL, 0, B_FALSE) == 0);
	}

	if (gota && gotb) {
	while (a != '\0' && (a == *b)) {
	a++;
	b++;
	}
	if (a == b)
	return (0);
	if (*a == '\0')
	return (-1);
	if (*b == '\0')
	return (1);
	if (*a == '/')
	return (-1);
	if (*b == '/')
	return (1);
	return (a < b ? -1 : a > b);
	}

	if (gota)
	return (-1);
	if (gotb)
	return (1);

	/*
	* If neither filesystem has a mountpoint, revert to sorting by
	* dataset name.
	*/
	return (strcmp(zfs_get_name(za), zfs_get_name(zb)));
	}

	/*
	* Return true if path2 is a child of path1 or path2 equals path1 or
	* path1 is "/" (path2 is always a child of "/").
	*/
	static boolean_t
	libzfs_path_contains(const char path1, const char path2)
	{
	return (strcmp(path1, path2) == 0 \|\| strcmp(path1, "/") == 0 \|\|
	(strstr(path2, path1) == path2 && path2[strlen(path1)] == '/'));
	}

	/*
	* Given a mountpoint specified by idx in the handles array, find the first
	* non-descendent of that mountpoint and return its index. Descendant paths
	* start with the parent's path. This function relies on the ordering
	* enforced by mountpoint_cmp().
	*/
	static int
	non_descendant_idx(zfs_handle_t **handles, size_t num_handles, int idx)
	{
	char parent[ZFS_MAXPROPLEN];
	char child[ZFS_MAXPROPLEN];
	int i;

	verify(zfs_prop_get(handles[idx], ZFS_PROP_MOUNTPOINT, parent,
	sizeof (parent), NULL, NULL, 0, B_FALSE) == 0);

	for (i = idx + 1; i < num_handles; i++) {
	verify(zfs_prop_get(handles[i], ZFS_PROP_MOUNTPOINT, child,
	sizeof (child), NULL, NULL, 0, B_FALSE) == 0);
	if (!libzfs_path_contains(parent, child))
	break;
	}
	return (i);
	}

	typedef struct mnt_param {
	libzfs_handle_t *mnt_hdl;
	tpool_t *mnt_tp;
	zfs_handle_t *mnt_zhps; / filesystems to mount */
	size_t mnt_num_handles;
	int mnt_idx; /* Index of selected entry to mount */
	zfs_iter_f mnt_func;
	void *mnt_data;
	} mnt_param_t;

	/*
	* Allocate and populate the parameter struct for mount function, and
	* schedule mounting of the entry selected by idx.
	*/
	static void
	zfs_dispatch_mount(libzfs_handle_t hdl, zfs_handle_t *handles,
	size_t num_handles, int idx, zfs_iter_f func, void data, tpool_t tp)
	{
	mnt_param_t *mnt_param = zfs_alloc(hdl, sizeof (mnt_param_t));

	mnt_param->mnt_hdl = hdl;
	mnt_param->mnt_tp = tp;
	mnt_param->mnt_zhps = handles;
	mnt_param->mnt_num_handles = num_handles;
	mnt_param->mnt_idx = idx;
	mnt_param->mnt_func = func;
	mnt_param->mnt_data = data;

	(void) tpool_dispatch(tp, zfs_mount_task, (void*)mnt_param);
	}

	/*
	* This is the structure used to keep state of mounting or sharing operations
	* during a call to zpool_enable_datasets().
	*/
	typedef struct mount_state {
	/*
	* ms_mntstatus is set to -1 if any mount fails. While multiple threads
	* could update this variable concurrently, no synchronization is
	* needed as it's only ever set to -1.
	*/
	int ms_mntstatus;
	int ms_mntflags;
	const char *ms_mntopts;
	} mount_state_t;

	static int
	zfs_mount_one(zfs_handle_t zhp, void arg)
	{
	mount_state_t *ms = arg;
	int ret = 0;

	/*
	* don't attempt to mount encrypted datasets with
	* unloaded keys
	*/
	if (zfs_prop_get_int(zhp, ZFS_PROP_KEYSTATUS) ==
	ZFS_KEYSTATUS_UNAVAILABLE)
	return (0);

	if (zfs_mount(zhp, ms->ms_mntopts, ms->ms_mntflags) != 0)
	ret = ms->ms_mntstatus = -1;
	return (ret);
	}

	static int
	zfs_share_one(zfs_handle_t zhp, void arg)
	{
	mount_state_t *ms = arg;
	int ret = 0;

	if (zfs_share(zhp) != 0)
	ret = ms->ms_mntstatus = -1;
	return (ret);
	}

	/*
	* Thread pool function to mount one file system. On completion, it finds and
	* schedules its children to be mounted. This depends on the sorting done in
	* zfs_foreach_mountpoint(). Note that the degenerate case (chain of entries
	* each descending from the previous) will have no parallelism since we always
	* have to wait for the parent to finish mounting before we can schedule
	* its children.
	*/
	static void
	zfs_mount_task(void *arg)
	{
	mnt_param_t *mp = arg;
	int idx = mp->mnt_idx;
	zfs_handle_t **handles = mp->mnt_zhps;
	size_t num_handles = mp->mnt_num_handles;
	char mountpoint[ZFS_MAXPROPLEN];

	verify(zfs_prop_get(handles[idx], ZFS_PROP_MOUNTPOINT, mountpoint,
	sizeof (mountpoint), NULL, NULL, 0, B_FALSE) == 0);

	if (mp->mnt_func(handles[idx], mp->mnt_data) != 0)
	return;

	/*
	* We dispatch tasks to mount filesystems with mountpoints underneath
	* this one. We do this by dispatching the next filesystem with a
	* descendant mountpoint of the one we just mounted, then skip all of
	* its descendants, dispatch the next descendant mountpoint, and so on.
	* The non_descendant_idx() function skips over filesystems that are
	* descendants of the filesystem we just dispatched.
	*/
	for (int i = idx + 1; i < num_handles;
	i = non_descendant_idx(handles, num_handles, i)) {
	char child[ZFS_MAXPROPLEN];
	verify(zfs_prop_get(handles[i], ZFS_PROP_MOUNTPOINT,
	child, sizeof (child), NULL, NULL, 0, B_FALSE) == 0);

	if (!libzfs_path_contains(mountpoint, child))
	break; /* not a descendant, return */
	zfs_dispatch_mount(mp->mnt_hdl, handles, num_handles, i,
	mp->mnt_func, mp->mnt_data, mp->mnt_tp);
	}
	free(mp);
	}

	/*
	* Issue the func callback for each ZFS handle contained in the handles
	* array. This function is used to mount all datasets, and so this function
	* guarantees that filesystems for parent mountpoints are called before their
	* children. As such, before issuing any callbacks, we first sort the array
	* of handles by mountpoint.
	*
	* Callbacks are issued in one of two ways:
	*
	* 1. Sequentially: If the parallel argument is B_FALSE or the ZFS_SERIAL_MOUNT
	* environment variable is set, then we issue callbacks sequentially.
	*
	* 2. In parallel: If the parallel argument is B_TRUE and the ZFS_SERIAL_MOUNT
	* environment variable is not set, then we use a tpool to dispatch threads
	* to mount filesystems in parallel. This function dispatches tasks to mount
	* the filesystems at the top-level mountpoints, and these tasks in turn
	* are responsible for recursively mounting filesystems in their children
	* mountpoints.
	*/
	void
	zfs_foreach_mountpoint(libzfs_handle_t hdl, zfs_handle_t *handles,
	size_t num_handles, zfs_iter_f func, void *data, boolean_t parallel)
	{
	zoneid_t zoneid = getzoneid();

	/*
	* The ZFS_SERIAL_MOUNT environment variable is an undocumented
	* variable that can be used as a convenience to do a/b comparison
	* of serial vs. parallel mounting.
	*/
	boolean_t serial_mount = !parallel \|\|
	(getenv("ZFS_SERIAL_MOUNT") != NULL);

	/*
	* Sort the datasets by mountpoint. See mountpoint_cmp for details
	* of how these are sorted.
	*/
	qsort(handles, num_handles, sizeof (zfs_handle_t *), mountpoint_cmp);

	if (serial_mount) {
	for (int i = 0; i < num_handles; i++) {
	func(handles[i], data);
	}
	return;
	}

	/*
	* Issue the callback function for each dataset using a parallel
	* algorithm that uses a thread pool to manage threads.
	*/
	tpool_t *tp = tpool_create(1, mount_tp_nthr, 0, NULL);

	/*
	* There may be multiple "top level" mountpoints outside of the pool's
	* root mountpoint, e.g.: /foo /bar. Dispatch a mount task for each of
	* these.
	*/
	for (int i = 0; i < num_handles;
	i = non_descendant_idx(handles, num_handles, i)) {
	/*
	* Since the mountpoints have been sorted so that the zoned
	* filesystems are at the end, a zoned filesystem seen from
	* the global zone means that we're done.
	*/
	if (zoneid == GLOBAL_ZONEID &&
	zfs_prop_get_int(handles[i], ZFS_PROP_ZONED))
	break;
	zfs_dispatch_mount(hdl, handles, num_handles, i, func, data,
	tp);
	}

	tpool_wait(tp); /* wait for all scheduled mounts to complete */
	tpool_destroy(tp);
	}

	/*
	* Mount and share all datasets within the given pool. This assumes that no
	* datasets within the pool are currently mounted.
	*/
	int
	zpool_enable_datasets(zpool_handle_t zhp, const char mntopts, int flags)
	{
	get_all_cb_t cb = { 0 };
	mount_state_t ms = { 0 };
	zfs_handle_t *zfsp;
	int ret = 0;

	if ((zfsp = zfs_open(zhp->zpool_hdl, zhp->zpool_name,
	ZFS_TYPE_DATASET)) == NULL)
	goto out;

	/*
	* Gather all non-snapshot datasets within the pool. Start by adding
	* the root filesystem for this pool to the list, and then iterate
	* over all child filesystems.
	*/
	libzfs_add_handle(&cb, zfsp);
	if (zfs_iter_filesystems(zfsp, zfs_iter_cb, &cb) != 0)
	goto out;

	/*
	* Mount all filesystems
	*/
	ms.ms_mntopts = mntopts;
	ms.ms_mntflags = flags;
	zfs_foreach_mountpoint(zhp->zpool_hdl, cb.cb_handles, cb.cb_used,
	zfs_mount_one, &ms, B_TRUE);
	if (ms.ms_mntstatus != 0)
	ret = ms.ms_mntstatus;

	/*
	* Share all filesystems that need to be shared. This needs to be
	* a separate pass because libshare is not mt-safe, and so we need
	* to share serially.
	*/
	ms.ms_mntstatus = 0;
	zfs_foreach_mountpoint(zhp->zpool_hdl, cb.cb_handles, cb.cb_used,
	zfs_share_one, &ms, B_FALSE);
	if (ms.ms_mntstatus != 0)
	ret = ms.ms_mntstatus;
	else
	zfs_commit_all_shares();

	out:
	for (int i = 0; i < cb.cb_used; i++)
	zfs_close(cb.cb_handles[i]);
	free(cb.cb_handles);

	return (ret);
	}

	struct sets_s {
	char *mountpoint;
	zfs_handle_t *dataset;
	};

	static int
	mountpoint_compare(const void a, const void b)
	{
	const struct sets_s mounta = (struct sets_s )a;
	const struct sets_s mountb = (struct sets_s )b;

	return (strcmp(mountb->mountpoint, mounta->mountpoint));
	}

	/*
	* Unshare and unmount all datasets within the given pool. We don't want to
	* rely on traversing the DSL to discover the filesystems within the pool,
	* because this may be expensive (if not all of them are mounted), and can fail
	* arbitrarily (on I/O error, for example). Instead, we walk /proc/self/mounts
	* and gather all the filesystems that are currently mounted.
	*/
	int
	zpool_disable_datasets(zpool_handle_t *zhp, boolean_t force)
	{
	int used, alloc;
	FILE *mnttab;
	struct mnttab entry;
	size_t namelen;
	struct sets_s *sets = NULL;
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	int i;
	int ret = -1;
	int flags = (force ? MS_FORCE : 0);

	namelen = strlen(zhp->zpool_name);

	if ((mnttab = fopen(MNTTAB, "re")) == NULL)
	return (ENOENT);

	used = alloc = 0;
	while (getmntent(mnttab, &entry) == 0) {
	/*
	* Ignore non-ZFS entries.
	*/
	if (entry.mnt_fstype == NULL \|\|
	strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0)
	continue;

	/*
	* Ignore filesystems not within this pool.
	*/
	if (entry.mnt_mountp == NULL \|\|
	strncmp(entry.mnt_special, zhp->zpool_name, namelen) != 0 \|\|
	(entry.mnt_special[namelen] != '/' &&
	entry.mnt_special[namelen] != '\0'))
	continue;

	/*
	* At this point we've found a filesystem within our pool. Add
	* it to our growing list.
	*/
	if (used == alloc) {
	if (alloc == 0) {

	if ((sets = zfs_alloc(hdl,
	8 * sizeof (struct sets_s))) == NULL)
	goto out;

	alloc = 8;
	} else {
	void *ptr;

	if ((ptr = zfs_realloc(hdl, sets,
	alloc * sizeof (struct sets_s),
	alloc * 2 * sizeof (struct sets_s)))
	== NULL)
	goto out;
	sets = ptr;

	alloc *= 2;
	}
	}

	if ((sets[used].mountpoint = zfs_strdup(hdl,
	entry.mnt_mountp)) == NULL)
	goto out;

	/*
	* This is allowed to fail, in case there is some I/O error. It
	* is only used to determine if we need to remove the underlying
	* mountpoint, so failure is not fatal.
	*/
	sets[used].dataset = make_dataset_handle(hdl,
	entry.mnt_special);

	used++;
	}

	/*
	* At this point, we have the entire list of filesystems, so sort it by
	* mountpoint.
	*/
	qsort(sets, used, sizeof (struct sets_s), mountpoint_compare);

	/*
	* Walk through and first unshare everything.
	*/
	for (i = 0; i < used; i++) {
	zfs_share_proto_t *curr_proto;
	for (curr_proto = share_all_proto; *curr_proto != PROTO_END;
	curr_proto++) {
	if (is_shared(sets[i].mountpoint, *curr_proto) &&
	unshare_one(hdl, sets[i].mountpoint,
	sets[i].mountpoint, *curr_proto) != 0)
	goto out;
	}
	}
	zfs_commit_all_shares();

	/*
	* Now unmount everything, removing the underlying directories as
	* appropriate.
	*/
	for (i = 0; i < used; i++) {
	if (unmount_one(sets[i].dataset, sets[i].mountpoint,
	flags) != 0)
	goto out;
	}

	for (i = 0; i < used; i++) {
	if (sets[i].dataset)
	remove_mountpoint(sets[i].dataset);
	}

	zpool_disable_datasets_os(zhp, force);

	ret = 0;
	out:
	(void) fclose(mnttab);
	for (i = 0; i < used; i++) {
	if (sets[i].dataset)
	zfs_close(sets[i].dataset);
	free(sets[i].mountpoint);
	}
	free(sets);

	return (ret);
	}
	diff --git a/sys/contrib/openzfs/lib/libzfs/libzfs_pool.c b/sys/contrib/openzfs/lib/libzfs/libzfs_pool.c
	index 8ed96275c4db..2179c03d76ed 100644
	--- a/sys/contrib/openzfs/lib/libzfs/libzfs_pool.c
	+++ b/sys/contrib/openzfs/lib/libzfs/libzfs_pool.c
	@@ -1,4961 +1,5413 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright 2015 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>
	* Copyright (c) 2018 Datto Inc.
	* Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
	* Copyright (c) 2017, Intel Corporation.
	* Copyright (c) 2018, loli10K <ezomori.nozomu@gmail.com>
	* Copyright (c) 2021, Colm Buckley <colm@tuatha.org>
	+ * Copyright (c) 2021, Klara Inc.
	*/

	#include <errno.h>
	#include <libintl.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <strings.h>
	#include <unistd.h>
	#include <libgen.h>
	#include <zone.h>
	#include <sys/stat.h>
	#include <sys/efi_partition.h>
	#include <sys/systeminfo.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/zfs_sysfs.h>
	#include <sys/vdev_disk.h>
	#include <sys/types.h>
	#include <dlfcn.h>
	#include <libzutil.h>
	#include <fcntl.h>

	#include "zfs_namecheck.h"
	#include "zfs_prop.h"
	#include "libzfs_impl.h"
	#include "zfs_comutil.h"
	#include "zfeature_common.h"

	static boolean_t zpool_vdev_is_interior(const char *name);

	typedef struct prop_flags {
	int create:1; /* Validate property on creation */
	int import:1; /* Validate property on import */
	+ int vdevprop:1; /* Validate property as a VDEV property */
	} prop_flags_t;

	/*
	* ====================================================================
	* zpool property functions
	* ====================================================================
	*/

	static int
	zpool_get_all_props(zpool_handle_t *zhp)
	{
	zfs_cmd_t zc = {"\0"};
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));

	if (zcmd_alloc_dst_nvlist(hdl, &zc, 0) != 0)
	return (-1);

	while (zfs_ioctl(hdl, ZFS_IOC_POOL_GET_PROPS, &zc) != 0) {
	if (errno == ENOMEM) {
	if (zcmd_expand_dst_nvlist(hdl, &zc) != 0) {
	zcmd_free_nvlists(&zc);
	return (-1);
	}
	} else {
	zcmd_free_nvlists(&zc);
	return (-1);
	}
	}

	if (zcmd_read_dst_nvlist(hdl, &zc, &zhp->zpool_props) != 0) {
	zcmd_free_nvlists(&zc);
	return (-1);
	}

	zcmd_free_nvlists(&zc);

	return (0);
	}

	int
	zpool_props_refresh(zpool_handle_t *zhp)
	{
	nvlist_t *old_props;

	old_props = zhp->zpool_props;

	if (zpool_get_all_props(zhp) != 0)
	return (-1);

	nvlist_free(old_props);
	return (0);
	}

	static const char *
	zpool_get_prop_string(zpool_handle_t *zhp, zpool_prop_t prop,
	zprop_source_t *src)
	{
	nvlist_t nv, nvl;
	uint64_t ival;
	char *value;
	zprop_source_t source;

	nvl = zhp->zpool_props;
	if (nvlist_lookup_nvlist(nvl, zpool_prop_to_name(prop), &nv) == 0) {
	verify(nvlist_lookup_uint64(nv, ZPROP_SOURCE, &ival) == 0);
	source = ival;
	verify(nvlist_lookup_string(nv, ZPROP_VALUE, &value) == 0);
	} else {
	source = ZPROP_SRC_DEFAULT;
	if ((value = (char *)zpool_prop_default_string(prop)) == NULL)
	value = "-";
	}

	if (src)
	*src = source;

	return (value);
	}

	uint64_t
	zpool_get_prop_int(zpool_handle_t zhp, zpool_prop_t prop, zprop_source_t src)
	{
	nvlist_t nv, nvl;
	uint64_t value;
	zprop_source_t source;

	if (zhp->zpool_props == NULL && zpool_get_all_props(zhp)) {
	/*
	* zpool_get_all_props() has most likely failed because
	* the pool is faulted, but if all we need is the top level
	* vdev's guid then get it from the zhp config nvlist.
	*/
	if ((prop == ZPOOL_PROP_GUID) &&
	(nvlist_lookup_nvlist(zhp->zpool_config,
	ZPOOL_CONFIG_VDEV_TREE, &nv) == 0) &&
	(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &value)
	== 0)) {
	return (value);
	}
	return (zpool_prop_default_numeric(prop));
	}

	nvl = zhp->zpool_props;
	if (nvlist_lookup_nvlist(nvl, zpool_prop_to_name(prop), &nv) == 0) {
	verify(nvlist_lookup_uint64(nv, ZPROP_SOURCE, &value) == 0);
	source = value;
	verify(nvlist_lookup_uint64(nv, ZPROP_VALUE, &value) == 0);
	} else {
	source = ZPROP_SRC_DEFAULT;
	value = zpool_prop_default_numeric(prop);
	}

	if (src)
	*src = source;

	return (value);
	}

	/*
	* Map VDEV STATE to printed strings.
	*/
	const char *
	zpool_state_to_name(vdev_state_t state, vdev_aux_t aux)
	{
	switch (state) {
	case VDEV_STATE_CLOSED:
	case VDEV_STATE_OFFLINE:
	return (gettext("OFFLINE"));
	case VDEV_STATE_REMOVED:
	return (gettext("REMOVED"));
	case VDEV_STATE_CANT_OPEN:
	if (aux == VDEV_AUX_CORRUPT_DATA \|\| aux == VDEV_AUX_BAD_LOG)
	return (gettext("FAULTED"));
	else if (aux == VDEV_AUX_SPLIT_POOL)
	return (gettext("SPLIT"));
	else
	return (gettext("UNAVAIL"));
	case VDEV_STATE_FAULTED:
	return (gettext("FAULTED"));
	case VDEV_STATE_DEGRADED:
	return (gettext("DEGRADED"));
	case VDEV_STATE_HEALTHY:
	return (gettext("ONLINE"));

	default:
	break;
	}

	return (gettext("UNKNOWN"));
	}

	/*
	* Map POOL STATE to printed strings.
	*/
	const char *
	zpool_pool_state_to_name(pool_state_t state)
	{
	switch (state) {
	default:
	break;
	case POOL_STATE_ACTIVE:
	return (gettext("ACTIVE"));
	case POOL_STATE_EXPORTED:
	return (gettext("EXPORTED"));
	case POOL_STATE_DESTROYED:
	return (gettext("DESTROYED"));
	case POOL_STATE_SPARE:
	return (gettext("SPARE"));
	case POOL_STATE_L2CACHE:
	return (gettext("L2CACHE"));
	case POOL_STATE_UNINITIALIZED:
	return (gettext("UNINITIALIZED"));
	case POOL_STATE_UNAVAIL:
	return (gettext("UNAVAIL"));
	case POOL_STATE_POTENTIALLY_ACTIVE:
	return (gettext("POTENTIALLY_ACTIVE"));
	}

	return (gettext("UNKNOWN"));
	}

	/*
	* Given a pool handle, return the pool health string ("ONLINE", "DEGRADED",
	* "SUSPENDED", etc).
	*/
	const char *
	zpool_get_state_str(zpool_handle_t *zhp)
	{
	zpool_errata_t errata;
	zpool_status_t status;
	nvlist_t *nvroot;
	vdev_stat_t *vs;
	uint_t vsc;
	const char *str;

	status = zpool_get_status(zhp, NULL, &errata);

	if (zpool_get_state(zhp) == POOL_STATE_UNAVAIL) {
	str = gettext("FAULTED");
	} else if (status == ZPOOL_STATUS_IO_FAILURE_WAIT \|\|
	status == ZPOOL_STATUS_IO_FAILURE_MMP) {
	str = gettext("SUSPENDED");
	} else {
	verify(nvlist_lookup_nvlist(zpool_get_config(zhp, NULL),
	ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0);
	verify(nvlist_lookup_uint64_array(nvroot,
	ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs, &vsc)
	== 0);
	str = zpool_state_to_name(vs->vs_state, vs->vs_aux);
	}
	return (str);
	}

	/*
	* Get a zpool property value for 'prop' and return the value in
	* a pre-allocated buffer.
	*/
	int
	zpool_get_prop(zpool_handle_t zhp, zpool_prop_t prop, char buf,
	size_t len, zprop_source_t *srctype, boolean_t literal)
	{
	uint64_t intval;
	const char *strval;
	zprop_source_t src = ZPROP_SRC_NONE;

	if (zpool_get_state(zhp) == POOL_STATE_UNAVAIL) {
	switch (prop) {
	case ZPOOL_PROP_NAME:
	(void) strlcpy(buf, zpool_get_name(zhp), len);
	break;

	case ZPOOL_PROP_HEALTH:
	(void) strlcpy(buf, zpool_get_state_str(zhp), len);
	break;

	case ZPOOL_PROP_GUID:
	intval = zpool_get_prop_int(zhp, prop, &src);
	(void) snprintf(buf, len, "%llu", (u_longlong_t)intval);
	break;

	case ZPOOL_PROP_ALTROOT:
	case ZPOOL_PROP_CACHEFILE:
	case ZPOOL_PROP_COMMENT:
	case ZPOOL_PROP_COMPATIBILITY:
	if (zhp->zpool_props != NULL \|\|
	zpool_get_all_props(zhp) == 0) {
	(void) strlcpy(buf,
	zpool_get_prop_string(zhp, prop, &src),
	len);
	break;
	}
	fallthrough;
	default:
	(void) strlcpy(buf, "-", len);
	break;
	}

	if (srctype != NULL)
	*srctype = src;
	return (0);
	}

	if (zhp->zpool_props == NULL && zpool_get_all_props(zhp) &&
	prop != ZPOOL_PROP_NAME)
	return (-1);

	switch (zpool_prop_get_type(prop)) {
	case PROP_TYPE_STRING:
	(void) strlcpy(buf, zpool_get_prop_string(zhp, prop, &src),
	len);
	break;

	case PROP_TYPE_NUMBER:
	intval = zpool_get_prop_int(zhp, prop, &src);

	switch (prop) {
	case ZPOOL_PROP_SIZE:
	case ZPOOL_PROP_ALLOCATED:
	case ZPOOL_PROP_FREE:
	case ZPOOL_PROP_FREEING:
	case ZPOOL_PROP_LEAKED:
	case ZPOOL_PROP_ASHIFT:
	if (literal)
	(void) snprintf(buf, len, "%llu",
	(u_longlong_t)intval);
	else
	(void) zfs_nicenum(intval, buf, len);
	break;

	case ZPOOL_PROP_EXPANDSZ:
	case ZPOOL_PROP_CHECKPOINT:
	if (intval == 0) {
	(void) strlcpy(buf, "-", len);
	} else if (literal) {
	(void) snprintf(buf, len, "%llu",
	(u_longlong_t)intval);
	} else {
	(void) zfs_nicebytes(intval, buf, len);
	}
	break;

	case ZPOOL_PROP_CAPACITY:
	if (literal) {
	(void) snprintf(buf, len, "%llu",
	(u_longlong_t)intval);
	} else {
	(void) snprintf(buf, len, "%llu%%",
	(u_longlong_t)intval);
	}
	break;

	case ZPOOL_PROP_FRAGMENTATION:
	if (intval == UINT64_MAX) {
	(void) strlcpy(buf, "-", len);
	} else if (literal) {
	(void) snprintf(buf, len, "%llu",
	(u_longlong_t)intval);
	} else {
	(void) snprintf(buf, len, "%llu%%",
	(u_longlong_t)intval);
	}
	break;

	case ZPOOL_PROP_DEDUPRATIO:
	if (literal)
	(void) snprintf(buf, len, "%llu.%02llu",
	(u_longlong_t)(intval / 100),
	(u_longlong_t)(intval % 100));
	else
	(void) snprintf(buf, len, "%llu.%02llux",
	(u_longlong_t)(intval / 100),
	(u_longlong_t)(intval % 100));
	break;

	case ZPOOL_PROP_HEALTH:
	(void) strlcpy(buf, zpool_get_state_str(zhp), len);
	break;
	case ZPOOL_PROP_VERSION:
	if (intval >= SPA_VERSION_FEATURES) {
	(void) snprintf(buf, len, "-");
	break;
	}
	fallthrough;
	default:
	(void) snprintf(buf, len, "%llu", (u_longlong_t)intval);
	}
	break;

	case PROP_TYPE_INDEX:
	intval = zpool_get_prop_int(zhp, prop, &src);
	if (zpool_prop_index_to_string(prop, intval, &strval)
	!= 0)
	return (-1);
	(void) strlcpy(buf, strval, len);
	break;

	default:
	abort();
	}

	if (srctype)
	*srctype = src;

	return (0);
	}

	/*
	* Check if the bootfs name has the same pool name as it is set to.
	* Assuming bootfs is a valid dataset name.
	*/
	static boolean_t
	bootfs_name_valid(const char pool, const char bootfs)
	{
	int len = strlen(pool);
	if (bootfs[0] == '\0')
	return (B_TRUE);

	if (!zfs_name_valid(bootfs, ZFS_TYPE_FILESYSTEM\|ZFS_TYPE_SNAPSHOT))
	return (B_FALSE);

	if (strncmp(pool, bootfs, len) == 0 &&
	(bootfs[len] == '/' \|\| bootfs[len] == '\0'))
	return (B_TRUE);

	return (B_FALSE);
	}

	/*
	* Given an nvlist of zpool properties to be set, validate that they are
	* correct, and parse any numeric properties (index, boolean, etc) if they are
	* specified as strings.
	*/
	static nvlist_t *
	zpool_valid_proplist(libzfs_handle_t hdl, const char poolname,
	nvlist_t props, uint64_t version, prop_flags_t flags, char errbuf)
	{
	nvpair_t *elem;
	nvlist_t *retprops;
	zpool_prop_t prop;
	char *strval;
	uint64_t intval;
	char slash, check;
	struct stat64 statbuf;
	zpool_handle_t *zhp;
	char report[1024];

	if (nvlist_alloc(&retprops, NV_UNIQUE_NAME, 0) != 0) {
	(void) no_memory(hdl);
	return (NULL);
	}

	elem = NULL;
	while ((elem = nvlist_next_nvpair(props, elem)) != NULL) {
	const char *propname = nvpair_name(elem);

	+ if (flags.vdevprop && zpool_prop_vdev(propname)) {
	+ vdev_prop_t vprop = vdev_name_to_prop(propname);
	+
	+ if (vdev_prop_readonly(vprop)) {
	+ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' "
	+ "is readonly"), propname);
	+ (void) zfs_error(hdl, EZFS_PROPREADONLY,
	+ errbuf);
	+ goto error;
	+ }
	+
	+ if (zprop_parse_value(hdl, elem, vprop, ZFS_TYPE_VDEV,
	+ retprops, &strval, &intval, errbuf) != 0)
	+ goto error;
	+
	+ continue;
	+ } else if (flags.vdevprop && vdev_prop_user(propname)) {
	+ if (nvlist_add_nvpair(retprops, elem) != 0) {
	+ (void) no_memory(hdl);
	+ goto error;
	+ }
	+ continue;
	+ } else if (flags.vdevprop) {
	+ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	+ "invalid property: '%s'"), propname);
	+ (void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	+ goto error;
	+ }
	+
	prop = zpool_name_to_prop(propname);
	if (prop == ZPOOL_PROP_INVAL && zpool_prop_feature(propname)) {
	int err;
	char *fname = strchr(propname, '@') + 1;

	err = zfeature_lookup_name(fname, NULL);
	if (err != 0) {
	ASSERT3U(err, ==, ENOENT);
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"feature '%s' unsupported by kernel"),
	fname);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}

	if (nvpair_type(elem) != DATA_TYPE_STRING) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"'%s' must be a string"), propname);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}

	(void) nvpair_value_string(elem, &strval);
	if (strcmp(strval, ZFS_FEATURE_ENABLED) != 0 &&
	strcmp(strval, ZFS_FEATURE_DISABLED) != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property '%s' can only be set to "
	"'enabled' or 'disabled'"), propname);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}

	if (!flags.create &&
	strcmp(strval, ZFS_FEATURE_DISABLED) == 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property '%s' can only be set to "
	"'disabled' at creation time"), propname);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}

	if (nvlist_add_uint64(retprops, propname, 0) != 0) {
	(void) no_memory(hdl);
	goto error;
	}
	continue;
	}

	/*
	* Make sure this property is valid and applies to this type.
	*/
	if (prop == ZPOOL_PROP_INVAL) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"invalid property '%s'"), propname);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}

	if (zpool_prop_readonly(prop)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' "
	"is readonly"), propname);
	(void) zfs_error(hdl, EZFS_PROPREADONLY, errbuf);
	goto error;
	}

	if (!flags.create && zpool_prop_setonce(prop)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property '%s' can only be set at "
	"creation time"), propname);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}

	if (zprop_parse_value(hdl, elem, prop, ZFS_TYPE_POOL, retprops,
	&strval, &intval, errbuf) != 0)
	goto error;

	/*
	* Perform additional checking for specific properties.
	*/
	switch (prop) {
	case ZPOOL_PROP_VERSION:
	if (intval < version \|\|
	!SPA_VERSION_IS_SUPPORTED(intval)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property '%s' number %llu is invalid."),
	propname, (unsigned long long)intval);
	(void) zfs_error(hdl, EZFS_BADVERSION, errbuf);
	goto error;
	}
	break;

	case ZPOOL_PROP_ASHIFT:
	if (intval != 0 &&
	(intval < ASHIFT_MIN \|\| intval > ASHIFT_MAX)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property '%s' number %llu is invalid, "
	"only values between %" PRId32 " and %"
	PRId32 " are allowed."),
	propname, (unsigned long long)intval,
	ASHIFT_MIN, ASHIFT_MAX);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}
	break;

	case ZPOOL_PROP_BOOTFS:
	if (flags.create \|\| flags.import) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property '%s' cannot be set at creation "
	"or import time"), propname);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}

	if (version < SPA_VERSION_BOOTFS) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"pool must be upgraded to support "
	"'%s' property"), propname);
	(void) zfs_error(hdl, EZFS_BADVERSION, errbuf);
	goto error;
	}

	/*
	* bootfs property value has to be a dataset name and
	* the dataset has to be in the same pool as it sets to.
	*/
	if (!bootfs_name_valid(poolname, strval)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "'%s' "
	"is an invalid name"), strval);
	(void) zfs_error(hdl, EZFS_INVALIDNAME, errbuf);
	goto error;
	}

	if ((zhp = zpool_open_canfail(hdl, poolname)) == NULL) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"could not open pool '%s'"), poolname);
	(void) zfs_error(hdl, EZFS_OPENFAILED, errbuf);
	goto error;
	}
	zpool_close(zhp);
	break;

	case ZPOOL_PROP_ALTROOT:
	if (!flags.create && !flags.import) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property '%s' can only be set during pool "
	"creation or import"), propname);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}

	if (strval[0] != '/') {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"bad alternate root '%s'"), strval);
	(void) zfs_error(hdl, EZFS_BADPATH, errbuf);
	goto error;
	}
	break;

	case ZPOOL_PROP_CACHEFILE:
	if (strval[0] == '\0')
	break;

	if (strcmp(strval, "none") == 0)
	break;

	if (strval[0] != '/') {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property '%s' must be empty, an "
	"absolute path, or 'none'"), propname);
	(void) zfs_error(hdl, EZFS_BADPATH, errbuf);
	goto error;
	}

	slash = strrchr(strval, '/');

	if (slash[1] == '\0' \|\| strcmp(slash, "/.") == 0 \|\|
	strcmp(slash, "/..") == 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"'%s' is not a valid file"), strval);
	(void) zfs_error(hdl, EZFS_BADPATH, errbuf);
	goto error;
	}

	*slash = '\0';

	if (strval[0] != '\0' &&
	(stat64(strval, &statbuf) != 0 \|\|
	!S_ISDIR(statbuf.st_mode))) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"'%s' is not a valid directory"),
	strval);
	(void) zfs_error(hdl, EZFS_BADPATH, errbuf);
	goto error;
	}

	*slash = '/';
	break;

	case ZPOOL_PROP_COMPATIBILITY:
	switch (zpool_load_compat(strval, NULL, report, 1024)) {
	case ZPOOL_COMPATIBILITY_OK:
	case ZPOOL_COMPATIBILITY_WARNTOKEN:
	break;
	case ZPOOL_COMPATIBILITY_BADFILE:
	case ZPOOL_COMPATIBILITY_BADTOKEN:
	case ZPOOL_COMPATIBILITY_NOFILES:
	zfs_error_aux(hdl, "%s", report);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}
	break;

	case ZPOOL_PROP_COMMENT:
	for (check = strval; *check != '\0'; check++) {
	if (!isprint(*check)) {
	zfs_error_aux(hdl,
	dgettext(TEXT_DOMAIN,
	"comment may only have printable "
	"characters"));
	(void) zfs_error(hdl, EZFS_BADPROP,
	errbuf);
	goto error;
	}
	}
	if (strlen(strval) > ZPROP_MAX_COMMENT) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"comment must not exceed %d characters"),
	ZPROP_MAX_COMMENT);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}
	break;
	case ZPOOL_PROP_READONLY:
	if (!flags.import) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property '%s' can only be set at "
	"import time"), propname);
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}
	break;
	case ZPOOL_PROP_MULTIHOST:
	if (get_system_hostid() == 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"requires a non-zero system hostid"));
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	goto error;
	}
	break;
	case ZPOOL_PROP_DEDUPDITTO:
	printf("Note: property '%s' no longer has "
	"any effect\n", propname);
	break;

	default:
	break;
	}
	}

	return (retprops);
	error:
	nvlist_free(retprops);
	return (NULL);
	}

	/*
	* Set zpool property : propname=propval.
	*/
	int
	zpool_set_prop(zpool_handle_t zhp, const char propname, const char *propval)
	{
	zfs_cmd_t zc = {"\0"};
	int ret = -1;
	char errbuf[1024];
	nvlist_t *nvl = NULL;
	nvlist_t *realprops;
	uint64_t version;
	prop_flags_t flags = { 0 };

	(void) snprintf(errbuf, sizeof (errbuf),
	dgettext(TEXT_DOMAIN, "cannot set property for '%s'"),
	zhp->zpool_name);

	if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0)
	return (no_memory(zhp->zpool_hdl));

	if (nvlist_add_string(nvl, propname, propval) != 0) {
	nvlist_free(nvl);
	return (no_memory(zhp->zpool_hdl));
	}

	version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL);
	if ((realprops = zpool_valid_proplist(zhp->zpool_hdl,
	zhp->zpool_name, nvl, version, flags, errbuf)) == NULL) {
	nvlist_free(nvl);
	return (-1);
	}

	nvlist_free(nvl);
	nvl = realprops;

	/*
	* Execute the corresponding ioctl() to set this property.
	*/
	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));

	if (zcmd_write_src_nvlist(zhp->zpool_hdl, &zc, nvl) != 0) {
	nvlist_free(nvl);
	return (-1);
	}

	ret = zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_POOL_SET_PROPS, &zc);

	zcmd_free_nvlists(&zc);
	nvlist_free(nvl);

	if (ret)
	(void) zpool_standard_error(zhp->zpool_hdl, errno, errbuf);
	else
	(void) zpool_props_refresh(zhp);

	return (ret);
	}

	int
	zpool_expand_proplist(zpool_handle_t zhp, zprop_list_t *plp,
	- boolean_t literal)
	+ zfs_type_t type, boolean_t literal)
	{
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	zprop_list_t *entry;
	char buf[ZFS_MAXPROPLEN];
	nvlist_t *features = NULL;
	nvpair_t *nvp;
	zprop_list_t **last;
	boolean_t firstexpand = (NULL == *plp);
	int i;

	- if (zprop_expand_list(hdl, plp, ZFS_TYPE_POOL) != 0)
	+ if (zprop_expand_list(hdl, plp, type) != 0)
	return (-1);

	+ if (type == ZFS_TYPE_VDEV)
	+ return (0);
	+
	last = plp;
	while (*last != NULL)
	last = &(*last)->pl_next;

	if ((*plp)->pl_all)
	features = zpool_get_features(zhp);

	if ((*plp)->pl_all && firstexpand) {
	for (i = 0; i < SPA_FEATURES; i++) {
	zprop_list_t *entry = zfs_alloc(hdl,
	sizeof (zprop_list_t));
	entry->pl_prop = ZPROP_INVAL;
	entry->pl_user_prop = zfs_asprintf(hdl, "feature@%s",
	spa_feature_table[i].fi_uname);
	entry->pl_width = strlen(entry->pl_user_prop);
	entry->pl_all = B_TRUE;

	*last = entry;
	last = &entry->pl_next;
	}
	}

	/* add any unsupported features */
	for (nvp = nvlist_next_nvpair(features, NULL);
	nvp != NULL; nvp = nvlist_next_nvpair(features, nvp)) {
	char *propname;
	boolean_t found;
	zprop_list_t *entry;

	if (zfeature_is_supported(nvpair_name(nvp)))
	continue;

	propname = zfs_asprintf(hdl, "unsupported@%s",
	nvpair_name(nvp));

	/*
	* Before adding the property to the list make sure that no
	* other pool already added the same property.
	*/
	found = B_FALSE;
	entry = *plp;
	while (entry != NULL) {
	if (entry->pl_user_prop != NULL &&
	strcmp(propname, entry->pl_user_prop) == 0) {
	found = B_TRUE;
	break;
	}
	entry = entry->pl_next;
	}
	if (found) {
	free(propname);
	continue;
	}

	entry = zfs_alloc(hdl, sizeof (zprop_list_t));
	entry->pl_prop = ZPROP_INVAL;
	entry->pl_user_prop = propname;
	entry->pl_width = strlen(entry->pl_user_prop);
	entry->pl_all = B_TRUE;

	*last = entry;
	last = &entry->pl_next;
	}

	for (entry = *plp; entry != NULL; entry = entry->pl_next) {
	if (entry->pl_fixed && !literal)
	continue;

	if (entry->pl_prop != ZPROP_INVAL &&
	zpool_get_prop(zhp, entry->pl_prop, buf, sizeof (buf),
	NULL, literal) == 0) {
	if (strlen(buf) > entry->pl_width)
	entry->pl_width = strlen(buf);
	}
	}

	return (0);
	}

	+int
	+vdev_expand_proplist(zpool_handle_t zhp, const char vdevname,
	+ zprop_list_t **plp)
	+{
	+ zprop_list_t *entry;
	+ char buf[ZFS_MAXPROPLEN];
	+ char *strval = NULL;
	+ int err = 0;
	+ nvpair_t *elem = NULL;
	+ nvlist_t *vprops = NULL;
	+ nvlist_t *propval = NULL;
	+ const char *propname;
	+ vdev_prop_t prop;
	+ zprop_list_t **last;
	+
	+ for (entry = *plp; entry != NULL; entry = entry->pl_next) {
	+ if (entry->pl_fixed)
	+ continue;
	+
	+ if (zpool_get_vdev_prop(zhp, vdevname, entry->pl_prop,
	+ entry->pl_user_prop, buf, sizeof (buf), NULL,
	+ B_FALSE) == 0) {
	+ if (strlen(buf) > entry->pl_width)
	+ entry->pl_width = strlen(buf);
	+ }
	+ if (entry->pl_prop == VDEV_PROP_NAME &&
	+ strlen(vdevname) > entry->pl_width)
	+ entry->pl_width = strlen(vdevname);
	+ }
	+
	+ /* Handle the all properties case */
	+ last = plp;
	+ if (last != NULL && (last)->pl_all == B_TRUE) {
	+ while (*last != NULL)
	+ last = &(*last)->pl_next;
	+
	+ err = zpool_get_all_vdev_props(zhp, vdevname, &vprops);
	+ if (err != 0)
	+ return (err);
	+
	+ while ((elem = nvlist_next_nvpair(vprops, elem)) != NULL) {
	+ propname = nvpair_name(elem);
	+
	+ /* Skip properties that are not user defined */
	+ if ((prop = vdev_name_to_prop(propname)) !=
	+ VDEV_PROP_USER)
	+ continue;
	+
	+ if (nvpair_value_nvlist(elem, &propval) != 0)
	+ continue;
	+
	+ verify(nvlist_lookup_string(propval, ZPROP_VALUE,
	+ &strval) == 0);
	+
	+ if ((entry = zfs_alloc(zhp->zpool_hdl,
	+ sizeof (zprop_list_t))) == NULL)
	+ return (ENOMEM);
	+
	+ entry->pl_prop = prop;
	+ entry->pl_user_prop = zfs_strdup(zhp->zpool_hdl,
	+ propname);
	+ entry->pl_width = strlen(strval);
	+ entry->pl_all = B_TRUE;
	+ *last = entry;
	+ last = &entry->pl_next;
	+ }
	+ }
	+
	+ return (0);
	+}
	+
	/*
	* Get the state for the given feature on the given ZFS pool.
	*/
	int
	zpool_prop_get_feature(zpool_handle_t zhp, const char propname, char *buf,
	size_t len)
	{
	uint64_t refcount;
	boolean_t found = B_FALSE;
	nvlist_t *features = zpool_get_features(zhp);
	boolean_t supported;
	const char *feature = strchr(propname, '@') + 1;

	supported = zpool_prop_feature(propname);
	ASSERT(supported \|\| zpool_prop_unsupported(propname));

	/*
	* Convert from feature name to feature guid. This conversion is
	* unnecessary for unsupported@... properties because they already
	* use guids.
	*/
	if (supported) {
	int ret;
	spa_feature_t fid;

	ret = zfeature_lookup_name(feature, &fid);
	if (ret != 0) {
	(void) strlcpy(buf, "-", len);
	return (ENOTSUP);
	}
	feature = spa_feature_table[fid].fi_guid;
	}

	if (nvlist_lookup_uint64(features, feature, &refcount) == 0)
	found = B_TRUE;

	if (supported) {
	if (!found) {
	(void) strlcpy(buf, ZFS_FEATURE_DISABLED, len);
	} else {
	if (refcount == 0)
	(void) strlcpy(buf, ZFS_FEATURE_ENABLED, len);
	else
	(void) strlcpy(buf, ZFS_FEATURE_ACTIVE, len);
	}
	} else {
	if (found) {
	if (refcount == 0) {
	(void) strcpy(buf, ZFS_UNSUPPORTED_INACTIVE);
	} else {
	(void) strcpy(buf, ZFS_UNSUPPORTED_READONLY);
	}
	} else {
	(void) strlcpy(buf, "-", len);
	return (ENOTSUP);
	}
	}

	return (0);
	}

	/*
	* Validate the given pool name, optionally putting an extended error message in
	* 'buf'.
	*/
	boolean_t
	zpool_name_valid(libzfs_handle_t hdl, boolean_t isopen, const char pool)
	{
	namecheck_err_t why;
	char what;
	int ret;

	ret = pool_namecheck(pool, &why, &what);

	/*
	* The rules for reserved pool names were extended at a later point.
	* But we need to support users with existing pools that may now be
	* invalid. So we only check for this expanded set of names during a
	* create (or import), and only in userland.
	*/
	if (ret == 0 && !isopen &&
	(strncmp(pool, "mirror", 6) == 0 \|\|
	strncmp(pool, "raidz", 5) == 0 \|\|
	strncmp(pool, "draid", 5) == 0 \|\|
	strncmp(pool, "spare", 5) == 0 \|\|
	strcmp(pool, "log") == 0)) {
	if (hdl != NULL)
	zfs_error_aux(hdl,
	dgettext(TEXT_DOMAIN, "name is reserved"));
	return (B_FALSE);
	}


	if (ret != 0) {
	if (hdl != NULL) {
	switch (why) {
	case NAME_ERR_TOOLONG:
	zfs_error_aux(hdl,
	dgettext(TEXT_DOMAIN, "name is too long"));
	break;

	case NAME_ERR_INVALCHAR:
	zfs_error_aux(hdl,
	dgettext(TEXT_DOMAIN, "invalid character "
	"'%c' in pool name"), what);
	break;

	case NAME_ERR_NOLETTER:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"name must begin with a letter"));
	break;

	case NAME_ERR_RESERVED:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"name is reserved"));
	break;

	case NAME_ERR_DISKLIKE:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"pool name is reserved"));
	break;

	case NAME_ERR_LEADING_SLASH:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"leading slash in name"));
	break;

	case NAME_ERR_EMPTY_COMPONENT:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"empty component in name"));
	break;

	case NAME_ERR_TRAILING_SLASH:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"trailing slash in name"));
	break;

	case NAME_ERR_MULTIPLE_DELIMITERS:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"multiple '@' and/or '#' delimiters in "
	"name"));
	break;

	case NAME_ERR_NO_AT:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"permission set is missing '@'"));
	break;

	default:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"(%d) not defined"), why);
	break;
	}
	}
	return (B_FALSE);
	}

	return (B_TRUE);
	}

	/*
	* Open a handle to the given pool, even if the pool is currently in the FAULTED
	* state.
	*/
	zpool_handle_t *
	zpool_open_canfail(libzfs_handle_t hdl, const char pool)
	{
	zpool_handle_t *zhp;
	boolean_t missing;

	/*
	* Make sure the pool name is valid.
	*/
	if (!zpool_name_valid(hdl, B_TRUE, pool)) {
	(void) zfs_error_fmt(hdl, EZFS_INVALIDNAME,
	dgettext(TEXT_DOMAIN, "cannot open '%s'"),
	pool);
	return (NULL);
	}

	if ((zhp = zfs_alloc(hdl, sizeof (zpool_handle_t))) == NULL)
	return (NULL);

	zhp->zpool_hdl = hdl;
	(void) strlcpy(zhp->zpool_name, pool, sizeof (zhp->zpool_name));

	if (zpool_refresh_stats(zhp, &missing) != 0) {
	zpool_close(zhp);
	return (NULL);
	}

	if (missing) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "no such pool"));
	(void) zfs_error_fmt(hdl, EZFS_NOENT,
	dgettext(TEXT_DOMAIN, "cannot open '%s'"), pool);
	zpool_close(zhp);
	return (NULL);
	}

	return (zhp);
	}

	/*
	* Like the above, but silent on error. Used when iterating over pools (because
	* the configuration cache may be out of date).
	*/
	int
	zpool_open_silent(libzfs_handle_t hdl, const char pool, zpool_handle_t **ret)
	{
	zpool_handle_t *zhp;
	boolean_t missing;

	if ((zhp = zfs_alloc(hdl, sizeof (zpool_handle_t))) == NULL)
	return (-1);

	zhp->zpool_hdl = hdl;
	(void) strlcpy(zhp->zpool_name, pool, sizeof (zhp->zpool_name));

	if (zpool_refresh_stats(zhp, &missing) != 0) {
	zpool_close(zhp);
	return (-1);
	}

	if (missing) {
	zpool_close(zhp);
	*ret = NULL;
	return (0);
	}

	*ret = zhp;
	return (0);
	}

	/*
	* Similar to zpool_open_canfail(), but refuses to open pools in the faulted
	* state.
	*/
	zpool_handle_t *
	zpool_open(libzfs_handle_t hdl, const char pool)
	{
	zpool_handle_t *zhp;

	if ((zhp = zpool_open_canfail(hdl, pool)) == NULL)
	return (NULL);

	if (zhp->zpool_state == POOL_STATE_UNAVAIL) {
	(void) zfs_error_fmt(hdl, EZFS_POOLUNAVAIL,
	dgettext(TEXT_DOMAIN, "cannot open '%s'"), zhp->zpool_name);
	zpool_close(zhp);
	return (NULL);
	}

	return (zhp);
	}

	/*
	* Close the handle. Simply frees the memory associated with the handle.
	*/
	void
	zpool_close(zpool_handle_t *zhp)
	{
	nvlist_free(zhp->zpool_config);
	nvlist_free(zhp->zpool_old_config);
	nvlist_free(zhp->zpool_props);
	free(zhp);
	}

	/*
	* Return the name of the pool.
	*/
	const char *
	zpool_get_name(zpool_handle_t *zhp)
	{
	return (zhp->zpool_name);
	}


	/*
	* Return the state of the pool (ACTIVE or UNAVAILABLE)
	*/
	int
	zpool_get_state(zpool_handle_t *zhp)
	{
	return (zhp->zpool_state);
	}

	/*
	* Check if vdev list contains a special vdev
	*/
	static boolean_t
	zpool_has_special_vdev(nvlist_t *nvroot)
	{
	nvlist_t **child;
	uint_t children;

	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, &child,
	&children) == 0) {
	for (uint_t c = 0; c < children; c++) {
	char *bias;

	if (nvlist_lookup_string(child[c],
	ZPOOL_CONFIG_ALLOCATION_BIAS, &bias) == 0 &&
	strcmp(bias, VDEV_ALLOC_BIAS_SPECIAL) == 0) {
	return (B_TRUE);
	}
	}
	}
	return (B_FALSE);
	}

	/*
	* Check if vdev list contains a dRAID vdev
	*/
	static boolean_t
	zpool_has_draid_vdev(nvlist_t *nvroot)
	{
	nvlist_t **child;
	uint_t children;

	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	&child, &children) == 0) {
	for (uint_t c = 0; c < children; c++) {
	char *type;

	if (nvlist_lookup_string(child[c],
	ZPOOL_CONFIG_TYPE, &type) == 0 &&
	strcmp(type, VDEV_TYPE_DRAID) == 0) {
	return (B_TRUE);
	}
	}
	}
	return (B_FALSE);
	}

	/*
	* Output a dRAID top-level vdev name in to the provided buffer.
	*/
	static char *
	zpool_draid_name(char *name, int len, uint64_t data, uint64_t parity,
	uint64_t spares, uint64_t children)
	{
	snprintf(name, len, "%s%llu:%llud:%lluc:%llus",
	VDEV_TYPE_DRAID, (u_longlong_t)parity, (u_longlong_t)data,
	(u_longlong_t)children, (u_longlong_t)spares);

	return (name);
	}

	/*
	* Return B_TRUE if the provided name is a dRAID spare name.
	*/
	boolean_t
	zpool_is_draid_spare(const char *name)
	{
	uint64_t spare_id, parity, vdev_id;

	if (sscanf(name, VDEV_TYPE_DRAID "%llu-%llu-%llu",
	(u_longlong_t )&parity, (u_longlong_t )&vdev_id,
	(u_longlong_t *)&spare_id) == 3) {
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	/*
	* Create the named pool, using the provided vdev list. It is assumed
	* that the consumer has already validated the contents of the nvlist, so we
	* don't have to worry about error semantics.
	*/
	int
	zpool_create(libzfs_handle_t hdl, const char pool, nvlist_t *nvroot,
	nvlist_t props, nvlist_t fsprops)
	{
	zfs_cmd_t zc = {"\0"};
	nvlist_t *zc_fsprops = NULL;
	nvlist_t *zc_props = NULL;
	nvlist_t *hidden_args = NULL;
	uint8_t *wkeydata = NULL;
	uint_t wkeylen = 0;
	char msg[1024];
	int ret = -1;

	(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
	"cannot create '%s'"), pool);

	if (!zpool_name_valid(hdl, B_FALSE, pool))
	return (zfs_error(hdl, EZFS_INVALIDNAME, msg));

	if (zcmd_write_conf_nvlist(hdl, &zc, nvroot) != 0)
	return (-1);

	if (props) {
	prop_flags_t flags = { .create = B_TRUE, .import = B_FALSE };

	if ((zc_props = zpool_valid_proplist(hdl, pool, props,
	SPA_VERSION_1, flags, msg)) == NULL) {
	goto create_failed;
	}
	}

	if (fsprops) {
	uint64_t zoned;
	char *zonestr;

	zoned = ((nvlist_lookup_string(fsprops,
	zfs_prop_to_name(ZFS_PROP_ZONED), &zonestr) == 0) &&
	strcmp(zonestr, "on") == 0);

	if ((zc_fsprops = zfs_valid_proplist(hdl, ZFS_TYPE_FILESYSTEM,
	fsprops, zoned, NULL, NULL, B_TRUE, msg)) == NULL) {
	goto create_failed;
	}

	if (nvlist_exists(zc_fsprops,
	zfs_prop_to_name(ZFS_PROP_SPECIAL_SMALL_BLOCKS)) &&
	!zpool_has_special_vdev(nvroot)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"%s property requires a special vdev"),
	zfs_prop_to_name(ZFS_PROP_SPECIAL_SMALL_BLOCKS));
	(void) zfs_error(hdl, EZFS_BADPROP, msg);
	goto create_failed;
	}

	if (!zc_props &&
	(nvlist_alloc(&zc_props, NV_UNIQUE_NAME, 0) != 0)) {
	goto create_failed;
	}
	if (zfs_crypto_create(hdl, NULL, zc_fsprops, props, B_TRUE,
	&wkeydata, &wkeylen) != 0) {
	zfs_error(hdl, EZFS_CRYPTOFAILED, msg);
	goto create_failed;
	}
	if (nvlist_add_nvlist(zc_props,
	ZPOOL_ROOTFS_PROPS, zc_fsprops) != 0) {
	goto create_failed;
	}
	if (wkeydata != NULL) {
	if (nvlist_alloc(&hidden_args, NV_UNIQUE_NAME, 0) != 0)
	goto create_failed;

	if (nvlist_add_uint8_array(hidden_args, "wkeydata",
	wkeydata, wkeylen) != 0)
	goto create_failed;

	if (nvlist_add_nvlist(zc_props, ZPOOL_HIDDEN_ARGS,
	hidden_args) != 0)
	goto create_failed;
	}
	}

	if (zc_props && zcmd_write_src_nvlist(hdl, &zc, zc_props) != 0)
	goto create_failed;

	(void) strlcpy(zc.zc_name, pool, sizeof (zc.zc_name));

	if ((ret = zfs_ioctl(hdl, ZFS_IOC_POOL_CREATE, &zc)) != 0) {

	zcmd_free_nvlists(&zc);
	nvlist_free(zc_props);
	nvlist_free(zc_fsprops);
	nvlist_free(hidden_args);
	if (wkeydata != NULL)
	free(wkeydata);

	switch (errno) {
	case EBUSY:
	/*
	* This can happen if the user has specified the same
	* device multiple times. We can't reliably detect this
	* until we try to add it and see we already have a
	* label. This can also happen under if the device is
	* part of an active md or lvm device.
	*/
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"one or more vdevs refer to the same device, or "
	"one of\nthe devices is part of an active md or "
	"lvm device"));
	return (zfs_error(hdl, EZFS_BADDEV, msg));

	case ERANGE:
	/*
	* This happens if the record size is smaller or larger
	* than the allowed size range, or not a power of 2.
	*
	* NOTE: although zfs_valid_proplist is called earlier,
	* this case may have slipped through since the
	* pool does not exist yet and it is therefore
	* impossible to read properties e.g. max blocksize
	* from the pool.
	*/
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"record size invalid"));
	return (zfs_error(hdl, EZFS_BADPROP, msg));

	case EOVERFLOW:
	/*
	* This occurs when one of the devices is below
	* SPA_MINDEVSIZE. Unfortunately, we can't detect which
	* device was the problem device since there's no
	* reliable way to determine device size from userland.
	*/
	{
	char buf[64];

	zfs_nicebytes(SPA_MINDEVSIZE, buf,
	sizeof (buf));

	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"one or more devices is less than the "
	"minimum size (%s)"), buf);
	}
	return (zfs_error(hdl, EZFS_BADDEV, msg));

	case ENOSPC:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"one or more devices is out of space"));
	return (zfs_error(hdl, EZFS_BADDEV, msg));

	case EINVAL:
	if (zpool_has_draid_vdev(nvroot) &&
	zfeature_lookup_name("draid", NULL) != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"dRAID vdevs are unsupported by the "
	"kernel"));
	return (zfs_error(hdl, EZFS_BADDEV, msg));
	} else {
	return (zpool_standard_error(hdl, errno, msg));
	}

	default:
	return (zpool_standard_error(hdl, errno, msg));
	}
	}

	create_failed:
	zcmd_free_nvlists(&zc);
	nvlist_free(zc_props);
	nvlist_free(zc_fsprops);
	nvlist_free(hidden_args);
	if (wkeydata != NULL)
	free(wkeydata);
	return (ret);
	}

	/*
	* Destroy the given pool. It is up to the caller to ensure that there are no
	* datasets left in the pool.
	*/
	int
	zpool_destroy(zpool_handle_t zhp, const char log_str)
	{
	zfs_cmd_t zc = {"\0"};
	zfs_handle_t *zfp = NULL;
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	char msg[1024];

	if (zhp->zpool_state == POOL_STATE_ACTIVE &&
	(zfp = zfs_open(hdl, zhp->zpool_name, ZFS_TYPE_FILESYSTEM)) == NULL)
	return (-1);

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	zc.zc_history = (uint64_t)(uintptr_t)log_str;

	if (zfs_ioctl(hdl, ZFS_IOC_POOL_DESTROY, &zc) != 0) {
	(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
	"cannot destroy '%s'"), zhp->zpool_name);

	if (errno == EROFS) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"one or more devices is read only"));
	(void) zfs_error(hdl, EZFS_BADDEV, msg);
	} else {
	(void) zpool_standard_error(hdl, errno, msg);
	}

	if (zfp)
	zfs_close(zfp);
	return (-1);
	}

	if (zfp) {
	remove_mountpoint(zfp);
	zfs_close(zfp);
	}

	return (0);
	}

	/*
	* Create a checkpoint in the given pool.
	*/
	int
	zpool_checkpoint(zpool_handle_t *zhp)
	{
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	char msg[1024];
	int error;

	error = lzc_pool_checkpoint(zhp->zpool_name);
	if (error != 0) {
	(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
	"cannot checkpoint '%s'"), zhp->zpool_name);
	(void) zpool_standard_error(hdl, error, msg);
	return (-1);
	}

	return (0);
	}

	/*
	* Discard the checkpoint from the given pool.
	*/
	int
	zpool_discard_checkpoint(zpool_handle_t *zhp)
	{
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	char msg[1024];
	int error;

	error = lzc_pool_checkpoint_discard(zhp->zpool_name);
	if (error != 0) {
	(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
	"cannot discard checkpoint in '%s'"), zhp->zpool_name);
	(void) zpool_standard_error(hdl, error, msg);
	return (-1);
	}

	return (0);
	}

	/*
	* Add the given vdevs to the pool. The caller must have already performed the
	* necessary verification to ensure that the vdev specification is well-formed.
	*/
	int
	zpool_add(zpool_handle_t zhp, nvlist_t nvroot)
	{
	zfs_cmd_t zc = {"\0"};
	int ret;
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	char msg[1024];
	nvlist_t spares, l2cache;
	uint_t nspares, nl2cache;

	(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
	"cannot add to '%s'"), zhp->zpool_name);

	if (zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL) <
	SPA_VERSION_SPARES &&
	nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	&spares, &nspares) == 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be "
	"upgraded to add hot spares"));
	return (zfs_error(hdl, EZFS_BADVERSION, msg));
	}

	if (zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL) <
	SPA_VERSION_L2CACHE &&
	nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
	&l2cache, &nl2cache) == 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool must be "
	"upgraded to add cache devices"));
	return (zfs_error(hdl, EZFS_BADVERSION, msg));
	}

	if (zcmd_write_conf_nvlist(hdl, &zc, nvroot) != 0)
	return (-1);
	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));

	if (zfs_ioctl(hdl, ZFS_IOC_VDEV_ADD, &zc) != 0) {
	switch (errno) {
	case EBUSY:
	/*
	* This can happen if the user has specified the same
	* device multiple times. We can't reliably detect this
	* until we try to add it and see we already have a
	* label.
	*/
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"one or more vdevs refer to the same device"));
	(void) zfs_error(hdl, EZFS_BADDEV, msg);
	break;

	case EINVAL:

	if (zpool_has_draid_vdev(nvroot) &&
	zfeature_lookup_name("draid", NULL) != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"dRAID vdevs are unsupported by the "
	"kernel"));
	} else {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"invalid config; a pool with removing/"
	"removed vdevs does not support adding "
	"raidz or dRAID vdevs"));
	}

	(void) zfs_error(hdl, EZFS_BADDEV, msg);
	break;

	case EOVERFLOW:
	/*
	* This occurs when one of the devices is below
	* SPA_MINDEVSIZE. Unfortunately, we can't detect which
	* device was the problem device since there's no
	* reliable way to determine device size from userland.
	*/
	{
	char buf[64];

	zfs_nicebytes(SPA_MINDEVSIZE, buf,
	sizeof (buf));

	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"device is less than the minimum "
	"size (%s)"), buf);
	}
	(void) zfs_error(hdl, EZFS_BADDEV, msg);
	break;

	case ENOTSUP:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"pool must be upgraded to add these vdevs"));
	(void) zfs_error(hdl, EZFS_BADVERSION, msg);
	break;

	default:
	(void) zpool_standard_error(hdl, errno, msg);
	}

	ret = -1;
	} else {
	ret = 0;
	}

	zcmd_free_nvlists(&zc);

	return (ret);
	}

	/*
	* Exports the pool from the system. The caller must ensure that there are no
	* mounted datasets in the pool.
	*/
	static int
	zpool_export_common(zpool_handle_t *zhp, boolean_t force, boolean_t hardforce,
	const char *log_str)
	{
	zfs_cmd_t zc = {"\0"};

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	zc.zc_cookie = force;
	zc.zc_guid = hardforce;
	zc.zc_history = (uint64_t)(uintptr_t)log_str;

	if (zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_POOL_EXPORT, &zc) != 0) {
	switch (errno) {
	case EXDEV:
	zfs_error_aux(zhp->zpool_hdl, dgettext(TEXT_DOMAIN,
	"use '-f' to override the following errors:\n"
	"'%s' has an active shared spare which could be"
	" used by other pools once '%s' is exported."),
	zhp->zpool_name, zhp->zpool_name);
	return (zfs_error_fmt(zhp->zpool_hdl, EZFS_ACTIVE_SPARE,
	dgettext(TEXT_DOMAIN, "cannot export '%s'"),
	zhp->zpool_name));
	default:
	return (zpool_standard_error_fmt(zhp->zpool_hdl, errno,
	dgettext(TEXT_DOMAIN, "cannot export '%s'"),
	zhp->zpool_name));
	}
	}

	return (0);
	}

	int
	zpool_export(zpool_handle_t zhp, boolean_t force, const char log_str)
	{
	return (zpool_export_common(zhp, force, B_FALSE, log_str));
	}

	int
	zpool_export_force(zpool_handle_t zhp, const char log_str)
	{
	return (zpool_export_common(zhp, B_TRUE, B_TRUE, log_str));
	}

	static void
	zpool_rewind_exclaim(libzfs_handle_t hdl, const char name, boolean_t dryrun,
	nvlist_t *config)
	{
	nvlist_t *nv = NULL;
	uint64_t rewindto;
	int64_t loss = -1;
	struct tm t;
	char timestr[128];

	if (!hdl->libzfs_printerr \|\| config == NULL)
	return;

	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, &nv) != 0 \|\|
	nvlist_lookup_nvlist(nv, ZPOOL_CONFIG_REWIND_INFO, &nv) != 0) {
	return;
	}

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_LOAD_TIME, &rewindto) != 0)
	return;
	(void) nvlist_lookup_int64(nv, ZPOOL_CONFIG_REWIND_TIME, &loss);

	if (localtime_r((time_t *)&rewindto, &t) != NULL &&
	strftime(timestr, 128, "%c", &t) != 0) {
	if (dryrun) {
	(void) printf(dgettext(TEXT_DOMAIN,
	"Would be able to return %s "
	"to its state as of %s.\n"),
	name, timestr);
	} else {
	(void) printf(dgettext(TEXT_DOMAIN,
	"Pool %s returned to its state as of %s.\n"),
	name, timestr);
	}
	if (loss > 120) {
	(void) printf(dgettext(TEXT_DOMAIN,
	"%s approximately %lld "),
	dryrun ? "Would discard" : "Discarded",
	((longlong_t)loss + 30) / 60);
	(void) printf(dgettext(TEXT_DOMAIN,
	"minutes of transactions.\n"));
	} else if (loss > 0) {
	(void) printf(dgettext(TEXT_DOMAIN,
	"%s approximately %lld "),
	dryrun ? "Would discard" : "Discarded",
	(longlong_t)loss);
	(void) printf(dgettext(TEXT_DOMAIN,
	"seconds of transactions.\n"));
	}
	}
	}

	void
	zpool_explain_recover(libzfs_handle_t hdl, const char name, int reason,
	nvlist_t *config)
	{
	nvlist_t *nv = NULL;
	int64_t loss = -1;
	uint64_t edata = UINT64_MAX;
	uint64_t rewindto;
	struct tm t;
	char timestr[128];

	if (!hdl->libzfs_printerr)
	return;

	if (reason >= 0)
	(void) printf(dgettext(TEXT_DOMAIN, "action: "));
	else
	(void) printf(dgettext(TEXT_DOMAIN, "\t"));

	/* All attempted rewinds failed if ZPOOL_CONFIG_LOAD_TIME missing */
	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, &nv) != 0 \|\|
	nvlist_lookup_nvlist(nv, ZPOOL_CONFIG_REWIND_INFO, &nv) != 0 \|\|
	nvlist_lookup_uint64(nv, ZPOOL_CONFIG_LOAD_TIME, &rewindto) != 0)
	goto no_info;

	(void) nvlist_lookup_int64(nv, ZPOOL_CONFIG_REWIND_TIME, &loss);
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_LOAD_DATA_ERRORS,
	&edata);

	(void) printf(dgettext(TEXT_DOMAIN,
	"Recovery is possible, but will result in some data loss.\n"));

	if (localtime_r((time_t *)&rewindto, &t) != NULL &&
	strftime(timestr, 128, "%c", &t) != 0) {
	(void) printf(dgettext(TEXT_DOMAIN,
	"\tReturning the pool to its state as of %s\n"
	"\tshould correct the problem. "),
	timestr);
	} else {
	(void) printf(dgettext(TEXT_DOMAIN,
	"\tReverting the pool to an earlier state "
	"should correct the problem.\n\t"));
	}

	if (loss > 120) {
	(void) printf(dgettext(TEXT_DOMAIN,
	"Approximately %lld minutes of data\n"
	"\tmust be discarded, irreversibly. "),
	((longlong_t)loss + 30) / 60);
	} else if (loss > 0) {
	(void) printf(dgettext(TEXT_DOMAIN,
	"Approximately %lld seconds of data\n"
	"\tmust be discarded, irreversibly. "),
	(longlong_t)loss);
	}
	if (edata != 0 && edata != UINT64_MAX) {
	if (edata == 1) {
	(void) printf(dgettext(TEXT_DOMAIN,
	"After rewind, at least\n"
	"\tone persistent user-data error will remain. "));
	} else {
	(void) printf(dgettext(TEXT_DOMAIN,
	"After rewind, several\n"
	"\tpersistent user-data errors will remain. "));
	}
	}
	(void) printf(dgettext(TEXT_DOMAIN,
	"Recovery can be attempted\n\tby executing 'zpool %s -F %s'. "),
	reason >= 0 ? "clear" : "import", name);

	(void) printf(dgettext(TEXT_DOMAIN,
	"A scrub of the pool\n"
	"\tis strongly recommended after recovery.\n"));
	return;

	no_info:
	(void) printf(dgettext(TEXT_DOMAIN,
	"Destroy and re-create the pool from\n\ta backup source.\n"));
	}

	/*
	* zpool_import() is a contracted interface. Should be kept the same
	* if possible.
	*
	* Applications should use zpool_import_props() to import a pool with
	* new properties value to be set.
	*/
	int
	zpool_import(libzfs_handle_t hdl, nvlist_t config, const char *newname,
	char *altroot)
	{
	nvlist_t *props = NULL;
	int ret;

	if (altroot != NULL) {
	if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0) {
	return (zfs_error_fmt(hdl, EZFS_NOMEM,
	dgettext(TEXT_DOMAIN, "cannot import '%s'"),
	newname));
	}

	if (nvlist_add_string(props,
	zpool_prop_to_name(ZPOOL_PROP_ALTROOT), altroot) != 0 \|\|
	nvlist_add_string(props,
	zpool_prop_to_name(ZPOOL_PROP_CACHEFILE), "none") != 0) {
	nvlist_free(props);
	return (zfs_error_fmt(hdl, EZFS_NOMEM,
	dgettext(TEXT_DOMAIN, "cannot import '%s'"),
	newname));
	}
	}

	ret = zpool_import_props(hdl, config, newname, props,
	ZFS_IMPORT_NORMAL);
	nvlist_free(props);
	return (ret);
	}

	static void
	print_vdev_tree(libzfs_handle_t hdl, const char name, nvlist_t *nv,
	int indent)
	{
	nvlist_t **child;
	uint_t c, children;
	char *vname;
	uint64_t is_log = 0;

	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG,
	&is_log);

	if (name != NULL)
	(void) printf("\t%*s%s%s\n", indent, "", name,
	is_log ? " [log]" : "");

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0)
	return;

	for (c = 0; c < children; c++) {
	vname = zpool_vdev_name(hdl, NULL, child[c], VDEV_NAME_TYPE_ID);
	print_vdev_tree(hdl, vname, child[c], indent + 2);
	free(vname);
	}
	}

	void
	zpool_print_unsup_feat(nvlist_t *config)
	{
	nvlist_t nvinfo, unsup_feat;
	nvpair_t *nvp;

	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, &nvinfo) ==
	0);
	verify(nvlist_lookup_nvlist(nvinfo, ZPOOL_CONFIG_UNSUP_FEAT,
	&unsup_feat) == 0);

	for (nvp = nvlist_next_nvpair(unsup_feat, NULL); nvp != NULL;
	nvp = nvlist_next_nvpair(unsup_feat, nvp)) {
	char *desc;

	verify(nvpair_type(nvp) == DATA_TYPE_STRING);
	verify(nvpair_value_string(nvp, &desc) == 0);

	if (strlen(desc) > 0)
	(void) printf("\t%s (%s)\n", nvpair_name(nvp), desc);
	else
	(void) printf("\t%s\n", nvpair_name(nvp));
	}
	}

	/*
	* Import the given pool using the known configuration and a list of
	* properties to be set. The configuration should have come from
	* zpool_find_import(). The 'newname' parameters control whether the pool
	* is imported with a different name.
	*/
	int
	zpool_import_props(libzfs_handle_t hdl, nvlist_t config, const char *newname,
	nvlist_t *props, int flags)
	{
	zfs_cmd_t zc = {"\0"};
	zpool_load_policy_t policy;
	nvlist_t *nv = NULL;
	nvlist_t *nvinfo = NULL;
	nvlist_t *missing = NULL;
	char *thename;
	char *origname;
	int ret;
	int error = 0;
	char errbuf[1024];

	verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME,
	&origname) == 0);

	(void) snprintf(errbuf, sizeof (errbuf), dgettext(TEXT_DOMAIN,
	"cannot import pool '%s'"), origname);

	if (newname != NULL) {
	if (!zpool_name_valid(hdl, B_FALSE, newname))
	return (zfs_error_fmt(hdl, EZFS_INVALIDNAME,
	dgettext(TEXT_DOMAIN, "cannot import '%s'"),
	newname));
	thename = (char *)newname;
	} else {
	thename = origname;
	}

	if (props != NULL) {
	uint64_t version;
	prop_flags_t flags = { .create = B_FALSE, .import = B_TRUE };

	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
	&version) == 0);

	if ((props = zpool_valid_proplist(hdl, origname,
	props, version, flags, errbuf)) == NULL)
	return (-1);
	if (zcmd_write_src_nvlist(hdl, &zc, props) != 0) {
	nvlist_free(props);
	return (-1);
	}
	nvlist_free(props);
	}

	(void) strlcpy(zc.zc_name, thename, sizeof (zc.zc_name));

	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
	&zc.zc_guid) == 0);

	if (zcmd_write_conf_nvlist(hdl, &zc, config) != 0) {
	zcmd_free_nvlists(&zc);
	return (-1);
	}
	if (zcmd_alloc_dst_nvlist(hdl, &zc, zc.zc_nvlist_conf_size * 2) != 0) {
	zcmd_free_nvlists(&zc);
	return (-1);
	}

	zc.zc_cookie = flags;
	while ((ret = zfs_ioctl(hdl, ZFS_IOC_POOL_IMPORT, &zc)) != 0 &&
	errno == ENOMEM) {
	if (zcmd_expand_dst_nvlist(hdl, &zc) != 0) {
	zcmd_free_nvlists(&zc);
	return (-1);
	}
	}
	if (ret != 0)
	error = errno;

	(void) zcmd_read_dst_nvlist(hdl, &zc, &nv);

	zcmd_free_nvlists(&zc);

	zpool_get_load_policy(config, &policy);

	if (error) {
	char desc[1024];
	char aux[256];

	/*
	* Dry-run failed, but we print out what success
	* looks like if we found a best txg
	*/
	if (policy.zlp_rewind & ZPOOL_TRY_REWIND) {
	zpool_rewind_exclaim(hdl, newname ? origname : thename,
	B_TRUE, nv);
	nvlist_free(nv);
	return (-1);
	}

	if (newname == NULL)
	(void) snprintf(desc, sizeof (desc),
	dgettext(TEXT_DOMAIN, "cannot import '%s'"),
	thename);
	else
	(void) snprintf(desc, sizeof (desc),
	dgettext(TEXT_DOMAIN, "cannot import '%s' as '%s'"),
	origname, thename);

	switch (error) {
	case ENOTSUP:
	if (nv != NULL && nvlist_lookup_nvlist(nv,
	ZPOOL_CONFIG_LOAD_INFO, &nvinfo) == 0 &&
	nvlist_exists(nvinfo, ZPOOL_CONFIG_UNSUP_FEAT)) {
	(void) printf(dgettext(TEXT_DOMAIN, "This "
	"pool uses the following feature(s) not "
	"supported by this system:\n"));
	zpool_print_unsup_feat(nv);
	if (nvlist_exists(nvinfo,
	ZPOOL_CONFIG_CAN_RDONLY)) {
	(void) printf(dgettext(TEXT_DOMAIN,
	"All unsupported features are only "
	"required for writing to the pool."
	"\nThe pool can be imported using "
	"'-o readonly=on'.\n"));
	}
	}
	/*
	* Unsupported version.
	*/
	(void) zfs_error(hdl, EZFS_BADVERSION, desc);
	break;

	case EREMOTEIO:
	if (nv != NULL && nvlist_lookup_nvlist(nv,
	ZPOOL_CONFIG_LOAD_INFO, &nvinfo) == 0) {
	char *hostname = "<unknown>";
	uint64_t hostid = 0;
	mmp_state_t mmp_state;

	mmp_state = fnvlist_lookup_uint64(nvinfo,
	ZPOOL_CONFIG_MMP_STATE);

	if (nvlist_exists(nvinfo,
	ZPOOL_CONFIG_MMP_HOSTNAME))
	hostname = fnvlist_lookup_string(nvinfo,
	ZPOOL_CONFIG_MMP_HOSTNAME);

	if (nvlist_exists(nvinfo,
	ZPOOL_CONFIG_MMP_HOSTID))
	hostid = fnvlist_lookup_uint64(nvinfo,
	ZPOOL_CONFIG_MMP_HOSTID);

	if (mmp_state == MMP_STATE_ACTIVE) {
	(void) snprintf(aux, sizeof (aux),
	dgettext(TEXT_DOMAIN, "pool is imp"
	"orted on host '%s' (hostid=%lx).\n"
	"Export the pool on the other "
	"system, then run 'zpool import'."),
	hostname, (unsigned long) hostid);
	} else if (mmp_state == MMP_STATE_NO_HOSTID) {
	(void) snprintf(aux, sizeof (aux),
	dgettext(TEXT_DOMAIN, "pool has "
	"the multihost property on and "
	"the\nsystem's hostid is not set. "
	"Set a unique system hostid with "
	"the zgenhostid(8) command.\n"));
	}

	(void) zfs_error_aux(hdl, "%s", aux);
	}
	(void) zfs_error(hdl, EZFS_ACTIVE_POOL, desc);
	break;

	case EINVAL:
	(void) zfs_error(hdl, EZFS_INVALCONFIG, desc);
	break;

	case EROFS:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"one or more devices is read only"));
	(void) zfs_error(hdl, EZFS_BADDEV, desc);
	break;

	case ENXIO:
	if (nv && nvlist_lookup_nvlist(nv,
	ZPOOL_CONFIG_LOAD_INFO, &nvinfo) == 0 &&
	nvlist_lookup_nvlist(nvinfo,
	ZPOOL_CONFIG_MISSING_DEVICES, &missing) == 0) {
	(void) printf(dgettext(TEXT_DOMAIN,
	"The devices below are missing or "
	"corrupted, use '-m' to import the pool "
	"anyway:\n"));
	print_vdev_tree(hdl, NULL, missing, 2);
	(void) printf("\n");
	}
	(void) zpool_standard_error(hdl, error, desc);
	break;

	case EEXIST:
	(void) zpool_standard_error(hdl, error, desc);
	break;

	case EBUSY:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"one or more devices are already in use\n"));
	(void) zfs_error(hdl, EZFS_BADDEV, desc);
	break;
	case ENAMETOOLONG:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"new name of at least one dataset is longer than "
	"the maximum allowable length"));
	(void) zfs_error(hdl, EZFS_NAMETOOLONG, desc);
	break;
	default:
	(void) zpool_standard_error(hdl, error, desc);
	zpool_explain_recover(hdl,
	newname ? origname : thename, -error, nv);
	break;
	}

	nvlist_free(nv);
	ret = -1;
	} else {
	zpool_handle_t *zhp;

	/*
	* This should never fail, but play it safe anyway.
	*/
	if (zpool_open_silent(hdl, thename, &zhp) != 0)
	ret = -1;
	else if (zhp != NULL)
	zpool_close(zhp);
	if (policy.zlp_rewind &
	(ZPOOL_DO_REWIND \| ZPOOL_TRY_REWIND)) {
	zpool_rewind_exclaim(hdl, newname ? origname : thename,
	((policy.zlp_rewind & ZPOOL_TRY_REWIND) != 0), nv);
	}
	nvlist_free(nv);
	return (0);
	}

	return (ret);
	}

	/*
	* Translate vdev names to guids. If a vdev_path is determined to be
	* unsuitable then a vd_errlist is allocated and the vdev path and errno
	* are added to it.
	*/
	static int
	zpool_translate_vdev_guids(zpool_handle_t zhp, nvlist_t vds,
	nvlist_t vdev_guids, nvlist_t guids_to_paths, nvlist_t **vd_errlist)
	{
	nvlist_t *errlist = NULL;
	int error = 0;

	for (nvpair_t *elem = nvlist_next_nvpair(vds, NULL); elem != NULL;
	elem = nvlist_next_nvpair(vds, elem)) {
	boolean_t spare, cache;

	char *vd_path = nvpair_name(elem);
	nvlist_t *tgt = zpool_find_vdev(zhp, vd_path, &spare, &cache,
	NULL);

	if ((tgt == NULL) \|\| cache \|\| spare) {
	if (errlist == NULL) {
	errlist = fnvlist_alloc();
	error = EINVAL;
	}

	uint64_t err = (tgt == NULL) ? EZFS_NODEVICE :
	(spare ? EZFS_ISSPARE : EZFS_ISL2CACHE);
	fnvlist_add_int64(errlist, vd_path, err);
	continue;
	}

	uint64_t guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID);
	fnvlist_add_uint64(vdev_guids, vd_path, guid);

	char msg[MAXNAMELEN];
	(void) snprintf(msg, sizeof (msg), "%llu", (u_longlong_t)guid);
	fnvlist_add_string(guids_to_paths, msg, vd_path);
	}

	if (error != 0) {
	verify(errlist != NULL);
	if (vd_errlist != NULL)
	*vd_errlist = errlist;
	else
	fnvlist_free(errlist);
	}

	return (error);
	}

	static int
	xlate_init_err(int err)
	{
	switch (err) {
	case ENODEV:
	return (EZFS_NODEVICE);
	case EINVAL:
	case EROFS:
	return (EZFS_BADDEV);
	case EBUSY:
	return (EZFS_INITIALIZING);
	case ESRCH:
	return (EZFS_NO_INITIALIZE);
	}
	return (err);
	}

	/*
	* Begin, suspend, or cancel the initialization (initializing of all free
	* blocks) for the given vdevs in the given pool.
	*/
	static int
	zpool_initialize_impl(zpool_handle_t *zhp, pool_initialize_func_t cmd_type,
	nvlist_t *vds, boolean_t wait)
	{
	int err;

	nvlist_t *vdev_guids = fnvlist_alloc();
	nvlist_t *guids_to_paths = fnvlist_alloc();
	nvlist_t *vd_errlist = NULL;
	nvlist_t *errlist;
	nvpair_t *elem;

	err = zpool_translate_vdev_guids(zhp, vds, vdev_guids,
	guids_to_paths, &vd_errlist);

	if (err != 0) {
	verify(vd_errlist != NULL);
	goto list_errors;
	}

	err = lzc_initialize(zhp->zpool_name, cmd_type,
	vdev_guids, &errlist);

	if (err != 0) {
	if (errlist != NULL) {
	vd_errlist = fnvlist_lookup_nvlist(errlist,
	ZPOOL_INITIALIZE_VDEVS);
	goto list_errors;
	}
	(void) zpool_standard_error(zhp->zpool_hdl, err,
	dgettext(TEXT_DOMAIN, "operation failed"));
	goto out;
	}

	if (wait) {
	for (elem = nvlist_next_nvpair(vdev_guids, NULL); elem != NULL;
	elem = nvlist_next_nvpair(vdev_guids, elem)) {

	uint64_t guid = fnvpair_value_uint64(elem);

	err = lzc_wait_tag(zhp->zpool_name,
	ZPOOL_WAIT_INITIALIZE, guid, NULL);
	if (err != 0) {
	(void) zpool_standard_error_fmt(zhp->zpool_hdl,
	err, dgettext(TEXT_DOMAIN, "error "
	"waiting for '%s' to initialize"),
	nvpair_name(elem));

	goto out;
	}
	}
	}
	goto out;

	list_errors:
	for (elem = nvlist_next_nvpair(vd_errlist, NULL); elem != NULL;
	elem = nvlist_next_nvpair(vd_errlist, elem)) {
	int64_t vd_error = xlate_init_err(fnvpair_value_int64(elem));
	char *path;

	if (nvlist_lookup_string(guids_to_paths, nvpair_name(elem),
	&path) != 0)
	path = nvpair_name(elem);

	(void) zfs_error_fmt(zhp->zpool_hdl, vd_error,
	"cannot initialize '%s'", path);
	}

	out:
	fnvlist_free(vdev_guids);
	fnvlist_free(guids_to_paths);

	if (vd_errlist != NULL)
	fnvlist_free(vd_errlist);

	return (err == 0 ? 0 : -1);
	}

	int
	zpool_initialize(zpool_handle_t *zhp, pool_initialize_func_t cmd_type,
	nvlist_t *vds)
	{
	return (zpool_initialize_impl(zhp, cmd_type, vds, B_FALSE));
	}

	int
	zpool_initialize_wait(zpool_handle_t *zhp, pool_initialize_func_t cmd_type,
	nvlist_t *vds)
	{
	return (zpool_initialize_impl(zhp, cmd_type, vds, B_TRUE));
	}

	static int
	xlate_trim_err(int err)
	{
	switch (err) {
	case ENODEV:
	return (EZFS_NODEVICE);
	case EINVAL:
	case EROFS:
	return (EZFS_BADDEV);
	case EBUSY:
	return (EZFS_TRIMMING);
	case ESRCH:
	return (EZFS_NO_TRIM);
	case EOPNOTSUPP:
	return (EZFS_TRIM_NOTSUP);
	}
	return (err);
	}

	static int
	zpool_trim_wait(zpool_handle_t zhp, nvlist_t vdev_guids)
	{
	int err;
	nvpair_t *elem;

	for (elem = nvlist_next_nvpair(vdev_guids, NULL); elem != NULL;
	elem = nvlist_next_nvpair(vdev_guids, elem)) {

	uint64_t guid = fnvpair_value_uint64(elem);

	err = lzc_wait_tag(zhp->zpool_name,
	ZPOOL_WAIT_TRIM, guid, NULL);
	if (err != 0) {
	(void) zpool_standard_error_fmt(zhp->zpool_hdl,
	err, dgettext(TEXT_DOMAIN, "error "
	"waiting to trim '%s'"), nvpair_name(elem));

	return (err);
	}
	}
	return (0);
	}

	/*
	* Check errlist and report any errors, omitting ones which should be
	* suppressed. Returns B_TRUE if any errors were reported.
	*/
	static boolean_t
	check_trim_errs(zpool_handle_t zhp, trimflags_t trim_flags,
	nvlist_t guids_to_paths, nvlist_t vds, nvlist_t *errlist)
	{
	nvpair_t *elem;
	boolean_t reported_errs = B_FALSE;
	int num_vds = 0;
	int num_suppressed_errs = 0;

	for (elem = nvlist_next_nvpair(vds, NULL);
	elem != NULL; elem = nvlist_next_nvpair(vds, elem)) {
	num_vds++;
	}

	for (elem = nvlist_next_nvpair(errlist, NULL);
	elem != NULL; elem = nvlist_next_nvpair(errlist, elem)) {
	int64_t vd_error = xlate_trim_err(fnvpair_value_int64(elem));
	char *path;

	/*
	* If only the pool was specified, and it was not a secure
	* trim then suppress warnings for individual vdevs which
	* do not support trimming.
	*/
	if (vd_error == EZFS_TRIM_NOTSUP &&
	trim_flags->fullpool &&
	!trim_flags->secure) {
	num_suppressed_errs++;
	continue;
	}

	reported_errs = B_TRUE;
	if (nvlist_lookup_string(guids_to_paths, nvpair_name(elem),
	&path) != 0)
	path = nvpair_name(elem);

	(void) zfs_error_fmt(zhp->zpool_hdl, vd_error,
	"cannot trim '%s'", path);
	}

	if (num_suppressed_errs == num_vds) {
	(void) zfs_error_aux(zhp->zpool_hdl, dgettext(TEXT_DOMAIN,
	"no devices in pool support trim operations"));
	(void) (zfs_error(zhp->zpool_hdl, EZFS_TRIM_NOTSUP,
	dgettext(TEXT_DOMAIN, "cannot trim")));
	reported_errs = B_TRUE;
	}

	return (reported_errs);
	}

	/*
	* Begin, suspend, or cancel the TRIM (discarding of all free blocks) for
	* the given vdevs in the given pool.
	*/
	int
	zpool_trim(zpool_handle_t zhp, pool_trim_func_t cmd_type, nvlist_t vds,
	trimflags_t *trim_flags)
	{
	int err;
	int retval = 0;

	nvlist_t *vdev_guids = fnvlist_alloc();
	nvlist_t *guids_to_paths = fnvlist_alloc();
	nvlist_t *errlist = NULL;

	err = zpool_translate_vdev_guids(zhp, vds, vdev_guids,
	guids_to_paths, &errlist);
	if (err != 0) {
	check_trim_errs(zhp, trim_flags, guids_to_paths, vds, errlist);
	retval = -1;
	goto out;
	}

	err = lzc_trim(zhp->zpool_name, cmd_type, trim_flags->rate,
	trim_flags->secure, vdev_guids, &errlist);
	if (err != 0) {
	nvlist_t *vd_errlist;
	if (errlist != NULL && nvlist_lookup_nvlist(errlist,
	ZPOOL_TRIM_VDEVS, &vd_errlist) == 0) {
	if (check_trim_errs(zhp, trim_flags, guids_to_paths,
	vds, vd_errlist)) {
	retval = -1;
	goto out;
	}
	} else {
	char msg[1024];

	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "operation failed"));
	zpool_standard_error(zhp->zpool_hdl, err, msg);
	retval = -1;
	goto out;
	}
	}


	if (trim_flags->wait)
	retval = zpool_trim_wait(zhp, vdev_guids);

	out:
	if (errlist != NULL)
	fnvlist_free(errlist);
	fnvlist_free(vdev_guids);
	fnvlist_free(guids_to_paths);
	return (retval);
	}

	/*
	* Scan the pool.
	*/
	int
	zpool_scan(zpool_handle_t *zhp, pool_scan_func_t func, pool_scrub_cmd_t cmd)
	{
	zfs_cmd_t zc = {"\0"};
	char msg[1024];
	int err;
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	zc.zc_cookie = func;
	zc.zc_flags = cmd;

	if (zfs_ioctl(hdl, ZFS_IOC_POOL_SCAN, &zc) == 0)
	return (0);

	err = errno;

	/* ECANCELED on a scrub means we resumed a paused scrub */
	if (err == ECANCELED && func == POOL_SCAN_SCRUB &&
	cmd == POOL_SCRUB_NORMAL)
	return (0);

	if (err == ENOENT && func != POOL_SCAN_NONE && cmd == POOL_SCRUB_NORMAL)
	return (0);

	if (func == POOL_SCAN_SCRUB) {
	if (cmd == POOL_SCRUB_PAUSE) {
	(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
	"cannot pause scrubbing %s"), zc.zc_name);
	} else {
	assert(cmd == POOL_SCRUB_NORMAL);
	(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
	"cannot scrub %s"), zc.zc_name);
	}
	} else if (func == POOL_SCAN_RESILVER) {
	assert(cmd == POOL_SCRUB_NORMAL);
	(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
	"cannot restart resilver on %s"), zc.zc_name);
	} else if (func == POOL_SCAN_NONE) {
	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot cancel scrubbing %s"),
	zc.zc_name);
	} else {
	assert(!"unexpected result");
	}

	if (err == EBUSY) {
	nvlist_t *nvroot;
	pool_scan_stat_t *ps = NULL;
	uint_t psc;

	verify(nvlist_lookup_nvlist(zhp->zpool_config,
	ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0);
	(void) nvlist_lookup_uint64_array(nvroot,
	ZPOOL_CONFIG_SCAN_STATS, (uint64_t **)&ps, &psc);
	if (ps && ps->pss_func == POOL_SCAN_SCRUB &&
	ps->pss_state == DSS_SCANNING) {
	if (cmd == POOL_SCRUB_PAUSE)
	return (zfs_error(hdl, EZFS_SCRUB_PAUSED, msg));
	else
	return (zfs_error(hdl, EZFS_SCRUBBING, msg));
	} else {
	return (zfs_error(hdl, EZFS_RESILVERING, msg));
	}
	} else if (err == ENOENT) {
	return (zfs_error(hdl, EZFS_NO_SCRUB, msg));
	} else if (err == ENOTSUP && func == POOL_SCAN_RESILVER) {
	return (zfs_error(hdl, EZFS_NO_RESILVER_DEFER, msg));
	} else {
	return (zpool_standard_error(hdl, err, msg));
	}
	}

	/*
	* Find a vdev that matches the search criteria specified. We use the
	* the nvpair name to determine how we should look for the device.
	* 'avail_spare' is set to TRUE if the provided guid refers to an AVAIL
	* spare; but FALSE if its an INUSE spare.
	*/
	static nvlist_t *
	vdev_to_nvlist_iter(nvlist_t nv, nvlist_t search, boolean_t *avail_spare,
	boolean_t l2cache, boolean_t log)
	{
	uint_t c, children;
	nvlist_t **child;
	nvlist_t *ret;
	uint64_t is_log;
	char *srchkey;
	nvpair_t *pair = nvlist_next_nvpair(search, NULL);

	/* Nothing to look for */
	if (search == NULL \|\| pair == NULL)
	return (NULL);

	/* Obtain the key we will use to search */
	srchkey = nvpair_name(pair);

	switch (nvpair_type(pair)) {
	case DATA_TYPE_UINT64:
	if (strcmp(srchkey, ZPOOL_CONFIG_GUID) == 0) {
	uint64_t srchval, theguid;

	verify(nvpair_value_uint64(pair, &srchval) == 0);
	verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID,
	&theguid) == 0);
	if (theguid == srchval)
	return (nv);
	}
	break;

	case DATA_TYPE_STRING: {
	char srchval, val;

	verify(nvpair_value_string(pair, &srchval) == 0);
	if (nvlist_lookup_string(nv, srchkey, &val) != 0)
	break;

	/*
	* Search for the requested value. Special cases:
	*
	* - ZPOOL_CONFIG_PATH for whole disk entries. These end in
	* "-part1", or "p1". The suffix is hidden from the user,
	* but included in the string, so this matches around it.
	* - ZPOOL_CONFIG_PATH for short names zfs_strcmp_shortname()
	* is used to check all possible expanded paths.
	* - looking for a top-level vdev name (i.e. ZPOOL_CONFIG_TYPE).
	*
	* Otherwise, all other searches are simple string compares.
	*/
	if (strcmp(srchkey, ZPOOL_CONFIG_PATH) == 0) {
	uint64_t wholedisk = 0;

	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK,
	&wholedisk);
	if (zfs_strcmp_pathname(srchval, val, wholedisk) == 0)
	return (nv);

	} else if (strcmp(srchkey, ZPOOL_CONFIG_TYPE) == 0 && val) {
	char type, idx, end, p;
	uint64_t id, vdev_id;

	/*
	* Determine our vdev type, keeping in mind
	* that the srchval is composed of a type and
	* vdev id pair (i.e. mirror-4).
	*/
	if ((type = strdup(srchval)) == NULL)
	return (NULL);

	if ((p = strrchr(type, '-')) == NULL) {
	free(type);
	break;
	}
	idx = p + 1;
	*p = '\0';

	/*
	* If the types don't match then keep looking.
	*/
	if (strncmp(val, type, strlen(val)) != 0) {
	free(type);
	break;
	}

	verify(zpool_vdev_is_interior(type));
	verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ID,
	&id) == 0);

	errno = 0;
	vdev_id = strtoull(idx, &end, 10);

	/*
	* If we are looking for a raidz and a parity is
	* specified, make sure it matches.
	*/
	int rzlen = strlen(VDEV_TYPE_RAIDZ);
	assert(rzlen == strlen(VDEV_TYPE_DRAID));
	int typlen = strlen(type);
	if ((strncmp(type, VDEV_TYPE_RAIDZ, rzlen) == 0 \|\|
	strncmp(type, VDEV_TYPE_DRAID, rzlen) == 0) &&
	typlen != rzlen) {
	uint64_t vdev_parity;
	int parity = *(type + rzlen) - '0';

	if (parity <= 0 \|\| parity > 3 \|\|
	(typlen - rzlen) != 1) {
	/*
	* Nonsense parity specified, can
	* never match
	*/
	free(type);
	return (NULL);
	}
	verify(nvlist_lookup_uint64(nv,
	ZPOOL_CONFIG_NPARITY, &vdev_parity) == 0);
	if ((int)vdev_parity != parity) {
	free(type);
	break;
	}
	}

	free(type);
	if (errno != 0)
	return (NULL);

	/*
	* Now verify that we have the correct vdev id.
	*/
	if (vdev_id == id)
	return (nv);
	}

	/*
	* Common case
	*/
	if (strcmp(srchval, val) == 0)
	return (nv);
	break;
	}

	default:
	break;
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0)
	return (NULL);

	for (c = 0; c < children; c++) {
	if ((ret = vdev_to_nvlist_iter(child[c], search,
	avail_spare, l2cache, NULL)) != NULL) {
	/*
	* The 'is_log' value is only set for the toplevel
	* vdev, not the leaf vdevs. So we always lookup the
	* log device from the root of the vdev tree (where
	* 'log' is non-NULL).
	*/
	if (log != NULL &&
	nvlist_lookup_uint64(child[c],
	ZPOOL_CONFIG_IS_LOG, &is_log) == 0 &&
	is_log) {
	*log = B_TRUE;
	}
	return (ret);
	}
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_SPARES,
	&child, &children) == 0) {
	for (c = 0; c < children; c++) {
	if ((ret = vdev_to_nvlist_iter(child[c], search,
	avail_spare, l2cache, NULL)) != NULL) {
	*avail_spare = B_TRUE;
	return (ret);
	}
	}
	}

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_L2CACHE,
	&child, &children) == 0) {
	for (c = 0; c < children; c++) {
	if ((ret = vdev_to_nvlist_iter(child[c], search,
	avail_spare, l2cache, NULL)) != NULL) {
	*l2cache = B_TRUE;
	return (ret);
	}
	}
	}

	return (NULL);
	}

	/*
	* Given a physical path or guid, find the associated vdev.
	*/
	nvlist_t *
	zpool_find_vdev_by_physpath(zpool_handle_t zhp, const char ppath,
	boolean_t avail_spare, boolean_t l2cache, boolean_t *log)
	{
	nvlist_t search, nvroot, *ret;
	uint64_t guid;
	char *end;

	verify(nvlist_alloc(&search, NV_UNIQUE_NAME, KM_SLEEP) == 0);

	guid = strtoull(ppath, &end, 0);
	if (guid != 0 && *end == '\0') {
	verify(nvlist_add_uint64(search, ZPOOL_CONFIG_GUID, guid) == 0);
	} else {
	verify(nvlist_add_string(search, ZPOOL_CONFIG_PHYS_PATH,
	ppath) == 0);
	}

	verify(nvlist_lookup_nvlist(zhp->zpool_config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) == 0);

	*avail_spare = B_FALSE;
	*l2cache = B_FALSE;
	if (log != NULL)
	*log = B_FALSE;
	ret = vdev_to_nvlist_iter(nvroot, search, avail_spare, l2cache, log);
	nvlist_free(search);

	return (ret);
	}

	/*
	* Determine if we have an "interior" top-level vdev (i.e mirror/raidz).
	*/
	static boolean_t
	zpool_vdev_is_interior(const char *name)
	{
	if (strncmp(name, VDEV_TYPE_RAIDZ, strlen(VDEV_TYPE_RAIDZ)) == 0 \|\|
	strncmp(name, VDEV_TYPE_SPARE, strlen(VDEV_TYPE_SPARE)) == 0 \|\|
	strncmp(name,
	VDEV_TYPE_REPLACING, strlen(VDEV_TYPE_REPLACING)) == 0 \|\|
	strncmp(name, VDEV_TYPE_MIRROR, strlen(VDEV_TYPE_MIRROR)) == 0)
	return (B_TRUE);

	if (strncmp(name, VDEV_TYPE_DRAID, strlen(VDEV_TYPE_DRAID)) == 0 &&
	!zpool_is_draid_spare(name))
	return (B_TRUE);

	return (B_FALSE);
	}

	nvlist_t *
	zpool_find_vdev(zpool_handle_t zhp, const char path, boolean_t *avail_spare,
	boolean_t l2cache, boolean_t log)
	{
	char *end;
	nvlist_t nvroot, search, *ret;
	uint64_t guid;

	verify(nvlist_alloc(&search, NV_UNIQUE_NAME, KM_SLEEP) == 0);

	guid = strtoull(path, &end, 0);
	if (guid != 0 && *end == '\0') {
	verify(nvlist_add_uint64(search, ZPOOL_CONFIG_GUID, guid) == 0);
	} else if (zpool_vdev_is_interior(path)) {
	verify(nvlist_add_string(search, ZPOOL_CONFIG_TYPE, path) == 0);
	} else {
	verify(nvlist_add_string(search, ZPOOL_CONFIG_PATH, path) == 0);
	}

	verify(nvlist_lookup_nvlist(zhp->zpool_config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) == 0);

	*avail_spare = B_FALSE;
	*l2cache = B_FALSE;
	if (log != NULL)
	*log = B_FALSE;
	ret = vdev_to_nvlist_iter(nvroot, search, avail_spare, l2cache, log);
	nvlist_free(search);

	return (ret);
	}

	static int
	vdev_is_online(nvlist_t *nv)
	{
	uint64_t ival;

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_OFFLINE, &ival) == 0 \|\|
	nvlist_lookup_uint64(nv, ZPOOL_CONFIG_FAULTED, &ival) == 0 \|\|
	nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REMOVED, &ival) == 0)
	return (0);

	return (1);
	}

	/*
	* Helper function for zpool_get_physpaths().
	*/
	static int
	vdev_get_one_physpath(nvlist_t config, char physpath, size_t physpath_size,
	size_t *bytes_written)
	{
	size_t bytes_left, pos, rsz;
	char *tmppath;
	const char *format;

	if (nvlist_lookup_string(config, ZPOOL_CONFIG_PHYS_PATH,
	&tmppath) != 0)
	return (EZFS_NODEVICE);

	pos = *bytes_written;
	bytes_left = physpath_size - pos;
	format = (pos == 0) ? "%s" : " %s";

	rsz = snprintf(physpath + pos, bytes_left, format, tmppath);
	*bytes_written += rsz;

	if (rsz >= bytes_left) {
	/* if physpath was not copied properly, clear it */
	if (bytes_left != 0) {
	physpath[pos] = 0;
	}
	return (EZFS_NOSPC);
	}
	return (0);
	}

	static int
	vdev_get_physpaths(nvlist_t nv, char physpath, size_t phypath_size,
	size_t *rsz, boolean_t is_spare)
	{
	char *type;
	int ret;

	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) != 0)
	return (EZFS_INVALCONFIG);

	if (strcmp(type, VDEV_TYPE_DISK) == 0) {
	/*
	* An active spare device has ZPOOL_CONFIG_IS_SPARE set.
	* For a spare vdev, we only want to boot from the active
	* spare device.
	*/
	if (is_spare) {
	uint64_t spare = 0;
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_SPARE,
	&spare);
	if (!spare)
	return (EZFS_INVALCONFIG);
	}

	if (vdev_is_online(nv)) {
	if ((ret = vdev_get_one_physpath(nv, physpath,
	phypath_size, rsz)) != 0)
	return (ret);
	}
	} else if (strcmp(type, VDEV_TYPE_MIRROR) == 0 \|\|
	strcmp(type, VDEV_TYPE_RAIDZ) == 0 \|\|
	strcmp(type, VDEV_TYPE_REPLACING) == 0 \|\|
	(is_spare = (strcmp(type, VDEV_TYPE_SPARE) == 0))) {
	nvlist_t **child;
	uint_t count;
	int i, ret;

	if (nvlist_lookup_nvlist_array(nv,
	ZPOOL_CONFIG_CHILDREN, &child, &count) != 0)
	return (EZFS_INVALCONFIG);

	for (i = 0; i < count; i++) {
	ret = vdev_get_physpaths(child[i], physpath,
	phypath_size, rsz, is_spare);
	if (ret == EZFS_NOSPC)
	return (ret);
	}
	}

	return (EZFS_POOL_INVALARG);
	}

	/*
	* Get phys_path for a root pool config.
	* Return 0 on success; non-zero on failure.
	*/
	static int
	zpool_get_config_physpath(nvlist_t config, char physpath, size_t phypath_size)
	{
	size_t rsz;
	nvlist_t *vdev_root;
	nvlist_t **child;
	uint_t count;
	char *type;

	rsz = 0;

	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&vdev_root) != 0)
	return (EZFS_INVALCONFIG);

	if (nvlist_lookup_string(vdev_root, ZPOOL_CONFIG_TYPE, &type) != 0 \|\|
	nvlist_lookup_nvlist_array(vdev_root, ZPOOL_CONFIG_CHILDREN,
	&child, &count) != 0)
	return (EZFS_INVALCONFIG);

	/*
	* root pool can only have a single top-level vdev.
	*/
	if (strcmp(type, VDEV_TYPE_ROOT) != 0 \|\| count != 1)
	return (EZFS_POOL_INVALARG);

	(void) vdev_get_physpaths(child[0], physpath, phypath_size, &rsz,
	B_FALSE);

	/* No online devices */
	if (rsz == 0)
	return (EZFS_NODEVICE);

	return (0);
	}

	/*
	* Get phys_path for a root pool
	* Return 0 on success; non-zero on failure.
	*/
	int
	zpool_get_physpath(zpool_handle_t zhp, char physpath, size_t phypath_size)
	{
	return (zpool_get_config_physpath(zhp->zpool_config, physpath,
	phypath_size));
	}

	/*
	* Convert a vdev path to a GUID. Returns GUID or 0 on error.
	*
	* If is_spare, is_l2cache, or is_log is non-NULL, then store within it
	* if the VDEV is a spare, l2cache, or log device. If they're NULL then
	* ignore them.
	*/
	static uint64_t
	zpool_vdev_path_to_guid_impl(zpool_handle_t zhp, const char path,
	boolean_t is_spare, boolean_t is_l2cache, boolean_t *is_log)
	{
	uint64_t guid;
	boolean_t spare = B_FALSE, l2cache = B_FALSE, log = B_FALSE;
	nvlist_t *tgt;

	if ((tgt = zpool_find_vdev(zhp, path, &spare, &l2cache,
	&log)) == NULL)
	return (0);

	verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &guid) == 0);
	if (is_spare != NULL)
	*is_spare = spare;
	if (is_l2cache != NULL)
	*is_l2cache = l2cache;
	if (is_log != NULL)
	*is_log = log;

	return (guid);
	}

	/* Convert a vdev path to a GUID. Returns GUID or 0 on error. */
	uint64_t
	zpool_vdev_path_to_guid(zpool_handle_t zhp, const char path)
	{
	return (zpool_vdev_path_to_guid_impl(zhp, path, NULL, NULL, NULL));
	}

	/*
	* Bring the specified vdev online. The 'flags' parameter is a set of the
	* ZFS_ONLINE_* flags.
	*/
	int
	zpool_vdev_online(zpool_handle_t zhp, const char path, int flags,
	vdev_state_t *newstate)
	{
	zfs_cmd_t zc = {"\0"};
	char msg[1024];
	char *pathname;
	nvlist_t *tgt;
	boolean_t avail_spare, l2cache, islog;
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	int error;

	if (flags & ZFS_ONLINE_EXPAND) {
	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot expand %s"), path);
	} else {
	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot online %s"), path);
	}

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache,
	&islog)) == NULL)
	return (zfs_error(hdl, EZFS_NODEVICE, msg));

	verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0);

	if (avail_spare)
	return (zfs_error(hdl, EZFS_ISSPARE, msg));

	if ((flags & ZFS_ONLINE_EXPAND \|\|
	zpool_get_prop_int(zhp, ZPOOL_PROP_AUTOEXPAND, NULL)) &&
	nvlist_lookup_string(tgt, ZPOOL_CONFIG_PATH, &pathname) == 0) {
	uint64_t wholedisk = 0;

	(void) nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_WHOLE_DISK,
	&wholedisk);

	/*
	* XXX - L2ARC 1.0 devices can't support expansion.
	*/
	if (l2cache) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"cannot expand cache devices"));
	return (zfs_error(hdl, EZFS_VDEVNOTSUP, msg));
	}

	if (wholedisk) {
	const char *fullpath = path;
	char buf[MAXPATHLEN];

	if (path[0] != '/') {
	error = zfs_resolve_shortname(path, buf,
	sizeof (buf));
	if (error != 0)
	return (zfs_error(hdl, EZFS_NODEVICE,
	msg));

	fullpath = buf;
	}

	error = zpool_relabel_disk(hdl, fullpath, msg);
	if (error != 0)
	return (error);
	}
	}

	zc.zc_cookie = VDEV_STATE_ONLINE;
	zc.zc_obj = flags;

	if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SET_STATE, &zc) != 0) {
	if (errno == EINVAL) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "was split "
	"from this pool into a new one. Use '%s' "
	"instead"), "zpool detach");
	return (zfs_error(hdl, EZFS_POSTSPLIT_ONLINE, msg));
	}
	return (zpool_standard_error(hdl, errno, msg));
	}

	*newstate = zc.zc_cookie;
	return (0);
	}

	/*
	* Take the specified vdev offline
	*/
	int
	zpool_vdev_offline(zpool_handle_t zhp, const char path, boolean_t istmp)
	{
	zfs_cmd_t zc = {"\0"};
	char msg[1024];
	nvlist_t *tgt;
	boolean_t avail_spare, l2cache;
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot offline %s"), path);

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache,
	NULL)) == NULL)
	return (zfs_error(hdl, EZFS_NODEVICE, msg));

	verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0);

	if (avail_spare)
	return (zfs_error(hdl, EZFS_ISSPARE, msg));

	zc.zc_cookie = VDEV_STATE_OFFLINE;
	zc.zc_obj = istmp ? ZFS_OFFLINE_TEMPORARY : 0;

	if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SET_STATE, &zc) == 0)
	return (0);

	switch (errno) {
	case EBUSY:

	/*
	* There are no other replicas of this device.
	*/
	return (zfs_error(hdl, EZFS_NOREPLICAS, msg));

	case EEXIST:
	/*
	* The log device has unplayed logs
	*/
	return (zfs_error(hdl, EZFS_UNPLAYED_LOGS, msg));

	default:
	return (zpool_standard_error(hdl, errno, msg));
	}
	}

	/*
	* Mark the given vdev faulted.
	*/
	int
	zpool_vdev_fault(zpool_handle_t *zhp, uint64_t guid, vdev_aux_t aux)
	{
	zfs_cmd_t zc = {"\0"};
	char msg[1024];
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot fault %llu"), (u_longlong_t)guid);

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	zc.zc_guid = guid;
	zc.zc_cookie = VDEV_STATE_FAULTED;
	zc.zc_obj = aux;

	if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SET_STATE, &zc) == 0)
	return (0);

	switch (errno) {
	case EBUSY:

	/*
	* There are no other replicas of this device.
	*/
	return (zfs_error(hdl, EZFS_NOREPLICAS, msg));

	default:
	return (zpool_standard_error(hdl, errno, msg));
	}

	}

	/*
	* Mark the given vdev degraded.
	*/
	int
	zpool_vdev_degrade(zpool_handle_t *zhp, uint64_t guid, vdev_aux_t aux)
	{
	zfs_cmd_t zc = {"\0"};
	char msg[1024];
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot degrade %llu"), (u_longlong_t)guid);

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	zc.zc_guid = guid;
	zc.zc_cookie = VDEV_STATE_DEGRADED;
	zc.zc_obj = aux;

	if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SET_STATE, &zc) == 0)
	return (0);

	return (zpool_standard_error(hdl, errno, msg));
	}

	/*
	* Returns TRUE if the given nvlist is a vdev that was originally swapped in as
	* a hot spare.
	*/
	static boolean_t
	is_replacing_spare(nvlist_t search, nvlist_t tgt, int which)
	{
	nvlist_t **child;
	uint_t c, children;
	char *type;

	if (nvlist_lookup_nvlist_array(search, ZPOOL_CONFIG_CHILDREN, &child,
	&children) == 0) {
	verify(nvlist_lookup_string(search, ZPOOL_CONFIG_TYPE,
	&type) == 0);

	if ((strcmp(type, VDEV_TYPE_SPARE) == 0 \|\|
	strcmp(type, VDEV_TYPE_DRAID_SPARE) == 0) &&
	children == 2 && child[which] == tgt)
	return (B_TRUE);

	for (c = 0; c < children; c++)
	if (is_replacing_spare(child[c], tgt, which))
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	/*
	* Attach new_disk (fully described by nvroot) to old_disk.
	* If 'replacing' is specified, the new disk will replace the old one.
	*/
	int
	zpool_vdev_attach(zpool_handle_t zhp, const char old_disk,
	const char new_disk, nvlist_t nvroot, int replacing, boolean_t rebuild)
	{
	zfs_cmd_t zc = {"\0"};
	char msg[1024];
	int ret;
	nvlist_t *tgt;
	boolean_t avail_spare, l2cache, islog;
	uint64_t val;
	char *newname;
	nvlist_t **child;
	uint_t children;
	nvlist_t *config_root;
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	if (replacing)
	(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
	"cannot replace %s with %s"), old_disk, new_disk);
	else
	(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,
	"cannot attach %s to %s"), new_disk, old_disk);

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	if ((tgt = zpool_find_vdev(zhp, old_disk, &avail_spare, &l2cache,
	&islog)) == NULL)
	return (zfs_error(hdl, EZFS_NODEVICE, msg));

	if (avail_spare)
	return (zfs_error(hdl, EZFS_ISSPARE, msg));

	if (l2cache)
	return (zfs_error(hdl, EZFS_ISL2CACHE, msg));

	verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0);
	zc.zc_cookie = replacing;
	zc.zc_simple = rebuild;

	if (rebuild &&
	zfeature_lookup_guid("org.openzfs:device_rebuild", NULL) != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"the loaded zfs module doesn't support device rebuilds"));
	return (zfs_error(hdl, EZFS_POOL_NOTSUP, msg));
	}

	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0 \|\| children != 1) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"new device must be a single disk"));
	return (zfs_error(hdl, EZFS_INVALCONFIG, msg));
	}

	verify(nvlist_lookup_nvlist(zpool_get_config(zhp, NULL),
	ZPOOL_CONFIG_VDEV_TREE, &config_root) == 0);

	if ((newname = zpool_vdev_name(NULL, NULL, child[0], 0)) == NULL)
	return (-1);

	/*
	* If the target is a hot spare that has been swapped in, we can only
	* replace it with another hot spare.
	*/
	if (replacing &&
	nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_IS_SPARE, &val) == 0 &&
	(zpool_find_vdev(zhp, newname, &avail_spare, &l2cache,
	NULL) == NULL \|\| !avail_spare) &&
	is_replacing_spare(config_root, tgt, 1)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"can only be replaced by another hot spare"));
	free(newname);
	return (zfs_error(hdl, EZFS_BADTARGET, msg));
	}

	free(newname);

	if (zcmd_write_conf_nvlist(hdl, &zc, nvroot) != 0)
	return (-1);

	ret = zfs_ioctl(hdl, ZFS_IOC_VDEV_ATTACH, &zc);

	zcmd_free_nvlists(&zc);

	if (ret == 0)
	return (0);

	switch (errno) {
	case ENOTSUP:
	/*
	* Can't attach to or replace this type of vdev.
	*/
	if (replacing) {
	uint64_t version = zpool_get_prop_int(zhp,
	ZPOOL_PROP_VERSION, NULL);

	if (islog) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"cannot replace a log with a spare"));
	} else if (rebuild) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"only mirror and dRAID vdevs support "
	"sequential reconstruction"));
	} else if (zpool_is_draid_spare(new_disk)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"dRAID spares can only replace child "
	"devices in their parent's dRAID vdev"));
	} else if (version >= SPA_VERSION_MULTI_REPLACE) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"already in replacing/spare config; wait "
	"for completion or use 'zpool detach'"));
	} else {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"cannot replace a replacing device"));
	}
	} else {
	char status[64] = {0};
	zpool_prop_get_feature(zhp,
	"feature@device_rebuild", status, 63);
	if (rebuild &&
	strncmp(status, ZFS_FEATURE_DISABLED, 64) == 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"device_rebuild feature must be enabled "
	"in order to use sequential "
	"reconstruction"));
	} else {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"can only attach to mirrors and top-level "
	"disks"));
	}
	}
	(void) zfs_error(hdl, EZFS_BADTARGET, msg);
	break;

	case EINVAL:
	/*
	* The new device must be a single disk.
	*/
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"new device must be a single disk"));
	(void) zfs_error(hdl, EZFS_INVALCONFIG, msg);
	break;

	case EBUSY:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "%s is busy, "
	"or device removal is in progress"),
	new_disk);
	(void) zfs_error(hdl, EZFS_BADDEV, msg);
	break;

	case EOVERFLOW:
	/*
	* The new device is too small.
	*/
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"device is too small"));
	(void) zfs_error(hdl, EZFS_BADDEV, msg);
	break;

	case EDOM:
	/*
	* The new device has a different optimal sector size.
	*/
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"new device has a different optimal sector size; use the "
	"option '-o ashift=N' to override the optimal size"));
	(void) zfs_error(hdl, EZFS_BADDEV, msg);
	break;

	case ENAMETOOLONG:
	/*
	* The resulting top-level vdev spec won't fit in the label.
	*/
	(void) zfs_error(hdl, EZFS_DEVOVERFLOW, msg);
	break;

	default:
	(void) zpool_standard_error(hdl, errno, msg);
	}

	return (-1);
	}

	/*
	* Detach the specified device.
	*/
	int
	zpool_vdev_detach(zpool_handle_t zhp, const char path)
	{
	zfs_cmd_t zc = {"\0"};
	char msg[1024];
	nvlist_t *tgt;
	boolean_t avail_spare, l2cache;
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot detach %s"), path);

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache,
	NULL)) == NULL)
	return (zfs_error(hdl, EZFS_NODEVICE, msg));

	if (avail_spare)
	return (zfs_error(hdl, EZFS_ISSPARE, msg));

	if (l2cache)
	return (zfs_error(hdl, EZFS_ISL2CACHE, msg));

	verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, &zc.zc_guid) == 0);

	if (zfs_ioctl(hdl, ZFS_IOC_VDEV_DETACH, &zc) == 0)
	return (0);

	switch (errno) {

	case ENOTSUP:
	/*
	* Can't detach from this type of vdev.
	*/
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "only "
	"applicable to mirror and replacing vdevs"));
	(void) zfs_error(hdl, EZFS_BADTARGET, msg);
	break;

	case EBUSY:
	/*
	* There are no other replicas of this device.
	*/
	(void) zfs_error(hdl, EZFS_NOREPLICAS, msg);
	break;

	default:
	(void) zpool_standard_error(hdl, errno, msg);
	}

	return (-1);
	}

	/*
	* Find a mirror vdev in the source nvlist.
	*
	* The mchild array contains a list of disks in one of the top-level mirrors
	* of the source pool. The schild array contains a list of disks that the
	* user specified on the command line. We loop over the mchild array to
	* see if any entry in the schild array matches.
	*
	* If a disk in the mchild array is found in the schild array, we return
	* the index of that entry. Otherwise we return -1.
	*/
	static int
	find_vdev_entry(zpool_handle_t zhp, nvlist_t *mchild, uint_t mchildren,
	nvlist_t **schild, uint_t schildren)
	{
	uint_t mc;

	for (mc = 0; mc < mchildren; mc++) {
	uint_t sc;
	char *mpath = zpool_vdev_name(zhp->zpool_hdl, zhp,
	mchild[mc], 0);

	for (sc = 0; sc < schildren; sc++) {
	char *spath = zpool_vdev_name(zhp->zpool_hdl, zhp,
	schild[sc], 0);
	boolean_t result = (strcmp(mpath, spath) == 0);

	free(spath);
	if (result) {
	free(mpath);
	return (mc);
	}
	}

	free(mpath);
	}

	return (-1);
	}

	/*
	* Split a mirror pool. If newroot points to null, then a new nvlist
	* is generated and it is the responsibility of the caller to free it.
	*/
	int
	zpool_vdev_split(zpool_handle_t zhp, char newname, nvlist_t **newroot,
	nvlist_t *props, splitflags_t flags)
	{
	zfs_cmd_t zc = {"\0"};
	char msg[1024], *bias;
	nvlist_t tree, config, child, newchild, *newconfig = NULL;
	nvlist_t *varray = NULL, zc_props = NULL;
	uint_t c, children, newchildren, lastlog = 0, vcount, found = 0;
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	uint64_t vers, readonly = B_FALSE;
	boolean_t freelist = B_FALSE, memory_err = B_TRUE;
	int retval = 0;

	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "Unable to split %s"), zhp->zpool_name);

	if (!zpool_name_valid(hdl, B_FALSE, newname))
	return (zfs_error(hdl, EZFS_INVALIDNAME, msg));

	if ((config = zpool_get_config(zhp, NULL)) == NULL) {
	(void) fprintf(stderr, gettext("Internal error: unable to "
	"retrieve pool configuration\n"));
	return (-1);
	}

	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &tree)
	== 0);
	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION, &vers) == 0);

	if (props) {
	prop_flags_t flags = { .create = B_FALSE, .import = B_TRUE };
	if ((zc_props = zpool_valid_proplist(hdl, zhp->zpool_name,
	props, vers, flags, msg)) == NULL)
	return (-1);
	(void) nvlist_lookup_uint64(zc_props,
	zpool_prop_to_name(ZPOOL_PROP_READONLY), &readonly);
	if (readonly) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property %s can only be set at import time"),
	zpool_prop_to_name(ZPOOL_PROP_READONLY));
	return (-1);
	}
	}

	if (nvlist_lookup_nvlist_array(tree, ZPOOL_CONFIG_CHILDREN, &child,
	&children) != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Source pool is missing vdev tree"));
	nvlist_free(zc_props);
	return (-1);
	}

	varray = zfs_alloc(hdl, children * sizeof (nvlist_t *));
	vcount = 0;

	if (*newroot == NULL \|\|
	nvlist_lookup_nvlist_array(*newroot, ZPOOL_CONFIG_CHILDREN,
	&newchild, &newchildren) != 0)
	newchildren = 0;

	for (c = 0; c < children; c++) {
	uint64_t is_log = B_FALSE, is_hole = B_FALSE;
	boolean_t is_special = B_FALSE, is_dedup = B_FALSE;
	char *type;
	nvlist_t *mchild, vdev;
	uint_t mchildren;
	int entry;

	/*
	* Unlike cache & spares, slogs are stored in the
	* ZPOOL_CONFIG_CHILDREN array. We filter them out here.
	*/
	(void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_LOG,
	&is_log);
	(void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_HOLE,
	&is_hole);
	if (is_log \|\| is_hole) {
	/*
	* Create a hole vdev and put it in the config.
	*/
	if (nvlist_alloc(&vdev, NV_UNIQUE_NAME, 0) != 0)
	goto out;
	if (nvlist_add_string(vdev, ZPOOL_CONFIG_TYPE,
	VDEV_TYPE_HOLE) != 0)
	goto out;
	if (nvlist_add_uint64(vdev, ZPOOL_CONFIG_IS_HOLE,
	1) != 0)
	goto out;
	if (lastlog == 0)
	lastlog = vcount;
	varray[vcount++] = vdev;
	continue;
	}
	lastlog = 0;
	verify(nvlist_lookup_string(child[c], ZPOOL_CONFIG_TYPE, &type)
	== 0);

	if (strcmp(type, VDEV_TYPE_INDIRECT) == 0) {
	vdev = child[c];
	if (nvlist_dup(vdev, &varray[vcount++], 0) != 0)
	goto out;
	continue;
	} else if (strcmp(type, VDEV_TYPE_MIRROR) != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Source pool must be composed only of mirrors\n"));
	retval = zfs_error(hdl, EZFS_INVALCONFIG, msg);
	goto out;
	}

	if (nvlist_lookup_string(child[c],
	ZPOOL_CONFIG_ALLOCATION_BIAS, &bias) == 0) {
	if (strcmp(bias, VDEV_ALLOC_BIAS_SPECIAL) == 0)
	is_special = B_TRUE;
	else if (strcmp(bias, VDEV_ALLOC_BIAS_DEDUP) == 0)
	is_dedup = B_TRUE;
	}
	verify(nvlist_lookup_nvlist_array(child[c],
	ZPOOL_CONFIG_CHILDREN, &mchild, &mchildren) == 0);

	/* find or add an entry for this top-level vdev */
	if (newchildren > 0 &&
	(entry = find_vdev_entry(zhp, mchild, mchildren,
	newchild, newchildren)) >= 0) {
	/* We found a disk that the user specified. */
	vdev = mchild[entry];
	++found;
	} else {
	/* User didn't specify a disk for this vdev. */
	vdev = mchild[mchildren - 1];
	}

	if (nvlist_dup(vdev, &varray[vcount++], 0) != 0)
	goto out;

	if (flags.dryrun != 0) {
	if (is_dedup == B_TRUE) {
	if (nvlist_add_string(varray[vcount - 1],
	ZPOOL_CONFIG_ALLOCATION_BIAS,
	VDEV_ALLOC_BIAS_DEDUP) != 0)
	goto out;
	} else if (is_special == B_TRUE) {
	if (nvlist_add_string(varray[vcount - 1],
	ZPOOL_CONFIG_ALLOCATION_BIAS,
	VDEV_ALLOC_BIAS_SPECIAL) != 0)
	goto out;
	}
	}
	}

	/* did we find every disk the user specified? */
	if (found != newchildren) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "Device list must "
	"include at most one disk from each mirror"));
	retval = zfs_error(hdl, EZFS_INVALCONFIG, msg);
	goto out;
	}

	/* Prepare the nvlist for populating. */
	if (*newroot == NULL) {
	if (nvlist_alloc(newroot, NV_UNIQUE_NAME, 0) != 0)
	goto out;
	freelist = B_TRUE;
	if (nvlist_add_string(*newroot, ZPOOL_CONFIG_TYPE,
	VDEV_TYPE_ROOT) != 0)
	goto out;
	} else {
	verify(nvlist_remove_all(*newroot, ZPOOL_CONFIG_CHILDREN) == 0);
	}

	/* Add all the children we found */
	- if (nvlist_add_nvlist_array(*newroot, ZPOOL_CONFIG_CHILDREN, varray,
	- lastlog == 0 ? vcount : lastlog) != 0)
	+ if (nvlist_add_nvlist_array(*newroot, ZPOOL_CONFIG_CHILDREN,
	+ (const nvlist_t **)varray, lastlog == 0 ? vcount : lastlog) != 0)
	goto out;

	/*
	* If we're just doing a dry run, exit now with success.
	*/
	if (flags.dryrun) {
	memory_err = B_FALSE;
	freelist = B_FALSE;
	goto out;
	}

	/* now build up the config list & call the ioctl */
	if (nvlist_alloc(&newconfig, NV_UNIQUE_NAME, 0) != 0)
	goto out;

	if (nvlist_add_nvlist(newconfig,
	ZPOOL_CONFIG_VDEV_TREE, *newroot) != 0 \|\|
	nvlist_add_string(newconfig,
	ZPOOL_CONFIG_POOL_NAME, newname) != 0 \|\|
	nvlist_add_uint64(newconfig, ZPOOL_CONFIG_VERSION, vers) != 0)
	goto out;

	/*
	* The new pool is automatically part of the namespace unless we
	* explicitly export it.
	*/
	if (!flags.import)
	zc.zc_cookie = ZPOOL_EXPORT_AFTER_SPLIT;
	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	(void) strlcpy(zc.zc_string, newname, sizeof (zc.zc_string));
	if (zcmd_write_conf_nvlist(hdl, &zc, newconfig) != 0)
	goto out;
	if (zc_props != NULL && zcmd_write_src_nvlist(hdl, &zc, zc_props) != 0)
	goto out;

	if (zfs_ioctl(hdl, ZFS_IOC_VDEV_SPLIT, &zc) != 0) {
	retval = zpool_standard_error(hdl, errno, msg);
	goto out;
	}

	freelist = B_FALSE;
	memory_err = B_FALSE;

	out:
	if (varray != NULL) {
	int v;

	for (v = 0; v < vcount; v++)
	nvlist_free(varray[v]);
	free(varray);
	}
	zcmd_free_nvlists(&zc);
	nvlist_free(zc_props);
	nvlist_free(newconfig);
	if (freelist) {
	nvlist_free(*newroot);
	*newroot = NULL;
	}

	if (retval != 0)
	return (retval);

	if (memory_err)
	return (no_memory(hdl));

	return (0);
	}

	/*
	* Remove the given device.
	*/
	int
	zpool_vdev_remove(zpool_handle_t zhp, const char path)
	{
	zfs_cmd_t zc = {"\0"};
	char msg[1024];
	nvlist_t *tgt;
	boolean_t avail_spare, l2cache, islog;
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	uint64_t version;

	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot remove %s"), path);

	if (zpool_is_draid_spare(path)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"dRAID spares cannot be removed"));
	return (zfs_error(hdl, EZFS_NODEVICE, msg));
	}

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache,
	&islog)) == NULL)
	return (zfs_error(hdl, EZFS_NODEVICE, msg));

	version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL);
	if (islog && version < SPA_VERSION_HOLES) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"pool must be upgraded to support log removal"));
	return (zfs_error(hdl, EZFS_BADVERSION, msg));
	}

	zc.zc_guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID);

	if (zfs_ioctl(hdl, ZFS_IOC_VDEV_REMOVE, &zc) == 0)
	return (0);

	switch (errno) {

	case EINVAL:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"invalid config; all top-level vdevs must "
	"have the same sector size and not be raidz."));
	(void) zfs_error(hdl, EZFS_INVALCONFIG, msg);
	break;

	case EBUSY:
	if (islog) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Mount encrypted datasets to replay logs."));
	} else {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Pool busy; removal may already be in progress"));
	}
	(void) zfs_error(hdl, EZFS_BUSY, msg);
	break;

	case EACCES:
	if (islog) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"Mount encrypted datasets to replay logs."));
	(void) zfs_error(hdl, EZFS_BUSY, msg);
	} else {
	(void) zpool_standard_error(hdl, errno, msg);
	}
	break;

	default:
	(void) zpool_standard_error(hdl, errno, msg);
	}
	return (-1);
	}

	int
	zpool_vdev_remove_cancel(zpool_handle_t *zhp)
	{
	zfs_cmd_t zc;
	char msg[1024];
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot cancel removal"));

	bzero(&zc, sizeof (zc));
	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	zc.zc_cookie = 1;

	if (zfs_ioctl(hdl, ZFS_IOC_VDEV_REMOVE, &zc) == 0)
	return (0);

	return (zpool_standard_error(hdl, errno, msg));
	}

	int
	zpool_vdev_indirect_size(zpool_handle_t zhp, const char path,
	uint64_t *sizep)
	{
	char msg[1024];
	nvlist_t *tgt;
	boolean_t avail_spare, l2cache, islog;
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot determine indirect size of %s"),
	path);

	if ((tgt = zpool_find_vdev(zhp, path, &avail_spare, &l2cache,
	&islog)) == NULL)
	return (zfs_error(hdl, EZFS_NODEVICE, msg));

	if (avail_spare \|\| l2cache \|\| islog) {
	*sizep = 0;
	return (0);
	}

	if (nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_INDIRECT_SIZE, sizep) != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"indirect size not available"));
	return (zfs_error(hdl, EINVAL, msg));
	}
	return (0);
	}

	/*
	* Clear the errors for the pool, or the particular device if specified.
	*/
	int
	zpool_clear(zpool_handle_t zhp, const char path, nvlist_t *rewindnvl)
	{
	zfs_cmd_t zc = {"\0"};
	char msg[1024];
	nvlist_t *tgt;
	zpool_load_policy_t policy;
	boolean_t avail_spare, l2cache;
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	nvlist_t *nvi = NULL;
	int error;

	if (path)
	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot clear errors for %s"),
	path);
	else
	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot clear errors for %s"),
	zhp->zpool_name);

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	if (path) {
	if ((tgt = zpool_find_vdev(zhp, path, &avail_spare,
	&l2cache, NULL)) == NULL)
	return (zfs_error(hdl, EZFS_NODEVICE, msg));

	/*
	* Don't allow error clearing for hot spares. Do allow
	* error clearing for l2cache devices.
	*/
	if (avail_spare)
	return (zfs_error(hdl, EZFS_ISSPARE, msg));

	verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID,
	&zc.zc_guid) == 0);
	}

	zpool_get_load_policy(rewindnvl, &policy);
	zc.zc_cookie = policy.zlp_rewind;

	if (zcmd_alloc_dst_nvlist(hdl, &zc, zhp->zpool_config_size * 2) != 0)
	return (-1);

	if (zcmd_write_src_nvlist(hdl, &zc, rewindnvl) != 0)
	return (-1);

	while ((error = zfs_ioctl(hdl, ZFS_IOC_CLEAR, &zc)) != 0 &&
	errno == ENOMEM) {
	if (zcmd_expand_dst_nvlist(hdl, &zc) != 0) {
	zcmd_free_nvlists(&zc);
	return (-1);
	}
	}

	if (!error \|\| ((policy.zlp_rewind & ZPOOL_TRY_REWIND) &&
	errno != EPERM && errno != EACCES)) {
	if (policy.zlp_rewind &
	(ZPOOL_DO_REWIND \| ZPOOL_TRY_REWIND)) {
	(void) zcmd_read_dst_nvlist(hdl, &zc, &nvi);
	zpool_rewind_exclaim(hdl, zc.zc_name,
	((policy.zlp_rewind & ZPOOL_TRY_REWIND) != 0),
	nvi);
	nvlist_free(nvi);
	}
	zcmd_free_nvlists(&zc);
	return (0);
	}

	zcmd_free_nvlists(&zc);
	return (zpool_standard_error(hdl, errno, msg));
	}

	/*
	* Similar to zpool_clear(), but takes a GUID (used by fmd).
	*/
	int
	zpool_vdev_clear(zpool_handle_t *zhp, uint64_t guid)
	{
	zfs_cmd_t zc = {"\0"};
	char msg[1024];
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot clear errors for %llx"),
	(u_longlong_t)guid);

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	zc.zc_guid = guid;
	zc.zc_cookie = ZPOOL_NO_REWIND;

	if (zfs_ioctl(hdl, ZFS_IOC_CLEAR, &zc) == 0)
	return (0);

	return (zpool_standard_error(hdl, errno, msg));
	}

	/*
	* Change the GUID for a pool.
	*/
	int
	zpool_reguid(zpool_handle_t *zhp)
	{
	char msg[1024];
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	zfs_cmd_t zc = {"\0"};

	(void) snprintf(msg, sizeof (msg),
	dgettext(TEXT_DOMAIN, "cannot reguid '%s'"), zhp->zpool_name);

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	if (zfs_ioctl(hdl, ZFS_IOC_POOL_REGUID, &zc) == 0)
	return (0);

	return (zpool_standard_error(hdl, errno, msg));
	}

	/*
	* Reopen the pool.
	*/
	int
	zpool_reopen_one(zpool_handle_t zhp, void data)
	{
	libzfs_handle_t *hdl = zpool_get_handle(zhp);
	const char *pool_name = zpool_get_name(zhp);
	boolean_t *scrub_restart = data;
	int error;

	error = lzc_reopen(pool_name, *scrub_restart);
	if (error) {
	return (zpool_standard_error_fmt(hdl, error,
	dgettext(TEXT_DOMAIN, "cannot reopen '%s'"), pool_name));
	}

	return (0);
	}

	/* call into libzfs_core to execute the sync IOCTL per pool */
	int
	zpool_sync_one(zpool_handle_t zhp, void data)
	{
	int ret;
	libzfs_handle_t *hdl = zpool_get_handle(zhp);
	const char *pool_name = zpool_get_name(zhp);
	boolean_t *force = data;
	nvlist_t *innvl = fnvlist_alloc();

	fnvlist_add_boolean_value(innvl, "force", *force);
	if ((ret = lzc_sync(pool_name, innvl, NULL)) != 0) {
	nvlist_free(innvl);
	return (zpool_standard_error_fmt(hdl, ret,
	dgettext(TEXT_DOMAIN, "sync '%s' failed"), pool_name));
	}
	nvlist_free(innvl);

	return (0);
	}

	#define PATH_BUF_LEN 64

	/*
	* Given a vdev, return the name to display in iostat. If the vdev has a path,
	* we use that, stripping off any leading "/dev/dsk/"; if not, we use the type.
	* We also check if this is a whole disk, in which case we strip off the
	* trailing 's0' slice name.
	*
	* This routine is also responsible for identifying when disks have been
	* reconfigured in a new location. The kernel will have opened the device by
	* devid, but the path will still refer to the old location. To catch this, we
	* first do a path -> devid translation (which is fast for the common case). If
	* the devid matches, we're done. If not, we do a reverse devid -> path
	* translation and issue the appropriate ioctl() to update the path of the vdev.
	* If 'zhp' is NULL, then this is an exported pool, and we don't need to do any
	* of these checks.
	*/
	char *
	zpool_vdev_name(libzfs_handle_t hdl, zpool_handle_t zhp, nvlist_t *nv,
	int name_flags)
	{
	char path, type, *env;
	uint64_t value;
	char buf[PATH_BUF_LEN];
	char tmpbuf[PATH_BUF_LEN];

	/*
	* vdev_name will be "root"/"root-0" for the root vdev, but it is the
	* zpool name that will be displayed to the user.
	*/
	verify(nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) == 0);
	if (zhp != NULL && strcmp(type, "root") == 0)
	return (zfs_strdup(hdl, zpool_get_name(zhp)));

	env = getenv("ZPOOL_VDEV_NAME_PATH");
	if (env && (strtoul(env, NULL, 0) > 0 \|\|
	!strncasecmp(env, "YES", 3) \|\| !strncasecmp(env, "ON", 2)))
	name_flags \|= VDEV_NAME_PATH;

	env = getenv("ZPOOL_VDEV_NAME_GUID");
	if (env && (strtoul(env, NULL, 0) > 0 \|\|
	!strncasecmp(env, "YES", 3) \|\| !strncasecmp(env, "ON", 2)))
	name_flags \|= VDEV_NAME_GUID;

	env = getenv("ZPOOL_VDEV_NAME_FOLLOW_LINKS");
	if (env && (strtoul(env, NULL, 0) > 0 \|\|
	!strncasecmp(env, "YES", 3) \|\| !strncasecmp(env, "ON", 2)))
	name_flags \|= VDEV_NAME_FOLLOW_LINKS;

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT, &value) == 0 \|\|
	name_flags & VDEV_NAME_GUID) {
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &value);
	(void) snprintf(buf, sizeof (buf), "%llu", (u_longlong_t)value);
	path = buf;
	} else if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0) {
	if (name_flags & VDEV_NAME_FOLLOW_LINKS) {
	char *rp = realpath(path, NULL);
	if (rp) {
	strlcpy(buf, rp, sizeof (buf));
	path = buf;
	free(rp);
	}
	}

	/*
	* For a block device only use the name.
	*/
	if ((strcmp(type, VDEV_TYPE_DISK) == 0) &&
	!(name_flags & VDEV_NAME_PATH)) {
	path = zfs_strip_path(path);
	}

	/*
	* Remove the partition from the path if this is a whole disk.
	*/
	if (strcmp(type, VDEV_TYPE_DRAID_SPARE) != 0 &&
	nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK, &value)
	== 0 && value && !(name_flags & VDEV_NAME_PATH)) {
	return (zfs_strip_partition(path));
	}
	} else {
	path = type;

	/*
	* If it's a raidz device, we need to stick in the parity level.
	*/
	if (strcmp(path, VDEV_TYPE_RAIDZ) == 0) {
	verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NPARITY,
	&value) == 0);
	(void) snprintf(buf, sizeof (buf), "%s%llu", path,
	(u_longlong_t)value);
	path = buf;
	}

	/*
	* If it's a dRAID device, we add parity, groups, and spares.
	*/
	if (strcmp(path, VDEV_TYPE_DRAID) == 0) {
	uint64_t ndata, nparity, nspares;
	nvlist_t **child;
	uint_t children;

	verify(nvlist_lookup_nvlist_array(nv,
	ZPOOL_CONFIG_CHILDREN, &child, &children) == 0);
	verify(nvlist_lookup_uint64(nv,
	ZPOOL_CONFIG_NPARITY, &nparity) == 0);
	verify(nvlist_lookup_uint64(nv,
	ZPOOL_CONFIG_DRAID_NDATA, &ndata) == 0);
	verify(nvlist_lookup_uint64(nv,
	ZPOOL_CONFIG_DRAID_NSPARES, &nspares) == 0);

	path = zpool_draid_name(buf, sizeof (buf), ndata,
	nparity, nspares, children);
	}

	/*
	* We identify each top-level vdev by using a <type-id>
	* naming convention.
	*/
	if (name_flags & VDEV_NAME_TYPE_ID) {
	uint64_t id;
	verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ID,
	&id) == 0);
	(void) snprintf(tmpbuf, sizeof (tmpbuf), "%s-%llu",
	path, (u_longlong_t)id);
	path = tmpbuf;
	}
	}

	return (zfs_strdup(hdl, path));
	}

	static int
	zbookmark_mem_compare(const void a, const void b)
	{
	return (memcmp(a, b, sizeof (zbookmark_phys_t)));
	}

	/*
	* Retrieve the persistent error log, uniquify the members, and return to the
	* caller.
	*/
	int
	zpool_get_errlog(zpool_handle_t zhp, nvlist_t *nverrlistp)
	{
	zfs_cmd_t zc = {"\0"};
	libzfs_handle_t *hdl = zhp->zpool_hdl;
	uint64_t count;
	zbookmark_phys_t *zb = NULL;
	int i;

	/*
	* Retrieve the raw error list from the kernel. If the number of errors
	* has increased, allocate more space and continue until we get the
	* entire list.
	*/
	verify(nvlist_lookup_uint64(zhp->zpool_config, ZPOOL_CONFIG_ERRCOUNT,
	&count) == 0);
	if (count == 0)
	return (0);
	zc.zc_nvlist_dst = (uintptr_t)zfs_alloc(zhp->zpool_hdl,
	count * sizeof (zbookmark_phys_t));
	zc.zc_nvlist_dst_size = count;
	(void) strcpy(zc.zc_name, zhp->zpool_name);
	for (;;) {
	if (zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_ERROR_LOG,
	&zc) != 0) {
	free((void *)(uintptr_t)zc.zc_nvlist_dst);
	if (errno == ENOMEM) {
	void *dst;

	count = zc.zc_nvlist_dst_size;
	dst = zfs_alloc(zhp->zpool_hdl, count *
	sizeof (zbookmark_phys_t));
	zc.zc_nvlist_dst = (uintptr_t)dst;
	} else {
	return (zpool_standard_error_fmt(hdl, errno,
	dgettext(TEXT_DOMAIN, "errors: List of "
	"errors unavailable")));
	}
	} else {
	break;
	}
	}

	/*
	* Sort the resulting bookmarks. This is a little confusing due to the
	* implementation of ZFS_IOC_ERROR_LOG. The bookmarks are copied last
	* to first, and 'zc_nvlist_dst_size' indicates the number of bookmarks
	* _not_ copied as part of the process. So we point the start of our
	* array appropriate and decrement the total number of elements.
	*/
	zb = ((zbookmark_phys_t *)(uintptr_t)zc.zc_nvlist_dst) +
	zc.zc_nvlist_dst_size;
	count -= zc.zc_nvlist_dst_size;

	qsort(zb, count, sizeof (zbookmark_phys_t), zbookmark_mem_compare);

	verify(nvlist_alloc(nverrlistp, 0, KM_SLEEP) == 0);

	/*
	* Fill in the nverrlistp with nvlist's of dataset and object numbers.
	*/
	for (i = 0; i < count; i++) {
	nvlist_t *nv;

	/* ignoring zb_blkid and zb_level for now */
	if (i > 0 && zb[i-1].zb_objset == zb[i].zb_objset &&
	zb[i-1].zb_object == zb[i].zb_object)
	continue;

	if (nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_SLEEP) != 0)
	goto nomem;
	if (nvlist_add_uint64(nv, ZPOOL_ERR_DATASET,
	zb[i].zb_objset) != 0) {
	nvlist_free(nv);
	goto nomem;
	}
	if (nvlist_add_uint64(nv, ZPOOL_ERR_OBJECT,
	zb[i].zb_object) != 0) {
	nvlist_free(nv);
	goto nomem;
	}
	if (nvlist_add_nvlist(*nverrlistp, "ejk", nv) != 0) {
	nvlist_free(nv);
	goto nomem;
	}
	nvlist_free(nv);
	}

	free((void *)(uintptr_t)zc.zc_nvlist_dst);
	return (0);

	nomem:
	free((void *)(uintptr_t)zc.zc_nvlist_dst);
	return (no_memory(zhp->zpool_hdl));
	}

	/*
	* Upgrade a ZFS pool to the latest on-disk version.
	*/
	int
	zpool_upgrade(zpool_handle_t *zhp, uint64_t new_version)
	{
	zfs_cmd_t zc = {"\0"};
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	(void) strcpy(zc.zc_name, zhp->zpool_name);
	zc.zc_cookie = new_version;

	if (zfs_ioctl(hdl, ZFS_IOC_POOL_UPGRADE, &zc) != 0)
	return (zpool_standard_error_fmt(hdl, errno,
	dgettext(TEXT_DOMAIN, "cannot upgrade '%s'"),
	zhp->zpool_name));
	return (0);
	}

	void
	zfs_save_arguments(int argc, char *argv, char string, int len)
	{
	int i;

	(void) strlcpy(string, zfs_basename(argv[0]), len);
	for (i = 1; i < argc; i++) {
	(void) strlcat(string, " ", len);
	(void) strlcat(string, argv[i], len);
	}
	}

	int
	zpool_log_history(libzfs_handle_t hdl, const char message)
	{
	zfs_cmd_t zc = {"\0"};
	nvlist_t *args;
	int err;

	args = fnvlist_alloc();
	fnvlist_add_string(args, "message", message);
	err = zcmd_write_src_nvlist(hdl, &zc, args);
	if (err == 0)
	err = zfs_ioctl(hdl, ZFS_IOC_LOG_HISTORY, &zc);
	nvlist_free(args);
	zcmd_free_nvlists(&zc);
	return (err);
	}

	/*
	* Perform ioctl to get some command history of a pool.
	*
	* 'buf' is the buffer to fill up to 'len' bytes. 'off' is the
	* logical offset of the history buffer to start reading from.
	*
	* Upon return, 'off' is the next logical offset to read from and
	* 'len' is the actual amount of bytes read into 'buf'.
	*/
	static int
	get_history(zpool_handle_t zhp, char buf, uint64_t off, uint64_t len)
	{
	zfs_cmd_t zc = {"\0"};
	libzfs_handle_t *hdl = zhp->zpool_hdl;

	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));

	zc.zc_history = (uint64_t)(uintptr_t)buf;
	zc.zc_history_len = *len;
	zc.zc_history_offset = *off;

	if (zfs_ioctl(hdl, ZFS_IOC_POOL_GET_HISTORY, &zc) != 0) {
	switch (errno) {
	case EPERM:
	return (zfs_error_fmt(hdl, EZFS_PERM,
	dgettext(TEXT_DOMAIN,
	"cannot show history for pool '%s'"),
	zhp->zpool_name));
	case ENOENT:
	return (zfs_error_fmt(hdl, EZFS_NOHISTORY,
	dgettext(TEXT_DOMAIN, "cannot get history for pool "
	"'%s'"), zhp->zpool_name));
	case ENOTSUP:
	return (zfs_error_fmt(hdl, EZFS_BADVERSION,
	dgettext(TEXT_DOMAIN, "cannot get history for pool "
	"'%s', pool must be upgraded"), zhp->zpool_name));
	default:
	return (zpool_standard_error_fmt(hdl, errno,
	dgettext(TEXT_DOMAIN,
	"cannot get history for '%s'"), zhp->zpool_name));
	}
	}

	*len = zc.zc_history_len;
	*off = zc.zc_history_offset;

	return (0);
	}

	/*
	* Retrieve the command history of a pool.
	*/
	int
	zpool_get_history(zpool_handle_t zhp, nvlist_t nvhisp, uint64_t off,
	boolean_t *eof)
	{
	char *buf;
	int buflen = 128 * 1024;
	nvlist_t **records = NULL;
	uint_t numrecords = 0;
	int err, i;
	uint64_t start = *off;

	buf = malloc(buflen);
	if (buf == NULL)
	return (ENOMEM);
	/* process about 1MB a time */
	while (off - start < 1024 1024) {
	uint64_t bytes_read = buflen;
	uint64_t leftover;

	if ((err = get_history(zhp, buf, off, &bytes_read)) != 0)
	break;

	/* if nothing else was read in, we're at EOF, just return */
	if (!bytes_read) {
	*eof = B_TRUE;
	break;
	}

	if ((err = zpool_history_unpack(buf, bytes_read,
	&leftover, &records, &numrecords)) != 0)
	break;
	*off -= leftover;
	if (leftover == bytes_read) {
	/*
	* no progress made, because buffer is not big enough
	* to hold this record; resize and retry.
	*/
	buflen *= 2;
	free(buf);
	buf = malloc(buflen);
	if (buf == NULL)
	return (ENOMEM);
	}
	}

	free(buf);

	if (!err) {
	verify(nvlist_alloc(nvhisp, NV_UNIQUE_NAME, 0) == 0);
	verify(nvlist_add_nvlist_array(*nvhisp, ZPOOL_HIST_RECORD,
	- records, numrecords) == 0);
	+ (const nvlist_t **)records, numrecords) == 0);
	}
	for (i = 0; i < numrecords; i++)
	nvlist_free(records[i]);
	free(records);

	return (err);
	}

	/*
	* Retrieve the next event given the passed 'zevent_fd' file descriptor.
	* If there is a new event available 'nvp' will contain a newly allocated
	* nvlist and 'dropped' will be set to the number of missed events since
	* the last call to this function. When 'nvp' is set to NULL it indicates
	* no new events are available. In either case the function returns 0 and
	* it is up to the caller to free 'nvp'. In the case of a fatal error the
	* function will return a non-zero value. When the function is called in
	* blocking mode (the default, unless the ZEVENT_NONBLOCK flag is passed),
	* it will not return until a new event is available.
	*/
	int
	zpool_events_next(libzfs_handle_t hdl, nvlist_t *nvp,
	int *dropped, unsigned flags, int zevent_fd)
	{
	zfs_cmd_t zc = {"\0"};
	int error = 0;

	*nvp = NULL;
	*dropped = 0;
	zc.zc_cleanup_fd = zevent_fd;

	if (flags & ZEVENT_NONBLOCK)
	zc.zc_guid = ZEVENT_NONBLOCK;

	if (zcmd_alloc_dst_nvlist(hdl, &zc, ZEVENT_SIZE) != 0)
	return (-1);

	retry:
	if (zfs_ioctl(hdl, ZFS_IOC_EVENTS_NEXT, &zc) != 0) {
	switch (errno) {
	case ESHUTDOWN:
	error = zfs_error_fmt(hdl, EZFS_POOLUNAVAIL,
	dgettext(TEXT_DOMAIN, "zfs shutdown"));
	goto out;
	case ENOENT:
	/* Blocking error case should not occur */
	if (!(flags & ZEVENT_NONBLOCK))
	error = zpool_standard_error_fmt(hdl, errno,
	dgettext(TEXT_DOMAIN, "cannot get event"));

	goto out;
	case ENOMEM:
	if (zcmd_expand_dst_nvlist(hdl, &zc) != 0) {
	error = zfs_error_fmt(hdl, EZFS_NOMEM,
	dgettext(TEXT_DOMAIN, "cannot get event"));
	goto out;
	} else {
	goto retry;
	}
	default:
	error = zpool_standard_error_fmt(hdl, errno,
	dgettext(TEXT_DOMAIN, "cannot get event"));
	goto out;
	}
	}

	error = zcmd_read_dst_nvlist(hdl, &zc, nvp);
	if (error != 0)
	goto out;

	*dropped = (int)zc.zc_cookie;
	out:
	zcmd_free_nvlists(&zc);

	return (error);
	}

	/*
	* Clear all events.
	*/
	int
	zpool_events_clear(libzfs_handle_t hdl, int count)
	{
	zfs_cmd_t zc = {"\0"};

	if (zfs_ioctl(hdl, ZFS_IOC_EVENTS_CLEAR, &zc) != 0)
	return (zpool_standard_error(hdl, errno,
	dgettext(TEXT_DOMAIN, "cannot clear events")));

	if (count != NULL)
	count = (int)zc.zc_cookie; / # of events cleared */

	return (0);
	}

	/*
	* Seek to a specific EID, ZEVENT_SEEK_START, or ZEVENT_SEEK_END for
	* the passed zevent_fd file handle. On success zero is returned,
	* otherwise -1 is returned and hdl->libzfs_error is set to the errno.
	*/
	int
	zpool_events_seek(libzfs_handle_t *hdl, uint64_t eid, int zevent_fd)
	{
	zfs_cmd_t zc = {"\0"};
	int error = 0;

	zc.zc_guid = eid;
	zc.zc_cleanup_fd = zevent_fd;

	if (zfs_ioctl(hdl, ZFS_IOC_EVENTS_SEEK, &zc) != 0) {
	switch (errno) {
	case ENOENT:
	error = zfs_error_fmt(hdl, EZFS_NOENT,
	dgettext(TEXT_DOMAIN, "cannot get event"));
	break;

	case ENOMEM:
	error = zfs_error_fmt(hdl, EZFS_NOMEM,
	dgettext(TEXT_DOMAIN, "cannot get event"));
	break;

	default:
	error = zpool_standard_error_fmt(hdl, errno,
	dgettext(TEXT_DOMAIN, "cannot get event"));
	break;
	}
	}

	return (error);
	}

	static void
	zpool_obj_to_path_impl(zpool_handle_t *zhp, uint64_t dsobj, uint64_t obj,
	char *pathname, size_t len, boolean_t always_unmounted)
	{
	zfs_cmd_t zc = {"\0"};
	boolean_t mounted = B_FALSE;
	char *mntpnt = NULL;
	char dsname[ZFS_MAX_DATASET_NAME_LEN];

	if (dsobj == 0) {
	/* special case for the MOS */
	(void) snprintf(pathname, len, "<metadata>:<0x%llx>",
	(longlong_t)obj);
	return;
	}

	/* get the dataset's name */
	(void) strlcpy(zc.zc_name, zhp->zpool_name, sizeof (zc.zc_name));
	zc.zc_obj = dsobj;
	if (zfs_ioctl(zhp->zpool_hdl,
	ZFS_IOC_DSOBJ_TO_DSNAME, &zc) != 0) {
	/* just write out a path of two object numbers */
	(void) snprintf(pathname, len, "<0x%llx>:<0x%llx>",
	(longlong_t)dsobj, (longlong_t)obj);
	return;
	}
	(void) strlcpy(dsname, zc.zc_value, sizeof (dsname));

	/* find out if the dataset is mounted */
	mounted = !always_unmounted && is_mounted(zhp->zpool_hdl, dsname,
	&mntpnt);

	/* get the corrupted object's path */
	(void) strlcpy(zc.zc_name, dsname, sizeof (zc.zc_name));
	zc.zc_obj = obj;
	if (zfs_ioctl(zhp->zpool_hdl, ZFS_IOC_OBJ_TO_PATH,
	&zc) == 0) {
	if (mounted) {
	(void) snprintf(pathname, len, "%s%s", mntpnt,
	zc.zc_value);
	} else {
	(void) snprintf(pathname, len, "%s:%s",
	dsname, zc.zc_value);
	}
	} else {
	(void) snprintf(pathname, len, "%s:<0x%llx>", dsname,
	(longlong_t)obj);
	}
	free(mntpnt);
	}

	void
	zpool_obj_to_path(zpool_handle_t *zhp, uint64_t dsobj, uint64_t obj,
	char *pathname, size_t len)
	{
	zpool_obj_to_path_impl(zhp, dsobj, obj, pathname, len, B_FALSE);
	}

	void
	zpool_obj_to_path_ds(zpool_handle_t *zhp, uint64_t dsobj, uint64_t obj,
	char *pathname, size_t len)
	{
	zpool_obj_to_path_impl(zhp, dsobj, obj, pathname, len, B_TRUE);
	}
	/*
	* Wait while the specified activity is in progress in the pool.
	*/
	int
	zpool_wait(zpool_handle_t *zhp, zpool_wait_activity_t activity)
	{
	boolean_t missing;

	int error = zpool_wait_status(zhp, activity, &missing, NULL);

	if (missing) {
	(void) zpool_standard_error_fmt(zhp->zpool_hdl, ENOENT,
	dgettext(TEXT_DOMAIN, "error waiting in pool '%s'"),
	zhp->zpool_name);
	return (ENOENT);
	} else {
	return (error);
	}
	}

	/*
	* Wait for the given activity and return the status of the wait (whether or not
	* any waiting was done) in the 'waited' parameter. Non-existent pools are
	* reported via the 'missing' parameter, rather than by printing an error
	* message. This is convenient when this function is called in a loop over a
	* long period of time (as it is, for example, by zpool's wait cmd). In that
	* scenario, a pool being exported or destroyed should be considered a normal
	* event, so we don't want to print an error when we find that the pool doesn't
	* exist.
	*/
	int
	zpool_wait_status(zpool_handle_t *zhp, zpool_wait_activity_t activity,
	boolean_t missing, boolean_t waited)
	{
	int error = lzc_wait(zhp->zpool_name, activity, waited);
	*missing = (error == ENOENT);
	if (*missing)
	return (0);

	if (error != 0) {
	(void) zpool_standard_error_fmt(zhp->zpool_hdl, error,
	dgettext(TEXT_DOMAIN, "error waiting in pool '%s'"),
	zhp->zpool_name);
	}

	return (error);
	}

	int
	zpool_set_bootenv(zpool_handle_t zhp, const nvlist_t envmap)
	{
	int error = lzc_set_bootenv(zhp->zpool_name, envmap);
	if (error != 0) {
	(void) zpool_standard_error_fmt(zhp->zpool_hdl, error,
	dgettext(TEXT_DOMAIN,
	"error setting bootenv in pool '%s'"), zhp->zpool_name);
	}

	return (error);
	}

	int
	zpool_get_bootenv(zpool_handle_t zhp, nvlist_t *nvlp)
	{
	nvlist_t *nvl;
	int error;

	nvl = NULL;
	error = lzc_get_bootenv(zhp->zpool_name, &nvl);
	if (error != 0) {
	(void) zpool_standard_error_fmt(zhp->zpool_hdl, error,
	dgettext(TEXT_DOMAIN,
	"error getting bootenv in pool '%s'"), zhp->zpool_name);
	} else {
	*nvlp = nvl;
	}

	return (error);
	}

	/*
	* Attempt to read and parse feature file(s) (from "compatibility" property).
	* Files contain zpool feature names, comma or whitespace-separated.
	* Comments (# character to next newline) are discarded.
	*
	* Arguments:
	* compatibility : string containing feature filenames
	* features : either NULL or pointer to array of boolean
	* report : either NULL or pointer to string buffer
	* rlen : length of "report" buffer
	*
	* compatibility is NULL (unset), "", "off", "legacy", or list of
	* comma-separated filenames. filenames should either be absolute,
	* or relative to:
	* 1) ZPOOL_SYSCONF_COMPAT_D (eg: /etc/zfs/compatibility.d) or
	* 2) ZPOOL_DATA_COMPAT_D (eg: /usr/share/zfs/compatibility.d).
	* (Unset), "" or "off" => enable all features
	* "legacy" => disable all features
	*
	* Any feature names read from files which match unames in spa_feature_table
	* will have the corresponding boolean set in the features array (if non-NULL).
	* If more than one feature set specified, only features present in all of
	* them will be set.
	*
	* "report" if not NULL will be populated with a suitable status message.
	*
	* Return values:
	* ZPOOL_COMPATIBILITY_OK : files read and parsed ok
	* ZPOOL_COMPATIBILITY_BADFILE : file too big or not a text file
	* ZPOOL_COMPATIBILITY_BADTOKEN : SYSCONF file contains invalid feature name
	* ZPOOL_COMPATIBILITY_WARNTOKEN : DATA file contains invalid feature name
	* ZPOOL_COMPATIBILITY_NOFILES : no feature files found
	*/
	zpool_compat_status_t
	zpool_load_compat(const char compat, boolean_t features, char *report,
	size_t rlen)
	{
	int sdirfd, ddirfd, featfd;
	struct stat fs;
	char *fc;
	char ps, ls, *ws;
	char file, line, *word;

	char l_compat[ZFS_MAXPROPLEN];

	boolean_t ret_nofiles = B_TRUE;
	boolean_t ret_badfile = B_FALSE;
	boolean_t ret_badtoken = B_FALSE;
	boolean_t ret_warntoken = B_FALSE;

	/* special cases (unset), "" and "off" => enable all features */
	if (compat == NULL \|\| compat[0] == '\0' \|\|
	strcmp(compat, ZPOOL_COMPAT_OFF) == 0) {
	if (features != NULL)
	for (uint_t i = 0; i < SPA_FEATURES; i++)
	features[i] = B_TRUE;
	if (report != NULL)
	strlcpy(report, gettext("all features enabled"), rlen);
	return (ZPOOL_COMPATIBILITY_OK);
	}

	/* Final special case "legacy" => disable all features */
	if (strcmp(compat, ZPOOL_COMPAT_LEGACY) == 0) {
	if (features != NULL)
	for (uint_t i = 0; i < SPA_FEATURES; i++)
	features[i] = B_FALSE;
	if (report != NULL)
	strlcpy(report, gettext("all features disabled"), rlen);
	return (ZPOOL_COMPATIBILITY_OK);
	}

	/*
	* Start with all true; will be ANDed with results from each file
	*/
	if (features != NULL)
	for (uint_t i = 0; i < SPA_FEATURES; i++)
	features[i] = B_TRUE;

	char err_badfile[1024] = "";
	char err_badtoken[1024] = "";

	/*
	* We ignore errors from the directory open()
	* as they're only needed if the filename is relative
	* which will be checked during the openat().
	*/

	/* O_PATH safer than O_RDONLY if system allows it */
	#if defined(O_PATH)
	#define ZC_DIR_FLAGS (O_DIRECTORY \| O_CLOEXEC \| O_PATH)
	#else
	#define ZC_DIR_FLAGS (O_DIRECTORY \| O_CLOEXEC \| O_RDONLY)
	#endif

	sdirfd = open(ZPOOL_SYSCONF_COMPAT_D, ZC_DIR_FLAGS);
	ddirfd = open(ZPOOL_DATA_COMPAT_D, ZC_DIR_FLAGS);

	(void) strlcpy(l_compat, compat, ZFS_MAXPROPLEN);

	for (file = strtok_r(l_compat, ",", &ps);
	file != NULL;
	file = strtok_r(NULL, ",", &ps)) {

	boolean_t l_features[SPA_FEATURES];

	enum { Z_SYSCONF, Z_DATA } source;

	/* try sysconfdir first, then datadir */
	source = Z_SYSCONF;
	if ((featfd = openat(sdirfd, file, O_RDONLY \| O_CLOEXEC)) < 0) {
	featfd = openat(ddirfd, file, O_RDONLY \| O_CLOEXEC);
	source = Z_DATA;
	}

	/* File readable and correct size? */
	if (featfd < 0 \|\|
	fstat(featfd, &fs) < 0 \|\|
	fs.st_size < 1 \|\|
	fs.st_size > ZPOOL_COMPAT_MAXSIZE) {
	(void) close(featfd);
	strlcat(err_badfile, file, ZFS_MAXPROPLEN);
	strlcat(err_badfile, " ", ZFS_MAXPROPLEN);
	ret_badfile = B_TRUE;
	continue;
	}

	/* Prefault the file if system allows */
	#if defined(MAP_POPULATE)
	#define ZC_MMAP_FLAGS (MAP_PRIVATE \| MAP_POPULATE)
	#elif defined(MAP_PREFAULT_READ)
	#define ZC_MMAP_FLAGS (MAP_PRIVATE \| MAP_PREFAULT_READ)
	#else
	#define ZC_MMAP_FLAGS (MAP_PRIVATE)
	#endif

	/* private mmap() so we can strtok safely */
	fc = (char *)mmap(NULL, fs.st_size, PROT_READ \| PROT_WRITE,
	ZC_MMAP_FLAGS, featfd, 0);
	(void) close(featfd);

	/* map ok, and last character == newline? */
	if (fc == MAP_FAILED \|\| fc[fs.st_size - 1] != '\n') {
	(void) munmap((void *) fc, fs.st_size);
	strlcat(err_badfile, file, ZFS_MAXPROPLEN);
	strlcat(err_badfile, " ", ZFS_MAXPROPLEN);
	ret_badfile = B_TRUE;
	continue;
	}

	ret_nofiles = B_FALSE;

	for (uint_t i = 0; i < SPA_FEATURES; i++)
	l_features[i] = B_FALSE;

	/* replace final newline with NULL to ensure string ends */
	fc[fs.st_size - 1] = '\0';

	for (line = strtok_r(fc, "\n", &ls);
	line != NULL;
	line = strtok_r(NULL, "\n", &ls)) {
	/* discard comments */
	char *r = strchr(line, '#');
	if (r != NULL)
	*r = '\0';

	for (word = strtok_r(line, ", \t", &ws);
	word != NULL;
	word = strtok_r(NULL, ", \t", &ws)) {
	/* Find matching feature name */
	uint_t f;
	for (f = 0; f < SPA_FEATURES; f++) {
	zfeature_info_t *fi =
	&spa_feature_table[f];
	if (strcmp(word, fi->fi_uname) == 0) {
	l_features[f] = B_TRUE;
	break;
	}
	}
	if (f < SPA_FEATURES)
	continue;

	/* found an unrecognized word */
	/* lightly sanitize it */
	if (strlen(word) > 32)
	word[32] = '\0';
	for (char c = word; c != '\0'; c++)
	if (!isprint(*c))
	*c = '?';

	strlcat(err_badtoken, word, ZFS_MAXPROPLEN);
	strlcat(err_badtoken, " ", ZFS_MAXPROPLEN);
	if (source == Z_SYSCONF)
	ret_badtoken = B_TRUE;
	else
	ret_warntoken = B_TRUE;
	}
	}
	(void) munmap((void *) fc, fs.st_size);

	if (features != NULL)
	for (uint_t i = 0; i < SPA_FEATURES; i++)
	features[i] &= l_features[i];
	}
	(void) close(sdirfd);
	(void) close(ddirfd);

	/* Return the most serious error */
	if (ret_badfile) {
	if (report != NULL)
	snprintf(report, rlen, gettext("could not read/"
	"parse feature file(s): %s"), err_badfile);
	return (ZPOOL_COMPATIBILITY_BADFILE);
	}
	if (ret_nofiles) {
	if (report != NULL)
	strlcpy(report,
	gettext("no valid compatibility files specified"),
	rlen);
	return (ZPOOL_COMPATIBILITY_NOFILES);
	}
	if (ret_badtoken) {
	if (report != NULL)
	snprintf(report, rlen, gettext("invalid feature "
	"name(s) in local compatibility files: %s"),
	err_badtoken);
	return (ZPOOL_COMPATIBILITY_BADTOKEN);
	}
	if (ret_warntoken) {
	if (report != NULL)
	snprintf(report, rlen, gettext("unrecognized feature "
	"name(s) in distribution compatibility files: %s"),
	err_badtoken);
	return (ZPOOL_COMPATIBILITY_WARNTOKEN);
	}
	if (report != NULL)
	strlcpy(report, gettext("compatibility set ok"), rlen);
	return (ZPOOL_COMPATIBILITY_OK);
	}
	+
	+static int
	+zpool_vdev_guid(zpool_handle_t zhp, const char vdevname, uint64_t *vdev_guid)
	+{
	+ nvlist_t *tgt;
	+ boolean_t avail_spare, l2cache;
	+
	+ verify(zhp != NULL);
	+ if (zpool_get_state(zhp) == POOL_STATE_UNAVAIL) {
	+ char errbuf[1024];
	+ (void) snprintf(errbuf, sizeof (errbuf),
	+ dgettext(TEXT_DOMAIN, "pool is in an unavailable state"));
	+ return (zfs_error(zhp->zpool_hdl, EZFS_POOLUNAVAIL, errbuf));
	+ }
	+
	+ if ((tgt = zpool_find_vdev(zhp, vdevname, &avail_spare, &l2cache,
	+ NULL)) == NULL) {
	+ char errbuf[1024];
	+ (void) snprintf(errbuf, sizeof (errbuf),
	+ dgettext(TEXT_DOMAIN, "can not find %s in %s"),
	+ vdevname, zhp->zpool_name);
	+ return (zfs_error(zhp->zpool_hdl, EZFS_NODEVICE, errbuf));
	+ }
	+
	+ verify(nvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID, vdev_guid) == 0);
	+ return (0);
	+}
	+
	+/*
	+ * Get a vdev property value for 'prop' and return the value in
	+ * a pre-allocated buffer.
	+ */
	+int
	+zpool_get_vdev_prop_value(nvlist_t nvprop, vdev_prop_t prop, char prop_name,
	+ char buf, size_t len, zprop_source_t srctype, boolean_t literal)
	+{
	+ nvlist_t *nv;
	+ uint64_t intval;
	+ char *strval;
	+ zprop_source_t src = ZPROP_SRC_NONE;
	+
	+ if (prop == VDEV_PROP_USER) {
	+ /* user property, prop_name must contain the property name */
	+ assert(prop_name != NULL);
	+ if (nvlist_lookup_nvlist(nvprop, prop_name, &nv) == 0) {
	+ verify(nvlist_lookup_uint64(nv, ZPROP_SOURCE,
	+ &intval) == 0);
	+ src = intval;
	+ verify(nvlist_lookup_string(nv, ZPROP_VALUE,
	+ &strval) == 0);
	+ } else {
	+ /* user prop not found */
	+ return (-1);
	+ }
	+ (void) strlcpy(buf, strval, len);
	+ if (srctype)
	+ *srctype = src;
	+ return (0);
	+ }
	+
	+ if (prop_name == NULL)
	+ prop_name = (char *)vdev_prop_to_name(prop);
	+
	+ switch (vdev_prop_get_type(prop)) {
	+ case PROP_TYPE_STRING:
	+ if (nvlist_lookup_nvlist(nvprop, prop_name, &nv) == 0) {
	+ verify(nvlist_lookup_uint64(nv, ZPROP_SOURCE,
	+ &intval) == 0);
	+ src = intval;
	+ verify(nvlist_lookup_string(nv, ZPROP_VALUE,
	+ &strval) == 0);
	+ } else {
	+ src = ZPROP_SRC_DEFAULT;
	+ if ((strval = (char *)vdev_prop_default_string(prop))
	+ == NULL)
	+ strval = "-";
	+ }
	+ (void) strlcpy(buf, strval, len);
	+ break;
	+
	+ case PROP_TYPE_NUMBER:
	+ if (nvlist_lookup_nvlist(nvprop, prop_name, &nv) == 0) {
	+ verify(nvlist_lookup_uint64(nv, ZPROP_SOURCE,
	+ &intval) == 0);
	+ src = intval;
	+ verify(nvlist_lookup_uint64(nv, ZPROP_VALUE,
	+ &intval) == 0);
	+ } else {
	+ src = ZPROP_SRC_DEFAULT;
	+ intval = vdev_prop_default_numeric(prop);
	+ }
	+
	+ switch (prop) {
	+ case VDEV_PROP_ASIZE:
	+ case VDEV_PROP_PSIZE:
	+ case VDEV_PROP_SIZE:
	+ case VDEV_PROP_ALLOCATED:
	+ case VDEV_PROP_FREE:
	+ case VDEV_PROP_READ_ERRORS:
	+ case VDEV_PROP_WRITE_ERRORS:
	+ case VDEV_PROP_CHECKSUM_ERRORS:
	+ case VDEV_PROP_INITIALIZE_ERRORS:
	+ case VDEV_PROP_OPS_NULL:
	+ case VDEV_PROP_OPS_READ:
	+ case VDEV_PROP_OPS_WRITE:
	+ case VDEV_PROP_OPS_FREE:
	+ case VDEV_PROP_OPS_CLAIM:
	+ case VDEV_PROP_OPS_TRIM:
	+ case VDEV_PROP_BYTES_NULL:
	+ case VDEV_PROP_BYTES_READ:
	+ case VDEV_PROP_BYTES_WRITE:
	+ case VDEV_PROP_BYTES_FREE:
	+ case VDEV_PROP_BYTES_CLAIM:
	+ case VDEV_PROP_BYTES_TRIM:
	+ if (literal) {
	+ (void) snprintf(buf, len, "%llu",
	+ (u_longlong_t)intval);
	+ } else {
	+ (void) zfs_nicenum(intval, buf, len);
	+ }
	+ break;
	+ case VDEV_PROP_EXPANDSZ:
	+ if (intval == 0) {
	+ (void) strlcpy(buf, "-", len);
	+ } else if (literal) {
	+ (void) snprintf(buf, len, "%llu",
	+ (u_longlong_t)intval);
	+ } else {
	+ (void) zfs_nicenum(intval, buf, len);
	+ }
	+ break;
	+ case VDEV_PROP_CAPACITY:
	+ if (literal) {
	+ (void) snprintf(buf, len, "%llu",
	+ (u_longlong_t)intval);
	+ } else {
	+ (void) snprintf(buf, len, "%llu%%",
	+ (u_longlong_t)intval);
	+ }
	+ break;
	+ case VDEV_PROP_FRAGMENTATION:
	+ if (intval == UINT64_MAX) {
	+ (void) strlcpy(buf, "-", len);
	+ } else {
	+ (void) snprintf(buf, len, "%llu%%",
	+ (u_longlong_t)intval);
	+ }
	+ break;
	+ case VDEV_PROP_STATE:
	+ if (literal) {
	+ (void) snprintf(buf, len, "%llu",
	+ (u_longlong_t)intval);
	+ } else {
	+ (void) strlcpy(buf, zpool_state_to_name(intval,
	+ VDEV_AUX_NONE), len);
	+ }
	+ break;
	+ default:
	+ (void) snprintf(buf, len, "%llu",
	+ (u_longlong_t)intval);
	+ }
	+ break;
	+
	+ case PROP_TYPE_INDEX:
	+ if (nvlist_lookup_nvlist(nvprop, prop_name, &nv) == 0) {
	+ verify(nvlist_lookup_uint64(nv, ZPROP_SOURCE,
	+ &intval) == 0);
	+ src = intval;
	+ verify(nvlist_lookup_uint64(nv, ZPROP_VALUE,
	+ &intval) == 0);
	+ } else {
	+ src = ZPROP_SRC_DEFAULT;
	+ intval = vdev_prop_default_numeric(prop);
	+ }
	+ if (vdev_prop_index_to_string(prop, intval,
	+ (const char **)&strval) != 0)
	+ return (-1);
	+ (void) strlcpy(buf, strval, len);
	+ break;
	+
	+ default:
	+ abort();
	+ }
	+
	+ if (srctype)
	+ *srctype = src;
	+
	+ return (0);
	+}
	+
	+/*
	+ * Get a vdev property value for 'prop_name' and return the value in
	+ * a pre-allocated buffer.
	+ */
	+int
	+zpool_get_vdev_prop(zpool_handle_t zhp, const char vdevname, vdev_prop_t prop,
	+ char prop_name, char buf, size_t len, zprop_source_t *srctype,
	+ boolean_t literal)
	+{
	+ nvlist_t reqnvl, reqprops;
	+ nvlist_t *retprops = NULL;
	+ uint64_t vdev_guid;
	+ int ret;
	+
	+ if ((ret = zpool_vdev_guid(zhp, vdevname, &vdev_guid)) != 0)
	+ return (ret);
	+
	+ if (nvlist_alloc(&reqnvl, NV_UNIQUE_NAME, 0) != 0)
	+ return (no_memory(zhp->zpool_hdl));
	+ if (nvlist_alloc(&reqprops, NV_UNIQUE_NAME, 0) != 0)
	+ return (no_memory(zhp->zpool_hdl));
	+
	+ fnvlist_add_uint64(reqnvl, ZPOOL_VDEV_PROPS_GET_VDEV, vdev_guid);
	+
	+ if (prop != VDEV_PROP_USER) {
	+ /* prop_name overrides prop value */
	+ if (prop_name != NULL)
	+ prop = vdev_name_to_prop(prop_name);
	+ else
	+ prop_name = (char *)vdev_prop_to_name(prop);
	+ assert(prop < VDEV_NUM_PROPS);
	+ }
	+
	+ assert(prop_name != NULL);
	+ if (nvlist_add_uint64(reqprops, prop_name, prop) != 0) {
	+ nvlist_free(reqnvl);
	+ nvlist_free(reqprops);
	+ return (no_memory(zhp->zpool_hdl));
	+ }
	+
	+ fnvlist_add_nvlist(reqnvl, ZPOOL_VDEV_PROPS_GET_PROPS, reqprops);
	+
	+ ret = lzc_get_vdev_prop(zhp->zpool_name, reqnvl, &retprops);
	+
	+ if (ret == 0) {
	+ ret = zpool_get_vdev_prop_value(retprops, prop, prop_name, buf,
	+ len, srctype, literal);
	+ } else {
	+ char errbuf[1024];
	+ (void) snprintf(errbuf, sizeof (errbuf),
	+ dgettext(TEXT_DOMAIN, "cannot get vdev property %s from"
	+ " %s in %s"), prop_name, vdevname, zhp->zpool_name);
	+ (void) zpool_standard_error(zhp->zpool_hdl, ret, errbuf);
	+ }
	+
	+ nvlist_free(reqnvl);
	+ nvlist_free(reqprops);
	+ nvlist_free(retprops);
	+
	+ return (ret);
	+}
	+
	+/*
	+ * Get all vdev properties
	+ */
	+int
	+zpool_get_all_vdev_props(zpool_handle_t zhp, const char vdevname,
	+ nvlist_t **outnvl)
	+{
	+ nvlist_t *nvl = NULL;
	+ uint64_t vdev_guid;
	+ int ret;
	+
	+ if ((ret = zpool_vdev_guid(zhp, vdevname, &vdev_guid)) != 0)
	+ return (ret);
	+
	+ if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0)
	+ return (no_memory(zhp->zpool_hdl));
	+
	+ fnvlist_add_uint64(nvl, ZPOOL_VDEV_PROPS_GET_VDEV, vdev_guid);
	+
	+ ret = lzc_get_vdev_prop(zhp->zpool_name, nvl, outnvl);
	+
	+ nvlist_free(nvl);
	+
	+ if (ret) {
	+ char errbuf[1024];
	+ (void) snprintf(errbuf, sizeof (errbuf),
	+ dgettext(TEXT_DOMAIN, "cannot get vdev properties for"
	+ " %s in %s"), vdevname, zhp->zpool_name);
	+ (void) zpool_standard_error(zhp->zpool_hdl, errno, errbuf);
	+ }
	+
	+ return (ret);
	+}
	+
	+/*
	+ * Set vdev property
	+ */
	+int
	+zpool_set_vdev_prop(zpool_handle_t zhp, const char vdevname,
	+ const char propname, const char propval)
	+{
	+ int ret;
	+ nvlist_t *nvl = NULL;
	+ nvlist_t *outnvl = NULL;
	+ nvlist_t *props;
	+ nvlist_t *realprops;
	+ prop_flags_t flags = { 0 };
	+ uint64_t version;
	+ uint64_t vdev_guid;
	+
	+ if ((ret = zpool_vdev_guid(zhp, vdevname, &vdev_guid)) != 0)
	+ return (ret);
	+
	+ if (nvlist_alloc(&nvl, NV_UNIQUE_NAME, 0) != 0)
	+ return (no_memory(zhp->zpool_hdl));
	+ if (nvlist_alloc(&props, NV_UNIQUE_NAME, 0) != 0)
	+ return (no_memory(zhp->zpool_hdl));
	+
	+ fnvlist_add_uint64(nvl, ZPOOL_VDEV_PROPS_SET_VDEV, vdev_guid);
	+
	+ if (nvlist_add_string(props, propname, propval) != 0) {
	+ nvlist_free(props);
	+ return (no_memory(zhp->zpool_hdl));
	+ }
	+
	+ char errbuf[1024];
	+ (void) snprintf(errbuf, sizeof (errbuf),
	+ dgettext(TEXT_DOMAIN, "cannot set property %s for %s on %s"),
	+ propname, vdevname, zhp->zpool_name);
	+
	+ flags.vdevprop = 1;
	+ version = zpool_get_prop_int(zhp, ZPOOL_PROP_VERSION, NULL);
	+ if ((realprops = zpool_valid_proplist(zhp->zpool_hdl,
	+ zhp->zpool_name, props, version, flags, errbuf)) == NULL) {
	+ nvlist_free(props);
	+ nvlist_free(nvl);
	+ return (-1);
	+ }
	+
	+ nvlist_free(props);
	+ props = realprops;
	+
	+ fnvlist_add_nvlist(nvl, ZPOOL_VDEV_PROPS_SET_PROPS, props);
	+
	+ ret = lzc_set_vdev_prop(zhp->zpool_name, nvl, &outnvl);
	+
	+ nvlist_free(props);
	+ nvlist_free(nvl);
	+ nvlist_free(outnvl);
	+
	+ if (ret)
	+ (void) zpool_standard_error(zhp->zpool_hdl, errno, errbuf);
	+
	+ return (ret);
	+}
	diff --git a/sys/contrib/openzfs/lib/libzfs/libzfs_util.c b/sys/contrib/openzfs/lib/libzfs/libzfs_util.c
	index c3c009ae3a10..b3f39afe2182 100644
	--- a/sys/contrib/openzfs/lib/libzfs/libzfs_util.c
	+++ b/sys/contrib/openzfs/lib/libzfs/libzfs_util.c
	@@ -1,2083 +1,2099 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright 2020 Joyent, Inc. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>
	* Copyright (c) 2017 Datto Inc.
	* Copyright (c) 2020 The FreeBSD Foundation
	*
	* Portions of this software were developed by Allan Jude
	* under sponsorship from the FreeBSD Foundation.
	*/

	/*
	* Internal utility routines for the ZFS library.
	*/

	#include <errno.h>
	#include <fcntl.h>
	#include <libintl.h>
	#include <stdarg.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <strings.h>
	#include <unistd.h>
	#include <math.h>
	#if LIBFETCH_DYNAMIC
	#include <dlfcn.h>
	#endif
	#include <sys/stat.h>
	#include <sys/mnttab.h>
	#include <sys/mntent.h>
	#include <sys/types.h>
	#include <sys/wait.h>

	#include <libzfs.h>
	#include <libzfs_core.h>

	#include "libzfs_impl.h"
	#include "zfs_prop.h"
	#include "zfeature_common.h"
	#include <zfs_fletcher.h>
	#include <libzutil.h>

	/*
	* We only care about the scheme in order to match the scheme
	* with the handler. Each handler should validate the full URI
	* as necessary.
	*/
	#define URI_REGEX "^\$[A-Za-z][A-Za-z0-9+.\\-]*\$:"

	int
	libzfs_errno(libzfs_handle_t *hdl)
	{
	return (hdl->libzfs_error);
	}

	const char *
	libzfs_error_action(libzfs_handle_t *hdl)
	{
	return (hdl->libzfs_action);
	}

	const char *
	libzfs_error_description(libzfs_handle_t *hdl)
	{
	if (hdl->libzfs_desc[0] != '\0')
	return (hdl->libzfs_desc);

	switch (hdl->libzfs_error) {
	case EZFS_NOMEM:
	return (dgettext(TEXT_DOMAIN, "out of memory"));
	case EZFS_BADPROP:
	return (dgettext(TEXT_DOMAIN, "invalid property value"));
	case EZFS_PROPREADONLY:
	return (dgettext(TEXT_DOMAIN, "read-only property"));
	case EZFS_PROPTYPE:
	return (dgettext(TEXT_DOMAIN, "property doesn't apply to "
	"datasets of this type"));
	case EZFS_PROPNONINHERIT:
	return (dgettext(TEXT_DOMAIN, "property cannot be inherited"));
	case EZFS_PROPSPACE:
	return (dgettext(TEXT_DOMAIN, "invalid quota or reservation"));
	case EZFS_BADTYPE:
	return (dgettext(TEXT_DOMAIN, "operation not applicable to "
	"datasets of this type"));
	case EZFS_BUSY:
	return (dgettext(TEXT_DOMAIN, "pool or dataset is busy"));
	case EZFS_EXISTS:
	return (dgettext(TEXT_DOMAIN, "pool or dataset exists"));
	case EZFS_NOENT:
	return (dgettext(TEXT_DOMAIN, "no such pool or dataset"));
	case EZFS_BADSTREAM:
	return (dgettext(TEXT_DOMAIN, "invalid backup stream"));
	case EZFS_DSREADONLY:
	return (dgettext(TEXT_DOMAIN, "dataset is read-only"));
	case EZFS_VOLTOOBIG:
	return (dgettext(TEXT_DOMAIN, "volume size exceeds limit for "
	"this system"));
	case EZFS_INVALIDNAME:
	return (dgettext(TEXT_DOMAIN, "invalid name"));
	case EZFS_BADRESTORE:
	return (dgettext(TEXT_DOMAIN, "unable to restore to "
	"destination"));
	case EZFS_BADBACKUP:
	return (dgettext(TEXT_DOMAIN, "backup failed"));
	case EZFS_BADTARGET:
	return (dgettext(TEXT_DOMAIN, "invalid target vdev"));
	case EZFS_NODEVICE:
	return (dgettext(TEXT_DOMAIN, "no such device in pool"));
	case EZFS_BADDEV:
	return (dgettext(TEXT_DOMAIN, "invalid device"));
	case EZFS_NOREPLICAS:
	return (dgettext(TEXT_DOMAIN, "no valid replicas"));
	case EZFS_RESILVERING:
	return (dgettext(TEXT_DOMAIN, "currently resilvering"));
	case EZFS_BADVERSION:
	return (dgettext(TEXT_DOMAIN, "unsupported version or "
	"feature"));
	case EZFS_POOLUNAVAIL:
	return (dgettext(TEXT_DOMAIN, "pool is unavailable"));
	case EZFS_DEVOVERFLOW:
	return (dgettext(TEXT_DOMAIN, "too many devices in one vdev"));
	case EZFS_BADPATH:
	return (dgettext(TEXT_DOMAIN, "must be an absolute path"));
	case EZFS_CROSSTARGET:
	return (dgettext(TEXT_DOMAIN, "operation crosses datasets or "
	"pools"));
	case EZFS_ZONED:
	return (dgettext(TEXT_DOMAIN, "dataset in use by local zone"));
	case EZFS_MOUNTFAILED:
	return (dgettext(TEXT_DOMAIN, "mount failed"));
	case EZFS_UMOUNTFAILED:
	return (dgettext(TEXT_DOMAIN, "unmount failed"));
	case EZFS_UNSHARENFSFAILED:
	return (dgettext(TEXT_DOMAIN, "NFS share removal failed"));
	case EZFS_SHARENFSFAILED:
	return (dgettext(TEXT_DOMAIN, "NFS share creation failed"));
	case EZFS_UNSHARESMBFAILED:
	return (dgettext(TEXT_DOMAIN, "SMB share removal failed"));
	case EZFS_SHARESMBFAILED:
	return (dgettext(TEXT_DOMAIN, "SMB share creation failed"));
	case EZFS_PERM:
	return (dgettext(TEXT_DOMAIN, "permission denied"));
	case EZFS_NOSPC:
	return (dgettext(TEXT_DOMAIN, "out of space"));
	case EZFS_FAULT:
	return (dgettext(TEXT_DOMAIN, "bad address"));
	case EZFS_IO:
	return (dgettext(TEXT_DOMAIN, "I/O error"));
	case EZFS_INTR:
	return (dgettext(TEXT_DOMAIN, "signal received"));
	case EZFS_ISSPARE:
	return (dgettext(TEXT_DOMAIN, "device is reserved as a hot "
	"spare"));
	case EZFS_INVALCONFIG:
	return (dgettext(TEXT_DOMAIN, "invalid vdev configuration"));
	case EZFS_RECURSIVE:
	return (dgettext(TEXT_DOMAIN, "recursive dataset dependency"));
	case EZFS_NOHISTORY:
	return (dgettext(TEXT_DOMAIN, "no history available"));
	case EZFS_POOLPROPS:
	return (dgettext(TEXT_DOMAIN, "failed to retrieve "
	"pool properties"));
	case EZFS_POOL_NOTSUP:
	return (dgettext(TEXT_DOMAIN, "operation not supported "
	"on this type of pool"));
	case EZFS_POOL_INVALARG:
	return (dgettext(TEXT_DOMAIN, "invalid argument for "
	"this pool operation"));
	case EZFS_NAMETOOLONG:
	return (dgettext(TEXT_DOMAIN, "dataset name is too long"));
	case EZFS_OPENFAILED:
	return (dgettext(TEXT_DOMAIN, "open failed"));
	case EZFS_NOCAP:
	return (dgettext(TEXT_DOMAIN,
	"disk capacity information could not be retrieved"));
	case EZFS_LABELFAILED:
	return (dgettext(TEXT_DOMAIN, "write of label failed"));
	case EZFS_BADWHO:
	return (dgettext(TEXT_DOMAIN, "invalid user/group"));
	case EZFS_BADPERM:
	return (dgettext(TEXT_DOMAIN, "invalid permission"));
	case EZFS_BADPERMSET:
	return (dgettext(TEXT_DOMAIN, "invalid permission set name"));
	case EZFS_NODELEGATION:
	return (dgettext(TEXT_DOMAIN, "delegated administration is "
	"disabled on pool"));
	case EZFS_BADCACHE:
	return (dgettext(TEXT_DOMAIN, "invalid or missing cache file"));
	case EZFS_ISL2CACHE:
	return (dgettext(TEXT_DOMAIN, "device is in use as a cache"));
	case EZFS_VDEVNOTSUP:
	return (dgettext(TEXT_DOMAIN, "vdev specification is not "
	"supported"));
	case EZFS_NOTSUP:
	return (dgettext(TEXT_DOMAIN, "operation not supported "
	"on this dataset"));
	case EZFS_IOC_NOTSUPPORTED:
	return (dgettext(TEXT_DOMAIN, "operation not supported by "
	"zfs kernel module"));
	case EZFS_ACTIVE_SPARE:
	return (dgettext(TEXT_DOMAIN, "pool has active shared spare "
	"device"));
	case EZFS_UNPLAYED_LOGS:
	return (dgettext(TEXT_DOMAIN, "log device has unplayed intent "
	"logs"));
	case EZFS_REFTAG_RELE:
	return (dgettext(TEXT_DOMAIN, "no such tag on this dataset"));
	case EZFS_REFTAG_HOLD:
	return (dgettext(TEXT_DOMAIN, "tag already exists on this "
	"dataset"));
	case EZFS_TAGTOOLONG:
	return (dgettext(TEXT_DOMAIN, "tag too long"));
	case EZFS_PIPEFAILED:
	return (dgettext(TEXT_DOMAIN, "pipe create failed"));
	case EZFS_THREADCREATEFAILED:
	return (dgettext(TEXT_DOMAIN, "thread create failed"));
	case EZFS_POSTSPLIT_ONLINE:
	return (dgettext(TEXT_DOMAIN, "disk was split from this pool "
	"into a new one"));
	case EZFS_SCRUB_PAUSED:
	return (dgettext(TEXT_DOMAIN, "scrub is paused; "
	"use 'zpool scrub' to resume"));
	case EZFS_SCRUBBING:
	return (dgettext(TEXT_DOMAIN, "currently scrubbing; "
	"use 'zpool scrub -s' to cancel current scrub"));
	case EZFS_NO_SCRUB:
	return (dgettext(TEXT_DOMAIN, "there is no active scrub"));
	case EZFS_DIFF:
	return (dgettext(TEXT_DOMAIN, "unable to generate diffs"));
	case EZFS_DIFFDATA:
	return (dgettext(TEXT_DOMAIN, "invalid diff data"));
	case EZFS_POOLREADONLY:
	return (dgettext(TEXT_DOMAIN, "pool is read-only"));
	case EZFS_NO_PENDING:
	return (dgettext(TEXT_DOMAIN, "operation is not "
	"in progress"));
	case EZFS_CHECKPOINT_EXISTS:
	return (dgettext(TEXT_DOMAIN, "checkpoint exists"));
	case EZFS_DISCARDING_CHECKPOINT:
	return (dgettext(TEXT_DOMAIN, "currently discarding "
	"checkpoint"));
	case EZFS_NO_CHECKPOINT:
	return (dgettext(TEXT_DOMAIN, "checkpoint does not exist"));
	case EZFS_DEVRM_IN_PROGRESS:
	return (dgettext(TEXT_DOMAIN, "device removal in progress"));
	case EZFS_VDEV_TOO_BIG:
	return (dgettext(TEXT_DOMAIN, "device exceeds supported size"));
	case EZFS_ACTIVE_POOL:
	return (dgettext(TEXT_DOMAIN, "pool is imported on a "
	"different host"));
	case EZFS_CRYPTOFAILED:
	return (dgettext(TEXT_DOMAIN, "encryption failure"));
	case EZFS_TOOMANY:
	return (dgettext(TEXT_DOMAIN, "argument list too long"));
	case EZFS_INITIALIZING:
	return (dgettext(TEXT_DOMAIN, "currently initializing"));
	case EZFS_NO_INITIALIZE:
	return (dgettext(TEXT_DOMAIN, "there is no active "
	"initialization"));
	case EZFS_WRONG_PARENT:
	return (dgettext(TEXT_DOMAIN, "invalid parent dataset"));
	case EZFS_TRIMMING:
	return (dgettext(TEXT_DOMAIN, "currently trimming"));
	case EZFS_NO_TRIM:
	return (dgettext(TEXT_DOMAIN, "there is no active trim"));
	case EZFS_TRIM_NOTSUP:
	return (dgettext(TEXT_DOMAIN, "trim operations are not "
	"supported by this device"));
	case EZFS_NO_RESILVER_DEFER:
	return (dgettext(TEXT_DOMAIN, "this action requires the "
	"resilver_defer feature"));
	case EZFS_EXPORT_IN_PROGRESS:
	return (dgettext(TEXT_DOMAIN, "pool export in progress"));
	case EZFS_REBUILDING:
	return (dgettext(TEXT_DOMAIN, "currently sequentially "
	"resilvering"));
	+ case EZFS_VDEV_NOTSUP:
	+ return (dgettext(TEXT_DOMAIN, "operation not supported "
	+ "on this type of vdev"));
	case EZFS_UNKNOWN:
	return (dgettext(TEXT_DOMAIN, "unknown error"));
	default:
	assert(hdl->libzfs_error == 0);
	return (dgettext(TEXT_DOMAIN, "no error"));
	}
	}

	void
	zfs_error_aux(libzfs_handle_t hdl, const char fmt, ...)
	{
	va_list ap;

	va_start(ap, fmt);

	(void) vsnprintf(hdl->libzfs_desc, sizeof (hdl->libzfs_desc),
	fmt, ap);
	hdl->libzfs_desc_active = 1;

	va_end(ap);
	}

	static void
	zfs_verror(libzfs_handle_t hdl, int error, const char fmt, va_list ap)
	{
	(void) vsnprintf(hdl->libzfs_action, sizeof (hdl->libzfs_action),
	fmt, ap);
	hdl->libzfs_error = error;

	if (hdl->libzfs_desc_active)
	hdl->libzfs_desc_active = 0;
	else
	hdl->libzfs_desc[0] = '\0';

	if (hdl->libzfs_printerr) {
	if (error == EZFS_UNKNOWN) {
	(void) fprintf(stderr, dgettext(TEXT_DOMAIN, "internal "
	"error: %s: %s\n"), hdl->libzfs_action,
	libzfs_error_description(hdl));
	abort();
	}

	(void) fprintf(stderr, "%s: %s\n", hdl->libzfs_action,
	libzfs_error_description(hdl));
	if (error == EZFS_NOMEM)
	exit(1);
	}
	}

	int
	zfs_error(libzfs_handle_t hdl, int error, const char msg)
	{
	return (zfs_error_fmt(hdl, error, "%s", msg));
	}

	int
	zfs_error_fmt(libzfs_handle_t hdl, int error, const char fmt, ...)
	{
	va_list ap;

	va_start(ap, fmt);

	zfs_verror(hdl, error, fmt, ap);

	va_end(ap);

	return (-1);
	}

	static int
	zfs_common_error(libzfs_handle_t hdl, int error, const char fmt,
	va_list ap)
	{
	switch (error) {
	case EPERM:
	case EACCES:
	zfs_verror(hdl, EZFS_PERM, fmt, ap);
	return (-1);

	case ECANCELED:
	zfs_verror(hdl, EZFS_NODELEGATION, fmt, ap);
	return (-1);

	case EIO:
	zfs_verror(hdl, EZFS_IO, fmt, ap);
	return (-1);

	case EFAULT:
	zfs_verror(hdl, EZFS_FAULT, fmt, ap);
	return (-1);

	case EINTR:
	zfs_verror(hdl, EZFS_INTR, fmt, ap);
	return (-1);
	}

	return (0);
	}

	int
	zfs_standard_error(libzfs_handle_t hdl, int error, const char msg)
	{
	return (zfs_standard_error_fmt(hdl, error, "%s", msg));
	}

	int
	zfs_standard_error_fmt(libzfs_handle_t hdl, int error, const char fmt, ...)
	{
	va_list ap;

	va_start(ap, fmt);

	if (zfs_common_error(hdl, error, fmt, ap) != 0) {
	va_end(ap);
	return (-1);
	}

	switch (error) {
	case ENXIO:
	case ENODEV:
	case EPIPE:
	zfs_verror(hdl, EZFS_IO, fmt, ap);
	break;

	case ENOENT:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"dataset does not exist"));
	zfs_verror(hdl, EZFS_NOENT, fmt, ap);
	break;

	case ENOSPC:
	case EDQUOT:
	zfs_verror(hdl, EZFS_NOSPC, fmt, ap);
	break;

	case EEXIST:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"dataset already exists"));
	zfs_verror(hdl, EZFS_EXISTS, fmt, ap);
	break;

	case EBUSY:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"dataset is busy"));
	zfs_verror(hdl, EZFS_BUSY, fmt, ap);
	break;
	case EROFS:
	zfs_verror(hdl, EZFS_POOLREADONLY, fmt, ap);
	break;
	case ENAMETOOLONG:
	zfs_verror(hdl, EZFS_NAMETOOLONG, fmt, ap);
	break;
	case ENOTSUP:
	zfs_verror(hdl, EZFS_BADVERSION, fmt, ap);
	break;
	case EAGAIN:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"pool I/O is currently suspended"));
	zfs_verror(hdl, EZFS_POOLUNAVAIL, fmt, ap);
	break;
	case EREMOTEIO:
	zfs_verror(hdl, EZFS_ACTIVE_POOL, fmt, ap);
	break;
	case ZFS_ERR_UNKNOWN_SEND_STREAM_FEATURE:
	case ZFS_ERR_IOC_CMD_UNAVAIL:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "the loaded zfs "
	"module does not support this operation. A reboot may "
	"be required to enable this operation."));
	zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap);
	break;
	case ZFS_ERR_IOC_ARG_UNAVAIL:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "the loaded zfs "
	"module does not support an option for this operation. "
	"A reboot may be required to enable this option."));
	zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap);
	break;
	case ZFS_ERR_IOC_ARG_REQUIRED:
	case ZFS_ERR_IOC_ARG_BADTYPE:
	zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap);
	break;
	case ZFS_ERR_WRONG_PARENT:
	zfs_verror(hdl, EZFS_WRONG_PARENT, fmt, ap);
	break;
	case ZFS_ERR_BADPROP:
	zfs_verror(hdl, EZFS_BADPROP, fmt, ap);
	break;
	default:
	zfs_error_aux(hdl, "%s", strerror(error));
	zfs_verror(hdl, EZFS_UNKNOWN, fmt, ap);
	break;
	}

	va_end(ap);
	return (-1);
	}

	void
	zfs_setprop_error(libzfs_handle_t *hdl, zfs_prop_t prop, int err,
	char *errbuf)
	{
	switch (err) {

	case ENOSPC:
	/*
	* For quotas and reservations, ENOSPC indicates
	* something different; setting a quota or reservation
	* doesn't use any disk space.
	*/
	switch (prop) {
	case ZFS_PROP_QUOTA:
	case ZFS_PROP_REFQUOTA:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"size is less than current used or "
	"reserved space"));
	(void) zfs_error(hdl, EZFS_PROPSPACE, errbuf);
	break;

	case ZFS_PROP_RESERVATION:
	case ZFS_PROP_REFRESERVATION:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"size is greater than available space"));
	(void) zfs_error(hdl, EZFS_PROPSPACE, errbuf);
	break;

	default:
	(void) zfs_standard_error(hdl, err, errbuf);
	break;
	}
	break;

	case EBUSY:
	(void) zfs_standard_error(hdl, EBUSY, errbuf);
	break;

	case EROFS:
	(void) zfs_error(hdl, EZFS_DSREADONLY, errbuf);
	break;

	case E2BIG:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property value too long"));
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	break;

	case ENOTSUP:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"pool and or dataset must be upgraded to set this "
	"property or value"));
	(void) zfs_error(hdl, EZFS_BADVERSION, errbuf);
	break;

	case ERANGE:
	if (prop == ZFS_PROP_COMPRESSION \|\|
	prop == ZFS_PROP_DNODESIZE \|\|
	prop == ZFS_PROP_RECORDSIZE) {
	(void) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property setting is not allowed on "
	"bootable datasets"));
	(void) zfs_error(hdl, EZFS_NOTSUP, errbuf);
	} else if (prop == ZFS_PROP_CHECKSUM \|\|
	prop == ZFS_PROP_DEDUP) {
	(void) zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"property setting is not allowed on "
	"root pools"));
	(void) zfs_error(hdl, EZFS_NOTSUP, errbuf);
	} else {
	(void) zfs_standard_error(hdl, err, errbuf);
	}
	break;

	case EINVAL:
	if (prop == ZPROP_INVAL) {
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	} else {
	(void) zfs_standard_error(hdl, err, errbuf);
	}
	break;

	case ZFS_ERR_BADPROP:
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	break;

	case EACCES:
	if (prop == ZFS_PROP_KEYLOCATION) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"keylocation may only be set on encryption roots"));
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	} else {
	(void) zfs_standard_error(hdl, err, errbuf);
	}
	break;

	case EOVERFLOW:
	/*
	* This platform can't address a volume this big.
	*/
	#ifdef _ILP32
	if (prop == ZFS_PROP_VOLSIZE) {
	(void) zfs_error(hdl, EZFS_VOLTOOBIG, errbuf);
	break;
	}
	#endif
	fallthrough;
	default:
	(void) zfs_standard_error(hdl, err, errbuf);
	}
	}

	int
	zpool_standard_error(libzfs_handle_t hdl, int error, const char msg)
	{
	return (zpool_standard_error_fmt(hdl, error, "%s", msg));
	}

	int
	zpool_standard_error_fmt(libzfs_handle_t hdl, int error, const char fmt, ...)
	{
	va_list ap;

	va_start(ap, fmt);

	if (zfs_common_error(hdl, error, fmt, ap) != 0) {
	va_end(ap);
	return (-1);
	}

	switch (error) {
	case ENODEV:
	zfs_verror(hdl, EZFS_NODEVICE, fmt, ap);
	break;

	case ENOENT:
	zfs_error_aux(hdl,
	dgettext(TEXT_DOMAIN, "no such pool or dataset"));
	zfs_verror(hdl, EZFS_NOENT, fmt, ap);
	break;

	case EEXIST:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"pool already exists"));
	zfs_verror(hdl, EZFS_EXISTS, fmt, ap);
	break;

	case EBUSY:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool is busy"));
	zfs_verror(hdl, EZFS_BUSY, fmt, ap);
	break;

	/* There is no pending operation to cancel */
	case ENOTACTIVE:
	zfs_verror(hdl, EZFS_NO_PENDING, fmt, ap);
	break;

	case ENXIO:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"one or more devices is currently unavailable"));
	zfs_verror(hdl, EZFS_BADDEV, fmt, ap);
	break;

	case ENAMETOOLONG:
	zfs_verror(hdl, EZFS_DEVOVERFLOW, fmt, ap);
	break;

	case ENOTSUP:
	zfs_verror(hdl, EZFS_POOL_NOTSUP, fmt, ap);
	break;

	case EINVAL:
	zfs_verror(hdl, EZFS_POOL_INVALARG, fmt, ap);
	break;

	case ENOSPC:
	case EDQUOT:
	zfs_verror(hdl, EZFS_NOSPC, fmt, ap);
	return (-1);

	case EAGAIN:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"pool I/O is currently suspended"));
	zfs_verror(hdl, EZFS_POOLUNAVAIL, fmt, ap);
	break;

	case EROFS:
	zfs_verror(hdl, EZFS_POOLREADONLY, fmt, ap);
	break;
	case EDOM:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"block size out of range or does not match"));
	zfs_verror(hdl, EZFS_BADPROP, fmt, ap);
	break;
	case EREMOTEIO:
	zfs_verror(hdl, EZFS_ACTIVE_POOL, fmt, ap);
	break;
	case ZFS_ERR_CHECKPOINT_EXISTS:
	zfs_verror(hdl, EZFS_CHECKPOINT_EXISTS, fmt, ap);
	break;
	case ZFS_ERR_DISCARDING_CHECKPOINT:
	zfs_verror(hdl, EZFS_DISCARDING_CHECKPOINT, fmt, ap);
	break;
	case ZFS_ERR_NO_CHECKPOINT:
	zfs_verror(hdl, EZFS_NO_CHECKPOINT, fmt, ap);
	break;
	case ZFS_ERR_DEVRM_IN_PROGRESS:
	zfs_verror(hdl, EZFS_DEVRM_IN_PROGRESS, fmt, ap);
	break;
	case ZFS_ERR_VDEV_TOO_BIG:
	zfs_verror(hdl, EZFS_VDEV_TOO_BIG, fmt, ap);
	break;
	case ZFS_ERR_EXPORT_IN_PROGRESS:
	zfs_verror(hdl, EZFS_EXPORT_IN_PROGRESS, fmt, ap);
	break;
	case ZFS_ERR_RESILVER_IN_PROGRESS:
	zfs_verror(hdl, EZFS_RESILVERING, fmt, ap);
	break;
	case ZFS_ERR_REBUILD_IN_PROGRESS:
	zfs_verror(hdl, EZFS_REBUILDING, fmt, ap);
	break;
	case ZFS_ERR_BADPROP:
	zfs_verror(hdl, EZFS_BADPROP, fmt, ap);
	break;
	+ case ZFS_ERR_VDEV_NOTSUP:
	+ zfs_verror(hdl, EZFS_VDEV_NOTSUP, fmt, ap);
	+ break;
	case ZFS_ERR_IOC_CMD_UNAVAIL:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "the loaded zfs "
	"module does not support this operation. A reboot may "
	"be required to enable this operation."));
	zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap);
	break;
	case ZFS_ERR_IOC_ARG_UNAVAIL:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "the loaded zfs "
	"module does not support an option for this operation. "
	"A reboot may be required to enable this option."));
	zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap);
	break;
	case ZFS_ERR_IOC_ARG_REQUIRED:
	case ZFS_ERR_IOC_ARG_BADTYPE:
	zfs_verror(hdl, EZFS_IOC_NOTSUPPORTED, fmt, ap);
	break;
	default:
	zfs_error_aux(hdl, "%s", strerror(error));
	zfs_verror(hdl, EZFS_UNKNOWN, fmt, ap);
	}

	va_end(ap);
	return (-1);
	}

	/*
	* Display an out of memory error message and abort the current program.
	*/
	int
	no_memory(libzfs_handle_t *hdl)
	{
	return (zfs_error(hdl, EZFS_NOMEM, "internal error"));
	}

	/*
	* A safe form of malloc() which will die if the allocation fails.
	*/
	void *
	zfs_alloc(libzfs_handle_t *hdl, size_t size)
	{
	void *data;

	if ((data = calloc(1, size)) == NULL)
	(void) no_memory(hdl);

	return (data);
	}

	/*
	* A safe form of asprintf() which will die if the allocation fails.
	*/
	char *
	zfs_asprintf(libzfs_handle_t hdl, const char fmt, ...)
	{
	va_list ap;
	char *ret;
	int err;

	va_start(ap, fmt);

	err = vasprintf(&ret, fmt, ap);

	va_end(ap);

	if (err < 0) {
	(void) no_memory(hdl);
	ret = NULL;
	}

	return (ret);
	}

	/*
	* A safe form of realloc(), which also zeroes newly allocated space.
	*/
	void *
	zfs_realloc(libzfs_handle_t hdl, void ptr, size_t oldsize, size_t newsize)
	{
	void *ret;

	if ((ret = realloc(ptr, newsize)) == NULL) {
	(void) no_memory(hdl);
	return (NULL);
	}

	bzero((char *)ret + oldsize, (newsize - oldsize));
	return (ret);
	}

	/*
	* A safe form of strdup() which will die if the allocation fails.
	*/
	char *
	zfs_strdup(libzfs_handle_t hdl, const char str)
	{
	char *ret;

	if ((ret = strdup(str)) == NULL)
	(void) no_memory(hdl);

	return (ret);
	}

	void
	libzfs_print_on_error(libzfs_handle_t *hdl, boolean_t printerr)
	{
	hdl->libzfs_printerr = printerr;
	}

	/*
	* Read lines from an open file descriptor and store them in an array of
	* strings until EOF. lines[] will be allocated and populated with all the
	* lines read. All newlines are replaced with NULL terminators for
	* convenience. lines[] must be freed after use with libzfs_free_str_array().
	*
	* Returns the number of lines read.
	*/
	static int
	libzfs_read_stdout_from_fd(int fd, char **lines[])
	{

	FILE *fp;
	int lines_cnt = 0;
	size_t len = 0;
	char *line = NULL;
	char tmp_lines = NULL, tmp;

	fp = fdopen(fd, "r");
	if (fp == NULL) {
	close(fd);
	return (0);
	}
	while (getline(&line, &len, fp) != -1) {
	tmp = realloc(tmp_lines, sizeof (tmp_lines) (lines_cnt + 1));
	if (tmp == NULL) {
	/* Return the lines we were able to process */
	break;
	}
	tmp_lines = tmp;

	/* Remove newline if not EOF */
	if (line[strlen(line) - 1] == '\n')
	line[strlen(line) - 1] = '\0';

	tmp_lines[lines_cnt] = strdup(line);
	if (tmp_lines[lines_cnt] == NULL)
	break;
	++lines_cnt;
	}
	free(line);
	fclose(fp);
	*lines = tmp_lines;
	return (lines_cnt);
	}

	static int
	libzfs_run_process_impl(const char path, char argv[], char *env[], int flags,
	char *lines[], int lines_cnt)
	{
	pid_t pid;
	int error, devnull_fd;
	int link[2];

	/*
	* Setup a pipe between our child and parent process if we're
	* reading stdout.
	*/
	if (lines != NULL && pipe2(link, O_NONBLOCK \| O_CLOEXEC) == -1)
	return (-EPIPE);

	pid = fork();
	if (pid == 0) {
	/* Child process */
	devnull_fd = open("/dev/null", O_WRONLY \| O_CLOEXEC);

	if (devnull_fd < 0)
	_exit(-1);

	if (!(flags & STDOUT_VERBOSE) && (lines == NULL))
	(void) dup2(devnull_fd, STDOUT_FILENO);
	else if (lines != NULL) {
	/* Save the output to lines[] */
	dup2(link[1], STDOUT_FILENO);
	}

	if (!(flags & STDERR_VERBOSE))
	(void) dup2(devnull_fd, STDERR_FILENO);

	if (flags & NO_DEFAULT_PATH) {
	if (env == NULL)
	execv(path, argv);
	else
	execve(path, argv, env);
	} else {
	if (env == NULL)
	execvp(path, argv);
	else
	execvpe(path, argv, env);
	}

	_exit(-1);
	} else if (pid > 0) {
	/* Parent process */
	int status;

	while ((error = waitpid(pid, &status, 0)) == -1 &&
	errno == EINTR)
	;
	if (error < 0 \|\| !WIFEXITED(status))
	return (-1);

	if (lines != NULL) {
	close(link[1]);
	*lines_cnt = libzfs_read_stdout_from_fd(link[0], lines);
	}
	return (WEXITSTATUS(status));
	}

	return (-1);
	}

	int
	libzfs_run_process(const char path, char argv[], int flags)
	{
	return (libzfs_run_process_impl(path, argv, NULL, flags, NULL, NULL));
	}

	/*
	* Run a command and store its stdout lines in an array of strings (lines[]).
	* lines[] is allocated and populated for you, and the number of lines is set in
	* lines_cnt. lines[] must be freed after use with libzfs_free_str_array().
	* All newlines (\n) in lines[] are terminated for convenience.
	*/
	int
	libzfs_run_process_get_stdout(const char path, char argv[], char *env[],
	char *lines[], int lines_cnt)
	{
	return (libzfs_run_process_impl(path, argv, env, 0, lines, lines_cnt));
	}

	/*
	* Same as libzfs_run_process_get_stdout(), but run without $PATH set. This
	* means that *path needs to be the full path to the executable.
	*/
	int
	libzfs_run_process_get_stdout_nopath(const char path, char argv[],
	char env[], char lines[], int lines_cnt)
	{
	return (libzfs_run_process_impl(path, argv, env, NO_DEFAULT_PATH,
	lines, lines_cnt));
	}

	/*
	* Free an array of strings. Free both the strings contained in the array and
	* the array itself.
	*/
	void
	libzfs_free_str_array(char **strs, int count)
	{
	while (--count >= 0)
	free(strs[count]);

	free(strs);
	}

	/*
	* Returns 1 if environment variable is set to "YES", "yes", "ON", "on", or
	* a non-zero number.
	*
	* Returns 0 otherwise.
	*/
	int
	libzfs_envvar_is_set(char *envvar)
	{
	char *env = getenv(envvar);
	if (env && (strtoul(env, NULL, 0) > 0 \|\|
	(!strncasecmp(env, "YES", 3) && strnlen(env, 4) == 3) \|\|
	(!strncasecmp(env, "ON", 2) && strnlen(env, 3) == 2)))
	return (1);

	return (0);
	}

	libzfs_handle_t *
	libzfs_init(void)
	{
	libzfs_handle_t *hdl;
	int error;
	char *env;

	if ((error = libzfs_load_module()) != 0) {
	errno = error;
	return (NULL);
	}

	if ((hdl = calloc(1, sizeof (libzfs_handle_t))) == NULL) {
	return (NULL);
	}

	if (regcomp(&hdl->libzfs_urire, URI_REGEX, 0) != 0) {
	free(hdl);
	return (NULL);
	}

	if ((hdl->libzfs_fd = open(ZFS_DEV, O_RDWR\|O_EXCL\|O_CLOEXEC)) < 0) {
	free(hdl);
	return (NULL);
	}

	if (libzfs_core_init() != 0) {
	(void) close(hdl->libzfs_fd);
	free(hdl);
	return (NULL);
	}

	zfs_prop_init();
	zpool_prop_init();
	zpool_feature_init();
	+ vdev_prop_init();
	libzfs_mnttab_init(hdl);
	fletcher_4_init();

	if (getenv("ZFS_PROP_DEBUG") != NULL) {
	hdl->libzfs_prop_debug = B_TRUE;
	}
	if ((env = getenv("ZFS_SENDRECV_MAX_NVLIST")) != NULL) {
	if ((error = zfs_nicestrtonum(hdl, env,
	&hdl->libzfs_max_nvlist))) {
	errno = error;
	(void) close(hdl->libzfs_fd);
	free(hdl);
	return (NULL);
	}
	} else {
	hdl->libzfs_max_nvlist = (SPA_MAXBLOCKSIZE * 4);
	}

	/*
	* For testing, remove some settable properties and features
	*/
	if (libzfs_envvar_is_set("ZFS_SYSFS_PROP_SUPPORT_TEST")) {
	zprop_desc_t *proptbl;

	proptbl = zpool_prop_get_table();
	proptbl[ZPOOL_PROP_COMMENT].pd_zfs_mod_supported = B_FALSE;

	proptbl = zfs_prop_get_table();
	proptbl[ZFS_PROP_DNODESIZE].pd_zfs_mod_supported = B_FALSE;

	zfeature_info_t *ftbl = spa_feature_table;
	ftbl[SPA_FEATURE_LARGE_BLOCKS].fi_zfs_mod_supported = B_FALSE;
	}

	return (hdl);
	}

	void
	libzfs_fini(libzfs_handle_t *hdl)
	{
	(void) close(hdl->libzfs_fd);
	zpool_free_handles(hdl);
	namespace_clear(hdl);
	libzfs_mnttab_fini(hdl);
	libzfs_core_fini();
	regfree(&hdl->libzfs_urire);
	fletcher_4_fini();
	#if LIBFETCH_DYNAMIC
	if (hdl->libfetch != (void *)-1 && hdl->libfetch != NULL)
	(void) dlclose(hdl->libfetch);
	free(hdl->libfetch_load_error);
	#endif
	free(hdl);
	}

	libzfs_handle_t *
	zpool_get_handle(zpool_handle_t *zhp)
	{
	return (zhp->zpool_hdl);
	}

	libzfs_handle_t *
	zfs_get_handle(zfs_handle_t *zhp)
	{
	return (zhp->zfs_hdl);
	}

	zpool_handle_t *
	zfs_get_pool_handle(const zfs_handle_t *zhp)
	{
	return (zhp->zpool_hdl);
	}

	/*
	* Given a name, determine whether or not it's a valid path
	* (starts with '/' or "./"). If so, walk the mnttab trying
	* to match the device number. If not, treat the path as an
	* fs/vol/snap/bkmark name.
	*/
	zfs_handle_t *
	zfs_path_to_zhandle(libzfs_handle_t hdl, const char path, zfs_type_t argtype)
	{
	struct stat64 statbuf;
	struct extmnttab entry;

	if (path[0] != '/' && strncmp(path, "./", strlen("./")) != 0) {
	/*
	* It's not a valid path, assume it's a name of type 'argtype'.
	*/
	return (zfs_open(hdl, path, argtype));
	}

	if (getextmntent(path, &entry, &statbuf) != 0)
	return (NULL);

	if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0) {
	(void) fprintf(stderr, gettext("'%s': not a ZFS filesystem\n"),
	path);
	return (NULL);
	}

	return (zfs_open(hdl, entry.mnt_special, ZFS_TYPE_FILESYSTEM));
	}

	/*
	* Initialize the zc_nvlist_dst member to prepare for receiving an nvlist from
	* an ioctl().
	*/
	int
	zcmd_alloc_dst_nvlist(libzfs_handle_t hdl, zfs_cmd_t zc, size_t len)
	{
	if (len == 0)
	len = 256 * 1024;
	zc->zc_nvlist_dst_size = len;
	zc->zc_nvlist_dst =
	(uint64_t)(uintptr_t)zfs_alloc(hdl, zc->zc_nvlist_dst_size);
	if (zc->zc_nvlist_dst == 0)
	return (-1);

	return (0);
	}

	/*
	* Called when an ioctl() which returns an nvlist fails with ENOMEM. This will
	* expand the nvlist to the size specified in 'zc_nvlist_dst_size', which was
	* filled in by the kernel to indicate the actual required size.
	*/
	int
	zcmd_expand_dst_nvlist(libzfs_handle_t hdl, zfs_cmd_t zc)
	{
	free((void *)(uintptr_t)zc->zc_nvlist_dst);
	zc->zc_nvlist_dst =
	(uint64_t)(uintptr_t)zfs_alloc(hdl, zc->zc_nvlist_dst_size);
	if (zc->zc_nvlist_dst == 0)
	return (-1);

	return (0);
	}

	/*
	* Called to free the src and dst nvlists stored in the command structure.
	*/
	void
	zcmd_free_nvlists(zfs_cmd_t *zc)
	{
	free((void *)(uintptr_t)zc->zc_nvlist_conf);
	free((void *)(uintptr_t)zc->zc_nvlist_src);
	free((void *)(uintptr_t)zc->zc_nvlist_dst);
	zc->zc_nvlist_conf = 0;
	zc->zc_nvlist_src = 0;
	zc->zc_nvlist_dst = 0;
	}

	static int
	zcmd_write_nvlist_com(libzfs_handle_t hdl, uint64_t outnv, uint64_t *outlen,
	nvlist_t *nvl)
	{
	char *packed;
	size_t len;

	verify(nvlist_size(nvl, &len, NV_ENCODE_NATIVE) == 0);

	if ((packed = zfs_alloc(hdl, len)) == NULL)
	return (-1);

	verify(nvlist_pack(nvl, &packed, &len, NV_ENCODE_NATIVE, 0) == 0);

	*outnv = (uint64_t)(uintptr_t)packed;
	*outlen = len;

	return (0);
	}

	int
	zcmd_write_conf_nvlist(libzfs_handle_t hdl, zfs_cmd_t zc, nvlist_t *nvl)
	{
	return (zcmd_write_nvlist_com(hdl, &zc->zc_nvlist_conf,
	&zc->zc_nvlist_conf_size, nvl));
	}

	int
	zcmd_write_src_nvlist(libzfs_handle_t hdl, zfs_cmd_t zc, nvlist_t *nvl)
	{
	return (zcmd_write_nvlist_com(hdl, &zc->zc_nvlist_src,
	&zc->zc_nvlist_src_size, nvl));
	}

	/*
	* Unpacks an nvlist from the ZFS ioctl command structure.
	*/
	int
	zcmd_read_dst_nvlist(libzfs_handle_t hdl, zfs_cmd_t zc, nvlist_t **nvlp)
	{
	if (nvlist_unpack((void *)(uintptr_t)zc->zc_nvlist_dst,
	zc->zc_nvlist_dst_size, nvlp, 0) != 0)
	return (no_memory(hdl));

	return (0);
	}

	/*
	* ================================================================
	* API shared by zfs and zpool property management
	* ================================================================
	*/

	static void
	zprop_print_headers(zprop_get_cbdata_t *cbp, zfs_type_t type)
	{
	zprop_list_t *pl = cbp->cb_proplist;
	int i;
	char *title;
	size_t len;

	cbp->cb_first = B_FALSE;
	if (cbp->cb_scripted)
	return;

	/*
	* Start with the length of the column headers.
	*/
	cbp->cb_colwidths[GET_COL_NAME] = strlen(dgettext(TEXT_DOMAIN, "NAME"));
	cbp->cb_colwidths[GET_COL_PROPERTY] = strlen(dgettext(TEXT_DOMAIN,
	"PROPERTY"));
	cbp->cb_colwidths[GET_COL_VALUE] = strlen(dgettext(TEXT_DOMAIN,
	"VALUE"));
	cbp->cb_colwidths[GET_COL_RECVD] = strlen(dgettext(TEXT_DOMAIN,
	"RECEIVED"));
	cbp->cb_colwidths[GET_COL_SOURCE] = strlen(dgettext(TEXT_DOMAIN,
	"SOURCE"));

	/* first property is always NAME */
	assert(cbp->cb_proplist->pl_prop ==
	- ((type == ZFS_TYPE_POOL) ? ZPOOL_PROP_NAME : ZFS_PROP_NAME));
	+ ((type == ZFS_TYPE_POOL) ? ZPOOL_PROP_NAME :
	+ ((type == ZFS_TYPE_VDEV) ? VDEV_PROP_NAME : ZFS_PROP_NAME)));

	/*
	* Go through and calculate the widths for each column. For the
	* 'source' column, we kludge it up by taking the worst-case scenario of
	* inheriting from the longest name. This is acceptable because in the
	* majority of cases 'SOURCE' is the last column displayed, and we don't
	* use the width anyway. Note that the 'VALUE' column can be oversized,
	* if the name of the property is much longer than any values we find.
	*/
	for (pl = cbp->cb_proplist; pl != NULL; pl = pl->pl_next) {
	/*
	* 'PROPERTY' column
	*/
	if (pl->pl_prop != ZPROP_INVAL) {
	const char *propname = (type == ZFS_TYPE_POOL) ?
	zpool_prop_to_name(pl->pl_prop) :
	- zfs_prop_to_name(pl->pl_prop);
	+ ((type == ZFS_TYPE_VDEV) ?
	+ vdev_prop_to_name(pl->pl_prop) :
	+ zfs_prop_to_name(pl->pl_prop));

	+ assert(propname != NULL);
	len = strlen(propname);
	if (len > cbp->cb_colwidths[GET_COL_PROPERTY])
	cbp->cb_colwidths[GET_COL_PROPERTY] = len;
	} else {
	+ assert(pl->pl_user_prop != NULL);
	len = strlen(pl->pl_user_prop);
	if (len > cbp->cb_colwidths[GET_COL_PROPERTY])
	cbp->cb_colwidths[GET_COL_PROPERTY] = len;
	}

	/*
	* 'VALUE' column. The first property is always the 'name'
	* property that was tacked on either by /sbin/zfs's
	* zfs_do_get() or when calling zprop_expand_list(), so we
	* ignore its width. If the user specified the name property
	* to display, then it will be later in the list in any case.
	*/
	if (pl != cbp->cb_proplist &&
	pl->pl_width > cbp->cb_colwidths[GET_COL_VALUE])
	cbp->cb_colwidths[GET_COL_VALUE] = pl->pl_width;

	/* 'RECEIVED' column. */
	if (pl != cbp->cb_proplist &&
	pl->pl_recvd_width > cbp->cb_colwidths[GET_COL_RECVD])
	cbp->cb_colwidths[GET_COL_RECVD] = pl->pl_recvd_width;

	/*
	* 'NAME' and 'SOURCE' columns
	*/
	- if (pl->pl_prop == (type == ZFS_TYPE_POOL ? ZPOOL_PROP_NAME :
	- ZFS_PROP_NAME) &&
	- pl->pl_width > cbp->cb_colwidths[GET_COL_NAME]) {
	+ if (pl->pl_prop == ((type == ZFS_TYPE_POOL) ? ZPOOL_PROP_NAME :
	+ ((type == ZFS_TYPE_VDEV) ? VDEV_PROP_NAME :
	+ ZFS_PROP_NAME)) && pl->pl_width >
	+ cbp->cb_colwidths[GET_COL_NAME]) {
	cbp->cb_colwidths[GET_COL_NAME] = pl->pl_width;
	cbp->cb_colwidths[GET_COL_SOURCE] = pl->pl_width +
	strlen(dgettext(TEXT_DOMAIN, "inherited from"));
	}
	}

	/*
	* Now go through and print the headers.
	*/
	for (i = 0; i < ZFS_GET_NCOLS; i++) {
	switch (cbp->cb_columns[i]) {
	case GET_COL_NAME:
	title = dgettext(TEXT_DOMAIN, "NAME");
	break;
	case GET_COL_PROPERTY:
	title = dgettext(TEXT_DOMAIN, "PROPERTY");
	break;
	case GET_COL_VALUE:
	title = dgettext(TEXT_DOMAIN, "VALUE");
	break;
	case GET_COL_RECVD:
	title = dgettext(TEXT_DOMAIN, "RECEIVED");
	break;
	case GET_COL_SOURCE:
	title = dgettext(TEXT_DOMAIN, "SOURCE");
	break;
	default:
	title = NULL;
	}

	if (title != NULL) {
	if (i == (ZFS_GET_NCOLS - 1) \|\|
	cbp->cb_columns[i + 1] == GET_COL_NONE)
	(void) printf("%s", title);
	else
	(void) printf("%-*s ",
	cbp->cb_colwidths[cbp->cb_columns[i]],
	title);
	}
	}
	(void) printf("\n");
	}

	/*
	* Display a single line of output, according to the settings in the callback
	* structure.
	*/
	void
	zprop_print_one_property(const char name, zprop_get_cbdata_t cbp,
	const char propname, const char value, zprop_source_t sourcetype,
	const char source, const char recvd_value)
	{
	int i;
	const char *str = NULL;
	char buf[128];

	/*
	* Ignore those source types that the user has chosen to ignore.
	*/
	if ((sourcetype & cbp->cb_sources) == 0)
	return;

	if (cbp->cb_first)
	zprop_print_headers(cbp, cbp->cb_type);

	for (i = 0; i < ZFS_GET_NCOLS; i++) {
	switch (cbp->cb_columns[i]) {
	case GET_COL_NAME:
	str = name;
	break;

	case GET_COL_PROPERTY:
	str = propname;
	break;

	case GET_COL_VALUE:
	str = value;
	break;

	case GET_COL_SOURCE:
	switch (sourcetype) {
	case ZPROP_SRC_NONE:
	str = "-";
	break;

	case ZPROP_SRC_DEFAULT:
	str = "default";
	break;

	case ZPROP_SRC_LOCAL:
	str = "local";
	break;

	case ZPROP_SRC_TEMPORARY:
	str = "temporary";
	break;

	case ZPROP_SRC_INHERITED:
	(void) snprintf(buf, sizeof (buf),
	"inherited from %s", source);
	str = buf;
	break;
	case ZPROP_SRC_RECEIVED:
	str = "received";
	break;

	default:
	str = NULL;
	assert(!"unhandled zprop_source_t");
	}
	break;

	case GET_COL_RECVD:
	str = (recvd_value == NULL ? "-" : recvd_value);
	break;

	default:
	continue;
	}

	if (i == (ZFS_GET_NCOLS - 1) \|\|
	cbp->cb_columns[i + 1] == GET_COL_NONE)
	(void) printf("%s", str);
	else if (cbp->cb_scripted)
	(void) printf("%s\t", str);
	else
	(void) printf("%-*s ",
	cbp->cb_colwidths[cbp->cb_columns[i]],
	str);
	}

	(void) printf("\n");
	}

	/*
	* Given a numeric suffix, convert the value into a number of bits that the
	* resulting value must be shifted.
	*/
	static int
	str2shift(libzfs_handle_t hdl, const char buf)
	{
	const char *ends = "BKMGTPEZ";
	int i;

	if (buf[0] == '\0')
	return (0);
	for (i = 0; i < strlen(ends); i++) {
	if (toupper(buf[0]) == ends[i])
	break;
	}
	if (i == strlen(ends)) {
	if (hdl)
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"invalid numeric suffix '%s'"), buf);
	return (-1);
	}

	/*
	* Allow 'G' = 'GB' = 'GiB', case-insensitively.
	* However, 'BB' and 'BiB' are disallowed.
	*/
	if (buf[1] == '\0' \|\|
	(toupper(buf[0]) != 'B' &&
	((toupper(buf[1]) == 'B' && buf[2] == '\0') \|\|
	(toupper(buf[1]) == 'I' && toupper(buf[2]) == 'B' &&
	buf[3] == '\0'))))
	return (10 * i);

	if (hdl)
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"invalid numeric suffix '%s'"), buf);
	return (-1);
	}

	/*
	* Convert a string of the form '100G' into a real number. Used when setting
	* properties or creating a volume. 'buf' is used to place an extended error
	* message for the caller to use.
	*/
	int
	zfs_nicestrtonum(libzfs_handle_t hdl, const char value, uint64_t *num)
	{
	char *end;
	int shift;

	*num = 0;

	/* Check to see if this looks like a number. */
	if ((value[0] < '0' \|\| value[0] > '9') && value[0] != '.') {
	if (hdl)
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"bad numeric value '%s'"), value);
	return (-1);
	}

	/* Rely on strtoull() to process the numeric portion. */
	errno = 0;
	*num = strtoull(value, &end, 10);

	/*
	* Check for ERANGE, which indicates that the value is too large to fit
	* in a 64-bit value.
	*/
	if (errno == ERANGE) {
	if (hdl)
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"numeric value is too large"));
	return (-1);
	}

	/*
	* If we have a decimal value, then do the computation with floating
	* point arithmetic. Otherwise, use standard arithmetic.
	*/
	if (*end == '.') {
	double fval = strtod(value, &end);

	if ((shift = str2shift(hdl, end)) == -1)
	return (-1);

	fval *= pow(2, shift);

	/*
	* UINT64_MAX is not exactly representable as a double.
	* The closest representation is UINT64_MAX + 1, so we
	* use a >= comparison instead of > for the bounds check.
	*/
	if (fval >= (double)UINT64_MAX) {
	if (hdl)
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"numeric value is too large"));
	return (-1);
	}

	*num = (uint64_t)fval;
	} else {
	if ((shift = str2shift(hdl, end)) == -1)
	return (-1);

	/* Check for overflow */
	if (shift >= 64 \|\| (num << shift) >> shift != num) {
	if (hdl)
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"numeric value is too large"));
	return (-1);
	}

	*num <<= shift;
	}

	return (0);
	}

	/*
	* Given a propname=value nvpair to set, parse any numeric properties
	* (index, boolean, etc) if they are specified as strings and add the
	* resulting nvpair to the returned nvlist.
	*
	* At the DSL layer, all properties are either 64-bit numbers or strings.
	* We want the user to be able to ignore this fact and specify properties
	* as native values (numbers, for example) or as strings (to simplify
	* command line utilities). This also handles converting index types
	* (compression, checksum, etc) from strings to their on-disk index.
	*/
	int
	zprop_parse_value(libzfs_handle_t hdl, nvpair_t elem, int prop,
	zfs_type_t type, nvlist_t ret, char svalp, uint64_t ivalp,
	const char *errbuf)
	{
	data_type_t datatype = nvpair_type(elem);
	zprop_type_t proptype;
	const char *propname;
	char *value;
	boolean_t isnone = B_FALSE;
	boolean_t isauto = B_FALSE;
	int err = 0;

	if (type == ZFS_TYPE_POOL) {
	proptype = zpool_prop_get_type(prop);
	propname = zpool_prop_to_name(prop);
	+ } else if (type == ZFS_TYPE_VDEV) {
	+ proptype = vdev_prop_get_type(prop);
	+ propname = vdev_prop_to_name(prop);
	} else {
	proptype = zfs_prop_get_type(prop);
	propname = zfs_prop_to_name(prop);
	}

	/*
	* Convert any properties to the internal DSL value types.
	*/
	*svalp = NULL;
	*ivalp = 0;

	switch (proptype) {
	case PROP_TYPE_STRING:
	if (datatype != DATA_TYPE_STRING) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"'%s' must be a string"), nvpair_name(elem));
	goto error;
	}
	err = nvpair_value_string(elem, svalp);
	if (err != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"'%s' is invalid"), nvpair_name(elem));
	goto error;
	}
	if (strlen(*svalp) >= ZFS_MAXPROPLEN) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"'%s' is too long"), nvpair_name(elem));
	goto error;
	}
	break;

	case PROP_TYPE_NUMBER:
	if (datatype == DATA_TYPE_STRING) {
	(void) nvpair_value_string(elem, &value);
	if (strcmp(value, "none") == 0) {
	isnone = B_TRUE;
	} else if (strcmp(value, "auto") == 0) {
	isauto = B_TRUE;
	} else if (zfs_nicestrtonum(hdl, value, ivalp) != 0) {
	goto error;
	}
	} else if (datatype == DATA_TYPE_UINT64) {
	(void) nvpair_value_uint64(elem, ivalp);
	} else {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"'%s' must be a number"), nvpair_name(elem));
	goto error;
	}

	/*
	* Quota special: force 'none' and don't allow 0.
	*/
	if ((type & ZFS_TYPE_DATASET) && *ivalp == 0 && !isnone &&
	(prop == ZFS_PROP_QUOTA \|\| prop == ZFS_PROP_REFQUOTA)) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"use 'none' to disable quota/refquota"));
	goto error;
	}

	/*
	* Special handling for "*_limit=none". In this case it's not
	* 0 but UINT64_MAX.
	*/
	if ((type & ZFS_TYPE_DATASET) && isnone &&
	(prop == ZFS_PROP_FILESYSTEM_LIMIT \|\|
	prop == ZFS_PROP_SNAPSHOT_LIMIT)) {
	*ivalp = UINT64_MAX;
	}

	/*
	* Special handling for setting 'refreservation' to 'auto'. Use
	* UINT64_MAX to tell the caller to use zfs_fix_auto_resv().
	* 'auto' is only allowed on volumes.
	*/
	if (isauto) {
	switch (prop) {
	case ZFS_PROP_REFRESERVATION:
	if ((type & ZFS_TYPE_VOLUME) == 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"'%s=auto' only allowed on "
	"volumes"), nvpair_name(elem));
	goto error;
	}
	*ivalp = UINT64_MAX;
	break;
	default:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"'auto' is invalid value for '%s'"),
	nvpair_name(elem));
	goto error;
	}
	}

	break;

	case PROP_TYPE_INDEX:
	if (datatype != DATA_TYPE_STRING) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"'%s' must be a string"), nvpair_name(elem));
	goto error;
	}

	(void) nvpair_value_string(elem, &value);

	if (zprop_string_to_index(prop, value, ivalp, type) != 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"'%s' must be one of '%s'"), propname,
	zprop_values(prop, type));
	goto error;
	}
	break;

	default:
	abort();
	}

	/*
	* Add the result to our return set of properties.
	*/
	if (*svalp != NULL) {
	if (nvlist_add_string(ret, propname, *svalp) != 0) {
	(void) no_memory(hdl);
	return (-1);
	}
	} else {
	if (nvlist_add_uint64(ret, propname, *ivalp) != 0) {
	(void) no_memory(hdl);
	return (-1);
	}
	}

	return (0);
	error:
	(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
	return (-1);
	}

	static int
	addlist(libzfs_handle_t hdl, char propname, zprop_list_t **listp,
	zfs_type_t type)
	{
	int prop;
	zprop_list_t *entry;

	prop = zprop_name_to_prop(propname, type);

	if (prop != ZPROP_INVAL && !zprop_valid_for_type(prop, type, B_FALSE))
	prop = ZPROP_INVAL;

	/*
	- * When no property table entry can be found, return failure if
	- * this is a pool property or if this isn't a user-defined
	- * dataset property,
	+ * Return failure if no property table entry was found and this isn't
	+ * a user-defined property.
	*/
	if (prop == ZPROP_INVAL && ((type == ZFS_TYPE_POOL &&
	!zpool_prop_feature(propname) &&
	!zpool_prop_unsupported(propname)) \|\|
	- (type == ZFS_TYPE_DATASET && !zfs_prop_user(propname) &&
	- !zfs_prop_userquota(propname) && !zfs_prop_written(propname)))) {
	+ ((type == ZFS_TYPE_DATASET) && !zfs_prop_user(propname) &&
	+ !zfs_prop_userquota(propname) && !zfs_prop_written(propname)) \|\|
	+ ((type == ZFS_TYPE_VDEV) && !vdev_prop_user(propname)))) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"invalid property '%s'"), propname);
	return (zfs_error(hdl, EZFS_BADPROP,
	dgettext(TEXT_DOMAIN, "bad property list")));
	}

	if ((entry = zfs_alloc(hdl, sizeof (zprop_list_t))) == NULL)
	return (-1);

	entry->pl_prop = prop;
	if (prop == ZPROP_INVAL) {
	if ((entry->pl_user_prop = zfs_strdup(hdl, propname)) ==
	NULL) {
	free(entry);
	return (-1);
	}
	entry->pl_width = strlen(propname);
	} else {
	entry->pl_width = zprop_width(prop, &entry->pl_fixed,
	type);
	}

	*listp = entry;

	return (0);
	}

	/*
	* Given a comma-separated list of properties, construct a property list
	* containing both user-defined and native properties. This function will
	* return a NULL list if 'all' is specified, which can later be expanded
	* by zprop_expand_list().
	*/
	int
	zprop_get_list(libzfs_handle_t hdl, char props, zprop_list_t **listp,
	zfs_type_t type)
	{
	*listp = NULL;

	/*
	* If 'all' is specified, return a NULL list.
	*/
	if (strcmp(props, "all") == 0)
	return (0);

	/*
	* If no props were specified, return an error.
	*/
	if (props[0] == '\0') {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"no properties specified"));
	return (zfs_error(hdl, EZFS_BADPROP, dgettext(TEXT_DOMAIN,
	"bad property list")));
	}

	/*
	* It would be nice to use getsubopt() here, but the inclusion of column
	* aliases makes this more effort than it's worth.
	*/
	while (*props != '\0') {
	size_t len;
	char *p;
	char c;

	if ((p = strchr(props, ',')) == NULL) {
	len = strlen(props);
	p = props + len;
	} else {
	len = p - props;
	}

	/*
	* Check for empty options.
	*/
	if (len == 0) {
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"empty property name"));
	return (zfs_error(hdl, EZFS_BADPROP,
	dgettext(TEXT_DOMAIN, "bad property list")));
	}

	/*
	* Check all regular property names.
	*/
	c = props[len];
	props[len] = '\0';

	if (strcmp(props, "space") == 0) {
	static char *spaceprops[] = {
	"name", "avail", "used", "usedbysnapshots",
	"usedbydataset", "usedbyrefreservation",
	"usedbychildren", NULL
	};
	int i;

	for (i = 0; spaceprops[i]; i++) {
	if (addlist(hdl, spaceprops[i], listp, type))
	return (-1);
	listp = &(*listp)->pl_next;
	}
	} else {
	if (addlist(hdl, props, listp, type))
	return (-1);
	listp = &(*listp)->pl_next;
	}

	props = p;
	if (c == ',')
	props++;
	}

	return (0);
	}

	void
	zprop_free_list(zprop_list_t *pl)
	{
	zprop_list_t *next;

	while (pl != NULL) {
	next = pl->pl_next;
	free(pl->pl_user_prop);
	free(pl);
	pl = next;
	}
	}

	typedef struct expand_data {
	zprop_list_t **last;
	libzfs_handle_t *hdl;
	zfs_type_t type;
	} expand_data_t;

	static int
	zprop_expand_list_cb(int prop, void *cb)
	{
	zprop_list_t *entry;
	expand_data_t *edp = cb;

	if ((entry = zfs_alloc(edp->hdl, sizeof (zprop_list_t))) == NULL)
	return (ZPROP_INVAL);

	entry->pl_prop = prop;
	entry->pl_width = zprop_width(prop, &entry->pl_fixed, edp->type);
	entry->pl_all = B_TRUE;

	*(edp->last) = entry;
	edp->last = &entry->pl_next;

	return (ZPROP_CONT);
	}

	int
	zprop_expand_list(libzfs_handle_t hdl, zprop_list_t *plp, zfs_type_t type)
	{
	zprop_list_t *entry;
	zprop_list_t **last;
	expand_data_t exp;

	if (*plp == NULL) {
	/*
	* If this is the very first time we've been called for an 'all'
	* specification, expand the list to include all native
	* properties.
	*/
	last = plp;

	exp.last = last;
	exp.hdl = hdl;
	exp.type = type;

	if (zprop_iter_common(zprop_expand_list_cb, &exp, B_FALSE,
	B_FALSE, type) == ZPROP_INVAL)
	return (-1);

	/*
	* Add 'name' to the beginning of the list, which is handled
	* specially.
	*/
	if ((entry = zfs_alloc(hdl, sizeof (zprop_list_t))) == NULL)
	return (-1);

	- entry->pl_prop = (type == ZFS_TYPE_POOL) ? ZPOOL_PROP_NAME :
	- ZFS_PROP_NAME;
	+ entry->pl_prop = ((type == ZFS_TYPE_POOL) ? ZPOOL_PROP_NAME :
	+ ((type == ZFS_TYPE_VDEV) ? VDEV_PROP_NAME : ZFS_PROP_NAME));
	entry->pl_width = zprop_width(entry->pl_prop,
	&entry->pl_fixed, type);
	entry->pl_all = B_TRUE;
	entry->pl_next = *plp;
	*plp = entry;
	}
	return (0);
	}

	int
	zprop_iter(zprop_func func, void *cb, boolean_t show_all, boolean_t ordered,
	zfs_type_t type)
	{
	return (zprop_iter_common(func, cb, show_all, ordered, type));
	}

	/*
	* Fill given version buffer with zfs userland version
	*/
	void
	zfs_version_userland(char *version, int len)
	{
	(void) strlcpy(version, ZFS_META_ALIAS, len);
	}

	/*
	* Prints both zfs userland and kernel versions
	* Returns 0 on success, and -1 on error (with errno set)
	*/
	int
	zfs_version_print(void)
	{
	char zver_userland[128];
	char zver_kernel[128];

	zfs_version_userland(zver_userland, sizeof (zver_userland));

	(void) printf("%s\n", zver_userland);

	if (zfs_version_kernel(zver_kernel, sizeof (zver_kernel)) == -1) {
	fprintf(stderr, "zfs_version_kernel() failed: %s\n",
	strerror(errno));
	return (-1);
	}

	(void) printf("zfs-kmod-%s\n", zver_kernel);

	return (0);
	}

	/*
	* Return 1 if the user requested ANSI color output, and our terminal supports
	* it. Return 0 for no color.
	*/
	static int
	use_color(void)
	{
	static int use_color = -1;
	char *term;

	/*
	* Optimization:
	*
	* For each zpool invocation, we do a single check to see if we should
	* be using color or not, and cache that value for the lifetime of the
	* the zpool command. That makes it cheap to call use_color() when
	* we're printing with color. We assume that the settings are not going
	* to change during the invocation of a zpool command (the user isn't
	* going to change the ZFS_COLOR value while zpool is running, for
	* example).
	*/
	if (use_color != -1) {
	/*
	* We've already figured out if we should be using color or
	* not. Return the cached value.
	*/
	return (use_color);
	}

	term = getenv("TERM");
	/*
	* The user sets the ZFS_COLOR env var set to enable zpool ANSI color
	* output. However if NO_COLOR is set (https://no-color.org/) then
	* don't use it. Also, don't use color if terminal doesn't support
	* it.
	*/
	if (libzfs_envvar_is_set("ZFS_COLOR") &&
	!libzfs_envvar_is_set("NO_COLOR") &&
	isatty(STDOUT_FILENO) && term && strcmp("dumb", term) != 0 &&
	strcmp("unknown", term) != 0) {
	/* Color supported */
	use_color = 1;
	} else {
	use_color = 0;
	}

	return (use_color);
	}

	/*
	* color_start() and color_end() are used for when you want to colorize a block
	* of text. For example:
	*
	* color_start(ANSI_RED_FG)
	* printf("hello");
	* printf("world");
	* color_end();
	*/
	void
	color_start(char *color)
	{
	if (use_color())
	printf("%s", color);
	}

	void
	color_end(void)
	{
	if (use_color())
	printf(ANSI_RESET);
	}

	/* printf() with a color. If color is NULL, then do a normal printf. */
	int
	printf_color(char color, char format, ...)
	{
	va_list aptr;
	int rc;

	if (color)
	color_start(color);

	va_start(aptr, format);
	rc = vprintf(format, aptr);
	va_end(aptr);

	if (color)
	color_end();

	return (rc);
	}
	diff --git a/sys/contrib/openzfs/lib/libzfs/os/freebsd/libzfs_compat.c b/sys/contrib/openzfs/lib/libzfs/os/freebsd/libzfs_compat.c
	index f143f9cb63b3..e3f17662a11e 100644
	--- a/sys/contrib/openzfs/lib/libzfs/os/freebsd/libzfs_compat.c
	+++ b/sys/contrib/openzfs/lib/libzfs/os/freebsd/libzfs_compat.c
	@@ -1,360 +1,364 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2013 Martin Matuska <mm@FreeBSD.org>. All rights reserved.
	*/
	#include "../../libzfs_impl.h"
	#include <libzfs.h>
	#include <libzutil.h>
	#include <sys/sysctl.h>
	#include <libintl.h>
	#include <sys/linker.h>
	#include <sys/module.h>
	#include <sys/stat.h>
	#include <sys/param.h>

	#ifdef IN_BASE
	#define ZFS_KMOD "zfs"
	#else
	#define ZFS_KMOD "openzfs"
	#endif

	void
	libzfs_set_pipe_max(int infd)
	{
	/* FreeBSD automatically resizes */
	}

	static int
	execvPe(const char name, const char path, char * const *argv,
	char * const *envp)
	{
	const char **memp;
	size_t cnt, lp, ln;
	int eacces, save_errno;
	char buf[MAXPATHLEN];
	const char bp, np, op, p;
	struct stat sb;

	eacces = 0;

	/* If it's an absolute or relative path name, it's easy. */
	if (strchr(name, '/')) {
	bp = name;
	op = NULL;
	goto retry;
	}
	bp = buf;

	/* If it's an empty path name, fail in the usual POSIX way. */
	if (*name == '\0') {
	errno = ENOENT;
	return (-1);
	}

	op = path;
	ln = strlen(name);
	while (op != NULL) {
	np = strchrnul(op, ':');

	/*
	* It's a SHELL path -- double, leading and trailing colons
	* mean the current directory.
	*/
	if (np == op) {
	/* Empty component. */
	p = ".";
	lp = 1;
	} else {
	/* Non-empty component. */
	p = op;
	lp = np - op;
	}

	/* Advance to the next component or terminate after this. */
	if (*np == '\0')
	op = NULL;
	else
	op = np + 1;

	/*
	* If the path is too long complain. This is a possible
	* security issue; given a way to make the path too long
	* the user may execute the wrong program.
	*/
	if (lp + ln + 2 > sizeof (buf)) {
	(void) write(STDERR_FILENO, "execvP: ", 8);
	(void) write(STDERR_FILENO, p, lp);
	(void) write(STDERR_FILENO, ": path too long\n",
	16);
	continue;
	}
	bcopy(p, buf, lp);
	buf[lp] = '/';
	bcopy(name, buf + lp + 1, ln);
	buf[lp + ln + 1] = '\0';

	retry: (void) execve(bp, argv, envp);
	switch (errno) {
	case E2BIG:
	goto done;
	case ELOOP:
	case ENAMETOOLONG:
	case ENOENT:
	break;
	case ENOEXEC:
	for (cnt = 0; argv[cnt]; ++cnt)
	;

	/*
	* cnt may be 0 above; always allocate at least
	* 3 entries so that we can at least fit "sh", bp, and
	* the NULL terminator. We can rely on cnt to take into
	* account the NULL terminator in all other scenarios,
	* as we drop argv[0].
	*/
	memp = alloca(MAX(3, cnt + 2) * sizeof (char *));
	if (memp == NULL) {
	/* errno = ENOMEM; XXX override ENOEXEC? */
	goto done;
	}
	if (cnt > 0) {
	memp[0] = argv[0];
	memp[1] = bp;
	bcopy(argv + 1, memp + 2,
	cnt * sizeof (char *));
	} else {
	memp[0] = "sh";
	memp[1] = bp;
	memp[2] = NULL;
	}
	(void) execve(_PATH_BSHELL,
	__DECONST(char **, memp), envp);
	goto done;
	case ENOMEM:
	goto done;
	case ENOTDIR:
	break;
	case ETXTBSY:
	/*
	* We used to retry here, but sh(1) doesn't.
	*/
	goto done;
	default:
	/*
	* EACCES may be for an inaccessible directory or
	* a non-executable file. Call stat() to decide
	* which. This also handles ambiguities for EFAULT
	* and EIO, and undocumented errors like ESTALE.
	* We hope that the race for a stat() is unimportant.
	*/
	save_errno = errno;
	if (stat(bp, &sb) != 0)
	break;
	if (save_errno == EACCES) {
	eacces = 1;
	continue;
	}
	errno = save_errno;
	goto done;
	}
	}
	if (eacces)
	errno = EACCES;
	else
	errno = ENOENT;
	done:
	return (-1);
	}

	int
	execvpe(const char name, char const argv[], char * const envp[])
	{
	const char *path;

	/* Get the path we're searching. */
	if ((path = getenv("PATH")) == NULL)
	path = _PATH_DEFPATH;

	return (execvPe(name, path, argv, envp));
	}

	#define ERRBUFLEN 256

	static __thread char errbuf[ERRBUFLEN];

	const char *
	libzfs_error_init(int error)
	{
	char *msg = errbuf;
	size_t len, msglen = ERRBUFLEN;

	if (modfind("zfs") < 0) {
	len = snprintf(msg, msglen, dgettext(TEXT_DOMAIN,
	"Failed to load %s module: "), ZFS_KMOD);
	msg += len;
	msglen -= len;
	}

	(void) snprintf(msg, msglen, "%s", strerror(error));

	return (errbuf);
	}

	int
	zfs_ioctl(libzfs_handle_t hdl, int request, zfs_cmd_t zc)
	{
	return (lzc_ioctl_fd(hdl->libzfs_fd, request, zc));
	}

	/*
	* Verify the required ZFS_DEV device is available and optionally attempt
	* to load the ZFS modules. Under normal circumstances the modules
	* should already have been loaded by some external mechanism.
	*/
	int
	libzfs_load_module(void)
	{
	/*
	* XXX: kldfind(ZFS_KMOD) would be nice here, but we retain
	* modfind("zfs") so out-of-base openzfs userland works with the
	* in-base module.
	*/
	if (modfind("zfs") < 0) {
	/* Not present in kernel, try loading it. */
	if (kldload(ZFS_KMOD) < 0 && errno != EEXIST) {
	return (errno);
	}
	}
	return (0);
	}

	int
	zpool_relabel_disk(libzfs_handle_t hdl, const char path, const char *msg)
	{
	return (0);
	}

	int
	zpool_label_disk(libzfs_handle_t hdl, zpool_handle_t zhp, const char *name)
	{
	return (0);
	}

	int
	find_shares_object(differ_info_t *di)
	{
	return (0);
	}

	int
	zfs_destroy_snaps_nvl_os(libzfs_handle_t hdl, nvlist_t snaps)
	{
	return (0);
	}

	/*
	* Attach/detach the given filesystem to/from the given jail.
	*/
	int
	zfs_jail(zfs_handle_t *zhp, int jailid, int attach)
	{
	libzfs_handle_t *hdl = zhp->zfs_hdl;
	zfs_cmd_t zc = {"\0"};
	char errbuf[1024];
	unsigned long cmd;
	int ret;

	if (attach) {
	(void) snprintf(errbuf, sizeof (errbuf),
	dgettext(TEXT_DOMAIN, "cannot jail '%s'"), zhp->zfs_name);
	} else {
	(void) snprintf(errbuf, sizeof (errbuf),
	dgettext(TEXT_DOMAIN, "cannot unjail '%s'"), zhp->zfs_name);
	}

	switch (zhp->zfs_type) {
	case ZFS_TYPE_VOLUME:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"volumes can not be jailed"));
	return (zfs_error(hdl, EZFS_BADTYPE, errbuf));
	case ZFS_TYPE_SNAPSHOT:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"snapshots can not be jailed"));
	return (zfs_error(hdl, EZFS_BADTYPE, errbuf));
	case ZFS_TYPE_BOOKMARK:
	zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	"bookmarks can not be jailed"));
	return (zfs_error(hdl, EZFS_BADTYPE, errbuf));
	+ case ZFS_TYPE_VDEV:
	+ zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
	+ "vdevs can not be jailed"));
	+ return (zfs_error(hdl, EZFS_BADTYPE, errbuf));
	case ZFS_TYPE_POOL:
	case ZFS_TYPE_FILESYSTEM:
	/* OK */
	;
	}
	assert(zhp->zfs_type == ZFS_TYPE_FILESYSTEM);

	(void) strlcpy(zc.zc_name, zhp->zfs_name, sizeof (zc.zc_name));
	zc.zc_objset_type = DMU_OST_ZFS;
	zc.zc_zoneid = jailid;

	cmd = attach ? ZFS_IOC_JAIL : ZFS_IOC_UNJAIL;
	if ((ret = zfs_ioctl(hdl, cmd, &zc)) != 0)
	zfs_standard_error(hdl, errno, errbuf);

	return (ret);
	}

	/*
	* Set loader options for next boot.
	*/
	int
	zpool_nextboot(libzfs_handle_t *hdl, uint64_t pool_guid, uint64_t dev_guid,
	const char *command)
	{
	zfs_cmd_t zc = {"\0"};
	nvlist_t *args;
	int error;

	args = fnvlist_alloc();
	fnvlist_add_uint64(args, ZPOOL_CONFIG_POOL_GUID, pool_guid);
	fnvlist_add_uint64(args, ZPOOL_CONFIG_GUID, dev_guid);
	fnvlist_add_string(args, "command", command);
	error = zcmd_write_src_nvlist(hdl, &zc, args);
	if (error == 0)
	error = zfs_ioctl(hdl, ZFS_IOC_NEXTBOOT, &zc);
	zcmd_free_nvlists(&zc);
	nvlist_free(args);
	return (error);
	}

	/*
	* Fill given version buffer with zfs kernel version.
	* Returns 0 on success, and -1 on error (with errno set)
	*/
	int
	zfs_version_kernel(char *version, int len)
	{
	size_t l = len;

	return (sysctlbyname("vfs.zfs.version.module",
	version, &l, NULL, 0));
	}
	diff --git a/sys/contrib/openzfs/lib/libzfs_core/libzfs_core.abi b/sys/contrib/openzfs/lib/libzfs_core/libzfs_core.abi
	index 5bed6c8e0f4f..5a3cbaf0b80d 100644
	--- a/sys/contrib/openzfs/lib/libzfs_core/libzfs_core.abi
	+++ b/sys/contrib/openzfs/lib/libzfs_core/libzfs_core.abi
	@@ -1,2139 +1,2158 @@
	<abi-corpus version='2.0' architecture='elf-amd-x86_64' soname='libzfs_core.so.3'>
	<elf-needed>
	<dependency name='libnvpair.so.3'/>
	<dependency name='libpthread.so.0'/>
	<dependency name='libc.so.6'/>
	<dependency name='ld-linux-x86-64.so.2'/>
	</elf-needed>
	<elf-function-symbols>
	+ <elf-symbol name='_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='_sol_getmntent' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_char' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_char_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_int' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_int_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_long' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_long_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_ptr_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_short' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_add_short_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_and_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_cas_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_clear_long_excl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_dec_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_inc_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uchar_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_uint_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ulong_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_or_ushort_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_set_long_excl' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_16_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_32_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_64_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_8_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_char' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_char_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_int' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_int_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_long' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_long_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_ptr_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_short' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_sub_short_nv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_16' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_32' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_64' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_8' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_ptr' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_uchar' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_uint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_ulong' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='atomic_swap_ushort' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='get_system_hostid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getexecname' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getextmntent' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getmntany' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='getzoneid' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libspl_assertf' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_core_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='libzfs_core_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_head' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_after' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_before' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_head' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_insert_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_is_empty' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_link_active' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_link_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_link_replace' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_move_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_next' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_prev' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_remove' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_remove_head' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_remove_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='list_tail' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_bookmark' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_change_key' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_channel_program' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_channel_program_nosync' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_clone' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_create' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_destroy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_destroy_bookmarks' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_destroy_snaps' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_exists' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_get_bookmark_props' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_get_bookmarks' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_get_bootenv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_get_holds' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='lzc_get_vdev_prop' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_hold' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_initialize' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_ioctl_fd' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_load_key' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_pool_checkpoint' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_pool_checkpoint_discard' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_promote' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_receive' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_receive_one' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_receive_resumable' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_receive_with_cmdprops' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_receive_with_header' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_redact' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_release' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_rename' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_reopen' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_rollback' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_rollback_to' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_send' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_send_redacted' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_send_resume' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_send_resume_redacted' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_send_space' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_send_space_resume_redacted' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_set_bootenv' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='lzc_set_vdev_prop' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_snaprange_space' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_snapshot' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_sync' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_trim' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_unload_key' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_wait' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_wait_fs' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzc_wait_tag' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_consumer' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_enter' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_exit' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='membar_producer' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='mkdirp' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='print_timestamp' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='spl_pagesize' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='strlcat' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='strlcpy' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	</elf-function-symbols>
	<elf-variable-symbols>
	<elf-symbol name='libspl_assert_ok' size='4' type='object-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	</elf-variable-symbols>
	<abi-instr address-size='64' path='assert.c' language='LANG_C99'>
	<type-decl name='variadic parameter type' id='2c1145c5'/>
	<var-decl name='libspl_assert_ok' type-id='95e97e5e' mangled-name='libspl_assert_ok' visibility='default' elf-symbol-id='libspl_assert_ok'/>
	<function-decl name='libspl_assertf' mangled-name='libspl_assertf' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libspl_assertf'>
	<parameter type-id='80f4b756' name='file'/>
	<parameter type-id='80f4b756' name='func'/>
	<parameter type-id='95e97e5e' name='line'/>
	<parameter type-id='80f4b756' name='format'/>
	<parameter is-variadic='yes'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='atomic.c' language='LANG_C99'>
	<type-decl name='long int' size-in-bits='64' id='bd54fe1a'/>
	<type-decl name='short int' size-in-bits='16' id='a2185560'/>
	<type-decl name='signed char' size-in-bits='8' id='28577a57'/>
	<type-decl name='unsigned short int' size-in-bits='16' id='8efea9e5'/>
	<typedef-decl name='ulong_t' type-id='7359adad' id='ee1f298e'/>
	<typedef-decl name='int8_t' type-id='2171a512' id='ee31ee44'/>
	<typedef-decl name='int16_t' type-id='03896e23' id='23bd8cb5'/>
	<typedef-decl name='uint16_t' type-id='253c2d2a' id='149c6638'/>
	<typedef-decl name='__int8_t' type-id='28577a57' id='2171a512'/>
	<typedef-decl name='__int16_t' type-id='a2185560' id='03896e23'/>
	<typedef-decl name='__uint16_t' type-id='8efea9e5' id='253c2d2a'/>
	<typedef-decl name='__ssize_t' type-id='bd54fe1a' id='41060289'/>
	<typedef-decl name='ssize_t' type-id='41060289' id='79a0948f'/>
	<pointer-type-def type-id='48b5725f' size-in-bits='64' id='eaa32e2f'/>
	<qualified-type-def type-id='149c6638' volatile='yes' id='5120c5f7'/>
	<pointer-type-def type-id='5120c5f7' size-in-bits='64' id='93977ae7'/>
	<qualified-type-def type-id='8f92235e' volatile='yes' id='430e0681'/>
	<pointer-type-def type-id='430e0681' size-in-bits='64' id='3a147f31'/>
	<qualified-type-def type-id='b96825af' volatile='yes' id='84ff7d66'/>
	<pointer-type-def type-id='84ff7d66' size-in-bits='64' id='aa323ea4'/>
	<qualified-type-def type-id='ee1f298e' volatile='yes' id='6f7e09cb'/>
	<pointer-type-def type-id='6f7e09cb' size-in-bits='64' id='64698d33'/>
	<qualified-type-def type-id='48b5725f' volatile='yes' id='b0b3cbf9'/>
	<pointer-type-def type-id='b0b3cbf9' size-in-bits='64' id='fe09dd29'/>
	<function-decl name='atomic_inc_8' mangled-name='atomic_inc_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_16' mangled-name='atomic_inc_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_16'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_32' mangled-name='atomic_inc_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_32'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_ulong' mangled-name='atomic_inc_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_8' mangled-name='atomic_dec_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_16' mangled-name='atomic_dec_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_16'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_32' mangled-name='atomic_dec_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_32'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_dec_ulong' mangled-name='atomic_dec_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_8' mangled-name='atomic_add_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_16' mangled-name='atomic_add_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_32' mangled-name='atomic_add_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_long' mangled-name='atomic_add_long' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_long'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_add_ptr' mangled-name='atomic_add_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_8' mangled-name='atomic_sub_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_16' mangled-name='atomic_sub_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_32' mangled-name='atomic_sub_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_long' mangled-name='atomic_sub_long' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_long'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_sub_ptr' mangled-name='atomic_sub_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_8' mangled-name='atomic_or_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_16' mangled-name='atomic_or_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_32' mangled-name='atomic_or_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_or_ulong' mangled-name='atomic_or_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_8' mangled-name='atomic_and_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_16' mangled-name='atomic_and_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_32' mangled-name='atomic_and_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_and_ulong' mangled-name='atomic_and_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='atomic_inc_8_nv' mangled-name='atomic_inc_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_inc_16_nv' mangled-name='atomic_inc_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_inc_32_nv' mangled-name='atomic_inc_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_inc_ulong_nv' mangled-name='atomic_inc_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_inc_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_dec_8_nv' mangled-name='atomic_dec_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_dec_16_nv' mangled-name='atomic_dec_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_dec_32_nv' mangled-name='atomic_dec_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_dec_ulong_nv' mangled-name='atomic_dec_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_dec_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_add_8_nv' mangled-name='atomic_add_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_add_16_nv' mangled-name='atomic_add_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_add_32_nv' mangled-name='atomic_add_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_add_long_nv' mangled-name='atomic_add_long_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_long_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_add_ptr_nv' mangled-name='atomic_add_ptr_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_add_ptr_nv'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_sub_8_nv' mangled-name='atomic_sub_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='ee31ee44' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_sub_16_nv' mangled-name='atomic_sub_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='23bd8cb5' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_sub_32_nv' mangled-name='atomic_sub_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='3ff5601b' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_sub_long_nv' mangled-name='atomic_sub_long_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_long_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='bd54fe1a' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_sub_ptr_nv' mangled-name='atomic_sub_ptr_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_sub_ptr_nv'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='79a0948f' name='bits'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_or_8_nv' mangled-name='atomic_or_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_or_16_nv' mangled-name='atomic_or_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_or_32_nv' mangled-name='atomic_or_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_or_ulong_nv' mangled-name='atomic_or_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_or_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_and_8_nv' mangled-name='atomic_and_8_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_8_nv'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_and_16_nv' mangled-name='atomic_and_16_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_16_nv'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_and_32_nv' mangled-name='atomic_and_32_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_32_nv'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_and_ulong_nv' mangled-name='atomic_and_ulong_nv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_and_ulong_nv'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_cas_8' mangled-name='atomic_cas_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='exp'/>
	<parameter type-id='b96825af' name='des'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_cas_16' mangled-name='atomic_cas_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='exp'/>
	<parameter type-id='149c6638' name='des'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_cas_32' mangled-name='atomic_cas_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='exp'/>
	<parameter type-id='8f92235e' name='des'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_cas_ulong' mangled-name='atomic_cas_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='exp'/>
	<parameter type-id='ee1f298e' name='des'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_cas_ptr' mangled-name='atomic_cas_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_cas_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='eaa32e2f' name='exp'/>
	<parameter type-id='eaa32e2f' name='des'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_swap_8' mangled-name='atomic_swap_8' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_8'>
	<parameter type-id='aa323ea4' name='target'/>
	<parameter type-id='b96825af' name='bits'/>
	<return type-id='b96825af'/>
	</function-decl>
	<function-decl name='atomic_swap_16' mangled-name='atomic_swap_16' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_16'>
	<parameter type-id='93977ae7' name='target'/>
	<parameter type-id='149c6638' name='bits'/>
	<return type-id='149c6638'/>
	</function-decl>
	<function-decl name='atomic_swap_32' mangled-name='atomic_swap_32' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_32'>
	<parameter type-id='3a147f31' name='target'/>
	<parameter type-id='8f92235e' name='bits'/>
	<return type-id='8f92235e'/>
	</function-decl>
	<function-decl name='atomic_swap_ulong' mangled-name='atomic_swap_ulong' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_ulong'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='ee1f298e' name='bits'/>
	<return type-id='ee1f298e'/>
	</function-decl>
	<function-decl name='atomic_swap_ptr' mangled-name='atomic_swap_ptr' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_swap_ptr'>
	<parameter type-id='fe09dd29' name='target'/>
	<parameter type-id='eaa32e2f' name='bits'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='atomic_set_long_excl' mangled-name='atomic_set_long_excl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_set_long_excl'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='3502e3ff' name='value'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='atomic_clear_long_excl' mangled-name='atomic_clear_long_excl' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='atomic_clear_long_excl'>
	<parameter type-id='64698d33' name='target'/>
	<parameter type-id='3502e3ff' name='value'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='membar_enter' mangled-name='membar_enter' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='membar_enter'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='membar_producer' mangled-name='membar_producer' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='membar_producer'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='membar_consumer' mangled-name='membar_consumer' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='membar_consumer'>
	<return type-id='48b5725f'/>
	</function-decl>
	<type-decl name='unsigned long int' size-in-bits='64' id='7359adad'/>
	</abi-instr>
	<abi-instr address-size='64' path='getexecname.c' language='LANG_C99'>
	<function-decl name='getexecname' mangled-name='getexecname' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getexecname'>
	<return type-id='80f4b756'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='list.c' language='LANG_C99'>
	<typedef-decl name='list_node_t' type-id='b0b5e45e' id='b21843b2'/>
	<typedef-decl name='list_t' type-id='e824dae9' id='0899125f'/>
	<class-decl name='list_node' size-in-bits='128' is-struct='yes' visibility='default' id='b0b5e45e'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='next' type-id='b03eadb4' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='prev' type-id='b03eadb4' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='list' size-in-bits='256' is-struct='yes' visibility='default' id='e824dae9'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='list_size' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='list_offset' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='list_head' type-id='b0b5e45e' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='size_t' type-id='7359adad' id='b59d7dce'/>
	<pointer-type-def type-id='b0b5e45e' size-in-bits='64' id='b03eadb4'/>
	<pointer-type-def type-id='b21843b2' size-in-bits='64' id='ccc38265'/>
	<pointer-type-def type-id='0899125f' size-in-bits='64' id='352ec160'/>
	<function-decl name='list_create' mangled-name='list_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_create'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='b59d7dce' name='size'/>
	<parameter type-id='b59d7dce' name='offset'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_destroy' mangled-name='list_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_destroy'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_after' mangled-name='list_insert_after' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_after'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<parameter type-id='eaa32e2f' name='nobject'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_before' mangled-name='list_insert_before' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_before'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<parameter type-id='eaa32e2f' name='nobject'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_head' mangled-name='list_insert_head' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_head'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_insert_tail' mangled-name='list_insert_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_insert_tail'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_remove' mangled-name='list_remove' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_remove'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_remove_head' mangled-name='list_remove_head' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_remove_head'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_remove_tail' mangled-name='list_remove_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_remove_tail'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_head' mangled-name='list_head' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_head'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_tail' mangled-name='list_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_tail'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_next' mangled-name='list_next' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_next'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_prev' mangled-name='list_prev' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_prev'>
	<parameter type-id='352ec160' name='list'/>
	<parameter type-id='eaa32e2f' name='object'/>
	<return type-id='eaa32e2f'/>
	</function-decl>
	<function-decl name='list_move_tail' mangled-name='list_move_tail' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_move_tail'>
	<parameter type-id='352ec160' name='dst'/>
	<parameter type-id='352ec160' name='src'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_link_replace' mangled-name='list_link_replace' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_link_replace'>
	<parameter type-id='ccc38265' name='lold'/>
	<parameter type-id='ccc38265' name='lnew'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_link_init' mangled-name='list_link_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_link_init'>
	<parameter type-id='ccc38265' name='ln'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='list_link_active' mangled-name='list_link_active' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_link_active'>
	<parameter type-id='ccc38265' name='ln'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='list_is_empty' mangled-name='list_is_empty' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='list_is_empty'>
	<parameter type-id='352ec160' name='list'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='mkdirp.c' language='LANG_C99'>
	<typedef-decl name='__mode_t' type-id='f0981eeb' id='e1c52942'/>
	<typedef-decl name='mode_t' type-id='e1c52942' id='d50d396c'/>
	<function-decl name='mkdirp' mangled-name='mkdirp' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='mkdirp'>
	<parameter type-id='80f4b756' name='d'/>
	<parameter type-id='d50d396c' name='mode'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/gethostid.c' language='LANG_C99'>
	<function-decl name='get_system_hostid' mangled-name='get_system_hostid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='get_system_hostid'>
	<return type-id='7359adad'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/getmntany.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='03085adc' size-in-bits='192' id='083f8d58'>
	<subrange length='3' type-id='7359adad' id='56f209d2'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='8' id='89feb1ec'>
	<subrange length='1' type-id='7359adad' id='52f813b4'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='160' id='664ac0b7'>
	<subrange length='20' type-id='7359adad' id='fdca39cf'/>
	</array-type-def>
	- <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
	- <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
	- <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
	<class-decl name='mnttab' size-in-bits='256' is-struct='yes' visibility='default' id='1b055409'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='mnt_special' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='mnt_mountp' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='mnt_fstype' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='mnt_mntopts' type-id='26a90f95' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='extmnttab' size-in-bits='320' is-struct='yes' visibility='default' id='0c544dc0'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='mnt_special' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='mnt_mountp' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='mnt_fstype' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='mnt_mntopts' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='mnt_major' type-id='3502e3ff' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='288'>
	<var-decl name='mnt_minor' type-id='3502e3ff' visibility='default'/>
	</data-member>
	</class-decl>
	- <class-decl name='stat64' size-in-bits='1152' is-struct='yes' visibility='default' id='0bbec9cd'>
	+ <typedef-decl name='_IO_lock_t' type-id='48b5725f' id='bb4788fa'/>
	+ <class-decl name='_IO_marker' size-in-bits='192' is-struct='yes' visibility='default' id='010ae0b9'>
	<data-member access='public' layout-offset-in-bits='0'>
	- <var-decl name='st_dev' type-id='35ed8932' visibility='default'/>
	+ <var-decl name='_next' type-id='e4c6fa61' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	- <var-decl name='st_ino' type-id='71288a47' visibility='default'/>
	+ <var-decl name='_sbuf' type-id='dca988a5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	- <var-decl name='st_nlink' type-id='80f0b9df' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='192'>
	- <var-decl name='st_mode' type-id='e1c52942' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='224'>
	- <var-decl name='st_uid' type-id='cc5fcceb' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='256'>
	- <var-decl name='st_gid' type-id='d94ec6d9' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='288'>
	- <var-decl name='__pad0' type-id='95e97e5e' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='320'>
	- <var-decl name='st_rdev' type-id='35ed8932' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='384'>
	- <var-decl name='st_size' type-id='79989e9c' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='448'>
	- <var-decl name='st_blksize' type-id='d3f10a7f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='512'>
	- <var-decl name='st_blocks' type-id='4e711bf1' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='576'>
	- <var-decl name='st_atim' type-id='a9c79a1f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='704'>
	- <var-decl name='st_mtim' type-id='a9c79a1f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='832'>
	- <var-decl name='st_ctim' type-id='a9c79a1f' visibility='default'/>
	- </data-member>
	- <data-member access='public' layout-offset-in-bits='960'>
	- <var-decl name='__glibc_reserved' type-id='083f8d58' visibility='default'/>
	+ <var-decl name='_pos' type-id='95e97e5e' visibility='default'/>
	</data-member>
	</class-decl>
	- <typedef-decl name='__dev_t' type-id='7359adad' id='35ed8932'/>
	- <typedef-decl name='__uid_t' type-id='f0981eeb' id='cc5fcceb'/>
	- <typedef-decl name='__gid_t' type-id='f0981eeb' id='d94ec6d9'/>
	- <typedef-decl name='__ino64_t' type-id='7359adad' id='71288a47'/>
	- <typedef-decl name='__nlink_t' type-id='7359adad' id='80f0b9df'/>
	- <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
	- <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
	- <typedef-decl name='__time_t' type-id='bd54fe1a' id='65eda9c0'/>
	- <typedef-decl name='__blksize_t' type-id='bd54fe1a' id='d3f10a7f'/>
	- <typedef-decl name='__blkcnt64_t' type-id='bd54fe1a' id='4e711bf1'/>
	- <typedef-decl name='__syscall_slong_t' type-id='bd54fe1a' id='03085adc'/>
	- <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
	- <typedef-decl name='_IO_lock_t' type-id='48b5725f' id='bb4788fa'/>
	<class-decl name='_IO_FILE' size-in-bits='1728' is-struct='yes' visibility='default' id='ec1ed955'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='_flags' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='_IO_read_ptr' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='_IO_read_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='_IO_read_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='_IO_write_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='_IO_write_ptr' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='_IO_write_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='_IO_buf_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='_IO_buf_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='576'>
	<var-decl name='_IO_save_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='640'>
	<var-decl name='_IO_backup_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='704'>
	<var-decl name='_IO_save_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='768'>
	<var-decl name='_markers' type-id='e4c6fa61' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='832'>
	<var-decl name='_chain' type-id='dca988a5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='896'>
	<var-decl name='_fileno' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='928'>
	<var-decl name='_flags2' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='960'>
	<var-decl name='_old_offset' type-id='79989e9c' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1024'>
	<var-decl name='_cur_column' type-id='8efea9e5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1040'>
	<var-decl name='_vtable_offset' type-id='28577a57' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1048'>
	<var-decl name='_shortbuf' type-id='89feb1ec' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1088'>
	<var-decl name='_lock' type-id='cecf4ea7' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1152'>
	<var-decl name='_offset' type-id='724e4de6' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1216'>
	- <var-decl name='_codecvt' type-id='570f8c59' visibility='default'/>
	+ <var-decl name='__pad1' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1280'>
	- <var-decl name='_wide_data' type-id='c65a1f29' visibility='default'/>
	+ <var-decl name='__pad2' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1344'>
	- <var-decl name='_freeres_list' type-id='dca988a5' visibility='default'/>
	+ <var-decl name='__pad3' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1408'>
	- <var-decl name='_freeres_buf' type-id='eaa32e2f' visibility='default'/>
	+ <var-decl name='__pad4' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1472'>
	<var-decl name='__pad5' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1536'>
	<var-decl name='_mode' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1568'>
	<var-decl name='_unused2' type-id='664ac0b7' visibility='default'/>
	</data-member>
	</class-decl>
	+ <class-decl name='stat64' size-in-bits='1152' is-struct='yes' visibility='default' id='0bbec9cd'>
	+ <data-member access='public' layout-offset-in-bits='0'>
	+ <var-decl name='st_dev' type-id='35ed8932' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='64'>
	+ <var-decl name='st_ino' type-id='71288a47' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='128'>
	+ <var-decl name='st_nlink' type-id='80f0b9df' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='192'>
	+ <var-decl name='st_mode' type-id='e1c52942' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='224'>
	+ <var-decl name='st_uid' type-id='cc5fcceb' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='256'>
	+ <var-decl name='st_gid' type-id='d94ec6d9' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='288'>
	+ <var-decl name='__pad0' type-id='95e97e5e' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='320'>
	+ <var-decl name='st_rdev' type-id='35ed8932' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='384'>
	+ <var-decl name='st_size' type-id='79989e9c' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='448'>
	+ <var-decl name='st_blksize' type-id='d3f10a7f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='512'>
	+ <var-decl name='st_blocks' type-id='4e711bf1' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='576'>
	+ <var-decl name='st_atim' type-id='a9c79a1f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='704'>
	+ <var-decl name='st_mtim' type-id='a9c79a1f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='832'>
	+ <var-decl name='st_ctim' type-id='a9c79a1f' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='960'>
	+ <var-decl name='__glibc_reserved' type-id='083f8d58' visibility='default'/>
	+ </data-member>
	+ </class-decl>
	+ <typedef-decl name='__dev_t' type-id='7359adad' id='35ed8932'/>
	+ <typedef-decl name='__uid_t' type-id='f0981eeb' id='cc5fcceb'/>
	+ <typedef-decl name='__gid_t' type-id='f0981eeb' id='d94ec6d9'/>
	+ <typedef-decl name='__ino64_t' type-id='7359adad' id='71288a47'/>
	+ <typedef-decl name='__nlink_t' type-id='7359adad' id='80f0b9df'/>
	+ <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
	+ <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
	+ <typedef-decl name='__time_t' type-id='bd54fe1a' id='65eda9c0'/>
	+ <typedef-decl name='__blksize_t' type-id='bd54fe1a' id='d3f10a7f'/>
	+ <typedef-decl name='__blkcnt64_t' type-id='bd54fe1a' id='4e711bf1'/>
	+ <typedef-decl name='__syscall_slong_t' type-id='bd54fe1a' id='03085adc'/>
	+ <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
	<class-decl name='timespec' size-in-bits='128' is-struct='yes' visibility='default' id='a9c79a1f'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='tv_sec' type-id='65eda9c0' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='tv_nsec' type-id='03085adc' visibility='default'/>
	</data-member>
	</class-decl>
	<pointer-type-def type-id='aa12d1ba' size-in-bits='64' id='822cd80b'/>
	<pointer-type-def type-id='ec1ed955' size-in-bits='64' id='dca988a5'/>
	- <pointer-type-def type-id='a4036571' size-in-bits='64' id='570f8c59'/>
	<pointer-type-def type-id='bb4788fa' size-in-bits='64' id='cecf4ea7'/>
	<pointer-type-def type-id='010ae0b9' size-in-bits='64' id='e4c6fa61'/>
	- <pointer-type-def type-id='79bd3751' size-in-bits='64' id='c65a1f29'/>
	<pointer-type-def type-id='0c544dc0' size-in-bits='64' id='394fc496'/>
	<pointer-type-def type-id='1b055409' size-in-bits='64' id='9d424d31'/>
	<pointer-type-def type-id='0bbec9cd' size-in-bits='64' id='62f7a03d'/>
	- <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
	- <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
	- <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
	<function-decl name='getmntany' mangled-name='getmntany' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getmntany'>
	<parameter type-id='822cd80b' name='fp'/>
	<parameter type-id='9d424d31' name='mgetp'/>
	<parameter type-id='9d424d31' name='mrefp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='_sol_getmntent' mangled-name='_sol_getmntent' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='_sol_getmntent'>
	<parameter type-id='822cd80b' name='fp'/>
	<parameter type-id='9d424d31' name='mgetp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='getextmntent' mangled-name='getextmntent' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getextmntent'>
	<parameter type-id='80f4b756' name='path'/>
	<parameter type-id='394fc496' name='entry'/>
	<parameter type-id='62f7a03d' name='statbuf'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/zone.c' language='LANG_C99'>
	<typedef-decl name='zoneid_t' type-id='95e97e5e' id='4da03624'/>
	<function-decl name='getzoneid' mangled-name='getzoneid' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='getzoneid'>
	<return type-id='4da03624'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='page.c' language='LANG_C99'>
	<function-decl name='spl_pagesize' mangled-name='spl_pagesize' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='spl_pagesize'>
	<return type-id='b59d7dce'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='strlcat.c' language='LANG_C99'>
	<function-decl name='strlcat' mangled-name='strlcat' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='strlcat'>
	<parameter type-id='26a90f95' name='dst'/>
	<parameter type-id='80f4b756' name='src'/>
	<parameter type-id='b59d7dce' name='dstsize'/>
	<return type-id='b59d7dce'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='strlcpy.c' language='LANG_C99'>
	<function-decl name='strlcpy' mangled-name='strlcpy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='strlcpy'>
	<parameter type-id='26a90f95' name='dst'/>
	<parameter type-id='80f4b756' name='src'/>
	<parameter type-id='b59d7dce' name='len'/>
	<return type-id='b59d7dce'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='timestamp.c' language='LANG_C99'>
	<function-decl name='print_timestamp' mangled-name='print_timestamp' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='print_timestamp'>
	<parameter type-id='3502e3ff' name='timestamp_fmt'/>
	<return type-id='48b5725f'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='libzfs_core.c' language='LANG_C99'>
	<type-decl name='char' size-in-bits='8' id='a84c031d'/>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='2048' id='d1617432'>
	<subrange length='256' type-id='7359adad' id='36e5b9fa'/>
	</array-type-def>
	<type-decl name='int' size-in-bits='32' id='95e97e5e'/>
	<array-type-def dimensions='1' type-id='9c313c2d' size-in-bits='2176' id='8c2bcad1'>
	<subrange length='34' type-id='7359adad' id='6a6a7e00'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='9c313c2d' size-in-bits='256' id='85c64d26'>
	<subrange length='4' type-id='7359adad' id='16fe7105'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='b96825af' size-in-bits='96' id='fa8ef949'>
	<subrange length='12' type-id='7359adad' id='84827bdc'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='b96825af' size-in-bits='128' id='fa9986a5'>
	<subrange length='16' type-id='7359adad' id='848d0938'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='b96825af' size-in-bits='24' id='d3490169'>
	<subrange length='3' type-id='7359adad' id='56f209d2'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='b96825af' size-in-bits='40' id='0f4ddd0b'>
	<subrange length='5' type-id='7359adad' id='53010e10'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='b96825af' size-in-bits='48' id='0f562bd0'>
	<subrange length='6' type-id='7359adad' id='52fa524b'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='b96825af' size-in-bits='64' id='13339fda'>
	<subrange length='8' type-id='7359adad' id='56e0c0b1'/>
	</array-type-def>
	<type-decl name='unnamed-enum-underlying-type-32' is-anonymous='yes' size-in-bits='32' alignment-in-bits='32' id='9cac1fee'/>
	<type-decl name='unsigned char' size-in-bits='8' id='002ac4a6'/>
	<type-decl name='unsigned int' size-in-bits='32' id='f0981eeb'/>
	<type-decl name='unsigned long int' size-in-bits='64' id='7359adad'/>
	- <type-decl name='void' id='48b5725f'/>
	<enum-decl name='lzc_dataset_type' id='bc9887f1'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='LZC_DATSET_TYPE_ZFS' value='2'/>
	<enumerator name='LZC_DATSET_TYPE_ZVOL' value='3'/>
	</enum-decl>
	<enum-decl name='lzc_send_flags' id='bfbd3c8e'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='LZC_SEND_FLAG_EMBED_DATA' value='1'/>
	<enumerator name='LZC_SEND_FLAG_LARGE_BLOCK' value='2'/>
	<enumerator name='LZC_SEND_FLAG_COMPRESS' value='4'/>
	<enumerator name='LZC_SEND_FLAG_RAW' value='8'/>
	<enumerator name='LZC_SEND_FLAG_SAVED' value='16'/>
	</enum-decl>
	<class-decl name='ddt_key' size-in-bits='320' is-struct='yes' visibility='default' id='e0a4a1cb'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='ddk_cksum' type-id='39730d0b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='ddk_prop' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='ddt_key_t' type-id='e0a4a1cb' id='67f6d2cf'/>
	<enum-decl name='dmu_object_type' id='04b3b0b9'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='DMU_OT_NONE' value='0'/>
	<enumerator name='DMU_OT_OBJECT_DIRECTORY' value='1'/>
	<enumerator name='DMU_OT_OBJECT_ARRAY' value='2'/>
	<enumerator name='DMU_OT_PACKED_NVLIST' value='3'/>
	<enumerator name='DMU_OT_PACKED_NVLIST_SIZE' value='4'/>
	<enumerator name='DMU_OT_BPOBJ' value='5'/>
	<enumerator name='DMU_OT_BPOBJ_HDR' value='6'/>
	<enumerator name='DMU_OT_SPACE_MAP_HEADER' value='7'/>
	<enumerator name='DMU_OT_SPACE_MAP' value='8'/>
	<enumerator name='DMU_OT_INTENT_LOG' value='9'/>
	<enumerator name='DMU_OT_DNODE' value='10'/>
	<enumerator name='DMU_OT_OBJSET' value='11'/>
	<enumerator name='DMU_OT_DSL_DIR' value='12'/>
	<enumerator name='DMU_OT_DSL_DIR_CHILD_MAP' value='13'/>
	<enumerator name='DMU_OT_DSL_DS_SNAP_MAP' value='14'/>
	<enumerator name='DMU_OT_DSL_PROPS' value='15'/>
	<enumerator name='DMU_OT_DSL_DATASET' value='16'/>
	<enumerator name='DMU_OT_ZNODE' value='17'/>
	<enumerator name='DMU_OT_OLDACL' value='18'/>
	<enumerator name='DMU_OT_PLAIN_FILE_CONTENTS' value='19'/>
	<enumerator name='DMU_OT_DIRECTORY_CONTENTS' value='20'/>
	<enumerator name='DMU_OT_MASTER_NODE' value='21'/>
	<enumerator name='DMU_OT_UNLINKED_SET' value='22'/>
	<enumerator name='DMU_OT_ZVOL' value='23'/>
	<enumerator name='DMU_OT_ZVOL_PROP' value='24'/>
	<enumerator name='DMU_OT_PLAIN_OTHER' value='25'/>
	<enumerator name='DMU_OT_UINT64_OTHER' value='26'/>
	<enumerator name='DMU_OT_ZAP_OTHER' value='27'/>
	<enumerator name='DMU_OT_ERROR_LOG' value='28'/>
	<enumerator name='DMU_OT_SPA_HISTORY' value='29'/>
	<enumerator name='DMU_OT_SPA_HISTORY_OFFSETS' value='30'/>
	<enumerator name='DMU_OT_POOL_PROPS' value='31'/>
	<enumerator name='DMU_OT_DSL_PERMS' value='32'/>
	<enumerator name='DMU_OT_ACL' value='33'/>
	<enumerator name='DMU_OT_SYSACL' value='34'/>
	<enumerator name='DMU_OT_FUID' value='35'/>
	<enumerator name='DMU_OT_FUID_SIZE' value='36'/>
	<enumerator name='DMU_OT_NEXT_CLONES' value='37'/>
	<enumerator name='DMU_OT_SCAN_QUEUE' value='38'/>
	<enumerator name='DMU_OT_USERGROUP_USED' value='39'/>
	<enumerator name='DMU_OT_USERGROUP_QUOTA' value='40'/>
	<enumerator name='DMU_OT_USERREFS' value='41'/>
	<enumerator name='DMU_OT_DDT_ZAP' value='42'/>
	<enumerator name='DMU_OT_DDT_STATS' value='43'/>
	<enumerator name='DMU_OT_SA' value='44'/>
	<enumerator name='DMU_OT_SA_MASTER_NODE' value='45'/>
	<enumerator name='DMU_OT_SA_ATTR_REGISTRATION' value='46'/>
	<enumerator name='DMU_OT_SA_ATTR_LAYOUTS' value='47'/>
	<enumerator name='DMU_OT_SCAN_XLATE' value='48'/>
	<enumerator name='DMU_OT_DEDUP' value='49'/>
	<enumerator name='DMU_OT_DEADLIST' value='50'/>
	<enumerator name='DMU_OT_DEADLIST_HDR' value='51'/>
	<enumerator name='DMU_OT_DSL_CLONES' value='52'/>
	<enumerator name='DMU_OT_BPOBJ_SUBOBJ' value='53'/>
	<enumerator name='DMU_OT_NUMTYPES' value='54'/>
	<enumerator name='DMU_OTN_UINT8_DATA' value='128'/>
	<enumerator name='DMU_OTN_UINT8_METADATA' value='192'/>
	<enumerator name='DMU_OTN_UINT16_DATA' value='129'/>
	<enumerator name='DMU_OTN_UINT16_METADATA' value='193'/>
	<enumerator name='DMU_OTN_UINT32_DATA' value='130'/>
	<enumerator name='DMU_OTN_UINT32_METADATA' value='194'/>
	<enumerator name='DMU_OTN_UINT64_DATA' value='131'/>
	<enumerator name='DMU_OTN_UINT64_METADATA' value='195'/>
	<enumerator name='DMU_OTN_ZAP_DATA' value='132'/>
	<enumerator name='DMU_OTN_ZAP_METADATA' value='196'/>
	<enumerator name='DMU_OTN_UINT8_ENC_DATA' value='160'/>
	<enumerator name='DMU_OTN_UINT8_ENC_METADATA' value='224'/>
	<enumerator name='DMU_OTN_UINT16_ENC_DATA' value='161'/>
	<enumerator name='DMU_OTN_UINT16_ENC_METADATA' value='225'/>
	<enumerator name='DMU_OTN_UINT32_ENC_DATA' value='162'/>
	<enumerator name='DMU_OTN_UINT32_ENC_METADATA' value='226'/>
	<enumerator name='DMU_OTN_UINT64_ENC_DATA' value='163'/>
	<enumerator name='DMU_OTN_UINT64_ENC_METADATA' value='227'/>
	<enumerator name='DMU_OTN_ZAP_ENC_DATA' value='164'/>
	<enumerator name='DMU_OTN_ZAP_ENC_METADATA' value='228'/>
	</enum-decl>
	<typedef-decl name='dmu_object_type_t' type-id='04b3b0b9' id='5c9d8906'/>
	<enum-decl name='dmu_objset_type' id='6b1b19f9'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='DMU_OST_NONE' value='0'/>
	<enumerator name='DMU_OST_META' value='1'/>
	<enumerator name='DMU_OST_ZFS' value='2'/>
	<enumerator name='DMU_OST_ZVOL' value='3'/>
	<enumerator name='DMU_OST_OTHER' value='4'/>
	<enumerator name='DMU_OST_ANY' value='5'/>
	<enumerator name='DMU_OST_NUMTYPES' value='6'/>
	</enum-decl>
	<typedef-decl name='dmu_objset_type_t' type-id='6b1b19f9' id='230f1e16'/>
	<enum-decl name='pool_initialize_func' id='5c246ad4'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='POOL_INITIALIZE_START' value='0'/>
	<enumerator name='POOL_INITIALIZE_CANCEL' value='1'/>
	<enumerator name='POOL_INITIALIZE_SUSPEND' value='2'/>
	<enumerator name='POOL_INITIALIZE_FUNCS' value='3'/>
	</enum-decl>
	<typedef-decl name='pool_initialize_func_t' type-id='5c246ad4' id='7063e1ab'/>
	<enum-decl name='pool_trim_func' id='54ed608a'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='POOL_TRIM_START' value='0'/>
	<enumerator name='POOL_TRIM_CANCEL' value='1'/>
	<enumerator name='POOL_TRIM_SUSPEND' value='2'/>
	<enumerator name='POOL_TRIM_FUNCS' value='3'/>
	</enum-decl>
	<typedef-decl name='pool_trim_func_t' type-id='54ed608a' id='b1146b8d'/>
	<enum-decl name='zpool_wait_activity_t' naming-typedef-id='73446457' id='849338e3'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZPOOL_WAIT_CKPT_DISCARD' value='0'/>
	<enumerator name='ZPOOL_WAIT_FREE' value='1'/>
	<enumerator name='ZPOOL_WAIT_INITIALIZE' value='2'/>
	<enumerator name='ZPOOL_WAIT_REPLACE' value='3'/>
	<enumerator name='ZPOOL_WAIT_REMOVE' value='4'/>
	<enumerator name='ZPOOL_WAIT_RESILVER' value='5'/>
	<enumerator name='ZPOOL_WAIT_SCRUB' value='6'/>
	<enumerator name='ZPOOL_WAIT_TRIM' value='7'/>
	<enumerator name='ZPOOL_WAIT_NUM_ACTIVITIES' value='8'/>
	</enum-decl>
	<typedef-decl name='zpool_wait_activity_t' type-id='849338e3' id='73446457'/>
	<enum-decl name='zfs_wait_activity_t' naming-typedef-id='3024501a' id='527d5dc6'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='ZFS_WAIT_DELETEQ' value='0'/>
	<enumerator name='ZFS_WAIT_NUM_ACTIVITIES' value='1'/>
	</enum-decl>
	<typedef-decl name='zfs_wait_activity_t' type-id='527d5dc6' id='3024501a'/>
	<class-decl name='nvlist' size-in-bits='192' is-struct='yes' visibility='default' id='ac266fd9'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='nvl_version' type-id='3ff5601b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='nvl_nvflag' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='nvl_priv' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='nvl_flag' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='160'>
	<var-decl name='nvl_pad' type-id='3ff5601b' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='nvlist_t' type-id='ac266fd9' id='8e8d4be3'/>
	<class-decl name='zio_cksum' size-in-bits='256' is-struct='yes' visibility='default' id='1d53e28b'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='zc_word' type-id='85c64d26' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zio_cksum_t' type-id='1d53e28b' id='39730d0b'/>
	<class-decl name='drr_begin' size-in-bits='2432' is-struct='yes' visibility='default' id='09fcdc01'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_magic' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_versioninfo' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='drr_creation_time' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='drr_type' type-id='230f1e16' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='224'>
	<var-decl name='drr_flags' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='drr_fromguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='drr_toname' type-id='d1617432' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='dmu_replay_record' size-in-bits='2496' is-struct='yes' visibility='default' id='781a52d7'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_type' type-id='08f5ca17' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32'>
	<var-decl name='drr_payloadlen' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_u' type-id='ac5ab595' visibility='default'/>
	</data-member>
	</class-decl>
	<enum-decl name='__anonymous_enum__' is-anonymous='yes' id='08f5ca17'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='DRR_BEGIN' value='0'/>
	<enumerator name='DRR_OBJECT' value='1'/>
	<enumerator name='DRR_FREEOBJECTS' value='2'/>
	<enumerator name='DRR_WRITE' value='3'/>
	<enumerator name='DRR_FREE' value='4'/>
	<enumerator name='DRR_END' value='5'/>
	<enumerator name='DRR_WRITE_BYREF' value='6'/>
	<enumerator name='DRR_SPILL' value='7'/>
	<enumerator name='DRR_WRITE_EMBEDDED' value='8'/>
	<enumerator name='DRR_OBJECT_RANGE' value='9'/>
	<enumerator name='DRR_REDACT' value='10'/>
	<enumerator name='DRR_NUMTYPES' value='11'/>
	</enum-decl>
	<union-decl name='__anonymous_union__' size-in-bits='2432' is-anonymous='yes' visibility='default' id='ac5ab595'>
	<data-member access='public'>
	<var-decl name='drr_begin' type-id='09fcdc01' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='drr_end' type-id='6ee25631' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='drr_object' type-id='f9ad530b' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='drr_freeobjects' type-id='a27d958e' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='drr_write' type-id='4cc69e4b' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='drr_free' type-id='c836cfd2' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='drr_write_byref' type-id='e511cdce' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='drr_spill' type-id='1e69a80a' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='drr_write_embedded' type-id='98b1345e' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='drr_object_range' type-id='aba1f9e1' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='drr_redact' type-id='50389039' visibility='default'/>
	</data-member>
	<data-member access='public'>
	<var-decl name='drr_checksum' type-id='a5fe3647' visibility='default'/>
	</data-member>
	</union-decl>
	<class-decl name='drr_end' size-in-bits='320' is-struct='yes' visibility='default' id='6ee25631'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_checksum' type-id='39730d0b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='drr_object' size-in-bits='448' is-struct='yes' visibility='default' id='f9ad530b'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_object' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_type' type-id='5c9d8906' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='96'>
	<var-decl name='drr_bonustype' type-id='5c9d8906' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='drr_blksz' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='160'>
	<var-decl name='drr_bonuslen' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='drr_checksumtype' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='200'>
	<var-decl name='drr_compress' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='208'>
	<var-decl name='drr_dn_slots' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='216'>
	<var-decl name='drr_flags' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='224'>
	<var-decl name='drr_raw_bonuslen' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='drr_indblkshift' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='328'>
	<var-decl name='drr_nlevels' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='336'>
	<var-decl name='drr_nblkptr' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='344'>
	<var-decl name='drr_pad' type-id='0f4ddd0b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='drr_maxblkid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='drr_freeobjects' size-in-bits='192' is-struct='yes' visibility='default' id='a27d958e'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_firstobj' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_numobjs' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='drr_write' size-in-bits='1088' is-struct='yes' visibility='default' id='4cc69e4b'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_object' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_type' type-id='5c9d8906' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='96'>
	<var-decl name='drr_pad' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='drr_offset' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='drr_logical_size' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='drr_checksumtype' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='328'>
	<var-decl name='drr_flags' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='336'>
	<var-decl name='drr_compressiontype' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='344'>
	<var-decl name='drr_pad2' type-id='0f4ddd0b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='drr_key' type-id='67f6d2cf' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='704'>
	<var-decl name='drr_compressed_size' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='768'>
	<var-decl name='drr_salt' type-id='13339fda' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='832'>
	<var-decl name='drr_iv' type-id='fa8ef949' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='928'>
	<var-decl name='drr_mac' type-id='fa9986a5' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='drr_free' size-in-bits='256' is-struct='yes' visibility='default' id='c836cfd2'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_object' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_offset' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='drr_length' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='drr_write_byref' size-in-bits='832' is-struct='yes' visibility='default' id='e511cdce'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_object' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_offset' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='drr_length' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='drr_refguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='drr_refobject' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='drr_refoffset' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='drr_checksumtype' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='456'>
	<var-decl name='drr_flags' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='464'>
	<var-decl name='drr_pad2' type-id='0f562bd0' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='drr_key' type-id='67f6d2cf' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='drr_spill' size-in-bits='640' is-struct='yes' visibility='default' id='1e69a80a'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_object' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_length' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='drr_flags' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='200'>
	<var-decl name='drr_compressiontype' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='208'>
	<var-decl name='drr_pad' type-id='0f562bd0' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='drr_compressed_size' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='drr_salt' type-id='13339fda' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='drr_iv' type-id='fa8ef949' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='480'>
	<var-decl name='drr_mac' type-id='fa9986a5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='608'>
	<var-decl name='drr_type' type-id='5c9d8906' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='drr_write_embedded' size-in-bits='384' is-struct='yes' visibility='default' id='98b1345e'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_object' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_offset' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='drr_length' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='drr_compression' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='264'>
	<var-decl name='drr_etype' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='272'>
	<var-decl name='drr_pad' type-id='0f562bd0' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='drr_lsize' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='352'>
	<var-decl name='drr_psize' type-id='8f92235e' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='drr_object_range' size-in-bits='512' is-struct='yes' visibility='default' id='aba1f9e1'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_firstobj' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_numslots' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='drr_salt' type-id='13339fda' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='drr_iv' type-id='fa8ef949' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='352'>
	<var-decl name='drr_mac' type-id='fa9986a5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='480'>
	<var-decl name='drr_flags' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='488'>
	<var-decl name='drr_pad' type-id='d3490169' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='drr_redact' size-in-bits='256' is-struct='yes' visibility='default' id='50389039'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_object' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='drr_offset' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='drr_length' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='drr_toguid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<class-decl name='drr_checksum' size-in-bits='2432' is-struct='yes' visibility='default' id='a5fe3647'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='drr_pad' type-id='8c2bcad1' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2176'>
	<var-decl name='drr_checksum' type-id='39730d0b' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='dmu_replay_record_t' type-id='781a52d7' id='8b8fc893'/>
	<enum-decl name='boolean_t' naming-typedef-id='c19b74c3' id='f58c8277'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='B_FALSE' value='0'/>
	<enumerator name='B_TRUE' value='1'/>
	</enum-decl>
	<typedef-decl name='boolean_t' type-id='f58c8277' id='c19b74c3'/>
	<typedef-decl name='uint_t' type-id='f0981eeb' id='3502e3ff'/>
	<typedef-decl name='int32_t' type-id='33f57a65' id='3ff5601b'/>
	<typedef-decl name='uint8_t' type-id='c51d6389' id='b96825af'/>
	<typedef-decl name='uint32_t' type-id='62f1140c' id='8f92235e'/>
	<typedef-decl name='uint64_t' type-id='8910171f' id='9c313c2d'/>
	<typedef-decl name='__uint8_t' type-id='002ac4a6' id='c51d6389'/>
	<typedef-decl name='__int32_t' type-id='95e97e5e' id='33f57a65'/>
	<typedef-decl name='__uint32_t' type-id='f0981eeb' id='62f1140c'/>
	<typedef-decl name='__uint64_t' type-id='7359adad' id='8910171f'/>
	<pointer-type-def type-id='c19b74c3' size-in-bits='64' id='37e3bd22'/>
	<pointer-type-def type-id='a84c031d' size-in-bits='64' id='26a90f95'/>
	<qualified-type-def type-id='a84c031d' const='yes' id='9b45d938'/>
	<pointer-type-def type-id='9b45d938' size-in-bits='64' id='80f4b756'/>
	<qualified-type-def type-id='8b8fc893' const='yes' id='9623bc03'/>
	<pointer-type-def type-id='9623bc03' size-in-bits='64' id='8341348b'/>
	<qualified-type-def type-id='8e8d4be3' const='yes' id='693c3853'/>
	<pointer-type-def type-id='693c3853' size-in-bits='64' id='22cce67b'/>
	<pointer-type-def type-id='8e8d4be3' size-in-bits='64' id='5ce45b60'/>
	<pointer-type-def type-id='5ce45b60' size-in-bits='64' id='857bb57e'/>
	<pointer-type-def type-id='9c313c2d' size-in-bits='64' id='5d6479ae'/>
	<pointer-type-def type-id='b96825af' size-in-bits='64' id='ae3e8ca6'/>
	<function-decl name='libzfs_core_init' mangled-name='libzfs_core_init' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_core_init'>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='libzfs_core_fini' mangled-name='libzfs_core_fini' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='libzfs_core_fini'>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='lzc_create' mangled-name='lzc_create' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_create'>
	<parameter type-id='80f4b756' name='fsname'/>
	<parameter type-id='bc9887f1' name='type'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='ae3e8ca6' name='wkeydata'/>
	<parameter type-id='3502e3ff' name='wkeylen'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_clone' mangled-name='lzc_clone' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_clone'>
	<parameter type-id='80f4b756' name='fsname'/>
	<parameter type-id='80f4b756' name='origin'/>
	<parameter type-id='5ce45b60' name='props'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_promote' mangled-name='lzc_promote' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_promote'>
	<parameter type-id='80f4b756' name='fsname'/>
	<parameter type-id='26a90f95' name='snapnamebuf'/>
	<parameter type-id='95e97e5e' name='snapnamelen'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_rename' mangled-name='lzc_rename' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_rename'>
	<parameter type-id='80f4b756' name='source'/>
	<parameter type-id='80f4b756' name='target'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_destroy' mangled-name='lzc_destroy' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_destroy'>
	<parameter type-id='80f4b756' name='fsname'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_snapshot' mangled-name='lzc_snapshot' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_snapshot'>
	<parameter type-id='5ce45b60' name='snaps'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='857bb57e' name='errlist'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_destroy_snaps' mangled-name='lzc_destroy_snaps' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_destroy_snaps'>
	<parameter type-id='5ce45b60' name='snaps'/>
	<parameter type-id='c19b74c3' name='defer'/>
	<parameter type-id='857bb57e' name='errlist'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_snaprange_space' mangled-name='lzc_snaprange_space' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_snaprange_space'>
	<parameter type-id='80f4b756' name='firstsnap'/>
	<parameter type-id='80f4b756' name='lastsnap'/>
	<parameter type-id='5d6479ae' name='usedp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_exists' mangled-name='lzc_exists' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_exists'>
	<parameter type-id='80f4b756' name='dataset'/>
	<return type-id='c19b74c3'/>
	</function-decl>
	<function-decl name='lzc_sync' mangled-name='lzc_sync' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_sync'>
	<parameter type-id='80f4b756' name='pool_name'/>
	<parameter type-id='5ce45b60' name='innvl'/>
	<parameter type-id='857bb57e' name='outnvl'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_hold' mangled-name='lzc_hold' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_hold'>
	<parameter type-id='5ce45b60' name='holds'/>
	<parameter type-id='95e97e5e' name='cleanup_fd'/>
	<parameter type-id='857bb57e' name='errlist'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_release' mangled-name='lzc_release' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_release'>
	<parameter type-id='5ce45b60' name='holds'/>
	<parameter type-id='857bb57e' name='errlist'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_get_holds' mangled-name='lzc_get_holds' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_get_holds'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='857bb57e' name='holdsp'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_send' mangled-name='lzc_send' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_send'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='80f4b756' name='from'/>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='bfbd3c8e' name='flags'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_send_redacted' mangled-name='lzc_send_redacted' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_send_redacted'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='80f4b756' name='from'/>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='bfbd3c8e' name='flags'/>
	<parameter type-id='80f4b756' name='redactbook'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_send_resume' mangled-name='lzc_send_resume' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_send_resume'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='80f4b756' name='from'/>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='bfbd3c8e' name='flags'/>
	<parameter type-id='9c313c2d' name='resumeobj'/>
	<parameter type-id='9c313c2d' name='resumeoff'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_send_resume_redacted' mangled-name='lzc_send_resume_redacted' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_send_resume_redacted'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='80f4b756' name='from'/>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='bfbd3c8e' name='flags'/>
	<parameter type-id='9c313c2d' name='resumeobj'/>
	<parameter type-id='9c313c2d' name='resumeoff'/>
	<parameter type-id='80f4b756' name='redactbook'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_send_space_resume_redacted' mangled-name='lzc_send_space_resume_redacted' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_send_space_resume_redacted'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='80f4b756' name='from'/>
	<parameter type-id='bfbd3c8e' name='flags'/>
	<parameter type-id='9c313c2d' name='resumeobj'/>
	<parameter type-id='9c313c2d' name='resumeoff'/>
	<parameter type-id='9c313c2d' name='resume_bytes'/>
	<parameter type-id='80f4b756' name='redactbook'/>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='5d6479ae' name='spacep'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_send_space' mangled-name='lzc_send_space' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_send_space'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='80f4b756' name='from'/>
	<parameter type-id='bfbd3c8e' name='flags'/>
	<parameter type-id='5d6479ae' name='spacep'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_receive' mangled-name='lzc_receive' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_receive'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='80f4b756' name='origin'/>
	<parameter type-id='c19b74c3' name='force'/>
	<parameter type-id='c19b74c3' name='raw'/>
	<parameter type-id='95e97e5e' name='fd'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_receive_resumable' mangled-name='lzc_receive_resumable' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_receive_resumable'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='80f4b756' name='origin'/>
	<parameter type-id='c19b74c3' name='force'/>
	<parameter type-id='c19b74c3' name='raw'/>
	<parameter type-id='95e97e5e' name='fd'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_receive_with_header' mangled-name='lzc_receive_with_header' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_receive_with_header'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='80f4b756' name='origin'/>
	<parameter type-id='c19b74c3' name='force'/>
	<parameter type-id='c19b74c3' name='resumable'/>
	<parameter type-id='c19b74c3' name='raw'/>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='8341348b' name='begin_record'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_receive_one' mangled-name='lzc_receive_one' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_receive_one'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='80f4b756' name='origin'/>
	<parameter type-id='c19b74c3' name='force'/>
	<parameter type-id='c19b74c3' name='resumable'/>
	<parameter type-id='c19b74c3' name='raw'/>
	<parameter type-id='95e97e5e' name='input_fd'/>
	<parameter type-id='8341348b' name='begin_record'/>
	<parameter type-id='95e97e5e' name='cleanup_fd'/>
	<parameter type-id='5d6479ae' name='read_bytes'/>
	<parameter type-id='5d6479ae' name='errflags'/>
	<parameter type-id='5d6479ae' name='action_handle'/>
	<parameter type-id='857bb57e' name='errors'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_receive_with_cmdprops' mangled-name='lzc_receive_with_cmdprops' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_receive_with_cmdprops'>
	<parameter type-id='80f4b756' name='snapname'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='5ce45b60' name='cmdprops'/>
	<parameter type-id='ae3e8ca6' name='wkeydata'/>
	<parameter type-id='3502e3ff' name='wkeylen'/>
	<parameter type-id='80f4b756' name='origin'/>
	<parameter type-id='c19b74c3' name='force'/>
	<parameter type-id='c19b74c3' name='resumable'/>
	<parameter type-id='c19b74c3' name='raw'/>
	<parameter type-id='95e97e5e' name='input_fd'/>
	<parameter type-id='8341348b' name='begin_record'/>
	<parameter type-id='95e97e5e' name='cleanup_fd'/>
	<parameter type-id='5d6479ae' name='read_bytes'/>
	<parameter type-id='5d6479ae' name='errflags'/>
	<parameter type-id='5d6479ae' name='action_handle'/>
	<parameter type-id='857bb57e' name='errors'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_rollback' mangled-name='lzc_rollback' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_rollback'>
	<parameter type-id='80f4b756' name='fsname'/>
	<parameter type-id='26a90f95' name='snapnamebuf'/>
	<parameter type-id='95e97e5e' name='snapnamelen'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_rollback_to' mangled-name='lzc_rollback_to' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_rollback_to'>
	<parameter type-id='80f4b756' name='fsname'/>
	<parameter type-id='80f4b756' name='snapname'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_bookmark' mangled-name='lzc_bookmark' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_bookmark'>
	<parameter type-id='5ce45b60' name='bookmarks'/>
	<parameter type-id='857bb57e' name='errlist'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_get_bookmarks' mangled-name='lzc_get_bookmarks' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_get_bookmarks'>
	<parameter type-id='80f4b756' name='fsname'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='857bb57e' name='bmarks'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_get_bookmark_props' mangled-name='lzc_get_bookmark_props' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_get_bookmark_props'>
	<parameter type-id='80f4b756' name='bookmark'/>
	<parameter type-id='857bb57e' name='props'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_destroy_bookmarks' mangled-name='lzc_destroy_bookmarks' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_destroy_bookmarks'>
	<parameter type-id='5ce45b60' name='bmarks'/>
	<parameter type-id='857bb57e' name='errlist'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_channel_program' mangled-name='lzc_channel_program' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_channel_program'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='80f4b756' name='program'/>
	<parameter type-id='9c313c2d' name='instrlimit'/>
	<parameter type-id='9c313c2d' name='memlimit'/>
	<parameter type-id='5ce45b60' name='argnvl'/>
	<parameter type-id='857bb57e' name='outnvl'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_pool_checkpoint' mangled-name='lzc_pool_checkpoint' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_pool_checkpoint'>
	<parameter type-id='80f4b756' name='pool'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_pool_checkpoint_discard' mangled-name='lzc_pool_checkpoint_discard' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_pool_checkpoint_discard'>
	<parameter type-id='80f4b756' name='pool'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_channel_program_nosync' mangled-name='lzc_channel_program_nosync' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_channel_program_nosync'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='80f4b756' name='program'/>
	<parameter type-id='9c313c2d' name='timeout'/>
	<parameter type-id='9c313c2d' name='memlimit'/>
	<parameter type-id='5ce45b60' name='argnvl'/>
	<parameter type-id='857bb57e' name='outnvl'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	+ <function-decl name='lzc_get_vdev_prop' mangled-name='lzc_get_vdev_prop' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_get_vdev_prop'>
	+ <parameter type-id='80f4b756' name='poolname'/>
	+ <parameter type-id='5ce45b60' name='innvl'/>
	+ <parameter type-id='857bb57e' name='outnvl'/>
	+ <return type-id='95e97e5e'/>
	+ </function-decl>
	+ <function-decl name='lzc_set_vdev_prop' mangled-name='lzc_set_vdev_prop' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_set_vdev_prop'>
	+ <parameter type-id='80f4b756' name='poolname'/>
	+ <parameter type-id='5ce45b60' name='innvl'/>
	+ <parameter type-id='857bb57e' name='outnvl'/>
	+ <return type-id='95e97e5e'/>
	+ </function-decl>
	<function-decl name='lzc_load_key' mangled-name='lzc_load_key' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_load_key'>
	<parameter type-id='80f4b756' name='fsname'/>
	<parameter type-id='c19b74c3' name='noop'/>
	<parameter type-id='ae3e8ca6' name='wkeydata'/>
	<parameter type-id='3502e3ff' name='wkeylen'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_unload_key' mangled-name='lzc_unload_key' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_unload_key'>
	<parameter type-id='80f4b756' name='fsname'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_change_key' mangled-name='lzc_change_key' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_change_key'>
	<parameter type-id='80f4b756' name='fsname'/>
	<parameter type-id='9c313c2d' name='crypt_cmd'/>
	<parameter type-id='5ce45b60' name='props'/>
	<parameter type-id='ae3e8ca6' name='wkeydata'/>
	<parameter type-id='3502e3ff' name='wkeylen'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_reopen' mangled-name='lzc_reopen' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_reopen'>
	<parameter type-id='80f4b756' name='pool_name'/>
	<parameter type-id='c19b74c3' name='scrub_restart'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_initialize' mangled-name='lzc_initialize' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_initialize'>
	<parameter type-id='80f4b756' name='poolname'/>
	<parameter type-id='7063e1ab' name='cmd_type'/>
	<parameter type-id='5ce45b60' name='vdevs'/>
	<parameter type-id='857bb57e' name='errlist'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_trim' mangled-name='lzc_trim' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_trim'>
	<parameter type-id='80f4b756' name='poolname'/>
	<parameter type-id='b1146b8d' name='cmd_type'/>
	<parameter type-id='9c313c2d' name='rate'/>
	<parameter type-id='c19b74c3' name='secure'/>
	<parameter type-id='5ce45b60' name='vdevs'/>
	<parameter type-id='857bb57e' name='errlist'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_redact' mangled-name='lzc_redact' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_redact'>
	<parameter type-id='80f4b756' name='snapshot'/>
	<parameter type-id='80f4b756' name='bookname'/>
	<parameter type-id='5ce45b60' name='snapnv'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_wait' mangled-name='lzc_wait' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_wait'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='73446457' name='activity'/>
	<parameter type-id='37e3bd22' name='waited'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_wait_tag' mangled-name='lzc_wait_tag' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_wait_tag'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='73446457' name='activity'/>
	<parameter type-id='9c313c2d' name='tag'/>
	<parameter type-id='37e3bd22' name='waited'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_wait_fs' mangled-name='lzc_wait_fs' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_wait_fs'>
	<parameter type-id='80f4b756' name='fs'/>
	<parameter type-id='3024501a' name='activity'/>
	<parameter type-id='37e3bd22' name='waited'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_set_bootenv' mangled-name='lzc_set_bootenv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_set_bootenv'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='22cce67b' name='env'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzc_get_bootenv' mangled-name='lzc_get_bootenv' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_get_bootenv'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='857bb57e' name='outnvl'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	+ <type-decl name='void' id='48b5725f'/>
	</abi-instr>
	<abi-instr address-size='64' path='os/linux/libzfs_core_ioctl.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='32768' id='d16c6df4'>
	<subrange length='4096' type-id='7359adad' id='bc1b5ddc'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='65536' id='163f6aa5'>
	<subrange length='8192' type-id='7359adad' id='c88f397d'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='9c313c2d' size-in-bits='128' id='c1c22e6c'>
	<subrange length='2' type-id='7359adad' id='52efc4ef'/>
	</array-type-def>
	<class-decl name='dmu_objset_stats' size-in-bits='2304' is-struct='yes' visibility='default' id='098f0221'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='dds_num_clones' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='dds_creation_txg' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='dds_guid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='dds_type' type-id='230f1e16' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='224'>
	<var-decl name='dds_is_snapshot' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='232'>
	<var-decl name='dds_inconsistent' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='240'>
	<var-decl name='dds_redacted' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='248'>
	<var-decl name='dds_origin' type-id='d1617432' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='dmu_objset_stats_t' type-id='098f0221' id='b2c14f17'/>
	<class-decl name='zinject_record' size-in-bits='2816' is-struct='yes' visibility='default' id='3216f820'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='zi_objset' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='zi_object' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='zi_start' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='zi_end' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='zi_guid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='zi_level' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='352'>
	<var-decl name='zi_error' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='zi_type' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='zi_freq' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='480'>
	<var-decl name='zi_failfast' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='zi_func' type-id='d1617432' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2560'>
	<var-decl name='zi_iotype' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2592'>
	<var-decl name='zi_duration' type-id='3ff5601b' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2624'>
	<var-decl name='zi_timer' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2688'>
	<var-decl name='zi_nlanes' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2752'>
	<var-decl name='zi_cmd' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='2784'>
	<var-decl name='zi_dvas' type-id='8f92235e' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zinject_record_t' type-id='3216f820' id='a4301ca6'/>
	<class-decl name='zfs_share' size-in-bits='256' is-struct='yes' visibility='default' id='feb6f2da'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='z_exportdata' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='z_sharedata' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='z_sharetype' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='z_sharemax' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfs_share_t' type-id='feb6f2da' id='ee5cec36'/>
	<class-decl name='zfs_cmd' size-in-bits='109952' is-struct='yes' visibility='default' id='3522cd69'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='zc_name' type-id='d16c6df4' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32768'>
	<var-decl name='zc_nvlist_src' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32832'>
	<var-decl name='zc_nvlist_src_size' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32896'>
	<var-decl name='zc_nvlist_dst' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='32960'>
	<var-decl name='zc_nvlist_dst_size' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='33024'>
	<var-decl name='zc_nvlist_dst_filled' type-id='c19b74c3' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='33056'>
	<var-decl name='zc_pad2' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='33088'>
	<var-decl name='zc_history' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='33152'>
	<var-decl name='zc_value' type-id='163f6aa5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='98688'>
	<var-decl name='zc_string' type-id='d1617432' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='100736'>
	<var-decl name='zc_guid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='100800'>
	<var-decl name='zc_nvlist_conf' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='100864'>
	<var-decl name='zc_nvlist_conf_size' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='100928'>
	<var-decl name='zc_cookie' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='100992'>
	<var-decl name='zc_objset_type' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101056'>
	<var-decl name='zc_perm_action' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101120'>
	<var-decl name='zc_history_len' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101184'>
	<var-decl name='zc_history_offset' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101248'>
	<var-decl name='zc_obj' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101312'>
	<var-decl name='zc_iflags' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101376'>
	<var-decl name='zc_share' type-id='ee5cec36' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='101632'>
	<var-decl name='zc_objset_stats' type-id='b2c14f17' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='103936'>
	<var-decl name='zc_begin_record' type-id='09fcdc01' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='106368'>
	<var-decl name='zc_inject_record' type-id='a4301ca6' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109184'>
	<var-decl name='zc_defer_destroy' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109216'>
	<var-decl name='zc_flags' type-id='8f92235e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109248'>
	<var-decl name='zc_action_handle' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109312'>
	<var-decl name='zc_cleanup_fd' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109344'>
	<var-decl name='zc_simple' type-id='b96825af' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109352'>
	<var-decl name='zc_pad' type-id='d3490169' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109376'>
	<var-decl name='zc_sendobj' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109440'>
	<var-decl name='zc_fromobj' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109504'>
	<var-decl name='zc_createtxg' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109568'>
	<var-decl name='zc_stat' type-id='0371a9c7' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='109888'>
	<var-decl name='zc_zoneid' type-id='9c313c2d' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfs_cmd_t' type-id='3522cd69' id='a5559cdd'/>
	<class-decl name='zfs_stat' size-in-bits='320' is-struct='yes' visibility='default' id='6417f0b9'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='zs_gen' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='zs_mode' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='zs_links' type-id='9c313c2d' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='zs_ctime' type-id='c1c22e6c' visibility='default'/>
	</data-member>
	</class-decl>
	<typedef-decl name='zfs_stat_t' type-id='6417f0b9' id='0371a9c7'/>
	<pointer-type-def type-id='a5559cdd' size-in-bits='64' id='e4ec4540'/>
	<function-decl name='lzc_ioctl_fd' mangled-name='lzc_ioctl_fd' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzc_ioctl_fd'>
	<parameter type-id='95e97e5e' name='fd'/>
	<parameter type-id='7359adad' name='request'/>
	<parameter type-id='e4ec4540' name='zc'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	</abi-corpus>
	diff --git a/sys/contrib/openzfs/lib/libzfs_core/libzfs_core.c b/sys/contrib/openzfs/lib/libzfs_core/libzfs_core.c
	index cbe486d08ba1..c7c4482a0cba 100644
	--- a/sys/contrib/openzfs/lib/libzfs_core/libzfs_core.c
	+++ b/sys/contrib/openzfs/lib/libzfs_core/libzfs_core.c
	@@ -1,1642 +1,1654 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2012, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2013 Steven Hartland. All rights reserved.
	* Copyright (c) 2017 Datto Inc.
	* Copyright 2017 RackTop Systems.
	* Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
	* Copyright (c) 2019, 2020 by Christian Schwarz. All rights reserved.
	*/

	/*
	* LibZFS_Core (lzc) is intended to replace most functionality in libzfs.
	* It has the following characteristics:
	*
	* - Thread Safe. libzfs_core is accessible concurrently from multiple
	* threads. This is accomplished primarily by avoiding global data
	* (e.g. caching). Since it's thread-safe, there is no reason for a
	* process to have multiple libzfs "instances". Therefore, we store
	* our few pieces of data (e.g. the file descriptor) in global
	* variables. The fd is reference-counted so that the libzfs_core
	* library can be "initialized" multiple times (e.g. by different
	* consumers within the same process).
	*
	* - Committed Interface. The libzfs_core interface will be committed,
	* therefore consumers can compile against it and be confident that
	* their code will continue to work on future releases of this code.
	* Currently, the interface is Evolving (not Committed), but we intend
	* to commit to it once it is more complete and we determine that it
	* meets the needs of all consumers.
	*
	* - Programmatic Error Handling. libzfs_core communicates errors with
	* defined error numbers, and doesn't print anything to stdout/stderr.
	*
	* - Thin Layer. libzfs_core is a thin layer, marshaling arguments
	* to/from the kernel ioctls. There is generally a 1:1 correspondence
	* between libzfs_core functions and ioctls to ZFS_DEV.
	*
	* - Clear Atomicity. Because libzfs_core functions are generally 1:1
	* with kernel ioctls, and kernel ioctls are general atomic, each
	* libzfs_core function is atomic. For example, creating multiple
	* snapshots with a single call to lzc_snapshot() is atomic -- it
	* can't fail with only some of the requested snapshots created, even
	* in the event of power loss or system crash.
	*
	* - Continued libzfs Support. Some higher-level operations (e.g.
	* support for "zfs send -R") are too complicated to fit the scope of
	* libzfs_core. This functionality will continue to live in libzfs.
	* Where appropriate, libzfs will use the underlying atomic operations
	* of libzfs_core. For example, libzfs may implement "zfs send -R \|
	* zfs receive" by using individual "send one snapshot", rename,
	* destroy, and "receive one snapshot" operations in libzfs_core.
	* /sbin/zfs and /sbin/zpool will link with both libzfs and
	* libzfs_core. Other consumers should aim to use only libzfs_core,
	* since that will be the supported, stable interface going forwards.
	*/

	#include <libzfs_core.h>
	#include <ctype.h>
	#include <unistd.h>
	#include <stdlib.h>
	#include <string.h>
	#ifdef ZFS_DEBUG
	#include <stdio.h>
	#endif
	#include <errno.h>
	#include <fcntl.h>
	#include <pthread.h>
	#include <libzutil.h>
	#include <sys/nvpair.h>
	#include <sys/param.h>
	#include <sys/types.h>
	#include <sys/stat.h>
	#include <sys/zfs_ioctl.h>

	static int g_fd = -1;
	static pthread_mutex_t g_lock = PTHREAD_MUTEX_INITIALIZER;
	static int g_refcount;

	#ifdef ZFS_DEBUG
	static zfs_ioc_t fail_ioc_cmd = ZFS_IOC_LAST;
	static zfs_errno_t fail_ioc_err;

	static void
	libzfs_core_debug_ioc(void)
	{
	/*
	* To test running newer user space binaries with kernel's
	* that don't yet support an ioctl or a new ioctl arg we
	* provide an override to intentionally fail an ioctl.
	*
	* USAGE:
	* The override variable, ZFS_IOC_TEST, is of the form "cmd:err"
	*
	* For example, to fail a ZFS_IOC_POOL_CHECKPOINT with a
	* ZFS_ERR_IOC_CMD_UNAVAIL, the string would be "0x5a4d:1029"
	*
	* $ sudo sh -c "ZFS_IOC_TEST=0x5a4d:1029 zpool checkpoint tank"
	* cannot checkpoint 'tank': the loaded zfs module does not support
	* this operation. A reboot may be required to enable this operation.
	*/
	if (fail_ioc_cmd == ZFS_IOC_LAST) {
	char *ioc_test = getenv("ZFS_IOC_TEST");
	unsigned int ioc_num = 0, ioc_err = 0;

	if (ioc_test != NULL &&
	sscanf(ioc_test, "%i:%i", &ioc_num, &ioc_err) == 2 &&
	ioc_num < ZFS_IOC_LAST) {
	fail_ioc_cmd = ioc_num;
	fail_ioc_err = ioc_err;
	}
	}
	}
	#endif

	int
	libzfs_core_init(void)
	{
	(void) pthread_mutex_lock(&g_lock);
	if (g_refcount == 0) {
	g_fd = open(ZFS_DEV, O_RDWR\|O_CLOEXEC);
	if (g_fd < 0) {
	(void) pthread_mutex_unlock(&g_lock);
	return (errno);
	}
	}
	g_refcount++;

	#ifdef ZFS_DEBUG
	libzfs_core_debug_ioc();
	#endif
	(void) pthread_mutex_unlock(&g_lock);
	return (0);
	}

	void
	libzfs_core_fini(void)
	{
	(void) pthread_mutex_lock(&g_lock);
	ASSERT3S(g_refcount, >, 0);

	g_refcount--;

	if (g_refcount == 0 && g_fd != -1) {
	(void) close(g_fd);
	g_fd = -1;
	}
	(void) pthread_mutex_unlock(&g_lock);
	}

	static int
	lzc_ioctl(zfs_ioc_t ioc, const char *name,
	nvlist_t source, nvlist_t *resultp)
	{
	zfs_cmd_t zc = {"\0"};
	int error = 0;
	char *packed = NULL;
	size_t size = 0;

	ASSERT3S(g_refcount, >, 0);
	VERIFY3S(g_fd, !=, -1);

	#ifdef ZFS_DEBUG
	if (ioc == fail_ioc_cmd)
	return (fail_ioc_err);
	#endif

	if (name != NULL)
	(void) strlcpy(zc.zc_name, name, sizeof (zc.zc_name));

	if (source != NULL) {
	packed = fnvlist_pack(source, &size);
	zc.zc_nvlist_src = (uint64_t)(uintptr_t)packed;
	zc.zc_nvlist_src_size = size;
	}

	if (resultp != NULL) {
	*resultp = NULL;
	if (ioc == ZFS_IOC_CHANNEL_PROGRAM) {
	zc.zc_nvlist_dst_size = fnvlist_lookup_uint64(source,
	ZCP_ARG_MEMLIMIT);
	} else {
	zc.zc_nvlist_dst_size = MAX(size * 2, 128 * 1024);
	}
	zc.zc_nvlist_dst = (uint64_t)(uintptr_t)
	malloc(zc.zc_nvlist_dst_size);
	if (zc.zc_nvlist_dst == (uint64_t)0) {
	error = ENOMEM;
	goto out;
	}
	}

	while (lzc_ioctl_fd(g_fd, ioc, &zc) != 0) {
	/*
	* If ioctl exited with ENOMEM, we retry the ioctl after
	* increasing the size of the destination nvlist.
	*
	* Channel programs that exit with ENOMEM ran over the
	* lua memory sandbox; they should not be retried.
	*/
	if (errno == ENOMEM && resultp != NULL &&
	ioc != ZFS_IOC_CHANNEL_PROGRAM) {
	free((void *)(uintptr_t)zc.zc_nvlist_dst);
	zc.zc_nvlist_dst_size *= 2;
	zc.zc_nvlist_dst = (uint64_t)(uintptr_t)
	malloc(zc.zc_nvlist_dst_size);
	if (zc.zc_nvlist_dst == (uint64_t)0) {
	error = ENOMEM;
	goto out;
	}
	} else {
	error = errno;
	break;
	}
	}
	if (zc.zc_nvlist_dst_filled) {
	resultp = fnvlist_unpack((void )(uintptr_t)zc.zc_nvlist_dst,
	zc.zc_nvlist_dst_size);
	}

	out:
	if (packed != NULL)
	fnvlist_pack_free(packed, size);
	free((void *)(uintptr_t)zc.zc_nvlist_dst);
	return (error);
	}

	int
	lzc_create(const char fsname, enum lzc_dataset_type type, nvlist_t props,
	uint8_t *wkeydata, uint_t wkeylen)
	{
	int error;
	nvlist_t *hidden_args = NULL;
	nvlist_t *args = fnvlist_alloc();

	fnvlist_add_int32(args, "type", (dmu_objset_type_t)type);
	if (props != NULL)
	fnvlist_add_nvlist(args, "props", props);

	if (wkeydata != NULL) {
	hidden_args = fnvlist_alloc();
	fnvlist_add_uint8_array(hidden_args, "wkeydata", wkeydata,
	wkeylen);
	fnvlist_add_nvlist(args, ZPOOL_HIDDEN_ARGS, hidden_args);
	}

	error = lzc_ioctl(ZFS_IOC_CREATE, fsname, args, NULL);
	nvlist_free(hidden_args);
	nvlist_free(args);
	return (error);
	}

	int
	lzc_clone(const char fsname, const char origin, nvlist_t *props)
	{
	int error;
	nvlist_t *hidden_args = NULL;
	nvlist_t *args = fnvlist_alloc();

	fnvlist_add_string(args, "origin", origin);
	if (props != NULL)
	fnvlist_add_nvlist(args, "props", props);
	error = lzc_ioctl(ZFS_IOC_CLONE, fsname, args, NULL);
	nvlist_free(hidden_args);
	nvlist_free(args);
	return (error);
	}

	int
	lzc_promote(const char fsname, char snapnamebuf, int snapnamelen)
	{
	/*
	* The promote ioctl is still legacy, so we need to construct our
	* own zfs_cmd_t rather than using lzc_ioctl().
	*/
	zfs_cmd_t zc = {"\0"};

	ASSERT3S(g_refcount, >, 0);
	VERIFY3S(g_fd, !=, -1);

	(void) strlcpy(zc.zc_name, fsname, sizeof (zc.zc_name));
	if (lzc_ioctl_fd(g_fd, ZFS_IOC_PROMOTE, &zc) != 0) {
	int error = errno;
	if (error == EEXIST && snapnamebuf != NULL)
	(void) strlcpy(snapnamebuf, zc.zc_string, snapnamelen);
	return (error);
	}
	return (0);
	}

	int
	lzc_rename(const char source, const char target)
	{
	zfs_cmd_t zc = {"\0"};
	int error;

	ASSERT3S(g_refcount, >, 0);
	VERIFY3S(g_fd, !=, -1);
	(void) strlcpy(zc.zc_name, source, sizeof (zc.zc_name));
	(void) strlcpy(zc.zc_value, target, sizeof (zc.zc_value));
	error = lzc_ioctl_fd(g_fd, ZFS_IOC_RENAME, &zc);
	if (error != 0)
	error = errno;
	return (error);
	}

	int
	lzc_destroy(const char *fsname)
	{
	int error;
	nvlist_t *args = fnvlist_alloc();
	error = lzc_ioctl(ZFS_IOC_DESTROY, fsname, args, NULL);
	nvlist_free(args);
	return (error);
	}

	/*
	* Creates snapshots.
	*
	* The keys in the snaps nvlist are the snapshots to be created.
	* They must all be in the same pool.
	*
	* The props nvlist is properties to set. Currently only user properties
	* are supported. { user:prop_name -> string value }
	*
	* The returned results nvlist will have an entry for each snapshot that failed.
	* The value will be the (int32) error code.
	*
	* The return value will be 0 if all snapshots were created, otherwise it will
	* be the errno of a (unspecified) snapshot that failed.
	*/
	int
	lzc_snapshot(nvlist_t snaps, nvlist_t props, nvlist_t **errlist)
	{
	nvpair_t *elem;
	nvlist_t *args;
	int error;
	char pool[ZFS_MAX_DATASET_NAME_LEN];

	*errlist = NULL;

	/* determine the pool name */
	elem = nvlist_next_nvpair(snaps, NULL);
	if (elem == NULL)
	return (0);
	(void) strlcpy(pool, nvpair_name(elem), sizeof (pool));
	pool[strcspn(pool, "/@")] = '\0';

	args = fnvlist_alloc();
	fnvlist_add_nvlist(args, "snaps", snaps);
	if (props != NULL)
	fnvlist_add_nvlist(args, "props", props);

	error = lzc_ioctl(ZFS_IOC_SNAPSHOT, pool, args, errlist);
	nvlist_free(args);

	return (error);
	}

	/*
	* Destroys snapshots.
	*
	* The keys in the snaps nvlist are the snapshots to be destroyed.
	* They must all be in the same pool.
	*
	* Snapshots that do not exist will be silently ignored.
	*
	* If 'defer' is not set, and a snapshot has user holds or clones, the
	* destroy operation will fail and none of the snapshots will be
	* destroyed.
	*
	* If 'defer' is set, and a snapshot has user holds or clones, it will be
	* marked for deferred destruction, and will be destroyed when the last hold
	* or clone is removed/destroyed.
	*
	* The return value will be 0 if all snapshots were destroyed (or marked for
	* later destruction if 'defer' is set) or didn't exist to begin with.
	*
	* Otherwise the return value will be the errno of a (unspecified) snapshot
	* that failed, no snapshots will be destroyed, and the errlist will have an
	* entry for each snapshot that failed. The value in the errlist will be
	* the (int32) error code.
	*/
	int
	lzc_destroy_snaps(nvlist_t snaps, boolean_t defer, nvlist_t *errlist)
	{
	nvpair_t *elem;
	nvlist_t *args;
	int error;
	char pool[ZFS_MAX_DATASET_NAME_LEN];

	/* determine the pool name */
	elem = nvlist_next_nvpair(snaps, NULL);
	if (elem == NULL)
	return (0);
	(void) strlcpy(pool, nvpair_name(elem), sizeof (pool));
	pool[strcspn(pool, "/@")] = '\0';

	args = fnvlist_alloc();
	fnvlist_add_nvlist(args, "snaps", snaps);
	if (defer)
	fnvlist_add_boolean(args, "defer");

	error = lzc_ioctl(ZFS_IOC_DESTROY_SNAPS, pool, args, errlist);
	nvlist_free(args);

	return (error);
	}

	int
	lzc_snaprange_space(const char firstsnap, const char lastsnap,
	uint64_t *usedp)
	{
	nvlist_t *args;
	nvlist_t *result;
	int err;
	char fs[ZFS_MAX_DATASET_NAME_LEN];
	char *atp;

	/* determine the fs name */
	(void) strlcpy(fs, firstsnap, sizeof (fs));
	atp = strchr(fs, '@');
	if (atp == NULL)
	return (EINVAL);
	*atp = '\0';

	args = fnvlist_alloc();
	fnvlist_add_string(args, "firstsnap", firstsnap);

	err = lzc_ioctl(ZFS_IOC_SPACE_SNAPS, lastsnap, args, &result);
	nvlist_free(args);
	if (err == 0)
	*usedp = fnvlist_lookup_uint64(result, "used");
	fnvlist_free(result);

	return (err);
	}

	boolean_t
	lzc_exists(const char *dataset)
	{
	/*
	* The objset_stats ioctl is still legacy, so we need to construct our
	* own zfs_cmd_t rather than using lzc_ioctl().
	*/
	zfs_cmd_t zc = {"\0"};

	ASSERT3S(g_refcount, >, 0);
	VERIFY3S(g_fd, !=, -1);

	(void) strlcpy(zc.zc_name, dataset, sizeof (zc.zc_name));
	return (lzc_ioctl_fd(g_fd, ZFS_IOC_OBJSET_STATS, &zc) == 0);
	}

	/*
	* outnvl is unused.
	* It was added to preserve the function signature in case it is
	* needed in the future.
	*/
	/ARGSUSED/
	int
	lzc_sync(const char pool_name, nvlist_t innvl, nvlist_t **outnvl)
	{
	return (lzc_ioctl(ZFS_IOC_POOL_SYNC, pool_name, innvl, NULL));
	}

	/*
	* Create "user holds" on snapshots. If there is a hold on a snapshot,
	* the snapshot can not be destroyed. (However, it can be marked for deletion
	* by lzc_destroy_snaps(defer=B_TRUE).)
	*
	* The keys in the nvlist are snapshot names.
	* The snapshots must all be in the same pool.
	* The value is the name of the hold (string type).
	*
	* If cleanup_fd is not -1, it must be the result of open(ZFS_DEV, O_EXCL).
	* In this case, when the cleanup_fd is closed (including on process
	* termination), the holds will be released. If the system is shut down
	* uncleanly, the holds will be released when the pool is next opened
	* or imported.
	*
	* Holds for snapshots which don't exist will be skipped and have an entry
	* added to errlist, but will not cause an overall failure.
	*
	* The return value will be 0 if all holds, for snapshots that existed,
	* were successfully created.
	*
	* Otherwise the return value will be the errno of a (unspecified) hold that
	* failed and no holds will be created.
	*
	* In all cases the errlist will have an entry for each hold that failed
	* (name = snapshot), with its value being the error code (int32).
	*/
	int
	lzc_hold(nvlist_t holds, int cleanup_fd, nvlist_t *errlist)
	{
	char pool[ZFS_MAX_DATASET_NAME_LEN];
	nvlist_t *args;
	nvpair_t *elem;
	int error;

	/* determine the pool name */
	elem = nvlist_next_nvpair(holds, NULL);
	if (elem == NULL)
	return (0);
	(void) strlcpy(pool, nvpair_name(elem), sizeof (pool));
	pool[strcspn(pool, "/@")] = '\0';

	args = fnvlist_alloc();
	fnvlist_add_nvlist(args, "holds", holds);
	if (cleanup_fd != -1)
	fnvlist_add_int32(args, "cleanup_fd", cleanup_fd);

	error = lzc_ioctl(ZFS_IOC_HOLD, pool, args, errlist);
	nvlist_free(args);
	return (error);
	}

	/*
	* Release "user holds" on snapshots. If the snapshot has been marked for
	* deferred destroy (by lzc_destroy_snaps(defer=B_TRUE)), it does not have
	* any clones, and all the user holds are removed, then the snapshot will be
	* destroyed.
	*
	* The keys in the nvlist are snapshot names.
	* The snapshots must all be in the same pool.
	* The value is an nvlist whose keys are the holds to remove.
	*
	* Holds which failed to release because they didn't exist will have an entry
	* added to errlist, but will not cause an overall failure.
	*
	* The return value will be 0 if the nvl holds was empty or all holds that
	* existed, were successfully removed.
	*
	* Otherwise the return value will be the errno of a (unspecified) hold that
	* failed to release and no holds will be released.
	*
	* In all cases the errlist will have an entry for each hold that failed to
	* to release.
	*/
	int
	lzc_release(nvlist_t holds, nvlist_t *errlist)
	{
	char pool[ZFS_MAX_DATASET_NAME_LEN];
	nvpair_t *elem;

	/* determine the pool name */
	elem = nvlist_next_nvpair(holds, NULL);
	if (elem == NULL)
	return (0);
	(void) strlcpy(pool, nvpair_name(elem), sizeof (pool));
	pool[strcspn(pool, "/@")] = '\0';

	return (lzc_ioctl(ZFS_IOC_RELEASE, pool, holds, errlist));
	}

	/*
	* Retrieve list of user holds on the specified snapshot.
	*
	* On success, *holdsp will be set to an nvlist which the caller must free.
	* The keys are the names of the holds, and the value is the creation time
	* of the hold (uint64) in seconds since the epoch.
	*/
	int
	lzc_get_holds(const char snapname, nvlist_t *holdsp)
	{
	return (lzc_ioctl(ZFS_IOC_GET_HOLDS, snapname, NULL, holdsp));
	}

	/*
	* Generate a zfs send stream for the specified snapshot and write it to
	* the specified file descriptor.
	*
	* "snapname" is the full name of the snapshot to send (e.g. "pool/fs@snap")
	*
	* If "from" is NULL, a full (non-incremental) stream will be sent.
	* If "from" is non-NULL, it must be the full name of a snapshot or
	* bookmark to send an incremental from (e.g. "pool/fs@earlier_snap" or
	* "pool/fs#earlier_bmark"). If non-NULL, the specified snapshot or
	* bookmark must represent an earlier point in the history of "snapname").
	* It can be an earlier snapshot in the same filesystem or zvol as "snapname",
	* or it can be the origin of "snapname"'s filesystem, or an earlier
	* snapshot in the origin, etc.
	*
	* "fd" is the file descriptor to write the send stream to.
	*
	* If "flags" contains LZC_SEND_FLAG_LARGE_BLOCK, the stream is permitted
	* to contain DRR_WRITE records with drr_length > 128K, and DRR_OBJECT
	* records with drr_blksz > 128K.
	*
	* If "flags" contains LZC_SEND_FLAG_EMBED_DATA, the stream is permitted
	* to contain DRR_WRITE_EMBEDDED records with drr_etype==BP_EMBEDDED_TYPE_DATA,
	* which the receiving system must support (as indicated by support
	* for the "embedded_data" feature).
	*
	* If "flags" contains LZC_SEND_FLAG_COMPRESS, the stream is generated by using
	* compressed WRITE records for blocks which are compressed on disk and in
	* memory. If the lz4_compress feature is active on the sending system, then
	* the receiving system must have that feature enabled as well.
	*
	* If "flags" contains LZC_SEND_FLAG_RAW, the stream is generated, for encrypted
	* datasets, by sending data exactly as it exists on disk. This allows backups
	* to be taken even if encryption keys are not currently loaded.
	*/
	int
	lzc_send(const char snapname, const char from, int fd,
	enum lzc_send_flags flags)
	{
	return (lzc_send_resume_redacted(snapname, from, fd, flags, 0, 0,
	NULL));
	}

	int
	lzc_send_redacted(const char snapname, const char from, int fd,
	enum lzc_send_flags flags, const char *redactbook)
	{
	return (lzc_send_resume_redacted(snapname, from, fd, flags, 0, 0,
	redactbook));
	}

	int
	lzc_send_resume(const char snapname, const char from, int fd,
	enum lzc_send_flags flags, uint64_t resumeobj, uint64_t resumeoff)
	{
	return (lzc_send_resume_redacted(snapname, from, fd, flags, resumeobj,
	resumeoff, NULL));
	}

	/*
	* snapname: The name of the "tosnap", or the snapshot whose contents we are
	* sending.
	* from: The name of the "fromsnap", or the incremental source.
	* fd: File descriptor to write the stream to.
	* flags: flags that determine features to be used by the stream.
	* resumeobj: Object to resume from, for resuming send
	* resumeoff: Offset to resume from, for resuming send.
	* redactnv: nvlist of string -> boolean(ignored) containing the names of all
	* the snapshots that we should redact with respect to.
	* redactbook: Name of the redaction bookmark to create.
	*/
	int
	lzc_send_resume_redacted(const char snapname, const char from, int fd,
	enum lzc_send_flags flags, uint64_t resumeobj, uint64_t resumeoff,
	const char *redactbook)
	{
	nvlist_t *args;
	int err;

	args = fnvlist_alloc();
	fnvlist_add_int32(args, "fd", fd);
	if (from != NULL)
	fnvlist_add_string(args, "fromsnap", from);
	if (flags & LZC_SEND_FLAG_LARGE_BLOCK)
	fnvlist_add_boolean(args, "largeblockok");
	if (flags & LZC_SEND_FLAG_EMBED_DATA)
	fnvlist_add_boolean(args, "embedok");
	if (flags & LZC_SEND_FLAG_COMPRESS)
	fnvlist_add_boolean(args, "compressok");
	if (flags & LZC_SEND_FLAG_RAW)
	fnvlist_add_boolean(args, "rawok");
	if (flags & LZC_SEND_FLAG_SAVED)
	fnvlist_add_boolean(args, "savedok");
	if (resumeobj != 0 \|\| resumeoff != 0) {
	fnvlist_add_uint64(args, "resume_object", resumeobj);
	fnvlist_add_uint64(args, "resume_offset", resumeoff);
	}
	if (redactbook != NULL)
	fnvlist_add_string(args, "redactbook", redactbook);

	err = lzc_ioctl(ZFS_IOC_SEND_NEW, snapname, args, NULL);
	nvlist_free(args);
	return (err);
	}

	/*
	* "from" can be NULL, a snapshot, or a bookmark.
	*
	* If from is NULL, a full (non-incremental) stream will be estimated. This
	* is calculated very efficiently.
	*
	* If from is a snapshot, lzc_send_space uses the deadlists attached to
	* each snapshot to efficiently estimate the stream size.
	*
	* If from is a bookmark, the indirect blocks in the destination snapshot
	* are traversed, looking for blocks with a birth time since the creation TXG of
	* the snapshot this bookmark was created from. This will result in
	* significantly more I/O and be less efficient than a send space estimation on
	* an equivalent snapshot. This process is also used if redact_snaps is
	* non-null.
	*/
	int
	lzc_send_space_resume_redacted(const char snapname, const char from,
	enum lzc_send_flags flags, uint64_t resumeobj, uint64_t resumeoff,
	uint64_t resume_bytes, const char redactbook, int fd, uint64_t spacep)
	{
	nvlist_t *args;
	nvlist_t *result;
	int err;

	args = fnvlist_alloc();
	if (from != NULL)
	fnvlist_add_string(args, "from", from);
	if (flags & LZC_SEND_FLAG_LARGE_BLOCK)
	fnvlist_add_boolean(args, "largeblockok");
	if (flags & LZC_SEND_FLAG_EMBED_DATA)
	fnvlist_add_boolean(args, "embedok");
	if (flags & LZC_SEND_FLAG_COMPRESS)
	fnvlist_add_boolean(args, "compressok");
	if (flags & LZC_SEND_FLAG_RAW)
	fnvlist_add_boolean(args, "rawok");
	if (resumeobj != 0 \|\| resumeoff != 0) {
	fnvlist_add_uint64(args, "resume_object", resumeobj);
	fnvlist_add_uint64(args, "resume_offset", resumeoff);
	fnvlist_add_uint64(args, "bytes", resume_bytes);
	}
	if (redactbook != NULL)
	fnvlist_add_string(args, "redactbook", redactbook);
	if (fd != -1)
	fnvlist_add_int32(args, "fd", fd);

	err = lzc_ioctl(ZFS_IOC_SEND_SPACE, snapname, args, &result);
	nvlist_free(args);
	if (err == 0)
	*spacep = fnvlist_lookup_uint64(result, "space");
	nvlist_free(result);
	return (err);
	}

	int
	lzc_send_space(const char snapname, const char from,
	enum lzc_send_flags flags, uint64_t *spacep)
	{
	return (lzc_send_space_resume_redacted(snapname, from, flags, 0, 0, 0,
	NULL, -1, spacep));
	}

	static int
	recv_read(int fd, void *buf, int ilen)
	{
	char *cp = buf;
	int rv;
	int len = ilen;

	do {
	rv = read(fd, cp, len);
	cp += rv;
	len -= rv;
	} while (rv > 0);

	if (rv < 0 \|\| len != 0)
	return (EIO);

	return (0);
	}

	/*
	* Linux adds ZFS_IOC_RECV_NEW for resumable and raw streams and preserves the
	* legacy ZFS_IOC_RECV user/kernel interface. The new interface supports all
	* stream options but is currently only used for resumable streams. This way
	* updated user space utilities will interoperate with older kernel modules.
	*
	* Non-Linux OpenZFS platforms have opted to modify the legacy interface.
	*/
	static int
	recv_impl(const char snapname, nvlist_t recvdprops, nvlist_t *localprops,
	uint8_t wkeydata, uint_t wkeylen, const char origin, boolean_t force,
	boolean_t resumable, boolean_t raw, int input_fd,
	const dmu_replay_record_t begin_record, uint64_t read_bytes,
	uint64_t errflags, nvlist_t *errors)
	{
	dmu_replay_record_t drr;
	char fsname[MAXPATHLEN];
	char *atp;
	int error;
	boolean_t payload = B_FALSE;

	ASSERT3S(g_refcount, >, 0);
	VERIFY3S(g_fd, !=, -1);

	/* Set 'fsname' to the name of containing filesystem */
	(void) strlcpy(fsname, snapname, sizeof (fsname));
	atp = strchr(fsname, '@');
	if (atp == NULL)
	return (EINVAL);
	*atp = '\0';

	/* If the fs does not exist, try its parent. */
	if (!lzc_exists(fsname)) {
	char *slashp = strrchr(fsname, '/');
	if (slashp == NULL)
	return (ENOENT);
	*slashp = '\0';
	}

	/*
	* The begin_record is normally a non-byteswapped BEGIN record.
	* For resumable streams it may be set to any non-byteswapped
	* dmu_replay_record_t.
	*/
	if (begin_record == NULL) {
	error = recv_read(input_fd, &drr, sizeof (drr));
	if (error != 0)
	return (error);
	} else {
	drr = *begin_record;
	payload = (begin_record->drr_payloadlen != 0);
	}

	/*
	* All receives with a payload should use the new interface.
	*/
	if (resumable \|\| raw \|\| wkeydata != NULL \|\| payload) {
	nvlist_t *outnvl = NULL;
	nvlist_t *innvl = fnvlist_alloc();

	fnvlist_add_string(innvl, "snapname", snapname);

	if (recvdprops != NULL)
	fnvlist_add_nvlist(innvl, "props", recvdprops);

	if (localprops != NULL)
	fnvlist_add_nvlist(innvl, "localprops", localprops);

	if (wkeydata != NULL) {
	/*
	* wkeydata must be placed in the special
	* ZPOOL_HIDDEN_ARGS nvlist so that it
	* will not be printed to the zpool history.
	*/
	nvlist_t *hidden_args = fnvlist_alloc();
	fnvlist_add_uint8_array(hidden_args, "wkeydata",
	wkeydata, wkeylen);
	fnvlist_add_nvlist(innvl, ZPOOL_HIDDEN_ARGS,
	hidden_args);
	nvlist_free(hidden_args);
	}

	if (origin != NULL && strlen(origin))
	fnvlist_add_string(innvl, "origin", origin);

	fnvlist_add_byte_array(innvl, "begin_record",
	(uchar_t *)&drr, sizeof (drr));

	fnvlist_add_int32(innvl, "input_fd", input_fd);

	if (force)
	fnvlist_add_boolean(innvl, "force");

	if (resumable)
	fnvlist_add_boolean(innvl, "resumable");


	error = lzc_ioctl(ZFS_IOC_RECV_NEW, fsname, innvl, &outnvl);

	if (error == 0 && read_bytes != NULL)
	error = nvlist_lookup_uint64(outnvl, "read_bytes",
	read_bytes);

	if (error == 0 && errflags != NULL)
	error = nvlist_lookup_uint64(outnvl, "error_flags",
	errflags);

	if (error == 0 && errors != NULL) {
	nvlist_t *nvl;
	error = nvlist_lookup_nvlist(outnvl, "errors", &nvl);
	if (error == 0)
	*errors = fnvlist_dup(nvl);
	}

	fnvlist_free(innvl);
	fnvlist_free(outnvl);
	} else {
	zfs_cmd_t zc = {"\0"};
	char *packed = NULL;
	size_t size;

	ASSERT3S(g_refcount, >, 0);

	(void) strlcpy(zc.zc_name, fsname, sizeof (zc.zc_name));
	(void) strlcpy(zc.zc_value, snapname, sizeof (zc.zc_value));

	if (recvdprops != NULL) {
	packed = fnvlist_pack(recvdprops, &size);
	zc.zc_nvlist_src = (uint64_t)(uintptr_t)packed;
	zc.zc_nvlist_src_size = size;
	}

	if (localprops != NULL) {
	packed = fnvlist_pack(localprops, &size);
	zc.zc_nvlist_conf = (uint64_t)(uintptr_t)packed;
	zc.zc_nvlist_conf_size = size;
	}

	if (origin != NULL)
	(void) strlcpy(zc.zc_string, origin,
	sizeof (zc.zc_string));

	ASSERT3S(drr.drr_type, ==, DRR_BEGIN);
	zc.zc_begin_record = drr.drr_u.drr_begin;
	zc.zc_guid = force;
	zc.zc_cookie = input_fd;
	zc.zc_cleanup_fd = -1;
	zc.zc_action_handle = 0;

	zc.zc_nvlist_dst_size = 128 * 1024;
	zc.zc_nvlist_dst = (uint64_t)(uintptr_t)
	malloc(zc.zc_nvlist_dst_size);

	error = lzc_ioctl_fd(g_fd, ZFS_IOC_RECV, &zc);
	if (error != 0) {
	error = errno;
	} else {
	if (read_bytes != NULL)
	*read_bytes = zc.zc_cookie;

	if (errflags != NULL)
	*errflags = zc.zc_obj;

	if (errors != NULL)
	VERIFY0(nvlist_unpack(
	(void *)(uintptr_t)zc.zc_nvlist_dst,
	zc.zc_nvlist_dst_size, errors, KM_SLEEP));
	}

	if (packed != NULL)
	fnvlist_pack_free(packed, size);
	free((void *)(uintptr_t)zc.zc_nvlist_dst);
	}

	return (error);
	}

	/*
	* The simplest receive case: receive from the specified fd, creating the
	* specified snapshot. Apply the specified properties as "received" properties
	* (which can be overridden by locally-set properties). If the stream is a
	* clone, its origin snapshot must be specified by 'origin'. The 'force'
	* flag will cause the target filesystem to be rolled back or destroyed if
	* necessary to receive.
	*
	* Return 0 on success or an errno on failure.
	*
	* Note: this interface does not work on dedup'd streams
	* (those with DMU_BACKUP_FEATURE_DEDUP).
	*/
	int
	lzc_receive(const char snapname, nvlist_t props, const char *origin,
	boolean_t force, boolean_t raw, int fd)
	{
	return (recv_impl(snapname, props, NULL, NULL, 0, origin, force,
	B_FALSE, raw, fd, NULL, NULL, NULL, NULL));
	}

	/*
	* Like lzc_receive, but if the receive fails due to premature stream
	* termination, the intermediate state will be preserved on disk. In this
	* case, ECKSUM will be returned. The receive may subsequently be resumed
	* with a resuming send stream generated by lzc_send_resume().
	*/
	int
	lzc_receive_resumable(const char snapname, nvlist_t props, const char *origin,
	boolean_t force, boolean_t raw, int fd)
	{
	return (recv_impl(snapname, props, NULL, NULL, 0, origin, force,
	B_TRUE, raw, fd, NULL, NULL, NULL, NULL));
	}

	/*
	* Like lzc_receive, but allows the caller to read the begin record and then to
	* pass it in. That could be useful if the caller wants to derive, for example,
	* the snapname or the origin parameters based on the information contained in
	* the begin record.
	* The begin record must be in its original form as read from the stream,
	* in other words, it should not be byteswapped.
	*
	* The 'resumable' parameter allows to obtain the same behavior as with
	* lzc_receive_resumable.
	*/
	int
	lzc_receive_with_header(const char snapname, nvlist_t props,
	const char *origin, boolean_t force, boolean_t resumable, boolean_t raw,
	int fd, const dmu_replay_record_t *begin_record)
	{
	if (begin_record == NULL)
	return (EINVAL);

	return (recv_impl(snapname, props, NULL, NULL, 0, origin, force,
	resumable, raw, fd, begin_record, NULL, NULL, NULL));
	}

	/*
	* Like lzc_receive, but allows the caller to pass all supported arguments
	* and retrieve all values returned. The only additional input parameter
	* is 'cleanup_fd' which is used to set a cleanup-on-exit file descriptor.
	*
	* The following parameters all provide return values. Several may be set
	* in the failure case and will contain additional information.
	*
	* The 'read_bytes' value will be set to the total number of bytes read.
	*
	* The 'errflags' value will contain zprop_errflags_t flags which are
	* used to describe any failures.
	*
	* The 'action_handle' and 'cleanup_fd' are no longer used, and are ignored.
	*
	* The 'errors' nvlist contains an entry for each unapplied received
	* property. Callers are responsible for freeing this nvlist.
	*/
	int
	lzc_receive_one(const char snapname, nvlist_t props,
	const char *origin, boolean_t force, boolean_t resumable, boolean_t raw,
	int input_fd, const dmu_replay_record_t *begin_record, int cleanup_fd,
	uint64_t read_bytes, uint64_t errflags, uint64_t *action_handle,
	nvlist_t **errors)
	{
	return (recv_impl(snapname, props, NULL, NULL, 0, origin, force,
	resumable, raw, input_fd, begin_record,
	read_bytes, errflags, errors));
	}

	/*
	* Like lzc_receive_one, but allows the caller to pass an additional 'cmdprops'
	* argument.
	*
	* The 'cmdprops' nvlist contains both override ('zfs receive -o') and
	* exclude ('zfs receive -x') properties. Callers are responsible for freeing
	* this nvlist
	*/
	int
	lzc_receive_with_cmdprops(const char snapname, nvlist_t props,
	nvlist_t cmdprops, uint8_t wkeydata, uint_t wkeylen, const char *origin,
	boolean_t force, boolean_t resumable, boolean_t raw, int input_fd,
	const dmu_replay_record_t *begin_record, int cleanup_fd,
	uint64_t read_bytes, uint64_t errflags, uint64_t *action_handle,
	nvlist_t **errors)
	{
	return (recv_impl(snapname, props, cmdprops, wkeydata, wkeylen, origin,
	force, resumable, raw, input_fd, begin_record,
	read_bytes, errflags, errors));
	}

	/*
	* Roll back this filesystem or volume to its most recent snapshot.
	* If snapnamebuf is not NULL, it will be filled in with the name
	* of the most recent snapshot.
	* Note that the latest snapshot may change if a new one is concurrently
	* created or the current one is destroyed. lzc_rollback_to can be used
	* to roll back to a specific latest snapshot.
	*
	* Return 0 on success or an errno on failure.
	*/
	int
	lzc_rollback(const char fsname, char snapnamebuf, int snapnamelen)
	{
	nvlist_t *args;
	nvlist_t *result;
	int err;

	args = fnvlist_alloc();
	err = lzc_ioctl(ZFS_IOC_ROLLBACK, fsname, args, &result);
	nvlist_free(args);
	if (err == 0 && snapnamebuf != NULL) {
	const char *snapname = fnvlist_lookup_string(result, "target");
	(void) strlcpy(snapnamebuf, snapname, snapnamelen);
	}
	nvlist_free(result);

	return (err);
	}

	/*
	* Roll back this filesystem or volume to the specified snapshot,
	* if possible.
	*
	* Return 0 on success or an errno on failure.
	*/
	int
	lzc_rollback_to(const char fsname, const char snapname)
	{
	nvlist_t *args;
	nvlist_t *result;
	int err;

	args = fnvlist_alloc();
	fnvlist_add_string(args, "target", snapname);
	err = lzc_ioctl(ZFS_IOC_ROLLBACK, fsname, args, &result);
	nvlist_free(args);
	nvlist_free(result);
	return (err);
	}

	/*
	* Creates new bookmarks from existing snapshot or bookmark.
	*
	* The bookmarks nvlist maps from the full name of the new bookmark to
	* the full name of the source snapshot or bookmark.
	* All the bookmarks and snapshots must be in the same pool.
	* The new bookmarks names must be unique.
	* => see function dsl_bookmark_create_nvl_validate
	*
	* The returned results nvlist will have an entry for each bookmark that failed.
	* The value will be the (int32) error code.
	*
	* The return value will be 0 if all bookmarks were created, otherwise it will
	* be the errno of a (undetermined) bookmarks that failed.
	*/
	int
	lzc_bookmark(nvlist_t bookmarks, nvlist_t *errlist)
	{
	nvpair_t *elem;
	int error;
	char pool[ZFS_MAX_DATASET_NAME_LEN];

	/* determine pool name from first bookmark */
	elem = nvlist_next_nvpair(bookmarks, NULL);
	if (elem == NULL)
	return (0);
	(void) strlcpy(pool, nvpair_name(elem), sizeof (pool));
	pool[strcspn(pool, "/#")] = '\0';

	error = lzc_ioctl(ZFS_IOC_BOOKMARK, pool, bookmarks, errlist);

	return (error);
	}

	/*
	* Retrieve bookmarks.
	*
	* Retrieve the list of bookmarks for the given file system. The props
	* parameter is an nvlist of property names (with no values) that will be
	* returned for each bookmark.
	*
	* The following are valid properties on bookmarks, most of which are numbers
	* (represented as uint64 in the nvlist), except redact_snaps, which is a
	* uint64 array, and redact_complete, which is a boolean
	*
	* "guid" - globally unique identifier of the snapshot it refers to
	* "createtxg" - txg when the snapshot it refers to was created
	* "creation" - timestamp when the snapshot it refers to was created
	* "ivsetguid" - IVset guid for identifying encrypted snapshots
	* "redact_snaps" - list of guids of the redaction snapshots for the specified
	* bookmark. If the bookmark is not a redaction bookmark, the nvlist will
	* not contain an entry for this value. If it is redacted with respect to
	* no snapshots, it will contain value -> NULL uint64 array
	* "redact_complete" - boolean value; true if the redaction bookmark is
	* complete, false otherwise.
	*
	* The format of the returned nvlist as follows:
	* <short name of bookmark> -> {
	* <name of property> -> {
	* "value" -> uint64
	* }
	* ...
	* "redact_snaps" -> {
	* "value" -> uint64 array
	* }
	* "redact_complete" -> {
	* "value" -> boolean value
	* }
	* }
	*/
	int
	lzc_get_bookmarks(const char fsname, nvlist_t props, nvlist_t **bmarks)
	{
	return (lzc_ioctl(ZFS_IOC_GET_BOOKMARKS, fsname, props, bmarks));
	}

	/*
	* Get bookmark properties.
	*
	* Given a bookmark's full name, retrieve all properties for the bookmark.
	*
	* The format of the returned property list is as follows:
	* {
	* <name of property> -> {
	* "value" -> uint64
	* }
	* ...
	* "redact_snaps" -> {
	* "value" -> uint64 array
	* }
	*/
	int
	lzc_get_bookmark_props(const char bookmark, nvlist_t *props)
	{
	int error;

	nvlist_t *innvl = fnvlist_alloc();
	error = lzc_ioctl(ZFS_IOC_GET_BOOKMARK_PROPS, bookmark, innvl, props);
	fnvlist_free(innvl);

	return (error);
	}

	/*
	* Destroys bookmarks.
	*
	* The keys in the bmarks nvlist are the bookmarks to be destroyed.
	* They must all be in the same pool. Bookmarks are specified as
	* <fs>#<bmark>.
	*
	* Bookmarks that do not exist will be silently ignored.
	*
	* The return value will be 0 if all bookmarks that existed were destroyed.
	*
	* Otherwise the return value will be the errno of a (undetermined) bookmark
	* that failed, no bookmarks will be destroyed, and the errlist will have an
	* entry for each bookmarks that failed. The value in the errlist will be
	* the (int32) error code.
	*/
	int
	lzc_destroy_bookmarks(nvlist_t bmarks, nvlist_t *errlist)
	{
	nvpair_t *elem;
	int error;
	char pool[ZFS_MAX_DATASET_NAME_LEN];

	/* determine the pool name */
	elem = nvlist_next_nvpair(bmarks, NULL);
	if (elem == NULL)
	return (0);
	(void) strlcpy(pool, nvpair_name(elem), sizeof (pool));
	pool[strcspn(pool, "/#")] = '\0';

	error = lzc_ioctl(ZFS_IOC_DESTROY_BOOKMARKS, pool, bmarks, errlist);

	return (error);
	}

	static int
	lzc_channel_program_impl(const char pool, const char program, boolean_t sync,
	uint64_t instrlimit, uint64_t memlimit, nvlist_t argnvl, nvlist_t *outnvl)
	{
	int error;
	nvlist_t *args;

	args = fnvlist_alloc();
	fnvlist_add_string(args, ZCP_ARG_PROGRAM, program);
	fnvlist_add_nvlist(args, ZCP_ARG_ARGLIST, argnvl);
	fnvlist_add_boolean_value(args, ZCP_ARG_SYNC, sync);
	fnvlist_add_uint64(args, ZCP_ARG_INSTRLIMIT, instrlimit);
	fnvlist_add_uint64(args, ZCP_ARG_MEMLIMIT, memlimit);
	error = lzc_ioctl(ZFS_IOC_CHANNEL_PROGRAM, pool, args, outnvl);
	fnvlist_free(args);

	return (error);
	}

	/*
	* Executes a channel program.
	*
	* If this function returns 0 the channel program was successfully loaded and
	* ran without failing. Note that individual commands the channel program ran
	* may have failed and the channel program is responsible for reporting such
	* errors through outnvl if they are important.
	*
	* This method may also return:
	*
	* EINVAL The program contains syntax errors, or an invalid memory or time
	* limit was given. No part of the channel program was executed.
	* If caused by syntax errors, 'outnvl' contains information about the
	* errors.
	*
	* ECHRNG The program was executed, but encountered a runtime error, such as
	* calling a function with incorrect arguments, invoking the error()
	* function directly, failing an assert() command, etc. Some portion
	* of the channel program may have executed and committed changes.
	* Information about the failure can be found in 'outnvl'.
	*
	* ENOMEM The program fully executed, but the output buffer was not large
	* enough to store the returned value. No output is returned through
	* 'outnvl'.
	*
	* ENOSPC The program was terminated because it exceeded its memory usage
	* limit. Some portion of the channel program may have executed and
	* committed changes to disk. No output is returned through 'outnvl'.
	*
	* ETIME The program was terminated because it exceeded its Lua instruction
	* limit. Some portion of the channel program may have executed and
	* committed changes to disk. No output is returned through 'outnvl'.
	*/
	int
	lzc_channel_program(const char pool, const char program, uint64_t instrlimit,
	uint64_t memlimit, nvlist_t argnvl, nvlist_t *outnvl)
	{
	return (lzc_channel_program_impl(pool, program, B_TRUE, instrlimit,
	memlimit, argnvl, outnvl));
	}

	/*
	* Creates a checkpoint for the specified pool.
	*
	* If this function returns 0 the pool was successfully checkpointed.
	*
	* This method may also return:
	*
	* ZFS_ERR_CHECKPOINT_EXISTS
	* The pool already has a checkpoint. A pools can only have one
	* checkpoint at most, at any given time.
	*
	* ZFS_ERR_DISCARDING_CHECKPOINT
	* ZFS is in the middle of discarding a checkpoint for this pool.
	* The pool can be checkpointed again once the discard is done.
	*
	* ZFS_DEVRM_IN_PROGRESS
	* A vdev is currently being removed. The pool cannot be
	* checkpointed until the device removal is done.
	*
	* ZFS_VDEV_TOO_BIG
	* One or more top-level vdevs exceed the maximum vdev size
	* supported for this feature.
	*/
	int
	lzc_pool_checkpoint(const char *pool)
	{
	int error;

	nvlist_t *result = NULL;
	nvlist_t *args = fnvlist_alloc();

	error = lzc_ioctl(ZFS_IOC_POOL_CHECKPOINT, pool, args, &result);

	fnvlist_free(args);
	fnvlist_free(result);

	return (error);
	}

	/*
	* Discard the checkpoint from the specified pool.
	*
	* If this function returns 0 the checkpoint was successfully discarded.
	*
	* This method may also return:
	*
	* ZFS_ERR_NO_CHECKPOINT
	* The pool does not have a checkpoint.
	*
	* ZFS_ERR_DISCARDING_CHECKPOINT
	* ZFS is already in the middle of discarding the checkpoint.
	*/
	int
	lzc_pool_checkpoint_discard(const char *pool)
	{
	int error;

	nvlist_t *result = NULL;
	nvlist_t *args = fnvlist_alloc();

	error = lzc_ioctl(ZFS_IOC_POOL_DISCARD_CHECKPOINT, pool, args, &result);

	fnvlist_free(args);
	fnvlist_free(result);

	return (error);
	}

	/*
	* Executes a read-only channel program.
	*
	* A read-only channel program works programmatically the same way as a
	* normal channel program executed with lzc_channel_program(). The only
	* difference is it runs exclusively in open-context and therefore can
	* return faster. The downside to that, is that the program cannot change
	* on-disk state by calling functions from the zfs.sync submodule.
	*
	* The return values of this function (and their meaning) are exactly the
	* same as the ones described in lzc_channel_program().
	*/
	int
	lzc_channel_program_nosync(const char pool, const char program,
	uint64_t timeout, uint64_t memlimit, nvlist_t argnvl, nvlist_t *outnvl)
	{
	return (lzc_channel_program_impl(pool, program, B_FALSE, timeout,
	memlimit, argnvl, outnvl));
	}

	+int
	+lzc_get_vdev_prop(const char poolname, nvlist_t innvl, nvlist_t **outnvl)
	+{
	+ return (lzc_ioctl(ZFS_IOC_VDEV_GET_PROPS, poolname, innvl, outnvl));
	+}
	+
	+int
	+lzc_set_vdev_prop(const char poolname, nvlist_t innvl, nvlist_t **outnvl)
	+{
	+ return (lzc_ioctl(ZFS_IOC_VDEV_SET_PROPS, poolname, innvl, outnvl));
	+}
	+
	/*
	* Performs key management functions
	*
	* crypto_cmd should be a value from dcp_cmd_t. If the command specifies to
	* load or change a wrapping key, the key should be specified in the
	* hidden_args nvlist so that it is not logged.
	*/
	int
	lzc_load_key(const char fsname, boolean_t noop, uint8_t wkeydata,
	uint_t wkeylen)
	{
	int error;
	nvlist_t *ioc_args;
	nvlist_t *hidden_args;

	if (wkeydata == NULL)
	return (EINVAL);

	ioc_args = fnvlist_alloc();
	hidden_args = fnvlist_alloc();
	fnvlist_add_uint8_array(hidden_args, "wkeydata", wkeydata, wkeylen);
	fnvlist_add_nvlist(ioc_args, ZPOOL_HIDDEN_ARGS, hidden_args);
	if (noop)
	fnvlist_add_boolean(ioc_args, "noop");
	error = lzc_ioctl(ZFS_IOC_LOAD_KEY, fsname, ioc_args, NULL);
	nvlist_free(hidden_args);
	nvlist_free(ioc_args);

	return (error);
	}

	int
	lzc_unload_key(const char *fsname)
	{
	return (lzc_ioctl(ZFS_IOC_UNLOAD_KEY, fsname, NULL, NULL));
	}

	int
	lzc_change_key(const char fsname, uint64_t crypt_cmd, nvlist_t props,
	uint8_t *wkeydata, uint_t wkeylen)
	{
	int error;
	nvlist_t *ioc_args = fnvlist_alloc();
	nvlist_t *hidden_args = NULL;

	fnvlist_add_uint64(ioc_args, "crypt_cmd", crypt_cmd);

	if (wkeydata != NULL) {
	hidden_args = fnvlist_alloc();
	fnvlist_add_uint8_array(hidden_args, "wkeydata", wkeydata,
	wkeylen);
	fnvlist_add_nvlist(ioc_args, ZPOOL_HIDDEN_ARGS, hidden_args);
	}

	if (props != NULL)
	fnvlist_add_nvlist(ioc_args, "props", props);

	error = lzc_ioctl(ZFS_IOC_CHANGE_KEY, fsname, ioc_args, NULL);
	nvlist_free(hidden_args);
	nvlist_free(ioc_args);

	return (error);
	}

	int
	lzc_reopen(const char *pool_name, boolean_t scrub_restart)
	{
	nvlist_t *args = fnvlist_alloc();
	int error;

	fnvlist_add_boolean_value(args, "scrub_restart", scrub_restart);

	error = lzc_ioctl(ZFS_IOC_POOL_REOPEN, pool_name, args, NULL);
	nvlist_free(args);
	return (error);
	}

	/*
	* Changes initializing state.
	*
	* vdevs should be a list of (<key>, guid) where guid is a uint64 vdev GUID.
	* The key is ignored.
	*
	* If there are errors related to vdev arguments, per-vdev errors are returned
	* in an nvlist with the key "vdevs". Each error is a (guid, errno) pair where
	* guid is stringified with PRIu64, and errno is one of the following as
	* an int64_t:
	* - ENODEV if the device was not found
	* - EINVAL if the devices is not a leaf or is not concrete (e.g. missing)
	* - EROFS if the device is not writeable
	* - EBUSY start requested but the device is already being either
	* initialized or trimmed
	* - ESRCH cancel/suspend requested but device is not being initialized
	*
	* If the errlist is empty, then return value will be:
	* - EINVAL if one or more arguments was invalid
	* - Other spa_open failures
	* - 0 if the operation succeeded
	*/
	int
	lzc_initialize(const char *poolname, pool_initialize_func_t cmd_type,
	nvlist_t vdevs, nvlist_t *errlist)
	{
	int error;

	nvlist_t *args = fnvlist_alloc();
	fnvlist_add_uint64(args, ZPOOL_INITIALIZE_COMMAND, (uint64_t)cmd_type);
	fnvlist_add_nvlist(args, ZPOOL_INITIALIZE_VDEVS, vdevs);

	error = lzc_ioctl(ZFS_IOC_POOL_INITIALIZE, poolname, args, errlist);

	fnvlist_free(args);

	return (error);
	}

	/*
	* Changes TRIM state.
	*
	* vdevs should be a list of (<key>, guid) where guid is a uint64 vdev GUID.
	* The key is ignored.
	*
	* If there are errors related to vdev arguments, per-vdev errors are returned
	* in an nvlist with the key "vdevs". Each error is a (guid, errno) pair where
	* guid is stringified with PRIu64, and errno is one of the following as
	* an int64_t:
	* - ENODEV if the device was not found
	* - EINVAL if the devices is not a leaf or is not concrete (e.g. missing)
	* - EROFS if the device is not writeable
	* - EBUSY start requested but the device is already being either trimmed
	* or initialized
	* - ESRCH cancel/suspend requested but device is not being initialized
	* - EOPNOTSUPP if the device does not support TRIM (or secure TRIM)
	*
	* If the errlist is empty, then return value will be:
	* - EINVAL if one or more arguments was invalid
	* - Other spa_open failures
	* - 0 if the operation succeeded
	*/
	int
	lzc_trim(const char *poolname, pool_trim_func_t cmd_type, uint64_t rate,
	boolean_t secure, nvlist_t vdevs, nvlist_t *errlist)
	{
	int error;

	nvlist_t *args = fnvlist_alloc();
	fnvlist_add_uint64(args, ZPOOL_TRIM_COMMAND, (uint64_t)cmd_type);
	fnvlist_add_nvlist(args, ZPOOL_TRIM_VDEVS, vdevs);
	fnvlist_add_uint64(args, ZPOOL_TRIM_RATE, rate);
	fnvlist_add_boolean_value(args, ZPOOL_TRIM_SECURE, secure);

	error = lzc_ioctl(ZFS_IOC_POOL_TRIM, poolname, args, errlist);

	fnvlist_free(args);

	return (error);
	}

	/*
	* Create a redaction bookmark named bookname by redacting snapshot with respect
	* to all the snapshots in snapnv.
	*/
	int
	lzc_redact(const char snapshot, const char bookname, nvlist_t *snapnv)
	{
	nvlist_t *args = fnvlist_alloc();
	fnvlist_add_string(args, "bookname", bookname);
	fnvlist_add_nvlist(args, "snapnv", snapnv);
	int error = lzc_ioctl(ZFS_IOC_REDACT, snapshot, args, NULL);
	fnvlist_free(args);
	return (error);
	}

	static int
	wait_common(const char *pool, zpool_wait_activity_t activity, boolean_t use_tag,
	uint64_t tag, boolean_t *waited)
	{
	nvlist_t *args = fnvlist_alloc();
	nvlist_t *result = NULL;

	fnvlist_add_int32(args, ZPOOL_WAIT_ACTIVITY, activity);
	if (use_tag)
	fnvlist_add_uint64(args, ZPOOL_WAIT_TAG, tag);

	int error = lzc_ioctl(ZFS_IOC_WAIT, pool, args, &result);

	if (error == 0 && waited != NULL)
	*waited = fnvlist_lookup_boolean_value(result,
	ZPOOL_WAIT_WAITED);

	fnvlist_free(args);
	fnvlist_free(result);

	return (error);
	}

	int
	lzc_wait(const char pool, zpool_wait_activity_t activity, boolean_t waited)
	{
	return (wait_common(pool, activity, B_FALSE, 0, waited));
	}

	int
	lzc_wait_tag(const char *pool, zpool_wait_activity_t activity, uint64_t tag,
	boolean_t *waited)
	{
	return (wait_common(pool, activity, B_TRUE, tag, waited));
	}

	int
	lzc_wait_fs(const char fs, zfs_wait_activity_t activity, boolean_t waited)
	{
	nvlist_t *args = fnvlist_alloc();
	nvlist_t *result = NULL;

	fnvlist_add_int32(args, ZFS_WAIT_ACTIVITY, activity);

	int error = lzc_ioctl(ZFS_IOC_WAIT_FS, fs, args, &result);

	if (error == 0 && waited != NULL)
	*waited = fnvlist_lookup_boolean_value(result,
	ZFS_WAIT_WAITED);

	fnvlist_free(args);
	fnvlist_free(result);

	return (error);
	}

	/*
	* Set the bootenv contents for the given pool.
	*/
	int
	lzc_set_bootenv(const char pool, const nvlist_t env)
	{
	return (lzc_ioctl(ZFS_IOC_SET_BOOTENV, pool, (nvlist_t *)env, NULL));
	}

	/*
	* Get the contents of the bootenv of the given pool.
	*/
	int
	lzc_get_bootenv(const char pool, nvlist_t *outnvl)
	{
	return (lzc_ioctl(ZFS_IOC_GET_BOOTENV, pool, NULL, outnvl));
	}
	diff --git a/sys/contrib/openzfs/lib/libzfsbootenv/libzfsbootenv.abi b/sys/contrib/openzfs/lib/libzfsbootenv/libzfsbootenv.abi
	index 86ec25cf8470..f1401a14f7a0 100644
	--- a/sys/contrib/openzfs/lib/libzfsbootenv/libzfsbootenv.abi
	+++ b/sys/contrib/openzfs/lib/libzfsbootenv/libzfsbootenv.abi
	@@ -1,201 +1,206 @@
	<abi-corpus version='2.0' architecture='elf-amd-x86_64' soname='libzfsbootenv.so.1'>
	<elf-needed>
	<dependency name='libzfs.so.4'/>
	<dependency name='libnvpair.so.3'/>
	<dependency name='libc.so.6'/>
	</elf-needed>
	<elf-function-symbols>
	+ <elf-symbol name='_fini' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	+ <elf-symbol name='_init' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzbe_add_pair' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzbe_bootenv_print' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzbe_get_boot_device' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzbe_nvlist_free' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzbe_nvlist_get' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzbe_nvlist_set' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzbe_remove_pair' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	<elf-symbol name='lzbe_set_boot_device' type='func-type' binding='global-binding' visibility='default-visibility' is-defined='yes'/>
	</elf-function-symbols>
	<abi-instr address-size='64' path='lzbe_device.c' language='LANG_C99'>
	<type-decl name='char' size-in-bits='8' id='a84c031d'/>
	<type-decl name='int' size-in-bits='32' id='95e97e5e'/>
	<type-decl name='unnamed-enum-underlying-type-32' is-anonymous='yes' size-in-bits='32' alignment-in-bits='32' id='9cac1fee'/>
	- <type-decl name='void' id='48b5725f'/>
	<enum-decl name='lzbe_flags' id='2b77720b'>
	<underlying-type type-id='9cac1fee'/>
	<enumerator name='lzbe_add' value='0'/>
	<enumerator name='lzbe_replace' value='1'/>
	</enum-decl>
	<typedef-decl name='lzbe_flags_t' type-id='2b77720b' id='a1936f04'/>
	<pointer-type-def type-id='a84c031d' size-in-bits='64' id='26a90f95'/>
	<pointer-type-def type-id='26a90f95' size-in-bits='64' id='9b23c9ad'/>
	<qualified-type-def type-id='a84c031d' const='yes' id='9b45d938'/>
	<pointer-type-def type-id='9b45d938' size-in-bits='64' id='80f4b756'/>
	<function-decl name='lzbe_set_boot_device' mangled-name='lzbe_set_boot_device' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzbe_set_boot_device'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='a1936f04' name='flag'/>
	<parameter type-id='80f4b756' name='device'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzbe_get_boot_device' mangled-name='lzbe_get_boot_device' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzbe_get_boot_device'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='9b23c9ad' name='device'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='lzbe_pair.c' language='LANG_C99'>
	<type-decl name='unsigned long int' size-in-bits='64' id='7359adad'/>
	<typedef-decl name='size_t' type-id='7359adad' id='b59d7dce'/>
	<pointer-type-def type-id='48b5725f' size-in-bits='64' id='eaa32e2f'/>
	<pointer-type-def type-id='eaa32e2f' size-in-bits='64' id='63e171df'/>
	<function-decl name='lzbe_nvlist_get' mangled-name='lzbe_nvlist_get' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzbe_nvlist_get'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='80f4b756' name='key'/>
	<parameter type-id='63e171df' name='ptr'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzbe_nvlist_set' mangled-name='lzbe_nvlist_set' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzbe_nvlist_set'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='80f4b756' name='key'/>
	<parameter type-id='eaa32e2f' name='ptr'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzbe_nvlist_free' mangled-name='lzbe_nvlist_free' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzbe_nvlist_free'>
	<parameter type-id='eaa32e2f' name='ptr'/>
	<return type-id='48b5725f'/>
	</function-decl>
	<function-decl name='lzbe_add_pair' mangled-name='lzbe_add_pair' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzbe_add_pair'>
	<parameter type-id='eaa32e2f' name='ptr'/>
	<parameter type-id='80f4b756' name='key'/>
	<parameter type-id='80f4b756' name='type'/>
	<parameter type-id='eaa32e2f' name='value'/>
	<parameter type-id='b59d7dce' name='size'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	<function-decl name='lzbe_remove_pair' mangled-name='lzbe_remove_pair' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzbe_remove_pair'>
	<parameter type-id='eaa32e2f' name='ptr'/>
	<parameter type-id='80f4b756' name='key'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	</abi-instr>
	<abi-instr address-size='64' path='lzbe_util.c' language='LANG_C99'>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='8' id='89feb1ec'>
	<subrange length='1' type-id='7359adad' id='52f813b4'/>
	</array-type-def>
	<array-type-def dimensions='1' type-id='a84c031d' size-in-bits='160' id='664ac0b7'>
	<subrange length='20' type-id='7359adad' id='fdca39cf'/>
	</array-type-def>
	- <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
	- <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
	- <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
	<type-decl name='long int' size-in-bits='64' id='bd54fe1a'/>
	<type-decl name='signed char' size-in-bits='8' id='28577a57'/>
	<type-decl name='unsigned short int' size-in-bits='16' id='8efea9e5'/>
	- <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
	- <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
	- <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
	<typedef-decl name='_IO_lock_t' type-id='48b5725f' id='bb4788fa'/>
	+ <class-decl name='_IO_marker' size-in-bits='192' is-struct='yes' visibility='default' id='010ae0b9'>
	+ <data-member access='public' layout-offset-in-bits='0'>
	+ <var-decl name='_next' type-id='e4c6fa61' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='64'>
	+ <var-decl name='_sbuf' type-id='dca988a5' visibility='default'/>
	+ </data-member>
	+ <data-member access='public' layout-offset-in-bits='128'>
	+ <var-decl name='_pos' type-id='95e97e5e' visibility='default'/>
	+ </data-member>
	+ </class-decl>
	<class-decl name='_IO_FILE' size-in-bits='1728' is-struct='yes' visibility='default' id='ec1ed955'>
	<data-member access='public' layout-offset-in-bits='0'>
	<var-decl name='_flags' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='64'>
	<var-decl name='_IO_read_ptr' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='128'>
	<var-decl name='_IO_read_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='192'>
	<var-decl name='_IO_read_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='256'>
	<var-decl name='_IO_write_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='320'>
	<var-decl name='_IO_write_ptr' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='384'>
	<var-decl name='_IO_write_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='448'>
	<var-decl name='_IO_buf_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='512'>
	<var-decl name='_IO_buf_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='576'>
	<var-decl name='_IO_save_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='640'>
	<var-decl name='_IO_backup_base' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='704'>
	<var-decl name='_IO_save_end' type-id='26a90f95' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='768'>
	<var-decl name='_markers' type-id='e4c6fa61' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='832'>
	<var-decl name='_chain' type-id='dca988a5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='896'>
	<var-decl name='_fileno' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='928'>
	<var-decl name='_flags2' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='960'>
	<var-decl name='_old_offset' type-id='79989e9c' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1024'>
	<var-decl name='_cur_column' type-id='8efea9e5' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1040'>
	<var-decl name='_vtable_offset' type-id='28577a57' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1048'>
	<var-decl name='_shortbuf' type-id='89feb1ec' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1088'>
	<var-decl name='_lock' type-id='cecf4ea7' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1152'>
	<var-decl name='_offset' type-id='724e4de6' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1216'>
	- <var-decl name='_codecvt' type-id='570f8c59' visibility='default'/>
	+ <var-decl name='__pad1' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1280'>
	- <var-decl name='_wide_data' type-id='c65a1f29' visibility='default'/>
	+ <var-decl name='__pad2' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1344'>
	- <var-decl name='_freeres_list' type-id='dca988a5' visibility='default'/>
	+ <var-decl name='__pad3' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1408'>
	- <var-decl name='_freeres_buf' type-id='eaa32e2f' visibility='default'/>
	+ <var-decl name='__pad4' type-id='eaa32e2f' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1472'>
	<var-decl name='__pad5' type-id='b59d7dce' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1536'>
	<var-decl name='_mode' type-id='95e97e5e' visibility='default'/>
	</data-member>
	<data-member access='public' layout-offset-in-bits='1568'>
	<var-decl name='_unused2' type-id='664ac0b7' visibility='default'/>
	</data-member>
	</class-decl>
	+ <typedef-decl name='__off_t' type-id='bd54fe1a' id='79989e9c'/>
	+ <typedef-decl name='__off64_t' type-id='bd54fe1a' id='724e4de6'/>
	+ <typedef-decl name='FILE' type-id='ec1ed955' id='aa12d1ba'/>
	<pointer-type-def type-id='aa12d1ba' size-in-bits='64' id='822cd80b'/>
	<pointer-type-def type-id='ec1ed955' size-in-bits='64' id='dca988a5'/>
	- <pointer-type-def type-id='a4036571' size-in-bits='64' id='570f8c59'/>
	<pointer-type-def type-id='bb4788fa' size-in-bits='64' id='cecf4ea7'/>
	<pointer-type-def type-id='010ae0b9' size-in-bits='64' id='e4c6fa61'/>
	- <pointer-type-def type-id='79bd3751' size-in-bits='64' id='c65a1f29'/>
	- <class-decl name='_IO_codecvt' is-struct='yes' visibility='default' is-declaration-only='yes' id='a4036571'/>
	- <class-decl name='_IO_marker' is-struct='yes' visibility='default' is-declaration-only='yes' id='010ae0b9'/>
	- <class-decl name='_IO_wide_data' is-struct='yes' visibility='default' is-declaration-only='yes' id='79bd3751'/>
	<function-decl name='lzbe_bootenv_print' mangled-name='lzbe_bootenv_print' visibility='default' binding='global' size-in-bits='64' elf-symbol-id='lzbe_bootenv_print'>
	<parameter type-id='80f4b756' name='pool'/>
	<parameter type-id='80f4b756' name='nvlist'/>
	<parameter type-id='822cd80b' name='of'/>
	<return type-id='95e97e5e'/>
	</function-decl>
	+ <type-decl name='void' id='48b5725f'/>
	</abi-instr>
	</abi-corpus>
	diff --git a/sys/contrib/openzfs/lib/libzfsbootenv/lzbe_pair.c b/sys/contrib/openzfs/lib/libzfsbootenv/lzbe_pair.c
	index 831355ba4b7c..b92225984517 100644
	--- a/sys/contrib/openzfs/lib/libzfsbootenv/lzbe_pair.c
	+++ b/sys/contrib/openzfs/lib/libzfsbootenv/lzbe_pair.c
	@@ -1,347 +1,348 @@
	/*
	* This file and its contents are supplied under the terms of the
	* Common Development and Distribution License ("CDDL"), version 1.0.
	* You may only use this file in accordance with the terms of version
	* 1.0 of the CDDL.
	*
	* A full copy of the text of the CDDL should have accompanied this
	* source. A copy of the CDDL is also available via the Internet at
	* http://www.illumos.org/license/CDDL.
	*/
	/*
	* Copyright 2020 Toomas Soome <tsoome@me.com>
	*/

	#include <sys/types.h>
	#include <string.h>
	#include <libzfs.h>
	#include <libzfsbootenv.h>
	#include <sys/zfs_bootenv.h>
	#include <sys/vdev_impl.h>

	/*
	* Get or create nvlist. If key is not NULL, get nvlist from bootenv,
	* otherwise return bootenv.
	*/
	int
	lzbe_nvlist_get(const char pool, const char key, void **ptr)
	{
	libzfs_handle_t *hdl;
	zpool_handle_t *zphdl;
	nvlist_t *nv;
	int rv = -1;

	if (pool == NULL \|\| *pool == '\0')
	return (rv);

	if ((hdl = libzfs_init()) == NULL) {
	return (rv);
	}

	zphdl = zpool_open(hdl, pool);
	if (zphdl == NULL) {
	libzfs_fini(hdl);
	return (rv);
	}

	rv = zpool_get_bootenv(zphdl, &nv);
	if (rv == 0) {
	nvlist_t nvl, dup;

	if (key != NULL) {
	rv = nvlist_lookup_nvlist(nv, key, &nvl);
	if (rv == 0) {
	rv = nvlist_dup(nvl, &dup, 0);
	nvlist_free(nv);
	if (rv == 0)
	nv = dup;
	else
	nv = NULL;
	} else {
	nvlist_free(nv);
	rv = nvlist_alloc(&nv, NV_UNIQUE_NAME, 0);
	}
	}
	*ptr = nv;
	}

	zpool_close(zphdl);
	libzfs_fini(hdl);
	return (rv);
	}

	int
	lzbe_nvlist_set(const char pool, const char key, void *ptr)
	{
	libzfs_handle_t *hdl;
	zpool_handle_t *zphdl;
	nvlist_t *nv;
	uint64_t version;
	int rv = -1;

	if (pool == NULL \|\| *pool == '\0')
	return (rv);

	if ((hdl = libzfs_init()) == NULL) {
	return (rv);
	}

	zphdl = zpool_open(hdl, pool);
	if (zphdl == NULL) {
	libzfs_fini(hdl);
	return (rv);
	}

	if (key != NULL) {
	rv = zpool_get_bootenv(zphdl, &nv);
	if (rv == 0) {
	/*
	* We got the nvlist, check for version.
	* if version is missing or is not VB_NVLIST,
	* create new list.
	*/
	rv = nvlist_lookup_uint64(nv, BOOTENV_VERSION,
	&version);
	if (rv != 0 \|\| version != VB_NVLIST) {
	/* Drop this nvlist */
	fnvlist_free(nv);
	/* Create and prepare new nvlist */
	nv = fnvlist_alloc();
	fnvlist_add_uint64(nv, BOOTENV_VERSION,
	VB_NVLIST);
	}
	rv = nvlist_add_nvlist(nv, key, ptr);
	if (rv == 0)
	rv = zpool_set_bootenv(zphdl, nv);
	nvlist_free(nv);
	}
	} else {
	rv = zpool_set_bootenv(zphdl, ptr);
	}

	zpool_close(zphdl);
	libzfs_fini(hdl);
	return (rv);
	}

	/*
	* free nvlist we got via lzbe_nvlist_get()
	*/
	void
	lzbe_nvlist_free(void *ptr)
	{
	nvlist_free(ptr);
	}

	static const char *typenames[] = {
	"DATA_TYPE_UNKNOWN",
	"DATA_TYPE_BOOLEAN",
	"DATA_TYPE_BYTE",
	"DATA_TYPE_INT16",
	"DATA_TYPE_UINT16",
	"DATA_TYPE_INT32",
	"DATA_TYPE_UINT32",
	"DATA_TYPE_INT64",
	"DATA_TYPE_UINT64",
	"DATA_TYPE_STRING",
	"DATA_TYPE_BYTE_ARRAY",
	"DATA_TYPE_INT16_ARRAY",
	"DATA_TYPE_UINT16_ARRAY",
	"DATA_TYPE_INT32_ARRAY",
	"DATA_TYPE_UINT32_ARRAY",
	"DATA_TYPE_INT64_ARRAY",
	"DATA_TYPE_UINT64_ARRAY",
	"DATA_TYPE_STRING_ARRAY",
	"DATA_TYPE_HRTIME",
	"DATA_TYPE_NVLIST",
	"DATA_TYPE_NVLIST_ARRAY",
	"DATA_TYPE_BOOLEAN_VALUE",
	"DATA_TYPE_INT8",
	"DATA_TYPE_UINT8",
	"DATA_TYPE_BOOLEAN_ARRAY",
	"DATA_TYPE_INT8_ARRAY",
	"DATA_TYPE_UINT8_ARRAY"
	};

	static int
	nvpair_type_from_name(const char *name)
	{
	unsigned i;

	for (i = 0; i < ARRAY_SIZE(typenames); i++) {
	if (strcmp(name, typenames[i]) == 0)
	return (i);
	}
	return (0);
	}

	/*
	* Add pair defined by key, type and value into nvlist.
	*/
	int
	lzbe_add_pair(void ptr, const char key, const char type, void value,
	size_t size)
	{
	nvlist_t *nv = ptr;
	data_type_t dt;
	int rv = 0;

	if (ptr == NULL \|\| key == NULL \|\| value == NULL)
	return (rv);

	if (type == NULL)
	type = "DATA_TYPE_STRING";
	dt = nvpair_type_from_name(type);
	if (dt == DATA_TYPE_UNKNOWN)
	return (EINVAL);

	switch (dt) {
	case DATA_TYPE_BYTE:
	if (size != sizeof (uint8_t)) {
	rv = EINVAL;
	break;
	}
	rv = nvlist_add_byte(nv, key, (uint8_t )value);
	break;

	case DATA_TYPE_INT16:
	if (size != sizeof (int16_t)) {
	rv = EINVAL;
	break;
	}
	rv = nvlist_add_int16(nv, key, (int16_t )value);
	break;

	case DATA_TYPE_UINT16:
	if (size != sizeof (uint16_t)) {
	rv = EINVAL;
	break;
	}
	rv = nvlist_add_uint16(nv, key, (uint16_t )value);
	break;

	case DATA_TYPE_INT32:
	if (size != sizeof (int32_t)) {
	rv = EINVAL;
	break;
	}
	rv = nvlist_add_int32(nv, key, (int32_t )value);
	break;

	case DATA_TYPE_UINT32:
	if (size != sizeof (uint32_t)) {
	rv = EINVAL;
	break;
	}
	rv = nvlist_add_uint32(nv, key, (uint32_t )value);
	break;

	case DATA_TYPE_INT64:
	if (size != sizeof (int64_t)) {
	rv = EINVAL;
	break;
	}
	rv = nvlist_add_int64(nv, key, (int64_t )value);
	break;

	case DATA_TYPE_UINT64:
	if (size != sizeof (uint64_t)) {
	rv = EINVAL;
	break;
	}
	rv = nvlist_add_uint64(nv, key, (uint64_t )value);
	break;

	case DATA_TYPE_STRING:
	rv = nvlist_add_string(nv, key, value);
	break;

	case DATA_TYPE_BYTE_ARRAY:
	rv = nvlist_add_byte_array(nv, key, value, size);
	break;

	case DATA_TYPE_INT16_ARRAY:
	rv = nvlist_add_int16_array(nv, key, value, size);
	break;

	case DATA_TYPE_UINT16_ARRAY:
	rv = nvlist_add_uint16_array(nv, key, value, size);
	break;

	case DATA_TYPE_INT32_ARRAY:
	rv = nvlist_add_int32_array(nv, key, value, size);
	break;

	case DATA_TYPE_UINT32_ARRAY:
	rv = nvlist_add_uint32_array(nv, key, value, size);
	break;

	case DATA_TYPE_INT64_ARRAY:
	rv = nvlist_add_int64_array(nv, key, value, size);
	break;

	case DATA_TYPE_UINT64_ARRAY:
	rv = nvlist_add_uint64_array(nv, key, value, size);
	break;

	case DATA_TYPE_STRING_ARRAY:
	rv = nvlist_add_string_array(nv, key, value, size);
	break;

	case DATA_TYPE_NVLIST:
	rv = nvlist_add_nvlist(nv, key, (nvlist_t *)value);
	break;

	case DATA_TYPE_NVLIST_ARRAY:
	- rv = nvlist_add_nvlist_array(nv, key, value, size);
	+ rv = nvlist_add_nvlist_array(nv, key, (const nvlist_t **)value,
	+ size);
	break;

	case DATA_TYPE_BOOLEAN_VALUE:
	if (size != sizeof (boolean_t)) {
	rv = EINVAL;
	break;
	}
	rv = nvlist_add_boolean_value(nv, key, (boolean_t )value);
	break;

	case DATA_TYPE_INT8:
	if (size != sizeof (int8_t)) {
	rv = EINVAL;
	break;
	}
	rv = nvlist_add_int8(nv, key, (int8_t )value);
	break;

	case DATA_TYPE_UINT8:
	if (size != sizeof (uint8_t)) {
	rv = EINVAL;
	break;
	}
	rv = nvlist_add_uint8(nv, key, (uint8_t )value);
	break;

	case DATA_TYPE_BOOLEAN_ARRAY:
	rv = nvlist_add_boolean_array(nv, key, value, size);
	break;

	case DATA_TYPE_INT8_ARRAY:
	rv = nvlist_add_int8_array(nv, key, value, size);
	break;

	case DATA_TYPE_UINT8_ARRAY:
	rv = nvlist_add_uint8_array(nv, key, value, size);
	break;

	default:
	return (ENOTSUP);
	}

	return (rv);
	}

	int
	lzbe_remove_pair(void ptr, const char key)
	{

	return (nvlist_remove_all(ptr, key));
	}
	diff --git a/sys/contrib/openzfs/lib/libzpool/Makefile.am b/sys/contrib/openzfs/lib/libzpool/Makefile.am
	index 3cc0c2f2ec05..dce3f81b829d 100644
	--- a/sys/contrib/openzfs/lib/libzpool/Makefile.am
	+++ b/sys/contrib/openzfs/lib/libzpool/Makefile.am
	@@ -1,240 +1,240 @@
	include $(top_srcdir)/config/Rules.am

	VPATH = \
	$(top_srcdir)/module/zfs \
	$(top_srcdir)/module/zcommon \
	$(top_srcdir)/module/lua \
	$(top_srcdir)/module/os/linux/zfs \
	$(top_srcdir)/lib/libzpool

	if BUILD_FREEBSD
	DEFAULT_INCLUDES += -I$(top_srcdir)/include/os/freebsd/zfs
	endif
	if BUILD_LINUX
	DEFAULT_INCLUDES += -I$(top_srcdir)/include/os/linux/zfs
	endif

	# Unconditionally enable debugging for libzpool
	AM_CPPFLAGS += -DDEBUG -UNDEBUG -DZFS_DEBUG

	# Suppress unused but set variable warnings often due to ASSERTs
	AM_CFLAGS += $(NO_UNUSED_BUT_SET_VARIABLE)

	# Includes kernel code generate warnings for large stack frames
	AM_CFLAGS += $(FRAME_LARGER_THAN)

	AM_CFLAGS += $(ZLIB_CFLAGS)

	AM_CFLAGS += -DLIB_ZPOOL_BUILD

	lib_LTLIBRARIES = libzpool.la

	USER_C = \
	kernel.c \
	taskq.c \
	util.c

	KERNEL_C = \
	zfeature_common.c \
	zfs_comutil.c \
	zfs_deleg.c \
	zfs_fletcher.c \
	zfs_fletcher_aarch64_neon.c \
	zfs_fletcher_avx512.c \
	zfs_fletcher_intel.c \
	zfs_fletcher_sse.c \
	zfs_fletcher_superscalar.c \
	zfs_fletcher_superscalar4.c \
	zfs_namecheck.c \
	zfs_prop.c \
	zpool_prop.c \
	zprop_common.c \
	abd.c \
	abd_os.c \
	aggsum.c \
	arc.c \
	arc_os.c \
	blkptr.c \
	bplist.c \
	bpobj.c \
	bptree.c \
	- btree.c \
	bqueue.c \
	+ btree.c \
	cityhash.c \
	dbuf.c \
	dbuf_stats.c \
	ddt.c \
	ddt_zap.c \
	dmu.c \
	dmu_diff.c \
	dmu_object.c \
	dmu_objset.c \
	dmu_recv.c \
	dmu_redact.c \
	dmu_send.c \
	dmu_traverse.c \
	dmu_tx.c \
	dmu_zfetch.c \
	dnode.c \
	dnode_sync.c \
	dsl_bookmark.c \
	+ dsl_crypt.c \
	dsl_dataset.c \
	dsl_deadlist.c \
	dsl_deleg.c \
	+ dsl_destroy.c \
	dsl_dir.c \
	- dsl_crypt.c \
	dsl_pool.c \
	dsl_prop.c \
	dsl_scan.c \
	dsl_synctask.c \
	- dsl_destroy.c \
	dsl_userhold.c \
	edonr_zfs.c \
	- hkdf.c \
	fm.c \
	gzip.c \
	- lzjb.c \
	+ hkdf.c \
	lz4.c \
	+ lzjb.c \
	metaslab.c \
	mmp.c \
	multilist.c \
	objlist.c \
	pathname.c \
	range_tree.c \
	refcount.c \
	rrwlock.c \
	sa.c \
	sha256.c \
	skein_zfs.c \
	spa.c \
	spa_boot.c \
	spa_checkpoint.c \
	spa_config.c \
	spa_errlog.c \
	spa_history.c \
	spa_log_spacemap.c \
	spa_misc.c \
	spa_stats.c \
	space_map.c \
	space_reftree.c \
	- txg.c \
	trace.c \
	+ txg.c \
	uberblock.c \
	unique.c \
	vdev.c \
	vdev_cache.c \
	vdev_draid.c \
	vdev_draid_rand.c \
	vdev_file.c \
	- vdev_indirect_births.c \
	vdev_indirect.c \
	+ vdev_indirect_births.c \
	vdev_indirect_mapping.c \
	vdev_initialize.c \
	vdev_label.c \
	vdev_mirror.c \
	vdev_missing.c \
	vdev_queue.c \
	vdev_raidz.c \
	+ vdev_raidz_math.c \
	vdev_raidz_math_aarch64_neon.c \
	vdev_raidz_math_aarch64_neonx2.c \
	vdev_raidz_math_avx2.c \
	vdev_raidz_math_avx512bw.c \
	vdev_raidz_math_avx512f.c \
	- vdev_raidz_math.c \
	+ vdev_raidz_math_powerpc_altivec.c \
	vdev_raidz_math_scalar.c \
	vdev_raidz_math_sse2.c \
	vdev_raidz_math_ssse3.c \
	- vdev_raidz_math_powerpc_altivec.c \
	vdev_rebuild.c \
	vdev_removal.c \
	vdev_root.c \
	vdev_trim.c \
	zap.c \
	zap_leaf.c \
	zap_micro.c \
	zcp.c \
	zcp_get.c \
	zcp_global.c \
	zcp_iter.c \
	zcp_set.c \
	zcp_synctask.c \
	zfeature.c \
	zfs_byteswap.c \
	zfs_debug.c \
	zfs_fm.c \
	zfs_fuid.c \
	zfs_racct.c \
	zfs_sa.c \
	- zfs_znode.c \
	zfs_ratelimit.c \
	zfs_rlock.c \
	+ zfs_znode.c \
	zil.c \
	zio.c \
	zio_checksum.c \
	zio_compress.c \
	zio_crypt.c \
	zio_inject.c \
	zle.c \
	zrlock.c \
	zthr.c

	LUA_C = \
	lapi.c \
	lauxlib.c \
	lbaselib.c \
	lcode.c \
	lcompat.c \
	lcorolib.c \
	lctype.c \
	ldebug.c \
	ldo.c \
	lfunc.c \
	lgc.c \
	llex.c \
	lmem.c \
	lobject.c \
	lopcodes.c \
	lparser.c \
	lstate.c \
	lstring.c \
	lstrlib.c \
	ltable.c \
	ltablib.c \
	ltm.c \
	lvm.c \
	lzio.c

	dist_libzpool_la_SOURCES = \
	$(USER_C)

	nodist_libzpool_la_SOURCES = \
	$(KERNEL_C) \
	$(LUA_C)

	libzpool_la_LIBADD = \
	$(abs_top_builddir)/lib/libicp/libicp.la \
	$(abs_top_builddir)/lib/libunicode/libunicode.la \
	$(abs_top_builddir)/lib/libnvpair/libnvpair.la \
	$(abs_top_builddir)/lib/libzstd/libzstd.la \
	$(abs_top_builddir)/lib/libzutil/libzutil.la

	libzpool_la_LIBADD += $(LIBCLOCK_GETTIME) $(ZLIB_LIBS) -ldl -lm

	libzpool_la_LDFLAGS = -pthread

	if !ASAN_ENABLED
	libzpool_la_LDFLAGS += -Wl,-z,defs
	endif

	if BUILD_FREEBSD
	libzpool_la_LIBADD += -lgeom
	endif

	libzpool_la_LDFLAGS += -version-info 5:0:0

	if TARGET_CPU_POWERPC
	vdev_raidz_math_powerpc_altivec.$(OBJEXT): CFLAGS += -maltivec
	vdev_raidz_math_powerpc_altivec.l$(OBJEXT): CFLAGS += -maltivec
	endif

	include $(top_srcdir)/config/CppCheck.am
	diff --git a/sys/contrib/openzfs/lib/libzutil/zutil_import.c b/sys/contrib/openzfs/lib/libzutil/zutil_import.c
	index 9eb55aaf77ce..4d956fdd8056 100644
	--- a/sys/contrib/openzfs/lib/libzutil/zutil_import.c
	+++ b/sys/contrib/openzfs/lib/libzutil/zutil_import.c
	@@ -1,1929 +1,1930 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright 2015 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2012, 2018 by Delphix. All rights reserved.
	* Copyright 2015 RackTop Systems.
	* Copyright (c) 2016, Intel Corporation.
	* Copyright (c) 2021, Colm Buckley <colm@tuatha.org>
	*/

	/*
	* Pool import support functions.
	*
	* Used by zpool, ztest, zdb, and zhack to locate importable configs. Since
	* these commands are expected to run in the global zone, we can assume
	* that the devices are all readable when called.
	*
	* To import a pool, we rely on reading the configuration information from the
	* ZFS label of each device. If we successfully read the label, then we
	* organize the configuration information in the following hierarchy:
	*
	* pool guid -> toplevel vdev guid -> label txg
	*
	* Duplicate entries matching this same tuple will be discarded. Once we have
	* examined every device, we pick the best label txg config for each toplevel
	* vdev. We then arrange these toplevel vdevs into a complete pool config, and
	* update any paths that have changed. Finally, we attempt to import the pool
	* using our derived config, and record the results.
	*/

	#include <aio.h>
	#include <ctype.h>
	#include <dirent.h>
	#include <errno.h>
	#include <libintl.h>
	#include <libgen.h>
	#include <stddef.h>
	#include <stdlib.h>
	#include <string.h>
	#include <sys/stat.h>
	#include <unistd.h>
	#include <fcntl.h>
	#include <sys/dktp/fdisk.h>
	#include <sys/vdev_impl.h>
	#include <sys/fs/zfs.h>

	#include <thread_pool.h>
	#include <libzutil.h>
	#include <libnvpair.h>

	#include "zutil_import.h"

	static __attribute__((format(printf, 2, 3))) void
	zutil_error_aux(libpc_handle_t hdl, const char fmt, ...)
	{
	va_list ap;

	va_start(ap, fmt);

	(void) vsnprintf(hdl->lpc_desc, sizeof (hdl->lpc_desc), fmt, ap);
	hdl->lpc_desc_active = B_TRUE;

	va_end(ap);
	}

	static void
	zutil_verror(libpc_handle_t hdl, const char error, const char *fmt,
	va_list ap)
	{
	char action[1024];

	(void) vsnprintf(action, sizeof (action), fmt, ap);

	if (hdl->lpc_desc_active)
	hdl->lpc_desc_active = B_FALSE;
	else
	hdl->lpc_desc[0] = '\0';

	if (hdl->lpc_printerr) {
	if (hdl->lpc_desc[0] != '\0')
	error = hdl->lpc_desc;

	(void) fprintf(stderr, "%s: %s\n", action, error);
	}
	}

	static __attribute__((format(printf, 3, 4))) int
	zutil_error_fmt(libpc_handle_t hdl, const char error, const char *fmt, ...)
	{
	va_list ap;

	va_start(ap, fmt);

	zutil_verror(hdl, error, fmt, ap);

	va_end(ap);

	return (-1);
	}

	static int
	zutil_error(libpc_handle_t hdl, const char error, const char *msg)
	{
	return (zutil_error_fmt(hdl, error, "%s", msg));
	}

	static int
	zutil_no_memory(libpc_handle_t *hdl)
	{
	zutil_error(hdl, EZFS_NOMEM, "internal error");
	exit(1);
	}

	void *
	zutil_alloc(libpc_handle_t *hdl, size_t size)
	{
	void *data;

	if ((data = calloc(1, size)) == NULL)
	(void) zutil_no_memory(hdl);

	return (data);
	}

	char *
	zutil_strdup(libpc_handle_t hdl, const char str)
	{
	char *ret;

	if ((ret = strdup(str)) == NULL)
	(void) zutil_no_memory(hdl);

	return (ret);
	}

	static char *
	zutil_strndup(libpc_handle_t hdl, const char str, size_t n)
	{
	char *ret;

	if ((ret = strndup(str, n)) == NULL)
	(void) zutil_no_memory(hdl);

	return (ret);
	}

	/*
	* Intermediate structures used to gather configuration information.
	*/
	typedef struct config_entry {
	uint64_t ce_txg;
	nvlist_t *ce_config;
	struct config_entry *ce_next;
	} config_entry_t;

	typedef struct vdev_entry {
	uint64_t ve_guid;
	config_entry_t *ve_configs;
	struct vdev_entry *ve_next;
	} vdev_entry_t;

	typedef struct pool_entry {
	uint64_t pe_guid;
	vdev_entry_t *pe_vdevs;
	struct pool_entry *pe_next;
	} pool_entry_t;

	typedef struct name_entry {
	char *ne_name;
	uint64_t ne_guid;
	uint64_t ne_order;
	uint64_t ne_num_labels;
	struct name_entry *ne_next;
	} name_entry_t;

	typedef struct pool_list {
	pool_entry_t *pools;
	name_entry_t *names;
	} pool_list_t;

	/*
	* Go through and fix up any path and/or devid information for the given vdev
	* configuration.
	*/
	static int
	fix_paths(libpc_handle_t hdl, nvlist_t nv, name_entry_t *names)
	{
	nvlist_t **child;
	uint_t c, children;
	uint64_t guid;
	name_entry_t ne, best;
	char *path;

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) == 0) {
	for (c = 0; c < children; c++)
	if (fix_paths(hdl, child[c], names) != 0)
	return (-1);
	return (0);
	}

	/*
	* This is a leaf (file or disk) vdev. In either case, go through
	* the name list and see if we find a matching guid. If so, replace
	* the path and see if we can calculate a new devid.
	*
	* There may be multiple names associated with a particular guid, in
	* which case we have overlapping partitions or multiple paths to the
	* same disk. In this case we prefer to use the path name which
	* matches the ZPOOL_CONFIG_PATH. If no matching entry is found we
	* use the lowest order device which corresponds to the first match
	* while traversing the ZPOOL_IMPORT_PATH search path.
	*/
	verify(nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) == 0);
	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) != 0)
	path = NULL;

	best = NULL;
	for (ne = names; ne != NULL; ne = ne->ne_next) {
	if (ne->ne_guid == guid) {
	if (path == NULL) {
	best = ne;
	break;
	}

	if ((strlen(path) == strlen(ne->ne_name)) &&
	strncmp(path, ne->ne_name, strlen(path)) == 0) {
	best = ne;
	break;
	}

	if (best == NULL) {
	best = ne;
	continue;
	}

	/* Prefer paths with move vdev labels. */
	if (ne->ne_num_labels > best->ne_num_labels) {
	best = ne;
	continue;
	}

	/* Prefer paths earlier in the search order. */
	if (ne->ne_num_labels == best->ne_num_labels &&
	ne->ne_order < best->ne_order) {
	best = ne;
	continue;
	}
	}
	}

	if (best == NULL)
	return (0);

	if (nvlist_add_string(nv, ZPOOL_CONFIG_PATH, best->ne_name) != 0)
	return (-1);

	update_vdev_config_dev_strs(nv);

	return (0);
	}

	/*
	* Add the given configuration to the list of known devices.
	*/
	static int
	add_config(libpc_handle_t hdl, pool_list_t pl, const char *path,
	int order, int num_labels, nvlist_t *config)
	{
	uint64_t pool_guid, vdev_guid, top_guid, txg, state;
	pool_entry_t *pe;
	vdev_entry_t *ve;
	config_entry_t *ce;
	name_entry_t *ne;

	/*
	* If this is a hot spare not currently in use or level 2 cache
	* device, add it to the list of names to translate, but don't do
	* anything else.
	*/
	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE,
	&state) == 0 &&
	(state == POOL_STATE_SPARE \|\| state == POOL_STATE_L2CACHE) &&
	nvlist_lookup_uint64(config, ZPOOL_CONFIG_GUID, &vdev_guid) == 0) {
	if ((ne = zutil_alloc(hdl, sizeof (name_entry_t))) == NULL)
	return (-1);

	if ((ne->ne_name = zutil_strdup(hdl, path)) == NULL) {
	free(ne);
	return (-1);
	}
	ne->ne_guid = vdev_guid;
	ne->ne_order = order;
	ne->ne_num_labels = num_labels;
	ne->ne_next = pl->names;
	pl->names = ne;

	return (0);
	}

	/*
	* If we have a valid config but cannot read any of these fields, then
	* it means we have a half-initialized label. In vdev_label_init()
	* we write a label with txg == 0 so that we can identify the device
	* in case the user refers to the same disk later on. If we fail to
	* create the pool, we'll be left with a label in this state
	* which should not be considered part of a valid pool.
	*/
	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
	&pool_guid) != 0 \|\|
	nvlist_lookup_uint64(config, ZPOOL_CONFIG_GUID,
	&vdev_guid) != 0 \|\|
	nvlist_lookup_uint64(config, ZPOOL_CONFIG_TOP_GUID,
	&top_guid) != 0 \|\|
	nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG,
	&txg) != 0 \|\| txg == 0) {
	return (0);
	}

	/*
	* First, see if we know about this pool. If not, then add it to the
	* list of known pools.
	*/
	for (pe = pl->pools; pe != NULL; pe = pe->pe_next) {
	if (pe->pe_guid == pool_guid)
	break;
	}

	if (pe == NULL) {
	if ((pe = zutil_alloc(hdl, sizeof (pool_entry_t))) == NULL) {
	return (-1);
	}
	pe->pe_guid = pool_guid;
	pe->pe_next = pl->pools;
	pl->pools = pe;
	}

	/*
	* Second, see if we know about this toplevel vdev. Add it if its
	* missing.
	*/
	for (ve = pe->pe_vdevs; ve != NULL; ve = ve->ve_next) {
	if (ve->ve_guid == top_guid)
	break;
	}

	if (ve == NULL) {
	if ((ve = zutil_alloc(hdl, sizeof (vdev_entry_t))) == NULL) {
	return (-1);
	}
	ve->ve_guid = top_guid;
	ve->ve_next = pe->pe_vdevs;
	pe->pe_vdevs = ve;
	}

	/*
	* Third, see if we have a config with a matching transaction group. If
	* so, then we do nothing. Otherwise, add it to the list of known
	* configs.
	*/
	for (ce = ve->ve_configs; ce != NULL; ce = ce->ce_next) {
	if (ce->ce_txg == txg)
	break;
	}

	if (ce == NULL) {
	if ((ce = zutil_alloc(hdl, sizeof (config_entry_t))) == NULL) {
	return (-1);
	}
	ce->ce_txg = txg;
	ce->ce_config = fnvlist_dup(config);
	ce->ce_next = ve->ve_configs;
	ve->ve_configs = ce;
	}

	/*
	* At this point we've successfully added our config to the list of
	* known configs. The last thing to do is add the vdev guid -> path
	* mappings so that we can fix up the configuration as necessary before
	* doing the import.
	*/
	if ((ne = zutil_alloc(hdl, sizeof (name_entry_t))) == NULL)
	return (-1);

	if ((ne->ne_name = zutil_strdup(hdl, path)) == NULL) {
	free(ne);
	return (-1);
	}

	ne->ne_guid = vdev_guid;
	ne->ne_order = order;
	ne->ne_num_labels = num_labels;
	ne->ne_next = pl->names;
	pl->names = ne;

	return (0);
	}

	static int
	zutil_pool_active(libpc_handle_t hdl, const char name, uint64_t guid,
	boolean_t *isactive)
	{
	ASSERT(hdl->lpc_ops->pco_pool_active != NULL);

	int error = hdl->lpc_ops->pco_pool_active(hdl->lpc_lib_handle, name,
	guid, isactive);

	return (error);
	}

	static nvlist_t *
	zutil_refresh_config(libpc_handle_t hdl, nvlist_t tryconfig)
	{
	ASSERT(hdl->lpc_ops->pco_refresh_config != NULL);

	return (hdl->lpc_ops->pco_refresh_config(hdl->lpc_lib_handle,
	tryconfig));
	}

	/*
	* Determine if the vdev id is a hole in the namespace.
	*/
	static boolean_t
	vdev_is_hole(uint64_t *hole_array, uint_t holes, uint_t id)
	{
	int c;

	for (c = 0; c < holes; c++) {

	/* Top-level is a hole */
	if (hole_array[c] == id)
	return (B_TRUE);
	}
	return (B_FALSE);
	}

	/*
	* Convert our list of pools into the definitive set of configurations. We
	* start by picking the best config for each toplevel vdev. Once that's done,
	* we assemble the toplevel vdevs into a full config for the pool. We make a
	* pass to fix up any incorrect paths, and then add it to the main list to
	* return to the user.
	*/
	static nvlist_t *
	get_configs(libpc_handle_t hdl, pool_list_t pl, boolean_t active_ok,
	nvlist_t *policy)
	{
	pool_entry_t *pe;
	vdev_entry_t *ve;
	config_entry_t *ce;
	nvlist_t ret = NULL, config = NULL, tmp = NULL, nvtop, *nvroot;
	nvlist_t spares, l2cache;
	uint_t i, nspares, nl2cache;
	boolean_t config_seen;
	uint64_t best_txg;
	char name, hostname = NULL;
	uint64_t guid;
	uint_t children = 0;
	nvlist_t **child = NULL;
	uint_t holes;
	uint64_t *hole_array, max_id;
	uint_t c;
	boolean_t isactive;
	uint64_t hostid;
	nvlist_t *nvl;
	boolean_t valid_top_config = B_FALSE;

	if (nvlist_alloc(&ret, 0, 0) != 0)
	goto nomem;

	for (pe = pl->pools; pe != NULL; pe = pe->pe_next) {
	uint64_t id, max_txg = 0;

	if (nvlist_alloc(&config, NV_UNIQUE_NAME, 0) != 0)
	goto nomem;
	config_seen = B_FALSE;

	/*
	* Iterate over all toplevel vdevs. Grab the pool configuration
	* from the first one we find, and then go through the rest and
	* add them as necessary to the 'vdevs' member of the config.
	*/
	for (ve = pe->pe_vdevs; ve != NULL; ve = ve->ve_next) {

	/*
	* Determine the best configuration for this vdev by
	* selecting the config with the latest transaction
	* group.
	*/
	best_txg = 0;
	for (ce = ve->ve_configs; ce != NULL;
	ce = ce->ce_next) {

	if (ce->ce_txg > best_txg) {
	tmp = ce->ce_config;
	best_txg = ce->ce_txg;
	}
	}

	/*
	* We rely on the fact that the max txg for the
	* pool will contain the most up-to-date information
	* about the valid top-levels in the vdev namespace.
	*/
	if (best_txg > max_txg) {
	(void) nvlist_remove(config,
	ZPOOL_CONFIG_VDEV_CHILDREN,
	DATA_TYPE_UINT64);
	(void) nvlist_remove(config,
	ZPOOL_CONFIG_HOLE_ARRAY,
	DATA_TYPE_UINT64_ARRAY);

	max_txg = best_txg;
	hole_array = NULL;
	holes = 0;
	max_id = 0;
	valid_top_config = B_FALSE;

	if (nvlist_lookup_uint64(tmp,
	ZPOOL_CONFIG_VDEV_CHILDREN, &max_id) == 0) {
	verify(nvlist_add_uint64(config,
	ZPOOL_CONFIG_VDEV_CHILDREN,
	max_id) == 0);
	valid_top_config = B_TRUE;
	}

	if (nvlist_lookup_uint64_array(tmp,
	ZPOOL_CONFIG_HOLE_ARRAY, &hole_array,
	&holes) == 0) {
	verify(nvlist_add_uint64_array(config,
	ZPOOL_CONFIG_HOLE_ARRAY,
	hole_array, holes) == 0);
	}
	}

	if (!config_seen) {
	/*
	* Copy the relevant pieces of data to the pool
	* configuration:
	*
	* version
	* pool guid
	* name
	* comment (if available)
	* compatibility features (if available)
	* pool state
	* hostid (if available)
	* hostname (if available)
	*/
	uint64_t state, version;
	char *comment = NULL;
	char *compatibility = NULL;

	version = fnvlist_lookup_uint64(tmp,
	ZPOOL_CONFIG_VERSION);
	fnvlist_add_uint64(config,
	ZPOOL_CONFIG_VERSION, version);
	guid = fnvlist_lookup_uint64(tmp,
	ZPOOL_CONFIG_POOL_GUID);
	fnvlist_add_uint64(config,
	ZPOOL_CONFIG_POOL_GUID, guid);
	name = fnvlist_lookup_string(tmp,
	ZPOOL_CONFIG_POOL_NAME);
	fnvlist_add_string(config,
	ZPOOL_CONFIG_POOL_NAME, name);

	if (nvlist_lookup_string(tmp,
	ZPOOL_CONFIG_COMMENT, &comment) == 0)
	fnvlist_add_string(config,
	ZPOOL_CONFIG_COMMENT, comment);

	if (nvlist_lookup_string(tmp,
	ZPOOL_CONFIG_COMPATIBILITY,
	&compatibility) == 0)
	fnvlist_add_string(config,
	ZPOOL_CONFIG_COMPATIBILITY,
	compatibility);

	state = fnvlist_lookup_uint64(tmp,
	ZPOOL_CONFIG_POOL_STATE);
	fnvlist_add_uint64(config,
	ZPOOL_CONFIG_POOL_STATE, state);

	hostid = 0;
	if (nvlist_lookup_uint64(tmp,
	ZPOOL_CONFIG_HOSTID, &hostid) == 0) {
	fnvlist_add_uint64(config,
	ZPOOL_CONFIG_HOSTID, hostid);
	hostname = fnvlist_lookup_string(tmp,
	ZPOOL_CONFIG_HOSTNAME);
	fnvlist_add_string(config,
	ZPOOL_CONFIG_HOSTNAME, hostname);
	}

	config_seen = B_TRUE;
	}

	/*
	* Add this top-level vdev to the child array.
	*/
	verify(nvlist_lookup_nvlist(tmp,
	ZPOOL_CONFIG_VDEV_TREE, &nvtop) == 0);
	verify(nvlist_lookup_uint64(nvtop, ZPOOL_CONFIG_ID,
	&id) == 0);

	if (id >= children) {
	nvlist_t **newchild;

	newchild = zutil_alloc(hdl, (id + 1) *
	sizeof (nvlist_t *));
	if (newchild == NULL)
	goto nomem;

	for (c = 0; c < children; c++)
	newchild[c] = child[c];

	free(child);
	child = newchild;
	children = id + 1;
	}
	if (nvlist_dup(nvtop, &child[id], 0) != 0)
	goto nomem;

	}

	/*
	* If we have information about all the top-levels then
	* clean up the nvlist which we've constructed. This
	* means removing any extraneous devices that are
	* beyond the valid range or adding devices to the end
	* of our array which appear to be missing.
	*/
	if (valid_top_config) {
	if (max_id < children) {
	for (c = max_id; c < children; c++)
	nvlist_free(child[c]);
	children = max_id;
	} else if (max_id > children) {
	nvlist_t **newchild;

	newchild = zutil_alloc(hdl, (max_id) *
	sizeof (nvlist_t *));
	if (newchild == NULL)
	goto nomem;

	for (c = 0; c < children; c++)
	newchild[c] = child[c];

	free(child);
	child = newchild;
	children = max_id;
	}
	}

	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
	&guid) == 0);

	/*
	* The vdev namespace may contain holes as a result of
	* device removal. We must add them back into the vdev
	* tree before we process any missing devices.
	*/
	if (holes > 0) {
	ASSERT(valid_top_config);

	for (c = 0; c < children; c++) {
	nvlist_t *holey;

	if (child[c] != NULL \|\|
	!vdev_is_hole(hole_array, holes, c))
	continue;

	if (nvlist_alloc(&holey, NV_UNIQUE_NAME,
	0) != 0)
	goto nomem;

	/*
	* Holes in the namespace are treated as
	* "hole" top-level vdevs and have a
	* special flag set on them.
	*/
	if (nvlist_add_string(holey,
	ZPOOL_CONFIG_TYPE,
	VDEV_TYPE_HOLE) != 0 \|\|
	nvlist_add_uint64(holey,
	ZPOOL_CONFIG_ID, c) != 0 \|\|
	nvlist_add_uint64(holey,
	ZPOOL_CONFIG_GUID, 0ULL) != 0) {
	nvlist_free(holey);
	goto nomem;
	}
	child[c] = holey;
	}
	}

	/*
	* Look for any missing top-level vdevs. If this is the case,
	* create a faked up 'missing' vdev as a placeholder. We cannot
	* simply compress the child array, because the kernel performs
	* certain checks to make sure the vdev IDs match their location
	* in the configuration.
	*/
	for (c = 0; c < children; c++) {
	if (child[c] == NULL) {
	nvlist_t *missing;
	if (nvlist_alloc(&missing, NV_UNIQUE_NAME,
	0) != 0)
	goto nomem;
	if (nvlist_add_string(missing,
	ZPOOL_CONFIG_TYPE,
	VDEV_TYPE_MISSING) != 0 \|\|
	nvlist_add_uint64(missing,
	ZPOOL_CONFIG_ID, c) != 0 \|\|
	nvlist_add_uint64(missing,
	ZPOOL_CONFIG_GUID, 0ULL) != 0) {
	nvlist_free(missing);
	goto nomem;
	}
	child[c] = missing;
	}
	}

	/*
	* Put all of this pool's top-level vdevs into a root vdev.
	*/
	if (nvlist_alloc(&nvroot, NV_UNIQUE_NAME, 0) != 0)
	goto nomem;
	if (nvlist_add_string(nvroot, ZPOOL_CONFIG_TYPE,
	VDEV_TYPE_ROOT) != 0 \|\|
	nvlist_add_uint64(nvroot, ZPOOL_CONFIG_ID, 0ULL) != 0 \|\|
	nvlist_add_uint64(nvroot, ZPOOL_CONFIG_GUID, guid) != 0 \|\|
	nvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	- child, children) != 0) {
	+ (const nvlist_t **)child, children) != 0) {
	nvlist_free(nvroot);
	goto nomem;
	}

	for (c = 0; c < children; c++)
	nvlist_free(child[c]);
	free(child);
	children = 0;
	child = NULL;

	/*
	* Go through and fix up any paths and/or devids based on our
	* known list of vdev GUID -> path mappings.
	*/
	if (fix_paths(hdl, nvroot, pl->names) != 0) {
	nvlist_free(nvroot);
	goto nomem;
	}

	/*
	* Add the root vdev to this pool's configuration.
	*/
	if (nvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	nvroot) != 0) {
	nvlist_free(nvroot);
	goto nomem;
	}
	nvlist_free(nvroot);

	/*
	* zdb uses this path to report on active pools that were
	* imported or created using -R.
	*/
	if (active_ok)
	goto add_pool;

	/*
	* Determine if this pool is currently active, in which case we
	* can't actually import it.
	*/
	verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME,
	&name) == 0);
	verify(nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID,
	&guid) == 0);

	if (zutil_pool_active(hdl, name, guid, &isactive) != 0)
	goto error;

	if (isactive) {
	nvlist_free(config);
	config = NULL;
	continue;
	}

	if (policy != NULL) {
	if (nvlist_add_nvlist(config, ZPOOL_LOAD_POLICY,
	policy) != 0)
	goto nomem;
	}

	if ((nvl = zutil_refresh_config(hdl, config)) == NULL) {
	nvlist_free(config);
	config = NULL;
	continue;
	}

	nvlist_free(config);
	config = nvl;

	/*
	* Go through and update the paths for spares, now that we have
	* them.
	*/
	verify(nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
	&nvroot) == 0);
	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	&spares, &nspares) == 0) {
	for (i = 0; i < nspares; i++) {
	if (fix_paths(hdl, spares[i], pl->names) != 0)
	goto nomem;
	}
	}

	/*
	* Update the paths for l2cache devices.
	*/
	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
	&l2cache, &nl2cache) == 0) {
	for (i = 0; i < nl2cache; i++) {
	if (fix_paths(hdl, l2cache[i], pl->names) != 0)
	goto nomem;
	}
	}

	/*
	* Restore the original information read from the actual label.
	*/
	(void) nvlist_remove(config, ZPOOL_CONFIG_HOSTID,
	DATA_TYPE_UINT64);
	(void) nvlist_remove(config, ZPOOL_CONFIG_HOSTNAME,
	DATA_TYPE_STRING);
	if (hostid != 0) {
	verify(nvlist_add_uint64(config, ZPOOL_CONFIG_HOSTID,
	hostid) == 0);
	verify(nvlist_add_string(config, ZPOOL_CONFIG_HOSTNAME,
	hostname) == 0);
	}

	add_pool:
	/*
	* Add this pool to the list of configs.
	*/
	verify(nvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME,
	&name) == 0);

	if (nvlist_add_nvlist(ret, name, config) != 0)
	goto nomem;

	nvlist_free(config);
	config = NULL;
	}

	return (ret);

	nomem:
	(void) zutil_no_memory(hdl);
	error:
	nvlist_free(config);
	nvlist_free(ret);
	for (c = 0; c < children; c++)
	nvlist_free(child[c]);
	free(child);

	return (NULL);
	}

	/*
	* Return the offset of the given label.
	*/
	static uint64_t
	label_offset(uint64_t size, int l)
	{
	ASSERT(P2PHASE_TYPED(size, sizeof (vdev_label_t), uint64_t) == 0);
	return (l * sizeof (vdev_label_t) + (l < VDEV_LABELS / 2 ?
	0 : size - VDEV_LABELS * sizeof (vdev_label_t)));
	}

	/*
	* The same description applies as to zpool_read_label below,
	* except here we do it without aio, presumably because an aio call
	* errored out in a way we think not using it could circumvent.
	*/
	static int
	zpool_read_label_slow(int fd, nvlist_t *config, int num_labels)
	{
	struct stat64 statbuf;
	int l, count = 0;
	vdev_phys_t *label;
	nvlist_t *expected_config = NULL;
	uint64_t expected_guid = 0, size;
	int error;

	*config = NULL;

	if (fstat64_blk(fd, &statbuf) == -1)
	return (0);
	size = P2ALIGN_TYPED(statbuf.st_size, sizeof (vdev_label_t), uint64_t);

	error = posix_memalign((void *)&label, PAGESIZE, sizeof (label));
	if (error)
	return (-1);

	for (l = 0; l < VDEV_LABELS; l++) {
	uint64_t state, guid, txg;
	off_t offset = label_offset(size, l) + VDEV_SKIP_SIZE;

	if (pread64(fd, label, sizeof (vdev_phys_t),
	offset) != sizeof (vdev_phys_t))
	continue;

	if (nvlist_unpack(label->vp_nvlist,
	sizeof (label->vp_nvlist), config, 0) != 0)
	continue;

	if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_GUID,
	&guid) != 0 \|\| guid == 0) {
	nvlist_free(*config);
	continue;
	}

	if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_STATE,
	&state) != 0 \|\| state > POOL_STATE_L2CACHE) {
	nvlist_free(*config);
	continue;
	}

	if (state != POOL_STATE_SPARE && state != POOL_STATE_L2CACHE &&
	(nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_TXG,
	&txg) != 0 \|\| txg == 0)) {
	nvlist_free(*config);
	continue;
	}

	if (expected_guid) {
	if (expected_guid == guid)
	count++;

	nvlist_free(*config);
	} else {
	expected_config = *config;
	expected_guid = guid;
	count++;
	}
	}

	if (num_labels != NULL)
	*num_labels = count;

	free(label);
	*config = expected_config;

	return (0);
	}

	/*
	* Given a file descriptor, read the label information and return an nvlist
	* describing the configuration, if there is one. The number of valid
	* labels found will be returned in num_labels when non-NULL.
	*/
	int
	zpool_read_label(int fd, nvlist_t *config, int num_labels)
	{
	struct stat64 statbuf;
	struct aiocb aiocbs[VDEV_LABELS];
	struct aiocb *aiocbps[VDEV_LABELS];
	vdev_phys_t *labels;
	nvlist_t *expected_config = NULL;
	uint64_t expected_guid = 0, size;
	int error, l, count = 0;

	*config = NULL;

	if (fstat64_blk(fd, &statbuf) == -1)
	return (0);
	size = P2ALIGN_TYPED(statbuf.st_size, sizeof (vdev_label_t), uint64_t);

	error = posix_memalign((void **)&labels, PAGESIZE,
	VDEV_LABELS * sizeof (*labels));
	if (error)
	return (-1);

	memset(aiocbs, 0, sizeof (aiocbs));
	for (l = 0; l < VDEV_LABELS; l++) {
	off_t offset = label_offset(size, l) + VDEV_SKIP_SIZE;

	aiocbs[l].aio_fildes = fd;
	aiocbs[l].aio_offset = offset;
	aiocbs[l].aio_buf = &labels[l];
	aiocbs[l].aio_nbytes = sizeof (vdev_phys_t);
	aiocbs[l].aio_lio_opcode = LIO_READ;
	aiocbps[l] = &aiocbs[l];
	}

	if (lio_listio(LIO_WAIT, aiocbps, VDEV_LABELS, NULL) != 0) {
	int saved_errno = errno;
	boolean_t do_slow = B_FALSE;
	error = -1;

	if (errno == EAGAIN \|\| errno == EINTR \|\| errno == EIO) {
	/*
	* A portion of the requests may have been submitted.
	* Clean them up.
	*/
	for (l = 0; l < VDEV_LABELS; l++) {
	errno = 0;
	switch (aio_error(&aiocbs[l])) {
	case EINVAL:
	break;
	case EINPROGRESS:
	// This shouldn't be possible to
	// encounter, die if we do.
	ASSERT(B_FALSE);
	fallthrough;
	case EOPNOTSUPP:
	case ENOSYS:
	do_slow = B_TRUE;
	fallthrough;
	case 0:
	default:
	(void) aio_return(&aiocbs[l]);
	}
	}
	}
	if (do_slow) {
	/*
	* At least some IO involved access unsafe-for-AIO
	* files. Let's try again, without AIO this time.
	*/
	error = zpool_read_label_slow(fd, config, num_labels);
	saved_errno = errno;
	}
	free(labels);
	errno = saved_errno;
	return (error);
	}

	for (l = 0; l < VDEV_LABELS; l++) {
	uint64_t state, guid, txg;

	if (aio_return(&aiocbs[l]) != sizeof (vdev_phys_t))
	continue;

	if (nvlist_unpack(labels[l].vp_nvlist,
	sizeof (labels[l].vp_nvlist), config, 0) != 0)
	continue;

	if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_GUID,
	&guid) != 0 \|\| guid == 0) {
	nvlist_free(*config);
	continue;
	}

	if (nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_STATE,
	&state) != 0 \|\| state > POOL_STATE_L2CACHE) {
	nvlist_free(*config);
	continue;
	}

	if (state != POOL_STATE_SPARE && state != POOL_STATE_L2CACHE &&
	(nvlist_lookup_uint64(*config, ZPOOL_CONFIG_POOL_TXG,
	&txg) != 0 \|\| txg == 0)) {
	nvlist_free(*config);
	continue;
	}

	if (expected_guid) {
	if (expected_guid == guid)
	count++;

	nvlist_free(*config);
	} else {
	expected_config = *config;
	expected_guid = guid;
	count++;
	}
	}

	if (num_labels != NULL)
	*num_labels = count;

	free(labels);
	*config = expected_config;

	return (0);
	}

	/*
	* Sorted by full path and then vdev guid to allow for multiple entries with
	* the same full path name. This is required because it's possible to
	* have multiple block devices with labels that refer to the same
	* ZPOOL_CONFIG_PATH yet have different vdev guids. In this case both
	* entries need to be added to the cache. Scenarios where this can occur
	* include overwritten pool labels, devices which are visible from multiple
	* hosts and multipath devices.
	*/
	int
	slice_cache_compare(const void arg1, const void arg2)
	{
	const char nm1 = ((rdsk_node_t )arg1)->rn_name;
	const char nm2 = ((rdsk_node_t )arg2)->rn_name;
	uint64_t guid1 = ((rdsk_node_t *)arg1)->rn_vdev_guid;
	uint64_t guid2 = ((rdsk_node_t *)arg2)->rn_vdev_guid;
	int rv;

	rv = TREE_ISIGN(strcmp(nm1, nm2));
	if (rv)
	return (rv);

	return (TREE_CMP(guid1, guid2));
	}

	static int
	label_paths_impl(libpc_handle_t hdl, nvlist_t nvroot, uint64_t pool_guid,
	uint64_t vdev_guid, char path, char devid)
	{
	nvlist_t **child;
	uint_t c, children;
	uint64_t guid;
	char *val;
	int error;

	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	&child, &children) == 0) {
	for (c = 0; c < children; c++) {
	error = label_paths_impl(hdl, child[c],
	pool_guid, vdev_guid, path, devid);
	if (error)
	return (error);
	}
	return (0);
	}

	if (nvroot == NULL)
	return (0);

	error = nvlist_lookup_uint64(nvroot, ZPOOL_CONFIG_GUID, &guid);
	if ((error != 0) \|\| (guid != vdev_guid))
	return (0);

	error = nvlist_lookup_string(nvroot, ZPOOL_CONFIG_PATH, &val);
	if (error == 0)
	*path = val;

	error = nvlist_lookup_string(nvroot, ZPOOL_CONFIG_DEVID, &val);
	if (error == 0)
	*devid = val;

	return (0);
	}

	/*
	* Given a disk label fetch the ZPOOL_CONFIG_PATH and ZPOOL_CONFIG_DEVID
	* and store these strings as config_path and devid_path respectively.
	* The returned pointers are only valid as long as label remains valid.
	*/
	int
	label_paths(libpc_handle_t hdl, nvlist_t label, char path, char devid)
	{
	nvlist_t *nvroot;
	uint64_t pool_guid;
	uint64_t vdev_guid;

	*path = NULL;
	*devid = NULL;

	if (nvlist_lookup_nvlist(label, ZPOOL_CONFIG_VDEV_TREE, &nvroot) \|\|
	nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_GUID, &pool_guid) \|\|
	nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &vdev_guid))
	return (ENOENT);

	return (label_paths_impl(hdl, nvroot, pool_guid, vdev_guid, path,
	devid));
	}

	static void
	zpool_find_import_scan_add_slice(libpc_handle_t hdl, pthread_mutex_t lock,
	avl_tree_t cache, const char path, const char *name, int order)
	{
	avl_index_t where;
	rdsk_node_t *slice;

	slice = zutil_alloc(hdl, sizeof (rdsk_node_t));
	if (asprintf(&slice->rn_name, "%s/%s", path, name) == -1) {
	free(slice);
	return;
	}
	slice->rn_vdev_guid = 0;
	slice->rn_lock = lock;
	slice->rn_avl = cache;
	slice->rn_hdl = hdl;
	slice->rn_order = order + IMPORT_ORDER_SCAN_OFFSET;
	slice->rn_labelpaths = B_FALSE;

	pthread_mutex_lock(lock);
	if (avl_find(cache, slice, &where)) {
	free(slice->rn_name);
	free(slice);
	} else {
	avl_insert(cache, slice, where);
	}
	pthread_mutex_unlock(lock);
	}

	static int
	zpool_find_import_scan_dir(libpc_handle_t hdl, pthread_mutex_t lock,
	avl_tree_t cache, const char dir, int order)
	{
	int error;
	char path[MAXPATHLEN];
	struct dirent64 *dp;
	DIR *dirp;

	if (realpath(dir, path) == NULL) {
	error = errno;
	if (error == ENOENT)
	return (0);

	zutil_error_aux(hdl, "%s", strerror(error));
	(void) zutil_error_fmt(hdl, EZFS_BADPATH, dgettext(
	TEXT_DOMAIN, "cannot resolve path '%s'"), dir);
	return (error);
	}

	dirp = opendir(path);
	if (dirp == NULL) {
	error = errno;
	zutil_error_aux(hdl, "%s", strerror(error));
	(void) zutil_error_fmt(hdl, EZFS_BADPATH,
	dgettext(TEXT_DOMAIN, "cannot open '%s'"), path);
	return (error);
	}

	while ((dp = readdir64(dirp)) != NULL) {
	const char *name = dp->d_name;
	if (strcmp(name, ".") == 0 \|\| strcmp(name, "..") == 0)
	continue;

	switch (dp->d_type) {
	case DT_UNKNOWN:
	case DT_BLK:
	case DT_LNK:
	#ifdef __FreeBSD__
	case DT_CHR:
	#endif
	case DT_REG:
	break;
	default:
	continue;
	}

	zpool_find_import_scan_add_slice(hdl, lock, cache, path, name,
	order);
	}

	(void) closedir(dirp);
	return (0);
	}

	static int
	zpool_find_import_scan_path(libpc_handle_t hdl, pthread_mutex_t lock,
	avl_tree_t cache, const char dir, int order)
	{
	int error = 0;
	char path[MAXPATHLEN];
	char *d = NULL;
	ssize_t dl;
	const char dpath, name;

	/*
	* Separate the directory and the basename.
	* We do this so that we can get the realpath of
	* the directory. We don't get the realpath on the
	* whole path because if it's a symlink, we want the
	* path of the symlink not where it points to.
	*/
	name = zfs_basename(dir);
	if ((dl = zfs_dirnamelen(dir)) == -1)
	dpath = ".";
	else
	dpath = d = zutil_strndup(hdl, dir, dl);

	if (realpath(dpath, path) == NULL) {
	error = errno;
	if (error == ENOENT) {
	error = 0;
	goto out;
	}

	zutil_error_aux(hdl, "%s", strerror(error));
	(void) zutil_error_fmt(hdl, EZFS_BADPATH, dgettext(
	TEXT_DOMAIN, "cannot resolve path '%s'"), dir);
	goto out;
	}

	zpool_find_import_scan_add_slice(hdl, lock, cache, path, name, order);

	out:
	free(d);
	return (error);
	}

	/*
	* Scan a list of directories for zfs devices.
	*/
	static int
	zpool_find_import_scan(libpc_handle_t hdl, pthread_mutex_t lock,
	avl_tree_t *slice_cache, const char const *dir, size_t dirs)
	{
	avl_tree_t *cache;
	rdsk_node_t *slice;
	void *cookie;
	int i, error;

	*slice_cache = NULL;
	cache = zutil_alloc(hdl, sizeof (avl_tree_t));
	avl_create(cache, slice_cache_compare, sizeof (rdsk_node_t),
	offsetof(rdsk_node_t, rn_node));

	for (i = 0; i < dirs; i++) {
	struct stat sbuf;

	if (stat(dir[i], &sbuf) != 0) {
	error = errno;
	if (error == ENOENT)
	continue;

	zutil_error_aux(hdl, "%s", strerror(error));
	(void) zutil_error_fmt(hdl, EZFS_BADPATH, dgettext(
	TEXT_DOMAIN, "cannot resolve path '%s'"), dir[i]);
	goto error;
	}

	/*
	* If dir[i] is a directory, we walk through it and add all
	* the entries to the cache. If it's not a directory, we just
	* add it to the cache.
	*/
	if (S_ISDIR(sbuf.st_mode)) {
	if ((error = zpool_find_import_scan_dir(hdl, lock,
	cache, dir[i], i)) != 0)
	goto error;
	} else {
	if ((error = zpool_find_import_scan_path(hdl, lock,
	cache, dir[i], i)) != 0)
	goto error;
	}
	}

	*slice_cache = cache;
	return (0);

	error:
	cookie = NULL;
	while ((slice = avl_destroy_nodes(cache, &cookie)) != NULL) {
	free(slice->rn_name);
	free(slice);
	}
	free(cache);

	return (error);
	}

	/*
	* Given a list of directories to search, find all pools stored on disk. This
	* includes partial pools which are not available to import. If no args are
	* given (argc is 0), then the default directory (/dev/dsk) is searched.
	* poolname or guid (but not both) are provided by the caller when trying
	* to import a specific pool.
	*/
	static nvlist_t *
	zpool_find_import_impl(libpc_handle_t hdl, importargs_t iarg,
	pthread_mutex_t lock, avl_tree_t cache)
	{
	nvlist_t *ret = NULL;
	pool_list_t pools = { 0 };
	pool_entry_t pe, penext;
	vdev_entry_t ve, venext;
	config_entry_t ce, cenext;
	name_entry_t ne, nenext;
	rdsk_node_t *slice;
	void *cookie;
	tpool_t *t;

	verify(iarg->poolname == NULL \|\| iarg->guid == 0);

	/*
	* Create a thread pool to parallelize the process of reading and
	* validating labels, a large number of threads can be used due to
	* minimal contention.
	*/
	t = tpool_create(1, 2 * sysconf(_SC_NPROCESSORS_ONLN), 0, NULL);
	for (slice = avl_first(cache); slice;
	(slice = avl_walk(cache, slice, AVL_AFTER)))
	(void) tpool_dispatch(t, zpool_open_func, slice);

	tpool_wait(t);
	tpool_destroy(t);

	/*
	* Process the cache, filtering out any entries which are not
	* for the specified pool then adding matching label configs.
	*/
	cookie = NULL;
	while ((slice = avl_destroy_nodes(cache, &cookie)) != NULL) {
	if (slice->rn_config != NULL) {
	nvlist_t *config = slice->rn_config;
	boolean_t matched = B_TRUE;
	boolean_t aux = B_FALSE;
	int fd;

	/*
	* Check if it's a spare or l2cache device. If it is,
	* we need to skip the name and guid check since they
	* don't exist on aux device label.
	*/
	if (iarg->poolname != NULL \|\| iarg->guid != 0) {
	uint64_t state;
	aux = nvlist_lookup_uint64(config,
	ZPOOL_CONFIG_POOL_STATE, &state) == 0 &&
	(state == POOL_STATE_SPARE \|\|
	state == POOL_STATE_L2CACHE);
	}

	if (iarg->poolname != NULL && !aux) {
	char *pname;

	matched = nvlist_lookup_string(config,
	ZPOOL_CONFIG_POOL_NAME, &pname) == 0 &&
	strcmp(iarg->poolname, pname) == 0;
	} else if (iarg->guid != 0 && !aux) {
	uint64_t this_guid;

	matched = nvlist_lookup_uint64(config,
	ZPOOL_CONFIG_POOL_GUID, &this_guid) == 0 &&
	iarg->guid == this_guid;
	}
	if (matched) {
	/*
	* Verify all remaining entries can be opened
	* exclusively. This will prune all underlying
	* multipath devices which otherwise could
	* result in the vdev appearing as UNAVAIL.
	*
	* Under zdb, this step isn't required and
	* would prevent a zdb -e of active pools with
	* no cachefile.
	*/
	fd = open(slice->rn_name,
	O_RDONLY \| O_EXCL \| O_CLOEXEC);
	if (fd >= 0 \|\| iarg->can_be_active) {
	if (fd >= 0)
	close(fd);
	add_config(hdl, &pools,
	slice->rn_name, slice->rn_order,
	slice->rn_num_labels, config);
	}
	}
	nvlist_free(config);
	}
	free(slice->rn_name);
	free(slice);
	}
	avl_destroy(cache);
	free(cache);

	ret = get_configs(hdl, &pools, iarg->can_be_active, iarg->policy);

	for (pe = pools.pools; pe != NULL; pe = penext) {
	penext = pe->pe_next;
	for (ve = pe->pe_vdevs; ve != NULL; ve = venext) {
	venext = ve->ve_next;
	for (ce = ve->ve_configs; ce != NULL; ce = cenext) {
	cenext = ce->ce_next;
	nvlist_free(ce->ce_config);
	free(ce);
	}
	free(ve);
	}
	free(pe);
	}

	for (ne = pools.names; ne != NULL; ne = nenext) {
	nenext = ne->ne_next;
	free(ne->ne_name);
	free(ne);
	}

	return (ret);
	}

	/*
	* Given a config, discover the paths for the devices which
	* exist in the config.
	*/
	static int
	discover_cached_paths(libpc_handle_t hdl, nvlist_t nv,
	avl_tree_t cache, pthread_mutex_t lock)
	{
	char *path = NULL;
	ssize_t dl;
	uint_t children;
	nvlist_t **child;

	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) == 0) {
	for (int c = 0; c < children; c++) {
	discover_cached_paths(hdl, child[c], cache, lock);
	}
	}

	/*
	* Once we have the path, we need to add the directory to
	* our directory cache.
	*/
	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &path) == 0) {
	if ((dl = zfs_dirnamelen(path)) == -1)
	path = ".";
	else
	path[dl] = '\0';
	return (zpool_find_import_scan_dir(hdl, lock, cache,
	path, 0));
	}
	return (0);
	}

	/*
	* Given a cache file, return the contents as a list of importable pools.
	* poolname or guid (but not both) are provided by the caller when trying
	* to import a specific pool.
	*/
	static nvlist_t *
	zpool_find_import_cached(libpc_handle_t hdl, importargs_t iarg)
	{
	char *buf;
	int fd;
	struct stat64 statbuf;
	nvlist_t raw, src, *dst;
	nvlist_t *pools;
	nvpair_t *elem;
	char *name;
	uint64_t this_guid;
	boolean_t active;

	verify(iarg->poolname == NULL \|\| iarg->guid == 0);

	if ((fd = open(iarg->cachefile, O_RDONLY \| O_CLOEXEC)) < 0) {
	zutil_error_aux(hdl, "%s", strerror(errno));
	(void) zutil_error(hdl, EZFS_BADCACHE,
	dgettext(TEXT_DOMAIN, "failed to open cache file"));
	return (NULL);
	}

	if (fstat64(fd, &statbuf) != 0) {
	zutil_error_aux(hdl, "%s", strerror(errno));
	(void) close(fd);
	(void) zutil_error(hdl, EZFS_BADCACHE,
	dgettext(TEXT_DOMAIN, "failed to get size of cache file"));
	return (NULL);
	}

	if ((buf = zutil_alloc(hdl, statbuf.st_size)) == NULL) {
	(void) close(fd);
	return (NULL);
	}

	if (read(fd, buf, statbuf.st_size) != statbuf.st_size) {
	(void) close(fd);
	free(buf);
	(void) zutil_error(hdl, EZFS_BADCACHE,
	dgettext(TEXT_DOMAIN,
	"failed to read cache file contents"));
	return (NULL);
	}

	(void) close(fd);

	if (nvlist_unpack(buf, statbuf.st_size, &raw, 0) != 0) {
	free(buf);
	(void) zutil_error(hdl, EZFS_BADCACHE,
	dgettext(TEXT_DOMAIN,
	"invalid or corrupt cache file contents"));
	return (NULL);
	}

	free(buf);

	/*
	* Go through and get the current state of the pools and refresh their
	* state.
	*/
	if (nvlist_alloc(&pools, 0, 0) != 0) {
	(void) zutil_no_memory(hdl);
	nvlist_free(raw);
	return (NULL);
	}

	elem = NULL;
	while ((elem = nvlist_next_nvpair(raw, elem)) != NULL) {
	src = fnvpair_value_nvlist(elem);

	name = fnvlist_lookup_string(src, ZPOOL_CONFIG_POOL_NAME);
	if (iarg->poolname != NULL && strcmp(iarg->poolname, name) != 0)
	continue;

	this_guid = fnvlist_lookup_uint64(src, ZPOOL_CONFIG_POOL_GUID);
	if (iarg->guid != 0 && iarg->guid != this_guid)
	continue;

	if (zutil_pool_active(hdl, name, this_guid, &active) != 0) {
	nvlist_free(raw);
	nvlist_free(pools);
	return (NULL);
	}

	if (active)
	continue;

	if (iarg->scan) {
	uint64_t saved_guid = iarg->guid;
	const char *saved_poolname = iarg->poolname;
	pthread_mutex_t lock;

	/*
	* Create the device cache that will hold the
	* devices we will scan based on the cachefile.
	* This will get destroyed and freed by
	* zpool_find_import_impl.
	*/
	avl_tree_t *cache = zutil_alloc(hdl,
	sizeof (avl_tree_t));
	avl_create(cache, slice_cache_compare,
	sizeof (rdsk_node_t),
	offsetof(rdsk_node_t, rn_node));
	nvlist_t *nvroot = fnvlist_lookup_nvlist(src,
	ZPOOL_CONFIG_VDEV_TREE);

	/*
	* We only want to find the pool with this_guid.
	* We will reset these values back later.
	*/
	iarg->guid = this_guid;
	iarg->poolname = NULL;

	/*
	* We need to build up a cache of devices that exists
	* in the paths pointed to by the cachefile. This allows
	* us to preserve the device namespace that was
	* originally specified by the user but also lets us
	* scan devices in those directories in case they had
	* been renamed.
	*/
	pthread_mutex_init(&lock, NULL);
	discover_cached_paths(hdl, nvroot, cache, &lock);
	nvlist_t *nv = zpool_find_import_impl(hdl, iarg,
	&lock, cache);
	pthread_mutex_destroy(&lock);

	/*
	* zpool_find_import_impl will return back
	* a list of pools that it found based on the
	* device cache. There should only be one pool
	* since we're looking for a specific guid.
	* We will use that pool to build up the final
	* pool nvlist which is returned back to the
	* caller.
	*/
	nvpair_t *pair = nvlist_next_nvpair(nv, NULL);
	fnvlist_add_nvlist(pools, nvpair_name(pair),
	fnvpair_value_nvlist(pair));

	VERIFY3P(nvlist_next_nvpair(nv, pair), ==, NULL);

	iarg->guid = saved_guid;
	iarg->poolname = saved_poolname;
	continue;
	}

	if (nvlist_add_string(src, ZPOOL_CONFIG_CACHEFILE,
	iarg->cachefile) != 0) {
	(void) zutil_no_memory(hdl);
	nvlist_free(raw);
	nvlist_free(pools);
	return (NULL);
	}

	update_vdevs_config_dev_sysfs_path(src);

	if ((dst = zutil_refresh_config(hdl, src)) == NULL) {
	nvlist_free(raw);
	nvlist_free(pools);
	return (NULL);
	}

	if (nvlist_add_nvlist(pools, nvpair_name(elem), dst) != 0) {
	(void) zutil_no_memory(hdl);
	nvlist_free(dst);
	nvlist_free(raw);
	nvlist_free(pools);
	return (NULL);
	}
	nvlist_free(dst);
	}
	nvlist_free(raw);
	return (pools);
	}

	static nvlist_t *
	zpool_find_import(libpc_handle_t hdl, importargs_t iarg)
	{
	pthread_mutex_t lock;
	avl_tree_t *cache;
	nvlist_t *pools = NULL;

	verify(iarg->poolname == NULL \|\| iarg->guid == 0);
	pthread_mutex_init(&lock, NULL);

	/*
	* Locate pool member vdevs by blkid or by directory scanning.
	* On success a newly allocated AVL tree which is populated with an
	* entry for each discovered vdev will be returned in the cache.
	* It's the caller's responsibility to consume and destroy this tree.
	*/
	if (iarg->scan \|\| iarg->paths != 0) {
	size_t dirs = iarg->paths;
	const char * const dir = (const char const *)iarg->path;

	if (dirs == 0)
	dir = zpool_default_search_paths(&dirs);

	if (zpool_find_import_scan(hdl, &lock, &cache,
	dir, dirs) != 0) {
	pthread_mutex_destroy(&lock);
	return (NULL);
	}
	} else {
	if (zpool_find_import_blkid(hdl, &lock, &cache) != 0) {
	pthread_mutex_destroy(&lock);
	return (NULL);
	}
	}

	pools = zpool_find_import_impl(hdl, iarg, &lock, cache);
	pthread_mutex_destroy(&lock);
	return (pools);
	}


	nvlist_t *
	zpool_search_import(void hdl, importargs_t import,
	const pool_config_ops_t *pco)
	{
	libpc_handle_t handle = { 0 };
	nvlist_t *pools = NULL;

	handle.lpc_lib_handle = hdl;
	handle.lpc_ops = pco;
	handle.lpc_printerr = B_TRUE;

	verify(import->poolname == NULL \|\| import->guid == 0);

	if (import->cachefile != NULL)
	pools = zpool_find_import_cached(&handle, import);
	else
	pools = zpool_find_import(&handle, import);

	if ((pools == NULL \|\| nvlist_empty(pools)) &&
	handle.lpc_open_access_error && geteuid() != 0) {
	(void) zutil_error(&handle, EZFS_EACESS, dgettext(TEXT_DOMAIN,
	"no pools found"));
	}

	return (pools);
	}

	static boolean_t
	pool_match(nvlist_t cfg, char tgt)
	{
	uint64_t v, guid = strtoull(tgt, NULL, 0);
	char *s;

	if (guid != 0) {
	if (nvlist_lookup_uint64(cfg, ZPOOL_CONFIG_POOL_GUID, &v) == 0)
	return (v == guid);
	} else {
	if (nvlist_lookup_string(cfg, ZPOOL_CONFIG_POOL_NAME, &s) == 0)
	return (strcmp(s, tgt) == 0);
	}
	return (B_FALSE);
	}

	int
	zpool_find_config(void hdl, const char target, nvlist_t **configp,
	importargs_t args, const pool_config_ops_t pco)
	{
	nvlist_t *pools;
	nvlist_t *match = NULL;
	nvlist_t *config = NULL;
	char *sepp = NULL;
	int count = 0;
	char *targetdup = strdup(target);

	*configp = NULL;

	if ((sepp = strpbrk(targetdup, "/@")) != NULL)
	*sepp = '\0';

	pools = zpool_search_import(hdl, args, pco);

	if (pools != NULL) {
	nvpair_t *elem = NULL;
	while ((elem = nvlist_next_nvpair(pools, elem)) != NULL) {
	VERIFY0(nvpair_value_nvlist(elem, &config));
	if (pool_match(config, targetdup)) {
	count++;
	if (match != NULL) {
	/* multiple matches found */
	continue;
	} else {
	match = fnvlist_dup(config);
	}
	}
	}
	fnvlist_free(pools);
	}

	if (count == 0) {
	free(targetdup);
	return (ENOENT);
	}

	if (count > 1) {
	free(targetdup);
	fnvlist_free(match);
	return (EINVAL);
	}

	*configp = match;
	free(targetdup);

	return (0);
	}

	/*
	* Internal function for iterating over the vdevs.
	*
	* For each vdev, func() will be called and will be passed 'zhp' (which is
	* typically the zpool_handle_t cast as a void pointer), the vdev's nvlist, and
	* a user-defined data pointer).
	*
	* The return values from all the func() calls will be OR'd together and
	* returned.
	*/
	int
	for_each_vdev_cb(void zhp, nvlist_t nv, pool_vdev_iter_f func,
	void *data)
	{
	nvlist_t **child;
	uint_t c, children;
	int ret = 0;
	int i;
	char *type;

	const char *list[] = {
	ZPOOL_CONFIG_SPARES,
	ZPOOL_CONFIG_L2CACHE,
	ZPOOL_CONFIG_CHILDREN
	};

	+ if (nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) != 0)
	+ return (ret);
	+
	+ /* Don't run our function on root or indirect vdevs */
	+ if ((strcmp(type, VDEV_TYPE_ROOT) != 0) &&
	+ (strcmp(type, VDEV_TYPE_INDIRECT) != 0)) {
	+ ret \|= func(zhp, nv, data);
	+ }
	+
	for (i = 0; i < ARRAY_SIZE(list); i++) {
	if (nvlist_lookup_nvlist_array(nv, list[i], &child,
	&children) == 0) {
	for (c = 0; c < children; c++) {
	uint64_t ishole = 0;

	(void) nvlist_lookup_uint64(child[c],
	ZPOOL_CONFIG_IS_HOLE, &ishole);

	if (ishole)
	continue;

	ret \|= for_each_vdev_cb(zhp, child[c],
	func, data);
	}
	}
	}

	- if (nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) != 0)
	- return (ret);
	-
	- /* Don't run our function on root vdevs */
	- if (strcmp(type, VDEV_TYPE_ROOT) != 0) {
	- ret \|= func(zhp, nv, data);
	- }
	-
	return (ret);
	}

	/*
	* Given an ZPOOL_CONFIG_VDEV_TREE nvpair, iterate over all the vdevs, calling
	* func() for each one. func() is passed the vdev's nvlist and an optional
	* user-defined 'data' pointer.
	*/
	int
	for_each_vdev_in_nvlist(nvlist_t nvroot, pool_vdev_iter_f func, void data)
	{
	return (for_each_vdev_cb(NULL, nvroot, func, data));
	}
	diff --git a/sys/contrib/openzfs/man/man4/zfs.4 b/sys/contrib/openzfs/man/man4/zfs.4
	index 9c21688705b8..ae10b2a37db8 100644
	--- a/sys/contrib/openzfs/man/man4/zfs.4
	+++ b/sys/contrib/openzfs/man/man4/zfs.4
	@@ -1,2399 +1,2399 @@
	.\"
	.\" Copyright (c) 2013 by Turbo Fredriksson <turbo@bayour.com>. All rights reserved.
	.\" Copyright (c) 2019, 2021 by Delphix. All rights reserved.
	.\" Copyright (c) 2019 Datto Inc.
	.\" The contents of this file are subject to the terms of the Common Development
	.\" and Distribution License (the "License"). You may not use this file except
	.\" in compliance with the License. You can obtain a copy of the license at
	.\" usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing.
	.\"
	.\" See the License for the specific language governing permissions and
	.\" limitations under the License. When distributing Covered Code, include this
	.\" CDDL HEADER in each file and include the License file at
	.\" usr/src/OPENSOLARIS.LICENSE. If applicable, add the following below this
	.\" CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your
	.\" own identifying information:
	.\" Portions Copyright [yyyy] [name of copyright owner]
	.\"
	.Dd June 1, 2021
	.Dt ZFS 4
	.Os
	.
	.Sh NAME
	.Nm zfs
	.Nd tuning of the ZFS kernel module
	.
	.Sh DESCRIPTION
	The ZFS module supports these parameters:
	.Bl -tag -width Ds
	.It Sy dbuf_cache_max_bytes Ns = Ns Sy ULONG_MAX Ns B Pq ulong
	Maximum size in bytes of the dbuf cache.
	The target size is determined by the MIN versus
	.No 1/2^ Ns Sy dbuf_cache_shift Pq 1/32nd
	of the target ARC size.
	The behavior of the dbuf cache and its associated settings
	can be observed via the
	.Pa /proc/spl/kstat/zfs/dbufstats
	kstat.
	.
	.It Sy dbuf_metadata_cache_max_bytes Ns = Ns Sy ULONG_MAX Ns B Pq ulong
	Maximum size in bytes of the metadata dbuf cache.
	The target size is determined by the MIN versus
	.No 1/2^ Ns Sy dbuf_metadata_cache_shift Pq 1/64th
	of the target ARC size.
	The behavior of the metadata dbuf cache and its associated settings
	can be observed via the
	.Pa /proc/spl/kstat/zfs/dbufstats
	kstat.
	.
	.It Sy dbuf_cache_hiwater_pct Ns = Ns Sy 10 Ns % Pq uint
	The percentage over
	.Sy dbuf_cache_max_bytes
	when dbufs must be evicted directly.
	.
	.It Sy dbuf_cache_lowater_pct Ns = Ns Sy 10 Ns % Pq uint
	The percentage below
	.Sy dbuf_cache_max_bytes
	when the evict thread stops evicting dbufs.
	.
	.It Sy dbuf_cache_shift Ns = Ns Sy 5 Pq int
	Set the size of the dbuf cache
	.Pq Sy dbuf_cache_max_bytes
	to a log2 fraction of the target ARC size.
	.
	.It Sy dbuf_metadata_cache_shift Ns = Ns Sy 6 Pq int
	Set the size of the dbuf metadata cache
	.Pq Sy dbuf_metadata_cache_max_bytes
	to a log2 fraction of the target ARC size.
	.
	.It Sy dmu_object_alloc_chunk_shift Ns = Ns Sy 7 Po 128 Pc Pq int
	dnode slots allocated in a single operation as a power of 2.
	The default value minimizes lock contention for the bulk operation performed.
	.
	.It Sy dmu_prefetch_max Ns = Ns Sy 134217728 Ns B Po 128MB Pc Pq int
	Limit the amount we can prefetch with one call to this amount in bytes.
	This helps to limit the amount of memory that can be used by prefetching.
	.
	.It Sy ignore_hole_birth Pq int
	Alias for
	.Sy send_holes_without_birth_time .
	.
	.It Sy l2arc_feed_again Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Turbo L2ARC warm-up.
	When the L2ARC is cold the fill interval will be set as fast as possible.
	.
	.It Sy l2arc_feed_min_ms Ns = Ns Sy 200 Pq ulong
	Min feed interval in milliseconds.
	Requires
	.Sy l2arc_feed_again Ns = Ns Ar 1
	and only applicable in related situations.
	.
	.It Sy l2arc_feed_secs Ns = Ns Sy 1 Pq ulong
	Seconds between L2ARC writing.
	.
	.It Sy l2arc_headroom Ns = Ns Sy 2 Pq ulong
	How far through the ARC lists to search for L2ARC cacheable content,
	expressed as a multiplier of
	.Sy l2arc_write_max .
	ARC persistence across reboots can be achieved with persistent L2ARC
	by setting this parameter to
	.Sy 0 ,
	allowing the full length of ARC lists to be searched for cacheable content.
	.
	.It Sy l2arc_headroom_boost Ns = Ns Sy 200 Ns % Pq ulong
	Scales
	.Sy l2arc_headroom
	by this percentage when L2ARC contents are being successfully compressed
	before writing.
	A value of
	.Sy 100
	disables this feature.
	.
	.It Sy l2arc_exclude_special Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Controls whether buffers present on special vdevs are eligibile for caching
	into L2ARC.
	If set to 1, exclude dbufs on special vdevs from being cached to L2ARC.
	.
	.It Sy l2arc_mfuonly Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Controls whether only MFU metadata and data are cached from ARC into L2ARC.
	This may be desired to avoid wasting space on L2ARC when reading/writing large
	amounts of data that are not expected to be accessed more than once.
	.Pp
	The default is off,
	meaning both MRU and MFU data and metadata are cached.
	When turning off this feature, some MRU buffers will still be present
	in ARC and eventually cached on L2ARC.
	.No If Sy l2arc_noprefetch Ns = Ns Sy 0 ,
	some prefetched buffers will be cached to L2ARC, and those might later
	transition to MRU, in which case the
	.Sy l2arc_mru_asize No arcstat will not be Sy 0 .
	.Pp
	Regardless of
	.Sy l2arc_noprefetch ,
	some MFU buffers might be evicted from ARC,
	accessed later on as prefetches and transition to MRU as prefetches.
	If accessed again they are counted as MRU and the
	.Sy l2arc_mru_asize No arcstat will not be Sy 0 .
	.Pp
	The ARC status of L2ARC buffers when they were first cached in
	L2ARC can be seen in the
	.Sy l2arc_mru_asize , Sy l2arc_mfu_asize , No and Sy l2arc_prefetch_asize
	arcstats when importing the pool or onlining a cache
	device if persistent L2ARC is enabled.
	.Pp
	The
	.Sy evict_l2_eligible_mru
	arcstat does not take into account if this option is enabled as the information
	provided by the
	.Sy evict_l2_eligible_m[rf]u
	arcstats can be used to decide if toggling this option is appropriate
	for the current workload.
	.
	.It Sy l2arc_meta_percent Ns = Ns Sy 33 Ns % Pq int
	Percent of ARC size allowed for L2ARC-only headers.
	Since L2ARC buffers are not evicted on memory pressure,
	too many headers on a system with an irrationally large L2ARC
	can render it slow or unusable.
	This parameter limits L2ARC writes and rebuilds to achieve the target.
	.
	.It Sy l2arc_trim_ahead Ns = Ns Sy 0 Ns % Pq ulong
	Trims ahead of the current write size
	.Pq Sy l2arc_write_max
	on L2ARC devices by this percentage of write size if we have filled the device.
	If set to
	.Sy 100
	we TRIM twice the space required to accommodate upcoming writes.
	A minimum of
	.Sy 64MB
	will be trimmed.
	It also enables TRIM of the whole L2ARC device upon creation
	or addition to an existing pool or if the header of the device is
	invalid upon importing a pool or onlining a cache device.
	A value of
	.Sy 0
	disables TRIM on L2ARC altogether and is the default as it can put significant
	stress on the underlying storage devices.
	This will vary depending of how well the specific device handles these commands.
	.
	.It Sy l2arc_noprefetch Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Do not write buffers to L2ARC if they were prefetched but not used by
	applications.
	In case there are prefetched buffers in L2ARC and this option
	is later set, we do not read the prefetched buffers from L2ARC.
	Unsetting this option is useful for caching sequential reads from the
	disks to L2ARC and serve those reads from L2ARC later on.
	This may be beneficial in case the L2ARC device is significantly faster
	in sequential reads than the disks of the pool.
	.Pp
	Use
	.Sy 1
	to disable and
	.Sy 0
	to enable caching/reading prefetches to/from L2ARC.
	.
	.It Sy l2arc_norw Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	No reads during writes.
	.
	.It Sy l2arc_write_boost Ns = Ns Sy 8388608 Ns B Po 8MB Pc Pq ulong
	Cold L2ARC devices will have
	.Sy l2arc_write_max
	increased by this amount while they remain cold.
	.
	.It Sy l2arc_write_max Ns = Ns Sy 8388608 Ns B Po 8MB Pc Pq ulong
	Max write bytes per interval.
	.
	.It Sy l2arc_rebuild_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Rebuild the L2ARC when importing a pool (persistent L2ARC).
	This can be disabled if there are problems importing a pool
	or attaching an L2ARC device (e.g. the L2ARC device is slow
	in reading stored log metadata, or the metadata
	has become somehow fragmented/unusable).
	.
	.It Sy l2arc_rebuild_blocks_min_l2size Ns = Ns Sy 1073741824 Ns B Po 1GB Pc Pq ulong
	Mininum size of an L2ARC device required in order to write log blocks in it.
	The log blocks are used upon importing the pool to rebuild the persistent L2ARC.
	.Pp
	For L2ARC devices less than 1GB, the amount of data
	.Fn l2arc_evict
	evicts is significant compared to the amount of restored L2ARC data.
	In this case, do not write log blocks in L2ARC in order not to waste space.
	.
	.It Sy metaslab_aliquot Ns = Ns Sy 524288 Ns B Po 512kB Pc Pq ulong
	Metaslab granularity, in bytes.
	This is roughly similar to what would be referred to as the "stripe size"
	in traditional RAID arrays.
	In normal operation, ZFS will try to write this amount of data
	to a top-level vdev before moving on to the next one.
	.
	.It Sy metaslab_bias_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Enable metaslab group biasing based on their vdevs' over- or under-utilization
	relative to the pool.
	.
	.It Sy metaslab_force_ganging Ns = Ns Sy 16777217 Ns B Ns B Po 16MB + 1B Pc Pq ulong
	Make some blocks above a certain size be gang blocks.
	This option is used by the test suite to facilitate testing.
	.
	.It Sy zfs_history_output_max Ns = Ns Sy 1048576 Ns B Ns B Po 1MB Pc Pq int
	When attempting to log an output nvlist of an ioctl in the on-disk history,
	the output will not be stored if it is larger than this size (in bytes).
	This must be less than
	.Sy DMU_MAX_ACCESS Pq 64MB .
	This applies primarily to
	.Fn zfs_ioc_channel_program Pq cf. Xr zfs-program 8 .
	.
	.It Sy zfs_keep_log_spacemaps_at_export Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Prevent log spacemaps from being destroyed during pool exports and destroys.
	.
	.It Sy zfs_metaslab_segment_weight_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Enable/disable segment-based metaslab selection.
	.
	.It Sy zfs_metaslab_switch_threshold Ns = Ns Sy 2 Pq int
	When using segment-based metaslab selection, continue allocating
	from the active metaslab until this option's
	worth of buckets have been exhausted.
	.
	.It Sy metaslab_debug_load Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Load all metaslabs during pool import.
	.
	.It Sy metaslab_debug_unload Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Prevent metaslabs from being unloaded.
	.
	.It Sy metaslab_fragmentation_factor_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Enable use of the fragmentation metric in computing metaslab weights.
	.
	.It Sy metaslab_df_max_search Ns = Ns Sy 16777216 Ns B Po 16MB Pc Pq int
	Maximum distance to search forward from the last offset.
	Without this limit, fragmented pools can see
	.Em >100`000
	iterations and
	.Fn metaslab_block_picker
	becomes the performance limiting factor on high-performance storage.
	.Pp
	With the default setting of
	.Sy 16MB ,
	we typically see less than
	.Em 500
	iterations, even with very fragmented
	.Sy ashift Ns = Ns Sy 9
	pools.
	The maximum number of iterations possible is
	.Sy metaslab_df_max_search / 2^(ashift+1) .
	With the default setting of
	.Sy 16MB
	this is
	.Em 16*1024 Pq with Sy ashift Ns = Ns Sy 9
	or
	.Em 2*1024 Pq with Sy ashift Ns = Ns Sy 12 .
	.
	.It Sy metaslab_df_use_largest_segment Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	If not searching forward (due to
	.Sy metaslab_df_max_search , metaslab_df_free_pct ,
	.No or Sy metaslab_df_alloc_threshold ) ,
	this tunable controls which segment is used.
	If set, we will use the largest free segment.
	If unset, we will use a segment of at least the requested size.
	.
	.It Sy zfs_metaslab_max_size_cache_sec Ns = Ns Sy 3600 Ns s Po 1h Pc Pq ulong
	When we unload a metaslab, we cache the size of the largest free chunk.
	We use that cached size to determine whether or not to load a metaslab
	for a given allocation.
	As more frees accumulate in that metaslab while it's unloaded,
	the cached max size becomes less and less accurate.
	After a number of seconds controlled by this tunable,
	we stop considering the cached max size and start
	considering only the histogram instead.
	.
	.It Sy zfs_metaslab_mem_limit Ns = Ns Sy 25 Ns % Pq int
	When we are loading a new metaslab, we check the amount of memory being used
	to store metaslab range trees.
	If it is over a threshold, we attempt to unload the least recently used metaslab
	to prevent the system from clogging all of its memory with range trees.
	This tunable sets the percentage of total system memory that is the threshold.
	.
	.It Sy zfs_metaslab_try_hard_before_gang Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	.Bl -item -compact
	.It
	If unset, we will first try normal allocation.
	.It
	If that fails then we will do a gang allocation.
	.It
	If that fails then we will do a "try hard" gang allocation.
	.It
	If that fails then we will have a multi-layer gang block.
	.El
	.Pp
	.Bl -item -compact
	.It
	If set, we will first try normal allocation.
	.It
	If that fails then we will do a "try hard" allocation.
	.It
	If that fails we will do a gang allocation.
	.It
	If that fails we will do a "try hard" gang allocation.
	.It
	If that fails then we will have a multi-layer gang block.
	.El
	.
	.It Sy zfs_metaslab_find_max_tries Ns = Ns Sy 100 Pq int
	When not trying hard, we only consider this number of the best metaslabs.
	This improves performance, especially when there are many metaslabs per vdev
	and the allocation can't actually be satisfied
	(so we would otherwise iterate all metaslabs).
	.
	.It Sy zfs_vdev_default_ms_count Ns = Ns Sy 200 Pq int
	When a vdev is added, target this number of metaslabs per top-level vdev.
	.
	.It Sy zfs_vdev_default_ms_shift Ns = Ns Sy 29 Po 512MB Pc Pq int
	Default limit for metaslab size.
	.
	.It Sy zfs_vdev_max_auto_ashift Ns = Ns Sy ASHIFT_MAX Po 16 Pc Pq ulong
	Maximum ashift used when optimizing for logical -> physical sector size on new
	top-level vdevs.
	.
	.It Sy zfs_vdev_min_auto_ashift Ns = Ns Sy ASHIFT_MIN Po 9 Pc Pq ulong
	Minimum ashift used when creating new top-level vdevs.
	.
	.It Sy zfs_vdev_min_ms_count Ns = Ns Sy 16 Pq int
	Minimum number of metaslabs to create in a top-level vdev.
	.
	.It Sy vdev_validate_skip Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Skip label validation steps during pool import.
	Changing is not recommended unless you know what you're doing
	and are recovering a damaged label.
	.
	.It Sy zfs_vdev_ms_count_limit Ns = Ns Sy 131072 Po 128k Pc Pq int
	Practical upper limit of total metaslabs per top-level vdev.
	.
	.It Sy metaslab_preload_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Enable metaslab group preloading.
	.
	.It Sy metaslab_lba_weighting_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Give more weight to metaslabs with lower LBAs,
	assuming they have greater bandwidth,
	as is typically the case on a modern constant angular velocity disk drive.
	.
	.It Sy metaslab_unload_delay Ns = Ns Sy 32 Pq int
	After a metaslab is used, we keep it loaded for this many TXGs, to attempt to
	reduce unnecessary reloading.
	Note that both this many TXGs and
	.Sy metaslab_unload_delay_ms
	milliseconds must pass before unloading will occur.
	.
	.It Sy metaslab_unload_delay_ms Ns = Ns Sy 600000 Ns ms Po 10min Pc Pq int
	After a metaslab is used, we keep it loaded for this many milliseconds,
	to attempt to reduce unnecessary reloading.
	Note, that both this many milliseconds and
	.Sy metaslab_unload_delay
	TXGs must pass before unloading will occur.
	.
	.It Sy reference_history Ns = Ns Sy 3 Pq int
	Maximum reference holders being tracked when reference_tracking_enable is active.
	.
	.It Sy reference_tracking_enable Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Track reference holders to
	.Sy refcount_t
	objects (debug builds only).
	.
	.It Sy send_holes_without_birth_time Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	When set, the
	.Sy hole_birth
	optimization will not be used, and all holes will always be sent during a
	.Nm zfs Cm send .
	This is useful if you suspect your datasets are affected by a bug in
	.Sy hole_birth .
	.
	.It Sy spa_config_path Ns = Ns Pa /etc/zfs/zpool.cache Pq charp
	SPA config file.
	.
	.It Sy spa_asize_inflation Ns = Ns Sy 24 Pq int
	Multiplication factor used to estimate actual disk consumption from the
	size of data being written.
	The default value is a worst case estimate,
	but lower values may be valid for a given pool depending on its configuration.
	Pool administrators who understand the factors involved
	may wish to specify a more realistic inflation factor,
	particularly if they operate close to quota or capacity limits.
	.
	.It Sy spa_load_print_vdev_tree Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Whether to print the vdev tree in the debugging message buffer during pool import.
	.
	.It Sy spa_load_verify_data Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Whether to traverse data blocks during an "extreme rewind"
	.Pq Fl X
	import.
	.Pp
	An extreme rewind import normally performs a full traversal of all
	blocks in the pool for verification.
	If this parameter is unset, the traversal skips non-metadata blocks.
	It can be toggled once the
	import has started to stop or start the traversal of non-metadata blocks.
	.
	.It Sy spa_load_verify_metadata Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Whether to traverse blocks during an "extreme rewind"
	.Pq Fl X
	pool import.
	.Pp
	An extreme rewind import normally performs a full traversal of all
	blocks in the pool for verification.
	If this parameter is unset, the traversal is not performed.
	It can be toggled once the import has started to stop or start the traversal.
	.
	.It Sy spa_load_verify_shift Ns = Ns Sy 4 Po 1/16th Pc Pq int
	Sets the maximum number of bytes to consume during pool import to the log2
	fraction of the target ARC size.
	.
	.It Sy spa_slop_shift Ns = Ns Sy 5 Po 1/32nd Pc Pq int
	Normally, we don't allow the last
	.Sy 3.2% Pq Sy 1/2^spa_slop_shift
	of space in the pool to be consumed.
	This ensures that we don't run the pool completely out of space,
	due to unaccounted changes (e.g. to the MOS).
	It also limits the worst-case time to allocate space.
	If we have less than this amount of free space,
	most ZPL operations (e.g. write, create) will return
	.Sy ENOSPC .
	.
	.It Sy vdev_removal_max_span Ns = Ns Sy 32768 Ns B Po 32kB Pc Pq int
	During top-level vdev removal, chunks of data are copied from the vdev
	which may include free space in order to trade bandwidth for IOPS.
	This parameter determines the maximum span of free space, in bytes,
	which will be included as "unnecessary" data in a chunk of copied data.
	.Pp
	The default value here was chosen to align with
	.Sy zfs_vdev_read_gap_limit ,
	which is a similar concept when doing
	regular reads (but there's no reason it has to be the same).
	.
	.It Sy vdev_file_logical_ashift Ns = Ns Sy 9 Po 512B Pc Pq ulong
	Logical ashift for file-based devices.
	.
	.It Sy vdev_file_physical_ashift Ns = Ns Sy 9 Po 512B Pc Pq ulong
	Physical ashift for file-based devices.
	.
	.It Sy zap_iterate_prefetch Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	If set, when we start iterating over a ZAP object,
	prefetch the entire object (all leaf blocks).
	However, this is limited by
	.Sy dmu_prefetch_max .
	.
	.It Sy zfetch_array_rd_sz Ns = Ns Sy 1048576 Ns B Po 1MB Pc Pq ulong
	If prefetching is enabled, disable prefetching for reads larger than this size.
	.
	.It Sy zfetch_max_distance Ns = Ns Sy 8388608 Ns B Po 8MB Pc Pq uint
	Max bytes to prefetch per stream.
	.
	.It Sy zfetch_max_idistance Ns = Ns Sy 67108864 Ns B Po 64MB Pc Pq uint
	Max bytes to prefetch indirects for per stream.
	.
	.It Sy zfetch_max_streams Ns = Ns Sy 8 Pq uint
	Max number of streams per zfetch (prefetch streams per file).
	.
	.It Sy zfetch_min_sec_reap Ns = Ns Sy 2 Pq uint
	Min time before an active prefetch stream can be reclaimed
	.
	.It Sy zfs_abd_scatter_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Enables ARC from using scatter/gather lists and forces all allocations to be
	linear in kernel memory.
	Disabling can improve performance in some code paths
	at the expense of fragmented kernel memory.
	.
	.It Sy zfs_abd_scatter_max_order Ns = Ns Sy MAX_ORDER-1 Pq uint
	Maximum number of consecutive memory pages allocated in a single block for
	scatter/gather lists.
	.Pp
	The value of
	.Sy MAX_ORDER
	depends on kernel configuration.
	.
	.It Sy zfs_abd_scatter_min_size Ns = Ns Sy 1536 Ns B Po 1.5kB Pc Pq uint
	This is the minimum allocation size that will use scatter (page-based) ABDs.
	Smaller allocations will use linear ABDs.
	.
	.It Sy zfs_arc_dnode_limit Ns = Ns Sy 0 Ns B Pq ulong
	When the number of bytes consumed by dnodes in the ARC exceeds this number of
	bytes, try to unpin some of it in response to demand for non-metadata.
	This value acts as a ceiling to the amount of dnode metadata, and defaults to
	.Sy 0 ,
	which indicates that a percent which is based on
	.Sy zfs_arc_dnode_limit_percent
	of the ARC meta buffers that may be used for dnodes.
	.Pp
	Also see
	.Sy zfs_arc_meta_prune
	which serves a similar purpose but is used
	when the amount of metadata in the ARC exceeds
	.Sy zfs_arc_meta_limit
	rather than in response to overall demand for non-metadata.
	.
	.It Sy zfs_arc_dnode_limit_percent Ns = Ns Sy 10 Ns % Pq ulong
	Percentage that can be consumed by dnodes of ARC meta buffers.
	.Pp
	See also
	.Sy zfs_arc_dnode_limit ,
	which serves a similar purpose but has a higher priority if nonzero.
	.
	.It Sy zfs_arc_dnode_reduce_percent Ns = Ns Sy 10 Ns % Pq ulong
	Percentage of ARC dnodes to try to scan in response to demand for non-metadata
	when the number of bytes consumed by dnodes exceeds
	.Sy zfs_arc_dnode_limit .
	.
	.It Sy zfs_arc_average_blocksize Ns = Ns Sy 8192 Ns B Po 8kB Pc Pq int
	The ARC's buffer hash table is sized based on the assumption of an average
	block size of this value.
	This works out to roughly 1MB of hash table per 1GB of physical memory
	with 8-byte pointers.
	For configurations with a known larger average block size,
	this value can be increased to reduce the memory footprint.
	.
	.It Sy zfs_arc_eviction_pct Ns = Ns Sy 200 Ns % Pq int
	When
	.Fn arc_is_overflowing ,
	.Fn arc_get_data_impl
	waits for this percent of the requested amount of data to be evicted.
	For example, by default, for every
	.Em 2kB
	that's evicted,
	.Em 1kB
	of it may be "reused" by a new allocation.
	Since this is above
	.Sy 100 Ns % ,
	it ensures that progress is made towards getting
	.Sy arc_size No under Sy arc_c .
	Since this is finite, it ensures that allocations can still happen,
	even during the potentially long time that
	.Sy arc_size No is more than Sy arc_c .
	.
	.It Sy zfs_arc_evict_batch_limit Ns = Ns Sy 10 Pq int
	Number ARC headers to evict per sub-list before proceeding to another sub-list.
	This batch-style operation prevents entire sub-lists from being evicted at once
	but comes at a cost of additional unlocking and locking.
	.
	.It Sy zfs_arc_grow_retry Ns = Ns Sy 0 Ns s Pq int
	If set to a non zero value, it will replace the
	.Sy arc_grow_retry
	value with this value.
	The
	.Sy arc_grow_retry
	.No value Pq default Sy 5 Ns s
	is the number of seconds the ARC will wait before
	trying to resume growth after a memory pressure event.
	.
	.It Sy zfs_arc_lotsfree_percent Ns = Ns Sy 10 Ns % Pq int
	Throttle I/O when free system memory drops below this percentage of total
	system memory.
	Setting this value to
	.Sy 0
	will disable the throttle.
	.
	.It Sy zfs_arc_max Ns = Ns Sy 0 Ns B Pq ulong
	Max size of ARC in bytes.
	If
	.Sy 0 ,
	then the max size of ARC is determined by the amount of system memory installed.
	Under Linux, half of system memory will be used as the limit.
	Under
	.Fx ,
	the larger of
	.Sy all_system_memory - 1GB No and Sy 5/8 * all_system_memory
	will be used as the limit.
	This value must be at least
	.Sy 67108864 Ns B Pq 64MB .
	.Pp
	This value can be changed dynamically, with some caveats.
	It cannot be set back to
	.Sy 0
	while running, and reducing it below the current ARC size will not cause
	the ARC to shrink without memory pressure to induce shrinking.
	.
	.It Sy zfs_arc_meta_adjust_restarts Ns = Ns Sy 4096 Pq ulong
	The number of restart passes to make while scanning the ARC attempting
	the free buffers in order to stay below the
	.Sy fs_arc_meta_limit .
	This value should not need to be tuned but is available to facilitate
	performance analysis.
	.
	.It Sy zfs_arc_meta_limit Ns = Ns Sy 0 Ns B Pq ulong
	The maximum allowed size in bytes that metadata buffers are allowed to
	consume in the ARC.
	When this limit is reached, metadata buffers will be reclaimed,
	even if the overall
	.Sy arc_c_max
	has not been reached.
	It defaults to
	.Sy 0 ,
	which indicates that a percentage based on
	.Sy zfs_arc_meta_limit_percent
	of the ARC may be used for metadata.
	.Pp
	This value my be changed dynamically, except that must be set to an explicit value
	.Pq cannot be set back to Sy 0 .
	.
	.It Sy zfs_arc_meta_limit_percent Ns = Ns Sy 75 Ns % Pq ulong
	Percentage of ARC buffers that can be used for metadata.
	.Pp
	See also
	.Sy zfs_arc_meta_limit ,
	which serves a similar purpose but has a higher priority if nonzero.
	.
	.It Sy zfs_arc_meta_min Ns = Ns Sy 0 Ns B Pq ulong
	The minimum allowed size in bytes that metadata buffers may consume in
	the ARC.
	.
	.It Sy zfs_arc_meta_prune Ns = Ns Sy 10000 Pq int
	The number of dentries and inodes to be scanned looking for entries
	which can be dropped.
	This may be required when the ARC reaches the
	.Sy zfs_arc_meta_limit
	because dentries and inodes can pin buffers in the ARC.
	Increasing this value will cause to dentry and inode caches
	to be pruned more aggressively.
	Setting this value to
	.Sy 0
	will disable pruning the inode and dentry caches.
	.
	.It Sy zfs_arc_meta_strategy Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Define the strategy for ARC metadata buffer eviction (meta reclaim strategy):
	.Bl -tag -compact -offset 4n -width "0 (META_ONLY)"
	.It Sy 0 Pq META_ONLY
	evict only the ARC metadata buffers
	.It Sy 1 Pq BALANCED
	additional data buffers may be evicted if required
	to evict the required number of metadata buffers.
	.El
	.
	.It Sy zfs_arc_min Ns = Ns Sy 0 Ns B Pq ulong
	Min size of ARC in bytes.
	.No If set to Sy 0 , arc_c_min
	will default to consuming the larger of
	.Sy 32MB No or Sy all_system_memory/32 .
	.
	.It Sy zfs_arc_min_prefetch_ms Ns = Ns Sy 0 Ns ms Ns Po Ns ≡ Ns 1s Pc Pq int
	Minimum time prefetched blocks are locked in the ARC.
	.
	.It Sy zfs_arc_min_prescient_prefetch_ms Ns = Ns Sy 0 Ns ms Ns Po Ns ≡ Ns 6s Pc Pq int
	Minimum time "prescient prefetched" blocks are locked in the ARC.
	These blocks are meant to be prefetched fairly aggressively ahead of
	the code that may use them.
	.
	.It Sy zfs_max_missing_tvds Ns = Ns Sy 0 Pq int
	Number of missing top-level vdevs which will be allowed during
	pool import (only in read-only mode).
	.
	.It Sy zfs_max_nvlist_src_size Ns = Sy 0 Pq ulong
	Maximum size in bytes allowed to be passed as
	.Sy zc_nvlist_src_size
	for ioctls on
	.Pa /dev/zfs .
	This prevents a user from causing the kernel to allocate
	an excessive amount of memory.
	When the limit is exceeded, the ioctl fails with
	.Sy EINVAL
	and a description of the error is sent to the
	.Pa zfs-dbgmsg
	log.
	This parameter should not need to be touched under normal circumstances.
	If
	.Sy 0 ,
	equivalent to a quarter of the user-wired memory limit under
	.Fx
	and to
	.Sy 134217728 Ns B Pq 128MB
	under Linux.
	.
	.It Sy zfs_multilist_num_sublists Ns = Ns Sy 0 Pq int
	To allow more fine-grained locking, each ARC state contains a series
	of lists for both data and metadata objects.
	Locking is performed at the level of these "sub-lists".
	This parameters controls the number of sub-lists per ARC state,
	and also applies to other uses of the multilist data structure.
	.Pp
	If
	.Sy 0 ,
	equivalent to the greater of the number of online CPUs and
	.Sy 4 .
	.
	.It Sy zfs_arc_overflow_shift Ns = Ns Sy 8 Pq int
	The ARC size is considered to be overflowing if it exceeds the current
	ARC target size
	.Pq Sy arc_c
	by thresholds determined by this parameter.
	Exceeding by
	.Sy ( arc_c >> zfs_arc_overflow_shift ) * 0.5
	starts ARC reclamation process.
	If that appears insufficient, exceeding by
	.Sy ( arc_c >> zfs_arc_overflow_shift ) * 1.5
	blocks new buffer allocation until the reclaim thread catches up.
	Started reclamation process continues till ARC size returns below the
	target size.
	.Pp
	The default value of
	.Sy 8
	causes the ARC to start reclamation if it exceeds the target size by
	.Em 0.2%
	of the target size, and block allocations by
	.Em 0.6% .
	.
	.It Sy zfs_arc_p_min_shift Ns = Ns Sy 0 Pq int
	If nonzero, this will update
	.Sy arc_p_min_shift Pq default Sy 4
	with the new value.
	.Sy arc_p_min_shift No is used as a shift of Sy arc_c
	when calculating the minumum
	.Sy arc_p No size.
	.
	.It Sy zfs_arc_p_dampener_disable Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Disable
	.Sy arc_p
	adapt dampener, which reduces the maximum single adjustment to
	.Sy arc_p .
	.
	.It Sy zfs_arc_shrink_shift Ns = Ns Sy 0 Pq int
	If nonzero, this will update
	.Sy arc_shrink_shift Pq default Sy 7
	with the new value.
	.
	.It Sy zfs_arc_pc_percent Ns = Ns Sy 0 Ns % Po off Pc Pq uint
	Percent of pagecache to reclaim ARC to.
	.Pp
	This tunable allows the ZFS ARC to play more nicely
	with the kernel's LRU pagecache.
	It can guarantee that the ARC size won't collapse under scanning
	pressure on the pagecache, yet still allows the ARC to be reclaimed down to
	.Sy zfs_arc_min
	if necessary.
	This value is specified as percent of pagecache size (as measured by
	.Sy NR_FILE_PAGES ) ,
	where that percent may exceed
	.Sy 100 .
	This
	only operates during memory pressure/reclaim.
	.
	.It Sy zfs_arc_shrinker_limit Ns = Ns Sy 10000 Pq int
	This is a limit on how many pages the ARC shrinker makes available for
	eviction in response to one page allocation attempt.
	Note that in practice, the kernel's shrinker can ask us to evict
	up to about four times this for one allocation attempt.
	.Pp
	The default limit of
	.Sy 10000 Pq in practice, Em 160MB No per allocation attempt with 4kB pages
	limits the amount of time spent attempting to reclaim ARC memory to
	less than 100ms per allocation attempt,
	even with a small average compressed block size of ~8kB.
	.Pp
	The parameter can be set to 0 (zero) to disable the limit,
	and only applies on Linux.
	.
	.It Sy zfs_arc_sys_free Ns = Ns Sy 0 Ns B Pq ulong
	The target number of bytes the ARC should leave as free memory on the system.
	If zero, equivalent to the bigger of
	.Sy 512kB No and Sy all_system_memory/64 .
	.
	.It Sy zfs_autoimport_disable Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Disable pool import at module load by ignoring the cache file
	.Pq Sy spa_config_path .
	.
	.It Sy zfs_checksum_events_per_second Ns = Ns Sy 20 Ns /s Pq uint
	Rate limit checksum events to this many per second.
	Note that this should not be set below the ZED thresholds
	(currently 10 checksums over 10 seconds)
	or else the daemon may not trigger any action.
	.
	.It Sy zfs_commit_timeout_pct Ns = Ns Sy 5 Ns % Pq int
	This controls the amount of time that a ZIL block (lwb) will remain "open"
	when it isn't "full", and it has a thread waiting for it to be committed to
	stable storage.
	The timeout is scaled based on a percentage of the last lwb
	latency to avoid significantly impacting the latency of each individual
	transaction record (itx).
	.
	.It Sy zfs_condense_indirect_commit_entry_delay_ms Ns = Ns Sy 0 Ns ms Pq int
	Vdev indirection layer (used for device removal) sleeps for this many
	milliseconds during mapping generation.
	Intended for use with the test suite to throttle vdev removal speed.
	.
	.It Sy zfs_condense_indirect_obsolete_pct Ns = Ns Sy 25 Ns % Pq int
	Minimum percent of obsolete bytes in vdev mapping required to attempt to condense
	.Pq see Sy zfs_condense_indirect_vdevs_enable .
	Intended for use with the test suite
	to facilitate triggering condensing as needed.
	.
	.It Sy zfs_condense_indirect_vdevs_enable Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Enable condensing indirect vdev mappings.
	When set, attempt to condense indirect vdev mappings
	if the mapping uses more than
	.Sy zfs_condense_min_mapping_bytes
	bytes of memory and if the obsolete space map object uses more than
	.Sy zfs_condense_max_obsolete_bytes
	bytes on-disk.
	The condensing process is an attempt to save memory by removing obsolete mappings.
	.
	.It Sy zfs_condense_max_obsolete_bytes Ns = Ns Sy 1073741824 Ns B Po 1GB Pc Pq ulong
	Only attempt to condense indirect vdev mappings if the on-disk size
	of the obsolete space map object is greater than this number of bytes
	.Pq see Sy zfs_condense_indirect_vdevs_enable .
	.
	.It Sy zfs_condense_min_mapping_bytes Ns = Ns Sy 131072 Ns B Po 128kB Pc Pq ulong
	Minimum size vdev mapping to attempt to condense
	.Pq see Sy zfs_condense_indirect_vdevs_enable .
	.
	.It Sy zfs_dbgmsg_enable Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Internally ZFS keeps a small log to facilitate debugging.
	The log is enabled by default, and can be disabled by unsetting this option.
	The contents of the log can be accessed by reading
	.Pa /proc/spl/kstat/zfs/dbgmsg .
	Writing
	.Sy 0
	to the file clears the log.
	.Pp
	This setting does not influence debug prints due to
	.Sy zfs_flags .
	.
	.It Sy zfs_dbgmsg_maxsize Ns = Ns Sy 4194304 Ns B Po 4MB Pc Pq int
	Maximum size of the internal ZFS debug log.
	.
	.It Sy zfs_dbuf_state_index Ns = Ns Sy 0 Pq int
	Historically used for controlling what reporting was available under
	.Pa /proc/spl/kstat/zfs .
	No effect.
	.
	.It Sy zfs_deadman_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	When a pool sync operation takes longer than
	.Sy zfs_deadman_synctime_ms ,
	or when an individual I/O operation takes longer than
	.Sy zfs_deadman_ziotime_ms ,
	then the operation is considered to be "hung".
	If
	.Sy zfs_deadman_enabled
	is set, then the deadman behavior is invoked as described by
	.Sy zfs_deadman_failmode .
	By default, the deadman is enabled and set to
	.Sy wait
	which results in "hung" I/Os only being logged.
	The deadman is automatically disabled when a pool gets suspended.
	.
	.It Sy zfs_deadman_failmode Ns = Ns Sy wait Pq charp
	Controls the failure behavior when the deadman detects a "hung" I/O operation.
	Valid values are:
	.Bl -tag -compact -offset 4n -width "continue"
	.It Sy wait
	Wait for a "hung" operation to complete.
	For each "hung" operation a "deadman" event will be posted
	describing that operation.
	.It Sy continue
	Attempt to recover from a "hung" operation by re-dispatching it
	to the I/O pipeline if possible.
	.It Sy panic
	Panic the system.
	This can be used to facilitate automatic fail-over
	to a properly configured fail-over partner.
	.El
	.
	.It Sy zfs_deadman_checktime_ms Ns = Ns Sy 60000 Ns ms Po 1min Pc Pq int
	Check time in milliseconds.
	This defines the frequency at which we check for hung I/O requests
	and potentially invoke the
	.Sy zfs_deadman_failmode
	behavior.
	.
	.It Sy zfs_deadman_synctime_ms Ns = Ns Sy 600000 Ns ms Po 10min Pc Pq ulong
	Interval in milliseconds after which the deadman is triggered and also
	the interval after which a pool sync operation is considered to be "hung".
	Once this limit is exceeded the deadman will be invoked every
	.Sy zfs_deadman_checktime_ms
	milliseconds until the pool sync completes.
	.
	.It Sy zfs_deadman_ziotime_ms Ns = Ns Sy 300000 Ns ms Po 5min Pc Pq ulong
	Interval in milliseconds after which the deadman is triggered and an
	individual I/O operation is considered to be "hung".
	As long as the operation remains "hung",
	the deadman will be invoked every
	.Sy zfs_deadman_checktime_ms
	milliseconds until the operation completes.
	.
	.It Sy zfs_dedup_prefetch Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Enable prefetching dedup-ed blocks which are going to be freed.
	.
	.It Sy zfs_delay_min_dirty_percent Ns = Ns Sy 60 Ns % Pq int
	Start to delay each transaction once there is this amount of dirty data,
	expressed as a percentage of
	.Sy zfs_dirty_data_max .
	This value should be at least
	.Sy zfs_vdev_async_write_active_max_dirty_percent .
	.No See Sx ZFS TRANSACTION DELAY .
	.
	.It Sy zfs_delay_scale Ns = Ns Sy 500000 Pq int
	This controls how quickly the transaction delay approaches infinity.
	Larger values cause longer delays for a given amount of dirty data.
	.Pp
	For the smoothest delay, this value should be about 1 billion divided
	by the maximum number of operations per second.
	This will smoothly handle between ten times and a tenth of this number.
	.No See Sx ZFS TRANSACTION DELAY .
	.Pp
	.Sy zfs_delay_scale * zfs_dirty_data_max Em must be smaller than Sy 2^64 .
	.
	.It Sy zfs_disable_ivset_guid_check Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Disables requirement for IVset GUIDs to be present and match when doing a raw
	receive of encrypted datasets.
	Intended for users whose pools were created with
	OpenZFS pre-release versions and now have compatibility issues.
	.
	.It Sy zfs_key_max_salt_uses Ns = Ns Sy 400000000 Po 4*10^8 Pc Pq ulong
	Maximum number of uses of a single salt value before generating a new one for
	encrypted datasets.
	The default value is also the maximum.
	.
	.It Sy zfs_object_mutex_size Ns = Ns Sy 64 Pq uint
	Size of the znode hashtable used for holds.
	.Pp
	Due to the need to hold locks on objects that may not exist yet, kernel mutexes
	are not created per-object and instead a hashtable is used where collisions
	will result in objects waiting when there is not actually contention on the
	same object.
	.
	.It Sy zfs_slow_io_events_per_second Ns = Ns Sy 20 Ns /s Pq int
	Rate limit delay and deadman zevents (which report slow I/Os) to this many per
	second.
	.
	.It Sy zfs_unflushed_max_mem_amt Ns = Ns Sy 1073741824 Ns B Po 1GB Pc Pq ulong
	Upper-bound limit for unflushed metadata changes to be held by the
	log spacemap in memory, in bytes.
	.
	.It Sy zfs_unflushed_max_mem_ppm Ns = Ns Sy 1000 Ns ppm Po 0.1% Pc Pq ulong
	Part of overall system memory that ZFS allows to be used
	for unflushed metadata changes by the log spacemap, in millionths.
	.
	.It Sy zfs_unflushed_log_block_max Ns = Ns Sy 262144 Po 256k Pc Pq ulong
	Describes the maximum number of log spacemap blocks allowed for each pool.
	The default value means that the space in all the log spacemaps
	can add up to no more than
	.Sy 262144
	blocks (which means
	.Em 32GB
	of logical space before compression and ditto blocks,
	assuming that blocksize is
	.Em 128kB ) .
	.Pp
	This tunable is important because it involves a trade-off between import
	time after an unclean export and the frequency of flushing metaslabs.
	The higher this number is, the more log blocks we allow when the pool is
	active which means that we flush metaslabs less often and thus decrease
	the number of I/Os for spacemap updates per TXG.
	At the same time though, that means that in the event of an unclean export,
	there will be more log spacemap blocks for us to read, inducing overhead
	in the import time of the pool.
	The lower the number, the amount of flushing increases, destroying log
	blocks quicker as they become obsolete faster, which leaves less blocks
	to be read during import time after a crash.
	.Pp
	Each log spacemap block existing during pool import leads to approximately
	one extra logical I/O issued.
	This is the reason why this tunable is exposed in terms of blocks rather
	than space used.
	.
	.It Sy zfs_unflushed_log_block_min Ns = Ns Sy 1000 Pq ulong
	If the number of metaslabs is small and our incoming rate is high,
	we could get into a situation that we are flushing all our metaslabs every TXG.
	Thus we always allow at least this many log blocks.
	.
	.It Sy zfs_unflushed_log_block_pct Ns = Ns Sy 400 Ns % Pq ulong
	Tunable used to determine the number of blocks that can be used for
	the spacemap log, expressed as a percentage of the total number of
	metaslabs in the pool.
	.
	.It Sy zfs_unlink_suspend_progress Ns = Ns Sy 0 Ns \| Ns 1 Pq uint
	When enabled, files will not be asynchronously removed from the list of pending
	unlinks and the space they consume will be leaked.
	Once this option has been disabled and the dataset is remounted,
	the pending unlinks will be processed and the freed space returned to the pool.
	This option is used by the test suite.
	.
	.It Sy zfs_delete_blocks Ns = Ns Sy 20480 Pq ulong
	This is the used to define a large file for the purposes of deletion.
	Files containing more than
	.Sy zfs_delete_blocks
	will be deleted asynchronously, while smaller files are deleted synchronously.
	Decreasing this value will reduce the time spent in an
	.Xr unlink 2
	system call, at the expense of a longer delay before the freed space is available.
	.
	.It Sy zfs_dirty_data_max Ns = Pq int
	Determines the dirty space limit in bytes.
	Once this limit is exceeded, new writes are halted until space frees up.
	This parameter takes precedence over
	.Sy zfs_dirty_data_max_percent .
	.No See Sx ZFS TRANSACTION DELAY .
	.Pp
	Defaults to
	.Sy physical_ram/10 ,
	capped at
	.Sy zfs_dirty_data_max_max .
	.
	.It Sy zfs_dirty_data_max_max Ns = Pq int
	Maximum allowable value of
	.Sy zfs_dirty_data_max ,
	expressed in bytes.
	This limit is only enforced at module load time, and will be ignored if
	.Sy zfs_dirty_data_max
	is later changed.
	This parameter takes precedence over
	.Sy zfs_dirty_data_max_max_percent .
	.No See Sx ZFS TRANSACTION DELAY .
	.Pp
	Defaults to
	.Sy physical_ram/4 ,
	.
	.It Sy zfs_dirty_data_max_max_percent Ns = Ns Sy 25 Ns % Pq int
	Maximum allowable value of
	.Sy zfs_dirty_data_max ,
	expressed as a percentage of physical RAM.
	This limit is only enforced at module load time, and will be ignored if
	.Sy zfs_dirty_data_max
	is later changed.
	The parameter
	.Sy zfs_dirty_data_max_max
	takes precedence over this one.
	.No See Sx ZFS TRANSACTION DELAY .
	.
	.It Sy zfs_dirty_data_max_percent Ns = Ns Sy 10 Ns % Pq int
	Determines the dirty space limit, expressed as a percentage of all memory.
	Once this limit is exceeded, new writes are halted until space frees up.
	The parameter
	.Sy zfs_dirty_data_max
	takes precedence over this one.
	.No See Sx ZFS TRANSACTION DELAY .
	.Pp
	Subject to
	.Sy zfs_dirty_data_max_max .
	.
	.It Sy zfs_dirty_data_sync_percent Ns = Ns Sy 20 Ns % Pq int
	Start syncing out a transaction group if there's at least this much dirty data
	.Pq as a percentage of Sy zfs_dirty_data_max .
	This should be less than
	.Sy zfs_vdev_async_write_active_min_dirty_percent .
	.
	.It Sy zfs_wrlog_data_max Ns = Pq int
	The upper limit of write-transaction zil log data size in bytes.
	Once it is reached, write operation is blocked, until log data is cleared out
	after transaction group sync. Because of some overhead, it should be set
	at least 2 times the size of
	.Sy zfs_dirty_data_max
	.No to prevent harming normal write throughput.
	It also should be smaller than the size of the slog device if slog is present.
	.Pp
	Defaults to
	.Sy zfs_dirty_data_max*2
	.
	.It Sy zfs_fallocate_reserve_percent Ns = Ns Sy 110 Ns % Pq uint
	Since ZFS is a copy-on-write filesystem with snapshots, blocks cannot be
	preallocated for a file in order to guarantee that later writes will not
	run out of space.
	Instead,
	.Xr fallocate 2
	space preallocation only checks that sufficient space is currently available
	in the pool or the user's project quota allocation,
	and then creates a sparse file of the requested size.
	The requested space is multiplied by
	.Sy zfs_fallocate_reserve_percent
	to allow additional space for indirect blocks and other internal metadata.
	Setting this to
	.Sy 0
	disables support for
	.Xr fallocate 2
	and causes it to return
	.Sy EOPNOTSUPP .
	.
	.It Sy zfs_fletcher_4_impl Ns = Ns Sy fastest Pq string
	Select a fletcher 4 implementation.
	.Pp
	Supported selectors are:
	.Sy fastest , scalar , sse2 , ssse3 , avx2 , avx512f , avx512bw ,
	.No and Sy aarch64_neon .
	All except
	.Sy fastest No and Sy scalar
	require instruction set extensions to be available,
	and will only appear if ZFS detects that they are present at runtime.
	If multiple implementations of fletcher 4 are available, the
	.Sy fastest
	will be chosen using a micro benchmark.
	Selecting
	.Sy scalar
	results in the original CPU-based calculation being used.
	Selecting any option other than
	.Sy fastest No or Sy scalar
	results in vector instructions
	from the respective CPU instruction set being used.
	.
	.It Sy zfs_free_bpobj_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Enable/disable the processing of the free_bpobj object.
	.
	.It Sy zfs_async_block_max_blocks Ns = Ns Sy ULONG_MAX Po unlimited Pc Pq ulong
	Maximum number of blocks freed in a single TXG.
	.
	.It Sy zfs_max_async_dedup_frees Ns = Ns Sy 100000 Po 10^5 Pc Pq ulong
	Maximum number of dedup blocks freed in a single TXG.
	.
	.It Sy zfs_override_estimate_recordsize Ns = Ns Sy 0 Pq ulong
	If nonzer, override record size calculation for
	.Nm zfs Cm send
	estimates.
	.
	.It Sy zfs_vdev_async_read_max_active Ns = Ns Sy 3 Pq int
	Maximum asynchronous read I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_async_read_min_active Ns = Ns Sy 1 Pq int
	Minimum asynchronous read I/O operation active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_async_write_active_max_dirty_percent Ns = Ns Sy 60 Ns % Pq int
	When the pool has more than this much dirty data, use
	.Sy zfs_vdev_async_write_max_active
	to limit active async writes.
	If the dirty data is between the minimum and maximum,
	the active I/O limit is linearly interpolated.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_async_write_active_min_dirty_percent Ns = Ns Sy 30 Ns % Pq int
	When the pool has less than this much dirty data, use
	.Sy zfs_vdev_async_write_min_active
	to limit active async writes.
	If the dirty data is between the minimum and maximum,
	the active I/O limit is linearly
	interpolated.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_async_write_max_active Ns = Ns Sy 30 Pq int
	Maximum asynchronous write I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_async_write_min_active Ns = Ns Sy 2 Pq int
	Minimum asynchronous write I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.Pp
	Lower values are associated with better latency on rotational media but poorer
	resilver performance.
	The default value of
	.Sy 2
	was chosen as a compromise.
	A value of
	.Sy 3
	has been shown to improve resilver performance further at a cost of
	further increasing latency.
	.
	.It Sy zfs_vdev_initializing_max_active Ns = Ns Sy 1 Pq int
	Maximum initializing I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_initializing_min_active Ns = Ns Sy 1 Pq int
	Minimum initializing I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_max_active Ns = Ns Sy 1000 Pq int
	The maximum number of I/O operations active to each device.
	Ideally, this will be at least the sum of each queue's
	.Sy max_active .
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_rebuild_max_active Ns = Ns Sy 3 Pq int
	Maximum sequential resilver I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_rebuild_min_active Ns = Ns Sy 1 Pq int
	Minimum sequential resilver I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_removal_max_active Ns = Ns Sy 2 Pq int
	Maximum removal I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_removal_min_active Ns = Ns Sy 1 Pq int
	Minimum removal I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_scrub_max_active Ns = Ns Sy 2 Pq int
	Maximum scrub I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_scrub_min_active Ns = Ns Sy 1 Pq int
	Minimum scrub I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_sync_read_max_active Ns = Ns Sy 10 Pq int
	Maximum synchronous read I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_sync_read_min_active Ns = Ns Sy 10 Pq int
	Minimum synchronous read I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_sync_write_max_active Ns = Ns Sy 10 Pq int
	Maximum synchronous write I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_sync_write_min_active Ns = Ns Sy 10 Pq int
	Minimum synchronous write I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_trim_max_active Ns = Ns Sy 2 Pq int
	Maximum trim/discard I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_trim_min_active Ns = Ns Sy 1 Pq int
	Minimum trim/discard I/O operations active to each device.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_nia_delay Ns = Ns Sy 5 Pq int
	For non-interactive I/O (scrub, resilver, removal, initialize and rebuild),
	the number of concurrently-active I/O operations is limited to
	.Sy zfs_*_min_active ,
	unless the vdev is "idle".
	When there are no interactive I/O operatinons active (synchronous or otherwise),
	and
	.Sy zfs_vdev_nia_delay
	operations have completed since the last interactive operation,
	then the vdev is considered to be "idle",
	and the number of concurrently-active non-interactive operations is increased to
	.Sy zfs_*_max_active .
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_nia_credit Ns = Ns Sy 5 Pq int
	Some HDDs tend to prioritize sequential I/O so strongly, that concurrent
	random I/O latency reaches several seconds.
	On some HDDs this happens even if sequential I/O operations
	are submitted one at a time, and so setting
	.Sy zfs_*_max_active Ns = Sy 1
	does not help.
	To prevent non-interactive I/O, like scrub,
	from monopolizing the device, no more than
	.Sy zfs_vdev_nia_credit operations can be sent
	while there are outstanding incomplete interactive operations.
	This enforced wait ensures the HDD services the interactive I/O
	within a reasonable amount of time.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_queue_depth_pct Ns = Ns Sy 1000 Ns % Pq int
	Maximum number of queued allocations per top-level vdev expressed as
	a percentage of
	.Sy zfs_vdev_async_write_max_active ,
	which allows the system to detect devices that are more capable
	of handling allocations and to allocate more blocks to those devices.
	This allows for dynamic allocation distribution when devices are imbalanced,
	as fuller devices will tend to be slower than empty devices.
	.Pp
	Also see
	.Sy zio_dva_throttle_enabled .
	.
	.It Sy zfs_expire_snapshot Ns = Ns Sy 300 Ns s Pq int
	Time before expiring
	.Pa .zfs/snapshot .
	.
	.It Sy zfs_admin_snapshot Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Allow the creation, removal, or renaming of entries in the
	.Sy .zfs/snapshot
	directory to cause the creation, destruction, or renaming of snapshots.
	When enabled, this functionality works both locally and over NFS exports
	which have the
	.Em no_root_squash
	option set.
	.
	.It Sy zfs_flags Ns = Ns Sy 0 Pq int
	Set additional debugging flags.
	The following flags may be bitwise-ored together:
	.TS
	box;
	lbz r l l .
	Value Symbolic Name Description
	_
	1 ZFS_DEBUG_DPRINTF Enable dprintf entries in the debug log.
	* 2 ZFS_DEBUG_DBUF_VERIFY Enable extra dbuf verifications.
	* 4 ZFS_DEBUG_DNODE_VERIFY Enable extra dnode verifications.
	8 ZFS_DEBUG_SNAPNAMES Enable snapshot name verification.
	16 ZFS_DEBUG_MODIFY Check for illegally modified ARC buffers.
	64 ZFS_DEBUG_ZIO_FREE Enable verification of block frees.
	128 ZFS_DEBUG_HISTOGRAM_VERIFY Enable extra spacemap histogram verifications.
	256 ZFS_DEBUG_METASLAB_VERIFY Verify space accounting on disk matches in-memory \fBrange_trees\fP.
	512 ZFS_DEBUG_SET_ERROR Enable \fBSET_ERROR\fP and dprintf entries in the debug log.
	1024 ZFS_DEBUG_INDIRECT_REMAP Verify split blocks created by device removal.
	2048 ZFS_DEBUG_TRIM Verify TRIM ranges are always within the allocatable range tree.
	4096 ZFS_DEBUG_LOG_SPACEMAP Verify that the log summary is consistent with the spacemap log
	and enable \fBzfs_dbgmsgs\fP for metaslab loading and flushing.
	.TE
	.Sy \& * No Requires debug build.
	.
	.It Sy zfs_free_leak_on_eio Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	If destroy encounters an
	.Sy EIO
	while reading metadata (e.g. indirect blocks),
	space referenced by the missing metadata can not be freed.
	Normally this causes the background destroy to become "stalled",
	as it is unable to make forward progress.
	While in this stalled state, all remaining space to free
	from the error-encountering filesystem is "temporarily leaked".
	Set this flag to cause it to ignore the
	.Sy EIO ,
	permanently leak the space from indirect blocks that can not be read,
	and continue to free everything else that it can.
	.Pp
	The default "stalling" behavior is useful if the storage partially
	fails (i.e. some but not all I/O operations fail), and then later recovers.
	In this case, we will be able to continue pool operations while it is
	partially failed, and when it recovers, we can continue to free the
	space, with no leaks.
	Note, however, that this case is actually fairly rare.
	.Pp
	Typically pools either
	.Bl -enum -compact -offset 4n -width "1."
	.It
	fail completely (but perhaps temporarily,
	e.g. due to a top-level vdev going offline), or
	.It
	have localized, permanent errors (e.g. disk returns the wrong data
	due to bit flip or firmware bug).
	.El
	In the former case, this setting does not matter because the
	pool will be suspended and the sync thread will not be able to make
	forward progress regardless.
	In the latter, because the error is permanent, the best we can do
	is leak the minimum amount of space,
	which is what setting this flag will do.
	It is therefore reasonable for this flag to normally be set,
	but we chose the more conservative approach of not setting it,
	so that there is no possibility of
	leaking space in the "partial temporary" failure case.
	.
	.It Sy zfs_free_min_time_ms Ns = Ns Sy 1000 Ns ms Po 1s Pc Pq int
	During a
	.Nm zfs Cm destroy
	operation using the
	.Sy async_destroy
	feature,
	a minimum of this much time will be spent working on freeing blocks per TXG.
	.
	.It Sy zfs_obsolete_min_time_ms Ns = Ns Sy 500 Ns ms Pq int
	Similar to
	.Sy zfs_free_min_time_ms ,
	but for cleanup of old indirection records for removed vdevs.
	.
	.It Sy zfs_immediate_write_sz Ns = Ns Sy 32768 Ns B Po 32kB Pc Pq long
	Largest data block to write to the ZIL.
	Larger blocks will be treated as if the dataset being written to had the
	.Sy logbias Ns = Ns Sy throughput
	property set.
	.
	.It Sy zfs_initialize_value Ns = Ns Sy 16045690984833335022 Po 0xDEADBEEFDEADBEEE Pc Pq ulong
	Pattern written to vdev free space by
	.Xr zpool-initialize 8 .
	.
	.It Sy zfs_initialize_chunk_size Ns = Ns Sy 1048576 Ns B Po 1MB Pc Pq ulong
	Size of writes used by
	.Xr zpool-initialize 8 .
	This option is used by the test suite.
	.
	.It Sy zfs_livelist_max_entries Ns = Ns Sy 500000 Po 5*10^5 Pc Pq ulong
	The threshold size (in block pointers) at which we create a new sub-livelist.
	Larger sublists are more costly from a memory perspective but the fewer
	sublists there are, the lower the cost of insertion.
	.
	.It Sy zfs_livelist_min_percent_shared Ns = Ns Sy 75 Ns % Pq int
	If the amount of shared space between a snapshot and its clone drops below
	this threshold, the clone turns off the livelist and reverts to the old
	deletion method.
	This is in place because livelists no long give us a benefit
	once a clone has been overwritten enough.
	.
	.It Sy zfs_livelist_condense_new_alloc Ns = Ns Sy 0 Pq int
	Incremented each time an extra ALLOC blkptr is added to a livelist entry while
	it is being condensed.
	This option is used by the test suite to track race conditions.
	.
	.It Sy zfs_livelist_condense_sync_cancel Ns = Ns Sy 0 Pq int
	Incremented each time livelist condensing is canceled while in
	.Fn spa_livelist_condense_sync .
	This option is used by the test suite to track race conditions.
	.
	.It Sy zfs_livelist_condense_sync_pause Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	When set, the livelist condense process pauses indefinitely before
	executing the synctask -
	.Fn spa_livelist_condense_sync .
	This option is used by the test suite to trigger race conditions.
	.
	.It Sy zfs_livelist_condense_zthr_cancel Ns = Ns Sy 0 Pq int
	Incremented each time livelist condensing is canceled while in
	.Fn spa_livelist_condense_cb .
	This option is used by the test suite to track race conditions.
	.
	.It Sy zfs_livelist_condense_zthr_pause Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	When set, the livelist condense process pauses indefinitely before
	executing the open context condensing work in
	.Fn spa_livelist_condense_cb .
	This option is used by the test suite to trigger race conditions.
	.
	.It Sy zfs_lua_max_instrlimit Ns = Ns Sy 100000000 Po 10^8 Pc Pq ulong
	The maximum execution time limit that can be set for a ZFS channel program,
	specified as a number of Lua instructions.
	.
	.It Sy zfs_lua_max_memlimit Ns = Ns Sy 104857600 Po 100MB Pc Pq ulong
	The maximum memory limit that can be set for a ZFS channel program, specified
	in bytes.
	.
	.It Sy zfs_max_dataset_nesting Ns = Ns Sy 50 Pq int
	The maximum depth of nested datasets.
	This value can be tuned temporarily to
	fix existing datasets that exceed the predefined limit.
	.
	.It Sy zfs_max_log_walking Ns = Ns Sy 5 Pq ulong
	The number of past TXGs that the flushing algorithm of the log spacemap
	feature uses to estimate incoming log blocks.
	.
	.It Sy zfs_max_logsm_summary_length Ns = Ns Sy 10 Pq ulong
	Maximum number of rows allowed in the summary of the spacemap log.
	.
	.It Sy zfs_max_recordsize Ns = Ns Sy 1048576 Po 1MB Pc Pq int
	We currently support block sizes from
	.Em 512B No to Em 16MB .
	The benefits of larger blocks, and thus larger I/O,
	need to be weighed against the cost of COWing a giant block to modify one byte.
	Additionally, very large blocks can have an impact on I/O latency,
	and also potentially on the memory allocator.
	Therefore, we do not allow the recordsize to be set larger than this tunable.
	Larger blocks can be created by changing it,
	and pools with larger blocks can always be imported and used,
	regardless of this setting.
	.
	.It Sy zfs_allow_redacted_dataset_mount Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Allow datasets received with redacted send/receive to be mounted.
	Normally disabled because these datasets may be missing key data.
	.
	.It Sy zfs_min_metaslabs_to_flush Ns = Ns Sy 1 Pq ulong
	Minimum number of metaslabs to flush per dirty TXG.
	.
	.It Sy zfs_metaslab_fragmentation_threshold Ns = Ns Sy 70 Ns % Pq int
	Allow metaslabs to keep their active state as long as their fragmentation
	percentage is no more than this value.
	An active metaslab that exceeds this threshold
	will no longer keep its active status allowing better metaslabs to be selected.
	.
	.It Sy zfs_mg_fragmentation_threshold Ns = Ns Sy 95 Ns % Pq int
	Metaslab groups are considered eligible for allocations if their
	fragmentation metric (measured as a percentage) is less than or equal to
	this value.
	If a metaslab group exceeds this threshold then it will be
	skipped unless all metaslab groups within the metaslab class have also
	crossed this threshold.
	.
	.It Sy zfs_mg_noalloc_threshold Ns = Ns Sy 0 Ns % Pq int
	Defines a threshold at which metaslab groups should be eligible for allocations.
	The value is expressed as a percentage of free space
	beyond which a metaslab group is always eligible for allocations.
	If a metaslab group's free space is less than or equal to the
	threshold, the allocator will avoid allocating to that group
	unless all groups in the pool have reached the threshold.
	Once all groups have reached the threshold, all groups are allowed to accept
	allocations.
	The default value of
	.Sy 0
	disables the feature and causes all metaslab groups to be eligible for allocations.
	.Pp
	This parameter allows one to deal with pools having heavily imbalanced
	vdevs such as would be the case when a new vdev has been added.
	Setting the threshold to a non-zero percentage will stop allocations
	from being made to vdevs that aren't filled to the specified percentage
	and allow lesser filled vdevs to acquire more allocations than they
	otherwise would under the old
	.Sy zfs_mg_alloc_failures
	facility.
	.
	.It Sy zfs_ddt_data_is_special Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	If enabled, ZFS will place DDT data into the special allocation class.
	.
	.It Sy zfs_user_indirect_is_special Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	If enabled, ZFS will place user data indirect blocks
	into the special allocation class.
	.
	.It Sy zfs_multihost_history Ns = Ns Sy 0 Pq int
	Historical statistics for this many latest multihost updates will be available in
	.Pa /proc/spl/kstat/zfs/ Ns Ao Ar pool Ac Ns Pa /multihost .
	.
	.It Sy zfs_multihost_interval Ns = Ns Sy 1000 Ns ms Po 1s Pc Pq ulong
	Used to control the frequency of multihost writes which are performed when the
	.Sy multihost
	pool property is on.
	This is one of the factors used to determine the
	length of the activity check during import.
	.Pp
	The multihost write period is
	.Sy zfs_multihost_interval / leaf-vdevs .
	On average a multihost write will be issued for each leaf vdev
	every
	.Sy zfs_multihost_interval
	milliseconds.
	In practice, the observed period can vary with the I/O load
	and this observed value is the delay which is stored in the uberblock.
	.
	.It Sy zfs_multihost_import_intervals Ns = Ns Sy 20 Pq uint
	Used to control the duration of the activity test on import.
	Smaller values of
	.Sy zfs_multihost_import_intervals
	will reduce the import time but increase
	the risk of failing to detect an active pool.
	The total activity check time is never allowed to drop below one second.
	.Pp
	On import the activity check waits a minimum amount of time determined by
	.Sy zfs_multihost_interval * zfs_multihost_import_intervals ,
	or the same product computed on the host which last had the pool imported,
	whichever is greater.
	The activity check time may be further extended if the value of MMP
	delay found in the best uberblock indicates actual multihost updates happened
	at longer intervals than
	.Sy zfs_multihost_interval .
	A minimum of
	.Em 100ms
	is enforced.
	.Pp
	.Sy 0 No is equivalent to Sy 1 .
	.
	.It Sy zfs_multihost_fail_intervals Ns = Ns Sy 10 Pq uint
	Controls the behavior of the pool when multihost write failures or delays are
	detected.
	.Pp
	When
	.Sy 0 ,
	multihost write failures or delays are ignored.
	The failures will still be reported to the ZED which depending on
	its configuration may take action such as suspending the pool or offlining a
	device.
	.Pp
	Otherwise, the pool will be suspended if
	.Sy zfs_multihost_fail_intervals * zfs_multihost_interval
	milliseconds pass without a successful MMP write.
	This guarantees the activity test will see MMP writes if the pool is imported.
	.Sy 1 No is equivalent to Sy 2 ;
	this is necessary to prevent the pool from being suspended
	due to normal, small I/O latency variations.
	.
	.It Sy zfs_no_scrub_io Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Set to disable scrub I/O.
	This results in scrubs not actually scrubbing data and
	simply doing a metadata crawl of the pool instead.
	.
	.It Sy zfs_no_scrub_prefetch Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Set to disable block prefetching for scrubs.
	.
	.It Sy zfs_nocacheflush Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Disable cache flush operations on disks when writing.
	Setting this will cause pool corruption on power loss
	if a volatile out-of-order write cache is enabled.
	.
	.It Sy zfs_nopwrite_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Allow no-operation writes.
	The occurrence of nopwrites will further depend on other pool properties
	.Pq i.a. the checksumming and compression algorithms .
	.
	-.It Sy zfs_dmu_offset_next_sync Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	+.It Sy zfs_dmu_offset_next_sync Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Enable forcing TXG sync to find holes.
	-When enabled forces ZFS to act like prior versions when
	+When enabled forces ZFS to sync data when
	.Sy SEEK_HOLE No or Sy SEEK_DATA
	-flags are used, which, when a dnode is dirty,
	-causes TXGs to be synced so that this data can be found.
	+flags are used allowing holes in a file to be accurately reported.
	+When disabled holes will not be reported in recently dirtied files.
	.
	.It Sy zfs_pd_bytes_max Ns = Ns Sy 52428800 Ns B Po 50MB Pc Pq int
	The number of bytes which should be prefetched during a pool traversal, like
	.Nm zfs Cm send
	or other data crawling operations.
	.
	.It Sy zfs_traverse_indirect_prefetch_limit Ns = Ns Sy 32 Pq int
	The number of blocks pointed by indirect (non-L0) block which should be
	prefetched during a pool traversal, like
	.Nm zfs Cm send
	or other data crawling operations.
	.
	.It Sy zfs_per_txg_dirty_frees_percent Ns = Ns Sy 5 Ns % Pq ulong
	Control percentage of dirtied indirect blocks from frees allowed into one TXG.
	After this threshold is crossed, additional frees will wait until the next TXG.
	.Sy 0 No disables this throttle.
	.
	.It Sy zfs_prefetch_disable Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Disable predictive prefetch.
	Note that it leaves "prescient" prefetch (for. e.g.\&
	.Nm zfs Cm send )
	intact.
	Unlike predictive prefetch, prescient prefetch never issues I/O
	that ends up not being needed, so it can't hurt performance.
	.
	.It Sy zfs_qat_checksum_disable Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Disable QAT hardware acceleration for SHA256 checksums.
	May be unset after the ZFS modules have been loaded to initialize the QAT
	hardware as long as support is compiled in and the QAT driver is present.
	.
	.It Sy zfs_qat_compress_disable Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Disable QAT hardware acceleration for gzip compression.
	May be unset after the ZFS modules have been loaded to initialize the QAT
	hardware as long as support is compiled in and the QAT driver is present.
	.
	.It Sy zfs_qat_encrypt_disable Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Disable QAT hardware acceleration for AES-GCM encryption.
	May be unset after the ZFS modules have been loaded to initialize the QAT
	hardware as long as support is compiled in and the QAT driver is present.
	.
	.It Sy zfs_vnops_read_chunk_size Ns = Ns Sy 1048576 Ns B Po 1MB Pc Pq long
	Bytes to read per chunk.
	.
	.It Sy zfs_read_history Ns = Ns Sy 0 Pq int
	Historical statistics for this many latest reads will be available in
	.Pa /proc/spl/kstat/zfs/ Ns Ao Ar pool Ac Ns Pa /reads .
	.
	.It Sy zfs_read_history_hits Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Include cache hits in read history
	.
	.It Sy zfs_rebuild_max_segment Ns = Ns Sy 1048576 Ns B Po 1MB Pc Pq ulong
	Maximum read segment size to issue when sequentially resilvering a
	top-level vdev.
	.
	.It Sy zfs_rebuild_scrub_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Automatically start a pool scrub when the last active sequential resilver
	completes in order to verify the checksums of all blocks which have been
	resilvered.
	This is enabled by default and strongly recommended.
	.
	.It Sy zfs_rebuild_vdev_limit Ns = Ns Sy 33554432 Ns B Po 32MB Pc Pq ulong
	Maximum amount of I/O that can be concurrently issued for a sequential
	resilver per leaf device, given in bytes.
	.
	.It Sy zfs_reconstruct_indirect_combinations_max Ns = Ns Sy 4096 Pq int
	If an indirect split block contains more than this many possible unique
	combinations when being reconstructed, consider it too computationally
	expensive to check them all.
	Instead, try at most this many randomly selected
	combinations each time the block is accessed.
	This allows all segment copies to participate fairly
	in the reconstruction when all combinations
	cannot be checked and prevents repeated use of one bad copy.
	.
	.It Sy zfs_recover Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Set to attempt to recover from fatal errors.
	This should only be used as a last resort,
	as it typically results in leaked space, or worse.
	.
	.It Sy zfs_removal_ignore_errors Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Ignore hard IO errors during device removal.
	When set, if a device encounters a hard IO error during the removal process
	the removal will not be cancelled.
	This can result in a normally recoverable block becoming permanently damaged
	and is hence not recommended.
	This should only be used as a last resort when the
	pool cannot be returned to a healthy state prior to removing the device.
	.
	.It Sy zfs_removal_suspend_progress Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	This is used by the test suite so that it can ensure that certain actions
	happen while in the middle of a removal.
	.
	.It Sy zfs_remove_max_segment Ns = Ns Sy 16777216 Ns B Po 16MB Pc Pq int
	The largest contiguous segment that we will attempt to allocate when removing
	a device.
	If there is a performance problem with attempting to allocate large blocks,
	consider decreasing this.
	The default value is also the maximum.
	.
	.It Sy zfs_resilver_disable_defer Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Ignore the
	.Sy resilver_defer
	feature, causing an operation that would start a resilver to
	immediately restart the one in progress.
	.
	.It Sy zfs_resilver_min_time_ms Ns = Ns Sy 3000 Ns ms Po 3s Pc Pq int
	Resilvers are processed by the sync thread.
	While resilvering, it will spend at least this much time
	working on a resilver between TXG flushes.
	.
	.It Sy zfs_scan_ignore_errors Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	If set, remove the DTL (dirty time list) upon completion of a pool scan (scrub),
	even if there were unrepairable errors.
	Intended to be used during pool repair or recovery to
	stop resilvering when the pool is next imported.
	.
	.It Sy zfs_scrub_min_time_ms Ns = Ns Sy 1000 Ns ms Po 1s Pc Pq int
	Scrubs are processed by the sync thread.
	While scrubbing, it will spend at least this much time
	working on a scrub between TXG flushes.
	.
	.It Sy zfs_scan_checkpoint_intval Ns = Ns Sy 7200 Ns s Po 2h Pc Pq int
	To preserve progress across reboots, the sequential scan algorithm periodically
	needs to stop metadata scanning and issue all the verification I/O to disk.
	The frequency of this flushing is determined by this tunable.
	.
	.It Sy zfs_scan_fill_weight Ns = Ns Sy 3 Pq int
	This tunable affects how scrub and resilver I/O segments are ordered.
	A higher number indicates that we care more about how filled in a segment is,
	while a lower number indicates we care more about the size of the extent without
	considering the gaps within a segment.
	This value is only tunable upon module insertion.
	-Changing the value afterwards will have no affect on scrub or resilver performance.
	+Changing the value afterwards will have no effect on scrub or resilver performance.
	.
	.It Sy zfs_scan_issue_strategy Ns = Ns Sy 0 Pq int
	Determines the order that data will be verified while scrubbing or resilvering:
	.Bl -tag -compact -offset 4n -width "a"
	.It Sy 1
	Data will be verified as sequentially as possible, given the
	amount of memory reserved for scrubbing
	.Pq see Sy zfs_scan_mem_lim_fact .
	This may improve scrub performance if the pool's data is very fragmented.
	.It Sy 2
	The largest mostly-contiguous chunk of found data will be verified first.
	By deferring scrubbing of small segments, we may later find adjacent data
	to coalesce and increase the segment size.
	.It Sy 0
	.No Use strategy Sy 1 No during normal verification
	.No and strategy Sy 2 No while taking a checkpoint.
	.El
	.
	.It Sy zfs_scan_legacy Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	If unset, indicates that scrubs and resilvers will gather metadata in
	memory before issuing sequential I/O.
	Otherwise indicates that the legacy algorithm will be used,
	where I/O is initiated as soon as it is discovered.
	Unsetting will not affect scrubs or resilvers that are already in progress.
	.
	.It Sy zfs_scan_max_ext_gap Ns = Ns Sy 2097152 Ns B Po 2MB Pc Pq int
	Sets the largest gap in bytes between scrub/resilver I/O operations
	that will still be considered sequential for sorting purposes.
	Changing this value will not
	affect scrubs or resilvers that are already in progress.
	.
	.It Sy zfs_scan_mem_lim_fact Ns = Ns Sy 20 Ns ^-1 Pq int
	Maximum fraction of RAM used for I/O sorting by sequential scan algorithm.
	This tunable determines the hard limit for I/O sorting memory usage.
	When the hard limit is reached we stop scanning metadata and start issuing
	data verification I/O.
	This is done until we get below the soft limit.
	.
	.It Sy zfs_scan_mem_lim_soft_fact Ns = Ns Sy 20 Ns ^-1 Pq int
	The fraction of the hard limit used to determined the soft limit for I/O sorting
	by the sequential scan algorithm.
	When we cross this limit from below no action is taken.
	When we cross this limit from above it is because we are issuing verification I/O.
	In this case (unless the metadata scan is done) we stop issuing verification I/O
	and start scanning metadata again until we get to the hard limit.
	.
	.It Sy zfs_scan_strict_mem_lim Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Enforce tight memory limits on pool scans when a sequential scan is in progress.
	When disabled, the memory limit may be exceeded by fast disks.
	.
	.It Sy zfs_scan_suspend_progress Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Freezes a scrub/resilver in progress without actually pausing it.
	Intended for testing/debugging.
	.
	.It Sy zfs_scan_vdev_limit Ns = Ns Sy 4194304 Ns B Po 4MB Pc Pq int
	Maximum amount of data that can be concurrently issued at once for scrubs and
	resilvers per leaf device, given in bytes.
	.
	.It Sy zfs_send_corrupt_data Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Allow sending of corrupt data (ignore read/checksum errors when sending).
	.
	.It Sy zfs_send_unmodified_spill_blocks Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Include unmodified spill blocks in the send stream.
	Under certain circumstances, previous versions of ZFS could incorrectly
	remove the spill block from an existing object.
	Including unmodified copies of the spill blocks creates a backwards-compatible
	stream which will recreate a spill block if it was incorrectly removed.
	.
	.It Sy zfs_send_no_prefetch_queue_ff Ns = Ns Sy 20 Ns ^-1 Pq int
	The fill fraction of the
	.Nm zfs Cm send
	internal queues.
	The fill fraction controls the timing with which internal threads are woken up.
	.
	.It Sy zfs_send_no_prefetch_queue_length Ns = Ns Sy 1048576 Ns B Po 1MB Pc Pq int
	The maximum number of bytes allowed in
	.Nm zfs Cm send Ns 's
	internal queues.
	.
	.It Sy zfs_send_queue_ff Ns = Ns Sy 20 Ns ^-1 Pq int
	The fill fraction of the
	.Nm zfs Cm send
	prefetch queue.
	The fill fraction controls the timing with which internal threads are woken up.
	.
	.It Sy zfs_send_queue_length Ns = Ns Sy 16777216 Ns B Po 16MB Pc Pq int
	The maximum number of bytes allowed that will be prefetched by
	.Nm zfs Cm send .
	This value must be at least twice the maximum block size in use.
	.
	.It Sy zfs_recv_queue_ff Ns = Ns Sy 20 Ns ^-1 Pq int
	The fill fraction of the
	.Nm zfs Cm receive
	queue.
	The fill fraction controls the timing with which internal threads are woken up.
	.
	.It Sy zfs_recv_queue_length Ns = Ns Sy 16777216 Ns B Po 16MB Pc Pq int
	The maximum number of bytes allowed in the
	.Nm zfs Cm receive
	queue.
	This value must be at least twice the maximum block size in use.
	.
	.It Sy zfs_recv_write_batch_size Ns = Ns Sy 1048576 Ns B Po 1MB Pc Pq int
	The maximum amount of data, in bytes, that
	.Nm zfs Cm receive
	will write in one DMU transaction.
	This is the uncompressed size, even when receiving a compressed send stream.
	This setting will not reduce the write size below a single block.
	Capped at a maximum of
	.Sy 32MB .
	.
	.It Sy zfs_override_estimate_recordsize Ns = Ns Sy 0 Ns \| Ns 1 Pq ulong
	Setting this variable overrides the default logic for estimating block
	sizes when doing a
	.Nm zfs Cm send .
	The default heuristic is that the average block size
	will be the current recordsize.
	Override this value if most data in your dataset is not of that size
	and you require accurate zfs send size estimates.
	.
	.It Sy zfs_sync_pass_deferred_free Ns = Ns Sy 2 Pq int
	Flushing of data to disk is done in passes.
	Defer frees starting in this pass.
	.
	.It Sy zfs_spa_discard_memory_limit Ns = Ns Sy 16777216 Ns B Po 16MB Pc Pq int
	Maximum memory used for prefetching a checkpoint's space map on each
	vdev while discarding the checkpoint.
	.
	.It Sy zfs_special_class_metadata_reserve_pct Ns = Ns Sy 25 Ns % Pq int
	Only allow small data blocks to be allocated on the special and dedup vdev
	types when the available free space percentage on these vdevs exceeds this value.
	This ensures reserved space is available for pool metadata as the
	special vdevs approach capacity.
	.
	.It Sy zfs_sync_pass_dont_compress Ns = Ns Sy 8 Pq int
	Starting in this sync pass, disable compression (including of metadata).
	With the default setting, in practice, we don't have this many sync passes,
	so this has no effect.
	.Pp
	The original intent was that disabling compression would help the sync passes
	to converge.
	However, in practice, disabling compression increases
	the average number of sync passes; because when we turn compression off,
	many blocks' size will change, and thus we have to re-allocate
	(not overwrite) them.
	It also increases the number of
	.Em 128kB
	allocations (e.g. for indirect blocks and spacemaps)
	because these will not be compressed.
	The
	.Em 128kB
	allocations are especially detrimental to performance
	on highly fragmented systems, which may have very few free segments of this size,
	and may need to load new metaslabs to satisfy these allocations.
	.
	.It Sy zfs_sync_pass_rewrite Ns = Ns Sy 2 Pq int
	Rewrite new block pointers starting in this pass.
	.
	.It Sy zfs_sync_taskq_batch_pct Ns = Ns Sy 75 Ns % Pq int
	This controls the number of threads used by
	.Sy dp_sync_taskq .
	The default value of
	.Sy 75%
	will create a maximum of one thread per CPU.
	.
	.It Sy zfs_trim_extent_bytes_max Ns = Ns Sy 134217728 Ns B Po 128MB Pc Pq uint
	Maximum size of TRIM command.
	Larger ranges will be split into chunks no larger than this value before issuing.
	.
	.It Sy zfs_trim_extent_bytes_min Ns = Ns Sy 32768 Ns B Po 32kB Pc Pq uint
	Minimum size of TRIM commands.
	TRIM ranges smaller than this will be skipped,
	unless they're part of a larger range which was chunked.
	This is done because it's common for these small TRIMs
	to negatively impact overall performance.
	.
	.It Sy zfs_trim_metaslab_skip Ns = Ns Sy 0 Ns \| Ns 1 Pq uint
	Skip uninitialized metaslabs during the TRIM process.
	This option is useful for pools constructed from large thinly-provisioned devices
	where TRIM operations are slow.
	As a pool ages, an increasing fraction of the pool's metaslabs
	will be initialized, progressively degrading the usefulness of this option.
	This setting is stored when starting a manual TRIM and will
	persist for the duration of the requested TRIM.
	.
	.It Sy zfs_trim_queue_limit Ns = Ns Sy 10 Pq uint
	Maximum number of queued TRIMs outstanding per leaf vdev.
	The number of concurrent TRIM commands issued to the device is controlled by
	.Sy zfs_vdev_trim_min_active No and Sy zfs_vdev_trim_max_active .
	.
	.It Sy zfs_trim_txg_batch Ns = Ns Sy 32 Pq uint
	The number of transaction groups' worth of frees which should be aggregated
	before TRIM operations are issued to the device.
	This setting represents a trade-off between issuing larger,
	more efficient TRIM operations and the delay
	before the recently trimmed space is available for use by the device.
	.Pp
	Increasing this value will allow frees to be aggregated for a longer time.
	This will result is larger TRIM operations and potentially increased memory usage.
	Decreasing this value will have the opposite effect.
	The default of
	.Sy 32
	was determined to be a reasonable compromise.
	.
	.It Sy zfs_txg_history Ns = Ns Sy 0 Pq int
	Historical statistics for this many latest TXGs will be available in
	.Pa /proc/spl/kstat/zfs/ Ns Ao Ar pool Ac Ns Pa /TXGs .
	.
	.It Sy zfs_txg_timeout Ns = Ns Sy 5 Ns s Pq int
	Flush dirty data to disk at least every this many seconds (maximum TXG duration).
	.
	.It Sy zfs_vdev_aggregate_trim Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Allow TRIM I/Os to be aggregated.
	This is normally not helpful because the extents to be trimmed
	will have been already been aggregated by the metaslab.
	This option is provided for debugging and performance analysis.
	.
	.It Sy zfs_vdev_aggregation_limit Ns = Ns Sy 1048576 Ns B Po 1MB Pc Pq int
	Max vdev I/O aggregation size.
	.
	.It Sy zfs_vdev_aggregation_limit_non_rotating Ns = Ns Sy 131072 Ns B Po 128kB Pc Pq int
	Max vdev I/O aggregation size for non-rotating media.
	.
	.It Sy zfs_vdev_cache_bshift Ns = Ns Sy 16 Po 64kB Pc Pq int
	Shift size to inflate reads to.
	.
	.It Sy zfs_vdev_cache_max Ns = Ns Sy 16384 Ns B Po 16kB Pc Pq int
	Inflate reads smaller than this value to meet the
	.Sy zfs_vdev_cache_bshift
	size
	.Pq default Sy 64kB .
	.
	.It Sy zfs_vdev_cache_size Ns = Ns Sy 0 Pq int
	Total size of the per-disk cache in bytes.
	.Pp
	Currently this feature is disabled, as it has been found to not be helpful
	for performance and in some cases harmful.
	.
	.It Sy zfs_vdev_mirror_rotating_inc Ns = Ns Sy 0 Pq int
	A number by which the balancing algorithm increments the load calculation for
	the purpose of selecting the least busy mirror member when an I/O operation
	immediately follows its predecessor on rotational vdevs
	for the purpose of making decisions based on load.
	.
	.It Sy zfs_vdev_mirror_rotating_seek_inc Ns = Ns Sy 5 Pq int
	A number by which the balancing algorithm increments the load calculation for
	the purpose of selecting the least busy mirror member when an I/O operation
	lacks locality as defined by
	.Sy zfs_vdev_mirror_rotating_seek_offset .
	Operations within this that are not immediately following the previous operation
	are incremented by half.
	.
	.It Sy zfs_vdev_mirror_rotating_seek_offset Ns = Ns Sy 1048576 Ns B Po 1MB Pc Pq int
	The maximum distance for the last queued I/O operation in which
	the balancing algorithm considers an operation to have locality.
	.No See Sx ZFS I/O SCHEDULER .
	.
	.It Sy zfs_vdev_mirror_non_rotating_inc Ns = Ns Sy 0 Pq int
	A number by which the balancing algorithm increments the load calculation for
	the purpose of selecting the least busy mirror member on non-rotational vdevs
	when I/O operations do not immediately follow one another.
	.
	.It Sy zfs_vdev_mirror_non_rotating_seek_inc Ns = Ns Sy 1 Pq int
	A number by which the balancing algorithm increments the load calculation for
	the purpose of selecting the least busy mirror member when an I/O operation lacks
	locality as defined by the
	.Sy zfs_vdev_mirror_rotating_seek_offset .
	Operations within this that are not immediately following the previous operation
	are incremented by half.
	.
	.It Sy zfs_vdev_read_gap_limit Ns = Ns Sy 32768 Ns B Po 32kB Pc Pq int
	Aggregate read I/O operations if the on-disk gap between them is within this
	threshold.
	.
	.It Sy zfs_vdev_write_gap_limit Ns = Ns Sy 4096 Ns B Po 4kB Pc Pq int
	Aggregate write I/O operations if the on-disk gap between them is within this
	threshold.
	.
	.It Sy zfs_vdev_raidz_impl Ns = Ns Sy fastest Pq string
	Select the raidz parity implementation to use.
	.Pp
	Variants that don't depend on CPU-specific features
	may be selected on module load, as they are supported on all systems.
	The remaining options may only be set after the module is loaded,
	as they are available only if the implementations are compiled in
	and supported on the running system.
	.Pp
	Once the module is loaded,
	.Pa /sys/module/zfs/parameters/zfs_vdev_raidz_impl
	will show the available options,
	with the currently selected one enclosed in square brackets.
	.Pp
	.TS
	lb l l .
	fastest selected by built-in benchmark
	original original implementation
	scalar scalar implementation
	sse2 SSE2 instruction set 64-bit x86
	ssse3 SSSE3 instruction set 64-bit x86
	avx2 AVX2 instruction set 64-bit x86
	avx512f AVX512F instruction set 64-bit x86
	avx512bw AVX512F & AVX512BW instruction sets 64-bit x86
	aarch64_neon NEON Aarch64/64-bit ARMv8
	aarch64_neonx2 NEON with more unrolling Aarch64/64-bit ARMv8
	powerpc_altivec Altivec PowerPC
	.TE
	.
	.It Sy zfs_vdev_scheduler Pq charp
	.Sy DEPRECATED .
	Prints warning to kernel log for compatibility.
	.
	.It Sy zfs_zevent_len_max Ns = Ns Sy 512 Pq int
	Max event queue length.
	Events in the queue can be viewed with
	.Xr zpool-events 8 .
	.
	.It Sy zfs_zevent_retain_max Ns = Ns Sy 2000 Pq int
	Maximum recent zevent records to retain for duplicate checking.
	Setting this to
	.Sy 0
	disables duplicate detection.
	.
	.It Sy zfs_zevent_retain_expire_secs Ns = Ns Sy 900 Ns s Po 15min Pc Pq int
	Lifespan for a recent ereport that was retained for duplicate checking.
	.
	.It Sy zfs_zil_clean_taskq_maxalloc Ns = Ns Sy 1048576 Pq int
	The maximum number of taskq entries that are allowed to be cached.
	When this limit is exceeded transaction records (itxs)
	will be cleaned synchronously.
	.
	.It Sy zfs_zil_clean_taskq_minalloc Ns = Ns Sy 1024 Pq int
	The number of taskq entries that are pre-populated when the taskq is first
	created and are immediately available for use.
	.
	.It Sy zfs_zil_clean_taskq_nthr_pct Ns = Ns Sy 100 Ns % Pq int
	This controls the number of threads used by
	.Sy dp_zil_clean_taskq .
	The default value of
	.Sy 100%
	will create a maximum of one thread per cpu.
	.
	.It Sy zil_maxblocksize Ns = Ns Sy 131072 Ns B Po 128kB Pc Pq int
	This sets the maximum block size used by the ZIL.
	On very fragmented pools, lowering this
	.Pq typically to Sy 36kB
	can improve performance.
	.
	.It Sy zil_nocacheflush Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Disable the cache flush commands that are normally sent to disk by
	the ZIL after an LWB write has completed.
	Setting this will cause ZIL corruption on power loss
	if a volatile out-of-order write cache is enabled.
	.
	.It Sy zil_replay_disable Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Disable intent logging replay.
	Can be disabled for recovery from corrupted ZIL.
	.
	.It Sy zil_slog_bulk Ns = Ns Sy 786432 Ns B Po 768kB Pc Pq ulong
	Limit SLOG write size per commit executed with synchronous priority.
	Any writes above that will be executed with lower (asynchronous) priority
	to limit potential SLOG device abuse by single active ZIL writer.
	.
	.It Sy zfs_embedded_slog_min_ms Ns = Ns Sy 64 Pq int
	Usually, one metaslab from each normal-class vdev is dedicated for use by
	the ZIL to log synchronous writes.
	However, if there are fewer than
	.Sy zfs_embedded_slog_min_ms
	metaslabs in the vdev, this functionality is disabled.
	This ensures that we don't set aside an unreasonable amount of space for the ZIL.
	.
	.It Sy zio_deadman_log_all Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	If non-zero, the zio deadman will produce debugging messages
	.Pq see Sy zfs_dbgmsg_enable
	for all zios, rather than only for leaf zios possessing a vdev.
	This is meant to be used by developers to gain
	diagnostic information for hang conditions which don't involve a mutex
	or other locking primitive: typically conditions in which a thread in
	the zio pipeline is looping indefinitely.
	.
	.It Sy zio_slow_io_ms Ns = Ns Sy 30000 Ns ms Po 30s Pc Pq int
	When an I/O operation takes more than this much time to complete,
	it's marked as slow.
	Each slow operation causes a delay zevent.
	Slow I/O counters can be seen with
	.Nm zpool Cm status Fl s .
	.
	.It Sy zio_dva_throttle_enabled Ns = Ns Sy 1 Ns \| Ns 0 Pq int
	Throttle block allocations in the I/O pipeline.
	This allows for dynamic allocation distribution when devices are imbalanced.
	When enabled, the maximum number of pending allocations per top-level vdev
	is limited by
	.Sy zfs_vdev_queue_depth_pct .
	.
	.It Sy zio_requeue_io_start_cut_in_line Ns = Ns Sy 0 Ns \| Ns 1 Pq int
	Prioritize requeued I/O.
	.
	.It Sy zio_taskq_batch_pct Ns = Ns Sy 80 Ns % Pq uint
	Percentage of online CPUs which will run a worker thread for I/O.
	These workers are responsible for I/O work such as compression and
	checksum calculations.
	Fractional number of CPUs will be rounded down.
	.Pp
	The default value of
	.Sy 80%
	was chosen to avoid using all CPUs which can result in
	latency issues and inconsistent application performance,
	especially when slower compression and/or checksumming is enabled.
	.
	.It Sy zio_taskq_batch_tpq Ns = Ns Sy 0 Pq uint
	Number of worker threads per taskq.
	Lower values improve I/O ordering and CPU utilization,
	while higher reduces lock contention.
	.Pp
	If
	.Sy 0 ,
	generate a system-dependent value close to 6 threads per taskq.
	.
	.It Sy zvol_inhibit_dev Ns = Ns Sy 0 Ns \| Ns 1 Pq uint
	Do not create zvol device nodes.
	This may slightly improve startup time on
	systems with a very large number of zvols.
	.
	.It Sy zvol_major Ns = Ns Sy 230 Pq uint
	Major number for zvol block devices.
	.
	.It Sy zvol_max_discard_blocks Ns = Ns Sy 16384 Pq ulong
	Discard (TRIM) operations done on zvols will be done in batches of this
	many blocks, where block size is determined by the
	.Sy volblocksize
	property of a zvol.
	.
	.It Sy zvol_prefetch_bytes Ns = Ns Sy 131072 Ns B Po 128kB Pc Pq uint
	When adding a zvol to the system, prefetch this many bytes
	from the start and end of the volume.
	Prefetching these regions of the volume is desirable,
	because they are likely to be accessed immediately by
	.Xr blkid 8
	or the kernel partitioner.
	.
	.It Sy zvol_request_sync Ns = Ns Sy 0 Ns \| Ns 1 Pq uint
	When processing I/O requests for a zvol, submit them synchronously.
	This effectively limits the queue depth to
	.Em 1
	for each I/O submitter.
	When unset, requests are handled asynchronously by a thread pool.
	The number of requests which can be handled concurrently is controlled by
	.Sy zvol_threads .
	.
	.It Sy zvol_threads Ns = Ns Sy 32 Pq uint
	Max number of threads which can handle zvol I/O requests concurrently.
	.
	.It Sy zvol_volmode Ns = Ns Sy 1 Pq uint
	Defines zvol block devices behaviour when
	.Sy volmode Ns = Ns Sy default :
	.Bl -tag -compact -offset 4n -width "a"
	.It Sy 1
	.No equivalent to Sy full
	.It Sy 2
	.No equivalent to Sy dev
	.It Sy 3
	.No equivalent to Sy none
	.El
	.El
	.
	.Sh ZFS I/O SCHEDULER
	ZFS issues I/O operations to leaf vdevs to satisfy and complete I/O operations.
	The scheduler determines when and in what order those operations are issued.
	The scheduler divides operations into five I/O classes,
	prioritized in the following order: sync read, sync write, async read,
	async write, and scrub/resilver.
	Each queue defines the minimum and maximum number of concurrent operations
	that may be issued to the device.
	In addition, the device has an aggregate maximum,
	.Sy zfs_vdev_max_active .
	Note that the sum of the per-queue minima must not exceed the aggregate maximum.
	If the sum of the per-queue maxima exceeds the aggregate maximum,
	then the number of active operations may reach
	.Sy zfs_vdev_max_active ,
	in which case no further operations will be issued,
	regardless of whether all per-queue minima have been met.
	.Pp
	For many physical devices, throughput increases with the number of
	concurrent operations, but latency typically suffers.
	Furthermore, physical devices typically have a limit
	at which more concurrent operations have no
	effect on throughput or can actually cause it to decrease.
	.Pp
	The scheduler selects the next operation to issue by first looking for an
	I/O class whose minimum has not been satisfied.
	Once all are satisfied and the aggregate maximum has not been hit,
	the scheduler looks for classes whose maximum has not been satisfied.
	Iteration through the I/O classes is done in the order specified above.
	No further operations are issued
	if the aggregate maximum number of concurrent operations has been hit,
	or if there are no operations queued for an I/O class that has not hit its maximum.
	Every time an I/O operation is queued or an operation completes,
	the scheduler looks for new operations to issue.
	.Pp
	In general, smaller
	.Sy max_active Ns s
	will lead to lower latency of synchronous operations.
	Larger
	.Sy max_active Ns s
	may lead to higher overall throughput, depending on underlying storage.
	.Pp
	The ratio of the queues'
	.Sy max_active Ns s
	determines the balance of performance between reads, writes, and scrubs.
	For example, increasing
	.Sy zfs_vdev_scrub_max_active
	will cause the scrub or resilver to complete more quickly,
	but reads and writes to have higher latency and lower throughput.
	.Pp
	All I/O classes have a fixed maximum number of outstanding operations,
	except for the async write class.
	Asynchronous writes represent the data that is committed to stable storage
	during the syncing stage for transaction groups.
	Transaction groups enter the syncing state periodically,
	so the number of queued async writes will quickly burst up
	and then bleed down to zero.
	Rather than servicing them as quickly as possible,
	the I/O scheduler changes the maximum number of active async write operations
	according to the amount of dirty data in the pool.
	Since both throughput and latency typically increase with the number of
	concurrent operations issued to physical devices, reducing the
	burstiness in the number of concurrent operations also stabilizes the
	response time of operations from other – and in particular synchronous – queues.
	In broad strokes, the I/O scheduler will issue more concurrent operations
	from the async write queue as there's more dirty data in the pool.
	.
	.Ss Async Writes
	The number of concurrent operations issued for the async write I/O class
	follows a piece-wise linear function defined by a few adjustable points:
	.Bd -literal
	\| o---------\| <-- \fBzfs_vdev_async_write_max_active\fP
	^ \| /^ \|
	\| \| / \| \|
	active \| / \| \|
	I/O \| / \| \|
	count \| / \| \|
	\| / \| \|
	\|-------o \| \| <-- \fBzfs_vdev_async_write_min_active\fP
	0\|_______^______\|_________\|
	0% \| \| 100% of \fBzfs_dirty_data_max\fP
	\| \|
	\| `-- \fBzfs_vdev_async_write_active_max_dirty_percent\fP
	`--------- \fBzfs_vdev_async_write_active_min_dirty_percent\fP
	.Ed
	.Pp
	Until the amount of dirty data exceeds a minimum percentage of the dirty
	data allowed in the pool, the I/O scheduler will limit the number of
	concurrent operations to the minimum.
	As that threshold is crossed, the number of concurrent operations issued
	increases linearly to the maximum at the specified maximum percentage
	of the dirty data allowed in the pool.
	.Pp
	Ideally, the amount of dirty data on a busy pool will stay in the sloped
	part of the function between
	.Sy zfs_vdev_async_write_active_min_dirty_percent
	and
	.Sy zfs_vdev_async_write_active_max_dirty_percent .
	If it exceeds the maximum percentage,
	this indicates that the rate of incoming data is
	greater than the rate that the backend storage can handle.
	In this case, we must further throttle incoming writes,
	as described in the next section.
	.
	.Sh ZFS TRANSACTION DELAY
	We delay transactions when we've determined that the backend storage
	isn't able to accommodate the rate of incoming writes.
	.Pp
	If there is already a transaction waiting, we delay relative to when
	that transaction will finish waiting.
	This way the calculated delay time
	is independent of the number of threads concurrently executing transactions.
	.Pp
	If we are the only waiter, wait relative to when the transaction started,
	rather than the current time.
	This credits the transaction for "time already served",
	e.g. reading indirect blocks.
	.Pp
	The minimum time for a transaction to take is calculated as
	.Dl min_time = min( Ns Sy zfs_delay_scale No * (dirty - min) / (max - dirty), 100ms)
	.Pp
	The delay has two degrees of freedom that can be adjusted via tunables.
	The percentage of dirty data at which we start to delay is defined by
	.Sy zfs_delay_min_dirty_percent .
	This should typically be at or above
	.Sy zfs_vdev_async_write_active_max_dirty_percent ,
	so that we only start to delay after writing at full speed
	has failed to keep up with the incoming write rate.
	The scale of the curve is defined by
	.Sy zfs_delay_scale .
	Roughly speaking, this variable determines the amount of delay at the midpoint of the curve.
	.Bd -literal
	delay
	10ms +-------------------------------------------------------------*+
	\| *\|
	9ms + *+
	\| *\|
	8ms + *+
	\| * \|
	7ms + * +
	\| * \|
	6ms + * +
	\| * \|
	5ms + * +
	\| * \|
	4ms + * +
	\| * \|
	3ms + * +
	\| * \|
	2ms + (midpoint) * +
	\| \| ** \|
	1ms + v *** +
	\| \fBzfs_delay_scale\fP ----------> ******** \|
	0 +-------------------------------------*********----------------+
	0% <- \fBzfs_dirty_data_max\fP -> 100%
	.Ed
	.Pp
	Note, that since the delay is added to the outstanding time remaining on the
	most recent transaction it's effectively the inverse of IOPS.
	Here, the midpoint of
	.Em 500us
	translates to
	.Em 2000 IOPS .
	The shape of the curve
	was chosen such that small changes in the amount of accumulated dirty data
	in the first three quarters of the curve yield relatively small differences
	in the amount of delay.
	.Pp
	The effects can be easier to understand when the amount of delay is
	represented on a logarithmic scale:
	.Bd -literal
	delay
	100ms +-------------------------------------------------------------++
	+ +
	\| \|
	+ *+
	10ms + *+
	+ ** +
	\| (midpoint) ** \|
	+ \| ** +
	1ms + v **** +
	+ \fBzfs_delay_scale\fP ----------> ***** +
	\| **** \|
	+ **** +
	100us + ** +
	+ * +
	\| * \|
	+ * +
	10us + * +
	+ +
	\| \|
	+ +
	+--------------------------------------------------------------+
	0% <- \fBzfs_dirty_data_max\fP -> 100%
	.Ed
	.Pp
	Note here that only as the amount of dirty data approaches its limit does
	the delay start to increase rapidly.
	The goal of a properly tuned system should be to keep the amount of dirty data
	out of that range by first ensuring that the appropriate limits are set
	for the I/O scheduler to reach optimal throughput on the back-end storage,
	and then by changing the value of
	.Sy zfs_delay_scale
	to increase the steepness of the curve.
	diff --git a/sys/contrib/openzfs/man/man7/vdevprops.7 b/sys/contrib/openzfs/man/man7/vdevprops.7
	new file mode 100644
	index 000000000000..ec7b52955e75
	--- /dev/null
	+++ b/sys/contrib/openzfs/man/man7/vdevprops.7
	@@ -0,0 +1,172 @@
	+.\"
	+.\" CDDL HEADER START
	+.\"
	+.\" The contents of this file are subject to the terms of the
	+.\" Common Development and Distribution License (the "License").
	+.\" You may not use this file except in compliance with the License.
	+.\"
	+.\" You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	+.\" or http://www.opensolaris.org/os/licensing.
	+.\" See the License for the specific language governing permissions
	+.\" and limitations under the License.
	+.\"
	+.\" When distributing Covered Code, include this CDDL HEADER in each
	+.\" file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	+.\" If applicable, add the following below this CDDL HEADER, with the
	+.\" fields enclosed by brackets "[]" replaced with your own identifying
	+.\" information: Portions Copyright [yyyy] [name of copyright owner]
	+.\"
	+.\" CDDL HEADER END
	+.\"
	+.\" Copyright (c) 2021 Klara, Inc.
	+.\"
	+.Dd November 27, 2021
	+.Dt VDEVPROPS 7
	+.Os
	+.
	+.Sh NAME
	+.Nm vdevprops
	+.Nd native and user-defined properties of ZFS vdevs
	+.
	+.Sh DESCRIPTION
	+Properties are divided into two types, native properties and user-defined
	+.Pq or Qq user
	+properties.
	+Native properties either export internal statistics or control ZFS behavior.
	+In addition, native properties are either editable or read-only.
	+User properties have no effect on ZFS behavior, but you can use them to annotate
	+vdevs in a way that is meaningful in your environment.
	+For more information about user properties, see the
	+.Sx User Properties
	+section, below.
	+.
	+.Ss Native Properties
	+Every vdev has a set of properties that export statistics about the vdev
	+as well as control various behaviors.
	+Properties are NOT inherited from top-level vdevs.
	+.Pp
	+The values of numeric properties can be specified using human-readable suffixes
	+.Po for example,
	+.Sy k , KB , M , Gb ,
	+and so forth, up to
	+.Sy Z
	+for zettabyte
	+.Pc .
	+The following are all valid
	+.Pq and equal
	+specifications:
	+.Li 1536M , 1.5g , 1.50GB .
	+.Pp
	+The values of non-numeric properties are case sensitive and must be lowercase.
	+.Pp
	+The following native properties consist of read-only statistics about the
	+vdev.
	+These properties can not be changed.
	+.Bl -tag -width "fragmentation"
	+.It Sy capacity
	+Percentage of vdev space used
	+.It Sy state
	+state of this vdev such as online, faulted, or offline
	+.It Sy guid
	+globaly unique id of this vdev
	+.It Sy asize
	+The allocable size of this vdev
	+.It Sy psize
	+The physical size of this vdev
	+.It Sy ashift
	+The physical sector size of this vdev expressed as the power of two
	+.It Sy size
	+The total size of this vdev
	+.It Sy free
	+The amount of remaining free space on this vdev
	+.It Sy allocated
	+The amount of allocated space on this vdev
	+.It Sy expandsize
	+How much this vdev can expand by
	+.It Sy fragmentation
	+Percent of fragmentation in this vdev
	+.It Sy parity
	+The level of parity for this vdev
	+.It Sy devid
	+The device id for this vdev
	+.It Sy physpath
	+The physical path to the device
	+.It Sy encpath
	+The enclosure path to the device
	+.It Sy fru
	+Field Replacable Unit, usually a model number
	+.It Sy parent
	+Parent of this vdev
	+.It Sy children
	+Comma separated list of children of this vdev
	+.It Sy numchildren
	+The number of children belonging to this vdev
	+.It Sy read_errors , write_errors , checksum_errors , initialize_errors
	+The number of errors of each type encountered by this vdev
	+.It Sy null_ops , read_ops , write_ops , free_ops , claim_ops , trim_ops
	+The number of I/O operations of each type performed by this vdev
	+.It Xo
	+.Sy null_bytes , read_bytes , write_bytes , free_bytes , claim_bytes ,
	+.Sy trim_bytes
	+.Xc
	+The cumulative size of all operations of each type performed by this vdev
	+.It Sy removing
	+If this device is currently being removed from the pool
	+.El
	+.Pp
	+The following native properties can be used to change the behavior of a ZFS
	+dataset.
	+.Bl -tag -width "allocating"
	+.It Sy comment
	+A text comment up to 8192 characters long
	+.It Sy bootsize
	+The amount of space to reserve for the EFI system partition
	+.It Sy path
	+The path to the device for this vdev
	+.It Sy allocating
	+If this device should perform new allocations, used to disable a device
	+when it is scheduled for later removal.
	+See
	+.Xr zpool-remove 8 .
	+.El
	+.Ss User Properties
	+In addition to the standard native properties, ZFS supports arbitrary user
	+properties.
	+User properties have no effect on ZFS behavior, but applications or
	+administrators can use them to annotate vdevs.
	+.Pp
	+User property names must contain a colon
	+.Pq Qq Sy \&:
	+character to distinguish them from native properties.
	+They may contain lowercase letters, numbers, and the following punctuation
	+characters: colon
	+.Pq Qq Sy \&: ,
	+dash
	+.Pq Qq Sy - ,
	+period
	+.Pq Qq Sy \&. ,
	+and underscore
	+.Pq Qq Sy _ .
	+The expected convention is that the property name is divided into two portions
	+such as
	+.Ar module : Ns Ar property ,
	+but this namespace is not enforced by ZFS.
	+User property names can be at most 256 characters, and cannot begin with a dash
	+.Pq Qq Sy - .
	+.Pp
	+When making programmatic use of user properties, it is strongly suggested to use
	+a reversed DNS domain name for the
	+.Ar module
	+component of property names to reduce the chance that two
	+independently-developed packages use the same property name for different
	+purposes.
	+.Pp
	+The values of user properties are arbitrary strings and
	+are never validated.
	+Use the
	+.Nm zpool Cm set
	+command with a blank value to clear a user property.
	+Property values are limited to 8192 bytes.
	+.Sh SEE ALSO
	+.Xr zpoolprops 7 ,
	+.Xr zpool-set 8
	diff --git a/sys/contrib/openzfs/man/man8/zdb.8 b/sys/contrib/openzfs/man/man8/zdb.8
	index a8a944219071..2b0471cc9ea4 100644
	--- a/sys/contrib/openzfs/man/man8/zdb.8
	+++ b/sys/contrib/openzfs/man/man8/zdb.8
	@@ -1,506 +1,504 @@
	.\"
	.\" This file and its contents are supplied under the terms of the
	.\" Common Development and Distribution License ("CDDL"), version 1.0.
	.\" You may only use this file in accordance with the terms of version
	.\" 1.0 of the CDDL.
	.\"
	.\" A full copy of the text of the CDDL should have accompanied this
	.\" source. A copy of the CDDL is also available via the Internet at
	.\" http://www.illumos.org/license/CDDL.
	.\"
	.\" Copyright 2012, Richard Lowe.
	.\" Copyright (c) 2012, 2019 by Delphix. All rights reserved.
	.\" Copyright 2017 Nexenta Systems, Inc.
	.\" Copyright (c) 2017 Lawrence Livermore National Security, LLC.
	.\" Copyright (c) 2017 Intel Corporation.
	.\"
	.Dd October 7, 2020
	.Dt ZDB 8
	.Os
	.
	.Sh NAME
	.Nm zdb
	.Nd display ZFS storage pool debugging and consistency information
	.Sh SYNOPSIS
	.Nm
	.Op Fl AbcdDFGhikLMPsvXYy
	.Op Fl e Oo Fl V Oc Oo Fl p Ar path Oc Ns …
	.Op Fl I Ar inflight I/Os
	.Oo Fl o Ar var Ns = Ns Ar value Oc Ns …
	.Op Fl t Ar txg
	.Op Fl U Ar cache
	.Op Fl x Ar dumpdir
	.Op Ar poolname Ns Op / Ns Ar dataset \| objset ID
	.Op Ar object Ns \| Ns Ar range Ns …
	.Nm
	.Op Fl AdiPv
	.Op Fl e Oo Fl V Oc Oo Fl p Ar path Oc Ns …
	.Op Fl U Ar cache
	.Ar poolname Ns Op Ar / Ns Ar dataset \| objset ID
	.Op Ar object Ns \| Ns Ar range Ns …
	.Nm
	.Fl C
	.Op Fl A
	.Op Fl U Ar cache
	.Nm
	.Fl E
	.Op Fl A
	.Ar word0 : Ns Ar word1 Ns :…: Ns Ar word15
	.Nm
	.Fl l
	.Op Fl Aqu
	.Ar device
	.Nm
	.Fl m
	.Op Fl AFLPXY
	.Op Fl e Oo Fl V Oc Oo Fl p Ar path Oc Ns …
	.Op Fl t Ar txg
	.Op Fl U Ar cache
	.Ar poolname Op Ar vdev Oo Ar metaslab Oc Ns …
	.Nm
	.Fl O
	.Ar dataset path
	.Nm
	.Fl r
	.Ar dataset path destination
	.Nm
	.Fl R
	.Op Fl A
	.Op Fl e Oo Fl V Oc Oo Fl p Ar path Oc Ns …
	.Op Fl U Ar cache
	.Ar poolname vdev : Ns Ar offset : Ns Oo Ar lsize Ns / Oc Ns Ar psize Ns Op : Ns Ar flags
	.Nm
	.Fl S
	.Op Fl AP
	.Op Fl e Oo Fl V Oc Oo Fl p Ar path Oc Ns …
	.Op Fl U Ar cache
	.Ar poolname
	.
	.Sh DESCRIPTION
	The
	.Nm
	utility displays information about a ZFS pool useful for debugging and performs
	some amount of consistency checking.
	It is a not a general purpose tool and options
	.Pq and facilities
	may change.
	This is not a
	.Xr fsck 8
	utility.
	.Pp
	The output of this command in general reflects the on-disk structure of a ZFS
	pool, and is inherently unstable.
	The precise output of most invocations is not documented, a knowledge of ZFS
	internals is assumed.
	.Pp
	If the
	.Ar dataset
	argument does not contain any
	.Qq Sy /
	or
	.Qq Sy @
	characters, it is interpreted as a pool name.
	The root dataset can be specified as
	.Qq Ar pool Ns / .
	.Pp
	When operating on an imported and active pool it is possible, though unlikely,
	that zdb may interpret inconsistent pool data and behave erratically.
	.
	.Sh OPTIONS
	Display options:
	.Bl -tag -width Ds
	.It Fl b
	Display statistics regarding the number, size
	.Pq logical, physical and allocated
	and deduplication of blocks.
	.It Fl c
	Verify the checksum of all metadata blocks while printing block statistics
	.Po see
	.Fl b
	.Pc .
	.Pp
	If specified multiple times, verify the checksums of all blocks.
	.It Fl C
	Display information about the configuration.
	If specified with no other options, instead display information about the cache
	file
	.Pq Pa /etc/zfs/zpool.cache .
	To specify the cache file to display, see
	.Fl U .
	.Pp
	If specified multiple times, and a pool name is also specified display both the
	cached configuration and the on-disk configuration.
	If specified multiple times with
	.Fl e
	also display the configuration that would be used were the pool to be imported.
	.It Fl d
	Display information about datasets.
	Specified once, displays basic dataset information: ID, create transaction,
	size, and object count.
	.Pp
	If specified multiple times provides greater and greater verbosity.
	.Pp
	If object IDs or object ID ranges are specified, display information about
	those specific objects or ranges only.
	.Pp
	An object ID range is specified in terms of a colon-separated tuple of
	the form
	.Ao start Ac : Ns Ao end Ac Ns Op : Ns Ao flags Ac .
	The fields
	.Ar start
	and
	.Ar end
	are integer object identifiers that denote the upper and lower bounds
	of the range.
	An
	.Ar end
	value of -1 specifies a range with no upper bound.
	The
	.Ar flags
	field optionally specifies a set of flags, described below, that control
	which object types are dumped.
	By default, all object types are dumped.
	A minus sign
	.Pq -
	negates the effect of the flag that follows it and has no effect unless
	preceded by the
	.Ar A
	flag.
	For example, the range 0:-1:A-d will dump all object types except for directories.
	.Pp
	.Bl -tag -compact -width Ds
	.It Sy A
	Dump all objects (this is the default)
	.It Sy d
	Dump ZFS directory objects
	.It Sy f
	Dump ZFS plain file objects
	.It Sy m
	Dump SPA space map objects
	.It Sy z
	Dump ZAP objects
	.It Sy -
	Negate the effect of next flag
	.El
	.It Fl D
	Display deduplication statistics, including the deduplication ratio
	.Pq Sy dedup ,
	compression ratio
	.Pq Sy compress ,
	inflation due to the zfs copies property
	.Pq Sy copies ,
	and an overall effective ratio
	.Pq Sy dedup No * Sy compress No / Sy copies .
	.It Fl DD
	Display a histogram of deduplication statistics, showing the allocated
	.Pq physically present on disk
	and referenced
	.Pq logically referenced in the pool
	block counts and sizes by reference count.
	.It Fl DDD
	Display the statistics independently for each deduplication table.
	.It Fl DDDD
	Dump the contents of the deduplication tables describing duplicate blocks.
	.It Fl DDDDD
	Also dump the contents of the deduplication tables describing unique blocks.
	.It Fl E Ar word0 : Ns Ar word1 Ns :…: Ns Ar word15
	Decode and display block from an embedded block pointer specified by the
	.Ar word
	arguments.
	.It Fl h
	Display pool history similar to
	.Nm zpool Cm history ,
	but include internal changes, transaction, and dataset information.
	.It Fl i
	Display information about intent log
	.Pq ZIL
	entries relating to each dataset.
	If specified multiple times, display counts of each intent log transaction type.
	.It Fl k
	Examine the checkpointed state of the pool.
	Note, the on disk format of the pool is not reverted to the checkpointed state.
	.It Fl l Ar device
	Read the vdev labels and L2ARC header from the specified device.
	.Nm Fl l
	will return 0 if valid label was found, 1 if error occurred, and 2 if no valid
	labels were found.
	The presence of L2ARC header is indicated by a specific
	sequence (L2ARC_DEV_HDR_MAGIC).
	If there is an accounting error in the size or the number of L2ARC log blocks
	.Nm Fl l
	will return 1.
	Each unique configuration is displayed only once.
	.It Fl ll Ar device
	In addition display label space usage stats.
	If a valid L2ARC header was found
	also display the properties of log blocks used for restoring L2ARC contents
	(persistent L2ARC).
	.It Fl lll Ar device
	Display every configuration, unique or not.
	If a valid L2ARC header was found
	also display the properties of log entries in log blocks used for restoring
	L2ARC contents (persistent L2ARC).
	.Pp
	If the
	.Fl q
	option is also specified, don't print the labels or the L2ARC header.
	.Pp
	If the
	.Fl u
	option is also specified, also display the uberblocks on this device.
	Specify multiple times to increase verbosity.
	.It Fl L
	Disable leak detection and the loading of space maps.
	By default,
	.Nm
	verifies that all non-free blocks are referenced, which can be very expensive.
	.It Fl m
	Display the offset, spacemap, free space of each metaslab, all the log
	spacemaps and their obsolete entry statistics.
	.It Fl mm
	Also display information about the on-disk free space histogram associated with
	each metaslab.
	.It Fl mmm
	Display the maximum contiguous free space, the in-core free space histogram, and
	the percentage of free space in each space map.
	.It Fl mmmm
	Display every spacemap record.
	.It Fl M
	-Display the offset, spacemap, and free space of each metaslab.
	+Display all "normal" vdev metaslab group information - per-vdev metaslab count, fragmentation,
	+and free space histogram, as well as overall pool fragmentation and histogram.
	.It Fl MM
	-Also display information about the maximum contiguous free space and the
	-percentage of free space in each space map.
	-.It Fl MMM
	-Display every spacemap record.
	+"Special" vdevs are added to -M's normal output.
	.It Fl O Ar dataset path
	Look up the specified
	.Ar path
	inside of the
	.Ar dataset
	and display its metadata and indirect blocks.
	Specified
	.Ar path
	must be relative to the root of
	.Ar dataset .
	This option can be combined with
	.Fl v
	for increasing verbosity.
	.It Fl r Ar dataset path destination
	Copy the specified
	.Ar path
	inside of the
	.Ar dataset
	to the specified destination.
	Specified
	.Ar path
	must be relative to the root of
	.Ar dataset .
	This option can be combined with
	.Fl v
	for increasing verbosity.
	.It Xo
	.Fl R Ar poolname vdev : Ns Ar offset : Ns Oo Ar lsize Ns / Oc Ns Ar psize Ns Op : Ns Ar flags
	.Xc
	Read and display a block from the specified device.
	By default the block is displayed as a hex dump, but see the description of the
	.Sy r
	flag, below.
	.Pp
	The block is specified in terms of a colon-separated tuple
	.Ar vdev
	.Pq an integer vdev identifier
	.Ar offset
	.Pq the offset within the vdev
	.Ar size
	.Pq the physical size, or logical size / physical size
	of the block to read and, optionally,
	.Ar flags
	.Pq a set of flags, described below .
	.Pp
	.Bl -tag -compact -width "b offset"
	.It Sy b Ar offset
	Print block pointer at hex offset
	.It Sy c
	Calculate and display checksums
	.It Sy d
	Decompress the block.
	Set environment variable
	.Nm ZDB_NO_ZLE
	to skip zle when guessing.
	.It Sy e
	Byte swap the block
	.It Sy g
	Dump gang block header
	.It Sy i
	Dump indirect block
	.It Sy r
	Dump raw uninterpreted block data
	.It Sy v
	Verbose output for guessing compression algorithm
	.El
	.It Fl s
	Report statistics on
	.Nm zdb
	I/O.
	Display operation counts, bandwidth, and error counts of I/O to the pool from
	.Nm .
	.It Fl S
	Simulate the effects of deduplication, constructing a DDT and then display
	that DDT as with
	.Fl DD .
	.It Fl u
	Display the current uberblock.
	.El
	.Pp
	Other options:
	.Bl -tag -width Ds
	.It Fl A
	Do not abort should any assertion fail.
	.It Fl AA
	Enable panic recovery, certain errors which would otherwise be fatal are
	demoted to warnings.
	.It Fl AAA
	Do not abort if asserts fail and also enable panic recovery.
	.It Fl e Oo Fl p Ar path Oc Ns …
	Operate on an exported pool, not present in
	.Pa /etc/zfs/zpool.cache .
	The
	.Fl p
	flag specifies the path under which devices are to be searched.
	.It Fl x Ar dumpdir
	All blocks accessed will be copied to files in the specified directory.
	The blocks will be placed in sparse files whose name is the same as
	that of the file or device read.
	.Nm
	can be then run on the generated files.
	Note that the
	.Fl bbc
	flags are sufficient to access
	.Pq and thus copy
	all metadata on the pool.
	.It Fl F
	Attempt to make an unreadable pool readable by trying progressively older
	transactions.
	.It Fl G
	Dump the contents of the zfs_dbgmsg buffer before exiting
	.Nm .
	zfs_dbgmsg is a buffer used by ZFS to dump advanced debug information.
	.It Fl I Ar inflight I/Os
	Limit the number of outstanding checksum I/Os to the specified value.
	The default value is 200.
	This option affects the performance of the
	.Fl c
	option.
	.It Fl o Ar var Ns = Ns Ar value …
	Set the given global libzpool variable to the provided value.
	The value must be an unsigned 32-bit integer.
	Currently only little-endian systems are supported to avoid accidentally setting
	the high 32 bits of 64-bit variables.
	.It Fl P
	Print numbers in an unscaled form more amenable to parsing, e.g.\&
	.Sy 1000000
	rather than
	.Sy 1M .
	.It Fl t Ar transaction
	Specify the highest transaction to use when searching for uberblocks.
	See also the
	.Fl u
	and
	.Fl l
	options for a means to see the available uberblocks and their associated
	transaction numbers.
	.It Fl U Ar cachefile
	Use a cache file other than
	.Pa /etc/zfs/zpool.cache .
	.It Fl v
	Enable verbosity.
	Specify multiple times for increased verbosity.
	.It Fl V
	Attempt verbatim import.
	This mimics the behavior of the kernel when loading a pool from a cachefile.
	Only usable with
	.Fl e .
	.It Fl X
	Attempt
	.Qq extreme
	transaction rewind, that is attempt the same recovery as
	.Fl F
	but read transactions otherwise deemed too old.
	.It Fl Y
	Attempt all possible combinations when reconstructing indirect split blocks.
	This flag disables the individual I/O deadman timer in order to allow as
	much time as required for the attempted reconstruction.
	.It Fl y
	Perform validation for livelists that are being deleted.
	Scans through the livelist and metaslabs, checking for duplicate entries
	and compares the two, checking for potential double frees.
	If it encounters issues, warnings will be printed, but the command will not
	necessarily fail.
	.El
	.Pp
	Specifying a display option more than once enables verbosity for only that
	option, with more occurrences enabling more verbosity.
	.Pp
	If no options are specified, all information about the named pool will be
	displayed at default verbosity.
	.
	.Sh EXAMPLES
	.Bl -tag -width Ds
	.It Xo
	.Sy Example 1 :
	Display the configuration of imported pool
	.Ar rpool
	.Xc
	.Bd -literal
	.No # Nm zdb Fl C Ar rpool
	MOS Configuration:
	version: 28
	name: 'rpool'
	…
	.Ed
	.It Xo
	.Sy Example 2 :
	Display basic dataset information about
	.Ar rpool
	.Xc
	.Bd -literal
	.No # Nm zdb Fl d Ar rpool
	Dataset mos [META], ID 0, cr_txg 4, 26.9M, 1051 objects
	Dataset rpool/swap [ZVOL], ID 59, cr_txg 356, 486M, 2 objects
	…
	.Ed
	.It Xo
	.Sy Example 3 :
	Display basic information about object 0 in
	.Ar rpool/export/home
	.Xc
	.Bd -literal
	.No # Nm zdb Fl d Ar rpool/export/home 0
	Dataset rpool/export/home [ZPL], ID 137, cr_txg 1546, 32K, 8 objects

	Object lvl iblk dblk dsize lsize %full type
	0 7 16K 16K 15.0K 16K 25.00 DMU dnode
	.Ed
	.It Xo
	.Sy Example 4 :
	Display the predicted effect of enabling deduplication on
	.Ar rpool
	.Xc
	.Bd -literal
	.No # Nm zdb Fl S Ar rpool
	Simulated DDT histogram:

	bucket allocated referenced
	______ ______________________________ ______________________________
	refcnt blocks LSIZE PSIZE DSIZE blocks LSIZE PSIZE DSIZE
	------ ------ ----- ----- ----- ------ ----- ----- -----
	1 694K 27.1G 15.0G 15.0G 694K 27.1G 15.0G 15.0G
	2 35.0K 1.33G 699M 699M 74.7K 2.79G 1.45G 1.45G
	…
	dedup = 1.11, compress = 1.80, copies = 1.00, dedup * compress / copies = 2.00
	.Ed
	.El
	.
	.Sh SEE ALSO
	.Xr zfs 8 ,
	.Xr zpool 8
	diff --git a/sys/contrib/openzfs/man/man8/zfs-diff.8 b/sys/contrib/openzfs/man/man8/zfs-diff.8
	index 49443bf47d17..a347f325203e 100644
	--- a/sys/contrib/openzfs/man/man8/zfs-diff.8
	+++ b/sys/contrib/openzfs/man/man8/zfs-diff.8
	@@ -1,98 +1,102 @@
	.\"
	.\" CDDL HEADER START
	.\"
	.\" The contents of this file are subject to the terms of the
	.\" Common Development and Distribution License (the "License").
	.\" You may not use this file except in compliance with the License.
	.\"
	.\" You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	.\" or http://www.opensolaris.org/os/licensing.
	.\" See the License for the specific language governing permissions
	.\" and limitations under the License.
	.\"
	.\" When distributing Covered Code, include this CDDL HEADER in each
	.\" file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	.\" If applicable, add the following below this CDDL HEADER, with the
	.\" fields enclosed by brackets "[]" replaced with your own identifying
	.\" information: Portions Copyright [yyyy] [name of copyright owner]
	.\"
	.\" CDDL HEADER END
	.\"
	.\" Copyright (c) 2009 Sun Microsystems, Inc. All Rights Reserved.
	.\" Copyright 2011 Joshua M. Clulow <josh@sysmgr.org>
	.\" Copyright (c) 2011, 2019 by Delphix. All rights reserved.
	.\" Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
	.\" Copyright (c) 2014, Joyent, Inc. All rights reserved.
	.\" Copyright (c) 2014 by Adam Stevko. All rights reserved.
	.\" Copyright (c) 2014 Integros [integros.com]
	.\" Copyright 2019 Richard Laager. All rights reserved.
	.\" Copyright 2018 Nexenta Systems, Inc.
	.\" Copyright 2019 Joyent, Inc.
	.\"
	.Dd May 29, 2021
	.Dt ZFS-DIFF 8
	.Os
	.
	.Sh NAME
	.Nm zfs-diff
	.Nd show difference between ZFS snapshots
	.Sh SYNOPSIS
	.Nm zfs
	.Cm diff
	-.Op Fl FHt
	+.Op Fl FHth
	.Ar snapshot Ar snapshot Ns \| Ns Ar filesystem
	.
	.Sh DESCRIPTION
	Display the difference between a snapshot of a given filesystem and another
	snapshot of that filesystem from a later time or the current contents of the
	filesystem.
	The first column is a character indicating the type of change, the other columns
	indicate pathname, new pathname
	.Pq in case of rename ,
	change in link count, and optionally file type and/or change time.
	The types of change are:
	.Bl -tag -compact -offset Ds -width "M"
	.It Sy -
	The path has been removed
	.It Sy +
	The path has been created
	.It Sy M
	The path has been modified
	.It Sy R
	The path has been renamed
	.El
	.Bl -tag -width "-F"
	.It Fl F
	Display an indication of the type of file, in a manner similar to the
	.Fl F
	option of
	.Xr ls 1 .
	.Bl -tag -compact -offset 2n -width "B"
	.It Sy B
	Block device
	.It Sy C
	Character device
	.It Sy /
	Directory
	.It Sy >
	Door
	.It Sy \|\&
	Named pipe
	.It Sy @
	Symbolic link
	.It Sy P
	Event port
	.It Sy =
	Socket
	.It Sy F
	Regular file
	.El
	.It Fl H
	Give more parsable tab-separated output, without header lines and without
	arrows.
	.It Fl t
	Display the path's inode change time as the first column of output.
	+.It Fl h
	+Do not
	+.Sy \e0 Ns Ar ooo Ns -escape
	+non-ASCII paths.
	.El
	.
	.Sh SEE ALSO
	.Xr zfs-snapshot 8
	diff --git a/sys/contrib/openzfs/man/man8/zfs-mount-generator.8.in b/sys/contrib/openzfs/man/man8/zfs-mount-generator.8.in
	index 7aa332ba8174..ae8937038e67 100644
	--- a/sys/contrib/openzfs/man/man8/zfs-mount-generator.8.in
	+++ b/sys/contrib/openzfs/man/man8/zfs-mount-generator.8.in
	@@ -1,192 +1,181 @@
	.\"
	.\" Copyright 2018 Antonio Russo <antonio.e.russo@gmail.com>
	.\" Copyright 2019 Kjeld Schouten-Lebbing <kjeld@schouten-lebbing.nl>
	.\" Copyright 2020 InsanePrawn <insane.prawny@gmail.com>
	.\"
	.\" Permission is hereby granted, free of charge, to any person obtaining
	.\" a copy of this software and associated documentation files (the
	.\" "Software"), to deal in the Software without restriction, including
	.\" without limitation the rights to use, copy, modify, merge, publish,
	.\" distribute, sublicense, and/or sell copies of the Software, and to
	.\" permit persons to whom the Software is furnished to do so, subject to
	.\" the following conditions:
	.\"
	.\" The above copyright notice and this permission notice shall be
	.\" included in all copies or substantial portions of the Software.
	.\"
	.\" THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
	.\" EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
	.\" MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
	.\" NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
	.\" LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
	.\" OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
	.\" WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
	.\"
	.Dd May 31, 2021
	.Dt ZFS-MOUNT-GENERATOR 8
	.Os
	.
	.Sh NAME
	.Nm zfs-mount-generator
	.Nd generate systemd mount units for ZFS filesystems
	.Sh SYNOPSIS
	.Pa @systemdgeneratordir@/zfs-mount-generator
	.
	.Sh DESCRIPTION
	.Nm
	is a
	.Xr systemd.generator 7
	that generates native
	.Xr systemd.mount 5
	units for configured ZFS datasets.
	.
	.Ss Properties
	.Bl -tag -compact -width "org.openzfs.systemd:required-by=unit[ unit]…"
	.It Sy mountpoint Ns =
	.No Skipped if Sy legacy No or Sy none .
	.
	.It Sy canmount Ns =
	.No Skipped if Sy off .
	.No Skipped if only Sy noauto
	datasets exist for a given mountpoint and there's more than one.
	.No Datasets with Sy yes No take precedence over ones with Sy noauto No for the same mountpoint.
	.No Sets logical Em noauto No flag if Sy noauto .
	Encryption roots always generate
	.Sy zfs-load-key@ Ns Ar root Ns Sy .service ,
	even if
	.Sy off .
	.
	.It Sy atime Ns = , Sy relatime Ns = , Sy devices Ns = , Sy exec Ns = , Sy readonly Ns = , Sy setuid Ns = , Sy nbmand Ns =
	Used to generate mount options equivalent to
	.Nm zfs Cm mount .
	.
	.It Sy encroot Ns = , Sy keylocation Ns =
	If the dataset is an encryption root, its mount unit will bind to
	.Sy zfs-load-key@ Ns Ar root Ns Sy .service ,
	with additional dependencies as follows:
	.Bl -tag -compact -offset Ds -width "keylocation=https://URL (et al.)"
	.It Sy keylocation Ns = Ns Sy prompt
	None, uses
	.Xr systemd-ask-password 1
	.It Sy keylocation Ns = Ns Sy https:// Ns Ar URL Pq et al.\&
	.Sy Wants Ns = , Sy After Ns = : Pa network-online.target
	.It Sy keylocation Ns = Ns Sy file:// Ns < Ns Ar path Ns >
	.Sy RequiresMountsFor Ns = Ns Ar path
	.El
	.
	The service also uses the same
	.Sy Wants Ns = ,
	.Sy After Ns = ,
	.Sy Requires Ns = , No and
	.Sy RequiresMountsFor Ns = ,
	as the mount unit.
	.
	.It Sy org.openzfs.systemd:requires Ns = Ns Pa path Ns Oo " " Ns Pa path Oc Ns …
	.No Sets Sy Requires Ns = for the mount- and key-loading unit.
	.
	.It Sy org.openzfs.systemd:requires-mounts-for Ns = Ns Pa path Ns Oo " " Ns Pa path Oc Ns …
	.No Sets Sy RequiresMountsFor Ns = for the mount- and key-loading unit.
	.
	.It Sy org.openzfs.systemd:before Ns = Ns Pa unit Ns Oo " " Ns Pa unit Oc Ns …
	.No Sets Sy Before Ns = for the mount unit.
	.
	.It Sy org.openzfs.systemd:after Ns = Ns Pa unit Ns Oo " " Ns Pa unit Oc Ns …
	.No Sets Sy After Ns = for the mount unit.
	.
	.It Sy org.openzfs.systemd:wanted-by Ns = Ns Pa unit Ns Oo " " Ns Pa unit Oc Ns …
	.No Sets logical Em noauto No flag (see below).
	.No If not Sy none , No sets Sy WantedBy Ns = for the mount unit.
	.It Sy org.openzfs.systemd:required-by Ns = Ns Pa unit Ns Oo " " Ns Pa unit Oc Ns …
	.No Sets logical Em noauto No flag (see below).
	.No If not Sy none , No sets Sy RequiredBy Ns = for the mount unit.
	.
	.It Sy org.openzfs.systemd:nofail Ns = Ns (unset) Ns \| Ns Sy on Ns \| Ns Sy off
	Waxes or wanes strength of default reverse dependencies of the mount unit, see below.
	.
	.It Sy org.openzfs.systemd:ignore Ns = Ns Sy on Ns \| Ns Sy off
	.No Skip if Sy on .
	.No Defaults to Sy off .
	.El
	.
	.Ss Unit Ordering And Dependencies
	Additionally, unless the pool the dataset resides on
	is imported at generation time, both units gain
	.Sy Wants Ns = Ns Pa zfs-import.target
	and
	.Sy After Ns = Ns Pa zfs-import.target .
	.Pp
	Additionally, unless the logical
	.Em noauto
	flag is set, the mount unit gains a reverse-dependency for
	.Pa local-fs.target
	of strength
	.Bl -tag -compact -offset Ds -width "(unset)"
	.It (unset)
	.Sy WantedBy Ns = No + Sy Before Ns =
	.It Sy on
	.Sy WantedBy Ns =
	.It Sy off
	.Sy RequiredBy Ns = No + Sy Before Ns =
	.El
	.
	.Ss Cache File
	Because ZFS pools may not be available very early in the boot process,
	information on ZFS mountpoints must be stored separately.
	The output of
	.Dl Nm zfs Cm list Fl Ho Ar name , Ns Aq every property above in order
	for datasets that should be mounted by systemd should be kept at
	.Pa @sysconfdir@/zfs/zfs-list.cache/ Ns Ar poolname ,
	and, if writeable, will be kept synchronized for the entire pool by the
	.Pa history_event-zfs-list-cacher.sh
	ZEDLET, if enabled
	.Pq see Xr zed 8 .
	.
	.Sh ENVIRONMENT
	-The
	+If the
	.Sy ZFS_DEBUG
	-environment variable can either be
	-.Sy 0
	-(default),
	-.Sy 1
	-(print summary accounting information at the end), or at least
	-.Sy 2
	-(print accounting information for each subprocess as it finishes).
	-.
	-If not present,
	-.Pa /proc/cmdline
	-is additionally checked for
	-.Qq debug ,
	-in which case the debug level is set to
	-.Sy 2 .
	+environment variable is nonzero
	+.Pq or unset and Pa /proc/cmdline No contains Qq Sy debug ,
	+print summary accounting information at the end.
	.
	.Sh EXAMPLES
	To begin, enable tracking for the pool:
	.Dl # Nm touch Pa @sysconfdir@/zfs/zfs-list.cache/ Ns Ar poolname
	Then enable the tracking ZEDLET:
	.Dl # Nm ln Fl s Pa @zfsexecdir@/zed.d/history_event-zfs-list-cacher.sh @sysconfdir@/zfs/zed.d
	.Dl # Nm systemctl Cm enable Pa zfs-zed.service
	.Dl # Nm systemctl Cm restart Pa zfs-zed.service
	.Pp
	If no history event is in the queue,
	inject one to ensure the ZEDLET runs to refresh the cache file
	by setting a monitored property somewhere on the pool:
	.Dl # Nm zfs Cm set Sy relatime Ns = Ns Sy off Ar poolname/dset
	.Dl # Nm zfs Cm inherit Sy relatime Ar poolname/dset
	.Pp
	To test the generator output:
	.Dl $ Nm mkdir Pa /tmp/zfs-mount-generator
	.Dl $ Nm @systemdgeneratordir@/zfs-mount-generator Pa /tmp/zfs-mount-generator
	.
	If the generated units are satisfactory, instruct
	.Nm systemd
	to re-run all generators:
	.Dl # Nm systemctl daemon-reload
	.
	.Sh SEE ALSO
	.Xr systemd.mount 5 ,
	.Xr systemd.target 5 ,
	.Xr zfs 5 ,
	.Xr systemd.generator 7 ,
	.Xr systemd.special 7 ,
	.Xr zed 8 ,
	.Xr zpool-events 8
	diff --git a/sys/contrib/openzfs/man/man8/zfs-receive.8 b/sys/contrib/openzfs/man/man8/zfs-receive.8
	index d2cec42a8e71..84ec657e5cd6 100644
	--- a/sys/contrib/openzfs/man/man8/zfs-receive.8
	+++ b/sys/contrib/openzfs/man/man8/zfs-receive.8
	@@ -1,400 +1,400 @@
	.\"
	.\" CDDL HEADER START
	.\"
	.\" The contents of this file are subject to the terms of the
	.\" Common Development and Distribution License (the "License").
	.\" You may not use this file except in compliance with the License.
	.\"
	.\" You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	.\" or http://www.opensolaris.org/os/licensing.
	.\" See the License for the specific language governing permissions
	.\" and limitations under the License.
	.\"
	.\" When distributing Covered Code, include this CDDL HEADER in each
	.\" file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	.\" If applicable, add the following below this CDDL HEADER, with the
	.\" fields enclosed by brackets "[]" replaced with your own identifying
	.\" information: Portions Copyright [yyyy] [name of copyright owner]
	.\"
	.\" CDDL HEADER END
	.\"
	.\" Copyright (c) 2009 Sun Microsystems, Inc. All Rights Reserved.
	.\" Copyright 2011 Joshua M. Clulow <josh@sysmgr.org>
	.\" Copyright (c) 2011, 2019 by Delphix. All rights reserved.
	.\" Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
	.\" Copyright (c) 2014, Joyent, Inc. All rights reserved.
	.\" Copyright (c) 2014 by Adam Stevko. All rights reserved.
	.\" Copyright (c) 2014 Integros [integros.com]
	.\" Copyright 2019 Richard Laager. All rights reserved.
	.\" Copyright 2018 Nexenta Systems, Inc.
	.\" Copyright 2019 Joyent, Inc.
	.\"
	.Dd February 16, 2020
	.Dt ZFS-RECEIVE 8
	.Os
	.
	.Sh NAME
	.Nm zfs-receive
	.Nd create snapshot from backup stream
	.Sh SYNOPSIS
	.Nm zfs
	.Cm receive
	.Op Fl FhMnsuv
	.Op Fl o Sy origin Ns = Ns Ar snapshot
	.Op Fl o Ar property Ns = Ns Ar value
	.Op Fl x Ar property
	.Ar filesystem Ns \| Ns Ar volume Ns \| Ns Ar snapshot
	.Nm zfs
	.Cm receive
	.Op Fl FhMnsuv
	.Op Fl d Ns \| Ns Fl e
	.Op Fl o Sy origin Ns = Ns Ar snapshot
	.Op Fl o Ar property Ns = Ns Ar value
	.Op Fl x Ar property
	.Ar filesystem
	.Nm zfs
	.Cm receive
	.Fl A
	.Ar filesystem Ns \| Ns Ar volume
	.
	.Sh DESCRIPTION
	.Bl -tag -width ""
	.It Xo
	.Nm zfs
	.Cm receive
	.Op Fl FhMnsuv
	.Op Fl o Sy origin Ns = Ns Ar snapshot
	.Op Fl o Ar property Ns = Ns Ar value
	.Op Fl x Ar property
	.Ar filesystem Ns \| Ns Ar volume Ns \| Ns Ar snapshot
	.Xc
	.It Xo
	.Nm zfs
	.Cm receive
	.Op Fl FhMnsuv
	.Op Fl d Ns \| Ns Fl e
	.Op Fl o Sy origin Ns = Ns Ar snapshot
	.Op Fl o Ar property Ns = Ns Ar value
	.Op Fl x Ar property
	.Ar filesystem
	.Xc
	Creates a snapshot whose contents are as specified in the stream provided on
	standard input.
	If a full stream is received, then a new file system is created as well.
	Streams are created using the
	.Nm zfs Cm send
	subcommand, which by default creates a full stream.
	.Nm zfs Cm recv
	can be used as an alias for
	.Nm zfs Cm receive .
	.Pp
	If an incremental stream is received, then the destination file system must
	already exist, and its most recent snapshot must match the incremental stream's
	source.
	For
	.Sy zvols ,
	the destination device link is destroyed and recreated, which means the
	.Sy zvol
	cannot be accessed during the
	.Cm receive
	operation.
	.Pp
	When a snapshot replication package stream that is generated by using the
	.Nm zfs Cm send Fl R
	command is received, any snapshots that do not exist on the sending location are
	destroyed by using the
	.Nm zfs Cm destroy Fl d
	command.
	.Pp
	The ability to send and receive deduplicated send streams has been removed.
	However, a deduplicated send stream created with older software can be converted
	to a regular (non-deduplicated) stream by using the
	.Nm zstream Cm redup
	command.
	.Pp
	If
	.Fl o Em property Ns = Ns Ar value
	or
	.Fl x Em property
	is specified, it applies to the effective value of the property throughout
	the entire subtree of replicated datasets.
	Effective property values will be set
	.Pq Fl o
	or inherited
	.Pq Fl x
	on the topmost in the replicated subtree.
	In descendant datasets, if the
	property is set by the send stream, it will be overridden by forcing the
	property to be inherited from the top‐most file system.
	Received properties are retained in spite of being overridden
	and may be restored with
	.Nm zfs Cm inherit Fl S .
	Specifying
	.Fl o Sy origin Ns = Ns Em snapshot
	is a special case because, even if
	.Sy origin
	is a read-only property and cannot be set, it's allowed to receive the send
	stream as a clone of the given snapshot.
	.Pp
	Raw encrypted send streams (created with
	.Nm zfs Cm send Fl w )
	may only be received as is, and cannot be re-encrypted, decrypted, or
	recompressed by the receive process.
	Unencrypted streams can be received as
	encrypted datasets, either through inheritance or by specifying encryption
	parameters with the
	.Fl o
	options.
	Note that the
	.Sy keylocation
	property cannot be overridden to
	.Sy prompt
	during a receive.
	This is because the receive process itself is already using
	the standard input for the send stream.
	Instead, the property can be overridden after the receive completes.
	.Pp
	The added security provided by raw sends adds some restrictions to the send
	and receive process.
	ZFS will not allow a mix of raw receives and non-raw receives.
	Specifically, any raw incremental receives that are attempted after
	a non-raw receive will fail.
	Non-raw receives do not have this restriction and,
	therefore, are always possible.
	Because of this, it is best practice to always
	use either raw sends for their security benefits or non-raw sends for their
	flexibility when working with encrypted datasets, but not a combination.
	.Pp
	The reason for this restriction stems from the inherent restrictions of the
	AEAD ciphers that ZFS uses to encrypt data.
	When using ZFS native encryption,
	each block of data is encrypted against a randomly generated number known as
	the "initialization vector" (IV), which is stored in the filesystem metadata.
	This number is required by the encryption algorithms whenever the data is to
	be decrypted.
	Together, all of the IVs provided for all of the blocks in a
	given snapshot are collectively called an "IV set".
	When ZFS performs a raw send, the IV set is transferred from the source
	to the destination in the send stream.
	When ZFS performs a non-raw send, the data is decrypted by the source
	system and re-encrypted by the destination system, creating a snapshot with
	effectively the same data, but a different IV set.
	In order for decryption to work after a raw send, ZFS must ensure that
	the IV set used on both the source and destination side match.
	When an incremental raw receive is performed on
	top of an existing snapshot, ZFS will check to confirm that the "from"
	snapshot on both the source and destination were using the same IV set,
	ensuring the new IV set is consistent.
	.Pp
	The name of the snapshot
	.Pq and file system, if a full stream is received
	that this subcommand creates depends on the argument type and the use of the
	.Fl d
	or
	.Fl e
	options.
	.Pp
	If the argument is a snapshot name, the specified
	.Ar snapshot
	is created.
	If the argument is a file system or volume name, a snapshot with the same name
	as the sent snapshot is created within the specified
	.Ar filesystem
	or
	.Ar volume .
	If neither of the
	.Fl d
	or
	.Fl e
	options are specified, the provided target snapshot name is used exactly as
	provided.
	.Pp
	The
	.Fl d
	and
	.Fl e
	options cause the file system name of the target snapshot to be determined by
	appending a portion of the sent snapshot's name to the specified target
	.Ar filesystem .
	If the
	.Fl d
	option is specified, all but the first element of the sent snapshot's file
	system path
	.Pq usually the pool name
	is used and any required intermediate file systems within the specified one are
	created.
	If the
	.Fl e
	option is specified, then only the last element of the sent snapshot's file
	system name
	.Pq i.e. the name of the source file system itself
	is used as the target file system name.
	.Bl -tag -width "-F"
	.It Fl F
	Force a rollback of the file system to the most recent snapshot before
	performing the receive operation.
	If receiving an incremental replication stream
	.Po for example, one generated by
	.Nm zfs Cm send Fl R Op Fl i Ns \| Ns Fl I
	.Pc ,
	destroy snapshots and file systems that do not exist on the sending side.
	.It Fl d
	Discard the first element of the sent snapshot's file system name, using the
	remaining elements to determine the name of the target file system for the new
	snapshot as described in the paragraph above.
	.It Fl e
	Discard all but the last element of the sent snapshot's file system name, using
	that element to determine the name of the target file system for the new
	snapshot as described in the paragraph above.
	.It Fl h
	Skip the receive of holds.
	There is no effect if holds are not sent.
	.It Fl M
	Force an unmount of the file system while receiving a snapshot.
	This option is not supported on Linux.
	.It Fl n
	Do not actually receive the stream.
	This can be useful in conjunction with the
	.Fl v
	option to verify the name the receive operation would use.
	.It Fl o Sy origin Ns = Ns Ar snapshot
	Forces the stream to be received as a clone of the given snapshot.
	If the stream is a full send stream, this will create the filesystem
	described by the stream as a clone of the specified snapshot.
	Which snapshot was specified will not affect the success or failure of the
	receive, as long as the snapshot does exist.
	If the stream is an incremental send stream, all the normal verification will be
	performed.
	.It Fl o Em property Ns = Ns Ar value
	Sets the specified property as if the command
	.Nm zfs Cm set Em property Ns = Ns Ar value
	was invoked immediately before the receive.
	When receiving a stream from
	.Nm zfs Cm send Fl R ,
	causes the property to be inherited by all descendant datasets, as through
	.Nm zfs Cm inherit Em property
	was run on any descendant datasets that have this property set on the
	sending system.
	.Pp
	If the send stream was sent with
	.Fl c
	then overriding the
	.Sy compression
	-property will have no affect on received data but the
	+property will have no effect on received data but the
	.Sy compression
	property will be set.
	To have the data recompressed on receive remove the
	.Fl c
	flag from the send stream.
	.Pp
	Any editable property can be set at receive time.
	Set-once properties bound
	to the received data, such as
	.Sy normalization
	and
	.Sy casesensitivity ,
	cannot be set at receive time even when the datasets are newly created by
	.Nm zfs Cm receive .
	Additionally both settable properties
	.Sy version
	and
	.Sy volsize
	cannot be set at receive time.
	.Pp
	The
	.Fl o
	option may be specified multiple times, for different properties.
	An error results if the same property is specified in multiple
	.Fl o
	or
	.Fl x
	options.
	.Pp
	The
	.Fl o
	option may also be used to override encryption properties upon initial receive.
	This allows unencrypted streams to be received as encrypted datasets.
	To cause the received dataset (or root dataset of a recursive stream) to be
	received as an encryption root, specify encryption properties in the same
	manner as is required for
	.Nm zfs Cm create .
	For instance:
	.Dl # Nm zfs Cm send Pa tank/test@snap1 \| Nm zfs Cm recv Fl o Sy encryption Ns = Ns Sy on Fl o keyformat=passphrase Fl o Sy keylocation Ns = Ns Pa file:///path/to/keyfile
	.Pp
	Note that
	.Fl o Sy keylocation Ns = Ns Sy prompt
	may not be specified here, since the standard input
	is already being utilized for the send stream.
	Once the receive has completed, you can use
	.Nm zfs Cm set
	to change this setting after the fact.
	Similarly, you can receive a dataset as an encrypted child by specifying
	.Op Fl x Ar encryption
	to force the property to be inherited.
	Overriding encryption properties (except for
	.Sy keylocation )
	is not possible with raw send streams.
	.It Fl s
	If the receive is interrupted, save the partially received state, rather
	than deleting it.
	Interruption may be due to premature termination of the stream
	.Po e.g. due to network failure or failure of the remote system
	if the stream is being read over a network connection
	.Pc ,
	a checksum error in the stream, termination of the
	.Nm zfs Cm receive
	process, or unclean shutdown of the system.
	.Pp
	The receive can be resumed with a stream generated by
	.Nm zfs Cm send Fl t Ar token ,
	where the
	.Ar token
	is the value of the
	.Sy receive_resume_token
	property of the filesystem or volume which is received into.
	.Pp
	To use this flag, the storage pool must have the
	.Sy extensible_dataset
	feature enabled.
	See
	.Xr zpool-features 7
	for details on ZFS feature flags.
	.It Fl u
	File system that is associated with the received stream is not mounted.
	.It Fl v
	Print verbose information about the stream and the time required to perform the
	receive operation.
	.It Fl x Em property
	Ensures that the effective value of the specified property after the
	receive is unaffected by the value of that property in the send stream (if any),
	as if the property had been excluded from the send stream.
	.Pp
	If the specified property is not present in the send stream, this option does
	nothing.
	.Pp
	If a received property needs to be overridden, the effective value will be
	set or inherited, depending on whether the property is inheritable or not.
	.Pp
	In the case of an incremental update,
	.Fl x
	leaves any existing local setting or explicit inheritance unchanged.
	.Pp
	All
	.Fl o
	restrictions (e.g. set-once) apply equally to
	.Fl x .
	.El
	.It Xo
	.Nm zfs
	.Cm receive
	.Fl A
	.Ar filesystem Ns \| Ns Ar volume
	.Xc
	Abort an interrupted
	.Nm zfs Cm receive Fl s ,
	deleting its saved partially received state.
	.El
	.
	.Sh SEE ALSO
	.Xr zfs-send 8 ,
	.Xr zstream 8
	diff --git a/sys/contrib/openzfs/man/man8/zpool-get.8 b/sys/contrib/openzfs/man/man8/zpool-get.8
	index 55904f169e24..4ef9a1b5ec44 100644
	--- a/sys/contrib/openzfs/man/man8/zpool-get.8
	+++ b/sys/contrib/openzfs/man/man8/zpool-get.8
	@@ -1,108 +1,187 @@
	.\"
	.\" CDDL HEADER START
	.\"
	.\" The contents of this file are subject to the terms of the
	.\" Common Development and Distribution License (the "License").
	.\" You may not use this file except in compliance with the License.
	.\"
	.\" You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	.\" or http://www.opensolaris.org/os/licensing.
	.\" See the License for the specific language governing permissions
	.\" and limitations under the License.
	.\"
	.\" When distributing Covered Code, include this CDDL HEADER in each
	.\" file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	.\" If applicable, add the following below this CDDL HEADER, with the
	.\" fields enclosed by brackets "[]" replaced with your own identifying
	.\" information: Portions Copyright [yyyy] [name of copyright owner]
	.\"
	.\" CDDL HEADER END
	.\"
	.\" Copyright (c) 2007, Sun Microsystems, Inc. All Rights Reserved.
	.\" Copyright (c) 2012, 2018 by Delphix. All rights reserved.
	.\" Copyright (c) 2012 Cyril Plisko. All Rights Reserved.
	.\" Copyright (c) 2017 Datto Inc.
	.\" Copyright (c) 2018 George Melikov. All Rights Reserved.
	.\" Copyright 2017 Nexenta Systems, Inc.
	.\" Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
	.\"
	.Dd August 9, 2019
	.Dt ZPOOL-GET 8
	.Os
	.
	.Sh NAME
	.Nm zpool-get
	.Nd retrieve properties of ZFS storage pools
	.Sh SYNOPSIS
	.Nm zpool
	.Cm get
	.Op Fl Hp
	.Op Fl o Ar field Ns Oo , Ns Ar field Oc Ns …
	.Sy all Ns \| Ns Ar property Ns Oo , Ns Ar property Oc Ns …
	.Oo Ar pool Oc Ns …
	+.
	+.Nm zpool
	+.Cm get
	+.Op Fl Hp
	+.Op Fl o Ar field Ns Oo , Ns Ar field Oc Ns …
	+.Sy all Ns \| Ns Ar property Ns Oo , Ns Ar property Oc Ns …
	+.Ar pool
	+.Oo Sy all-vdevs Ns \| Ns
	+.Ar vdev Oc Ns …
	+.
	.Nm zpool
	.Cm set
	.Ar property Ns = Ns Ar value
	.Ar pool
	.
	+.Nm zpool
	+.Cm set
	+.Ar property Ns = Ns Ar value
	+.Ar pool
	+.Ar vdev
	+.
	.Sh DESCRIPTION
	.Bl -tag -width Ds
	.It Xo
	.Nm zpool
	.Cm get
	.Op Fl Hp
	.Op Fl o Ar field Ns Oo , Ns Ar field Oc Ns …
	.Sy all Ns \| Ns Ar property Ns Oo , Ns Ar property Oc Ns …
	.Oo Ar pool Oc Ns …
	.Xc
	Retrieves the given list of properties
	.Po
	or all properties if
	.Sy all
	is used
	.Pc
	for the specified storage pool(s).
	These properties are displayed with the following fields:
	.Bl -tag -compact -offset Ds -width "property"
	.It Sy name
	Name of storage pool.
	.It Sy property
	Property name.
	.It Sy value
	Property value.
	.It Sy source
	Property source, either
	.Sy default No or Sy local .
	.El
	.Pp
	See the
	.Xr zpoolprops 7
	manual page for more information on the available pool properties.
	.Bl -tag -compact -offset Ds -width "-o field"
	.It Fl H
	Scripted mode.
	Do not display headers, and separate fields by a single tab instead of arbitrary
	space.
	.It Fl o Ar field
	A comma-separated list of columns to display, defaults to
	.Sy name , Ns Sy property , Ns Sy value , Ns Sy source .
	.It Fl p
	Display numbers in parsable (exact) values.
	.El
	.It Xo
	.Nm zpool
	+.Cm get
	+.Op Fl Hp
	+.Op Fl o Ar field Ns Oo , Ns Ar field Oc Ns …
	+.Sy all Ns \| Ns Ar property Ns Oo , Ns Ar property Oc Ns …
	+.Ar pool
	+.Oo Sy all-vdevs Ns \| Ns
	+.Ar vdev Oc Ns …
	+.Xc
	+Retrieves the given list of properties
	+.Po
	+or all properties if
	+.Sy all
	+is used
	+.Pc
	+for the specified vdevs
	+.Po
	+or all vdevs if
	+.Sy all-vdevs
	+is used
	+.Pc
	+in the specified pool.
	+These properties are displayed with the following fields:
	+.Bl -tag -compact -offset Ds -width "property"
	+.It Sy name
	+Name of vdev.
	+.It Sy property
	+Property name.
	+.It Sy value
	+Property value.
	+.It Sy source
	+Property source, either
	+.Sy default No or Sy local .
	+.El
	+.Pp
	+See the
	+.Xr vdevprops 7
	+manual page for more information on the available pool properties.
	+.Bl -tag -compact -offset Ds -width "-o field"
	+.It Fl H
	+Scripted mode.
	+Do not display headers, and separate fields by a single tab instead of arbitrary
	+space.
	+.It Fl o Ar field
	+A comma-separated list of columns to display, defaults to
	+.Sy name , Ns Sy property , Ns Sy value , Ns Sy source .
	+.It Fl p
	+Display numbers in parsable (exact) values.
	+.El
	+.It Xo
	+.Nm zpool
	.Cm set
	.Ar property Ns = Ns Ar value
	.Ar pool
	.Xc
	Sets the given property on the specified pool.
	See the
	.Xr zpoolprops 7
	manual page for more information on what properties can be set and acceptable
	values.
	+.It Xo
	+.Nm zpool
	+.Cm set
	+.Ar property Ns = Ns Ar value
	+.Ar pool
	+.Ar vdev
	+.Xc
	+Sets the given property on the specified vdev in the specified pool.
	+See the
	+.Xr vdevprops 7
	+manual page for more information on what properties can be set and acceptable
	+values.
	.El
	.
	.Sh SEE ALSO
	+.Xr vdevprops 7 ,
	.Xr zpool-features 7 ,
	.Xr zpoolprops 7 ,
	.Xr zpool-list 8
	diff --git a/sys/contrib/openzfs/man/man8/zpool.8 b/sys/contrib/openzfs/man/man8/zpool.8
	index 192a8e2eac8d..e2de528a301f 100644
	--- a/sys/contrib/openzfs/man/man8/zpool.8
	+++ b/sys/contrib/openzfs/man/man8/zpool.8
	@@ -1,560 +1,560 @@
	.\"
	.\" CDDL HEADER START
	.\"
	.\" The contents of this file are subject to the terms of the
	.\" Common Development and Distribution License (the "License").
	.\" You may not use this file except in compliance with the License.
	.\"
	.\" You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	.\" or http://www.opensolaris.org/os/licensing.
	.\" See the License for the specific language governing permissions
	.\" and limitations under the License.
	.\"
	.\" When distributing Covered Code, include this CDDL HEADER in each
	.\" file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	.\" If applicable, add the following below this CDDL HEADER, with the
	.\" fields enclosed by brackets "[]" replaced with your own identifying
	.\" information: Portions Copyright [yyyy] [name of copyright owner]
	.\"
	.\" CDDL HEADER END
	.\"
	.\" Copyright (c) 2007, Sun Microsystems, Inc. All Rights Reserved.
	.\" Copyright (c) 2012, 2018 by Delphix. All rights reserved.
	.\" Copyright (c) 2012 Cyril Plisko. All Rights Reserved.
	.\" Copyright (c) 2017 Datto Inc.
	.\" Copyright (c) 2018 George Melikov. All Rights Reserved.
	.\" Copyright 2017 Nexenta Systems, Inc.
	.\" Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
	.\"
	.Dd June 2, 2021
	.Dt ZPOOL 8
	.Os
	.
	.Sh NAME
	.Nm zpool
	.Nd configure ZFS storage pools
	.Sh SYNOPSIS
	.Nm
	.Fl ?V
	.Nm
	.Cm version
	.Nm
	.Cm subcommand
	-.Op Ar argumentss
	+.Op Ar arguments
	.
	.Sh DESCRIPTION
	The
	.Nm
	command configures ZFS storage pools.
	A storage pool is a collection of devices that provides physical storage and
	data replication for ZFS datasets.
	All datasets within a storage pool share the same space.
	See
	.Xr zfs 8
	for information on managing datasets.
	.Pp
	For an overview of creating and managing ZFS storage pools see the
	.Xr zpoolconcepts 7
	manual page.
	.
	.Sh SUBCOMMANDS
	All subcommands that modify state are logged persistently to the pool in their
	original form.
	.Pp
	The
	.Nm
	command provides subcommands to create and destroy storage pools, add capacity
	to storage pools, and provide information about the storage pools.
	The following subcommands are supported:
	.Bl -tag -width Ds
	.It Xo
	.Nm
	.Fl ?\&
	.Xc
	Displays a help message.
	.It Xo
	.Nm
	.Fl V , -version
	.Xc
	.It Xo
	.Nm
	.Cm version
	.Xc
	Displays the software version of the
	.Nm
	userland utility and the ZFS kernel module.
	.El
	.
	.Ss Creation
	.Bl -tag -width Ds
	.It Xr zpool-create 8
	Creates a new storage pool containing the virtual devices specified on the
	command line.
	.It Xr zpool-initialize 8
	Begins initializing by writing to all unallocated regions on the specified
	devices, or all eligible devices in the pool if no individual devices are
	specified.
	.El
	.
	.Ss Destruction
	.Bl -tag -width Ds
	.It Xr zpool-destroy 8
	Destroys the given pool, freeing up any devices for other use.
	.It Xr zpool-labelclear 8
	Removes ZFS label information from the specified
	.Ar device .
	.El
	.
	.Ss Virtual Devices
	.Bl -tag -width Ds
	.It Xo
	.Xr zpool-attach 8 Ns / Ns Xr zpool-detach 8
	.Xc
	Increases or decreases redundancy by
	.Cm attach Ns ing or
	.Cm detach Ns ing a device on an existing vdev (virtual device).
	.It Xo
	.Xr zpool-add 8 Ns / Ns Xr zpool-remove 8
	.Xc
	Adds the specified virtual devices to the given pool,
	or removes the specified device from the pool.
	.It Xr zpool-replace 8
	Replaces an existing device (which may be faulted) with a new one.
	.It Xr zpool-split 8
	Creates a new pool by splitting all mirrors in an existing pool (which decreases its redundancy).
	.El
	.
	.Ss Properties
	Available pool properties listed in the
	.Xr zpoolprops 7
	manual page.
	.Bl -tag -width Ds
	.It Xr zpool-list 8
	Lists the given pools along with a health status and space usage.
	.It Xo
	.Xr zpool-get 8 Ns / Ns Xr zpool-set 8
	.Xc
	Retrieves the given list of properties
	.Po
	or all properties if
	.Sy all
	is used
	.Pc
	for the specified storage pool(s).
	.El
	.
	.Ss Monitoring
	.Bl -tag -width Ds
	.It Xr zpool-status 8
	Displays the detailed health status for the given pools.
	.It Xr zpool-iostat 8
	Displays logical I/O statistics for the given pools/vdevs. Physical I/Os may
	be observed via
	.Xr iostat 1 .
	.It Xr zpool-events 8
	Lists all recent events generated by the ZFS kernel modules.
	These events are consumed by the
	.Xr zed 8
	and used to automate administrative tasks such as replacing a failed device
	with a hot spare.
	That manual page also describes the subclasses and event payloads
	that can be generated.
	.It Xr zpool-history 8
	Displays the command history of the specified pool(s) or all pools if no pool is
	specified.
	.El
	.
	.Ss Maintenance
	.Bl -tag -width Ds
	.It Xr zpool-scrub 8
	Begins a scrub or resumes a paused scrub.
	.It Xr zpool-checkpoint 8
	Checkpoints the current state of
	.Ar pool ,
	which can be later restored by
	.Nm zpool Cm import Fl -rewind-to-checkpoint .
	.It Xr zpool-trim 8
	Initiates an immediate on-demand TRIM operation for all of the free space in a pool.
	This operation informs the underlying storage devices of all blocks
	in the pool which are no longer allocated and allows thinly provisioned
	devices to reclaim the space.
	.It Xr zpool-sync 8
	This command forces all in-core dirty data to be written to the primary
	pool storage and not the ZIL.
	It will also update administrative information including quota reporting.
	Without arguments,
	.Nm zpool Cm sync
	will sync all pools on the system.
	Otherwise, it will sync only the specified pool(s).
	.It Xr zpool-upgrade 8
	Manage the on-disk format version of storage pools.
	.It Xr zpool-wait 8
	Waits until all background activity of the given types has ceased in the given
	pool.
	.El
	.
	.Ss Fault Resolution
	.Bl -tag -width Ds
	.It Xo
	.Xr zpool-offline 8 Ns / Ns Xr zpool-online 8
	.Xc
	Takes the specified physical device offline or brings it online.
	.It Xr zpool-resilver 8
	Starts a resilver.
	If an existing resilver is already running it will be restarted from the beginning.
	.It Xr zpool-reopen 8
	Reopen all the vdevs associated with the pool.
	.It Xr zpool-clear 8
	Clears device errors in a pool.
	.El
	.
	.Ss Import & Export
	.Bl -tag -width Ds
	.It Xr zpool-import 8
	Make disks containing ZFS storage pools available for use on the system.
	.It Xr zpool-export 8
	Exports the given pools from the system.
	.It Xr zpool-reguid 8
	Generates a new unique identifier for the pool.
	.El
	.
	.Sh EXIT STATUS
	The following exit values are returned:
	.Bl -tag -compact -offset 4n -width "a"
	.It Sy 0
	Successful completion.
	.It Sy 1
	An error occurred.
	.It Sy 2
	Invalid command line options were specified.
	.El
	.
	.Sh EXAMPLES
	.Bl -tag -width "Exam"
	.It Sy Example 1 : No Creating a RAID-Z Storage Pool
	The following command creates a pool with a single raidz root vdev that
	consists of six disks:
	.Dl # Nm zpool Cm create Ar tank Sy raidz Ar sda sdb sdc sdd sde sdf
	.
	.It Sy Example 2 : No Creating a Mirrored Storage Pool
	The following command creates a pool with two mirrors, where each mirror
	contains two disks:
	.Dl # Nm zpool Cm create Ar tank Sy mirror Ar sda sdb Sy mirror Ar sdc sdd
	.
	.It Sy Example 3 : No Creating a ZFS Storage Pool by Using Partitions
	The following command creates an unmirrored pool using two disk partitions:
	.Dl # Nm zpool Cm create Ar tank sda1 sdb2
	.
	.It Sy Example 4 : No Creating a ZFS Storage Pool by Using Files
	The following command creates an unmirrored pool using files.
	While not recommended, a pool based on files can be useful for experimental
	purposes.
	.Dl # Nm zpool Cm create Ar tank /path/to/file/a /path/to/file/b
	.
	.It Sy Example 5 : No Adding a Mirror to a ZFS Storage Pool
	The following command adds two mirrored disks to the pool
	.Ar tank ,
	assuming the pool is already made up of two-way mirrors.
	The additional space is immediately available to any datasets within the pool.
	.Dl # Nm zpool Cm add Ar tank Sy mirror Ar sda sdb
	.
	.It Sy Example 6 : No Listing Available ZFS Storage Pools
	The following command lists all available pools on the system.
	In this case, the pool
	.Ar zion
	is faulted due to a missing device.
	The results from this command are similar to the following:
	.Bd -literal -compact -offset Ds
	.No # Nm zpool Cm list
	NAME SIZE ALLOC FREE EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT
	rpool 19.9G 8.43G 11.4G - 33% 42% 1.00x ONLINE -
	tank 61.5G 20.0G 41.5G - 48% 32% 1.00x ONLINE -
	zion - - - - - - - FAULTED -
	.Ed
	.
	.It Sy Example 7 : No Destroying a ZFS Storage Pool
	The following command destroys the pool
	.Ar tank
	and any datasets contained within:
	.Dl # Nm zpool Cm destroy Fl f Ar tank
	.
	.It Sy Example 8 : No Exporting a ZFS Storage Pool
	The following command exports the devices in pool
	.Ar tank
	so that they can be relocated or later imported:
	.Dl # Nm zpool Cm export Ar tank
	.
	.It Sy Example 9 : No Importing a ZFS Storage Pool
	The following command displays available pools, and then imports the pool
	.Ar tank
	for use on the system.
	The results from this command are similar to the following:
	.Bd -literal -compact -offset Ds
	.No # Nm zpool Cm import
	pool: tank
	id: 15451357997522795478
	state: ONLINE
	action: The pool can be imported using its name or numeric identifier.
	config:

	tank ONLINE
	mirror ONLINE
	sda ONLINE
	sdb ONLINE

	.No # Nm zpool Cm import Ar tank
	.Ed
	.
	.It Sy Example 10 : No Upgrading All ZFS Storage Pools to the Current Version
	The following command upgrades all ZFS Storage pools to the current version of
	the software:
	.Bd -literal -compact -offset Ds
	.No # Nm zpool Cm upgrade Fl a
	This system is currently running ZFS version 2.
	.Ed
	.
	.It Sy Example 11 : No Managing Hot Spares
	The following command creates a new pool with an available hot spare:
	.Dl # Nm zpool Cm create Ar tank Sy mirror Ar sda sdb Sy spare Ar sdc
	.Pp
	If one of the disks were to fail, the pool would be reduced to the degraded
	state.
	The failed device can be replaced using the following command:
	.Dl # Nm zpool Cm replace Ar tank sda sdd
	.Pp
	Once the data has been resilvered, the spare is automatically removed and is
	made available for use should another device fail.
	The hot spare can be permanently removed from the pool using the following
	command:
	.Dl # Nm zpool Cm remove Ar tank sdc
	.
	.It Sy Example 12 : No Creating a ZFS Pool with Mirrored Separate Intent Logs
	The following command creates a ZFS storage pool consisting of two, two-way
	mirrors and mirrored log devices:
	.Dl # Nm zpool Cm create Ar pool Sy mirror Ar sda sdb Sy mirror Ar sdc sdd Sy log mirror Ar sde sdf
	.
	.It Sy Example 13 : No Adding Cache Devices to a ZFS Pool
	The following command adds two disks for use as cache devices to a ZFS storage
	pool:
	.Dl # Nm zpool Cm add Ar pool Sy cache Ar sdc sdd
	.Pp
	Once added, the cache devices gradually fill with content from main memory.
	Depending on the size of your cache devices, it could take over an hour for
	them to fill.
	Capacity and reads can be monitored using the
	.Cm iostat
	subcommand as follows:
	.Dl # Nm zpool Cm iostat Fl v Ar pool 5
	.
	.It Sy Example 14 : No Removing a Mirrored top-level (Log or Data) Device
	The following commands remove the mirrored log device
	.Sy mirror-2
	and mirrored top-level data device
	.Sy mirror-1 .
	.Pp
	Given this configuration:
	.Bd -literal -compact -offset Ds
	pool: tank
	state: ONLINE
	scrub: none requested
	config:

	NAME STATE READ WRITE CKSUM
	tank ONLINE 0 0 0
	mirror-0 ONLINE 0 0 0
	sda ONLINE 0 0 0
	sdb ONLINE 0 0 0
	mirror-1 ONLINE 0 0 0
	sdc ONLINE 0 0 0
	sdd ONLINE 0 0 0
	logs
	mirror-2 ONLINE 0 0 0
	sde ONLINE 0 0 0
	sdf ONLINE 0 0 0
	.Ed
	.Pp
	The command to remove the mirrored log
	.Ar mirror-2 No is:
	.Dl # Nm zpool Cm remove Ar tank mirror-2
	.Pp
	The command to remove the mirrored data
	.Ar mirror-1 No is:
	.Dl # Nm zpool Cm remove Ar tank mirror-1
	.
	.It Sy Example 15 : No Displaying expanded space on a device
	The following command displays the detailed information for the pool
	.Ar data .
	This pool is comprised of a single raidz vdev where one of its devices
	increased its capacity by 10GB.
	In this example, the pool will not be able to utilize this extra capacity until
	all the devices under the raidz vdev have been expanded.
	.Bd -literal -compact -offset Ds
	.No # Nm zpool Cm list Fl v Ar data
	NAME SIZE ALLOC FREE EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT
	data 23.9G 14.6G 9.30G - 48% 61% 1.00x ONLINE -
	raidz1 23.9G 14.6G 9.30G - 48%
	sda - - - - -
	sdb - - - 10G -
	sdc - - - - -
	.Ed
	.
	.It Sy Example 16 : No Adding output columns
	Additional columns can be added to the
	.Nm zpool Cm status No and Nm zpool Cm iostat No output with Fl c .
	.Bd -literal -compact -offset Ds
	.No # Nm zpool Cm status Fl c Ar vendor , Ns Ar model , Ns Ar size
	NAME STATE READ WRITE CKSUM vendor model size
	tank ONLINE 0 0 0
	mirror-0 ONLINE 0 0 0
	U1 ONLINE 0 0 0 SEAGATE ST8000NM0075 7.3T
	U10 ONLINE 0 0 0 SEAGATE ST8000NM0075 7.3T
	U11 ONLINE 0 0 0 SEAGATE ST8000NM0075 7.3T
	U12 ONLINE 0 0 0 SEAGATE ST8000NM0075 7.3T
	U13 ONLINE 0 0 0 SEAGATE ST8000NM0075 7.3T
	U14 ONLINE 0 0 0 SEAGATE ST8000NM0075 7.3T

	.No # Nm zpool Cm iostat Fl vc Ar size
	capacity operations bandwidth
	pool alloc free read write read write size
	---------- ----- ----- ----- ----- ----- ----- ----
	rpool 14.6G 54.9G 4 55 250K 2.69M
	sda1 14.6G 54.9G 4 55 250K 2.69M 70G
	---------- ----- ----- ----- ----- ----- ----- ----
	.Ed
	.El
	.
	.Sh ENVIRONMENT VARIABLES
	.Bl -tag -compact -width "ZPOOL_IMPORT_UDEV_TIMEOUT_MS"
	.It Sy ZFS_ABORT
	Cause
	.Nm
	to dump core on exit for the purposes of running
	.Sy ::findleaks .
	.It Sy ZFS_COLOR
	Use ANSI color in
	.Nm zpool status
	output.
	.It Sy ZPOOL_IMPORT_PATH
	The search path for devices or files to use with the pool.
	This is a colon-separated list of directories in which
	.Nm
	looks for device nodes and files.
	Similar to the
	.Fl d
	option in
	.Nm zpool import .
	.It Sy ZPOOL_IMPORT_UDEV_TIMEOUT_MS
	The maximum time in milliseconds that
	.Nm zpool import
	will wait for an expected device to be available.
	.It Sy ZPOOL_STATUS_NON_NATIVE_ASHIFT_IGNORE
	If set, suppress warning about non-native vdev ashift in
	.Nm zpool status .
	The value is not used, only the presence or absence of the variable matters.
	.It Sy ZPOOL_VDEV_NAME_GUID
	Cause
	.Nm
	subcommands to output vdev guids by default.
	This behavior is identical to the
	.Nm zpool Cm status Fl g
	command line option.
	.It Sy ZPOOL_VDEV_NAME_FOLLOW_LINKS
	Cause
	.Nm
	subcommands to follow links for vdev names by default.
	This behavior is identical to the
	.Nm zpool Cm status Fl L
	command line option.
	.It Sy ZPOOL_VDEV_NAME_PATH
	Cause
	.Nm
	subcommands to output full vdev path names by default.
	This behavior is identical to the
	.Nm zpool Cm status Fl P
	command line option.
	.It Sy ZFS_VDEV_DEVID_OPT_OUT
	Older OpenZFS implementations had issues when attempting to display pool
	config VDEV names if a
	.Sy devid
	NVP value is present in the pool's config.
	.Pp
	For example, a pool that originated on illumos platform would have a
	.Sy devid
	value in the config and
	.Nm zpool status
	would fail when listing the config.
	This would also be true for future Linux-based pools.
	.Pp
	A pool can be stripped of any
	.Sy devid
	values on import or prevented from adding
	them on
	.Nm zpool Cm create
	or
	.Nm zpool Cm add
	by setting
	.Sy ZFS_VDEV_DEVID_OPT_OUT .
	.Pp
	.It Sy ZPOOL_SCRIPTS_AS_ROOT
	Allow a privileged user to run
	.Nm zpool status/iostat Fl c .
	Normally, only unprivileged users are allowed to run
	.Fl c .
	.It Sy ZPOOL_SCRIPTS_PATH
	The search path for scripts when running
	.Nm zpool status/iostat Fl c .
	This is a colon-separated list of directories and overrides the default
	.Pa ~/.zpool.d
	and
	.Pa /etc/zfs/zpool.d
	search paths.
	.It Sy ZPOOL_SCRIPTS_ENABLED
	Allow a user to run
	.Nm zpool status/iostat Fl c .
	If
	.Sy ZPOOL_SCRIPTS_ENABLED
	is not set, it is assumed that the user is allowed to run
	.Nm zpool Cm status Ns / Ns Cm iostat Fl c .
	.El
	.
	.Sh INTERFACE STABILITY
	.Sy Evolving
	.
	.Sh SEE ALSO
	.Xr zfs 4 ,
	.Xr zpool-features 7 ,
	.Xr zpoolconcepts 7 ,
	.Xr zpoolprops 7 ,
	.Xr zed 8 ,
	.Xr zfs 8 ,
	.Xr zpool-add 8 ,
	.Xr zpool-attach 8 ,
	.Xr zpool-checkpoint 8 ,
	.Xr zpool-clear 8 ,
	.Xr zpool-create 8 ,
	.Xr zpool-destroy 8 ,
	.Xr zpool-detach 8 ,
	.Xr zpool-events 8 ,
	.Xr zpool-export 8 ,
	.Xr zpool-get 8 ,
	.Xr zpool-history 8 ,
	.Xr zpool-import 8 ,
	.Xr zpool-initialize 8 ,
	.Xr zpool-iostat 8 ,
	.Xr zpool-labelclear 8 ,
	.Xr zpool-list 8 ,
	.Xr zpool-offline 8 ,
	.Xr zpool-online 8 ,
	.Xr zpool-reguid 8 ,
	.Xr zpool-remove 8 ,
	.Xr zpool-reopen 8 ,
	.Xr zpool-replace 8 ,
	.Xr zpool-resilver 8 ,
	.Xr zpool-scrub 8 ,
	.Xr zpool-set 8 ,
	.Xr zpool-split 8 ,
	.Xr zpool-status 8 ,
	.Xr zpool-sync 8 ,
	.Xr zpool-trim 8 ,
	.Xr zpool-upgrade 8 ,
	.Xr zpool-wait 8
	diff --git a/sys/contrib/openzfs/module/Makefile.bsd b/sys/contrib/openzfs/module/Makefile.bsd
	index 63aacb04b35b..315be2808605 100644
	--- a/sys/contrib/openzfs/module/Makefile.bsd
	+++ b/sys/contrib/openzfs/module/Makefile.bsd
	@@ -1,375 +1,375 @@
	.if !defined(WITH_CTF)
	WITH_CTF=1
	.endif

	.include <bsd.sys.mk>

	SRCDIR=${.CURDIR}
	INCDIR=${.CURDIR:H}/include

	KMOD= openzfs

	.PATH: ${SRCDIR}/avl \
	${SRCDIR}/lua \
	${SRCDIR}/nvpair \
	${SRCDIR}/icp/algs/edonr \
	${SRCDIR}/os/freebsd/spl \
	${SRCDIR}/os/freebsd/zfs \
	${SRCDIR}/unicode \
	${SRCDIR}/zcommon \
	${SRCDIR}/zfs \
	${SRCDIR}/zstd \
	${SRCDIR}/zstd/lib



	CFLAGS+= -I${.OBJDIR:H}/include
	CFLAGS+= -I${INCDIR}
	CFLAGS+= -I${INCDIR}/os/freebsd
	CFLAGS+= -I${INCDIR}/os/freebsd/spl
	CFLAGS+= -I${INCDIR}/os/freebsd/zfs
	CFLAGS+= -I${SRCDIR}/zstd/include
	CFLAGS+= -include ${INCDIR}/os/freebsd/spl/sys/ccompile.h

	CFLAGS+= -D__KERNEL__ -DFREEBSD_NAMECACHE -DBUILDING_ZFS -D__BSD_VISIBLE=1 \
	-DHAVE_UIO_ZEROCOPY -DWITHOUT_NETDUMP -D__KERNEL -D_SYS_CONDVAR_H_ \
	-D_SYS_VMEM_H_ -DKDTRACE_HOOKS -DSMP -DHAVE_KSID -DCOMPAT_FREEBSD11

	.if ${MACHINE_ARCH} == "amd64"
	CFLAGS+= -DHAVE_AVX2 -DHAVE_AVX -D__x86_64 -DHAVE_SSE2 -DHAVE_AVX512F -DHAVE_SSSE3
	.endif

	.if defined(WITH_DEBUG) && ${WITH_DEBUG} == "true"
	CFLAGS+= -DZFS_DEBUG -g
	.if defined(WITH_INVARIANTS) && ${WITH_INVARIANTS} == "true"
	CFLAGS+= -DINVARIANTS -DWITNESS -DOPENSOLARIS_WITNESS
	.endif
	.if defined(WITH_O0) && ${WITH_O0} == "true"
	CFLAGS+= -O0
	.endif
	.else
	CFLAGS += -DNDEBUG
	.endif

	.if defined(WITH_VFS_DEBUG) && ${WITH_VFS_DEBUG} == "true"
	# kernel must also be built with this option for this to work
	CFLAGS+= -DDEBUG_VFS_LOCKS
	.endif

	.if defined(WITH_GCOV) && ${WITH_GCOV} == "true"
	CFLAGS+= -fprofile-arcs -ftest-coverage
	.endif

	DEBUG_FLAGS=-g

	.if ${MACHINE_ARCH} == "i386" \|\| ${MACHINE_ARCH} == "powerpc" \|\| \
	${MACHINE_ARCH} == "arm"
	CFLAGS+= -DBITS_PER_LONG=32
	.else
	CFLAGS+= -DBITS_PER_LONG=64
	.endif

	SRCS= vnode_if.h device_if.h bus_if.h

	# avl
	SRCS+= avl.c

	# icp
	SRCS+= edonr.c

	#lua
	SRCS+= lapi.c \
	lauxlib.c \
	lbaselib.c \
	lcode.c \
	lcompat.c \
	lcorolib.c \
	lctype.c \
	ldebug.c \
	ldo.c \
	lfunc.c \
	lgc.c \
	llex.c \
	lmem.c \
	lobject.c \
	lopcodes.c \
	lparser.c \
	lstate.c \
	lstring.c \
	lstrlib.c \
	ltable.c \
	ltablib.c \
	ltm.c \
	lvm.c \
	lzio.c

	#nvpair
	SRCS+= nvpair.c \
	fnvpair.c \
	nvpair_alloc_spl.c \
	nvpair_alloc_fixed.c

	#os/freebsd/spl
	SRCS+= acl_common.c \
	callb.c \
	list.c \
	sha256c.c \
	sha512c.c \
	spl_acl.c \
	spl_cmn_err.c \
	spl_dtrace.c \
	spl_kmem.c \
	spl_kstat.c \
	spl_misc.c \
	spl_policy.c \
	spl_procfs_list.c \
	spl_string.c \
	spl_sunddi.c \
	spl_sysevent.c \
	spl_taskq.c \
	spl_uio.c \
	spl_vfs.c \
	spl_vm.c \
	spl_zlib.c \
	spl_zone.c


	.if ${MACHINE_ARCH} == "i386" \|\| ${MACHINE_ARCH} == "powerpc" \|\| \
	${MACHINE_ARCH} == "arm"
	SRCS+= spl_atomic.c
	.endif

	#os/freebsd/zfs
	SRCS+= abd_os.c \
	arc_os.c \
	crypto_os.c \
	dmu_os.c \
	hkdf.c \
	kmod_core.c \
	spa_os.c \
	sysctl_os.c \
	vdev_file.c \
	vdev_geom.c \
	vdev_label_os.c \
	zfs_acl.c \
	zfs_ctldir.c \
	zfs_debug.c \
	zfs_dir.c \
	zfs_ioctl_compat.c \
	zfs_ioctl_os.c \
	zfs_racct.c \
	zfs_vfsops.c \
	zfs_vnops_os.c \
	zfs_znode.c \
	zio_crypt.c \
	zvol_os.c

	#unicode
	SRCS+= uconv.c \
	u8_textprep.c

	#zcommon
	SRCS+= zfeature_common.c \
	zfs_comutil.c \
	zfs_deleg.c \
	zfs_fletcher.c \
	zfs_fletcher_avx512.c \
	zfs_fletcher_intel.c \
	zfs_fletcher_sse.c \
	zfs_fletcher_superscalar.c \
	zfs_fletcher_superscalar4.c \
	zfs_namecheck.c \
	zfs_prop.c \
	zpool_prop.c \
	zprop_common.c

	#zfs
	SRCS+= abd.c \
	aggsum.c \
	arc.c \
	blkptr.c \
	bplist.c \
	bpobj.c \
	btree.c \
	cityhash.c \
	dbuf.c \
	dbuf_stats.c \
	bptree.c \
	bqueue.c \
	dataset_kstats.c \
	ddt.c \
	ddt_zap.c \
	dmu.c \
	dmu_diff.c \
	dmu_object.c \
	dmu_objset.c \
	dmu_recv.c \
	dmu_redact.c \
	dmu_send.c \
	dmu_traverse.c \
	dmu_tx.c \
	dmu_zfetch.c \
	dnode.c \
	dnode_sync.c \
	dsl_dataset.c \
	dsl_deadlist.c \
	dsl_deleg.c \
	dsl_bookmark.c \
	dsl_dir.c \
	dsl_crypt.c \
	dsl_destroy.c \
	dsl_pool.c \
	dsl_prop.c \
	dsl_scan.c \
	dsl_synctask.c \
	dsl_userhold.c \
	edonr_zfs.c \
	fm.c \
	gzip.c \
	lzjb.c \
	lz4.c \
	metaslab.c \
	mmp.c \
	multilist.c \
	objlist.c \
	pathname.c \
	range_tree.c \
	refcount.c \
	rrwlock.c \
	sa.c \
	sha256.c \
	skein_zfs.c \
	spa.c \
	spa_boot.c \
	spa_checkpoint.c \
	spa_config.c \
	spa_errlog.c \
	spa_history.c \
	spa_log_spacemap.c \
	spa_misc.c \
	spa_stats.c \
	space_map.c \
	space_reftree.c \
	txg.c \
	uberblock.c \
	unique.c \
	vdev.c \
	vdev_cache.c \
	vdev_draid.c \
	vdev_draid_rand.c \
	vdev_indirect.c \
	vdev_indirect_births.c \
	vdev_indirect_mapping.c \
	vdev_initialize.c \
	vdev_label.c \
	vdev_mirror.c \
	vdev_missing.c \
	vdev_queue.c \
	vdev_raidz.c \
	vdev_raidz_math.c \
	vdev_raidz_math_scalar.c \
	- vdev_rebuild.c \
	vdev_raidz_math_avx2.c \
	vdev_raidz_math_avx512bw.c \
	vdev_raidz_math_avx512f.c \
	vdev_raidz_math_sse2.c \
	vdev_raidz_math_ssse3.c \
	+ vdev_rebuild.c \
	vdev_removal.c \
	vdev_root.c \
	vdev_trim.c \
	zap.c \
	zap_leaf.c \
	zap_micro.c \
	zcp.c \
	zcp_get.c \
	zcp_global.c \
	zcp_iter.c \
	zcp_set.c \
	zcp_synctask.c \
	zfeature.c \
	zfs_byteswap.c \
	zfs_file_os.c \
	zfs_fm.c \
	zfs_fuid.c \
	zfs_ioctl.c \
	zfs_log.c \
	zfs_onexit.c \
	zfs_quota.c \
	zfs_ratelimit.c \
	zfs_replay.c \
	zfs_rlock.c \
	zfs_sa.c \
	zfs_vnops.c \
	zil.c \
	zio.c \
	zio_checksum.c \
	zio_compress.c \
	zio_inject.c \
	zle.c \
	zrlock.c \
	zthr.c \
	zvol.c

	#zstd
	SRCS+= zfs_zstd.c \
	zstd.c

	beforeinstall:
	.if ${MK_DEBUG_FILES} != "no"
	mtree -eu \
	-f /etc/mtree/BSD.debug.dist \
	-p ${DESTDIR}/usr/lib
	.endif

	.include <bsd.kmod.mk>


	CFLAGS.gcc+= -Wno-pointer-to-int-cast

	CFLAGS.lapi.c= -Wno-cast-qual
	CFLAGS.lcompat.c= -Wno-cast-qual
	CFLAGS.lobject.c= -Wno-cast-qual
	CFLAGS.ltable.c= -Wno-cast-qual
	CFLAGS.lvm.c= -Wno-cast-qual
	CFLAGS.nvpair.c= -DHAVE_RPC_TYPES -Wno-cast-qual
	CFLAGS.spl_string.c= -Wno-cast-qual
	CFLAGS.spl_vm.c= -Wno-cast-qual
	CFLAGS.spl_zlib.c= -Wno-cast-qual
	CFLAGS.abd.c= -Wno-cast-qual
	CFLAGS.zfs_log.c= -Wno-cast-qual
	CFLAGS.zfs_vnops_os.c= -Wno-pointer-arith
	CFLAGS.u8_textprep.c= -Wno-cast-qual
	CFLAGS.zfs_fletcher.c= -Wno-cast-qual -Wno-pointer-arith
	CFLAGS.zfs_fletcher_intel.c= -Wno-cast-qual -Wno-pointer-arith
	CFLAGS.zfs_fletcher_sse.c= -Wno-cast-qual -Wno-pointer-arith
	CFLAGS.zfs_fletcher_avx512.c= -Wno-cast-qual -Wno-pointer-arith
	CFLAGS.zprop_common.c= -Wno-cast-qual
	CFLAGS.ddt.c= -Wno-cast-qual
	CFLAGS.dmu.c= -Wno-cast-qual
	CFLAGS.dmu_traverse.c= -Wno-cast-qual
	CFLAGS.dsl_dir.c= -Wno-cast-qual
	CFLAGS.dsl_deadlist.c= -Wno-cast-qual
	CFLAGS.dsl_prop.c= -Wno-cast-qual
	CFLAGS.edonr.c=-Wno-cast-qual
	CFLAGS.fm.c= -Wno-cast-qual
	CFLAGS.lz4.c= -Wno-cast-qual
	CFLAGS.spa.c= -Wno-cast-qual
	CFLAGS.spa_misc.c= -Wno-cast-qual
	CFLAGS.sysctl_os.c= -include ../zfs_config.h
	CFLAGS.vdev_draid.c= -Wno-cast-qual
	CFLAGS.vdev_raidz.c= -Wno-cast-qual
	CFLAGS.vdev_raidz_math.c= -Wno-cast-qual
	CFLAGS.vdev_raidz_math_scalar.c= -Wno-cast-qual
	CFLAGS.vdev_raidz_math_avx2.c= -Wno-cast-qual -Wno-duplicate-decl-specifier
	CFLAGS.vdev_raidz_math_avx512f.c= -Wno-cast-qual -Wno-duplicate-decl-specifier
	CFLAGS.vdev_raidz_math_sse2.c= -Wno-cast-qual -Wno-duplicate-decl-specifier
	CFLAGS.zap_leaf.c= -Wno-cast-qual
	CFLAGS.zap_micro.c= -Wno-cast-qual
	CFLAGS.zcp.c= -Wno-cast-qual
	CFLAGS.zfs_fm.c= -Wno-cast-qual
	CFLAGS.zfs_ioctl.c= -Wno-cast-qual
	CFLAGS.zil.c= -Wno-cast-qual
	CFLAGS.zio.c= -Wno-cast-qual
	CFLAGS.zrlock.c= -Wno-cast-qual
	CFLAGS.zfs_zstd.c= -Wno-cast-qual -Wno-pointer-arith
	CFLAGS.zstd.c= -fno-tree-vectorize -U__BMI__
	diff --git a/sys/contrib/openzfs/module/nvpair/fnvpair.c b/sys/contrib/openzfs/module/nvpair/fnvpair.c
	index dc8257e48594..3a4f07486808 100644
	--- a/sys/contrib/openzfs/module/nvpair/fnvpair.c
	+++ b/sys/contrib/openzfs/module/nvpair/fnvpair.c
	@@ -1,660 +1,666 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2012, 2018 by Delphix. All rights reserved.
	*/

	#include <sys/nvpair.h>
	#include <sys/kmem.h>
	#include <sys/debug.h>
	#include <sys/param.h>
	#ifndef _KERNEL
	#include <stdlib.h>
	#endif

	/*
	* "Force" nvlist wrapper.
	*
	* These functions wrap the nvlist_* functions with assertions that assume
	* the operation is successful. This allows the caller's code to be much
	* more readable, especially for the fnvlist_lookup_* and fnvpair_value_*
	* functions, which can return the requested value (rather than filling in
	* a pointer).
	*
	* These functions use NV_UNIQUE_NAME, encoding NV_ENCODE_NATIVE, and allocate
	* with KM_SLEEP.
	*
	* More wrappers should be added as needed -- for example
	* nvlist_lookup__array and nvpair_value__array.
	*/

	nvlist_t *
	fnvlist_alloc(void)
	{
	nvlist_t *nvl;
	VERIFY0(nvlist_alloc(&nvl, NV_UNIQUE_NAME, KM_SLEEP));
	return (nvl);
	}

	void
	fnvlist_free(nvlist_t *nvl)
	{
	nvlist_free(nvl);
	}

	size_t
	fnvlist_size(nvlist_t *nvl)
	{
	size_t size;
	VERIFY0(nvlist_size(nvl, &size, NV_ENCODE_NATIVE));
	return (size);
	}

	/*
	* Returns allocated buffer of size *sizep. Caller must free the buffer with
	* fnvlist_pack_free().
	*/
	char *
	fnvlist_pack(nvlist_t nvl, size_t sizep)
	{
	char *packed = 0;
	VERIFY3U(nvlist_pack(nvl, &packed, sizep, NV_ENCODE_NATIVE,
	KM_SLEEP), ==, 0);
	return (packed);
	}

	/ARGSUSED/
	void
	fnvlist_pack_free(char *pack, size_t size)
	{
	#ifdef _KERNEL
	kmem_free(pack, size);
	#else
	free(pack);
	#endif
	}

	nvlist_t *
	fnvlist_unpack(char *buf, size_t buflen)
	{
	nvlist_t *rv;
	VERIFY0(nvlist_unpack(buf, buflen, &rv, KM_SLEEP));
	return (rv);
	}

	nvlist_t *
	-fnvlist_dup(nvlist_t *nvl)
	+fnvlist_dup(const nvlist_t *nvl)
	{
	nvlist_t *rv;
	VERIFY0(nvlist_dup(nvl, &rv, KM_SLEEP));
	return (rv);
	}

	void
	fnvlist_merge(nvlist_t dst, nvlist_t src)
	{
	VERIFY0(nvlist_merge(dst, src, KM_SLEEP));
	}

	size_t
	fnvlist_num_pairs(nvlist_t *nvl)
	{
	size_t count = 0;
	nvpair_t *pair;

	for (pair = nvlist_next_nvpair(nvl, 0); pair != NULL;
	pair = nvlist_next_nvpair(nvl, pair))
	count++;
	return (count);
	}

	void
	fnvlist_add_boolean(nvlist_t nvl, const char name)
	{
	VERIFY0(nvlist_add_boolean(nvl, name));
	}

	void
	fnvlist_add_boolean_value(nvlist_t nvl, const char name, boolean_t val)
	{
	VERIFY0(nvlist_add_boolean_value(nvl, name, val));
	}

	void
	fnvlist_add_byte(nvlist_t nvl, const char name, uchar_t val)
	{
	VERIFY0(nvlist_add_byte(nvl, name, val));
	}

	void
	fnvlist_add_int8(nvlist_t nvl, const char name, int8_t val)
	{
	VERIFY0(nvlist_add_int8(nvl, name, val));
	}

	void
	fnvlist_add_uint8(nvlist_t nvl, const char name, uint8_t val)
	{
	VERIFY0(nvlist_add_uint8(nvl, name, val));
	}

	void
	fnvlist_add_int16(nvlist_t nvl, const char name, int16_t val)
	{
	VERIFY0(nvlist_add_int16(nvl, name, val));
	}

	void
	fnvlist_add_uint16(nvlist_t nvl, const char name, uint16_t val)
	{
	VERIFY0(nvlist_add_uint16(nvl, name, val));
	}

	void
	fnvlist_add_int32(nvlist_t nvl, const char name, int32_t val)
	{
	VERIFY0(nvlist_add_int32(nvl, name, val));
	}

	void
	fnvlist_add_uint32(nvlist_t nvl, const char name, uint32_t val)
	{
	VERIFY0(nvlist_add_uint32(nvl, name, val));
	}

	void
	fnvlist_add_int64(nvlist_t nvl, const char name, int64_t val)
	{
	VERIFY0(nvlist_add_int64(nvl, name, val));
	}

	void
	fnvlist_add_uint64(nvlist_t nvl, const char name, uint64_t val)
	{
	VERIFY0(nvlist_add_uint64(nvl, name, val));
	}

	void
	fnvlist_add_string(nvlist_t nvl, const char name, const char *val)
	{
	VERIFY0(nvlist_add_string(nvl, name, val));
	}

	void
	fnvlist_add_nvlist(nvlist_t nvl, const char name, nvlist_t *val)
	{
	VERIFY0(nvlist_add_nvlist(nvl, name, val));
	}

	void
	fnvlist_add_nvpair(nvlist_t nvl, nvpair_t pair)
	{
	VERIFY0(nvlist_add_nvpair(nvl, pair));
	}

	void
	fnvlist_add_boolean_array(nvlist_t nvl, const char name,
	- boolean_t *val, uint_t n)
	+ const boolean_t *val, uint_t n)
	{
	VERIFY0(nvlist_add_boolean_array(nvl, name, val, n));
	}

	void
	-fnvlist_add_byte_array(nvlist_t nvl, const char name, uchar_t *val, uint_t n)
	+fnvlist_add_byte_array(nvlist_t nvl, const char name, const uchar_t *val,
	+ uint_t n)
	{
	VERIFY0(nvlist_add_byte_array(nvl, name, val, n));
	}

	void
	-fnvlist_add_int8_array(nvlist_t nvl, const char name, int8_t *val, uint_t n)
	+fnvlist_add_int8_array(nvlist_t nvl, const char name, const int8_t *val,
	+ uint_t n)
	{
	VERIFY0(nvlist_add_int8_array(nvl, name, val, n));
	}

	void
	-fnvlist_add_uint8_array(nvlist_t nvl, const char name, uint8_t *val, uint_t n)
	+fnvlist_add_uint8_array(nvlist_t nvl, const char name, const uint8_t *val,
	+ uint_t n)
	{
	VERIFY0(nvlist_add_uint8_array(nvl, name, val, n));
	}

	void
	-fnvlist_add_int16_array(nvlist_t nvl, const char name, int16_t *val, uint_t n)
	+fnvlist_add_int16_array(nvlist_t nvl, const char name, const int16_t *val,
	+ uint_t n)
	{
	VERIFY0(nvlist_add_int16_array(nvl, name, val, n));
	}

	void
	fnvlist_add_uint16_array(nvlist_t nvl, const char name,
	- uint16_t *val, uint_t n)
	+ const uint16_t *val, uint_t n)
	{
	VERIFY0(nvlist_add_uint16_array(nvl, name, val, n));
	}

	void
	-fnvlist_add_int32_array(nvlist_t nvl, const char name, int32_t *val, uint_t n)
	+fnvlist_add_int32_array(nvlist_t nvl, const char name, const int32_t *val,
	+ uint_t n)
	{
	VERIFY0(nvlist_add_int32_array(nvl, name, val, n));
	}

	void
	fnvlist_add_uint32_array(nvlist_t nvl, const char name,
	- uint32_t *val, uint_t n)
	+ const uint32_t *val, uint_t n)
	{
	VERIFY0(nvlist_add_uint32_array(nvl, name, val, n));
	}

	void
	-fnvlist_add_int64_array(nvlist_t nvl, const char name, int64_t *val, uint_t n)
	+fnvlist_add_int64_array(nvlist_t nvl, const char name, const int64_t *val,
	+ uint_t n)
	{
	VERIFY0(nvlist_add_int64_array(nvl, name, val, n));
	}

	void
	fnvlist_add_uint64_array(nvlist_t nvl, const char name,
	- uint64_t *val, uint_t n)
	+ const uint64_t *val, uint_t n)
	{
	VERIFY0(nvlist_add_uint64_array(nvl, name, val, n));
	}

	void
	fnvlist_add_string_array(nvlist_t nvl, const char name,
	- char * const *val, uint_t n)
	+ const char * const *val, uint_t n)
	{
	VERIFY0(nvlist_add_string_array(nvl, name, val, n));
	}

	void
	fnvlist_add_nvlist_array(nvlist_t nvl, const char name,
	- nvlist_t **val, uint_t n)
	+ const nvlist_t * const *val, uint_t n)
	{
	VERIFY0(nvlist_add_nvlist_array(nvl, name, val, n));
	}

	void
	fnvlist_remove(nvlist_t nvl, const char name)
	{
	VERIFY0(nvlist_remove_all(nvl, name));
	}

	void
	fnvlist_remove_nvpair(nvlist_t nvl, nvpair_t pair)
	{
	VERIFY0(nvlist_remove_nvpair(nvl, pair));
	}

	nvpair_t *
	fnvlist_lookup_nvpair(nvlist_t nvl, const char name)
	{
	nvpair_t *rv;
	VERIFY0(nvlist_lookup_nvpair(nvl, name, &rv));
	return (rv);
	}

	/* returns B_TRUE if the entry exists */
	boolean_t
	-fnvlist_lookup_boolean(nvlist_t nvl, const char name)
	+fnvlist_lookup_boolean(const nvlist_t nvl, const char name)
	{
	return (nvlist_lookup_boolean(nvl, name) == 0);
	}

	boolean_t
	-fnvlist_lookup_boolean_value(nvlist_t nvl, const char name)
	+fnvlist_lookup_boolean_value(const nvlist_t nvl, const char name)
	{
	boolean_t rv;
	VERIFY0(nvlist_lookup_boolean_value(nvl, name, &rv));
	return (rv);
	}

	uchar_t
	-fnvlist_lookup_byte(nvlist_t nvl, const char name)
	+fnvlist_lookup_byte(const nvlist_t nvl, const char name)
	{
	uchar_t rv;
	VERIFY0(nvlist_lookup_byte(nvl, name, &rv));
	return (rv);
	}

	int8_t
	-fnvlist_lookup_int8(nvlist_t nvl, const char name)
	+fnvlist_lookup_int8(const nvlist_t nvl, const char name)
	{
	int8_t rv;
	VERIFY0(nvlist_lookup_int8(nvl, name, &rv));
	return (rv);
	}

	int16_t
	-fnvlist_lookup_int16(nvlist_t nvl, const char name)
	+fnvlist_lookup_int16(const nvlist_t nvl, const char name)
	{
	int16_t rv;
	VERIFY0(nvlist_lookup_int16(nvl, name, &rv));
	return (rv);
	}

	int32_t
	-fnvlist_lookup_int32(nvlist_t nvl, const char name)
	+fnvlist_lookup_int32(const nvlist_t nvl, const char name)
	{
	int32_t rv;
	VERIFY0(nvlist_lookup_int32(nvl, name, &rv));
	return (rv);
	}

	int64_t
	-fnvlist_lookup_int64(nvlist_t nvl, const char name)
	+fnvlist_lookup_int64(const nvlist_t nvl, const char name)
	{
	int64_t rv;
	VERIFY0(nvlist_lookup_int64(nvl, name, &rv));
	return (rv);
	}

	uint8_t
	-fnvlist_lookup_uint8(nvlist_t nvl, const char name)
	+fnvlist_lookup_uint8(const nvlist_t nvl, const char name)
	{
	uint8_t rv;
	VERIFY0(nvlist_lookup_uint8(nvl, name, &rv));
	return (rv);
	}

	uint16_t
	-fnvlist_lookup_uint16(nvlist_t nvl, const char name)
	+fnvlist_lookup_uint16(const nvlist_t nvl, const char name)
	{
	uint16_t rv;
	VERIFY0(nvlist_lookup_uint16(nvl, name, &rv));
	return (rv);
	}

	uint32_t
	-fnvlist_lookup_uint32(nvlist_t nvl, const char name)
	+fnvlist_lookup_uint32(const nvlist_t nvl, const char name)
	{
	uint32_t rv;
	VERIFY0(nvlist_lookup_uint32(nvl, name, &rv));
	return (rv);
	}

	uint64_t
	-fnvlist_lookup_uint64(nvlist_t nvl, const char name)
	+fnvlist_lookup_uint64(const nvlist_t nvl, const char name)
	{
	uint64_t rv;
	VERIFY0(nvlist_lookup_uint64(nvl, name, &rv));
	return (rv);
	}

	char *
	fnvlist_lookup_string(nvlist_t nvl, const char name)
	{
	char *rv;
	VERIFY0(nvlist_lookup_string(nvl, name, &rv));
	return (rv);
	}

	nvlist_t *
	fnvlist_lookup_nvlist(nvlist_t nvl, const char name)
	{
	nvlist_t *rv;
	VERIFY0(nvlist_lookup_nvlist(nvl, name, &rv));
	return (rv);
	}
	boolean_t *
	fnvlist_lookup_boolean_array(nvlist_t nvl, const char name, uint_t *n)
	{
	boolean_t *rv;
	VERIFY0(nvlist_lookup_boolean_array(nvl, name, &rv, n));
	return (rv);
	}

	uchar_t *
	fnvlist_lookup_byte_array(nvlist_t nvl, const char name, uint_t *n)
	{
	uchar_t *rv;
	VERIFY0(nvlist_lookup_byte_array(nvl, name, &rv, n));
	return (rv);
	}

	int8_t *
	fnvlist_lookup_int8_array(nvlist_t nvl, const char name, uint_t *n)
	{
	int8_t *rv;
	VERIFY0(nvlist_lookup_int8_array(nvl, name, &rv, n));
	return (rv);
	}

	uint8_t *
	fnvlist_lookup_uint8_array(nvlist_t nvl, const char name, uint_t *n)
	{
	uint8_t *rv;
	VERIFY0(nvlist_lookup_uint8_array(nvl, name, &rv, n));
	return (rv);
	}

	int16_t *
	fnvlist_lookup_int16_array(nvlist_t nvl, const char name, uint_t *n)
	{
	int16_t *rv;
	VERIFY0(nvlist_lookup_int16_array(nvl, name, &rv, n));
	return (rv);
	}

	uint16_t *
	fnvlist_lookup_uint16_array(nvlist_t nvl, const char name, uint_t *n)
	{
	uint16_t *rv;
	VERIFY0(nvlist_lookup_uint16_array(nvl, name, &rv, n));
	return (rv);
	}

	int32_t *
	fnvlist_lookup_int32_array(nvlist_t nvl, const char name, uint_t *n)
	{
	int32_t *rv;
	VERIFY0(nvlist_lookup_int32_array(nvl, name, &rv, n));
	return (rv);
	}

	uint32_t *
	fnvlist_lookup_uint32_array(nvlist_t nvl, const char name, uint_t *n)
	{
	uint32_t *rv;
	VERIFY0(nvlist_lookup_uint32_array(nvl, name, &rv, n));
	return (rv);
	}

	int64_t *
	fnvlist_lookup_int64_array(nvlist_t nvl, const char name, uint_t *n)
	{
	int64_t *rv;
	VERIFY0(nvlist_lookup_int64_array(nvl, name, &rv, n));
	return (rv);
	}

	uint64_t *
	fnvlist_lookup_uint64_array(nvlist_t nvl, const char name, uint_t *n)
	{
	uint64_t *rv;
	VERIFY0(nvlist_lookup_uint64_array(nvl, name, &rv, n));
	return (rv);
	}

	boolean_t
	-fnvpair_value_boolean_value(nvpair_t *nvp)
	+fnvpair_value_boolean_value(const nvpair_t *nvp)
	{
	boolean_t rv;
	VERIFY0(nvpair_value_boolean_value(nvp, &rv));
	return (rv);
	}

	uchar_t
	-fnvpair_value_byte(nvpair_t *nvp)
	+fnvpair_value_byte(const nvpair_t *nvp)
	{
	uchar_t rv;
	VERIFY0(nvpair_value_byte(nvp, &rv));
	return (rv);
	}

	int8_t
	-fnvpair_value_int8(nvpair_t *nvp)
	+fnvpair_value_int8(const nvpair_t *nvp)
	{
	int8_t rv;
	VERIFY0(nvpair_value_int8(nvp, &rv));
	return (rv);
	}

	int16_t
	-fnvpair_value_int16(nvpair_t *nvp)
	+fnvpair_value_int16(const nvpair_t *nvp)
	{
	int16_t rv;
	VERIFY0(nvpair_value_int16(nvp, &rv));
	return (rv);
	}

	int32_t
	-fnvpair_value_int32(nvpair_t *nvp)
	+fnvpair_value_int32(const nvpair_t *nvp)
	{
	int32_t rv;
	VERIFY0(nvpair_value_int32(nvp, &rv));
	return (rv);
	}

	int64_t
	-fnvpair_value_int64(nvpair_t *nvp)
	+fnvpair_value_int64(const nvpair_t *nvp)
	{
	int64_t rv;
	VERIFY0(nvpair_value_int64(nvp, &rv));
	return (rv);
	}

	uint8_t
	-fnvpair_value_uint8(nvpair_t *nvp)
	+fnvpair_value_uint8(const nvpair_t *nvp)
	{
	uint8_t rv;
	VERIFY0(nvpair_value_uint8(nvp, &rv));
	return (rv);
	}

	uint16_t
	-fnvpair_value_uint16(nvpair_t *nvp)
	+fnvpair_value_uint16(const nvpair_t *nvp)
	{
	uint16_t rv;
	VERIFY0(nvpair_value_uint16(nvp, &rv));
	return (rv);
	}

	uint32_t
	-fnvpair_value_uint32(nvpair_t *nvp)
	+fnvpair_value_uint32(const nvpair_t *nvp)
	{
	uint32_t rv;
	VERIFY0(nvpair_value_uint32(nvp, &rv));
	return (rv);
	}

	uint64_t
	-fnvpair_value_uint64(nvpair_t *nvp)
	+fnvpair_value_uint64(const nvpair_t *nvp)
	{
	uint64_t rv;
	VERIFY0(nvpair_value_uint64(nvp, &rv));
	return (rv);
	}

	char *
	fnvpair_value_string(nvpair_t *nvp)
	{
	char *rv;
	VERIFY0(nvpair_value_string(nvp, &rv));
	return (rv);
	}

	nvlist_t *
	fnvpair_value_nvlist(nvpair_t *nvp)
	{
	nvlist_t *rv;
	VERIFY0(nvpair_value_nvlist(nvp, &rv));
	return (rv);
	}

	#if defined(_KERNEL)

	EXPORT_SYMBOL(fnvlist_alloc);
	EXPORT_SYMBOL(fnvlist_free);
	EXPORT_SYMBOL(fnvlist_size);
	EXPORT_SYMBOL(fnvlist_pack);
	EXPORT_SYMBOL(fnvlist_pack_free);
	EXPORT_SYMBOL(fnvlist_unpack);
	EXPORT_SYMBOL(fnvlist_dup);
	EXPORT_SYMBOL(fnvlist_merge);

	EXPORT_SYMBOL(fnvlist_add_nvpair);
	EXPORT_SYMBOL(fnvlist_add_boolean);
	EXPORT_SYMBOL(fnvlist_add_boolean_value);
	EXPORT_SYMBOL(fnvlist_add_byte);
	EXPORT_SYMBOL(fnvlist_add_int8);
	EXPORT_SYMBOL(fnvlist_add_uint8);
	EXPORT_SYMBOL(fnvlist_add_int16);
	EXPORT_SYMBOL(fnvlist_add_uint16);
	EXPORT_SYMBOL(fnvlist_add_int32);
	EXPORT_SYMBOL(fnvlist_add_uint32);
	EXPORT_SYMBOL(fnvlist_add_int64);
	EXPORT_SYMBOL(fnvlist_add_uint64);
	EXPORT_SYMBOL(fnvlist_add_string);
	EXPORT_SYMBOL(fnvlist_add_nvlist);
	EXPORT_SYMBOL(fnvlist_add_boolean_array);
	EXPORT_SYMBOL(fnvlist_add_byte_array);
	EXPORT_SYMBOL(fnvlist_add_int8_array);
	EXPORT_SYMBOL(fnvlist_add_uint8_array);
	EXPORT_SYMBOL(fnvlist_add_int16_array);
	EXPORT_SYMBOL(fnvlist_add_uint16_array);
	EXPORT_SYMBOL(fnvlist_add_int32_array);
	EXPORT_SYMBOL(fnvlist_add_uint32_array);
	EXPORT_SYMBOL(fnvlist_add_int64_array);
	EXPORT_SYMBOL(fnvlist_add_uint64_array);
	EXPORT_SYMBOL(fnvlist_add_string_array);
	EXPORT_SYMBOL(fnvlist_add_nvlist_array);

	EXPORT_SYMBOL(fnvlist_remove);
	EXPORT_SYMBOL(fnvlist_remove_nvpair);

	EXPORT_SYMBOL(fnvlist_lookup_nvpair);
	EXPORT_SYMBOL(fnvlist_lookup_boolean);
	EXPORT_SYMBOL(fnvlist_lookup_boolean_value);
	EXPORT_SYMBOL(fnvlist_lookup_byte);
	EXPORT_SYMBOL(fnvlist_lookup_int8);
	EXPORT_SYMBOL(fnvlist_lookup_uint8);
	EXPORT_SYMBOL(fnvlist_lookup_int16);
	EXPORT_SYMBOL(fnvlist_lookup_uint16);
	EXPORT_SYMBOL(fnvlist_lookup_int32);
	EXPORT_SYMBOL(fnvlist_lookup_uint32);
	EXPORT_SYMBOL(fnvlist_lookup_int64);
	EXPORT_SYMBOL(fnvlist_lookup_uint64);
	EXPORT_SYMBOL(fnvlist_lookup_string);
	EXPORT_SYMBOL(fnvlist_lookup_nvlist);

	EXPORT_SYMBOL(fnvpair_value_boolean_value);
	EXPORT_SYMBOL(fnvpair_value_byte);
	EXPORT_SYMBOL(fnvpair_value_int8);
	EXPORT_SYMBOL(fnvpair_value_uint8);
	EXPORT_SYMBOL(fnvpair_value_int16);
	EXPORT_SYMBOL(fnvpair_value_uint16);
	EXPORT_SYMBOL(fnvpair_value_int32);
	EXPORT_SYMBOL(fnvpair_value_uint32);
	EXPORT_SYMBOL(fnvpair_value_int64);
	EXPORT_SYMBOL(fnvpair_value_uint64);
	EXPORT_SYMBOL(fnvpair_value_string);
	EXPORT_SYMBOL(fnvpair_value_nvlist);
	EXPORT_SYMBOL(fnvlist_num_pairs);

	#endif
	diff --git a/sys/contrib/openzfs/module/nvpair/nvpair.c b/sys/contrib/openzfs/module/nvpair/nvpair.c
	index 9834dedd859d..668e4a978105 100644
	--- a/sys/contrib/openzfs/module/nvpair/nvpair.c
	+++ b/sys/contrib/openzfs/module/nvpair/nvpair.c
	@@ -1,3792 +1,3812 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2000, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2015, 2017 by Delphix. All rights reserved.
	* Copyright 2018 RackTop Systems.
	*/

	/*
	* Links to Illumos.org for more information on Interface Libraries:
	* [1] https://illumos.org/man/3lib/libnvpair
	* [2] https://illumos.org/man/3nvpair/nvlist_alloc
	* [3] https://illumos.org/man/9f/nvlist_alloc
	* [4] https://illumos.org/man/9f/nvlist_next_nvpair
	* [5] https://illumos.org/man/9f/nvpair_value_byte
	*/

	#include <sys/debug.h>
	#include <sys/isa_defs.h>
	#include <sys/nvpair.h>
	#include <sys/nvpair_impl.h>
	#include <sys/types.h>
	#include <sys/param.h>
	#include <sys/strings.h>
	#include <rpc/xdr.h>
	#include <sys/mod.h>

	#if defined(_KERNEL)
	#include <sys/sunddi.h>
	#include <sys/sysmacros.h>
	#else
	#include <stdarg.h>
	#include <stdlib.h>
	#include <stddef.h>
	#endif

	#define skip_whitespace(p) while (((p) == ' ') \|\| ((p) == '\t')) p++

	/*
	* nvpair.c - Provides kernel & userland interfaces for manipulating
	* name-value pairs.
	*
	* Overview Diagram
	*
	* +--------------+
	* \| nvlist_t \|
	* \|--------------\|
	* \| nvl_version \|
	* \| nvl_nvflag \|
	* \| nvl_priv -+-+
	* \| nvl_flag \| \|
	* \| nvl_pad \| \|
	* +--------------+ \|
	* V
	* +--------------+ last i_nvp in list
	* \| nvpriv_t \| +--------------------->
	* \|--------------\| \|
	* +--+- nvp_list \| \| +------------+
	* \| \| nvp_last -+--+ + nv_alloc_t \|
	* \| \| nvp_curr \| \|------------\|
	* \| \| nvp_nva -+----> \| nva_ops \|
	* \| \| nvp_stat \| \| nva_arg \|
	* \| +--------------+ +------------+
	* \|
	* +-------+
	* V
	* +---------------------+ +-------------------+
	* \| i_nvp_t \| +-->\| i_nvp_t \| +-->
	* \|---------------------\| \| \|-------------------\| \|
	* \| nvi_next -+--+ \| nvi_next -+--+
	* \| nvi_prev (NULL) \| <----+ nvi_prev \|
	* \| . . . . . . . . . . \| \| . . . . . . . . . \|
	* \| nvp (nvpair_t) \| \| nvp (nvpair_t) \|
	* \| - nvp_size \| \| - nvp_size \|
	* \| - nvp_name_sz \| \| - nvp_name_sz \|
	* \| - nvp_value_elem \| \| - nvp_value_elem \|
	* \| - nvp_type \| \| - nvp_type \|
	* \| - data ... \| \| - data ... \|
	* +---------------------+ +-------------------+
	*
	*
	*
	* +---------------------+ +---------------------+
	* \| i_nvp_t \| +--> +-->\| i_nvp_t (last) \|
	* \|---------------------\| \| \| \|---------------------\|
	* \| nvi_next -+--+ ... --+ \| nvi_next (NULL) \|
	* <-+- nvi_prev \|<-- ... <----+ nvi_prev \|
	* \| . . . . . . . . . \| \| . . . . . . . . . \|
	* \| nvp (nvpair_t) \| \| nvp (nvpair_t) \|
	* \| - nvp_size \| \| - nvp_size \|
	* \| - nvp_name_sz \| \| - nvp_name_sz \|
	* \| - nvp_value_elem \| \| - nvp_value_elem \|
	* \| - DATA_TYPE_NVLIST \| \| - nvp_type \|
	* \| - data (embedded) \| \| - data ... \|
	* \| nvlist name \| +---------------------+
	* \| +--------------+ \|
	* \| \| nvlist_t \| \|
	* \| \|--------------\| \|
	* \| \| nvl_version \| \|
	* \| \| nvl_nvflag \| \|
	* \| \| nvl_priv --+---+---->
	* \| \| nvl_flag \| \|
	* \| \| nvl_pad \| \|
	* \| +--------------+ \|
	* +---------------------+
	*
	*
	* N.B. nvpair_t may be aligned on 4 byte boundary, so +4 will
	* allow value to be aligned on 8 byte boundary
	*
	* name_len is the length of the name string including the null terminator
	* so it must be >= 1
	*/
	#define NVP_SIZE_CALC(name_len, data_len) \
	(NV_ALIGN((sizeof (nvpair_t)) + name_len) + NV_ALIGN(data_len))

	static int i_get_value_size(data_type_t type, const void *data, uint_t nelem);
	static int nvlist_add_common(nvlist_t nvl, const char name, data_type_t type,
	uint_t nelem, const void *data);

	#define NV_STAT_EMBEDDED 0x1
	#define EMBEDDED_NVL(nvp) ((nvlist_t )(void )NVP_VALUE(nvp))
	#define EMBEDDED_NVL_ARRAY(nvp) ((nvlist_t *)(void )NVP_VALUE(nvp))

	#define NVP_VALOFF(nvp) (NV_ALIGN(sizeof (nvpair_t) + (nvp)->nvp_name_sz))
	#define NVPAIR2I_NVP(nvp) \
	((i_nvp_t *)((size_t)(nvp) - offsetof(i_nvp_t, nvi_nvp)))

	#ifdef _KERNEL
	int nvpair_max_recursion = 20;
	#else
	int nvpair_max_recursion = 100;
	#endif

	uint64_t nvlist_hashtable_init_size = (1 << 4);

	int
	nv_alloc_init(nv_alloc_t nva, const nv_alloc_ops_t nvo, /* args */ ...)
	{
	va_list valist;
	int err = 0;

	nva->nva_ops = nvo;
	nva->nva_arg = NULL;

	va_start(valist, nvo);
	if (nva->nva_ops->nv_ao_init != NULL)
	err = nva->nva_ops->nv_ao_init(nva, valist);
	va_end(valist);

	return (err);
	}

	void
	nv_alloc_reset(nv_alloc_t *nva)
	{
	if (nva->nva_ops->nv_ao_reset != NULL)
	nva->nva_ops->nv_ao_reset(nva);
	}

	void
	nv_alloc_fini(nv_alloc_t *nva)
	{
	if (nva->nva_ops->nv_ao_fini != NULL)
	nva->nva_ops->nv_ao_fini(nva);
	}

	nv_alloc_t *
	nvlist_lookup_nv_alloc(nvlist_t *nvl)
	{
	nvpriv_t *priv;

	if (nvl == NULL \|\|
	(priv = (nvpriv_t *)(uintptr_t)nvl->nvl_priv) == NULL)
	return (NULL);

	return (priv->nvp_nva);
	}

	static void *
	nv_mem_zalloc(nvpriv_t *nvp, size_t size)
	{
	nv_alloc_t *nva = nvp->nvp_nva;
	void *buf;

	if ((buf = nva->nva_ops->nv_ao_alloc(nva, size)) != NULL)
	bzero(buf, size);

	return (buf);
	}

	static void
	nv_mem_free(nvpriv_t nvp, void buf, size_t size)
	{
	nv_alloc_t *nva = nvp->nvp_nva;

	nva->nva_ops->nv_ao_free(nva, buf, size);
	}

	static void
	nv_priv_init(nvpriv_t priv, nv_alloc_t nva, uint32_t stat)
	{
	bzero(priv, sizeof (nvpriv_t));

	priv->nvp_nva = nva;
	priv->nvp_stat = stat;
	}

	static nvpriv_t *
	nv_priv_alloc(nv_alloc_t *nva)
	{
	nvpriv_t *priv;

	/*
	* nv_mem_alloc() cannot called here because it needs the priv
	* argument.
	*/
	if ((priv = nva->nva_ops->nv_ao_alloc(nva, sizeof (nvpriv_t))) == NULL)
	return (NULL);

	nv_priv_init(priv, nva, 0);

	return (priv);
	}

	/*
	* Embedded lists need their own nvpriv_t's. We create a new
	* nvpriv_t using the parameters and allocator from the parent
	* list's nvpriv_t.
	*/
	static nvpriv_t *
	nv_priv_alloc_embedded(nvpriv_t *priv)
	{
	nvpriv_t *emb_priv;

	if ((emb_priv = nv_mem_zalloc(priv, sizeof (nvpriv_t))) == NULL)
	return (NULL);

	nv_priv_init(emb_priv, priv->nvp_nva, NV_STAT_EMBEDDED);

	return (emb_priv);
	}

	static int
	nvt_tab_alloc(nvpriv_t *priv, uint64_t buckets)
	{
	ASSERT3P(priv->nvp_hashtable, ==, NULL);
	ASSERT0(priv->nvp_nbuckets);
	ASSERT0(priv->nvp_nentries);

	i_nvp_t *tab = nv_mem_zalloc(priv, buckets sizeof (i_nvp_t *));
	if (tab == NULL)
	return (ENOMEM);

	priv->nvp_hashtable = tab;
	priv->nvp_nbuckets = buckets;
	return (0);
	}

	static void
	nvt_tab_free(nvpriv_t *priv)
	{
	i_nvp_t **tab = priv->nvp_hashtable;
	if (tab == NULL) {
	ASSERT0(priv->nvp_nbuckets);
	ASSERT0(priv->nvp_nentries);
	return;
	}

	nv_mem_free(priv, tab, priv->nvp_nbuckets * sizeof (i_nvp_t *));

	priv->nvp_hashtable = NULL;
	priv->nvp_nbuckets = 0;
	priv->nvp_nentries = 0;
	}

	static uint32_t
	nvt_hash(const char *p)
	{
	uint32_t g, hval = 0;

	while (*p) {
	hval = (hval << 4) + *p++;
	if ((g = (hval & 0xf0000000)) != 0)
	hval ^= g >> 24;
	hval &= ~g;
	}
	return (hval);
	}

	static boolean_t
	-nvt_nvpair_match(nvpair_t nvp1, nvpair_t nvp2, uint32_t nvflag)
	+nvt_nvpair_match(const nvpair_t nvp1, const nvpair_t nvp2, uint32_t nvflag)
	{
	boolean_t match = B_FALSE;
	if (nvflag & NV_UNIQUE_NAME_TYPE) {
	if (strcmp(NVP_NAME(nvp1), NVP_NAME(nvp2)) == 0 &&
	NVP_TYPE(nvp1) == NVP_TYPE(nvp2))
	match = B_TRUE;
	} else {
	ASSERT(nvflag == 0 \|\| nvflag & NV_UNIQUE_NAME);
	if (strcmp(NVP_NAME(nvp1), NVP_NAME(nvp2)) == 0)
	match = B_TRUE;
	}
	return (match);
	}

	static nvpair_t *
	-nvt_lookup_name_type(nvlist_t nvl, const char name, data_type_t type)
	+nvt_lookup_name_type(const nvlist_t nvl, const char name, data_type_t type)
	{
	- nvpriv_t priv = (nvpriv_t )(uintptr_t)nvl->nvl_priv;
	+ const nvpriv_t priv = (const nvpriv_t )(uintptr_t)nvl->nvl_priv;
	ASSERT(priv != NULL);

	i_nvp_t **tab = priv->nvp_hashtable;

	if (tab == NULL) {
	ASSERT3P(priv->nvp_list, ==, NULL);
	ASSERT0(priv->nvp_nbuckets);
	ASSERT0(priv->nvp_nentries);
	return (NULL);
	} else {
	ASSERT(priv->nvp_nbuckets != 0);
	}

	uint64_t hash = nvt_hash(name);
	uint64_t index = hash & (priv->nvp_nbuckets - 1);

	ASSERT3U(index, <, priv->nvp_nbuckets);
	i_nvp_t *entry = tab[index];

	for (i_nvp_t *e = entry; e != NULL; e = e->nvi_hashtable_next) {
	if (strcmp(NVP_NAME(&e->nvi_nvp), name) == 0 &&
	(type == DATA_TYPE_DONTCARE \|\|
	NVP_TYPE(&e->nvi_nvp) == type))
	return (&e->nvi_nvp);
	}
	return (NULL);
	}

	static nvpair_t *
	-nvt_lookup_name(nvlist_t nvl, const char name)
	+nvt_lookup_name(const nvlist_t nvl, const char name)
	{
	return (nvt_lookup_name_type(nvl, name, DATA_TYPE_DONTCARE));
	}

	static int
	nvt_resize(nvpriv_t *priv, uint32_t new_size)
	{
	i_nvp_t **tab = priv->nvp_hashtable;

	/*
	* Migrate all the entries from the current table
	* to a newly-allocated table with the new size by
	* re-adjusting the pointers of their entries.
	*/
	uint32_t size = priv->nvp_nbuckets;
	uint32_t new_mask = new_size - 1;
	ASSERT(ISP2(new_size));

	i_nvp_t *new_tab = nv_mem_zalloc(priv, new_size sizeof (i_nvp_t *));
	if (new_tab == NULL)
	return (ENOMEM);

	uint32_t nentries = 0;
	for (uint32_t i = 0; i < size; i++) {
	i_nvp_t next, e = tab[i];

	while (e != NULL) {
	next = e->nvi_hashtable_next;

	uint32_t hash = nvt_hash(NVP_NAME(&e->nvi_nvp));
	uint32_t index = hash & new_mask;

	e->nvi_hashtable_next = new_tab[index];
	new_tab[index] = e;
	nentries++;

	e = next;
	}
	tab[i] = NULL;
	}
	ASSERT3U(nentries, ==, priv->nvp_nentries);

	nvt_tab_free(priv);

	priv->nvp_hashtable = new_tab;
	priv->nvp_nbuckets = new_size;
	priv->nvp_nentries = nentries;

	return (0);
	}

	static boolean_t
	nvt_needs_togrow(nvpriv_t *priv)
	{
	/*
	* Grow only when we have more elements than buckets
	* and the # of buckets doesn't overflow.
	*/
	return (priv->nvp_nentries > priv->nvp_nbuckets &&
	(UINT32_MAX >> 1) >= priv->nvp_nbuckets);
	}

	/*
	* Allocate a new table that's twice the size of the old one,
	* and migrate all the entries from the old one to the new
	* one by re-adjusting their pointers.
	*/
	static int
	nvt_grow(nvpriv_t *priv)
	{
	uint32_t current_size = priv->nvp_nbuckets;
	/* ensure we won't overflow */
	ASSERT3U(UINT32_MAX >> 1, >=, current_size);
	return (nvt_resize(priv, current_size << 1));
	}

	static boolean_t
	nvt_needs_toshrink(nvpriv_t *priv)
	{
	/*
	* Shrink only when the # of elements is less than or
	* equal to 1/4 the # of buckets. Never shrink less than
	* nvlist_hashtable_init_size.
	*/
	ASSERT3U(priv->nvp_nbuckets, >=, nvlist_hashtable_init_size);
	if (priv->nvp_nbuckets == nvlist_hashtable_init_size)
	return (B_FALSE);
	return (priv->nvp_nentries <= (priv->nvp_nbuckets >> 2));
	}

	/*
	* Allocate a new table that's half the size of the old one,
	* and migrate all the entries from the old one to the new
	* one by re-adjusting their pointers.
	*/
	static int
	nvt_shrink(nvpriv_t *priv)
	{
	uint32_t current_size = priv->nvp_nbuckets;
	/* ensure we won't overflow */
	ASSERT3U(current_size, >=, nvlist_hashtable_init_size);
	return (nvt_resize(priv, current_size >> 1));
	}

	static int
	-nvt_remove_nvpair(nvlist_t nvl, nvpair_t nvp)
	+nvt_remove_nvpair(nvlist_t nvl, const nvpair_t nvp)
	{
	nvpriv_t priv = (nvpriv_t )(uintptr_t)nvl->nvl_priv;

	if (nvt_needs_toshrink(priv)) {
	int err = nvt_shrink(priv);
	if (err != 0)
	return (err);
	}
	i_nvp_t **tab = priv->nvp_hashtable;

	char *name = NVP_NAME(nvp);
	uint64_t hash = nvt_hash(name);
	uint64_t index = hash & (priv->nvp_nbuckets - 1);

	ASSERT3U(index, <, priv->nvp_nbuckets);
	i_nvp_t *bucket = tab[index];

	for (i_nvp_t prev = NULL, e = bucket;
	e != NULL; prev = e, e = e->nvi_hashtable_next) {
	if (nvt_nvpair_match(&e->nvi_nvp, nvp, nvl->nvl_nvflag)) {
	if (prev != NULL) {
	prev->nvi_hashtable_next =
	e->nvi_hashtable_next;
	} else {
	ASSERT3P(e, ==, bucket);
	tab[index] = e->nvi_hashtable_next;
	}
	e->nvi_hashtable_next = NULL;
	priv->nvp_nentries--;
	break;
	}
	}

	return (0);
	}

	static int
	nvt_add_nvpair(nvlist_t nvl, nvpair_t nvp)
	{
	nvpriv_t priv = (nvpriv_t )(uintptr_t)nvl->nvl_priv;

	/* initialize nvpair table now if it doesn't exist. */
	if (priv->nvp_hashtable == NULL) {
	int err = nvt_tab_alloc(priv, nvlist_hashtable_init_size);
	if (err != 0)
	return (err);
	}

	/*
	* if we don't allow duplicate entries, make sure to
	* unlink any existing entries from the table.
	*/
	if (nvl->nvl_nvflag != 0) {
	int err = nvt_remove_nvpair(nvl, nvp);
	if (err != 0)
	return (err);
	}

	if (nvt_needs_togrow(priv)) {
	int err = nvt_grow(priv);
	if (err != 0)
	return (err);
	}
	i_nvp_t **tab = priv->nvp_hashtable;

	char *name = NVP_NAME(nvp);
	uint64_t hash = nvt_hash(name);
	uint64_t index = hash & (priv->nvp_nbuckets - 1);

	ASSERT3U(index, <, priv->nvp_nbuckets);
	// cppcheck-suppress nullPointerRedundantCheck
	i_nvp_t *bucket = tab[index];

	/* insert link at the beginning of the bucket */
	i_nvp_t *new_entry = NVPAIR2I_NVP(nvp);
	ASSERT3P(new_entry->nvi_hashtable_next, ==, NULL);
	new_entry->nvi_hashtable_next = bucket;
	// cppcheck-suppress nullPointerRedundantCheck
	tab[index] = new_entry;

	priv->nvp_nentries++;
	return (0);
	}

	static void
	nvlist_init(nvlist_t nvl, uint32_t nvflag, nvpriv_t priv)
	{
	nvl->nvl_version = NV_VERSION;
	nvl->nvl_nvflag = nvflag & (NV_UNIQUE_NAME\|NV_UNIQUE_NAME_TYPE);
	nvl->nvl_priv = (uint64_t)(uintptr_t)priv;
	nvl->nvl_flag = 0;
	nvl->nvl_pad = 0;
	}

	uint_t
	nvlist_nvflag(nvlist_t *nvl)
	{
	return (nvl->nvl_nvflag);
	}

	static nv_alloc_t *
	nvlist_nv_alloc(int kmflag)
	{
	#if defined(_KERNEL)
	switch (kmflag) {
	case KM_SLEEP:
	return (nv_alloc_sleep);
	case KM_NOSLEEP:
	return (nv_alloc_nosleep);
	default:
	return (nv_alloc_pushpage);
	}
	#else
	return (nv_alloc_nosleep);
	#endif /* _KERNEL */
	}

	/*
	* nvlist_alloc - Allocate nvlist.
	*/
	int
	nvlist_alloc(nvlist_t **nvlp, uint_t nvflag, int kmflag)
	{
	return (nvlist_xalloc(nvlp, nvflag, nvlist_nv_alloc(kmflag)));
	}

	int
	nvlist_xalloc(nvlist_t *nvlp, uint_t nvflag, nv_alloc_t nva)
	{
	nvpriv_t *priv;

	if (nvlp == NULL \|\| nva == NULL)
	return (EINVAL);

	if ((priv = nv_priv_alloc(nva)) == NULL)
	return (ENOMEM);

	if ((*nvlp = nv_mem_zalloc(priv,
	NV_ALIGN(sizeof (nvlist_t)))) == NULL) {
	nv_mem_free(priv, priv, sizeof (nvpriv_t));
	return (ENOMEM);
	}

	nvlist_init(*nvlp, nvflag, priv);

	return (0);
	}

	/*
	* nvp_buf_alloc - Allocate i_nvp_t for storing a new nv pair.
	*/
	static nvpair_t *
	nvp_buf_alloc(nvlist_t *nvl, size_t len)
	{
	nvpriv_t priv = (nvpriv_t )(uintptr_t)nvl->nvl_priv;
	i_nvp_t *buf;
	nvpair_t *nvp;
	size_t nvsize;

	/*
	* Allocate the buffer
	*/
	nvsize = len + offsetof(i_nvp_t, nvi_nvp);

	if ((buf = nv_mem_zalloc(priv, nvsize)) == NULL)
	return (NULL);

	nvp = &buf->nvi_nvp;
	nvp->nvp_size = len;

	return (nvp);
	}

	/*
	* nvp_buf_free - de-Allocate an i_nvp_t.
	*/
	static void
	nvp_buf_free(nvlist_t nvl, nvpair_t nvp)
	{
	nvpriv_t priv = (nvpriv_t )(uintptr_t)nvl->nvl_priv;
	size_t nvsize = nvp->nvp_size + offsetof(i_nvp_t, nvi_nvp);

	nv_mem_free(priv, NVPAIR2I_NVP(nvp), nvsize);
	}

	/*
	* nvp_buf_link - link a new nv pair into the nvlist.
	*/
	static void
	nvp_buf_link(nvlist_t nvl, nvpair_t nvp)
	{
	nvpriv_t priv = (nvpriv_t )(uintptr_t)nvl->nvl_priv;
	i_nvp_t *curr = NVPAIR2I_NVP(nvp);

	/* Put element at end of nvlist */
	if (priv->nvp_list == NULL) {
	priv->nvp_list = priv->nvp_last = curr;
	} else {
	curr->nvi_prev = priv->nvp_last;
	priv->nvp_last->nvi_next = curr;
	priv->nvp_last = curr;
	}
	}

	/*
	* nvp_buf_unlink - unlink an removed nvpair out of the nvlist.
	*/
	static void
	nvp_buf_unlink(nvlist_t nvl, nvpair_t nvp)
	{
	nvpriv_t priv = (nvpriv_t )(uintptr_t)nvl->nvl_priv;
	i_nvp_t *curr = NVPAIR2I_NVP(nvp);

	/*
	* protect nvlist_next_nvpair() against walking on freed memory.
	*/
	if (priv->nvp_curr == curr)
	priv->nvp_curr = curr->nvi_next;

	if (curr == priv->nvp_list)
	priv->nvp_list = curr->nvi_next;
	else
	curr->nvi_prev->nvi_next = curr->nvi_next;

	if (curr == priv->nvp_last)
	priv->nvp_last = curr->nvi_prev;
	else
	curr->nvi_next->nvi_prev = curr->nvi_prev;
	}

	/*
	* take a nvpair type and number of elements and make sure the are valid
	*/
	static int
	i_validate_type_nelem(data_type_t type, uint_t nelem)
	{
	switch (type) {
	case DATA_TYPE_BOOLEAN:
	if (nelem != 0)
	return (EINVAL);
	break;
	case DATA_TYPE_BOOLEAN_VALUE:
	case DATA_TYPE_BYTE:
	case DATA_TYPE_INT8:
	case DATA_TYPE_UINT8:
	case DATA_TYPE_INT16:
	case DATA_TYPE_UINT16:
	case DATA_TYPE_INT32:
	case DATA_TYPE_UINT32:
	case DATA_TYPE_INT64:
	case DATA_TYPE_UINT64:
	case DATA_TYPE_STRING:
	case DATA_TYPE_HRTIME:
	case DATA_TYPE_NVLIST:
	#if !defined(_KERNEL)
	case DATA_TYPE_DOUBLE:
	#endif
	if (nelem != 1)
	return (EINVAL);
	break;
	case DATA_TYPE_BOOLEAN_ARRAY:
	case DATA_TYPE_BYTE_ARRAY:
	case DATA_TYPE_INT8_ARRAY:
	case DATA_TYPE_UINT8_ARRAY:
	case DATA_TYPE_INT16_ARRAY:
	case DATA_TYPE_UINT16_ARRAY:
	case DATA_TYPE_INT32_ARRAY:
	case DATA_TYPE_UINT32_ARRAY:
	case DATA_TYPE_INT64_ARRAY:
	case DATA_TYPE_UINT64_ARRAY:
	case DATA_TYPE_STRING_ARRAY:
	case DATA_TYPE_NVLIST_ARRAY:
	/* we allow arrays with 0 elements */
	break;
	default:
	return (EINVAL);
	}
	return (0);
	}

	/*
	* Verify nvp_name_sz and check the name string length.
	*/
	static int
	i_validate_nvpair_name(nvpair_t *nvp)
	{
	if ((nvp->nvp_name_sz <= 0) \|\|
	(nvp->nvp_size < NVP_SIZE_CALC(nvp->nvp_name_sz, 0)))
	return (EFAULT);

	/* verify the name string, make sure its terminated */
	if (NVP_NAME(nvp)[nvp->nvp_name_sz - 1] != '\0')
	return (EFAULT);

	return (strlen(NVP_NAME(nvp)) == nvp->nvp_name_sz - 1 ? 0 : EFAULT);
	}

	static int
	i_validate_nvpair_value(data_type_t type, uint_t nelem, const void *data)
	{
	switch (type) {
	case DATA_TYPE_BOOLEAN_VALUE:
	if ((boolean_t )data != B_TRUE &&
	(boolean_t )data != B_FALSE)
	return (EINVAL);
	break;
	case DATA_TYPE_BOOLEAN_ARRAY: {
	int i;

	for (i = 0; i < nelem; i++)
	if (((boolean_t *)data)[i] != B_TRUE &&
	((boolean_t *)data)[i] != B_FALSE)
	return (EINVAL);
	break;
	}
	default:
	break;
	}

	return (0);
	}

	/*
	* This function takes a pointer to what should be a nvpair and it's size
	* and then verifies that all the nvpair fields make sense and can be
	* trusted. This function is used when decoding packed nvpairs.
	*/
	static int
	i_validate_nvpair(nvpair_t *nvp)
	{
	data_type_t type = NVP_TYPE(nvp);
	int size1, size2;

	/* verify nvp_name_sz, check the name string length */
	if (i_validate_nvpair_name(nvp) != 0)
	return (EFAULT);

	if (i_validate_nvpair_value(type, NVP_NELEM(nvp), NVP_VALUE(nvp)) != 0)
	return (EFAULT);

	/*
	* verify nvp_type, nvp_value_elem, and also possibly
	* verify string values and get the value size.
	*/
	size2 = i_get_value_size(type, NVP_VALUE(nvp), NVP_NELEM(nvp));
	size1 = nvp->nvp_size - NVP_VALOFF(nvp);
	if (size2 < 0 \|\| size1 != NV_ALIGN(size2))
	return (EFAULT);

	return (0);
	}

	static int
	-nvlist_copy_pairs(nvlist_t snvl, nvlist_t dnvl)
	+nvlist_copy_pairs(const nvlist_t snvl, nvlist_t dnvl)
	{
	- nvpriv_t *priv;
	- i_nvp_t *curr;
	+ const nvpriv_t *priv;
	+ const i_nvp_t *curr;

	- if ((priv = (nvpriv_t *)(uintptr_t)snvl->nvl_priv) == NULL)
	+ if ((priv = (const nvpriv_t *)(uintptr_t)snvl->nvl_priv) == NULL)
	return (EINVAL);

	for (curr = priv->nvp_list; curr != NULL; curr = curr->nvi_next) {
	- nvpair_t *nvp = &curr->nvi_nvp;
	+ const nvpair_t *nvp = &curr->nvi_nvp;
	int err;

	if ((err = nvlist_add_common(dnvl, NVP_NAME(nvp), NVP_TYPE(nvp),
	NVP_NELEM(nvp), NVP_VALUE(nvp))) != 0)
	return (err);
	}

	return (0);
	}

	/*
	* Frees all memory allocated for an nvpair (like embedded lists) with
	* the exception of the nvpair buffer itself.
	*/
	static void
	nvpair_free(nvpair_t *nvp)
	{
	switch (NVP_TYPE(nvp)) {
	case DATA_TYPE_NVLIST:
	nvlist_free(EMBEDDED_NVL(nvp));
	break;
	case DATA_TYPE_NVLIST_ARRAY: {
	nvlist_t **nvlp = EMBEDDED_NVL_ARRAY(nvp);
	int i;

	for (i = 0; i < NVP_NELEM(nvp); i++)
	if (nvlp[i] != NULL)
	nvlist_free(nvlp[i]);
	break;
	}
	default:
	break;
	}
	}

	/*
	* nvlist_free - free an unpacked nvlist
	*/
	void
	nvlist_free(nvlist_t *nvl)
	{
	nvpriv_t *priv;
	i_nvp_t *curr;

	if (nvl == NULL \|\|
	(priv = (nvpriv_t *)(uintptr_t)nvl->nvl_priv) == NULL)
	return;

	/*
	* Unpacked nvlist are linked through i_nvp_t
	*/
	curr = priv->nvp_list;
	while (curr != NULL) {
	nvpair_t *nvp = &curr->nvi_nvp;
	curr = curr->nvi_next;

	nvpair_free(nvp);
	nvp_buf_free(nvl, nvp);
	}

	if (!(priv->nvp_stat & NV_STAT_EMBEDDED))
	nv_mem_free(priv, nvl, NV_ALIGN(sizeof (nvlist_t)));
	else
	nvl->nvl_priv = 0;

	nvt_tab_free(priv);
	nv_mem_free(priv, priv, sizeof (nvpriv_t));
	}

	static int
	-nvlist_contains_nvp(nvlist_t nvl, nvpair_t nvp)
	+nvlist_contains_nvp(const nvlist_t nvl, const nvpair_t nvp)
	{
	- nvpriv_t priv = (nvpriv_t )(uintptr_t)nvl->nvl_priv;
	- i_nvp_t *curr;
	+ const nvpriv_t priv = (const nvpriv_t )(uintptr_t)nvl->nvl_priv;
	+ const i_nvp_t *curr;

	if (nvp == NULL)
	return (0);

	for (curr = priv->nvp_list; curr != NULL; curr = curr->nvi_next)
	if (&curr->nvi_nvp == nvp)
	return (1);

	return (0);
	}

	/*
	* Make a copy of nvlist
	*/
	int
	-nvlist_dup(nvlist_t nvl, nvlist_t *nvlp, int kmflag)
	+nvlist_dup(const nvlist_t nvl, nvlist_t *nvlp, int kmflag)
	{
	return (nvlist_xdup(nvl, nvlp, nvlist_nv_alloc(kmflag)));
	}

	int
	-nvlist_xdup(nvlist_t nvl, nvlist_t nvlp, nv_alloc_t nva)
	+nvlist_xdup(const nvlist_t nvl, nvlist_t nvlp, nv_alloc_t nva)
	{
	int err;
	nvlist_t *ret;

	if (nvl == NULL \|\| nvlp == NULL)
	return (EINVAL);

	if ((err = nvlist_xalloc(&ret, nvl->nvl_nvflag, nva)) != 0)
	return (err);

	if ((err = nvlist_copy_pairs(nvl, ret)) != 0)
	nvlist_free(ret);
	else
	*nvlp = ret;

	return (err);
	}

	/*
	* Remove all with matching name
	*/
	int
	nvlist_remove_all(nvlist_t nvl, const char name)
	{
	int error = ENOENT;

	if (nvl == NULL \|\| name == NULL \|\| nvl->nvl_priv == 0)
	return (EINVAL);

	nvpair_t *nvp;
	while ((nvp = nvt_lookup_name(nvl, name)) != NULL) {
	VERIFY0(nvlist_remove_nvpair(nvl, nvp));
	error = 0;
	}

	return (error);
	}

	/*
	* Remove first one with matching name and type
	*/
	int
	nvlist_remove(nvlist_t nvl, const char name, data_type_t type)
	{
	if (nvl == NULL \|\| name == NULL \|\| nvl->nvl_priv == 0)
	return (EINVAL);

	nvpair_t *nvp = nvt_lookup_name_type(nvl, name, type);
	if (nvp == NULL)
	return (ENOENT);

	return (nvlist_remove_nvpair(nvl, nvp));
	}

	int
	nvlist_remove_nvpair(nvlist_t nvl, nvpair_t nvp)
	{
	if (nvl == NULL \|\| nvp == NULL)
	return (EINVAL);

	int err = nvt_remove_nvpair(nvl, nvp);
	if (err != 0)
	return (err);

	nvp_buf_unlink(nvl, nvp);
	nvpair_free(nvp);
	nvp_buf_free(nvl, nvp);
	return (0);
	}

	/*
	* This function calculates the size of an nvpair value.
	*
	* The data argument controls the behavior in case of the data types
	* DATA_TYPE_STRING and
	* DATA_TYPE_STRING_ARRAY
	* Is data == NULL then the size of the string(s) is excluded.
	*/
	static int
	i_get_value_size(data_type_t type, const void *data, uint_t nelem)
	{
	uint64_t value_sz;

	if (i_validate_type_nelem(type, nelem) != 0)
	return (-1);

	/* Calculate required size for holding value */
	switch (type) {
	case DATA_TYPE_BOOLEAN:
	value_sz = 0;
	break;
	case DATA_TYPE_BOOLEAN_VALUE:
	value_sz = sizeof (boolean_t);
	break;
	case DATA_TYPE_BYTE:
	value_sz = sizeof (uchar_t);
	break;
	case DATA_TYPE_INT8:
	value_sz = sizeof (int8_t);
	break;
	case DATA_TYPE_UINT8:
	value_sz = sizeof (uint8_t);
	break;
	case DATA_TYPE_INT16:
	value_sz = sizeof (int16_t);
	break;
	case DATA_TYPE_UINT16:
	value_sz = sizeof (uint16_t);
	break;
	case DATA_TYPE_INT32:
	value_sz = sizeof (int32_t);
	break;
	case DATA_TYPE_UINT32:
	value_sz = sizeof (uint32_t);
	break;
	case DATA_TYPE_INT64:
	value_sz = sizeof (int64_t);
	break;
	case DATA_TYPE_UINT64:
	value_sz = sizeof (uint64_t);
	break;
	#if !defined(_KERNEL)
	case DATA_TYPE_DOUBLE:
	value_sz = sizeof (double);
	break;
	#endif
	case DATA_TYPE_STRING:
	if (data == NULL)
	value_sz = 0;
	else
	value_sz = strlen(data) + 1;
	break;
	case DATA_TYPE_BOOLEAN_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (boolean_t);
	break;
	case DATA_TYPE_BYTE_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (uchar_t);
	break;
	case DATA_TYPE_INT8_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (int8_t);
	break;
	case DATA_TYPE_UINT8_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (uint8_t);
	break;
	case DATA_TYPE_INT16_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (int16_t);
	break;
	case DATA_TYPE_UINT16_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (uint16_t);
	break;
	case DATA_TYPE_INT32_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (int32_t);
	break;
	case DATA_TYPE_UINT32_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (uint32_t);
	break;
	case DATA_TYPE_INT64_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (int64_t);
	break;
	case DATA_TYPE_UINT64_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (uint64_t);
	break;
	case DATA_TYPE_STRING_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (uint64_t);

	if (data != NULL) {
	char const strs = data;
	uint_t i;

	/* no alignment requirement for strings */
	for (i = 0; i < nelem; i++) {
	if (strs[i] == NULL)
	return (-1);
	value_sz += strlen(strs[i]) + 1;
	}
	}
	break;
	case DATA_TYPE_HRTIME:
	value_sz = sizeof (hrtime_t);
	break;
	case DATA_TYPE_NVLIST:
	value_sz = NV_ALIGN(sizeof (nvlist_t));
	break;
	case DATA_TYPE_NVLIST_ARRAY:
	value_sz = (uint64_t)nelem * sizeof (uint64_t) +
	(uint64_t)nelem * NV_ALIGN(sizeof (nvlist_t));
	break;
	default:
	return (-1);
	}

	return (value_sz > INT32_MAX ? -1 : (int)value_sz);
	}

	static int
	nvlist_copy_embedded(nvlist_t nvl, nvlist_t onvl, nvlist_t *emb_nvl)
	{
	nvpriv_t *priv;
	int err;

	if ((priv = nv_priv_alloc_embedded((nvpriv_t *)(uintptr_t)
	nvl->nvl_priv)) == NULL)
	return (ENOMEM);

	nvlist_init(emb_nvl, onvl->nvl_nvflag, priv);

	if ((err = nvlist_copy_pairs(onvl, emb_nvl)) != 0) {
	nvlist_free(emb_nvl);
	emb_nvl->nvl_priv = 0;
	}

	return (err);
	}

	/*
	* nvlist_add_common - Add new <name,value> pair to nvlist
	*/
	static int
	nvlist_add_common(nvlist_t nvl, const char name,
	data_type_t type, uint_t nelem, const void *data)
	{
	nvpair_t *nvp;
	uint_t i;

	int nvp_sz, name_sz, value_sz;
	int err = 0;

	if (name == NULL \|\| nvl == NULL \|\| nvl->nvl_priv == 0)
	return (EINVAL);

	if (nelem != 0 && data == NULL)
	return (EINVAL);

	/*
	* Verify type and nelem and get the value size.
	* In case of data types DATA_TYPE_STRING and DATA_TYPE_STRING_ARRAY
	* is the size of the string(s) included.
	*/
	if ((value_sz = i_get_value_size(type, data, nelem)) < 0)
	return (EINVAL);

	if (i_validate_nvpair_value(type, nelem, data) != 0)
	return (EINVAL);

	/*
	* If we're adding an nvlist or nvlist array, ensure that we are not
	* adding the input nvlist to itself, which would cause recursion,
	* and ensure that no NULL nvlist pointers are present.
	*/
	switch (type) {
	case DATA_TYPE_NVLIST:
	if (data == nvl \|\| data == NULL)
	return (EINVAL);
	break;
	case DATA_TYPE_NVLIST_ARRAY: {
	nvlist_t onvlp = (nvlist_t )data;
	for (i = 0; i < nelem; i++) {
	if (onvlp[i] == nvl \|\| onvlp[i] == NULL)
	return (EINVAL);
	}
	break;
	}
	default:
	break;
	}

	/* calculate sizes of the nvpair elements and the nvpair itself */
	name_sz = strlen(name) + 1;
	if (name_sz >= 1ULL << (sizeof (nvp->nvp_name_sz) * NBBY - 1))
	return (EINVAL);

	nvp_sz = NVP_SIZE_CALC(name_sz, value_sz);

	if ((nvp = nvp_buf_alloc(nvl, nvp_sz)) == NULL)
	return (ENOMEM);

	ASSERT(nvp->nvp_size == nvp_sz);
	nvp->nvp_name_sz = name_sz;
	nvp->nvp_value_elem = nelem;
	nvp->nvp_type = type;
	bcopy(name, NVP_NAME(nvp), name_sz);

	switch (type) {
	case DATA_TYPE_BOOLEAN:
	break;
	case DATA_TYPE_STRING_ARRAY: {
	char const strs = data;
	char *buf = NVP_VALUE(nvp);
	char *cstrs = (void )buf;

	/* skip pre-allocated space for pointer array */
	buf += nelem * sizeof (uint64_t);
	for (i = 0; i < nelem; i++) {
	int slen = strlen(strs[i]) + 1;
	bcopy(strs[i], buf, slen);
	cstrs[i] = buf;
	buf += slen;
	}
	break;
	}
	case DATA_TYPE_NVLIST: {
	nvlist_t *nnvl = EMBEDDED_NVL(nvp);
	nvlist_t onvl = (nvlist_t )data;

	if ((err = nvlist_copy_embedded(nvl, onvl, nnvl)) != 0) {
	nvp_buf_free(nvl, nvp);
	return (err);
	}
	break;
	}
	case DATA_TYPE_NVLIST_ARRAY: {
	nvlist_t onvlp = (nvlist_t )data;
	nvlist_t **nvlp = EMBEDDED_NVL_ARRAY(nvp);
	nvlist_t embedded = (nvlist_t )
	((uintptr_t)nvlp + nelem * sizeof (uint64_t));

	for (i = 0; i < nelem; i++) {
	if ((err = nvlist_copy_embedded(nvl,
	onvlp[i], embedded)) != 0) {
	/*
	* Free any successfully created lists
	*/
	nvpair_free(nvp);
	nvp_buf_free(nvl, nvp);
	return (err);
	}

	nvlp[i] = embedded++;
	}
	break;
	}
	default:
	bcopy(data, NVP_VALUE(nvp), value_sz);
	}

	/* if unique name, remove before add */
	if (nvl->nvl_nvflag & NV_UNIQUE_NAME)
	(void) nvlist_remove_all(nvl, name);
	else if (nvl->nvl_nvflag & NV_UNIQUE_NAME_TYPE)
	(void) nvlist_remove(nvl, name, type);

	err = nvt_add_nvpair(nvl, nvp);
	if (err != 0) {
	nvpair_free(nvp);
	nvp_buf_free(nvl, nvp);
	return (err);
	}
	nvp_buf_link(nvl, nvp);

	return (0);
	}

	int
	nvlist_add_boolean(nvlist_t nvl, const char name)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_BOOLEAN, 0, NULL));
	}

	int
	nvlist_add_boolean_value(nvlist_t nvl, const char name, boolean_t val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_BOOLEAN_VALUE, 1, &val));
	}

	int
	nvlist_add_byte(nvlist_t nvl, const char name, uchar_t val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_BYTE, 1, &val));
	}

	int
	nvlist_add_int8(nvlist_t nvl, const char name, int8_t val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_INT8, 1, &val));
	}

	int
	nvlist_add_uint8(nvlist_t nvl, const char name, uint8_t val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_UINT8, 1, &val));
	}

	int
	nvlist_add_int16(nvlist_t nvl, const char name, int16_t val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_INT16, 1, &val));
	}

	int
	nvlist_add_uint16(nvlist_t nvl, const char name, uint16_t val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_UINT16, 1, &val));
	}

	int
	nvlist_add_int32(nvlist_t nvl, const char name, int32_t val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_INT32, 1, &val));
	}

	int
	nvlist_add_uint32(nvlist_t nvl, const char name, uint32_t val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_UINT32, 1, &val));
	}

	int
	nvlist_add_int64(nvlist_t nvl, const char name, int64_t val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_INT64, 1, &val));
	}

	int
	nvlist_add_uint64(nvlist_t nvl, const char name, uint64_t val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_UINT64, 1, &val));
	}

	#if !defined(_KERNEL)
	int
	nvlist_add_double(nvlist_t nvl, const char name, double val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_DOUBLE, 1, &val));
	}
	#endif

	int
	nvlist_add_string(nvlist_t nvl, const char name, const char *val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_STRING, 1, (void *)val));
	}

	int
	nvlist_add_boolean_array(nvlist_t nvl, const char name,
	- boolean_t *a, uint_t n)
	+ const boolean_t *a, uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_BOOLEAN_ARRAY, n, a));
	}

	int
	-nvlist_add_byte_array(nvlist_t nvl, const char name, uchar_t *a, uint_t n)
	+nvlist_add_byte_array(nvlist_t nvl, const char name, const uchar_t *a,
	+ uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_BYTE_ARRAY, n, a));
	}

	int
	-nvlist_add_int8_array(nvlist_t nvl, const char name, int8_t *a, uint_t n)
	+nvlist_add_int8_array(nvlist_t nvl, const char name, const int8_t *a,
	+ uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_INT8_ARRAY, n, a));
	}

	int
	-nvlist_add_uint8_array(nvlist_t nvl, const char name, uint8_t *a, uint_t n)
	+nvlist_add_uint8_array(nvlist_t nvl, const char name, const uint8_t *a,
	+ uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_UINT8_ARRAY, n, a));
	}

	int
	-nvlist_add_int16_array(nvlist_t nvl, const char name, int16_t *a, uint_t n)
	+nvlist_add_int16_array(nvlist_t nvl, const char name, const int16_t *a,
	+ uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_INT16_ARRAY, n, a));
	}

	int
	-nvlist_add_uint16_array(nvlist_t nvl, const char name, uint16_t *a, uint_t n)
	+nvlist_add_uint16_array(nvlist_t nvl, const char name, const uint16_t *a,
	+ uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_UINT16_ARRAY, n, a));
	}

	int
	-nvlist_add_int32_array(nvlist_t nvl, const char name, int32_t *a, uint_t n)
	+nvlist_add_int32_array(nvlist_t nvl, const char name, const int32_t *a,
	+ uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_INT32_ARRAY, n, a));
	}

	int
	-nvlist_add_uint32_array(nvlist_t nvl, const char name, uint32_t *a, uint_t n)
	+nvlist_add_uint32_array(nvlist_t nvl, const char name, const uint32_t *a,
	+ uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_UINT32_ARRAY, n, a));
	}

	int
	-nvlist_add_int64_array(nvlist_t nvl, const char name, int64_t *a, uint_t n)
	+nvlist_add_int64_array(nvlist_t nvl, const char name, const int64_t *a,
	+ uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_INT64_ARRAY, n, a));
	}

	int
	-nvlist_add_uint64_array(nvlist_t nvl, const char name, uint64_t *a, uint_t n)
	+nvlist_add_uint64_array(nvlist_t nvl, const char name, const uint64_t *a,
	+ uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_UINT64_ARRAY, n, a));
	}

	int
	nvlist_add_string_array(nvlist_t nvl, const char name,
	- char const a, uint_t n)
	+ const char const a, uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_STRING_ARRAY, n, a));
	}

	int
	nvlist_add_hrtime(nvlist_t nvl, const char name, hrtime_t val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_HRTIME, 1, &val));
	}

	int
	-nvlist_add_nvlist(nvlist_t nvl, const char name, nvlist_t *val)
	+nvlist_add_nvlist(nvlist_t nvl, const char name, const nvlist_t *val)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_NVLIST, 1, val));
	}

	int
	-nvlist_add_nvlist_array(nvlist_t nvl, const char name, nvlist_t **a, uint_t n)
	+nvlist_add_nvlist_array(nvlist_t nvl, const char name,
	+ const nvlist_t * const *a, uint_t n)
	{
	return (nvlist_add_common(nvl, name, DATA_TYPE_NVLIST_ARRAY, n, a));
	}

	/* reading name-value pairs */
	nvpair_t *
	-nvlist_next_nvpair(nvlist_t nvl, nvpair_t nvp)
	+nvlist_next_nvpair(nvlist_t nvl, const nvpair_t nvp)
	{
	nvpriv_t *priv;
	i_nvp_t *curr;

	if (nvl == NULL \|\|
	(priv = (nvpriv_t *)(uintptr_t)nvl->nvl_priv) == NULL)
	return (NULL);

	curr = NVPAIR2I_NVP(nvp);

	/*
	* Ensure that nvp is a valid nvpair on this nvlist.
	* NB: nvp_curr is used only as a hint so that we don't always
	* have to walk the list to determine if nvp is still on the list.
	*/
	if (nvp == NULL)
	curr = priv->nvp_list;
	else if (priv->nvp_curr == curr \|\| nvlist_contains_nvp(nvl, nvp))
	curr = curr->nvi_next;
	else
	curr = NULL;

	priv->nvp_curr = curr;

	return (curr != NULL ? &curr->nvi_nvp : NULL);
	}

	nvpair_t *
	-nvlist_prev_nvpair(nvlist_t nvl, nvpair_t nvp)
	+nvlist_prev_nvpair(nvlist_t nvl, const nvpair_t nvp)
	{
	nvpriv_t *priv;
	i_nvp_t *curr;

	if (nvl == NULL \|\|
	(priv = (nvpriv_t *)(uintptr_t)nvl->nvl_priv) == NULL)
	return (NULL);

	curr = NVPAIR2I_NVP(nvp);

	if (nvp == NULL)
	curr = priv->nvp_last;
	else if (priv->nvp_curr == curr \|\| nvlist_contains_nvp(nvl, nvp))
	curr = curr->nvi_prev;
	else
	curr = NULL;

	priv->nvp_curr = curr;

	return (curr != NULL ? &curr->nvi_nvp : NULL);
	}

	boolean_t
	-nvlist_empty(nvlist_t *nvl)
	+nvlist_empty(const nvlist_t *nvl)
	{
	- nvpriv_t *priv;
	+ const nvpriv_t *priv;

	if (nvl == NULL \|\|
	- (priv = (nvpriv_t *)(uintptr_t)nvl->nvl_priv) == NULL)
	+ (priv = (const nvpriv_t *)(uintptr_t)nvl->nvl_priv) == NULL)
	return (B_TRUE);

	return (priv->nvp_list == NULL);
	}

	char *
	-nvpair_name(nvpair_t *nvp)
	+nvpair_name(const nvpair_t *nvp)
	{
	return (NVP_NAME(nvp));
	}

	data_type_t
	-nvpair_type(nvpair_t *nvp)
	+nvpair_type(const nvpair_t *nvp)
	{
	return (NVP_TYPE(nvp));
	}

	int
	-nvpair_type_is_array(nvpair_t *nvp)
	+nvpair_type_is_array(const nvpair_t *nvp)
	{
	data_type_t type = NVP_TYPE(nvp);

	if ((type == DATA_TYPE_BYTE_ARRAY) \|\|
	(type == DATA_TYPE_INT8_ARRAY) \|\|
	(type == DATA_TYPE_UINT8_ARRAY) \|\|
	(type == DATA_TYPE_INT16_ARRAY) \|\|
	(type == DATA_TYPE_UINT16_ARRAY) \|\|
	(type == DATA_TYPE_INT32_ARRAY) \|\|
	(type == DATA_TYPE_UINT32_ARRAY) \|\|
	(type == DATA_TYPE_INT64_ARRAY) \|\|
	(type == DATA_TYPE_UINT64_ARRAY) \|\|
	(type == DATA_TYPE_BOOLEAN_ARRAY) \|\|
	(type == DATA_TYPE_STRING_ARRAY) \|\|
	(type == DATA_TYPE_NVLIST_ARRAY))
	return (1);
	return (0);

	}

	static int
	-nvpair_value_common(nvpair_t nvp, data_type_t type, uint_t nelem, void *data)
	+nvpair_value_common(const nvpair_t nvp, data_type_t type, uint_t nelem,
	+ void *data)
	{
	int value_sz;

	if (nvp == NULL \|\| nvpair_type(nvp) != type)
	return (EINVAL);

	/*
	* For non-array types, we copy the data.
	* For array types (including string), we set a pointer.
	*/
	switch (type) {
	case DATA_TYPE_BOOLEAN:
	if (nelem != NULL)
	*nelem = 0;
	break;

	case DATA_TYPE_BOOLEAN_VALUE:
	case DATA_TYPE_BYTE:
	case DATA_TYPE_INT8:
	case DATA_TYPE_UINT8:
	case DATA_TYPE_INT16:
	case DATA_TYPE_UINT16:
	case DATA_TYPE_INT32:
	case DATA_TYPE_UINT32:
	case DATA_TYPE_INT64:
	case DATA_TYPE_UINT64:
	case DATA_TYPE_HRTIME:
	#if !defined(_KERNEL)
	case DATA_TYPE_DOUBLE:
	#endif
	if (data == NULL)
	return (EINVAL);
	if ((value_sz = i_get_value_size(type, NULL, 1)) < 0)
	return (EINVAL);
	bcopy(NVP_VALUE(nvp), data, (size_t)value_sz);
	if (nelem != NULL)
	*nelem = 1;
	break;

	case DATA_TYPE_NVLIST:
	case DATA_TYPE_STRING:
	if (data == NULL)
	return (EINVAL);
	+ /*
	+ * This discards the const from nvp, so all callers for these
	+ * types must not accept const nvpairs.
	+ */
	(void )data = (void )NVP_VALUE(nvp);
	if (nelem != NULL)
	*nelem = 1;
	break;

	case DATA_TYPE_BOOLEAN_ARRAY:
	case DATA_TYPE_BYTE_ARRAY:
	case DATA_TYPE_INT8_ARRAY:
	case DATA_TYPE_UINT8_ARRAY:
	case DATA_TYPE_INT16_ARRAY:
	case DATA_TYPE_UINT16_ARRAY:
	case DATA_TYPE_INT32_ARRAY:
	case DATA_TYPE_UINT32_ARRAY:
	case DATA_TYPE_INT64_ARRAY:
	case DATA_TYPE_UINT64_ARRAY:
	case DATA_TYPE_STRING_ARRAY:
	case DATA_TYPE_NVLIST_ARRAY:
	if (nelem == NULL \|\| data == NULL)
	return (EINVAL);
	+ /*
	+ * This discards the const from nvp, so all callers for these
	+ * types must not accept const nvpairs.
	+ */
	if ((*nelem = NVP_NELEM(nvp)) != 0)
	(void )data = (void )NVP_VALUE(nvp);
	else
	(void *)data = NULL;
	break;

	default:
	return (ENOTSUP);
	}

	return (0);
	}

	static int
	-nvlist_lookup_common(nvlist_t nvl, const char name, data_type_t type,
	+nvlist_lookup_common(const nvlist_t nvl, const char name, data_type_t type,
	uint_t nelem, void data)
	{
	if (name == NULL \|\| nvl == NULL \|\| nvl->nvl_priv == 0)
	return (EINVAL);

	if (!(nvl->nvl_nvflag & (NV_UNIQUE_NAME \| NV_UNIQUE_NAME_TYPE)))
	return (ENOTSUP);

	nvpair_t *nvp = nvt_lookup_name_type(nvl, name, type);
	if (nvp == NULL)
	return (ENOENT);

	return (nvpair_value_common(nvp, type, nelem, data));
	}

	int
	-nvlist_lookup_boolean(nvlist_t nvl, const char name)
	+nvlist_lookup_boolean(const nvlist_t nvl, const char name)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_BOOLEAN, NULL, NULL));
	}

	int
	-nvlist_lookup_boolean_value(nvlist_t nvl, const char name, boolean_t *val)
	+nvlist_lookup_boolean_value(const nvlist_t nvl, const char name,
	+ boolean_t *val)
	{
	return (nvlist_lookup_common(nvl, name,
	DATA_TYPE_BOOLEAN_VALUE, NULL, val));
	}

	int
	-nvlist_lookup_byte(nvlist_t nvl, const char name, uchar_t *val)
	+nvlist_lookup_byte(const nvlist_t nvl, const char name, uchar_t *val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_BYTE, NULL, val));
	}

	int
	-nvlist_lookup_int8(nvlist_t nvl, const char name, int8_t *val)
	+nvlist_lookup_int8(const nvlist_t nvl, const char name, int8_t *val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_INT8, NULL, val));
	}

	int
	-nvlist_lookup_uint8(nvlist_t nvl, const char name, uint8_t *val)
	+nvlist_lookup_uint8(const nvlist_t nvl, const char name, uint8_t *val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_UINT8, NULL, val));
	}

	int
	-nvlist_lookup_int16(nvlist_t nvl, const char name, int16_t *val)
	+nvlist_lookup_int16(const nvlist_t nvl, const char name, int16_t *val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_INT16, NULL, val));
	}

	int
	-nvlist_lookup_uint16(nvlist_t nvl, const char name, uint16_t *val)
	+nvlist_lookup_uint16(const nvlist_t nvl, const char name, uint16_t *val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_UINT16, NULL, val));
	}

	int
	-nvlist_lookup_int32(nvlist_t nvl, const char name, int32_t *val)
	+nvlist_lookup_int32(const nvlist_t nvl, const char name, int32_t *val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_INT32, NULL, val));
	}

	int
	-nvlist_lookup_uint32(nvlist_t nvl, const char name, uint32_t *val)
	+nvlist_lookup_uint32(const nvlist_t nvl, const char name, uint32_t *val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_UINT32, NULL, val));
	}

	int
	-nvlist_lookup_int64(nvlist_t nvl, const char name, int64_t *val)
	+nvlist_lookup_int64(const nvlist_t nvl, const char name, int64_t *val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_INT64, NULL, val));
	}

	int
	-nvlist_lookup_uint64(nvlist_t nvl, const char name, uint64_t *val)
	+nvlist_lookup_uint64(const nvlist_t nvl, const char name, uint64_t *val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_UINT64, NULL, val));
	}

	#if !defined(_KERNEL)
	int
	-nvlist_lookup_double(nvlist_t nvl, const char name, double *val)
	+nvlist_lookup_double(const nvlist_t nvl, const char name, double *val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_DOUBLE, NULL, val));
	}
	#endif

	int
	nvlist_lookup_string(nvlist_t nvl, const char name, char **val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_STRING, NULL, val));
	}

	int
	nvlist_lookup_nvlist(nvlist_t nvl, const char name, nvlist_t **val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_NVLIST, NULL, val));
	}

	int
	nvlist_lookup_boolean_array(nvlist_t nvl, const char name,
	boolean_t *a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name,
	DATA_TYPE_BOOLEAN_ARRAY, n, a));
	}

	int
	nvlist_lookup_byte_array(nvlist_t nvl, const char name,
	uchar_t *a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_BYTE_ARRAY, n, a));
	}

	int
	nvlist_lookup_int8_array(nvlist_t nvl, const char name, int8_t *a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_INT8_ARRAY, n, a));
	}

	int
	nvlist_lookup_uint8_array(nvlist_t nvl, const char name,
	uint8_t *a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_UINT8_ARRAY, n, a));
	}

	int
	nvlist_lookup_int16_array(nvlist_t nvl, const char name,
	int16_t *a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_INT16_ARRAY, n, a));
	}

	int
	nvlist_lookup_uint16_array(nvlist_t nvl, const char name,
	uint16_t *a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_UINT16_ARRAY, n, a));
	}

	int
	nvlist_lookup_int32_array(nvlist_t nvl, const char name,
	int32_t *a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_INT32_ARRAY, n, a));
	}

	int
	nvlist_lookup_uint32_array(nvlist_t nvl, const char name,
	uint32_t *a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_UINT32_ARRAY, n, a));
	}

	int
	nvlist_lookup_int64_array(nvlist_t nvl, const char name,
	int64_t *a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_INT64_ARRAY, n, a));
	}

	int
	nvlist_lookup_uint64_array(nvlist_t nvl, const char name,
	uint64_t *a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_UINT64_ARRAY, n, a));
	}

	int
	nvlist_lookup_string_array(nvlist_t nvl, const char name,
	char **a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_STRING_ARRAY, n, a));
	}

	int
	nvlist_lookup_nvlist_array(nvlist_t nvl, const char name,
	nvlist_t **a, uint_t n)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_NVLIST_ARRAY, n, a));
	}

	int
	nvlist_lookup_hrtime(nvlist_t nvl, const char name, hrtime_t *val)
	{
	return (nvlist_lookup_common(nvl, name, DATA_TYPE_HRTIME, NULL, val));
	}

	int
	nvlist_lookup_pairs(nvlist_t *nvl, int flag, ...)
	{
	va_list ap;
	char *name;
	int noentok = (flag & NV_FLAG_NOENTOK ? 1 : 0);
	int ret = 0;

	va_start(ap, flag);
	while (ret == 0 && (name = va_arg(ap, char *)) != NULL) {
	data_type_t type;
	void *val;
	uint_t *nelem;

	switch (type = va_arg(ap, data_type_t)) {
	case DATA_TYPE_BOOLEAN:
	ret = nvlist_lookup_common(nvl, name, type, NULL, NULL);
	break;

	case DATA_TYPE_BOOLEAN_VALUE:
	case DATA_TYPE_BYTE:
	case DATA_TYPE_INT8:
	case DATA_TYPE_UINT8:
	case DATA_TYPE_INT16:
	case DATA_TYPE_UINT16:
	case DATA_TYPE_INT32:
	case DATA_TYPE_UINT32:
	case DATA_TYPE_INT64:
	case DATA_TYPE_UINT64:
	case DATA_TYPE_HRTIME:
	case DATA_TYPE_STRING:
	case DATA_TYPE_NVLIST:
	#if !defined(_KERNEL)
	case DATA_TYPE_DOUBLE:
	#endif
	val = va_arg(ap, void *);
	ret = nvlist_lookup_common(nvl, name, type, NULL, val);
	break;

	case DATA_TYPE_BYTE_ARRAY:
	case DATA_TYPE_BOOLEAN_ARRAY:
	case DATA_TYPE_INT8_ARRAY:
	case DATA_TYPE_UINT8_ARRAY:
	case DATA_TYPE_INT16_ARRAY:
	case DATA_TYPE_UINT16_ARRAY:
	case DATA_TYPE_INT32_ARRAY:
	case DATA_TYPE_UINT32_ARRAY:
	case DATA_TYPE_INT64_ARRAY:
	case DATA_TYPE_UINT64_ARRAY:
	case DATA_TYPE_STRING_ARRAY:
	case DATA_TYPE_NVLIST_ARRAY:
	val = va_arg(ap, void *);
	nelem = va_arg(ap, uint_t *);
	ret = nvlist_lookup_common(nvl, name, type, nelem, val);
	break;

	default:
	ret = EINVAL;
	}

	if (ret == ENOENT && noentok)
	ret = 0;
	}
	va_end(ap);

	return (ret);
	}

	/*
	* Find the 'name'ed nvpair in the nvlist 'nvl'. If 'name' found, the function
	* returns zero and a pointer to the matching nvpair is returned in '*ret'
	* (given 'ret' is non-NULL). If 'sep' is specified then 'name' will penitrate
	* multiple levels of embedded nvlists, with 'sep' as the separator. As an
	* example, if sep is '.', name might look like: "a" or "a.b" or "a.c[3]" or
	* "a.d[3].e[1]". This matches the C syntax for array embed (for convenience,
	* code also supports "a.d[3]e[1]" syntax).
	*
	* If 'ip' is non-NULL and the last name component is an array, return the
	* value of the "...[index]" array index in *ip. For an array reference that
	* is not indexed, *ip will be returned as -1. If there is a syntax error in
	* 'name', and 'ep' is non-NULL then *ep will be set to point to the location
	* inside the 'name' string where the syntax error was detected.
	*/
	static int
	nvlist_lookup_nvpair_ei_sep(nvlist_t nvl, const char name, const char sep,
	nvpair_t *ret, int ip, char **ep)
	{
	nvpair_t *nvp;
	const char *np;
	char *sepp = NULL;
	char idxp, idxep;
	nvlist_t **nva;
	long idx = 0;
	int n;

	if (ip)
	ip = -1; / not indexed */
	if (ep)
	*ep = NULL;

	if ((nvl == NULL) \|\| (name == NULL))
	return (EINVAL);

	sepp = NULL;
	idx = 0;
	/* step through components of name */
	for (np = name; np && *np; np = sepp) {
	/* ensure unique names */
	if (!(nvl->nvl_nvflag & NV_UNIQUE_NAME))
	return (ENOTSUP);

	/* skip white space */
	skip_whitespace(np);
	if (*np == 0)
	break;

	/* set 'sepp' to end of current component 'np' */
	if (sep)
	sepp = strchr(np, sep);
	else
	sepp = NULL;

	/* find start of next "[ index ]..." */
	idxp = strchr(np, '[');

	/* if sepp comes first, set idxp to NULL */
	if (sepp && idxp && (sepp < idxp))
	idxp = NULL;

	/*
	* At this point 'idxp' is set if there is an index
	* expected for the current component.
	*/
	if (idxp) {
	/* set 'n' to length of current 'np' name component */
	n = idxp++ - np;

	/* keep sepp up to date for ep use as we advance /
	skip_whitespace(idxp);
	sepp = idxp;

	/* determine the index value */
	#if defined(_KERNEL)
	if (ddi_strtol(idxp, &idxep, 0, &idx))
	goto fail;
	#else
	idx = strtol(idxp, &idxep, 0);
	#endif
	if (idxep == idxp)
	goto fail;

	/* keep sepp up to date for ep use as we advance /
	sepp = idxep;

	/* skip white space index value and check for ']' */
	skip_whitespace(sepp);
	if (*sepp++ != ']')
	goto fail;

	/* for embedded arrays, support C syntax: "a[1].b" */
	skip_whitespace(sepp);
	if (sep && (*sepp == sep))
	sepp++;
	} else if (sepp) {
	n = sepp++ - np;
	} else {
	n = strlen(np);
	}

	/* trim trailing whitespace by reducing length of 'np' */
	if (n == 0)
	goto fail;
	for (n--; (np[n] == ' ') \|\| (np[n] == '\t'); n--)
	;
	n++;

	/* skip whitespace, and set sepp to NULL if complete */
	if (sepp) {
	skip_whitespace(sepp);
	if (*sepp == 0)
	sepp = NULL;
	}

	/*
	* At this point:
	* o 'n' is the length of current 'np' component.
	* o 'idxp' is set if there was an index, and value 'idx'.
	* o 'sepp' is set to the beginning of the next component,
	* and set to NULL if we have no more components.
	*
	* Search for nvpair with matching component name.
	*/
	for (nvp = nvlist_next_nvpair(nvl, NULL); nvp != NULL;
	nvp = nvlist_next_nvpair(nvl, nvp)) {

	/* continue if no match on name */
	if (strncmp(np, nvpair_name(nvp), n) \|\|
	(strlen(nvpair_name(nvp)) != n))
	continue;

	/* if indexed, verify type is array oriented */
	if (idxp && !nvpair_type_is_array(nvp))
	goto fail;

	/*
	* Full match found, return nvp and idx if this
	* was the last component.
	*/
	if (sepp == NULL) {
	if (ret)
	*ret = nvp;
	if (ip && idxp)
	ip = (int)idx; / return index */
	return (0); /* found */
	}

	/*
	* More components: current match must be
	* of DATA_TYPE_NVLIST or DATA_TYPE_NVLIST_ARRAY
	* to support going deeper.
	*/
	if (nvpair_type(nvp) == DATA_TYPE_NVLIST) {
	nvl = EMBEDDED_NVL(nvp);
	break;
	} else if (nvpair_type(nvp) == DATA_TYPE_NVLIST_ARRAY) {
	(void) nvpair_value_nvlist_array(nvp,
	&nva, (uint_t *)&n);
	if ((n < 0) \|\| (idx >= n))
	goto fail;
	nvl = nva[idx];
	break;
	}

	/* type does not support more levels */
	goto fail;
	}
	if (nvp == NULL)
	goto fail; /* 'name' not found */

	/* search for match of next component in embedded 'nvl' list */
	}

	fail: if (ep && sepp)
	*ep = sepp;
	return (EINVAL);
	}

	/*
	* Return pointer to nvpair with specified 'name'.
	*/
	int
	nvlist_lookup_nvpair(nvlist_t nvl, const char name, nvpair_t **ret)
	{
	return (nvlist_lookup_nvpair_ei_sep(nvl, name, 0, ret, NULL, NULL));
	}

	/*
	* Determine if named nvpair exists in nvlist (use embedded separator of '.'
	* and return array index). See nvlist_lookup_nvpair_ei_sep for more detailed
	* description.
	*/
	int nvlist_lookup_nvpair_embedded_index(nvlist_t *nvl,
	const char name, nvpair_t ret, int ip, char **ep)
	{
	return (nvlist_lookup_nvpair_ei_sep(nvl, name, '.', ret, ip, ep));
	}

	boolean_t
	-nvlist_exists(nvlist_t nvl, const char name)
	+nvlist_exists(const nvlist_t nvl, const char name)
	{
	nvpriv_t *priv;
	nvpair_t *nvp;
	i_nvp_t *curr;

	if (name == NULL \|\| nvl == NULL \|\|
	(priv = (nvpriv_t *)(uintptr_t)nvl->nvl_priv) == NULL)
	return (B_FALSE);

	for (curr = priv->nvp_list; curr != NULL; curr = curr->nvi_next) {
	nvp = &curr->nvi_nvp;

	if (strcmp(name, NVP_NAME(nvp)) == 0)
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	int
	-nvpair_value_boolean_value(nvpair_t nvp, boolean_t val)
	+nvpair_value_boolean_value(const nvpair_t nvp, boolean_t val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_BOOLEAN_VALUE, NULL, val));
	}

	int
	-nvpair_value_byte(nvpair_t nvp, uchar_t val)
	+nvpair_value_byte(const nvpair_t nvp, uchar_t val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_BYTE, NULL, val));
	}

	int
	-nvpair_value_int8(nvpair_t nvp, int8_t val)
	+nvpair_value_int8(const nvpair_t nvp, int8_t val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_INT8, NULL, val));
	}

	int
	-nvpair_value_uint8(nvpair_t nvp, uint8_t val)
	+nvpair_value_uint8(const nvpair_t nvp, uint8_t val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_UINT8, NULL, val));
	}

	int
	-nvpair_value_int16(nvpair_t nvp, int16_t val)
	+nvpair_value_int16(const nvpair_t nvp, int16_t val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_INT16, NULL, val));
	}

	int
	-nvpair_value_uint16(nvpair_t nvp, uint16_t val)
	+nvpair_value_uint16(const nvpair_t nvp, uint16_t val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_UINT16, NULL, val));
	}

	int
	-nvpair_value_int32(nvpair_t nvp, int32_t val)
	+nvpair_value_int32(const nvpair_t nvp, int32_t val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_INT32, NULL, val));
	}

	int
	-nvpair_value_uint32(nvpair_t nvp, uint32_t val)
	+nvpair_value_uint32(const nvpair_t nvp, uint32_t val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_UINT32, NULL, val));
	}

	int
	-nvpair_value_int64(nvpair_t nvp, int64_t val)
	+nvpair_value_int64(const nvpair_t nvp, int64_t val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_INT64, NULL, val));
	}

	int
	-nvpair_value_uint64(nvpair_t nvp, uint64_t val)
	+nvpair_value_uint64(const nvpair_t nvp, uint64_t val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_UINT64, NULL, val));
	}

	#if !defined(_KERNEL)
	int
	-nvpair_value_double(nvpair_t nvp, double val)
	+nvpair_value_double(const nvpair_t nvp, double val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_DOUBLE, NULL, val));
	}
	#endif

	int
	nvpair_value_string(nvpair_t nvp, char *val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_STRING, NULL, val));
	}

	int
	nvpair_value_nvlist(nvpair_t nvp, nvlist_t *val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_NVLIST, NULL, val));
	}

	int
	nvpair_value_boolean_array(nvpair_t nvp, boolean_t val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_BOOLEAN_ARRAY, nelem, val));
	}

	int
	nvpair_value_byte_array(nvpair_t nvp, uchar_t val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_BYTE_ARRAY, nelem, val));
	}

	int
	nvpair_value_int8_array(nvpair_t nvp, int8_t val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_INT8_ARRAY, nelem, val));
	}

	int
	nvpair_value_uint8_array(nvpair_t nvp, uint8_t val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_UINT8_ARRAY, nelem, val));
	}

	int
	nvpair_value_int16_array(nvpair_t nvp, int16_t val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_INT16_ARRAY, nelem, val));
	}

	int
	nvpair_value_uint16_array(nvpair_t nvp, uint16_t val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_UINT16_ARRAY, nelem, val));
	}

	int
	nvpair_value_int32_array(nvpair_t nvp, int32_t val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_INT32_ARRAY, nelem, val));
	}

	int
	nvpair_value_uint32_array(nvpair_t nvp, uint32_t val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_UINT32_ARRAY, nelem, val));
	}

	int
	nvpair_value_int64_array(nvpair_t nvp, int64_t val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_INT64_ARRAY, nelem, val));
	}

	int
	nvpair_value_uint64_array(nvpair_t nvp, uint64_t val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_UINT64_ARRAY, nelem, val));
	}

	int
	nvpair_value_string_array(nvpair_t nvp, char *val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_STRING_ARRAY, nelem, val));
	}

	int
	nvpair_value_nvlist_array(nvpair_t nvp, nvlist_t *val, uint_t nelem)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_NVLIST_ARRAY, nelem, val));
	}

	int
	nvpair_value_hrtime(nvpair_t nvp, hrtime_t val)
	{
	return (nvpair_value_common(nvp, DATA_TYPE_HRTIME, NULL, val));
	}

	/*
	* Add specified pair to the list.
	*/
	int
	nvlist_add_nvpair(nvlist_t nvl, nvpair_t nvp)
	{
	if (nvl == NULL \|\| nvp == NULL)
	return (EINVAL);

	return (nvlist_add_common(nvl, NVP_NAME(nvp), NVP_TYPE(nvp),
	NVP_NELEM(nvp), NVP_VALUE(nvp)));
	}

	/*
	* Merge the supplied nvlists and put the result in dst.
	* The merged list will contain all names specified in both lists,
	* the values are taken from nvl in the case of duplicates.
	* Return 0 on success.
	*/
	/ARGSUSED/
	int
	nvlist_merge(nvlist_t dst, nvlist_t nvl, int flag)
	{
	if (nvl == NULL \|\| dst == NULL)
	return (EINVAL);

	if (dst != nvl)
	return (nvlist_copy_pairs(nvl, dst));

	return (0);
	}

	/*
	* Encoding related routines
	*/
	#define NVS_OP_ENCODE 0
	#define NVS_OP_DECODE 1
	#define NVS_OP_GETSIZE 2

	typedef struct nvs_ops nvs_ops_t;

	typedef struct {
	int nvs_op;
	const nvs_ops_t *nvs_ops;
	void *nvs_private;
	nvpriv_t *nvs_priv;
	int nvs_recursion;
	} nvstream_t;

	/*
	* nvs operations are:
	* - nvs_nvlist
	* encoding / decoding of an nvlist header (nvlist_t)
	* calculates the size used for header and end detection
	*
	* - nvs_nvpair
	* responsible for the first part of encoding / decoding of an nvpair
	* calculates the decoded size of an nvpair
	*
	* - nvs_nvp_op
	* second part of encoding / decoding of an nvpair
	*
	* - nvs_nvp_size
	* calculates the encoding size of an nvpair
	*
	* - nvs_nvl_fini
	* encodes the end detection mark (zeros).
	*/
	struct nvs_ops {
	int (nvs_nvlist)(nvstream_t , nvlist_t , size_t );
	int (nvs_nvpair)(nvstream_t , nvpair_t , size_t );
	int (nvs_nvp_op)(nvstream_t , nvpair_t *);
	int (nvs_nvp_size)(nvstream_t , nvpair_t , size_t );
	int (nvs_nvl_fini)(nvstream_t );
	};

	typedef struct {
	char nvh_encoding; /* nvs encoding method */
	char nvh_endian; /* nvs endian */
	char nvh_reserved1; /* reserved for future use */
	char nvh_reserved2; /* reserved for future use */
	} nvs_header_t;

	static int
	nvs_encode_pairs(nvstream_t nvs, nvlist_t nvl)
	{
	nvpriv_t priv = (nvpriv_t )(uintptr_t)nvl->nvl_priv;
	i_nvp_t *curr;

	/*
	* Walk nvpair in list and encode each nvpair
	*/
	for (curr = priv->nvp_list; curr != NULL; curr = curr->nvi_next)
	if (nvs->nvs_ops->nvs_nvpair(nvs, &curr->nvi_nvp, NULL) != 0)
	return (EFAULT);

	return (nvs->nvs_ops->nvs_nvl_fini(nvs));
	}

	static int
	nvs_decode_pairs(nvstream_t nvs, nvlist_t nvl)
	{
	nvpair_t *nvp;
	size_t nvsize;
	int err;

	/*
	* Get decoded size of next pair in stream, alloc
	* memory for nvpair_t, then decode the nvpair
	*/
	while ((err = nvs->nvs_ops->nvs_nvpair(nvs, NULL, &nvsize)) == 0) {
	if (nvsize == 0) /* end of list */
	break;

	/* make sure len makes sense */
	if (nvsize < NVP_SIZE_CALC(1, 0))
	return (EFAULT);

	if ((nvp = nvp_buf_alloc(nvl, nvsize)) == NULL)
	return (ENOMEM);

	if ((err = nvs->nvs_ops->nvs_nvp_op(nvs, nvp)) != 0) {
	nvp_buf_free(nvl, nvp);
	return (err);
	}

	if (i_validate_nvpair(nvp) != 0) {
	nvpair_free(nvp);
	nvp_buf_free(nvl, nvp);
	return (EFAULT);
	}

	err = nvt_add_nvpair(nvl, nvp);
	if (err != 0) {
	nvpair_free(nvp);
	nvp_buf_free(nvl, nvp);
	return (err);
	}
	nvp_buf_link(nvl, nvp);
	}
	return (err);
	}

	static int
	nvs_getsize_pairs(nvstream_t nvs, nvlist_t nvl, size_t *buflen)
	{
	nvpriv_t priv = (nvpriv_t )(uintptr_t)nvl->nvl_priv;
	i_nvp_t *curr;
	uint64_t nvsize = *buflen;
	size_t size;

	/*
	* Get encoded size of nvpairs in nvlist
	*/
	for (curr = priv->nvp_list; curr != NULL; curr = curr->nvi_next) {
	if (nvs->nvs_ops->nvs_nvp_size(nvs, &curr->nvi_nvp, &size) != 0)
	return (EINVAL);

	if ((nvsize += size) > INT32_MAX)
	return (EINVAL);
	}

	*buflen = nvsize;
	return (0);
	}

	static int
	nvs_operation(nvstream_t nvs, nvlist_t nvl, size_t *buflen)
	{
	int err;

	if (nvl->nvl_priv == 0)
	return (EFAULT);

	/*
	* Perform the operation, starting with header, then each nvpair
	*/
	if ((err = nvs->nvs_ops->nvs_nvlist(nvs, nvl, buflen)) != 0)
	return (err);

	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE:
	err = nvs_encode_pairs(nvs, nvl);
	break;

	case NVS_OP_DECODE:
	err = nvs_decode_pairs(nvs, nvl);
	break;

	case NVS_OP_GETSIZE:
	err = nvs_getsize_pairs(nvs, nvl, buflen);
	break;

	default:
	err = EINVAL;
	}

	return (err);
	}

	static int
	nvs_embedded(nvstream_t nvs, nvlist_t embedded)
	{
	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE: {
	int err;

	if (nvs->nvs_recursion >= nvpair_max_recursion)
	return (EINVAL);
	nvs->nvs_recursion++;
	err = nvs_operation(nvs, embedded, NULL);
	nvs->nvs_recursion--;
	return (err);
	}
	case NVS_OP_DECODE: {
	nvpriv_t *priv;
	int err;

	if (embedded->nvl_version != NV_VERSION)
	return (ENOTSUP);

	if ((priv = nv_priv_alloc_embedded(nvs->nvs_priv)) == NULL)
	return (ENOMEM);

	nvlist_init(embedded, embedded->nvl_nvflag, priv);

	if (nvs->nvs_recursion >= nvpair_max_recursion) {
	nvlist_free(embedded);
	return (EINVAL);
	}
	nvs->nvs_recursion++;
	if ((err = nvs_operation(nvs, embedded, NULL)) != 0)
	nvlist_free(embedded);
	nvs->nvs_recursion--;
	return (err);
	}
	default:
	break;
	}

	return (EINVAL);
	}

	static int
	nvs_embedded_nvl_array(nvstream_t nvs, nvpair_t nvp, size_t *size)
	{
	size_t nelem = NVP_NELEM(nvp);
	nvlist_t **nvlp = EMBEDDED_NVL_ARRAY(nvp);
	int i;

	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE:
	for (i = 0; i < nelem; i++)
	if (nvs_embedded(nvs, nvlp[i]) != 0)
	return (EFAULT);
	break;

	case NVS_OP_DECODE: {
	size_t len = nelem * sizeof (uint64_t);
	nvlist_t embedded = (nvlist_t )((uintptr_t)nvlp + len);

	bzero(nvlp, len); /* don't trust packed data */
	for (i = 0; i < nelem; i++) {
	if (nvs_embedded(nvs, embedded) != 0) {
	nvpair_free(nvp);
	return (EFAULT);
	}

	nvlp[i] = embedded++;
	}
	break;
	}
	case NVS_OP_GETSIZE: {
	uint64_t nvsize = 0;

	for (i = 0; i < nelem; i++) {
	size_t nvp_sz = 0;

	if (nvs_operation(nvs, nvlp[i], &nvp_sz) != 0)
	return (EINVAL);

	if ((nvsize += nvp_sz) > INT32_MAX)
	return (EINVAL);
	}

	*size = nvsize;
	break;
	}
	default:
	return (EINVAL);
	}

	return (0);
	}

	static int nvs_native(nvstream_t , nvlist_t , char , size_t );
	static int nvs_xdr(nvstream_t , nvlist_t , char , size_t );

	/*
	* Common routine for nvlist operations:
	* encode, decode, getsize (encoded size).
	*/
	static int
	nvlist_common(nvlist_t nvl, char buf, size_t *buflen, int encoding,
	int nvs_op)
	{
	int err = 0;
	nvstream_t nvs;
	int nvl_endian;
	#if defined(_ZFS_LITTLE_ENDIAN)
	int host_endian = 1;
	#elif defined(_ZFS_BIG_ENDIAN)
	int host_endian = 0;
	#else
	#error "No endian defined!"
	#endif /* _ZFS_LITTLE_ENDIAN */
	nvs_header_t *nvh;

	if (buflen == NULL \|\| nvl == NULL \|\|
	(nvs.nvs_priv = (nvpriv_t *)(uintptr_t)nvl->nvl_priv) == NULL)
	return (EINVAL);

	nvs.nvs_op = nvs_op;
	nvs.nvs_recursion = 0;

	/*
	* For NVS_OP_ENCODE and NVS_OP_DECODE make sure an nvlist and
	* a buffer is allocated. The first 4 bytes in the buffer are
	* used for encoding method and host endian.
	*/
	switch (nvs_op) {
	case NVS_OP_ENCODE:
	if (buf == NULL \|\| *buflen < sizeof (nvs_header_t))
	return (EINVAL);

	nvh = (void *)buf;
	nvh->nvh_encoding = encoding;
	nvh->nvh_endian = nvl_endian = host_endian;
	nvh->nvh_reserved1 = 0;
	nvh->nvh_reserved2 = 0;
	break;

	case NVS_OP_DECODE:
	if (buf == NULL \|\| *buflen < sizeof (nvs_header_t))
	return (EINVAL);

	/* get method of encoding from first byte */
	nvh = (void *)buf;
	encoding = nvh->nvh_encoding;
	nvl_endian = nvh->nvh_endian;
	break;

	case NVS_OP_GETSIZE:
	nvl_endian = host_endian;

	/*
	* add the size for encoding
	*/
	*buflen = sizeof (nvs_header_t);
	break;

	default:
	return (ENOTSUP);
	}

	/*
	* Create an nvstream with proper encoding method
	*/
	switch (encoding) {
	case NV_ENCODE_NATIVE:
	/*
	* check endianness, in case we are unpacking
	* from a file
	*/
	if (nvl_endian != host_endian)
	return (ENOTSUP);
	err = nvs_native(&nvs, nvl, buf, buflen);
	break;
	case NV_ENCODE_XDR:
	err = nvs_xdr(&nvs, nvl, buf, buflen);
	break;
	default:
	err = ENOTSUP;
	break;
	}

	return (err);
	}

	int
	nvlist_size(nvlist_t nvl, size_t size, int encoding)
	{
	return (nvlist_common(nvl, NULL, size, encoding, NVS_OP_GETSIZE));
	}

	/*
	* Pack nvlist into contiguous memory
	*/
	int
	nvlist_pack(nvlist_t nvl, char bufp, size_t buflen, int encoding,
	int kmflag)
	{
	return (nvlist_xpack(nvl, bufp, buflen, encoding,
	nvlist_nv_alloc(kmflag)));
	}

	int
	nvlist_xpack(nvlist_t nvl, char bufp, size_t buflen, int encoding,
	nv_alloc_t *nva)
	{
	nvpriv_t nvpriv;
	size_t alloc_size;
	char *buf;
	int err;

	if (nva == NULL \|\| nvl == NULL \|\| bufp == NULL \|\| buflen == NULL)
	return (EINVAL);

	if (*bufp != NULL)
	return (nvlist_common(nvl, *bufp, buflen, encoding,
	NVS_OP_ENCODE));

	/*
	* Here is a difficult situation:
	* 1. The nvlist has fixed allocator properties.
	* All other nvlist routines (like nvlist_add_*, ...) use
	* these properties.
	* 2. When using nvlist_pack() the user can specify their own
	* allocator properties (e.g. by using KM_NOSLEEP).
	*
	* We use the user specified properties (2). A clearer solution
	* will be to remove the kmflag from nvlist_pack(), but we will
	* not change the interface.
	*/
	nv_priv_init(&nvpriv, nva, 0);

	if ((err = nvlist_size(nvl, &alloc_size, encoding)))
	return (err);

	if ((buf = nv_mem_zalloc(&nvpriv, alloc_size)) == NULL)
	return (ENOMEM);

	if ((err = nvlist_common(nvl, buf, &alloc_size, encoding,
	NVS_OP_ENCODE)) != 0) {
	nv_mem_free(&nvpriv, buf, alloc_size);
	} else {
	*buflen = alloc_size;
	*bufp = buf;
	}

	return (err);
	}

	/*
	* Unpack buf into an nvlist_t
	*/
	int
	nvlist_unpack(char buf, size_t buflen, nvlist_t *nvlp, int kmflag)
	{
	return (nvlist_xunpack(buf, buflen, nvlp, nvlist_nv_alloc(kmflag)));
	}

	int
	nvlist_xunpack(char buf, size_t buflen, nvlist_t nvlp, nv_alloc_t nva)
	{
	nvlist_t *nvl;
	int err;

	if (nvlp == NULL)
	return (EINVAL);

	if ((err = nvlist_xalloc(&nvl, 0, nva)) != 0)
	return (err);

	if ((err = nvlist_common(nvl, buf, &buflen, NV_ENCODE_NATIVE,
	NVS_OP_DECODE)) != 0)
	nvlist_free(nvl);
	else
	*nvlp = nvl;

	return (err);
	}

	/*
	* Native encoding functions
	*/
	typedef struct {
	/*
	* This structure is used when decoding a packed nvpair in
	* the native format. n_base points to a buffer containing the
	* packed nvpair. n_end is a pointer to the end of the buffer.
	* (n_end actually points to the first byte past the end of the
	* buffer.) n_curr is a pointer that lies between n_base and n_end.
	* It points to the current data that we are decoding.
	* The amount of data left in the buffer is equal to n_end - n_curr.
	* n_flag is used to recognize a packed embedded list.
	*/
	caddr_t n_base;
	caddr_t n_end;
	caddr_t n_curr;
	uint_t n_flag;
	} nvs_native_t;

	static int
	nvs_native_create(nvstream_t nvs, nvs_native_t native, char *buf,
	size_t buflen)
	{
	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE:
	case NVS_OP_DECODE:
	nvs->nvs_private = native;
	native->n_curr = native->n_base = buf;
	native->n_end = buf + buflen;
	native->n_flag = 0;
	return (0);

	case NVS_OP_GETSIZE:
	nvs->nvs_private = native;
	native->n_curr = native->n_base = native->n_end = NULL;
	native->n_flag = 0;
	return (0);
	default:
	return (EINVAL);
	}
	}

	/ARGSUSED/
	static void
	nvs_native_destroy(nvstream_t *nvs)
	{
	}

	static int
	native_cp(nvstream_t nvs, void buf, size_t size)
	{
	nvs_native_t native = (nvs_native_t )nvs->nvs_private;

	if (native->n_curr + size > native->n_end)
	return (EFAULT);

	/*
	* The bcopy() below eliminates alignment requirement
	* on the buffer (stream) and is preferred over direct access.
	*/
	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE:
	bcopy(buf, native->n_curr, size);
	break;
	case NVS_OP_DECODE:
	bcopy(native->n_curr, buf, size);
	break;
	default:
	return (EINVAL);
	}

	native->n_curr += size;
	return (0);
	}

	/*
	* operate on nvlist_t header
	*/
	static int
	nvs_native_nvlist(nvstream_t nvs, nvlist_t nvl, size_t *size)
	{
	nvs_native_t *native = nvs->nvs_private;

	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE:
	case NVS_OP_DECODE:
	if (native->n_flag)
	return (0); /* packed embedded list */

	native->n_flag = 1;

	/* copy version and nvflag of the nvlist_t */
	if (native_cp(nvs, &nvl->nvl_version, sizeof (int32_t)) != 0 \|\|
	native_cp(nvs, &nvl->nvl_nvflag, sizeof (int32_t)) != 0)
	return (EFAULT);

	return (0);

	case NVS_OP_GETSIZE:
	/*
	* if calculate for packed embedded list
	* 4 for end of the embedded list
	* else
	* 2 * sizeof (int32_t) for nvl_version and nvl_nvflag
	* and 4 for end of the entire list
	*/
	if (native->n_flag) {
	*size += 4;
	} else {
	native->n_flag = 1;
	size += 2 sizeof (int32_t) + 4;
	}

	return (0);

	default:
	return (EINVAL);
	}
	}

	static int
	nvs_native_nvl_fini(nvstream_t *nvs)
	{
	if (nvs->nvs_op == NVS_OP_ENCODE) {
	nvs_native_t native = (nvs_native_t )nvs->nvs_private;
	/*
	* Add 4 zero bytes at end of nvlist. They are used
	* for end detection by the decode routine.
	*/
	if (native->n_curr + sizeof (int) > native->n_end)
	return (EFAULT);

	bzero(native->n_curr, sizeof (int));
	native->n_curr += sizeof (int);
	}

	return (0);
	}

	static int
	nvpair_native_embedded(nvstream_t nvs, nvpair_t nvp)
	{
	if (nvs->nvs_op == NVS_OP_ENCODE) {
	nvs_native_t native = (nvs_native_t )nvs->nvs_private;
	nvlist_t packed = (void )
	(native->n_curr - nvp->nvp_size + NVP_VALOFF(nvp));
	/*
	* Null out the pointer that is meaningless in the packed
	* structure. The address may not be aligned, so we have
	* to use bzero.
	*/
	bzero((char *)packed + offsetof(nvlist_t, nvl_priv),
	sizeof (uint64_t));
	}

	return (nvs_embedded(nvs, EMBEDDED_NVL(nvp)));
	}

	static int
	nvpair_native_embedded_array(nvstream_t nvs, nvpair_t nvp)
	{
	if (nvs->nvs_op == NVS_OP_ENCODE) {
	nvs_native_t native = (nvs_native_t )nvs->nvs_private;
	char *value = native->n_curr - nvp->nvp_size + NVP_VALOFF(nvp);
	size_t len = NVP_NELEM(nvp) * sizeof (uint64_t);
	nvlist_t packed = (nvlist_t )((uintptr_t)value + len);
	int i;
	/*
	* Null out pointers that are meaningless in the packed
	* structure. The addresses may not be aligned, so we have
	* to use bzero.
	*/
	bzero(value, len);

	for (i = 0; i < NVP_NELEM(nvp); i++, packed++)
	/*
	* Null out the pointer that is meaningless in the
	* packed structure. The address may not be aligned,
	* so we have to use bzero.
	*/
	bzero((char *)packed + offsetof(nvlist_t, nvl_priv),
	sizeof (uint64_t));
	}

	return (nvs_embedded_nvl_array(nvs, nvp, NULL));
	}

	static void
	nvpair_native_string_array(nvstream_t nvs, nvpair_t nvp)
	{
	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE: {
	nvs_native_t native = (nvs_native_t )nvs->nvs_private;
	uint64_t strp = (void )
	(native->n_curr - nvp->nvp_size + NVP_VALOFF(nvp));
	/*
	* Null out pointers that are meaningless in the packed
	* structure. The addresses may not be aligned, so we have
	* to use bzero.
	*/
	bzero(strp, NVP_NELEM(nvp) * sizeof (uint64_t));
	break;
	}
	case NVS_OP_DECODE: {
	char *strp = (void )NVP_VALUE(nvp);
	char buf = ((char )strp + NVP_NELEM(nvp) * sizeof (uint64_t));
	int i;

	for (i = 0; i < NVP_NELEM(nvp); i++) {
	strp[i] = buf;
	buf += strlen(buf) + 1;
	}
	break;
	}
	}
	}

	static int
	nvs_native_nvp_op(nvstream_t nvs, nvpair_t nvp)
	{
	data_type_t type;
	int value_sz;
	int ret = 0;

	/*
	* We do the initial bcopy of the data before we look at
	* the nvpair type, because when we're decoding, we won't
	* have the correct values for the pair until we do the bcopy.
	*/
	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE:
	case NVS_OP_DECODE:
	if (native_cp(nvs, nvp, nvp->nvp_size) != 0)
	return (EFAULT);
	break;
	default:
	return (EINVAL);
	}

	/* verify nvp_name_sz, check the name string length */
	if (i_validate_nvpair_name(nvp) != 0)
	return (EFAULT);

	type = NVP_TYPE(nvp);

	/*
	* Verify type and nelem and get the value size.
	* In case of data types DATA_TYPE_STRING and DATA_TYPE_STRING_ARRAY
	* is the size of the string(s) excluded.
	*/
	if ((value_sz = i_get_value_size(type, NULL, NVP_NELEM(nvp))) < 0)
	return (EFAULT);

	if (NVP_SIZE_CALC(nvp->nvp_name_sz, value_sz) > nvp->nvp_size)
	return (EFAULT);

	switch (type) {
	case DATA_TYPE_NVLIST:
	ret = nvpair_native_embedded(nvs, nvp);
	break;
	case DATA_TYPE_NVLIST_ARRAY:
	ret = nvpair_native_embedded_array(nvs, nvp);
	break;
	case DATA_TYPE_STRING_ARRAY:
	nvpair_native_string_array(nvs, nvp);
	break;
	default:
	break;
	}

	return (ret);
	}

	static int
	nvs_native_nvp_size(nvstream_t nvs, nvpair_t nvp, size_t *size)
	{
	uint64_t nvp_sz = nvp->nvp_size;

	switch (NVP_TYPE(nvp)) {
	case DATA_TYPE_NVLIST: {
	size_t nvsize = 0;

	if (nvs_operation(nvs, EMBEDDED_NVL(nvp), &nvsize) != 0)
	return (EINVAL);

	nvp_sz += nvsize;
	break;
	}
	case DATA_TYPE_NVLIST_ARRAY: {
	size_t nvsize;

	if (nvs_embedded_nvl_array(nvs, nvp, &nvsize) != 0)
	return (EINVAL);

	nvp_sz += nvsize;
	break;
	}
	default:
	break;
	}

	if (nvp_sz > INT32_MAX)
	return (EINVAL);

	*size = nvp_sz;

	return (0);
	}

	static int
	nvs_native_nvpair(nvstream_t nvs, nvpair_t nvp, size_t *size)
	{
	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE:
	return (nvs_native_nvp_op(nvs, nvp));

	case NVS_OP_DECODE: {
	nvs_native_t native = (nvs_native_t )nvs->nvs_private;
	int32_t decode_len;

	/* try to read the size value from the stream */
	if (native->n_curr + sizeof (int32_t) > native->n_end)
	return (EFAULT);
	bcopy(native->n_curr, &decode_len, sizeof (int32_t));

	/* sanity check the size value */
	if (decode_len < 0 \|\|
	decode_len > native->n_end - native->n_curr)
	return (EFAULT);

	*size = decode_len;

	/*
	* If at the end of the stream then move the cursor
	* forward, otherwise nvpair_native_op() will read
	* the entire nvpair at the same cursor position.
	*/
	if (*size == 0)
	native->n_curr += sizeof (int32_t);
	break;
	}

	default:
	return (EINVAL);
	}

	return (0);
	}

	static const nvs_ops_t nvs_native_ops = {
	.nvs_nvlist = nvs_native_nvlist,
	.nvs_nvpair = nvs_native_nvpair,
	.nvs_nvp_op = nvs_native_nvp_op,
	.nvs_nvp_size = nvs_native_nvp_size,
	.nvs_nvl_fini = nvs_native_nvl_fini
	};

	static int
	nvs_native(nvstream_t nvs, nvlist_t nvl, char buf, size_t buflen)
	{
	nvs_native_t native;
	int err;

	nvs->nvs_ops = &nvs_native_ops;

	if ((err = nvs_native_create(nvs, &native, buf + sizeof (nvs_header_t),
	*buflen - sizeof (nvs_header_t))) != 0)
	return (err);

	err = nvs_operation(nvs, nvl, buflen);

	nvs_native_destroy(nvs);

	return (err);
	}

	/*
	* XDR encoding functions
	*
	* An xdr packed nvlist is encoded as:
	*
	* - encoding method and host endian (4 bytes)
	* - nvl_version (4 bytes)
	* - nvl_nvflag (4 bytes)
	*
	* - encoded nvpairs, the format of one xdr encoded nvpair is:
	* - encoded size of the nvpair (4 bytes)
	* - decoded size of the nvpair (4 bytes)
	* - name string, (4 + sizeof(NV_ALIGN4(string))
	* a string is coded as size (4 bytes) and data
	* - data type (4 bytes)
	* - number of elements in the nvpair (4 bytes)
	* - data
	*
	* - 2 zero's for end of the entire list (8 bytes)
	*/
	static int
	nvs_xdr_create(nvstream_t nvs, XDR xdr, char *buf, size_t buflen)
	{
	/* xdr data must be 4 byte aligned */
	if ((ulong_t)buf % 4 != 0)
	return (EFAULT);

	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE:
	xdrmem_create(xdr, buf, (uint_t)buflen, XDR_ENCODE);
	nvs->nvs_private = xdr;
	return (0);
	case NVS_OP_DECODE:
	xdrmem_create(xdr, buf, (uint_t)buflen, XDR_DECODE);
	nvs->nvs_private = xdr;
	return (0);
	case NVS_OP_GETSIZE:
	nvs->nvs_private = NULL;
	return (0);
	default:
	return (EINVAL);
	}
	}

	static void
	nvs_xdr_destroy(nvstream_t *nvs)
	{
	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE:
	case NVS_OP_DECODE:
	xdr_destroy((XDR *)nvs->nvs_private);
	break;
	default:
	break;
	}
	}

	static int
	nvs_xdr_nvlist(nvstream_t nvs, nvlist_t nvl, size_t *size)
	{
	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE:
	case NVS_OP_DECODE: {
	XDR *xdr = nvs->nvs_private;

	if (!xdr_int(xdr, &nvl->nvl_version) \|\|
	!xdr_u_int(xdr, &nvl->nvl_nvflag))
	return (EFAULT);
	break;
	}
	case NVS_OP_GETSIZE: {
	/*
	* 2 * 4 for nvl_version + nvl_nvflag
	* and 8 for end of the entire list
	*/
	size += 2 4 + 8;
	break;
	}
	default:
	return (EINVAL);
	}
	return (0);
	}

	static int
	nvs_xdr_nvl_fini(nvstream_t *nvs)
	{
	if (nvs->nvs_op == NVS_OP_ENCODE) {
	XDR *xdr = nvs->nvs_private;
	int zero = 0;

	if (!xdr_int(xdr, &zero) \|\| !xdr_int(xdr, &zero))
	return (EFAULT);
	}

	return (0);
	}

	/*
	* xdrproc_t-compatible callbacks for xdr_array()
	*/

	#if defined(_KERNEL) && defined(__linux__) /* Linux kernel */

	#define NVS_BUILD_XDRPROC_T(type) \
	static bool_t \
	nvs_xdr_nvp_##type(XDR xdrs, void ptr) \
	{ \
	return (xdr_##type(xdrs, ptr)); \
	}

	#elif !defined(_KERNEL) && defined(XDR_CONTROL) /* tirpc */

	#define NVS_BUILD_XDRPROC_T(type) \
	static bool_t \
	nvs_xdr_nvp_##type(XDR *xdrs, ...) \
	{ \
	va_list args; \
	void *ptr; \
	\
	va_start(args, xdrs); \
	ptr = va_arg(args, void *); \
	va_end(args); \
	\
	return (xdr_##type(xdrs, ptr)); \
	}

	#else /* FreeBSD, sunrpc */

	#define NVS_BUILD_XDRPROC_T(type) \
	static bool_t \
	nvs_xdr_nvp_##type(XDR xdrs, void ptr, ...) \
	{ \
	return (xdr_##type(xdrs, ptr)); \
	}

	#endif

	/* BEGIN CSTYLED */
	NVS_BUILD_XDRPROC_T(char);
	NVS_BUILD_XDRPROC_T(short);
	NVS_BUILD_XDRPROC_T(u_short);
	NVS_BUILD_XDRPROC_T(int);
	NVS_BUILD_XDRPROC_T(u_int);
	NVS_BUILD_XDRPROC_T(longlong_t);
	NVS_BUILD_XDRPROC_T(u_longlong_t);
	/* END CSTYLED */

	/*
	* The format of xdr encoded nvpair is:
	* encode_size, decode_size, name string, data type, nelem, data
	*/
	static int
	nvs_xdr_nvp_op(nvstream_t nvs, nvpair_t nvp)
	{
	ASSERT(nvs != NULL && nvp != NULL);

	data_type_t type;
	char *buf;
	char buf_end = (char )nvp + nvp->nvp_size;
	int value_sz;
	uint_t nelem, buflen;
	bool_t ret = FALSE;
	XDR *xdr = nvs->nvs_private;

	ASSERT(xdr != NULL);

	/* name string */
	if ((buf = NVP_NAME(nvp)) >= buf_end)
	return (EFAULT);
	buflen = buf_end - buf;

	if (!xdr_string(xdr, &buf, buflen - 1))
	return (EFAULT);
	nvp->nvp_name_sz = strlen(buf) + 1;

	/* type and nelem */
	if (!xdr_int(xdr, (int *)&nvp->nvp_type) \|\|
	!xdr_int(xdr, &nvp->nvp_value_elem))
	return (EFAULT);

	type = NVP_TYPE(nvp);
	nelem = nvp->nvp_value_elem;

	/*
	* Verify type and nelem and get the value size.
	* In case of data types DATA_TYPE_STRING and DATA_TYPE_STRING_ARRAY
	* is the size of the string(s) excluded.
	*/
	if ((value_sz = i_get_value_size(type, NULL, nelem)) < 0)
	return (EFAULT);

	/* if there is no data to extract then return */
	if (nelem == 0)
	return (0);

	/* value */
	if ((buf = NVP_VALUE(nvp)) >= buf_end)
	return (EFAULT);
	buflen = buf_end - buf;

	if (buflen < value_sz)
	return (EFAULT);

	switch (type) {
	case DATA_TYPE_NVLIST:
	if (nvs_embedded(nvs, (void *)buf) == 0)
	return (0);
	break;

	case DATA_TYPE_NVLIST_ARRAY:
	if (nvs_embedded_nvl_array(nvs, nvp, NULL) == 0)
	return (0);
	break;

	case DATA_TYPE_BOOLEAN:
	ret = TRUE;
	break;

	case DATA_TYPE_BYTE:
	case DATA_TYPE_INT8:
	case DATA_TYPE_UINT8:
	ret = xdr_char(xdr, buf);
	break;

	case DATA_TYPE_INT16:
	ret = xdr_short(xdr, (void *)buf);
	break;

	case DATA_TYPE_UINT16:
	ret = xdr_u_short(xdr, (void *)buf);
	break;

	case DATA_TYPE_BOOLEAN_VALUE:
	case DATA_TYPE_INT32:
	ret = xdr_int(xdr, (void *)buf);
	break;

	case DATA_TYPE_UINT32:
	ret = xdr_u_int(xdr, (void *)buf);
	break;

	case DATA_TYPE_INT64:
	ret = xdr_longlong_t(xdr, (void *)buf);
	break;

	case DATA_TYPE_UINT64:
	ret = xdr_u_longlong_t(xdr, (void *)buf);
	break;

	case DATA_TYPE_HRTIME:
	/*
	* NOTE: must expose the definition of hrtime_t here
	*/
	ret = xdr_longlong_t(xdr, (void *)buf);
	break;
	#if !defined(_KERNEL)
	case DATA_TYPE_DOUBLE:
	ret = xdr_double(xdr, (void *)buf);
	break;
	#endif
	case DATA_TYPE_STRING:
	ret = xdr_string(xdr, &buf, buflen - 1);
	break;

	case DATA_TYPE_BYTE_ARRAY:
	ret = xdr_opaque(xdr, buf, nelem);
	break;

	case DATA_TYPE_INT8_ARRAY:
	case DATA_TYPE_UINT8_ARRAY:
	ret = xdr_array(xdr, &buf, &nelem, buflen, sizeof (int8_t),
	nvs_xdr_nvp_char);
	break;

	case DATA_TYPE_INT16_ARRAY:
	ret = xdr_array(xdr, &buf, &nelem, buflen / sizeof (int16_t),
	sizeof (int16_t), nvs_xdr_nvp_short);
	break;

	case DATA_TYPE_UINT16_ARRAY:
	ret = xdr_array(xdr, &buf, &nelem, buflen / sizeof (uint16_t),
	sizeof (uint16_t), nvs_xdr_nvp_u_short);
	break;

	case DATA_TYPE_BOOLEAN_ARRAY:
	case DATA_TYPE_INT32_ARRAY:
	ret = xdr_array(xdr, &buf, &nelem, buflen / sizeof (int32_t),
	sizeof (int32_t), nvs_xdr_nvp_int);
	break;

	case DATA_TYPE_UINT32_ARRAY:
	ret = xdr_array(xdr, &buf, &nelem, buflen / sizeof (uint32_t),
	sizeof (uint32_t), nvs_xdr_nvp_u_int);
	break;

	case DATA_TYPE_INT64_ARRAY:
	ret = xdr_array(xdr, &buf, &nelem, buflen / sizeof (int64_t),
	sizeof (int64_t), nvs_xdr_nvp_longlong_t);
	break;

	case DATA_TYPE_UINT64_ARRAY:
	ret = xdr_array(xdr, &buf, &nelem, buflen / sizeof (uint64_t),
	sizeof (uint64_t), nvs_xdr_nvp_u_longlong_t);
	break;

	case DATA_TYPE_STRING_ARRAY: {
	size_t len = nelem * sizeof (uint64_t);
	char *strp = (void )buf;
	int i;

	if (nvs->nvs_op == NVS_OP_DECODE)
	bzero(buf, len); /* don't trust packed data */

	for (i = 0; i < nelem; i++) {
	if (buflen <= len)
	return (EFAULT);

	buf += len;
	buflen -= len;

	if (xdr_string(xdr, &buf, buflen - 1) != TRUE)
	return (EFAULT);

	if (nvs->nvs_op == NVS_OP_DECODE)
	strp[i] = buf;
	len = strlen(buf) + 1;
	}
	ret = TRUE;
	break;
	}
	default:
	break;
	}

	return (ret == TRUE ? 0 : EFAULT);
	}

	static int
	nvs_xdr_nvp_size(nvstream_t nvs, nvpair_t nvp, size_t *size)
	{
	data_type_t type = NVP_TYPE(nvp);
	/*
	* encode_size + decode_size + name string size + data type + nelem
	* where name string size = 4 + NV_ALIGN4(strlen(NVP_NAME(nvp)))
	*/
	uint64_t nvp_sz = 4 + 4 + 4 + NV_ALIGN4(strlen(NVP_NAME(nvp))) + 4 + 4;

	switch (type) {
	case DATA_TYPE_BOOLEAN:
	break;

	case DATA_TYPE_BOOLEAN_VALUE:
	case DATA_TYPE_BYTE:
	case DATA_TYPE_INT8:
	case DATA_TYPE_UINT8:
	case DATA_TYPE_INT16:
	case DATA_TYPE_UINT16:
	case DATA_TYPE_INT32:
	case DATA_TYPE_UINT32:
	nvp_sz += 4; /* 4 is the minimum xdr unit */
	break;

	case DATA_TYPE_INT64:
	case DATA_TYPE_UINT64:
	case DATA_TYPE_HRTIME:
	#if !defined(_KERNEL)
	case DATA_TYPE_DOUBLE:
	#endif
	nvp_sz += 8;
	break;

	case DATA_TYPE_STRING:
	nvp_sz += 4 + NV_ALIGN4(strlen((char *)NVP_VALUE(nvp)));
	break;

	case DATA_TYPE_BYTE_ARRAY:
	nvp_sz += NV_ALIGN4(NVP_NELEM(nvp));
	break;

	case DATA_TYPE_BOOLEAN_ARRAY:
	case DATA_TYPE_INT8_ARRAY:
	case DATA_TYPE_UINT8_ARRAY:
	case DATA_TYPE_INT16_ARRAY:
	case DATA_TYPE_UINT16_ARRAY:
	case DATA_TYPE_INT32_ARRAY:
	case DATA_TYPE_UINT32_ARRAY:
	nvp_sz += 4 + 4 * (uint64_t)NVP_NELEM(nvp);
	break;

	case DATA_TYPE_INT64_ARRAY:
	case DATA_TYPE_UINT64_ARRAY:
	nvp_sz += 4 + 8 * (uint64_t)NVP_NELEM(nvp);
	break;

	case DATA_TYPE_STRING_ARRAY: {
	int i;
	char *strs = (void )NVP_VALUE(nvp);

	for (i = 0; i < NVP_NELEM(nvp); i++)
	nvp_sz += 4 + NV_ALIGN4(strlen(strs[i]));

	break;
	}

	case DATA_TYPE_NVLIST:
	case DATA_TYPE_NVLIST_ARRAY: {
	size_t nvsize = 0;
	int old_nvs_op = nvs->nvs_op;
	int err;

	nvs->nvs_op = NVS_OP_GETSIZE;
	if (type == DATA_TYPE_NVLIST)
	err = nvs_operation(nvs, EMBEDDED_NVL(nvp), &nvsize);
	else
	err = nvs_embedded_nvl_array(nvs, nvp, &nvsize);
	nvs->nvs_op = old_nvs_op;

	if (err != 0)
	return (EINVAL);

	nvp_sz += nvsize;
	break;
	}

	default:
	return (EINVAL);
	}

	if (nvp_sz > INT32_MAX)
	return (EINVAL);

	*size = nvp_sz;

	return (0);
	}


	/*
	* The NVS_XDR_MAX_LEN macro takes a packed xdr buffer of size x and estimates
	* the largest nvpair that could be encoded in the buffer.
	*
	* See comments above nvpair_xdr_op() for the format of xdr encoding.
	* The size of a xdr packed nvpair without any data is 5 words.
	*
	* Using the size of the data directly as an estimate would be ok
	* in all cases except one. If the data type is of DATA_TYPE_STRING_ARRAY
	* then the actual nvpair has space for an array of pointers to index
	* the strings. These pointers are not encoded into the packed xdr buffer.
	*
	* If the data is of type DATA_TYPE_STRING_ARRAY and all the strings are
	* of length 0, then each string is encoded in xdr format as a single word.
	* Therefore when expanded to an nvpair there will be 2.25 word used for
	* each string. (a int64_t allocated for pointer usage, and a single char
	* for the null termination.)
	*
	* This is the calculation performed by the NVS_XDR_MAX_LEN macro.
	*/
	#define NVS_XDR_HDR_LEN ((size_t)(5 * 4))
	#define NVS_XDR_DATA_LEN(y) (((size_t)(y) <= NVS_XDR_HDR_LEN) ? \
	0 : ((size_t)(y) - NVS_XDR_HDR_LEN))
	#define NVS_XDR_MAX_LEN(x) (NVP_SIZE_CALC(1, 0) + \
	(NVS_XDR_DATA_LEN(x) * 2) + \
	NV_ALIGN4((NVS_XDR_DATA_LEN(x) / 4)))

	static int
	nvs_xdr_nvpair(nvstream_t nvs, nvpair_t nvp, size_t *size)
	{
	XDR *xdr = nvs->nvs_private;
	int32_t encode_len, decode_len;

	switch (nvs->nvs_op) {
	case NVS_OP_ENCODE: {
	size_t nvsize;

	if (nvs_xdr_nvp_size(nvs, nvp, &nvsize) != 0)
	return (EFAULT);

	decode_len = nvp->nvp_size;
	encode_len = nvsize;
	if (!xdr_int(xdr, &encode_len) \|\| !xdr_int(xdr, &decode_len))
	return (EFAULT);

	return (nvs_xdr_nvp_op(nvs, nvp));
	}
	case NVS_OP_DECODE: {
	struct xdr_bytesrec bytesrec;

	/* get the encode and decode size */
	if (!xdr_int(xdr, &encode_len) \|\| !xdr_int(xdr, &decode_len))
	return (EFAULT);
	*size = decode_len;

	/* are we at the end of the stream? */
	if (*size == 0)
	return (0);

	/* sanity check the size parameter */
	if (!xdr_control(xdr, XDR_GET_BYTES_AVAIL, &bytesrec))
	return (EFAULT);

	if (*size > NVS_XDR_MAX_LEN(bytesrec.xc_num_avail))
	return (EFAULT);
	break;
	}

	default:
	return (EINVAL);
	}
	return (0);
	}

	static const struct nvs_ops nvs_xdr_ops = {
	.nvs_nvlist = nvs_xdr_nvlist,
	.nvs_nvpair = nvs_xdr_nvpair,
	.nvs_nvp_op = nvs_xdr_nvp_op,
	.nvs_nvp_size = nvs_xdr_nvp_size,
	.nvs_nvl_fini = nvs_xdr_nvl_fini
	};

	static int
	nvs_xdr(nvstream_t nvs, nvlist_t nvl, char buf, size_t buflen)
	{
	XDR xdr;
	int err;

	nvs->nvs_ops = &nvs_xdr_ops;

	if ((err = nvs_xdr_create(nvs, &xdr, buf + sizeof (nvs_header_t),
	*buflen - sizeof (nvs_header_t))) != 0)
	return (err);

	err = nvs_operation(nvs, nvl, buflen);

	nvs_xdr_destroy(nvs);

	return (err);
	}

	#if defined(_KERNEL)
	static int __init
	nvpair_init(void)
	{
	return (0);
	}

	static void __exit
	nvpair_fini(void)
	{
	}

	module_init(nvpair_init);
	module_exit(nvpair_fini);
	#endif

	ZFS_MODULE_DESCRIPTION("Generic name/value pair implementation");
	ZFS_MODULE_AUTHOR(ZFS_META_AUTHOR);
	ZFS_MODULE_LICENSE(ZFS_META_LICENSE);
	ZFS_MODULE_VERSION(ZFS_META_VERSION "-" ZFS_META_RELEASE);

	EXPORT_SYMBOL(nv_alloc_init);
	EXPORT_SYMBOL(nv_alloc_reset);
	EXPORT_SYMBOL(nv_alloc_fini);

	/* list management */
	EXPORT_SYMBOL(nvlist_alloc);
	EXPORT_SYMBOL(nvlist_free);
	EXPORT_SYMBOL(nvlist_size);
	EXPORT_SYMBOL(nvlist_pack);
	EXPORT_SYMBOL(nvlist_unpack);
	EXPORT_SYMBOL(nvlist_dup);
	EXPORT_SYMBOL(nvlist_merge);

	EXPORT_SYMBOL(nvlist_xalloc);
	EXPORT_SYMBOL(nvlist_xpack);
	EXPORT_SYMBOL(nvlist_xunpack);
	EXPORT_SYMBOL(nvlist_xdup);
	EXPORT_SYMBOL(nvlist_lookup_nv_alloc);

	EXPORT_SYMBOL(nvlist_add_nvpair);
	EXPORT_SYMBOL(nvlist_add_boolean);
	EXPORT_SYMBOL(nvlist_add_boolean_value);
	EXPORT_SYMBOL(nvlist_add_byte);
	EXPORT_SYMBOL(nvlist_add_int8);
	EXPORT_SYMBOL(nvlist_add_uint8);
	EXPORT_SYMBOL(nvlist_add_int16);
	EXPORT_SYMBOL(nvlist_add_uint16);
	EXPORT_SYMBOL(nvlist_add_int32);
	EXPORT_SYMBOL(nvlist_add_uint32);
	EXPORT_SYMBOL(nvlist_add_int64);
	EXPORT_SYMBOL(nvlist_add_uint64);
	EXPORT_SYMBOL(nvlist_add_string);
	EXPORT_SYMBOL(nvlist_add_nvlist);
	EXPORT_SYMBOL(nvlist_add_boolean_array);
	EXPORT_SYMBOL(nvlist_add_byte_array);
	EXPORT_SYMBOL(nvlist_add_int8_array);
	EXPORT_SYMBOL(nvlist_add_uint8_array);
	EXPORT_SYMBOL(nvlist_add_int16_array);
	EXPORT_SYMBOL(nvlist_add_uint16_array);
	EXPORT_SYMBOL(nvlist_add_int32_array);
	EXPORT_SYMBOL(nvlist_add_uint32_array);
	EXPORT_SYMBOL(nvlist_add_int64_array);
	EXPORT_SYMBOL(nvlist_add_uint64_array);
	EXPORT_SYMBOL(nvlist_add_string_array);
	EXPORT_SYMBOL(nvlist_add_nvlist_array);
	EXPORT_SYMBOL(nvlist_next_nvpair);
	EXPORT_SYMBOL(nvlist_prev_nvpair);
	EXPORT_SYMBOL(nvlist_empty);
	EXPORT_SYMBOL(nvlist_add_hrtime);

	EXPORT_SYMBOL(nvlist_remove);
	EXPORT_SYMBOL(nvlist_remove_nvpair);
	EXPORT_SYMBOL(nvlist_remove_all);

	EXPORT_SYMBOL(nvlist_lookup_boolean);
	EXPORT_SYMBOL(nvlist_lookup_boolean_value);
	EXPORT_SYMBOL(nvlist_lookup_byte);
	EXPORT_SYMBOL(nvlist_lookup_int8);
	EXPORT_SYMBOL(nvlist_lookup_uint8);
	EXPORT_SYMBOL(nvlist_lookup_int16);
	EXPORT_SYMBOL(nvlist_lookup_uint16);
	EXPORT_SYMBOL(nvlist_lookup_int32);
	EXPORT_SYMBOL(nvlist_lookup_uint32);
	EXPORT_SYMBOL(nvlist_lookup_int64);
	EXPORT_SYMBOL(nvlist_lookup_uint64);
	EXPORT_SYMBOL(nvlist_lookup_string);
	EXPORT_SYMBOL(nvlist_lookup_nvlist);
	EXPORT_SYMBOL(nvlist_lookup_boolean_array);
	EXPORT_SYMBOL(nvlist_lookup_byte_array);
	EXPORT_SYMBOL(nvlist_lookup_int8_array);
	EXPORT_SYMBOL(nvlist_lookup_uint8_array);
	EXPORT_SYMBOL(nvlist_lookup_int16_array);
	EXPORT_SYMBOL(nvlist_lookup_uint16_array);
	EXPORT_SYMBOL(nvlist_lookup_int32_array);
	EXPORT_SYMBOL(nvlist_lookup_uint32_array);
	EXPORT_SYMBOL(nvlist_lookup_int64_array);
	EXPORT_SYMBOL(nvlist_lookup_uint64_array);
	EXPORT_SYMBOL(nvlist_lookup_string_array);
	EXPORT_SYMBOL(nvlist_lookup_nvlist_array);
	EXPORT_SYMBOL(nvlist_lookup_hrtime);
	EXPORT_SYMBOL(nvlist_lookup_pairs);

	EXPORT_SYMBOL(nvlist_lookup_nvpair);
	EXPORT_SYMBOL(nvlist_exists);

	/* processing nvpair */
	EXPORT_SYMBOL(nvpair_name);
	EXPORT_SYMBOL(nvpair_type);
	EXPORT_SYMBOL(nvpair_value_boolean_value);
	EXPORT_SYMBOL(nvpair_value_byte);
	EXPORT_SYMBOL(nvpair_value_int8);
	EXPORT_SYMBOL(nvpair_value_uint8);
	EXPORT_SYMBOL(nvpair_value_int16);
	EXPORT_SYMBOL(nvpair_value_uint16);
	EXPORT_SYMBOL(nvpair_value_int32);
	EXPORT_SYMBOL(nvpair_value_uint32);
	EXPORT_SYMBOL(nvpair_value_int64);
	EXPORT_SYMBOL(nvpair_value_uint64);
	EXPORT_SYMBOL(nvpair_value_string);
	EXPORT_SYMBOL(nvpair_value_nvlist);
	EXPORT_SYMBOL(nvpair_value_boolean_array);
	EXPORT_SYMBOL(nvpair_value_byte_array);
	EXPORT_SYMBOL(nvpair_value_int8_array);
	EXPORT_SYMBOL(nvpair_value_uint8_array);
	EXPORT_SYMBOL(nvpair_value_int16_array);
	EXPORT_SYMBOL(nvpair_value_uint16_array);
	EXPORT_SYMBOL(nvpair_value_int32_array);
	EXPORT_SYMBOL(nvpair_value_uint32_array);
	EXPORT_SYMBOL(nvpair_value_int64_array);
	EXPORT_SYMBOL(nvpair_value_uint64_array);
	EXPORT_SYMBOL(nvpair_value_string_array);
	EXPORT_SYMBOL(nvpair_value_nvlist_array);
	EXPORT_SYMBOL(nvpair_value_hrtime);
	diff --git a/sys/contrib/openzfs/module/os/freebsd/spl/spl_taskq.c b/sys/contrib/openzfs/module/os/freebsd/spl/spl_taskq.c
	index 8ad6de9b5e9f..b31810d57f59 100644
	--- a/sys/contrib/openzfs/module/os/freebsd/spl/spl_taskq.c
	+++ b/sys/contrib/openzfs/module/os/freebsd/spl/spl_taskq.c
	@@ -1,444 +1,449 @@
	/*
	* Copyright (c) 2009 Pawel Jakub Dawidek <pjd@FreeBSD.org>
	* All rights reserved.
	*
	* Copyright (c) 2012 Spectra Logic Corporation. All rights reserved.
	*
	* Redistribution and use in source and binary forms, with or without
	* modification, are permitted provided that the following conditions
	* are met:
	* 1. Redistributions of source code must retain the above copyright
	* notice, this list of conditions and the following disclaimer.
	* 2. Redistributions in binary form must reproduce the above copyright
	* notice, this list of conditions and the following disclaimer in the
	* documentation and/or other materials provided with the distribution.
	*
	* THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
	* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
	* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
	* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
	* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
	* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
	* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
	* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
	* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
	* SUCH DAMAGE.
	*/

	#include <sys/cdefs.h>
	__FBSDID("$FreeBSD$");

	#include <sys/param.h>
	#include <sys/ck.h>
	#include <sys/epoch.h>
	#include <sys/kernel.h>
	#include <sys/kmem.h>
	#include <sys/lock.h>
	#include <sys/mutex.h>
	#include <sys/queue.h>
	#include <sys/taskq.h>
	#include <sys/taskqueue.h>
	#include <sys/zfs_context.h>

	#if defined(__i386__) \|\| defined(__amd64__) \|\| defined(__aarch64__)
	#include <machine/pcb.h>
	#endif

	#include <vm/uma.h>

	#if __FreeBSD_version < 1201522
	#define taskqueue_start_threads_in_proc(tqp, count, pri, proc, name, ...) \
	taskqueue_start_threads(tqp, count, pri, name, __VA_ARGS__)
	#endif

	static uint_t taskq_tsd;
	static uma_zone_t taskq_zone;

	+/*
	+ * Global system-wide dynamic task queue available for all consumers. This
	+ * taskq is not intended for long-running tasks; instead, a dedicated taskq
	+ * should be created.
	+ */
	taskq_t *system_taskq = NULL;
	taskq_t *system_delay_taskq = NULL;
	taskq_t *dynamic_taskq = NULL;

	proc_t *system_proc;

	extern int uma_align_cache;

	static MALLOC_DEFINE(M_TASKQ, "taskq", "taskq structures");

	static CK_LIST_HEAD(tqenthashhead, taskq_ent) *tqenthashtbl;
	static unsigned long tqenthash;
	static unsigned long tqenthashlock;
	static struct sx *tqenthashtbl_lock;

	static taskqid_t tqidnext;

	#define TQIDHASH(tqid) (&tqenthashtbl[(tqid) & tqenthash])
	#define TQIDHASHLOCK(tqid) (&tqenthashtbl_lock[((tqid) & tqenthashlock)])

	#define TIMEOUT_TASK 1
	#define NORMAL_TASK 2

	static void
	system_taskq_init(void *arg)
	{
	int i;

	tsd_create(&taskq_tsd, NULL);
	tqenthashtbl = hashinit(mp_ncpus * 8, M_TASKQ, &tqenthash);
	tqenthashlock = (tqenthash + 1) / 8;
	if (tqenthashlock > 0)
	tqenthashlock--;
	tqenthashtbl_lock =
	malloc(sizeof (tqenthashtbl_lock) (tqenthashlock + 1),
	M_TASKQ, M_WAITOK \| M_ZERO);
	for (i = 0; i < tqenthashlock + 1; i++)
	sx_init_flags(&tqenthashtbl_lock[i], "tqenthash", SX_DUPOK);
	taskq_zone = uma_zcreate("taskq_zone", sizeof (taskq_ent_t),
	NULL, NULL, NULL, NULL,
	UMA_ALIGN_CACHE, 0);
	system_taskq = taskq_create("system_taskq", mp_ncpus, minclsyspri,
	0, 0, 0);
	system_delay_taskq = taskq_create("system_delay_taskq", mp_ncpus,
	minclsyspri, 0, 0, 0);
	}
	SYSINIT(system_taskq_init, SI_SUB_CONFIGURE, SI_ORDER_ANY, system_taskq_init,
	NULL);

	static void
	system_taskq_fini(void *arg)
	{
	int i;

	taskq_destroy(system_delay_taskq);
	taskq_destroy(system_taskq);
	uma_zdestroy(taskq_zone);
	tsd_destroy(&taskq_tsd);
	for (i = 0; i < tqenthashlock + 1; i++)
	sx_destroy(&tqenthashtbl_lock[i]);
	for (i = 0; i < tqenthash + 1; i++)
	VERIFY(CK_LIST_EMPTY(&tqenthashtbl[i]));
	free(tqenthashtbl_lock, M_TASKQ);
	free(tqenthashtbl, M_TASKQ);
	}
	SYSUNINIT(system_taskq_fini, SI_SUB_CONFIGURE, SI_ORDER_ANY, system_taskq_fini,
	NULL);

	#ifdef __LP64__
	static taskqid_t
	__taskq_genid(void)
	{
	taskqid_t tqid;

	/*
	* Assume a 64-bit counter will not wrap in practice.
	*/
	tqid = atomic_add_64_nv(&tqidnext, 1);
	VERIFY(tqid);
	return (tqid);
	}
	#else
	static taskqid_t
	__taskq_genid(void)
	{
	taskqid_t tqid;

	for (;;) {
	tqid = atomic_add_32_nv(&tqidnext, 1);
	if (__predict_true(tqid != 0))
	break;
	}
	VERIFY(tqid);
	return (tqid);
	}
	#endif

	static taskq_ent_t *
	taskq_lookup(taskqid_t tqid)
	{
	taskq_ent_t *ent = NULL;

	sx_xlock(TQIDHASHLOCK(tqid));
	CK_LIST_FOREACH(ent, TQIDHASH(tqid), tqent_hash) {
	if (ent->tqent_id == tqid)
	break;
	}
	if (ent != NULL)
	refcount_acquire(&ent->tqent_rc);
	sx_xunlock(TQIDHASHLOCK(tqid));
	return (ent);
	}

	static taskqid_t
	taskq_insert(taskq_ent_t *ent)
	{
	taskqid_t tqid;

	tqid = __taskq_genid();
	ent->tqent_id = tqid;
	ent->tqent_registered = B_TRUE;
	sx_xlock(TQIDHASHLOCK(tqid));
	CK_LIST_INSERT_HEAD(TQIDHASH(tqid), ent, tqent_hash);
	sx_xunlock(TQIDHASHLOCK(tqid));
	return (tqid);
	}

	static void
	taskq_remove(taskq_ent_t *ent)
	{
	taskqid_t tqid = ent->tqent_id;

	if (!ent->tqent_registered)
	return;

	sx_xlock(TQIDHASHLOCK(tqid));
	CK_LIST_REMOVE(ent, tqent_hash);
	sx_xunlock(TQIDHASHLOCK(tqid));
	ent->tqent_registered = B_FALSE;
	}

	static void
	taskq_tsd_set(void *context)
	{
	taskq_t *tq = context;

	#if defined(__amd64__) \|\| defined(__aarch64__)
	if (context != NULL && tsd_get(taskq_tsd) == NULL)
	fpu_kern_thread(FPU_KERN_NORMAL);
	#endif
	tsd_set(taskq_tsd, tq);
	}

	static taskq_t *
	taskq_create_impl(const char *name, int nthreads, pri_t pri,
	proc_t *proc __maybe_unused, uint_t flags)
	{
	taskq_t *tq;

	if ((flags & TASKQ_THREADS_CPU_PCT) != 0)
	nthreads = MAX((mp_ncpus * nthreads) / 100, 1);

	tq = kmem_alloc(sizeof (*tq), KM_SLEEP);
	tq->tq_queue = taskqueue_create(name, M_WAITOK,
	taskqueue_thread_enqueue, &tq->tq_queue);
	taskqueue_set_callback(tq->tq_queue, TASKQUEUE_CALLBACK_TYPE_INIT,
	taskq_tsd_set, tq);
	taskqueue_set_callback(tq->tq_queue, TASKQUEUE_CALLBACK_TYPE_SHUTDOWN,
	taskq_tsd_set, NULL);
	(void) taskqueue_start_threads_in_proc(&tq->tq_queue, nthreads, pri,
	proc, "%s", name);

	return ((taskq_t *)tq);
	}

	taskq_t *
	taskq_create(const char *name, int nthreads, pri_t pri, int minalloc __unused,
	int maxalloc __unused, uint_t flags)
	{
	return (taskq_create_impl(name, nthreads, pri, system_proc, flags));
	}

	taskq_t *
	taskq_create_proc(const char *name, int nthreads, pri_t pri,
	int minalloc __unused, int maxalloc __unused, proc_t *proc, uint_t flags)
	{
	return (taskq_create_impl(name, nthreads, pri, proc, flags));
	}

	void
	taskq_destroy(taskq_t *tq)
	{

	taskqueue_free(tq->tq_queue);
	kmem_free(tq, sizeof (*tq));
	}

	int
	taskq_member(taskq_t tq, kthread_t thread)
	{

	return (taskqueue_member(tq->tq_queue, thread));
	}

	taskq_t *
	taskq_of_curthread(void)
	{
	return (tsd_get(taskq_tsd));
	}

	static void
	taskq_free(taskq_ent_t *task)
	{
	taskq_remove(task);
	if (refcount_release(&task->tqent_rc))
	uma_zfree(taskq_zone, task);
	}

	int
	taskq_cancel_id(taskq_t *tq, taskqid_t tid)
	{
	uint32_t pend;
	int rc;
	taskq_ent_t *ent;

	if (tid == 0)
	return (0);

	if ((ent = taskq_lookup(tid)) == NULL)
	return (0);

	ent->tqent_cancelled = B_TRUE;
	if (ent->tqent_type == TIMEOUT_TASK) {
	rc = taskqueue_cancel_timeout(tq->tq_queue,
	&ent->tqent_timeout_task, &pend);
	} else
	rc = taskqueue_cancel(tq->tq_queue, &ent->tqent_task, &pend);
	if (rc == EBUSY) {
	taskqueue_drain(tq->tq_queue, &ent->tqent_task);
	} else if (pend) {
	/*
	* Tasks normally free themselves when run, but here the task
	* was cancelled so it did not free itself.
	*/
	taskq_free(ent);
	}
	/* Free the extra reference we added with taskq_lookup. */
	taskq_free(ent);
	return (rc);
	}

	static void
	taskq_run(void *arg, int pending __unused)
	{
	taskq_ent_t *task = arg;

	if (!task->tqent_cancelled)
	task->tqent_func(task->tqent_arg);
	taskq_free(task);
	}

	taskqid_t
	taskq_dispatch_delay(taskq_t tq, task_func_t func, void arg,
	uint_t flags, clock_t expire_time)
	{
	taskq_ent_t *task;
	taskqid_t tqid;
	clock_t timo;
	int mflag;

	timo = expire_time - ddi_get_lbolt();
	if (timo <= 0)
	return (taskq_dispatch(tq, func, arg, flags));

	if ((flags & (TQ_SLEEP \| TQ_NOQUEUE)) == TQ_SLEEP)
	mflag = M_WAITOK;
	else
	mflag = M_NOWAIT;

	task = uma_zalloc(taskq_zone, mflag);
	if (task == NULL)
	return (0);
	task->tqent_func = func;
	task->tqent_arg = arg;
	task->tqent_type = TIMEOUT_TASK;
	task->tqent_cancelled = B_FALSE;
	refcount_init(&task->tqent_rc, 1);
	tqid = taskq_insert(task);
	TIMEOUT_TASK_INIT(tq->tq_queue, &task->tqent_timeout_task, 0,
	taskq_run, task);

	taskqueue_enqueue_timeout(tq->tq_queue, &task->tqent_timeout_task,
	timo);
	return (tqid);
	}

	taskqid_t
	taskq_dispatch(taskq_t tq, task_func_t func, void arg, uint_t flags)
	{
	taskq_ent_t *task;
	int mflag, prio;
	taskqid_t tqid;

	if ((flags & (TQ_SLEEP \| TQ_NOQUEUE)) == TQ_SLEEP)
	mflag = M_WAITOK;
	else
	mflag = M_NOWAIT;
	/*
	* If TQ_FRONT is given, we want higher priority for this task, so it
	* can go at the front of the queue.
	*/
	prio = !!(flags & TQ_FRONT);

	task = uma_zalloc(taskq_zone, mflag);
	if (task == NULL)
	return (0);
	refcount_init(&task->tqent_rc, 1);
	task->tqent_func = func;
	task->tqent_arg = arg;
	task->tqent_cancelled = B_FALSE;
	task->tqent_type = NORMAL_TASK;
	tqid = taskq_insert(task);
	TASK_INIT(&task->tqent_task, prio, taskq_run, task);
	taskqueue_enqueue(tq->tq_queue, &task->tqent_task);
	return (tqid);
	}

	static void
	taskq_run_ent(void *arg, int pending __unused)
	{
	taskq_ent_t *task = arg;

	task->tqent_func(task->tqent_arg);
	}

	void
	taskq_dispatch_ent(taskq_t tq, task_func_t func, void arg, uint32_t flags,
	taskq_ent_t *task)
	{
	int prio;

	/*
	* If TQ_FRONT is given, we want higher priority for this task, so it
	* can go at the front of the queue.
	*/
	prio = !!(flags & TQ_FRONT);
	task->tqent_cancelled = B_FALSE;
	task->tqent_registered = B_FALSE;
	task->tqent_id = 0;
	task->tqent_func = func;
	task->tqent_arg = arg;

	TASK_INIT(&task->tqent_task, prio, taskq_run_ent, task);
	taskqueue_enqueue(tq->tq_queue, &task->tqent_task);
	}

	void
	taskq_wait(taskq_t *tq)
	{
	taskqueue_quiesce(tq->tq_queue);
	}

	void
	taskq_wait_id(taskq_t *tq, taskqid_t tid)
	{
	taskq_ent_t *ent;

	if (tid == 0)
	return;
	if ((ent = taskq_lookup(tid)) == NULL)
	return;

	taskqueue_drain(tq->tq_queue, &ent->tqent_task);
	taskq_free(ent);
	}

	void
	taskq_wait_outstanding(taskq_t *tq, taskqid_t id __unused)
	{
	taskqueue_drain_all(tq->tq_queue);
	}

	int
	taskq_empty_ent(taskq_ent_t *t)
	{
	return (t->tqent_task.ta_pending == 0);
	}
	diff --git a/sys/contrib/openzfs/module/os/freebsd/zfs/spa_os.c b/sys/contrib/openzfs/module/os/freebsd/zfs/spa_os.c
	index 070e7a5b9f1b..85aecaacefdd 100644
	--- a/sys/contrib/openzfs/module/os/freebsd/zfs/spa_os.c
	+++ b/sys/contrib/openzfs/module/os/freebsd/zfs/spa_os.c
	@@ -1,272 +1,272 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011 by Delphix. All rights reserved.
	* Copyright (c) 2013 Martin Matuska <mm@FreeBSD.org>. All rights reserved.
	*/


	#include <sys/zfs_context.h>
	#include <sys/fm/fs/zfs.h>
	#include <sys/spa_impl.h>
	#include <sys/zio.h>
	#include <sys/zio_checksum.h>
	#include <sys/dmu.h>
	#include <sys/dmu_tx.h>
	#include <sys/zap.h>
	#include <sys/zil.h>
	#include <sys/ddt.h>
	#include <sys/vdev_impl.h>
	#include <sys/vdev_os.h>
	#include <sys/vdev_removal.h>
	#include <sys/vdev_indirect_mapping.h>
	#include <sys/vdev_indirect_births.h>
	#include <sys/metaslab.h>
	#include <sys/metaslab_impl.h>
	#include <sys/uberblock_impl.h>
	#include <sys/txg.h>
	#include <sys/avl.h>
	#include <sys/bpobj.h>
	#include <sys/dmu_traverse.h>
	#include <sys/dmu_objset.h>
	#include <sys/unique.h>
	#include <sys/dsl_pool.h>
	#include <sys/dsl_dataset.h>
	#include <sys/dsl_dir.h>
	#include <sys/dsl_prop.h>
	#include <sys/dsl_synctask.h>
	#include <sys/fs/zfs.h>
	#include <sys/arc.h>
	#include <sys/callb.h>
	#include <sys/spa_boot.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/dsl_scan.h>
	#include <sys/dmu_send.h>
	#include <sys/dsl_destroy.h>
	#include <sys/dsl_userhold.h>
	#include <sys/zfeature.h>
	#include <sys/zvol.h>
	#include <sys/abd.h>
	#include <sys/callb.h>
	#include <sys/zone.h>

	#include "zfs_prop.h"
	#include "zfs_comutil.h"

	static nvlist_t *
	spa_generate_rootconf(const char *name)
	{
	nvlist_t configs, tops;
	nvlist_t *config;
	nvlist_t best_cfg, nvtop, *nvroot;
	uint64_t *holes;
	uint64_t best_txg;
	uint64_t nchildren;
	uint64_t pgid;
	uint64_t count;
	uint64_t i;
	uint_t nholes;

	if (vdev_geom_read_pool_label(name, &configs, &count) != 0)
	return (NULL);

	ASSERT3U(count, !=, 0);
	best_txg = 0;
	for (i = 0; i < count; i++) {
	uint64_t txg;

	txg = fnvlist_lookup_uint64(configs[i], ZPOOL_CONFIG_POOL_TXG);
	if (txg > best_txg) {
	best_txg = txg;
	best_cfg = configs[i];
	}
	}

	nchildren = 1;
	nvlist_lookup_uint64(best_cfg, ZPOOL_CONFIG_VDEV_CHILDREN, &nchildren);
	holes = NULL;
	nvlist_lookup_uint64_array(best_cfg, ZPOOL_CONFIG_HOLE_ARRAY,
	&holes, &nholes);

	tops = kmem_zalloc(nchildren * sizeof (void *), KM_SLEEP);
	for (i = 0; i < nchildren; i++) {
	if (i >= count)
	break;
	if (configs[i] == NULL)
	continue;
	nvtop = fnvlist_lookup_nvlist(configs[i],
	ZPOOL_CONFIG_VDEV_TREE);
	tops[i] = fnvlist_dup(nvtop);
	}
	for (i = 0; holes != NULL && i < nholes; i++) {
	if (i >= nchildren)
	continue;
	if (tops[holes[i]] != NULL)
	continue;
	tops[holes[i]] = fnvlist_alloc();
	fnvlist_add_string(tops[holes[i]], ZPOOL_CONFIG_TYPE,
	VDEV_TYPE_HOLE);
	fnvlist_add_uint64(tops[holes[i]], ZPOOL_CONFIG_ID, holes[i]);
	fnvlist_add_uint64(tops[holes[i]], ZPOOL_CONFIG_GUID, 0);
	}
	for (i = 0; i < nchildren; i++) {
	if (tops[i] != NULL)
	continue;
	tops[i] = fnvlist_alloc();
	fnvlist_add_string(tops[i], ZPOOL_CONFIG_TYPE,
	VDEV_TYPE_MISSING);
	fnvlist_add_uint64(tops[i], ZPOOL_CONFIG_ID, i);
	fnvlist_add_uint64(tops[i], ZPOOL_CONFIG_GUID, 0);
	}

	/*
	* Create pool config based on the best vdev config.
	*/
	config = fnvlist_dup(best_cfg);

	/*
	* Put this pool's top-level vdevs into a root vdev.
	*/
	pgid = fnvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID);
	nvroot = fnvlist_alloc();
	fnvlist_add_string(nvroot, ZPOOL_CONFIG_TYPE, VDEV_TYPE_ROOT);
	fnvlist_add_uint64(nvroot, ZPOOL_CONFIG_ID, 0ULL);
	fnvlist_add_uint64(nvroot, ZPOOL_CONFIG_GUID, pgid);
	- fnvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, tops,
	- nchildren);
	+ fnvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN,
	+ (const nvlist_t * const *)tops, nchildren);

	/*
	* Replace the existing vdev_tree with the new root vdev in
	* this pool's configuration (remove the old, add the new).
	*/
	fnvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot);

	/*
	* Drop vdev config elements that should not be present at pool level.
	*/
	fnvlist_remove(config, ZPOOL_CONFIG_GUID);
	fnvlist_remove(config, ZPOOL_CONFIG_TOP_GUID);

	for (i = 0; i < count; i++)
	fnvlist_free(configs[i]);
	kmem_free(configs, count * sizeof (void *));
	for (i = 0; i < nchildren; i++)
	fnvlist_free(tops[i]);
	kmem_free(tops, nchildren * sizeof (void *));
	fnvlist_free(nvroot);
	return (config);
	}

	int
	spa_import_rootpool(const char *name, bool checkpointrewind)
	{
	spa_t *spa;
	vdev_t *rvd;
	nvlist_t config, nvtop;
	uint64_t txg;
	char *pname;
	int error;

	/*
	* Read the label from the boot device and generate a configuration.
	*/
	config = spa_generate_rootconf(name);

	mutex_enter(&spa_namespace_lock);
	if (config != NULL) {
	pname = fnvlist_lookup_string(config, ZPOOL_CONFIG_POOL_NAME);
	VERIFY0(strcmp(name, pname));
	txg = fnvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG);

	if ((spa = spa_lookup(pname)) != NULL) {
	/*
	* The pool could already be imported,
	* e.g., after reboot -r.
	*/
	if (spa->spa_state == POOL_STATE_ACTIVE) {
	mutex_exit(&spa_namespace_lock);
	fnvlist_free(config);
	return (0);
	}

	/*
	* Remove the existing root pool from the namespace so
	* that we can replace it with the correct config
	* we just read in.
	*/
	spa_remove(spa);
	}
	spa = spa_add(pname, config, NULL);

	/*
	* Set spa_ubsync.ub_version as it can be used in vdev_alloc()
	* via spa_version().
	*/
	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
	&spa->spa_ubsync.ub_version) != 0)
	spa->spa_ubsync.ub_version = SPA_VERSION_INITIAL;
	} else if ((spa = spa_lookup(name)) == NULL) {
	mutex_exit(&spa_namespace_lock);
	fnvlist_free(config);
	cmn_err(CE_NOTE, "Cannot find the pool label for '%s'",
	name);
	return (EIO);
	} else {
	config = fnvlist_dup(spa->spa_config);
	}
	spa->spa_is_root = B_TRUE;
	spa->spa_import_flags = ZFS_IMPORT_VERBATIM;
	if (checkpointrewind) {
	spa->spa_import_flags \|= ZFS_IMPORT_CHECKPOINT;
	}

	/*
	* Build up a vdev tree based on the boot device's label config.
	*/
	nvtop = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE);
	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	error = spa_config_parse(spa, &rvd, nvtop, NULL, 0,
	VDEV_ALLOC_ROOTPOOL);
	spa_config_exit(spa, SCL_ALL, FTAG);
	if (error) {
	mutex_exit(&spa_namespace_lock);
	fnvlist_free(config);
	cmn_err(CE_NOTE, "Can not parse the config for pool '%s'",
	pname);
	return (error);
	}

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	vdev_free(rvd);
	spa_config_exit(spa, SCL_ALL, FTAG);
	mutex_exit(&spa_namespace_lock);

	fnvlist_free(config);
	return (0);
	}

	const char *
	spa_history_zone(void)
	{
	return ("freebsd");
	}
	diff --git a/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_ctldir.c b/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_ctldir.c
	index 28d98e788263..11620949dec6 100644
	--- a/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_ctldir.c
	+++ b/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_ctldir.c
	@@ -1,1364 +1,1370 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2012, 2015 by Delphix. All rights reserved.
	* Copyright 2015, OmniTI Computer Consulting, Inc. All rights reserved.
	*/

	/*
	* ZFS control directory (a.k.a. ".zfs")
	*
	* This directory provides a common location for all ZFS meta-objects.
	* Currently, this is only the 'snapshot' directory, but this may expand in the
	* future. The elements are built using the GFS primitives, as the hierarchy
	* does not actually exist on disk.
	*
	* For 'snapshot', we don't want to have all snapshots always mounted, because
	* this would take up a huge amount of space in /etc/mnttab. We have three
	* types of objects:
	*
	* ctldir ------> snapshotdir -------> snapshot
	* \|
	* \|
	* V
	* mounted fs
	*
	* The 'snapshot' node contains just enough information to lookup '..' and act
	* as a mountpoint for the snapshot. Whenever we lookup a specific snapshot, we
	* perform an automount of the underlying filesystem and return the
	* corresponding vnode.
	*
	* All mounts are handled automatically by the kernel, but unmounts are
	* (currently) handled from user land. The main reason is that there is no
	* reliable way to auto-unmount the filesystem when it's "no longer in use".
	* When the user unmounts a filesystem, we call zfsctl_unmount(), which
	* unmounts any snapshots within the snapshot directory.
	*
	* The '.zfs', '.zfs/snapshot', and all directories created under
	* '.zfs/snapshot' (ie: '.zfs/snapshot/<snapname>') are all GFS nodes and
	* share the same vfs_t as the head filesystem (what '.zfs' lives under).
	*
	* File systems mounted ontop of the GFS nodes '.zfs/snapshot/<snapname>'
	* (ie: snapshots) are ZFS nodes and have their own unique vfs_t.
	* However, vnodes within these mounted on file systems have their v_vfsp
	* fields set to the head filesystem to make NFS happy (see
	* zfsctl_snapdir_lookup()). We VFS_HOLD the head filesystem's vfs_t
	* so that it cannot be freed until all snapshots have been unmounted.
	*/

	#include <sys/types.h>
	#include <sys/param.h>
	#include <sys/libkern.h>
	#include <sys/dirent.h>
	#include <sys/zfs_context.h>
	#include <sys/zfs_ctldir.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/zfs_vfsops.h>
	#include <sys/namei.h>
	#include <sys/stat.h>
	#include <sys/dmu.h>
	#include <sys/dsl_dataset.h>
	#include <sys/dsl_destroy.h>
	#include <sys/dsl_deleg.h>
	#include <sys/mount.h>
	#include <sys/zap.h>
	#include <sys/sysproto.h>

	#include "zfs_namecheck.h"

	#include <sys/kernel.h>
	#include <sys/ccompat.h>

	/* Common access mode for all virtual directories under the ctldir */
	const uint16_t zfsctl_ctldir_mode = S_IRUSR \| S_IXUSR \| S_IRGRP \| S_IXGRP \|
	S_IROTH \| S_IXOTH;

	/*
	* "Synthetic" filesystem implementation.
	*/

	/*
	* Assert that A implies B.
	*/
	#define KASSERT_IMPLY(A, B, msg) KASSERT(!(A) \|\| (B), (msg));

	static MALLOC_DEFINE(M_SFSNODES, "sfs_nodes", "synthetic-fs nodes");

	typedef struct sfs_node {
	char sn_name[ZFS_MAX_DATASET_NAME_LEN];
	uint64_t sn_parent_id;
	uint64_t sn_id;
	} sfs_node_t;

	/*
	* Check the parent's ID as well as the node's to account for a chance
	* that IDs originating from different domains (snapshot IDs, artificial
	* IDs, znode IDs) may clash.
	*/
	static int
	sfs_compare_ids(struct vnode vp, void arg)
	{
	sfs_node_t *n1 = vp->v_data;
	sfs_node_t *n2 = arg;
	bool equal;

	equal = n1->sn_id == n2->sn_id &&
	n1->sn_parent_id == n2->sn_parent_id;

	/* Zero means equality. */
	return (!equal);
	}

	static int
	sfs_vnode_get(const struct mount *mp, int flags, uint64_t parent_id,
	uint64_t id, struct vnode **vpp)
	{
	sfs_node_t search;
	int err;

	search.sn_id = id;
	search.sn_parent_id = parent_id;
	err = vfs_hash_get(mp, (uint32_t)id, flags, curthread, vpp,
	sfs_compare_ids, &search);
	return (err);
	}

	static int
	sfs_vnode_insert(struct vnode *vp, int flags, uint64_t parent_id,
	uint64_t id, struct vnode **vpp)
	{
	int err;

	KASSERT(vp->v_data != NULL, ("sfs_vnode_insert with NULL v_data"));
	err = vfs_hash_insert(vp, (uint32_t)id, flags, curthread, vpp,
	sfs_compare_ids, vp->v_data);
	return (err);
	}

	static void
	sfs_vnode_remove(struct vnode *vp)
	{
	vfs_hash_remove(vp);
	}

	typedef void sfs_vnode_setup_fn(vnode_t vp, void arg);

	static int
	sfs_vgetx(struct mount *mp, int flags, uint64_t parent_id, uint64_t id,
	const char tag, struct vop_vector vops,
	sfs_vnode_setup_fn setup, void *arg,
	struct vnode **vpp)
	{
	struct vnode *vp;
	int error;

	error = sfs_vnode_get(mp, flags, parent_id, id, vpp);
	if (error != 0 \|\| *vpp != NULL) {
	KASSERT_IMPLY(error == 0, (*vpp)->v_data != NULL,
	"sfs vnode with no data");
	return (error);
	}

	/* Allocate a new vnode/inode. */
	error = getnewvnode(tag, mp, vops, &vp);
	if (error != 0) {
	*vpp = NULL;
	return (error);
	}

	/*
	* Exclusively lock the vnode vnode while it's being constructed.
	*/
	lockmgr(vp->v_vnlock, LK_EXCLUSIVE, NULL);
	error = insmntque(vp, mp);
	if (error != 0) {
	*vpp = NULL;
	return (error);
	}

	setup(vp, arg);

	error = sfs_vnode_insert(vp, flags, parent_id, id, vpp);
	if (error != 0 \|\| *vpp != NULL) {
	KASSERT_IMPLY(error == 0, (*vpp)->v_data != NULL,
	"sfs vnode with no data");
	return (error);
	}

	*vpp = vp;
	return (0);
	}

	static void
	sfs_print_node(sfs_node_t *node)
	{
	printf("\tname = %s\n", node->sn_name);
	printf("\tparent_id = %ju\n", (uintmax_t)node->sn_parent_id);
	printf("\tid = %ju\n", (uintmax_t)node->sn_id);
	}

	static sfs_node_t *
	sfs_alloc_node(size_t size, const char *name, uint64_t parent_id, uint64_t id)
	{
	struct sfs_node *node;

	KASSERT(strlen(name) < sizeof (node->sn_name),
	("sfs node name is too long"));
	KASSERT(size >= sizeof (*node), ("sfs node size is too small"));
	node = malloc(size, M_SFSNODES, M_WAITOK \| M_ZERO);
	strlcpy(node->sn_name, name, sizeof (node->sn_name));
	node->sn_parent_id = parent_id;
	node->sn_id = id;

	return (node);
	}

	static void
	sfs_destroy_node(sfs_node_t *node)
	{
	free(node, M_SFSNODES);
	}

	static void *
	sfs_reclaim_vnode(vnode_t *vp)
	{
	void *data;

	sfs_vnode_remove(vp);
	data = vp->v_data;
	vp->v_data = NULL;
	return (data);
	}

	static int
	sfs_readdir_common(uint64_t parent_id, uint64_t id, struct vop_readdir_args *ap,
	zfs_uio_t uio, off_t offp)
	{
	struct dirent entry;
	int error;

	/* Reset ncookies for subsequent use of vfs_read_dirent. */
	if (ap->a_ncookies != NULL)
	*ap->a_ncookies = 0;

	if (zfs_uio_resid(uio) < sizeof (entry))
	return (SET_ERROR(EINVAL));

	if (zfs_uio_offset(uio) < 0)
	return (SET_ERROR(EINVAL));
	if (zfs_uio_offset(uio) == 0) {
	entry.d_fileno = id;
	entry.d_type = DT_DIR;
	entry.d_name[0] = '.';
	entry.d_name[1] = '\0';
	entry.d_namlen = 1;
	entry.d_reclen = sizeof (entry);
	error = vfs_read_dirent(ap, &entry, zfs_uio_offset(uio));
	if (error != 0)
	return (SET_ERROR(error));
	}

	if (zfs_uio_offset(uio) < sizeof (entry))
	return (SET_ERROR(EINVAL));
	if (zfs_uio_offset(uio) == sizeof (entry)) {
	entry.d_fileno = parent_id;
	entry.d_type = DT_DIR;
	entry.d_name[0] = '.';
	entry.d_name[1] = '.';
	entry.d_name[2] = '\0';
	entry.d_namlen = 2;
	entry.d_reclen = sizeof (entry);
	error = vfs_read_dirent(ap, &entry, zfs_uio_offset(uio));
	if (error != 0)
	return (SET_ERROR(error));
	}

	if (offp != NULL)
	offp = 2 sizeof (entry);
	return (0);
	}


	/*
	* .zfs inode namespace
	*
	* We need to generate unique inode numbers for all files and directories
	* within the .zfs pseudo-filesystem. We use the following scheme:
	*
	* ENTRY ZFSCTL_INODE
	* .zfs 1
	* .zfs/snapshot 2
	* .zfs/snapshot/<snap> objectid(snap)
	*/
	#define ZFSCTL_INO_SNAP(id) (id)

	static struct vop_vector zfsctl_ops_root;
	static struct vop_vector zfsctl_ops_snapdir;
	static struct vop_vector zfsctl_ops_snapshot;

	void
	zfsctl_init(void)
	{
	}

	void
	zfsctl_fini(void)
	{
	}

	boolean_t
	zfsctl_is_node(vnode_t *vp)
	{
	return (vn_matchops(vp, zfsctl_ops_root) \|\|
	vn_matchops(vp, zfsctl_ops_snapdir) \|\|
	vn_matchops(vp, zfsctl_ops_snapshot));

	}

	typedef struct zfsctl_root {
	sfs_node_t node;
	sfs_node_t *snapdir;
	timestruc_t cmtime;
	} zfsctl_root_t;


	/*
	* Create the '.zfs' directory.
	*/
	void
	zfsctl_create(zfsvfs_t *zfsvfs)
	{
	zfsctl_root_t *dot_zfs;
	sfs_node_t *snapdir;
	vnode_t *rvp;
	uint64_t crtime[2];

	ASSERT3P(zfsvfs->z_ctldir, ==, NULL);

	snapdir = sfs_alloc_node(sizeof (*snapdir), "snapshot", ZFSCTL_INO_ROOT,
	ZFSCTL_INO_SNAPDIR);
	dot_zfs = (zfsctl_root_t )sfs_alloc_node(sizeof (dot_zfs), ".zfs", 0,
	ZFSCTL_INO_ROOT);
	dot_zfs->snapdir = snapdir;

	VERIFY0(VFS_ROOT(zfsvfs->z_vfs, LK_EXCLUSIVE, &rvp));
	VERIFY0(sa_lookup(VTOZ(rvp)->z_sa_hdl, SA_ZPL_CRTIME(zfsvfs),
	&crtime, sizeof (crtime)));
	ZFS_TIME_DECODE(&dot_zfs->cmtime, crtime);
	vput(rvp);

	zfsvfs->z_ctldir = dot_zfs;
	}

	/*
	* Destroy the '.zfs' directory. Only called when the filesystem is unmounted.
	* The nodes must not have any associated vnodes by now as they should be
	* vflush-ed.
	*/
	void
	zfsctl_destroy(zfsvfs_t *zfsvfs)
	{
	sfs_destroy_node(zfsvfs->z_ctldir->snapdir);
	sfs_destroy_node((sfs_node_t *)zfsvfs->z_ctldir);
	zfsvfs->z_ctldir = NULL;
	}

	static int
	zfsctl_fs_root_vnode(struct mount mp, void arg __unused, int flags,
	struct vnode **vpp)
	{
	return (VFS_ROOT(mp, flags, vpp));
	}

	static void
	zfsctl_common_vnode_setup(vnode_t vp, void arg)
	{
	ASSERT_VOP_ELOCKED(vp, __func__);

	/* We support shared locking. */
	VN_LOCK_ASHARE(vp);
	vp->v_type = VDIR;
	vp->v_data = arg;
	}

	static int
	zfsctl_root_vnode(struct mount mp, void arg __unused, int flags,
	struct vnode **vpp)
	{
	void *node;
	int err;

	node = ((zfsvfs_t *)mp->mnt_data)->z_ctldir;
	err = sfs_vgetx(mp, flags, 0, ZFSCTL_INO_ROOT, "zfs", &zfsctl_ops_root,
	zfsctl_common_vnode_setup, node, vpp);
	return (err);
	}

	static int
	zfsctl_snapdir_vnode(struct mount mp, void arg __unused, int flags,
	struct vnode **vpp)
	{
	void *node;
	int err;

	node = ((zfsvfs_t *)mp->mnt_data)->z_ctldir->snapdir;
	err = sfs_vgetx(mp, flags, ZFSCTL_INO_ROOT, ZFSCTL_INO_SNAPDIR, "zfs",
	&zfsctl_ops_snapdir, zfsctl_common_vnode_setup, node, vpp);
	return (err);
	}

	/*
	* Given a root znode, retrieve the associated .zfs directory.
	* Add a hold to the vnode and return it.
	*/
	int
	zfsctl_root(zfsvfs_t zfsvfs, int flags, vnode_t *vpp)
	{
	int error;

	error = zfsctl_root_vnode(zfsvfs->z_vfs, NULL, flags, vpp);
	return (error);
	}

	/*
	* Common open routine. Disallow any write access.
	*/
	static int
	zfsctl_common_open(struct vop_open_args *ap)
	{
	int flags = ap->a_mode;

	if (flags & FWRITE)
	return (SET_ERROR(EACCES));

	return (0);
	}

	/*
	* Common close routine. Nothing to do here.
	*/
	/* ARGSUSED */
	static int
	zfsctl_common_close(struct vop_close_args *ap)
	{
	return (0);
	}

	/*
	* Common access routine. Disallow writes.
	*/
	static int
	zfsctl_common_access(struct vop_access_args *ap)
	{
	accmode_t accmode = ap->a_accmode;

	if (accmode & VWRITE)
	return (SET_ERROR(EACCES));
	return (0);
	}

	/*
	* Common getattr function. Fill in basic information.
	*/
	static void
	zfsctl_common_getattr(vnode_t vp, vattr_t vap)
	{
	timestruc_t now;
	sfs_node_t *node;

	node = vp->v_data;

	vap->va_uid = 0;
	vap->va_gid = 0;
	vap->va_rdev = 0;
	/*
	* We are a purely virtual object, so we have no
	* blocksize or allocated blocks.
	*/
	vap->va_blksize = 0;
	vap->va_nblocks = 0;
	vap->va_seq = 0;
	vn_fsid(vp, vap);
	vap->va_mode = zfsctl_ctldir_mode;
	vap->va_type = VDIR;
	/*
	* We live in the now (for atime).
	*/
	gethrestime(&now);
	vap->va_atime = now;
	/* FreeBSD: Reset chflags(2) flags. */
	vap->va_flags = 0;

	vap->va_nodeid = node->sn_id;

	/* At least '.' and '..'. */
	vap->va_nlink = 2;
	}

	#ifndef _OPENSOLARIS_SYS_VNODE_H_
	struct vop_fid_args {
	struct vnode *a_vp;
	struct fid *a_fid;
	};
	#endif

	static int
	zfsctl_common_fid(struct vop_fid_args *ap)
	{
	vnode_t *vp = ap->a_vp;
	fid_t fidp = (void )ap->a_fid;
	sfs_node_t *node = vp->v_data;
	uint64_t object = node->sn_id;
	zfid_short_t *zfid;
	int i;

	zfid = (zfid_short_t *)fidp;
	zfid->zf_len = SHORT_FID_LEN;

	for (i = 0; i < sizeof (zfid->zf_object); i++)
	zfid->zf_object[i] = (uint8_t)(object >> (8 * i));

	/* .zfs nodes always have a generation number of 0 */
	for (i = 0; i < sizeof (zfid->zf_gen); i++)
	zfid->zf_gen[i] = 0;

	return (0);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_reclaim_args {
	struct vnode *a_vp;
	struct thread *a_td;
	};
	#endif

	static int
	zfsctl_common_reclaim(struct vop_reclaim_args *ap)
	{
	vnode_t *vp = ap->a_vp;

	(void) sfs_reclaim_vnode(vp);
	return (0);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_print_args {
	struct vnode *a_vp;
	};
	#endif

	static int
	zfsctl_common_print(struct vop_print_args *ap)
	{
	sfs_print_node(ap->a_vp->v_data);
	return (0);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_getattr_args {
	struct vnode *a_vp;
	struct vattr *a_vap;
	struct ucred *a_cred;
	};
	#endif

	/*
	* Get root directory attributes.
	*/
	static int
	zfsctl_root_getattr(struct vop_getattr_args *ap)
	{
	struct vnode *vp = ap->a_vp;
	struct vattr *vap = ap->a_vap;
	zfsctl_root_t *node = vp->v_data;

	zfsctl_common_getattr(vp, vap);
	vap->va_ctime = node->cmtime;
	vap->va_mtime = vap->va_ctime;
	vap->va_birthtime = vap->va_ctime;
	vap->va_nlink += 1; /* snapdir */
	vap->va_size = vap->va_nlink;
	return (0);
	}

	/*
	* When we lookup "." we still can be asked to lock it
	* differently, can't we?
	*/
	static int
	zfsctl_relock_dot(vnode_t *dvp, int ltype)
	{
	vref(dvp);
	if (ltype != VOP_ISLOCKED(dvp)) {
	if (ltype == LK_EXCLUSIVE)
	vn_lock(dvp, LK_UPGRADE \| LK_RETRY);
	else /* if (ltype == LK_SHARED) */
	vn_lock(dvp, LK_DOWNGRADE \| LK_RETRY);

	/* Relock for the "." case may left us with reclaimed vnode. */
	if (VN_IS_DOOMED(dvp)) {
	vrele(dvp);
	return (SET_ERROR(ENOENT));
	}
	}
	return (0);
	}

	/*
	* Special case the handling of "..".
	*/
	static int
	zfsctl_root_lookup(struct vop_lookup_args *ap)
	{
	struct componentname *cnp = ap->a_cnp;
	vnode_t *dvp = ap->a_dvp;
	vnode_t **vpp = ap->a_vpp;
	int flags = ap->a_cnp->cn_flags;
	int lkflags = ap->a_cnp->cn_lkflags;
	int nameiop = ap->a_cnp->cn_nameiop;
	int err;

	ASSERT3S(dvp->v_type, ==, VDIR);

	if ((flags & ISLASTCN) != 0 && nameiop != LOOKUP)
	return (SET_ERROR(ENOTSUP));

	if (cnp->cn_namelen == 1 && *cnp->cn_nameptr == '.') {
	err = zfsctl_relock_dot(dvp, lkflags & LK_TYPE_MASK);
	if (err == 0)
	*vpp = dvp;
	} else if ((flags & ISDOTDOT) != 0) {
	err = vn_vget_ino_gen(dvp, zfsctl_fs_root_vnode, NULL,
	lkflags, vpp);
	} else if (strncmp(cnp->cn_nameptr, "snapshot", cnp->cn_namelen) == 0) {
	err = zfsctl_snapdir_vnode(dvp->v_mount, NULL, lkflags, vpp);
	} else {
	err = SET_ERROR(ENOENT);
	}
	if (err != 0)
	*vpp = NULL;
	return (err);
	}

	static int
	zfsctl_root_readdir(struct vop_readdir_args *ap)
	{
	struct dirent entry;
	vnode_t *vp = ap->a_vp;
	zfsvfs_t *zfsvfs = vp->v_vfsp->vfs_data;
	zfsctl_root_t *node = vp->v_data;
	zfs_uio_t uio;
	int *eofp = ap->a_eofflag;
	off_t dots_offset;
	int error;

	zfs_uio_init(&uio, ap->a_uio);

	ASSERT3S(vp->v_type, ==, VDIR);

	error = sfs_readdir_common(zfsvfs->z_root, ZFSCTL_INO_ROOT, ap, &uio,
	&dots_offset);
	if (error != 0) {
	if (error == ENAMETOOLONG) /* ran out of destination space */
	error = 0;
	return (error);
	}
	if (zfs_uio_offset(&uio) != dots_offset)
	return (SET_ERROR(EINVAL));

	CTASSERT(sizeof (node->snapdir->sn_name) <= sizeof (entry.d_name));
	entry.d_fileno = node->snapdir->sn_id;
	entry.d_type = DT_DIR;
	strcpy(entry.d_name, node->snapdir->sn_name);
	entry.d_namlen = strlen(entry.d_name);
	entry.d_reclen = sizeof (entry);
	error = vfs_read_dirent(ap, &entry, zfs_uio_offset(&uio));
	if (error != 0) {
	if (error == ENAMETOOLONG)
	error = 0;
	return (SET_ERROR(error));
	}
	if (eofp != NULL)
	*eofp = 1;
	return (0);
	}

	static int
	zfsctl_root_vptocnp(struct vop_vptocnp_args *ap)
	{
	static const char dotzfs_name[4] = ".zfs";
	vnode_t *dvp;
	int error;

	if (*ap->a_buflen < sizeof (dotzfs_name))
	return (SET_ERROR(ENOMEM));

	error = vn_vget_ino_gen(ap->a_vp, zfsctl_fs_root_vnode, NULL,
	LK_SHARED, &dvp);
	if (error != 0)
	return (SET_ERROR(error));

	VOP_UNLOCK1(dvp);
	*ap->a_vpp = dvp;
	*ap->a_buflen -= sizeof (dotzfs_name);
	bcopy(dotzfs_name, ap->a_buf + *ap->a_buflen, sizeof (dotzfs_name));
	return (0);
	}

	static int
	zfsctl_common_pathconf(struct vop_pathconf_args *ap)
	{
	/*
	* We care about ACL variables so that user land utilities like ls
	* can display them correctly. Since the ctldir's st_dev is set to be
	* the same as the parent dataset, we must support all variables that
	* it supports.
	*/
	switch (ap->a_name) {
	case _PC_LINK_MAX:
	*ap->a_retval = MIN(LONG_MAX, ZFS_LINK_MAX);
	return (0);

	case _PC_FILESIZEBITS:
	*ap->a_retval = 64;
	return (0);

	case _PC_MIN_HOLE_SIZE:
	*ap->a_retval = (int)SPA_MINBLOCKSIZE;
	return (0);

	case _PC_ACL_EXTENDED:
	*ap->a_retval = 0;
	return (0);

	case _PC_ACL_NFS4:
	*ap->a_retval = 1;
	return (0);

	case _PC_ACL_PATH_MAX:
	*ap->a_retval = ACL_MAX_ENTRIES;
	return (0);

	case _PC_NAME_MAX:
	*ap->a_retval = NAME_MAX;
	return (0);

	default:
	return (vop_stdpathconf(ap));
	}
	}

	/*
	* Returns a trivial ACL
	*/
	static int
	zfsctl_common_getacl(struct vop_getacl_args *ap)
	{
	int i;

	if (ap->a_type != ACL_TYPE_NFS4)
	return (EINVAL);

	acl_nfs4_sync_acl_from_mode(ap->a_aclp, zfsctl_ctldir_mode, 0);
	/*
	* acl_nfs4_sync_acl_from_mode assumes that the owner can always modify
	* attributes. That is not the case for the ctldir, so we must clear
	* those bits. We also must clear ACL_READ_NAMED_ATTRS, because xattrs
	* aren't supported by the ctldir.
	*/
	for (i = 0; i < ap->a_aclp->acl_cnt; i++) {
	struct acl_entry *entry;
	entry = &(ap->a_aclp->acl_entry[i]);
	entry->ae_perm &= ~(ACL_WRITE_ACL \| ACL_WRITE_OWNER \|
	ACL_WRITE_ATTRIBUTES \| ACL_WRITE_NAMED_ATTRS \|
	ACL_READ_NAMED_ATTRS);
	}

	return (0);
	}

	static struct vop_vector zfsctl_ops_root = {
	.vop_default = &default_vnodeops,
	#if __FreeBSD_version >= 1300121
	.vop_fplookup_vexec = VOP_EAGAIN,
	#endif
	.vop_open = zfsctl_common_open,
	.vop_close = zfsctl_common_close,
	.vop_ioctl = VOP_EINVAL,
	.vop_getattr = zfsctl_root_getattr,
	.vop_access = zfsctl_common_access,
	.vop_readdir = zfsctl_root_readdir,
	.vop_lookup = zfsctl_root_lookup,
	.vop_inactive = VOP_NULL,
	.vop_reclaim = zfsctl_common_reclaim,
	.vop_fid = zfsctl_common_fid,
	.vop_print = zfsctl_common_print,
	.vop_vptocnp = zfsctl_root_vptocnp,
	.vop_pathconf = zfsctl_common_pathconf,
	.vop_getacl = zfsctl_common_getacl,
	+#if __FreeBSD_version >= 1400043
	.vop_add_writecount = vop_stdadd_writecount_nomsync,
	+#endif
	};
	VFS_VOP_VECTOR_REGISTER(zfsctl_ops_root);

	static int
	zfsctl_snapshot_zname(vnode_t vp, const char name, int len, char *zname)
	{
	objset_t os = ((zfsvfs_t )((vp)->v_vfsp->vfs_data))->z_os;

	dmu_objset_name(os, zname);
	if (strlen(zname) + 1 + strlen(name) >= len)
	return (SET_ERROR(ENAMETOOLONG));
	(void) strcat(zname, "@");
	(void) strcat(zname, name);
	return (0);
	}

	static int
	zfsctl_snapshot_lookup(vnode_t vp, const char name, uint64_t *id)
	{
	objset_t os = ((zfsvfs_t )((vp)->v_vfsp->vfs_data))->z_os;
	int err;

	err = dsl_dataset_snap_lookup(dmu_objset_ds(os), name, id);
	return (err);
	}

	/*
	* Given a vnode get a root vnode of a filesystem mounted on top of
	* the vnode, if any. The root vnode is referenced and locked.
	* If no filesystem is mounted then the orinal vnode remains referenced
	* and locked. If any error happens the orinal vnode is unlocked and
	* released.
	*/
	static int
	zfsctl_mounted_here(vnode_t **vpp, int flags)
	{
	struct mount *mp;
	int err;

	ASSERT_VOP_LOCKED(*vpp, __func__);
	ASSERT3S((*vpp)->v_type, ==, VDIR);

	if ((mp = (*vpp)->v_mountedhere) != NULL) {
	err = vfs_busy(mp, 0);
	KASSERT(err == 0, ("vfs_busy(mp, 0) failed with %d", err));
	KASSERT(vrefcnt(*vpp) > 1, ("unreferenced mountpoint"));
	vput(*vpp);
	err = VFS_ROOT(mp, flags, vpp);
	vfs_unbusy(mp);
	return (err);
	}
	return (EJUSTRETURN);
	}

	typedef struct {
	const char *snap_name;
	uint64_t snap_id;
	} snapshot_setup_arg_t;

	static void
	zfsctl_snapshot_vnode_setup(vnode_t vp, void arg)
	{
	snapshot_setup_arg_t *ssa = arg;
	sfs_node_t *node;

	ASSERT_VOP_ELOCKED(vp, __func__);

	node = sfs_alloc_node(sizeof (sfs_node_t),
	ssa->snap_name, ZFSCTL_INO_SNAPDIR, ssa->snap_id);
	zfsctl_common_vnode_setup(vp, node);

	/* We have to support recursive locking. */
	VN_LOCK_AREC(vp);
	}

	/*
	* Lookup entry point for the 'snapshot' directory. Try to open the
	* snapshot if it exist, creating the pseudo filesystem vnode as necessary.
	* Perform a mount of the associated dataset on top of the vnode.
	* There are four possibilities:
	* - the snapshot node and vnode do not exist
	* - the snapshot vnode is covered by the mounted snapshot
	* - the snapshot vnode is not covered yet, the mount operation is in progress
	* - the snapshot vnode is not covered, because the snapshot has been unmounted
	* The last two states are transient and should be relatively short-lived.
	*/
	static int
	zfsctl_snapdir_lookup(struct vop_lookup_args *ap)
	{
	vnode_t *dvp = ap->a_dvp;
	vnode_t **vpp = ap->a_vpp;
	struct componentname *cnp = ap->a_cnp;
	char name[NAME_MAX + 1];
	char fullname[ZFS_MAX_DATASET_NAME_LEN];
	char *mountpoint;
	size_t mountpoint_len;
	zfsvfs_t *zfsvfs = dvp->v_vfsp->vfs_data;
	uint64_t snap_id;
	int nameiop = cnp->cn_nameiop;
	int lkflags = cnp->cn_lkflags;
	int flags = cnp->cn_flags;
	int err;

	ASSERT3S(dvp->v_type, ==, VDIR);

	if ((flags & ISLASTCN) != 0 && nameiop != LOOKUP)
	return (SET_ERROR(ENOTSUP));

	if (cnp->cn_namelen == 1 && *cnp->cn_nameptr == '.') {
	err = zfsctl_relock_dot(dvp, lkflags & LK_TYPE_MASK);
	if (err == 0)
	*vpp = dvp;
	return (err);
	}
	if (flags & ISDOTDOT) {
	err = vn_vget_ino_gen(dvp, zfsctl_root_vnode, NULL, lkflags,
	vpp);
	return (err);
	}

	if (cnp->cn_namelen >= sizeof (name))
	return (SET_ERROR(ENAMETOOLONG));

	strlcpy(name, ap->a_cnp->cn_nameptr, ap->a_cnp->cn_namelen + 1);
	err = zfsctl_snapshot_lookup(dvp, name, &snap_id);
	if (err != 0)
	return (SET_ERROR(ENOENT));

	for (;;) {
	snapshot_setup_arg_t ssa;

	ssa.snap_name = name;
	ssa.snap_id = snap_id;
	err = sfs_vgetx(dvp->v_mount, LK_SHARED, ZFSCTL_INO_SNAPDIR,
	snap_id, "zfs", &zfsctl_ops_snapshot,
	zfsctl_snapshot_vnode_setup, &ssa, vpp);
	if (err != 0)
	return (err);

	/* Check if a new vnode has just been created. */
	if (VOP_ISLOCKED(*vpp) == LK_EXCLUSIVE)
	break;

	/*
	* Check if a snapshot is already mounted on top of the vnode.
	*/
	err = zfsctl_mounted_here(vpp, lkflags);
	if (err != EJUSTRETURN)
	return (err);

	/*
	* If the vnode is not covered, then either the mount operation
	* is in progress or the snapshot has already been unmounted
	* but the vnode hasn't been inactivated and reclaimed yet.
	* We can try to re-use the vnode in the latter case.
	*/
	VI_LOCK(*vpp);
	if (((*vpp)->v_iflag & VI_MOUNT) == 0) {
	/*
	* Upgrade to exclusive lock in order to:
	* - avoid race conditions
	* - satisfy the contract of mount_snapshot()
	*/
	err = VOP_LOCK(*vpp, LK_TRYUPGRADE \| LK_INTERLOCK);
	if (err == 0)
	break;
	} else {
	VI_UNLOCK(*vpp);
	}

	/*
	* In this state we can loop on uncontested locks and starve
	* the thread doing the lengthy, non-trivial mount operation.
	* So, yield to prevent that from happening.
	*/
	vput(*vpp);
	kern_yield(PRI_USER);
	}

	VERIFY0(zfsctl_snapshot_zname(dvp, name, sizeof (fullname), fullname));

	mountpoint_len = strlen(dvp->v_vfsp->mnt_stat.f_mntonname) +
	strlen("/" ZFS_CTLDIR_NAME "/snapshot/") + strlen(name) + 1;
	mountpoint = kmem_alloc(mountpoint_len, KM_SLEEP);
	(void) snprintf(mountpoint, mountpoint_len,
	"%s/" ZFS_CTLDIR_NAME "/snapshot/%s",
	dvp->v_vfsp->mnt_stat.f_mntonname, name);

	err = mount_snapshot(curthread, vpp, "zfs", mountpoint, fullname, 0);
	kmem_free(mountpoint, mountpoint_len);
	if (err == 0) {
	/*
	* Fix up the root vnode mounted on .zfs/snapshot/<snapname>.
	*
	* This is where we lie about our v_vfsp in order to
	* make .zfs/snapshot/<snapname> accessible over NFS
	* without requiring manual mounts of <snapname>.
	*/
	ASSERT3P(VTOZ(*vpp)->z_zfsvfs, !=, zfsvfs);
	VTOZ(*vpp)->z_zfsvfs->z_parent = zfsvfs;

	/* Clear the root flag (set via VFS_ROOT) as well. */
	(*vpp)->v_vflag &= ~VV_ROOT;
	}

	if (err != 0)
	*vpp = NULL;
	return (err);
	}

	static int
	zfsctl_snapdir_readdir(struct vop_readdir_args *ap)
	{
	char snapname[ZFS_MAX_DATASET_NAME_LEN];
	struct dirent entry;
	vnode_t *vp = ap->a_vp;
	zfsvfs_t *zfsvfs = vp->v_vfsp->vfs_data;
	zfs_uio_t uio;
	int *eofp = ap->a_eofflag;
	off_t dots_offset;
	int error;

	zfs_uio_init(&uio, ap->a_uio);

	ASSERT3S(vp->v_type, ==, VDIR);

	error = sfs_readdir_common(ZFSCTL_INO_ROOT, ZFSCTL_INO_SNAPDIR, ap,
	&uio, &dots_offset);
	if (error != 0) {
	if (error == ENAMETOOLONG) /* ran out of destination space */
	error = 0;
	return (error);
	}

	ZFS_ENTER(zfsvfs);
	for (;;) {
	uint64_t cookie;
	uint64_t id;

	cookie = zfs_uio_offset(&uio) - dots_offset;

	dsl_pool_config_enter(dmu_objset_pool(zfsvfs->z_os), FTAG);
	error = dmu_snapshot_list_next(zfsvfs->z_os, sizeof (snapname),
	snapname, &id, &cookie, NULL);
	dsl_pool_config_exit(dmu_objset_pool(zfsvfs->z_os), FTAG);
	if (error != 0) {
	if (error == ENOENT) {
	if (eofp != NULL)
	*eofp = 1;
	error = 0;
	}
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	entry.d_fileno = id;
	entry.d_type = DT_DIR;
	strcpy(entry.d_name, snapname);
	entry.d_namlen = strlen(entry.d_name);
	entry.d_reclen = sizeof (entry);
	error = vfs_read_dirent(ap, &entry, zfs_uio_offset(&uio));
	if (error != 0) {
	if (error == ENAMETOOLONG)
	error = 0;
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(error));
	}
	zfs_uio_setoffset(&uio, cookie + dots_offset);
	}
	__builtin_unreachable();
	}

	static int
	zfsctl_snapdir_getattr(struct vop_getattr_args *ap)
	{
	vnode_t *vp = ap->a_vp;
	vattr_t *vap = ap->a_vap;
	zfsvfs_t *zfsvfs = vp->v_vfsp->vfs_data;
	dsl_dataset_t *ds;
	uint64_t snap_count;
	int err;

	ZFS_ENTER(zfsvfs);
	ds = dmu_objset_ds(zfsvfs->z_os);
	zfsctl_common_getattr(vp, vap);
	vap->va_ctime = dmu_objset_snap_cmtime(zfsvfs->z_os);
	vap->va_mtime = vap->va_ctime;
	vap->va_birthtime = vap->va_ctime;
	if (dsl_dataset_phys(ds)->ds_snapnames_zapobj != 0) {
	err = zap_count(dmu_objset_pool(ds->ds_objset)->dp_meta_objset,
	dsl_dataset_phys(ds)->ds_snapnames_zapobj, &snap_count);
	if (err != 0) {
	ZFS_EXIT(zfsvfs);
	return (err);
	}
	vap->va_nlink += snap_count;
	}
	vap->va_size = vap->va_nlink;

	ZFS_EXIT(zfsvfs);
	return (0);
	}

	static struct vop_vector zfsctl_ops_snapdir = {
	.vop_default = &default_vnodeops,
	#if __FreeBSD_version >= 1300121
	.vop_fplookup_vexec = VOP_EAGAIN,
	#endif
	.vop_open = zfsctl_common_open,
	.vop_close = zfsctl_common_close,
	.vop_getattr = zfsctl_snapdir_getattr,
	.vop_access = zfsctl_common_access,
	.vop_readdir = zfsctl_snapdir_readdir,
	.vop_lookup = zfsctl_snapdir_lookup,
	.vop_reclaim = zfsctl_common_reclaim,
	.vop_fid = zfsctl_common_fid,
	.vop_print = zfsctl_common_print,
	.vop_pathconf = zfsctl_common_pathconf,
	.vop_getacl = zfsctl_common_getacl,
	+#if __FreeBSD_version >= 1400043
	.vop_add_writecount = vop_stdadd_writecount_nomsync,
	+#endif
	};
	VFS_VOP_VECTOR_REGISTER(zfsctl_ops_snapdir);


	static int
	zfsctl_snapshot_inactive(struct vop_inactive_args *ap)
	{
	vnode_t *vp = ap->a_vp;

	VERIFY3S(vrecycle(vp), ==, 1);
	return (0);
	}

	static int
	zfsctl_snapshot_reclaim(struct vop_reclaim_args *ap)
	{
	vnode_t *vp = ap->a_vp;
	void *data = vp->v_data;

	sfs_reclaim_vnode(vp);
	sfs_destroy_node(data);
	return (0);
	}

	static int
	zfsctl_snapshot_vptocnp(struct vop_vptocnp_args *ap)
	{
	struct mount *mp;
	vnode_t *dvp;
	vnode_t *vp;
	sfs_node_t *node;
	size_t len;
	int locked;
	int error;

	vp = ap->a_vp;
	node = vp->v_data;
	len = strlen(node->sn_name);
	if (*ap->a_buflen < len)
	return (SET_ERROR(ENOMEM));

	/*
	* Prevent unmounting of the snapshot while the vnode lock
	* is not held. That is not strictly required, but allows
	* us to assert that an uncovered snapshot vnode is never
	* "leaked".
	*/
	mp = vp->v_mountedhere;
	if (mp == NULL)
	return (SET_ERROR(ENOENT));
	error = vfs_busy(mp, 0);
	KASSERT(error == 0, ("vfs_busy(mp, 0) failed with %d", error));

	/*
	* We can vput the vnode as we can now depend on the reference owned
	* by the busied mp. But we also need to hold the vnode, because
	* the reference may go after vfs_unbusy() which has to be called
	* before we can lock the vnode again.
	*/
	locked = VOP_ISLOCKED(vp);
	#if __FreeBSD_version >= 1300045
	enum vgetstate vs = vget_prep(vp);
	#else
	vhold(vp);
	#endif
	vput(vp);

	/* Look up .zfs/snapshot, our parent. */
	error = zfsctl_snapdir_vnode(vp->v_mount, NULL, LK_SHARED, &dvp);
	if (error == 0) {
	VOP_UNLOCK1(dvp);
	*ap->a_vpp = dvp;
	*ap->a_buflen -= len;
	bcopy(node->sn_name, ap->a_buf + *ap->a_buflen, len);
	}
	vfs_unbusy(mp);
	#if __FreeBSD_version >= 1300045
	vget_finish(vp, locked \| LK_RETRY, vs);
	#else
	vget(vp, locked \| LK_VNHELD \| LK_RETRY, curthread);
	#endif
	return (error);
	}

	/*
	* These VP's should never see the light of day. They should always
	* be covered.
	*/
	static struct vop_vector zfsctl_ops_snapshot = {
	.vop_default = NULL, /* ensure very restricted access */
	#if __FreeBSD_version >= 1300121
	.vop_fplookup_vexec = VOP_EAGAIN,
	#endif
	.vop_inactive = zfsctl_snapshot_inactive,
	#if __FreeBSD_version >= 1300045
	.vop_need_inactive = vop_stdneed_inactive,
	#endif
	.vop_reclaim = zfsctl_snapshot_reclaim,
	.vop_vptocnp = zfsctl_snapshot_vptocnp,
	.vop_lock1 = vop_stdlock,
	.vop_unlock = vop_stdunlock,
	.vop_islocked = vop_stdislocked,
	.vop_advlockpurge = vop_stdadvlockpurge, /* called by vgone */
	.vop_print = zfsctl_common_print,
	+#if __FreeBSD_version >= 1400043
	.vop_add_writecount = vop_stdadd_writecount_nomsync,
	+#endif
	};
	VFS_VOP_VECTOR_REGISTER(zfsctl_ops_snapshot);

	int
	zfsctl_lookup_objset(vfs_t vfsp, uint64_t objsetid, zfsvfs_t *zfsvfsp)
	{
	zfsvfs_t *zfsvfs __unused = vfsp->vfs_data;
	vnode_t *vp;
	int error;

	ASSERT3P(zfsvfs->z_ctldir, !=, NULL);
	*zfsvfsp = NULL;
	error = sfs_vnode_get(vfsp, LK_EXCLUSIVE,
	ZFSCTL_INO_SNAPDIR, objsetid, &vp);
	if (error == 0 && vp != NULL) {
	/*
	* XXX Probably need to at least reference, if not busy, the mp.
	*/
	if (vp->v_mountedhere != NULL)
	*zfsvfsp = vp->v_mountedhere->mnt_data;
	vput(vp);
	}
	if (*zfsvfsp == NULL)
	return (SET_ERROR(EINVAL));
	return (0);
	}

	/*
	* Unmount any snapshots for the given filesystem. This is called from
	* zfs_umount() - if we have a ctldir, then go through and unmount all the
	* snapshots.
	*/
	int
	zfsctl_umount_snapshots(vfs_t vfsp, int fflags, cred_t cr)
	{
	char snapname[ZFS_MAX_DATASET_NAME_LEN];
	zfsvfs_t *zfsvfs = vfsp->vfs_data;
	struct mount *mp;
	vnode_t *vp;
	uint64_t cookie;
	int error;

	ASSERT3P(zfsvfs->z_ctldir, !=, NULL);

	cookie = 0;
	for (;;) {
	uint64_t id;

	dsl_pool_config_enter(dmu_objset_pool(zfsvfs->z_os), FTAG);
	error = dmu_snapshot_list_next(zfsvfs->z_os, sizeof (snapname),
	snapname, &id, &cookie, NULL);
	dsl_pool_config_exit(dmu_objset_pool(zfsvfs->z_os), FTAG);
	if (error != 0) {
	if (error == ENOENT)
	error = 0;
	break;
	}

	for (;;) {
	error = sfs_vnode_get(vfsp, LK_EXCLUSIVE,
	ZFSCTL_INO_SNAPDIR, id, &vp);
	if (error != 0 \|\| vp == NULL)
	break;

	mp = vp->v_mountedhere;

	/*
	* v_mountedhere being NULL means that the
	* (uncovered) vnode is in a transient state
	* (mounting or unmounting), so loop until it
	* settles down.
	*/
	if (mp != NULL)
	break;
	vput(vp);
	}
	if (error != 0)
	break;
	if (vp == NULL)
	continue; /* no mountpoint, nothing to do */

	/*
	* The mount-point vnode is kept locked to avoid spurious EBUSY
	* from a concurrent umount.
	* The vnode lock must have recursive locking enabled.
	*/
	vfs_ref(mp);
	error = dounmount(mp, fflags, curthread);
	KASSERT_IMPLY(error == 0, vrefcnt(vp) == 1,
	("extra references after unmount"));
	vput(vp);
	if (error != 0)
	break;
	}
	KASSERT_IMPLY((fflags & MS_FORCE) != 0, error == 0,
	("force unmounting failed"));
	return (error);
	}

	int
	zfsctl_snapshot_unmount(const char *snapname, int flags __unused)
	{
	vfs_t *vfsp = NULL;
	zfsvfs_t *zfsvfs = NULL;

	if (strchr(snapname, '@') == NULL)
	return (0);

	int err = getzfsvfs(snapname, &zfsvfs);
	if (err != 0) {
	ASSERT3P(zfsvfs, ==, NULL);
	return (0);
	}
	vfsp = zfsvfs->z_vfs;

	ASSERT(!dsl_pool_config_held(dmu_objset_pool(zfsvfs->z_os)));

	vfs_ref(vfsp);
	vfs_unbusy(vfsp);
	return (dounmount(vfsp, MS_FORCE, curthread));
	}
	diff --git a/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_vnops_os.c b/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_vnops_os.c
	index 1dde0ac43f1a..5aa54c6d2745 100644
	--- a/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_vnops_os.c
	+++ b/sys/contrib/openzfs/module/os/freebsd/zfs/zfs_vnops_os.c
	@@ -1,6222 +1,6245 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2012, 2015 by Delphix. All rights reserved.
	* Copyright (c) 2014 Integros [integros.com]
	* Copyright 2017 Nexenta Systems, Inc.
	*/

	/* Portions Copyright 2007 Jeremy Teo */
	/* Portions Copyright 2010 Robert Milkowski */


	#include <sys/types.h>
	#include <sys/param.h>
	#include <sys/time.h>
	#include <sys/systm.h>
	#include <sys/sysmacros.h>
	#include <sys/resource.h>
	#include <sys/vfs.h>
	#include <sys/endian.h>
	#include <sys/vm.h>
	#include <sys/vnode.h>
	#if __FreeBSD_version >= 1300102
	#include <sys/smr.h>
	#endif
	#include <sys/dirent.h>
	#include <sys/file.h>
	#include <sys/stat.h>
	#include <sys/kmem.h>
	#include <sys/taskq.h>
	#include <sys/uio.h>
	#include <sys/atomic.h>
	#include <sys/namei.h>
	#include <sys/mman.h>
	#include <sys/cmn_err.h>
	#include <sys/kdb.h>
	#include <sys/sysproto.h>
	#include <sys/errno.h>
	#include <sys/unistd.h>
	#include <sys/zfs_dir.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/fs/zfs.h>
	#include <sys/dmu.h>
	#include <sys/dmu_objset.h>
	#include <sys/spa.h>
	#include <sys/txg.h>
	#include <sys/dbuf.h>
	#include <sys/zap.h>
	#include <sys/sa.h>
	#include <sys/policy.h>
	#include <sys/sunddi.h>
	#include <sys/filio.h>
	#include <sys/sid.h>
	#include <sys/zfs_ctldir.h>
	#include <sys/zfs_fuid.h>
	#include <sys/zfs_quota.h>
	#include <sys/zfs_sa.h>
	#include <sys/zfs_rlock.h>
	#include <sys/extdirent.h>
	#include <sys/bio.h>
	#include <sys/buf.h>
	#include <sys/sched.h>
	#include <sys/acl.h>
	#include <sys/vmmeter.h>
	#include <vm/vm_param.h>
	#include <sys/zil.h>
	#include <sys/zfs_vnops.h>

	#include <vm/vm_object.h>

	#include <sys/extattr.h>
	#include <sys/priv.h>

	#ifndef VN_OPEN_INVFS
	#define VN_OPEN_INVFS 0x0
	#endif

	VFS_SMR_DECLARE;

	#if __FreeBSD_version >= 1300047
	#define vm_page_wire_lock(pp)
	#define vm_page_wire_unlock(pp)
	#else
	#define vm_page_wire_lock(pp) vm_page_lock(pp)
	#define vm_page_wire_unlock(pp) vm_page_unlock(pp)
	#endif

	#ifdef DEBUG_VFS_LOCKS
	#define VNCHECKREF(vp) \
	VNASSERT((vp)->v_holdcnt > 0 && (vp)->v_usecount > 0, vp, \
	("%s: wrong ref counts", __func__));
	#else
	#define VNCHECKREF(vp)
	#endif

	/*
	* Programming rules.
	*
	* Each vnode op performs some logical unit of work. To do this, the ZPL must
	* properly lock its in-core state, create a DMU transaction, do the work,
	* record this work in the intent log (ZIL), commit the DMU transaction,
	* and wait for the intent log to commit if it is a synchronous operation.
	* Moreover, the vnode ops must work in both normal and log replay context.
	* The ordering of events is important to avoid deadlocks and references
	* to freed memory. The example below illustrates the following Big Rules:
	*
	* (1) A check must be made in each zfs thread for a mounted file system.
	* This is done avoiding races using ZFS_ENTER(zfsvfs).
	* A ZFS_EXIT(zfsvfs) is needed before all returns. Any znodes
	* must be checked with ZFS_VERIFY_ZP(zp). Both of these macros
	* can return EIO from the calling function.
	*
	* (2) VN_RELE() should always be the last thing except for zil_commit()
	* (if necessary) and ZFS_EXIT(). This is for 3 reasons:
	* First, if it's the last reference, the vnode/znode
	* can be freed, so the zp may point to freed memory. Second, the last
	* reference will call zfs_zinactive(), which may induce a lot of work --
	* pushing cached pages (which acquires range locks) and syncing out
	* cached atime changes. Third, zfs_zinactive() may require a new tx,
	* which could deadlock the system if you were already holding one.
	* If you must call VN_RELE() within a tx then use VN_RELE_ASYNC().
	*
	* (3) All range locks must be grabbed before calling dmu_tx_assign(),
	* as they can span dmu_tx_assign() calls.
	*
	* (4) If ZPL locks are held, pass TXG_NOWAIT as the second argument to
	* dmu_tx_assign(). This is critical because we don't want to block
	* while holding locks.
	*
	* If no ZPL locks are held (aside from ZFS_ENTER()), use TXG_WAIT. This
	* reduces lock contention and CPU usage when we must wait (note that if
	* throughput is constrained by the storage, nearly every transaction
	* must wait).
	*
	* Note, in particular, that if a lock is sometimes acquired before
	* the tx assigns, and sometimes after (e.g. z_lock), then failing
	* to use a non-blocking assign can deadlock the system. The scenario:
	*
	* Thread A has grabbed a lock before calling dmu_tx_assign().
	* Thread B is in an already-assigned tx, and blocks for this lock.
	* Thread A calls dmu_tx_assign(TXG_WAIT) and blocks in txg_wait_open()
	* forever, because the previous txg can't quiesce until B's tx commits.
	*
	* If dmu_tx_assign() returns ERESTART and zfsvfs->z_assign is TXG_NOWAIT,
	* then drop all locks, call dmu_tx_wait(), and try again. On subsequent
	* calls to dmu_tx_assign(), pass TXG_NOTHROTTLE in addition to TXG_NOWAIT,
	* to indicate that this operation has already called dmu_tx_wait().
	* This will ensure that we don't retry forever, waiting a short bit
	* each time.
	*
	* (5) If the operation succeeded, generate the intent log entry for it
	* before dropping locks. This ensures that the ordering of events
	* in the intent log matches the order in which they actually occurred.
	* During ZIL replay the zfs_log_* functions will update the sequence
	* number to indicate the zil transaction has replayed.
	*
	* (6) At the end of each vnode op, the DMU tx must always commit,
	* regardless of whether there were any errors.
	*
	* (7) After dropping all locks, invoke zil_commit(zilog, foid)
	* to ensure that synchronous semantics are provided when necessary.
	*
	* In general, this is how things should be ordered in each vnode op:
	*
	* ZFS_ENTER(zfsvfs); // exit if unmounted
	* top:
	* zfs_dirent_lookup(&dl, ...) // lock directory entry (may VN_HOLD())
	* rw_enter(...); // grab any other locks you need
	* tx = dmu_tx_create(...); // get DMU tx
	* dmu_tx_hold_*(); // hold each object you might modify
	* error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) \| TXG_NOWAIT);
	* if (error) {
	* rw_exit(...); // drop locks
	* zfs_dirent_unlock(dl); // unlock directory entry
	* VN_RELE(...); // release held vnodes
	* if (error == ERESTART) {
	* waited = B_TRUE;
	* dmu_tx_wait(tx);
	* dmu_tx_abort(tx);
	* goto top;
	* }
	* dmu_tx_abort(tx); // abort DMU tx
	* ZFS_EXIT(zfsvfs); // finished in zfs
	* return (error); // really out of space
	* }
	* error = do_real_work(); // do whatever this VOP does
	* if (error == 0)
	* zfs_log_*(...); // on success, make ZIL entry
	* dmu_tx_commit(tx); // commit DMU tx -- error or not
	* rw_exit(...); // drop locks
	* zfs_dirent_unlock(dl); // unlock directory entry
	* VN_RELE(...); // release held vnodes
	* zil_commit(zilog, foid); // synchronous when necessary
	* ZFS_EXIT(zfsvfs); // finished in zfs
	* return (error); // done, report error
	*/

	/* ARGSUSED */
	static int
	zfs_open(vnode_t *vpp, int flag, cred_t cr)
	{
	znode_t zp = VTOZ(vpp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	if ((flag & FWRITE) && (zp->z_pflags & ZFS_APPENDONLY) &&
	((flag & FAPPEND) == 0)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	/* Keep a count of the synchronous opens in the znode */
	if (flag & (FSYNC \| FDSYNC))
	atomic_inc_32(&zp->z_sync_cnt);

	ZFS_EXIT(zfsvfs);
	return (0);
	}

	/* ARGSUSED */
	static int
	zfs_close(vnode_t vp, int flag, int count, offset_t offset, cred_t cr)
	{
	znode_t *zp = VTOZ(vp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	/* Decrement the synchronous opens in the znode */
	if ((flag & (FSYNC \| FDSYNC)) && (count == 1))
	atomic_dec_32(&zp->z_sync_cnt);

	ZFS_EXIT(zfsvfs);
	return (0);
	}

	/* ARGSUSED */
	static int
	zfs_ioctl(vnode_t vp, ulong_t com, intptr_t data, int flag, cred_t cred,
	int *rvalp)
	{
	loff_t off;
	int error;

	switch (com) {
	case _FIOFFS:
	{
	return (0);

	/*
	* The following two ioctls are used by bfu. Faking out,
	* necessary to avoid bfu errors.
	*/
	}
	case _FIOGDIO:
	case _FIOSDIO:
	{
	return (0);
	}

	case F_SEEK_DATA:
	case F_SEEK_HOLE:
	{
	off = (offset_t )data;
	/* offset parameter is in/out */
	error = zfs_holey(VTOZ(vp), com, &off);
	if (error)
	return (error);
	(offset_t )data = off;
	return (0);
	}
	}
	return (SET_ERROR(ENOTTY));
	}

	static vm_page_t
	page_busy(vnode_t *vp, int64_t start, int64_t off, int64_t nbytes)
	{
	vm_object_t obj;
	vm_page_t pp;
	int64_t end;

	/*
	* At present vm_page_clear_dirty extends the cleared range to DEV_BSIZE
	* aligned boundaries, if the range is not aligned. As a result a
	* DEV_BSIZE subrange with partially dirty data may get marked as clean.
	* It may happen that all DEV_BSIZE subranges are marked clean and thus
	* the whole page would be considered clean despite have some
	* dirty data.
	* For this reason we should shrink the range to DEV_BSIZE aligned
	* boundaries before calling vm_page_clear_dirty.
	*/
	end = rounddown2(off + nbytes, DEV_BSIZE);
	off = roundup2(off, DEV_BSIZE);
	nbytes = end - off;

	obj = vp->v_object;
	zfs_vmobject_assert_wlocked_12(obj);
	#if __FreeBSD_version < 1300050
	for (;;) {
	if ((pp = vm_page_lookup(obj, OFF_TO_IDX(start))) != NULL &&
	pp->valid) {
	if (vm_page_xbusied(pp)) {
	/*
	* Reference the page before unlocking and
	* sleeping so that the page daemon is less
	* likely to reclaim it.
	*/
	vm_page_reference(pp);
	vm_page_lock(pp);
	zfs_vmobject_wunlock(obj);
	vm_page_busy_sleep(pp, "zfsmwb", true);
	zfs_vmobject_wlock(obj);
	continue;
	}
	vm_page_sbusy(pp);
	} else if (pp != NULL) {
	ASSERT(!pp->valid);
	pp = NULL;
	}
	if (pp != NULL) {
	ASSERT3U(pp->valid, ==, VM_PAGE_BITS_ALL);
	vm_object_pip_add(obj, 1);
	pmap_remove_write(pp);
	if (nbytes != 0)
	vm_page_clear_dirty(pp, off, nbytes);
	}
	break;
	}
	#else
	vm_page_grab_valid_unlocked(&pp, obj, OFF_TO_IDX(start),
	VM_ALLOC_NOCREAT \| VM_ALLOC_SBUSY \| VM_ALLOC_NORMAL \|
	VM_ALLOC_IGN_SBUSY);
	if (pp != NULL) {
	ASSERT3U(pp->valid, ==, VM_PAGE_BITS_ALL);
	vm_object_pip_add(obj, 1);
	pmap_remove_write(pp);
	if (nbytes != 0)
	vm_page_clear_dirty(pp, off, nbytes);
	}
	#endif
	return (pp);
	}

	static void
	page_unbusy(vm_page_t pp)
	{

	vm_page_sunbusy(pp);
	#if __FreeBSD_version >= 1300041
	vm_object_pip_wakeup(pp->object);
	#else
	vm_object_pip_subtract(pp->object, 1);
	#endif
	}

	#if __FreeBSD_version > 1300051
	static vm_page_t
	page_hold(vnode_t *vp, int64_t start)
	{
	vm_object_t obj;
	vm_page_t m;

	obj = vp->v_object;
	vm_page_grab_valid_unlocked(&m, obj, OFF_TO_IDX(start),
	VM_ALLOC_NOCREAT \| VM_ALLOC_WIRED \| VM_ALLOC_IGN_SBUSY \|
	VM_ALLOC_NOBUSY);
	return (m);
	}
	#else
	static vm_page_t
	page_hold(vnode_t *vp, int64_t start)
	{
	vm_object_t obj;
	vm_page_t pp;

	obj = vp->v_object;
	zfs_vmobject_assert_wlocked(obj);

	for (;;) {
	if ((pp = vm_page_lookup(obj, OFF_TO_IDX(start))) != NULL &&
	pp->valid) {
	if (vm_page_xbusied(pp)) {
	/*
	* Reference the page before unlocking and
	* sleeping so that the page daemon is less
	* likely to reclaim it.
	*/
	vm_page_reference(pp);
	vm_page_lock(pp);
	zfs_vmobject_wunlock(obj);
	vm_page_busy_sleep(pp, "zfsmwb", true);
	zfs_vmobject_wlock(obj);
	continue;
	}

	ASSERT3U(pp->valid, ==, VM_PAGE_BITS_ALL);
	vm_page_wire_lock(pp);
	vm_page_hold(pp);
	vm_page_wire_unlock(pp);

	} else
	pp = NULL;
	break;
	}
	return (pp);
	}
	#endif

	static void
	page_unhold(vm_page_t pp)
	{

	vm_page_wire_lock(pp);
	#if __FreeBSD_version >= 1300035
	vm_page_unwire(pp, PQ_ACTIVE);
	#else
	vm_page_unhold(pp);
	#endif
	vm_page_wire_unlock(pp);
	}

	/*
	* When a file is memory mapped, we must keep the IO data synchronized
	* between the DMU cache and the memory mapped pages. What this means:
	*
	* On Write: If we find a memory mapped page, we write to both
	* the page and the dmu buffer.
	*/
	void
	update_pages(znode_t zp, int64_t start, int len, objset_t os)
	{
	vm_object_t obj;
	struct sf_buf *sf;
	vnode_t *vp = ZTOV(zp);
	caddr_t va;
	int off;

	ASSERT3P(vp->v_mount, !=, NULL);
	obj = vp->v_object;
	ASSERT3P(obj, !=, NULL);

	off = start & PAGEOFFSET;
	zfs_vmobject_wlock_12(obj);
	#if __FreeBSD_version >= 1300041
	vm_object_pip_add(obj, 1);
	#endif
	for (start &= PAGEMASK; len > 0; start += PAGESIZE) {
	vm_page_t pp;
	int nbytes = imin(PAGESIZE - off, len);

	if ((pp = page_busy(vp, start, off, nbytes)) != NULL) {
	zfs_vmobject_wunlock_12(obj);

	va = zfs_map_page(pp, &sf);
	(void) dmu_read(os, zp->z_id, start + off, nbytes,
	va + off, DMU_READ_PREFETCH);
	zfs_unmap_page(sf);

	zfs_vmobject_wlock_12(obj);
	page_unbusy(pp);
	}
	len -= nbytes;
	off = 0;
	}
	#if __FreeBSD_version >= 1300041
	vm_object_pip_wakeup(obj);
	#else
	vm_object_pip_wakeupn(obj, 0);
	#endif
	zfs_vmobject_wunlock_12(obj);
	}

	/*
	* Read with UIO_NOCOPY flag means that sendfile(2) requests
	* ZFS to populate a range of page cache pages with data.
	*
	* NOTE: this function could be optimized to pre-allocate
	* all pages in advance, drain exclusive busy on all of them,
	* map them into contiguous KVA region and populate them
	* in one single dmu_read() call.
	*/
	int
	mappedread_sf(znode_t zp, int nbytes, zfs_uio_t uio)
	{
	vnode_t *vp = ZTOV(zp);
	objset_t *os = zp->z_zfsvfs->z_os;
	struct sf_buf *sf;
	vm_object_t obj;
	vm_page_t pp;
	int64_t start;
	caddr_t va;
	int len = nbytes;
	int error = 0;

	ASSERT3U(zfs_uio_segflg(uio), ==, UIO_NOCOPY);
	ASSERT3P(vp->v_mount, !=, NULL);
	obj = vp->v_object;
	ASSERT3P(obj, !=, NULL);
	ASSERT0(zfs_uio_offset(uio) & PAGEOFFSET);

	zfs_vmobject_wlock_12(obj);
	for (start = zfs_uio_offset(uio); len > 0; start += PAGESIZE) {
	int bytes = MIN(PAGESIZE, len);

	pp = vm_page_grab_unlocked(obj, OFF_TO_IDX(start),
	VM_ALLOC_SBUSY \| VM_ALLOC_NORMAL \| VM_ALLOC_IGN_SBUSY);
	if (vm_page_none_valid(pp)) {
	zfs_vmobject_wunlock_12(obj);
	va = zfs_map_page(pp, &sf);
	error = dmu_read(os, zp->z_id, start, bytes, va,
	DMU_READ_PREFETCH);
	if (bytes != PAGESIZE && error == 0)
	bzero(va + bytes, PAGESIZE - bytes);
	zfs_unmap_page(sf);
	zfs_vmobject_wlock_12(obj);
	#if __FreeBSD_version >= 1300081
	if (error == 0) {
	vm_page_valid(pp);
	vm_page_activate(pp);
	vm_page_do_sunbusy(pp);
	} else {
	zfs_vmobject_wlock(obj);
	if (!vm_page_wired(pp) && pp->valid == 0 &&
	vm_page_busy_tryupgrade(pp))
	vm_page_free(pp);
	else
	vm_page_sunbusy(pp);
	zfs_vmobject_wunlock(obj);
	}
	#else
	vm_page_do_sunbusy(pp);
	vm_page_lock(pp);
	if (error) {
	if (pp->wire_count == 0 && pp->valid == 0 &&
	!vm_page_busied(pp))
	vm_page_free(pp);
	} else {
	pp->valid = VM_PAGE_BITS_ALL;
	vm_page_activate(pp);
	}
	vm_page_unlock(pp);
	#endif
	} else {
	ASSERT3U(pp->valid, ==, VM_PAGE_BITS_ALL);
	vm_page_do_sunbusy(pp);
	}
	if (error)
	break;
	zfs_uio_advance(uio, bytes);
	len -= bytes;
	}
	zfs_vmobject_wunlock_12(obj);
	return (error);
	}

	/*
	* When a file is memory mapped, we must keep the IO data synchronized
	* between the DMU cache and the memory mapped pages. What this means:
	*
	* On Read: We "read" preferentially from memory mapped pages,
	* else we default from the dmu buffer.
	*
	* NOTE: We will always "break up" the IO into PAGESIZE uiomoves when
	* the file is memory mapped.
	*/
	int
	mappedread(znode_t zp, int nbytes, zfs_uio_t uio)
	{
	vnode_t *vp = ZTOV(zp);
	vm_object_t obj;
	int64_t start;
	int len = nbytes;
	int off;
	int error = 0;

	ASSERT3P(vp->v_mount, !=, NULL);
	obj = vp->v_object;
	ASSERT3P(obj, !=, NULL);

	start = zfs_uio_offset(uio);
	off = start & PAGEOFFSET;
	zfs_vmobject_wlock_12(obj);
	for (start &= PAGEMASK; len > 0; start += PAGESIZE) {
	vm_page_t pp;
	uint64_t bytes = MIN(PAGESIZE - off, len);

	if ((pp = page_hold(vp, start))) {
	struct sf_buf *sf;
	caddr_t va;

	zfs_vmobject_wunlock_12(obj);
	va = zfs_map_page(pp, &sf);
	error = vn_io_fault_uiomove(va + off, bytes,
	GET_UIO_STRUCT(uio));
	zfs_unmap_page(sf);
	zfs_vmobject_wlock_12(obj);
	page_unhold(pp);
	} else {
	zfs_vmobject_wunlock_12(obj);
	error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl),
	uio, bytes);
	zfs_vmobject_wlock_12(obj);
	}
	len -= bytes;
	off = 0;
	if (error)
	break;
	}
	zfs_vmobject_wunlock_12(obj);
	return (error);
	}

	int
	zfs_write_simple(znode_t zp, const void data, size_t len,
	loff_t pos, size_t *presid)
	{
	int error = 0;
	ssize_t resid;

	error = vn_rdwr(UIO_WRITE, ZTOV(zp), __DECONST(void *, data), len, pos,
	UIO_SYSSPACE, IO_SYNC, kcred, NOCRED, &resid, curthread);

	if (error) {
	return (SET_ERROR(error));
	} else if (presid == NULL) {
	if (resid != 0) {
	error = SET_ERROR(EIO);
	}
	} else {
	*presid = resid;
	}
	return (error);
	}

	void
	zfs_zrele_async(znode_t *zp)
	{
	vnode_t *vp = ZTOV(zp);
	objset_t *os = ITOZSB(vp)->z_os;

	VN_RELE_ASYNC(vp, dsl_pool_zrele_taskq(dmu_objset_pool(os)));
	}

	static int
	zfs_dd_callback(struct mount mp, void arg, int lkflags, struct vnode **vpp)
	{
	int error;

	*vpp = arg;
	error = vn_lock(*vpp, lkflags);
	if (error != 0)
	vrele(*vpp);
	return (error);
	}

	static int
	zfs_lookup_lock(vnode_t dvp, vnode_t vp, const char *name, int lkflags)
	{
	znode_t *zdp = VTOZ(dvp);
	zfsvfs_t *zfsvfs __unused = zdp->z_zfsvfs;
	int error;
	int ltype;

	if (zfsvfs->z_replay == B_FALSE)
	ASSERT_VOP_LOCKED(dvp, __func__);

	if (name[0] == 0 \|\| (name[0] == '.' && name[1] == 0)) {
	ASSERT3P(dvp, ==, vp);
	vref(dvp);
	ltype = lkflags & LK_TYPE_MASK;
	if (ltype != VOP_ISLOCKED(dvp)) {
	if (ltype == LK_EXCLUSIVE)
	vn_lock(dvp, LK_UPGRADE \| LK_RETRY);
	else /* if (ltype == LK_SHARED) */
	vn_lock(dvp, LK_DOWNGRADE \| LK_RETRY);

	/*
	* Relock for the "." case could leave us with
	* reclaimed vnode.
	*/
	if (VN_IS_DOOMED(dvp)) {
	vrele(dvp);
	return (SET_ERROR(ENOENT));
	}
	}
	return (0);
	} else if (name[0] == '.' && name[1] == '.' && name[2] == 0) {
	/*
	* Note that in this case, dvp is the child vnode, and we
	* are looking up the parent vnode - exactly reverse from
	* normal operation. Unlocking dvp requires some rather
	* tricky unlock/relock dance to prevent mp from being freed;
	* use vn_vget_ino_gen() which takes care of all that.
	*
	* XXX Note that there is a time window when both vnodes are
	* unlocked. It is possible, although highly unlikely, that
	* during that window the parent-child relationship between
	* the vnodes may change, for example, get reversed.
	* In that case we would have a wrong lock order for the vnodes.
	* All other filesystems seem to ignore this problem, so we
	* do the same here.
	* A potential solution could be implemented as follows:
	* - using LK_NOWAIT when locking the second vnode and retrying
	* if necessary
	* - checking that the parent-child relationship still holds
	* after locking both vnodes and retrying if it doesn't
	*/
	error = vn_vget_ino_gen(dvp, zfs_dd_callback, vp, lkflags, &vp);
	return (error);
	} else {
	error = vn_lock(vp, lkflags);
	if (error != 0)
	vrele(vp);
	return (error);
	}
	}

	/*
	* Lookup an entry in a directory, or an extended attribute directory.
	* If it exists, return a held vnode reference for it.
	*
	* IN: dvp - vnode of directory to search.
	* nm - name of entry to lookup.
	* pnp - full pathname to lookup [UNUSED].
	* flags - LOOKUP_XATTR set if looking for an attribute.
	* rdir - root directory vnode [UNUSED].
	* cr - credentials of caller.
	* ct - caller context
	*
	* OUT: vpp - vnode of located entry, NULL if not found.
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* NA
	*/
	/* ARGSUSED */
	static int
	zfs_lookup(vnode_t dvp, const char nm, vnode_t **vpp,
	struct componentname cnp, int nameiop, cred_t cr, int flags,
	boolean_t cached)
	{
	znode_t *zdp = VTOZ(dvp);
	znode_t *zp;
	zfsvfs_t *zfsvfs = zdp->z_zfsvfs;
	#if __FreeBSD_version > 1300124
	seqc_t dvp_seqc;
	#endif
	int error = 0;

	/*
	* Fast path lookup, however we must skip DNLC lookup
	* for case folding or normalizing lookups because the
	* DNLC code only stores the passed in name. This means
	* creating 'a' and removing 'A' on a case insensitive
	* file system would work, but DNLC still thinks 'a'
	* exists and won't let you create it again on the next
	* pass through fast path.
	*/
	if (!(flags & LOOKUP_XATTR)) {
	if (dvp->v_type != VDIR) {
	return (SET_ERROR(ENOTDIR));
	} else if (zdp->z_sa_hdl == NULL) {
	return (SET_ERROR(EIO));
	}
	}

	DTRACE_PROBE2(zfs__fastpath__lookup__miss, vnode_t *, dvp,
	const char *, nm);

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zdp);

	#if __FreeBSD_version > 1300124
	dvp_seqc = vn_seqc_read_notmodify(dvp);
	#endif

	*vpp = NULL;

	if (flags & LOOKUP_XATTR) {
	/*
	* If the xattr property is off, refuse the lookup request.
	*/
	if (!(zfsvfs->z_flags & ZSB_XATTR)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EOPNOTSUPP));
	}

	/*
	* We don't allow recursive attributes..
	* Maybe someday we will.
	*/
	if (zdp->z_pflags & ZFS_XATTR) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	if ((error = zfs_get_xattrdir(VTOZ(dvp), &zp, cr, flags))) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	*vpp = ZTOV(zp);

	/*
	* Do we have permission to get into attribute directory?
	*/
	error = zfs_zaccess(zp, ACE_EXECUTE, 0, B_FALSE, cr);
	if (error) {
	vrele(ZTOV(zp));
	}

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Check accessibility of directory if we're not coming in via
	* VOP_CACHEDLOOKUP.
	*/
	if (!cached) {
	#ifdef NOEXECCHECK
	if ((cnp->cn_flags & NOEXECCHECK) != 0) {
	cnp->cn_flags &= ~NOEXECCHECK;
	} else
	#endif
	if ((error = zfs_zaccess(zdp, ACE_EXECUTE, 0, B_FALSE, cr))) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	}

	if (zfsvfs->z_utf8 && u8_validate(nm, strlen(nm),
	NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EILSEQ));
	}


	/*
	* First handle the special cases.
	*/
	if ((cnp->cn_flags & ISDOTDOT) != 0) {
	/*
	* If we are a snapshot mounted under .zfs, return
	* the vp for the snapshot directory.
	*/
	if (zdp->z_id == zfsvfs->z_root && zfsvfs->z_parent != zfsvfs) {
	struct componentname cn;
	vnode_t *zfsctl_vp;
	int ltype;

	ZFS_EXIT(zfsvfs);
	ltype = VOP_ISLOCKED(dvp);
	VOP_UNLOCK1(dvp);
	error = zfsctl_root(zfsvfs->z_parent, LK_SHARED,
	&zfsctl_vp);
	if (error == 0) {
	cn.cn_nameptr = "snapshot";
	cn.cn_namelen = strlen(cn.cn_nameptr);
	cn.cn_nameiop = cnp->cn_nameiop;
	cn.cn_flags = cnp->cn_flags & ~ISDOTDOT;
	cn.cn_lkflags = cnp->cn_lkflags;
	error = VOP_LOOKUP(zfsctl_vp, vpp, &cn);
	vput(zfsctl_vp);
	}
	vn_lock(dvp, ltype \| LK_RETRY);
	return (error);
	}
	}
	if (zfs_has_ctldir(zdp) && strcmp(nm, ZFS_CTLDIR_NAME) == 0) {
	ZFS_EXIT(zfsvfs);
	if ((cnp->cn_flags & ISLASTCN) != 0 && nameiop != LOOKUP)
	return (SET_ERROR(ENOTSUP));
	error = zfsctl_root(zfsvfs, cnp->cn_lkflags, vpp);
	return (error);
	}

	/*
	* The loop is retry the lookup if the parent-child relationship
	* changes during the dot-dot locking complexities.
	*/
	for (;;) {
	uint64_t parent;

	error = zfs_dirlook(zdp, nm, &zp);
	if (error == 0)
	*vpp = ZTOV(zp);

	ZFS_EXIT(zfsvfs);
	if (error != 0)
	break;

	error = zfs_lookup_lock(dvp, *vpp, nm, cnp->cn_lkflags);
	if (error != 0) {
	/*
	* If we've got a locking error, then the vnode
	* got reclaimed because of a force unmount.
	* We never enter doomed vnodes into the name cache.
	*/
	*vpp = NULL;
	return (error);
	}

	if ((cnp->cn_flags & ISDOTDOT) == 0)
	break;

	ZFS_ENTER(zfsvfs);
	if (zdp->z_sa_hdl == NULL) {
	error = SET_ERROR(EIO);
	} else {
	error = sa_lookup(zdp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
	&parent, sizeof (parent));
	}
	if (error != 0) {
	ZFS_EXIT(zfsvfs);
	vput(ZTOV(zp));
	break;
	}
	if (zp->z_id == parent) {
	ZFS_EXIT(zfsvfs);
	break;
	}
	vput(ZTOV(zp));
	}

	if (error != 0)
	*vpp = NULL;

	/* Translate errors and add SAVENAME when needed. */
	if (cnp->cn_flags & ISLASTCN) {
	switch (nameiop) {
	case CREATE:
	case RENAME:
	if (error == ENOENT) {
	error = EJUSTRETURN;
	cnp->cn_flags \|= SAVENAME;
	break;
	}
	fallthrough;
	case DELETE:
	if (error == 0)
	cnp->cn_flags \|= SAVENAME;
	break;
	}
	}

	#if __FreeBSD_version > 1300124
	if ((cnp->cn_flags & ISDOTDOT) != 0) {
	/*
	* FIXME: zfs_lookup_lock relocks vnodes and does nothing to
	* handle races. In particular different callers may end up
	* with different vnodes and will try to add conflicting
	* entries to the namecache.
	*
	* While finding different result may be acceptable in face
	* of concurrent modification, adding conflicting entries
	* trips over an assert in the namecache.
	*
	* Ultimately let an entry through once everything settles.
	*/
	if (!vn_seqc_consistent(dvp, dvp_seqc)) {
	cnp->cn_flags &= ~MAKEENTRY;
	}
	}
	#endif

	/* Insert name into cache (as non-existent) if appropriate. */
	if (zfsvfs->z_use_namecache && !zfsvfs->z_replay &&
	error == ENOENT && (cnp->cn_flags & MAKEENTRY) != 0)
	cache_enter(dvp, NULL, cnp);

	/* Insert name into cache if appropriate. */
	if (zfsvfs->z_use_namecache && !zfsvfs->z_replay &&
	error == 0 && (cnp->cn_flags & MAKEENTRY)) {
	if (!(cnp->cn_flags & ISLASTCN) \|\|
	(nameiop != DELETE && nameiop != RENAME)) {
	cache_enter(dvp, *vpp, cnp);
	}
	}

	return (error);
	}

	/*
	* Attempt to create a new entry in a directory. If the entry
	* already exists, truncate the file if permissible, else return
	* an error. Return the vp of the created or trunc'd file.
	*
	* IN: dvp - vnode of directory to put new file entry in.
	* name - name of new file entry.
	* vap - attributes of new file.
	* excl - flag indicating exclusive or non-exclusive mode.
	* mode - mode to open file with.
	* cr - credentials of caller.
	* flag - large file flag [UNUSED].
	* ct - caller context
	* vsecp - ACL to be set
	*
	* OUT: vpp - vnode of created or trunc'd entry.
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* dvp - ctime\|mtime updated if new entry created
	* vp - ctime\|mtime always, atime if new
	*/

	/* ARGSUSED */
	int
	zfs_create(znode_t dzp, const char name, vattr_t *vap, int excl, int mode,
	znode_t *zpp, cred_t cr, int flag, vsecattr_t *vsecp)
	{
	znode_t *zp;
	zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
	zilog_t *zilog;
	objset_t *os;
	dmu_tx_t *tx;
	int error;
	ksid_t *ksid;
	uid_t uid;
	gid_t gid = crgetgid(cr);
	uint64_t projid = ZFS_DEFAULT_PROJID;
	zfs_acl_ids_t acl_ids;
	boolean_t fuid_dirtied;
	uint64_t txtype;
	#ifdef DEBUG_VFS_LOCKS
	vnode_t *dvp = ZTOV(dzp);
	#endif

	/*
	* If we have an ephemeral id, ACL, or XVATTR then
	* make sure file system is at proper version
	*/

	ksid = crgetsid(cr, KSID_OWNER);
	if (ksid)
	uid = ksid_getid(ksid);
	else
	uid = crgetuid(cr);

	if (zfsvfs->z_use_fuids == B_FALSE &&
	(vsecp \|\| (vap->va_mask & AT_XVATTR) \|\|
	IS_EPHEMERAL(uid) \|\| IS_EPHEMERAL(gid)))
	return (SET_ERROR(EINVAL));

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(dzp);
	os = zfsvfs->z_os;
	zilog = zfsvfs->z_log;

	if (zfsvfs->z_utf8 && u8_validate(name, strlen(name),
	NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EILSEQ));
	}

	if (vap->va_mask & AT_XVATTR) {
	if ((error = secpolicy_xvattr(ZTOV(dzp), (xvattr_t *)vap,
	crgetuid(cr), cr, vap->va_type)) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	}

	*zpp = NULL;

	if ((vap->va_mode & S_ISVTX) && secpolicy_vnode_stky_modify(cr))
	vap->va_mode &= ~S_ISVTX;

	error = zfs_dirent_lookup(dzp, name, &zp, ZNEW);
	if (error) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	ASSERT3P(zp, ==, NULL);

	/*
	* Create a new file object and update the directory
	* to reference it.
	*/
	if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
	goto out;
	}

	/*
	* We only support the creation of regular files in
	* extended attribute directories.
	*/

	if ((dzp->z_pflags & ZFS_XATTR) &&
	(vap->va_type != VREG)) {
	error = SET_ERROR(EINVAL);
	goto out;
	}

	if ((error = zfs_acl_ids_create(dzp, 0, vap,
	cr, vsecp, &acl_ids)) != 0)
	goto out;

	if (S_ISREG(vap->va_mode) \|\| S_ISDIR(vap->va_mode))
	projid = zfs_inherit_projid(dzp);
	if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, projid)) {
	zfs_acl_ids_free(&acl_ids);
	error = SET_ERROR(EDQUOT);
	goto out;
	}

	getnewvnode_reserve_();

	tx = dmu_tx_create(os);

	dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
	ZFS_SA_BASE_ATTR_SIZE);

	fuid_dirtied = zfsvfs->z_fuid_dirty;
	if (fuid_dirtied)
	zfs_fuid_txhold(zfsvfs, tx);
	dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
	dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
	if (!zfsvfs->z_use_sa &&
	acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
	0, acl_ids.z_aclp->z_acl_bytes);
	}
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	zfs_acl_ids_free(&acl_ids);
	dmu_tx_abort(tx);
	getnewvnode_drop_reserve();
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
	if (fuid_dirtied)
	zfs_fuid_sync(zfsvfs, tx);

	(void) zfs_link_create(dzp, name, zp, tx, ZNEW);
	txtype = zfs_log_create_txtype(Z_FILE, vsecp, vap);
	zfs_log_create(zilog, tx, txtype, dzp, zp, name,
	vsecp, acl_ids.z_fuidp, vap);
	zfs_acl_ids_free(&acl_ids);
	dmu_tx_commit(tx);

	getnewvnode_drop_reserve();

	out:
	VNCHECKREF(dvp);
	if (error == 0) {
	*zpp = zp;
	}

	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Remove an entry from a directory.
	*
	* IN: dvp - vnode of directory to remove entry from.
	* name - name of entry to remove.
	* cr - credentials of caller.
	* ct - caller context
	* flags - case flags
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* dvp - ctime\|mtime
	* vp - ctime (if nlink > 0)
	*/

	/ARGSUSED/
	static int
	zfs_remove_(vnode_t dvp, vnode_t vp, const char name, cred_t cr)
	{
	znode_t *dzp = VTOZ(dvp);
	znode_t *zp;
	znode_t *xzp;
	zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
	zilog_t *zilog;
	uint64_t xattr_obj;
	uint64_t obj = 0;
	dmu_tx_t *tx;
	boolean_t unlinked;
	uint64_t txtype;
	int error;


	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(dzp);
	zp = VTOZ(vp);
	ZFS_VERIFY_ZP(zp);
	zilog = zfsvfs->z_log;

	xattr_obj = 0;
	xzp = NULL;

	if ((error = zfs_zaccess_delete(dzp, zp, cr))) {
	goto out;
	}

	/*
	* Need to use rmdir for removing directories.
	*/
	if (vp->v_type == VDIR) {
	error = SET_ERROR(EPERM);
	goto out;
	}

	vnevent_remove(vp, dvp, name, ct);

	obj = zp->z_id;

	/* are there any extended attributes? */
	error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
	&xattr_obj, sizeof (xattr_obj));
	if (error == 0 && xattr_obj) {
	error = zfs_zget(zfsvfs, xattr_obj, &xzp);
	ASSERT0(error);
	}

	/*
	* We may delete the znode now, or we may put it in the unlinked set;
	* it depends on whether we're the last link, and on whether there are
	* other holds on the vnode. So we dmu_tx_hold() the right things to
	* allow for either case.
	*/
	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name);
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
	zfs_sa_upgrade_txholds(tx, zp);
	zfs_sa_upgrade_txholds(tx, dzp);

	if (xzp) {
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
	dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
	}

	/* charge as an update -- would be nice not to charge at all */
	dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);

	/*
	* Mark this transaction as typically resulting in a net free of space
	*/
	dmu_tx_mark_netfree(tx);

	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	dmu_tx_abort(tx);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Remove the directory entry.
	*/
	error = zfs_link_destroy(dzp, name, zp, tx, ZEXISTS, &unlinked);

	if (error) {
	dmu_tx_commit(tx);
	goto out;
	}

	if (unlinked) {
	zfs_unlinked_add(zp, tx);
	vp->v_vflag \|= VV_NOSYNC;
	}
	/* XXX check changes to linux vnops */
	txtype = TX_REMOVE;
	zfs_log_remove(zilog, tx, txtype, dzp, name, obj, unlinked);

	dmu_tx_commit(tx);
	out:

	if (xzp)
	vrele(ZTOV(xzp));

	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);


	ZFS_EXIT(zfsvfs);
	return (error);
	}


	static int
	zfs_lookup_internal(znode_t dzp, const char name, vnode_t **vpp,
	struct componentname *cnp, int nameiop)
	{
	zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
	int error;

	cnp->cn_nameptr = __DECONST(char *, name);
	cnp->cn_namelen = strlen(name);
	cnp->cn_nameiop = nameiop;
	cnp->cn_flags = ISLASTCN \| SAVENAME;
	cnp->cn_lkflags = LK_EXCLUSIVE \| LK_RETRY;
	cnp->cn_cred = kcred;
	#if __FreeBSD_version < 1400037
	cnp->cn_thread = curthread;
	#endif

	if (zfsvfs->z_use_namecache && !zfsvfs->z_replay) {
	struct vop_lookup_args a;

	a.a_gen.a_desc = &vop_lookup_desc;
	a.a_dvp = ZTOV(dzp);
	a.a_vpp = vpp;
	a.a_cnp = cnp;
	error = vfs_cache_lookup(&a);
	} else {
	error = zfs_lookup(ZTOV(dzp), name, vpp, cnp, nameiop, kcred, 0,
	B_FALSE);
	}
	#ifdef ZFS_DEBUG
	if (error) {
	printf("got error %d on name %s on op %d\n", error, name,
	nameiop);
	kdb_backtrace();
	}
	#endif
	return (error);
	}

	int
	zfs_remove(znode_t dzp, const char name, cred_t *cr, int flags)
	{
	vnode_t *vp;
	int error;
	struct componentname cn;

	if ((error = zfs_lookup_internal(dzp, name, &vp, &cn, DELETE)))
	return (error);

	error = zfs_remove_(ZTOV(dzp), vp, name, cr);
	vput(vp);
	return (error);
	}
	/*
	* Create a new directory and insert it into dvp using the name
	* provided. Return a pointer to the inserted directory.
	*
	* IN: dvp - vnode of directory to add subdir to.
	* dirname - name of new directory.
	* vap - attributes of new directory.
	* cr - credentials of caller.
	* ct - caller context
	* flags - case flags
	* vsecp - ACL to be set
	*
	* OUT: vpp - vnode of created directory.
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* dvp - ctime\|mtime updated
	* vp - ctime\|mtime\|atime updated
	*/
	/ARGSUSED/
	int
	zfs_mkdir(znode_t dzp, const char dirname, vattr_t vap, znode_t *zpp,
	cred_t cr, int flags, vsecattr_t vsecp)
	{
	znode_t *zp;
	zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
	zilog_t *zilog;
	uint64_t txtype;
	dmu_tx_t *tx;
	int error;
	ksid_t *ksid;
	uid_t uid;
	gid_t gid = crgetgid(cr);
	zfs_acl_ids_t acl_ids;
	boolean_t fuid_dirtied;

	ASSERT3U(vap->va_type, ==, VDIR);

	/*
	* If we have an ephemeral id, ACL, or XVATTR then
	* make sure file system is at proper version
	*/

	ksid = crgetsid(cr, KSID_OWNER);
	if (ksid)
	uid = ksid_getid(ksid);
	else
	uid = crgetuid(cr);
	if (zfsvfs->z_use_fuids == B_FALSE &&
	((vap->va_mask & AT_XVATTR) \|\|
	IS_EPHEMERAL(uid) \|\| IS_EPHEMERAL(gid)))
	return (SET_ERROR(EINVAL));

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(dzp);
	zilog = zfsvfs->z_log;

	if (dzp->z_pflags & ZFS_XATTR) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	if (zfsvfs->z_utf8 && u8_validate(dirname,
	strlen(dirname), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EILSEQ));
	}

	if (vap->va_mask & AT_XVATTR) {
	if ((error = secpolicy_xvattr(ZTOV(dzp), (xvattr_t *)vap,
	crgetuid(cr), cr, vap->va_type)) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	}

	if ((error = zfs_acl_ids_create(dzp, 0, vap, cr,
	NULL, &acl_ids)) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* First make sure the new directory doesn't exist.
	*
	* Existence is checked first to make sure we don't return
	* EACCES instead of EEXIST which can cause some applications
	* to fail.
	*/
	*zpp = NULL;

	if ((error = zfs_dirent_lookup(dzp, dirname, &zp, ZNEW))) {
	zfs_acl_ids_free(&acl_ids);
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	ASSERT3P(zp, ==, NULL);

	if ((error = zfs_zaccess(dzp, ACE_ADD_SUBDIRECTORY, 0, B_FALSE, cr))) {
	zfs_acl_ids_free(&acl_ids);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, zfs_inherit_projid(dzp))) {
	zfs_acl_ids_free(&acl_ids);
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EDQUOT));
	}

	/*
	* Add a new entry to the directory.
	*/
	getnewvnode_reserve_();
	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_zap(tx, dzp->z_id, TRUE, dirname);
	dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
	fuid_dirtied = zfsvfs->z_fuid_dirty;
	if (fuid_dirtied)
	zfs_fuid_txhold(zfsvfs, tx);
	if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
	acl_ids.z_aclp->z_acl_bytes);
	}

	dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
	ZFS_SA_BASE_ATTR_SIZE);

	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	zfs_acl_ids_free(&acl_ids);
	dmu_tx_abort(tx);
	getnewvnode_drop_reserve();
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Create new node.
	*/
	zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);

	if (fuid_dirtied)
	zfs_fuid_sync(zfsvfs, tx);

	/*
	* Now put new name in parent dir.
	*/
	(void) zfs_link_create(dzp, dirname, zp, tx, ZNEW);

	*zpp = zp;

	txtype = zfs_log_create_txtype(Z_DIR, NULL, vap);
	zfs_log_create(zilog, tx, txtype, dzp, zp, dirname, NULL,
	acl_ids.z_fuidp, vap);

	zfs_acl_ids_free(&acl_ids);

	dmu_tx_commit(tx);

	getnewvnode_drop_reserve();

	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	ZFS_EXIT(zfsvfs);
	return (0);
	}

	#if __FreeBSD_version < 1300124
	static void
	cache_vop_rmdir(struct vnode dvp, struct vnode vp)
	{

	cache_purge(dvp);
	cache_purge(vp);
	}
	#endif

	/*
	* Remove a directory subdir entry. If the current working
	* directory is the same as the subdir to be removed, the
	* remove will fail.
	*
	* IN: dvp - vnode of directory to remove from.
	* name - name of directory to be removed.
	* cwd - vnode of current working directory.
	* cr - credentials of caller.
	* ct - caller context
	* flags - case flags
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* dvp - ctime\|mtime updated
	*/
	/ARGSUSED/
	static int
	zfs_rmdir_(vnode_t dvp, vnode_t vp, const char name, cred_t cr)
	{
	znode_t *dzp = VTOZ(dvp);
	znode_t *zp = VTOZ(vp);
	zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
	zilog_t *zilog;
	dmu_tx_t *tx;
	int error;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(dzp);
	ZFS_VERIFY_ZP(zp);
	zilog = zfsvfs->z_log;


	if ((error = zfs_zaccess_delete(dzp, zp, cr))) {
	goto out;
	}

	if (vp->v_type != VDIR) {
	error = SET_ERROR(ENOTDIR);
	goto out;
	}

	vnevent_rmdir(vp, dvp, name, ct);

	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name);
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
	dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
	zfs_sa_upgrade_txholds(tx, zp);
	zfs_sa_upgrade_txholds(tx, dzp);
	dmu_tx_mark_netfree(tx);
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	dmu_tx_abort(tx);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	error = zfs_link_destroy(dzp, name, zp, tx, ZEXISTS, NULL);

	if (error == 0) {
	uint64_t txtype = TX_RMDIR;
	zfs_log_remove(zilog, tx, txtype, dzp, name,
	ZFS_NO_OBJECT, B_FALSE);
	}

	dmu_tx_commit(tx);

	cache_vop_rmdir(dvp, vp);
	out:
	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	int
	zfs_rmdir(znode_t dzp, const char name, znode_t cwd, cred_t cr, int flags)
	{
	struct componentname cn;
	vnode_t *vp;
	int error;

	if ((error = zfs_lookup_internal(dzp, name, &vp, &cn, DELETE)))
	return (error);

	error = zfs_rmdir_(ZTOV(dzp), vp, name, cr);
	vput(vp);
	return (error);
	}

	/*
	* Read as many directory entries as will fit into the provided
	* buffer from the given directory cursor position (specified in
	* the uio structure).
	*
	* IN: vp - vnode of directory to read.
	* uio - structure supplying read location, range info,
	* and return buffer.
	* cr - credentials of caller.
	* ct - caller context
	* flags - case flags
	*
	* OUT: uio - updated offset and range, buffer filled.
	* eofp - set to true if end-of-file detected.
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* vp - atime updated
	*
	* Note that the low 4 bits of the cookie returned by zap is always zero.
	* This allows us to use the low range for "special" directory entries:
	* We use 0 for '.', and 1 for '..'. If this is the root of the filesystem,
	* we use the offset 2 for the '.zfs' directory.
	*/
	/* ARGSUSED */
	static int
	zfs_readdir(vnode_t vp, zfs_uio_t uio, cred_t cr, int eofp,
	int ncookies, uint64_t *cookies)
	{
	znode_t *zp = VTOZ(vp);
	iovec_t *iovp;
	edirent_t *eodp;
	dirent64_t *odp;
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
	objset_t *os;
	caddr_t outbuf;
	size_t bufsize;
	zap_cursor_t zc;
	zap_attribute_t zap;
	uint_t bytes_wanted;
	uint64_t offset; /* must be unsigned; checks for < 1 */
	uint64_t parent;
	int local_eof;
	int outcount;
	int error;
	uint8_t prefetch;
	boolean_t check_sysattrs;
	uint8_t type;
	int ncooks;
	uint64_t *cooks = NULL;
	int flags = 0;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
	&parent, sizeof (parent))) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* If we are not given an eof variable,
	* use a local one.
	*/
	if (eofp == NULL)
	eofp = &local_eof;

	/*
	* Check for valid iov_len.
	*/
	if (GET_UIO_STRUCT(uio)->uio_iov->iov_len <= 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	/*
	* Quit if directory has been removed (posix)
	*/
	if ((*eofp = zp->z_unlinked) != 0) {
	ZFS_EXIT(zfsvfs);
	return (0);
	}

	error = 0;
	os = zfsvfs->z_os;
	offset = zfs_uio_offset(uio);
	prefetch = zp->z_zn_prefetch;

	/*
	* Initialize the iterator cursor.
	*/
	if (offset <= 3) {
	/*
	* Start iteration from the beginning of the directory.
	*/
	zap_cursor_init(&zc, os, zp->z_id);
	} else {
	/*
	* The offset is a serialized cursor.
	*/
	zap_cursor_init_serialized(&zc, os, zp->z_id, offset);
	}

	/*
	* Get space to change directory entries into fs independent format.
	*/
	iovp = GET_UIO_STRUCT(uio)->uio_iov;
	bytes_wanted = iovp->iov_len;
	if (zfs_uio_segflg(uio) != UIO_SYSSPACE \|\| zfs_uio_iovcnt(uio) != 1) {
	bufsize = bytes_wanted;
	outbuf = kmem_alloc(bufsize, KM_SLEEP);
	odp = (struct dirent64 *)outbuf;
	} else {
	bufsize = bytes_wanted;
	outbuf = NULL;
	odp = (struct dirent64 *)iovp->iov_base;
	}
	eodp = (struct edirent *)odp;

	if (ncookies != NULL) {
	/*
	* Minimum entry size is dirent size and 1 byte for a file name.
	*/
	ncooks = zfs_uio_resid(uio) / (sizeof (struct dirent) -
	sizeof (((struct dirent *)NULL)->d_name) + 1);
	cooks = malloc(ncooks * sizeof (*cooks), M_TEMP, M_WAITOK);
	*cookies = cooks;
	*ncookies = ncooks;
	}
	/*
	* If this VFS supports the system attribute view interface; and
	* we're looking at an extended attribute directory; and we care
	* about normalization conflicts on this vfs; then we must check
	* for normalization conflicts with the sysattr name space.
	*/
	#ifdef TODO
	check_sysattrs = vfs_has_feature(vp->v_vfsp, VFSFT_SYSATTR_VIEWS) &&
	(vp->v_flag & V_XATTRDIR) && zfsvfs->z_norm &&
	(flags & V_RDDIR_ENTFLAGS);
	#else
	check_sysattrs = 0;
	#endif

	/*
	* Transform to file-system independent format
	*/
	outcount = 0;
	while (outcount < bytes_wanted) {
	ino64_t objnum;
	ushort_t reclen;
	off64_t *next = NULL;

	/*
	* Special case `.', `..', and `.zfs'.
	*/
	if (offset == 0) {
	(void) strcpy(zap.za_name, ".");
	zap.za_normalization_conflict = 0;
	objnum = zp->z_id;
	type = DT_DIR;
	} else if (offset == 1) {
	(void) strcpy(zap.za_name, "..");
	zap.za_normalization_conflict = 0;
	objnum = parent;
	type = DT_DIR;
	} else if (offset == 2 && zfs_show_ctldir(zp)) {
	(void) strcpy(zap.za_name, ZFS_CTLDIR_NAME);
	zap.za_normalization_conflict = 0;
	objnum = ZFSCTL_INO_ROOT;
	type = DT_DIR;
	} else {
	/*
	* Grab next entry.
	*/
	if ((error = zap_cursor_retrieve(&zc, &zap))) {
	if ((*eofp = (error == ENOENT)) != 0)
	break;
	else
	goto update;
	}

	if (zap.za_integer_length != 8 \|\|
	zap.za_num_integers != 1) {
	cmn_err(CE_WARN, "zap_readdir: bad directory "
	"entry, obj = %lld, offset = %lld\n",
	(u_longlong_t)zp->z_id,
	(u_longlong_t)offset);
	error = SET_ERROR(ENXIO);
	goto update;
	}

	objnum = ZFS_DIRENT_OBJ(zap.za_first_integer);
	/*
	* MacOS X can extract the object type here such as:
	* uint8_t type = ZFS_DIRENT_TYPE(zap.za_first_integer);
	*/
	type = ZFS_DIRENT_TYPE(zap.za_first_integer);

	if (check_sysattrs && !zap.za_normalization_conflict) {
	#ifdef TODO
	zap.za_normalization_conflict =
	xattr_sysattr_casechk(zap.za_name);
	#else
	panic("%s:%u: TODO", __func__, __LINE__);
	#endif
	}
	}

	if (flags & V_RDDIR_ACCFILTER) {
	/*
	* If we have no access at all, don't include
	* this entry in the returned information
	*/
	znode_t *ezp;
	if (zfs_zget(zp->z_zfsvfs, objnum, &ezp) != 0)
	goto skip_entry;
	if (!zfs_has_access(ezp, cr)) {
	vrele(ZTOV(ezp));
	goto skip_entry;
	}
	vrele(ZTOV(ezp));
	}

	if (flags & V_RDDIR_ENTFLAGS)
	reclen = EDIRENT_RECLEN(strlen(zap.za_name));
	else
	reclen = DIRENT64_RECLEN(strlen(zap.za_name));

	/*
	* Will this entry fit in the buffer?
	*/
	if (outcount + reclen > bufsize) {
	/*
	* Did we manage to fit anything in the buffer?
	*/
	if (!outcount) {
	error = SET_ERROR(EINVAL);
	goto update;
	}
	break;
	}
	if (flags & V_RDDIR_ENTFLAGS) {
	/*
	* Add extended flag entry:
	*/
	eodp->ed_ino = objnum;
	eodp->ed_reclen = reclen;
	/* NOTE: ed_off is the offset for the next entry */
	next = &(eodp->ed_off);
	eodp->ed_eflags = zap.za_normalization_conflict ?
	ED_CASE_CONFLICT : 0;
	(void) strncpy(eodp->ed_name, zap.za_name,
	EDIRENT_NAMELEN(reclen));
	eodp = (edirent_t *)((intptr_t)eodp + reclen);
	} else {
	/*
	* Add normal entry:
	*/
	odp->d_ino = objnum;
	odp->d_reclen = reclen;
	odp->d_namlen = strlen(zap.za_name);
	/* NOTE: d_off is the offset for the next entry. */
	next = &odp->d_off;
	strlcpy(odp->d_name, zap.za_name, odp->d_namlen + 1);
	odp->d_type = type;
	dirent_terminate(odp);
	odp = (dirent64_t *)((intptr_t)odp + reclen);
	}
	outcount += reclen;

	ASSERT3S(outcount, <=, bufsize);

	/* Prefetch znode */
	if (prefetch)
	dmu_prefetch(os, objnum, 0, 0, 0,
	ZIO_PRIORITY_SYNC_READ);

	skip_entry:
	/*
	* Move to the next entry, fill in the previous offset.
	*/
	if (offset > 2 \|\| (offset == 2 && !zfs_show_ctldir(zp))) {
	zap_cursor_advance(&zc);
	offset = zap_cursor_serialize(&zc);
	} else {
	offset += 1;
	}

	/* Fill the offset right after advancing the cursor. */
	if (next != NULL)
	*next = offset;
	if (cooks != NULL) {
	*cooks++ = offset;
	ncooks--;
	KASSERT(ncooks >= 0, ("ncookies=%d", ncooks));
	}
	}
	zp->z_zn_prefetch = B_FALSE; /* a lookup will re-enable pre-fetching */

	/* Subtract unused cookies */
	if (ncookies != NULL)
	*ncookies -= ncooks;

	if (zfs_uio_segflg(uio) == UIO_SYSSPACE && zfs_uio_iovcnt(uio) == 1) {
	iovp->iov_base += outcount;
	iovp->iov_len -= outcount;
	zfs_uio_resid(uio) -= outcount;
	} else if ((error =
	zfs_uiomove(outbuf, (long)outcount, UIO_READ, uio))) {
	/*
	* Reset the pointer.
	*/
	offset = zfs_uio_offset(uio);
	}

	update:
	zap_cursor_fini(&zc);
	if (zfs_uio_segflg(uio) != UIO_SYSSPACE \|\| zfs_uio_iovcnt(uio) != 1)
	kmem_free(outbuf, bufsize);

	if (error == ENOENT)
	error = 0;

	ZFS_ACCESSTIME_STAMP(zfsvfs, zp);

	zfs_uio_setoffset(uio, offset);
	ZFS_EXIT(zfsvfs);
	if (error != 0 && cookies != NULL) {
	free(*cookies, M_TEMP);
	*cookies = NULL;
	*ncookies = 0;
	}
	return (error);
	}

	/*
	* Get the requested file attributes and place them in the provided
	* vattr structure.
	*
	* IN: vp - vnode of file.
	* vap - va_mask identifies requested attributes.
	* If AT_XVATTR set, then optional attrs are requested
	* flags - ATTR_NOACLCHECK (CIFS server context)
	* cr - credentials of caller.
	*
	* OUT: vap - attribute values.
	*
	* RETURN: 0 (always succeeds).
	*/
	/* ARGSUSED */
	static int
	zfs_getattr(vnode_t vp, vattr_t vap, int flags, cred_t *cr)
	{
	znode_t *zp = VTOZ(vp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
	int error = 0;
	uint32_t blksize;
	u_longlong_t nblocks;
	uint64_t mtime[2], ctime[2], crtime[2], rdev;
	xvattr_t xvap = (xvattr_t )vap; /* vap may be an xvattr_t * */
	xoptattr_t *xoap = NULL;
	boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
	sa_bulk_attr_t bulk[4];
	int count = 0;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	zfs_fuid_map_ids(zp, cr, &vap->va_uid, &vap->va_gid);

	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CRTIME(zfsvfs), NULL, &crtime, 16);
	if (vp->v_type == VBLK \|\| vp->v_type == VCHR)
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_RDEV(zfsvfs), NULL,
	&rdev, 8);

	if ((error = sa_bulk_lookup(zp->z_sa_hdl, bulk, count)) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* If ACL is trivial don't bother looking for ACE_READ_ATTRIBUTES.
	* Also, if we are the owner don't bother, since owner should
	* always be allowed to read basic attributes of file.
	*/
	if (!(zp->z_pflags & ZFS_ACL_TRIVIAL) &&
	(vap->va_uid != crgetuid(cr))) {
	if ((error = zfs_zaccess(zp, ACE_READ_ATTRIBUTES, 0,
	skipaclchk, cr))) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	}

	/*
	* Return all attributes. It's cheaper to provide the answer
	* than to determine whether we were asked the question.
	*/

	vap->va_type = IFTOVT(zp->z_mode);
	vap->va_mode = zp->z_mode & ~S_IFMT;
	vn_fsid(vp, vap);
	vap->va_nodeid = zp->z_id;
	vap->va_nlink = zp->z_links;
	if ((vp->v_flag & VROOT) && zfs_show_ctldir(zp) &&
	zp->z_links < ZFS_LINK_MAX)
	vap->va_nlink++;
	vap->va_size = zp->z_size;
	if (vp->v_type == VBLK \|\| vp->v_type == VCHR)
	vap->va_rdev = zfs_cmpldev(rdev);
	vap->va_seq = zp->z_seq;
	vap->va_flags = 0; /* FreeBSD: Reset chflags(2) flags. */
	vap->va_filerev = zp->z_seq;

	/*
	* Add in any requested optional attributes and the create time.
	* Also set the corresponding bits in the returned attribute bitmap.
	*/
	if ((xoap = xva_getxoptattr(xvap)) != NULL && zfsvfs->z_use_fuids) {
	if (XVA_ISSET_REQ(xvap, XAT_ARCHIVE)) {
	xoap->xoa_archive =
	((zp->z_pflags & ZFS_ARCHIVE) != 0);
	XVA_SET_RTN(xvap, XAT_ARCHIVE);
	}

	if (XVA_ISSET_REQ(xvap, XAT_READONLY)) {
	xoap->xoa_readonly =
	((zp->z_pflags & ZFS_READONLY) != 0);
	XVA_SET_RTN(xvap, XAT_READONLY);
	}

	if (XVA_ISSET_REQ(xvap, XAT_SYSTEM)) {
	xoap->xoa_system =
	((zp->z_pflags & ZFS_SYSTEM) != 0);
	XVA_SET_RTN(xvap, XAT_SYSTEM);
	}

	if (XVA_ISSET_REQ(xvap, XAT_HIDDEN)) {
	xoap->xoa_hidden =
	((zp->z_pflags & ZFS_HIDDEN) != 0);
	XVA_SET_RTN(xvap, XAT_HIDDEN);
	}

	if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) {
	xoap->xoa_nounlink =
	((zp->z_pflags & ZFS_NOUNLINK) != 0);
	XVA_SET_RTN(xvap, XAT_NOUNLINK);
	}

	if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) {
	xoap->xoa_immutable =
	((zp->z_pflags & ZFS_IMMUTABLE) != 0);
	XVA_SET_RTN(xvap, XAT_IMMUTABLE);
	}

	if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) {
	xoap->xoa_appendonly =
	((zp->z_pflags & ZFS_APPENDONLY) != 0);
	XVA_SET_RTN(xvap, XAT_APPENDONLY);
	}

	if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) {
	xoap->xoa_nodump =
	((zp->z_pflags & ZFS_NODUMP) != 0);
	XVA_SET_RTN(xvap, XAT_NODUMP);
	}

	if (XVA_ISSET_REQ(xvap, XAT_OPAQUE)) {
	xoap->xoa_opaque =
	((zp->z_pflags & ZFS_OPAQUE) != 0);
	XVA_SET_RTN(xvap, XAT_OPAQUE);
	}

	if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) {
	xoap->xoa_av_quarantined =
	((zp->z_pflags & ZFS_AV_QUARANTINED) != 0);
	XVA_SET_RTN(xvap, XAT_AV_QUARANTINED);
	}

	if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) {
	xoap->xoa_av_modified =
	((zp->z_pflags & ZFS_AV_MODIFIED) != 0);
	XVA_SET_RTN(xvap, XAT_AV_MODIFIED);
	}

	if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP) &&
	vp->v_type == VREG) {
	zfs_sa_get_scanstamp(zp, xvap);
	}

	if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) {
	xoap->xoa_reparse = ((zp->z_pflags & ZFS_REPARSE) != 0);
	XVA_SET_RTN(xvap, XAT_REPARSE);
	}
	if (XVA_ISSET_REQ(xvap, XAT_GEN)) {
	xoap->xoa_generation = zp->z_gen;
	XVA_SET_RTN(xvap, XAT_GEN);
	}

	if (XVA_ISSET_REQ(xvap, XAT_OFFLINE)) {
	xoap->xoa_offline =
	((zp->z_pflags & ZFS_OFFLINE) != 0);
	XVA_SET_RTN(xvap, XAT_OFFLINE);
	}

	if (XVA_ISSET_REQ(xvap, XAT_SPARSE)) {
	xoap->xoa_sparse =
	((zp->z_pflags & ZFS_SPARSE) != 0);
	XVA_SET_RTN(xvap, XAT_SPARSE);
	}

	if (XVA_ISSET_REQ(xvap, XAT_PROJINHERIT)) {
	xoap->xoa_projinherit =
	((zp->z_pflags & ZFS_PROJINHERIT) != 0);
	XVA_SET_RTN(xvap, XAT_PROJINHERIT);
	}

	if (XVA_ISSET_REQ(xvap, XAT_PROJID)) {
	xoap->xoa_projid = zp->z_projid;
	XVA_SET_RTN(xvap, XAT_PROJID);
	}
	}

	ZFS_TIME_DECODE(&vap->va_atime, zp->z_atime);
	ZFS_TIME_DECODE(&vap->va_mtime, mtime);
	ZFS_TIME_DECODE(&vap->va_ctime, ctime);
	ZFS_TIME_DECODE(&vap->va_birthtime, crtime);


	sa_object_size(zp->z_sa_hdl, &blksize, &nblocks);
	vap->va_blksize = blksize;
	vap->va_bytes = nblocks << 9; /* nblocks * 512 */

	if (zp->z_blksz == 0) {
	/*
	* Block size hasn't been set; suggest maximal I/O transfers.
	*/
	vap->va_blksize = zfsvfs->z_max_blksz;
	}

	ZFS_EXIT(zfsvfs);
	return (0);
	}

	/*
	* Set the file attributes to the values contained in the
	* vattr structure.
	*
	* IN: zp - znode of file to be modified.
	* vap - new attribute values.
	* If AT_XVATTR set, then optional attrs are being set
	* flags - ATTR_UTIME set if non-default time values provided.
	* - ATTR_NOACLCHECK (CIFS context only).
	* cr - credentials of caller.
	* ct - caller context
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* vp - ctime updated, mtime updated if size changed.
	*/
	/* ARGSUSED */
	int
	zfs_setattr(znode_t zp, vattr_t vap, int flags, cred_t *cr)
	{
	vnode_t *vp = ZTOV(zp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
	objset_t *os;
	zilog_t *zilog;
	dmu_tx_t *tx;
	vattr_t oldva;
	xvattr_t tmpxvattr;
	uint_t mask = vap->va_mask;
	uint_t saved_mask = 0;
	uint64_t saved_mode;
	int trim_mask = 0;
	uint64_t new_mode;
	uint64_t new_uid, new_gid;
	uint64_t xattr_obj;
	uint64_t mtime[2], ctime[2];
	uint64_t projid = ZFS_INVALID_PROJID;
	znode_t *attrzp;
	int need_policy = FALSE;
	int err, err2;
	zfs_fuid_info_t *fuidp = NULL;
	xvattr_t xvap = (xvattr_t )vap; /* vap may be an xvattr_t * */
	xoptattr_t *xoap;
	zfs_acl_t *aclp;
	boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
	boolean_t fuid_dirtied = B_FALSE;
	sa_bulk_attr_t bulk[7], xattr_bulk[7];
	int count = 0, xattr_count = 0;

	if (mask == 0)
	return (0);

	if (mask & AT_NOSET)
	return (SET_ERROR(EINVAL));

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	os = zfsvfs->z_os;
	zilog = zfsvfs->z_log;

	/*
	* Make sure that if we have ephemeral uid/gid or xvattr specified
	* that file system is at proper version level
	*/

	if (zfsvfs->z_use_fuids == B_FALSE &&
	(((mask & AT_UID) && IS_EPHEMERAL(vap->va_uid)) \|\|
	((mask & AT_GID) && IS_EPHEMERAL(vap->va_gid)) \|\|
	(mask & AT_XVATTR))) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	if (mask & AT_SIZE && vp->v_type == VDIR) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EISDIR));
	}

	if (mask & AT_SIZE && vp->v_type != VREG && vp->v_type != VFIFO) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	/*
	* If this is an xvattr_t, then get a pointer to the structure of
	* optional attributes. If this is NULL, then we have a vattr_t.
	*/
	xoap = xva_getxoptattr(xvap);

	xva_init(&tmpxvattr);

	/*
	* Immutable files can only alter immutable bit and atime
	*/
	if ((zp->z_pflags & ZFS_IMMUTABLE) &&
	((mask & (AT_SIZE\|AT_UID\|AT_GID\|AT_MTIME\|AT_MODE)) \|\|
	((mask & AT_XVATTR) && XVA_ISSET_REQ(xvap, XAT_CREATETIME)))) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	/*
	* Note: ZFS_READONLY is handled in zfs_zaccess_common.
	*/

	/*
	* Verify timestamps doesn't overflow 32 bits.
	* ZFS can handle large timestamps, but 32bit syscalls can't
	* handle times greater than 2039. This check should be removed
	* once large timestamps are fully supported.
	*/
	if (mask & (AT_ATIME \| AT_MTIME)) {
	if (((mask & AT_ATIME) && TIMESPEC_OVERFLOW(&vap->va_atime)) \|\|
	((mask & AT_MTIME) && TIMESPEC_OVERFLOW(&vap->va_mtime))) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EOVERFLOW));
	}
	}
	if (xoap != NULL && (mask & AT_XVATTR)) {
	if (XVA_ISSET_REQ(xvap, XAT_CREATETIME) &&
	TIMESPEC_OVERFLOW(&vap->va_birthtime)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EOVERFLOW));
	}

	if (XVA_ISSET_REQ(xvap, XAT_PROJID)) {
	if (!dmu_objset_projectquota_enabled(os) \|\|
	(!S_ISREG(zp->z_mode) && !S_ISDIR(zp->z_mode))) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EOPNOTSUPP));
	}

	projid = xoap->xoa_projid;
	if (unlikely(projid == ZFS_INVALID_PROJID)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	if (projid == zp->z_projid && zp->z_pflags & ZFS_PROJID)
	projid = ZFS_INVALID_PROJID;
	else
	need_policy = TRUE;
	}

	if (XVA_ISSET_REQ(xvap, XAT_PROJINHERIT) &&
	(xoap->xoa_projinherit !=
	((zp->z_pflags & ZFS_PROJINHERIT) != 0)) &&
	(!dmu_objset_projectquota_enabled(os) \|\|
	(!S_ISREG(zp->z_mode) && !S_ISDIR(zp->z_mode)))) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EOPNOTSUPP));
	}
	}

	attrzp = NULL;
	aclp = NULL;

	if (zfsvfs->z_vfs->vfs_flag & VFS_RDONLY) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EROFS));
	}

	/*
	* First validate permissions
	*/

	if (mask & AT_SIZE) {
	/*
	* XXX - Note, we are not providing any open
	* mode flags here (like FNDELAY), so we may
	* block if there are locks present... this
	* should be addressed in openat().
	*/
	/* XXX - would it be OK to generate a log record here? */
	err = zfs_freesp(zp, vap->va_size, 0, 0, FALSE);
	if (err) {
	ZFS_EXIT(zfsvfs);
	return (err);
	}
	}

	if (mask & (AT_ATIME\|AT_MTIME) \|\|
	((mask & AT_XVATTR) && (XVA_ISSET_REQ(xvap, XAT_HIDDEN) \|\|
	XVA_ISSET_REQ(xvap, XAT_READONLY) \|\|
	XVA_ISSET_REQ(xvap, XAT_ARCHIVE) \|\|
	XVA_ISSET_REQ(xvap, XAT_OFFLINE) \|\|
	XVA_ISSET_REQ(xvap, XAT_SPARSE) \|\|
	XVA_ISSET_REQ(xvap, XAT_CREATETIME) \|\|
	XVA_ISSET_REQ(xvap, XAT_SYSTEM)))) {
	need_policy = zfs_zaccess(zp, ACE_WRITE_ATTRIBUTES, 0,
	skipaclchk, cr);
	}

	if (mask & (AT_UID\|AT_GID)) {
	int idmask = (mask & (AT_UID\|AT_GID));
	int take_owner;
	int take_group;

	/*
	* NOTE: even if a new mode is being set,
	* we may clear S_ISUID/S_ISGID bits.
	*/

	if (!(mask & AT_MODE))
	vap->va_mode = zp->z_mode;

	/*
	* Take ownership or chgrp to group we are a member of
	*/

	take_owner = (mask & AT_UID) && (vap->va_uid == crgetuid(cr));
	take_group = (mask & AT_GID) &&
	zfs_groupmember(zfsvfs, vap->va_gid, cr);

	/*
	* If both AT_UID and AT_GID are set then take_owner and
	* take_group must both be set in order to allow taking
	* ownership.
	*
	* Otherwise, send the check through secpolicy_vnode_setattr()
	*
	*/

	if (((idmask == (AT_UID\|AT_GID)) && take_owner && take_group) \|\|
	((idmask == AT_UID) && take_owner) \|\|
	((idmask == AT_GID) && take_group)) {
	if (zfs_zaccess(zp, ACE_WRITE_OWNER, 0,
	skipaclchk, cr) == 0) {
	/*
	* Remove setuid/setgid for non-privileged users
	*/
	secpolicy_setid_clear(vap, vp, cr);
	trim_mask = (mask & (AT_UID\|AT_GID));
	} else {
	need_policy = TRUE;
	}
	} else {
	need_policy = TRUE;
	}
	}

	oldva.va_mode = zp->z_mode;
	zfs_fuid_map_ids(zp, cr, &oldva.va_uid, &oldva.va_gid);
	if (mask & AT_XVATTR) {
	/*
	* Update xvattr mask to include only those attributes
	* that are actually changing.
	*
	* the bits will be restored prior to actually setting
	* the attributes so the caller thinks they were set.
	*/
	if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) {
	if (xoap->xoa_appendonly !=
	((zp->z_pflags & ZFS_APPENDONLY) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_APPENDONLY);
	XVA_SET_REQ(&tmpxvattr, XAT_APPENDONLY);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_PROJINHERIT)) {
	if (xoap->xoa_projinherit !=
	((zp->z_pflags & ZFS_PROJINHERIT) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_PROJINHERIT);
	XVA_SET_REQ(&tmpxvattr, XAT_PROJINHERIT);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) {
	if (xoap->xoa_nounlink !=
	((zp->z_pflags & ZFS_NOUNLINK) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_NOUNLINK);
	XVA_SET_REQ(&tmpxvattr, XAT_NOUNLINK);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) {
	if (xoap->xoa_immutable !=
	((zp->z_pflags & ZFS_IMMUTABLE) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_IMMUTABLE);
	XVA_SET_REQ(&tmpxvattr, XAT_IMMUTABLE);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) {
	if (xoap->xoa_nodump !=
	((zp->z_pflags & ZFS_NODUMP) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_NODUMP);
	XVA_SET_REQ(&tmpxvattr, XAT_NODUMP);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) {
	if (xoap->xoa_av_modified !=
	((zp->z_pflags & ZFS_AV_MODIFIED) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_AV_MODIFIED);
	XVA_SET_REQ(&tmpxvattr, XAT_AV_MODIFIED);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) {
	if ((vp->v_type != VREG &&
	xoap->xoa_av_quarantined) \|\|
	xoap->xoa_av_quarantined !=
	((zp->z_pflags & ZFS_AV_QUARANTINED) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_AV_QUARANTINED);
	XVA_SET_REQ(&tmpxvattr, XAT_AV_QUARANTINED);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	if (need_policy == FALSE &&
	(XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP) \|\|
	XVA_ISSET_REQ(xvap, XAT_OPAQUE))) {
	need_policy = TRUE;
	}
	}

	if (mask & AT_MODE) {
	if (zfs_zaccess(zp, ACE_WRITE_ACL, 0, skipaclchk, cr) == 0) {
	err = secpolicy_setid_setsticky_clear(vp, vap,
	&oldva, cr);
	if (err) {
	ZFS_EXIT(zfsvfs);
	return (err);
	}
	trim_mask \|= AT_MODE;
	} else {
	need_policy = TRUE;
	}
	}

	if (need_policy) {
	/*
	* If trim_mask is set then take ownership
	* has been granted or write_acl is present and user
	* has the ability to modify mode. In that case remove
	* UID\|GID and or MODE from mask so that
	* secpolicy_vnode_setattr() doesn't revoke it.
	*/

	if (trim_mask) {
	saved_mask = vap->va_mask;
	vap->va_mask &= ~trim_mask;
	if (trim_mask & AT_MODE) {
	/*
	* Save the mode, as secpolicy_vnode_setattr()
	* will overwrite it with ova.va_mode.
	*/
	saved_mode = vap->va_mode;
	}
	}
	err = secpolicy_vnode_setattr(cr, vp, vap, &oldva, flags,
	(int ()(void , int, cred_t *))zfs_zaccess_unix, zp);
	if (err) {
	ZFS_EXIT(zfsvfs);
	return (err);
	}

	if (trim_mask) {
	vap->va_mask \|= saved_mask;
	if (trim_mask & AT_MODE) {
	/*
	* Recover the mode after
	* secpolicy_vnode_setattr().
	*/
	vap->va_mode = saved_mode;
	}
	}
	}

	/*
	* secpolicy_vnode_setattr, or take ownership may have
	* changed va_mask
	*/
	mask = vap->va_mask;

	if ((mask & (AT_UID \| AT_GID)) \|\| projid != ZFS_INVALID_PROJID) {
	err = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
	&xattr_obj, sizeof (xattr_obj));

	if (err == 0 && xattr_obj) {
	err = zfs_zget(zp->z_zfsvfs, xattr_obj, &attrzp);
	if (err == 0) {
	err = vn_lock(ZTOV(attrzp), LK_EXCLUSIVE);
	if (err != 0)
	vrele(ZTOV(attrzp));
	}
	if (err)
	goto out2;
	}
	if (mask & AT_UID) {
	new_uid = zfs_fuid_create(zfsvfs,
	(uint64_t)vap->va_uid, cr, ZFS_OWNER, &fuidp);
	if (new_uid != zp->z_uid &&
	zfs_id_overquota(zfsvfs, DMU_USERUSED_OBJECT,
	new_uid)) {
	if (attrzp)
	vput(ZTOV(attrzp));
	err = SET_ERROR(EDQUOT);
	goto out2;
	}
	}

	if (mask & AT_GID) {
	new_gid = zfs_fuid_create(zfsvfs, (uint64_t)vap->va_gid,
	cr, ZFS_GROUP, &fuidp);
	if (new_gid != zp->z_gid &&
	zfs_id_overquota(zfsvfs, DMU_GROUPUSED_OBJECT,
	new_gid)) {
	if (attrzp)
	vput(ZTOV(attrzp));
	err = SET_ERROR(EDQUOT);
	goto out2;
	}
	}

	if (projid != ZFS_INVALID_PROJID &&
	zfs_id_overquota(zfsvfs, DMU_PROJECTUSED_OBJECT, projid)) {
	if (attrzp)
	vput(ZTOV(attrzp));
	err = SET_ERROR(EDQUOT);
	goto out2;
	}
	}
	tx = dmu_tx_create(os);

	if (mask & AT_MODE) {
	uint64_t pmode = zp->z_mode;
	uint64_t acl_obj;
	new_mode = (pmode & S_IFMT) \| (vap->va_mode & ~S_IFMT);

	if (zp->z_zfsvfs->z_acl_mode == ZFS_ACL_RESTRICTED &&
	!(zp->z_pflags & ZFS_ACL_TRIVIAL)) {
	err = SET_ERROR(EPERM);
	goto out;
	}

	if ((err = zfs_acl_chmod_setattr(zp, &aclp, new_mode)))
	goto out;

	if (!zp->z_is_sa && ((acl_obj = zfs_external_acl(zp)) != 0)) {
	/*
	* Are we upgrading ACL from old V0 format
	* to V1 format?
	*/
	if (zfsvfs->z_version >= ZPL_VERSION_FUID &&
	zfs_znode_acl_version(zp) ==
	ZFS_ACL_VERSION_INITIAL) {
	dmu_tx_hold_free(tx, acl_obj, 0,
	DMU_OBJECT_END);
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
	0, aclp->z_acl_bytes);
	} else {
	dmu_tx_hold_write(tx, acl_obj, 0,
	aclp->z_acl_bytes);
	}
	} else if (!zp->z_is_sa && aclp->z_acl_bytes > ZFS_ACE_SPACE) {
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
	0, aclp->z_acl_bytes);
	}
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
	} else {
	if (((mask & AT_XVATTR) &&
	XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP)) \|\|
	(projid != ZFS_INVALID_PROJID &&
	!(zp->z_pflags & ZFS_PROJID)))
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
	else
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
	}

	if (attrzp) {
	dmu_tx_hold_sa(tx, attrzp->z_sa_hdl, B_FALSE);
	}

	fuid_dirtied = zfsvfs->z_fuid_dirty;
	if (fuid_dirtied)
	zfs_fuid_txhold(zfsvfs, tx);

	zfs_sa_upgrade_txholds(tx, zp);

	err = dmu_tx_assign(tx, TXG_WAIT);
	if (err)
	goto out;

	count = 0;
	/*
	* Set each attribute requested.
	* We group settings according to the locks they need to acquire.
	*
	* Note: you cannot set ctime directly, although it will be
	* updated as a side-effect of calling this function.
	*/

	if (projid != ZFS_INVALID_PROJID && !(zp->z_pflags & ZFS_PROJID)) {
	/*
	* For the existed object that is upgraded from old system,
	* its on-disk layout has no slot for the project ID attribute.
	* But quota accounting logic needs to access related slots by
	* offset directly. So we need to adjust old objects' layout
	* to make the project ID to some unified and fixed offset.
	*/
	if (attrzp)
	err = sa_add_projid(attrzp->z_sa_hdl, tx, projid);
	if (err == 0)
	err = sa_add_projid(zp->z_sa_hdl, tx, projid);

	if (unlikely(err == EEXIST))
	err = 0;
	else if (err != 0)
	goto out;
	else
	projid = ZFS_INVALID_PROJID;
	}

	if (mask & (AT_UID\|AT_GID\|AT_MODE))
	mutex_enter(&zp->z_acl_lock);

	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
	&zp->z_pflags, sizeof (zp->z_pflags));

	if (attrzp) {
	if (mask & (AT_UID\|AT_GID\|AT_MODE))
	mutex_enter(&attrzp->z_acl_lock);
	SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
	SA_ZPL_FLAGS(zfsvfs), NULL, &attrzp->z_pflags,
	sizeof (attrzp->z_pflags));
	if (projid != ZFS_INVALID_PROJID) {
	attrzp->z_projid = projid;
	SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
	SA_ZPL_PROJID(zfsvfs), NULL, &attrzp->z_projid,
	sizeof (attrzp->z_projid));
	}
	}

	if (mask & (AT_UID\|AT_GID)) {

	if (mask & AT_UID) {
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
	&new_uid, sizeof (new_uid));
	zp->z_uid = new_uid;
	if (attrzp) {
	SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
	SA_ZPL_UID(zfsvfs), NULL, &new_uid,
	sizeof (new_uid));
	attrzp->z_uid = new_uid;
	}
	}

	if (mask & AT_GID) {
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs),
	NULL, &new_gid, sizeof (new_gid));
	zp->z_gid = new_gid;
	if (attrzp) {
	SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
	SA_ZPL_GID(zfsvfs), NULL, &new_gid,
	sizeof (new_gid));
	attrzp->z_gid = new_gid;
	}
	}
	if (!(mask & AT_MODE)) {
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs),
	NULL, &new_mode, sizeof (new_mode));
	new_mode = zp->z_mode;
	}
	err = zfs_acl_chown_setattr(zp);
	ASSERT0(err);
	if (attrzp) {
	vn_seqc_write_begin(ZTOV(attrzp));
	err = zfs_acl_chown_setattr(attrzp);
	vn_seqc_write_end(ZTOV(attrzp));
	ASSERT0(err);
	}
	}

	if (mask & AT_MODE) {
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL,
	&new_mode, sizeof (new_mode));
	zp->z_mode = new_mode;
	ASSERT3P(aclp, !=, NULL);
	err = zfs_aclset_common(zp, aclp, cr, tx);
	ASSERT0(err);
	if (zp->z_acl_cached)
	zfs_acl_free(zp->z_acl_cached);
	zp->z_acl_cached = aclp;
	aclp = NULL;
	}


	if (mask & AT_ATIME) {
	ZFS_TIME_ENCODE(&vap->va_atime, zp->z_atime);
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL,
	&zp->z_atime, sizeof (zp->z_atime));
	}

	if (mask & AT_MTIME) {
	ZFS_TIME_ENCODE(&vap->va_mtime, mtime);
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL,
	mtime, sizeof (mtime));
	}

	if (projid != ZFS_INVALID_PROJID) {
	zp->z_projid = projid;
	SA_ADD_BULK_ATTR(bulk, count,
	SA_ZPL_PROJID(zfsvfs), NULL, &zp->z_projid,
	sizeof (zp->z_projid));
	}

	/* XXX - shouldn't this be done before the ATIME/MTIME checks? */
	if (mask & AT_SIZE && !(mask & AT_MTIME)) {
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs),
	NULL, mtime, sizeof (mtime));
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
	&ctime, sizeof (ctime));
	zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime);
	} else if (mask != 0) {
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
	&ctime, sizeof (ctime));
	zfs_tstamp_update_setup(zp, STATE_CHANGED, mtime, ctime);
	if (attrzp) {
	SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
	SA_ZPL_CTIME(zfsvfs), NULL,
	&ctime, sizeof (ctime));
	zfs_tstamp_update_setup(attrzp, STATE_CHANGED,
	mtime, ctime);
	}
	}

	/*
	* Do this after setting timestamps to prevent timestamp
	* update from toggling bit
	*/

	if (xoap && (mask & AT_XVATTR)) {

	if (XVA_ISSET_REQ(xvap, XAT_CREATETIME))
	xoap->xoa_createtime = vap->va_birthtime;
	/*
	* restore trimmed off masks
	* so that return masks can be set for caller.
	*/

	if (XVA_ISSET_REQ(&tmpxvattr, XAT_APPENDONLY)) {
	XVA_SET_REQ(xvap, XAT_APPENDONLY);
	}
	if (XVA_ISSET_REQ(&tmpxvattr, XAT_NOUNLINK)) {
	XVA_SET_REQ(xvap, XAT_NOUNLINK);
	}
	if (XVA_ISSET_REQ(&tmpxvattr, XAT_IMMUTABLE)) {
	XVA_SET_REQ(xvap, XAT_IMMUTABLE);
	}
	if (XVA_ISSET_REQ(&tmpxvattr, XAT_NODUMP)) {
	XVA_SET_REQ(xvap, XAT_NODUMP);
	}
	if (XVA_ISSET_REQ(&tmpxvattr, XAT_AV_MODIFIED)) {
	XVA_SET_REQ(xvap, XAT_AV_MODIFIED);
	}
	if (XVA_ISSET_REQ(&tmpxvattr, XAT_AV_QUARANTINED)) {
	XVA_SET_REQ(xvap, XAT_AV_QUARANTINED);
	}
	if (XVA_ISSET_REQ(&tmpxvattr, XAT_PROJINHERIT)) {
	XVA_SET_REQ(xvap, XAT_PROJINHERIT);
	}

	if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP))
	ASSERT3S(vp->v_type, ==, VREG);

	zfs_xvattr_set(zp, xvap, tx);
	}

	if (fuid_dirtied)
	zfs_fuid_sync(zfsvfs, tx);

	if (mask != 0)
	zfs_log_setattr(zilog, tx, TX_SETATTR, zp, vap, mask, fuidp);

	if (mask & (AT_UID\|AT_GID\|AT_MODE))
	mutex_exit(&zp->z_acl_lock);

	if (attrzp) {
	if (mask & (AT_UID\|AT_GID\|AT_MODE))
	mutex_exit(&attrzp->z_acl_lock);
	}
	out:
	if (err == 0 && attrzp) {
	err2 = sa_bulk_update(attrzp->z_sa_hdl, xattr_bulk,
	xattr_count, tx);
	ASSERT0(err2);
	}

	if (attrzp)
	vput(ZTOV(attrzp));

	if (aclp)
	zfs_acl_free(aclp);

	if (fuidp) {
	zfs_fuid_info_free(fuidp);
	fuidp = NULL;
	}

	if (err) {
	dmu_tx_abort(tx);
	} else {
	err2 = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
	dmu_tx_commit(tx);
	}

	out2:
	if (os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	ZFS_EXIT(zfsvfs);
	return (err);
	}

	+/*
	+ * Look up the directory entries corresponding to the source and target
	+ * directory/name pairs.
	+ */
	+static int
	+zfs_rename_relock_lookup(znode_t sdzp, const struct componentname scnp,
	+ znode_t *szpp, znode_t tdzp, const struct componentname *tcnp,
	+ znode_t **tzpp)
	+{
	+ zfsvfs_t *zfsvfs;
	+ znode_t szp, tzp;
	+ int error;
	+
	+ /*
	+ * Before using sdzp and tdzp we must ensure that they are live.
	+ * As a porting legacy from illumos we have two things to worry
	+ * about. One is typical for FreeBSD and it is that the vnode is
	+ * not reclaimed (doomed). The other is that the znode is live.
	+ * The current code can invalidate the znode without acquiring the
	+ * corresponding vnode lock if the object represented by the znode
	+ * and vnode is no longer valid after a rollback or receive operation.
	+ * z_teardown_lock hidden behind ZFS_ENTER and ZFS_EXIT is the lock
	+ * that protects the znodes from the invalidation.
	+ */
	+ zfsvfs = sdzp->z_zfsvfs;
	+ ASSERT3P(zfsvfs, ==, tdzp->z_zfsvfs);
	+ ZFS_ENTER(zfsvfs);
	+ ZFS_VERIFY_ZP(sdzp);
	+ ZFS_VERIFY_ZP(tdzp);
	+
	+ /*
	+ * Re-resolve svp to be certain it still exists and fetch the
	+ * correct vnode.
	+ */
	+ error = zfs_dirent_lookup(sdzp, scnp->cn_nameptr, &szp, ZEXISTS);
	+ if (error != 0) {
	+ /* Source entry invalid or not there. */
	+ if ((scnp->cn_flags & ISDOTDOT) != 0 \|\|
	+ (scnp->cn_namelen == 1 && scnp->cn_nameptr[0] == '.'))
	+ error = SET_ERROR(EINVAL);
	+ goto out;
	+ }
	+ *szpp = szp;
	+
	+ /*
	+ * Re-resolve tvp, if it disappeared we just carry on.
	+ */
	+ error = zfs_dirent_lookup(tdzp, tcnp->cn_nameptr, &tzp, 0);
	+ if (error != 0) {
	+ vrele(ZTOV(szp));
	+ if ((tcnp->cn_flags & ISDOTDOT) != 0)
	+ error = SET_ERROR(EINVAL);
	+ goto out;
	+ }
	+ *tzpp = tzp;
	+out:
	+ ZFS_EXIT(zfsvfs);
	+ return (error);
	+}
	+
	/*
	* We acquire all but fdvp locks using non-blocking acquisitions. If we
	* fail to acquire any lock in the path we will drop all held locks,
	* acquire the new lock in a blocking fashion, and then release it and
	* restart the rename. This acquire/release step ensures that we do not
	* spin on a lock waiting for release. On error release all vnode locks
	* and decrement references the way tmpfs_rename() would do.
	*/
	static int
	zfs_rename_relock(struct vnode sdvp, struct vnode *svpp,
	struct vnode tdvp, struct vnode *tvpp,
	const struct componentname scnp, const struct componentname tcnp)
	{
	- zfsvfs_t *zfsvfs;
	struct vnode nvp, svp, *tvp;
	znode_t sdzp, tdzp, szp, tzp;
	- const char *snm = scnp->cn_nameptr;
	- const char *tnm = tcnp->cn_nameptr;
	- int error;
	+ int error;

	VOP_UNLOCK1(tdvp);
	if (tvpp != NULL && tvpp != tdvp)
	VOP_UNLOCK1(*tvpp);

	relock:
	error = vn_lock(sdvp, LK_EXCLUSIVE);
	if (error)
	goto out;
	- sdzp = VTOZ(sdvp);
	-
	error = vn_lock(tdvp, LK_EXCLUSIVE \| LK_NOWAIT);
	if (error != 0) {
	VOP_UNLOCK1(sdvp);
	if (error != EBUSY)
	goto out;
	error = vn_lock(tdvp, LK_EXCLUSIVE);
	if (error)
	goto out;
	VOP_UNLOCK1(tdvp);
	goto relock;
	}
	tdzp = VTOZ(tdvp);
	+ sdzp = VTOZ(sdvp);

	- /*
	- * Before using sdzp and tdzp we must ensure that they are live.
	- * As a porting legacy from illumos we have two things to worry
	- * about. One is typical for FreeBSD and it is that the vnode is
	- * not reclaimed (doomed). The other is that the znode is live.
	- * The current code can invalidate the znode without acquiring the
	- * corresponding vnode lock if the object represented by the znode
	- * and vnode is no longer valid after a rollback or receive operation.
	- * z_teardown_lock hidden behind ZFS_ENTER and ZFS_EXIT is the lock
	- * that protects the znodes from the invalidation.
	- */
	- zfsvfs = sdzp->z_zfsvfs;
	- ASSERT3P(zfsvfs, ==, tdzp->z_zfsvfs);
	- ZFS_ENTER(zfsvfs);
	-
	- /*
	- * We can not use ZFS_VERIFY_ZP() here because it could directly return
	- * bypassing the cleanup code in the case of an error.
	- */
	- if (tdzp->z_sa_hdl == NULL \|\| sdzp->z_sa_hdl == NULL) {
	- ZFS_EXIT(zfsvfs);
	- VOP_UNLOCK1(sdvp);
	- VOP_UNLOCK1(tdvp);
	- error = SET_ERROR(EIO);
	- goto out;
	- }
	-
	- /*
	- * Re-resolve svp to be certain it still exists and fetch the
	- * correct vnode.
	- */
	- error = zfs_dirent_lookup(sdzp, snm, &szp, ZEXISTS);
	+ error = zfs_rename_relock_lookup(sdzp, scnp, &szp, tdzp, tcnp, &tzp);
	if (error != 0) {
	- /* Source entry invalid or not there. */
	- ZFS_EXIT(zfsvfs);
	VOP_UNLOCK1(sdvp);
	VOP_UNLOCK1(tdvp);
	- if ((scnp->cn_flags & ISDOTDOT) != 0 \|\|
	- (scnp->cn_namelen == 1 && scnp->cn_nameptr[0] == '.'))
	- error = SET_ERROR(EINVAL);
	goto out;
	}
	svp = ZTOV(szp);
	-
	- /*
	- * Re-resolve tvp, if it disappeared we just carry on.
	- */
	- error = zfs_dirent_lookup(tdzp, tnm, &tzp, 0);
	- if (error != 0) {
	- ZFS_EXIT(zfsvfs);
	- VOP_UNLOCK1(sdvp);
	- VOP_UNLOCK1(tdvp);
	- vrele(svp);
	- if ((tcnp->cn_flags & ISDOTDOT) != 0)
	- error = SET_ERROR(EINVAL);
	- goto out;
	- }
	- if (tzp != NULL)
	- tvp = ZTOV(tzp);
	- else
	- tvp = NULL;
	-
	- /*
	- * At present the vnode locks must be acquired before z_teardown_lock,
	- * although it would be more logical to use the opposite order.
	- */
	- ZFS_EXIT(zfsvfs);
	+ tvp = tzp != NULL ? ZTOV(tzp) : NULL;

	/*
	* Now try acquire locks on svp and tvp.
	*/
	nvp = svp;
	error = vn_lock(nvp, LK_EXCLUSIVE \| LK_NOWAIT);
	if (error != 0) {
	VOP_UNLOCK1(sdvp);
	VOP_UNLOCK1(tdvp);
	if (tvp != NULL)
	vrele(tvp);
	if (error != EBUSY) {
	vrele(nvp);
	goto out;
	}
	error = vn_lock(nvp, LK_EXCLUSIVE);
	if (error != 0) {
	vrele(nvp);
	goto out;
	}
	VOP_UNLOCK1(nvp);
	/*
	* Concurrent rename race.
	* XXX ?
	*/
	if (nvp == tdvp) {
	vrele(nvp);
	error = SET_ERROR(EINVAL);
	goto out;
	}
	vrele(*svpp);
	*svpp = nvp;
	goto relock;
	}
	vrele(*svpp);
	*svpp = nvp;

	if (*tvpp != NULL)
	vrele(*tvpp);
	*tvpp = NULL;
	if (tvp != NULL) {
	nvp = tvp;
	error = vn_lock(nvp, LK_EXCLUSIVE \| LK_NOWAIT);
	if (error != 0) {
	VOP_UNLOCK1(sdvp);
	VOP_UNLOCK1(tdvp);
	VOP_UNLOCK1(*svpp);
	if (error != EBUSY) {
	vrele(nvp);
	goto out;
	}
	error = vn_lock(nvp, LK_EXCLUSIVE);
	if (error != 0) {
	vrele(nvp);
	goto out;
	}
	vput(nvp);
	goto relock;
	}
	*tvpp = nvp;
	}

	return (0);

	out:
	return (error);
	}

	/*
	* Note that we must use VRELE_ASYNC in this function as it walks
	* up the directory tree and vrele may need to acquire an exclusive
	* lock if a last reference to a vnode is dropped.
	*/
	static int
	zfs_rename_check(znode_t szp, znode_t sdzp, znode_t *tdzp)
	{
	zfsvfs_t *zfsvfs;
	znode_t zp, zp1;
	uint64_t parent;
	int error;

	zfsvfs = tdzp->z_zfsvfs;
	if (tdzp == szp)
	return (SET_ERROR(EINVAL));
	if (tdzp == sdzp)
	return (0);
	if (tdzp->z_id == zfsvfs->z_root)
	return (0);
	zp = tdzp;
	for (;;) {
	ASSERT(!zp->z_unlinked);
	if ((error = sa_lookup(zp->z_sa_hdl,
	SA_ZPL_PARENT(zfsvfs), &parent, sizeof (parent))) != 0)
	break;

	if (parent == szp->z_id) {
	error = SET_ERROR(EINVAL);
	break;
	}
	if (parent == zfsvfs->z_root)
	break;
	if (parent == sdzp->z_id)
	break;

	error = zfs_zget(zfsvfs, parent, &zp1);
	if (error != 0)
	break;

	if (zp != tdzp)
	VN_RELE_ASYNC(ZTOV(zp),
	dsl_pool_zrele_taskq(
	dmu_objset_pool(zfsvfs->z_os)));
	zp = zp1;
	}

	if (error == ENOTDIR)
	panic("checkpath: .. not a directory\n");
	if (zp != tdzp)
	VN_RELE_ASYNC(ZTOV(zp),
	dsl_pool_zrele_taskq(dmu_objset_pool(zfsvfs->z_os)));
	return (error);
	}

	#if __FreeBSD_version < 1300124
	static void
	cache_vop_rename(struct vnode fdvp, struct vnode fvp, struct vnode *tdvp,
	struct vnode tvp, struct componentname fcnp, struct componentname *tcnp)
	{

	cache_purge(fvp);
	if (tvp != NULL)
	cache_purge(tvp);
	cache_purge_negative(tdvp);
	}
	#endif

	+static int
	+zfs_do_rename_impl(vnode_t sdvp, vnode_t svpp, struct componentname scnp,
	+ vnode_t tdvp, vnode_t tvpp, struct componentname tcnp,
	+ cred_t *cr);
	+
	/*
	* Move an entry from the provided source directory to the target
	* directory. Change the entry name as indicated.
	*
	* IN: sdvp - Source directory containing the "old entry".
	- * snm - Old entry name.
	+ * scnp - Old entry name.
	* tdvp - Target directory to contain the "new entry".
	- * tnm - New entry name.
	+ * tcnp - New entry name.
	* cr - credentials of caller.
	- * ct - caller context
	- * flags - case flags
	+ * INOUT: svpp - Source file
	+ * tvpp - Target file, may point to NULL initially
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* sdvp,tdvp - ctime\|mtime updated
	*/
	/ARGSUSED/
	static int
	-zfs_rename_(vnode_t sdvp, vnode_t svpp, struct componentname scnp,
	+zfs_do_rename(vnode_t sdvp, vnode_t svpp, struct componentname scnp,
	vnode_t tdvp, vnode_t tvpp, struct componentname tcnp,
	- cred_t *cr, int log)
	+ cred_t *cr)
	{
	- zfsvfs_t *zfsvfs;
	- znode_t sdzp, tdzp, szp, tzp;
	- zilog_t *zilog = NULL;
	- dmu_tx_t *tx;
	- const char *snm = scnp->cn_nameptr;
	- const char *tnm = tcnp->cn_nameptr;
	- int error = 0;
	- bool want_seqc_end __maybe_unused = false;
	+ int error;
	+
	+ ASSERT_VOP_ELOCKED(tdvp, __func__);
	+ if (*tvpp != NULL)
	+ ASSERT_VOP_ELOCKED(*tvpp, __func__);

	/* Reject renames across filesystems. */
	if ((*svpp)->v_mount != tdvp->v_mount \|\|
	((tvpp) != NULL && (svpp)->v_mount != (*tvpp)->v_mount)) {
	error = SET_ERROR(EXDEV);
	goto out;
	}

	if (zfsctl_is_node(tdvp)) {
	error = SET_ERROR(EXDEV);
	goto out;
	}

	/*
	* Lock all four vnodes to ensure safety and semantics of renaming.
	*/
	error = zfs_rename_relock(sdvp, svpp, tdvp, tvpp, scnp, tcnp);
	if (error != 0) {
	/* no vnodes are locked in the case of error here */
	return (error);
	}

	+ error = zfs_do_rename_impl(sdvp, svpp, scnp, tdvp, tvpp, tcnp, cr);
	+ VOP_UNLOCK1(sdvp);
	+ VOP_UNLOCK1(*svpp);
	+out:
	+ if (*tvpp != NULL)
	+ VOP_UNLOCK1(*tvpp);
	+ if (tdvp != *tvpp)
	+ VOP_UNLOCK1(tdvp);
	+
	+ return (error);
	+}
	+
	+static int
	+zfs_do_rename_impl(vnode_t sdvp, vnode_t svpp, struct componentname scnp,
	+ vnode_t tdvp, vnode_t tvpp, struct componentname tcnp,
	+ cred_t *cr)
	+{
	+ dmu_tx_t *tx;
	+ zfsvfs_t *zfsvfs;
	+ zilog_t *zilog;
	+ znode_t tdzp, sdzp, tzp, szp;
	+ const char *snm = scnp->cn_nameptr;
	+ const char *tnm = tcnp->cn_nameptr;
	+ int error;
	+
	tdzp = VTOZ(tdvp);
	sdzp = VTOZ(sdvp);
	zfsvfs = tdzp->z_zfsvfs;
	- zilog = zfsvfs->z_log;

	- /*
	- * After we re-enter ZFS_ENTER() we will have to revalidate all
	- * znodes involved.
	- */
	ZFS_ENTER(zfsvfs);
	+ ZFS_VERIFY_ZP(tdzp);
	+ ZFS_VERIFY_ZP(sdzp);
	+ zilog = zfsvfs->z_log;

	if (zfsvfs->z_utf8 && u8_validate(tnm,
	strlen(tnm), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	error = SET_ERROR(EILSEQ);
	- goto unlockout;
	+ goto out;
	}

	/* If source and target are the same file, there is nothing to do. */
	if ((svpp) == (tvpp)) {
	error = 0;
	- goto unlockout;
	+ goto out;
	}

	if (((svpp)->v_type == VDIR && (svpp)->v_mountedhere != NULL) \|\|
	((tvpp) != NULL && (tvpp)->v_type == VDIR &&
	(*tvpp)->v_mountedhere != NULL)) {
	error = SET_ERROR(EXDEV);
	- goto unlockout;
	- }
	-
	- /*
	- * We can not use ZFS_VERIFY_ZP() here because it could directly return
	- * bypassing the cleanup code in the case of an error.
	- */
	- if (tdzp->z_sa_hdl == NULL \|\| sdzp->z_sa_hdl == NULL) {
	- error = SET_ERROR(EIO);
	- goto unlockout;
	+ goto out;
	}

	szp = VTOZ(*svpp);
	+ ZFS_VERIFY_ZP(szp);
	tzp = tvpp == NULL ? NULL : VTOZ(tvpp);
	- if (szp->z_sa_hdl == NULL \|\| (tzp != NULL && tzp->z_sa_hdl == NULL)) {
	- error = SET_ERROR(EIO);
	- goto unlockout;
	- }
	+ if (tzp != NULL)
	+ ZFS_VERIFY_ZP(tzp);

	/*
	* This is to prevent the creation of links into attribute space
	* by renaming a linked file into/outof an attribute directory.
	* See the comment in zfs_link() for why this is considered bad.
	*/
	if ((tdzp->z_pflags & ZFS_XATTR) != (sdzp->z_pflags & ZFS_XATTR)) {
	error = SET_ERROR(EINVAL);
	- goto unlockout;
	+ goto out;
	}

	/*
	* If we are using project inheritance, means if the directory has
	* ZFS_PROJINHERIT set, then its descendant directories will inherit
	* not only the project ID, but also the ZFS_PROJINHERIT flag. Under
	* such case, we only allow renames into our tree when the project
	* IDs are the same.
	*/
	if (tdzp->z_pflags & ZFS_PROJINHERIT &&
	tdzp->z_projid != szp->z_projid) {
	error = SET_ERROR(EXDEV);
	- goto unlockout;
	+ goto out;
	}

	/*
	* Must have write access at the source to remove the old entry
	* and write access at the target to create the new entry.
	* Note that if target and source are the same, this can be
	* done in a single check.
	*/
	if ((error = zfs_zaccess_rename(sdzp, szp, tdzp, tzp, cr)))
	- goto unlockout;
	+ goto out;

	if ((*svpp)->v_type == VDIR) {
	/*
	* Avoid ".", "..", and aliases of "." for obvious reasons.
	*/
	if ((scnp->cn_namelen == 1 && scnp->cn_nameptr[0] == '.') \|\|
	sdzp == szp \|\|
	(scnp->cn_flags \| tcnp->cn_flags) & ISDOTDOT) {
	error = EINVAL;
	- goto unlockout;
	+ goto out;
	}

	/*
	* Check to make sure rename is valid.
	* Can't do a move like this: /usr/a/b to /usr/a/b/c/d
	*/
	if ((error = zfs_rename_check(szp, sdzp, tdzp)))
	- goto unlockout;
	+ goto out;
	}

	/*
	* Does target exist?
	*/
	if (tzp) {
	/*
	* Source and target must be the same type.
	*/
	if ((*svpp)->v_type == VDIR) {
	if ((*tvpp)->v_type != VDIR) {
	error = SET_ERROR(ENOTDIR);
	- goto unlockout;
	+ goto out;
	} else {
	cache_purge(tdvp);
	if (sdvp != tdvp)
	cache_purge(sdvp);
	}
	} else {
	if ((*tvpp)->v_type == VDIR) {
	error = SET_ERROR(EISDIR);
	- goto unlockout;
	+ goto out;
	}
	}
	}

	vn_seqc_write_begin(*svpp);
	vn_seqc_write_begin(sdvp);
	if (*tvpp != NULL)
	vn_seqc_write_begin(*tvpp);
	if (tdvp != *tvpp)
	vn_seqc_write_begin(tdvp);
	-#if __FreeBSD_version >= 1300102
	- want_seqc_end = true;
	-#endif
	+
	vnevent_rename_src(*svpp, sdvp, scnp->cn_nameptr, ct);
	if (tzp)
	vnevent_rename_dest(*tvpp, tdvp, tnm, ct);

	/*
	* notify the target directory if it is not the same
	* as source directory.
	*/
	if (tdvp != sdvp) {
	vnevent_rename_dest_dir(tdvp, ct);
	}

	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE);
	dmu_tx_hold_sa(tx, sdzp->z_sa_hdl, B_FALSE);
	dmu_tx_hold_zap(tx, sdzp->z_id, FALSE, snm);
	dmu_tx_hold_zap(tx, tdzp->z_id, TRUE, tnm);
	if (sdzp != tdzp) {
	dmu_tx_hold_sa(tx, tdzp->z_sa_hdl, B_FALSE);
	zfs_sa_upgrade_txholds(tx, tdzp);
	}
	if (tzp) {
	dmu_tx_hold_sa(tx, tzp->z_sa_hdl, B_FALSE);
	zfs_sa_upgrade_txholds(tx, tzp);
	}

	zfs_sa_upgrade_txholds(tx, szp);
	dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	dmu_tx_abort(tx);
	- goto unlockout;
	+ goto out_seq;
	}

	-
	if (tzp) /* Attempt to remove the existing target */
	error = zfs_link_destroy(tdzp, tnm, tzp, tx, 0, NULL);

	if (error == 0) {
	error = zfs_link_create(tdzp, tnm, szp, tx, ZRENAMING);
	if (error == 0) {
	szp->z_pflags \|= ZFS_AV_MODIFIED;

	error = sa_update(szp->z_sa_hdl, SA_ZPL_FLAGS(zfsvfs),
	(void *)&szp->z_pflags, sizeof (uint64_t), tx);
	ASSERT0(error);

	error = zfs_link_destroy(sdzp, snm, szp, tx, ZRENAMING,
	NULL);
	if (error == 0) {
	zfs_log_rename(zilog, tx, TX_RENAME, sdzp,
	snm, tdzp, tnm, szp);

	/*
	* Update path information for the target vnode
	*/
	vn_renamepath(tdvp, *svpp, tnm, strlen(tnm));
	} else {
	/*
	* At this point, we have successfully created
	* the target name, but have failed to remove
	* the source name. Since the create was done
	* with the ZRENAMING flag, there are
	* complications; for one, the link count is
	* wrong. The easiest way to deal with this
	* is to remove the newly created target, and
	* return the original error. This must
	* succeed; fortunately, it is very unlikely to
	* fail, since we just created it.
	*/
	VERIFY0(zfs_link_destroy(tdzp, tnm, szp, tx,
	ZRENAMING, NULL));
	}
	}
	if (error == 0) {
	cache_vop_rename(sdvp, svpp, tdvp, tvpp, scnp, tcnp);
	}
	}

	dmu_tx_commit(tx);

	-unlockout: /* all 4 vnodes are locked, ZFS_ENTER called */
	- if (want_seqc_end) {
	- vn_seqc_write_end(*svpp);
	- vn_seqc_write_end(sdvp);
	- if (*tvpp != NULL)
	- vn_seqc_write_end(*tvpp);
	- if (tdvp != *tvpp)
	- vn_seqc_write_end(tdvp);
	- want_seqc_end = false;
	- }
	- VOP_UNLOCK1(*svpp);
	- VOP_UNLOCK1(sdvp);
	+out_seq:
	+ vn_seqc_write_end(*svpp);
	+ vn_seqc_write_end(sdvp);
	+ if (*tvpp != NULL)
	+ vn_seqc_write_end(*tvpp);
	+ if (tdvp != *tvpp)
	+ vn_seqc_write_end(tdvp);

	+out:
	if (error == 0 && zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);
	ZFS_EXIT(zfsvfs);

	-out: /* original two vnodes are locked */
	- MPASS(!want_seqc_end);
	-
	- if (*tvpp != NULL)
	- VOP_UNLOCK1(*tvpp);
	- if (tdvp != *tvpp)
	- VOP_UNLOCK1(tdvp);
	return (error);
	}

	int
	zfs_rename(znode_t sdzp, const char sname, znode_t tdzp, const char tname,
	cred_t *cr, int flags)
	{
	struct componentname scn, tcn;
	vnode_t sdvp, tdvp;
	vnode_t svp, tvp;
	int error;
	svp = tvp = NULL;

	sdvp = ZTOV(sdzp);
	tdvp = ZTOV(tdzp);
	error = zfs_lookup_internal(sdzp, sname, &svp, &scn, DELETE);
	if (sdzp->z_zfsvfs->z_replay == B_FALSE)
	VOP_UNLOCK1(sdvp);
	if (error != 0)
	goto fail;
	VOP_UNLOCK1(svp);

	vn_lock(tdvp, LK_EXCLUSIVE \| LK_RETRY);
	error = zfs_lookup_internal(tdzp, tname, &tvp, &tcn, RENAME);
	if (error == EJUSTRETURN)
	tvp = NULL;
	else if (error != 0) {
	VOP_UNLOCK1(tdvp);
	goto fail;
	}

	- error = zfs_rename_(sdvp, &svp, &scn, tdvp, &tvp, &tcn, cr, 0);
	+ error = zfs_do_rename(sdvp, &svp, &scn, tdvp, &tvp, &tcn, cr);
	fail:
	if (svp != NULL)
	vrele(svp);
	if (tvp != NULL)
	vrele(tvp);

	return (error);
	}

	/*
	* Insert the indicated symbolic reference entry into the directory.
	*
	* IN: dvp - Directory to contain new symbolic link.
	* link - Name for new symlink entry.
	* vap - Attributes of new entry.
	* cr - credentials of caller.
	* ct - caller context
	* flags - case flags
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* dvp - ctime\|mtime updated
	*/
	/ARGSUSED/
	int
	zfs_symlink(znode_t dzp, const char name, vattr_t *vap,
	const char link, znode_t zpp, cred_t cr, int flags)
	{
	znode_t *zp;
	dmu_tx_t *tx;
	zfsvfs_t *zfsvfs = dzp->z_zfsvfs;
	zilog_t *zilog;
	uint64_t len = strlen(link);
	int error;
	zfs_acl_ids_t acl_ids;
	boolean_t fuid_dirtied;
	uint64_t txtype = TX_SYMLINK;

	ASSERT3S(vap->va_type, ==, VLNK);

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(dzp);
	zilog = zfsvfs->z_log;

	if (zfsvfs->z_utf8 && u8_validate(name, strlen(name),
	NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EILSEQ));
	}

	if (len > MAXPATHLEN) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(ENAMETOOLONG));
	}

	if ((error = zfs_acl_ids_create(dzp, 0,
	vap, cr, NULL, &acl_ids)) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Attempt to lock directory; fail if entry already exists.
	*/
	error = zfs_dirent_lookup(dzp, name, &zp, ZNEW);
	if (error) {
	zfs_acl_ids_free(&acl_ids);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
	zfs_acl_ids_free(&acl_ids);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if (zfs_acl_ids_overquota(zfsvfs, &acl_ids,
	0 /* projid */)) {
	zfs_acl_ids_free(&acl_ids);
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EDQUOT));
	}

	getnewvnode_reserve_();
	tx = dmu_tx_create(zfsvfs->z_os);
	fuid_dirtied = zfsvfs->z_fuid_dirty;
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, MAX(1, len));
	dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
	dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
	ZFS_SA_BASE_ATTR_SIZE + len);
	dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
	if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
	acl_ids.z_aclp->z_acl_bytes);
	}
	if (fuid_dirtied)
	zfs_fuid_txhold(zfsvfs, tx);
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	zfs_acl_ids_free(&acl_ids);
	dmu_tx_abort(tx);
	getnewvnode_drop_reserve();
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Create a new object for the symlink.
	* for version 4 ZPL datasets the symlink will be an SA attribute
	*/
	zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);

	if (fuid_dirtied)
	zfs_fuid_sync(zfsvfs, tx);

	if (zp->z_is_sa)
	error = sa_update(zp->z_sa_hdl, SA_ZPL_SYMLINK(zfsvfs),
	__DECONST(void *, link), len, tx);
	else
	zfs_sa_symlink(zp, __DECONST(char *, link), len, tx);

	zp->z_size = len;
	(void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs),
	&zp->z_size, sizeof (zp->z_size), tx);
	/*
	* Insert the new object into the directory.
	*/
	(void) zfs_link_create(dzp, name, zp, tx, ZNEW);

	zfs_log_symlink(zilog, tx, txtype, dzp, zp, name, link);
	*zpp = zp;

	zfs_acl_ids_free(&acl_ids);

	dmu_tx_commit(tx);

	getnewvnode_drop_reserve();

	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Return, in the buffer contained in the provided uio structure,
	* the symbolic path referred to by vp.
	*
	* IN: vp - vnode of symbolic link.
	* uio - structure to contain the link path.
	* cr - credentials of caller.
	* ct - caller context
	*
	* OUT: uio - structure containing the link path.
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* vp - atime updated
	*/
	/* ARGSUSED */
	static int
	zfs_readlink(vnode_t vp, zfs_uio_t uio, cred_t cr, caller_context_t ct)
	{
	znode_t *zp = VTOZ(vp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
	int error;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	if (zp->z_is_sa)
	error = sa_lookup_uio(zp->z_sa_hdl,
	SA_ZPL_SYMLINK(zfsvfs), uio);
	else
	error = zfs_sa_readlink(zp, uio);

	ZFS_ACCESSTIME_STAMP(zfsvfs, zp);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Insert a new entry into directory tdvp referencing svp.
	*
	* IN: tdvp - Directory to contain new entry.
	* svp - vnode of new entry.
	* name - name of new entry.
	* cr - credentials of caller.
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* tdvp - ctime\|mtime updated
	* svp - ctime updated
	*/
	/* ARGSUSED */
	int
	zfs_link(znode_t tdzp, znode_t szp, const char name, cred_t cr,
	int flags)
	{
	znode_t *tzp;
	zfsvfs_t *zfsvfs = tdzp->z_zfsvfs;
	zilog_t *zilog;
	dmu_tx_t *tx;
	int error;
	uint64_t parent;
	uid_t owner;

	ASSERT3S(ZTOV(tdzp)->v_type, ==, VDIR);

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(tdzp);
	zilog = zfsvfs->z_log;

	/*
	* POSIX dictates that we return EPERM here.
	* Better choices include ENOTSUP or EISDIR.
	*/
	if (ZTOV(szp)->v_type == VDIR) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	ZFS_VERIFY_ZP(szp);

	/*
	* If we are using project inheritance, means if the directory has
	* ZFS_PROJINHERIT set, then its descendant directories will inherit
	* not only the project ID, but also the ZFS_PROJINHERIT flag. Under
	* such case, we only allow hard link creation in our tree when the
	* project IDs are the same.
	*/
	if (tdzp->z_pflags & ZFS_PROJINHERIT &&
	tdzp->z_projid != szp->z_projid) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EXDEV));
	}

	if (szp->z_pflags & (ZFS_APPENDONLY \|
	ZFS_IMMUTABLE \| ZFS_READONLY)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	/* Prevent links to .zfs/shares files */

	if ((error = sa_lookup(szp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
	&parent, sizeof (uint64_t))) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	if (parent == zfsvfs->z_shares_dir) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	if (zfsvfs->z_utf8 && u8_validate(name,
	strlen(name), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EILSEQ));
	}

	/*
	* We do not support links between attributes and non-attributes
	* because of the potential security risk of creating links
	* into "normal" file space in order to circumvent restrictions
	* imposed in attribute space.
	*/
	if ((szp->z_pflags & ZFS_XATTR) != (tdzp->z_pflags & ZFS_XATTR)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}


	owner = zfs_fuid_map_id(zfsvfs, szp->z_uid, cr, ZFS_OWNER);
	if (owner != crgetuid(cr) && secpolicy_basic_link(ZTOV(szp), cr) != 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	if ((error = zfs_zaccess(tdzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Attempt to lock directory; fail if entry already exists.
	*/
	error = zfs_dirent_lookup(tdzp, name, &tzp, ZNEW);
	if (error) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE);
	dmu_tx_hold_zap(tx, tdzp->z_id, TRUE, name);
	zfs_sa_upgrade_txholds(tx, szp);
	zfs_sa_upgrade_txholds(tx, tdzp);
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	dmu_tx_abort(tx);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	error = zfs_link_create(tdzp, name, szp, tx, 0);

	if (error == 0) {
	uint64_t txtype = TX_LINK;
	zfs_log_link(zilog, tx, txtype, tdzp, szp, name);
	}

	dmu_tx_commit(tx);

	if (error == 0) {
	vnevent_link(ZTOV(szp), ct);
	}

	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Free or allocate space in a file. Currently, this function only
	* supports the `F_FREESP' command. However, this command is somewhat
	* misnamed, as its functionality includes the ability to allocate as
	* well as free space.
	*
	* IN: ip - inode of file to free data in.
	* cmd - action to take (only F_FREESP supported).
	* bfp - section of file to free/alloc.
	* flag - current file open mode flags.
	* offset - current file offset.
	* cr - credentials of caller.
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* ip - ctime\|mtime updated
	*/
	/* ARGSUSED */
	int
	zfs_space(znode_t zp, int cmd, flock64_t bfp, int flag,
	offset_t offset, cred_t *cr)
	{
	zfsvfs_t *zfsvfs = ZTOZSB(zp);
	uint64_t off, len;
	int error;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	if (cmd != F_FREESP) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	/*
	* Callers might not be able to detect properly that we are read-only,
	* so check it explicitly here.
	*/
	if (zfs_is_readonly(zfsvfs)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EROFS));
	}

	if (bfp->l_len < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	/*
	* Permissions aren't checked on Solaris because on this OS
	* zfs_space() can only be called with an opened file handle.
	* On Linux we can get here through truncate_range() which
	* operates directly on inodes, so we need to check access rights.
	*/
	if ((error = zfs_zaccess(zp, ACE_WRITE_DATA, 0, B_FALSE, cr))) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	off = bfp->l_start;
	len = bfp->l_len; /* 0 means from off to end of file */

	error = zfs_freesp(zp, off, len, flag, TRUE);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/ARGSUSED/
	static void
	zfs_inactive(vnode_t vp, cred_t cr, caller_context_t *ct)
	{
	znode_t *zp = VTOZ(vp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
	int error;

	ZFS_TEARDOWN_INACTIVE_ENTER_READ(zfsvfs);
	if (zp->z_sa_hdl == NULL) {
	/*
	* The fs has been unmounted, or we did a
	* suspend/resume and this file no longer exists.
	*/
	ZFS_TEARDOWN_INACTIVE_EXIT_READ(zfsvfs);
	vrecycle(vp);
	return;
	}

	if (zp->z_unlinked) {
	/*
	* Fast path to recycle a vnode of a removed file.
	*/
	ZFS_TEARDOWN_INACTIVE_EXIT_READ(zfsvfs);
	vrecycle(vp);
	return;
	}

	if (zp->z_atime_dirty && zp->z_unlinked == 0) {
	dmu_tx_t *tx = dmu_tx_create(zfsvfs->z_os);

	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
	zfs_sa_upgrade_txholds(tx, zp);
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	dmu_tx_abort(tx);
	} else {
	(void) sa_update(zp->z_sa_hdl, SA_ZPL_ATIME(zfsvfs),
	(void *)&zp->z_atime, sizeof (zp->z_atime), tx);
	zp->z_atime_dirty = 0;
	dmu_tx_commit(tx);
	}
	}
	ZFS_TEARDOWN_INACTIVE_EXIT_READ(zfsvfs);
	}


	CTASSERT(sizeof (struct zfid_short) <= sizeof (struct fid));
	CTASSERT(sizeof (struct zfid_long) <= sizeof (struct fid));

	/ARGSUSED/
	static int
	zfs_fid(vnode_t vp, fid_t fidp, caller_context_t *ct)
	{
	znode_t *zp = VTOZ(vp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
	uint32_t gen;
	uint64_t gen64;
	uint64_t object = zp->z_id;
	zfid_short_t *zfid;
	int size, i, error;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zfsvfs),
	&gen64, sizeof (uint64_t))) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	gen = (uint32_t)gen64;

	size = (zfsvfs->z_parent != zfsvfs) ? LONG_FID_LEN : SHORT_FID_LEN;
	fidp->fid_len = size;

	zfid = (zfid_short_t *)fidp;

	zfid->zf_len = size;

	for (i = 0; i < sizeof (zfid->zf_object); i++)
	zfid->zf_object[i] = (uint8_t)(object >> (8 * i));

	/* Must have a non-zero generation number to distinguish from .zfs */
	if (gen == 0)
	gen = 1;
	for (i = 0; i < sizeof (zfid->zf_gen); i++)
	zfid->zf_gen[i] = (uint8_t)(gen >> (8 * i));

	if (size == LONG_FID_LEN) {
	uint64_t objsetid = dmu_objset_id(zfsvfs->z_os);
	zfid_long_t *zlfid;

	zlfid = (zfid_long_t *)fidp;

	for (i = 0; i < sizeof (zlfid->zf_setid); i++)
	zlfid->zf_setid[i] = (uint8_t)(objsetid >> (8 * i));

	/* XXX - this should be the generation number for the objset */
	for (i = 0; i < sizeof (zlfid->zf_setgen); i++)
	zlfid->zf_setgen[i] = 0;
	}

	ZFS_EXIT(zfsvfs);
	return (0);
	}

	static int
	zfs_pathconf(vnode_t vp, int cmd, ulong_t valp, cred_t *cr,
	caller_context_t *ct)
	{
	znode_t *zp;
	zfsvfs_t *zfsvfs;

	switch (cmd) {
	case _PC_LINK_MAX:
	*valp = MIN(LONG_MAX, ZFS_LINK_MAX);
	return (0);

	case _PC_FILESIZEBITS:
	*valp = 64;
	return (0);
	case _PC_MIN_HOLE_SIZE:
	*valp = (int)SPA_MINBLOCKSIZE;
	return (0);
	case _PC_ACL_EXTENDED:
	#if 0 /* POSIX ACLs are not implemented for ZFS on FreeBSD yet. */
	zp = VTOZ(vp);
	zfsvfs = zp->z_zfsvfs;
	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);
	*valp = zfsvfs->z_acl_type == ZFSACLTYPE_POSIX ? 1 : 0;
	ZFS_EXIT(zfsvfs);
	#else
	*valp = 0;
	#endif
	return (0);

	case _PC_ACL_NFS4:
	zp = VTOZ(vp);
	zfsvfs = zp->z_zfsvfs;
	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);
	*valp = zfsvfs->z_acl_type == ZFS_ACLTYPE_NFSV4 ? 1 : 0;
	ZFS_EXIT(zfsvfs);
	return (0);

	case _PC_ACL_PATH_MAX:
	*valp = ACL_MAX_ENTRIES;
	return (0);

	default:
	return (EOPNOTSUPP);
	}
	}

	static int
	zfs_getpages(struct vnode vp, vm_page_t ma, int count, int *rbehind,
	int *rahead)
	{
	znode_t *zp = VTOZ(vp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
	zfs_locked_range_t *lr;
	vm_object_t object;
	off_t start, end, obj_size;
	uint_t blksz;
	int pgsin_b, pgsin_a;
	int error;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	start = IDX_TO_OFF(ma[0]->pindex);
	end = IDX_TO_OFF(ma[count - 1]->pindex + 1);

	/*
	* Lock a range covering all required and optional pages.
	* Note that we need to handle the case of the block size growing.
	*/
	for (;;) {
	blksz = zp->z_blksz;
	lr = zfs_rangelock_tryenter(&zp->z_rangelock,
	rounddown(start, blksz),
	roundup(end, blksz) - rounddown(start, blksz), RL_READER);
	if (lr == NULL) {
	if (rahead != NULL) {
	*rahead = 0;
	rahead = NULL;
	}
	if (rbehind != NULL) {
	*rbehind = 0;
	rbehind = NULL;
	}
	break;
	}
	if (blksz == zp->z_blksz)
	break;
	zfs_rangelock_exit(lr);
	}

	object = ma[0]->object;
	zfs_vmobject_wlock(object);
	obj_size = object->un_pager.vnp.vnp_size;
	zfs_vmobject_wunlock(object);
	if (IDX_TO_OFF(ma[count - 1]->pindex) >= obj_size) {
	if (lr != NULL)
	zfs_rangelock_exit(lr);
	ZFS_EXIT(zfsvfs);
	return (zfs_vm_pagerret_bad);
	}

	pgsin_b = 0;
	if (rbehind != NULL) {
	pgsin_b = OFF_TO_IDX(start - rounddown(start, blksz));
	pgsin_b = MIN(*rbehind, pgsin_b);
	}

	pgsin_a = 0;
	if (rahead != NULL) {
	pgsin_a = OFF_TO_IDX(roundup(end, blksz) - end);
	if (end + IDX_TO_OFF(pgsin_a) >= obj_size)
	pgsin_a = OFF_TO_IDX(round_page(obj_size) - end);
	pgsin_a = MIN(*rahead, pgsin_a);
	}

	/*
	* NB: we need to pass the exact byte size of the data that we expect
	* to read after accounting for the file size. This is required because
	* ZFS will panic if we request DMU to read beyond the end of the last
	* allocated block.
	*/
	error = dmu_read_pages(zfsvfs->z_os, zp->z_id, ma, count, &pgsin_b,
	&pgsin_a, MIN(end, obj_size) - (end - PAGE_SIZE));

	if (lr != NULL)
	zfs_rangelock_exit(lr);
	ZFS_ACCESSTIME_STAMP(zfsvfs, zp);
	ZFS_EXIT(zfsvfs);

	if (error != 0)
	return (zfs_vm_pagerret_error);

	VM_CNT_INC(v_vnodein);
	VM_CNT_ADD(v_vnodepgsin, count + pgsin_b + pgsin_a);
	if (rbehind != NULL)
	*rbehind = pgsin_b;
	if (rahead != NULL)
	*rahead = pgsin_a;
	return (zfs_vm_pagerret_ok);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_getpages_args {
	struct vnode *a_vp;
	vm_page_t *a_m;
	int a_count;
	int *a_rbehind;
	int *a_rahead;
	};
	#endif

	static int
	zfs_freebsd_getpages(struct vop_getpages_args *ap)
	{

	return (zfs_getpages(ap->a_vp, ap->a_m, ap->a_count, ap->a_rbehind,
	ap->a_rahead));
	}

	static int
	zfs_putpages(struct vnode vp, vm_page_t ma, size_t len, int flags,
	int *rtvals)
	{
	znode_t *zp = VTOZ(vp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
	zfs_locked_range_t *lr;
	dmu_tx_t *tx;
	struct sf_buf *sf;
	vm_object_t object;
	vm_page_t m;
	caddr_t va;
	size_t tocopy;
	size_t lo_len;
	vm_ooffset_t lo_off;
	vm_ooffset_t off;
	uint_t blksz;
	int ncount;
	int pcount;
	int err;
	int i;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	object = vp->v_object;
	pcount = btoc(len);
	ncount = pcount;

	KASSERT(ma[0]->object == object, ("mismatching object"));
	KASSERT(len > 0 && (len & PAGE_MASK) == 0, ("unexpected length"));

	for (i = 0; i < pcount; i++)
	rtvals[i] = zfs_vm_pagerret_error;

	off = IDX_TO_OFF(ma[0]->pindex);
	blksz = zp->z_blksz;
	lo_off = rounddown(off, blksz);
	lo_len = roundup(len + (off - lo_off), blksz);
	lr = zfs_rangelock_enter(&zp->z_rangelock, lo_off, lo_len, RL_WRITER);

	zfs_vmobject_wlock(object);
	if (len + off > object->un_pager.vnp.vnp_size) {
	if (object->un_pager.vnp.vnp_size > off) {
	int pgoff;

	len = object->un_pager.vnp.vnp_size - off;
	ncount = btoc(len);
	if ((pgoff = (int)len & PAGE_MASK) != 0) {
	/*
	* If the object is locked and the following
	* conditions hold, then the page's dirty
	* field cannot be concurrently changed by a
	* pmap operation.
	*/
	m = ma[ncount - 1];
	vm_page_assert_sbusied(m);
	KASSERT(!pmap_page_is_write_mapped(m),
	("zfs_putpages: page %p is not read-only",
	m));
	vm_page_clear_dirty(m, pgoff, PAGE_SIZE -
	pgoff);
	}
	} else {
	len = 0;
	ncount = 0;
	}
	if (ncount < pcount) {
	for (i = ncount; i < pcount; i++) {
	rtvals[i] = zfs_vm_pagerret_bad;
	}
	}
	}
	zfs_vmobject_wunlock(object);

	if (ncount == 0)
	goto out;

	if (zfs_id_overblockquota(zfsvfs, DMU_USERUSED_OBJECT, zp->z_uid) \|\|
	zfs_id_overblockquota(zfsvfs, DMU_GROUPUSED_OBJECT, zp->z_gid) \|\|
	(zp->z_projid != ZFS_DEFAULT_PROJID &&
	zfs_id_overblockquota(zfsvfs, DMU_PROJECTUSED_OBJECT,
	zp->z_projid))) {
	goto out;
	}

	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_write(tx, zp->z_id, off, len);

	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
	zfs_sa_upgrade_txholds(tx, zp);
	err = dmu_tx_assign(tx, TXG_WAIT);
	if (err != 0) {
	dmu_tx_abort(tx);
	goto out;
	}

	if (zp->z_blksz < PAGE_SIZE) {
	for (i = 0; len > 0; off += tocopy, len -= tocopy, i++) {
	tocopy = len > PAGE_SIZE ? PAGE_SIZE : len;
	va = zfs_map_page(ma[i], &sf);
	dmu_write(zfsvfs->z_os, zp->z_id, off, tocopy, va, tx);
	zfs_unmap_page(sf);
	}
	} else {
	err = dmu_write_pages(zfsvfs->z_os, zp->z_id, off, len, ma, tx);
	}

	if (err == 0) {
	uint64_t mtime[2], ctime[2];
	sa_bulk_attr_t bulk[3];
	int count = 0;

	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL,
	&mtime, 16);
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
	&ctime, 16);
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
	&zp->z_pflags, 8);
	zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime);
	err = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
	ASSERT0(err);
	/*
	* XXX we should be passing a callback to undirty
	* but that would make the locking messier
	*/
	zfs_log_write(zfsvfs->z_log, tx, TX_WRITE, zp, off,
	len, 0, NULL, NULL);

	zfs_vmobject_wlock(object);
	for (i = 0; i < ncount; i++) {
	rtvals[i] = zfs_vm_pagerret_ok;
	vm_page_undirty(ma[i]);
	}
	zfs_vmobject_wunlock(object);
	VM_CNT_INC(v_vnodeout);
	VM_CNT_ADD(v_vnodepgsout, ncount);
	}
	dmu_tx_commit(tx);

	out:
	zfs_rangelock_exit(lr);
	if ((flags & (zfs_vm_pagerput_sync \| zfs_vm_pagerput_inval)) != 0 \|\|
	zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zfsvfs->z_log, zp->z_id);
	ZFS_EXIT(zfsvfs);
	return (rtvals[0]);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_putpages_args {
	struct vnode *a_vp;
	vm_page_t *a_m;
	int a_count;
	int a_sync;
	int *a_rtvals;
	};
	#endif

	static int
	zfs_freebsd_putpages(struct vop_putpages_args *ap)
	{

	return (zfs_putpages(ap->a_vp, ap->a_m, ap->a_count, ap->a_sync,
	ap->a_rtvals));
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_bmap_args {
	struct vnode *a_vp;
	daddr_t a_bn;
	struct bufobj **a_bop;
	daddr_t *a_bnp;
	int *a_runp;
	int *a_runb;
	};
	#endif

	static int
	zfs_freebsd_bmap(struct vop_bmap_args *ap)
	{

	if (ap->a_bop != NULL)
	*ap->a_bop = &ap->a_vp->v_bufobj;
	if (ap->a_bnp != NULL)
	*ap->a_bnp = ap->a_bn;
	if (ap->a_runp != NULL)
	*ap->a_runp = 0;
	if (ap->a_runb != NULL)
	*ap->a_runb = 0;

	return (0);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_open_args {
	struct vnode *a_vp;
	int a_mode;
	struct ucred *a_cred;
	struct thread *a_td;
	};
	#endif

	static int
	zfs_freebsd_open(struct vop_open_args *ap)
	{
	vnode_t *vp = ap->a_vp;
	znode_t *zp = VTOZ(vp);
	int error;

	error = zfs_open(&vp, ap->a_mode, ap->a_cred);
	if (error == 0)
	vnode_create_vobject(vp, zp->z_size, ap->a_td);
	return (error);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_close_args {
	struct vnode *a_vp;
	int a_fflag;
	struct ucred *a_cred;
	struct thread *a_td;
	};
	#endif

	static int
	zfs_freebsd_close(struct vop_close_args *ap)
	{

	return (zfs_close(ap->a_vp, ap->a_fflag, 1, 0, ap->a_cred));
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_ioctl_args {
	struct vnode *a_vp;
	ulong_t a_command;
	caddr_t a_data;
	int a_fflag;
	struct ucred *cred;
	struct thread *td;
	};
	#endif

	static int
	zfs_freebsd_ioctl(struct vop_ioctl_args *ap)
	{

	return (zfs_ioctl(ap->a_vp, ap->a_command, (intptr_t)ap->a_data,
	ap->a_fflag, ap->a_cred, NULL));
	}

	static int
	ioflags(int ioflags)
	{
	int flags = 0;

	if (ioflags & IO_APPEND)
	flags \|= FAPPEND;
	if (ioflags & IO_NDELAY)
	flags \|= FNONBLOCK;
	if (ioflags & IO_SYNC)
	flags \|= (FSYNC \| FDSYNC \| FRSYNC);

	return (flags);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_read_args {
	struct vnode *a_vp;
	struct uio *a_uio;
	int a_ioflag;
	struct ucred *a_cred;
	};
	#endif

	static int
	zfs_freebsd_read(struct vop_read_args *ap)
	{
	zfs_uio_t uio;
	zfs_uio_init(&uio, ap->a_uio);
	return (zfs_read(VTOZ(ap->a_vp), &uio, ioflags(ap->a_ioflag),
	ap->a_cred));
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_write_args {
	struct vnode *a_vp;
	struct uio *a_uio;
	int a_ioflag;
	struct ucred *a_cred;
	};
	#endif

	static int
	zfs_freebsd_write(struct vop_write_args *ap)
	{
	zfs_uio_t uio;
	zfs_uio_init(&uio, ap->a_uio);
	return (zfs_write(VTOZ(ap->a_vp), &uio, ioflags(ap->a_ioflag),
	ap->a_cred));
	}

	#if __FreeBSD_version >= 1300102
	/*
	* VOP_FPLOOKUP_VEXEC routines are subject to special circumstances, see
	* the comment above cache_fplookup for details.
	*/
	static int
	zfs_freebsd_fplookup_vexec(struct vop_fplookup_vexec_args *v)
	{
	vnode_t *vp;
	znode_t *zp;
	uint64_t pflags;

	vp = v->a_vp;
	zp = VTOZ_SMR(vp);
	if (__predict_false(zp == NULL))
	return (EAGAIN);
	pflags = atomic_load_64(&zp->z_pflags);
	if (pflags & ZFS_AV_QUARANTINED)
	return (EAGAIN);
	if (pflags & ZFS_XATTR)
	return (EAGAIN);
	if ((pflags & ZFS_NO_EXECS_DENIED) == 0)
	return (EAGAIN);
	return (0);
	}
	#endif

	#if __FreeBSD_version >= 1300139
	static int
	zfs_freebsd_fplookup_symlink(struct vop_fplookup_symlink_args *v)
	{
	vnode_t *vp;
	znode_t *zp;
	char *target;

	vp = v->a_vp;
	zp = VTOZ_SMR(vp);
	if (__predict_false(zp == NULL)) {
	return (EAGAIN);
	}

	target = atomic_load_consume_ptr(&zp->z_cached_symlink);
	if (target == NULL) {
	return (EAGAIN);
	}
	return (cache_symlink_resolve(v->a_fpl, target, strlen(target)));
	}
	#endif

	#ifndef _SYS_SYSPROTO_H_
	struct vop_access_args {
	struct vnode *a_vp;
	accmode_t a_accmode;
	struct ucred *a_cred;
	struct thread *a_td;
	};
	#endif

	static int
	zfs_freebsd_access(struct vop_access_args *ap)
	{
	vnode_t *vp = ap->a_vp;
	znode_t *zp = VTOZ(vp);
	accmode_t accmode;
	int error = 0;


	if (ap->a_accmode == VEXEC) {
	if (zfs_fastaccesschk_execute(zp, ap->a_cred) == 0)
	return (0);
	}

	/*
	* ZFS itself only knowns about VREAD, VWRITE, VEXEC and VAPPEND,
	*/
	accmode = ap->a_accmode & (VREAD\|VWRITE\|VEXEC\|VAPPEND);
	if (accmode != 0)
	error = zfs_access(zp, accmode, 0, ap->a_cred);

	/*
	* VADMIN has to be handled by vaccess().
	*/
	if (error == 0) {
	accmode = ap->a_accmode & ~(VREAD\|VWRITE\|VEXEC\|VAPPEND);
	if (accmode != 0) {
	#if __FreeBSD_version >= 1300105
	error = vaccess(vp->v_type, zp->z_mode, zp->z_uid,
	zp->z_gid, accmode, ap->a_cred);
	#else
	error = vaccess(vp->v_type, zp->z_mode, zp->z_uid,
	zp->z_gid, accmode, ap->a_cred, NULL);
	#endif
	}
	}

	/*
	* For VEXEC, ensure that at least one execute bit is set for
	* non-directories.
	*/
	if (error == 0 && (ap->a_accmode & VEXEC) != 0 && vp->v_type != VDIR &&
	(zp->z_mode & (S_IXUSR \| S_IXGRP \| S_IXOTH)) == 0) {
	error = EACCES;
	}

	return (error);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_lookup_args {
	struct vnode *a_dvp;
	struct vnode **a_vpp;
	struct componentname *a_cnp;
	};
	#endif

	static int
	zfs_freebsd_lookup(struct vop_lookup_args *ap, boolean_t cached)
	{
	struct componentname *cnp = ap->a_cnp;
	char nm[NAME_MAX + 1];

	ASSERT3U(cnp->cn_namelen, <, sizeof (nm));
	strlcpy(nm, cnp->cn_nameptr, MIN(cnp->cn_namelen + 1, sizeof (nm)));

	return (zfs_lookup(ap->a_dvp, nm, ap->a_vpp, cnp, cnp->cn_nameiop,
	cnp->cn_cred, 0, cached));
	}

	static int
	zfs_freebsd_cachedlookup(struct vop_cachedlookup_args *ap)
	{

	return (zfs_freebsd_lookup((struct vop_lookup_args *)ap, B_TRUE));
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_lookup_args {
	struct vnode *a_dvp;
	struct vnode **a_vpp;
	struct componentname *a_cnp;
	};
	#endif

	static int
	zfs_cache_lookup(struct vop_lookup_args *ap)
	{
	zfsvfs_t *zfsvfs;

	zfsvfs = ap->a_dvp->v_mount->mnt_data;
	if (zfsvfs->z_use_namecache)
	return (vfs_cache_lookup(ap));
	else
	return (zfs_freebsd_lookup(ap, B_FALSE));
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_create_args {
	struct vnode *a_dvp;
	struct vnode **a_vpp;
	struct componentname *a_cnp;
	struct vattr *a_vap;
	};
	#endif

	static int
	zfs_freebsd_create(struct vop_create_args *ap)
	{
	zfsvfs_t *zfsvfs;
	struct componentname *cnp = ap->a_cnp;
	vattr_t *vap = ap->a_vap;
	znode_t *zp = NULL;
	int rc, mode;

	ASSERT(cnp->cn_flags & SAVENAME);

	vattr_init_mask(vap);
	mode = vap->va_mode & ALLPERMS;
	zfsvfs = ap->a_dvp->v_mount->mnt_data;
	*ap->a_vpp = NULL;

	rc = zfs_create(VTOZ(ap->a_dvp), cnp->cn_nameptr, vap, !EXCL, mode,
	&zp, cnp->cn_cred, 0 /* flag /, NULL / vsecattr */);
	if (rc == 0)
	*ap->a_vpp = ZTOV(zp);
	if (zfsvfs->z_use_namecache &&
	rc == 0 && (cnp->cn_flags & MAKEENTRY) != 0)
	cache_enter(ap->a_dvp, *ap->a_vpp, cnp);

	return (rc);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_remove_args {
	struct vnode *a_dvp;
	struct vnode *a_vp;
	struct componentname *a_cnp;
	};
	#endif

	static int
	zfs_freebsd_remove(struct vop_remove_args *ap)
	{

	ASSERT(ap->a_cnp->cn_flags & SAVENAME);

	return (zfs_remove_(ap->a_dvp, ap->a_vp, ap->a_cnp->cn_nameptr,
	ap->a_cnp->cn_cred));
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_mkdir_args {
	struct vnode *a_dvp;
	struct vnode **a_vpp;
	struct componentname *a_cnp;
	struct vattr *a_vap;
	};
	#endif

	static int
	zfs_freebsd_mkdir(struct vop_mkdir_args *ap)
	{
	vattr_t *vap = ap->a_vap;
	znode_t *zp = NULL;
	int rc;

	ASSERT(ap->a_cnp->cn_flags & SAVENAME);

	vattr_init_mask(vap);
	*ap->a_vpp = NULL;

	rc = zfs_mkdir(VTOZ(ap->a_dvp), ap->a_cnp->cn_nameptr, vap, &zp,
	ap->a_cnp->cn_cred, 0, NULL);

	if (rc == 0)
	*ap->a_vpp = ZTOV(zp);
	return (rc);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_rmdir_args {
	struct vnode *a_dvp;
	struct vnode *a_vp;
	struct componentname *a_cnp;
	};
	#endif

	static int
	zfs_freebsd_rmdir(struct vop_rmdir_args *ap)
	{
	struct componentname *cnp = ap->a_cnp;

	ASSERT(cnp->cn_flags & SAVENAME);

	return (zfs_rmdir_(ap->a_dvp, ap->a_vp, cnp->cn_nameptr, cnp->cn_cred));
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_readdir_args {
	struct vnode *a_vp;
	struct uio *a_uio;
	struct ucred *a_cred;
	int *a_eofflag;
	int *a_ncookies;
	uint64_t **a_cookies;
	};
	#endif

	static int
	zfs_freebsd_readdir(struct vop_readdir_args *ap)
	{
	zfs_uio_t uio;
	zfs_uio_init(&uio, ap->a_uio);
	return (zfs_readdir(ap->a_vp, &uio, ap->a_cred, ap->a_eofflag,
	ap->a_ncookies, ap->a_cookies));
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_fsync_args {
	struct vnode *a_vp;
	int a_waitfor;
	struct thread *a_td;
	};
	#endif

	static int
	zfs_freebsd_fsync(struct vop_fsync_args *ap)
	{

	vop_stdfsync(ap);
	return (zfs_fsync(VTOZ(ap->a_vp), 0, ap->a_td->td_ucred));
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_getattr_args {
	struct vnode *a_vp;
	struct vattr *a_vap;
	struct ucred *a_cred;
	};
	#endif

	static int
	zfs_freebsd_getattr(struct vop_getattr_args *ap)
	{
	vattr_t *vap = ap->a_vap;
	xvattr_t xvap;
	ulong_t fflags = 0;
	int error;

	xva_init(&xvap);
	xvap.xva_vattr = *vap;
	xvap.xva_vattr.va_mask \|= AT_XVATTR;

	/* Convert chflags into ZFS-type flags. */
	/* XXX: what about SF_SETTABLE?. */
	XVA_SET_REQ(&xvap, XAT_IMMUTABLE);
	XVA_SET_REQ(&xvap, XAT_APPENDONLY);
	XVA_SET_REQ(&xvap, XAT_NOUNLINK);
	XVA_SET_REQ(&xvap, XAT_NODUMP);
	XVA_SET_REQ(&xvap, XAT_READONLY);
	XVA_SET_REQ(&xvap, XAT_ARCHIVE);
	XVA_SET_REQ(&xvap, XAT_SYSTEM);
	XVA_SET_REQ(&xvap, XAT_HIDDEN);
	XVA_SET_REQ(&xvap, XAT_REPARSE);
	XVA_SET_REQ(&xvap, XAT_OFFLINE);
	XVA_SET_REQ(&xvap, XAT_SPARSE);

	error = zfs_getattr(ap->a_vp, (vattr_t *)&xvap, 0, ap->a_cred);
	if (error != 0)
	return (error);

	/* Convert ZFS xattr into chflags. */
	#define FLAG_CHECK(fflag, xflag, xfield) do { \
	if (XVA_ISSET_RTN(&xvap, (xflag)) && (xfield) != 0) \
	fflags \|= (fflag); \
	} while (0)
	FLAG_CHECK(SF_IMMUTABLE, XAT_IMMUTABLE,
	xvap.xva_xoptattrs.xoa_immutable);
	FLAG_CHECK(SF_APPEND, XAT_APPENDONLY,
	xvap.xva_xoptattrs.xoa_appendonly);
	FLAG_CHECK(SF_NOUNLINK, XAT_NOUNLINK,
	xvap.xva_xoptattrs.xoa_nounlink);
	FLAG_CHECK(UF_ARCHIVE, XAT_ARCHIVE,
	xvap.xva_xoptattrs.xoa_archive);
	FLAG_CHECK(UF_NODUMP, XAT_NODUMP,
	xvap.xva_xoptattrs.xoa_nodump);
	FLAG_CHECK(UF_READONLY, XAT_READONLY,
	xvap.xva_xoptattrs.xoa_readonly);
	FLAG_CHECK(UF_SYSTEM, XAT_SYSTEM,
	xvap.xva_xoptattrs.xoa_system);
	FLAG_CHECK(UF_HIDDEN, XAT_HIDDEN,
	xvap.xva_xoptattrs.xoa_hidden);
	FLAG_CHECK(UF_REPARSE, XAT_REPARSE,
	xvap.xva_xoptattrs.xoa_reparse);
	FLAG_CHECK(UF_OFFLINE, XAT_OFFLINE,
	xvap.xva_xoptattrs.xoa_offline);
	FLAG_CHECK(UF_SPARSE, XAT_SPARSE,
	xvap.xva_xoptattrs.xoa_sparse);

	#undef FLAG_CHECK
	*vap = xvap.xva_vattr;
	vap->va_flags = fflags;
	return (0);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_setattr_args {
	struct vnode *a_vp;
	struct vattr *a_vap;
	struct ucred *a_cred;
	};
	#endif

	static int
	zfs_freebsd_setattr(struct vop_setattr_args *ap)
	{
	vnode_t *vp = ap->a_vp;
	vattr_t *vap = ap->a_vap;
	cred_t *cred = ap->a_cred;
	xvattr_t xvap;
	ulong_t fflags;
	uint64_t zflags;

	vattr_init_mask(vap);
	vap->va_mask &= ~AT_NOSET;

	xva_init(&xvap);
	xvap.xva_vattr = *vap;

	zflags = VTOZ(vp)->z_pflags;

	if (vap->va_flags != VNOVAL) {
	zfsvfs_t *zfsvfs = VTOZ(vp)->z_zfsvfs;
	int error;

	if (zfsvfs->z_use_fuids == B_FALSE)
	return (EOPNOTSUPP);

	fflags = vap->va_flags;
	/*
	* XXX KDM
	* We need to figure out whether it makes sense to allow
	* UF_REPARSE through, since we don't really have other
	* facilities to handle reparse points and zfs_setattr()
	* doesn't currently allow setting that attribute anyway.
	*/
	if ((fflags & ~(SF_IMMUTABLE\|SF_APPEND\|SF_NOUNLINK\|UF_ARCHIVE\|
	UF_NODUMP\|UF_SYSTEM\|UF_HIDDEN\|UF_READONLY\|UF_REPARSE\|
	UF_OFFLINE\|UF_SPARSE)) != 0)
	return (EOPNOTSUPP);
	/*
	* Unprivileged processes are not permitted to unset system
	* flags, or modify flags if any system flags are set.
	* Privileged non-jail processes may not modify system flags
	* if securelevel > 0 and any existing system flags are set.
	* Privileged jail processes behave like privileged non-jail
	* processes if the PR_ALLOW_CHFLAGS permission bit is set;
	* otherwise, they behave like unprivileged processes.
	*/
	if (secpolicy_fs_owner(vp->v_mount, cred) == 0 \|\|
	spl_priv_check_cred(cred, PRIV_VFS_SYSFLAGS) == 0) {
	if (zflags &
	(ZFS_IMMUTABLE \| ZFS_APPENDONLY \| ZFS_NOUNLINK)) {
	error = securelevel_gt(cred, 0);
	if (error != 0)
	return (error);
	}
	} else {
	/*
	* Callers may only modify the file flags on
	* objects they have VADMIN rights for.
	*/
	if ((error = VOP_ACCESS(vp, VADMIN, cred,
	curthread)) != 0)
	return (error);
	if (zflags &
	(ZFS_IMMUTABLE \| ZFS_APPENDONLY \|
	ZFS_NOUNLINK)) {
	return (EPERM);
	}
	if (fflags &
	(SF_IMMUTABLE \| SF_APPEND \| SF_NOUNLINK)) {
	return (EPERM);
	}
	}

	#define FLAG_CHANGE(fflag, zflag, xflag, xfield) do { \
	if (((fflags & (fflag)) && !(zflags & (zflag))) \|\| \
	((zflags & (zflag)) && !(fflags & (fflag)))) { \
	XVA_SET_REQ(&xvap, (xflag)); \
	(xfield) = ((fflags & (fflag)) != 0); \
	} \
	} while (0)
	/* Convert chflags into ZFS-type flags. */
	/* XXX: what about SF_SETTABLE?. */
	FLAG_CHANGE(SF_IMMUTABLE, ZFS_IMMUTABLE, XAT_IMMUTABLE,
	xvap.xva_xoptattrs.xoa_immutable);
	FLAG_CHANGE(SF_APPEND, ZFS_APPENDONLY, XAT_APPENDONLY,
	xvap.xva_xoptattrs.xoa_appendonly);
	FLAG_CHANGE(SF_NOUNLINK, ZFS_NOUNLINK, XAT_NOUNLINK,
	xvap.xva_xoptattrs.xoa_nounlink);
	FLAG_CHANGE(UF_ARCHIVE, ZFS_ARCHIVE, XAT_ARCHIVE,
	xvap.xva_xoptattrs.xoa_archive);
	FLAG_CHANGE(UF_NODUMP, ZFS_NODUMP, XAT_NODUMP,
	xvap.xva_xoptattrs.xoa_nodump);
	FLAG_CHANGE(UF_READONLY, ZFS_READONLY, XAT_READONLY,
	xvap.xva_xoptattrs.xoa_readonly);
	FLAG_CHANGE(UF_SYSTEM, ZFS_SYSTEM, XAT_SYSTEM,
	xvap.xva_xoptattrs.xoa_system);
	FLAG_CHANGE(UF_HIDDEN, ZFS_HIDDEN, XAT_HIDDEN,
	xvap.xva_xoptattrs.xoa_hidden);
	FLAG_CHANGE(UF_REPARSE, ZFS_REPARSE, XAT_REPARSE,
	xvap.xva_xoptattrs.xoa_reparse);
	FLAG_CHANGE(UF_OFFLINE, ZFS_OFFLINE, XAT_OFFLINE,
	xvap.xva_xoptattrs.xoa_offline);
	FLAG_CHANGE(UF_SPARSE, ZFS_SPARSE, XAT_SPARSE,
	xvap.xva_xoptattrs.xoa_sparse);
	#undef FLAG_CHANGE
	}
	if (vap->va_birthtime.tv_sec != VNOVAL) {
	xvap.xva_vattr.va_mask \|= AT_XVATTR;
	XVA_SET_REQ(&xvap, XAT_CREATETIME);
	}
	return (zfs_setattr(VTOZ(vp), (vattr_t *)&xvap, 0, cred));
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_rename_args {
	struct vnode *a_fdvp;
	struct vnode *a_fvp;
	struct componentname *a_fcnp;
	struct vnode *a_tdvp;
	struct vnode *a_tvp;
	struct componentname *a_tcnp;
	};
	#endif

	static int
	zfs_freebsd_rename(struct vop_rename_args *ap)
	{
	vnode_t *fdvp = ap->a_fdvp;
	vnode_t *fvp = ap->a_fvp;
	vnode_t *tdvp = ap->a_tdvp;
	vnode_t *tvp = ap->a_tvp;
	int error;

	ASSERT(ap->a_fcnp->cn_flags & (SAVENAME\|SAVESTART));
	ASSERT(ap->a_tcnp->cn_flags & (SAVENAME\|SAVESTART));

	- error = zfs_rename_(fdvp, &fvp, ap->a_fcnp, tdvp, &tvp,
	- ap->a_tcnp, ap->a_fcnp->cn_cred, 1);
	+ error = zfs_do_rename(fdvp, &fvp, ap->a_fcnp, tdvp, &tvp,
	+ ap->a_tcnp, ap->a_fcnp->cn_cred);

	vrele(fdvp);
	vrele(fvp);
	vrele(tdvp);
	if (tvp != NULL)
	vrele(tvp);

	return (error);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_symlink_args {
	struct vnode *a_dvp;
	struct vnode **a_vpp;
	struct componentname *a_cnp;
	struct vattr *a_vap;
	char *a_target;
	};
	#endif

	static int
	zfs_freebsd_symlink(struct vop_symlink_args *ap)
	{
	struct componentname *cnp = ap->a_cnp;
	vattr_t *vap = ap->a_vap;
	znode_t *zp = NULL;
	#if __FreeBSD_version >= 1300139
	char *symlink;
	size_t symlink_len;
	#endif
	int rc;

	ASSERT(cnp->cn_flags & SAVENAME);

	vap->va_type = VLNK; /* FreeBSD: Syscall only sets va_mode. */
	vattr_init_mask(vap);
	*ap->a_vpp = NULL;

	rc = zfs_symlink(VTOZ(ap->a_dvp), cnp->cn_nameptr, vap,
	ap->a_target, &zp, cnp->cn_cred, 0 /* flags */);
	if (rc == 0) {
	*ap->a_vpp = ZTOV(zp);
	ASSERT_VOP_ELOCKED(ZTOV(zp), __func__);
	#if __FreeBSD_version >= 1300139
	MPASS(zp->z_cached_symlink == NULL);
	symlink_len = strlen(ap->a_target);
	symlink = cache_symlink_alloc(symlink_len + 1, M_WAITOK);
	if (symlink != NULL) {
	memcpy(symlink, ap->a_target, symlink_len);
	symlink[symlink_len] = '\0';
	atomic_store_rel_ptr((uintptr_t *)&zp->z_cached_symlink,
	(uintptr_t)symlink);
	}
	#endif
	}
	return (rc);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_readlink_args {
	struct vnode *a_vp;
	struct uio *a_uio;
	struct ucred *a_cred;
	};
	#endif

	static int
	zfs_freebsd_readlink(struct vop_readlink_args *ap)
	{
	zfs_uio_t uio;
	int error;
	#if __FreeBSD_version >= 1300139
	znode_t *zp = VTOZ(ap->a_vp);
	char symlink, base;
	size_t symlink_len;
	bool trycache;
	#endif

	zfs_uio_init(&uio, ap->a_uio);
	#if __FreeBSD_version >= 1300139
	trycache = false;
	if (zfs_uio_segflg(&uio) == UIO_SYSSPACE &&
	zfs_uio_iovcnt(&uio) == 1) {
	base = zfs_uio_iovbase(&uio, 0);
	symlink_len = zfs_uio_iovlen(&uio, 0);
	trycache = true;
	}
	#endif
	error = zfs_readlink(ap->a_vp, &uio, ap->a_cred, NULL);
	#if __FreeBSD_version >= 1300139
	if (atomic_load_ptr(&zp->z_cached_symlink) != NULL \|\|
	error != 0 \|\| !trycache) {
	return (error);
	}
	symlink_len -= zfs_uio_resid(&uio);
	symlink = cache_symlink_alloc(symlink_len + 1, M_WAITOK);
	if (symlink != NULL) {
	memcpy(symlink, base, symlink_len);
	symlink[symlink_len] = '\0';
	if (!atomic_cmpset_rel_ptr((uintptr_t *)&zp->z_cached_symlink,
	(uintptr_t)NULL, (uintptr_t)symlink)) {
	cache_symlink_free(symlink, symlink_len + 1);
	}
	}
	#endif
	return (error);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_link_args {
	struct vnode *a_tdvp;
	struct vnode *a_vp;
	struct componentname *a_cnp;
	};
	#endif

	static int
	zfs_freebsd_link(struct vop_link_args *ap)
	{
	struct componentname *cnp = ap->a_cnp;
	vnode_t *vp = ap->a_vp;
	vnode_t *tdvp = ap->a_tdvp;

	if (tdvp->v_mount != vp->v_mount)
	return (EXDEV);

	ASSERT(cnp->cn_flags & SAVENAME);

	return (zfs_link(VTOZ(tdvp), VTOZ(vp),
	cnp->cn_nameptr, cnp->cn_cred, 0));
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_inactive_args {
	struct vnode *a_vp;
	struct thread *a_td;
	};
	#endif

	static int
	zfs_freebsd_inactive(struct vop_inactive_args *ap)
	{
	vnode_t *vp = ap->a_vp;

	#if __FreeBSD_version >= 1300123
	zfs_inactive(vp, curthread->td_ucred, NULL);
	#else
	zfs_inactive(vp, ap->a_td->td_ucred, NULL);
	#endif
	return (0);
	}

	#if __FreeBSD_version >= 1300042
	#ifndef _SYS_SYSPROTO_H_
	struct vop_need_inactive_args {
	struct vnode *a_vp;
	struct thread *a_td;
	};
	#endif

	static int
	zfs_freebsd_need_inactive(struct vop_need_inactive_args *ap)
	{
	vnode_t *vp = ap->a_vp;
	znode_t *zp = VTOZ(vp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
	int need;

	if (vn_need_pageq_flush(vp))
	return (1);

	if (!ZFS_TEARDOWN_INACTIVE_TRY_ENTER_READ(zfsvfs))
	return (1);
	need = (zp->z_sa_hdl == NULL \|\| zp->z_unlinked \|\| zp->z_atime_dirty);
	ZFS_TEARDOWN_INACTIVE_EXIT_READ(zfsvfs);

	return (need);
	}
	#endif

	#ifndef _SYS_SYSPROTO_H_
	struct vop_reclaim_args {
	struct vnode *a_vp;
	struct thread *a_td;
	};
	#endif

	static int
	zfs_freebsd_reclaim(struct vop_reclaim_args *ap)
	{
	vnode_t *vp = ap->a_vp;
	znode_t *zp = VTOZ(vp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;

	ASSERT3P(zp, !=, NULL);

	#if __FreeBSD_version < 1300042
	/* Destroy the vm object and flush associated pages. */
	vnode_destroy_vobject(vp);
	#endif
	/*
	* z_teardown_inactive_lock protects from a race with
	* zfs_znode_dmu_fini in zfsvfs_teardown during
	* force unmount.
	*/
	ZFS_TEARDOWN_INACTIVE_ENTER_READ(zfsvfs);
	if (zp->z_sa_hdl == NULL)
	zfs_znode_free(zp);
	else
	zfs_zinactive(zp);
	ZFS_TEARDOWN_INACTIVE_EXIT_READ(zfsvfs);

	vp->v_data = NULL;
	return (0);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_fid_args {
	struct vnode *a_vp;
	struct fid *a_fid;
	};
	#endif

	static int
	zfs_freebsd_fid(struct vop_fid_args *ap)
	{

	return (zfs_fid(ap->a_vp, (void *)ap->a_fid, NULL));
	}


	#ifndef _SYS_SYSPROTO_H_
	struct vop_pathconf_args {
	struct vnode *a_vp;
	int a_name;
	register_t *a_retval;
	} *ap;
	#endif

	static int
	zfs_freebsd_pathconf(struct vop_pathconf_args *ap)
	{
	ulong_t val;
	int error;

	error = zfs_pathconf(ap->a_vp, ap->a_name, &val,
	curthread->td_ucred, NULL);
	if (error == 0) {
	*ap->a_retval = val;
	return (error);
	}
	if (error != EOPNOTSUPP)
	return (error);

	switch (ap->a_name) {
	case _PC_NAME_MAX:
	*ap->a_retval = NAME_MAX;
	return (0);
	#if __FreeBSD_version >= 1400032
	case _PC_DEALLOC_PRESENT:
	*ap->a_retval = 1;
	return (0);
	#endif
	case _PC_PIPE_BUF:
	if (ap->a_vp->v_type == VDIR \|\| ap->a_vp->v_type == VFIFO) {
	*ap->a_retval = PIPE_BUF;
	return (0);
	}
	return (EINVAL);
	default:
	return (vop_stdpathconf(ap));
	}
	}

	/*
	* FreeBSD's extended attributes namespace defines file name prefix for ZFS'
	* extended attribute name:
	*
	* NAMESPACE PREFIX
	* system freebsd:system:
	* user (none, can be used to access ZFS fsattr(5) attributes
	* created on Solaris)
	*/
	static int
	zfs_create_attrname(int attrnamespace, const char name, char attrname,
	size_t size)
	{
	const char namespace, prefix, *suffix;

	/* We don't allow '/' character in attribute name. */
	if (strchr(name, '/') != NULL)
	return (SET_ERROR(EINVAL));
	/* We don't allow attribute names that start with "freebsd:" string. */
	if (strncmp(name, "freebsd:", 8) == 0)
	return (SET_ERROR(EINVAL));

	bzero(attrname, size);

	switch (attrnamespace) {
	case EXTATTR_NAMESPACE_USER:
	#if 0
	prefix = "freebsd:";
	namespace = EXTATTR_NAMESPACE_USER_STRING;
	suffix = ":";
	#else
	/*
	* This is the default namespace by which we can access all
	* attributes created on Solaris.
	*/
	prefix = namespace = suffix = "";
	#endif
	break;
	case EXTATTR_NAMESPACE_SYSTEM:
	prefix = "freebsd:";
	namespace = EXTATTR_NAMESPACE_SYSTEM_STRING;
	suffix = ":";
	break;
	case EXTATTR_NAMESPACE_EMPTY:
	default:
	return (SET_ERROR(EINVAL));
	}
	if (snprintf(attrname, size, "%s%s%s%s", prefix, namespace, suffix,
	name) >= size) {
	return (SET_ERROR(ENAMETOOLONG));
	}
	return (0);
	}

	static int
	zfs_ensure_xattr_cached(znode_t *zp)
	{
	int error = 0;

	ASSERT(RW_LOCK_HELD(&zp->z_xattr_lock));

	if (zp->z_xattr_cached != NULL)
	return (0);

	if (rw_write_held(&zp->z_xattr_lock))
	return (zfs_sa_get_xattr(zp));

	if (!rw_tryupgrade(&zp->z_xattr_lock)) {
	rw_exit(&zp->z_xattr_lock);
	rw_enter(&zp->z_xattr_lock, RW_WRITER);
	}
	if (zp->z_xattr_cached == NULL)
	error = zfs_sa_get_xattr(zp);
	rw_downgrade(&zp->z_xattr_lock);
	return (error);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_getextattr {
	IN struct vnode *a_vp;
	IN int a_attrnamespace;
	IN const char *a_name;
	INOUT struct uio *a_uio;
	OUT size_t *a_size;
	IN struct ucred *a_cred;
	IN struct thread *a_td;
	};
	#endif

	static int
	zfs_getextattr_dir(struct vop_getextattr_args ap, const char attrname)
	{
	struct thread *td = ap->a_td;
	struct nameidata nd;
	struct vattr va;
	vnode_t xvp = NULL, vp;
	int error, flags;

	error = zfs_lookup(ap->a_vp, NULL, &xvp, NULL, 0, ap->a_cred,
	LOOKUP_XATTR, B_FALSE);
	if (error != 0)
	return (error);

	flags = FREAD;
	+#if __FreeBSD_version < 1400043
	+ NDINIT_ATVP(&nd, LOOKUP, NOFOLLOW, UIO_SYSSPACE, attrname,
	+ xvp, td);
	+#else
	NDINIT_ATVP(&nd, LOOKUP, NOFOLLOW, UIO_SYSSPACE, attrname, xvp);
	+#endif
	error = vn_open_cred(&nd, &flags, 0, VN_OPEN_INVFS, ap->a_cred, NULL);
	vp = nd.ni_vp;
	NDFREE(&nd, NDF_ONLY_PNBUF);
	if (error != 0)
	return (SET_ERROR(error));

	if (ap->a_size != NULL) {
	error = VOP_GETATTR(vp, &va, ap->a_cred);
	if (error == 0)
	*ap->a_size = (size_t)va.va_size;
	} else if (ap->a_uio != NULL)
	error = VOP_READ(vp, ap->a_uio, IO_UNIT, ap->a_cred);

	VOP_UNLOCK1(vp);
	vn_close(vp, flags, ap->a_cred, td);
	return (error);
	}

	static int
	zfs_getextattr_sa(struct vop_getextattr_args ap, const char attrname)
	{
	znode_t *zp = VTOZ(ap->a_vp);
	uchar_t *nv_value;
	uint_t nv_size;
	int error;

	error = zfs_ensure_xattr_cached(zp);
	if (error != 0)
	return (error);

	ASSERT(RW_LOCK_HELD(&zp->z_xattr_lock));
	ASSERT3P(zp->z_xattr_cached, !=, NULL);

	error = nvlist_lookup_byte_array(zp->z_xattr_cached, attrname,
	&nv_value, &nv_size);
	if (error != 0)
	return (SET_ERROR(error));

	if (ap->a_size != NULL)
	*ap->a_size = nv_size;
	else if (ap->a_uio != NULL)
	error = uiomove(nv_value, nv_size, ap->a_uio);
	if (error != 0)
	return (SET_ERROR(error));

	return (0);
	}

	/*
	* Vnode operation to retrieve a named extended attribute.
	*/
	static int
	zfs_getextattr(struct vop_getextattr_args *ap)
	{
	znode_t *zp = VTOZ(ap->a_vp);
	zfsvfs_t *zfsvfs = ZTOZSB(zp);
	char attrname[EXTATTR_MAXNAMELEN+1];
	int error;

	/*
	* If the xattr property is off, refuse the request.
	*/
	if (!(zfsvfs->z_flags & ZSB_XATTR))
	return (SET_ERROR(EOPNOTSUPP));

	error = extattr_check_cred(ap->a_vp, ap->a_attrnamespace,
	ap->a_cred, ap->a_td, VREAD);
	if (error != 0)
	return (SET_ERROR(error));

	error = zfs_create_attrname(ap->a_attrnamespace, ap->a_name, attrname,
	sizeof (attrname));
	if (error != 0)
	return (error);

	error = ENOENT;
	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp)
	rw_enter(&zp->z_xattr_lock, RW_READER);
	if (zfsvfs->z_use_sa && zp->z_is_sa)
	error = zfs_getextattr_sa(ap, attrname);
	if (error == ENOENT)
	error = zfs_getextattr_dir(ap, attrname);
	rw_exit(&zp->z_xattr_lock);
	ZFS_EXIT(zfsvfs);
	if (error == ENOENT)
	error = SET_ERROR(ENOATTR);
	return (error);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_deleteextattr {
	IN struct vnode *a_vp;
	IN int a_attrnamespace;
	IN const char *a_name;
	IN struct ucred *a_cred;
	IN struct thread *a_td;
	};
	#endif

	static int
	zfs_deleteextattr_dir(struct vop_deleteextattr_args ap, const char attrname)
	{
	struct nameidata nd;
	vnode_t xvp = NULL, vp;
	int error;

	error = zfs_lookup(ap->a_vp, NULL, &xvp, NULL, 0, ap->a_cred,
	LOOKUP_XATTR, B_FALSE);
	if (error != 0)
	return (error);

	+#if __FreeBSD_version < 1400043
	+ NDINIT_ATVP(&nd, DELETE, NOFOLLOW \| LOCKPARENT \| LOCKLEAF,
	+ UIO_SYSSPACE, attrname, xvp, ap->a_td);
	+#else
	NDINIT_ATVP(&nd, DELETE, NOFOLLOW \| LOCKPARENT \| LOCKLEAF,
	UIO_SYSSPACE, attrname, xvp);
	+#endif
	error = namei(&nd);
	vp = nd.ni_vp;
	if (error != 0) {
	NDFREE(&nd, NDF_ONLY_PNBUF);
	return (SET_ERROR(error));
	}

	error = VOP_REMOVE(nd.ni_dvp, vp, &nd.ni_cnd);
	NDFREE(&nd, NDF_ONLY_PNBUF);

	vput(nd.ni_dvp);
	if (vp == nd.ni_dvp)
	vrele(vp);
	else
	vput(vp);

	return (error);
	}

	static int
	zfs_deleteextattr_sa(struct vop_deleteextattr_args ap, const char attrname)
	{
	znode_t *zp = VTOZ(ap->a_vp);
	nvlist_t *nvl;
	int error;

	error = zfs_ensure_xattr_cached(zp);
	if (error != 0)
	return (error);

	ASSERT(RW_WRITE_HELD(&zp->z_xattr_lock));
	ASSERT3P(zp->z_xattr_cached, !=, NULL);

	nvl = zp->z_xattr_cached;
	error = nvlist_remove(nvl, attrname, DATA_TYPE_BYTE_ARRAY);
	if (error != 0)
	error = SET_ERROR(error);
	else
	error = zfs_sa_set_xattr(zp);
	if (error != 0) {
	zp->z_xattr_cached = NULL;
	nvlist_free(nvl);
	}
	return (error);
	}

	/*
	* Vnode operation to remove a named attribute.
	*/
	static int
	zfs_deleteextattr(struct vop_deleteextattr_args *ap)
	{
	znode_t *zp = VTOZ(ap->a_vp);
	zfsvfs_t *zfsvfs = ZTOZSB(zp);
	char attrname[EXTATTR_MAXNAMELEN+1];
	int error;

	/*
	* If the xattr property is off, refuse the request.
	*/
	if (!(zfsvfs->z_flags & ZSB_XATTR))
	return (SET_ERROR(EOPNOTSUPP));

	error = extattr_check_cred(ap->a_vp, ap->a_attrnamespace,
	ap->a_cred, ap->a_td, VWRITE);
	if (error != 0)
	return (SET_ERROR(error));

	error = zfs_create_attrname(ap->a_attrnamespace, ap->a_name, attrname,
	sizeof (attrname));
	if (error != 0)
	return (error);

	size_t size = 0;
	struct vop_getextattr_args vga = {
	.a_vp = ap->a_vp,
	.a_size = &size,
	.a_cred = ap->a_cred,
	.a_td = ap->a_td,
	};
	error = ENOENT;
	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);
	rw_enter(&zp->z_xattr_lock, RW_WRITER);
	if (zfsvfs->z_use_sa && zp->z_is_sa) {
	error = zfs_getextattr_sa(&vga, attrname);
	if (error == 0)
	error = zfs_deleteextattr_sa(ap, attrname);
	}
	if (error == ENOENT) {
	error = zfs_getextattr_dir(&vga, attrname);
	if (error == 0)
	error = zfs_deleteextattr_dir(ap, attrname);
	}
	rw_exit(&zp->z_xattr_lock);
	ZFS_EXIT(zfsvfs);
	if (error == ENOENT)
	error = SET_ERROR(ENOATTR);
	return (error);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_setextattr {
	IN struct vnode *a_vp;
	IN int a_attrnamespace;
	IN const char *a_name;
	INOUT struct uio *a_uio;
	IN struct ucred *a_cred;
	IN struct thread *a_td;
	};
	#endif

	static int
	zfs_setextattr_dir(struct vop_setextattr_args ap, const char attrname)
	{
	struct thread *td = ap->a_td;
	struct nameidata nd;
	struct vattr va;
	vnode_t xvp = NULL, vp;
	int error, flags;

	error = zfs_lookup(ap->a_vp, NULL, &xvp, NULL, 0, ap->a_cred,
	LOOKUP_XATTR \| CREATE_XATTR_DIR, B_FALSE);
	if (error != 0)
	return (error);

	flags = FFLAGS(O_WRONLY \| O_CREAT);
	+#if __FreeBSD_version < 1400043
	+ NDINIT_ATVP(&nd, LOOKUP, NOFOLLOW, UIO_SYSSPACE, attrname, xvp, td);
	+#else
	NDINIT_ATVP(&nd, LOOKUP, NOFOLLOW, UIO_SYSSPACE, attrname, xvp);
	+#endif
	error = vn_open_cred(&nd, &flags, 0600, VN_OPEN_INVFS, ap->a_cred,
	NULL);
	vp = nd.ni_vp;
	NDFREE(&nd, NDF_ONLY_PNBUF);
	if (error != 0)
	return (SET_ERROR(error));

	VATTR_NULL(&va);
	va.va_size = 0;
	error = VOP_SETATTR(vp, &va, ap->a_cred);
	if (error == 0)
	VOP_WRITE(vp, ap->a_uio, IO_UNIT, ap->a_cred);

	VOP_UNLOCK1(vp);
	vn_close(vp, flags, ap->a_cred, td);
	return (error);
	}

	static int
	zfs_setextattr_sa(struct vop_setextattr_args ap, const char attrname)
	{
	znode_t *zp = VTOZ(ap->a_vp);
	nvlist_t *nvl;
	size_t sa_size;
	int error;

	error = zfs_ensure_xattr_cached(zp);
	if (error != 0)
	return (error);

	ASSERT(RW_WRITE_HELD(&zp->z_xattr_lock));
	ASSERT3P(zp->z_xattr_cached, !=, NULL);

	nvl = zp->z_xattr_cached;
	size_t entry_size = ap->a_uio->uio_resid;
	if (entry_size > DXATTR_MAX_ENTRY_SIZE)
	return (SET_ERROR(EFBIG));
	error = nvlist_size(nvl, &sa_size, NV_ENCODE_XDR);
	if (error != 0)
	return (SET_ERROR(error));
	if (sa_size > DXATTR_MAX_SA_SIZE)
	return (SET_ERROR(EFBIG));
	uchar_t *buf = kmem_alloc(entry_size, KM_SLEEP);
	error = uiomove(buf, entry_size, ap->a_uio);
	if (error != 0) {
	error = SET_ERROR(error);
	} else {
	error = nvlist_add_byte_array(nvl, attrname, buf, entry_size);
	if (error != 0)
	error = SET_ERROR(error);
	}
	kmem_free(buf, entry_size);
	if (error == 0)
	error = zfs_sa_set_xattr(zp);
	if (error != 0) {
	zp->z_xattr_cached = NULL;
	nvlist_free(nvl);
	}
	return (error);
	}

	/*
	* Vnode operation to set a named attribute.
	*/
	static int
	zfs_setextattr(struct vop_setextattr_args *ap)
	{
	znode_t *zp = VTOZ(ap->a_vp);
	zfsvfs_t *zfsvfs = ZTOZSB(zp);
	char attrname[EXTATTR_MAXNAMELEN+1];
	int error;

	/*
	* If the xattr property is off, refuse the request.
	*/
	if (!(zfsvfs->z_flags & ZSB_XATTR))
	return (SET_ERROR(EOPNOTSUPP));

	error = extattr_check_cred(ap->a_vp, ap->a_attrnamespace,
	ap->a_cred, ap->a_td, VWRITE);
	if (error != 0)
	return (SET_ERROR(error));

	error = zfs_create_attrname(ap->a_attrnamespace, ap->a_name, attrname,
	sizeof (attrname));
	if (error != 0)
	return (error);

	struct vop_deleteextattr_args vda = {
	.a_vp = ap->a_vp,
	.a_cred = ap->a_cred,
	.a_td = ap->a_td,
	};
	error = ENOENT;
	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);
	rw_enter(&zp->z_xattr_lock, RW_WRITER);
	if (zfsvfs->z_use_sa && zp->z_is_sa && zfsvfs->z_xattr_sa) {
	error = zfs_setextattr_sa(ap, attrname);
	if (error == 0)
	/*
	* Successfully put into SA, we need to clear the one
	* in dir if present.
	*/
	zfs_deleteextattr_dir(&vda, attrname);
	}
	if (error) {
	error = zfs_setextattr_dir(ap, attrname);
	if (error == 0 && zp->z_is_sa)
	/*
	* Successfully put into dir, we need to clear the one
	* in SA if present.
	*/
	zfs_deleteextattr_sa(&vda, attrname);
	}
	rw_exit(&zp->z_xattr_lock);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_listextattr {
	IN struct vnode *a_vp;
	IN int a_attrnamespace;
	INOUT struct uio *a_uio;
	OUT size_t *a_size;
	IN struct ucred *a_cred;
	IN struct thread *a_td;
	};
	#endif

	static int
	zfs_listextattr_dir(struct vop_listextattr_args ap, const char attrprefix)
	{
	struct thread *td = ap->a_td;
	struct nameidata nd;
	uint8_t dirbuf[sizeof (struct dirent)];
	struct iovec aiov;
	struct uio auio;
	vnode_t xvp = NULL, vp;
	int error, eof;

	error = zfs_lookup(ap->a_vp, NULL, &xvp, NULL, 0, ap->a_cred,
	LOOKUP_XATTR, B_FALSE);
	if (error != 0) {
	/*
	* ENOATTR means that the EA directory does not yet exist,
	* i.e. there are no extended attributes there.
	*/
	if (error == ENOATTR)
	error = 0;
	return (error);
	}

	+#if __FreeBSD_version < 1400043
	+ NDINIT_ATVP(&nd, LOOKUP, NOFOLLOW \| LOCKLEAF \| LOCKSHARED,
	+ UIO_SYSSPACE, ".", xvp, td);
	+#else
	NDINIT_ATVP(&nd, LOOKUP, NOFOLLOW \| LOCKLEAF \| LOCKSHARED,
	UIO_SYSSPACE, ".", xvp);
	+#endif
	error = namei(&nd);
	vp = nd.ni_vp;
	NDFREE(&nd, NDF_ONLY_PNBUF);
	if (error != 0)
	return (SET_ERROR(error));

	auio.uio_iov = &aiov;
	auio.uio_iovcnt = 1;
	auio.uio_segflg = UIO_SYSSPACE;
	auio.uio_td = td;
	auio.uio_rw = UIO_READ;
	auio.uio_offset = 0;

	size_t plen = strlen(attrprefix);

	do {
	aiov.iov_base = (void *)dirbuf;
	aiov.iov_len = sizeof (dirbuf);
	auio.uio_resid = sizeof (dirbuf);
	error = VOP_READDIR(vp, &auio, ap->a_cred, &eof, NULL, NULL);
	if (error != 0)
	break;
	int done = sizeof (dirbuf) - auio.uio_resid;
	for (int pos = 0; pos < done; ) {
	struct dirent dp = (struct dirent )(dirbuf + pos);
	pos += dp->d_reclen;
	/*
	* XXX: Temporarily we also accept DT_UNKNOWN, as this
	* is what we get when attribute was created on Solaris.
	*/
	if (dp->d_type != DT_REG && dp->d_type != DT_UNKNOWN)
	continue;
	else if (plen == 0 &&
	strncmp(dp->d_name, "freebsd:", 8) == 0)
	continue;
	else if (strncmp(dp->d_name, attrprefix, plen) != 0)
	continue;
	uint8_t nlen = dp->d_namlen - plen;
	if (ap->a_size != NULL) {
	*ap->a_size += 1 + nlen;
	} else if (ap->a_uio != NULL) {
	/*
	* Format of extattr name entry is one byte for
	* length and the rest for name.
	*/
	error = uiomove(&nlen, 1, ap->a_uio);
	if (error == 0) {
	char *namep = dp->d_name + plen;
	error = uiomove(namep, nlen, ap->a_uio);
	}
	if (error != 0) {
	error = SET_ERROR(error);
	break;
	}
	}
	}
	} while (!eof && error == 0);

	vput(vp);
	return (error);
	}

	static int
	zfs_listextattr_sa(struct vop_listextattr_args ap, const char attrprefix)
	{
	znode_t *zp = VTOZ(ap->a_vp);
	int error;

	error = zfs_ensure_xattr_cached(zp);
	if (error != 0)
	return (error);

	ASSERT(RW_LOCK_HELD(&zp->z_xattr_lock));
	ASSERT3P(zp->z_xattr_cached, !=, NULL);

	size_t plen = strlen(attrprefix);
	nvpair_t *nvp = NULL;
	while ((nvp = nvlist_next_nvpair(zp->z_xattr_cached, nvp)) != NULL) {
	ASSERT3U(nvpair_type(nvp), ==, DATA_TYPE_BYTE_ARRAY);

	const char *name = nvpair_name(nvp);
	if (plen == 0 && strncmp(name, "freebsd:", 8) == 0)
	continue;
	else if (strncmp(name, attrprefix, plen) != 0)
	continue;
	uint8_t nlen = strlen(name) - plen;
	if (ap->a_size != NULL) {
	*ap->a_size += 1 + nlen;
	} else if (ap->a_uio != NULL) {
	/*
	* Format of extattr name entry is one byte for
	* length and the rest for name.
	*/
	error = uiomove(&nlen, 1, ap->a_uio);
	if (error == 0) {
	char namep = __DECONST(char , name) + plen;
	error = uiomove(namep, nlen, ap->a_uio);
	}
	if (error != 0) {
	error = SET_ERROR(error);
	break;
	}
	}
	}

	return (error);
	}

	/*
	* Vnode operation to retrieve extended attributes on a vnode.
	*/
	static int
	zfs_listextattr(struct vop_listextattr_args *ap)
	{
	znode_t *zp = VTOZ(ap->a_vp);
	zfsvfs_t *zfsvfs = ZTOZSB(zp);
	char attrprefix[16];
	int error;

	if (ap->a_size != NULL)
	*ap->a_size = 0;

	/*
	* If the xattr property is off, refuse the request.
	*/
	if (!(zfsvfs->z_flags & ZSB_XATTR))
	return (SET_ERROR(EOPNOTSUPP));

	error = extattr_check_cred(ap->a_vp, ap->a_attrnamespace,
	ap->a_cred, ap->a_td, VREAD);
	if (error != 0)
	return (SET_ERROR(error));

	error = zfs_create_attrname(ap->a_attrnamespace, "", attrprefix,
	sizeof (attrprefix));
	if (error != 0)
	return (error);

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);
	rw_enter(&zp->z_xattr_lock, RW_READER);
	if (zfsvfs->z_use_sa && zp->z_is_sa)
	error = zfs_listextattr_sa(ap, attrprefix);
	if (error == 0)
	error = zfs_listextattr_dir(ap, attrprefix);
	rw_exit(&zp->z_xattr_lock);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_getacl_args {
	struct vnode *vp;
	acl_type_t type;
	struct acl *aclp;
	struct ucred *cred;
	struct thread *td;
	};
	#endif

	static int
	zfs_freebsd_getacl(struct vop_getacl_args *ap)
	{
	int error;
	vsecattr_t vsecattr;

	if (ap->a_type != ACL_TYPE_NFS4)
	return (EINVAL);

	vsecattr.vsa_mask = VSA_ACE \| VSA_ACECNT;
	if ((error = zfs_getsecattr(VTOZ(ap->a_vp),
	&vsecattr, 0, ap->a_cred)))
	return (error);

	error = acl_from_aces(ap->a_aclp, vsecattr.vsa_aclentp,
	vsecattr.vsa_aclcnt);
	if (vsecattr.vsa_aclentp != NULL)
	kmem_free(vsecattr.vsa_aclentp, vsecattr.vsa_aclentsz);

	return (error);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_setacl_args {
	struct vnode *vp;
	acl_type_t type;
	struct acl *aclp;
	struct ucred *cred;
	struct thread *td;
	};
	#endif

	static int
	zfs_freebsd_setacl(struct vop_setacl_args *ap)
	{
	int error;
	vsecattr_t vsecattr;
	int aclbsize; /* size of acl list in bytes */
	aclent_t *aaclp;

	if (ap->a_type != ACL_TYPE_NFS4)
	return (EINVAL);

	if (ap->a_aclp == NULL)
	return (EINVAL);

	if (ap->a_aclp->acl_cnt < 1 \|\| ap->a_aclp->acl_cnt > MAX_ACL_ENTRIES)
	return (EINVAL);

	/*
	* With NFSv4 ACLs, chmod(2) may need to add additional entries,
	* splitting every entry into two and appending "canonical six"
	* entries at the end. Don't allow for setting an ACL that would
	* cause chmod(2) to run out of ACL entries.
	*/
	if (ap->a_aclp->acl_cnt * 2 + 6 > ACL_MAX_ENTRIES)
	return (ENOSPC);

	error = acl_nfs4_check(ap->a_aclp, ap->a_vp->v_type == VDIR);
	if (error != 0)
	return (error);

	vsecattr.vsa_mask = VSA_ACE;
	aclbsize = ap->a_aclp->acl_cnt * sizeof (ace_t);
	vsecattr.vsa_aclentp = kmem_alloc(aclbsize, KM_SLEEP);
	aaclp = vsecattr.vsa_aclentp;
	vsecattr.vsa_aclentsz = aclbsize;

	aces_from_acl(vsecattr.vsa_aclentp, &vsecattr.vsa_aclcnt, ap->a_aclp);
	error = zfs_setsecattr(VTOZ(ap->a_vp), &vsecattr, 0, ap->a_cred);
	kmem_free(aaclp, aclbsize);

	return (error);
	}

	#ifndef _SYS_SYSPROTO_H_
	struct vop_aclcheck_args {
	struct vnode *vp;
	acl_type_t type;
	struct acl *aclp;
	struct ucred *cred;
	struct thread *td;
	};
	#endif

	static int
	zfs_freebsd_aclcheck(struct vop_aclcheck_args *ap)
	{

	return (EOPNOTSUPP);
	}

	static int
	zfs_vptocnp(struct vop_vptocnp_args *ap)
	{
	vnode_t *covered_vp;
	vnode_t *vp = ap->a_vp;
	zfsvfs_t *zfsvfs = vp->v_vfsp->vfs_data;
	znode_t *zp = VTOZ(vp);
	int ltype;
	int error;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	/*
	* If we are a snapshot mounted under .zfs, run the operation
	* on the covered vnode.
	*/
	if (zp->z_id != zfsvfs->z_root \|\| zfsvfs->z_parent == zfsvfs) {
	char name[MAXNAMLEN + 1];
	znode_t *dzp;
	size_t len;

	error = zfs_znode_parent_and_name(zp, &dzp, name);
	if (error == 0) {
	len = strlen(name);
	if (*ap->a_buflen < len)
	error = SET_ERROR(ENOMEM);
	}
	if (error == 0) {
	*ap->a_buflen -= len;
	bcopy(name, ap->a_buf + *ap->a_buflen, len);
	*ap->a_vpp = ZTOV(dzp);
	}
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	ZFS_EXIT(zfsvfs);

	covered_vp = vp->v_mount->mnt_vnodecovered;
	#if __FreeBSD_version >= 1300045
	enum vgetstate vs = vget_prep(covered_vp);
	#else
	vhold(covered_vp);
	#endif
	ltype = VOP_ISLOCKED(vp);
	VOP_UNLOCK1(vp);
	#if __FreeBSD_version >= 1300045
	error = vget_finish(covered_vp, LK_SHARED, vs);
	#else
	error = vget(covered_vp, LK_SHARED \| LK_VNHELD, curthread);
	#endif
	if (error == 0) {
	#if __FreeBSD_version >= 1300123
	error = VOP_VPTOCNP(covered_vp, ap->a_vpp, ap->a_buf,
	ap->a_buflen);
	#else
	error = VOP_VPTOCNP(covered_vp, ap->a_vpp, ap->a_cred,
	ap->a_buf, ap->a_buflen);
	#endif
	vput(covered_vp);
	}
	vn_lock(vp, ltype \| LK_RETRY);
	if (VN_IS_DOOMED(vp))
	error = SET_ERROR(ENOENT);
	return (error);
	}

	#if __FreeBSD_version >= 1400032
	static int
	zfs_deallocate(struct vop_deallocate_args *ap)
	{
	znode_t *zp = VTOZ(ap->a_vp);
	zfsvfs_t *zfsvfs = zp->z_zfsvfs;
	zilog_t *zilog;
	off_t off, len, file_sz;
	int error;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	/*
	* Callers might not be able to detect properly that we are read-only,
	* so check it explicitly here.
	*/
	if (zfs_is_readonly(zfsvfs)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EROFS));
	}

	zilog = zfsvfs->z_log;
	off = *ap->a_offset;
	len = *ap->a_len;
	file_sz = zp->z_size;
	if (off + len > file_sz)
	len = file_sz - off;
	/* Fast path for out-of-range request. */
	if (len <= 0) {
	*ap->a_len = 0;
	ZFS_EXIT(zfsvfs);
	return (0);
	}

	error = zfs_freesp(zp, off, len, O_RDWR, TRUE);
	if (error == 0) {
	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS \|\|
	(ap->a_ioflag & IO_SYNC) != 0)
	zil_commit(zilog, zp->z_id);
	*ap->a_offset = off + len;
	*ap->a_len = 0;
	}

	ZFS_EXIT(zfsvfs);
	return (error);
	}
	#endif

	struct vop_vector zfs_vnodeops;
	struct vop_vector zfs_fifoops;
	struct vop_vector zfs_shareops;

	struct vop_vector zfs_vnodeops = {
	.vop_default = &default_vnodeops,
	.vop_inactive = zfs_freebsd_inactive,
	#if __FreeBSD_version >= 1300042
	.vop_need_inactive = zfs_freebsd_need_inactive,
	#endif
	.vop_reclaim = zfs_freebsd_reclaim,
	#if __FreeBSD_version >= 1300102
	.vop_fplookup_vexec = zfs_freebsd_fplookup_vexec,
	#endif
	#if __FreeBSD_version >= 1300139
	.vop_fplookup_symlink = zfs_freebsd_fplookup_symlink,
	#endif
	.vop_access = zfs_freebsd_access,
	.vop_allocate = VOP_EINVAL,
	#if __FreeBSD_version >= 1400032
	.vop_deallocate = zfs_deallocate,
	#endif
	.vop_lookup = zfs_cache_lookup,
	.vop_cachedlookup = zfs_freebsd_cachedlookup,
	.vop_getattr = zfs_freebsd_getattr,
	.vop_setattr = zfs_freebsd_setattr,
	.vop_create = zfs_freebsd_create,
	.vop_mknod = (vop_mknod_t *)zfs_freebsd_create,
	.vop_mkdir = zfs_freebsd_mkdir,
	.vop_readdir = zfs_freebsd_readdir,
	.vop_fsync = zfs_freebsd_fsync,
	.vop_open = zfs_freebsd_open,
	.vop_close = zfs_freebsd_close,
	.vop_rmdir = zfs_freebsd_rmdir,
	.vop_ioctl = zfs_freebsd_ioctl,
	.vop_link = zfs_freebsd_link,
	.vop_symlink = zfs_freebsd_symlink,
	.vop_readlink = zfs_freebsd_readlink,
	.vop_read = zfs_freebsd_read,
	.vop_write = zfs_freebsd_write,
	.vop_remove = zfs_freebsd_remove,
	.vop_rename = zfs_freebsd_rename,
	.vop_pathconf = zfs_freebsd_pathconf,
	.vop_bmap = zfs_freebsd_bmap,
	.vop_fid = zfs_freebsd_fid,
	.vop_getextattr = zfs_getextattr,
	.vop_deleteextattr = zfs_deleteextattr,
	.vop_setextattr = zfs_setextattr,
	.vop_listextattr = zfs_listextattr,
	.vop_getacl = zfs_freebsd_getacl,
	.vop_setacl = zfs_freebsd_setacl,
	.vop_aclcheck = zfs_freebsd_aclcheck,
	.vop_getpages = zfs_freebsd_getpages,
	.vop_putpages = zfs_freebsd_putpages,
	.vop_vptocnp = zfs_vptocnp,
	#if __FreeBSD_version >= 1300064
	.vop_lock1 = vop_lock,
	.vop_unlock = vop_unlock,
	.vop_islocked = vop_islocked,
	#endif
	+#if __FreeBSD_version >= 1400043
	.vop_add_writecount = vop_stdadd_writecount_nomsync,
	+#endif
	};
	VFS_VOP_VECTOR_REGISTER(zfs_vnodeops);

	struct vop_vector zfs_fifoops = {
	.vop_default = &fifo_specops,
	.vop_fsync = zfs_freebsd_fsync,
	#if __FreeBSD_version >= 1300102
	.vop_fplookup_vexec = zfs_freebsd_fplookup_vexec,
	#endif
	#if __FreeBSD_version >= 1300139
	.vop_fplookup_symlink = zfs_freebsd_fplookup_symlink,
	#endif
	.vop_access = zfs_freebsd_access,
	.vop_getattr = zfs_freebsd_getattr,
	.vop_inactive = zfs_freebsd_inactive,
	.vop_read = VOP_PANIC,
	.vop_reclaim = zfs_freebsd_reclaim,
	.vop_setattr = zfs_freebsd_setattr,
	.vop_write = VOP_PANIC,
	.vop_pathconf = zfs_freebsd_pathconf,
	.vop_fid = zfs_freebsd_fid,
	.vop_getacl = zfs_freebsd_getacl,
	.vop_setacl = zfs_freebsd_setacl,
	.vop_aclcheck = zfs_freebsd_aclcheck,
	+#if __FreeBSD_version >= 1400043
	.vop_add_writecount = vop_stdadd_writecount_nomsync,
	+#endif
	};
	VFS_VOP_VECTOR_REGISTER(zfs_fifoops);

	/*
	* special share hidden files vnode operations template
	*/
	struct vop_vector zfs_shareops = {
	.vop_default = &default_vnodeops,
	#if __FreeBSD_version >= 1300121
	.vop_fplookup_vexec = VOP_EAGAIN,
	#endif
	#if __FreeBSD_version >= 1300139
	.vop_fplookup_symlink = VOP_EAGAIN,
	#endif
	.vop_access = zfs_freebsd_access,
	.vop_inactive = zfs_freebsd_inactive,
	.vop_reclaim = zfs_freebsd_reclaim,
	.vop_fid = zfs_freebsd_fid,
	.vop_pathconf = zfs_freebsd_pathconf,
	+#if __FreeBSD_version >= 1400043
	.vop_add_writecount = vop_stdadd_writecount_nomsync,
	+#endif
	};
	VFS_VOP_VECTOR_REGISTER(zfs_shareops);
	diff --git a/sys/contrib/openzfs/module/os/linux/spl/spl-taskq.c b/sys/contrib/openzfs/module/os/linux/spl/spl-taskq.c
	index fb25a4154485..a879c2856302 100644
	--- a/sys/contrib/openzfs/module/os/linux/spl/spl-taskq.c
	+++ b/sys/contrib/openzfs/module/os/linux/spl/spl-taskq.c
	@@ -1,1429 +1,1433 @@
	/*
	* Copyright (C) 2007-2010 Lawrence Livermore National Security, LLC.
	* Copyright (C) 2007 The Regents of the University of California.
	* Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
	* Written by Brian Behlendorf <behlendorf1@llnl.gov>.
	* UCRL-CODE-235197
	*
	* This file is part of the SPL, Solaris Porting Layer.
	*
	* The SPL is free software; you can redistribute it and/or modify it
	* under the terms of the GNU General Public License as published by the
	* Free Software Foundation; either version 2 of the License, or (at your
	* option) any later version.
	*
	* The SPL is distributed in the hope that it will be useful, but WITHOUT
	* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
	* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
	* for more details.
	*
	* You should have received a copy of the GNU General Public License along
	* with the SPL. If not, see <http://www.gnu.org/licenses/>.
	*
	* Solaris Porting Layer (SPL) Task Queue Implementation.
	*/

	#include <sys/timer.h>
	#include <sys/taskq.h>
	#include <sys/kmem.h>
	#include <sys/tsd.h>
	#include <sys/trace_spl.h>
	#ifdef HAVE_CPU_HOTPLUG
	#include <linux/cpuhotplug.h>
	#endif

	int spl_taskq_thread_bind = 0;
	module_param(spl_taskq_thread_bind, int, 0644);
	MODULE_PARM_DESC(spl_taskq_thread_bind, "Bind taskq thread to CPU by default");


	int spl_taskq_thread_dynamic = 1;
	module_param(spl_taskq_thread_dynamic, int, 0444);
	MODULE_PARM_DESC(spl_taskq_thread_dynamic, "Allow dynamic taskq threads");

	int spl_taskq_thread_priority = 1;
	module_param(spl_taskq_thread_priority, int, 0644);
	MODULE_PARM_DESC(spl_taskq_thread_priority,
	"Allow non-default priority for taskq threads");

	int spl_taskq_thread_sequential = 4;
	module_param(spl_taskq_thread_sequential, int, 0644);
	MODULE_PARM_DESC(spl_taskq_thread_sequential,
	"Create new taskq threads after N sequential tasks");

	-/* Global system-wide dynamic task queue available for all consumers */
	+/*
	+ * Global system-wide dynamic task queue available for all consumers. This
	+ * taskq is not intended for long-running tasks; instead, a dedicated taskq
	+ * should be created.
	+ */
	taskq_t *system_taskq;
	EXPORT_SYMBOL(system_taskq);
	/* Global dynamic task queue for long delay */
	taskq_t *system_delay_taskq;
	EXPORT_SYMBOL(system_delay_taskq);

	/* Private dedicated taskq for creating new taskq threads on demand. */
	static taskq_t *dynamic_taskq;
	static taskq_thread_t taskq_thread_create(taskq_t );

	#ifdef HAVE_CPU_HOTPLUG
	/* Multi-callback id for cpu hotplugging. */
	static int spl_taskq_cpuhp_state;
	#endif

	/* List of all taskqs */
	LIST_HEAD(tq_list);
	struct rw_semaphore tq_list_sem;
	static uint_t taskq_tsd;

	static int
	task_km_flags(uint_t flags)
	{
	if (flags & TQ_NOSLEEP)
	return (KM_NOSLEEP);

	if (flags & TQ_PUSHPAGE)
	return (KM_PUSHPAGE);

	return (KM_SLEEP);
	}

	/*
	* taskq_find_by_name - Find the largest instance number of a named taskq.
	*/
	static int
	taskq_find_by_name(const char *name)
	{
	struct list_head *tql = NULL;
	taskq_t *tq;

	list_for_each_prev(tql, &tq_list) {
	tq = list_entry(tql, taskq_t, tq_taskqs);
	if (strcmp(name, tq->tq_name) == 0)
	return (tq->tq_instance);
	}
	return (-1);
	}

	/*
	* NOTE: Must be called with tq->tq_lock held, returns a list_t which
	* is not attached to the free, work, or pending taskq lists.
	*/
	static taskq_ent_t *
	task_alloc(taskq_t tq, uint_t flags, unsigned long irqflags)
	{
	taskq_ent_t *t;
	int count = 0;

	ASSERT(tq);
	retry:
	/* Acquire taskq_ent_t's from free list if available */
	if (!list_empty(&tq->tq_free_list) && !(flags & TQ_NEW)) {
	t = list_entry(tq->tq_free_list.next, taskq_ent_t, tqent_list);

	ASSERT(!(t->tqent_flags & TQENT_FLAG_PREALLOC));
	ASSERT(!(t->tqent_flags & TQENT_FLAG_CANCEL));
	ASSERT(!timer_pending(&t->tqent_timer));

	list_del_init(&t->tqent_list);
	return (t);
	}

	/* Free list is empty and memory allocations are prohibited */
	if (flags & TQ_NOALLOC)
	return (NULL);

	/* Hit maximum taskq_ent_t pool size */
	if (tq->tq_nalloc >= tq->tq_maxalloc) {
	if (flags & TQ_NOSLEEP)
	return (NULL);

	/*
	* Sleep periodically polling the free list for an available
	* taskq_ent_t. Dispatching with TQ_SLEEP should always succeed
	* but we cannot block forever waiting for an taskq_ent_t to
	* show up in the free list, otherwise a deadlock can happen.
	*
	* Therefore, we need to allocate a new task even if the number
	* of allocated tasks is above tq->tq_maxalloc, but we still
	* end up delaying the task allocation by one second, thereby
	* throttling the task dispatch rate.
	*/
	spin_unlock_irqrestore(&tq->tq_lock, *irqflags);
	schedule_timeout(HZ / 100);
	spin_lock_irqsave_nested(&tq->tq_lock, *irqflags,
	tq->tq_lock_class);
	if (count < 100) {
	count++;
	goto retry;
	}
	}

	spin_unlock_irqrestore(&tq->tq_lock, *irqflags);
	t = kmem_alloc(sizeof (taskq_ent_t), task_km_flags(flags));
	spin_lock_irqsave_nested(&tq->tq_lock, *irqflags, tq->tq_lock_class);

	if (t) {
	taskq_init_ent(t);
	tq->tq_nalloc++;
	}

	return (t);
	}

	/*
	* NOTE: Must be called with tq->tq_lock held, expects the taskq_ent_t
	* to already be removed from the free, work, or pending taskq lists.
	*/
	static void
	task_free(taskq_t tq, taskq_ent_t t)
	{
	ASSERT(tq);
	ASSERT(t);
	ASSERT(list_empty(&t->tqent_list));
	ASSERT(!timer_pending(&t->tqent_timer));

	kmem_free(t, sizeof (taskq_ent_t));
	tq->tq_nalloc--;
	}

	/*
	* NOTE: Must be called with tq->tq_lock held, either destroys the
	* taskq_ent_t if too many exist or moves it to the free list for later use.
	*/
	static void
	task_done(taskq_t tq, taskq_ent_t t)
	{
	ASSERT(tq);
	ASSERT(t);

	/* Wake tasks blocked in taskq_wait_id() */
	wake_up_all(&t->tqent_waitq);

	list_del_init(&t->tqent_list);

	if (tq->tq_nalloc <= tq->tq_minalloc) {
	t->tqent_id = TASKQID_INVALID;
	t->tqent_func = NULL;
	t->tqent_arg = NULL;
	t->tqent_flags = 0;

	list_add_tail(&t->tqent_list, &tq->tq_free_list);
	} else {
	task_free(tq, t);
	}
	}

	/*
	* When a delayed task timer expires remove it from the delay list and
	* add it to the priority list in order for immediate processing.
	*/
	static void
	task_expire_impl(taskq_ent_t *t)
	{
	taskq_ent_t *w;
	taskq_t *tq = t->tqent_taskq;
	struct list_head *l = NULL;
	unsigned long flags;

	spin_lock_irqsave_nested(&tq->tq_lock, flags, tq->tq_lock_class);

	if (t->tqent_flags & TQENT_FLAG_CANCEL) {
	ASSERT(list_empty(&t->tqent_list));
	spin_unlock_irqrestore(&tq->tq_lock, flags);
	return;
	}

	t->tqent_birth = jiffies;
	DTRACE_PROBE1(taskq_ent__birth, taskq_ent_t *, t);

	/*
	* The priority list must be maintained in strict task id order
	* from lowest to highest for lowest_id to be easily calculable.
	*/
	list_del(&t->tqent_list);
	list_for_each_prev(l, &tq->tq_prio_list) {
	w = list_entry(l, taskq_ent_t, tqent_list);
	if (w->tqent_id < t->tqent_id) {
	list_add(&t->tqent_list, l);
	break;
	}
	}
	if (l == &tq->tq_prio_list)
	list_add(&t->tqent_list, &tq->tq_prio_list);

	spin_unlock_irqrestore(&tq->tq_lock, flags);

	wake_up(&tq->tq_work_waitq);
	}

	static void
	task_expire(spl_timer_list_t tl)
	{
	struct timer_list tmr = (struct timer_list )tl;
	taskq_ent_t *t = from_timer(t, tmr, tqent_timer);
	task_expire_impl(t);
	}

	/*
	* Returns the lowest incomplete taskqid_t. The taskqid_t may
	* be queued on the pending list, on the priority list, on the
	* delay list, or on the work list currently being handled, but
	* it is not 100% complete yet.
	*/
	static taskqid_t
	taskq_lowest_id(taskq_t *tq)
	{
	taskqid_t lowest_id = tq->tq_next_id;
	taskq_ent_t *t;
	taskq_thread_t *tqt;

	if (!list_empty(&tq->tq_pend_list)) {
	t = list_entry(tq->tq_pend_list.next, taskq_ent_t, tqent_list);
	lowest_id = MIN(lowest_id, t->tqent_id);
	}

	if (!list_empty(&tq->tq_prio_list)) {
	t = list_entry(tq->tq_prio_list.next, taskq_ent_t, tqent_list);
	lowest_id = MIN(lowest_id, t->tqent_id);
	}

	if (!list_empty(&tq->tq_delay_list)) {
	t = list_entry(tq->tq_delay_list.next, taskq_ent_t, tqent_list);
	lowest_id = MIN(lowest_id, t->tqent_id);
	}

	if (!list_empty(&tq->tq_active_list)) {
	tqt = list_entry(tq->tq_active_list.next, taskq_thread_t,
	tqt_active_list);
	ASSERT(tqt->tqt_id != TASKQID_INVALID);
	lowest_id = MIN(lowest_id, tqt->tqt_id);
	}

	return (lowest_id);
	}

	/*
	* Insert a task into a list keeping the list sorted by increasing taskqid.
	*/
	static void
	taskq_insert_in_order(taskq_t tq, taskq_thread_t tqt)
	{
	taskq_thread_t *w;
	struct list_head *l = NULL;

	ASSERT(tq);
	ASSERT(tqt);

	list_for_each_prev(l, &tq->tq_active_list) {
	w = list_entry(l, taskq_thread_t, tqt_active_list);
	if (w->tqt_id < tqt->tqt_id) {
	list_add(&tqt->tqt_active_list, l);
	break;
	}
	}
	if (l == &tq->tq_active_list)
	list_add(&tqt->tqt_active_list, &tq->tq_active_list);
	}

	/*
	* Find and return a task from the given list if it exists. The list
	* must be in lowest to highest task id order.
	*/
	static taskq_ent_t *
	taskq_find_list(taskq_t tq, struct list_head lh, taskqid_t id)
	{
	struct list_head *l = NULL;
	taskq_ent_t *t;

	list_for_each(l, lh) {
	t = list_entry(l, taskq_ent_t, tqent_list);

	if (t->tqent_id == id)
	return (t);

	if (t->tqent_id > id)
	break;
	}

	return (NULL);
	}

	/*
	* Find an already dispatched task given the task id regardless of what
	* state it is in. If a task is still pending it will be returned.
	* If a task is executing, then -EBUSY will be returned instead.
	* If the task has already been run then NULL is returned.
	*/
	static taskq_ent_t *
	taskq_find(taskq_t *tq, taskqid_t id)
	{
	taskq_thread_t *tqt;
	struct list_head *l = NULL;
	taskq_ent_t *t;

	t = taskq_find_list(tq, &tq->tq_delay_list, id);
	if (t)
	return (t);

	t = taskq_find_list(tq, &tq->tq_prio_list, id);
	if (t)
	return (t);

	t = taskq_find_list(tq, &tq->tq_pend_list, id);
	if (t)
	return (t);

	list_for_each(l, &tq->tq_active_list) {
	tqt = list_entry(l, taskq_thread_t, tqt_active_list);
	if (tqt->tqt_id == id) {
	/*
	* Instead of returning tqt_task, we just return a non
	* NULL value to prevent misuse, since tqt_task only
	* has two valid fields.
	*/
	return (ERR_PTR(-EBUSY));
	}
	}

	return (NULL);
	}

	/*
	* Theory for the taskq_wait_id(), taskq_wait_outstanding(), and
	* taskq_wait() functions below.
	*
	* Taskq waiting is accomplished by tracking the lowest outstanding task
	* id and the next available task id. As tasks are dispatched they are
	* added to the tail of the pending, priority, or delay lists. As worker
	* threads become available the tasks are removed from the heads of these
	* lists and linked to the worker threads. This ensures the lists are
	* kept sorted by lowest to highest task id.
	*
	* Therefore the lowest outstanding task id can be quickly determined by
	* checking the head item from all of these lists. This value is stored
	* with the taskq as the lowest id. It only needs to be recalculated when
	* either the task with the current lowest id completes or is canceled.
	*
	* By blocking until the lowest task id exceeds the passed task id the
	* taskq_wait_outstanding() function can be easily implemented. Similarly,
	* by blocking until the lowest task id matches the next task id taskq_wait()
	* can be implemented.
	*
	* Callers should be aware that when there are multiple worked threads it
	* is possible for larger task ids to complete before smaller ones. Also
	* when the taskq contains delay tasks with small task ids callers may
	* block for a considerable length of time waiting for them to expire and
	* execute.
	*/
	static int
	taskq_wait_id_check(taskq_t *tq, taskqid_t id)
	{
	int rc;
	unsigned long flags;

	spin_lock_irqsave_nested(&tq->tq_lock, flags, tq->tq_lock_class);
	rc = (taskq_find(tq, id) == NULL);
	spin_unlock_irqrestore(&tq->tq_lock, flags);

	return (rc);
	}

	/*
	* The taskq_wait_id() function blocks until the passed task id completes.
	* This does not guarantee that all lower task ids have completed.
	*/
	void
	taskq_wait_id(taskq_t *tq, taskqid_t id)
	{
	wait_event(tq->tq_wait_waitq, taskq_wait_id_check(tq, id));
	}
	EXPORT_SYMBOL(taskq_wait_id);

	static int
	taskq_wait_outstanding_check(taskq_t *tq, taskqid_t id)
	{
	int rc;
	unsigned long flags;

	spin_lock_irqsave_nested(&tq->tq_lock, flags, tq->tq_lock_class);
	rc = (id < tq->tq_lowest_id);
	spin_unlock_irqrestore(&tq->tq_lock, flags);

	return (rc);
	}

	/*
	* The taskq_wait_outstanding() function will block until all tasks with a
	* lower taskqid than the passed 'id' have been completed. Note that all
	* task id's are assigned monotonically at dispatch time. Zero may be
	* passed for the id to indicate all tasks dispatch up to this point,
	* but not after, should be waited for.
	*/
	void
	taskq_wait_outstanding(taskq_t *tq, taskqid_t id)
	{
	id = id ? id : tq->tq_next_id - 1;
	wait_event(tq->tq_wait_waitq, taskq_wait_outstanding_check(tq, id));
	}
	EXPORT_SYMBOL(taskq_wait_outstanding);

	static int
	taskq_wait_check(taskq_t *tq)
	{
	int rc;
	unsigned long flags;

	spin_lock_irqsave_nested(&tq->tq_lock, flags, tq->tq_lock_class);
	rc = (tq->tq_lowest_id == tq->tq_next_id);
	spin_unlock_irqrestore(&tq->tq_lock, flags);

	return (rc);
	}

	/*
	* The taskq_wait() function will block until the taskq is empty.
	* This means that if a taskq re-dispatches work to itself taskq_wait()
	* callers will block indefinitely.
	*/
	void
	taskq_wait(taskq_t *tq)
	{
	wait_event(tq->tq_wait_waitq, taskq_wait_check(tq));
	}
	EXPORT_SYMBOL(taskq_wait);

	int
	taskq_member(taskq_t tq, kthread_t t)
	{
	return (tq == (taskq_t *)tsd_get_by_thread(taskq_tsd, t));
	}
	EXPORT_SYMBOL(taskq_member);

	taskq_t *
	taskq_of_curthread(void)
	{
	return (tsd_get(taskq_tsd));
	}
	EXPORT_SYMBOL(taskq_of_curthread);

	/*
	* Cancel an already dispatched task given the task id. Still pending tasks
	* will be immediately canceled, and if the task is active the function will
	* block until it completes. Preallocated tasks which are canceled must be
	* freed by the caller.
	*/
	int
	taskq_cancel_id(taskq_t *tq, taskqid_t id)
	{
	taskq_ent_t *t;
	int rc = ENOENT;
	unsigned long flags;

	ASSERT(tq);

	spin_lock_irqsave_nested(&tq->tq_lock, flags, tq->tq_lock_class);
	t = taskq_find(tq, id);
	if (t && t != ERR_PTR(-EBUSY)) {
	list_del_init(&t->tqent_list);
	t->tqent_flags \|= TQENT_FLAG_CANCEL;

	/*
	* When canceling the lowest outstanding task id we
	* must recalculate the new lowest outstanding id.
	*/
	if (tq->tq_lowest_id == t->tqent_id) {
	tq->tq_lowest_id = taskq_lowest_id(tq);
	ASSERT3S(tq->tq_lowest_id, >, t->tqent_id);
	}

	/*
	* The task_expire() function takes the tq->tq_lock so drop
	* drop the lock before synchronously cancelling the timer.
	*/
	if (timer_pending(&t->tqent_timer)) {
	spin_unlock_irqrestore(&tq->tq_lock, flags);
	del_timer_sync(&t->tqent_timer);
	spin_lock_irqsave_nested(&tq->tq_lock, flags,
	tq->tq_lock_class);
	}

	if (!(t->tqent_flags & TQENT_FLAG_PREALLOC))
	task_done(tq, t);

	rc = 0;
	}
	spin_unlock_irqrestore(&tq->tq_lock, flags);

	if (t == ERR_PTR(-EBUSY)) {
	taskq_wait_id(tq, id);
	rc = EBUSY;
	}

	return (rc);
	}
	EXPORT_SYMBOL(taskq_cancel_id);

	static int taskq_thread_spawn(taskq_t *tq);

	taskqid_t
	taskq_dispatch(taskq_t tq, task_func_t func, void arg, uint_t flags)
	{
	taskq_ent_t *t;
	taskqid_t rc = TASKQID_INVALID;
	unsigned long irqflags;

	ASSERT(tq);
	ASSERT(func);

	spin_lock_irqsave_nested(&tq->tq_lock, irqflags, tq->tq_lock_class);

	/* Taskq being destroyed and all tasks drained */
	if (!(tq->tq_flags & TASKQ_ACTIVE))
	goto out;

	/* Do not queue the task unless there is idle thread for it */
	ASSERT(tq->tq_nactive <= tq->tq_nthreads);
	if ((flags & TQ_NOQUEUE) && (tq->tq_nactive == tq->tq_nthreads)) {
	/* Dynamic taskq may be able to spawn another thread */
	if (!(tq->tq_flags & TASKQ_DYNAMIC) \|\|
	taskq_thread_spawn(tq) == 0)
	goto out;
	}

	if ((t = task_alloc(tq, flags, &irqflags)) == NULL)
	goto out;

	spin_lock(&t->tqent_lock);

	/* Queue to the front of the list to enforce TQ_NOQUEUE semantics */
	if (flags & TQ_NOQUEUE)
	list_add(&t->tqent_list, &tq->tq_prio_list);
	/* Queue to the priority list instead of the pending list */
	else if (flags & TQ_FRONT)
	list_add_tail(&t->tqent_list, &tq->tq_prio_list);
	else
	list_add_tail(&t->tqent_list, &tq->tq_pend_list);

	t->tqent_id = rc = tq->tq_next_id;
	tq->tq_next_id++;
	t->tqent_func = func;
	t->tqent_arg = arg;
	t->tqent_taskq = tq;
	t->tqent_timer.function = NULL;
	t->tqent_timer.expires = 0;

	t->tqent_birth = jiffies;
	DTRACE_PROBE1(taskq_ent__birth, taskq_ent_t *, t);

	ASSERT(!(t->tqent_flags & TQENT_FLAG_PREALLOC));

	spin_unlock(&t->tqent_lock);

	wake_up(&tq->tq_work_waitq);
	out:
	/* Spawn additional taskq threads if required. */
	if (!(flags & TQ_NOQUEUE) && tq->tq_nactive == tq->tq_nthreads)
	(void) taskq_thread_spawn(tq);

	spin_unlock_irqrestore(&tq->tq_lock, irqflags);
	return (rc);
	}
	EXPORT_SYMBOL(taskq_dispatch);

	taskqid_t
	taskq_dispatch_delay(taskq_t tq, task_func_t func, void arg,
	uint_t flags, clock_t expire_time)
	{
	taskqid_t rc = TASKQID_INVALID;
	taskq_ent_t *t;
	unsigned long irqflags;

	ASSERT(tq);
	ASSERT(func);

	spin_lock_irqsave_nested(&tq->tq_lock, irqflags, tq->tq_lock_class);

	/* Taskq being destroyed and all tasks drained */
	if (!(tq->tq_flags & TASKQ_ACTIVE))
	goto out;

	if ((t = task_alloc(tq, flags, &irqflags)) == NULL)
	goto out;

	spin_lock(&t->tqent_lock);

	/* Queue to the delay list for subsequent execution */
	list_add_tail(&t->tqent_list, &tq->tq_delay_list);

	t->tqent_id = rc = tq->tq_next_id;
	tq->tq_next_id++;
	t->tqent_func = func;
	t->tqent_arg = arg;
	t->tqent_taskq = tq;
	t->tqent_timer.function = task_expire;
	t->tqent_timer.expires = (unsigned long)expire_time;
	add_timer(&t->tqent_timer);

	ASSERT(!(t->tqent_flags & TQENT_FLAG_PREALLOC));

	spin_unlock(&t->tqent_lock);
	out:
	/* Spawn additional taskq threads if required. */
	if (tq->tq_nactive == tq->tq_nthreads)
	(void) taskq_thread_spawn(tq);
	spin_unlock_irqrestore(&tq->tq_lock, irqflags);
	return (rc);
	}
	EXPORT_SYMBOL(taskq_dispatch_delay);

	void
	taskq_dispatch_ent(taskq_t tq, task_func_t func, void arg, uint_t flags,
	taskq_ent_t *t)
	{
	unsigned long irqflags;
	ASSERT(tq);
	ASSERT(func);

	spin_lock_irqsave_nested(&tq->tq_lock, irqflags,
	tq->tq_lock_class);

	/* Taskq being destroyed and all tasks drained */
	if (!(tq->tq_flags & TASKQ_ACTIVE)) {
	t->tqent_id = TASKQID_INVALID;
	goto out;
	}

	if ((flags & TQ_NOQUEUE) && (tq->tq_nactive == tq->tq_nthreads)) {
	/* Dynamic taskq may be able to spawn another thread */
	if (!(tq->tq_flags & TASKQ_DYNAMIC) \|\|
	taskq_thread_spawn(tq) == 0)
	goto out2;
	flags \|= TQ_FRONT;
	}

	spin_lock(&t->tqent_lock);

	/*
	* Make sure the entry is not on some other taskq; it is important to
	* ASSERT() under lock
	*/
	ASSERT(taskq_empty_ent(t));

	/*
	* Mark it as a prealloc'd task. This is important
	* to ensure that we don't free it later.
	*/
	t->tqent_flags \|= TQENT_FLAG_PREALLOC;

	/* Queue to the priority list instead of the pending list */
	if (flags & TQ_FRONT)
	list_add_tail(&t->tqent_list, &tq->tq_prio_list);
	else
	list_add_tail(&t->tqent_list, &tq->tq_pend_list);

	t->tqent_id = tq->tq_next_id;
	tq->tq_next_id++;
	t->tqent_func = func;
	t->tqent_arg = arg;
	t->tqent_taskq = tq;

	t->tqent_birth = jiffies;
	DTRACE_PROBE1(taskq_ent__birth, taskq_ent_t *, t);

	spin_unlock(&t->tqent_lock);

	wake_up(&tq->tq_work_waitq);
	out:
	/* Spawn additional taskq threads if required. */
	if (tq->tq_nactive == tq->tq_nthreads)
	(void) taskq_thread_spawn(tq);
	out2:
	spin_unlock_irqrestore(&tq->tq_lock, irqflags);
	}
	EXPORT_SYMBOL(taskq_dispatch_ent);

	int
	taskq_empty_ent(taskq_ent_t *t)
	{
	return (list_empty(&t->tqent_list));
	}
	EXPORT_SYMBOL(taskq_empty_ent);

	void
	taskq_init_ent(taskq_ent_t *t)
	{
	spin_lock_init(&t->tqent_lock);
	init_waitqueue_head(&t->tqent_waitq);
	timer_setup(&t->tqent_timer, NULL, 0);
	INIT_LIST_HEAD(&t->tqent_list);
	t->tqent_id = 0;
	t->tqent_func = NULL;
	t->tqent_arg = NULL;
	t->tqent_flags = 0;
	t->tqent_taskq = NULL;
	}
	EXPORT_SYMBOL(taskq_init_ent);

	/*
	* Return the next pending task, preference is given to tasks on the
	* priority list which were dispatched with TQ_FRONT.
	*/
	static taskq_ent_t *
	taskq_next_ent(taskq_t *tq)
	{
	struct list_head *list;

	if (!list_empty(&tq->tq_prio_list))
	list = &tq->tq_prio_list;
	else if (!list_empty(&tq->tq_pend_list))
	list = &tq->tq_pend_list;
	else
	return (NULL);

	return (list_entry(list->next, taskq_ent_t, tqent_list));
	}

	/*
	* Spawns a new thread for the specified taskq.
	*/
	static void
	taskq_thread_spawn_task(void *arg)
	{
	taskq_t tq = (taskq_t )arg;
	unsigned long flags;

	if (taskq_thread_create(tq) == NULL) {
	/* restore spawning count if failed */
	spin_lock_irqsave_nested(&tq->tq_lock, flags,
	tq->tq_lock_class);
	tq->tq_nspawn--;
	spin_unlock_irqrestore(&tq->tq_lock, flags);
	}
	}

	/*
	* Spawn addition threads for dynamic taskqs (TASKQ_DYNAMIC) the current
	* number of threads is insufficient to handle the pending tasks. These
	* new threads must be created by the dedicated dynamic_taskq to avoid
	* deadlocks between thread creation and memory reclaim. The system_taskq
	* which is also a dynamic taskq cannot be safely used for this.
	*/
	static int
	taskq_thread_spawn(taskq_t *tq)
	{
	int spawning = 0;

	if (!(tq->tq_flags & TASKQ_DYNAMIC))
	return (0);

	if ((tq->tq_nthreads + tq->tq_nspawn < tq->tq_maxthreads) &&
	(tq->tq_flags & TASKQ_ACTIVE)) {
	spawning = (++tq->tq_nspawn);
	taskq_dispatch(dynamic_taskq, taskq_thread_spawn_task,
	tq, TQ_NOSLEEP);
	}

	return (spawning);
	}

	/*
	* Threads in a dynamic taskq should only exit once it has been completely
	* drained and no other threads are actively servicing tasks. This prevents
	* threads from being created and destroyed more than is required.
	*
	* The first thread is the thread list is treated as the primary thread.
	* There is nothing special about the primary thread but in order to avoid
	* all the taskq pids from changing we opt to make it long running.
	*/
	static int
	taskq_thread_should_stop(taskq_t tq, taskq_thread_t tqt)
	{
	if (!(tq->tq_flags & TASKQ_DYNAMIC))
	return (0);

	if (list_first_entry(&(tq->tq_thread_list), taskq_thread_t,
	tqt_thread_list) == tqt)
	return (0);

	return
	((tq->tq_nspawn == 0) && /* No threads are being spawned */
	(tq->tq_nactive == 0) && /* No threads are handling tasks */
	(tq->tq_nthreads > 1) && /* More than 1 thread is running */
	(!taskq_next_ent(tq)) && /* There are no pending tasks */
	(spl_taskq_thread_dynamic)); /* Dynamic taskqs are allowed */
	}

	static int
	taskq_thread(void *args)
	{
	DECLARE_WAITQUEUE(wait, current);
	sigset_t blocked;
	taskq_thread_t *tqt = args;
	taskq_t *tq;
	taskq_ent_t *t;
	int seq_tasks = 0;
	unsigned long flags;
	taskq_ent_t dup_task = {};

	ASSERT(tqt);
	ASSERT(tqt->tqt_tq);
	tq = tqt->tqt_tq;
	current->flags \|= PF_NOFREEZE;

	(void) spl_fstrans_mark();

	sigfillset(&blocked);
	sigprocmask(SIG_BLOCK, &blocked, NULL);
	flush_signals(current);

	tsd_set(taskq_tsd, tq);
	spin_lock_irqsave_nested(&tq->tq_lock, flags, tq->tq_lock_class);
	/*
	* If we are dynamically spawned, decrease spawning count. Note that
	* we could be created during taskq_create, in which case we shouldn't
	* do the decrement. But it's fine because taskq_create will reset
	* tq_nspawn later.
	*/
	if (tq->tq_flags & TASKQ_DYNAMIC)
	tq->tq_nspawn--;

	/* Immediately exit if more threads than allowed were created. */
	if (tq->tq_nthreads >= tq->tq_maxthreads)
	goto error;

	tq->tq_nthreads++;
	list_add_tail(&tqt->tqt_thread_list, &tq->tq_thread_list);
	wake_up(&tq->tq_wait_waitq);
	set_current_state(TASK_INTERRUPTIBLE);

	while (!kthread_should_stop()) {

	if (list_empty(&tq->tq_pend_list) &&
	list_empty(&tq->tq_prio_list)) {

	if (taskq_thread_should_stop(tq, tqt)) {
	wake_up_all(&tq->tq_wait_waitq);
	break;
	}

	add_wait_queue_exclusive(&tq->tq_work_waitq, &wait);
	spin_unlock_irqrestore(&tq->tq_lock, flags);

	schedule();
	seq_tasks = 0;

	spin_lock_irqsave_nested(&tq->tq_lock, flags,
	tq->tq_lock_class);
	remove_wait_queue(&tq->tq_work_waitq, &wait);
	} else {
	__set_current_state(TASK_RUNNING);
	}

	if ((t = taskq_next_ent(tq)) != NULL) {
	list_del_init(&t->tqent_list);

	/*
	* A TQENT_FLAG_PREALLOC task may be reused or freed
	* during the task function call. Store tqent_id and
	* tqent_flags here.
	*
	* Also use an on stack taskq_ent_t for tqt_task
	* assignment in this case; we want to make sure
	* to duplicate all fields, so the values are
	* correct when it's accessed via DTRACE_PROBE*.
	*/
	tqt->tqt_id = t->tqent_id;
	tqt->tqt_flags = t->tqent_flags;

	if (t->tqent_flags & TQENT_FLAG_PREALLOC) {
	dup_task = *t;
	t = &dup_task;
	}
	tqt->tqt_task = t;

	taskq_insert_in_order(tq, tqt);
	tq->tq_nactive++;
	spin_unlock_irqrestore(&tq->tq_lock, flags);

	DTRACE_PROBE1(taskq_ent__start, taskq_ent_t *, t);

	/* Perform the requested task */
	t->tqent_func(t->tqent_arg);

	DTRACE_PROBE1(taskq_ent__finish, taskq_ent_t *, t);

	spin_lock_irqsave_nested(&tq->tq_lock, flags,
	tq->tq_lock_class);
	tq->tq_nactive--;
	list_del_init(&tqt->tqt_active_list);
	tqt->tqt_task = NULL;

	/* For prealloc'd tasks, we don't free anything. */
	if (!(tqt->tqt_flags & TQENT_FLAG_PREALLOC))
	task_done(tq, t);

	/*
	* When the current lowest outstanding taskqid is
	* done calculate the new lowest outstanding id
	*/
	if (tq->tq_lowest_id == tqt->tqt_id) {
	tq->tq_lowest_id = taskq_lowest_id(tq);
	ASSERT3S(tq->tq_lowest_id, >, tqt->tqt_id);
	}

	/* Spawn additional taskq threads if required. */
	if ((++seq_tasks) > spl_taskq_thread_sequential &&
	taskq_thread_spawn(tq))
	seq_tasks = 0;

	tqt->tqt_id = TASKQID_INVALID;
	tqt->tqt_flags = 0;
	wake_up_all(&tq->tq_wait_waitq);
	} else {
	if (taskq_thread_should_stop(tq, tqt))
	break;
	}

	set_current_state(TASK_INTERRUPTIBLE);

	}

	__set_current_state(TASK_RUNNING);
	tq->tq_nthreads--;
	list_del_init(&tqt->tqt_thread_list);
	error:
	kmem_free(tqt, sizeof (taskq_thread_t));
	spin_unlock_irqrestore(&tq->tq_lock, flags);

	tsd_set(taskq_tsd, NULL);
	thread_exit();

	return (0);
	}

	static taskq_thread_t *
	taskq_thread_create(taskq_t *tq)
	{
	static int last_used_cpu = 0;
	taskq_thread_t *tqt;

	tqt = kmem_alloc(sizeof (*tqt), KM_PUSHPAGE);
	INIT_LIST_HEAD(&tqt->tqt_thread_list);
	INIT_LIST_HEAD(&tqt->tqt_active_list);
	tqt->tqt_tq = tq;
	tqt->tqt_id = TASKQID_INVALID;

	tqt->tqt_thread = spl_kthread_create(taskq_thread, tqt,
	"%s", tq->tq_name);
	if (tqt->tqt_thread == NULL) {
	kmem_free(tqt, sizeof (taskq_thread_t));
	return (NULL);
	}

	if (spl_taskq_thread_bind) {
	last_used_cpu = (last_used_cpu + 1) % num_online_cpus();
	kthread_bind(tqt->tqt_thread, last_used_cpu);
	}

	if (spl_taskq_thread_priority)
	set_user_nice(tqt->tqt_thread, PRIO_TO_NICE(tq->tq_pri));

	wake_up_process(tqt->tqt_thread);

	return (tqt);
	}

	taskq_t *
	taskq_create(const char *name, int threads_arg, pri_t pri,
	int minalloc, int maxalloc, uint_t flags)
	{
	taskq_t *tq;
	taskq_thread_t *tqt;
	int count = 0, rc = 0, i;
	unsigned long irqflags;
	int nthreads = threads_arg;

	ASSERT(name != NULL);
	ASSERT(minalloc >= 0);
	ASSERT(maxalloc <= INT_MAX);
	ASSERT(!(flags & (TASKQ_CPR_SAFE))); /* Unsupported */

	/* Scale the number of threads using nthreads as a percentage */
	if (flags & TASKQ_THREADS_CPU_PCT) {
	ASSERT(nthreads <= 100);
	ASSERT(nthreads >= 0);
	nthreads = MIN(threads_arg, 100);
	nthreads = MAX(nthreads, 0);
	nthreads = MAX((num_online_cpus() * nthreads) /100, 1);
	}

	tq = kmem_alloc(sizeof (*tq), KM_PUSHPAGE);
	if (tq == NULL)
	return (NULL);

	tq->tq_hp_support = B_FALSE;
	#ifdef HAVE_CPU_HOTPLUG
	if (flags & TASKQ_THREADS_CPU_PCT) {
	tq->tq_hp_support = B_TRUE;
	if (cpuhp_state_add_instance_nocalls(spl_taskq_cpuhp_state,
	&tq->tq_hp_cb_node) != 0) {
	kmem_free(tq, sizeof (*tq));
	return (NULL);
	}
	}
	#endif

	spin_lock_init(&tq->tq_lock);
	INIT_LIST_HEAD(&tq->tq_thread_list);
	INIT_LIST_HEAD(&tq->tq_active_list);
	tq->tq_name = kmem_strdup(name);
	tq->tq_nactive = 0;
	tq->tq_nthreads = 0;
	tq->tq_nspawn = 0;
	tq->tq_maxthreads = nthreads;
	tq->tq_cpu_pct = threads_arg;
	tq->tq_pri = pri;
	tq->tq_minalloc = minalloc;
	tq->tq_maxalloc = maxalloc;
	tq->tq_nalloc = 0;
	tq->tq_flags = (flags \| TASKQ_ACTIVE);
	tq->tq_next_id = TASKQID_INITIAL;
	tq->tq_lowest_id = TASKQID_INITIAL;
	INIT_LIST_HEAD(&tq->tq_free_list);
	INIT_LIST_HEAD(&tq->tq_pend_list);
	INIT_LIST_HEAD(&tq->tq_prio_list);
	INIT_LIST_HEAD(&tq->tq_delay_list);
	init_waitqueue_head(&tq->tq_work_waitq);
	init_waitqueue_head(&tq->tq_wait_waitq);
	tq->tq_lock_class = TQ_LOCK_GENERAL;
	INIT_LIST_HEAD(&tq->tq_taskqs);

	if (flags & TASKQ_PREPOPULATE) {
	spin_lock_irqsave_nested(&tq->tq_lock, irqflags,
	tq->tq_lock_class);

	for (i = 0; i < minalloc; i++)
	task_done(tq, task_alloc(tq, TQ_PUSHPAGE \| TQ_NEW,
	&irqflags));

	spin_unlock_irqrestore(&tq->tq_lock, irqflags);
	}

	if ((flags & TASKQ_DYNAMIC) && spl_taskq_thread_dynamic)
	nthreads = 1;

	for (i = 0; i < nthreads; i++) {
	tqt = taskq_thread_create(tq);
	if (tqt == NULL)
	rc = 1;
	else
	count++;
	}

	/* Wait for all threads to be started before potential destroy */
	wait_event(tq->tq_wait_waitq, tq->tq_nthreads == count);
	/*
	* taskq_thread might have touched nspawn, but we don't want them to
	* because they're not dynamically spawned. So we reset it to 0
	*/
	tq->tq_nspawn = 0;

	if (rc) {
	taskq_destroy(tq);
	tq = NULL;
	} else {
	down_write(&tq_list_sem);
	tq->tq_instance = taskq_find_by_name(name) + 1;
	list_add_tail(&tq->tq_taskqs, &tq_list);
	up_write(&tq_list_sem);
	}

	return (tq);
	}
	EXPORT_SYMBOL(taskq_create);

	void
	taskq_destroy(taskq_t *tq)
	{
	struct task_struct *thread;
	taskq_thread_t *tqt;
	taskq_ent_t *t;
	unsigned long flags;

	ASSERT(tq);
	spin_lock_irqsave_nested(&tq->tq_lock, flags, tq->tq_lock_class);
	tq->tq_flags &= ~TASKQ_ACTIVE;
	spin_unlock_irqrestore(&tq->tq_lock, flags);

	#ifdef HAVE_CPU_HOTPLUG
	if (tq->tq_hp_support) {
	VERIFY0(cpuhp_state_remove_instance_nocalls(
	spl_taskq_cpuhp_state, &tq->tq_hp_cb_node));
	}
	#endif
	/*
	* When TASKQ_ACTIVE is clear new tasks may not be added nor may
	* new worker threads be spawned for dynamic taskq.
	*/
	if (dynamic_taskq != NULL)
	taskq_wait_outstanding(dynamic_taskq, 0);

	taskq_wait(tq);

	/* remove taskq from global list used by the kstats */
	down_write(&tq_list_sem);
	list_del(&tq->tq_taskqs);
	up_write(&tq_list_sem);

	spin_lock_irqsave_nested(&tq->tq_lock, flags, tq->tq_lock_class);
	/* wait for spawning threads to insert themselves to the list */
	while (tq->tq_nspawn) {
	spin_unlock_irqrestore(&tq->tq_lock, flags);
	schedule_timeout_interruptible(1);
	spin_lock_irqsave_nested(&tq->tq_lock, flags,
	tq->tq_lock_class);
	}

	/*
	* Signal each thread to exit and block until it does. Each thread
	* is responsible for removing itself from the list and freeing its
	* taskq_thread_t. This allows for idle threads to opt to remove
	* themselves from the taskq. They can be recreated as needed.
	*/
	while (!list_empty(&tq->tq_thread_list)) {
	tqt = list_entry(tq->tq_thread_list.next,
	taskq_thread_t, tqt_thread_list);
	thread = tqt->tqt_thread;
	spin_unlock_irqrestore(&tq->tq_lock, flags);

	kthread_stop(thread);

	spin_lock_irqsave_nested(&tq->tq_lock, flags,
	tq->tq_lock_class);
	}

	while (!list_empty(&tq->tq_free_list)) {
	t = list_entry(tq->tq_free_list.next, taskq_ent_t, tqent_list);

	ASSERT(!(t->tqent_flags & TQENT_FLAG_PREALLOC));

	list_del_init(&t->tqent_list);
	task_free(tq, t);
	}

	ASSERT0(tq->tq_nthreads);
	ASSERT0(tq->tq_nalloc);
	ASSERT0(tq->tq_nspawn);
	ASSERT(list_empty(&tq->tq_thread_list));
	ASSERT(list_empty(&tq->tq_active_list));
	ASSERT(list_empty(&tq->tq_free_list));
	ASSERT(list_empty(&tq->tq_pend_list));
	ASSERT(list_empty(&tq->tq_prio_list));
	ASSERT(list_empty(&tq->tq_delay_list));

	spin_unlock_irqrestore(&tq->tq_lock, flags);

	kmem_strfree(tq->tq_name);
	kmem_free(tq, sizeof (taskq_t));
	}
	EXPORT_SYMBOL(taskq_destroy);

	static unsigned int spl_taskq_kick = 0;

	/*
	* 2.6.36 API Change
	* module_param_cb is introduced to take kernel_param_ops and
	* module_param_call is marked as obsolete. Also set and get operations
	* were changed to take a 'const struct kernel_param *'.
	*/
	static int
	#ifdef module_param_cb
	param_set_taskq_kick(const char val, const struct kernel_param kp)
	#else
	param_set_taskq_kick(const char val, struct kernel_param kp)
	#endif
	{
	int ret;
	taskq_t *tq = NULL;
	taskq_ent_t *t;
	unsigned long flags;

	ret = param_set_uint(val, kp);
	if (ret < 0 \|\| !spl_taskq_kick)
	return (ret);
	/* reset value */
	spl_taskq_kick = 0;

	down_read(&tq_list_sem);
	list_for_each_entry(tq, &tq_list, tq_taskqs) {
	spin_lock_irqsave_nested(&tq->tq_lock, flags,
	tq->tq_lock_class);
	/* Check if the first pending is older than 5 seconds */
	t = taskq_next_ent(tq);
	if (t && time_after(jiffies, t->tqent_birth + 5*HZ)) {
	(void) taskq_thread_spawn(tq);
	printk(KERN_INFO "spl: Kicked taskq %s/%d\n",
	tq->tq_name, tq->tq_instance);
	}
	spin_unlock_irqrestore(&tq->tq_lock, flags);
	}
	up_read(&tq_list_sem);
	return (ret);
	}

	#ifdef module_param_cb
	static const struct kernel_param_ops param_ops_taskq_kick = {
	.set = param_set_taskq_kick,
	.get = param_get_uint,
	};
	module_param_cb(spl_taskq_kick, &param_ops_taskq_kick, &spl_taskq_kick, 0644);
	#else
	module_param_call(spl_taskq_kick, param_set_taskq_kick, param_get_uint,
	&spl_taskq_kick, 0644);
	#endif
	MODULE_PARM_DESC(spl_taskq_kick,
	"Write nonzero to kick stuck taskqs to spawn more threads");

	#ifdef HAVE_CPU_HOTPLUG
	/*
	* This callback will be called exactly once for each core that comes online,
	* for each dynamic taskq. We attempt to expand taskqs that have
	* TASKQ_THREADS_CPU_PCT set. We need to redo the percentage calculation every
	* time, to correctly determine whether or not to add a thread.
	*/
	static int
	spl_taskq_expand(unsigned int cpu, struct hlist_node *node)
	{
	taskq_t *tq = list_entry(node, taskq_t, tq_hp_cb_node);
	unsigned long flags;
	int err = 0;

	ASSERT(tq);
	spin_lock_irqsave_nested(&tq->tq_lock, flags, tq->tq_lock_class);

	if (!(tq->tq_flags & TASKQ_ACTIVE)) {
	spin_unlock_irqrestore(&tq->tq_lock, flags);
	return (err);
	}

	ASSERT(tq->tq_flags & TASKQ_THREADS_CPU_PCT);
	int nthreads = MIN(tq->tq_cpu_pct, 100);
	nthreads = MAX(((num_online_cpus() + 1) * nthreads) / 100, 1);
	tq->tq_maxthreads = nthreads;

	if (!((tq->tq_flags & TASKQ_DYNAMIC) && spl_taskq_thread_dynamic) &&
	tq->tq_maxthreads > tq->tq_nthreads) {
	spin_unlock_irqrestore(&tq->tq_lock, flags);
	taskq_thread_t *tqt = taskq_thread_create(tq);
	if (tqt == NULL)
	err = -1;
	return (err);
	}
	spin_unlock_irqrestore(&tq->tq_lock, flags);
	return (err);
	}

	/*
	* While we don't support offlining CPUs, it is possible that CPUs will fail
	* to online successfully. We do need to be able to handle this case
	* gracefully.
	*/
	static int
	spl_taskq_prepare_down(unsigned int cpu, struct hlist_node *node)
	{
	taskq_t *tq = list_entry(node, taskq_t, tq_hp_cb_node);
	unsigned long flags;

	ASSERT(tq);
	spin_lock_irqsave_nested(&tq->tq_lock, flags, tq->tq_lock_class);

	if (!(tq->tq_flags & TASKQ_ACTIVE))
	goto out;

	ASSERT(tq->tq_flags & TASKQ_THREADS_CPU_PCT);
	int nthreads = MIN(tq->tq_cpu_pct, 100);
	nthreads = MAX(((num_online_cpus()) * nthreads) / 100, 1);
	tq->tq_maxthreads = nthreads;

	if (!((tq->tq_flags & TASKQ_DYNAMIC) && spl_taskq_thread_dynamic) &&
	tq->tq_maxthreads < tq->tq_nthreads) {
	ASSERT3U(tq->tq_maxthreads, ==, tq->tq_nthreads - 1);
	taskq_thread_t *tqt = list_entry(tq->tq_thread_list.next,
	taskq_thread_t, tqt_thread_list);
	struct task_struct *thread = tqt->tqt_thread;
	spin_unlock_irqrestore(&tq->tq_lock, flags);

	kthread_stop(thread);

	return (0);
	}

	out:
	spin_unlock_irqrestore(&tq->tq_lock, flags);
	return (0);
	}
	#endif

	int
	spl_taskq_init(void)
	{
	init_rwsem(&tq_list_sem);
	tsd_create(&taskq_tsd, NULL);

	#ifdef HAVE_CPU_HOTPLUG
	spl_taskq_cpuhp_state = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
	"fs/spl_taskq:online", spl_taskq_expand, spl_taskq_prepare_down);
	#endif

	system_taskq = taskq_create("spl_system_taskq", MAX(boot_ncpus, 64),
	maxclsyspri, boot_ncpus, INT_MAX, TASKQ_PREPOPULATE\|TASKQ_DYNAMIC);
	if (system_taskq == NULL)
	return (1);

	system_delay_taskq = taskq_create("spl_delay_taskq", MAX(boot_ncpus, 4),
	maxclsyspri, boot_ncpus, INT_MAX, TASKQ_PREPOPULATE\|TASKQ_DYNAMIC);
	if (system_delay_taskq == NULL) {
	#ifdef HAVE_CPU_HOTPLUG
	cpuhp_remove_multi_state(spl_taskq_cpuhp_state);
	#endif
	taskq_destroy(system_taskq);
	return (1);
	}

	dynamic_taskq = taskq_create("spl_dynamic_taskq", 1,
	maxclsyspri, boot_ncpus, INT_MAX, TASKQ_PREPOPULATE);
	if (dynamic_taskq == NULL) {
	#ifdef HAVE_CPU_HOTPLUG
	cpuhp_remove_multi_state(spl_taskq_cpuhp_state);
	#endif
	taskq_destroy(system_taskq);
	taskq_destroy(system_delay_taskq);
	return (1);
	}

	/*
	* This is used to annotate tq_lock, so
	* taskq_dispatch -> taskq_thread_spawn -> taskq_dispatch
	* does not trigger a lockdep warning re: possible recursive locking
	*/
	dynamic_taskq->tq_lock_class = TQ_LOCK_DYNAMIC;

	return (0);
	}

	void
	spl_taskq_fini(void)
	{
	taskq_destroy(dynamic_taskq);
	dynamic_taskq = NULL;

	taskq_destroy(system_delay_taskq);
	system_delay_taskq = NULL;

	taskq_destroy(system_taskq);
	system_taskq = NULL;

	tsd_destroy(&taskq_tsd);

	#ifdef HAVE_CPU_HOTPLUG
	cpuhp_remove_multi_state(spl_taskq_cpuhp_state);
	spl_taskq_cpuhp_state = 0;
	#endif
	}
	diff --git a/sys/contrib/openzfs/module/os/linux/zfs/vdev_disk.c b/sys/contrib/openzfs/module/os/linux/zfs/vdev_disk.c
	index 42144322a38f..46b459f5c80b 100644
	--- a/sys/contrib/openzfs/module/os/linux/zfs/vdev_disk.c
	+++ b/sys/contrib/openzfs/module/os/linux/zfs/vdev_disk.c
	@@ -1,930 +1,965 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (C) 2008-2010 Lawrence Livermore National Security, LLC.
	* Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
	* Rewritten for Linux by Brian Behlendorf <behlendorf1@llnl.gov>.
	* LLNL-CODE-403049.
	* Copyright (c) 2012, 2019 by Delphix. All rights reserved.
	*/

	#include <sys/zfs_context.h>
	#include <sys/spa_impl.h>
	#include <sys/vdev_disk.h>
	#include <sys/vdev_impl.h>
	#include <sys/vdev_trim.h>
	#include <sys/abd.h>
	#include <sys/fs/zfs.h>
	#include <sys/zio.h>
	#include <linux/blkpg.h>
	#include <linux/msdos_fs.h>
	#include <linux/vfs_compat.h>
	+#ifdef HAVE_LINUX_BLK_CGROUP_HEADER
	+#include <linux/blk-cgroup.h>
	+#endif

	typedef struct vdev_disk {
	struct block_device *vd_bdev;
	krwlock_t vd_lock;
	} vdev_disk_t;

	/*
	* Unique identifier for the exclusive vdev holder.
	*/
	static void *zfs_vdev_holder = VDEV_HOLDER;

	/*
	* Wait up to zfs_vdev_open_timeout_ms milliseconds before determining the
	* device is missing. The missing path may be transient since the links
	* can be briefly removed and recreated in response to udev events.
	*/
	static unsigned zfs_vdev_open_timeout_ms = 1000;

	/*
	* Size of the "reserved" partition, in blocks.
	*/
	#define EFI_MIN_RESV_SIZE (16 * 1024)

	/*
	* Virtual device vector for disks.
	*/
	typedef struct dio_request {
	zio_t dr_zio; / Parent ZIO */
	atomic_t dr_ref; /* References */
	int dr_error; /* Bio error */
	int dr_bio_count; /* Count of bio's */
	struct bio dr_bio[0]; / Attached bio's */
	} dio_request_t;

	static fmode_t
	vdev_bdev_mode(spa_mode_t spa_mode)
	{
	fmode_t mode = 0;

	if (spa_mode & SPA_MODE_READ)
	mode \|= FMODE_READ;

	if (spa_mode & SPA_MODE_WRITE)
	mode \|= FMODE_WRITE;

	return (mode);
	}

	/*
	* Returns the usable capacity (in bytes) for the partition or disk.
	*/
	static uint64_t
	bdev_capacity(struct block_device *bdev)
	{
	return (i_size_read(bdev->bd_inode));
	}

	#if !defined(HAVE_BDEV_WHOLE)
	static inline struct block_device *
	bdev_whole(struct block_device *bdev)
	{
	return (bdev->bd_contains);
	}
	#endif

	/*
	* Returns the maximum expansion capacity of the block device (in bytes).
	*
	* It is possible to expand a vdev when it has been created as a wholedisk
	* and the containing block device has increased in capacity. Or when the
	* partition containing the pool has been manually increased in size.
	*
	* This function is only responsible for calculating the potential expansion
	* size so it can be reported by 'zpool list'. The efi_use_whole_disk() is
	* responsible for verifying the expected partition layout in the wholedisk
	* case, and updating the partition table if appropriate. Once the partition
	* size has been increased the additional capacity will be visible using
	* bdev_capacity().
	*
	* The returned maximum expansion capacity is always expected to be larger, or
	* at the very least equal, to its usable capacity to prevent overestimating
	* the pool expandsize.
	*/
	static uint64_t
	bdev_max_capacity(struct block_device *bdev, uint64_t wholedisk)
	{
	uint64_t psize;
	int64_t available;

	if (wholedisk && bdev != bdev_whole(bdev)) {
	/*
	* When reporting maximum expansion capacity for a wholedisk
	* deduct any capacity which is expected to be lost due to
	* alignment restrictions. Over reporting this value isn't
	* harmful and would only result in slightly less capacity
	* than expected post expansion.
	* The estimated available space may be slightly smaller than
	* bdev_capacity() for devices where the number of sectors is
	* not a multiple of the alignment size and the partition layout
	* is keeping less than PARTITION_END_ALIGNMENT bytes after the
	* "reserved" EFI partition: in such cases return the device
	* usable capacity.
	*/
	available = i_size_read(bdev_whole(bdev)->bd_inode) -
	((EFI_MIN_RESV_SIZE + NEW_START_BLOCK +
	PARTITION_END_ALIGNMENT) << SECTOR_BITS);
	psize = MAX(available, bdev_capacity(bdev));
	} else {
	psize = bdev_capacity(bdev);
	}

	return (psize);
	}

	static void
	vdev_disk_error(zio_t *zio)
	{
	/*
	* This function can be called in interrupt context, for instance while
	* handling IRQs coming from a misbehaving disk device; use printk()
	* which is safe from any context.
	*/
	printk(KERN_WARNING "zio pool=%s vdev=%s error=%d type=%d "
	"offset=%llu size=%llu flags=%x\n", spa_name(zio->io_spa),
	zio->io_vd->vdev_path, zio->io_error, zio->io_type,
	(u_longlong_t)zio->io_offset, (u_longlong_t)zio->io_size,
	zio->io_flags);
	}

	static int
	vdev_disk_open(vdev_t v, uint64_t psize, uint64_t *max_psize,
	uint64_t logical_ashift, uint64_t physical_ashift)
	{
	struct block_device *bdev;
	fmode_t mode = vdev_bdev_mode(spa_mode(v->vdev_spa));
	hrtime_t timeout = MSEC2NSEC(zfs_vdev_open_timeout_ms);
	vdev_disk_t *vd;

	/* Must have a pathname and it must be absolute. */
	if (v->vdev_path == NULL \|\| v->vdev_path[0] != '/') {
	v->vdev_stat.vs_aux = VDEV_AUX_BAD_LABEL;
	vdev_dbgmsg(v, "invalid vdev_path");
	return (SET_ERROR(EINVAL));
	}

	/*
	* Reopen the device if it is currently open. When expanding a
	* partition force re-scanning the partition table if userland
	* did not take care of this already. We need to do this while closed
	* in order to get an accurate updated block device size. Then
	* since udev may need to recreate the device links increase the
	* open retry timeout before reporting the device as unavailable.
	*/
	vd = v->vdev_tsd;
	if (vd) {
	char disk_name[BDEVNAME_SIZE + 6] = "/dev/";
	boolean_t reread_part = B_FALSE;

	rw_enter(&vd->vd_lock, RW_WRITER);
	bdev = vd->vd_bdev;
	vd->vd_bdev = NULL;

	if (bdev) {
	if (v->vdev_expanding && bdev != bdev_whole(bdev)) {
	bdevname(bdev_whole(bdev), disk_name + 5);
	/*
	* If userland has BLKPG_RESIZE_PARTITION,
	* then it should have updated the partition
	* table already. We can detect this by
	* comparing our current physical size
	* with that of the device. If they are
	* the same, then we must not have
	* BLKPG_RESIZE_PARTITION or it failed to
	* update the partition table online. We
	* fallback to rescanning the partition
	* table from the kernel below. However,
	* if the capacity already reflects the
	* updated partition, then we skip
	* rescanning the partition table here.
	*/
	if (v->vdev_psize == bdev_capacity(bdev))
	reread_part = B_TRUE;
	}

	blkdev_put(bdev, mode \| FMODE_EXCL);
	}

	if (reread_part) {
	bdev = blkdev_get_by_path(disk_name, mode \| FMODE_EXCL,
	zfs_vdev_holder);
	if (!IS_ERR(bdev)) {
	int error = vdev_bdev_reread_part(bdev);
	blkdev_put(bdev, mode \| FMODE_EXCL);
	if (error == 0) {
	timeout = MSEC2NSEC(
	zfs_vdev_open_timeout_ms * 2);
	}
	}
	}
	} else {
	vd = kmem_zalloc(sizeof (vdev_disk_t), KM_SLEEP);

	rw_init(&vd->vd_lock, NULL, RW_DEFAULT, NULL);
	rw_enter(&vd->vd_lock, RW_WRITER);
	}

	/*
	* Devices are always opened by the path provided at configuration
	* time. This means that if the provided path is a udev by-id path
	* then drives may be re-cabled without an issue. If the provided
	* path is a udev by-path path, then the physical location information
	* will be preserved. This can be critical for more complicated
	* configurations where drives are located in specific physical
	* locations to maximize the systems tolerance to component failure.
	*
	* Alternatively, you can provide your own udev rule to flexibly map
	* the drives as you see fit. It is not advised that you use the
	* /dev/[hd]d devices which may be reordered due to probing order.
	* Devices in the wrong locations will be detected by the higher
	* level vdev validation.
	*
	* The specified paths may be briefly removed and recreated in
	* response to udev events. This should be exceptionally unlikely
	* because the zpool command makes every effort to verify these paths
	* have already settled prior to reaching this point. Therefore,
	* a ENOENT failure at this point is highly likely to be transient
	* and it is reasonable to sleep and retry before giving up. In
	* practice delays have been observed to be on the order of 100ms.
	+ *
	+ * When ERESTARTSYS is returned it indicates the block device is
	+ * a zvol which could not be opened due to the deadlock detection
	+ * logic in zvol_open(). Extend the timeout and retry the open
	+ * subsequent attempts are expected to eventually succeed.
	*/
	hrtime_t start = gethrtime();
	bdev = ERR_PTR(-ENXIO);
	while (IS_ERR(bdev) && ((gethrtime() - start) < timeout)) {
	bdev = blkdev_get_by_path(v->vdev_path, mode \| FMODE_EXCL,
	zfs_vdev_holder);
	if (unlikely(PTR_ERR(bdev) == -ENOENT)) {
	schedule_timeout(MSEC_TO_TICK(10));
	+ } else if (unlikely(PTR_ERR(bdev) == -ERESTARTSYS)) {
	+ timeout = MSEC2NSEC(zfs_vdev_open_timeout_ms * 10);
	+ continue;
	} else if (IS_ERR(bdev)) {
	break;
	}
	}

	if (IS_ERR(bdev)) {
	int error = -PTR_ERR(bdev);
	vdev_dbgmsg(v, "open error=%d timeout=%llu/%llu", error,
	(u_longlong_t)(gethrtime() - start),
	(u_longlong_t)timeout);
	vd->vd_bdev = NULL;
	v->vdev_tsd = vd;
	rw_exit(&vd->vd_lock);
	return (SET_ERROR(error));
	} else {
	vd->vd_bdev = bdev;
	v->vdev_tsd = vd;
	rw_exit(&vd->vd_lock);
	}

	struct request_queue *q = bdev_get_queue(vd->vd_bdev);

	/* Determine the physical block size */
	int physical_block_size = bdev_physical_block_size(vd->vd_bdev);

	/* Determine the logical block size */
	int logical_block_size = bdev_logical_block_size(vd->vd_bdev);

	/* Clear the nowritecache bit, causes vdev_reopen() to try again. */
	v->vdev_nowritecache = B_FALSE;

	/* Set when device reports it supports TRIM. */
	v->vdev_has_trim = !!blk_queue_discard(q);

	/* Set when device reports it supports secure TRIM. */
	v->vdev_has_securetrim = !!blk_queue_discard_secure(q);

	/* Inform the ZIO pipeline that we are non-rotational */
	v->vdev_nonrot = blk_queue_nonrot(q);

	/* Physical volume size in bytes for the partition */
	*psize = bdev_capacity(vd->vd_bdev);

	/* Physical volume size in bytes including possible expansion space */
	*max_psize = bdev_max_capacity(vd->vd_bdev, v->vdev_wholedisk);

	/* Based on the minimum sector size set the block size */
	*physical_ashift = highbit64(MAX(physical_block_size,
	SPA_MINBLOCKSIZE)) - 1;

	*logical_ashift = highbit64(MAX(logical_block_size,
	SPA_MINBLOCKSIZE)) - 1;

	return (0);
	}

	static void
	vdev_disk_close(vdev_t *v)
	{
	vdev_disk_t *vd = v->vdev_tsd;

	if (v->vdev_reopening \|\| vd == NULL)
	return;

	if (vd->vd_bdev != NULL) {
	blkdev_put(vd->vd_bdev,
	vdev_bdev_mode(spa_mode(v->vdev_spa)) \| FMODE_EXCL);
	}

	rw_destroy(&vd->vd_lock);
	kmem_free(vd, sizeof (vdev_disk_t));
	v->vdev_tsd = NULL;
	}

	static dio_request_t *
	vdev_disk_dio_alloc(int bio_count)
	{
	dio_request_t *dr = kmem_zalloc(sizeof (dio_request_t) +
	sizeof (struct bio ) bio_count, KM_SLEEP);
	atomic_set(&dr->dr_ref, 0);
	dr->dr_bio_count = bio_count;
	dr->dr_error = 0;

	for (int i = 0; i < dr->dr_bio_count; i++)
	dr->dr_bio[i] = NULL;

	return (dr);
	}

	static void
	vdev_disk_dio_free(dio_request_t *dr)
	{
	int i;

	for (i = 0; i < dr->dr_bio_count; i++)
	if (dr->dr_bio[i])
	bio_put(dr->dr_bio[i]);

	kmem_free(dr, sizeof (dio_request_t) +
	sizeof (struct bio ) dr->dr_bio_count);
	}

	static void
	vdev_disk_dio_get(dio_request_t *dr)
	{
	atomic_inc(&dr->dr_ref);
	}

	static int
	vdev_disk_dio_put(dio_request_t *dr)
	{
	int rc = atomic_dec_return(&dr->dr_ref);

	/*
	* Free the dio_request when the last reference is dropped and
	* ensure zio_interpret is called only once with the correct zio
	*/
	if (rc == 0) {
	zio_t *zio = dr->dr_zio;
	int error = dr->dr_error;

	vdev_disk_dio_free(dr);

	if (zio) {
	zio->io_error = error;
	ASSERT3S(zio->io_error, >=, 0);
	if (zio->io_error)
	vdev_disk_error(zio);

	zio_delay_interrupt(zio);
	}
	}

	return (rc);
	}

	BIO_END_IO_PROTO(vdev_disk_physio_completion, bio, error)
	{
	dio_request_t *dr = bio->bi_private;
	int rc;

	if (dr->dr_error == 0) {
	#ifdef HAVE_1ARG_BIO_END_IO_T
	dr->dr_error = BIO_END_IO_ERROR(bio);
	#else
	if (error)
	dr->dr_error = -(error);
	else if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
	dr->dr_error = EIO;
	#endif
	}

	/* Drop reference acquired by __vdev_disk_physio */
	rc = vdev_disk_dio_put(dr);
	}

	static inline void
	vdev_submit_bio_impl(struct bio *bio)
	{
	#ifdef HAVE_1ARG_SUBMIT_BIO
	(void) submit_bio(bio);
	#else
	(void) submit_bio(0, bio);
	#endif
	}

	/*
	* preempt_schedule_notrace is GPL-only which breaks the ZFS build, so
	* replace it with preempt_schedule under the following condition:
	*/
	#if defined(CONFIG_ARM64) && \
	defined(CONFIG_PREEMPTION) && \
	defined(CONFIG_BLK_CGROUP)
	#define preempt_schedule_notrace(x) preempt_schedule(x)
	#endif

	#ifdef HAVE_BIO_SET_DEV
	#if defined(CONFIG_BLK_CGROUP) && defined(HAVE_BIO_SET_DEV_GPL_ONLY)
	/*
	* The Linux 5.5 kernel updated percpu_ref_tryget() which is inlined by
	* blkg_tryget() to use rcu_read_lock() instead of rcu_read_lock_sched().
	* As a side effect the function was converted to GPL-only. Define our
	* own version when needed which uses rcu_read_lock_sched().
	*/
	#if defined(HAVE_BLKG_TRYGET_GPL_ONLY)
	static inline bool
	vdev_blkg_tryget(struct blkcg_gq *blkg)
	{
	struct percpu_ref *ref = &blkg->refcnt;
	unsigned long __percpu *count;
	bool rc;

	rcu_read_lock_sched();

	if (__ref_is_percpu(ref, &count)) {
	this_cpu_inc(*count);
	rc = true;
	} else {
	#ifdef ZFS_PERCPU_REF_COUNT_IN_DATA
	rc = atomic_long_inc_not_zero(&ref->data->count);
	#else
	rc = atomic_long_inc_not_zero(&ref->count);
	#endif
	}

	rcu_read_unlock_sched();

	return (rc);
	}
	#elif defined(HAVE_BLKG_TRYGET)
	#define vdev_blkg_tryget(bg) blkg_tryget(bg)
	#endif
	+#ifdef HAVE_BIO_SET_DEV_MACRO
	/*
	* The Linux 5.0 kernel updated the bio_set_dev() macro so it calls the
	* GPL-only bio_associate_blkg() symbol thus inadvertently converting
	* the entire macro. Provide a minimal version which always assigns the
	* request queue's root_blkg to the bio.
	*/
	static inline void
	vdev_bio_associate_blkg(struct bio *bio)
	{
	#if defined(HAVE_BIO_BDEV_DISK)
	struct request_queue *q = bio->bi_bdev->bd_disk->queue;
	#else
	struct request_queue *q = bio->bi_disk->queue;
	#endif

	ASSERT3P(q, !=, NULL);
	ASSERT3P(bio->bi_blkg, ==, NULL);

	if (q->root_blkg && vdev_blkg_tryget(q->root_blkg))
	bio->bi_blkg = q->root_blkg;
	}
	+
	#define bio_associate_blkg vdev_bio_associate_blkg
	+#else
	+static inline void
	+vdev_bio_set_dev(struct bio bio, struct block_device bdev)
	+{
	+#if defined(HAVE_BIO_BDEV_DISK)
	+ struct request_queue *q = bdev->bd_disk->queue;
	+#else
	+ struct request_queue *q = bio->bi_disk->queue;
	+#endif
	+ bio_clear_flag(bio, BIO_REMAPPED);
	+ if (bio->bi_bdev != bdev)
	+ bio_clear_flag(bio, BIO_THROTTLED);
	+ bio->bi_bdev = bdev;
	+
	+ ASSERT3P(q, !=, NULL);
	+ ASSERT3P(bio->bi_blkg, ==, NULL);
	+
	+ if (q->root_blkg && vdev_blkg_tryget(q->root_blkg))
	+ bio->bi_blkg = q->root_blkg;
	+}
	+#define bio_set_dev vdev_bio_set_dev
	+#endif
	#endif
	#else
	/*
	* Provide a bio_set_dev() helper macro for pre-Linux 4.14 kernels.
	*/
	static inline void
	bio_set_dev(struct bio bio, struct block_device bdev)
	{
	bio->bi_bdev = bdev;
	}
	#endif /* HAVE_BIO_SET_DEV */

	static inline void
	vdev_submit_bio(struct bio *bio)
	{
	struct bio_list *bio_list = current->bio_list;
	current->bio_list = NULL;
	vdev_submit_bio_impl(bio);
	current->bio_list = bio_list;
	}

	static int
	__vdev_disk_physio(struct block_device bdev, zio_t zio,
	size_t io_size, uint64_t io_offset, int rw, int flags)
	{
	dio_request_t *dr;
	uint64_t abd_offset;
	uint64_t bio_offset;
	int bio_size;
	int bio_count = 16;
	int error = 0;
	struct blk_plug plug;

	/*
	* Accessing outside the block device is never allowed.
	*/
	if (io_offset + io_size > bdev->bd_inode->i_size) {
	vdev_dbgmsg(zio->io_vd,
	"Illegal access %llu size %llu, device size %llu",
	(u_longlong_t)io_offset,
	(u_longlong_t)io_size,
	(u_longlong_t)i_size_read(bdev->bd_inode));
	return (SET_ERROR(EIO));
	}

	retry:
	dr = vdev_disk_dio_alloc(bio_count);

	if (zio && !(zio->io_flags & (ZIO_FLAG_IO_RETRY \| ZIO_FLAG_TRYHARD)))
	bio_set_flags_failfast(bdev, &flags);

	dr->dr_zio = zio;

	/*
	* Since bio's can have up to BIO_MAX_PAGES=256 iovec's, each of which
	* is at least 512 bytes and at most PAGESIZE (typically 4K), one bio
	* can cover at least 128KB and at most 1MB. When the required number
	* of iovec's exceeds this, we are forced to break the IO in multiple
	* bio's and wait for them all to complete. This is likely if the
	* recordsize property is increased beyond 1MB. The default
	* bio_count=16 should typically accommodate the maximum-size zio of
	* 16MB.
	*/

	abd_offset = 0;
	bio_offset = io_offset;
	bio_size = io_size;
	for (int i = 0; i <= dr->dr_bio_count; i++) {

	/* Finished constructing bio's for given buffer */
	if (bio_size <= 0)
	break;

	/*
	* If additional bio's are required, we have to retry, but
	* this should be rare - see the comment above.
	*/
	if (dr->dr_bio_count == i) {
	vdev_disk_dio_free(dr);
	bio_count *= 2;
	goto retry;
	}

	/* bio_alloc() with __GFP_WAIT never returns NULL */
	#ifdef HAVE_BIO_MAX_SEGS
	dr->dr_bio[i] = bio_alloc(GFP_NOIO, bio_max_segs(
	abd_nr_pages_off(zio->io_abd, bio_size, abd_offset)));
	#else
	dr->dr_bio[i] = bio_alloc(GFP_NOIO,
	MIN(abd_nr_pages_off(zio->io_abd, bio_size, abd_offset),
	BIO_MAX_PAGES));
	#endif
	if (unlikely(dr->dr_bio[i] == NULL)) {
	vdev_disk_dio_free(dr);
	return (SET_ERROR(ENOMEM));
	}

	/* Matching put called by vdev_disk_physio_completion */
	vdev_disk_dio_get(dr);

	bio_set_dev(dr->dr_bio[i], bdev);
	BIO_BI_SECTOR(dr->dr_bio[i]) = bio_offset >> 9;
	dr->dr_bio[i]->bi_end_io = vdev_disk_physio_completion;
	dr->dr_bio[i]->bi_private = dr;
	bio_set_op_attrs(dr->dr_bio[i], rw, flags);

	/* Remaining size is returned to become the new size */
	bio_size = abd_bio_map_off(dr->dr_bio[i], zio->io_abd,
	bio_size, abd_offset);

	/* Advance in buffer and construct another bio if needed */
	abd_offset += BIO_BI_SIZE(dr->dr_bio[i]);
	bio_offset += BIO_BI_SIZE(dr->dr_bio[i]);
	}

	/* Extra reference to protect dio_request during vdev_submit_bio */
	vdev_disk_dio_get(dr);

	if (dr->dr_bio_count > 1)
	blk_start_plug(&plug);

	/* Submit all bio's associated with this dio */
	for (int i = 0; i < dr->dr_bio_count; i++) {
	if (dr->dr_bio[i])
	vdev_submit_bio(dr->dr_bio[i]);
	}

	if (dr->dr_bio_count > 1)
	blk_finish_plug(&plug);

	(void) vdev_disk_dio_put(dr);

	return (error);
	}

	BIO_END_IO_PROTO(vdev_disk_io_flush_completion, bio, error)
	{
	zio_t *zio = bio->bi_private;
	#ifdef HAVE_1ARG_BIO_END_IO_T
	zio->io_error = BIO_END_IO_ERROR(bio);
	#else
	zio->io_error = -error;
	#endif

	if (zio->io_error && (zio->io_error == EOPNOTSUPP))
	zio->io_vd->vdev_nowritecache = B_TRUE;

	bio_put(bio);
	ASSERT3S(zio->io_error, >=, 0);
	if (zio->io_error)
	vdev_disk_error(zio);
	zio_interrupt(zio);
	}

	static int
	vdev_disk_io_flush(struct block_device bdev, zio_t zio)
	{
	struct request_queue *q;
	struct bio *bio;

	q = bdev_get_queue(bdev);
	if (!q)
	return (SET_ERROR(ENXIO));

	bio = bio_alloc(GFP_NOIO, 0);
	/* bio_alloc() with __GFP_WAIT never returns NULL */
	if (unlikely(bio == NULL))
	return (SET_ERROR(ENOMEM));

	bio->bi_end_io = vdev_disk_io_flush_completion;
	bio->bi_private = zio;
	bio_set_dev(bio, bdev);
	bio_set_flush(bio);
	vdev_submit_bio(bio);
	invalidate_bdev(bdev);

	return (0);
	}

	static void
	vdev_disk_io_start(zio_t *zio)
	{
	vdev_t *v = zio->io_vd;
	vdev_disk_t *vd = v->vdev_tsd;
	unsigned long trim_flags = 0;
	int rw, error;

	/*
	* If the vdev is closed, it's likely in the REMOVED or FAULTED state.
	* Nothing to be done here but return failure.
	*/
	if (vd == NULL) {
	zio->io_error = ENXIO;
	zio_interrupt(zio);
	return;
	}

	rw_enter(&vd->vd_lock, RW_READER);

	/*
	* If the vdev is closed, it's likely due to a failed reopen and is
	* in the UNAVAIL state. Nothing to be done here but return failure.
	*/
	if (vd->vd_bdev == NULL) {
	rw_exit(&vd->vd_lock);
	zio->io_error = ENXIO;
	zio_interrupt(zio);
	return;
	}

	switch (zio->io_type) {
	case ZIO_TYPE_IOCTL:

	if (!vdev_readable(v)) {
	rw_exit(&vd->vd_lock);
	zio->io_error = SET_ERROR(ENXIO);
	zio_interrupt(zio);
	return;
	}

	switch (zio->io_cmd) {
	case DKIOCFLUSHWRITECACHE:

	if (zfs_nocacheflush)
	break;

	if (v->vdev_nowritecache) {
	zio->io_error = SET_ERROR(ENOTSUP);
	break;
	}

	error = vdev_disk_io_flush(vd->vd_bdev, zio);
	if (error == 0) {
	rw_exit(&vd->vd_lock);
	return;
	}

	zio->io_error = error;

	break;

	default:
	zio->io_error = SET_ERROR(ENOTSUP);
	}

	rw_exit(&vd->vd_lock);
	zio_execute(zio);
	return;
	case ZIO_TYPE_WRITE:
	rw = WRITE;
	break;

	case ZIO_TYPE_READ:
	rw = READ;
	break;

	case ZIO_TYPE_TRIM:
	#if defined(BLKDEV_DISCARD_SECURE)
	if (zio->io_trim_flags & ZIO_TRIM_SECURE)
	trim_flags \|= BLKDEV_DISCARD_SECURE;
	#endif
	zio->io_error = -blkdev_issue_discard(vd->vd_bdev,
	zio->io_offset >> 9, zio->io_size >> 9, GFP_NOFS,
	trim_flags);

	rw_exit(&vd->vd_lock);
	zio_interrupt(zio);
	return;

	default:
	rw_exit(&vd->vd_lock);
	zio->io_error = SET_ERROR(ENOTSUP);
	zio_interrupt(zio);
	return;
	}

	zio->io_target_timestamp = zio_handle_io_delay(zio);
	error = __vdev_disk_physio(vd->vd_bdev, zio,
	zio->io_size, zio->io_offset, rw, 0);
	rw_exit(&vd->vd_lock);

	if (error) {
	zio->io_error = error;
	zio_interrupt(zio);
	return;
	}
	}

	static void
	vdev_disk_io_done(zio_t *zio)
	{
	/*
	* If the device returned EIO, we revalidate the media. If it is
	* determined the media has changed this triggers the asynchronous
	* removal of the device from the configuration.
	*/
	if (zio->io_error == EIO) {
	vdev_t *v = zio->io_vd;
	vdev_disk_t *vd = v->vdev_tsd;

	if (zfs_check_media_change(vd->vd_bdev)) {
	invalidate_bdev(vd->vd_bdev);
	v->vdev_remove_wanted = B_TRUE;
	spa_async_request(zio->io_spa, SPA_ASYNC_REMOVE);
	}
	}
	}

	static void
	vdev_disk_hold(vdev_t *vd)
	{
	ASSERT(spa_config_held(vd->vdev_spa, SCL_STATE, RW_WRITER));

	/* We must have a pathname, and it must be absolute. */
	if (vd->vdev_path == NULL \|\| vd->vdev_path[0] != '/')
	return;

	/*
	* Only prefetch path and devid info if the device has
	* never been opened.
	*/
	if (vd->vdev_tsd != NULL)
	return;

	}

	static void
	vdev_disk_rele(vdev_t *vd)
	{
	ASSERT(spa_config_held(vd->vdev_spa, SCL_STATE, RW_WRITER));

	/* XXX: Implement me as a vnode rele for the device */
	}

	vdev_ops_t vdev_disk_ops = {
	.vdev_op_init = NULL,
	.vdev_op_fini = NULL,
	.vdev_op_open = vdev_disk_open,
	.vdev_op_close = vdev_disk_close,
	.vdev_op_asize = vdev_default_asize,
	.vdev_op_min_asize = vdev_default_min_asize,
	.vdev_op_min_alloc = NULL,
	.vdev_op_io_start = vdev_disk_io_start,
	.vdev_op_io_done = vdev_disk_io_done,
	.vdev_op_state_change = NULL,
	.vdev_op_need_resilver = NULL,
	.vdev_op_hold = vdev_disk_hold,
	.vdev_op_rele = vdev_disk_rele,
	.vdev_op_remap = NULL,
	.vdev_op_xlate = vdev_default_xlate,
	.vdev_op_rebuild_asize = NULL,
	.vdev_op_metaslab_init = NULL,
	.vdev_op_config_generate = NULL,
	.vdev_op_nparity = NULL,
	.vdev_op_ndisks = NULL,
	.vdev_op_type = VDEV_TYPE_DISK, /* name of this vdev type */
	.vdev_op_leaf = B_TRUE /* leaf vdev */
	};

	/*
	* The zfs_vdev_scheduler module option has been deprecated. Setting this
	* value no longer has any effect. It has not yet been entirely removed
	* to allow the module to be loaded if this option is specified in the
	* /etc/modprobe.d/zfs.conf file. The following warning will be logged.
	*/
	static int
	param_set_vdev_scheduler(const char val, zfs_kernel_param_t kp)
	{
	int error = param_set_charp(val, kp);
	if (error == 0) {
	printk(KERN_INFO "The 'zfs_vdev_scheduler' module option "
	"is not supported.\n");
	}

	return (error);
	}

	char *zfs_vdev_scheduler = "unused";
	module_param_call(zfs_vdev_scheduler, param_set_vdev_scheduler,
	param_get_charp, &zfs_vdev_scheduler, 0644);
	MODULE_PARM_DESC(zfs_vdev_scheduler, "I/O scheduler");

	int
	param_set_min_auto_ashift(const char buf, zfs_kernel_param_t kp)
	{
	uint64_t val;
	int error;

	error = kstrtoull(buf, 0, &val);
	if (error < 0)
	return (SET_ERROR(error));

	if (val < ASHIFT_MIN \|\| val > zfs_vdev_max_auto_ashift)
	return (SET_ERROR(-EINVAL));

	error = param_set_ulong(buf, kp);
	if (error < 0)
	return (SET_ERROR(error));

	return (0);
	}

	int
	param_set_max_auto_ashift(const char buf, zfs_kernel_param_t kp)
	{
	uint64_t val;
	int error;

	error = kstrtoull(buf, 0, &val);
	if (error < 0)
	return (SET_ERROR(error));

	if (val > ASHIFT_MAX \|\| val < zfs_vdev_min_auto_ashift)
	return (SET_ERROR(-EINVAL));

	error = param_set_ulong(buf, kp);
	if (error < 0)
	return (SET_ERROR(error));

	return (0);
	}
	diff --git a/sys/contrib/openzfs/module/os/linux/zfs/zfs_sysfs.c b/sys/contrib/openzfs/module/os/linux/zfs/zfs_sysfs.c
	index fb7c68987360..6f71382cf74e 100644
	--- a/sys/contrib/openzfs/module/os/linux/zfs/zfs_sysfs.c
	+++ b/sys/contrib/openzfs/module/os/linux/zfs/zfs_sysfs.c
	@@ -1,662 +1,696 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2018, 2019 by Delphix. All rights reserved.
	*/

	#include <sys/types.h>
	#include <sys/param.h>
	#include <sys/zfeature.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/zfs_sysfs.h>
	#include <sys/kmem.h>
	#include <sys/fs/zfs.h>
	#include <linux/kobject.h>

	#include "zfs_prop.h"

	#if !defined(_KERNEL)
	#error kernel builds only
	#endif

	/*
	* ZFS Module sysfs support
	*
	* This extends our sysfs '/sys/module/zfs' entry to include feature
	* and property attributes. The primary consumer of this information
	* is user processes, like the zfs CLI, that need to know what the
	* current loaded ZFS module supports. The libzfs binary will consult
	* this information when instantiating the zfs\|zpool property tables
	* and the pool features table.
	*
	* The added top-level directories are:
	* /sys/module/zfs
	* ├── features.kernel
	* ├── features.pool
	* ├── properties.dataset
	* └── properties.pool
	*
	* The local interface for the zfs kobjects includes:
	* zfs_kobj_init()
	* zfs_kobj_add()
	* zfs_kobj_release()
	* zfs_kobj_add_attr()
	* zfs_kobj_fini()
	*/

	/*
	* A zfs_mod_kobj_t represents a zfs kobject under '/sys/module/zfs'
	*/
	struct zfs_mod_kobj;
	typedef struct zfs_mod_kobj zfs_mod_kobj_t;

	struct zfs_mod_kobj {
	struct kobject zko_kobj;
	struct kobj_type zko_kobj_type;
	struct sysfs_ops zko_sysfs_ops;
	size_t zko_attr_count;
	struct attribute zko_attr_list; / allocated */
	struct attribute *zko_default_attrs; / allocated */
	size_t zko_child_count;
	zfs_mod_kobj_t zko_children; / allocated */
	};

	#define ATTR_TABLE_SIZE(cnt) (sizeof (struct attribute) * (cnt))
	/* Note +1 for NULL terminator slot */
	#define DEFAULT_ATTR_SIZE(cnt) (sizeof (struct attribute ) (cnt + 1))
	#define CHILD_TABLE_SIZE(cnt) (sizeof (zfs_mod_kobj_t) * (cnt))

	/*
	* These are the top-level kobjects under '/sys/module/zfs/'
	*/
	static zfs_mod_kobj_t kernel_features_kobj;
	static zfs_mod_kobj_t pool_features_kobj;
	static zfs_mod_kobj_t dataset_props_kobj;
	+static zfs_mod_kobj_t vdev_props_kobj;
	static zfs_mod_kobj_t pool_props_kobj;

	/*
	* The show function is used to provide the content
	* of an attribute into a PAGE_SIZE buffer.
	*/
	typedef ssize_t (sysfs_show_func)(struct kobject , struct attribute *,
	char *);

	static void
	zfs_kobj_fini(zfs_mod_kobj_t *zkobj)
	{
	/* finalize any child kobjects */
	if (zkobj->zko_child_count != 0) {
	ASSERT(zkobj->zko_children);
	for (int i = 0; i < zkobj->zko_child_count; i++)
	zfs_kobj_fini(&zkobj->zko_children[i]);
	}

	/* kobject_put() will call zfs_kobj_release() to release memory */
	kobject_del(&zkobj->zko_kobj);
	kobject_put(&zkobj->zko_kobj);
	}

	static void
	zfs_kobj_release(struct kobject *kobj)
	{
	zfs_mod_kobj_t *zkobj = container_of(kobj, zfs_mod_kobj_t, zko_kobj);

	if (zkobj->zko_attr_list != NULL) {
	ASSERT3S(zkobj->zko_attr_count, !=, 0);
	kmem_free(zkobj->zko_attr_list,
	ATTR_TABLE_SIZE(zkobj->zko_attr_count));
	zkobj->zko_attr_list = NULL;
	}

	if (zkobj->zko_default_attrs != NULL) {
	kmem_free(zkobj->zko_default_attrs,
	DEFAULT_ATTR_SIZE(zkobj->zko_attr_count));
	zkobj->zko_default_attrs = NULL;
	}

	if (zkobj->zko_child_count != 0) {
	ASSERT(zkobj->zko_children);

	kmem_free(zkobj->zko_children,
	CHILD_TABLE_SIZE(zkobj->zko_child_count));
	zkobj->zko_child_count = 0;
	zkobj->zko_children = NULL;
	}

	zkobj->zko_attr_count = 0;
	}

	#ifndef sysfs_attr_init
	#define sysfs_attr_init(attr) do {} while (0)
	#endif

	static void
	zfs_kobj_add_attr(zfs_mod_kobj_t zkobj, int attr_num, const char attr_name)
	{
	VERIFY3U(attr_num, <, zkobj->zko_attr_count);
	ASSERT(zkobj->zko_attr_list);
	ASSERT(zkobj->zko_default_attrs);

	zkobj->zko_attr_list[attr_num].name = attr_name;
	zkobj->zko_attr_list[attr_num].mode = 0444;
	zkobj->zko_default_attrs[attr_num] = &zkobj->zko_attr_list[attr_num];
	sysfs_attr_init(&zkobj->zko_attr_list[attr_num]);
	}

	static int
	zfs_kobj_init(zfs_mod_kobj_t *zkobj, int attr_cnt, int child_cnt,
	sysfs_show_func show_func)
	{
	/*
	* Initialize object's attributes. Count can be zero.
	*/
	if (attr_cnt > 0) {
	zkobj->zko_attr_list = kmem_zalloc(ATTR_TABLE_SIZE(attr_cnt),
	KM_SLEEP);
	if (zkobj->zko_attr_list == NULL)
	return (ENOMEM);
	}
	/* this will always have at least one slot for NULL termination */
	zkobj->zko_default_attrs = kmem_zalloc(DEFAULT_ATTR_SIZE(attr_cnt),
	KM_SLEEP);
	if (zkobj->zko_default_attrs == NULL) {
	if (zkobj->zko_attr_list != NULL) {
	kmem_free(zkobj->zko_attr_list,
	ATTR_TABLE_SIZE(attr_cnt));
	}
	return (ENOMEM);
	}
	zkobj->zko_attr_count = attr_cnt;
	zkobj->zko_kobj_type.default_attrs = zkobj->zko_default_attrs;

	if (child_cnt > 0) {
	zkobj->zko_children = kmem_zalloc(CHILD_TABLE_SIZE(child_cnt),
	KM_SLEEP);
	if (zkobj->zko_children == NULL) {
	if (zkobj->zko_default_attrs != NULL) {
	kmem_free(zkobj->zko_default_attrs,
	DEFAULT_ATTR_SIZE(attr_cnt));
	}
	if (zkobj->zko_attr_list != NULL) {
	kmem_free(zkobj->zko_attr_list,
	ATTR_TABLE_SIZE(attr_cnt));
	}
	return (ENOMEM);
	}
	zkobj->zko_child_count = child_cnt;
	}

	zkobj->zko_sysfs_ops.show = show_func;
	zkobj->zko_kobj_type.sysfs_ops = &zkobj->zko_sysfs_ops;
	zkobj->zko_kobj_type.release = zfs_kobj_release;

	return (0);
	}

	static int
	zfs_kobj_add(zfs_mod_kobj_t zkobj, struct kobject parent, const char *name)
	{
	/* zko_default_attrs must be NULL terminated */
	ASSERT(zkobj->zko_default_attrs != NULL);
	ASSERT(zkobj->zko_default_attrs[zkobj->zko_attr_count] == NULL);

	kobject_init(&zkobj->zko_kobj, &zkobj->zko_kobj_type);
	return (kobject_add(&zkobj->zko_kobj, parent, name));
	}

	/*
	* Each zfs property has these common attributes
	*/
	static const char *zprop_attrs[] = {
	"type",
	"readonly",
	"setonce",
	"visible",
	"values",
	"default",
	"datasets" /* zfs properties only */
	};

	#define ZFS_PROP_ATTR_COUNT ARRAY_SIZE(zprop_attrs)
	#define ZPOOL_PROP_ATTR_COUNT (ZFS_PROP_ATTR_COUNT - 1)

	static const char *zprop_types[] = {
	"number",
	"string",
	"index",
	};

	typedef struct zfs_type_map {
	zfs_type_t ztm_type;
	const char *ztm_name;
	} zfs_type_map_t;

	static zfs_type_map_t type_map[] = {
	{ZFS_TYPE_FILESYSTEM, "filesystem"},
	{ZFS_TYPE_SNAPSHOT, "snapshot"},
	{ZFS_TYPE_VOLUME, "volume"},
	{ZFS_TYPE_BOOKMARK, "bookmark"}
	};

	/*
	* Show the content for a zfs property attribute
	*/
	static ssize_t
	zprop_sysfs_show(const char attr_name, const zprop_desc_t property,
	char *buf, size_t buflen)
	{
	const char *show_str;
	char number[32];

	/* For dataset properties list the dataset types that apply */
	if (strcmp(attr_name, "datasets") == 0 &&
	property->pd_types != ZFS_TYPE_POOL) {
	int len = 0;

	for (int i = 0; i < ARRAY_SIZE(type_map); i++) {
	if (type_map[i].ztm_type & property->pd_types) {
	len += snprintf(buf + len, buflen - len, "%s ",
	type_map[i].ztm_name);
	}
	}
	len += snprintf(buf + len, buflen - len, "\n");
	return (len);
	}

	if (strcmp(attr_name, "type") == 0) {
	show_str = zprop_types[property->pd_proptype];
	} else if (strcmp(attr_name, "readonly") == 0) {
	show_str = property->pd_attr == PROP_READONLY ? "1" : "0";
	} else if (strcmp(attr_name, "setonce") == 0) {
	show_str = property->pd_attr == PROP_ONETIME ? "1" : "0";
	} else if (strcmp(attr_name, "visible") == 0) {
	show_str = property->pd_visible ? "1" : "0";
	} else if (strcmp(attr_name, "values") == 0) {
	show_str = property->pd_values ? property->pd_values : "";
	} else if (strcmp(attr_name, "default") == 0) {
	switch (property->pd_proptype) {
	case PROP_TYPE_NUMBER:
	(void) snprintf(number, sizeof (number), "%llu",
	(u_longlong_t)property->pd_numdefault);
	show_str = number;
	break;
	case PROP_TYPE_STRING:
	show_str = property->pd_strdefault ?
	property->pd_strdefault : "";
	break;
	case PROP_TYPE_INDEX:
	if (zprop_index_to_string(property->pd_propnum,
	property->pd_numdefault, &show_str,
	property->pd_types) != 0) {
	show_str = "";
	}
	break;
	default:
	return (0);
	}
	} else {
	return (0);
	}

	return (snprintf(buf, buflen, "%s\n", show_str));
	}

	static ssize_t
	dataset_property_show(struct kobject kobj, struct attribute attr, char *buf)
	{
	zfs_prop_t prop = zfs_name_to_prop(kobject_name(kobj));
	zprop_desc_t *prop_tbl = zfs_prop_get_table();
	ssize_t len;

	ASSERT3U(prop, <, ZFS_NUM_PROPS);

	len = zprop_sysfs_show(attr->name, &prop_tbl[prop], buf, PAGE_SIZE);

	return (len);
	}

	+static ssize_t
	+vdev_property_show(struct kobject kobj, struct attribute attr, char *buf)
	+{
	+ vdev_prop_t prop = vdev_name_to_prop(kobject_name(kobj));
	+ zprop_desc_t *prop_tbl = vdev_prop_get_table();
	+ ssize_t len;
	+
	+ ASSERT3U(prop, <, VDEV_NUM_PROPS);
	+
	+ len = zprop_sysfs_show(attr->name, &prop_tbl[prop], buf, PAGE_SIZE);
	+
	+ return (len);
	+}
	+
	static ssize_t
	pool_property_show(struct kobject kobj, struct attribute attr, char *buf)
	{
	zpool_prop_t prop = zpool_name_to_prop(kobject_name(kobj));
	zprop_desc_t *prop_tbl = zpool_prop_get_table();
	ssize_t len;

	ASSERT3U(prop, <, ZPOOL_NUM_PROPS);

	len = zprop_sysfs_show(attr->name, &prop_tbl[prop], buf, PAGE_SIZE);

	return (len);
	}

	/*
	* ZFS kernel feature attributes for '/sys/module/zfs/features.kernel'
	*
	* This list is intended for kernel features that don't have a pool feature
	* association or that extend existing user kernel interfaces.
	*
	* A user process can easily check if the running zfs kernel module
	* supports the new feature.
	*/
	static const char *zfs_kernel_features[] = {
	/* --> Add new kernel features here */
	"com.delphix:vdev_initialize",
	"org.zfsonlinux:vdev_trim",
	"org.openzfs:l2arc_persistent",
	};

	#define KERNEL_FEATURE_COUNT ARRAY_SIZE(zfs_kernel_features)

	static ssize_t
	kernel_feature_show(struct kobject kobj, struct attribute attr, char *buf)
	{
	if (strcmp(attr->name, "supported") == 0)
	return (snprintf(buf, PAGE_SIZE, "yes\n"));
	return (0);
	}

	static void
	kernel_feature_to_kobj(zfs_mod_kobj_t parent, int slot, const char name)
	{
	zfs_mod_kobj_t *zfs_kobj = &parent->zko_children[slot];

	ASSERT3U(slot, <, KERNEL_FEATURE_COUNT);
	ASSERT(name);

	int err = zfs_kobj_init(zfs_kobj, 1, 0, kernel_feature_show);
	if (err)
	return;

	zfs_kobj_add_attr(zfs_kobj, 0, "supported");

	err = zfs_kobj_add(zfs_kobj, &parent->zko_kobj, name);
	if (err)
	zfs_kobj_release(&zfs_kobj->zko_kobj);
	}

	static int
	zfs_kernel_features_init(zfs_mod_kobj_t zfs_kobj, struct kobject parent)
	{
	/*
	* Create a parent kobject to host kernel features.
	*
	* '/sys/module/zfs/features.kernel'
	*/
	int err = zfs_kobj_init(zfs_kobj, 0, KERNEL_FEATURE_COUNT,
	kernel_feature_show);
	if (err)
	return (err);
	err = zfs_kobj_add(zfs_kobj, parent, ZFS_SYSFS_KERNEL_FEATURES);
	if (err) {
	zfs_kobj_release(&zfs_kobj->zko_kobj);
	return (err);
	}

	/*
	* Now create a kobject for each feature.
	*
	* '/sys/module/zfs/features.kernel/<feature>'
	*/
	for (int f = 0; f < KERNEL_FEATURE_COUNT; f++)
	kernel_feature_to_kobj(zfs_kobj, f, zfs_kernel_features[f]);

	return (0);
	}

	/*
	* Each pool feature has these common attributes
	*/
	static const char *pool_feature_attrs[] = {
	"description",
	"guid",
	"uname",
	"readonly_compatible",
	"required_for_mos",
	"activate_on_enable",
	"per_dataset"
	};

	#define ZPOOL_FEATURE_ATTR_COUNT ARRAY_SIZE(pool_feature_attrs)

	/*
	* Show the content for the given zfs pool feature attribute
	*/
	static ssize_t
	pool_feature_show(struct kobject kobj, struct attribute attr, char *buf)
	{
	spa_feature_t fid;

	if (zfeature_lookup_guid(kobject_name(kobj), &fid) != 0)
	return (0);

	ASSERT3U(fid, <, SPA_FEATURES);

	zfeature_flags_t flags = spa_feature_table[fid].fi_flags;
	const char *show_str = NULL;

	if (strcmp(attr->name, "description") == 0) {
	show_str = spa_feature_table[fid].fi_desc;
	} else if (strcmp(attr->name, "guid") == 0) {
	show_str = spa_feature_table[fid].fi_guid;
	} else if (strcmp(attr->name, "uname") == 0) {
	show_str = spa_feature_table[fid].fi_uname;
	} else if (strcmp(attr->name, "readonly_compatible") == 0) {
	show_str = flags & ZFEATURE_FLAG_READONLY_COMPAT ? "1" : "0";
	} else if (strcmp(attr->name, "required_for_mos") == 0) {
	show_str = flags & ZFEATURE_FLAG_MOS ? "1" : "0";
	} else if (strcmp(attr->name, "activate_on_enable") == 0) {
	show_str = flags & ZFEATURE_FLAG_ACTIVATE_ON_ENABLE ? "1" : "0";
	} else if (strcmp(attr->name, "per_dataset") == 0) {
	show_str = flags & ZFEATURE_FLAG_PER_DATASET ? "1" : "0";
	}
	if (show_str == NULL)
	return (0);

	return (snprintf(buf, PAGE_SIZE, "%s\n", show_str));
	}

	static void
	pool_feature_to_kobj(zfs_mod_kobj_t *parent, spa_feature_t fid,
	const char *name)
	{
	zfs_mod_kobj_t *zfs_kobj = &parent->zko_children[fid];

	ASSERT3U(fid, <, SPA_FEATURES);
	ASSERT(name);

	int err = zfs_kobj_init(zfs_kobj, ZPOOL_FEATURE_ATTR_COUNT, 0,
	pool_feature_show);
	if (err)
	return;

	for (int i = 0; i < ZPOOL_FEATURE_ATTR_COUNT; i++)
	zfs_kobj_add_attr(zfs_kobj, i, pool_feature_attrs[i]);

	err = zfs_kobj_add(zfs_kobj, &parent->zko_kobj, name);
	if (err)
	zfs_kobj_release(&zfs_kobj->zko_kobj);
	}

	static int
	zfs_pool_features_init(zfs_mod_kobj_t zfs_kobj, struct kobject parent)
	{
	/*
	* Create a parent kobject to host pool features.
	*
	* '/sys/module/zfs/features.pool'
	*/
	int err = zfs_kobj_init(zfs_kobj, 0, SPA_FEATURES, pool_feature_show);
	if (err)
	return (err);
	err = zfs_kobj_add(zfs_kobj, parent, ZFS_SYSFS_POOL_FEATURES);
	if (err) {
	zfs_kobj_release(&zfs_kobj->zko_kobj);
	return (err);
	}

	/*
	* Now create a kobject for each feature.
	*
	* '/sys/module/zfs/features.pool/<feature>'
	*/
	for (spa_feature_t i = 0; i < SPA_FEATURES; i++)
	pool_feature_to_kobj(zfs_kobj, i, spa_feature_table[i].fi_guid);

	return (0);
	}

	typedef struct prop_to_kobj_arg {
	zprop_desc_t *p2k_table;
	zfs_mod_kobj_t *p2k_parent;
	sysfs_show_func p2k_show_func;
	int p2k_attr_count;
	} prop_to_kobj_arg_t;

	static int
	zprop_to_kobj(int prop, void *args)
	{
	prop_to_kobj_arg_t *data = args;
	zfs_mod_kobj_t *parent = data->p2k_parent;
	zfs_mod_kobj_t *zfs_kobj = &parent->zko_children[prop];
	const char *name = data->p2k_table[prop].pd_name;
	int err;

	ASSERT(name);

	err = zfs_kobj_init(zfs_kobj, data->p2k_attr_count, 0,
	data->p2k_show_func);
	if (err)
	return (ZPROP_CONT);

	for (int i = 0; i < data->p2k_attr_count; i++)
	zfs_kobj_add_attr(zfs_kobj, i, zprop_attrs[i]);

	err = zfs_kobj_add(zfs_kobj, &parent->zko_kobj, name);
	if (err)
	zfs_kobj_release(&zfs_kobj->zko_kobj);

	return (ZPROP_CONT);
	}

	static int
	zfs_sysfs_properties_init(zfs_mod_kobj_t zfs_kobj, struct kobject parent,
	zfs_type_t type)
	{
	prop_to_kobj_arg_t context;
	const char *name;
	int err;

	/*
	* Create a parent kobject to host properties.
	*
	* '/sys/module/zfs/properties.<type>'
	*/
	if (type == ZFS_TYPE_POOL) {
	name = ZFS_SYSFS_POOL_PROPERTIES;
	context.p2k_table = zpool_prop_get_table();
	context.p2k_attr_count = ZPOOL_PROP_ATTR_COUNT;
	context.p2k_parent = zfs_kobj;
	context.p2k_show_func = pool_property_show;
	err = zfs_kobj_init(zfs_kobj, 0, ZPOOL_NUM_PROPS,
	pool_property_show);
	+ } else if (type == ZFS_TYPE_VDEV) {
	+ name = ZFS_SYSFS_VDEV_PROPERTIES;
	+ context.p2k_table = vdev_prop_get_table();
	+ context.p2k_attr_count = ZPOOL_PROP_ATTR_COUNT;
	+ context.p2k_parent = zfs_kobj;
	+ context.p2k_show_func = vdev_property_show;
	+ err = zfs_kobj_init(zfs_kobj, 0, VDEV_NUM_PROPS,
	+ vdev_property_show);
	} else {
	name = ZFS_SYSFS_DATASET_PROPERTIES;
	context.p2k_table = zfs_prop_get_table();
	context.p2k_attr_count = ZFS_PROP_ATTR_COUNT;
	context.p2k_parent = zfs_kobj;
	context.p2k_show_func = dataset_property_show;
	err = zfs_kobj_init(zfs_kobj, 0, ZFS_NUM_PROPS,
	dataset_property_show);
	}

	if (err)
	return (err);

	err = zfs_kobj_add(zfs_kobj, parent, name);
	if (err) {
	zfs_kobj_release(&zfs_kobj->zko_kobj);
	return (err);
	}

	/*
	* Create a kobject for each property.
	*
	* '/sys/module/zfs/properties.<type>/<property>'
	*/
	(void) zprop_iter_common(zprop_to_kobj, &context, B_TRUE,
	B_FALSE, type);

	return (err);
	}

	void
	zfs_sysfs_init(void)
	{
	struct kobject *parent;
	#if defined(CONFIG_ZFS) && !defined(CONFIG_ZFS_MODULE)
	parent = kobject_create_and_add("zfs", fs_kobj);
	#else
	parent = &(((struct module *)(THIS_MODULE))->mkobj).kobj;
	#endif
	int err;

	if (parent == NULL)
	return;

	err = zfs_kernel_features_init(&kernel_features_kobj, parent);
	if (err)
	return;

	err = zfs_pool_features_init(&pool_features_kobj, parent);
	if (err) {
	zfs_kobj_fini(&kernel_features_kobj);
	return;
	}

	err = zfs_sysfs_properties_init(&pool_props_kobj, parent,
	ZFS_TYPE_POOL);
	if (err) {
	zfs_kobj_fini(&kernel_features_kobj);
	zfs_kobj_fini(&pool_features_kobj);
	return;
	}

	+ err = zfs_sysfs_properties_init(&vdev_props_kobj, parent,
	+ ZFS_TYPE_VDEV);
	+ if (err) {
	+ zfs_kobj_fini(&kernel_features_kobj);
	+ zfs_kobj_fini(&pool_features_kobj);
	+ zfs_kobj_fini(&pool_props_kobj);
	+ return;
	+ }
	+
	err = zfs_sysfs_properties_init(&dataset_props_kobj, parent,
	ZFS_TYPE_FILESYSTEM);
	if (err) {
	zfs_kobj_fini(&kernel_features_kobj);
	zfs_kobj_fini(&pool_features_kobj);
	zfs_kobj_fini(&pool_props_kobj);
	+ zfs_kobj_fini(&vdev_props_kobj);
	return;
	}
	}

	void
	zfs_sysfs_fini(void)
	{
	/*
	* Remove top-level kobjects; each will remove any children kobjects
	*/
	zfs_kobj_fini(&kernel_features_kobj);
	zfs_kobj_fini(&pool_features_kobj);
	- zfs_kobj_fini(&dataset_props_kobj);
	zfs_kobj_fini(&pool_props_kobj);
	+ zfs_kobj_fini(&vdev_props_kobj);
	+ zfs_kobj_fini(&dataset_props_kobj);
	}
	diff --git a/sys/contrib/openzfs/module/os/linux/zfs/zfs_vnops_os.c b/sys/contrib/openzfs/module/os/linux/zfs/zfs_vnops_os.c
	index e0dc6ed95747..6c7de9830c15 100644
	--- a/sys/contrib/openzfs/module/os/linux/zfs/zfs_vnops_os.c
	+++ b/sys/contrib/openzfs/module/os/linux/zfs/zfs_vnops_os.c
	@@ -1,3999 +1,4003 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2012, 2018 by Delphix. All rights reserved.
	* Copyright (c) 2015 by Chunwei Chen. All rights reserved.
	* Copyright 2017 Nexenta Systems, Inc.
	*/

	/* Portions Copyright 2007 Jeremy Teo */
	/* Portions Copyright 2010 Robert Milkowski */


	#include <sys/types.h>
	#include <sys/param.h>
	#include <sys/time.h>
	#include <sys/sysmacros.h>
	#include <sys/vfs.h>
	#include <sys/file.h>
	#include <sys/stat.h>
	#include <sys/kmem.h>
	#include <sys/taskq.h>
	#include <sys/uio.h>
	#include <sys/vmsystm.h>
	#include <sys/atomic.h>
	#include <sys/pathname.h>
	#include <sys/cmn_err.h>
	#include <sys/errno.h>
	#include <sys/zfs_dir.h>
	#include <sys/zfs_acl.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/fs/zfs.h>
	#include <sys/dmu.h>
	#include <sys/dmu_objset.h>
	#include <sys/spa.h>
	#include <sys/txg.h>
	#include <sys/dbuf.h>
	#include <sys/zap.h>
	#include <sys/sa.h>
	#include <sys/policy.h>
	#include <sys/sunddi.h>
	#include <sys/sid.h>
	#include <sys/zfs_ctldir.h>
	#include <sys/zfs_fuid.h>
	#include <sys/zfs_quota.h>
	#include <sys/zfs_sa.h>
	#include <sys/zfs_vnops.h>
	#include <sys/zfs_rlock.h>
	#include <sys/cred.h>
	#include <sys/zpl.h>
	#include <sys/zil.h>
	#include <sys/sa_impl.h>

	/*
	* Programming rules.
	*
	* Each vnode op performs some logical unit of work. To do this, the ZPL must
	* properly lock its in-core state, create a DMU transaction, do the work,
	* record this work in the intent log (ZIL), commit the DMU transaction,
	* and wait for the intent log to commit if it is a synchronous operation.
	* Moreover, the vnode ops must work in both normal and log replay context.
	* The ordering of events is important to avoid deadlocks and references
	* to freed memory. The example below illustrates the following Big Rules:
	*
	* (1) A check must be made in each zfs thread for a mounted file system.
	* This is done avoiding races using ZFS_ENTER(zfsvfs).
	* A ZFS_EXIT(zfsvfs) is needed before all returns. Any znodes
	* must be checked with ZFS_VERIFY_ZP(zp). Both of these macros
	* can return EIO from the calling function.
	*
	* (2) zrele() should always be the last thing except for zil_commit() (if
	* necessary) and ZFS_EXIT(). This is for 3 reasons: First, if it's the
	* last reference, the vnode/znode can be freed, so the zp may point to
	* freed memory. Second, the last reference will call zfs_zinactive(),
	* which may induce a lot of work -- pushing cached pages (which acquires
	* range locks) and syncing out cached atime changes. Third,
	* zfs_zinactive() may require a new tx, which could deadlock the system
	* if you were already holding one. This deadlock occurs because the tx
	* currently being operated on prevents a txg from syncing, which
	* prevents the new tx from progressing, resulting in a deadlock. If you
	* must call zrele() within a tx, use zfs_zrele_async(). Note that iput()
	* is a synonym for zrele().
	*
	* (3) All range locks must be grabbed before calling dmu_tx_assign(),
	* as they can span dmu_tx_assign() calls.
	*
	* (4) If ZPL locks are held, pass TXG_NOWAIT as the second argument to
	* dmu_tx_assign(). This is critical because we don't want to block
	* while holding locks.
	*
	* If no ZPL locks are held (aside from ZFS_ENTER()), use TXG_WAIT. This
	* reduces lock contention and CPU usage when we must wait (note that if
	* throughput is constrained by the storage, nearly every transaction
	* must wait).
	*
	* Note, in particular, that if a lock is sometimes acquired before
	* the tx assigns, and sometimes after (e.g. z_lock), then failing
	* to use a non-blocking assign can deadlock the system. The scenario:
	*
	* Thread A has grabbed a lock before calling dmu_tx_assign().
	* Thread B is in an already-assigned tx, and blocks for this lock.
	* Thread A calls dmu_tx_assign(TXG_WAIT) and blocks in txg_wait_open()
	* forever, because the previous txg can't quiesce until B's tx commits.
	*
	* If dmu_tx_assign() returns ERESTART and zfsvfs->z_assign is TXG_NOWAIT,
	* then drop all locks, call dmu_tx_wait(), and try again. On subsequent
	* calls to dmu_tx_assign(), pass TXG_NOTHROTTLE in addition to TXG_NOWAIT,
	* to indicate that this operation has already called dmu_tx_wait().
	* This will ensure that we don't retry forever, waiting a short bit
	* each time.
	*
	* (5) If the operation succeeded, generate the intent log entry for it
	* before dropping locks. This ensures that the ordering of events
	* in the intent log matches the order in which they actually occurred.
	* During ZIL replay the zfs_log_* functions will update the sequence
	* number to indicate the zil transaction has replayed.
	*
	* (6) At the end of each vnode op, the DMU tx must always commit,
	* regardless of whether there were any errors.
	*
	* (7) After dropping all locks, invoke zil_commit(zilog, foid)
	* to ensure that synchronous semantics are provided when necessary.
	*
	* In general, this is how things should be ordered in each vnode op:
	*
	* ZFS_ENTER(zfsvfs); // exit if unmounted
	* top:
	* zfs_dirent_lock(&dl, ...) // lock directory entry (may igrab())
	* rw_enter(...); // grab any other locks you need
	* tx = dmu_tx_create(...); // get DMU tx
	* dmu_tx_hold_*(); // hold each object you might modify
	* error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) \| TXG_NOWAIT);
	* if (error) {
	* rw_exit(...); // drop locks
	* zfs_dirent_unlock(dl); // unlock directory entry
	* zrele(...); // release held znodes
	* if (error == ERESTART) {
	* waited = B_TRUE;
	* dmu_tx_wait(tx);
	* dmu_tx_abort(tx);
	* goto top;
	* }
	* dmu_tx_abort(tx); // abort DMU tx
	* ZFS_EXIT(zfsvfs); // finished in zfs
	* return (error); // really out of space
	* }
	* error = do_real_work(); // do whatever this VOP does
	* if (error == 0)
	* zfs_log_*(...); // on success, make ZIL entry
	* dmu_tx_commit(tx); // commit DMU tx -- error or not
	* rw_exit(...); // drop locks
	* zfs_dirent_unlock(dl); // unlock directory entry
	* zrele(...); // release held znodes
	* zil_commit(zilog, foid); // synchronous when necessary
	* ZFS_EXIT(zfsvfs); // finished in zfs
	* return (error); // done, report error
	*/

	/* ARGSUSED */
	int
	zfs_open(struct inode ip, int mode, int flag, cred_t cr)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	/* Honor ZFS_APPENDONLY file attribute */
	if ((mode & FMODE_WRITE) && (zp->z_pflags & ZFS_APPENDONLY) &&
	((flag & O_APPEND) == 0)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	/* Keep a count of the synchronous opens in the znode */
	if (flag & O_SYNC)
	atomic_inc_32(&zp->z_sync_cnt);

	ZFS_EXIT(zfsvfs);
	return (0);
	}

	/* ARGSUSED */
	int
	zfs_close(struct inode ip, int flag, cred_t cr)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	/* Decrement the synchronous opens in the znode */
	if (flag & O_SYNC)
	atomic_dec_32(&zp->z_sync_cnt);

	ZFS_EXIT(zfsvfs);
	return (0);
	}

	#if defined(_KERNEL)
	/*
	* When a file is memory mapped, we must keep the IO data synchronized
	* between the DMU cache and the memory mapped pages. What this means:
	*
	* On Write: If we find a memory mapped page, we write to both
	* the page and the dmu buffer.
	*/
	void
	update_pages(znode_t zp, int64_t start, int len, objset_t os)
	{
	struct inode *ip = ZTOI(zp);
	struct address_space *mp = ip->i_mapping;
	struct page *pp;
	uint64_t nbytes;
	int64_t off;
	void *pb;

	off = start & (PAGE_SIZE-1);
	for (start &= PAGE_MASK; len > 0; start += PAGE_SIZE) {
	nbytes = MIN(PAGE_SIZE - off, len);

	pp = find_lock_page(mp, start >> PAGE_SHIFT);
	if (pp) {
	if (mapping_writably_mapped(mp))
	flush_dcache_page(pp);

	pb = kmap(pp);
	(void) dmu_read(os, zp->z_id, start + off, nbytes,
	pb + off, DMU_READ_PREFETCH);
	kunmap(pp);

	if (mapping_writably_mapped(mp))
	flush_dcache_page(pp);

	mark_page_accessed(pp);
	SetPageUptodate(pp);
	ClearPageError(pp);
	unlock_page(pp);
	put_page(pp);
	}

	len -= nbytes;
	off = 0;
	}
	}

	/*
	* When a file is memory mapped, we must keep the IO data synchronized
	* between the DMU cache and the memory mapped pages. What this means:
	*
	* On Read: We "read" preferentially from memory mapped pages,
	* else we default from the dmu buffer.
	*
	* NOTE: We will always "break up" the IO into PAGESIZE uiomoves when
	* the file is memory mapped.
	*/
	int
	mappedread(znode_t zp, int nbytes, zfs_uio_t uio)
	{
	struct inode *ip = ZTOI(zp);
	struct address_space *mp = ip->i_mapping;
	struct page *pp;
	int64_t start, off;
	uint64_t bytes;
	int len = nbytes;
	int error = 0;
	void *pb;

	start = uio->uio_loffset;
	off = start & (PAGE_SIZE-1);
	for (start &= PAGE_MASK; len > 0; start += PAGE_SIZE) {
	bytes = MIN(PAGE_SIZE - off, len);

	pp = find_lock_page(mp, start >> PAGE_SHIFT);
	if (pp) {
	ASSERT(PageUptodate(pp));
	unlock_page(pp);

	pb = kmap(pp);
	error = zfs_uiomove(pb + off, bytes, UIO_READ, uio);
	kunmap(pp);

	if (mapping_writably_mapped(mp))
	flush_dcache_page(pp);

	mark_page_accessed(pp);
	put_page(pp);
	} else {
	error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl),
	uio, bytes);
	}

	len -= bytes;
	off = 0;
	if (error)
	break;
	}
	return (error);
	}
	#endif /* _KERNEL */

	unsigned long zfs_delete_blocks = DMU_MAX_DELETEBLKCNT;

	/*
	* Write the bytes to a file.
	*
	* IN: zp - znode of file to be written to
	* data - bytes to write
	* len - number of bytes to write
	* pos - offset to start writing at
	*
	* OUT: resid - remaining bytes to write
	*
	* RETURN: 0 if success
	* positive error code if failure. EIO is returned
	* for a short write when residp isn't provided.
	*
	* Timestamps:
	* zp - ctime\|mtime updated if byte count > 0
	*/
	int
	zfs_write_simple(znode_t zp, const void data, size_t len,
	loff_t pos, size_t *residp)
	{
	fstrans_cookie_t cookie;
	int error;

	struct iovec iov;
	iov.iov_base = (void *)data;
	iov.iov_len = len;

	zfs_uio_t uio;
	zfs_uio_iovec_init(&uio, &iov, 1, pos, UIO_SYSSPACE, len, 0);

	cookie = spl_fstrans_mark();
	error = zfs_write(zp, &uio, 0, kcred);
	spl_fstrans_unmark(cookie);

	if (error == 0) {
	if (residp != NULL)
	*residp = zfs_uio_resid(&uio);
	else if (zfs_uio_resid(&uio) != 0)
	error = SET_ERROR(EIO);
	}

	return (error);
	}

	static void
	zfs_rele_async_task(void *arg)
	{
	iput(arg);
	}

	void
	zfs_zrele_async(znode_t *zp)
	{
	struct inode *ip = ZTOI(zp);
	objset_t *os = ITOZSB(ip)->z_os;

	ASSERT(atomic_read(&ip->i_count) > 0);
	ASSERT(os != NULL);

	/*
	* If decrementing the count would put us at 0, we can't do it inline
	* here, because that would be synchronous. Instead, dispatch an iput
	* to run later.
	*
	* For more information on the dangers of a synchronous iput, see the
	* header comment of this file.
	*/
	if (!atomic_add_unless(&ip->i_count, -1, 1)) {
	VERIFY(taskq_dispatch(dsl_pool_zrele_taskq(dmu_objset_pool(os)),
	zfs_rele_async_task, ip, TQ_SLEEP) != TASKQID_INVALID);
	}
	}


	/*
	* Lookup an entry in a directory, or an extended attribute directory.
	* If it exists, return a held inode reference for it.
	*
	* IN: zdp - znode of directory to search.
	* nm - name of entry to lookup.
	* flags - LOOKUP_XATTR set if looking for an attribute.
	* cr - credentials of caller.
	* direntflags - directory lookup flags
	* realpnp - returned pathname.
	*
	* OUT: zpp - znode of located entry, NULL if not found.
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* NA
	*/
	/* ARGSUSED */
	int
	zfs_lookup(znode_t zdp, char nm, znode_t *zpp, int flags, cred_t cr,
	int direntflags, pathname_t realpnp)
	{
	zfsvfs_t *zfsvfs = ZTOZSB(zdp);
	int error = 0;

	/*
	* Fast path lookup, however we must skip DNLC lookup
	* for case folding or normalizing lookups because the
	* DNLC code only stores the passed in name. This means
	* creating 'a' and removing 'A' on a case insensitive
	* file system would work, but DNLC still thinks 'a'
	* exists and won't let you create it again on the next
	* pass through fast path.
	*/
	if (!(flags & (LOOKUP_XATTR \| FIGNORECASE))) {

	if (!S_ISDIR(ZTOI(zdp)->i_mode)) {
	return (SET_ERROR(ENOTDIR));
	} else if (zdp->z_sa_hdl == NULL) {
	return (SET_ERROR(EIO));
	}

	if (nm[0] == 0 \|\| (nm[0] == '.' && nm[1] == '\0')) {
	error = zfs_fastaccesschk_execute(zdp, cr);
	if (!error) {
	*zpp = zdp;
	zhold(*zpp);
	return (0);
	}
	return (error);
	}
	}

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zdp);

	*zpp = NULL;

	if (flags & LOOKUP_XATTR) {
	/*
	* We don't allow recursive attributes..
	* Maybe someday we will.
	*/
	if (zdp->z_pflags & ZFS_XATTR) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	if ((error = zfs_get_xattrdir(zdp, zpp, cr, flags))) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Do we have permission to get into attribute directory?
	*/

	if ((error = zfs_zaccess(*zpp, ACE_EXECUTE, 0,
	B_FALSE, cr))) {
	zrele(*zpp);
	*zpp = NULL;
	}

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if (!S_ISDIR(ZTOI(zdp)->i_mode)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(ENOTDIR));
	}

	/*
	* Check accessibility of directory.
	*/

	if ((error = zfs_zaccess(zdp, ACE_EXECUTE, 0, B_FALSE, cr))) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if (zfsvfs->z_utf8 && u8_validate(nm, strlen(nm),
	NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EILSEQ));
	}

	error = zfs_dirlook(zdp, nm, zpp, flags, direntflags, realpnp);
	if ((error == 0) && (*zpp))
	zfs_znode_update_vfs(*zpp);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Attempt to create a new entry in a directory. If the entry
	* already exists, truncate the file if permissible, else return
	* an error. Return the ip of the created or trunc'd file.
	*
	* IN: dzp - znode of directory to put new file entry in.
	* name - name of new file entry.
	* vap - attributes of new file.
	* excl - flag indicating exclusive or non-exclusive mode.
	* mode - mode to open file with.
	* cr - credentials of caller.
	* flag - file flag.
	* vsecp - ACL to be set
	*
	* OUT: zpp - znode of created or trunc'd entry.
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* dzp - ctime\|mtime updated if new entry created
	* zp - ctime\|mtime always, atime if new
	*/

	/* ARGSUSED */
	int
	zfs_create(znode_t dzp, char name, vattr_t *vap, int excl,
	int mode, znode_t *zpp, cred_t cr, int flag, vsecattr_t *vsecp)
	{
	znode_t *zp;
	zfsvfs_t *zfsvfs = ZTOZSB(dzp);
	zilog_t *zilog;
	objset_t *os;
	zfs_dirlock_t *dl;
	dmu_tx_t *tx;
	int error;
	uid_t uid;
	gid_t gid;
	zfs_acl_ids_t acl_ids;
	boolean_t fuid_dirtied;
	boolean_t have_acl = B_FALSE;
	boolean_t waited = B_FALSE;

	/*
	* If we have an ephemeral id, ACL, or XVATTR then
	* make sure file system is at proper version
	*/

	gid = crgetgid(cr);
	uid = crgetuid(cr);

	if (zfsvfs->z_use_fuids == B_FALSE &&
	(vsecp \|\| IS_EPHEMERAL(uid) \|\| IS_EPHEMERAL(gid)))
	return (SET_ERROR(EINVAL));

	if (name == NULL)
	return (SET_ERROR(EINVAL));

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(dzp);
	os = zfsvfs->z_os;
	zilog = zfsvfs->z_log;

	if (zfsvfs->z_utf8 && u8_validate(name, strlen(name),
	NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EILSEQ));
	}

	if (vap->va_mask & ATTR_XVATTR) {
	if ((error = secpolicy_xvattr((xvattr_t *)vap,
	crgetuid(cr), cr, vap->va_mode)) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	}

	top:
	*zpp = NULL;
	if (*name == '\0') {
	/*
	* Null component name refers to the directory itself.
	*/
	zhold(dzp);
	zp = dzp;
	dl = NULL;
	error = 0;
	} else {
	/* possible igrab(zp) */
	int zflg = 0;

	if (flag & FIGNORECASE)
	zflg \|= ZCILOOK;

	error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
	NULL, NULL);
	if (error) {
	if (have_acl)
	zfs_acl_ids_free(&acl_ids);
	if (strcmp(name, "..") == 0)
	error = SET_ERROR(EISDIR);
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	}

	if (zp == NULL) {
	uint64_t txtype;
	uint64_t projid = ZFS_DEFAULT_PROJID;

	/*
	* Create a new file object and update the directory
	* to reference it.
	*/
	if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
	if (have_acl)
	zfs_acl_ids_free(&acl_ids);
	goto out;
	}

	/*
	* We only support the creation of regular files in
	* extended attribute directories.
	*/

	if ((dzp->z_pflags & ZFS_XATTR) && !S_ISREG(vap->va_mode)) {
	if (have_acl)
	zfs_acl_ids_free(&acl_ids);
	error = SET_ERROR(EINVAL);
	goto out;
	}

	if (!have_acl && (error = zfs_acl_ids_create(dzp, 0, vap,
	cr, vsecp, &acl_ids)) != 0)
	goto out;
	have_acl = B_TRUE;

	if (S_ISREG(vap->va_mode) \|\| S_ISDIR(vap->va_mode))
	projid = zfs_inherit_projid(dzp);
	if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, projid)) {
	zfs_acl_ids_free(&acl_ids);
	error = SET_ERROR(EDQUOT);
	goto out;
	}

	tx = dmu_tx_create(os);

	dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
	ZFS_SA_BASE_ATTR_SIZE);

	fuid_dirtied = zfsvfs->z_fuid_dirty;
	if (fuid_dirtied)
	zfs_fuid_txhold(zfsvfs, tx);
	dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
	dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
	if (!zfsvfs->z_use_sa &&
	acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
	0, acl_ids.z_aclp->z_acl_bytes);
	}

	error = dmu_tx_assign(tx,
	(waited ? TXG_NOTHROTTLE : 0) \| TXG_NOWAIT);
	if (error) {
	zfs_dirent_unlock(dl);
	if (error == ERESTART) {
	waited = B_TRUE;
	dmu_tx_wait(tx);
	dmu_tx_abort(tx);
	goto top;
	}
	zfs_acl_ids_free(&acl_ids);
	dmu_tx_abort(tx);
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);

	error = zfs_link_create(dl, zp, tx, ZNEW);
	if (error != 0) {
	/*
	* Since, we failed to add the directory entry for it,
	* delete the newly created dnode.
	*/
	zfs_znode_delete(zp, tx);
	remove_inode_hash(ZTOI(zp));
	zfs_acl_ids_free(&acl_ids);
	dmu_tx_commit(tx);
	goto out;
	}

	if (fuid_dirtied)
	zfs_fuid_sync(zfsvfs, tx);

	txtype = zfs_log_create_txtype(Z_FILE, vsecp, vap);
	if (flag & FIGNORECASE)
	txtype \|= TX_CI;
	zfs_log_create(zilog, tx, txtype, dzp, zp, name,
	vsecp, acl_ids.z_fuidp, vap);
	zfs_acl_ids_free(&acl_ids);
	dmu_tx_commit(tx);
	} else {
	int aflags = (flag & O_APPEND) ? V_APPEND : 0;

	if (have_acl)
	zfs_acl_ids_free(&acl_ids);
	have_acl = B_FALSE;

	/*
	* A directory entry already exists for this name.
	*/
	/*
	* Can't truncate an existing file if in exclusive mode.
	*/
	if (excl) {
	error = SET_ERROR(EEXIST);
	goto out;
	}
	/*
	* Can't open a directory for writing.
	*/
	if (S_ISDIR(ZTOI(zp)->i_mode)) {
	error = SET_ERROR(EISDIR);
	goto out;
	}
	/*
	* Verify requested access to file.
	*/
	if (mode && (error = zfs_zaccess_rwx(zp, mode, aflags, cr))) {
	goto out;
	}

	mutex_enter(&dzp->z_lock);
	dzp->z_seq++;
	mutex_exit(&dzp->z_lock);

	/*
	* Truncate regular files if requested.
	*/
	if (S_ISREG(ZTOI(zp)->i_mode) &&
	(vap->va_mask & ATTR_SIZE) && (vap->va_size == 0)) {
	/* we can't hold any locks when calling zfs_freesp() */
	if (dl) {
	zfs_dirent_unlock(dl);
	dl = NULL;
	}
	error = zfs_freesp(zp, 0, 0, mode, TRUE);
	}
	}
	out:

	if (dl)
	zfs_dirent_unlock(dl);

	if (error) {
	if (zp)
	zrele(zp);
	} else {
	zfs_znode_update_vfs(dzp);
	zfs_znode_update_vfs(zp);
	*zpp = zp;
	}

	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/* ARGSUSED */
	int
	zfs_tmpfile(struct inode dip, vattr_t vap, int excl,
	int mode, struct inode *ipp, cred_t cr, int flag, vsecattr_t *vsecp)
	{
	znode_t zp = NULL, dzp = ITOZ(dip);
	zfsvfs_t *zfsvfs = ITOZSB(dip);
	objset_t *os;
	dmu_tx_t *tx;
	int error;
	uid_t uid;
	gid_t gid;
	zfs_acl_ids_t acl_ids;
	uint64_t projid = ZFS_DEFAULT_PROJID;
	boolean_t fuid_dirtied;
	boolean_t have_acl = B_FALSE;
	boolean_t waited = B_FALSE;

	/*
	* If we have an ephemeral id, ACL, or XVATTR then
	* make sure file system is at proper version
	*/

	gid = crgetgid(cr);
	uid = crgetuid(cr);

	if (zfsvfs->z_use_fuids == B_FALSE &&
	(vsecp \|\| IS_EPHEMERAL(uid) \|\| IS_EPHEMERAL(gid)))
	return (SET_ERROR(EINVAL));

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(dzp);
	os = zfsvfs->z_os;

	if (vap->va_mask & ATTR_XVATTR) {
	if ((error = secpolicy_xvattr((xvattr_t *)vap,
	crgetuid(cr), cr, vap->va_mode)) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	}

	top:
	*ipp = NULL;

	/*
	* Create a new file object and update the directory
	* to reference it.
	*/
	if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
	if (have_acl)
	zfs_acl_ids_free(&acl_ids);
	goto out;
	}

	if (!have_acl && (error = zfs_acl_ids_create(dzp, 0, vap,
	cr, vsecp, &acl_ids)) != 0)
	goto out;
	have_acl = B_TRUE;

	if (S_ISREG(vap->va_mode) \|\| S_ISDIR(vap->va_mode))
	projid = zfs_inherit_projid(dzp);
	if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, projid)) {
	zfs_acl_ids_free(&acl_ids);
	error = SET_ERROR(EDQUOT);
	goto out;
	}

	tx = dmu_tx_create(os);

	dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
	ZFS_SA_BASE_ATTR_SIZE);
	dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);

	fuid_dirtied = zfsvfs->z_fuid_dirty;
	if (fuid_dirtied)
	zfs_fuid_txhold(zfsvfs, tx);
	if (!zfsvfs->z_use_sa &&
	acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
	0, acl_ids.z_aclp->z_acl_bytes);
	}
	error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) \| TXG_NOWAIT);
	if (error) {
	if (error == ERESTART) {
	waited = B_TRUE;
	dmu_tx_wait(tx);
	dmu_tx_abort(tx);
	goto top;
	}
	zfs_acl_ids_free(&acl_ids);
	dmu_tx_abort(tx);
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	zfs_mknode(dzp, vap, tx, cr, IS_TMPFILE, &zp, &acl_ids);

	if (fuid_dirtied)
	zfs_fuid_sync(zfsvfs, tx);

	/* Add to unlinked set */
	zp->z_unlinked = B_TRUE;
	zfs_unlinked_add(zp, tx);
	zfs_acl_ids_free(&acl_ids);
	dmu_tx_commit(tx);
	out:

	if (error) {
	if (zp)
	zrele(zp);
	} else {
	zfs_znode_update_vfs(dzp);
	zfs_znode_update_vfs(zp);
	*ipp = ZTOI(zp);
	}

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Remove an entry from a directory.
	*
	* IN: dzp - znode of directory to remove entry from.
	* name - name of entry to remove.
	* cr - credentials of caller.
	* flags - case flags.
	*
	* RETURN: 0 if success
	* error code if failure
	*
	* Timestamps:
	* dzp - ctime\|mtime
	* ip - ctime (if nlink > 0)
	*/

	uint64_t null_xattr = 0;

	/ARGSUSED/
	int
	zfs_remove(znode_t dzp, char name, cred_t *cr, int flags)
	{
	znode_t *zp;
	znode_t *xzp;
	zfsvfs_t *zfsvfs = ZTOZSB(dzp);
	zilog_t *zilog;
	uint64_t acl_obj, xattr_obj;
	uint64_t xattr_obj_unlinked = 0;
	uint64_t obj = 0;
	uint64_t links;
	zfs_dirlock_t *dl;
	dmu_tx_t *tx;
	boolean_t may_delete_now, delete_now = FALSE;
	boolean_t unlinked, toobig = FALSE;
	uint64_t txtype;
	pathname_t *realnmp = NULL;
	pathname_t realnm;
	int error;
	int zflg = ZEXISTS;
	boolean_t waited = B_FALSE;

	if (name == NULL)
	return (SET_ERROR(EINVAL));

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(dzp);
	zilog = zfsvfs->z_log;

	if (flags & FIGNORECASE) {
	zflg \|= ZCILOOK;
	pn_alloc(&realnm);
	realnmp = &realnm;
	}

	top:
	xattr_obj = 0;
	xzp = NULL;
	/*
	* Attempt to lock directory; fail if entry doesn't exist.
	*/
	if ((error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
	NULL, realnmp))) {
	if (realnmp)
	pn_free(realnmp);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if ((error = zfs_zaccess_delete(dzp, zp, cr))) {
	goto out;
	}

	/*
	* Need to use rmdir for removing directories.
	*/
	if (S_ISDIR(ZTOI(zp)->i_mode)) {
	error = SET_ERROR(EPERM);
	goto out;
	}

	mutex_enter(&zp->z_lock);
	may_delete_now = atomic_read(&ZTOI(zp)->i_count) == 1 &&
	!(zp->z_is_mapped);
	mutex_exit(&zp->z_lock);

	/*
	* We may delete the znode now, or we may put it in the unlinked set;
	* it depends on whether we're the last link, and on whether there are
	* other holds on the inode. So we dmu_tx_hold() the right things to
	* allow for either case.
	*/
	obj = zp->z_id;
	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name);
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
	zfs_sa_upgrade_txholds(tx, zp);
	zfs_sa_upgrade_txholds(tx, dzp);
	if (may_delete_now) {
	toobig = zp->z_size > zp->z_blksz * zfs_delete_blocks;
	/* if the file is too big, only hold_free a token amount */
	dmu_tx_hold_free(tx, zp->z_id, 0,
	(toobig ? DMU_MAX_ACCESS : DMU_OBJECT_END));
	}

	/* are there any extended attributes? */
	error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
	&xattr_obj, sizeof (xattr_obj));
	if (error == 0 && xattr_obj) {
	error = zfs_zget(zfsvfs, xattr_obj, &xzp);
	ASSERT0(error);
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
	dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
	}

	mutex_enter(&zp->z_lock);
	if ((acl_obj = zfs_external_acl(zp)) != 0 && may_delete_now)
	dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END);
	mutex_exit(&zp->z_lock);

	/* charge as an update -- would be nice not to charge at all */
	dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);

	/*
	* Mark this transaction as typically resulting in a net free of space
	*/
	dmu_tx_mark_netfree(tx);

	error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) \| TXG_NOWAIT);
	if (error) {
	zfs_dirent_unlock(dl);
	if (error == ERESTART) {
	waited = B_TRUE;
	dmu_tx_wait(tx);
	dmu_tx_abort(tx);
	zrele(zp);
	if (xzp)
	zrele(xzp);
	goto top;
	}
	if (realnmp)
	pn_free(realnmp);
	dmu_tx_abort(tx);
	zrele(zp);
	if (xzp)
	zrele(xzp);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Remove the directory entry.
	*/
	error = zfs_link_destroy(dl, zp, tx, zflg, &unlinked);

	if (error) {
	dmu_tx_commit(tx);
	goto out;
	}

	if (unlinked) {
	/*
	* Hold z_lock so that we can make sure that the ACL obj
	* hasn't changed. Could have been deleted due to
	* zfs_sa_upgrade().
	*/
	mutex_enter(&zp->z_lock);
	(void) sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
	&xattr_obj_unlinked, sizeof (xattr_obj_unlinked));
	delete_now = may_delete_now && !toobig &&
	atomic_read(&ZTOI(zp)->i_count) == 1 &&
	!(zp->z_is_mapped) && xattr_obj == xattr_obj_unlinked &&
	zfs_external_acl(zp) == acl_obj;
	}

	if (delete_now) {
	if (xattr_obj_unlinked) {
	ASSERT3U(ZTOI(xzp)->i_nlink, ==, 2);
	mutex_enter(&xzp->z_lock);
	xzp->z_unlinked = B_TRUE;
	clear_nlink(ZTOI(xzp));
	links = 0;
	error = sa_update(xzp->z_sa_hdl, SA_ZPL_LINKS(zfsvfs),
	&links, sizeof (links), tx);
	ASSERT3U(error, ==, 0);
	mutex_exit(&xzp->z_lock);
	zfs_unlinked_add(xzp, tx);

	if (zp->z_is_sa)
	error = sa_remove(zp->z_sa_hdl,
	SA_ZPL_XATTR(zfsvfs), tx);
	else
	error = sa_update(zp->z_sa_hdl,
	SA_ZPL_XATTR(zfsvfs), &null_xattr,
	sizeof (uint64_t), tx);
	ASSERT0(error);
	}
	/*
	* Add to the unlinked set because a new reference could be
	* taken concurrently resulting in a deferred destruction.
	*/
	zfs_unlinked_add(zp, tx);
	mutex_exit(&zp->z_lock);
	} else if (unlinked) {
	mutex_exit(&zp->z_lock);
	zfs_unlinked_add(zp, tx);
	}

	txtype = TX_REMOVE;
	if (flags & FIGNORECASE)
	txtype \|= TX_CI;
	zfs_log_remove(zilog, tx, txtype, dzp, name, obj, unlinked);

	dmu_tx_commit(tx);
	out:
	if (realnmp)
	pn_free(realnmp);

	zfs_dirent_unlock(dl);
	zfs_znode_update_vfs(dzp);
	zfs_znode_update_vfs(zp);

	if (delete_now)
	zrele(zp);
	else
	zfs_zrele_async(zp);

	if (xzp) {
	zfs_znode_update_vfs(xzp);
	zfs_zrele_async(xzp);
	}

	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Create a new directory and insert it into dzp using the name
	* provided. Return a pointer to the inserted directory.
	*
	* IN: dzp - znode of directory to add subdir to.
	* dirname - name of new directory.
	* vap - attributes of new directory.
	* cr - credentials of caller.
	* flags - case flags.
	* vsecp - ACL to be set
	*
	* OUT: zpp - znode of created directory.
	*
	* RETURN: 0 if success
	* error code if failure
	*
	* Timestamps:
	* dzp - ctime\|mtime updated
	* zpp - ctime\|mtime\|atime updated
	*/
	/ARGSUSED/
	int
	zfs_mkdir(znode_t dzp, char dirname, vattr_t vap, znode_t *zpp,
	cred_t cr, int flags, vsecattr_t vsecp)
	{
	znode_t *zp;
	zfsvfs_t *zfsvfs = ZTOZSB(dzp);
	zilog_t *zilog;
	zfs_dirlock_t *dl;
	uint64_t txtype;
	dmu_tx_t *tx;
	int error;
	int zf = ZNEW;
	uid_t uid;
	gid_t gid = crgetgid(cr);
	zfs_acl_ids_t acl_ids;
	boolean_t fuid_dirtied;
	boolean_t waited = B_FALSE;

	ASSERT(S_ISDIR(vap->va_mode));

	/*
	* If we have an ephemeral id, ACL, or XVATTR then
	* make sure file system is at proper version
	*/

	uid = crgetuid(cr);
	if (zfsvfs->z_use_fuids == B_FALSE &&
	(vsecp \|\| IS_EPHEMERAL(uid) \|\| IS_EPHEMERAL(gid)))
	return (SET_ERROR(EINVAL));

	if (dirname == NULL)
	return (SET_ERROR(EINVAL));

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(dzp);
	zilog = zfsvfs->z_log;

	if (dzp->z_pflags & ZFS_XATTR) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	if (zfsvfs->z_utf8 && u8_validate(dirname,
	strlen(dirname), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EILSEQ));
	}
	if (flags & FIGNORECASE)
	zf \|= ZCILOOK;

	if (vap->va_mask & ATTR_XVATTR) {
	if ((error = secpolicy_xvattr((xvattr_t *)vap,
	crgetuid(cr), cr, vap->va_mode)) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	}

	if ((error = zfs_acl_ids_create(dzp, 0, vap, cr,
	vsecp, &acl_ids)) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	/*
	* First make sure the new directory doesn't exist.
	*
	* Existence is checked first to make sure we don't return
	* EACCES instead of EEXIST which can cause some applications
	* to fail.
	*/
	top:
	*zpp = NULL;

	if ((error = zfs_dirent_lock(&dl, dzp, dirname, &zp, zf,
	NULL, NULL))) {
	zfs_acl_ids_free(&acl_ids);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if ((error = zfs_zaccess(dzp, ACE_ADD_SUBDIRECTORY, 0, B_FALSE, cr))) {
	zfs_acl_ids_free(&acl_ids);
	zfs_dirent_unlock(dl);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, zfs_inherit_projid(dzp))) {
	zfs_acl_ids_free(&acl_ids);
	zfs_dirent_unlock(dl);
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EDQUOT));
	}

	/*
	* Add a new entry to the directory.
	*/
	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_zap(tx, dzp->z_id, TRUE, dirname);
	dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
	fuid_dirtied = zfsvfs->z_fuid_dirty;
	if (fuid_dirtied)
	zfs_fuid_txhold(zfsvfs, tx);
	if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
	acl_ids.z_aclp->z_acl_bytes);
	}

	dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
	ZFS_SA_BASE_ATTR_SIZE);

	error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) \| TXG_NOWAIT);
	if (error) {
	zfs_dirent_unlock(dl);
	if (error == ERESTART) {
	waited = B_TRUE;
	dmu_tx_wait(tx);
	dmu_tx_abort(tx);
	goto top;
	}
	zfs_acl_ids_free(&acl_ids);
	dmu_tx_abort(tx);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Create new node.
	*/
	zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);

	/*
	* Now put new name in parent dir.
	*/
	error = zfs_link_create(dl, zp, tx, ZNEW);
	if (error != 0) {
	zfs_znode_delete(zp, tx);
	remove_inode_hash(ZTOI(zp));
	goto out;
	}

	if (fuid_dirtied)
	zfs_fuid_sync(zfsvfs, tx);

	*zpp = zp;

	txtype = zfs_log_create_txtype(Z_DIR, vsecp, vap);
	if (flags & FIGNORECASE)
	txtype \|= TX_CI;
	zfs_log_create(zilog, tx, txtype, dzp, zp, dirname, vsecp,
	acl_ids.z_fuidp, vap);

	out:
	zfs_acl_ids_free(&acl_ids);

	dmu_tx_commit(tx);

	zfs_dirent_unlock(dl);

	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	if (error != 0) {
	zrele(zp);
	} else {
	zfs_znode_update_vfs(dzp);
	zfs_znode_update_vfs(zp);
	}
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Remove a directory subdir entry. If the current working
	* directory is the same as the subdir to be removed, the
	* remove will fail.
	*
	* IN: dzp - znode of directory to remove from.
	* name - name of directory to be removed.
	* cwd - inode of current working directory.
	* cr - credentials of caller.
	* flags - case flags
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* dzp - ctime\|mtime updated
	*/
	/ARGSUSED/
	int
	zfs_rmdir(znode_t dzp, char name, znode_t cwd, cred_t cr,
	int flags)
	{
	znode_t *zp;
	zfsvfs_t *zfsvfs = ZTOZSB(dzp);
	zilog_t *zilog;
	zfs_dirlock_t *dl;
	dmu_tx_t *tx;
	int error;
	int zflg = ZEXISTS;
	boolean_t waited = B_FALSE;

	if (name == NULL)
	return (SET_ERROR(EINVAL));

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(dzp);
	zilog = zfsvfs->z_log;

	if (flags & FIGNORECASE)
	zflg \|= ZCILOOK;
	top:
	zp = NULL;

	/*
	* Attempt to lock directory; fail if entry doesn't exist.
	*/
	if ((error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
	NULL, NULL))) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if ((error = zfs_zaccess_delete(dzp, zp, cr))) {
	goto out;
	}

	if (!S_ISDIR(ZTOI(zp)->i_mode)) {
	error = SET_ERROR(ENOTDIR);
	goto out;
	}

	if (zp == cwd) {
	error = SET_ERROR(EINVAL);
	goto out;
	}

	/*
	* Grab a lock on the directory to make sure that no one is
	* trying to add (or lookup) entries while we are removing it.
	*/
	rw_enter(&zp->z_name_lock, RW_WRITER);

	/*
	* Grab a lock on the parent pointer to make sure we play well
	* with the treewalk and directory rename code.
	*/
	rw_enter(&zp->z_parent_lock, RW_WRITER);

	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name);
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
	dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
	zfs_sa_upgrade_txholds(tx, zp);
	zfs_sa_upgrade_txholds(tx, dzp);
	dmu_tx_mark_netfree(tx);
	error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) \| TXG_NOWAIT);
	if (error) {
	rw_exit(&zp->z_parent_lock);
	rw_exit(&zp->z_name_lock);
	zfs_dirent_unlock(dl);
	if (error == ERESTART) {
	waited = B_TRUE;
	dmu_tx_wait(tx);
	dmu_tx_abort(tx);
	zrele(zp);
	goto top;
	}
	dmu_tx_abort(tx);
	zrele(zp);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	error = zfs_link_destroy(dl, zp, tx, zflg, NULL);

	if (error == 0) {
	uint64_t txtype = TX_RMDIR;
	if (flags & FIGNORECASE)
	txtype \|= TX_CI;
	zfs_log_remove(zilog, tx, txtype, dzp, name, ZFS_NO_OBJECT,
	B_FALSE);
	}

	dmu_tx_commit(tx);

	rw_exit(&zp->z_parent_lock);
	rw_exit(&zp->z_name_lock);
	out:
	zfs_dirent_unlock(dl);

	zfs_znode_update_vfs(dzp);
	zfs_znode_update_vfs(zp);
	zrele(zp);

	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Read directory entries from the given directory cursor position and emit
	* name and position for each entry.
	*
	* IN: ip - inode of directory to read.
	* ctx - directory entry context.
	* cr - credentials of caller.
	*
	* RETURN: 0 if success
	* error code if failure
	*
	* Timestamps:
	* ip - atime updated
	*
	* Note that the low 4 bits of the cookie returned by zap is always zero.
	* This allows us to use the low range for "special" directory entries:
	* We use 0 for '.', and 1 for '..'. If this is the root of the filesystem,
	* we use the offset 2 for the '.zfs' directory.
	*/
	/* ARGSUSED */
	int
	zfs_readdir(struct inode ip, zpl_dir_context_t ctx, cred_t *cr)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);
	objset_t *os;
	zap_cursor_t zc;
	zap_attribute_t zap;
	int error;
	uint8_t prefetch;
	uint8_t type;
	int done = 0;
	uint64_t parent;
	uint64_t offset; /* must be unsigned; checks for < 1 */

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
	&parent, sizeof (parent))) != 0)
	goto out;

	/*
	* Quit if directory has been removed (posix)
	*/
	if (zp->z_unlinked)
	goto out;

	error = 0;
	os = zfsvfs->z_os;
	offset = ctx->pos;
	prefetch = zp->z_zn_prefetch;

	/*
	* Initialize the iterator cursor.
	*/
	if (offset <= 3) {
	/*
	* Start iteration from the beginning of the directory.
	*/
	zap_cursor_init(&zc, os, zp->z_id);
	} else {
	/*
	* The offset is a serialized cursor.
	*/
	zap_cursor_init_serialized(&zc, os, zp->z_id, offset);
	}

	/*
	* Transform to file-system independent format
	*/
	while (!done) {
	uint64_t objnum;
	/*
	* Special case `.', `..', and `.zfs'.
	*/
	if (offset == 0) {
	(void) strcpy(zap.za_name, ".");
	zap.za_normalization_conflict = 0;
	objnum = zp->z_id;
	type = DT_DIR;
	} else if (offset == 1) {
	(void) strcpy(zap.za_name, "..");
	zap.za_normalization_conflict = 0;
	objnum = parent;
	type = DT_DIR;
	} else if (offset == 2 && zfs_show_ctldir(zp)) {
	(void) strcpy(zap.za_name, ZFS_CTLDIR_NAME);
	zap.za_normalization_conflict = 0;
	objnum = ZFSCTL_INO_ROOT;
	type = DT_DIR;
	} else {
	/*
	* Grab next entry.
	*/
	if ((error = zap_cursor_retrieve(&zc, &zap))) {
	if (error == ENOENT)
	break;
	else
	goto update;
	}

	/*
	* Allow multiple entries provided the first entry is
	* the object id. Non-zpl consumers may safely make
	* use of the additional space.
	*
	* XXX: This should be a feature flag for compatibility
	*/
	if (zap.za_integer_length != 8 \|\|
	zap.za_num_integers == 0) {
	cmn_err(CE_WARN, "zap_readdir: bad directory "
	"entry, obj = %lld, offset = %lld, "
	"length = %d, num = %lld\n",
	(u_longlong_t)zp->z_id,
	(u_longlong_t)offset,
	zap.za_integer_length,
	(u_longlong_t)zap.za_num_integers);
	error = SET_ERROR(ENXIO);
	goto update;
	}

	objnum = ZFS_DIRENT_OBJ(zap.za_first_integer);
	type = ZFS_DIRENT_TYPE(zap.za_first_integer);
	}

	done = !zpl_dir_emit(ctx, zap.za_name, strlen(zap.za_name),
	objnum, type);
	if (done)
	break;

	/* Prefetch znode */
	if (prefetch) {
	dmu_prefetch(os, objnum, 0, 0, 0,
	ZIO_PRIORITY_SYNC_READ);
	}

	/*
	* Move to the next entry, fill in the previous offset.
	*/
	if (offset > 2 \|\| (offset == 2 && !zfs_show_ctldir(zp))) {
	zap_cursor_advance(&zc);
	offset = zap_cursor_serialize(&zc);
	} else {
	offset += 1;
	}
	ctx->pos = offset;
	}
	zp->z_zn_prefetch = B_FALSE; /* a lookup will re-enable pre-fetching */

	update:
	zap_cursor_fini(&zc);
	if (error == ENOENT)
	error = 0;
	out:
	ZFS_EXIT(zfsvfs);

	return (error);
	}

	/*
	* Get the basic file attributes and place them in the provided kstat
	* structure. The inode is assumed to be the authoritative source
	* for most of the attributes. However, the znode currently has the
	* authoritative atime, blksize, and block count.
	*
	* IN: ip - inode of file.
	*
	* OUT: sp - kstat values.
	*
	* RETURN: 0 (always succeeds)
	*/
	/* ARGSUSED */
	int
	zfs_getattr_fast(struct user_namespace user_ns, struct inode ip,
	struct kstat *sp)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);
	uint32_t blksize;
	u_longlong_t nblocks;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	mutex_enter(&zp->z_lock);

	zpl_generic_fillattr(user_ns, ip, sp);
	/*
	* +1 link count for root inode with visible '.zfs' directory.
	*/
	if ((zp->z_id == zfsvfs->z_root) && zfs_show_ctldir(zp))
	if (sp->nlink < ZFS_LINK_MAX)
	sp->nlink++;

	sa_object_size(zp->z_sa_hdl, &blksize, &nblocks);
	sp->blksize = blksize;
	sp->blocks = nblocks;

	if (unlikely(zp->z_blksz == 0)) {
	/*
	* Block size hasn't been set; suggest maximal I/O transfers.
	*/
	sp->blksize = zfsvfs->z_max_blksz;
	}

	mutex_exit(&zp->z_lock);

	/*
	* Required to prevent NFS client from detecting different inode
	* numbers of snapshot root dentry before and after snapshot mount.
	*/
	if (zfsvfs->z_issnap) {
	if (ip->i_sb->s_root->d_inode == ip)
	sp->ino = ZFSCTL_INO_SNAPDIRS -
	dmu_objset_id(zfsvfs->z_os);
	}

	ZFS_EXIT(zfsvfs);

	return (0);
	}

	/*
	* For the operation of changing file's user/group/project, we need to
	* handle not only the main object that is assigned to the file directly,
	* but also the ones that are used by the file via hidden xattr directory.
	*
	* Because the xattr directory may contains many EA entries, as to it may
	* be impossible to change all of them via the transaction of changing the
	* main object's user/group/project attributes. Then we have to change them
	* via other multiple independent transactions one by one. It may be not good
	* solution, but we have no better idea yet.
	*/
	static int
	zfs_setattr_dir(znode_t *dzp)
	{
	struct inode *dxip = ZTOI(dzp);
	struct inode *xip = NULL;
	zfsvfs_t *zfsvfs = ZTOZSB(dzp);
	objset_t *os = zfsvfs->z_os;
	zap_cursor_t zc;
	zap_attribute_t zap;
	zfs_dirlock_t *dl;
	znode_t *zp = NULL;
	dmu_tx_t *tx = NULL;
	uint64_t uid, gid;
	sa_bulk_attr_t bulk[4];
	int count;
	int err;

	zap_cursor_init(&zc, os, dzp->z_id);
	while ((err = zap_cursor_retrieve(&zc, &zap)) == 0) {
	count = 0;
	if (zap.za_integer_length != 8 \|\| zap.za_num_integers != 1) {
	err = ENXIO;
	break;
	}

	err = zfs_dirent_lock(&dl, dzp, (char *)zap.za_name, &zp,
	ZEXISTS, NULL, NULL);
	if (err == ENOENT)
	goto next;
	if (err)
	break;

	xip = ZTOI(zp);
	if (KUID_TO_SUID(xip->i_uid) == KUID_TO_SUID(dxip->i_uid) &&
	KGID_TO_SGID(xip->i_gid) == KGID_TO_SGID(dxip->i_gid) &&
	zp->z_projid == dzp->z_projid)
	goto next;

	tx = dmu_tx_create(os);
	if (!(zp->z_pflags & ZFS_PROJID))
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
	else
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);

	err = dmu_tx_assign(tx, TXG_WAIT);
	if (err)
	break;

	mutex_enter(&dzp->z_lock);

	if (KUID_TO_SUID(xip->i_uid) != KUID_TO_SUID(dxip->i_uid)) {
	xip->i_uid = dxip->i_uid;
	uid = zfs_uid_read(dxip);
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
	&uid, sizeof (uid));
	}

	if (KGID_TO_SGID(xip->i_gid) != KGID_TO_SGID(dxip->i_gid)) {
	xip->i_gid = dxip->i_gid;
	gid = zfs_gid_read(dxip);
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs), NULL,
	&gid, sizeof (gid));
	}

	if (zp->z_projid != dzp->z_projid) {
	if (!(zp->z_pflags & ZFS_PROJID)) {
	zp->z_pflags \|= ZFS_PROJID;
	SA_ADD_BULK_ATTR(bulk, count,
	SA_ZPL_FLAGS(zfsvfs), NULL, &zp->z_pflags,
	sizeof (zp->z_pflags));
	}

	zp->z_projid = dzp->z_projid;
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PROJID(zfsvfs),
	NULL, &zp->z_projid, sizeof (zp->z_projid));
	}

	mutex_exit(&dzp->z_lock);

	if (likely(count > 0)) {
	err = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
	dmu_tx_commit(tx);
	} else {
	dmu_tx_abort(tx);
	}
	tx = NULL;
	if (err != 0 && err != ENOENT)
	break;

	next:
	if (zp) {
	zrele(zp);
	zp = NULL;
	zfs_dirent_unlock(dl);
	}
	zap_cursor_advance(&zc);
	}

	if (tx)
	dmu_tx_abort(tx);
	if (zp) {
	zrele(zp);
	zfs_dirent_unlock(dl);
	}
	zap_cursor_fini(&zc);

	return (err == ENOENT ? 0 : err);
	}

	/*
	* Set the file attributes to the values contained in the
	* vattr structure.
	*
	* IN: zp - znode of file to be modified.
	* vap - new attribute values.
	* If ATTR_XVATTR set, then optional attrs are being set
	* flags - ATTR_UTIME set if non-default time values provided.
	* - ATTR_NOACLCHECK (CIFS context only).
	* cr - credentials of caller.
	*
	* RETURN: 0 if success
	* error code if failure
	*
	* Timestamps:
	* ip - ctime updated, mtime updated if size changed.
	*/
	/* ARGSUSED */
	int
	zfs_setattr(znode_t zp, vattr_t vap, int flags, cred_t *cr)
	{
	struct inode *ip;
	zfsvfs_t *zfsvfs = ZTOZSB(zp);
	objset_t *os = zfsvfs->z_os;
	zilog_t *zilog;
	dmu_tx_t *tx;
	vattr_t oldva;
	xvattr_t *tmpxvattr;
	uint_t mask = vap->va_mask;
	uint_t saved_mask = 0;
	int trim_mask = 0;
	uint64_t new_mode;
	uint64_t new_kuid = 0, new_kgid = 0, new_uid, new_gid;
	uint64_t xattr_obj;
	uint64_t mtime[2], ctime[2], atime[2];
	uint64_t projid = ZFS_INVALID_PROJID;
	znode_t *attrzp;
	int need_policy = FALSE;
	int err, err2 = 0;
	zfs_fuid_info_t *fuidp = NULL;
	xvattr_t xvap = (xvattr_t )vap; /* vap may be an xvattr_t * */
	xoptattr_t *xoap;
	zfs_acl_t *aclp;
	boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
	boolean_t fuid_dirtied = B_FALSE;
	boolean_t handle_eadir = B_FALSE;
	sa_bulk_attr_t bulk, xattr_bulk;
	int count = 0, xattr_count = 0, bulks = 8;

	if (mask == 0)
	return (0);

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);
	ip = ZTOI(zp);

	/*
	* If this is a xvattr_t, then get a pointer to the structure of
	* optional attributes. If this is NULL, then we have a vattr_t.
	*/
	xoap = xva_getxoptattr(xvap);
	if (xoap != NULL && (mask & ATTR_XVATTR)) {
	if (XVA_ISSET_REQ(xvap, XAT_PROJID)) {
	if (!dmu_objset_projectquota_enabled(os) \|\|
	(!S_ISREG(ip->i_mode) && !S_ISDIR(ip->i_mode))) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(ENOTSUP));
	}

	projid = xoap->xoa_projid;
	if (unlikely(projid == ZFS_INVALID_PROJID)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	if (projid == zp->z_projid && zp->z_pflags & ZFS_PROJID)
	projid = ZFS_INVALID_PROJID;
	else
	need_policy = TRUE;
	}

	if (XVA_ISSET_REQ(xvap, XAT_PROJINHERIT) &&
	(xoap->xoa_projinherit !=
	((zp->z_pflags & ZFS_PROJINHERIT) != 0)) &&
	(!dmu_objset_projectquota_enabled(os) \|\|
	(!S_ISREG(ip->i_mode) && !S_ISDIR(ip->i_mode)))) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(ENOTSUP));
	}
	}

	zilog = zfsvfs->z_log;

	/*
	* Make sure that if we have ephemeral uid/gid or xvattr specified
	* that file system is at proper version level
	*/

	if (zfsvfs->z_use_fuids == B_FALSE &&
	(((mask & ATTR_UID) && IS_EPHEMERAL(vap->va_uid)) \|\|
	((mask & ATTR_GID) && IS_EPHEMERAL(vap->va_gid)) \|\|
	(mask & ATTR_XVATTR))) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	if (mask & ATTR_SIZE && S_ISDIR(ip->i_mode)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EISDIR));
	}

	if (mask & ATTR_SIZE && !S_ISREG(ip->i_mode) && !S_ISFIFO(ip->i_mode)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	tmpxvattr = kmem_alloc(sizeof (xvattr_t), KM_SLEEP);
	xva_init(tmpxvattr);

	bulk = kmem_alloc(sizeof (sa_bulk_attr_t) * bulks, KM_SLEEP);
	xattr_bulk = kmem_alloc(sizeof (sa_bulk_attr_t) * bulks, KM_SLEEP);

	/*
	* Immutable files can only alter immutable bit and atime
	*/
	if ((zp->z_pflags & ZFS_IMMUTABLE) &&
	((mask & (ATTR_SIZE\|ATTR_UID\|ATTR_GID\|ATTR_MTIME\|ATTR_MODE)) \|\|
	((mask & ATTR_XVATTR) && XVA_ISSET_REQ(xvap, XAT_CREATETIME)))) {
	err = SET_ERROR(EPERM);
	goto out3;
	}

	if ((mask & ATTR_SIZE) && (zp->z_pflags & ZFS_READONLY)) {
	err = SET_ERROR(EPERM);
	goto out3;
	}

	/*
	* Verify timestamps doesn't overflow 32 bits.
	* ZFS can handle large timestamps, but 32bit syscalls can't
	* handle times greater than 2039. This check should be removed
	* once large timestamps are fully supported.
	*/
	if (mask & (ATTR_ATIME \| ATTR_MTIME)) {
	if (((mask & ATTR_ATIME) &&
	TIMESPEC_OVERFLOW(&vap->va_atime)) \|\|
	((mask & ATTR_MTIME) &&
	TIMESPEC_OVERFLOW(&vap->va_mtime))) {
	err = SET_ERROR(EOVERFLOW);
	goto out3;
	}
	}

	top:
	attrzp = NULL;
	aclp = NULL;

	/* Can this be moved to before the top label? */
	if (zfs_is_readonly(zfsvfs)) {
	err = SET_ERROR(EROFS);
	goto out3;
	}

	/*
	* First validate permissions
	*/

	if (mask & ATTR_SIZE) {
	err = zfs_zaccess(zp, ACE_WRITE_DATA, 0, skipaclchk, cr);
	if (err)
	goto out3;

	/*
	* XXX - Note, we are not providing any open
	* mode flags here (like FNDELAY), so we may
	* block if there are locks present... this
	* should be addressed in openat().
	*/
	/* XXX - would it be OK to generate a log record here? */
	err = zfs_freesp(zp, vap->va_size, 0, 0, FALSE);
	if (err)
	goto out3;
	}

	if (mask & (ATTR_ATIME\|ATTR_MTIME) \|\|
	((mask & ATTR_XVATTR) && (XVA_ISSET_REQ(xvap, XAT_HIDDEN) \|\|
	XVA_ISSET_REQ(xvap, XAT_READONLY) \|\|
	XVA_ISSET_REQ(xvap, XAT_ARCHIVE) \|\|
	XVA_ISSET_REQ(xvap, XAT_OFFLINE) \|\|
	XVA_ISSET_REQ(xvap, XAT_SPARSE) \|\|
	XVA_ISSET_REQ(xvap, XAT_CREATETIME) \|\|
	XVA_ISSET_REQ(xvap, XAT_SYSTEM)))) {
	need_policy = zfs_zaccess(zp, ACE_WRITE_ATTRIBUTES, 0,
	skipaclchk, cr);
	}

	if (mask & (ATTR_UID\|ATTR_GID)) {
	int idmask = (mask & (ATTR_UID\|ATTR_GID));
	int take_owner;
	int take_group;

	/*
	* NOTE: even if a new mode is being set,
	* we may clear S_ISUID/S_ISGID bits.
	*/

	if (!(mask & ATTR_MODE))
	vap->va_mode = zp->z_mode;

	/*
	* Take ownership or chgrp to group we are a member of
	*/

	take_owner = (mask & ATTR_UID) && (vap->va_uid == crgetuid(cr));
	take_group = (mask & ATTR_GID) &&
	zfs_groupmember(zfsvfs, vap->va_gid, cr);

	/*
	* If both ATTR_UID and ATTR_GID are set then take_owner and
	* take_group must both be set in order to allow taking
	* ownership.
	*
	* Otherwise, send the check through secpolicy_vnode_setattr()
	*
	*/

	if (((idmask == (ATTR_UID\|ATTR_GID)) &&
	take_owner && take_group) \|\|
	((idmask == ATTR_UID) && take_owner) \|\|
	((idmask == ATTR_GID) && take_group)) {
	if (zfs_zaccess(zp, ACE_WRITE_OWNER, 0,
	skipaclchk, cr) == 0) {
	/*
	* Remove setuid/setgid for non-privileged users
	*/
	(void) secpolicy_setid_clear(vap, cr);
	trim_mask = (mask & (ATTR_UID\|ATTR_GID));
	} else {
	need_policy = TRUE;
	}
	} else {
	need_policy = TRUE;
	}
	}

	mutex_enter(&zp->z_lock);
	oldva.va_mode = zp->z_mode;
	zfs_fuid_map_ids(zp, cr, &oldva.va_uid, &oldva.va_gid);
	if (mask & ATTR_XVATTR) {
	/*
	* Update xvattr mask to include only those attributes
	* that are actually changing.
	*
	* the bits will be restored prior to actually setting
	* the attributes so the caller thinks they were set.
	*/
	if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) {
	if (xoap->xoa_appendonly !=
	((zp->z_pflags & ZFS_APPENDONLY) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_APPENDONLY);
	XVA_SET_REQ(tmpxvattr, XAT_APPENDONLY);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_PROJINHERIT)) {
	if (xoap->xoa_projinherit !=
	((zp->z_pflags & ZFS_PROJINHERIT) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_PROJINHERIT);
	XVA_SET_REQ(tmpxvattr, XAT_PROJINHERIT);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) {
	if (xoap->xoa_nounlink !=
	((zp->z_pflags & ZFS_NOUNLINK) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_NOUNLINK);
	XVA_SET_REQ(tmpxvattr, XAT_NOUNLINK);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) {
	if (xoap->xoa_immutable !=
	((zp->z_pflags & ZFS_IMMUTABLE) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_IMMUTABLE);
	XVA_SET_REQ(tmpxvattr, XAT_IMMUTABLE);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) {
	if (xoap->xoa_nodump !=
	((zp->z_pflags & ZFS_NODUMP) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_NODUMP);
	XVA_SET_REQ(tmpxvattr, XAT_NODUMP);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) {
	if (xoap->xoa_av_modified !=
	((zp->z_pflags & ZFS_AV_MODIFIED) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_AV_MODIFIED);
	XVA_SET_REQ(tmpxvattr, XAT_AV_MODIFIED);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) {
	if ((!S_ISREG(ip->i_mode) &&
	xoap->xoa_av_quarantined) \|\|
	xoap->xoa_av_quarantined !=
	((zp->z_pflags & ZFS_AV_QUARANTINED) != 0)) {
	need_policy = TRUE;
	} else {
	XVA_CLR_REQ(xvap, XAT_AV_QUARANTINED);
	XVA_SET_REQ(tmpxvattr, XAT_AV_QUARANTINED);
	}
	}

	if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) {
	mutex_exit(&zp->z_lock);
	err = SET_ERROR(EPERM);
	goto out3;
	}

	if (need_policy == FALSE &&
	(XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP) \|\|
	XVA_ISSET_REQ(xvap, XAT_OPAQUE))) {
	need_policy = TRUE;
	}
	}

	mutex_exit(&zp->z_lock);

	if (mask & ATTR_MODE) {
	if (zfs_zaccess(zp, ACE_WRITE_ACL, 0, skipaclchk, cr) == 0) {
	err = secpolicy_setid_setsticky_clear(ip, vap,
	&oldva, cr);
	if (err)
	goto out3;

	trim_mask \|= ATTR_MODE;
	} else {
	need_policy = TRUE;
	}
	}

	if (need_policy) {
	/*
	* If trim_mask is set then take ownership
	* has been granted or write_acl is present and user
	* has the ability to modify mode. In that case remove
	* UID\|GID and or MODE from mask so that
	* secpolicy_vnode_setattr() doesn't revoke it.
	*/

	if (trim_mask) {
	saved_mask = vap->va_mask;
	vap->va_mask &= ~trim_mask;
	}
	err = secpolicy_vnode_setattr(cr, ip, vap, &oldva, flags,
	(int ()(void , int, cred_t *))zfs_zaccess_unix, zp);
	if (err)
	goto out3;

	if (trim_mask)
	vap->va_mask \|= saved_mask;
	}

	/*
	* secpolicy_vnode_setattr, or take ownership may have
	* changed va_mask
	*/
	mask = vap->va_mask;

	if ((mask & (ATTR_UID \| ATTR_GID)) \|\| projid != ZFS_INVALID_PROJID) {
	handle_eadir = B_TRUE;
	err = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
	&xattr_obj, sizeof (xattr_obj));

	if (err == 0 && xattr_obj) {
	err = zfs_zget(ZTOZSB(zp), xattr_obj, &attrzp);
	if (err)
	goto out2;
	}
	if (mask & ATTR_UID) {
	new_kuid = zfs_fuid_create(zfsvfs,
	(uint64_t)vap->va_uid, cr, ZFS_OWNER, &fuidp);
	if (new_kuid != KUID_TO_SUID(ZTOI(zp)->i_uid) &&
	zfs_id_overquota(zfsvfs, DMU_USERUSED_OBJECT,
	new_kuid)) {
	if (attrzp)
	zrele(attrzp);
	err = SET_ERROR(EDQUOT);
	goto out2;
	}
	}

	if (mask & ATTR_GID) {
	new_kgid = zfs_fuid_create(zfsvfs,
	(uint64_t)vap->va_gid, cr, ZFS_GROUP, &fuidp);
	if (new_kgid != KGID_TO_SGID(ZTOI(zp)->i_gid) &&
	zfs_id_overquota(zfsvfs, DMU_GROUPUSED_OBJECT,
	new_kgid)) {
	if (attrzp)
	zrele(attrzp);
	err = SET_ERROR(EDQUOT);
	goto out2;
	}
	}

	if (projid != ZFS_INVALID_PROJID &&
	zfs_id_overquota(zfsvfs, DMU_PROJECTUSED_OBJECT, projid)) {
	if (attrzp)
	zrele(attrzp);
	err = EDQUOT;
	goto out2;
	}
	}
	tx = dmu_tx_create(os);

	if (mask & ATTR_MODE) {
	uint64_t pmode = zp->z_mode;
	uint64_t acl_obj;
	new_mode = (pmode & S_IFMT) \| (vap->va_mode & ~S_IFMT);

	if (ZTOZSB(zp)->z_acl_mode == ZFS_ACL_RESTRICTED &&
	!(zp->z_pflags & ZFS_ACL_TRIVIAL)) {
	err = EPERM;
	goto out;
	}

	if ((err = zfs_acl_chmod_setattr(zp, &aclp, new_mode)))
	goto out;

	mutex_enter(&zp->z_lock);
	if (!zp->z_is_sa && ((acl_obj = zfs_external_acl(zp)) != 0)) {
	/*
	* Are we upgrading ACL from old V0 format
	* to V1 format?
	*/
	if (zfsvfs->z_version >= ZPL_VERSION_FUID &&
	zfs_znode_acl_version(zp) ==
	ZFS_ACL_VERSION_INITIAL) {
	dmu_tx_hold_free(tx, acl_obj, 0,
	DMU_OBJECT_END);
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
	0, aclp->z_acl_bytes);
	} else {
	dmu_tx_hold_write(tx, acl_obj, 0,
	aclp->z_acl_bytes);
	}
	} else if (!zp->z_is_sa && aclp->z_acl_bytes > ZFS_ACE_SPACE) {
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
	0, aclp->z_acl_bytes);
	}
	mutex_exit(&zp->z_lock);
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
	} else {
	if (((mask & ATTR_XVATTR) &&
	XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP)) \|\|
	(projid != ZFS_INVALID_PROJID &&
	!(zp->z_pflags & ZFS_PROJID)))
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
	else
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
	}

	if (attrzp) {
	dmu_tx_hold_sa(tx, attrzp->z_sa_hdl, B_FALSE);
	}

	fuid_dirtied = zfsvfs->z_fuid_dirty;
	if (fuid_dirtied)
	zfs_fuid_txhold(zfsvfs, tx);

	zfs_sa_upgrade_txholds(tx, zp);

	err = dmu_tx_assign(tx, TXG_WAIT);
	if (err)
	goto out;

	count = 0;
	/*
	* Set each attribute requested.
	* We group settings according to the locks they need to acquire.
	*
	* Note: you cannot set ctime directly, although it will be
	* updated as a side-effect of calling this function.
	*/

	if (projid != ZFS_INVALID_PROJID && !(zp->z_pflags & ZFS_PROJID)) {
	/*
	* For the existed object that is upgraded from old system,
	* its on-disk layout has no slot for the project ID attribute.
	* But quota accounting logic needs to access related slots by
	* offset directly. So we need to adjust old objects' layout
	* to make the project ID to some unified and fixed offset.
	*/
	if (attrzp)
	err = sa_add_projid(attrzp->z_sa_hdl, tx, projid);
	if (err == 0)
	err = sa_add_projid(zp->z_sa_hdl, tx, projid);

	if (unlikely(err == EEXIST))
	err = 0;
	else if (err != 0)
	goto out;
	else
	projid = ZFS_INVALID_PROJID;
	}

	if (mask & (ATTR_UID\|ATTR_GID\|ATTR_MODE))
	mutex_enter(&zp->z_acl_lock);
	mutex_enter(&zp->z_lock);

	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
	&zp->z_pflags, sizeof (zp->z_pflags));

	if (attrzp) {
	if (mask & (ATTR_UID\|ATTR_GID\|ATTR_MODE))
	mutex_enter(&attrzp->z_acl_lock);
	mutex_enter(&attrzp->z_lock);
	SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
	SA_ZPL_FLAGS(zfsvfs), NULL, &attrzp->z_pflags,
	sizeof (attrzp->z_pflags));
	if (projid != ZFS_INVALID_PROJID) {
	attrzp->z_projid = projid;
	SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
	SA_ZPL_PROJID(zfsvfs), NULL, &attrzp->z_projid,
	sizeof (attrzp->z_projid));
	}
	}

	if (mask & (ATTR_UID\|ATTR_GID)) {

	if (mask & ATTR_UID) {
	ZTOI(zp)->i_uid = SUID_TO_KUID(new_kuid);
	new_uid = zfs_uid_read(ZTOI(zp));
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
	&new_uid, sizeof (new_uid));
	if (attrzp) {
	SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
	SA_ZPL_UID(zfsvfs), NULL, &new_uid,
	sizeof (new_uid));
	ZTOI(attrzp)->i_uid = SUID_TO_KUID(new_uid);
	}
	}

	if (mask & ATTR_GID) {
	ZTOI(zp)->i_gid = SGID_TO_KGID(new_kgid);
	new_gid = zfs_gid_read(ZTOI(zp));
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs),
	NULL, &new_gid, sizeof (new_gid));
	if (attrzp) {
	SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
	SA_ZPL_GID(zfsvfs), NULL, &new_gid,
	sizeof (new_gid));
	ZTOI(attrzp)->i_gid = SGID_TO_KGID(new_kgid);
	}
	}
	if (!(mask & ATTR_MODE)) {
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs),
	NULL, &new_mode, sizeof (new_mode));
	new_mode = zp->z_mode;
	}
	err = zfs_acl_chown_setattr(zp);
	ASSERT(err == 0);
	if (attrzp) {
	err = zfs_acl_chown_setattr(attrzp);
	ASSERT(err == 0);
	}
	}

	if (mask & ATTR_MODE) {
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL,
	&new_mode, sizeof (new_mode));
	zp->z_mode = ZTOI(zp)->i_mode = new_mode;
	ASSERT3P(aclp, !=, NULL);
	err = zfs_aclset_common(zp, aclp, cr, tx);
	ASSERT0(err);
	if (zp->z_acl_cached)
	zfs_acl_free(zp->z_acl_cached);
	zp->z_acl_cached = aclp;
	aclp = NULL;
	}

	if ((mask & ATTR_ATIME) \|\| zp->z_atime_dirty) {
	zp->z_atime_dirty = B_FALSE;
	ZFS_TIME_ENCODE(&ip->i_atime, atime);
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL,
	&atime, sizeof (atime));
	}

	if (mask & (ATTR_MTIME \| ATTR_SIZE)) {
	ZFS_TIME_ENCODE(&vap->va_mtime, mtime);
	ZTOI(zp)->i_mtime = zpl_inode_timestamp_truncate(
	vap->va_mtime, ZTOI(zp));

	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL,
	mtime, sizeof (mtime));
	}

	if (mask & (ATTR_CTIME \| ATTR_SIZE)) {
	ZFS_TIME_ENCODE(&vap->va_ctime, ctime);
	ZTOI(zp)->i_ctime = zpl_inode_timestamp_truncate(vap->va_ctime,
	ZTOI(zp));
	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
	ctime, sizeof (ctime));
	}

	if (projid != ZFS_INVALID_PROJID) {
	zp->z_projid = projid;
	SA_ADD_BULK_ATTR(bulk, count,
	SA_ZPL_PROJID(zfsvfs), NULL, &zp->z_projid,
	sizeof (zp->z_projid));
	}

	if (attrzp && mask) {
	SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
	SA_ZPL_CTIME(zfsvfs), NULL, &ctime,
	sizeof (ctime));
	}

	/*
	* Do this after setting timestamps to prevent timestamp
	* update from toggling bit
	*/

	if (xoap && (mask & ATTR_XVATTR)) {

	/*
	* restore trimmed off masks
	* so that return masks can be set for caller.
	*/

	if (XVA_ISSET_REQ(tmpxvattr, XAT_APPENDONLY)) {
	XVA_SET_REQ(xvap, XAT_APPENDONLY);
	}
	if (XVA_ISSET_REQ(tmpxvattr, XAT_NOUNLINK)) {
	XVA_SET_REQ(xvap, XAT_NOUNLINK);
	}
	if (XVA_ISSET_REQ(tmpxvattr, XAT_IMMUTABLE)) {
	XVA_SET_REQ(xvap, XAT_IMMUTABLE);
	}
	if (XVA_ISSET_REQ(tmpxvattr, XAT_NODUMP)) {
	XVA_SET_REQ(xvap, XAT_NODUMP);
	}
	if (XVA_ISSET_REQ(tmpxvattr, XAT_AV_MODIFIED)) {
	XVA_SET_REQ(xvap, XAT_AV_MODIFIED);
	}
	if (XVA_ISSET_REQ(tmpxvattr, XAT_AV_QUARANTINED)) {
	XVA_SET_REQ(xvap, XAT_AV_QUARANTINED);
	}
	if (XVA_ISSET_REQ(tmpxvattr, XAT_PROJINHERIT)) {
	XVA_SET_REQ(xvap, XAT_PROJINHERIT);
	}

	if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP))
	ASSERT(S_ISREG(ip->i_mode));

	zfs_xvattr_set(zp, xvap, tx);
	}

	if (fuid_dirtied)
	zfs_fuid_sync(zfsvfs, tx);

	if (mask != 0)
	zfs_log_setattr(zilog, tx, TX_SETATTR, zp, vap, mask, fuidp);

	mutex_exit(&zp->z_lock);
	if (mask & (ATTR_UID\|ATTR_GID\|ATTR_MODE))
	mutex_exit(&zp->z_acl_lock);

	if (attrzp) {
	if (mask & (ATTR_UID\|ATTR_GID\|ATTR_MODE))
	mutex_exit(&attrzp->z_acl_lock);
	mutex_exit(&attrzp->z_lock);
	}
	out:
	if (err == 0 && xattr_count > 0) {
	err2 = sa_bulk_update(attrzp->z_sa_hdl, xattr_bulk,
	xattr_count, tx);
	ASSERT(err2 == 0);
	}

	if (aclp)
	zfs_acl_free(aclp);

	if (fuidp) {
	zfs_fuid_info_free(fuidp);
	fuidp = NULL;
	}

	if (err) {
	dmu_tx_abort(tx);
	if (attrzp)
	zrele(attrzp);
	if (err == ERESTART)
	goto top;
	} else {
	if (count > 0)
	err2 = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
	dmu_tx_commit(tx);
	if (attrzp) {
	if (err2 == 0 && handle_eadir)
	err2 = zfs_setattr_dir(attrzp);
	zrele(attrzp);
	}
	zfs_znode_update_vfs(zp);
	}

	out2:
	if (os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	out3:
	kmem_free(xattr_bulk, sizeof (sa_bulk_attr_t) * bulks);
	kmem_free(bulk, sizeof (sa_bulk_attr_t) * bulks);
	kmem_free(tmpxvattr, sizeof (xvattr_t));
	ZFS_EXIT(zfsvfs);
	return (err);
	}

	typedef struct zfs_zlock {
	krwlock_t zl_rwlock; / lock we acquired */
	znode_t zl_znode; / znode we held */
	struct zfs_zlock zl_next; / next in list */
	} zfs_zlock_t;

	/*
	* Drop locks and release vnodes that were held by zfs_rename_lock().
	*/
	static void
	zfs_rename_unlock(zfs_zlock_t **zlpp)
	{
	zfs_zlock_t *zl;

	while ((zl = *zlpp) != NULL) {
	if (zl->zl_znode != NULL)
	zfs_zrele_async(zl->zl_znode);
	rw_exit(zl->zl_rwlock);
	*zlpp = zl->zl_next;
	kmem_free(zl, sizeof (*zl));
	}
	}

	/*
	* Search back through the directory tree, using the ".." entries.
	* Lock each directory in the chain to prevent concurrent renames.
	* Fail any attempt to move a directory into one of its own descendants.
	* XXX - z_parent_lock can overlap with map or grow locks
	*/
	static int
	zfs_rename_lock(znode_t szp, znode_t tdzp, znode_t sdzp, zfs_zlock_t *zlpp)
	{
	zfs_zlock_t *zl;
	znode_t *zp = tdzp;
	uint64_t rootid = ZTOZSB(zp)->z_root;
	uint64_t oidp = zp->z_id;
	krwlock_t *rwlp = &szp->z_parent_lock;
	krw_t rw = RW_WRITER;

	/*
	* First pass write-locks szp and compares to zp->z_id.
	* Later passes read-lock zp and compare to zp->z_parent.
	*/
	do {
	if (!rw_tryenter(rwlp, rw)) {
	/*
	* Another thread is renaming in this path.
	* Note that if we are a WRITER, we don't have any
	* parent_locks held yet.
	*/
	if (rw == RW_READER && zp->z_id > szp->z_id) {
	/*
	* Drop our locks and restart
	*/
	zfs_rename_unlock(&zl);
	*zlpp = NULL;
	zp = tdzp;
	oidp = zp->z_id;
	rwlp = &szp->z_parent_lock;
	rw = RW_WRITER;
	continue;
	} else {
	/*
	* Wait for other thread to drop its locks
	*/
	rw_enter(rwlp, rw);
	}
	}

	zl = kmem_alloc(sizeof (*zl), KM_SLEEP);
	zl->zl_rwlock = rwlp;
	zl->zl_znode = NULL;
	zl->zl_next = *zlpp;
	*zlpp = zl;

	if (oidp == szp->z_id) /* We're a descendant of szp */
	return (SET_ERROR(EINVAL));

	if (oidp == rootid) /* We've hit the top */
	return (0);

	if (rw == RW_READER) { /* i.e. not the first pass */
	int error = zfs_zget(ZTOZSB(zp), oidp, &zp);
	if (error)
	return (error);
	zl->zl_znode = zp;
	}
	(void) sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(ZTOZSB(zp)),
	&oidp, sizeof (oidp));
	rwlp = &zp->z_parent_lock;
	rw = RW_READER;

	} while (zp->z_id != sdzp->z_id);

	return (0);
	}

	/*
	* Move an entry from the provided source directory to the target
	* directory. Change the entry name as indicated.
	*
	* IN: sdzp - Source directory containing the "old entry".
	* snm - Old entry name.
	* tdzp - Target directory to contain the "new entry".
	* tnm - New entry name.
	* cr - credentials of caller.
	* flags - case flags
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* sdzp,tdzp - ctime\|mtime updated
	*/
	/ARGSUSED/
	int
	zfs_rename(znode_t sdzp, char snm, znode_t tdzp, char tnm,
	cred_t *cr, int flags)
	{
	znode_t szp, tzp;
	zfsvfs_t *zfsvfs = ZTOZSB(sdzp);
	zilog_t *zilog;
	zfs_dirlock_t sdl, tdl;
	dmu_tx_t *tx;
	zfs_zlock_t *zl;
	int cmp, serr, terr;
	int error = 0;
	int zflg = 0;
	boolean_t waited = B_FALSE;

	if (snm == NULL \|\| tnm == NULL)
	return (SET_ERROR(EINVAL));

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(sdzp);
	zilog = zfsvfs->z_log;

	ZFS_VERIFY_ZP(tdzp);

	/*
	* We check i_sb because snapshots and the ctldir must have different
	* super blocks.
	*/
	if (ZTOI(tdzp)->i_sb != ZTOI(sdzp)->i_sb \|\|
	zfsctl_is_node(ZTOI(tdzp))) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EXDEV));
	}

	if (zfsvfs->z_utf8 && u8_validate(tnm,
	strlen(tnm), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EILSEQ));
	}

	if (flags & FIGNORECASE)
	zflg \|= ZCILOOK;

	top:
	szp = NULL;
	tzp = NULL;
	zl = NULL;

	/*
	* This is to prevent the creation of links into attribute space
	* by renaming a linked file into/outof an attribute directory.
	* See the comment in zfs_link() for why this is considered bad.
	*/
	if ((tdzp->z_pflags & ZFS_XATTR) != (sdzp->z_pflags & ZFS_XATTR)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	/*
	* Lock source and target directory entries. To prevent deadlock,
	* a lock ordering must be defined. We lock the directory with
	* the smallest object id first, or if it's a tie, the one with
	* the lexically first name.
	*/
	if (sdzp->z_id < tdzp->z_id) {
	cmp = -1;
	} else if (sdzp->z_id > tdzp->z_id) {
	cmp = 1;
	} else {
	/*
	* First compare the two name arguments without
	* considering any case folding.
	*/
	int nofold = (zfsvfs->z_norm & ~U8_TEXTPREP_TOUPPER);

	cmp = u8_strcmp(snm, tnm, 0, nofold, U8_UNICODE_LATEST, &error);
	ASSERT(error == 0 \|\| !zfsvfs->z_utf8);
	if (cmp == 0) {
	/*
	* POSIX: "If the old argument and the new argument
	* both refer to links to the same existing file,
	* the rename() function shall return successfully
	* and perform no other action."
	*/
	ZFS_EXIT(zfsvfs);
	return (0);
	}
	/*
	* If the file system is case-folding, then we may
	* have some more checking to do. A case-folding file
	* system is either supporting mixed case sensitivity
	* access or is completely case-insensitive. Note
	* that the file system is always case preserving.
	*
	* In mixed sensitivity mode case sensitive behavior
	* is the default. FIGNORECASE must be used to
	* explicitly request case insensitive behavior.
	*
	* If the source and target names provided differ only
	* by case (e.g., a request to rename 'tim' to 'Tim'),
	* we will treat this as a special case in the
	* case-insensitive mode: as long as the source name
	* is an exact match, we will allow this to proceed as
	* a name-change request.
	*/
	if ((zfsvfs->z_case == ZFS_CASE_INSENSITIVE \|\|
	(zfsvfs->z_case == ZFS_CASE_MIXED &&
	flags & FIGNORECASE)) &&
	u8_strcmp(snm, tnm, 0, zfsvfs->z_norm, U8_UNICODE_LATEST,
	&error) == 0) {
	/*
	* case preserving rename request, require exact
	* name matches
	*/
	zflg \|= ZCIEXACT;
	zflg &= ~ZCILOOK;
	}
	}

	/*
	* If the source and destination directories are the same, we should
	* grab the z_name_lock of that directory only once.
	*/
	if (sdzp == tdzp) {
	zflg \|= ZHAVELOCK;
	rw_enter(&sdzp->z_name_lock, RW_READER);
	}

	if (cmp < 0) {
	serr = zfs_dirent_lock(&sdl, sdzp, snm, &szp,
	ZEXISTS \| zflg, NULL, NULL);
	terr = zfs_dirent_lock(&tdl,
	tdzp, tnm, &tzp, ZRENAMING \| zflg, NULL, NULL);
	} else {
	terr = zfs_dirent_lock(&tdl,
	tdzp, tnm, &tzp, zflg, NULL, NULL);
	serr = zfs_dirent_lock(&sdl,
	sdzp, snm, &szp, ZEXISTS \| ZRENAMING \| zflg,
	NULL, NULL);
	}

	if (serr) {
	/*
	* Source entry invalid or not there.
	*/
	if (!terr) {
	zfs_dirent_unlock(tdl);
	if (tzp)
	zrele(tzp);
	}

	if (sdzp == tdzp)
	rw_exit(&sdzp->z_name_lock);

	if (strcmp(snm, "..") == 0)
	serr = EINVAL;
	ZFS_EXIT(zfsvfs);
	return (serr);
	}
	if (terr) {
	zfs_dirent_unlock(sdl);
	zrele(szp);

	if (sdzp == tdzp)
	rw_exit(&sdzp->z_name_lock);

	if (strcmp(tnm, "..") == 0)
	terr = EINVAL;
	ZFS_EXIT(zfsvfs);
	return (terr);
	}

	/*
	* If we are using project inheritance, means if the directory has
	* ZFS_PROJINHERIT set, then its descendant directories will inherit
	* not only the project ID, but also the ZFS_PROJINHERIT flag. Under
	* such case, we only allow renames into our tree when the project
	* IDs are the same.
	*/
	if (tdzp->z_pflags & ZFS_PROJINHERIT &&
	tdzp->z_projid != szp->z_projid) {
	error = SET_ERROR(EXDEV);
	goto out;
	}

	/*
	* Must have write access at the source to remove the old entry
	* and write access at the target to create the new entry.
	* Note that if target and source are the same, this can be
	* done in a single check.
	*/

	if ((error = zfs_zaccess_rename(sdzp, szp, tdzp, tzp, cr)))
	goto out;

	if (S_ISDIR(ZTOI(szp)->i_mode)) {
	/*
	* Check to make sure rename is valid.
	* Can't do a move like this: /usr/a/b to /usr/a/b/c/d
	*/
	if ((error = zfs_rename_lock(szp, tdzp, sdzp, &zl)))
	goto out;
	}

	/*
	* Does target exist?
	*/
	if (tzp) {
	/*
	* Source and target must be the same type.
	*/
	if (S_ISDIR(ZTOI(szp)->i_mode)) {
	if (!S_ISDIR(ZTOI(tzp)->i_mode)) {
	error = SET_ERROR(ENOTDIR);
	goto out;
	}
	} else {
	if (S_ISDIR(ZTOI(tzp)->i_mode)) {
	error = SET_ERROR(EISDIR);
	goto out;
	}
	}
	/*
	* POSIX dictates that when the source and target
	* entries refer to the same file object, rename
	* must do nothing and exit without error.
	*/
	if (szp->z_id == tzp->z_id) {
	error = 0;
	goto out;
	}
	}

	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE);
	dmu_tx_hold_sa(tx, sdzp->z_sa_hdl, B_FALSE);
	dmu_tx_hold_zap(tx, sdzp->z_id, FALSE, snm);
	dmu_tx_hold_zap(tx, tdzp->z_id, TRUE, tnm);
	if (sdzp != tdzp) {
	dmu_tx_hold_sa(tx, tdzp->z_sa_hdl, B_FALSE);
	zfs_sa_upgrade_txholds(tx, tdzp);
	}
	if (tzp) {
	dmu_tx_hold_sa(tx, tzp->z_sa_hdl, B_FALSE);
	zfs_sa_upgrade_txholds(tx, tzp);
	}

	zfs_sa_upgrade_txholds(tx, szp);
	dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
	error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) \| TXG_NOWAIT);
	if (error) {
	if (zl != NULL)
	zfs_rename_unlock(&zl);
	zfs_dirent_unlock(sdl);
	zfs_dirent_unlock(tdl);

	if (sdzp == tdzp)
	rw_exit(&sdzp->z_name_lock);

	if (error == ERESTART) {
	waited = B_TRUE;
	dmu_tx_wait(tx);
	dmu_tx_abort(tx);
	zrele(szp);
	if (tzp)
	zrele(tzp);
	goto top;
	}
	dmu_tx_abort(tx);
	zrele(szp);
	if (tzp)
	zrele(tzp);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if (tzp) /* Attempt to remove the existing target */
	error = zfs_link_destroy(tdl, tzp, tx, zflg, NULL);

	if (error == 0) {
	error = zfs_link_create(tdl, szp, tx, ZRENAMING);
	if (error == 0) {
	szp->z_pflags \|= ZFS_AV_MODIFIED;
	if (tdzp->z_pflags & ZFS_PROJINHERIT)
	szp->z_pflags \|= ZFS_PROJINHERIT;

	error = sa_update(szp->z_sa_hdl, SA_ZPL_FLAGS(zfsvfs),
	(void *)&szp->z_pflags, sizeof (uint64_t), tx);
	ASSERT0(error);

	error = zfs_link_destroy(sdl, szp, tx, ZRENAMING, NULL);
	if (error == 0) {
	zfs_log_rename(zilog, tx, TX_RENAME \|
	(flags & FIGNORECASE ? TX_CI : 0), sdzp,
	sdl->dl_name, tdzp, tdl->dl_name, szp);
	} else {
	/*
	* At this point, we have successfully created
	* the target name, but have failed to remove
	* the source name. Since the create was done
	* with the ZRENAMING flag, there are
	* complications; for one, the link count is
	* wrong. The easiest way to deal with this
	* is to remove the newly created target, and
	* return the original error. This must
	* succeed; fortunately, it is very unlikely to
	* fail, since we just created it.
	*/
	VERIFY3U(zfs_link_destroy(tdl, szp, tx,
	ZRENAMING, NULL), ==, 0);
	}
	} else {
	/*
	* If we had removed the existing target, subsequent
	* call to zfs_link_create() to add back the same entry
	* but, the new dnode (szp) should not fail.
	*/
	ASSERT(tzp == NULL);
	}
	}

	dmu_tx_commit(tx);
	out:
	if (zl != NULL)
	zfs_rename_unlock(&zl);

	zfs_dirent_unlock(sdl);
	zfs_dirent_unlock(tdl);

	zfs_znode_update_vfs(sdzp);
	if (sdzp == tdzp)
	rw_exit(&sdzp->z_name_lock);

	if (sdzp != tdzp)
	zfs_znode_update_vfs(tdzp);

	zfs_znode_update_vfs(szp);
	zrele(szp);
	if (tzp) {
	zfs_znode_update_vfs(tzp);
	zrele(tzp);
	}

	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Insert the indicated symbolic reference entry into the directory.
	*
	* IN: dzp - Directory to contain new symbolic link.
	* name - Name of directory entry in dip.
	* vap - Attributes of new entry.
	* link - Name for new symlink entry.
	* cr - credentials of caller.
	* flags - case flags
	*
	* OUT: zpp - Znode for new symbolic link.
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* dip - ctime\|mtime updated
	*/
	/ARGSUSED/
	int
	zfs_symlink(znode_t dzp, char name, vattr_t vap, char link,
	znode_t *zpp, cred_t cr, int flags)
	{
	znode_t *zp;
	zfs_dirlock_t *dl;
	dmu_tx_t *tx;
	zfsvfs_t *zfsvfs = ZTOZSB(dzp);
	zilog_t *zilog;
	uint64_t len = strlen(link);
	int error;
	int zflg = ZNEW;
	zfs_acl_ids_t acl_ids;
	boolean_t fuid_dirtied;
	uint64_t txtype = TX_SYMLINK;
	boolean_t waited = B_FALSE;

	ASSERT(S_ISLNK(vap->va_mode));

	if (name == NULL)
	return (SET_ERROR(EINVAL));

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(dzp);
	zilog = zfsvfs->z_log;

	if (zfsvfs->z_utf8 && u8_validate(name, strlen(name),
	NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EILSEQ));
	}
	if (flags & FIGNORECASE)
	zflg \|= ZCILOOK;

	if (len > MAXPATHLEN) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(ENAMETOOLONG));
	}

	if ((error = zfs_acl_ids_create(dzp, 0,
	vap, cr, NULL, &acl_ids)) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	top:
	*zpp = NULL;

	/*
	* Attempt to lock directory; fail if entry already exists.
	*/
	error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, NULL);
	if (error) {
	zfs_acl_ids_free(&acl_ids);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
	zfs_acl_ids_free(&acl_ids);
	zfs_dirent_unlock(dl);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, ZFS_DEFAULT_PROJID)) {
	zfs_acl_ids_free(&acl_ids);
	zfs_dirent_unlock(dl);
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EDQUOT));
	}
	tx = dmu_tx_create(zfsvfs->z_os);
	fuid_dirtied = zfsvfs->z_fuid_dirty;
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, MAX(1, len));
	dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
	dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
	ZFS_SA_BASE_ATTR_SIZE + len);
	dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
	if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
	acl_ids.z_aclp->z_acl_bytes);
	}
	if (fuid_dirtied)
	zfs_fuid_txhold(zfsvfs, tx);
	error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) \| TXG_NOWAIT);
	if (error) {
	zfs_dirent_unlock(dl);
	if (error == ERESTART) {
	waited = B_TRUE;
	dmu_tx_wait(tx);
	dmu_tx_abort(tx);
	goto top;
	}
	zfs_acl_ids_free(&acl_ids);
	dmu_tx_abort(tx);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Create a new object for the symlink.
	* for version 4 ZPL datasets the symlink will be an SA attribute
	*/
	zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);

	if (fuid_dirtied)
	zfs_fuid_sync(zfsvfs, tx);

	mutex_enter(&zp->z_lock);
	if (zp->z_is_sa)
	error = sa_update(zp->z_sa_hdl, SA_ZPL_SYMLINK(zfsvfs),
	link, len, tx);
	else
	zfs_sa_symlink(zp, link, len, tx);
	mutex_exit(&zp->z_lock);

	zp->z_size = len;
	(void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs),
	&zp->z_size, sizeof (zp->z_size), tx);
	/*
	* Insert the new object into the directory.
	*/
	error = zfs_link_create(dl, zp, tx, ZNEW);
	if (error != 0) {
	zfs_znode_delete(zp, tx);
	remove_inode_hash(ZTOI(zp));
	} else {
	if (flags & FIGNORECASE)
	txtype \|= TX_CI;
	zfs_log_symlink(zilog, tx, txtype, dzp, zp, name, link);

	zfs_znode_update_vfs(dzp);
	zfs_znode_update_vfs(zp);
	}

	zfs_acl_ids_free(&acl_ids);

	dmu_tx_commit(tx);

	zfs_dirent_unlock(dl);

	if (error == 0) {
	*zpp = zp;

	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);
	} else {
	zrele(zp);
	}

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Return, in the buffer contained in the provided uio structure,
	* the symbolic path referred to by ip.
	*
	* IN: ip - inode of symbolic link
	* uio - structure to contain the link path.
	* cr - credentials of caller.
	*
	* RETURN: 0 if success
	* error code if failure
	*
	* Timestamps:
	* ip - atime updated
	*/
	/* ARGSUSED */
	int
	zfs_readlink(struct inode ip, zfs_uio_t uio, cred_t *cr)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);
	int error;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	mutex_enter(&zp->z_lock);
	if (zp->z_is_sa)
	error = sa_lookup_uio(zp->z_sa_hdl,
	SA_ZPL_SYMLINK(zfsvfs), uio);
	else
	error = zfs_sa_readlink(zp, uio);
	mutex_exit(&zp->z_lock);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/*
	* Insert a new entry into directory tdzp referencing szp.
	*
	* IN: tdzp - Directory to contain new entry.
	* szp - znode of new entry.
	* name - name of new entry.
	* cr - credentials of caller.
	* flags - case flags.
	*
	* RETURN: 0 if success
	* error code if failure
	*
	* Timestamps:
	* tdzp - ctime\|mtime updated
	* szp - ctime updated
	*/
	/* ARGSUSED */
	int
	zfs_link(znode_t tdzp, znode_t szp, char name, cred_t cr,
	int flags)
	{
	struct inode *sip = ZTOI(szp);
	znode_t *tzp;
	zfsvfs_t *zfsvfs = ZTOZSB(tdzp);
	zilog_t *zilog;
	zfs_dirlock_t *dl;
	dmu_tx_t *tx;
	int error;
	int zf = ZNEW;
	uint64_t parent;
	uid_t owner;
	boolean_t waited = B_FALSE;
	boolean_t is_tmpfile = 0;
	uint64_t txg;
	#ifdef HAVE_TMPFILE
	is_tmpfile = (sip->i_nlink == 0 && (sip->i_state & I_LINKABLE));
	#endif
	ASSERT(S_ISDIR(ZTOI(tdzp)->i_mode));

	if (name == NULL)
	return (SET_ERROR(EINVAL));

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(tdzp);
	zilog = zfsvfs->z_log;

	/*
	* POSIX dictates that we return EPERM here.
	* Better choices include ENOTSUP or EISDIR.
	*/
	if (S_ISDIR(sip->i_mode)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	ZFS_VERIFY_ZP(szp);

	/*
	* If we are using project inheritance, means if the directory has
	* ZFS_PROJINHERIT set, then its descendant directories will inherit
	* not only the project ID, but also the ZFS_PROJINHERIT flag. Under
	* such case, we only allow hard link creation in our tree when the
	* project IDs are the same.
	*/
	if (tdzp->z_pflags & ZFS_PROJINHERIT &&
	tdzp->z_projid != szp->z_projid) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EXDEV));
	}

	/*
	* We check i_sb because snapshots and the ctldir must have different
	* super blocks.
	*/
	if (sip->i_sb != ZTOI(tdzp)->i_sb \|\| zfsctl_is_node(sip)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EXDEV));
	}

	/* Prevent links to .zfs/shares files */

	if ((error = sa_lookup(szp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
	&parent, sizeof (uint64_t))) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	if (parent == zfsvfs->z_shares_dir) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	if (zfsvfs->z_utf8 && u8_validate(name,
	strlen(name), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EILSEQ));
	}
	if (flags & FIGNORECASE)
	zf \|= ZCILOOK;

	/*
	* We do not support links between attributes and non-attributes
	* because of the potential security risk of creating links
	* into "normal" file space in order to circumvent restrictions
	* imposed in attribute space.
	*/
	if ((szp->z_pflags & ZFS_XATTR) != (tdzp->z_pflags & ZFS_XATTR)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	owner = zfs_fuid_map_id(zfsvfs, KUID_TO_SUID(sip->i_uid),
	cr, ZFS_OWNER);
	if (owner != crgetuid(cr) && secpolicy_basic_link(cr) != 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	if ((error = zfs_zaccess(tdzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	top:
	/*
	* Attempt to lock directory; fail if entry already exists.
	*/
	error = zfs_dirent_lock(&dl, tdzp, name, &tzp, zf, NULL, NULL);
	if (error) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE);
	dmu_tx_hold_zap(tx, tdzp->z_id, TRUE, name);
	if (is_tmpfile)
	dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);

	zfs_sa_upgrade_txholds(tx, szp);
	zfs_sa_upgrade_txholds(tx, tdzp);
	error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) \| TXG_NOWAIT);
	if (error) {
	zfs_dirent_unlock(dl);
	if (error == ERESTART) {
	waited = B_TRUE;
	dmu_tx_wait(tx);
	dmu_tx_abort(tx);
	goto top;
	}
	dmu_tx_abort(tx);
	ZFS_EXIT(zfsvfs);
	return (error);
	}
	/* unmark z_unlinked so zfs_link_create will not reject */
	if (is_tmpfile)
	szp->z_unlinked = B_FALSE;
	error = zfs_link_create(dl, szp, tx, 0);

	if (error == 0) {
	uint64_t txtype = TX_LINK;
	/*
	* tmpfile is created to be in z_unlinkedobj, so remove it.
	* Also, we don't log in ZIL, because all previous file
	* operation on the tmpfile are ignored by ZIL. Instead we
	* always wait for txg to sync to make sure all previous
	* operation are sync safe.
	*/
	if (is_tmpfile) {
	VERIFY(zap_remove_int(zfsvfs->z_os,
	zfsvfs->z_unlinkedobj, szp->z_id, tx) == 0);
	} else {
	if (flags & FIGNORECASE)
	txtype \|= TX_CI;
	zfs_log_link(zilog, tx, txtype, tdzp, szp, name);
	}
	} else if (is_tmpfile) {
	/* restore z_unlinked since when linking failed */
	szp->z_unlinked = B_TRUE;
	}
	txg = dmu_tx_get_txg(tx);
	dmu_tx_commit(tx);

	zfs_dirent_unlock(dl);

	if (!is_tmpfile && zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	zil_commit(zilog, 0);

	if (is_tmpfile && zfsvfs->z_os->os_sync != ZFS_SYNC_DISABLED)
	txg_wait_synced(dmu_objset_pool(zfsvfs->z_os), txg);

	zfs_znode_update_vfs(tdzp);
	zfs_znode_update_vfs(szp);
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	static void
	zfs_putpage_commit_cb(void *arg)
	{
	struct page *pp = arg;

	ClearPageError(pp);
	end_page_writeback(pp);
	}

	/*
	* Push a page out to disk, once the page is on stable storage the
	* registered commit callback will be run as notification of completion.
	*
	* IN: ip - page mapped for inode.
	* pp - page to push (page is locked)
	* wbc - writeback control data
	*
	* RETURN: 0 if success
	* error code if failure
	*
	* Timestamps:
	* ip - ctime\|mtime updated
	*/
	/* ARGSUSED */
	int
	zfs_putpage(struct inode ip, struct page pp, struct writeback_control *wbc)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);
	loff_t offset;
	loff_t pgoff;
	unsigned int pglen;
	dmu_tx_t *tx;
	caddr_t va;
	int err = 0;
	uint64_t mtime[2], ctime[2];
	sa_bulk_attr_t bulk[3];
	int cnt = 0;
	struct address_space *mapping;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	ASSERT(PageLocked(pp));

	pgoff = page_offset(pp); /* Page byte-offset in file */
	offset = i_size_read(ip); /* File length in bytes */
	pglen = MIN(PAGE_SIZE, /* Page length in bytes */
	P2ROUNDUP(offset, PAGE_SIZE)-pgoff);

	/* Page is beyond end of file */
	if (pgoff >= offset) {
	unlock_page(pp);
	ZFS_EXIT(zfsvfs);
	return (0);
	}

	/* Truncate page length to end of file */
	if (pgoff + pglen > offset)
	pglen = offset - pgoff;

	#if 0
	/*
	* FIXME: Allow mmap writes past its quota. The correct fix
	* is to register a page_mkwrite() handler to count the page
	* against its quota when it is about to be dirtied.
	*/
	if (zfs_id_overblockquota(zfsvfs, DMU_USERUSED_OBJECT,
	KUID_TO_SUID(ip->i_uid)) \|\|
	zfs_id_overblockquota(zfsvfs, DMU_GROUPUSED_OBJECT,
	KGID_TO_SGID(ip->i_gid)) \|\|
	(zp->z_projid != ZFS_DEFAULT_PROJID &&
	zfs_id_overblockquota(zfsvfs, DMU_PROJECTUSED_OBJECT,
	zp->z_projid))) {
	err = EDQUOT;
	}
	#endif

	/*
	* The ordering here is critical and must adhere to the following
	* rules in order to avoid deadlocking in either zfs_read() or
	* zfs_free_range() due to a lock inversion.
	*
	* 1) The page must be unlocked prior to acquiring the range lock.
	* This is critical because zfs_read() calls find_lock_page()
	* which may block on the page lock while holding the range lock.
	*
	* 2) Before setting or clearing write back on a page the range lock
	* must be held in order to prevent a lock inversion with the
	* zfs_free_range() function.
	*
	* This presents a problem because upon entering this function the
	* page lock is already held. To safely acquire the range lock the
	* page lock must be dropped. This creates a window where another
	* process could truncate, invalidate, dirty, or write out the page.
	*
	* Therefore, after successfully reacquiring the range and page locks
	* the current page state is checked. In the common case everything
	* will be as is expected and it can be written out. However, if
	* the page state has changed it must be handled accordingly.
	*/
	mapping = pp->mapping;
	redirty_page_for_writepage(wbc, pp);
	unlock_page(pp);

	zfs_locked_range_t *lr = zfs_rangelock_enter(&zp->z_rangelock,
	pgoff, pglen, RL_WRITER);
	lock_page(pp);

	/* Page mapping changed or it was no longer dirty, we're done */
	if (unlikely((mapping != pp->mapping) \|\| !PageDirty(pp))) {
	unlock_page(pp);
	zfs_rangelock_exit(lr);
	ZFS_EXIT(zfsvfs);
	return (0);
	}

	/* Another process started write block if required */
	if (PageWriteback(pp)) {
	unlock_page(pp);
	zfs_rangelock_exit(lr);

	if (wbc->sync_mode != WB_SYNC_NONE) {
	if (PageWriteback(pp))
	+#ifdef HAVE_PAGEMAP_FOLIO_WAIT_BIT
	+ folio_wait_bit(page_folio(pp), PG_writeback);
	+#else
	wait_on_page_bit(pp, PG_writeback);
	+#endif
	}

	ZFS_EXIT(zfsvfs);
	return (0);
	}

	/* Clear the dirty flag the required locks are held */
	if (!clear_page_dirty_for_io(pp)) {
	unlock_page(pp);
	zfs_rangelock_exit(lr);
	ZFS_EXIT(zfsvfs);
	return (0);
	}

	/*
	* Counterpart for redirty_page_for_writepage() above. This page
	* was in fact not skipped and should not be counted as if it were.
	*/
	wbc->pages_skipped--;
	set_page_writeback(pp);
	unlock_page(pp);

	tx = dmu_tx_create(zfsvfs->z_os);
	dmu_tx_hold_write(tx, zp->z_id, pgoff, pglen);
	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
	zfs_sa_upgrade_txholds(tx, zp);

	err = dmu_tx_assign(tx, TXG_NOWAIT);
	if (err != 0) {
	if (err == ERESTART)
	dmu_tx_wait(tx);

	dmu_tx_abort(tx);
	__set_page_dirty_nobuffers(pp);
	ClearPageError(pp);
	end_page_writeback(pp);
	zfs_rangelock_exit(lr);
	ZFS_EXIT(zfsvfs);
	return (err);
	}

	va = kmap(pp);
	ASSERT3U(pglen, <=, PAGE_SIZE);
	dmu_write(zfsvfs->z_os, zp->z_id, pgoff, pglen, va, tx);
	kunmap(pp);

	SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
	SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);
	SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_FLAGS(zfsvfs), NULL,
	&zp->z_pflags, 8);

	/* Preserve the mtime and ctime provided by the inode */
	ZFS_TIME_ENCODE(&ip->i_mtime, mtime);
	ZFS_TIME_ENCODE(&ip->i_ctime, ctime);
	zp->z_atime_dirty = B_FALSE;
	zp->z_seq++;

	err = sa_bulk_update(zp->z_sa_hdl, bulk, cnt, tx);

	zfs_log_write(zfsvfs->z_log, tx, TX_WRITE, zp, pgoff, pglen, 0,
	zfs_putpage_commit_cb, pp);
	dmu_tx_commit(tx);

	zfs_rangelock_exit(lr);

	if (wbc->sync_mode != WB_SYNC_NONE) {
	/*
	* Note that this is rarely called under writepages(), because
	* writepages() normally handles the entire commit for
	* performance reasons.
	*/
	zil_commit(zfsvfs->z_log, zp->z_id);
	}

	ZFS_EXIT(zfsvfs);
	return (err);
	}

	/*
	* Update the system attributes when the inode has been dirtied. For the
	* moment we only update the mode, atime, mtime, and ctime.
	*/
	int
	zfs_dirty_inode(struct inode *ip, int flags)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);
	dmu_tx_t *tx;
	uint64_t mode, atime[2], mtime[2], ctime[2];
	sa_bulk_attr_t bulk[4];
	int error = 0;
	int cnt = 0;

	if (zfs_is_readonly(zfsvfs) \|\| dmu_objset_is_snapshot(zfsvfs->z_os))
	return (0);

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	#ifdef I_DIRTY_TIME
	/*
	* This is the lazytime semantic introduced in Linux 4.0
	* This flag will only be called from update_time when lazytime is set.
	* (Note, I_DIRTY_SYNC will also set if not lazytime)
	* Fortunately mtime and ctime are managed within ZFS itself, so we
	* only need to dirty atime.
	*/
	if (flags == I_DIRTY_TIME) {
	zp->z_atime_dirty = B_TRUE;
	goto out;
	}
	#endif

	tx = dmu_tx_create(zfsvfs->z_os);

	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
	zfs_sa_upgrade_txholds(tx, zp);

	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	dmu_tx_abort(tx);
	goto out;
	}

	mutex_enter(&zp->z_lock);
	zp->z_atime_dirty = B_FALSE;

	SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MODE(zfsvfs), NULL, &mode, 8);
	SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_ATIME(zfsvfs), NULL, &atime, 16);
	SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
	SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);

	/* Preserve the mode, mtime and ctime provided by the inode */
	ZFS_TIME_ENCODE(&ip->i_atime, atime);
	ZFS_TIME_ENCODE(&ip->i_mtime, mtime);
	ZFS_TIME_ENCODE(&ip->i_ctime, ctime);
	mode = ip->i_mode;

	zp->z_mode = mode;

	error = sa_bulk_update(zp->z_sa_hdl, bulk, cnt, tx);
	mutex_exit(&zp->z_lock);

	dmu_tx_commit(tx);
	out:
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/ARGSUSED/
	void
	zfs_inactive(struct inode *ip)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);
	uint64_t atime[2];
	int error;
	int need_unlock = 0;

	/* Only read lock if we haven't already write locked, e.g. rollback */
	if (!RW_WRITE_HELD(&zfsvfs->z_teardown_inactive_lock)) {
	need_unlock = 1;
	rw_enter(&zfsvfs->z_teardown_inactive_lock, RW_READER);
	}
	if (zp->z_sa_hdl == NULL) {
	if (need_unlock)
	rw_exit(&zfsvfs->z_teardown_inactive_lock);
	return;
	}

	if (zp->z_atime_dirty && zp->z_unlinked == B_FALSE) {
	dmu_tx_t *tx = dmu_tx_create(zfsvfs->z_os);

	dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
	zfs_sa_upgrade_txholds(tx, zp);
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	dmu_tx_abort(tx);
	} else {
	ZFS_TIME_ENCODE(&ip->i_atime, atime);
	mutex_enter(&zp->z_lock);
	(void) sa_update(zp->z_sa_hdl, SA_ZPL_ATIME(zfsvfs),
	(void *)&atime, sizeof (atime), tx);
	zp->z_atime_dirty = B_FALSE;
	mutex_exit(&zp->z_lock);
	dmu_tx_commit(tx);
	}
	}

	zfs_zinactive(zp);
	if (need_unlock)
	rw_exit(&zfsvfs->z_teardown_inactive_lock);
	}

	/*
	* Fill pages with data from the disk.
	*/
	static int
	zfs_fillpage(struct inode ip, struct page pl[], int nr_pages)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);
	objset_t *os;
	struct page *cur_pp;
	u_offset_t io_off, total;
	size_t io_len;
	loff_t i_size;
	unsigned page_idx;
	int err;

	os = zfsvfs->z_os;
	io_len = nr_pages << PAGE_SHIFT;
	i_size = i_size_read(ip);
	io_off = page_offset(pl[0]);

	if (io_off + io_len > i_size)
	io_len = i_size - io_off;

	/*
	* Iterate over list of pages and read each page individually.
	*/
	page_idx = 0;
	for (total = io_off + io_len; io_off < total; io_off += PAGESIZE) {
	caddr_t va;

	cur_pp = pl[page_idx++];
	va = kmap(cur_pp);
	err = dmu_read(os, zp->z_id, io_off, PAGESIZE, va,
	DMU_READ_PREFETCH);
	kunmap(cur_pp);
	if (err) {
	/* convert checksum errors into IO errors */
	if (err == ECKSUM)
	err = SET_ERROR(EIO);
	return (err);
	}
	}

	return (0);
	}

	/*
	* Uses zfs_fillpage to read data from the file and fill the pages.
	*
	* IN: ip - inode of file to get data from.
	* pl - list of pages to read
	* nr_pages - number of pages to read
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* vp - atime updated
	*/
	/* ARGSUSED */
	int
	zfs_getpage(struct inode ip, struct page pl[], int nr_pages)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);
	int err;

	if (pl == NULL)
	return (0);

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	err = zfs_fillpage(ip, pl, nr_pages);

	ZFS_EXIT(zfsvfs);
	return (err);
	}

	/*
	* Check ZFS specific permissions to memory map a section of a file.
	*
	* IN: ip - inode of the file to mmap
	* off - file offset
	* addrp - start address in memory region
	* len - length of memory region
	* vm_flags- address flags
	*
	* RETURN: 0 if success
	* error code if failure
	*/
	/ARGSUSED/
	int
	zfs_map(struct inode ip, offset_t off, caddr_t addrp, size_t len,
	unsigned long vm_flags)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	if ((vm_flags & VM_WRITE) && (zp->z_pflags &
	(ZFS_IMMUTABLE \| ZFS_READONLY \| ZFS_APPENDONLY))) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EPERM));
	}

	if ((vm_flags & (VM_READ \| VM_EXEC)) &&
	(zp->z_pflags & ZFS_AV_QUARANTINED)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EACCES));
	}

	if (off < 0 \|\| len > MAXOFFSET_T - off) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(ENXIO));
	}

	ZFS_EXIT(zfsvfs);
	return (0);
	}

	/*
	* Free or allocate space in a file. Currently, this function only
	* supports the `F_FREESP' command. However, this command is somewhat
	* misnamed, as its functionality includes the ability to allocate as
	* well as free space.
	*
	* IN: zp - znode of file to free data in.
	* cmd - action to take (only F_FREESP supported).
	* bfp - section of file to free/alloc.
	* flag - current file open mode flags.
	* offset - current file offset.
	* cr - credentials of caller.
	*
	* RETURN: 0 on success, error code on failure.
	*
	* Timestamps:
	* zp - ctime\|mtime updated
	*/
	/* ARGSUSED */
	int
	zfs_space(znode_t zp, int cmd, flock64_t bfp, int flag,
	offset_t offset, cred_t *cr)
	{
	zfsvfs_t *zfsvfs = ZTOZSB(zp);
	uint64_t off, len;
	int error;

	ZFS_ENTER(zfsvfs);
	ZFS_VERIFY_ZP(zp);

	if (cmd != F_FREESP) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	/*
	* Callers might not be able to detect properly that we are read-only,
	* so check it explicitly here.
	*/
	if (zfs_is_readonly(zfsvfs)) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EROFS));
	}

	if (bfp->l_len < 0) {
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(EINVAL));
	}

	/*
	* Permissions aren't checked on Solaris because on this OS
	* zfs_space() can only be called with an opened file handle.
	* On Linux we can get here through truncate_range() which
	* operates directly on inodes, so we need to check access rights.
	*/
	if ((error = zfs_zaccess(zp, ACE_WRITE_DATA, 0, B_FALSE, cr))) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	off = bfp->l_start;
	len = bfp->l_len; /* 0 means from off to end of file */

	error = zfs_freesp(zp, off, len, flag, TRUE);

	ZFS_EXIT(zfsvfs);
	return (error);
	}

	/ARGSUSED/
	int
	zfs_fid(struct inode ip, fid_t fidp)
	{
	znode_t *zp = ITOZ(ip);
	zfsvfs_t *zfsvfs = ITOZSB(ip);
	uint32_t gen;
	uint64_t gen64;
	uint64_t object = zp->z_id;
	zfid_short_t *zfid;
	int size, i, error;

	ZFS_ENTER(zfsvfs);

	if (fidp->fid_len < SHORT_FID_LEN) {
	fidp->fid_len = SHORT_FID_LEN;
	ZFS_EXIT(zfsvfs);
	return (SET_ERROR(ENOSPC));
	}

	ZFS_VERIFY_ZP(zp);

	if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zfsvfs),
	&gen64, sizeof (uint64_t))) != 0) {
	ZFS_EXIT(zfsvfs);
	return (error);
	}

	gen = (uint32_t)gen64;

	size = SHORT_FID_LEN;

	zfid = (zfid_short_t *)fidp;

	zfid->zf_len = size;

	for (i = 0; i < sizeof (zfid->zf_object); i++)
	zfid->zf_object[i] = (uint8_t)(object >> (8 * i));

	/* Must have a non-zero generation number to distinguish from .zfs */
	if (gen == 0)
	gen = 1;
	for (i = 0; i < sizeof (zfid->zf_gen); i++)
	zfid->zf_gen[i] = (uint8_t)(gen >> (8 * i));

	ZFS_EXIT(zfsvfs);
	return (0);
	}

	#if defined(_KERNEL)
	EXPORT_SYMBOL(zfs_open);
	EXPORT_SYMBOL(zfs_close);
	EXPORT_SYMBOL(zfs_lookup);
	EXPORT_SYMBOL(zfs_create);
	EXPORT_SYMBOL(zfs_tmpfile);
	EXPORT_SYMBOL(zfs_remove);
	EXPORT_SYMBOL(zfs_mkdir);
	EXPORT_SYMBOL(zfs_rmdir);
	EXPORT_SYMBOL(zfs_readdir);
	EXPORT_SYMBOL(zfs_getattr_fast);
	EXPORT_SYMBOL(zfs_setattr);
	EXPORT_SYMBOL(zfs_rename);
	EXPORT_SYMBOL(zfs_symlink);
	EXPORT_SYMBOL(zfs_readlink);
	EXPORT_SYMBOL(zfs_link);
	EXPORT_SYMBOL(zfs_inactive);
	EXPORT_SYMBOL(zfs_space);
	EXPORT_SYMBOL(zfs_fid);
	EXPORT_SYMBOL(zfs_getpage);
	EXPORT_SYMBOL(zfs_putpage);
	EXPORT_SYMBOL(zfs_dirty_inode);
	EXPORT_SYMBOL(zfs_map);

	/* BEGIN CSTYLED */
	module_param(zfs_delete_blocks, ulong, 0644);
	MODULE_PARM_DESC(zfs_delete_blocks, "Delete files larger than N blocks async");
	/* END CSTYLED */

	#endif
	diff --git a/sys/contrib/openzfs/module/os/linux/zfs/zpl_file.c b/sys/contrib/openzfs/module/os/linux/zfs/zpl_file.c
	index 63002fe3b932..7e88eae33711 100644
	--- a/sys/contrib/openzfs/module/os/linux/zfs/zpl_file.c
	+++ b/sys/contrib/openzfs/module/os/linux/zfs/zpl_file.c
	@@ -1,1083 +1,1089 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2011, Lawrence Livermore National Security, LLC.
	* Copyright (c) 2015 by Chunwei Chen. All rights reserved.
	*/


	#ifdef CONFIG_COMPAT
	#include <linux/compat.h>
	#endif
	#include <sys/file.h>
	#include <sys/dmu_objset.h>
	#include <sys/zfs_znode.h>
	#include <sys/zfs_vfsops.h>
	#include <sys/zfs_vnops.h>
	#include <sys/zfs_project.h>
	#ifdef HAVE_VFS_SET_PAGE_DIRTY_NOBUFFERS
	#include <linux/pagemap.h>
	#endif

	/*
	* When using fallocate(2) to preallocate space, inflate the requested
	* capacity check by 10% to account for the required metadata blocks.
	*/
	unsigned int zfs_fallocate_reserve_percent = 110;

	static int
	zpl_open(struct inode ip, struct file filp)
	{
	cred_t *cr = CRED();
	int error;
	fstrans_cookie_t cookie;

	error = generic_file_open(ip, filp);
	if (error)
	return (error);

	crhold(cr);
	cookie = spl_fstrans_mark();
	error = -zfs_open(ip, filp->f_mode, filp->f_flags, cr);
	spl_fstrans_unmark(cookie);
	crfree(cr);
	ASSERT3S(error, <=, 0);

	return (error);
	}

	static int
	zpl_release(struct inode ip, struct file filp)
	{
	cred_t *cr = CRED();
	int error;
	fstrans_cookie_t cookie;

	cookie = spl_fstrans_mark();
	if (ITOZ(ip)->z_atime_dirty)
	zfs_mark_inode_dirty(ip);

	crhold(cr);
	error = -zfs_close(ip, filp->f_flags, cr);
	spl_fstrans_unmark(cookie);
	crfree(cr);
	ASSERT3S(error, <=, 0);

	return (error);
	}

	static int
	zpl_iterate(struct file filp, zpl_dir_context_t ctx)
	{
	cred_t *cr = CRED();
	int error;
	fstrans_cookie_t cookie;

	crhold(cr);
	cookie = spl_fstrans_mark();
	error = -zfs_readdir(file_inode(filp), ctx, cr);
	spl_fstrans_unmark(cookie);
	crfree(cr);
	ASSERT3S(error, <=, 0);

	return (error);
	}

	#if !defined(HAVE_VFS_ITERATE) && !defined(HAVE_VFS_ITERATE_SHARED)
	static int
	zpl_readdir(struct file filp, void dirent, filldir_t filldir)
	{
	zpl_dir_context_t ctx =
	ZPL_DIR_CONTEXT_INIT(dirent, filldir, filp->f_pos);
	int error;

	error = zpl_iterate(filp, &ctx);
	filp->f_pos = ctx.pos;

	return (error);
	}
	#endif /* !HAVE_VFS_ITERATE && !HAVE_VFS_ITERATE_SHARED */

	#if defined(HAVE_FSYNC_WITHOUT_DENTRY)
	/*
	* Linux 2.6.35 - 3.0 API,
	* As of 2.6.35 the dentry argument to the fops->fsync() hook was deemed
	* redundant. The dentry is still accessible via filp->f_path.dentry,
	* and we are guaranteed that filp will never be NULL.
	*/
	static int
	zpl_fsync(struct file *filp, int datasync)
	{
	struct inode *inode = filp->f_mapping->host;
	cred_t *cr = CRED();
	int error;
	fstrans_cookie_t cookie;

	crhold(cr);
	cookie = spl_fstrans_mark();
	error = -zfs_fsync(ITOZ(inode), datasync, cr);
	spl_fstrans_unmark(cookie);
	crfree(cr);
	ASSERT3S(error, <=, 0);

	return (error);
	}

	#ifdef HAVE_FILE_AIO_FSYNC
	static int
	zpl_aio_fsync(struct kiocb *kiocb, int datasync)
	{
	return (zpl_fsync(kiocb->ki_filp, datasync));
	}
	#endif

	#elif defined(HAVE_FSYNC_RANGE)
	/*
	* Linux 3.1 API,
	* As of 3.1 the responsibility to call filemap_write_and_wait_range() has
	* been pushed down in to the .fsync() vfs hook. Additionally, the i_mutex
	* lock is no longer held by the caller, for zfs we don't require the lock
	* to be held so we don't acquire it.
	*/
	static int
	zpl_fsync(struct file *filp, loff_t start, loff_t end, int datasync)
	{
	struct inode *inode = filp->f_mapping->host;
	cred_t *cr = CRED();
	int error;
	fstrans_cookie_t cookie;

	error = filemap_write_and_wait_range(inode->i_mapping, start, end);
	if (error)
	return (error);

	crhold(cr);
	cookie = spl_fstrans_mark();
	error = -zfs_fsync(ITOZ(inode), datasync, cr);
	spl_fstrans_unmark(cookie);
	crfree(cr);
	ASSERT3S(error, <=, 0);

	return (error);
	}

	#ifdef HAVE_FILE_AIO_FSYNC
	static int
	zpl_aio_fsync(struct kiocb *kiocb, int datasync)
	{
	return (zpl_fsync(kiocb->ki_filp, kiocb->ki_pos, -1, datasync));
	}
	#endif

	#else
	#error "Unsupported fops->fsync() implementation"
	#endif

	static inline int
	zfs_io_flags(struct kiocb *kiocb)
	{
	int flags = 0;

	#if defined(IOCB_DSYNC)
	if (kiocb->ki_flags & IOCB_DSYNC)
	flags \|= O_DSYNC;
	#endif
	#if defined(IOCB_SYNC)
	if (kiocb->ki_flags & IOCB_SYNC)
	flags \|= O_SYNC;
	#endif
	#if defined(IOCB_APPEND)
	if (kiocb->ki_flags & IOCB_APPEND)
	flags \|= O_APPEND;
	#endif
	#if defined(IOCB_DIRECT)
	if (kiocb->ki_flags & IOCB_DIRECT)
	flags \|= O_DIRECT;
	#endif
	return (flags);
	}

	/*
	* If relatime is enabled, call file_accessed() if zfs_relatime_need_update()
	* is true. This is needed since datasets with inherited "relatime" property
	* aren't necessarily mounted with the MNT_RELATIME flag (e.g. after
	* `zfs set relatime=...`), which is what relatime test in VFS by
	* relatime_need_update() is based on.
	*/
	static inline void
	zpl_file_accessed(struct file *filp)
	{
	struct inode *ip = filp->f_mapping->host;

	if (!IS_NOATIME(ip) && ITOZSB(ip)->z_relatime) {
	if (zfs_relatime_need_update(ip))
	file_accessed(filp);
	} else {
	file_accessed(filp);
	}
	}

	#if defined(HAVE_VFS_RW_ITERATE)

	/*
	* When HAVE_VFS_IOV_ITER is defined the iov_iter structure supports
	* iovecs, kvevs, bvecs and pipes, plus all the required interfaces to
	* manipulate the iov_iter are available. In which case the full iov_iter
	* can be attached to the uio and correctly handled in the lower layers.
	* Otherwise, for older kernels extract the iovec and pass it instead.
	*/
	static void
	zpl_uio_init(zfs_uio_t uio, struct kiocb kiocb, struct iov_iter *to,
	loff_t pos, ssize_t count, size_t skip)
	{
	#if defined(HAVE_VFS_IOV_ITER)
	zfs_uio_iov_iter_init(uio, to, pos, count, skip);
	+#else
	+#ifdef HAVE_IOV_ITER_TYPE
	+ zfs_uio_iovec_init(uio, to->iov, to->nr_segs, pos,
	+ iov_iter_type(to) & ITER_KVEC ? UIO_SYSSPACE : UIO_USERSPACE,
	+ count, skip);
	#else
	zfs_uio_iovec_init(uio, to->iov, to->nr_segs, pos,
	to->type & ITER_KVEC ? UIO_SYSSPACE : UIO_USERSPACE,
	count, skip);
	#endif
	+#endif
	}

	static ssize_t
	zpl_iter_read(struct kiocb kiocb, struct iov_iter to)
	{
	cred_t *cr = CRED();
	fstrans_cookie_t cookie;
	struct file *filp = kiocb->ki_filp;
	ssize_t count = iov_iter_count(to);
	zfs_uio_t uio;

	zpl_uio_init(&uio, kiocb, to, kiocb->ki_pos, count, 0);

	crhold(cr);
	cookie = spl_fstrans_mark();

	int error = -zfs_read(ITOZ(filp->f_mapping->host), &uio,
	filp->f_flags \| zfs_io_flags(kiocb), cr);

	spl_fstrans_unmark(cookie);
	crfree(cr);

	if (error < 0)
	return (error);

	ssize_t read = count - uio.uio_resid;
	kiocb->ki_pos += read;

	zpl_file_accessed(filp);

	return (read);
	}

	static inline ssize_t
	zpl_generic_write_checks(struct kiocb kiocb, struct iov_iter from,
	size_t *countp)
	{
	#ifdef HAVE_GENERIC_WRITE_CHECKS_KIOCB
	ssize_t ret = generic_write_checks(kiocb, from);
	if (ret <= 0)
	return (ret);

	*countp = ret;
	#else
	struct file *file = kiocb->ki_filp;
	struct address_space *mapping = file->f_mapping;
	struct inode *ip = mapping->host;
	int isblk = S_ISBLK(ip->i_mode);

	*countp = iov_iter_count(from);
	ssize_t ret = generic_write_checks(file, &kiocb->ki_pos, countp, isblk);
	if (ret)
	return (ret);
	#endif

	return (0);
	}

	static ssize_t
	zpl_iter_write(struct kiocb kiocb, struct iov_iter from)
	{
	cred_t *cr = CRED();
	fstrans_cookie_t cookie;
	struct file *filp = kiocb->ki_filp;
	struct inode *ip = filp->f_mapping->host;
	zfs_uio_t uio;
	size_t count = 0;
	ssize_t ret;

	ret = zpl_generic_write_checks(kiocb, from, &count);
	if (ret)
	return (ret);

	zpl_uio_init(&uio, kiocb, from, kiocb->ki_pos, count, from->iov_offset);

	crhold(cr);
	cookie = spl_fstrans_mark();

	int error = -zfs_write(ITOZ(ip), &uio,
	filp->f_flags \| zfs_io_flags(kiocb), cr);

	spl_fstrans_unmark(cookie);
	crfree(cr);

	if (error < 0)
	return (error);

	ssize_t wrote = count - uio.uio_resid;
	kiocb->ki_pos += wrote;

	return (wrote);
	}

	#else /* !HAVE_VFS_RW_ITERATE */

	static ssize_t
	zpl_aio_read(struct kiocb kiocb, const struct iovec iov,
	unsigned long nr_segs, loff_t pos)
	{
	cred_t *cr = CRED();
	fstrans_cookie_t cookie;
	struct file *filp = kiocb->ki_filp;
	size_t count;
	ssize_t ret;

	ret = generic_segment_checks(iov, &nr_segs, &count, VERIFY_WRITE);
	if (ret)
	return (ret);

	zfs_uio_t uio;
	zfs_uio_iovec_init(&uio, iov, nr_segs, kiocb->ki_pos, UIO_USERSPACE,
	count, 0);

	crhold(cr);
	cookie = spl_fstrans_mark();

	int error = -zfs_read(ITOZ(filp->f_mapping->host), &uio,
	filp->f_flags \| zfs_io_flags(kiocb), cr);

	spl_fstrans_unmark(cookie);
	crfree(cr);

	if (error < 0)
	return (error);

	ssize_t read = count - uio.uio_resid;
	kiocb->ki_pos += read;

	zpl_file_accessed(filp);

	return (read);
	}

	static ssize_t
	zpl_aio_write(struct kiocb kiocb, const struct iovec iov,
	unsigned long nr_segs, loff_t pos)
	{
	cred_t *cr = CRED();
	fstrans_cookie_t cookie;
	struct file *filp = kiocb->ki_filp;
	struct inode *ip = filp->f_mapping->host;
	size_t count;
	ssize_t ret;

	ret = generic_segment_checks(iov, &nr_segs, &count, VERIFY_READ);
	if (ret)
	return (ret);

	ret = generic_write_checks(filp, &pos, &count, S_ISBLK(ip->i_mode));
	if (ret)
	return (ret);

	zfs_uio_t uio;
	zfs_uio_iovec_init(&uio, iov, nr_segs, kiocb->ki_pos, UIO_USERSPACE,
	count, 0);

	crhold(cr);
	cookie = spl_fstrans_mark();

	int error = -zfs_write(ITOZ(ip), &uio,
	filp->f_flags \| zfs_io_flags(kiocb), cr);

	spl_fstrans_unmark(cookie);
	crfree(cr);

	if (error < 0)
	return (error);

	ssize_t wrote = count - uio.uio_resid;
	kiocb->ki_pos += wrote;

	return (wrote);
	}
	#endif /* HAVE_VFS_RW_ITERATE */

	#if defined(HAVE_VFS_RW_ITERATE)
	static ssize_t
	zpl_direct_IO_impl(int rw, struct kiocb kiocb, struct iov_iter iter)
	{
	if (rw == WRITE)
	return (zpl_iter_write(kiocb, iter));
	else
	return (zpl_iter_read(kiocb, iter));
	}
	#if defined(HAVE_VFS_DIRECT_IO_ITER)
	static ssize_t
	zpl_direct_IO(struct kiocb kiocb, struct iov_iter iter)
	{
	return (zpl_direct_IO_impl(iov_iter_rw(iter), kiocb, iter));
	}
	#elif defined(HAVE_VFS_DIRECT_IO_ITER_OFFSET)
	static ssize_t
	zpl_direct_IO(struct kiocb kiocb, struct iov_iter iter, loff_t pos)
	{
	ASSERT3S(pos, ==, kiocb->ki_pos);
	return (zpl_direct_IO_impl(iov_iter_rw(iter), kiocb, iter));
	}
	#elif defined(HAVE_VFS_DIRECT_IO_ITER_RW_OFFSET)
	static ssize_t
	zpl_direct_IO(int rw, struct kiocb kiocb, struct iov_iter iter, loff_t pos)
	{
	ASSERT3S(pos, ==, kiocb->ki_pos);
	return (zpl_direct_IO_impl(rw, kiocb, iter));
	}
	#else
	#error "Unknown direct IO interface"
	#endif

	#else /* HAVE_VFS_RW_ITERATE */

	#if defined(HAVE_VFS_DIRECT_IO_IOVEC)
	static ssize_t
	zpl_direct_IO(int rw, struct kiocb kiocb, const struct iovec iov,
	loff_t pos, unsigned long nr_segs)
	{
	if (rw == WRITE)
	return (zpl_aio_write(kiocb, iov, nr_segs, pos));
	else
	return (zpl_aio_read(kiocb, iov, nr_segs, pos));
	}
	#elif defined(HAVE_VFS_DIRECT_IO_ITER_RW_OFFSET)
	static ssize_t
	zpl_direct_IO(int rw, struct kiocb kiocb, struct iov_iter iter, loff_t pos)
	{
	const struct iovec *iovp = iov_iter_iovec(iter);
	unsigned long nr_segs = iter->nr_segs;

	ASSERT3S(pos, ==, kiocb->ki_pos);
	if (rw == WRITE)
	return (zpl_aio_write(kiocb, iovp, nr_segs, pos));
	else
	return (zpl_aio_read(kiocb, iovp, nr_segs, pos));
	}
	#else
	#error "Unknown direct IO interface"
	#endif

	#endif /* HAVE_VFS_RW_ITERATE */

	static loff_t
	zpl_llseek(struct file *filp, loff_t offset, int whence)
	{
	#if defined(SEEK_HOLE) && defined(SEEK_DATA)
	fstrans_cookie_t cookie;

	if (whence == SEEK_DATA \|\| whence == SEEK_HOLE) {
	struct inode *ip = filp->f_mapping->host;
	loff_t maxbytes = ip->i_sb->s_maxbytes;
	loff_t error;

	spl_inode_lock_shared(ip);
	cookie = spl_fstrans_mark();
	error = -zfs_holey(ITOZ(ip), whence, &offset);
	spl_fstrans_unmark(cookie);
	if (error == 0)
	error = lseek_execute(filp, ip, offset, maxbytes);
	spl_inode_unlock_shared(ip);

	return (error);
	}
	#endif /* SEEK_HOLE && SEEK_DATA */

	return (generic_file_llseek(filp, offset, whence));
	}

	/*
	* It's worth taking a moment to describe how mmap is implemented
	* for zfs because it differs considerably from other Linux filesystems.
	* However, this issue is handled the same way under OpenSolaris.
	*
	* The issue is that by design zfs bypasses the Linux page cache and
	* leaves all caching up to the ARC. This has been shown to work
	* well for the common read(2)/write(2) case. However, mmap(2)
	* is problem because it relies on being tightly integrated with the
	* page cache. To handle this we cache mmap'ed files twice, once in
	* the ARC and a second time in the page cache. The code is careful
	* to keep both copies synchronized.
	*
	* When a file with an mmap'ed region is written to using write(2)
	* both the data in the ARC and existing pages in the page cache
	* are updated. For a read(2) data will be read first from the page
	* cache then the ARC if needed. Neither a write(2) or read(2) will
	* will ever result in new pages being added to the page cache.
	*
	* New pages are added to the page cache only via .readpage() which
	* is called when the vfs needs to read a page off disk to back the
	* virtual memory region. These pages may be modified without
	* notifying the ARC and will be written out periodically via
	* .writepage(). This will occur due to either a sync or the usual
	* page aging behavior. Note because a read(2) of a mmap'ed file
	* will always check the page cache first even when the ARC is out
	* of date correct data will still be returned.
	*
	* While this implementation ensures correct behavior it does have
	* have some drawbacks. The most obvious of which is that it
	* increases the required memory footprint when access mmap'ed
	* files. It also adds additional complexity to the code keeping
	* both caches synchronized.
	*
	* Longer term it may be possible to cleanly resolve this wart by
	* mapping page cache pages directly on to the ARC buffers. The
	* Linux address space operations are flexible enough to allow
	* selection of which pages back a particular index. The trick
	* would be working out the details of which subsystem is in
	* charge, the ARC, the page cache, or both. It may also prove
	* helpful to move the ARC buffers to a scatter-gather lists
	* rather than a vmalloc'ed region.
	*/
	static int
	zpl_mmap(struct file filp, struct vm_area_struct vma)
	{
	struct inode *ip = filp->f_mapping->host;
	znode_t *zp = ITOZ(ip);
	int error;
	fstrans_cookie_t cookie;

	cookie = spl_fstrans_mark();
	error = -zfs_map(ip, vma->vm_pgoff, (caddr_t *)vma->vm_start,
	(size_t)(vma->vm_end - vma->vm_start), vma->vm_flags);
	spl_fstrans_unmark(cookie);
	if (error)
	return (error);

	error = generic_file_mmap(filp, vma);
	if (error)
	return (error);

	mutex_enter(&zp->z_lock);
	zp->z_is_mapped = B_TRUE;
	mutex_exit(&zp->z_lock);

	return (error);
	}

	/*
	* Populate a page with data for the Linux page cache. This function is
	* only used to support mmap(2). There will be an identical copy of the
	* data in the ARC which is kept up to date via .write() and .writepage().
	*/
	static inline int
	zpl_readpage_common(struct page *pp)
	{
	struct inode *ip;
	struct page *pl[1];
	int error = 0;
	fstrans_cookie_t cookie;

	ASSERT(PageLocked(pp));
	ip = pp->mapping->host;
	pl[0] = pp;

	cookie = spl_fstrans_mark();
	error = -zfs_getpage(ip, pl, 1);
	spl_fstrans_unmark(cookie);

	if (error) {
	SetPageError(pp);
	ClearPageUptodate(pp);
	} else {
	ClearPageError(pp);
	SetPageUptodate(pp);
	flush_dcache_page(pp);
	}

	unlock_page(pp);
	return (error);
	}

	static int
	zpl_readpage(struct file filp, struct page pp)
	{
	return (zpl_readpage_common(pp));
	}

	static int
	zpl_readpage_filler(void data, struct page pp)
	{
	return (zpl_readpage_common(pp));
	}

	/*
	* Populate a set of pages with data for the Linux page cache. This
	* function will only be called for read ahead and never for demand
	* paging. For simplicity, the code relies on read_cache_pages() to
	* correctly lock each page for IO and call zpl_readpage().
	*/
	static int
	zpl_readpages(struct file filp, struct address_space mapping,
	struct list_head *pages, unsigned nr_pages)
	{
	return (read_cache_pages(mapping, pages, zpl_readpage_filler, NULL));
	}

	static int
	zpl_putpage(struct page pp, struct writeback_control wbc, void *data)
	{
	struct address_space *mapping = data;
	fstrans_cookie_t cookie;

	ASSERT(PageLocked(pp));
	ASSERT(!PageWriteback(pp));

	cookie = spl_fstrans_mark();
	(void) zfs_putpage(mapping->host, pp, wbc);
	spl_fstrans_unmark(cookie);

	return (0);
	}

	static int
	zpl_writepages(struct address_space mapping, struct writeback_control wbc)
	{
	znode_t *zp = ITOZ(mapping->host);
	zfsvfs_t *zfsvfs = ITOZSB(mapping->host);
	enum writeback_sync_modes sync_mode;
	int result;

	ZPL_ENTER(zfsvfs);
	if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
	wbc->sync_mode = WB_SYNC_ALL;
	ZPL_EXIT(zfsvfs);
	sync_mode = wbc->sync_mode;

	/*
	* We don't want to run write_cache_pages() in SYNC mode here, because
	* that would make putpage() wait for a single page to be committed to
	* disk every single time, resulting in atrocious performance. Instead
	* we run it once in non-SYNC mode so that the ZIL gets all the data,
	* and then we commit it all in one go.
	*/
	wbc->sync_mode = WB_SYNC_NONE;
	result = write_cache_pages(mapping, wbc, zpl_putpage, mapping);
	if (sync_mode != wbc->sync_mode) {
	ZPL_ENTER(zfsvfs);
	ZPL_VERIFY_ZP(zp);
	if (zfsvfs->z_log != NULL)
	zil_commit(zfsvfs->z_log, zp->z_id);
	ZPL_EXIT(zfsvfs);

	/*
	* We need to call write_cache_pages() again (we can't just
	* return after the commit) because the previous call in
	* non-SYNC mode does not guarantee that we got all the dirty
	* pages (see the implementation of write_cache_pages() for
	* details). That being said, this is a no-op in most cases.
	*/
	wbc->sync_mode = sync_mode;
	result = write_cache_pages(mapping, wbc, zpl_putpage, mapping);
	}
	return (result);
	}

	/*
	* Write out dirty pages to the ARC, this function is only required to
	* support mmap(2). Mapped pages may be dirtied by memory operations
	* which never call .write(). These dirty pages are kept in sync with
	* the ARC buffers via this hook.
	*/
	static int
	zpl_writepage(struct page pp, struct writeback_control wbc)
	{
	if (ITOZSB(pp->mapping->host)->z_os->os_sync == ZFS_SYNC_ALWAYS)
	wbc->sync_mode = WB_SYNC_ALL;

	return (zpl_putpage(pp, wbc, pp->mapping));
	}

	/*
	* The flag combination which matches the behavior of zfs_space() is
	* FALLOC_FL_KEEP_SIZE \| FALLOC_FL_PUNCH_HOLE. The FALLOC_FL_PUNCH_HOLE
	* flag was introduced in the 2.6.38 kernel.
	*
	* The original mode=0 (allocate space) behavior can be reasonably emulated
	* by checking if enough space exists and creating a sparse file, as real
	* persistent space reservation is not possible due to COW, snapshots, etc.
	*/
	static long
	zpl_fallocate_common(struct inode *ip, int mode, loff_t offset, loff_t len)
	{
	cred_t *cr = CRED();
	loff_t olen;
	fstrans_cookie_t cookie;
	int error = 0;

	if ((mode & ~(FALLOC_FL_KEEP_SIZE \| FALLOC_FL_PUNCH_HOLE)) != 0)
	return (-EOPNOTSUPP);

	if (offset < 0 \|\| len <= 0)
	return (-EINVAL);

	spl_inode_lock(ip);
	olen = i_size_read(ip);

	crhold(cr);
	cookie = spl_fstrans_mark();
	if (mode & FALLOC_FL_PUNCH_HOLE) {
	flock64_t bf;

	if (offset > olen)
	goto out_unmark;

	if (offset + len > olen)
	len = olen - offset;
	bf.l_type = F_WRLCK;
	bf.l_whence = SEEK_SET;
	bf.l_start = offset;
	bf.l_len = len;
	bf.l_pid = 0;

	error = -zfs_space(ITOZ(ip), F_FREESP, &bf, O_RDWR, offset, cr);
	} else if ((mode & ~FALLOC_FL_KEEP_SIZE) == 0) {
	unsigned int percent = zfs_fallocate_reserve_percent;
	struct kstatfs statfs;

	/* Legacy mode, disable fallocate compatibility. */
	if (percent == 0) {
	error = -EOPNOTSUPP;
	goto out_unmark;
	}

	/*
	* Use zfs_statvfs() instead of dmu_objset_space() since it
	* also checks project quota limits, which are relevant here.
	*/
	error = zfs_statvfs(ip, &statfs);
	if (error)
	goto out_unmark;

	/*
	* Shrink available space a bit to account for overhead/races.
	* We know the product previously fit into availbytes from
	* dmu_objset_space(), so the smaller product will also fit.
	*/
	if (len > statfs.f_bavail * (statfs.f_bsize * 100 / percent)) {
	error = -ENOSPC;
	goto out_unmark;
	}
	if (!(mode & FALLOC_FL_KEEP_SIZE) && offset + len > olen)
	error = zfs_freesp(ITOZ(ip), offset + len, 0, 0, FALSE);
	}
	out_unmark:
	spl_fstrans_unmark(cookie);
	spl_inode_unlock(ip);

	crfree(cr);

	return (error);
	}

	static long
	zpl_fallocate(struct file *filp, int mode, loff_t offset, loff_t len)
	{
	return zpl_fallocate_common(file_inode(filp),
	mode, offset, len);
	}

	#define ZFS_FL_USER_VISIBLE (FS_FL_USER_VISIBLE \| ZFS_PROJINHERIT_FL)
	#define ZFS_FL_USER_MODIFIABLE (FS_FL_USER_MODIFIABLE \| ZFS_PROJINHERIT_FL)

	static uint32_t
	__zpl_ioctl_getflags(struct inode *ip)
	{
	uint64_t zfs_flags = ITOZ(ip)->z_pflags;
	uint32_t ioctl_flags = 0;

	if (zfs_flags & ZFS_IMMUTABLE)
	ioctl_flags \|= FS_IMMUTABLE_FL;

	if (zfs_flags & ZFS_APPENDONLY)
	ioctl_flags \|= FS_APPEND_FL;

	if (zfs_flags & ZFS_NODUMP)
	ioctl_flags \|= FS_NODUMP_FL;

	if (zfs_flags & ZFS_PROJINHERIT)
	ioctl_flags \|= ZFS_PROJINHERIT_FL;

	return (ioctl_flags & ZFS_FL_USER_VISIBLE);
	}

	/*
	* Map zfs file z_pflags (xvattr_t) to linux file attributes. Only file
	* attributes common to both Linux and Solaris are mapped.
	*/
	static int
	zpl_ioctl_getflags(struct file filp, void __user arg)
	{
	uint32_t flags;
	int err;

	flags = __zpl_ioctl_getflags(file_inode(filp));
	err = copy_to_user(arg, &flags, sizeof (flags));

	return (err);
	}

	/*
	* fchange() is a helper macro to detect if we have been asked to change a
	* flag. This is ugly, but the requirement that we do this is a consequence of
	* how the Linux file attribute interface was designed. Another consequence is
	* that concurrent modification of files suffers from a TOCTOU race. Neither
	* are things we can fix without modifying the kernel-userland interface, which
	* is outside of our jurisdiction.
	*/

	#define fchange(f0, f1, b0, b1) (!((f0) & (b0)) != !((f1) & (b1)))

	static int
	__zpl_ioctl_setflags(struct inode ip, uint32_t ioctl_flags, xvattr_t xva)
	{
	uint64_t zfs_flags = ITOZ(ip)->z_pflags;
	xoptattr_t *xoap;

	if (ioctl_flags & ~(FS_IMMUTABLE_FL \| FS_APPEND_FL \| FS_NODUMP_FL \|
	ZFS_PROJINHERIT_FL))
	return (-EOPNOTSUPP);

	if (ioctl_flags & ~ZFS_FL_USER_MODIFIABLE)
	return (-EACCES);

	if ((fchange(ioctl_flags, zfs_flags, FS_IMMUTABLE_FL, ZFS_IMMUTABLE) \|\|
	fchange(ioctl_flags, zfs_flags, FS_APPEND_FL, ZFS_APPENDONLY)) &&
	!capable(CAP_LINUX_IMMUTABLE))
	return (-EPERM);

	if (!zpl_inode_owner_or_capable(kcred->user_ns, ip))
	return (-EACCES);

	xva_init(xva);
	xoap = xva_getxoptattr(xva);

	XVA_SET_REQ(xva, XAT_IMMUTABLE);
	if (ioctl_flags & FS_IMMUTABLE_FL)
	xoap->xoa_immutable = B_TRUE;

	XVA_SET_REQ(xva, XAT_APPENDONLY);
	if (ioctl_flags & FS_APPEND_FL)
	xoap->xoa_appendonly = B_TRUE;

	XVA_SET_REQ(xva, XAT_NODUMP);
	if (ioctl_flags & FS_NODUMP_FL)
	xoap->xoa_nodump = B_TRUE;

	XVA_SET_REQ(xva, XAT_PROJINHERIT);
	if (ioctl_flags & ZFS_PROJINHERIT_FL)
	xoap->xoa_projinherit = B_TRUE;

	return (0);
	}

	static int
	zpl_ioctl_setflags(struct file filp, void __user arg)
	{
	struct inode *ip = file_inode(filp);
	uint32_t flags;
	cred_t *cr = CRED();
	xvattr_t xva;
	int err;
	fstrans_cookie_t cookie;

	if (copy_from_user(&flags, arg, sizeof (flags)))
	return (-EFAULT);

	err = __zpl_ioctl_setflags(ip, flags, &xva);
	if (err)
	return (err);

	crhold(cr);
	cookie = spl_fstrans_mark();
	err = -zfs_setattr(ITOZ(ip), (vattr_t *)&xva, 0, cr);
	spl_fstrans_unmark(cookie);
	crfree(cr);

	return (err);
	}

	static int
	zpl_ioctl_getxattr(struct file filp, void __user arg)
	{
	zfsxattr_t fsx = { 0 };
	struct inode *ip = file_inode(filp);
	int err;

	fsx.fsx_xflags = __zpl_ioctl_getflags(ip);
	fsx.fsx_projid = ITOZ(ip)->z_projid;
	err = copy_to_user(arg, &fsx, sizeof (fsx));

	return (err);
	}

	static int
	zpl_ioctl_setxattr(struct file filp, void __user arg)
	{
	struct inode *ip = file_inode(filp);
	zfsxattr_t fsx;
	cred_t *cr = CRED();
	xvattr_t xva;
	xoptattr_t *xoap;
	int err;
	fstrans_cookie_t cookie;

	if (copy_from_user(&fsx, arg, sizeof (fsx)))
	return (-EFAULT);

	if (!zpl_is_valid_projid(fsx.fsx_projid))
	return (-EINVAL);

	err = __zpl_ioctl_setflags(ip, fsx.fsx_xflags, &xva);
	if (err)
	return (err);

	xoap = xva_getxoptattr(&xva);
	XVA_SET_REQ(&xva, XAT_PROJID);
	xoap->xoa_projid = fsx.fsx_projid;

	crhold(cr);
	cookie = spl_fstrans_mark();
	err = -zfs_setattr(ITOZ(ip), (vattr_t *)&xva, 0, cr);
	spl_fstrans_unmark(cookie);
	crfree(cr);

	return (err);
	}

	static long
	zpl_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
	{
	switch (cmd) {
	case FS_IOC_GETFLAGS:
	return (zpl_ioctl_getflags(filp, (void *)arg));
	case FS_IOC_SETFLAGS:
	return (zpl_ioctl_setflags(filp, (void *)arg));
	case ZFS_IOC_FSGETXATTR:
	return (zpl_ioctl_getxattr(filp, (void *)arg));
	case ZFS_IOC_FSSETXATTR:
	return (zpl_ioctl_setxattr(filp, (void *)arg));
	default:
	return (-ENOTTY);
	}
	}

	#ifdef CONFIG_COMPAT
	static long
	zpl_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
	{
	switch (cmd) {
	case FS_IOC32_GETFLAGS:
	cmd = FS_IOC_GETFLAGS;
	break;
	case FS_IOC32_SETFLAGS:
	cmd = FS_IOC_SETFLAGS;
	break;
	default:
	return (-ENOTTY);
	}
	return (zpl_ioctl(filp, cmd, (unsigned long)compat_ptr(arg)));
	}
	#endif /* CONFIG_COMPAT */


	const struct address_space_operations zpl_address_space_operations = {
	.readpages = zpl_readpages,
	.readpage = zpl_readpage,
	.writepage = zpl_writepage,
	.writepages = zpl_writepages,
	.direct_IO = zpl_direct_IO,
	#ifdef HAVE_VFS_SET_PAGE_DIRTY_NOBUFFERS
	.set_page_dirty = __set_page_dirty_nobuffers,
	#endif
	};

	const struct file_operations zpl_file_operations = {
	.open = zpl_open,
	.release = zpl_release,
	.llseek = zpl_llseek,
	#ifdef HAVE_VFS_RW_ITERATE
	#ifdef HAVE_NEW_SYNC_READ
	.read = new_sync_read,
	.write = new_sync_write,
	#endif
	.read_iter = zpl_iter_read,
	.write_iter = zpl_iter_write,
	#ifdef HAVE_VFS_IOV_ITER
	.splice_read = generic_file_splice_read,
	.splice_write = iter_file_splice_write,
	#endif
	#else
	.read = do_sync_read,
	.write = do_sync_write,
	.aio_read = zpl_aio_read,
	.aio_write = zpl_aio_write,
	#endif
	.mmap = zpl_mmap,
	.fsync = zpl_fsync,
	#ifdef HAVE_FILE_AIO_FSYNC
	.aio_fsync = zpl_aio_fsync,
	#endif
	.fallocate = zpl_fallocate,
	.unlocked_ioctl = zpl_ioctl,
	#ifdef CONFIG_COMPAT
	.compat_ioctl = zpl_compat_ioctl,
	#endif
	};

	const struct file_operations zpl_dir_file_operations = {
	.llseek = generic_file_llseek,
	.read = generic_read_dir,
	#if defined(HAVE_VFS_ITERATE_SHARED)
	.iterate_shared = zpl_iterate,
	#elif defined(HAVE_VFS_ITERATE)
	.iterate = zpl_iterate,
	#else
	.readdir = zpl_readdir,
	#endif
	.fsync = zpl_fsync,
	.unlocked_ioctl = zpl_ioctl,
	#ifdef CONFIG_COMPAT
	.compat_ioctl = zpl_compat_ioctl,
	#endif
	};

	/* BEGIN CSTYLED */
	module_param(zfs_fallocate_reserve_percent, uint, 0644);
	MODULE_PARM_DESC(zfs_fallocate_reserve_percent,
	"Percentage of length to use for the available capacity check");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c b/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c
	index c17423426319..44caadd587f7 100644
	--- a/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c
	+++ b/sys/contrib/openzfs/module/os/linux/zfs/zvol_os.c
	@@ -1,1174 +1,1231 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2012, 2020 by Delphix. All rights reserved.
	*/

	#include <sys/dataset_kstats.h>
	#include <sys/dbuf.h>
	#include <sys/dmu_traverse.h>
	#include <sys/dsl_dataset.h>
	#include <sys/dsl_prop.h>
	#include <sys/dsl_dir.h>
	#include <sys/zap.h>
	#include <sys/zfeature.h>
	#include <sys/zil_impl.h>
	#include <sys/dmu_tx.h>
	#include <sys/zio.h>
	#include <sys/zfs_rlock.h>
	#include <sys/spa_impl.h>
	#include <sys/zvol.h>
	#include <sys/zvol_impl.h>

	#include <linux/blkdev_compat.h>
	#include <linux/task_io_accounting_ops.h>

	unsigned int zvol_major = ZVOL_MAJOR;
	unsigned int zvol_request_sync = 0;
	unsigned int zvol_prefetch_bytes = (128 * 1024);
	unsigned long zvol_max_discard_blocks = 16384;
	unsigned int zvol_threads = 32;
	+unsigned int zvol_open_timeout_ms = 1000;

	struct zvol_state_os {
	struct gendisk zvo_disk; / generic disk */
	struct request_queue zvo_queue; / request queue */
	dev_t zvo_dev; /* device id */
	};

	taskq_t *zvol_taskq;
	static struct ida zvol_ida;

	typedef struct zv_request_stack {
	zvol_state_t *zv;
	struct bio *bio;
	} zv_request_t;

	typedef struct zv_request_task {
	zv_request_t zvr;
	taskq_ent_t ent;
	} zv_request_task_t;

	static zv_request_task_t *
	zv_request_task_create(zv_request_t zvr)
	{
	zv_request_task_t *task;
	task = kmem_alloc(sizeof (zv_request_task_t), KM_SLEEP);
	taskq_init_ent(&task->ent);
	task->zvr = zvr;
	return (task);
	}

	static void
	zv_request_task_free(zv_request_task_t *task)
	{
	kmem_free(task, sizeof (*task));
	}

	/*
	* Given a path, return TRUE if path is a ZVOL.
	*/
	static boolean_t
	zvol_is_zvol_impl(const char *path)
	{
	dev_t dev = 0;

	if (vdev_lookup_bdev(path, &dev) != 0)
	return (B_FALSE);

	if (MAJOR(dev) == zvol_major)
	return (B_TRUE);

	return (B_FALSE);
	}

	static void
	zvol_write(zv_request_t *zvr)
	{
	struct bio *bio = zvr->bio;
	int error = 0;
	zfs_uio_t uio;

	zfs_uio_bvec_init(&uio, bio);

	zvol_state_t *zv = zvr->zv;
	ASSERT3P(zv, !=, NULL);
	ASSERT3U(zv->zv_open_count, >, 0);
	ASSERT3P(zv->zv_zilog, !=, NULL);

	/* bio marked as FLUSH need to flush before write */
	if (bio_is_flush(bio))
	zil_commit(zv->zv_zilog, ZVOL_OBJ);

	/* Some requests are just for flush and nothing else. */
	if (uio.uio_resid == 0) {
	rw_exit(&zv->zv_suspend_lock);
	BIO_END_IO(bio, 0);
	return;
	}

	struct request_queue *q = zv->zv_zso->zvo_queue;
	struct gendisk *disk = zv->zv_zso->zvo_disk;
	ssize_t start_resid = uio.uio_resid;
	unsigned long start_time;

	boolean_t acct = blk_queue_io_stat(q);
	if (acct)
	start_time = blk_generic_start_io_acct(q, disk, WRITE, bio);

	boolean_t sync =
	bio_is_fua(bio) \|\| zv->zv_objset->os_sync == ZFS_SYNC_ALWAYS;

	zfs_locked_range_t *lr = zfs_rangelock_enter(&zv->zv_rangelock,
	uio.uio_loffset, uio.uio_resid, RL_WRITER);

	uint64_t volsize = zv->zv_volsize;
	while (uio.uio_resid > 0 && uio.uio_loffset < volsize) {
	uint64_t bytes = MIN(uio.uio_resid, DMU_MAX_ACCESS >> 1);
	uint64_t off = uio.uio_loffset;
	dmu_tx_t *tx = dmu_tx_create(zv->zv_objset);

	if (bytes > volsize - off) /* don't write past the end */
	bytes = volsize - off;

	dmu_tx_hold_write_by_dnode(tx, zv->zv_dn, off, bytes);

	/* This will only fail for ENOSPC */
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	dmu_tx_abort(tx);
	break;
	}
	error = dmu_write_uio_dnode(zv->zv_dn, &uio, bytes, tx);
	if (error == 0) {
	zvol_log_write(zv, tx, off, bytes, sync);
	}
	dmu_tx_commit(tx);

	if (error)
	break;
	}
	zfs_rangelock_exit(lr);

	int64_t nwritten = start_resid - uio.uio_resid;
	dataset_kstats_update_write_kstats(&zv->zv_kstat, nwritten);
	task_io_account_write(nwritten);

	if (sync)
	zil_commit(zv->zv_zilog, ZVOL_OBJ);

	rw_exit(&zv->zv_suspend_lock);

	if (acct)
	blk_generic_end_io_acct(q, disk, WRITE, bio, start_time);

	BIO_END_IO(bio, -error);
	}

	static void
	zvol_write_task(void *arg)
	{
	zv_request_task_t *task = arg;
	zvol_write(&task->zvr);
	zv_request_task_free(task);
	}

	static void
	zvol_discard(zv_request_t *zvr)
	{
	struct bio *bio = zvr->bio;
	zvol_state_t *zv = zvr->zv;
	uint64_t start = BIO_BI_SECTOR(bio) << 9;
	uint64_t size = BIO_BI_SIZE(bio);
	uint64_t end = start + size;
	boolean_t sync;
	int error = 0;
	dmu_tx_t *tx;

	ASSERT3P(zv, !=, NULL);
	ASSERT3U(zv->zv_open_count, >, 0);
	ASSERT3P(zv->zv_zilog, !=, NULL);

	struct request_queue *q = zv->zv_zso->zvo_queue;
	struct gendisk *disk = zv->zv_zso->zvo_disk;
	unsigned long start_time;

	boolean_t acct = blk_queue_io_stat(q);
	if (acct)
	start_time = blk_generic_start_io_acct(q, disk, WRITE, bio);

	sync = bio_is_fua(bio) \|\| zv->zv_objset->os_sync == ZFS_SYNC_ALWAYS;

	if (end > zv->zv_volsize) {
	error = SET_ERROR(EIO);
	goto unlock;
	}

	/*
	* Align the request to volume block boundaries when a secure erase is
	* not required. This will prevent dnode_free_range() from zeroing out
	* the unaligned parts which is slow (read-modify-write) and useless
	* since we are not freeing any space by doing so.
	*/
	if (!bio_is_secure_erase(bio)) {
	start = P2ROUNDUP(start, zv->zv_volblocksize);
	end = P2ALIGN(end, zv->zv_volblocksize);
	size = end - start;
	}

	if (start >= end)
	goto unlock;

	zfs_locked_range_t *lr = zfs_rangelock_enter(&zv->zv_rangelock,
	start, size, RL_WRITER);

	tx = dmu_tx_create(zv->zv_objset);
	dmu_tx_mark_netfree(tx);
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error != 0) {
	dmu_tx_abort(tx);
	} else {
	zvol_log_truncate(zv, tx, start, size, B_TRUE);
	dmu_tx_commit(tx);
	error = dmu_free_long_range(zv->zv_objset,
	ZVOL_OBJ, start, size);
	}
	zfs_rangelock_exit(lr);

	if (error == 0 && sync)
	zil_commit(zv->zv_zilog, ZVOL_OBJ);

	unlock:
	rw_exit(&zv->zv_suspend_lock);

	if (acct)
	blk_generic_end_io_acct(q, disk, WRITE, bio, start_time);

	BIO_END_IO(bio, -error);
	}

	static void
	zvol_discard_task(void *arg)
	{
	zv_request_task_t *task = arg;
	zvol_discard(&task->zvr);
	zv_request_task_free(task);
	}

	static void
	zvol_read(zv_request_t *zvr)
	{
	struct bio *bio = zvr->bio;
	int error = 0;
	zfs_uio_t uio;

	zfs_uio_bvec_init(&uio, bio);

	zvol_state_t *zv = zvr->zv;
	ASSERT3P(zv, !=, NULL);
	ASSERT3U(zv->zv_open_count, >, 0);

	struct request_queue *q = zv->zv_zso->zvo_queue;
	struct gendisk *disk = zv->zv_zso->zvo_disk;
	ssize_t start_resid = uio.uio_resid;
	unsigned long start_time;

	boolean_t acct = blk_queue_io_stat(q);
	if (acct)
	start_time = blk_generic_start_io_acct(q, disk, READ, bio);

	zfs_locked_range_t *lr = zfs_rangelock_enter(&zv->zv_rangelock,
	uio.uio_loffset, uio.uio_resid, RL_READER);

	uint64_t volsize = zv->zv_volsize;
	while (uio.uio_resid > 0 && uio.uio_loffset < volsize) {
	uint64_t bytes = MIN(uio.uio_resid, DMU_MAX_ACCESS >> 1);

	/* don't read past the end */
	if (bytes > volsize - uio.uio_loffset)
	bytes = volsize - uio.uio_loffset;

	error = dmu_read_uio_dnode(zv->zv_dn, &uio, bytes);
	if (error) {
	/* convert checksum errors into IO errors */
	if (error == ECKSUM)
	error = SET_ERROR(EIO);
	break;
	}
	}
	zfs_rangelock_exit(lr);

	int64_t nread = start_resid - uio.uio_resid;
	dataset_kstats_update_read_kstats(&zv->zv_kstat, nread);
	task_io_account_read(nread);

	rw_exit(&zv->zv_suspend_lock);

	if (acct)
	blk_generic_end_io_acct(q, disk, READ, bio, start_time);

	BIO_END_IO(bio, -error);
	}

	static void
	zvol_read_task(void *arg)
	{
	zv_request_task_t *task = arg;
	zvol_read(&task->zvr);
	zv_request_task_free(task);
	}

	#ifdef HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS
	+#ifdef HAVE_BDEV_SUBMIT_BIO_RETURNS_VOID
	+static void
	+zvol_submit_bio(struct bio *bio)
	+#else
	static blk_qc_t
	zvol_submit_bio(struct bio *bio)
	+#endif
	#else
	static MAKE_REQUEST_FN_RET
	zvol_request(struct request_queue q, struct bio bio)
	#endif
	{
	#ifdef HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS
	#if defined(HAVE_BIO_BDEV_DISK)
	struct request_queue *q = bio->bi_bdev->bd_disk->queue;
	#else
	struct request_queue *q = bio->bi_disk->queue;
	#endif
	#endif
	zvol_state_t *zv = q->queuedata;
	fstrans_cookie_t cookie = spl_fstrans_mark();
	uint64_t offset = BIO_BI_SECTOR(bio) << 9;
	uint64_t size = BIO_BI_SIZE(bio);
	int rw = bio_data_dir(bio);

	if (bio_has_data(bio) && offset + size > zv->zv_volsize) {
	printk(KERN_INFO
	"%s: bad access: offset=%llu, size=%lu\n",
	zv->zv_zso->zvo_disk->disk_name,
	(long long unsigned)offset,
	(long unsigned)size);

	BIO_END_IO(bio, -SET_ERROR(EIO));
	goto out;
	}

	zv_request_t zvr = {
	.zv = zv,
	.bio = bio,
	};
	zv_request_task_t *task;

	if (rw == WRITE) {
	if (unlikely(zv->zv_flags & ZVOL_RDONLY)) {
	BIO_END_IO(bio, -SET_ERROR(EROFS));
	goto out;
	}

	/*
	* Prevents the zvol from being suspended, or the ZIL being
	* concurrently opened. Will be released after the i/o
	* completes.
	*/
	rw_enter(&zv->zv_suspend_lock, RW_READER);

	/*
	* Open a ZIL if this is the first time we have written to this
	* zvol. We protect zv->zv_zilog with zv_suspend_lock rather
	* than zv_state_lock so that we don't need to acquire an
	* additional lock in this path.
	*/
	if (zv->zv_zilog == NULL) {
	rw_exit(&zv->zv_suspend_lock);
	rw_enter(&zv->zv_suspend_lock, RW_WRITER);
	if (zv->zv_zilog == NULL) {
	zv->zv_zilog = zil_open(zv->zv_objset,
	zvol_get_data);
	zv->zv_flags \|= ZVOL_WRITTEN_TO;
	/* replay / destroy done in zvol_create_minor */
	VERIFY0((zv->zv_zilog->zl_header->zh_flags &
	ZIL_REPLAY_NEEDED));
	}
	rw_downgrade(&zv->zv_suspend_lock);
	}

	/*
	* We don't want this thread to be blocked waiting for i/o to
	* complete, so we instead wait from a taskq callback. The
	* i/o may be a ZIL write (via zil_commit()), or a read of an
	* indirect block, or a read of a data block (if this is a
	* partial-block write). We will indicate that the i/o is
	* complete by calling BIO_END_IO() from the taskq callback.
	*
	* This design allows the calling thread to continue and
	* initiate more concurrent operations by calling
	* zvol_request() again. There are typically only a small
	* number of threads available to call zvol_request() (e.g.
	* one per iSCSI target), so keeping the latency of
	* zvol_request() low is important for performance.
	*
	* The zvol_request_sync module parameter allows this
	* behavior to be altered, for performance evaluation
	* purposes. If the callback blocks, setting
	* zvol_request_sync=1 will result in much worse performance.
	*
	* We can have up to zvol_threads concurrent i/o's being
	* processed for all zvols on the system. This is typically
	* a vast improvement over the zvol_request_sync=1 behavior
	* of one i/o at a time per zvol. However, an even better
	* design would be for zvol_request() to initiate the zio
	* directly, and then be notified by the zio_done callback,
	* which would call BIO_END_IO(). Unfortunately, the DMU/ZIL
	* interfaces lack this functionality (they block waiting for
	* the i/o to complete).
	*/
	if (bio_is_discard(bio) \|\| bio_is_secure_erase(bio)) {
	if (zvol_request_sync) {
	zvol_discard(&zvr);
	} else {
	task = zv_request_task_create(zvr);
	taskq_dispatch_ent(zvol_taskq,
	zvol_discard_task, task, 0, &task->ent);
	}
	} else {
	if (zvol_request_sync) {
	zvol_write(&zvr);
	} else {
	task = zv_request_task_create(zvr);
	taskq_dispatch_ent(zvol_taskq,
	zvol_write_task, task, 0, &task->ent);
	}
	}
	} else {
	/*
	* The SCST driver, and possibly others, may issue READ I/Os
	* with a length of zero bytes. These empty I/Os contain no
	* data and require no additional handling.
	*/
	if (size == 0) {
	BIO_END_IO(bio, 0);
	goto out;
	}

	rw_enter(&zv->zv_suspend_lock, RW_READER);

	/* See comment in WRITE case above. */
	if (zvol_request_sync) {
	zvol_read(&zvr);
	} else {
	task = zv_request_task_create(zvr);
	taskq_dispatch_ent(zvol_taskq,
	zvol_read_task, task, 0, &task->ent);
	}
	}

	out:
	spl_fstrans_unmark(cookie);
	-#if defined(HAVE_MAKE_REQUEST_FN_RET_QC) \|\| \
	- defined(HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS)
	+#if (defined(HAVE_MAKE_REQUEST_FN_RET_QC) \|\| \
	+ defined(HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS)) && \
	+ !defined(HAVE_BDEV_SUBMIT_BIO_RETURNS_VOID)
	return (BLK_QC_T_NONE);
	#endif
	}

	static int
	zvol_open(struct block_device *bdev, fmode_t flag)
	{
	zvol_state_t *zv;
	int error = 0;
	boolean_t drop_suspend = B_TRUE;
	+ boolean_t drop_namespace = B_FALSE;
	+#ifndef HAVE_BLKDEV_GET_ERESTARTSYS
	+ hrtime_t timeout = MSEC2NSEC(zvol_open_timeout_ms);
	+ hrtime_t start = gethrtime();

	+retry:
	+#endif
	rw_enter(&zvol_state_lock, RW_READER);
	/*
	* Obtain a copy of private_data under the zvol_state_lock to make
	* sure that either the result of zvol free code path setting
	* bdev->bd_disk->private_data to NULL is observed, or zvol_free()
	* is not called on this zv because of the positive zv_open_count.
	*/
	zv = bdev->bd_disk->private_data;
	if (zv == NULL) {
	rw_exit(&zvol_state_lock);
	return (SET_ERROR(-ENXIO));
	}

	+ if (zv->zv_open_count == 0 && !mutex_owned(&spa_namespace_lock)) {
	+ /*
	+ * In all other call paths the spa_namespace_lock is taken
	+ * before the bdev->bd_mutex lock. However, on open(2)
	+ * the __blkdev_get() function calls fops->open() with the
	+ * bdev->bd_mutex lock held. This can result in a deadlock
	+ * when zvols from one pool are used as vdevs in another.
	+ *
	+ * To prevent a lock inversion deadlock we preemptively
	+ * take the spa_namespace_lock. Normally the lock will not
	+ * be contended and this is safe because spa_open_common()
	+ * handles the case where the caller already holds the
	+ * spa_namespace_lock.
	+ *
	+ * When the lock cannot be aquired after multiple retries
	+ * this must be the vdev on zvol deadlock case and we have
	+ * no choice but to return an error. For 5.12 and older
	+ * kernels returning -ERESTARTSYS will result in the
	+ * bdev->bd_mutex being dropped, then reacquired, and
	+ * fops->open() being called again. This process can be
	+ * repeated safely until both locks are acquired. For 5.13
	+ * and newer the -ERESTARTSYS retry logic was removed from
	+ * the kernel so the only option is to return the error for
	+ * the caller to handle it.
	+ */
	+ if (!mutex_tryenter(&spa_namespace_lock)) {
	+ rw_exit(&zvol_state_lock);
	+
	+#ifdef HAVE_BLKDEV_GET_ERESTARTSYS
	+ schedule();
	+ return (SET_ERROR(-ERESTARTSYS));
	+#else
	+ if ((gethrtime() - start) > timeout)
	+ return (SET_ERROR(-ERESTARTSYS));
	+
	+ schedule_timeout(MSEC_TO_TICK(10));
	+ goto retry;
	+#endif
	+ } else {
	+ drop_namespace = B_TRUE;
	+ }
	+ }
	+
	mutex_enter(&zv->zv_state_lock);
	/*
	* make sure zvol is not suspended during first open
	* (hold zv_suspend_lock) and respect proper lock acquisition
	* ordering - zv_suspend_lock before zv_state_lock
	*/
	if (zv->zv_open_count == 0) {
	if (!rw_tryenter(&zv->zv_suspend_lock, RW_READER)) {
	mutex_exit(&zv->zv_state_lock);
	rw_enter(&zv->zv_suspend_lock, RW_READER);
	mutex_enter(&zv->zv_state_lock);
	/* check to see if zv_suspend_lock is needed */
	if (zv->zv_open_count != 0) {
	rw_exit(&zv->zv_suspend_lock);
	drop_suspend = B_FALSE;
	}
	}
	} else {
	drop_suspend = B_FALSE;
	}
	rw_exit(&zvol_state_lock);

	ASSERT(MUTEX_HELD(&zv->zv_state_lock));

	if (zv->zv_open_count == 0) {
	ASSERT(RW_READ_HELD(&zv->zv_suspend_lock));
	error = -zvol_first_open(zv, !(flag & FMODE_WRITE));
	if (error)
	goto out_mutex;
	}

	if ((flag & FMODE_WRITE) && (zv->zv_flags & ZVOL_RDONLY)) {
	error = -EROFS;
	goto out_open_count;
	}

	zv->zv_open_count++;

	mutex_exit(&zv->zv_state_lock);
	+ if (drop_namespace)
	+ mutex_exit(&spa_namespace_lock);
	if (drop_suspend)
	rw_exit(&zv->zv_suspend_lock);

	zfs_check_media_change(bdev);

	return (0);

	out_open_count:
	if (zv->zv_open_count == 0)
	zvol_last_close(zv);

	out_mutex:
	mutex_exit(&zv->zv_state_lock);
	+ if (drop_namespace)
	+ mutex_exit(&spa_namespace_lock);
	if (drop_suspend)
	rw_exit(&zv->zv_suspend_lock);
	- if (error == -EINTR) {
	- error = -ERESTARTSYS;
	- schedule();
	- }
	+
	return (SET_ERROR(error));
	}

	static void
	zvol_release(struct gendisk *disk, fmode_t mode)
	{
	zvol_state_t *zv;
	boolean_t drop_suspend = B_TRUE;

	rw_enter(&zvol_state_lock, RW_READER);
	zv = disk->private_data;

	mutex_enter(&zv->zv_state_lock);
	ASSERT3U(zv->zv_open_count, >, 0);
	/*
	* make sure zvol is not suspended during last close
	* (hold zv_suspend_lock) and respect proper lock acquisition
	* ordering - zv_suspend_lock before zv_state_lock
	*/
	if (zv->zv_open_count == 1) {
	if (!rw_tryenter(&zv->zv_suspend_lock, RW_READER)) {
	mutex_exit(&zv->zv_state_lock);
	rw_enter(&zv->zv_suspend_lock, RW_READER);
	mutex_enter(&zv->zv_state_lock);
	/* check to see if zv_suspend_lock is needed */
	if (zv->zv_open_count != 1) {
	rw_exit(&zv->zv_suspend_lock);
	drop_suspend = B_FALSE;
	}
	}
	} else {
	drop_suspend = B_FALSE;
	}
	rw_exit(&zvol_state_lock);

	ASSERT(MUTEX_HELD(&zv->zv_state_lock));

	zv->zv_open_count--;
	if (zv->zv_open_count == 0) {
	ASSERT(RW_READ_HELD(&zv->zv_suspend_lock));
	zvol_last_close(zv);
	}

	mutex_exit(&zv->zv_state_lock);

	if (drop_suspend)
	rw_exit(&zv->zv_suspend_lock);
	}

	static int
	zvol_ioctl(struct block_device *bdev, fmode_t mode,
	unsigned int cmd, unsigned long arg)
	{
	zvol_state_t *zv = bdev->bd_disk->private_data;
	int error = 0;

	ASSERT3U(zv->zv_open_count, >, 0);

	switch (cmd) {
	case BLKFLSBUF:
	fsync_bdev(bdev);
	invalidate_bdev(bdev);
	rw_enter(&zv->zv_suspend_lock, RW_READER);

	if (!(zv->zv_flags & ZVOL_RDONLY))
	txg_wait_synced(dmu_objset_pool(zv->zv_objset), 0);

	rw_exit(&zv->zv_suspend_lock);
	break;

	case BLKZNAME:
	mutex_enter(&zv->zv_state_lock);
	error = copy_to_user((void *)arg, zv->zv_name, MAXNAMELEN);
	mutex_exit(&zv->zv_state_lock);
	break;

	default:
	error = -ENOTTY;
	break;
	}

	return (SET_ERROR(error));
	}

	#ifdef CONFIG_COMPAT
	static int
	zvol_compat_ioctl(struct block_device *bdev, fmode_t mode,
	unsigned cmd, unsigned long arg)
	{
	return (zvol_ioctl(bdev, mode, cmd, arg));
	}
	#else
	#define zvol_compat_ioctl NULL
	#endif

	static unsigned int
	zvol_check_events(struct gendisk *disk, unsigned int clearing)
	{
	unsigned int mask = 0;

	rw_enter(&zvol_state_lock, RW_READER);

	zvol_state_t *zv = disk->private_data;
	if (zv != NULL) {
	mutex_enter(&zv->zv_state_lock);
	mask = zv->zv_changed ? DISK_EVENT_MEDIA_CHANGE : 0;
	zv->zv_changed = 0;
	mutex_exit(&zv->zv_state_lock);
	}

	rw_exit(&zvol_state_lock);

	return (mask);
	}

	static int
	zvol_revalidate_disk(struct gendisk *disk)
	{
	rw_enter(&zvol_state_lock, RW_READER);

	zvol_state_t *zv = disk->private_data;
	if (zv != NULL) {
	mutex_enter(&zv->zv_state_lock);
	set_capacity(zv->zv_zso->zvo_disk,
	zv->zv_volsize >> SECTOR_BITS);
	mutex_exit(&zv->zv_state_lock);
	}

	rw_exit(&zvol_state_lock);

	return (0);
	}

	static int
	zvol_update_volsize(zvol_state_t *zv, uint64_t volsize)
	{
	struct gendisk *disk = zv->zv_zso->zvo_disk;

	#if defined(HAVE_REVALIDATE_DISK_SIZE)
	revalidate_disk_size(disk, zvol_revalidate_disk(disk) == 0);
	#elif defined(HAVE_REVALIDATE_DISK)
	revalidate_disk(disk);
	#else
	zvol_revalidate_disk(disk);
	#endif
	return (0);
	}

	static void
	zvol_clear_private(zvol_state_t *zv)
	{
	/*
	* Cleared while holding zvol_state_lock as a writer
	* which will prevent zvol_open() from opening it.
	*/
	zv->zv_zso->zvo_disk->private_data = NULL;
	}

	/*
	* Provide a simple virtual geometry for legacy compatibility. For devices
	* smaller than 1 MiB a small head and sector count is used to allow very
	* tiny devices. For devices over 1 Mib a standard head and sector count
	* is used to keep the cylinders count reasonable.
	*/
	static int
	zvol_getgeo(struct block_device bdev, struct hd_geometry geo)
	{
	zvol_state_t *zv = bdev->bd_disk->private_data;
	sector_t sectors;

	ASSERT3U(zv->zv_open_count, >, 0);

	sectors = get_capacity(zv->zv_zso->zvo_disk);

	if (sectors > 2048) {
	geo->heads = 16;
	geo->sectors = 63;
	} else {
	geo->heads = 2;
	geo->sectors = 4;
	}

	geo->start = 0;
	geo->cylinders = sectors / (geo->heads * geo->sectors);

	return (0);
	}

	static struct block_device_operations zvol_ops = {
	.open = zvol_open,
	.release = zvol_release,
	.ioctl = zvol_ioctl,
	.compat_ioctl = zvol_compat_ioctl,
	.check_events = zvol_check_events,
	#ifdef HAVE_BLOCK_DEVICE_OPERATIONS_REVALIDATE_DISK
	.revalidate_disk = zvol_revalidate_disk,
	#endif
	.getgeo = zvol_getgeo,
	.owner = THIS_MODULE,
	#ifdef HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS
	.submit_bio = zvol_submit_bio,
	#endif
	};

	/*
	* Allocate memory for a new zvol_state_t and setup the required
	* request queue and generic disk structures for the block device.
	*/
	static zvol_state_t *
	zvol_alloc(dev_t dev, const char *name)
	{
	zvol_state_t *zv;
	struct zvol_state_os *zso;
	uint64_t volmode;

	if (dsl_prop_get_integer(name, "volmode", &volmode, NULL) != 0)
	return (NULL);

	if (volmode == ZFS_VOLMODE_DEFAULT)
	volmode = zvol_volmode;

	if (volmode == ZFS_VOLMODE_NONE)
	return (NULL);

	zv = kmem_zalloc(sizeof (zvol_state_t), KM_SLEEP);
	zso = kmem_zalloc(sizeof (struct zvol_state_os), KM_SLEEP);
	zv->zv_zso = zso;
	zv->zv_volmode = volmode;

	list_link_init(&zv->zv_next);
	mutex_init(&zv->zv_state_lock, NULL, MUTEX_DEFAULT, NULL);

	#ifdef HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS
	#ifdef HAVE_BLK_ALLOC_DISK
	zso->zvo_disk = blk_alloc_disk(NUMA_NO_NODE);
	if (zso->zvo_disk == NULL)
	goto out_kmem;

	zso->zvo_disk->minors = ZVOL_MINORS;
	zso->zvo_queue = zso->zvo_disk->queue;
	#else
	zso->zvo_queue = blk_alloc_queue(NUMA_NO_NODE);
	if (zso->zvo_queue == NULL)
	goto out_kmem;

	zso->zvo_disk = alloc_disk(ZVOL_MINORS);
	if (zso->zvo_disk == NULL) {
	blk_cleanup_queue(zso->zvo_queue);
	goto out_kmem;
	}

	zso->zvo_disk->queue = zso->zvo_queue;
	#endif /* HAVE_BLK_ALLOC_DISK */
	#else
	zso->zvo_queue = blk_generic_alloc_queue(zvol_request, NUMA_NO_NODE);
	if (zso->zvo_queue == NULL)
	goto out_kmem;

	zso->zvo_disk = alloc_disk(ZVOL_MINORS);
	if (zso->zvo_disk == NULL) {
	blk_cleanup_queue(zso->zvo_queue);
	goto out_kmem;
	}

	zso->zvo_disk->queue = zso->zvo_queue;
	#endif /* HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS */

	blk_queue_set_write_cache(zso->zvo_queue, B_TRUE, B_TRUE);

	/* Limit read-ahead to a single page to prevent over-prefetching. */
	blk_queue_set_read_ahead(zso->zvo_queue, 1);

	/* Disable write merging in favor of the ZIO pipeline. */
	blk_queue_flag_set(QUEUE_FLAG_NOMERGES, zso->zvo_queue);

	/* Enable /proc/diskstats */
	blk_queue_flag_set(QUEUE_FLAG_IO_STAT, zso->zvo_queue);

	zso->zvo_queue->queuedata = zv;
	zso->zvo_dev = dev;
	zv->zv_open_count = 0;
	strlcpy(zv->zv_name, name, MAXNAMELEN);

	zfs_rangelock_init(&zv->zv_rangelock, NULL, NULL);
	rw_init(&zv->zv_suspend_lock, NULL, RW_DEFAULT, NULL);

	zso->zvo_disk->major = zvol_major;
	zso->zvo_disk->events = DISK_EVENT_MEDIA_CHANGE;

	if (volmode == ZFS_VOLMODE_DEV) {
	/*
	* ZFS_VOLMODE_DEV disable partitioning on ZVOL devices: set
	* gendisk->minors = 1 as noted in include/linux/genhd.h.
	* Also disable extended partition numbers (GENHD_FL_EXT_DEVT)
	* and suppresses partition scanning (GENHD_FL_NO_PART_SCAN)
	* setting gendisk->flags accordingly.
	*/
	zso->zvo_disk->minors = 1;
	#if defined(GENHD_FL_EXT_DEVT)
	zso->zvo_disk->flags &= ~GENHD_FL_EXT_DEVT;
	#endif
	#if defined(GENHD_FL_NO_PART_SCAN)
	zso->zvo_disk->flags \|= GENHD_FL_NO_PART_SCAN;
	#endif
	}
	zso->zvo_disk->first_minor = (dev & MINORMASK);
	zso->zvo_disk->fops = &zvol_ops;
	zso->zvo_disk->private_data = zv;
	snprintf(zso->zvo_disk->disk_name, DISK_NAME_LEN, "%s%d",
	ZVOL_DEV_NAME, (dev & MINORMASK));

	return (zv);

	out_kmem:
	kmem_free(zso, sizeof (struct zvol_state_os));
	kmem_free(zv, sizeof (zvol_state_t));
	return (NULL);
	}

	/*
	* Cleanup then free a zvol_state_t which was created by zvol_alloc().
	* At this time, the structure is not opened by anyone, is taken off
	* the zvol_state_list, and has its private data set to NULL.
	* The zvol_state_lock is dropped.
	*
	* This function may take many milliseconds to complete (e.g. we've seen
	* it take over 256ms), due to the calls to "blk_cleanup_queue" and
	* "del_gendisk". Thus, consumers need to be careful to account for this
	* latency when calling this function.
	*/
	static void
	zvol_free(zvol_state_t *zv)
	{

	ASSERT(!RW_LOCK_HELD(&zv->zv_suspend_lock));
	ASSERT(!MUTEX_HELD(&zv->zv_state_lock));
	ASSERT0(zv->zv_open_count);
	ASSERT3P(zv->zv_zso->zvo_disk->private_data, ==, NULL);

	rw_destroy(&zv->zv_suspend_lock);
	zfs_rangelock_fini(&zv->zv_rangelock);

	del_gendisk(zv->zv_zso->zvo_disk);
	#if defined(HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS) && \
	defined(HAVE_BLK_ALLOC_DISK)
	blk_cleanup_disk(zv->zv_zso->zvo_disk);
	#else
	blk_cleanup_queue(zv->zv_zso->zvo_queue);
	put_disk(zv->zv_zso->zvo_disk);
	#endif

	ida_simple_remove(&zvol_ida,
	MINOR(zv->zv_zso->zvo_dev) >> ZVOL_MINOR_BITS);

	mutex_destroy(&zv->zv_state_lock);
	dataset_kstats_destroy(&zv->zv_kstat);

	kmem_free(zv->zv_zso, sizeof (struct zvol_state_os));
	kmem_free(zv, sizeof (zvol_state_t));
	}

	void
	zvol_wait_close(zvol_state_t *zv)
	{
	}

	/*
	* Create a block device minor node and setup the linkage between it
	* and the specified volume. Once this function returns the block
	* device is live and ready for use.
	*/
	static int
	zvol_os_create_minor(const char *name)
	{
	zvol_state_t *zv;
	objset_t *os;
	dmu_object_info_t *doi;
	uint64_t volsize;
	uint64_t len;
	unsigned minor = 0;
	int error = 0;
	int idx;
	uint64_t hash = zvol_name_hash(name);

	if (zvol_inhibit_dev)
	return (0);

	idx = ida_simple_get(&zvol_ida, 0, 0, kmem_flags_convert(KM_SLEEP));
	if (idx < 0)
	return (SET_ERROR(-idx));
	minor = idx << ZVOL_MINOR_BITS;

	zv = zvol_find_by_name_hash(name, hash, RW_NONE);
	if (zv) {
	ASSERT(MUTEX_HELD(&zv->zv_state_lock));
	mutex_exit(&zv->zv_state_lock);
	ida_simple_remove(&zvol_ida, idx);
	return (SET_ERROR(EEXIST));
	}

	doi = kmem_alloc(sizeof (dmu_object_info_t), KM_SLEEP);

	error = dmu_objset_own(name, DMU_OST_ZVOL, B_TRUE, B_TRUE, FTAG, &os);
	if (error)
	goto out_doi;

	error = dmu_object_info(os, ZVOL_OBJ, doi);
	if (error)
	goto out_dmu_objset_disown;

	error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize);
	if (error)
	goto out_dmu_objset_disown;

	zv = zvol_alloc(MKDEV(zvol_major, minor), name);
	if (zv == NULL) {
	error = SET_ERROR(EAGAIN);
	goto out_dmu_objset_disown;
	}
	zv->zv_hash = hash;

	if (dmu_objset_is_snapshot(os))
	zv->zv_flags \|= ZVOL_RDONLY;

	zv->zv_volblocksize = doi->doi_data_block_size;
	zv->zv_volsize = volsize;
	zv->zv_objset = os;

	set_capacity(zv->zv_zso->zvo_disk, zv->zv_volsize >> 9);

	blk_queue_max_hw_sectors(zv->zv_zso->zvo_queue,
	(DMU_MAX_ACCESS / 4) >> 9);
	blk_queue_max_segments(zv->zv_zso->zvo_queue, UINT16_MAX);
	blk_queue_max_segment_size(zv->zv_zso->zvo_queue, UINT_MAX);
	blk_queue_physical_block_size(zv->zv_zso->zvo_queue,
	zv->zv_volblocksize);
	blk_queue_io_opt(zv->zv_zso->zvo_queue, zv->zv_volblocksize);
	blk_queue_max_discard_sectors(zv->zv_zso->zvo_queue,
	(zvol_max_discard_blocks * zv->zv_volblocksize) >> 9);
	blk_queue_discard_granularity(zv->zv_zso->zvo_queue,
	zv->zv_volblocksize);
	blk_queue_flag_set(QUEUE_FLAG_DISCARD, zv->zv_zso->zvo_queue);
	#ifdef QUEUE_FLAG_NONROT
	blk_queue_flag_set(QUEUE_FLAG_NONROT, zv->zv_zso->zvo_queue);
	#endif
	#ifdef QUEUE_FLAG_ADD_RANDOM
	blk_queue_flag_clear(QUEUE_FLAG_ADD_RANDOM, zv->zv_zso->zvo_queue);
	#endif
	/* This flag was introduced in kernel version 4.12. */
	#ifdef QUEUE_FLAG_SCSI_PASSTHROUGH
	blk_queue_flag_set(QUEUE_FLAG_SCSI_PASSTHROUGH, zv->zv_zso->zvo_queue);
	#endif

	ASSERT3P(zv->zv_zilog, ==, NULL);
	zv->zv_zilog = zil_open(os, zvol_get_data);
	if (spa_writeable(dmu_objset_spa(os))) {
	if (zil_replay_disable)
	zil_destroy(zv->zv_zilog, B_FALSE);
	else
	zil_replay(os, zv, zvol_replay_vector);
	}
	zil_close(zv->zv_zilog);
	zv->zv_zilog = NULL;
	ASSERT3P(zv->zv_kstat.dk_kstats, ==, NULL);
	dataset_kstats_create(&zv->zv_kstat, zv->zv_objset);

	/*
	* When udev detects the addition of the device it will immediately
	* invoke blkid(8) to determine the type of content on the device.
	* Prefetching the blocks commonly scanned by blkid(8) will speed
	* up this process.
	*/
	len = MIN(MAX(zvol_prefetch_bytes, 0), SPA_MAXBLOCKSIZE);
	if (len > 0) {
	dmu_prefetch(os, ZVOL_OBJ, 0, 0, len, ZIO_PRIORITY_SYNC_READ);
	dmu_prefetch(os, ZVOL_OBJ, 0, volsize - len, len,
	ZIO_PRIORITY_SYNC_READ);
	}

	zv->zv_objset = NULL;
	out_dmu_objset_disown:
	dmu_objset_disown(os, B_TRUE, FTAG);
	out_doi:
	kmem_free(doi, sizeof (dmu_object_info_t));

	/*
	* Keep in mind that once add_disk() is called, the zvol is
	* announced to the world, and zvol_open()/zvol_release() can
	* be called at any time. Incidentally, add_disk() itself calls
	* zvol_open()->zvol_first_open() and zvol_release()->zvol_last_close()
	* directly as well.
	*/
	if (error == 0) {
	rw_enter(&zvol_state_lock, RW_WRITER);
	zvol_insert(zv);
	rw_exit(&zvol_state_lock);
	add_disk(zv->zv_zso->zvo_disk);
	} else {
	ida_simple_remove(&zvol_ida, idx);
	}

	return (error);
	}

	static void
	zvol_rename_minor(zvol_state_t zv, const char newname)
	{
	int readonly = get_disk_ro(zv->zv_zso->zvo_disk);

	ASSERT(RW_LOCK_HELD(&zvol_state_lock));
	ASSERT(MUTEX_HELD(&zv->zv_state_lock));

	strlcpy(zv->zv_name, newname, sizeof (zv->zv_name));

	/* move to new hashtable entry */
	zv->zv_hash = zvol_name_hash(zv->zv_name);
	hlist_del(&zv->zv_hlink);
	hlist_add_head(&zv->zv_hlink, ZVOL_HT_HEAD(zv->zv_hash));

	/*
	* The block device's read-only state is briefly changed causing
	* a KOBJ_CHANGE uevent to be issued. This ensures udev detects
	* the name change and fixes the symlinks. This does not change
	* ZVOL_RDONLY in zv->zv_flags so the actual read-only state never
	* changes. This would normally be done using kobject_uevent() but
	* that is a GPL-only symbol which is why we need this workaround.
	*/
	set_disk_ro(zv->zv_zso->zvo_disk, !readonly);
	set_disk_ro(zv->zv_zso->zvo_disk, readonly);
	}

	static void
	zvol_set_disk_ro_impl(zvol_state_t *zv, int flags)
	{

	set_disk_ro(zv->zv_zso->zvo_disk, flags);
	}

	static void
	zvol_set_capacity_impl(zvol_state_t *zv, uint64_t capacity)
	{

	set_capacity(zv->zv_zso->zvo_disk, capacity);
	}

	const static zvol_platform_ops_t zvol_linux_ops = {
	.zv_free = zvol_free,
	.zv_rename_minor = zvol_rename_minor,
	.zv_create_minor = zvol_os_create_minor,
	.zv_update_volsize = zvol_update_volsize,
	.zv_clear_private = zvol_clear_private,
	.zv_is_zvol = zvol_is_zvol_impl,
	.zv_set_disk_ro = zvol_set_disk_ro_impl,
	.zv_set_capacity = zvol_set_capacity_impl,
	};

	int
	zvol_init(void)
	{
	int error;
	int threads = MIN(MAX(zvol_threads, 1), 1024);

	error = register_blkdev(zvol_major, ZVOL_DRIVER);
	if (error) {
	printk(KERN_INFO "ZFS: register_blkdev() failed %d\n", error);
	return (error);
	}
	zvol_taskq = taskq_create(ZVOL_DRIVER, threads, maxclsyspri,
	threads * 2, INT_MAX, TASKQ_PREPOPULATE \| TASKQ_DYNAMIC);
	if (zvol_taskq == NULL) {
	unregister_blkdev(zvol_major, ZVOL_DRIVER);
	return (-ENOMEM);
	}
	zvol_init_impl();
	ida_init(&zvol_ida);
	zvol_register_ops(&zvol_linux_ops);
	return (0);
	}

	void
	zvol_fini(void)
	{
	zvol_fini_impl();
	unregister_blkdev(zvol_major, ZVOL_DRIVER);
	taskq_destroy(zvol_taskq);
	ida_destroy(&zvol_ida);
	}

	/* BEGIN CSTYLED */
	module_param(zvol_inhibit_dev, uint, 0644);
	MODULE_PARM_DESC(zvol_inhibit_dev, "Do not create zvol device nodes");

	module_param(zvol_major, uint, 0444);
	MODULE_PARM_DESC(zvol_major, "Major number for zvol device");

	module_param(zvol_threads, uint, 0444);
	MODULE_PARM_DESC(zvol_threads, "Max number of threads to handle I/O requests");

	module_param(zvol_request_sync, uint, 0644);
	MODULE_PARM_DESC(zvol_request_sync, "Synchronously handle bio requests");

	module_param(zvol_max_discard_blocks, ulong, 0444);
	MODULE_PARM_DESC(zvol_max_discard_blocks, "Max number of blocks to discard");

	module_param(zvol_prefetch_bytes, uint, 0644);
	MODULE_PARM_DESC(zvol_prefetch_bytes, "Prefetch N bytes at zvol start+end");

	module_param(zvol_volmode, uint, 0644);
	MODULE_PARM_DESC(zvol_volmode, "Default volmode property value");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/zcommon/zfeature_common.c b/sys/contrib/openzfs/module/zcommon/zfeature_common.c
	index da7454c118d7..8d84c6d15148 100644
	--- a/sys/contrib/openzfs/module/zcommon/zfeature_common.c
	+++ b/sys/contrib/openzfs/module/zcommon/zfeature_common.c
	@@ -1,607 +1,608 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2011, 2018 by Delphix. All rights reserved.
	* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
	* Copyright (c) 2013, Joyent, Inc. All rights reserved.
	* Copyright (c) 2014, Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2017, Intel Corporation.
	* Copyright (c) 2019, Klara Inc.
	* Copyright (c) 2019, Allan Jude
	*/

	#ifndef _KERNEL
	#include <errno.h>
	#include <string.h>
	#include <sys/stat.h>
	#endif
	#include <sys/debug.h>
	#include <sys/fs/zfs.h>
	#include <sys/inttypes.h>
	#include <sys/types.h>
	#include <sys/param.h>
	#include <sys/zfs_sysfs.h>
	#include "zfeature_common.h"

	/*
	* Set to disable all feature checks while opening pools, allowing pools with
	* unsupported features to be opened. Set for testing only.
	*/
	boolean_t zfeature_checks_disable = B_FALSE;

	zfeature_info_t spa_feature_table[SPA_FEATURES];

	/*
	* Valid characters for feature guids. This list is mainly for aesthetic
	* purposes and could be expanded in the future. There are different allowed
	* characters in the guids reverse dns portion (before the colon) and its
	* short name (after the colon).
	*/
	static int
	valid_char(char c, boolean_t after_colon)
	{
	return ((c >= 'a' && c <= 'z') \|\|
	(c >= '0' && c <= '9') \|\|
	(after_colon && c == '_') \|\|
	(!after_colon && (c == '.' \|\| c == '-')));
	}

	/*
	* Every feature guid must contain exactly one colon which separates a reverse
	* dns organization name from the feature's "short" name (e.g.
	* "com.company:feature_name").
	*/
	boolean_t
	zfeature_is_valid_guid(const char *name)
	{
	int i;
	boolean_t has_colon = B_FALSE;

	i = 0;
	while (name[i] != '\0') {
	char c = name[i++];
	if (c == ':') {
	if (has_colon)
	return (B_FALSE);
	has_colon = B_TRUE;
	continue;
	}
	if (!valid_char(c, has_colon))
	return (B_FALSE);
	}

	return (has_colon);
	}

	boolean_t
	zfeature_is_supported(const char *guid)
	{
	if (zfeature_checks_disable)
	return (B_TRUE);

	for (spa_feature_t i = 0; i < SPA_FEATURES; i++) {
	zfeature_info_t *feature = &spa_feature_table[i];
	if (!feature->fi_zfs_mod_supported)
	continue;
	if (strcmp(guid, feature->fi_guid) == 0)
	return (B_TRUE);
	}
	return (B_FALSE);
	}

	int
	zfeature_lookup_guid(const char guid, spa_feature_t res)
	{
	for (spa_feature_t i = 0; i < SPA_FEATURES; i++) {
	zfeature_info_t *feature = &spa_feature_table[i];
	if (!feature->fi_zfs_mod_supported)
	continue;
	if (strcmp(guid, feature->fi_guid) == 0) {
	if (res != NULL)
	*res = i;
	return (0);
	}
	}

	return (ENOENT);
	}

	int
	zfeature_lookup_name(const char name, spa_feature_t res)
	{
	for (spa_feature_t i = 0; i < SPA_FEATURES; i++) {
	zfeature_info_t *feature = &spa_feature_table[i];
	if (!feature->fi_zfs_mod_supported)
	continue;
	if (strcmp(name, feature->fi_uname) == 0) {
	if (res != NULL)
	*res = i;
	return (0);
	}
	}

	return (ENOENT);
	}

	boolean_t
	zfeature_depends_on(spa_feature_t fid, spa_feature_t check)
	{
	zfeature_info_t *feature = &spa_feature_table[fid];

	for (int i = 0; feature->fi_depends[i] != SPA_FEATURE_NONE; i++) {
	if (feature->fi_depends[i] == check)
	return (B_TRUE);
	}
	return (B_FALSE);
	}

	static boolean_t
	deps_contains_feature(const spa_feature_t *deps, const spa_feature_t feature)
	{
	for (int i = 0; deps[i] != SPA_FEATURE_NONE; i++)
	if (deps[i] == feature)
	return (B_TRUE);

	return (B_FALSE);
	}

	#if !defined(_KERNEL) && !defined(LIB_ZPOOL_BUILD)
	static boolean_t
	zfs_mod_supported_impl(const char scope, const char name, const char *sysfs)
	{
	boolean_t supported = B_FALSE;
	char *path;

	int len = asprintf(&path, "%s%s%s%s%s", sysfs,
	scope == NULL ? "" : "/", scope == NULL ? "" : scope,
	name == NULL ? "" : "/", name == NULL ? "" : name);
	if (len > 0) {
	struct stat64 statbuf;
	supported = !!(stat64(path, &statbuf) == 0);
	free(path);
	}

	return (supported);
	}

	boolean_t
	zfs_mod_supported(const char scope, const char name)
	{
	boolean_t supported;

	/*
	* Check both the primary and alternate sysfs locations to determine
	* if the required functionality is supported.
	*/
	supported = (zfs_mod_supported_impl(scope, name, ZFS_SYSFS_DIR) \|\|
	zfs_mod_supported_impl(scope, name, ZFS_SYSFS_ALT_DIR));

	/*
	* For backwards compatibility with kernel modules that predate
	* supported feature/property checking. Report the feature/property
	* as supported if the kernel module is loaded but the requested
	* scope directory does not exist.
	*/
	if (supported == B_FALSE) {
	struct stat64 statbuf;
	if ((stat64(ZFS_SYSFS_DIR, &statbuf) == 0) &&
	!zfs_mod_supported_impl(scope, NULL, ZFS_SYSFS_DIR) &&
	!zfs_mod_supported_impl(scope, NULL, ZFS_SYSFS_ALT_DIR)) {
	supported = B_TRUE;
	}
	}

	return (supported);
	}
	#endif

	static boolean_t
	zfs_mod_supported_feature(const char *name)
	{
	/*
	* The zfs module spa_feature_table[], whether in-kernel or in
	* libzpool, always supports all the features. libzfs needs to
	* query the running module, via sysfs, to determine which
	* features are supported.
	*
	* The equivalent _can_ be done on FreeBSD by way of the sysctl
	* tree, but this has not been done yet. Therefore, we return
	- * that all features except edonr are supported.
	+ * that all features are supported.
	*/

	#if defined(_KERNEL) \|\| defined(LIB_ZPOOL_BUILD) \|\| defined(__FreeBSD__)
	+ (void) name;
	return (B_TRUE);
	#else
	return (zfs_mod_supported(ZFS_SYSFS_POOL_FEATURES, name));
	#endif
	}

	static void
	zfeature_register(spa_feature_t fid, const char guid, const char name,
	const char *desc, zfeature_flags_t flags, zfeature_type_t type,
	const spa_feature_t *deps)
	{
	zfeature_info_t *feature = &spa_feature_table[fid];
	static spa_feature_t nodeps[] = { SPA_FEATURE_NONE };

	ASSERT(name != NULL);
	ASSERT(desc != NULL);
	ASSERT((flags & ZFEATURE_FLAG_READONLY_COMPAT) == 0 \|\|
	(flags & ZFEATURE_FLAG_MOS) == 0);
	ASSERT3U(fid, <, SPA_FEATURES);
	ASSERT(zfeature_is_valid_guid(guid));

	if (deps == NULL)
	deps = nodeps;

	VERIFY(((flags & ZFEATURE_FLAG_PER_DATASET) == 0) \|\|
	(deps_contains_feature(deps, SPA_FEATURE_EXTENSIBLE_DATASET)));

	feature->fi_feature = fid;
	feature->fi_guid = guid;
	feature->fi_uname = name;
	feature->fi_desc = desc;
	feature->fi_flags = flags;
	feature->fi_type = type;
	feature->fi_depends = deps;
	feature->fi_zfs_mod_supported = zfs_mod_supported_feature(guid);
	}

	/*
	* Every feature has a GUID of the form com.example:feature_name. The
	* reversed DNS name ensures that the feature's GUID is unique across all ZFS
	* implementations. This allows companies to independently develop and
	* release features. Examples include org.delphix and org.datto. Previously,
	* features developed on one implementation have used that implementation's
	* domain name (e.g. org.illumos and org.zfsonlinux). Use of the org.openzfs
	* domain name is recommended for new features which are developed by the
	* OpenZFS community and its platforms. This domain may optionally be used by
	* companies developing features for initial release through an OpenZFS
	* implementation. Use of the org.openzfs domain requires reserving the
	* feature name in advance with the OpenZFS project.
	*/
	void
	zpool_feature_init(void)
	{
	zfeature_register(SPA_FEATURE_ASYNC_DESTROY,
	"com.delphix:async_destroy", "async_destroy",
	"Destroy filesystems asynchronously.",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL);

	zfeature_register(SPA_FEATURE_EMPTY_BPOBJ,
	"com.delphix:empty_bpobj", "empty_bpobj",
	"Snapshots use less space.",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL);

	zfeature_register(SPA_FEATURE_LZ4_COMPRESS,
	"org.illumos:lz4_compress", "lz4_compress",
	"LZ4 compression algorithm support.",
	ZFEATURE_FLAG_ACTIVATE_ON_ENABLE, ZFEATURE_TYPE_BOOLEAN, NULL);

	zfeature_register(SPA_FEATURE_MULTI_VDEV_CRASH_DUMP,
	"com.joyent:multi_vdev_crash_dump", "multi_vdev_crash_dump",
	"Crash dumps to multiple vdev pools.",
	0, ZFEATURE_TYPE_BOOLEAN, NULL);

	zfeature_register(SPA_FEATURE_SPACEMAP_HISTOGRAM,
	"com.delphix:spacemap_histogram", "spacemap_histogram",
	"Spacemaps maintain space histograms.",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL);

	zfeature_register(SPA_FEATURE_ENABLED_TXG,
	"com.delphix:enabled_txg", "enabled_txg",
	"Record txg at which a feature is enabled",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL);

	{
	static const spa_feature_t hole_birth_deps[] = {
	SPA_FEATURE_ENABLED_TXG,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_HOLE_BIRTH,
	"com.delphix:hole_birth", "hole_birth",
	"Retain hole birth txg for more precise zfs send",
	ZFEATURE_FLAG_MOS \| ZFEATURE_FLAG_ACTIVATE_ON_ENABLE,
	ZFEATURE_TYPE_BOOLEAN, hole_birth_deps);
	}

	zfeature_register(SPA_FEATURE_POOL_CHECKPOINT,
	"com.delphix:zpool_checkpoint", "zpool_checkpoint",
	"Pool state can be checkpointed, allowing rewind later.",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL);

	zfeature_register(SPA_FEATURE_SPACEMAP_V2,
	"com.delphix:spacemap_v2", "spacemap_v2",
	"Space maps representing large segments are more efficient.",
	ZFEATURE_FLAG_READONLY_COMPAT \| ZFEATURE_FLAG_ACTIVATE_ON_ENABLE,
	ZFEATURE_TYPE_BOOLEAN, NULL);

	zfeature_register(SPA_FEATURE_EXTENSIBLE_DATASET,
	"com.delphix:extensible_dataset", "extensible_dataset",
	"Enhanced dataset functionality, used by other features.",
	0, ZFEATURE_TYPE_BOOLEAN, NULL);

	{
	static const spa_feature_t bookmarks_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};

	zfeature_register(SPA_FEATURE_BOOKMARKS,
	"com.delphix:bookmarks", "bookmarks",
	"\"zfs bookmark\" command",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN,
	bookmarks_deps);
	}

	{
	static const spa_feature_t filesystem_limits_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_FS_SS_LIMIT,
	"com.joyent:filesystem_limits", "filesystem_limits",
	"Filesystem and snapshot limits.",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN,
	filesystem_limits_deps);
	}

	zfeature_register(SPA_FEATURE_EMBEDDED_DATA,
	"com.delphix:embedded_data", "embedded_data",
	"Blocks which compress very well use even less space.",
	ZFEATURE_FLAG_MOS \| ZFEATURE_FLAG_ACTIVATE_ON_ENABLE,
	ZFEATURE_TYPE_BOOLEAN, NULL);

	{
	static const spa_feature_t livelist_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_LIVELIST,
	"com.delphix:livelist", "livelist",
	"Improved clone deletion performance.",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN,
	livelist_deps);
	}

	{
	static const spa_feature_t log_spacemap_deps[] = {
	SPA_FEATURE_SPACEMAP_V2,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_LOG_SPACEMAP,
	"com.delphix:log_spacemap", "log_spacemap",
	"Log metaslab changes on a single spacemap and "
	"flush them periodically.",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN,
	log_spacemap_deps);
	}

	{
	static const spa_feature_t large_blocks_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_LARGE_BLOCKS,
	"org.open-zfs:large_blocks", "large_blocks",
	"Support for blocks larger than 128KB.",
	ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN,
	large_blocks_deps);
	}

	{
	static const spa_feature_t large_dnode_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_LARGE_DNODE,
	"org.zfsonlinux:large_dnode", "large_dnode",
	"Variable on-disk size of dnodes.",
	ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN,
	large_dnode_deps);
	}

	{
	static const spa_feature_t sha512_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_SHA512,
	"org.illumos:sha512", "sha512",
	"SHA-512/256 hash algorithm.",
	ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN,
	sha512_deps);
	}

	{
	static const spa_feature_t skein_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_SKEIN,
	"org.illumos:skein", "skein",
	"Skein hash algorithm.",
	ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN,
	skein_deps);
	}

	{
	static const spa_feature_t edonr_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_EDONR,
	"org.illumos:edonr", "edonr",
	"Edon-R hash algorithm.",
	ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN,
	edonr_deps);
	}

	{
	static const spa_feature_t redact_books_deps[] = {
	SPA_FEATURE_BOOKMARK_V2,
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_BOOKMARKS,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_REDACTION_BOOKMARKS,
	"com.delphix:redaction_bookmarks", "redaction_bookmarks",
	"Support for bookmarks which store redaction lists for zfs "
	"redacted send/recv.", 0, ZFEATURE_TYPE_BOOLEAN,
	redact_books_deps);
	}

	{
	static const spa_feature_t redact_datasets_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_REDACTED_DATASETS,
	"com.delphix:redacted_datasets", "redacted_datasets", "Support for "
	"redacted datasets, produced by receiving a redacted zfs send "
	"stream.", ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_UINT64_ARRAY,
	redact_datasets_deps);
	}

	{
	static const spa_feature_t bookmark_written_deps[] = {
	SPA_FEATURE_BOOKMARK_V2,
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_BOOKMARKS,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_BOOKMARK_WRITTEN,
	"com.delphix:bookmark_written", "bookmark_written",
	"Additional accounting, enabling the written#<bookmark> property"
	"(space written since a bookmark), and estimates of send stream "
	"sizes for incrementals from bookmarks.",
	0, ZFEATURE_TYPE_BOOLEAN, bookmark_written_deps);
	}

	zfeature_register(SPA_FEATURE_DEVICE_REMOVAL,
	"com.delphix:device_removal", "device_removal",
	"Top-level vdevs can be removed, reducing logical pool size.",
	ZFEATURE_FLAG_MOS, ZFEATURE_TYPE_BOOLEAN, NULL);

	{
	static const spa_feature_t obsolete_counts_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_DEVICE_REMOVAL,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_OBSOLETE_COUNTS,
	"com.delphix:obsolete_counts", "obsolete_counts",
	"Reduce memory used by removed devices when their blocks are "
	"freed or remapped.",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN,
	obsolete_counts_deps);
	}

	{
	static const spa_feature_t userobj_accounting_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_USEROBJ_ACCOUNTING,
	"org.zfsonlinux:userobj_accounting", "userobj_accounting",
	"User/Group object accounting.",
	ZFEATURE_FLAG_READONLY_COMPAT \| ZFEATURE_FLAG_PER_DATASET,
	ZFEATURE_TYPE_BOOLEAN, userobj_accounting_deps);
	}

	{
	static const spa_feature_t bookmark_v2_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_BOOKMARKS,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_BOOKMARK_V2,
	"com.datto:bookmark_v2", "bookmark_v2",
	"Support for larger bookmarks",
	0, ZFEATURE_TYPE_BOOLEAN, bookmark_v2_deps);
	}

	{
	static const spa_feature_t encryption_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_BOOKMARK_V2,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_ENCRYPTION,
	"com.datto:encryption", "encryption",
	"Support for dataset level encryption",
	ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN,
	encryption_deps);
	}

	{
	static const spa_feature_t project_quota_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_PROJECT_QUOTA,
	"org.zfsonlinux:project_quota", "project_quota",
	"space/object accounting based on project ID.",
	ZFEATURE_FLAG_READONLY_COMPAT \| ZFEATURE_FLAG_PER_DATASET,
	ZFEATURE_TYPE_BOOLEAN, project_quota_deps);
	}

	zfeature_register(SPA_FEATURE_ALLOCATION_CLASSES,
	"org.zfsonlinux:allocation_classes", "allocation_classes",
	"Support for separate allocation classes.",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL);

	zfeature_register(SPA_FEATURE_RESILVER_DEFER,
	"com.datto:resilver_defer", "resilver_defer",
	"Support for deferring new resilvers when one is already running.",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL);

	zfeature_register(SPA_FEATURE_DEVICE_REBUILD,
	"org.openzfs:device_rebuild", "device_rebuild",
	"Support for sequential mirror/dRAID device rebuilds",
	ZFEATURE_FLAG_READONLY_COMPAT, ZFEATURE_TYPE_BOOLEAN, NULL);

	{
	static const spa_feature_t zstd_deps[] = {
	SPA_FEATURE_EXTENSIBLE_DATASET,
	SPA_FEATURE_NONE
	};
	zfeature_register(SPA_FEATURE_ZSTD_COMPRESS,
	"org.freebsd:zstd_compress", "zstd_compress",
	"zstd compression algorithm support.",
	ZFEATURE_FLAG_PER_DATASET, ZFEATURE_TYPE_BOOLEAN, zstd_deps);
	}

	zfeature_register(SPA_FEATURE_DRAID,
	"org.openzfs:draid", "draid", "Support for distributed spare RAID",
	ZFEATURE_FLAG_MOS, ZFEATURE_TYPE_BOOLEAN, NULL);
	}

	#if defined(_KERNEL)
	EXPORT_SYMBOL(zfeature_lookup_guid);
	EXPORT_SYMBOL(zfeature_lookup_name);
	EXPORT_SYMBOL(zfeature_is_supported);
	EXPORT_SYMBOL(zfeature_is_valid_guid);
	EXPORT_SYMBOL(zfeature_depends_on);
	EXPORT_SYMBOL(zpool_feature_init);
	EXPORT_SYMBOL(spa_feature_table);
	#endif
	diff --git a/sys/contrib/openzfs/module/zcommon/zfs_prop.c b/sys/contrib/openzfs/module/zcommon/zfs_prop.c
	index 260bf185a3f8..2a0e26eca9c8 100644
	--- a/sys/contrib/openzfs/module/zcommon/zfs_prop.c
	+++ b/sys/contrib/openzfs/module/zcommon/zfs_prop.c
	@@ -1,1042 +1,1030 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2018 by Delphix. All rights reserved.
	* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
	* Copyright 2016, Joyent, Inc.
	* Copyright (c) 2019, Klara Inc.
	* Copyright (c) 2019, Allan Jude
	*/

	/* Portions Copyright 2010 Robert Milkowski */

	#include <sys/zio.h>
	#include <sys/spa.h>
	#include <sys/u8_textprep.h>
	#include <sys/zfs_acl.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/zfs_znode.h>
	#include <sys/dsl_crypt.h>

	#include "zfs_prop.h"
	#include "zfs_deleg.h"
	#include "zfs_fletcher.h"

	#if !defined(_KERNEL)
	#include <stdlib.h>
	#include <string.h>
	#include <ctype.h>
	#endif

	static zprop_desc_t zfs_prop_table[ZFS_NUM_PROPS];

	/* Note this is indexed by zfs_userquota_prop_t, keep the order the same */
	const char *zfs_userquota_prop_prefixes[] = {
	"userused@",
	"userquota@",
	"groupused@",
	"groupquota@",
	"userobjused@",
	"userobjquota@",
	"groupobjused@",
	"groupobjquota@",
	"projectused@",
	"projectquota@",
	"projectobjused@",
	"projectobjquota@"
	};

	zprop_desc_t *
	zfs_prop_get_table(void)
	{
	return (zfs_prop_table);
	}

	void
	zfs_prop_init(void)
	{
	static zprop_index_t checksum_table[] = {
	{ "on", ZIO_CHECKSUM_ON },
	{ "off", ZIO_CHECKSUM_OFF },
	{ "fletcher2", ZIO_CHECKSUM_FLETCHER_2 },
	{ "fletcher4", ZIO_CHECKSUM_FLETCHER_4 },
	{ "sha256", ZIO_CHECKSUM_SHA256 },
	{ "noparity", ZIO_CHECKSUM_NOPARITY },
	{ "sha512", ZIO_CHECKSUM_SHA512 },
	{ "skein", ZIO_CHECKSUM_SKEIN },
	{ "edonr", ZIO_CHECKSUM_EDONR },
	{ NULL }
	};

	static zprop_index_t dedup_table[] = {
	{ "on", ZIO_CHECKSUM_ON },
	{ "off", ZIO_CHECKSUM_OFF },
	{ "verify", ZIO_CHECKSUM_ON \| ZIO_CHECKSUM_VERIFY },
	{ "sha256", ZIO_CHECKSUM_SHA256 },
	{ "sha256,verify",
	ZIO_CHECKSUM_SHA256 \| ZIO_CHECKSUM_VERIFY },
	{ "sha512", ZIO_CHECKSUM_SHA512 },
	{ "sha512,verify",
	ZIO_CHECKSUM_SHA512 \| ZIO_CHECKSUM_VERIFY },
	{ "skein", ZIO_CHECKSUM_SKEIN },
	{ "skein,verify",
	ZIO_CHECKSUM_SKEIN \| ZIO_CHECKSUM_VERIFY },
	{ "edonr,verify",
	ZIO_CHECKSUM_EDONR \| ZIO_CHECKSUM_VERIFY },
	{ NULL }
	};

	static zprop_index_t compress_table[] = {
	{ "on", ZIO_COMPRESS_ON },
	{ "off", ZIO_COMPRESS_OFF },
	{ "lzjb", ZIO_COMPRESS_LZJB },
	{ "gzip", ZIO_COMPRESS_GZIP_6 }, /* gzip default */
	{ "gzip-1", ZIO_COMPRESS_GZIP_1 },
	{ "gzip-2", ZIO_COMPRESS_GZIP_2 },
	{ "gzip-3", ZIO_COMPRESS_GZIP_3 },
	{ "gzip-4", ZIO_COMPRESS_GZIP_4 },
	{ "gzip-5", ZIO_COMPRESS_GZIP_5 },
	{ "gzip-6", ZIO_COMPRESS_GZIP_6 },
	{ "gzip-7", ZIO_COMPRESS_GZIP_7 },
	{ "gzip-8", ZIO_COMPRESS_GZIP_8 },
	{ "gzip-9", ZIO_COMPRESS_GZIP_9 },
	{ "zle", ZIO_COMPRESS_ZLE },
	{ "lz4", ZIO_COMPRESS_LZ4 },
	{ "zstd", ZIO_COMPRESS_ZSTD },
	{ "zstd-fast",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_DEFAULT) },

	/*
	* ZSTD 1-19 are synthetic. We store the compression level in a
	* separate hidden property to avoid wasting a large amount of
	* space in the ZIO_COMPRESS enum.
	*
	* The compression level is also stored within the header of the
	* compressed block since we may need it for later recompression
	* to avoid checksum errors (L2ARC).
	*
	* Note that the level here is defined as bit shifted mask on
	* top of the method.
	*/
	{ "zstd-1", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_1) },
	{ "zstd-2", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_2) },
	{ "zstd-3", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_3) },
	{ "zstd-4", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_4) },
	{ "zstd-5", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_5) },
	{ "zstd-6", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_6) },
	{ "zstd-7", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_7) },
	{ "zstd-8", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_8) },
	{ "zstd-9", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_9) },
	{ "zstd-10", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_10) },
	{ "zstd-11", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_11) },
	{ "zstd-12", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_12) },
	{ "zstd-13", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_13) },
	{ "zstd-14", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_14) },
	{ "zstd-15", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_15) },
	{ "zstd-16", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_16) },
	{ "zstd-17", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_17) },
	{ "zstd-18", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_18) },
	{ "zstd-19", ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_19) },

	/*
	* The ZSTD-Fast levels are also synthetic.
	*/
	{ "zstd-fast-1",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_1) },
	{ "zstd-fast-2",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_2) },
	{ "zstd-fast-3",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_3) },
	{ "zstd-fast-4",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_4) },
	{ "zstd-fast-5",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_5) },
	{ "zstd-fast-6",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_6) },
	{ "zstd-fast-7",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_7) },
	{ "zstd-fast-8",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_8) },
	{ "zstd-fast-9",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_9) },
	{ "zstd-fast-10",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_10) },
	{ "zstd-fast-20",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_20) },
	{ "zstd-fast-30",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_30) },
	{ "zstd-fast-40",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_40) },
	{ "zstd-fast-50",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_50) },
	{ "zstd-fast-60",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_60) },
	{ "zstd-fast-70",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_70) },
	{ "zstd-fast-80",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_80) },
	{ "zstd-fast-90",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_90) },
	{ "zstd-fast-100",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_100) },
	{ "zstd-fast-500",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_500) },
	{ "zstd-fast-1000",
	ZIO_COMPLEVEL_ZSTD(ZIO_ZSTD_LEVEL_FAST_1000) },
	{ NULL }
	};

	static zprop_index_t crypto_table[] = {
	{ "on", ZIO_CRYPT_ON },
	{ "off", ZIO_CRYPT_OFF },
	{ "aes-128-ccm", ZIO_CRYPT_AES_128_CCM },
	{ "aes-192-ccm", ZIO_CRYPT_AES_192_CCM },
	{ "aes-256-ccm", ZIO_CRYPT_AES_256_CCM },
	{ "aes-128-gcm", ZIO_CRYPT_AES_128_GCM },
	{ "aes-192-gcm", ZIO_CRYPT_AES_192_GCM },
	{ "aes-256-gcm", ZIO_CRYPT_AES_256_GCM },
	{ NULL }
	};

	static zprop_index_t keyformat_table[] = {
	{ "none", ZFS_KEYFORMAT_NONE },
	{ "raw", ZFS_KEYFORMAT_RAW },
	{ "hex", ZFS_KEYFORMAT_HEX },
	{ "passphrase", ZFS_KEYFORMAT_PASSPHRASE },
	{ NULL }
	};

	static zprop_index_t snapdir_table[] = {
	{ "hidden", ZFS_SNAPDIR_HIDDEN },
	{ "visible", ZFS_SNAPDIR_VISIBLE },
	{ NULL }
	};

	static zprop_index_t snapdev_table[] = {
	{ "hidden", ZFS_SNAPDEV_HIDDEN },
	{ "visible", ZFS_SNAPDEV_VISIBLE },
	{ NULL }
	};

	static zprop_index_t acl_mode_table[] = {
	{ "discard", ZFS_ACL_DISCARD },
	{ "groupmask", ZFS_ACL_GROUPMASK },
	{ "passthrough", ZFS_ACL_PASSTHROUGH },
	{ "restricted", ZFS_ACL_RESTRICTED },
	{ NULL }
	};

	static zprop_index_t acltype_table[] = {
	{ "off", ZFS_ACLTYPE_OFF },
	{ "posix", ZFS_ACLTYPE_POSIX },
	{ "nfsv4", ZFS_ACLTYPE_NFSV4 },
	{ "disabled", ZFS_ACLTYPE_OFF }, /* bkwrd compatibility */
	{ "noacl", ZFS_ACLTYPE_OFF }, /* bkwrd compatibility */
	{ "posixacl", ZFS_ACLTYPE_POSIX }, /* bkwrd compatibility */
	{ NULL }
	};

	static zprop_index_t acl_inherit_table[] = {
	{ "discard", ZFS_ACL_DISCARD },
	{ "noallow", ZFS_ACL_NOALLOW },
	{ "restricted", ZFS_ACL_RESTRICTED },
	{ "passthrough", ZFS_ACL_PASSTHROUGH },
	{ "secure", ZFS_ACL_RESTRICTED }, /* bkwrd compatibility */
	{ "passthrough-x", ZFS_ACL_PASSTHROUGH_X },
	{ NULL }
	};

	static zprop_index_t case_table[] = {
	{ "sensitive", ZFS_CASE_SENSITIVE },
	{ "insensitive", ZFS_CASE_INSENSITIVE },
	{ "mixed", ZFS_CASE_MIXED },
	{ NULL }
	};

	static zprop_index_t copies_table[] = {
	{ "1", 1 },
	{ "2", 2 },
	{ "3", 3 },
	{ NULL }
	};

	/*
	* Use the unique flags we have to send to u8_strcmp() and/or
	* u8_textprep() to represent the various normalization property
	* values.
	*/
	static zprop_index_t normalize_table[] = {
	{ "none", 0 },
	{ "formD", U8_TEXTPREP_NFD },
	{ "formKC", U8_TEXTPREP_NFKC },
	{ "formC", U8_TEXTPREP_NFC },
	{ "formKD", U8_TEXTPREP_NFKD },
	{ NULL }
	};

	static zprop_index_t version_table[] = {
	{ "1", 1 },
	{ "2", 2 },
	{ "3", 3 },
	{ "4", 4 },
	{ "5", 5 },
	{ "current", ZPL_VERSION },
	{ NULL }
	};

	static zprop_index_t boolean_table[] = {
	{ "off", 0 },
	{ "on", 1 },
	{ NULL }
	};

	static zprop_index_t keystatus_table[] = {
	{ "none", ZFS_KEYSTATUS_NONE},
	{ "unavailable", ZFS_KEYSTATUS_UNAVAILABLE},
	{ "available", ZFS_KEYSTATUS_AVAILABLE},
	{ NULL }
	};

	static zprop_index_t logbias_table[] = {
	{ "latency", ZFS_LOGBIAS_LATENCY },
	{ "throughput", ZFS_LOGBIAS_THROUGHPUT },
	{ NULL }
	};

	static zprop_index_t canmount_table[] = {
	{ "off", ZFS_CANMOUNT_OFF },
	{ "on", ZFS_CANMOUNT_ON },
	{ "noauto", ZFS_CANMOUNT_NOAUTO },
	{ NULL }
	};

	static zprop_index_t cache_table[] = {
	{ "none", ZFS_CACHE_NONE },
	{ "metadata", ZFS_CACHE_METADATA },
	{ "all", ZFS_CACHE_ALL },
	{ NULL }
	};

	static zprop_index_t sync_table[] = {
	{ "standard", ZFS_SYNC_STANDARD },
	{ "always", ZFS_SYNC_ALWAYS },
	{ "disabled", ZFS_SYNC_DISABLED },
	{ NULL }
	};

	static zprop_index_t xattr_table[] = {
	{ "off", ZFS_XATTR_OFF },
	{ "on", ZFS_XATTR_DIR },
	{ "sa", ZFS_XATTR_SA },
	{ "dir", ZFS_XATTR_DIR },
	{ NULL }
	};

	static zprop_index_t dnsize_table[] = {
	{ "legacy", ZFS_DNSIZE_LEGACY },
	{ "auto", ZFS_DNSIZE_AUTO },
	{ "1k", ZFS_DNSIZE_1K },
	{ "2k", ZFS_DNSIZE_2K },
	{ "4k", ZFS_DNSIZE_4K },
	{ "8k", ZFS_DNSIZE_8K },
	{ "16k", ZFS_DNSIZE_16K },
	{ NULL }
	};

	static zprop_index_t redundant_metadata_table[] = {
	{ "all", ZFS_REDUNDANT_METADATA_ALL },
	{ "most", ZFS_REDUNDANT_METADATA_MOST },
	{ NULL }
	};

	static zprop_index_t volmode_table[] = {
	{ "default", ZFS_VOLMODE_DEFAULT },
	{ "full", ZFS_VOLMODE_GEOM },
	{ "geom", ZFS_VOLMODE_GEOM },
	{ "dev", ZFS_VOLMODE_DEV },
	{ "none", ZFS_VOLMODE_NONE },
	{ NULL }
	};

	/* inherit index properties */
	zprop_register_index(ZFS_PROP_REDUNDANT_METADATA, "redundant_metadata",
	ZFS_REDUNDANT_METADATA_ALL,
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"all \| most", "REDUND_MD",
	redundant_metadata_table);
	zprop_register_index(ZFS_PROP_SYNC, "sync", ZFS_SYNC_STANDARD,
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"standard \| always \| disabled", "SYNC",
	sync_table);
	zprop_register_index(ZFS_PROP_CHECKSUM, "checksum",
	ZIO_CHECKSUM_DEFAULT, PROP_INHERIT, ZFS_TYPE_FILESYSTEM \|
	ZFS_TYPE_VOLUME,
	"on \| off \| fletcher2 \| fletcher4 \| sha256 \| sha512 \| skein"
	" \| edonr",
	"CHECKSUM", checksum_table);
	zprop_register_index(ZFS_PROP_DEDUP, "dedup", ZIO_CHECKSUM_OFF,
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"on \| off \| verify \| sha256[,verify] \| sha512[,verify] \| "
	"skein[,verify] \| edonr,verify",
	"DEDUP", dedup_table);
	zprop_register_index(ZFS_PROP_COMPRESSION, "compression",
	ZIO_COMPRESS_DEFAULT, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"on \| off \| lzjb \| gzip \| gzip-[1-9] \| zle \| lz4 \| "
	"zstd \| zstd-[1-19] \| "
	"zstd-fast \| zstd-fast-[1-10,20,30,40,50,60,70,80,90,100,500,1000]",
	"COMPRESS", compress_table);
	zprop_register_index(ZFS_PROP_SNAPDIR, "snapdir", ZFS_SNAPDIR_HIDDEN,
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM,
	"hidden \| visible", "SNAPDIR", snapdir_table);
	zprop_register_index(ZFS_PROP_SNAPDEV, "snapdev", ZFS_SNAPDEV_HIDDEN,
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"hidden \| visible", "SNAPDEV", snapdev_table);
	zprop_register_index(ZFS_PROP_ACLMODE, "aclmode", ZFS_ACL_DISCARD,
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM,
	"discard \| groupmask \| passthrough \| restricted", "ACLMODE",
	acl_mode_table);
	zprop_register_index(ZFS_PROP_ACLTYPE, "acltype",
	#ifdef __linux__
	/* Linux doesn't natively support ZFS's NFSv4-style ACLs. */
	ZFS_ACLTYPE_OFF,
	#else
	ZFS_ACLTYPE_NFSV4,
	#endif
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_SNAPSHOT,
	"off \| nfsv4 \| posix", "ACLTYPE", acltype_table);
	zprop_register_index(ZFS_PROP_ACLINHERIT, "aclinherit",
	ZFS_ACL_RESTRICTED, PROP_INHERIT, ZFS_TYPE_FILESYSTEM,
	"discard \| noallow \| restricted \| passthrough \| passthrough-x",
	"ACLINHERIT", acl_inherit_table);
	zprop_register_index(ZFS_PROP_COPIES, "copies", 1, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"1 \| 2 \| 3", "COPIES", copies_table);
	zprop_register_index(ZFS_PROP_PRIMARYCACHE, "primarycache",
	ZFS_CACHE_ALL, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_SNAPSHOT \| ZFS_TYPE_VOLUME,
	"all \| none \| metadata", "PRIMARYCACHE", cache_table);
	zprop_register_index(ZFS_PROP_SECONDARYCACHE, "secondarycache",
	ZFS_CACHE_ALL, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_SNAPSHOT \| ZFS_TYPE_VOLUME,
	"all \| none \| metadata", "SECONDARYCACHE", cache_table);
	zprop_register_index(ZFS_PROP_LOGBIAS, "logbias", ZFS_LOGBIAS_LATENCY,
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"latency \| throughput", "LOGBIAS", logbias_table);
	zprop_register_index(ZFS_PROP_XATTR, "xattr", ZFS_XATTR_DIR,
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_SNAPSHOT,
	"on \| off \| dir \| sa", "XATTR", xattr_table);
	zprop_register_index(ZFS_PROP_DNODESIZE, "dnodesize",
	ZFS_DNSIZE_LEGACY, PROP_INHERIT, ZFS_TYPE_FILESYSTEM,
	"legacy \| auto \| 1k \| 2k \| 4k \| 8k \| 16k", "DNSIZE", dnsize_table);
	zprop_register_index(ZFS_PROP_VOLMODE, "volmode",
	ZFS_VOLMODE_DEFAULT, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"default \| full \| geom \| dev \| none", "VOLMODE", volmode_table);

	/* inherit index (boolean) properties */
	zprop_register_index(ZFS_PROP_ATIME, "atime", 1, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM, "on \| off", "ATIME", boolean_table);
	zprop_register_index(ZFS_PROP_RELATIME, "relatime", 0, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM, "on \| off", "RELATIME", boolean_table);
	zprop_register_index(ZFS_PROP_DEVICES, "devices", 1, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_SNAPSHOT, "on \| off", "DEVICES",
	boolean_table);
	zprop_register_index(ZFS_PROP_EXEC, "exec", 1, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_SNAPSHOT, "on \| off", "EXEC",
	boolean_table);
	zprop_register_index(ZFS_PROP_SETUID, "setuid", 1, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_SNAPSHOT, "on \| off", "SETUID",
	boolean_table);
	zprop_register_index(ZFS_PROP_READONLY, "readonly", 0, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME, "on \| off", "RDONLY",
	boolean_table);
	#ifdef __FreeBSD__
	zprop_register_index(ZFS_PROP_ZONED, "jailed", 0, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM, "on \| off", "JAILED", boolean_table);
	#else
	zprop_register_index(ZFS_PROP_ZONED, "zoned", 0, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM, "on \| off", "ZONED", boolean_table);
	#endif
	zprop_register_index(ZFS_PROP_VSCAN, "vscan", 0, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM, "on \| off", "VSCAN", boolean_table);
	zprop_register_index(ZFS_PROP_NBMAND, "nbmand", 0, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_SNAPSHOT, "on \| off", "NBMAND",
	boolean_table);
	zprop_register_index(ZFS_PROP_OVERLAY, "overlay", 1, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM, "on \| off", "OVERLAY", boolean_table);

	/* default index properties */
	zprop_register_index(ZFS_PROP_VERSION, "version", 0, PROP_DEFAULT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_SNAPSHOT,
	"1 \| 2 \| 3 \| 4 \| 5 \| current", "VERSION", version_table);
	zprop_register_index(ZFS_PROP_CANMOUNT, "canmount", ZFS_CANMOUNT_ON,
	PROP_DEFAULT, ZFS_TYPE_FILESYSTEM, "on \| off \| noauto",
	"CANMOUNT", canmount_table);

	/* readonly index properties */
	zprop_register_index(ZFS_PROP_MOUNTED, "mounted", 0, PROP_READONLY,
	ZFS_TYPE_FILESYSTEM, "yes \| no", "MOUNTED", boolean_table);
	zprop_register_index(ZFS_PROP_DEFER_DESTROY, "defer_destroy", 0,
	PROP_READONLY, ZFS_TYPE_SNAPSHOT, "yes \| no", "DEFER_DESTROY",
	boolean_table);
	zprop_register_index(ZFS_PROP_KEYSTATUS, "keystatus",
	ZFS_KEYSTATUS_NONE, PROP_READONLY, ZFS_TYPE_DATASET,
	"none \| unavailable \| available",
	"KEYSTATUS", keystatus_table);

	/* set once index properties */
	zprop_register_index(ZFS_PROP_NORMALIZE, "normalization", 0,
	PROP_ONETIME, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_SNAPSHOT,
	"none \| formC \| formD \| formKC \| formKD", "NORMALIZATION",
	normalize_table);
	zprop_register_index(ZFS_PROP_CASE, "casesensitivity",
	ZFS_CASE_SENSITIVE, PROP_ONETIME, ZFS_TYPE_FILESYSTEM \|
	ZFS_TYPE_SNAPSHOT,
	"sensitive \| insensitive \| mixed", "CASE", case_table);
	zprop_register_index(ZFS_PROP_KEYFORMAT, "keyformat",
	ZFS_KEYFORMAT_NONE, PROP_ONETIME_DEFAULT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"none \| raw \| hex \| passphrase", "KEYFORMAT", keyformat_table);
	zprop_register_index(ZFS_PROP_ENCRYPTION, "encryption",
	ZIO_CRYPT_DEFAULT, PROP_ONETIME, ZFS_TYPE_DATASET,
	"on \| off \| aes-128-ccm \| aes-192-ccm \| aes-256-ccm \| "
	"aes-128-gcm \| aes-192-gcm \| aes-256-gcm", "ENCRYPTION",
	crypto_table);

	/* set once index (boolean) properties */
	zprop_register_index(ZFS_PROP_UTF8ONLY, "utf8only", 0, PROP_ONETIME,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_SNAPSHOT,
	"on \| off", "UTF8ONLY", boolean_table);

	/* string properties */
	zprop_register_string(ZFS_PROP_ORIGIN, "origin", NULL, PROP_READONLY,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME, "<snapshot>", "ORIGIN");
	zprop_register_string(ZFS_PROP_CLONES, "clones", NULL, PROP_READONLY,
	ZFS_TYPE_SNAPSHOT, "<dataset>[,...]", "CLONES");
	zprop_register_string(ZFS_PROP_MOUNTPOINT, "mountpoint", "/",
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "<path> \| legacy \| none",
	"MOUNTPOINT");
	zprop_register_string(ZFS_PROP_SHARENFS, "sharenfs", "off",
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM, "on \| off \| NFS share options",
	"SHARENFS");
	zprop_register_string(ZFS_PROP_TYPE, "type", NULL, PROP_READONLY,
	ZFS_TYPE_DATASET \| ZFS_TYPE_BOOKMARK,
	"filesystem \| volume \| snapshot \| bookmark", "TYPE");
	zprop_register_string(ZFS_PROP_SHARESMB, "sharesmb", "off",
	PROP_INHERIT, ZFS_TYPE_FILESYSTEM,
	"on \| off \| SMB share options", "SHARESMB");
	zprop_register_string(ZFS_PROP_MLSLABEL, "mlslabel",
	ZFS_MLSLABEL_DEFAULT, PROP_INHERIT, ZFS_TYPE_DATASET,
	"<sensitivity label>", "MLSLABEL");
	zprop_register_string(ZFS_PROP_SELINUX_CONTEXT, "context",
	"none", PROP_DEFAULT, ZFS_TYPE_DATASET, "<selinux context>",
	"CONTEXT");
	zprop_register_string(ZFS_PROP_SELINUX_FSCONTEXT, "fscontext",
	"none", PROP_DEFAULT, ZFS_TYPE_DATASET, "<selinux fscontext>",
	"FSCONTEXT");
	zprop_register_string(ZFS_PROP_SELINUX_DEFCONTEXT, "defcontext",
	"none", PROP_DEFAULT, ZFS_TYPE_DATASET, "<selinux defcontext>",
	"DEFCONTEXT");
	zprop_register_string(ZFS_PROP_SELINUX_ROOTCONTEXT, "rootcontext",
	"none", PROP_DEFAULT, ZFS_TYPE_DATASET, "<selinux rootcontext>",
	"ROOTCONTEXT");
	zprop_register_string(ZFS_PROP_RECEIVE_RESUME_TOKEN,
	"receive_resume_token",
	NULL, PROP_READONLY, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"<string token>", "RESUMETOK");
	zprop_register_string(ZFS_PROP_ENCRYPTION_ROOT, "encryptionroot", NULL,
	PROP_READONLY, ZFS_TYPE_DATASET, "<filesystem \| volume>",
	"ENCROOT");
	zprop_register_string(ZFS_PROP_KEYLOCATION, "keylocation",
	"none", PROP_DEFAULT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"prompt \| <file URI> \| <https URL> \| <http URL>", "KEYLOCATION");
	zprop_register_string(ZFS_PROP_REDACT_SNAPS,
	"redact_snaps", NULL, PROP_READONLY,
	ZFS_TYPE_DATASET \| ZFS_TYPE_BOOKMARK, "<snapshot>[,...]",
	"RSNAPS");

	/* readonly number properties */
	zprop_register_number(ZFS_PROP_USED, "used", 0, PROP_READONLY,
	ZFS_TYPE_DATASET, "<size>", "USED");
	zprop_register_number(ZFS_PROP_AVAILABLE, "available", 0, PROP_READONLY,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME, "<size>", "AVAIL");
	zprop_register_number(ZFS_PROP_REFERENCED, "referenced", 0,
	PROP_READONLY, ZFS_TYPE_DATASET \| ZFS_TYPE_BOOKMARK, "<size>",
	"REFER");
	zprop_register_number(ZFS_PROP_COMPRESSRATIO, "compressratio", 0,
	PROP_READONLY, ZFS_TYPE_DATASET \| ZFS_TYPE_BOOKMARK,
	"<1.00x or higher if compressed>", "RATIO");
	zprop_register_number(ZFS_PROP_REFRATIO, "refcompressratio", 0,
	PROP_READONLY, ZFS_TYPE_DATASET,
	"<1.00x or higher if compressed>", "REFRATIO");
	zprop_register_number(ZFS_PROP_VOLBLOCKSIZE, "volblocksize",
	ZVOL_DEFAULT_BLOCKSIZE, PROP_ONETIME,
	ZFS_TYPE_VOLUME, "512 to 128k, power of 2", "VOLBLOCK");
	zprop_register_number(ZFS_PROP_USEDSNAP, "usedbysnapshots", 0,
	PROP_READONLY, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME, "<size>",
	"USEDSNAP");
	zprop_register_number(ZFS_PROP_USEDDS, "usedbydataset", 0,
	PROP_READONLY, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME, "<size>",
	"USEDDS");
	zprop_register_number(ZFS_PROP_USEDCHILD, "usedbychildren", 0,
	PROP_READONLY, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME, "<size>",
	"USEDCHILD");
	zprop_register_number(ZFS_PROP_USEDREFRESERV, "usedbyrefreservation", 0,
	PROP_READONLY,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME, "<size>", "USEDREFRESERV");
	zprop_register_number(ZFS_PROP_USERREFS, "userrefs", 0, PROP_READONLY,
	ZFS_TYPE_SNAPSHOT, "<count>", "USERREFS");
	zprop_register_number(ZFS_PROP_WRITTEN, "written", 0, PROP_READONLY,
	ZFS_TYPE_DATASET, "<size>", "WRITTEN");
	zprop_register_number(ZFS_PROP_LOGICALUSED, "logicalused", 0,
	PROP_READONLY, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME, "<size>",
	"LUSED");
	zprop_register_number(ZFS_PROP_LOGICALREFERENCED, "logicalreferenced",
	0, PROP_READONLY, ZFS_TYPE_DATASET \| ZFS_TYPE_BOOKMARK, "<size>",
	"LREFER");
	zprop_register_number(ZFS_PROP_FILESYSTEM_COUNT, "filesystem_count",
	UINT64_MAX, PROP_READONLY, ZFS_TYPE_FILESYSTEM,
	"<count>", "FSCOUNT");
	zprop_register_number(ZFS_PROP_SNAPSHOT_COUNT, "snapshot_count",
	UINT64_MAX, PROP_READONLY, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"<count>", "SSCOUNT");
	zprop_register_number(ZFS_PROP_GUID, "guid", 0, PROP_READONLY,
	ZFS_TYPE_DATASET \| ZFS_TYPE_BOOKMARK, "<uint64>", "GUID");
	zprop_register_number(ZFS_PROP_CREATETXG, "createtxg", 0, PROP_READONLY,
	ZFS_TYPE_DATASET \| ZFS_TYPE_BOOKMARK, "<uint64>", "CREATETXG");
	zprop_register_number(ZFS_PROP_PBKDF2_ITERS, "pbkdf2iters",
	0, PROP_ONETIME_DEFAULT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"<iters>", "PBKDF2ITERS");
	zprop_register_number(ZFS_PROP_OBJSETID, "objsetid", 0,
	PROP_READONLY, ZFS_TYPE_DATASET, "<uint64>", "OBJSETID");

	/* default number properties */
	zprop_register_number(ZFS_PROP_QUOTA, "quota", 0, PROP_DEFAULT,
	ZFS_TYPE_FILESYSTEM, "<size> \| none", "QUOTA");
	zprop_register_number(ZFS_PROP_RESERVATION, "reservation", 0,
	PROP_DEFAULT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"<size> \| none", "RESERV");
	zprop_register_number(ZFS_PROP_VOLSIZE, "volsize", 0, PROP_DEFAULT,
	ZFS_TYPE_SNAPSHOT \| ZFS_TYPE_VOLUME, "<size>", "VOLSIZE");
	zprop_register_number(ZFS_PROP_REFQUOTA, "refquota", 0, PROP_DEFAULT,
	ZFS_TYPE_FILESYSTEM, "<size> \| none", "REFQUOTA");
	zprop_register_number(ZFS_PROP_REFRESERVATION, "refreservation", 0,
	PROP_DEFAULT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"<size> \| none", "REFRESERV");
	zprop_register_number(ZFS_PROP_FILESYSTEM_LIMIT, "filesystem_limit",
	UINT64_MAX, PROP_DEFAULT, ZFS_TYPE_FILESYSTEM,
	"<count> \| none", "FSLIMIT");
	zprop_register_number(ZFS_PROP_SNAPSHOT_LIMIT, "snapshot_limit",
	UINT64_MAX, PROP_DEFAULT, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME,
	"<count> \| none", "SSLIMIT");

	/* inherit number properties */
	zprop_register_number(ZFS_PROP_RECORDSIZE, "recordsize",
	SPA_OLD_MAXBLOCKSIZE, PROP_INHERIT,
	ZFS_TYPE_FILESYSTEM, "512 to 1M, power of 2", "RECSIZE");
	zprop_register_number(ZFS_PROP_SPECIAL_SMALL_BLOCKS,
	"special_small_blocks", 0, PROP_INHERIT, ZFS_TYPE_FILESYSTEM,
	"zero or 512 to 1M, power of 2", "SPECIAL_SMALL_BLOCKS");

	/* hidden properties */
	zprop_register_hidden(ZFS_PROP_NUMCLONES, "numclones", PROP_TYPE_NUMBER,
	PROP_READONLY, ZFS_TYPE_SNAPSHOT, "NUMCLONES");
	zprop_register_hidden(ZFS_PROP_NAME, "name", PROP_TYPE_STRING,
	PROP_READONLY, ZFS_TYPE_DATASET \| ZFS_TYPE_BOOKMARK, "NAME");
	zprop_register_hidden(ZFS_PROP_ISCSIOPTIONS, "iscsioptions",
	PROP_TYPE_STRING, PROP_INHERIT, ZFS_TYPE_VOLUME, "ISCSIOPTIONS");
	zprop_register_hidden(ZFS_PROP_STMF_SHAREINFO, "stmf_sbd_lu",
	PROP_TYPE_STRING, PROP_INHERIT, ZFS_TYPE_VOLUME,
	"STMF_SBD_LU");
	zprop_register_hidden(ZFS_PROP_USERACCOUNTING, "useraccounting",
	PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_DATASET,
	"USERACCOUNTING");
	zprop_register_hidden(ZFS_PROP_UNIQUE, "unique", PROP_TYPE_NUMBER,
	PROP_READONLY, ZFS_TYPE_DATASET, "UNIQUE");
	zprop_register_hidden(ZFS_PROP_INCONSISTENT, "inconsistent",
	PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_DATASET, "INCONSISTENT");
	zprop_register_hidden(ZFS_PROP_IVSET_GUID, "ivsetguid",
	PROP_TYPE_NUMBER, PROP_READONLY,
	ZFS_TYPE_DATASET \| ZFS_TYPE_BOOKMARK, "IVSETGUID");
	zprop_register_hidden(ZFS_PROP_PREV_SNAP, "prevsnap", PROP_TYPE_STRING,
	PROP_READONLY, ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME, "PREVSNAP");
	zprop_register_hidden(ZFS_PROP_PBKDF2_SALT, "pbkdf2salt",
	PROP_TYPE_NUMBER, PROP_ONETIME_DEFAULT,
	ZFS_TYPE_FILESYSTEM \| ZFS_TYPE_VOLUME, "PBKDF2SALT");
	zprop_register_hidden(ZFS_PROP_KEY_GUID, "keyguid", PROP_TYPE_NUMBER,
	PROP_READONLY, ZFS_TYPE_DATASET, "KEYGUID");
	zprop_register_hidden(ZFS_PROP_REDACTED, "redacted", PROP_TYPE_NUMBER,
	PROP_READONLY, ZFS_TYPE_DATASET, "REDACTED");

	/*
	* Properties that are obsolete and not used. These are retained so
	* that we don't have to change the values of the zfs_prop_t enum, or
	* have NULL pointers in the zfs_prop_table[].
	*/
	zprop_register_hidden(ZFS_PROP_REMAPTXG, "remaptxg", PROP_TYPE_NUMBER,
	PROP_READONLY, ZFS_TYPE_DATASET, "REMAPTXG");

	/* oddball properties */
	zprop_register_impl(ZFS_PROP_CREATION, "creation", PROP_TYPE_NUMBER, 0,
	NULL, PROP_READONLY, ZFS_TYPE_DATASET \| ZFS_TYPE_BOOKMARK,
	"<date>", "CREATION", B_FALSE, B_TRUE, NULL);
	}

	boolean_t
	zfs_prop_delegatable(zfs_prop_t prop)
	{
	zprop_desc_t *pd = &zfs_prop_table[prop];

	/* The mlslabel property is never delegatable. */
	if (prop == ZFS_PROP_MLSLABEL)
	return (B_FALSE);

	return (pd->pd_attr != PROP_READONLY);
	}

	/*
	* Given a zfs dataset property name, returns the corresponding property ID.
	*/
	zfs_prop_t
	zfs_name_to_prop(const char *propname)
	{
	return (zprop_name_to_prop(propname, ZFS_TYPE_DATASET));
	}

	-/*
	- * For user property names, we allow all lowercase alphanumeric characters, plus
	- * a few useful punctuation characters.
	- */
	-static int
	-valid_char(char c)
	-{
	- return ((c >= 'a' && c <= 'z') \|\|
	- (c >= '0' && c <= '9') \|\|
	- c == '-' \|\| c == '_' \|\| c == '.' \|\| c == ':');
	-}
	-
	/*
	* Returns true if this is a valid user-defined property (one with a ':').
	*/
	boolean_t
	zfs_prop_user(const char *name)
	{
	int i;
	char c;
	boolean_t foundsep = B_FALSE;

	for (i = 0; i < strlen(name); i++) {
	c = name[i];
	- if (!valid_char(c))
	+ if (!zprop_valid_char(c))
	return (B_FALSE);
	if (c == ':')
	foundsep = B_TRUE;
	}

	if (!foundsep)
	return (B_FALSE);

	return (B_TRUE);
	}

	/*
	* Returns true if this is a valid userspace-type property (one with a '@').
	* Note that after the @, any character is valid (eg, another @, for SID
	* user@domain).
	*/
	boolean_t
	zfs_prop_userquota(const char *name)
	{
	zfs_userquota_prop_t prop;

	for (prop = 0; prop < ZFS_NUM_USERQUOTA_PROPS; prop++) {
	if (strncmp(name, zfs_userquota_prop_prefixes[prop],
	strlen(zfs_userquota_prop_prefixes[prop])) == 0) {
	return (B_TRUE);
	}
	}

	return (B_FALSE);
	}

	/*
	* Returns true if this is a valid written@ property.
	* Note that after the @, any character is valid (eg, another @, for
	* written@pool/fs@origin).
	*/
	boolean_t
	zfs_prop_written(const char *name)
	{
	static const char *prop_prefix = "written@";
	static const char *book_prefix = "written#";
	return (strncmp(name, prop_prefix, strlen(prop_prefix)) == 0 \|\|
	strncmp(name, book_prefix, strlen(book_prefix)) == 0);
	}

	/*
	* Tables of index types, plus functions to convert between the user view
	* (strings) and internal representation (uint64_t).
	*/
	int
	zfs_prop_string_to_index(zfs_prop_t prop, const char string, uint64_t index)
	{
	return (zprop_string_to_index(prop, string, index, ZFS_TYPE_DATASET));
	}

	int
	zfs_prop_index_to_string(zfs_prop_t prop, uint64_t index, const char **string)
	{
	return (zprop_index_to_string(prop, index, string, ZFS_TYPE_DATASET));
	}

	uint64_t
	zfs_prop_random_value(zfs_prop_t prop, uint64_t seed)
	{
	return (zprop_random_value(prop, seed, ZFS_TYPE_DATASET));
	}

	/*
	* Returns TRUE if the property applies to any of the given dataset types.
	*/
	boolean_t
	zfs_prop_valid_for_type(int prop, zfs_type_t types, boolean_t headcheck)
	{
	return (zprop_valid_for_type(prop, types, headcheck));
	}

	zprop_type_t
	zfs_prop_get_type(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_proptype);
	}

	/*
	* Returns TRUE if the property is readonly.
	*/
	boolean_t
	zfs_prop_readonly(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_attr == PROP_READONLY \|\|
	zfs_prop_table[prop].pd_attr == PROP_ONETIME \|\|
	zfs_prop_table[prop].pd_attr == PROP_ONETIME_DEFAULT);
	}

	/*
	* Returns TRUE if the property is visible (not hidden).
	*/
	boolean_t
	zfs_prop_visible(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_visible &&
	zfs_prop_table[prop].pd_zfs_mod_supported);
	}

	/*
	* Returns TRUE if the property is only allowed to be set once.
	*/
	boolean_t
	zfs_prop_setonce(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_attr == PROP_ONETIME \|\|
	zfs_prop_table[prop].pd_attr == PROP_ONETIME_DEFAULT);
	}

	const char *
	zfs_prop_default_string(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_strdefault);
	}

	uint64_t
	zfs_prop_default_numeric(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_numdefault);
	}

	/*
	* Given a dataset property ID, returns the corresponding name.
	* Assuming the zfs dataset property ID is valid.
	*/
	const char *
	zfs_prop_to_name(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_name);
	}

	/*
	* Returns TRUE if the property is inheritable.
	*/
	boolean_t
	zfs_prop_inheritable(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_attr == PROP_INHERIT \|\|
	zfs_prop_table[prop].pd_attr == PROP_ONETIME);
	}

	/*
	* Returns TRUE if property is one of the encryption properties that requires
	* a loaded encryption key to modify.
	*/
	boolean_t
	zfs_prop_encryption_key_param(zfs_prop_t prop)
	{
	/*
	* keylocation does not count as an encryption property. It can be
	* changed at will without needing the master keys.
	*/
	return (prop == ZFS_PROP_PBKDF2_SALT \|\| prop == ZFS_PROP_PBKDF2_ITERS \|\|
	prop == ZFS_PROP_KEYFORMAT);
	}

	/*
	* Helper function used by both kernelspace and userspace to check the
	* keylocation property. If encrypted is set, the keylocation must be valid
	* for an encrypted dataset.
	*/
	boolean_t
	zfs_prop_valid_keylocation(const char *str, boolean_t encrypted)
	{
	if (strcmp("none", str) == 0)
	return (!encrypted);
	else if (strcmp("prompt", str) == 0)
	return (B_TRUE);
	else if (strlen(str) > 8 && strncmp("file:///", str, 8) == 0)
	return (B_TRUE);
	else if (strlen(str) > 8 && strncmp("https://", str, 8) == 0)
	return (B_TRUE);
	else if (strlen(str) > 7 && strncmp("http://", str, 7) == 0)
	return (B_TRUE);

	return (B_FALSE);
	}


	#ifndef _KERNEL
	#include <libzfs.h>

	/*
	* Returns a string describing the set of acceptable values for the given
	* zfs property, or NULL if it cannot be set.
	*/
	const char *
	zfs_prop_values(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_values);
	}

	/*
	* Returns TRUE if this property is a string type. Note that index types
	* (compression, checksum) are treated as strings in userland, even though they
	* are stored numerically on disk.
	*/
	int
	zfs_prop_is_string(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_proptype == PROP_TYPE_STRING \|\|
	zfs_prop_table[prop].pd_proptype == PROP_TYPE_INDEX);
	}

	/*
	* Returns the column header for the given property. Used only in
	* 'zfs list -o', but centralized here with the other property information.
	*/
	const char *
	zfs_prop_column_name(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_colname);
	}

	/*
	* Returns whether the given property should be displayed right-justified for
	* 'zfs list'.
	*/
	boolean_t
	zfs_prop_align_right(zfs_prop_t prop)
	{
	return (zfs_prop_table[prop].pd_rightalign);
	}

	#endif

	#if defined(_KERNEL)

	#include <sys/simd.h>

	#if defined(HAVE_KERNEL_FPU_INTERNAL)
	union fpregs_state **zfs_kfpu_fpregs;
	EXPORT_SYMBOL(zfs_kfpu_fpregs);
	#endif /* HAVE_KERNEL_FPU_INTERNAL */

	static int __init
	zcommon_init(void)
	{
	int error = kfpu_init();
	if (error)
	return (error);

	fletcher_4_init();

	return (0);
	}

	static void __exit
	zcommon_fini(void)
	{
	fletcher_4_fini();
	kfpu_fini();
	}

	module_init_early(zcommon_init);
	module_exit(zcommon_fini);

	#endif

	ZFS_MODULE_DESCRIPTION("Generic ZFS support");
	ZFS_MODULE_AUTHOR(ZFS_META_AUTHOR);
	ZFS_MODULE_LICENSE(ZFS_META_LICENSE);
	ZFS_MODULE_VERSION(ZFS_META_VERSION "-" ZFS_META_RELEASE);

	/* zfs dataset property functions */
	EXPORT_SYMBOL(zfs_userquota_prop_prefixes);
	EXPORT_SYMBOL(zfs_prop_init);
	EXPORT_SYMBOL(zfs_prop_get_type);
	EXPORT_SYMBOL(zfs_prop_get_table);
	EXPORT_SYMBOL(zfs_prop_delegatable);
	EXPORT_SYMBOL(zfs_prop_visible);

	/* Dataset property functions shared between libzfs and kernel. */
	EXPORT_SYMBOL(zfs_prop_default_string);
	EXPORT_SYMBOL(zfs_prop_default_numeric);
	EXPORT_SYMBOL(zfs_prop_readonly);
	EXPORT_SYMBOL(zfs_prop_inheritable);
	EXPORT_SYMBOL(zfs_prop_encryption_key_param);
	EXPORT_SYMBOL(zfs_prop_valid_keylocation);
	EXPORT_SYMBOL(zfs_prop_setonce);
	EXPORT_SYMBOL(zfs_prop_to_name);
	EXPORT_SYMBOL(zfs_name_to_prop);
	EXPORT_SYMBOL(zfs_prop_user);
	EXPORT_SYMBOL(zfs_prop_userquota);
	EXPORT_SYMBOL(zfs_prop_index_to_string);
	EXPORT_SYMBOL(zfs_prop_string_to_index);
	EXPORT_SYMBOL(zfs_prop_valid_for_type);
	EXPORT_SYMBOL(zfs_prop_written);
	diff --git a/sys/contrib/openzfs/module/zcommon/zpool_prop.c b/sys/contrib/openzfs/module/zcommon/zpool_prop.c
	index 6299d371f25d..8e7a20e8efdc 100644
	--- a/sys/contrib/openzfs/module/zcommon/zpool_prop.c
	+++ b/sys/contrib/openzfs/module/zcommon/zpool_prop.c
	@@ -1,279 +1,529 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2012, 2018 by Delphix. All rights reserved.
	* Copyright (c) 2021, Colm Buckley <colm@tuatha.org>
	+ * Copyright (c) 2021, Klara Inc.
	*/

	#include <sys/zio.h>
	#include <sys/spa.h>
	#include <sys/zfs_acl.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/fs/zfs.h>

	#include "zfs_prop.h"

	#if !defined(_KERNEL)
	#include <stdlib.h>
	#include <string.h>
	#include <ctype.h>
	#endif

	static zprop_desc_t zpool_prop_table[ZPOOL_NUM_PROPS];
	+static zprop_desc_t vdev_prop_table[VDEV_NUM_PROPS];

	zprop_desc_t *
	zpool_prop_get_table(void)
	{
	return (zpool_prop_table);
	}

	void
	zpool_prop_init(void)
	{
	static zprop_index_t boolean_table[] = {
	{ "off", 0},
	{ "on", 1},
	{ NULL }
	};

	static zprop_index_t failuremode_table[] = {
	{ "wait", ZIO_FAILURE_MODE_WAIT },
	{ "continue", ZIO_FAILURE_MODE_CONTINUE },
	{ "panic", ZIO_FAILURE_MODE_PANIC },
	{ NULL }
	};

	/* string properties */
	zprop_register_string(ZPOOL_PROP_ALTROOT, "altroot", NULL, PROP_DEFAULT,
	ZFS_TYPE_POOL, "<path>", "ALTROOT");
	zprop_register_string(ZPOOL_PROP_BOOTFS, "bootfs", NULL, PROP_DEFAULT,
	ZFS_TYPE_POOL, "<filesystem>", "BOOTFS");
	zprop_register_string(ZPOOL_PROP_CACHEFILE, "cachefile", NULL,
	PROP_DEFAULT, ZFS_TYPE_POOL, "<file> \| none", "CACHEFILE");
	zprop_register_string(ZPOOL_PROP_COMMENT, "comment", NULL,
	PROP_DEFAULT, ZFS_TYPE_POOL, "<comment-string>", "COMMENT");
	zprop_register_string(ZPOOL_PROP_COMPATIBILITY, "compatibility",
	"off", PROP_DEFAULT, ZFS_TYPE_POOL,
	"<file[,file...]> \| off \| legacy", "COMPATIBILITY");

	/* readonly number properties */
	zprop_register_number(ZPOOL_PROP_SIZE, "size", 0, PROP_READONLY,
	ZFS_TYPE_POOL, "<size>", "SIZE");
	zprop_register_number(ZPOOL_PROP_FREE, "free", 0, PROP_READONLY,
	ZFS_TYPE_POOL, "<size>", "FREE");
	zprop_register_number(ZPOOL_PROP_FREEING, "freeing", 0, PROP_READONLY,
	ZFS_TYPE_POOL, "<size>", "FREEING");
	zprop_register_number(ZPOOL_PROP_CHECKPOINT, "checkpoint", 0,
	PROP_READONLY, ZFS_TYPE_POOL, "<size>", "CKPOINT");
	zprop_register_number(ZPOOL_PROP_LEAKED, "leaked", 0, PROP_READONLY,
	ZFS_TYPE_POOL, "<size>", "LEAKED");
	zprop_register_number(ZPOOL_PROP_ALLOCATED, "allocated", 0,
	PROP_READONLY, ZFS_TYPE_POOL, "<size>", "ALLOC");
	zprop_register_number(ZPOOL_PROP_EXPANDSZ, "expandsize", 0,
	PROP_READONLY, ZFS_TYPE_POOL, "<size>", "EXPANDSZ");
	zprop_register_number(ZPOOL_PROP_FRAGMENTATION, "fragmentation", 0,
	PROP_READONLY, ZFS_TYPE_POOL, "<percent>", "FRAG");
	zprop_register_number(ZPOOL_PROP_CAPACITY, "capacity", 0, PROP_READONLY,
	ZFS_TYPE_POOL, "<size>", "CAP");
	zprop_register_number(ZPOOL_PROP_GUID, "guid", 0, PROP_READONLY,
	ZFS_TYPE_POOL, "<guid>", "GUID");
	zprop_register_number(ZPOOL_PROP_LOAD_GUID, "load_guid", 0,
	PROP_READONLY, ZFS_TYPE_POOL, "<load_guid>", "LOAD_GUID");
	zprop_register_number(ZPOOL_PROP_HEALTH, "health", 0, PROP_READONLY,
	ZFS_TYPE_POOL, "<state>", "HEALTH");
	zprop_register_number(ZPOOL_PROP_DEDUPRATIO, "dedupratio", 0,
	PROP_READONLY, ZFS_TYPE_POOL, "<1.00x or higher if deduped>",
	"DEDUP");

	/* default number properties */
	zprop_register_number(ZPOOL_PROP_VERSION, "version", SPA_VERSION,
	PROP_DEFAULT, ZFS_TYPE_POOL, "<version>", "VERSION");
	zprop_register_number(ZPOOL_PROP_ASHIFT, "ashift", 0, PROP_DEFAULT,
	ZFS_TYPE_POOL, "<ashift, 9-16, or 0=default>", "ASHIFT");

	/* default index (boolean) properties */
	zprop_register_index(ZPOOL_PROP_DELEGATION, "delegation", 1,
	PROP_DEFAULT, ZFS_TYPE_POOL, "on \| off", "DELEGATION",
	boolean_table);
	zprop_register_index(ZPOOL_PROP_AUTOREPLACE, "autoreplace", 0,
	PROP_DEFAULT, ZFS_TYPE_POOL, "on \| off", "REPLACE", boolean_table);
	zprop_register_index(ZPOOL_PROP_LISTSNAPS, "listsnapshots", 0,
	PROP_DEFAULT, ZFS_TYPE_POOL, "on \| off", "LISTSNAPS",
	boolean_table);
	zprop_register_index(ZPOOL_PROP_AUTOEXPAND, "autoexpand", 0,
	PROP_DEFAULT, ZFS_TYPE_POOL, "on \| off", "EXPAND", boolean_table);
	zprop_register_index(ZPOOL_PROP_READONLY, "readonly", 0,
	PROP_DEFAULT, ZFS_TYPE_POOL, "on \| off", "RDONLY", boolean_table);
	zprop_register_index(ZPOOL_PROP_MULTIHOST, "multihost", 0,
	PROP_DEFAULT, ZFS_TYPE_POOL, "on \| off", "MULTIHOST",
	boolean_table);

	/* default index properties */
	zprop_register_index(ZPOOL_PROP_FAILUREMODE, "failmode",
	ZIO_FAILURE_MODE_WAIT, PROP_DEFAULT, ZFS_TYPE_POOL,
	"wait \| continue \| panic", "FAILMODE", failuremode_table);
	zprop_register_index(ZPOOL_PROP_AUTOTRIM, "autotrim",
	SPA_AUTOTRIM_DEFAULT, PROP_DEFAULT, ZFS_TYPE_POOL,
	"on \| off", "AUTOTRIM", boolean_table);

	/* hidden properties */
	zprop_register_hidden(ZPOOL_PROP_NAME, "name", PROP_TYPE_STRING,
	PROP_READONLY, ZFS_TYPE_POOL, "NAME");
	zprop_register_hidden(ZPOOL_PROP_MAXBLOCKSIZE, "maxblocksize",
	PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_POOL, "MAXBLOCKSIZE");
	zprop_register_hidden(ZPOOL_PROP_TNAME, "tname", PROP_TYPE_STRING,
	PROP_ONETIME, ZFS_TYPE_POOL, "TNAME");
	zprop_register_hidden(ZPOOL_PROP_MAXDNODESIZE, "maxdnodesize",
	PROP_TYPE_NUMBER, PROP_READONLY, ZFS_TYPE_POOL, "MAXDNODESIZE");
	zprop_register_hidden(ZPOOL_PROP_DEDUPDITTO, "dedupditto",
	PROP_TYPE_NUMBER, PROP_DEFAULT, ZFS_TYPE_POOL, "DEDUPDITTO");
	}

	/*
	* Given a property name and its type, returns the corresponding property ID.
	*/
	zpool_prop_t
	zpool_name_to_prop(const char *propname)
	{
	return (zprop_name_to_prop(propname, ZFS_TYPE_POOL));
	}

	/*
	* Given a pool property ID, returns the corresponding name.
	* Assuming the pool property ID is valid.
	*/
	const char *
	zpool_prop_to_name(zpool_prop_t prop)
	{
	return (zpool_prop_table[prop].pd_name);
	}

	zprop_type_t
	zpool_prop_get_type(zpool_prop_t prop)
	{
	return (zpool_prop_table[prop].pd_proptype);
	}

	boolean_t
	zpool_prop_readonly(zpool_prop_t prop)
	{
	return (zpool_prop_table[prop].pd_attr == PROP_READONLY);
	}

	boolean_t
	zpool_prop_setonce(zpool_prop_t prop)
	{
	return (zpool_prop_table[prop].pd_attr == PROP_ONETIME);
	}

	const char *
	zpool_prop_default_string(zpool_prop_t prop)
	{
	return (zpool_prop_table[prop].pd_strdefault);
	}

	uint64_t
	zpool_prop_default_numeric(zpool_prop_t prop)
	{
	return (zpool_prop_table[prop].pd_numdefault);
	}

	/*
	* Returns true if this is a valid feature@ property.
	*/
	boolean_t
	zpool_prop_feature(const char *name)
	{
	static const char *prefix = "feature@";
	return (strncmp(name, prefix, strlen(prefix)) == 0);
	}

	/*
	* Returns true if this is a valid unsupported@ property.
	*/
	boolean_t
	zpool_prop_unsupported(const char *name)
	{
	static const char *prefix = "unsupported@";
	return (strncmp(name, prefix, strlen(prefix)) == 0);
	}

	int
	zpool_prop_string_to_index(zpool_prop_t prop, const char *string,
	uint64_t *index)
	{
	return (zprop_string_to_index(prop, string, index, ZFS_TYPE_POOL));
	}

	int
	zpool_prop_index_to_string(zpool_prop_t prop, uint64_t index,
	const char **string)
	{
	return (zprop_index_to_string(prop, index, string, ZFS_TYPE_POOL));
	}

	uint64_t
	zpool_prop_random_value(zpool_prop_t prop, uint64_t seed)
	{
	return (zprop_random_value(prop, seed, ZFS_TYPE_POOL));
	}

	#ifndef _KERNEL
	#include <libzfs.h>

	const char *
	zpool_prop_values(zpool_prop_t prop)
	{
	return (zpool_prop_table[prop].pd_values);
	}

	const char *
	zpool_prop_column_name(zpool_prop_t prop)
	{
	return (zpool_prop_table[prop].pd_colname);
	}

	boolean_t
	zpool_prop_align_right(zpool_prop_t prop)
	{
	return (zpool_prop_table[prop].pd_rightalign);
	}
	#endif

	+zprop_desc_t *
	+vdev_prop_get_table(void)
	+{
	+ return (vdev_prop_table);
	+}
	+
	+void
	+vdev_prop_init(void)
	+{
	+ static zprop_index_t boolean_table[] = {
	+ { "off", 0},
	+ { "on", 1},
	+ { NULL }
	+ };
	+ static zprop_index_t boolean_na_table[] = {
	+ { "off", 0},
	+ { "on", 1},
	+ { "-", 2}, /* ZPROP_BOOLEAN_NA */
	+ { NULL }
	+ };
	+
	+ /* string properties */
	+ zprop_register_string(VDEV_PROP_COMMENT, "comment", NULL,
	+ PROP_DEFAULT, ZFS_TYPE_VDEV, "<comment-string>", "COMMENT");
	+ zprop_register_string(VDEV_PROP_PATH, "path", NULL,
	+ PROP_DEFAULT, ZFS_TYPE_VDEV, "<device-path>", "PATH");
	+ zprop_register_string(VDEV_PROP_DEVID, "devid", NULL,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<devid>", "DEVID");
	+ zprop_register_string(VDEV_PROP_PHYS_PATH, "physpath", NULL,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<physpath>", "PHYSPATH");
	+ zprop_register_string(VDEV_PROP_ENC_PATH, "encpath", NULL,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<encpath>", "ENCPATH");
	+ zprop_register_string(VDEV_PROP_FRU, "fru", NULL,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<fru>", "FRU");
	+ zprop_register_string(VDEV_PROP_PARENT, "parent", NULL,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<parent>", "PARENT");
	+ zprop_register_string(VDEV_PROP_CHILDREN, "children", NULL,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<child[,...]>", "CHILDREN");
	+
	+ /* readonly number properties */
	+ zprop_register_number(VDEV_PROP_SIZE, "size", 0, PROP_READONLY,
	+ ZFS_TYPE_VDEV, "<size>", "SIZE");
	+ zprop_register_number(VDEV_PROP_FREE, "free", 0, PROP_READONLY,
	+ ZFS_TYPE_VDEV, "<size>", "FREE");
	+ zprop_register_number(VDEV_PROP_ALLOCATED, "allocated", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<size>", "ALLOC");
	+ zprop_register_number(VDEV_PROP_EXPANDSZ, "expandsize", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<size>", "EXPANDSZ");
	+ zprop_register_number(VDEV_PROP_FRAGMENTATION, "fragmentation", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<percent>", "FRAG");
	+ zprop_register_number(VDEV_PROP_CAPACITY, "capacity", 0, PROP_READONLY,
	+ ZFS_TYPE_VDEV, "<size>", "CAP");
	+ zprop_register_number(VDEV_PROP_GUID, "guid", 0, PROP_READONLY,
	+ ZFS_TYPE_VDEV, "<guid>", "GUID");
	+ zprop_register_number(VDEV_PROP_STATE, "state", 0, PROP_READONLY,
	+ ZFS_TYPE_VDEV, "<state>", "STATE");
	+ zprop_register_number(VDEV_PROP_BOOTSIZE, "bootsize", 0, PROP_READONLY,
	+ ZFS_TYPE_VDEV, "<size>", "BOOTSIZE");
	+ zprop_register_number(VDEV_PROP_ASIZE, "asize", 0, PROP_READONLY,
	+ ZFS_TYPE_VDEV, "<asize>", "ASIZE");
	+ zprop_register_number(VDEV_PROP_PSIZE, "psize", 0, PROP_READONLY,
	+ ZFS_TYPE_VDEV, "<psize>", "PSIZE");
	+ zprop_register_number(VDEV_PROP_ASHIFT, "ashift", 0, PROP_READONLY,
	+ ZFS_TYPE_VDEV, "<ashift>", "ASHIFT");
	+ zprop_register_number(VDEV_PROP_PARITY, "parity", 0, PROP_READONLY,
	+ ZFS_TYPE_VDEV, "<parity>", "PARITY");
	+ zprop_register_number(VDEV_PROP_NUMCHILDREN, "numchildren", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<number-of-children>", "NUMCHILD");
	+ zprop_register_number(VDEV_PROP_READ_ERRORS, "read_errors", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<errors>", "RDERR");
	+ zprop_register_number(VDEV_PROP_WRITE_ERRORS, "write_errors", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<errors>", "WRERR");
	+ zprop_register_number(VDEV_PROP_CHECKSUM_ERRORS, "checksum_errors", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<errors>", "CKERR");
	+ zprop_register_number(VDEV_PROP_INITIALIZE_ERRORS,
	+ "initialize_errors", 0, PROP_READONLY, ZFS_TYPE_VDEV, "<errors>",
	+ "INITERR");
	+ zprop_register_number(VDEV_PROP_OPS_NULL, "null_ops", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<operations>", "NULLOP");
	+ zprop_register_number(VDEV_PROP_OPS_READ, "read_ops", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<operations>", "READOP");
	+ zprop_register_number(VDEV_PROP_OPS_WRITE, "write_ops", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<operations>", "WRITEOP");
	+ zprop_register_number(VDEV_PROP_OPS_FREE, "free_ops", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<operations>", "FREEOP");
	+ zprop_register_number(VDEV_PROP_OPS_CLAIM, "claim_ops", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<operations>", "CLAIMOP");
	+ zprop_register_number(VDEV_PROP_OPS_TRIM, "trim_ops", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<operations>", "TRIMOP");
	+ zprop_register_number(VDEV_PROP_BYTES_NULL, "null_bytes", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<bytes>", "NULLBYTE");
	+ zprop_register_number(VDEV_PROP_BYTES_READ, "read_bytes", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<bytes>", "READBYTE");
	+ zprop_register_number(VDEV_PROP_BYTES_WRITE, "write_bytes", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<bytes>", "WRITEBYTE");
	+ zprop_register_number(VDEV_PROP_BYTES_FREE, "free_bytes", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<bytes>", "FREEBYTE");
	+ zprop_register_number(VDEV_PROP_BYTES_CLAIM, "claim_bytes", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<bytes>", "CLAIMBYTE");
	+ zprop_register_number(VDEV_PROP_BYTES_TRIM, "trim_bytes", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "<bytes>", "TRIMBYTE");
	+
	+ /* default numeric properties */
	+
	+ /* default index (boolean) properties */
	+ zprop_register_index(VDEV_PROP_REMOVING, "removing", 0,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "on \| off", "REMOVING",
	+ boolean_table);
	+ zprop_register_index(VDEV_PROP_ALLOCATING, "allocating", 1,
	+ PROP_DEFAULT, ZFS_TYPE_VDEV, "on \| off", "ALLOCATING",
	+ boolean_na_table);
	+
	+ /* default index properties */
	+
	+ /* hidden properties */
	+ zprop_register_hidden(VDEV_PROP_NAME, "name", PROP_TYPE_STRING,
	+ PROP_READONLY, ZFS_TYPE_VDEV, "NAME");
	+}
	+
	+/*
	+ * Given a property name and its type, returns the corresponding property ID.
	+ */
	+vdev_prop_t
	+vdev_name_to_prop(const char *propname)
	+{
	+ return (zprop_name_to_prop(propname, ZFS_TYPE_VDEV));
	+}
	+
	+/*
	+ * Returns true if this is a valid user-defined property (one with a ':').
	+ */
	+boolean_t
	+vdev_prop_user(const char *name)
	+{
	+ int i;
	+ char c;
	+ boolean_t foundsep = B_FALSE;
	+
	+ for (i = 0; i < strlen(name); i++) {
	+ c = name[i];
	+ if (!zprop_valid_char(c))
	+ return (B_FALSE);
	+ if (c == ':')
	+ foundsep = B_TRUE;
	+ }
	+
	+ return (foundsep);
	+}
	+
	+/*
	+ * Given a pool property ID, returns the corresponding name.
	+ * Assuming the pool property ID is valid.
	+ */
	+const char *
	+vdev_prop_to_name(vdev_prop_t prop)
	+{
	+ return (vdev_prop_table[prop].pd_name);
	+}
	+
	+zprop_type_t
	+vdev_prop_get_type(vdev_prop_t prop)
	+{
	+ return (vdev_prop_table[prop].pd_proptype);
	+}
	+
	+boolean_t
	+vdev_prop_readonly(vdev_prop_t prop)
	+{
	+ return (vdev_prop_table[prop].pd_attr == PROP_READONLY);
	+}
	+
	+const char *
	+vdev_prop_default_string(vdev_prop_t prop)
	+{
	+ return (vdev_prop_table[prop].pd_strdefault);
	+}
	+
	+uint64_t
	+vdev_prop_default_numeric(vdev_prop_t prop)
	+{
	+ return (vdev_prop_table[prop].pd_numdefault);
	+}
	+
	+int
	+vdev_prop_string_to_index(vdev_prop_t prop, const char *string,
	+ uint64_t *index)
	+{
	+ return (zprop_string_to_index(prop, string, index, ZFS_TYPE_VDEV));
	+}
	+
	+int
	+vdev_prop_index_to_string(vdev_prop_t prop, uint64_t index,
	+ const char **string)
	+{
	+ return (zprop_index_to_string(prop, index, string, ZFS_TYPE_VDEV));
	+}
	+
	+/*
	+ * Returns true if this is a valid vdev property.
	+ */
	+boolean_t
	+zpool_prop_vdev(const char *name)
	+{
	+ return (vdev_name_to_prop(name) != VDEV_PROP_INVAL);
	+}
	+
	+uint64_t
	+vdev_prop_random_value(vdev_prop_t prop, uint64_t seed)
	+{
	+ return (zprop_random_value(prop, seed, ZFS_TYPE_VDEV));
	+}
	+
	+#ifndef _KERNEL
	+const char *
	+vdev_prop_values(vdev_prop_t prop)
	+{
	+ return (vdev_prop_table[prop].pd_values);
	+}
	+
	+const char *
	+vdev_prop_column_name(vdev_prop_t prop)
	+{
	+ return (vdev_prop_table[prop].pd_colname);
	+}
	+
	+boolean_t
	+vdev_prop_align_right(vdev_prop_t prop)
	+{
	+ return (vdev_prop_table[prop].pd_rightalign);
	+}
	+#endif
	+
	#if defined(_KERNEL)
	/* zpool property functions */
	EXPORT_SYMBOL(zpool_prop_init);
	EXPORT_SYMBOL(zpool_prop_get_type);
	EXPORT_SYMBOL(zpool_prop_get_table);

	+/* vdev property functions */
	+EXPORT_SYMBOL(vdev_prop_init);
	+EXPORT_SYMBOL(vdev_prop_get_type);
	+EXPORT_SYMBOL(vdev_prop_get_table);
	+
	/* Pool property functions shared between libzfs and kernel. */
	EXPORT_SYMBOL(zpool_name_to_prop);
	EXPORT_SYMBOL(zpool_prop_to_name);
	EXPORT_SYMBOL(zpool_prop_default_string);
	EXPORT_SYMBOL(zpool_prop_default_numeric);
	EXPORT_SYMBOL(zpool_prop_readonly);
	EXPORT_SYMBOL(zpool_prop_feature);
	EXPORT_SYMBOL(zpool_prop_unsupported);
	EXPORT_SYMBOL(zpool_prop_index_to_string);
	EXPORT_SYMBOL(zpool_prop_string_to_index);
	+EXPORT_SYMBOL(zpool_prop_vdev);
	+
	+/* vdev property functions shared between libzfs and kernel. */
	+EXPORT_SYMBOL(vdev_name_to_prop);
	+EXPORT_SYMBOL(vdev_prop_user);
	+EXPORT_SYMBOL(vdev_prop_to_name);
	+EXPORT_SYMBOL(vdev_prop_default_string);
	+EXPORT_SYMBOL(vdev_prop_default_numeric);
	+EXPORT_SYMBOL(vdev_prop_readonly);
	+EXPORT_SYMBOL(vdev_prop_index_to_string);
	+EXPORT_SYMBOL(vdev_prop_string_to_index);
	#endif
	diff --git a/sys/contrib/openzfs/module/zcommon/zprop_common.c b/sys/contrib/openzfs/module/zcommon/zprop_common.c
	index faab9d9a74fd..17a48361f96e 100644
	--- a/sys/contrib/openzfs/module/zcommon/zprop_common.c
	+++ b/sys/contrib/openzfs/module/zcommon/zprop_common.c
	@@ -1,480 +1,500 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright 2010 Sun Microsystems, Inc. All rights reserved.
	* Use is subject to license terms.
	*/
	/*
	* Copyright (c) 2012 by Delphix. All rights reserved.
	*/

	/*
	* Common routines used by zfs and zpool property management.
	*/

	#include <sys/zio.h>
	#include <sys/spa.h>
	#include <sys/zfs_acl.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/zfs_sysfs.h>
	#include <sys/zfs_znode.h>
	#include <sys/fs/zfs.h>

	#include "zfs_prop.h"
	#include "zfs_deleg.h"

	#if !defined(_KERNEL)
	#include <stdlib.h>
	#include <string.h>
	#include <ctype.h>
	#include <sys/stat.h>
	#endif

	static zprop_desc_t *
	zprop_get_proptable(zfs_type_t type)
	{
	if (type == ZFS_TYPE_POOL)
	return (zpool_prop_get_table());
	+ else if (type == ZFS_TYPE_VDEV)
	+ return (vdev_prop_get_table());
	else
	return (zfs_prop_get_table());
	}

	static int
	zprop_get_numprops(zfs_type_t type)
	{
	if (type == ZFS_TYPE_POOL)
	return (ZPOOL_NUM_PROPS);
	+ else if (type == ZFS_TYPE_VDEV)
	+ return (VDEV_NUM_PROPS);
	else
	return (ZFS_NUM_PROPS);
	}

	static boolean_t
	zfs_mod_supported_prop(const char *name, zfs_type_t type)
	{
	/*
	* The zfs module spa_feature_table[], whether in-kernel or in libzpool,
	* always supports all the properties. libzfs needs to query the running
	* module, via sysfs, to determine which properties are supported.
	*
	* The equivalent _can_ be done on FreeBSD by way of the sysctl
	* tree, but this has not been done yet.
	*/
	#if defined(_KERNEL) \|\| defined(LIB_ZPOOL_BUILD) \|\| defined(__FreeBSD__)
	return (B_TRUE);
	#else
	return (zfs_mod_supported(type == ZFS_TYPE_POOL ?
	- ZFS_SYSFS_POOL_PROPERTIES : ZFS_SYSFS_DATASET_PROPERTIES, name));
	+ ZFS_SYSFS_POOL_PROPERTIES : (type == ZFS_TYPE_VDEV ?
	+ ZFS_SYSFS_VDEV_PROPERTIES : ZFS_SYSFS_DATASET_PROPERTIES), name));
	#endif
	}

	void
	zprop_register_impl(int prop, const char *name, zprop_type_t type,
	uint64_t numdefault, const char *strdefault, zprop_attr_t attr,
	int objset_types, const char values, const char colname,
	boolean_t rightalign, boolean_t visible, const zprop_index_t *idx_tbl)
	{
	zprop_desc_t *prop_tbl = zprop_get_proptable(objset_types);
	zprop_desc_t *pd;

	pd = &prop_tbl[prop];

	ASSERT(pd->pd_name == NULL \|\| pd->pd_name == name);
	ASSERT(name != NULL);
	ASSERT(colname != NULL);

	pd->pd_name = name;
	pd->pd_propnum = prop;
	pd->pd_proptype = type;
	pd->pd_numdefault = numdefault;
	pd->pd_strdefault = strdefault;
	pd->pd_attr = attr;
	pd->pd_types = objset_types;
	pd->pd_values = values;
	pd->pd_colname = colname;
	pd->pd_rightalign = rightalign;
	pd->pd_visible = visible;
	pd->pd_zfs_mod_supported = zfs_mod_supported_prop(name, objset_types);
	pd->pd_table = idx_tbl;
	pd->pd_table_size = 0;
	while (idx_tbl && (idx_tbl++)->pi_name != NULL)
	pd->pd_table_size++;
	}

	void
	zprop_register_string(int prop, const char name, const char def,
	zprop_attr_t attr, int objset_types, const char *values,
	const char *colname)
	{
	zprop_register_impl(prop, name, PROP_TYPE_STRING, 0, def, attr,
	objset_types, values, colname, B_FALSE, B_TRUE, NULL);

	}

	void
	zprop_register_number(int prop, const char *name, uint64_t def,
	zprop_attr_t attr, int objset_types, const char *values,
	const char *colname)
	{
	zprop_register_impl(prop, name, PROP_TYPE_NUMBER, def, NULL, attr,
	objset_types, values, colname, B_TRUE, B_TRUE, NULL);
	}

	void
	zprop_register_index(int prop, const char *name, uint64_t def,
	zprop_attr_t attr, int objset_types, const char *values,
	const char colname, const zprop_index_t idx_tbl)
	{
	zprop_register_impl(prop, name, PROP_TYPE_INDEX, def, NULL, attr,
	objset_types, values, colname, B_FALSE, B_TRUE, idx_tbl);
	}

	void
	zprop_register_hidden(int prop, const char *name, zprop_type_t type,
	zprop_attr_t attr, int objset_types, const char *colname)
	{
	zprop_register_impl(prop, name, type, 0, NULL, attr,
	objset_types, NULL, colname,
	type == PROP_TYPE_NUMBER, B_FALSE, NULL);
	}


	/*
	* A comparison function we can use to order indexes into property tables.
	*/
	static int
	zprop_compare(const void arg1, const void arg2)
	{
	const zprop_desc_t p1 = ((zprop_desc_t **)arg1);
	const zprop_desc_t p2 = ((zprop_desc_t **)arg2);
	boolean_t p1ro, p2ro;

	p1ro = (p1->pd_attr == PROP_READONLY);
	p2ro = (p2->pd_attr == PROP_READONLY);

	if (p1ro == p2ro)
	return (strcmp(p1->pd_name, p2->pd_name));

	return (p1ro ? -1 : 1);
	}

	/*
	* Iterate over all properties in the given property table, calling back
	* into the specified function for each property. We will continue to
	* iterate until we either reach the end or the callback function returns
	* something other than ZPROP_CONT.
	*/
	int
	zprop_iter_common(zprop_func func, void *cb, boolean_t show_all,
	boolean_t ordered, zfs_type_t type)
	{
	int i, num_props, size, prop;
	zprop_desc_t *prop_tbl;
	zprop_desc_t **order;

	prop_tbl = zprop_get_proptable(type);
	num_props = zprop_get_numprops(type);
	size = num_props * sizeof (zprop_desc_t *);

	#if defined(_KERNEL)
	order = kmem_alloc(size, KM_SLEEP);
	#else
	if ((order = malloc(size)) == NULL)
	return (ZPROP_CONT);
	#endif

	for (int j = 0; j < num_props; j++)
	order[j] = &prop_tbl[j];

	if (ordered) {
	qsort((void )order, num_props, sizeof (zprop_desc_t ),
	zprop_compare);
	}

	prop = ZPROP_CONT;
	for (i = 0; i < num_props; i++) {
	if ((order[i]->pd_visible \|\| show_all) &&
	order[i]->pd_zfs_mod_supported &&
	(func(order[i]->pd_propnum, cb) != ZPROP_CONT)) {
	prop = order[i]->pd_propnum;
	break;
	}
	}

	#if defined(_KERNEL)
	kmem_free(order, size);
	#else
	free(order);
	#endif
	return (prop);
	}

	static boolean_t
	propname_match(const char p, size_t len, zprop_desc_t prop_entry)
	{
	const char *propname = prop_entry->pd_name;
	#ifndef _KERNEL
	const char *colname = prop_entry->pd_colname;
	int c;
	#endif

	+ ASSERT(propname != NULL);
	+
	if (len == strlen(propname) &&
	strncmp(p, propname, len) == 0)
	return (B_TRUE);

	#ifndef _KERNEL
	if (colname == NULL \|\| len != strlen(colname))
	return (B_FALSE);

	for (c = 0; c < len; c++)
	if (p[c] != tolower(colname[c]))
	break;

	return (colname[c] == '\0');
	#else
	return (B_FALSE);
	#endif
	}

	typedef struct name_to_prop_cb {
	const char *propname;
	zprop_desc_t *prop_tbl;
	} name_to_prop_cb_t;

	static int
	zprop_name_to_prop_cb(int prop, void *cb_data)
	{
	name_to_prop_cb_t *data = cb_data;

	if (propname_match(data->propname, strlen(data->propname),
	&data->prop_tbl[prop]))
	return (prop);

	return (ZPROP_CONT);
	}

	int
	zprop_name_to_prop(const char *propname, zfs_type_t type)
	{
	int prop;
	name_to_prop_cb_t cb_data;

	cb_data.propname = propname;
	cb_data.prop_tbl = zprop_get_proptable(type);

	prop = zprop_iter_common(zprop_name_to_prop_cb, &cb_data,
	B_TRUE, B_FALSE, type);

	return (prop == ZPROP_CONT ? ZPROP_INVAL : prop);
	}

	int
	zprop_string_to_index(int prop, const char string, uint64_t index,
	zfs_type_t type)
	{
	zprop_desc_t *prop_tbl;
	const zprop_index_t *idx_tbl;
	int i;

	if (prop == ZPROP_INVAL \|\| prop == ZPROP_CONT)
	return (-1);

	ASSERT(prop < zprop_get_numprops(type));
	prop_tbl = zprop_get_proptable(type);
	if ((idx_tbl = prop_tbl[prop].pd_table) == NULL)
	return (-1);

	for (i = 0; idx_tbl[i].pi_name != NULL; i++) {
	if (strcmp(string, idx_tbl[i].pi_name) == 0) {
	*index = idx_tbl[i].pi_value;
	return (0);
	}
	}

	return (-1);
	}

	int
	zprop_index_to_string(int prop, uint64_t index, const char **string,
	zfs_type_t type)
	{
	zprop_desc_t *prop_tbl;
	const zprop_index_t *idx_tbl;
	int i;

	if (prop == ZPROP_INVAL \|\| prop == ZPROP_CONT)
	return (-1);

	ASSERT(prop < zprop_get_numprops(type));
	prop_tbl = zprop_get_proptable(type);
	if ((idx_tbl = prop_tbl[prop].pd_table) == NULL)
	return (-1);

	for (i = 0; idx_tbl[i].pi_name != NULL; i++) {
	if (idx_tbl[i].pi_value == index) {
	*string = idx_tbl[i].pi_name;
	return (0);
	}
	}

	return (-1);
	}

	/*
	* Return a random valid property value. Used by ztest.
	*/
	uint64_t
	zprop_random_value(int prop, uint64_t seed, zfs_type_t type)
	{
	zprop_desc_t *prop_tbl;
	const zprop_index_t *idx_tbl;

	ASSERT((uint_t)prop < zprop_get_numprops(type));
	prop_tbl = zprop_get_proptable(type);
	idx_tbl = prop_tbl[prop].pd_table;

	if (idx_tbl == NULL)
	return (seed);

	return (idx_tbl[seed % prop_tbl[prop].pd_table_size].pi_value);
	}

	const char *
	zprop_values(int prop, zfs_type_t type)
	{
	zprop_desc_t *prop_tbl;

	ASSERT(prop != ZPROP_INVAL && prop != ZPROP_CONT);
	ASSERT(prop < zprop_get_numprops(type));

	prop_tbl = zprop_get_proptable(type);

	return (prop_tbl[prop].pd_values);
	}

	/*
	* Returns TRUE if the property applies to any of the given dataset types.
	*
	* If headcheck is set, the check is being made against the head dataset
	* type of a snapshot which requires to return B_TRUE when the property
	* is only valid for snapshots.
	*/
	boolean_t
	zprop_valid_for_type(int prop, zfs_type_t type, boolean_t headcheck)
	{
	zprop_desc_t *prop_tbl;

	if (prop == ZPROP_INVAL \|\| prop == ZPROP_CONT)
	return (B_FALSE);

	ASSERT(prop < zprop_get_numprops(type));
	prop_tbl = zprop_get_proptable(type);
	if (headcheck && prop_tbl[prop].pd_types == ZFS_TYPE_SNAPSHOT)
	return (B_TRUE);
	return ((prop_tbl[prop].pd_types & type) != 0);
	}

	+/*
	+ * For user property names, we allow all lowercase alphanumeric characters, plus
	+ * a few useful punctuation characters.
	+ */
	+int
	+zprop_valid_char(char c)
	+{
	+ return ((c >= 'a' && c <= 'z') \|\|
	+ (c >= '0' && c <= '9') \|\|
	+ c == '-' \|\| c == '_' \|\| c == '.' \|\| c == ':');
	+}
	+
	#ifndef _KERNEL

	/*
	* Determines the minimum width for the column, and indicates whether it's fixed
	* or not. Only string columns are non-fixed.
	*/
	size_t
	zprop_width(int prop, boolean_t *fixed, zfs_type_t type)
	{
	zprop_desc_t prop_tbl, pd;
	const zprop_index_t *idx;
	size_t ret;
	int i;

	ASSERT(prop != ZPROP_INVAL && prop != ZPROP_CONT);
	ASSERT(prop < zprop_get_numprops(type));

	prop_tbl = zprop_get_proptable(type);
	pd = &prop_tbl[prop];

	*fixed = B_TRUE;

	/*
	* Start with the width of the column name.
	*/
	ret = strlen(pd->pd_colname);

	/*
	* For fixed-width values, make sure the width is large enough to hold
	* any possible value.
	*/
	switch (pd->pd_proptype) {
	case PROP_TYPE_NUMBER:
	/*
	* The maximum length of a human-readable number is 5 characters
	* ("20.4M", for example).
	*/
	if (ret < 5)
	ret = 5;
	/*
	* 'creation' is handled specially because it's a number
	* internally, but displayed as a date string.
	*/
	if (prop == ZFS_PROP_CREATION)
	*fixed = B_FALSE;
	/*
	* 'health' is handled specially because it's a number
	* internally, but displayed as a fixed 8 character string.
	*/
	if (prop == ZPOOL_PROP_HEALTH)
	ret = 8;
	break;
	case PROP_TYPE_INDEX:
	idx = prop_tbl[prop].pd_table;
	for (i = 0; idx[i].pi_name != NULL; i++) {
	if (strlen(idx[i].pi_name) > ret)
	ret = strlen(idx[i].pi_name);
	}
	break;

	case PROP_TYPE_STRING:
	*fixed = B_FALSE;
	break;
	}

	return (ret);
	}

	#endif

	#if defined(_KERNEL)
	/* Common routines to initialize property tables */
	EXPORT_SYMBOL(zprop_register_impl);
	EXPORT_SYMBOL(zprop_register_string);
	EXPORT_SYMBOL(zprop_register_number);
	EXPORT_SYMBOL(zprop_register_index);
	EXPORT_SYMBOL(zprop_register_hidden);

	/* Common routines for zfs and zpool property management */
	EXPORT_SYMBOL(zprop_iter_common);
	EXPORT_SYMBOL(zprop_name_to_prop);
	EXPORT_SYMBOL(zprop_string_to_index);
	EXPORT_SYMBOL(zprop_index_to_string);
	EXPORT_SYMBOL(zprop_random_value);
	EXPORT_SYMBOL(zprop_values);
	EXPORT_SYMBOL(zprop_valid_for_type);
	+EXPORT_SYMBOL(zprop_valid_char);
	#endif
	diff --git a/sys/contrib/openzfs/module/zfs/arc.c b/sys/contrib/openzfs/module/zfs/arc.c
	index 79e2d4381830..5427f2a71a4f 100644
	--- a/sys/contrib/openzfs/module/zfs/arc.c
	+++ b/sys/contrib/openzfs/module/zfs/arc.c
	@@ -1,11133 +1,11135 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2018, Joyent, Inc.
	* Copyright (c) 2011, 2020, Delphix. All rights reserved.
	* Copyright (c) 2014, Saso Kiselkov. All rights reserved.
	* Copyright (c) 2017, Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2019, loli10K <ezomori.nozomu@gmail.com>. All rights reserved.
	* Copyright (c) 2020, George Amanakis. All rights reserved.
	* Copyright (c) 2019, Klara Inc.
	* Copyright (c) 2019, Allan Jude
	* Copyright (c) 2020, The FreeBSD Foundation [1]
	*
	* [1] Portions of this software were developed by Allan Jude
	* under sponsorship from the FreeBSD Foundation.
	*/

	/*
	* DVA-based Adjustable Replacement Cache
	*
	* While much of the theory of operation used here is
	* based on the self-tuning, low overhead replacement cache
	* presented by Megiddo and Modha at FAST 2003, there are some
	* significant differences:
	*
	* 1. The Megiddo and Modha model assumes any page is evictable.
	* Pages in its cache cannot be "locked" into memory. This makes
	* the eviction algorithm simple: evict the last page in the list.
	* This also make the performance characteristics easy to reason
	* about. Our cache is not so simple. At any given moment, some
	* subset of the blocks in the cache are un-evictable because we
	* have handed out a reference to them. Blocks are only evictable
	* when there are no external references active. This makes
	* eviction far more problematic: we choose to evict the evictable
	* blocks that are the "lowest" in the list.
	*
	* There are times when it is not possible to evict the requested
	* space. In these circumstances we are unable to adjust the cache
	* size. To prevent the cache growing unbounded at these times we
	* implement a "cache throttle" that slows the flow of new data
	* into the cache until we can make space available.
	*
	* 2. The Megiddo and Modha model assumes a fixed cache size.
	* Pages are evicted when the cache is full and there is a cache
	* miss. Our model has a variable sized cache. It grows with
	* high use, but also tries to react to memory pressure from the
	* operating system: decreasing its size when system memory is
	* tight.
	*
	* 3. The Megiddo and Modha model assumes a fixed page size. All
	* elements of the cache are therefore exactly the same size. So
	* when adjusting the cache size following a cache miss, its simply
	* a matter of choosing a single page to evict. In our model, we
	* have variable sized cache blocks (ranging from 512 bytes to
	* 128K bytes). We therefore choose a set of blocks to evict to make
	* space for a cache miss that approximates as closely as possible
	* the space used by the new block.
	*
	* See also: "ARC: A Self-Tuning, Low Overhead Replacement Cache"
	* by N. Megiddo & D. Modha, FAST 2003
	*/

	/*
	* The locking model:
	*
	* A new reference to a cache buffer can be obtained in two
	* ways: 1) via a hash table lookup using the DVA as a key,
	* or 2) via one of the ARC lists. The arc_read() interface
	* uses method 1, while the internal ARC algorithms for
	* adjusting the cache use method 2. We therefore provide two
	* types of locks: 1) the hash table lock array, and 2) the
	* ARC list locks.
	*
	* Buffers do not have their own mutexes, rather they rely on the
	* hash table mutexes for the bulk of their protection (i.e. most
	* fields in the arc_buf_hdr_t are protected by these mutexes).
	*
	* buf_hash_find() returns the appropriate mutex (held) when it
	* locates the requested buffer in the hash table. It returns
	* NULL for the mutex if the buffer was not in the table.
	*
	* buf_hash_remove() expects the appropriate hash mutex to be
	* already held before it is invoked.
	*
	* Each ARC state also has a mutex which is used to protect the
	* buffer list associated with the state. When attempting to
	* obtain a hash table lock while holding an ARC list lock you
	* must use: mutex_tryenter() to avoid deadlock. Also note that
	* the active state mutex must be held before the ghost state mutex.
	*
	* It as also possible to register a callback which is run when the
	* arc_meta_limit is reached and no buffers can be safely evicted. In
	* this case the arc user should drop a reference on some arc buffers so
	* they can be reclaimed and the arc_meta_limit honored. For example,
	* when using the ZPL each dentry holds a references on a znode. These
	* dentries must be pruned before the arc buffer holding the znode can
	* be safely evicted.
	*
	* Note that the majority of the performance stats are manipulated
	* with atomic operations.
	*
	* The L2ARC uses the l2ad_mtx on each vdev for the following:
	*
	* - L2ARC buflist creation
	* - L2ARC buflist eviction
	* - L2ARC write completion, which walks L2ARC buflists
	* - ARC header destruction, as it removes from L2ARC buflists
	* - ARC header release, as it removes from L2ARC buflists
	*/

	/*
	* ARC operation:
	*
	* Every block that is in the ARC is tracked by an arc_buf_hdr_t structure.
	* This structure can point either to a block that is still in the cache or to
	* one that is only accessible in an L2 ARC device, or it can provide
	* information about a block that was recently evicted. If a block is
	* only accessible in the L2ARC, then the arc_buf_hdr_t only has enough
	* information to retrieve it from the L2ARC device. This information is
	* stored in the l2arc_buf_hdr_t sub-structure of the arc_buf_hdr_t. A block
	* that is in this state cannot access the data directly.
	*
	* Blocks that are actively being referenced or have not been evicted
	* are cached in the L1ARC. The L1ARC (l1arc_buf_hdr_t) is a structure within
	* the arc_buf_hdr_t that will point to the data block in memory. A block can
	* only be read by a consumer if it has an l1arc_buf_hdr_t. The L1ARC
	* caches data in two ways -- in a list of ARC buffers (arc_buf_t) and
	* also in the arc_buf_hdr_t's private physical data block pointer (b_pabd).
	*
	* The L1ARC's data pointer may or may not be uncompressed. The ARC has the
	* ability to store the physical data (b_pabd) associated with the DVA of the
	* arc_buf_hdr_t. Since the b_pabd is a copy of the on-disk physical block,
	* it will match its on-disk compression characteristics. This behavior can be
	* disabled by setting 'zfs_compressed_arc_enabled' to B_FALSE. When the
	* compressed ARC functionality is disabled, the b_pabd will point to an
	* uncompressed version of the on-disk data.
	*
	* Data in the L1ARC is not accessed by consumers of the ARC directly. Each
	* arc_buf_hdr_t can have multiple ARC buffers (arc_buf_t) which reference it.
	* Each ARC buffer (arc_buf_t) is being actively accessed by a specific ARC
	* consumer. The ARC will provide references to this data and will keep it
	* cached until it is no longer in use. The ARC caches only the L1ARC's physical
	* data block and will evict any arc_buf_t that is no longer referenced. The
	* amount of memory consumed by the arc_buf_ts' data buffers can be seen via the
	* "overhead_size" kstat.
	*
	* Depending on the consumer, an arc_buf_t can be requested in uncompressed or
	* compressed form. The typical case is that consumers will want uncompressed
	* data, and when that happens a new data buffer is allocated where the data is
	* decompressed for them to use. Currently the only consumer who wants
	* compressed arc_buf_t's is "zfs send", when it streams data exactly as it
	* exists on disk. When this happens, the arc_buf_t's data buffer is shared
	* with the arc_buf_hdr_t.
	*
	* Here is a diagram showing an arc_buf_hdr_t referenced by two arc_buf_t's. The
	* first one is owned by a compressed send consumer (and therefore references
	* the same compressed data buffer as the arc_buf_hdr_t) and the second could be
	* used by any other consumer (and has its own uncompressed copy of the data
	* buffer).
	*
	* arc_buf_hdr_t
	* +-----------+
	* \| fields \|
	* \| common to \|
	* \| L1- and \|
	* \| L2ARC \|
	* +-----------+
	* \| l2arc_buf_hdr_t
	* \| \|
	* +-----------+
	* \| l1arc_buf_hdr_t
	* \| \| arc_buf_t
	* \| b_buf +------------>+-----------+ arc_buf_t
	* \| b_pabd +-+ \|b_next +---->+-----------+
	* +-----------+ \| \|-----------\| \|b_next +-->NULL
	* \| \|b_comp = T \| +-----------+
	* \| \|b_data +-+ \|b_comp = F \|
	* \| +-----------+ \| \|b_data +-+
	* +->+------+ \| +-----------+ \|
	* compressed \| \| \| \|
	* data \| \|<--------------+ \| uncompressed
	* +------+ compressed, \| data
	* shared +-->+------+
	* data \| \|
	* \| \|
	* +------+
	*
	* When a consumer reads a block, the ARC must first look to see if the
	* arc_buf_hdr_t is cached. If the hdr is cached then the ARC allocates a new
	* arc_buf_t and either copies uncompressed data into a new data buffer from an
	* existing uncompressed arc_buf_t, decompresses the hdr's b_pabd buffer into a
	* new data buffer, or shares the hdr's b_pabd buffer, depending on whether the
	* hdr is compressed and the desired compression characteristics of the
	* arc_buf_t consumer. If the arc_buf_t ends up sharing data with the
	* arc_buf_hdr_t and both of them are uncompressed then the arc_buf_t must be
	* the last buffer in the hdr's b_buf list, however a shared compressed buf can
	* be anywhere in the hdr's list.
	*
	* The diagram below shows an example of an uncompressed ARC hdr that is
	* sharing its data with an arc_buf_t (note that the shared uncompressed buf is
	* the last element in the buf list):
	*
	* arc_buf_hdr_t
	* +-----------+
	* \| \|
	* \| \|
	* \| \|
	* +-----------+
	* l2arc_buf_hdr_t\| \|
	* \| \|
	* +-----------+
	* l1arc_buf_hdr_t\| \|
	* \| \| arc_buf_t (shared)
	* \| b_buf +------------>+---------+ arc_buf_t
	* \| \| \|b_next +---->+---------+
	* \| b_pabd +-+ \|---------\| \|b_next +-->NULL
	* +-----------+ \| \| \| +---------+
	* \| \|b_data +-+ \| \|
	* \| +---------+ \| \|b_data +-+
	* +->+------+ \| +---------+ \|
	* \| \| \| \|
	* uncompressed \| \| \| \|
	* data +------+ \| \|
	* ^ +->+------+ \|
	* \| uncompressed \| \| \|
	* \| data \| \| \|
	* \| +------+ \|
	* +---------------------------------+
	*
	* Writing to the ARC requires that the ARC first discard the hdr's b_pabd
	* since the physical block is about to be rewritten. The new data contents
	* will be contained in the arc_buf_t. As the I/O pipeline performs the write,
	* it may compress the data before writing it to disk. The ARC will be called
	* with the transformed data and will bcopy the transformed on-disk block into
	* a newly allocated b_pabd. Writes are always done into buffers which have
	* either been loaned (and hence are new and don't have other readers) or
	* buffers which have been released (and hence have their own hdr, if there
	* were originally other readers of the buf's original hdr). This ensures that
	* the ARC only needs to update a single buf and its hdr after a write occurs.
	*
	* When the L2ARC is in use, it will also take advantage of the b_pabd. The
	* L2ARC will always write the contents of b_pabd to the L2ARC. This means
	* that when compressed ARC is enabled that the L2ARC blocks are identical
	* to the on-disk block in the main data pool. This provides a significant
	* advantage since the ARC can leverage the bp's checksum when reading from the
	* L2ARC to determine if the contents are valid. However, if the compressed
	* ARC is disabled, then the L2ARC's block must be transformed to look
	* like the physical block in the main data pool before comparing the
	* checksum and determining its validity.
	*
	* The L1ARC has a slightly different system for storing encrypted data.
	* Raw (encrypted + possibly compressed) data has a few subtle differences from
	* data that is just compressed. The biggest difference is that it is not
	* possible to decrypt encrypted data (or vice-versa) if the keys aren't loaded.
	* The other difference is that encryption cannot be treated as a suggestion.
	* If a caller would prefer compressed data, but they actually wind up with
	* uncompressed data the worst thing that could happen is there might be a
	* performance hit. If the caller requests encrypted data, however, we must be
	* sure they actually get it or else secret information could be leaked. Raw
	* data is stored in hdr->b_crypt_hdr.b_rabd. An encrypted header, therefore,
	* may have both an encrypted version and a decrypted version of its data at
	* once. When a caller needs a raw arc_buf_t, it is allocated and the data is
	* copied out of this header. To avoid complications with b_pabd, raw buffers
	* cannot be shared.
	*/

	#include <sys/spa.h>
	#include <sys/zio.h>
	#include <sys/spa_impl.h>
	#include <sys/zio_compress.h>
	#include <sys/zio_checksum.h>
	#include <sys/zfs_context.h>
	#include <sys/arc.h>
	#include <sys/zfs_refcount.h>
	#include <sys/vdev.h>
	#include <sys/vdev_impl.h>
	#include <sys/dsl_pool.h>
	#include <sys/multilist.h>
	#include <sys/abd.h>
	#include <sys/zil.h>
	#include <sys/fm/fs/zfs.h>
	#include <sys/callb.h>
	#include <sys/kstat.h>
	#include <sys/zthr.h>
	#include <zfs_fletcher.h>
	#include <sys/arc_impl.h>
	#include <sys/trace_zfs.h>
	#include <sys/aggsum.h>
	#include <sys/wmsum.h>
	#include <cityhash.h>
	#include <sys/vdev_trim.h>
	#include <sys/zfs_racct.h>
	#include <sys/zstd/zstd.h>

	#ifndef _KERNEL
	/* set with ZFS_DEBUG=watch, to enable watchpoints on frozen buffers */
	boolean_t arc_watch = B_FALSE;
	#endif

	/*
	* This thread's job is to keep enough free memory in the system, by
	* calling arc_kmem_reap_soon() plus arc_reduce_target_size(), which improves
	* arc_available_memory().
	*/
	static zthr_t *arc_reap_zthr;

	/*
	* This thread's job is to keep arc_size under arc_c, by calling
	* arc_evict(), which improves arc_is_overflowing().
	*/
	static zthr_t *arc_evict_zthr;

	static kmutex_t arc_evict_lock;
	static boolean_t arc_evict_needed = B_FALSE;

	/*
	* Count of bytes evicted since boot.
	*/
	static uint64_t arc_evict_count;

	/*
	* List of arc_evict_waiter_t's, representing threads waiting for the
	* arc_evict_count to reach specific values.
	*/
	static list_t arc_evict_waiters;

	/*
	* When arc_is_overflowing(), arc_get_data_impl() waits for this percent of
	* the requested amount of data to be evicted. For example, by default for
	* every 2KB that's evicted, 1KB of it may be "reused" by a new allocation.
	* Since this is above 100%, it ensures that progress is made towards getting
	* arc_size under arc_c. Since this is finite, it ensures that allocations
	* can still happen, even during the potentially long time that arc_size is
	* more than arc_c.
	*/
	int zfs_arc_eviction_pct = 200;

	/*
	* The number of headers to evict in arc_evict_state_impl() before
	* dropping the sublist lock and evicting from another sublist. A lower
	* value means we're more likely to evict the "correct" header (i.e. the
	* oldest header in the arc state), but comes with higher overhead
	* (i.e. more invocations of arc_evict_state_impl()).
	*/
	int zfs_arc_evict_batch_limit = 10;

	/* number of seconds before growing cache again */
	int arc_grow_retry = 5;

	/*
	* Minimum time between calls to arc_kmem_reap_soon().
	*/
	int arc_kmem_cache_reap_retry_ms = 1000;

	/* shift of arc_c for calculating overflow limit in arc_get_data_impl */
	int zfs_arc_overflow_shift = 8;

	/* shift of arc_c for calculating both min and max arc_p */
	int arc_p_min_shift = 4;

	/* log2(fraction of arc to reclaim) */
	int arc_shrink_shift = 7;

	/* percent of pagecache to reclaim arc to */
	#ifdef _KERNEL
	uint_t zfs_arc_pc_percent = 0;
	#endif

	/*
	* log2(fraction of ARC which must be free to allow growing).
	* I.e. If there is less than arc_c >> arc_no_grow_shift free memory,
	* when reading a new block into the ARC, we will evict an equal-sized block
	* from the ARC.
	*
	* This must be less than arc_shrink_shift, so that when we shrink the ARC,
	* we will still not allow it to grow.
	*/
	int arc_no_grow_shift = 5;


	/*
	* minimum lifespan of a prefetch block in clock ticks
	* (initialized in arc_init())
	*/
	static int arc_min_prefetch_ms;
	static int arc_min_prescient_prefetch_ms;

	/*
	* If this percent of memory is free, don't throttle.
	*/
	int arc_lotsfree_percent = 10;

	/*
	* The arc has filled available memory and has now warmed up.
	*/
	boolean_t arc_warm;

	/*
	* These tunables are for performance analysis.
	*/
	unsigned long zfs_arc_max = 0;
	unsigned long zfs_arc_min = 0;
	unsigned long zfs_arc_meta_limit = 0;
	unsigned long zfs_arc_meta_min = 0;
	unsigned long zfs_arc_dnode_limit = 0;
	unsigned long zfs_arc_dnode_reduce_percent = 10;
	int zfs_arc_grow_retry = 0;
	int zfs_arc_shrink_shift = 0;
	int zfs_arc_p_min_shift = 0;
	int zfs_arc_average_blocksize = 8 * 1024; /* 8KB */

	/*
	* ARC dirty data constraints for arc_tempreserve_space() throttle.
	*/
	unsigned long zfs_arc_dirty_limit_percent = 50; /* total dirty data limit */
	unsigned long zfs_arc_anon_limit_percent = 25; /* anon block dirty limit */
	unsigned long zfs_arc_pool_dirty_percent = 20; /* each pool's anon allowance */

	/*
	* Enable or disable compressed arc buffers.
	*/
	int zfs_compressed_arc_enabled = B_TRUE;

	/*
	* ARC will evict meta buffers that exceed arc_meta_limit. This
	* tunable make arc_meta_limit adjustable for different workloads.
	*/
	unsigned long zfs_arc_meta_limit_percent = 75;

	/*
	* Percentage that can be consumed by dnodes of ARC meta buffers.
	*/
	unsigned long zfs_arc_dnode_limit_percent = 10;

	/*
	* These tunables are Linux specific
	*/
	unsigned long zfs_arc_sys_free = 0;
	int zfs_arc_min_prefetch_ms = 0;
	int zfs_arc_min_prescient_prefetch_ms = 0;
	int zfs_arc_p_dampener_disable = 1;
	int zfs_arc_meta_prune = 10000;
	int zfs_arc_meta_strategy = ARC_STRATEGY_META_BALANCED;
	int zfs_arc_meta_adjust_restarts = 4096;
	int zfs_arc_lotsfree_percent = 10;

	/* The 6 states: */
	arc_state_t ARC_anon;
	arc_state_t ARC_mru;
	arc_state_t ARC_mru_ghost;
	arc_state_t ARC_mfu;
	arc_state_t ARC_mfu_ghost;
	arc_state_t ARC_l2c_only;

	arc_stats_t arc_stats = {
	{ "hits", KSTAT_DATA_UINT64 },
	{ "misses", KSTAT_DATA_UINT64 },
	{ "demand_data_hits", KSTAT_DATA_UINT64 },
	{ "demand_data_misses", KSTAT_DATA_UINT64 },
	{ "demand_metadata_hits", KSTAT_DATA_UINT64 },
	{ "demand_metadata_misses", KSTAT_DATA_UINT64 },
	{ "prefetch_data_hits", KSTAT_DATA_UINT64 },
	{ "prefetch_data_misses", KSTAT_DATA_UINT64 },
	{ "prefetch_metadata_hits", KSTAT_DATA_UINT64 },
	{ "prefetch_metadata_misses", KSTAT_DATA_UINT64 },
	{ "mru_hits", KSTAT_DATA_UINT64 },
	{ "mru_ghost_hits", KSTAT_DATA_UINT64 },
	{ "mfu_hits", KSTAT_DATA_UINT64 },
	{ "mfu_ghost_hits", KSTAT_DATA_UINT64 },
	{ "deleted", KSTAT_DATA_UINT64 },
	{ "mutex_miss", KSTAT_DATA_UINT64 },
	{ "access_skip", KSTAT_DATA_UINT64 },
	{ "evict_skip", KSTAT_DATA_UINT64 },
	{ "evict_not_enough", KSTAT_DATA_UINT64 },
	{ "evict_l2_cached", KSTAT_DATA_UINT64 },
	{ "evict_l2_eligible", KSTAT_DATA_UINT64 },
	{ "evict_l2_eligible_mfu", KSTAT_DATA_UINT64 },
	{ "evict_l2_eligible_mru", KSTAT_DATA_UINT64 },
	{ "evict_l2_ineligible", KSTAT_DATA_UINT64 },
	{ "evict_l2_skip", KSTAT_DATA_UINT64 },
	{ "hash_elements", KSTAT_DATA_UINT64 },
	{ "hash_elements_max", KSTAT_DATA_UINT64 },
	{ "hash_collisions", KSTAT_DATA_UINT64 },
	{ "hash_chains", KSTAT_DATA_UINT64 },
	{ "hash_chain_max", KSTAT_DATA_UINT64 },
	{ "p", KSTAT_DATA_UINT64 },
	{ "c", KSTAT_DATA_UINT64 },
	{ "c_min", KSTAT_DATA_UINT64 },
	{ "c_max", KSTAT_DATA_UINT64 },
	{ "size", KSTAT_DATA_UINT64 },
	{ "compressed_size", KSTAT_DATA_UINT64 },
	{ "uncompressed_size", KSTAT_DATA_UINT64 },
	{ "overhead_size", KSTAT_DATA_UINT64 },
	{ "hdr_size", KSTAT_DATA_UINT64 },
	{ "data_size", KSTAT_DATA_UINT64 },
	{ "metadata_size", KSTAT_DATA_UINT64 },
	{ "dbuf_size", KSTAT_DATA_UINT64 },
	{ "dnode_size", KSTAT_DATA_UINT64 },
	{ "bonus_size", KSTAT_DATA_UINT64 },
	#if defined(COMPAT_FREEBSD11)
	{ "other_size", KSTAT_DATA_UINT64 },
	#endif
	{ "anon_size", KSTAT_DATA_UINT64 },
	{ "anon_evictable_data", KSTAT_DATA_UINT64 },
	{ "anon_evictable_metadata", KSTAT_DATA_UINT64 },
	{ "mru_size", KSTAT_DATA_UINT64 },
	{ "mru_evictable_data", KSTAT_DATA_UINT64 },
	{ "mru_evictable_metadata", KSTAT_DATA_UINT64 },
	{ "mru_ghost_size", KSTAT_DATA_UINT64 },
	{ "mru_ghost_evictable_data", KSTAT_DATA_UINT64 },
	{ "mru_ghost_evictable_metadata", KSTAT_DATA_UINT64 },
	{ "mfu_size", KSTAT_DATA_UINT64 },
	{ "mfu_evictable_data", KSTAT_DATA_UINT64 },
	{ "mfu_evictable_metadata", KSTAT_DATA_UINT64 },
	{ "mfu_ghost_size", KSTAT_DATA_UINT64 },
	{ "mfu_ghost_evictable_data", KSTAT_DATA_UINT64 },
	{ "mfu_ghost_evictable_metadata", KSTAT_DATA_UINT64 },
	{ "l2_hits", KSTAT_DATA_UINT64 },
	{ "l2_misses", KSTAT_DATA_UINT64 },
	{ "l2_prefetch_asize", KSTAT_DATA_UINT64 },
	{ "l2_mru_asize", KSTAT_DATA_UINT64 },
	{ "l2_mfu_asize", KSTAT_DATA_UINT64 },
	{ "l2_bufc_data_asize", KSTAT_DATA_UINT64 },
	{ "l2_bufc_metadata_asize", KSTAT_DATA_UINT64 },
	{ "l2_feeds", KSTAT_DATA_UINT64 },
	{ "l2_rw_clash", KSTAT_DATA_UINT64 },
	{ "l2_read_bytes", KSTAT_DATA_UINT64 },
	{ "l2_write_bytes", KSTAT_DATA_UINT64 },
	{ "l2_writes_sent", KSTAT_DATA_UINT64 },
	{ "l2_writes_done", KSTAT_DATA_UINT64 },
	{ "l2_writes_error", KSTAT_DATA_UINT64 },
	{ "l2_writes_lock_retry", KSTAT_DATA_UINT64 },
	{ "l2_evict_lock_retry", KSTAT_DATA_UINT64 },
	{ "l2_evict_reading", KSTAT_DATA_UINT64 },
	{ "l2_evict_l1cached", KSTAT_DATA_UINT64 },
	{ "l2_free_on_write", KSTAT_DATA_UINT64 },
	{ "l2_abort_lowmem", KSTAT_DATA_UINT64 },
	{ "l2_cksum_bad", KSTAT_DATA_UINT64 },
	{ "l2_io_error", KSTAT_DATA_UINT64 },
	{ "l2_size", KSTAT_DATA_UINT64 },
	{ "l2_asize", KSTAT_DATA_UINT64 },
	{ "l2_hdr_size", KSTAT_DATA_UINT64 },
	{ "l2_log_blk_writes", KSTAT_DATA_UINT64 },
	{ "l2_log_blk_avg_asize", KSTAT_DATA_UINT64 },
	{ "l2_log_blk_asize", KSTAT_DATA_UINT64 },
	{ "l2_log_blk_count", KSTAT_DATA_UINT64 },
	{ "l2_data_to_meta_ratio", KSTAT_DATA_UINT64 },
	{ "l2_rebuild_success", KSTAT_DATA_UINT64 },
	{ "l2_rebuild_unsupported", KSTAT_DATA_UINT64 },
	{ "l2_rebuild_io_errors", KSTAT_DATA_UINT64 },
	{ "l2_rebuild_dh_errors", KSTAT_DATA_UINT64 },
	{ "l2_rebuild_cksum_lb_errors", KSTAT_DATA_UINT64 },
	{ "l2_rebuild_lowmem", KSTAT_DATA_UINT64 },
	{ "l2_rebuild_size", KSTAT_DATA_UINT64 },
	{ "l2_rebuild_asize", KSTAT_DATA_UINT64 },
	{ "l2_rebuild_bufs", KSTAT_DATA_UINT64 },
	{ "l2_rebuild_bufs_precached", KSTAT_DATA_UINT64 },
	{ "l2_rebuild_log_blks", KSTAT_DATA_UINT64 },
	{ "memory_throttle_count", KSTAT_DATA_UINT64 },
	{ "memory_direct_count", KSTAT_DATA_UINT64 },
	{ "memory_indirect_count", KSTAT_DATA_UINT64 },
	{ "memory_all_bytes", KSTAT_DATA_UINT64 },
	{ "memory_free_bytes", KSTAT_DATA_UINT64 },
	{ "memory_available_bytes", KSTAT_DATA_INT64 },
	{ "arc_no_grow", KSTAT_DATA_UINT64 },
	{ "arc_tempreserve", KSTAT_DATA_UINT64 },
	{ "arc_loaned_bytes", KSTAT_DATA_UINT64 },
	{ "arc_prune", KSTAT_DATA_UINT64 },
	{ "arc_meta_used", KSTAT_DATA_UINT64 },
	{ "arc_meta_limit", KSTAT_DATA_UINT64 },
	{ "arc_dnode_limit", KSTAT_DATA_UINT64 },
	{ "arc_meta_max", KSTAT_DATA_UINT64 },
	{ "arc_meta_min", KSTAT_DATA_UINT64 },
	{ "async_upgrade_sync", KSTAT_DATA_UINT64 },
	{ "demand_hit_predictive_prefetch", KSTAT_DATA_UINT64 },
	{ "demand_hit_prescient_prefetch", KSTAT_DATA_UINT64 },
	{ "arc_need_free", KSTAT_DATA_UINT64 },
	{ "arc_sys_free", KSTAT_DATA_UINT64 },
	{ "arc_raw_size", KSTAT_DATA_UINT64 },
	{ "cached_only_in_progress", KSTAT_DATA_UINT64 },
	{ "abd_chunk_waste_size", KSTAT_DATA_UINT64 },
	};

	arc_sums_t arc_sums;

	#define ARCSTAT_MAX(stat, val) { \
	uint64_t m; \
	while ((val) > (m = arc_stats.stat.value.ui64) && \
	(m != atomic_cas_64(&arc_stats.stat.value.ui64, m, (val)))) \
	continue; \
	}

	/*
	* We define a macro to allow ARC hits/misses to be easily broken down by
	* two separate conditions, giving a total of four different subtypes for
	* each of hits and misses (so eight statistics total).
	*/
	#define ARCSTAT_CONDSTAT(cond1, stat1, notstat1, cond2, stat2, notstat2, stat) \
	if (cond1) { \
	if (cond2) { \
	ARCSTAT_BUMP(arcstat_##stat1##_##stat2##_##stat); \
	} else { \
	ARCSTAT_BUMP(arcstat_##stat1##_##notstat2##_##stat); \
	} \
	} else { \
	if (cond2) { \
	ARCSTAT_BUMP(arcstat_##notstat1##_##stat2##_##stat); \
	} else { \
	ARCSTAT_BUMP(arcstat_##notstat1##_##notstat2##_##stat);\
	} \
	}

	/*
	* This macro allows us to use kstats as floating averages. Each time we
	* update this kstat, we first factor it and the update value by
	* ARCSTAT_AVG_FACTOR to shrink the new value's contribution to the overall
	* average. This macro assumes that integer loads and stores are atomic, but
	* is not safe for multiple writers updating the kstat in parallel (only the
	* last writer's update will remain).
	*/
	#define ARCSTAT_F_AVG_FACTOR 3
	#define ARCSTAT_F_AVG(stat, value) \
	do { \
	uint64_t x = ARCSTAT(stat); \
	x = x - x / ARCSTAT_F_AVG_FACTOR + \
	(value) / ARCSTAT_F_AVG_FACTOR; \
	ARCSTAT(stat) = x; \
	} while (0)

	kstat_t *arc_ksp;

	/*
	* There are several ARC variables that are critical to export as kstats --
	* but we don't want to have to grovel around in the kstat whenever we wish to
	* manipulate them. For these variables, we therefore define them to be in
	* terms of the statistic variable. This assures that we are not introducing
	* the possibility of inconsistency by having shadow copies of the variables,
	* while still allowing the code to be readable.
	*/
	#define arc_tempreserve ARCSTAT(arcstat_tempreserve)
	#define arc_loaned_bytes ARCSTAT(arcstat_loaned_bytes)
	#define arc_meta_limit ARCSTAT(arcstat_meta_limit) /* max size for metadata */
	/* max size for dnodes */
	#define arc_dnode_size_limit ARCSTAT(arcstat_dnode_limit)
	#define arc_meta_min ARCSTAT(arcstat_meta_min) /* min size for metadata */
	#define arc_need_free ARCSTAT(arcstat_need_free) /* waiting to be evicted */

	hrtime_t arc_growtime;
	list_t arc_prune_list;
	kmutex_t arc_prune_mtx;
	taskq_t *arc_prune_taskq;

	#define GHOST_STATE(state) \
	((state) == arc_mru_ghost \|\| (state) == arc_mfu_ghost \|\| \
	(state) == arc_l2c_only)

	#define HDR_IN_HASH_TABLE(hdr) ((hdr)->b_flags & ARC_FLAG_IN_HASH_TABLE)
	#define HDR_IO_IN_PROGRESS(hdr) ((hdr)->b_flags & ARC_FLAG_IO_IN_PROGRESS)
	#define HDR_IO_ERROR(hdr) ((hdr)->b_flags & ARC_FLAG_IO_ERROR)
	#define HDR_PREFETCH(hdr) ((hdr)->b_flags & ARC_FLAG_PREFETCH)
	#define HDR_PRESCIENT_PREFETCH(hdr) \
	((hdr)->b_flags & ARC_FLAG_PRESCIENT_PREFETCH)
	#define HDR_COMPRESSION_ENABLED(hdr) \
	((hdr)->b_flags & ARC_FLAG_COMPRESSED_ARC)

	#define HDR_L2CACHE(hdr) ((hdr)->b_flags & ARC_FLAG_L2CACHE)
	#define HDR_L2_READING(hdr) \
	(((hdr)->b_flags & ARC_FLAG_IO_IN_PROGRESS) && \
	((hdr)->b_flags & ARC_FLAG_HAS_L2HDR))
	#define HDR_L2_WRITING(hdr) ((hdr)->b_flags & ARC_FLAG_L2_WRITING)
	#define HDR_L2_EVICTED(hdr) ((hdr)->b_flags & ARC_FLAG_L2_EVICTED)
	#define HDR_L2_WRITE_HEAD(hdr) ((hdr)->b_flags & ARC_FLAG_L2_WRITE_HEAD)
	#define HDR_PROTECTED(hdr) ((hdr)->b_flags & ARC_FLAG_PROTECTED)
	#define HDR_NOAUTH(hdr) ((hdr)->b_flags & ARC_FLAG_NOAUTH)
	#define HDR_SHARED_DATA(hdr) ((hdr)->b_flags & ARC_FLAG_SHARED_DATA)

	#define HDR_ISTYPE_METADATA(hdr) \
	((hdr)->b_flags & ARC_FLAG_BUFC_METADATA)
	#define HDR_ISTYPE_DATA(hdr) (!HDR_ISTYPE_METADATA(hdr))

	#define HDR_HAS_L1HDR(hdr) ((hdr)->b_flags & ARC_FLAG_HAS_L1HDR)
	#define HDR_HAS_L2HDR(hdr) ((hdr)->b_flags & ARC_FLAG_HAS_L2HDR)
	#define HDR_HAS_RABD(hdr) \
	(HDR_HAS_L1HDR(hdr) && HDR_PROTECTED(hdr) && \
	(hdr)->b_crypt_hdr.b_rabd != NULL)
	#define HDR_ENCRYPTED(hdr) \
	(HDR_PROTECTED(hdr) && DMU_OT_IS_ENCRYPTED((hdr)->b_crypt_hdr.b_ot))
	#define HDR_AUTHENTICATED(hdr) \
	(HDR_PROTECTED(hdr) && !DMU_OT_IS_ENCRYPTED((hdr)->b_crypt_hdr.b_ot))

	/* For storing compression mode in b_flags */
	#define HDR_COMPRESS_OFFSET (highbit64(ARC_FLAG_COMPRESS_0) - 1)

	#define HDR_GET_COMPRESS(hdr) ((enum zio_compress)BF32_GET((hdr)->b_flags, \
	HDR_COMPRESS_OFFSET, SPA_COMPRESSBITS))
	#define HDR_SET_COMPRESS(hdr, cmp) BF32_SET((hdr)->b_flags, \
	HDR_COMPRESS_OFFSET, SPA_COMPRESSBITS, (cmp));

	#define ARC_BUF_LAST(buf) ((buf)->b_next == NULL)
	#define ARC_BUF_SHARED(buf) ((buf)->b_flags & ARC_BUF_FLAG_SHARED)
	#define ARC_BUF_COMPRESSED(buf) ((buf)->b_flags & ARC_BUF_FLAG_COMPRESSED)
	#define ARC_BUF_ENCRYPTED(buf) ((buf)->b_flags & ARC_BUF_FLAG_ENCRYPTED)

	/*
	* Other sizes
	*/

	#define HDR_FULL_CRYPT_SIZE ((int64_t)sizeof (arc_buf_hdr_t))
	#define HDR_FULL_SIZE ((int64_t)offsetof(arc_buf_hdr_t, b_crypt_hdr))
	#define HDR_L2ONLY_SIZE ((int64_t)offsetof(arc_buf_hdr_t, b_l1hdr))

	/*
	* Hash table routines
	*/

	#define BUF_LOCKS 2048
	typedef struct buf_hash_table {
	uint64_t ht_mask;
	arc_buf_hdr_t **ht_table;
	kmutex_t ht_locks[BUF_LOCKS] ____cacheline_aligned;
	} buf_hash_table_t;

	static buf_hash_table_t buf_hash_table;

	#define BUF_HASH_INDEX(spa, dva, birth) \
	(buf_hash(spa, dva, birth) & buf_hash_table.ht_mask)
	#define BUF_HASH_LOCK(idx) (&buf_hash_table.ht_locks[idx & (BUF_LOCKS-1)])
	#define HDR_LOCK(hdr) \
	(BUF_HASH_LOCK(BUF_HASH_INDEX(hdr->b_spa, &hdr->b_dva, hdr->b_birth)))

	uint64_t zfs_crc64_table[256];

	/*
	* Level 2 ARC
	*/

	#define L2ARC_WRITE_SIZE (8 * 1024 * 1024) /* initial write max */
	#define L2ARC_HEADROOM 2 /* num of writes */

	/*
	* If we discover during ARC scan any buffers to be compressed, we boost
	* our headroom for the next scanning cycle by this percentage multiple.
	*/
	#define L2ARC_HEADROOM_BOOST 200
	#define L2ARC_FEED_SECS 1 /* caching interval secs */
	#define L2ARC_FEED_MIN_MS 200 /* min caching interval ms */

	/*
	* We can feed L2ARC from two states of ARC buffers, mru and mfu,
	* and each of the state has two types: data and metadata.
	*/
	#define L2ARC_FEED_TYPES 4

	/* L2ARC Performance Tunables */
	unsigned long l2arc_write_max = L2ARC_WRITE_SIZE; /* def max write size */
	unsigned long l2arc_write_boost = L2ARC_WRITE_SIZE; /* extra warmup write */
	unsigned long l2arc_headroom = L2ARC_HEADROOM; /* # of dev writes */
	unsigned long l2arc_headroom_boost = L2ARC_HEADROOM_BOOST;
	unsigned long l2arc_feed_secs = L2ARC_FEED_SECS; /* interval seconds */
	unsigned long l2arc_feed_min_ms = L2ARC_FEED_MIN_MS; /* min interval msecs */
	int l2arc_noprefetch = B_TRUE; /* don't cache prefetch bufs */
	int l2arc_feed_again = B_TRUE; /* turbo warmup */
	int l2arc_norw = B_FALSE; /* no reads during writes */
	int l2arc_meta_percent = 33; /* limit on headers size */

	/*
	* L2ARC Internals
	*/
	static list_t L2ARC_dev_list; /* device list */
	static list_t l2arc_dev_list; / device list pointer */
	static kmutex_t l2arc_dev_mtx; /* device list mutex */
	static l2arc_dev_t l2arc_dev_last; / last device used */
	static list_t L2ARC_free_on_write; /* free after write buf list */
	static list_t l2arc_free_on_write; / free after write list ptr */
	static kmutex_t l2arc_free_on_write_mtx; /* mutex for list */
	static uint64_t l2arc_ndev; /* number of devices */

	typedef struct l2arc_read_callback {
	arc_buf_hdr_t l2rcb_hdr; / read header */
	blkptr_t l2rcb_bp; /* original blkptr */
	zbookmark_phys_t l2rcb_zb; /* original bookmark */
	int l2rcb_flags; /* original flags */
	abd_t l2rcb_abd; / temporary buffer */
	} l2arc_read_callback_t;

	typedef struct l2arc_data_free {
	/* protected by l2arc_free_on_write_mtx */
	abd_t *l2df_abd;
	size_t l2df_size;
	arc_buf_contents_t l2df_type;
	list_node_t l2df_list_node;
	} l2arc_data_free_t;

	typedef enum arc_fill_flags {
	ARC_FILL_LOCKED = 1 << 0, /* hdr lock is held */
	ARC_FILL_COMPRESSED = 1 << 1, /* fill with compressed data */
	ARC_FILL_ENCRYPTED = 1 << 2, /* fill with encrypted data */
	ARC_FILL_NOAUTH = 1 << 3, /* don't attempt to authenticate */
	ARC_FILL_IN_PLACE = 1 << 4 /* fill in place (special case) */
	} arc_fill_flags_t;

	typedef enum arc_ovf_level {
	ARC_OVF_NONE, /* ARC within target size. */
	ARC_OVF_SOME, /* ARC is slightly overflowed. */
	ARC_OVF_SEVERE /* ARC is severely overflowed. */
	} arc_ovf_level_t;

	static kmutex_t l2arc_feed_thr_lock;
	static kcondvar_t l2arc_feed_thr_cv;
	static uint8_t l2arc_thread_exit;

	static kmutex_t l2arc_rebuild_thr_lock;
	static kcondvar_t l2arc_rebuild_thr_cv;

	enum arc_hdr_alloc_flags {
	ARC_HDR_ALLOC_RDATA = 0x1,
	ARC_HDR_DO_ADAPT = 0x2,
	ARC_HDR_USE_RESERVE = 0x4,
	};


	static abd_t arc_get_data_abd(arc_buf_hdr_t , uint64_t, void *, int);
	static void arc_get_data_buf(arc_buf_hdr_t , uint64_t, void *);
	static void arc_get_data_impl(arc_buf_hdr_t , uint64_t, void , int);
	static void arc_free_data_abd(arc_buf_hdr_t , abd_t , uint64_t, void *);
	static void arc_free_data_buf(arc_buf_hdr_t , void , uint64_t, void *);
	static void arc_free_data_impl(arc_buf_hdr_t hdr, uint64_t size, void tag);
	static void arc_hdr_free_abd(arc_buf_hdr_t *, boolean_t);
	static void arc_hdr_alloc_abd(arc_buf_hdr_t *, int);
	static void arc_access(arc_buf_hdr_t , kmutex_t );
	static void arc_buf_watch(arc_buf_t *);

	static arc_buf_contents_t arc_buf_type(arc_buf_hdr_t *);
	static uint32_t arc_bufc_to_flags(arc_buf_contents_t);
	static inline void arc_hdr_set_flags(arc_buf_hdr_t *hdr, arc_flags_t flags);
	static inline void arc_hdr_clear_flags(arc_buf_hdr_t *hdr, arc_flags_t flags);

	static boolean_t l2arc_write_eligible(uint64_t, arc_buf_hdr_t *);
	static void l2arc_read_done(zio_t *);
	static void l2arc_do_free_on_write(void);
	static void l2arc_hdr_arcstats_update(arc_buf_hdr_t *hdr, boolean_t incr,
	boolean_t state_only);

	#define l2arc_hdr_arcstats_increment(hdr) \
	l2arc_hdr_arcstats_update((hdr), B_TRUE, B_FALSE)
	#define l2arc_hdr_arcstats_decrement(hdr) \
	l2arc_hdr_arcstats_update((hdr), B_FALSE, B_FALSE)
	#define l2arc_hdr_arcstats_increment_state(hdr) \
	l2arc_hdr_arcstats_update((hdr), B_TRUE, B_TRUE)
	#define l2arc_hdr_arcstats_decrement_state(hdr) \
	l2arc_hdr_arcstats_update((hdr), B_FALSE, B_TRUE)

	/*
	* l2arc_exclude_special : A zfs module parameter that controls whether buffers
	* present on special vdevs are eligibile for caching in L2ARC. If
	* set to 1, exclude dbufs on special vdevs from being cached to
	* L2ARC.
	*/
	int l2arc_exclude_special = 0;

	/*
	* l2arc_mfuonly : A ZFS module parameter that controls whether only MFU
	* metadata and data are cached from ARC into L2ARC.
	*/
	int l2arc_mfuonly = 0;

	/*
	* L2ARC TRIM
	* l2arc_trim_ahead : A ZFS module parameter that controls how much ahead of
	* the current write size (l2arc_write_max) we should TRIM if we
	* have filled the device. It is defined as a percentage of the
	* write size. If set to 100 we trim twice the space required to
	* accommodate upcoming writes. A minimum of 64MB will be trimmed.
	* It also enables TRIM of the whole L2ARC device upon creation or
	* addition to an existing pool or if the header of the device is
	* invalid upon importing a pool or onlining a cache device. The
	* default is 0, which disables TRIM on L2ARC altogether as it can
	* put significant stress on the underlying storage devices. This
	* will vary depending of how well the specific device handles
	* these commands.
	*/
	unsigned long l2arc_trim_ahead = 0;

	/*
	* Performance tuning of L2ARC persistence:
	*
	* l2arc_rebuild_enabled : A ZFS module parameter that controls whether adding
	* an L2ARC device (either at pool import or later) will attempt
	* to rebuild L2ARC buffer contents.
	* l2arc_rebuild_blocks_min_l2size : A ZFS module parameter that controls
	* whether log blocks are written to the L2ARC device. If the L2ARC
	* device is less than 1GB, the amount of data l2arc_evict()
	* evicts is significant compared to the amount of restored L2ARC
	* data. In this case do not write log blocks in L2ARC in order
	* not to waste space.
	*/
	int l2arc_rebuild_enabled = B_TRUE;
	unsigned long l2arc_rebuild_blocks_min_l2size = 1024 * 1024 * 1024;

	/* L2ARC persistence rebuild control routines. */
	void l2arc_rebuild_vdev(vdev_t *vd, boolean_t reopen);
	static void l2arc_dev_rebuild_thread(void *arg);
	static int l2arc_rebuild(l2arc_dev_t *dev);

	/* L2ARC persistence read I/O routines. */
	static int l2arc_dev_hdr_read(l2arc_dev_t *dev);
	static int l2arc_log_blk_read(l2arc_dev_t *dev,
	const l2arc_log_blkptr_t this_lp, const l2arc_log_blkptr_t next_lp,
	l2arc_log_blk_phys_t this_lb, l2arc_log_blk_phys_t next_lb,
	zio_t this_io, zio_t *next_io);
	static zio_t l2arc_log_blk_fetch(vdev_t vd,
	const l2arc_log_blkptr_t lp, l2arc_log_blk_phys_t lb);
	static void l2arc_log_blk_fetch_abort(zio_t *zio);

	/* L2ARC persistence block restoration routines. */
	static void l2arc_log_blk_restore(l2arc_dev_t *dev,
	const l2arc_log_blk_phys_t *lb, uint64_t lb_asize);
	static void l2arc_hdr_restore(const l2arc_log_ent_phys_t *le,
	l2arc_dev_t *dev);

	/* L2ARC persistence write I/O routines. */
	static void l2arc_log_blk_commit(l2arc_dev_t dev, zio_t pio,
	l2arc_write_callback_t *cb);

	/* L2ARC persistence auxiliary routines. */
	boolean_t l2arc_log_blkptr_valid(l2arc_dev_t *dev,
	const l2arc_log_blkptr_t *lbp);
	static boolean_t l2arc_log_blk_insert(l2arc_dev_t *dev,
	const arc_buf_hdr_t *ab);
	boolean_t l2arc_range_check_overlap(uint64_t bottom,
	uint64_t top, uint64_t check);
	static void l2arc_blk_fetch_done(zio_t *zio);
	static inline uint64_t
	l2arc_log_blk_overhead(uint64_t write_sz, l2arc_dev_t *dev);

	/*
	* We use Cityhash for this. It's fast, and has good hash properties without
	* requiring any large static buffers.
	*/
	static uint64_t
	buf_hash(uint64_t spa, const dva_t *dva, uint64_t birth)
	{
	return (cityhash4(spa, dva->dva_word[0], dva->dva_word[1], birth));
	}

	#define HDR_EMPTY(hdr) \
	((hdr)->b_dva.dva_word[0] == 0 && \
	(hdr)->b_dva.dva_word[1] == 0)

	#define HDR_EMPTY_OR_LOCKED(hdr) \
	(HDR_EMPTY(hdr) \|\| MUTEX_HELD(HDR_LOCK(hdr)))

	#define HDR_EQUAL(spa, dva, birth, hdr) \
	((hdr)->b_dva.dva_word[0] == (dva)->dva_word[0]) && \
	((hdr)->b_dva.dva_word[1] == (dva)->dva_word[1]) && \
	((hdr)->b_birth == birth) && ((hdr)->b_spa == spa)

	static void
	buf_discard_identity(arc_buf_hdr_t *hdr)
	{
	hdr->b_dva.dva_word[0] = 0;
	hdr->b_dva.dva_word[1] = 0;
	hdr->b_birth = 0;
	}

	static arc_buf_hdr_t *
	buf_hash_find(uint64_t spa, const blkptr_t bp, kmutex_t *lockp)
	{
	const dva_t *dva = BP_IDENTITY(bp);
	uint64_t birth = BP_PHYSICAL_BIRTH(bp);
	uint64_t idx = BUF_HASH_INDEX(spa, dva, birth);
	kmutex_t *hash_lock = BUF_HASH_LOCK(idx);
	arc_buf_hdr_t *hdr;

	mutex_enter(hash_lock);
	for (hdr = buf_hash_table.ht_table[idx]; hdr != NULL;
	hdr = hdr->b_hash_next) {
	if (HDR_EQUAL(spa, dva, birth, hdr)) {
	*lockp = hash_lock;
	return (hdr);
	}
	}
	mutex_exit(hash_lock);
	*lockp = NULL;
	return (NULL);
	}

	/*
	* Insert an entry into the hash table. If there is already an element
	* equal to elem in the hash table, then the already existing element
	* will be returned and the new element will not be inserted.
	* Otherwise returns NULL.
	* If lockp == NULL, the caller is assumed to already hold the hash lock.
	*/
	static arc_buf_hdr_t *
	buf_hash_insert(arc_buf_hdr_t hdr, kmutex_t *lockp)
	{
	uint64_t idx = BUF_HASH_INDEX(hdr->b_spa, &hdr->b_dva, hdr->b_birth);
	kmutex_t *hash_lock = BUF_HASH_LOCK(idx);
	arc_buf_hdr_t *fhdr;
	uint32_t i;

	ASSERT(!DVA_IS_EMPTY(&hdr->b_dva));
	ASSERT(hdr->b_birth != 0);
	ASSERT(!HDR_IN_HASH_TABLE(hdr));

	if (lockp != NULL) {
	*lockp = hash_lock;
	mutex_enter(hash_lock);
	} else {
	ASSERT(MUTEX_HELD(hash_lock));
	}

	for (fhdr = buf_hash_table.ht_table[idx], i = 0; fhdr != NULL;
	fhdr = fhdr->b_hash_next, i++) {
	if (HDR_EQUAL(hdr->b_spa, &hdr->b_dva, hdr->b_birth, fhdr))
	return (fhdr);
	}

	hdr->b_hash_next = buf_hash_table.ht_table[idx];
	buf_hash_table.ht_table[idx] = hdr;
	arc_hdr_set_flags(hdr, ARC_FLAG_IN_HASH_TABLE);

	/* collect some hash table performance data */
	if (i > 0) {
	ARCSTAT_BUMP(arcstat_hash_collisions);
	if (i == 1)
	ARCSTAT_BUMP(arcstat_hash_chains);

	ARCSTAT_MAX(arcstat_hash_chain_max, i);
	}
	uint64_t he = atomic_inc_64_nv(
	&arc_stats.arcstat_hash_elements.value.ui64);
	ARCSTAT_MAX(arcstat_hash_elements_max, he);

	return (NULL);
	}

	static void
	buf_hash_remove(arc_buf_hdr_t *hdr)
	{
	arc_buf_hdr_t fhdr, *hdrp;
	uint64_t idx = BUF_HASH_INDEX(hdr->b_spa, &hdr->b_dva, hdr->b_birth);

	ASSERT(MUTEX_HELD(BUF_HASH_LOCK(idx)));
	ASSERT(HDR_IN_HASH_TABLE(hdr));

	hdrp = &buf_hash_table.ht_table[idx];
	while ((fhdr = *hdrp) != hdr) {
	ASSERT3P(fhdr, !=, NULL);
	hdrp = &fhdr->b_hash_next;
	}
	*hdrp = hdr->b_hash_next;
	hdr->b_hash_next = NULL;
	arc_hdr_clear_flags(hdr, ARC_FLAG_IN_HASH_TABLE);

	/* collect some hash table performance data */
	atomic_dec_64(&arc_stats.arcstat_hash_elements.value.ui64);

	if (buf_hash_table.ht_table[idx] &&
	buf_hash_table.ht_table[idx]->b_hash_next == NULL)
	ARCSTAT_BUMPDOWN(arcstat_hash_chains);
	}

	/*
	* Global data structures and functions for the buf kmem cache.
	*/

	static kmem_cache_t *hdr_full_cache;
	static kmem_cache_t *hdr_full_crypt_cache;
	static kmem_cache_t *hdr_l2only_cache;
	static kmem_cache_t *buf_cache;

	static void
	buf_fini(void)
	{
	int i;

	#if defined(_KERNEL)
	/*
	* Large allocations which do not require contiguous pages
	* should be using vmem_free() in the linux kernel\
	*/
	vmem_free(buf_hash_table.ht_table,
	(buf_hash_table.ht_mask + 1) * sizeof (void *));
	#else
	kmem_free(buf_hash_table.ht_table,
	(buf_hash_table.ht_mask + 1) * sizeof (void *));
	#endif
	for (i = 0; i < BUF_LOCKS; i++)
	mutex_destroy(BUF_HASH_LOCK(i));
	kmem_cache_destroy(hdr_full_cache);
	kmem_cache_destroy(hdr_full_crypt_cache);
	kmem_cache_destroy(hdr_l2only_cache);
	kmem_cache_destroy(buf_cache);
	}

	/*
	* Constructor callback - called when the cache is empty
	* and a new buf is requested.
	*/
	/* ARGSUSED */
	static int
	hdr_full_cons(void vbuf, void unused, int kmflag)
	{
	arc_buf_hdr_t *hdr = vbuf;

	bzero(hdr, HDR_FULL_SIZE);
	hdr->b_l1hdr.b_byteswap = DMU_BSWAP_NUMFUNCS;
	cv_init(&hdr->b_l1hdr.b_cv, NULL, CV_DEFAULT, NULL);
	zfs_refcount_create(&hdr->b_l1hdr.b_refcnt);
	mutex_init(&hdr->b_l1hdr.b_freeze_lock, NULL, MUTEX_DEFAULT, NULL);
	list_link_init(&hdr->b_l1hdr.b_arc_node);
	list_link_init(&hdr->b_l2hdr.b_l2node);
	multilist_link_init(&hdr->b_l1hdr.b_arc_node);
	arc_space_consume(HDR_FULL_SIZE, ARC_SPACE_HDRS);

	return (0);
	}

	/* ARGSUSED */
	static int
	hdr_full_crypt_cons(void vbuf, void unused, int kmflag)
	{
	arc_buf_hdr_t *hdr = vbuf;

	hdr_full_cons(vbuf, unused, kmflag);
	bzero(&hdr->b_crypt_hdr, sizeof (hdr->b_crypt_hdr));
	arc_space_consume(sizeof (hdr->b_crypt_hdr), ARC_SPACE_HDRS);

	return (0);
	}

	/* ARGSUSED */
	static int
	hdr_l2only_cons(void vbuf, void unused, int kmflag)
	{
	arc_buf_hdr_t *hdr = vbuf;

	bzero(hdr, HDR_L2ONLY_SIZE);
	arc_space_consume(HDR_L2ONLY_SIZE, ARC_SPACE_L2HDRS);

	return (0);
	}

	/* ARGSUSED */
	static int
	buf_cons(void vbuf, void unused, int kmflag)
	{
	arc_buf_t *buf = vbuf;

	bzero(buf, sizeof (arc_buf_t));
	mutex_init(&buf->b_evict_lock, NULL, MUTEX_DEFAULT, NULL);
	arc_space_consume(sizeof (arc_buf_t), ARC_SPACE_HDRS);

	return (0);
	}

	/*
	* Destructor callback - called when a cached buf is
	* no longer required.
	*/
	/* ARGSUSED */
	static void
	hdr_full_dest(void vbuf, void unused)
	{
	arc_buf_hdr_t *hdr = vbuf;

	ASSERT(HDR_EMPTY(hdr));
	cv_destroy(&hdr->b_l1hdr.b_cv);
	zfs_refcount_destroy(&hdr->b_l1hdr.b_refcnt);
	mutex_destroy(&hdr->b_l1hdr.b_freeze_lock);
	ASSERT(!multilist_link_active(&hdr->b_l1hdr.b_arc_node));
	arc_space_return(HDR_FULL_SIZE, ARC_SPACE_HDRS);
	}

	/* ARGSUSED */
	static void
	hdr_full_crypt_dest(void vbuf, void unused)
	{
	arc_buf_hdr_t *hdr = vbuf;

	hdr_full_dest(vbuf, unused);
	arc_space_return(sizeof (hdr->b_crypt_hdr), ARC_SPACE_HDRS);
	}

	/* ARGSUSED */
	static void
	hdr_l2only_dest(void vbuf, void unused)
	{
	arc_buf_hdr_t *hdr __maybe_unused = vbuf;

	ASSERT(HDR_EMPTY(hdr));
	arc_space_return(HDR_L2ONLY_SIZE, ARC_SPACE_L2HDRS);
	}

	/* ARGSUSED */
	static void
	buf_dest(void vbuf, void unused)
	{
	arc_buf_t *buf = vbuf;

	mutex_destroy(&buf->b_evict_lock);
	arc_space_return(sizeof (arc_buf_t), ARC_SPACE_HDRS);
	}

	static void
	buf_init(void)
	{
	uint64_t *ct = NULL;
	uint64_t hsize = 1ULL << 12;
	int i, j;

	/*
	* The hash table is big enough to fill all of physical memory
	* with an average block size of zfs_arc_average_blocksize (default 8K).
	* By default, the table will take up
	* totalmem * sizeof(void*) / 8K (1MB per GB with 8-byte pointers).
	*/
	while (hsize * zfs_arc_average_blocksize < arc_all_memory())
	hsize <<= 1;
	retry:
	buf_hash_table.ht_mask = hsize - 1;
	#if defined(_KERNEL)
	/*
	* Large allocations which do not require contiguous pages
	* should be using vmem_alloc() in the linux kernel
	*/
	buf_hash_table.ht_table =
	vmem_zalloc(hsize * sizeof (void*), KM_SLEEP);
	#else
	buf_hash_table.ht_table =
	kmem_zalloc(hsize * sizeof (void*), KM_NOSLEEP);
	#endif
	if (buf_hash_table.ht_table == NULL) {
	ASSERT(hsize > (1ULL << 8));
	hsize >>= 1;
	goto retry;
	}

	hdr_full_cache = kmem_cache_create("arc_buf_hdr_t_full", HDR_FULL_SIZE,
	0, hdr_full_cons, hdr_full_dest, NULL, NULL, NULL, 0);
	hdr_full_crypt_cache = kmem_cache_create("arc_buf_hdr_t_full_crypt",
	HDR_FULL_CRYPT_SIZE, 0, hdr_full_crypt_cons, hdr_full_crypt_dest,
	NULL, NULL, NULL, 0);
	hdr_l2only_cache = kmem_cache_create("arc_buf_hdr_t_l2only",
	HDR_L2ONLY_SIZE, 0, hdr_l2only_cons, hdr_l2only_dest, NULL,
	NULL, NULL, 0);
	buf_cache = kmem_cache_create("arc_buf_t", sizeof (arc_buf_t),
	0, buf_cons, buf_dest, NULL, NULL, NULL, 0);

	for (i = 0; i < 256; i++)
	for (ct = zfs_crc64_table + i, *ct = i, j = 8; j > 0; j--)
	ct = (ct >> 1) ^ (-(*ct & 1) & ZFS_CRC64_POLY);

	for (i = 0; i < BUF_LOCKS; i++)
	mutex_init(BUF_HASH_LOCK(i), NULL, MUTEX_DEFAULT, NULL);
	}

	#define ARC_MINTIME (hz>>4) /* 62 ms */

	/*
	* This is the size that the buf occupies in memory. If the buf is compressed,
	* it will correspond to the compressed size. You should use this method of
	* getting the buf size unless you explicitly need the logical size.
	*/
	uint64_t
	arc_buf_size(arc_buf_t *buf)
	{
	return (ARC_BUF_COMPRESSED(buf) ?
	HDR_GET_PSIZE(buf->b_hdr) : HDR_GET_LSIZE(buf->b_hdr));
	}

	uint64_t
	arc_buf_lsize(arc_buf_t *buf)
	{
	return (HDR_GET_LSIZE(buf->b_hdr));
	}

	/*
	* This function will return B_TRUE if the buffer is encrypted in memory.
	* This buffer can be decrypted by calling arc_untransform().
	*/
	boolean_t
	arc_is_encrypted(arc_buf_t *buf)
	{
	return (ARC_BUF_ENCRYPTED(buf) != 0);
	}

	/*
	* Returns B_TRUE if the buffer represents data that has not had its MAC
	* verified yet.
	*/
	boolean_t
	arc_is_unauthenticated(arc_buf_t *buf)
	{
	return (HDR_NOAUTH(buf->b_hdr) != 0);
	}

	void
	arc_get_raw_params(arc_buf_t buf, boolean_t byteorder, uint8_t *salt,
	uint8_t iv, uint8_t mac)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	ASSERT(HDR_PROTECTED(hdr));

	bcopy(hdr->b_crypt_hdr.b_salt, salt, ZIO_DATA_SALT_LEN);
	bcopy(hdr->b_crypt_hdr.b_iv, iv, ZIO_DATA_IV_LEN);
	bcopy(hdr->b_crypt_hdr.b_mac, mac, ZIO_DATA_MAC_LEN);
	*byteorder = (hdr->b_l1hdr.b_byteswap == DMU_BSWAP_NUMFUNCS) ?
	ZFS_HOST_BYTEORDER : !ZFS_HOST_BYTEORDER;
	}

	/*
	* Indicates how this buffer is compressed in memory. If it is not compressed
	* the value will be ZIO_COMPRESS_OFF. It can be made normally readable with
	* arc_untransform() as long as it is also unencrypted.
	*/
	enum zio_compress
	arc_get_compression(arc_buf_t *buf)
	{
	return (ARC_BUF_COMPRESSED(buf) ?
	HDR_GET_COMPRESS(buf->b_hdr) : ZIO_COMPRESS_OFF);
	}

	/*
	* Return the compression algorithm used to store this data in the ARC. If ARC
	* compression is enabled or this is an encrypted block, this will be the same
	* as what's used to store it on-disk. Otherwise, this will be ZIO_COMPRESS_OFF.
	*/
	static inline enum zio_compress
	arc_hdr_get_compress(arc_buf_hdr_t *hdr)
	{
	return (HDR_COMPRESSION_ENABLED(hdr) ?
	HDR_GET_COMPRESS(hdr) : ZIO_COMPRESS_OFF);
	}

	uint8_t
	arc_get_complevel(arc_buf_t *buf)
	{
	return (buf->b_hdr->b_complevel);
	}

	static inline boolean_t
	arc_buf_is_shared(arc_buf_t *buf)
	{
	boolean_t shared = (buf->b_data != NULL &&
	buf->b_hdr->b_l1hdr.b_pabd != NULL &&
	abd_is_linear(buf->b_hdr->b_l1hdr.b_pabd) &&
	buf->b_data == abd_to_buf(buf->b_hdr->b_l1hdr.b_pabd));
	IMPLY(shared, HDR_SHARED_DATA(buf->b_hdr));
	IMPLY(shared, ARC_BUF_SHARED(buf));
	IMPLY(shared, ARC_BUF_COMPRESSED(buf) \|\| ARC_BUF_LAST(buf));

	/*
	* It would be nice to assert arc_can_share() too, but the "hdr isn't
	* already being shared" requirement prevents us from doing that.
	*/

	return (shared);
	}

	/*
	* Free the checksum associated with this header. If there is no checksum, this
	* is a no-op.
	*/
	static inline void
	arc_cksum_free(arc_buf_hdr_t *hdr)
	{
	ASSERT(HDR_HAS_L1HDR(hdr));

	mutex_enter(&hdr->b_l1hdr.b_freeze_lock);
	if (hdr->b_l1hdr.b_freeze_cksum != NULL) {
	kmem_free(hdr->b_l1hdr.b_freeze_cksum, sizeof (zio_cksum_t));
	hdr->b_l1hdr.b_freeze_cksum = NULL;
	}
	mutex_exit(&hdr->b_l1hdr.b_freeze_lock);
	}

	/*
	* Return true iff at least one of the bufs on hdr is not compressed.
	* Encrypted buffers count as compressed.
	*/
	static boolean_t
	arc_hdr_has_uncompressed_buf(arc_buf_hdr_t *hdr)
	{
	ASSERT(hdr->b_l1hdr.b_state == arc_anon \|\| HDR_EMPTY_OR_LOCKED(hdr));

	for (arc_buf_t *b = hdr->b_l1hdr.b_buf; b != NULL; b = b->b_next) {
	if (!ARC_BUF_COMPRESSED(b)) {
	return (B_TRUE);
	}
	}
	return (B_FALSE);
	}


	/*
	* If we've turned on the ZFS_DEBUG_MODIFY flag, verify that the buf's data
	* matches the checksum that is stored in the hdr. If there is no checksum,
	* or if the buf is compressed, this is a no-op.
	*/
	static void
	arc_cksum_verify(arc_buf_t *buf)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;
	zio_cksum_t zc;

	if (!(zfs_flags & ZFS_DEBUG_MODIFY))
	return;

	if (ARC_BUF_COMPRESSED(buf))
	return;

	ASSERT(HDR_HAS_L1HDR(hdr));

	mutex_enter(&hdr->b_l1hdr.b_freeze_lock);

	if (hdr->b_l1hdr.b_freeze_cksum == NULL \|\| HDR_IO_ERROR(hdr)) {
	mutex_exit(&hdr->b_l1hdr.b_freeze_lock);
	return;
	}

	fletcher_2_native(buf->b_data, arc_buf_size(buf), NULL, &zc);
	if (!ZIO_CHECKSUM_EQUAL(*hdr->b_l1hdr.b_freeze_cksum, zc))
	panic("buffer modified while frozen!");
	mutex_exit(&hdr->b_l1hdr.b_freeze_lock);
	}

	/*
	* This function makes the assumption that data stored in the L2ARC
	* will be transformed exactly as it is in the main pool. Because of
	* this we can verify the checksum against the reading process's bp.
	*/
	static boolean_t
	arc_cksum_is_equal(arc_buf_hdr_t hdr, zio_t zio)
	{
	ASSERT(!BP_IS_EMBEDDED(zio->io_bp));
	VERIFY3U(BP_GET_PSIZE(zio->io_bp), ==, HDR_GET_PSIZE(hdr));

	/*
	* Block pointers always store the checksum for the logical data.
	* If the block pointer has the gang bit set, then the checksum
	* it represents is for the reconstituted data and not for an
	* individual gang member. The zio pipeline, however, must be able to
	* determine the checksum of each of the gang constituents so it
	* treats the checksum comparison differently than what we need
	* for l2arc blocks. This prevents us from using the
	* zio_checksum_error() interface directly. Instead we must call the
	* zio_checksum_error_impl() so that we can ensure the checksum is
	* generated using the correct checksum algorithm and accounts for the
	* logical I/O size and not just a gang fragment.
	*/
	return (zio_checksum_error_impl(zio->io_spa, zio->io_bp,
	BP_GET_CHECKSUM(zio->io_bp), zio->io_abd, zio->io_size,
	zio->io_offset, NULL) == 0);
	}

	/*
	* Given a buf full of data, if ZFS_DEBUG_MODIFY is enabled this computes a
	* checksum and attaches it to the buf's hdr so that we can ensure that the buf
	* isn't modified later on. If buf is compressed or there is already a checksum
	* on the hdr, this is a no-op (we only checksum uncompressed bufs).
	*/
	static void
	arc_cksum_compute(arc_buf_t *buf)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	if (!(zfs_flags & ZFS_DEBUG_MODIFY))
	return;

	ASSERT(HDR_HAS_L1HDR(hdr));

	mutex_enter(&buf->b_hdr->b_l1hdr.b_freeze_lock);
	if (hdr->b_l1hdr.b_freeze_cksum != NULL \|\| ARC_BUF_COMPRESSED(buf)) {
	mutex_exit(&hdr->b_l1hdr.b_freeze_lock);
	return;
	}

	ASSERT(!ARC_BUF_ENCRYPTED(buf));
	ASSERT(!ARC_BUF_COMPRESSED(buf));
	hdr->b_l1hdr.b_freeze_cksum = kmem_alloc(sizeof (zio_cksum_t),
	KM_SLEEP);
	fletcher_2_native(buf->b_data, arc_buf_size(buf), NULL,
	hdr->b_l1hdr.b_freeze_cksum);
	mutex_exit(&hdr->b_l1hdr.b_freeze_lock);
	arc_buf_watch(buf);
	}

	#ifndef _KERNEL
	void
	arc_buf_sigsegv(int sig, siginfo_t si, void unused)
	{
	panic("Got SIGSEGV at address: 0x%lx\n", (long)si->si_addr);
	}
	#endif

	/* ARGSUSED */
	static void
	arc_buf_unwatch(arc_buf_t *buf)
	{
	#ifndef _KERNEL
	if (arc_watch) {
	ASSERT0(mprotect(buf->b_data, arc_buf_size(buf),
	PROT_READ \| PROT_WRITE));
	}
	#endif
	}

	/* ARGSUSED */
	static void
	arc_buf_watch(arc_buf_t *buf)
	{
	#ifndef _KERNEL
	if (arc_watch)
	ASSERT0(mprotect(buf->b_data, arc_buf_size(buf),
	PROT_READ));
	#endif
	}

	static arc_buf_contents_t
	arc_buf_type(arc_buf_hdr_t *hdr)
	{
	arc_buf_contents_t type;
	if (HDR_ISTYPE_METADATA(hdr)) {
	type = ARC_BUFC_METADATA;
	} else {
	type = ARC_BUFC_DATA;
	}
	VERIFY3U(hdr->b_type, ==, type);
	return (type);
	}

	boolean_t
	arc_is_metadata(arc_buf_t *buf)
	{
	return (HDR_ISTYPE_METADATA(buf->b_hdr) != 0);
	}

	static uint32_t
	arc_bufc_to_flags(arc_buf_contents_t type)
	{
	switch (type) {
	case ARC_BUFC_DATA:
	/* metadata field is 0 if buffer contains normal data */
	return (0);
	case ARC_BUFC_METADATA:
	return (ARC_FLAG_BUFC_METADATA);
	default:
	break;
	}
	panic("undefined ARC buffer type!");
	return ((uint32_t)-1);
	}

	void
	arc_buf_thaw(arc_buf_t *buf)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	ASSERT3P(hdr->b_l1hdr.b_state, ==, arc_anon);
	ASSERT(!HDR_IO_IN_PROGRESS(hdr));

	arc_cksum_verify(buf);

	/*
	* Compressed buffers do not manipulate the b_freeze_cksum.
	*/
	if (ARC_BUF_COMPRESSED(buf))
	return;

	ASSERT(HDR_HAS_L1HDR(hdr));
	arc_cksum_free(hdr);
	arc_buf_unwatch(buf);
	}

	void
	arc_buf_freeze(arc_buf_t *buf)
	{
	if (!(zfs_flags & ZFS_DEBUG_MODIFY))
	return;

	if (ARC_BUF_COMPRESSED(buf))
	return;

	ASSERT(HDR_HAS_L1HDR(buf->b_hdr));
	arc_cksum_compute(buf);
	}

	/*
	* The arc_buf_hdr_t's b_flags should never be modified directly. Instead,
	* the following functions should be used to ensure that the flags are
	* updated in a thread-safe way. When manipulating the flags either
	* the hash_lock must be held or the hdr must be undiscoverable. This
	* ensures that we're not racing with any other threads when updating
	* the flags.
	*/
	static inline void
	arc_hdr_set_flags(arc_buf_hdr_t *hdr, arc_flags_t flags)
	{
	ASSERT(HDR_EMPTY_OR_LOCKED(hdr));
	hdr->b_flags \|= flags;
	}

	static inline void
	arc_hdr_clear_flags(arc_buf_hdr_t *hdr, arc_flags_t flags)
	{
	ASSERT(HDR_EMPTY_OR_LOCKED(hdr));
	hdr->b_flags &= ~flags;
	}

	/*
	* Setting the compression bits in the arc_buf_hdr_t's b_flags is
	* done in a special way since we have to clear and set bits
	* at the same time. Consumers that wish to set the compression bits
	* must use this function to ensure that the flags are updated in
	* thread-safe manner.
	*/
	static void
	arc_hdr_set_compress(arc_buf_hdr_t *hdr, enum zio_compress cmp)
	{
	ASSERT(HDR_EMPTY_OR_LOCKED(hdr));

	/*
	* Holes and embedded blocks will always have a psize = 0 so
	* we ignore the compression of the blkptr and set the
	* want to uncompress them. Mark them as uncompressed.
	*/
	if (!zfs_compressed_arc_enabled \|\| HDR_GET_PSIZE(hdr) == 0) {
	arc_hdr_clear_flags(hdr, ARC_FLAG_COMPRESSED_ARC);
	ASSERT(!HDR_COMPRESSION_ENABLED(hdr));
	} else {
	arc_hdr_set_flags(hdr, ARC_FLAG_COMPRESSED_ARC);
	ASSERT(HDR_COMPRESSION_ENABLED(hdr));
	}

	HDR_SET_COMPRESS(hdr, cmp);
	ASSERT3U(HDR_GET_COMPRESS(hdr), ==, cmp);
	}

	/*
	* Looks for another buf on the same hdr which has the data decompressed, copies
	* from it, and returns true. If no such buf exists, returns false.
	*/
	static boolean_t
	arc_buf_try_copy_decompressed_data(arc_buf_t *buf)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;
	boolean_t copied = B_FALSE;

	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT3P(buf->b_data, !=, NULL);
	ASSERT(!ARC_BUF_COMPRESSED(buf));

	for (arc_buf_t *from = hdr->b_l1hdr.b_buf; from != NULL;
	from = from->b_next) {
	/* can't use our own data buffer */
	if (from == buf) {
	continue;
	}

	if (!ARC_BUF_COMPRESSED(from)) {
	bcopy(from->b_data, buf->b_data, arc_buf_size(buf));
	copied = B_TRUE;
	break;
	}
	}

	/*
	* There were no decompressed bufs, so there should not be a
	* checksum on the hdr either.
	*/
	if (zfs_flags & ZFS_DEBUG_MODIFY)
	EQUIV(!copied, hdr->b_l1hdr.b_freeze_cksum == NULL);

	return (copied);
	}

	/*
	* Allocates an ARC buf header that's in an evicted & L2-cached state.
	* This is used during l2arc reconstruction to make empty ARC buffers
	* which circumvent the regular disk->arc->l2arc path and instead come
	* into being in the reverse order, i.e. l2arc->arc.
	*/
	static arc_buf_hdr_t *
	arc_buf_alloc_l2only(size_t size, arc_buf_contents_t type, l2arc_dev_t *dev,
	dva_t dva, uint64_t daddr, int32_t psize, uint64_t birth,
	enum zio_compress compress, uint8_t complevel, boolean_t protected,
	boolean_t prefetch, arc_state_type_t arcs_state)
	{
	arc_buf_hdr_t *hdr;

	ASSERT(size != 0);
	hdr = kmem_cache_alloc(hdr_l2only_cache, KM_SLEEP);
	hdr->b_birth = birth;
	hdr->b_type = type;
	hdr->b_flags = 0;
	arc_hdr_set_flags(hdr, arc_bufc_to_flags(type) \| ARC_FLAG_HAS_L2HDR);
	HDR_SET_LSIZE(hdr, size);
	HDR_SET_PSIZE(hdr, psize);
	arc_hdr_set_compress(hdr, compress);
	hdr->b_complevel = complevel;
	if (protected)
	arc_hdr_set_flags(hdr, ARC_FLAG_PROTECTED);
	if (prefetch)
	arc_hdr_set_flags(hdr, ARC_FLAG_PREFETCH);
	hdr->b_spa = spa_load_guid(dev->l2ad_vdev->vdev_spa);

	hdr->b_dva = dva;

	hdr->b_l2hdr.b_dev = dev;
	hdr->b_l2hdr.b_daddr = daddr;
	hdr->b_l2hdr.b_arcs_state = arcs_state;

	return (hdr);
	}

	/*
	* Return the size of the block, b_pabd, that is stored in the arc_buf_hdr_t.
	*/
	static uint64_t
	arc_hdr_size(arc_buf_hdr_t *hdr)
	{
	uint64_t size;

	if (arc_hdr_get_compress(hdr) != ZIO_COMPRESS_OFF &&
	HDR_GET_PSIZE(hdr) > 0) {
	size = HDR_GET_PSIZE(hdr);
	} else {
	ASSERT3U(HDR_GET_LSIZE(hdr), !=, 0);
	size = HDR_GET_LSIZE(hdr);
	}
	return (size);
	}

	static int
	arc_hdr_authenticate(arc_buf_hdr_t hdr, spa_t spa, uint64_t dsobj)
	{
	int ret;
	uint64_t csize;
	uint64_t lsize = HDR_GET_LSIZE(hdr);
	uint64_t psize = HDR_GET_PSIZE(hdr);
	void *tmpbuf = NULL;
	abd_t *abd = hdr->b_l1hdr.b_pabd;

	ASSERT(HDR_EMPTY_OR_LOCKED(hdr));
	ASSERT(HDR_AUTHENTICATED(hdr));
	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);

	/*
	* The MAC is calculated on the compressed data that is stored on disk.
	* However, if compressed arc is disabled we will only have the
	* decompressed data available to us now. Compress it into a temporary
	* abd so we can verify the MAC. The performance overhead of this will
	* be relatively low, since most objects in an encrypted objset will
	* be encrypted (instead of authenticated) anyway.
	*/
	if (HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF &&
	!HDR_COMPRESSION_ENABLED(hdr)) {
	tmpbuf = zio_buf_alloc(lsize);
	abd = abd_get_from_buf(tmpbuf, lsize);
	abd_take_ownership_of_buf(abd, B_TRUE);
	csize = zio_compress_data(HDR_GET_COMPRESS(hdr),
	hdr->b_l1hdr.b_pabd, tmpbuf, lsize, hdr->b_complevel);
	ASSERT3U(csize, <=, psize);
	abd_zero_off(abd, csize, psize - csize);
	}

	/*
	* Authentication is best effort. We authenticate whenever the key is
	* available. If we succeed we clear ARC_FLAG_NOAUTH.
	*/
	if (hdr->b_crypt_hdr.b_ot == DMU_OT_OBJSET) {
	ASSERT3U(HDR_GET_COMPRESS(hdr), ==, ZIO_COMPRESS_OFF);
	ASSERT3U(lsize, ==, psize);
	ret = spa_do_crypt_objset_mac_abd(B_FALSE, spa, dsobj, abd,
	psize, hdr->b_l1hdr.b_byteswap != DMU_BSWAP_NUMFUNCS);
	} else {
	ret = spa_do_crypt_mac_abd(B_FALSE, spa, dsobj, abd, psize,
	hdr->b_crypt_hdr.b_mac);
	}

	if (ret == 0)
	arc_hdr_clear_flags(hdr, ARC_FLAG_NOAUTH);
	else if (ret != ENOENT)
	goto error;

	if (tmpbuf != NULL)
	abd_free(abd);

	return (0);

	error:
	if (tmpbuf != NULL)
	abd_free(abd);

	return (ret);
	}

	/*
	* This function will take a header that only has raw encrypted data in
	* b_crypt_hdr.b_rabd and decrypt it into a new buffer which is stored in
	* b_l1hdr.b_pabd. If designated in the header flags, this function will
	* also decompress the data.
	*/
	static int
	arc_hdr_decrypt(arc_buf_hdr_t hdr, spa_t spa, const zbookmark_phys_t *zb)
	{
	int ret;
	abd_t *cabd = NULL;
	void *tmp = NULL;
	boolean_t no_crypt = B_FALSE;
	boolean_t bswap = (hdr->b_l1hdr.b_byteswap != DMU_BSWAP_NUMFUNCS);

	ASSERT(HDR_EMPTY_OR_LOCKED(hdr));
	ASSERT(HDR_ENCRYPTED(hdr));

	arc_hdr_alloc_abd(hdr, ARC_HDR_DO_ADAPT);

	ret = spa_do_crypt_abd(B_FALSE, spa, zb, hdr->b_crypt_hdr.b_ot,
	B_FALSE, bswap, hdr->b_crypt_hdr.b_salt, hdr->b_crypt_hdr.b_iv,
	hdr->b_crypt_hdr.b_mac, HDR_GET_PSIZE(hdr), hdr->b_l1hdr.b_pabd,
	hdr->b_crypt_hdr.b_rabd, &no_crypt);
	if (ret != 0)
	goto error;

	if (no_crypt) {
	abd_copy(hdr->b_l1hdr.b_pabd, hdr->b_crypt_hdr.b_rabd,
	HDR_GET_PSIZE(hdr));
	}

	/*
	* If this header has disabled arc compression but the b_pabd is
	* compressed after decrypting it, we need to decompress the newly
	* decrypted data.
	*/
	if (HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF &&
	!HDR_COMPRESSION_ENABLED(hdr)) {
	/*
	* We want to make sure that we are correctly honoring the
	* zfs_abd_scatter_enabled setting, so we allocate an abd here
	* and then loan a buffer from it, rather than allocating a
	* linear buffer and wrapping it in an abd later.
	*/
	cabd = arc_get_data_abd(hdr, arc_hdr_size(hdr), hdr,
	ARC_HDR_DO_ADAPT);
	tmp = abd_borrow_buf(cabd, arc_hdr_size(hdr));

	ret = zio_decompress_data(HDR_GET_COMPRESS(hdr),
	hdr->b_l1hdr.b_pabd, tmp, HDR_GET_PSIZE(hdr),
	HDR_GET_LSIZE(hdr), &hdr->b_complevel);
	if (ret != 0) {
	abd_return_buf(cabd, tmp, arc_hdr_size(hdr));
	goto error;
	}

	abd_return_buf_copy(cabd, tmp, arc_hdr_size(hdr));
	arc_free_data_abd(hdr, hdr->b_l1hdr.b_pabd,
	arc_hdr_size(hdr), hdr);
	hdr->b_l1hdr.b_pabd = cabd;
	}

	return (0);

	error:
	arc_hdr_free_abd(hdr, B_FALSE);
	if (cabd != NULL)
	arc_free_data_buf(hdr, cabd, arc_hdr_size(hdr), hdr);

	return (ret);
	}

	/*
	* This function is called during arc_buf_fill() to prepare the header's
	* abd plaintext pointer for use. This involves authenticated protected
	* data and decrypting encrypted data into the plaintext abd.
	*/
	static int
	arc_fill_hdr_crypt(arc_buf_hdr_t hdr, kmutex_t hash_lock, spa_t *spa,
	const zbookmark_phys_t *zb, boolean_t noauth)
	{
	int ret;

	ASSERT(HDR_PROTECTED(hdr));

	if (hash_lock != NULL)
	mutex_enter(hash_lock);

	if (HDR_NOAUTH(hdr) && !noauth) {
	/*
	* The caller requested authenticated data but our data has
	* not been authenticated yet. Verify the MAC now if we can.
	*/
	ret = arc_hdr_authenticate(hdr, spa, zb->zb_objset);
	if (ret != 0)
	goto error;
	} else if (HDR_HAS_RABD(hdr) && hdr->b_l1hdr.b_pabd == NULL) {
	/*
	* If we only have the encrypted version of the data, but the
	* unencrypted version was requested we take this opportunity
	* to store the decrypted version in the header for future use.
	*/
	ret = arc_hdr_decrypt(hdr, spa, zb);
	if (ret != 0)
	goto error;
	}

	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);

	if (hash_lock != NULL)
	mutex_exit(hash_lock);

	return (0);

	error:
	if (hash_lock != NULL)
	mutex_exit(hash_lock);

	return (ret);
	}

	/*
	* This function is used by the dbuf code to decrypt bonus buffers in place.
	* The dbuf code itself doesn't have any locking for decrypting a shared dnode
	* block, so we use the hash lock here to protect against concurrent calls to
	* arc_buf_fill().
	*/
	static void
	arc_buf_untransform_in_place(arc_buf_t buf, kmutex_t hash_lock)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	ASSERT(HDR_ENCRYPTED(hdr));
	ASSERT3U(hdr->b_crypt_hdr.b_ot, ==, DMU_OT_DNODE);
	ASSERT(HDR_EMPTY_OR_LOCKED(hdr));
	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);

	zio_crypt_copy_dnode_bonus(hdr->b_l1hdr.b_pabd, buf->b_data,
	arc_buf_size(buf));
	buf->b_flags &= ~ARC_BUF_FLAG_ENCRYPTED;
	buf->b_flags &= ~ARC_BUF_FLAG_COMPRESSED;
	hdr->b_crypt_hdr.b_ebufcnt -= 1;
	}

	/*
	* Given a buf that has a data buffer attached to it, this function will
	* efficiently fill the buf with data of the specified compression setting from
	* the hdr and update the hdr's b_freeze_cksum if necessary. If the buf and hdr
	* are already sharing a data buf, no copy is performed.
	*
	* If the buf is marked as compressed but uncompressed data was requested, this
	* will allocate a new data buffer for the buf, remove that flag, and fill the
	* buf with uncompressed data. You can't request a compressed buf on a hdr with
	* uncompressed data, and (since we haven't added support for it yet) if you
	* want compressed data your buf must already be marked as compressed and have
	* the correct-sized data buffer.
	*/
	static int
	arc_buf_fill(arc_buf_t buf, spa_t spa, const zbookmark_phys_t *zb,
	arc_fill_flags_t flags)
	{
	int error = 0;
	arc_buf_hdr_t *hdr = buf->b_hdr;
	boolean_t hdr_compressed =
	(arc_hdr_get_compress(hdr) != ZIO_COMPRESS_OFF);
	boolean_t compressed = (flags & ARC_FILL_COMPRESSED) != 0;
	boolean_t encrypted = (flags & ARC_FILL_ENCRYPTED) != 0;
	dmu_object_byteswap_t bswap = hdr->b_l1hdr.b_byteswap;
	kmutex_t *hash_lock = (flags & ARC_FILL_LOCKED) ? NULL : HDR_LOCK(hdr);

	ASSERT3P(buf->b_data, !=, NULL);
	IMPLY(compressed, hdr_compressed \|\| ARC_BUF_ENCRYPTED(buf));
	IMPLY(compressed, ARC_BUF_COMPRESSED(buf));
	IMPLY(encrypted, HDR_ENCRYPTED(hdr));
	IMPLY(encrypted, ARC_BUF_ENCRYPTED(buf));
	IMPLY(encrypted, ARC_BUF_COMPRESSED(buf));
	IMPLY(encrypted, !ARC_BUF_SHARED(buf));

	/*
	* If the caller wanted encrypted data we just need to copy it from
	* b_rabd and potentially byteswap it. We won't be able to do any
	* further transforms on it.
	*/
	if (encrypted) {
	ASSERT(HDR_HAS_RABD(hdr));
	abd_copy_to_buf(buf->b_data, hdr->b_crypt_hdr.b_rabd,
	HDR_GET_PSIZE(hdr));
	goto byteswap;
	}

	/*
	* Adjust encrypted and authenticated headers to accommodate
	* the request if needed. Dnode blocks (ARC_FILL_IN_PLACE) are
	* allowed to fail decryption due to keys not being loaded
	* without being marked as an IO error.
	*/
	if (HDR_PROTECTED(hdr)) {
	error = arc_fill_hdr_crypt(hdr, hash_lock, spa,
	zb, !!(flags & ARC_FILL_NOAUTH));
	if (error == EACCES && (flags & ARC_FILL_IN_PLACE) != 0) {
	return (error);
	} else if (error != 0) {
	if (hash_lock != NULL)
	mutex_enter(hash_lock);
	arc_hdr_set_flags(hdr, ARC_FLAG_IO_ERROR);
	if (hash_lock != NULL)
	mutex_exit(hash_lock);
	return (error);
	}
	}

	/*
	* There is a special case here for dnode blocks which are
	* decrypting their bonus buffers. These blocks may request to
	* be decrypted in-place. This is necessary because there may
	* be many dnodes pointing into this buffer and there is
	* currently no method to synchronize replacing the backing
	* b_data buffer and updating all of the pointers. Here we use
	* the hash lock to ensure there are no races. If the need
	* arises for other types to be decrypted in-place, they must
	* add handling here as well.
	*/
	if ((flags & ARC_FILL_IN_PLACE) != 0) {
	ASSERT(!hdr_compressed);
	ASSERT(!compressed);
	ASSERT(!encrypted);

	if (HDR_ENCRYPTED(hdr) && ARC_BUF_ENCRYPTED(buf)) {
	ASSERT3U(hdr->b_crypt_hdr.b_ot, ==, DMU_OT_DNODE);

	if (hash_lock != NULL)
	mutex_enter(hash_lock);
	arc_buf_untransform_in_place(buf, hash_lock);
	if (hash_lock != NULL)
	mutex_exit(hash_lock);

	/* Compute the hdr's checksum if necessary */
	arc_cksum_compute(buf);
	}

	return (0);
	}

	if (hdr_compressed == compressed) {
	if (!arc_buf_is_shared(buf)) {
	abd_copy_to_buf(buf->b_data, hdr->b_l1hdr.b_pabd,
	arc_buf_size(buf));
	}
	} else {
	ASSERT(hdr_compressed);
	ASSERT(!compressed);
	- ASSERT3U(HDR_GET_LSIZE(hdr), !=, HDR_GET_PSIZE(hdr));

	/*
	* If the buf is sharing its data with the hdr, unlink it and
	* allocate a new data buffer for the buf.
	*/
	if (arc_buf_is_shared(buf)) {
	ASSERT(ARC_BUF_COMPRESSED(buf));

	/* We need to give the buf its own b_data */
	buf->b_flags &= ~ARC_BUF_FLAG_SHARED;
	buf->b_data =
	arc_get_data_buf(hdr, HDR_GET_LSIZE(hdr), buf);
	arc_hdr_clear_flags(hdr, ARC_FLAG_SHARED_DATA);

	/* Previously overhead was 0; just add new overhead */
	ARCSTAT_INCR(arcstat_overhead_size, HDR_GET_LSIZE(hdr));
	} else if (ARC_BUF_COMPRESSED(buf)) {
	/* We need to reallocate the buf's b_data */
	arc_free_data_buf(hdr, buf->b_data, HDR_GET_PSIZE(hdr),
	buf);
	buf->b_data =
	arc_get_data_buf(hdr, HDR_GET_LSIZE(hdr), buf);

	/* We increased the size of b_data; update overhead */
	ARCSTAT_INCR(arcstat_overhead_size,
	HDR_GET_LSIZE(hdr) - HDR_GET_PSIZE(hdr));
	}

	/*
	* Regardless of the buf's previous compression settings, it
	* should not be compressed at the end of this function.
	*/
	buf->b_flags &= ~ARC_BUF_FLAG_COMPRESSED;

	/*
	* Try copying the data from another buf which already has a
	* decompressed version. If that's not possible, it's time to
	* bite the bullet and decompress the data from the hdr.
	*/
	if (arc_buf_try_copy_decompressed_data(buf)) {
	/* Skip byteswapping and checksumming (already done) */
	return (0);
	} else {
	error = zio_decompress_data(HDR_GET_COMPRESS(hdr),
	hdr->b_l1hdr.b_pabd, buf->b_data,
	HDR_GET_PSIZE(hdr), HDR_GET_LSIZE(hdr),
	&hdr->b_complevel);

	/*
	* Absent hardware errors or software bugs, this should
	* be impossible, but log it anyway so we can debug it.
	*/
	if (error != 0) {
	zfs_dbgmsg(
	"hdr %px, compress %d, psize %d, lsize %d",
	hdr, arc_hdr_get_compress(hdr),
	HDR_GET_PSIZE(hdr), HDR_GET_LSIZE(hdr));
	if (hash_lock != NULL)
	mutex_enter(hash_lock);
	arc_hdr_set_flags(hdr, ARC_FLAG_IO_ERROR);
	if (hash_lock != NULL)
	mutex_exit(hash_lock);
	return (SET_ERROR(EIO));
	}
	}
	}

	byteswap:
	/* Byteswap the buf's data if necessary */
	if (bswap != DMU_BSWAP_NUMFUNCS) {
	ASSERT(!HDR_SHARED_DATA(hdr));
	ASSERT3U(bswap, <, DMU_BSWAP_NUMFUNCS);
	dmu_ot_byteswap[bswap].ob_func(buf->b_data, HDR_GET_LSIZE(hdr));
	}

	/* Compute the hdr's checksum if necessary */
	arc_cksum_compute(buf);

	return (0);
	}

	/*
	* If this function is being called to decrypt an encrypted buffer or verify an
	* authenticated one, the key must be loaded and a mapping must be made
	* available in the keystore via spa_keystore_create_mapping() or one of its
	* callers.
	*/
	int
	arc_untransform(arc_buf_t buf, spa_t spa, const zbookmark_phys_t *zb,
	boolean_t in_place)
	{
	int ret;
	arc_fill_flags_t flags = 0;

	if (in_place)
	flags \|= ARC_FILL_IN_PLACE;

	ret = arc_buf_fill(buf, spa, zb, flags);
	if (ret == ECKSUM) {
	/*
	* Convert authentication and decryption errors to EIO
	* (and generate an ereport) before leaving the ARC.
	*/
	ret = SET_ERROR(EIO);
	spa_log_error(spa, zb);
	(void) zfs_ereport_post(FM_EREPORT_ZFS_AUTHENTICATION,
	spa, NULL, zb, NULL, 0);
	}

	return (ret);
	}

	/*
	* Increment the amount of evictable space in the arc_state_t's refcount.
	* We account for the space used by the hdr and the arc buf individually
	* so that we can add and remove them from the refcount individually.
	*/
	static void
	arc_evictable_space_increment(arc_buf_hdr_t hdr, arc_state_t state)
	{
	arc_buf_contents_t type = arc_buf_type(hdr);

	ASSERT(HDR_HAS_L1HDR(hdr));

	if (GHOST_STATE(state)) {
	ASSERT0(hdr->b_l1hdr.b_bufcnt);
	ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	ASSERT(!HDR_HAS_RABD(hdr));
	(void) zfs_refcount_add_many(&state->arcs_esize[type],
	HDR_GET_LSIZE(hdr), hdr);
	return;
	}

	if (hdr->b_l1hdr.b_pabd != NULL) {
	(void) zfs_refcount_add_many(&state->arcs_esize[type],
	arc_hdr_size(hdr), hdr);
	}
	if (HDR_HAS_RABD(hdr)) {
	(void) zfs_refcount_add_many(&state->arcs_esize[type],
	HDR_GET_PSIZE(hdr), hdr);
	}

	for (arc_buf_t *buf = hdr->b_l1hdr.b_buf; buf != NULL;
	buf = buf->b_next) {
	if (arc_buf_is_shared(buf))
	continue;
	(void) zfs_refcount_add_many(&state->arcs_esize[type],
	arc_buf_size(buf), buf);
	}
	}

	/*
	* Decrement the amount of evictable space in the arc_state_t's refcount.
	* We account for the space used by the hdr and the arc buf individually
	* so that we can add and remove them from the refcount individually.
	*/
	static void
	arc_evictable_space_decrement(arc_buf_hdr_t hdr, arc_state_t state)
	{
	arc_buf_contents_t type = arc_buf_type(hdr);

	ASSERT(HDR_HAS_L1HDR(hdr));

	if (GHOST_STATE(state)) {
	ASSERT0(hdr->b_l1hdr.b_bufcnt);
	ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	ASSERT(!HDR_HAS_RABD(hdr));
	(void) zfs_refcount_remove_many(&state->arcs_esize[type],
	HDR_GET_LSIZE(hdr), hdr);
	return;
	}

	if (hdr->b_l1hdr.b_pabd != NULL) {
	(void) zfs_refcount_remove_many(&state->arcs_esize[type],
	arc_hdr_size(hdr), hdr);
	}
	if (HDR_HAS_RABD(hdr)) {
	(void) zfs_refcount_remove_many(&state->arcs_esize[type],
	HDR_GET_PSIZE(hdr), hdr);
	}

	for (arc_buf_t *buf = hdr->b_l1hdr.b_buf; buf != NULL;
	buf = buf->b_next) {
	if (arc_buf_is_shared(buf))
	continue;
	(void) zfs_refcount_remove_many(&state->arcs_esize[type],
	arc_buf_size(buf), buf);
	}
	}

	/*
	* Add a reference to this hdr indicating that someone is actively
	* referencing that memory. When the refcount transitions from 0 to 1,
	* we remove it from the respective arc_state_t list to indicate that
	* it is not evictable.
	*/
	static void
	add_reference(arc_buf_hdr_t hdr, void tag)
	{
	arc_state_t *state;

	ASSERT(HDR_HAS_L1HDR(hdr));
	if (!HDR_EMPTY(hdr) && !MUTEX_HELD(HDR_LOCK(hdr))) {
	ASSERT(hdr->b_l1hdr.b_state == arc_anon);
	ASSERT(zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
	ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
	}

	state = hdr->b_l1hdr.b_state;

	if ((zfs_refcount_add(&hdr->b_l1hdr.b_refcnt, tag) == 1) &&
	(state != arc_anon)) {
	/* We don't use the L2-only state list. */
	if (state != arc_l2c_only) {
	multilist_remove(&state->arcs_list[arc_buf_type(hdr)],
	hdr);
	arc_evictable_space_decrement(hdr, state);
	}
	/* remove the prefetch flag if we get a reference */
	if (HDR_HAS_L2HDR(hdr))
	l2arc_hdr_arcstats_decrement_state(hdr);
	arc_hdr_clear_flags(hdr, ARC_FLAG_PREFETCH);
	if (HDR_HAS_L2HDR(hdr))
	l2arc_hdr_arcstats_increment_state(hdr);
	}
	}

	/*
	* Remove a reference from this hdr. When the reference transitions from
	* 1 to 0 and we're not anonymous, then we add this hdr to the arc_state_t's
	* list making it eligible for eviction.
	*/
	static int
	remove_reference(arc_buf_hdr_t hdr, kmutex_t hash_lock, void *tag)
	{
	int cnt;
	arc_state_t *state = hdr->b_l1hdr.b_state;

	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT(state == arc_anon \|\| MUTEX_HELD(hash_lock));
	ASSERT(!GHOST_STATE(state));

	/*
	* arc_l2c_only counts as a ghost state so we don't need to explicitly
	* check to prevent usage of the arc_l2c_only list.
	*/
	if (((cnt = zfs_refcount_remove(&hdr->b_l1hdr.b_refcnt, tag)) == 0) &&
	(state != arc_anon)) {
	multilist_insert(&state->arcs_list[arc_buf_type(hdr)], hdr);
	ASSERT3U(hdr->b_l1hdr.b_bufcnt, >, 0);
	arc_evictable_space_increment(hdr, state);
	}
	return (cnt);
	}

	/*
	* Returns detailed information about a specific arc buffer. When the
	* state_index argument is set the function will calculate the arc header
	* list position for its arc state. Since this requires a linear traversal
	* callers are strongly encourage not to do this. However, it can be helpful
	* for targeted analysis so the functionality is provided.
	*/
	void
	arc_buf_info(arc_buf_t ab, arc_buf_info_t abi, int state_index)
	{
	arc_buf_hdr_t *hdr = ab->b_hdr;
	l1arc_buf_hdr_t *l1hdr = NULL;
	l2arc_buf_hdr_t *l2hdr = NULL;
	arc_state_t *state = NULL;

	memset(abi, 0, sizeof (arc_buf_info_t));

	if (hdr == NULL)
	return;

	abi->abi_flags = hdr->b_flags;

	if (HDR_HAS_L1HDR(hdr)) {
	l1hdr = &hdr->b_l1hdr;
	state = l1hdr->b_state;
	}
	if (HDR_HAS_L2HDR(hdr))
	l2hdr = &hdr->b_l2hdr;

	if (l1hdr) {
	abi->abi_bufcnt = l1hdr->b_bufcnt;
	abi->abi_access = l1hdr->b_arc_access;
	abi->abi_mru_hits = l1hdr->b_mru_hits;
	abi->abi_mru_ghost_hits = l1hdr->b_mru_ghost_hits;
	abi->abi_mfu_hits = l1hdr->b_mfu_hits;
	abi->abi_mfu_ghost_hits = l1hdr->b_mfu_ghost_hits;
	abi->abi_holds = zfs_refcount_count(&l1hdr->b_refcnt);
	}

	if (l2hdr) {
	abi->abi_l2arc_dattr = l2hdr->b_daddr;
	abi->abi_l2arc_hits = l2hdr->b_hits;
	}

	abi->abi_state_type = state ? state->arcs_state : ARC_STATE_ANON;
	abi->abi_state_contents = arc_buf_type(hdr);
	abi->abi_size = arc_hdr_size(hdr);
	}

	/*
	* Move the supplied buffer to the indicated state. The hash lock
	* for the buffer must be held by the caller.
	*/
	static void
	arc_change_state(arc_state_t new_state, arc_buf_hdr_t hdr,
	kmutex_t *hash_lock)
	{
	arc_state_t *old_state;
	int64_t refcnt;
	uint32_t bufcnt;
	boolean_t update_old, update_new;
	arc_buf_contents_t buftype = arc_buf_type(hdr);

	/*
	* We almost always have an L1 hdr here, since we call arc_hdr_realloc()
	* in arc_read() when bringing a buffer out of the L2ARC. However, the
	* L1 hdr doesn't always exist when we change state to arc_anon before
	* destroying a header, in which case reallocating to add the L1 hdr is
	* pointless.
	*/
	if (HDR_HAS_L1HDR(hdr)) {
	old_state = hdr->b_l1hdr.b_state;
	refcnt = zfs_refcount_count(&hdr->b_l1hdr.b_refcnt);
	bufcnt = hdr->b_l1hdr.b_bufcnt;
	update_old = (bufcnt > 0 \|\| hdr->b_l1hdr.b_pabd != NULL \|\|
	HDR_HAS_RABD(hdr));
	} else {
	old_state = arc_l2c_only;
	refcnt = 0;
	bufcnt = 0;
	update_old = B_FALSE;
	}
	update_new = update_old;

	ASSERT(MUTEX_HELD(hash_lock));
	ASSERT3P(new_state, !=, old_state);
	ASSERT(!GHOST_STATE(new_state) \|\| bufcnt == 0);
	ASSERT(old_state != arc_anon \|\| bufcnt <= 1);

	/*
	* If this buffer is evictable, transfer it from the
	* old state list to the new state list.
	*/
	if (refcnt == 0) {
	if (old_state != arc_anon && old_state != arc_l2c_only) {
	ASSERT(HDR_HAS_L1HDR(hdr));
	multilist_remove(&old_state->arcs_list[buftype], hdr);

	if (GHOST_STATE(old_state)) {
	ASSERT0(bufcnt);
	ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
	update_old = B_TRUE;
	}
	arc_evictable_space_decrement(hdr, old_state);
	}
	if (new_state != arc_anon && new_state != arc_l2c_only) {
	/*
	* An L1 header always exists here, since if we're
	* moving to some L1-cached state (i.e. not l2c_only or
	* anonymous), we realloc the header to add an L1hdr
	* beforehand.
	*/
	ASSERT(HDR_HAS_L1HDR(hdr));
	multilist_insert(&new_state->arcs_list[buftype], hdr);

	if (GHOST_STATE(new_state)) {
	ASSERT0(bufcnt);
	ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
	update_new = B_TRUE;
	}
	arc_evictable_space_increment(hdr, new_state);
	}
	}

	ASSERT(!HDR_EMPTY(hdr));
	if (new_state == arc_anon && HDR_IN_HASH_TABLE(hdr))
	buf_hash_remove(hdr);

	/* adjust state sizes (ignore arc_l2c_only) */

	if (update_new && new_state != arc_l2c_only) {
	ASSERT(HDR_HAS_L1HDR(hdr));
	if (GHOST_STATE(new_state)) {
	ASSERT0(bufcnt);

	/*
	* When moving a header to a ghost state, we first
	* remove all arc buffers. Thus, we'll have a
	* bufcnt of zero, and no arc buffer to use for
	* the reference. As a result, we use the arc
	* header pointer for the reference.
	*/
	(void) zfs_refcount_add_many(&new_state->arcs_size,
	HDR_GET_LSIZE(hdr), hdr);
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	ASSERT(!HDR_HAS_RABD(hdr));
	} else {
	uint32_t buffers = 0;

	/*
	* Each individual buffer holds a unique reference,
	* thus we must remove each of these references one
	* at a time.
	*/
	for (arc_buf_t *buf = hdr->b_l1hdr.b_buf; buf != NULL;
	buf = buf->b_next) {
	ASSERT3U(bufcnt, !=, 0);
	buffers++;

	/*
	* When the arc_buf_t is sharing the data
	* block with the hdr, the owner of the
	* reference belongs to the hdr. Only
	* add to the refcount if the arc_buf_t is
	* not shared.
	*/
	if (arc_buf_is_shared(buf))
	continue;

	(void) zfs_refcount_add_many(
	&new_state->arcs_size,
	arc_buf_size(buf), buf);
	}
	ASSERT3U(bufcnt, ==, buffers);

	if (hdr->b_l1hdr.b_pabd != NULL) {
	(void) zfs_refcount_add_many(
	&new_state->arcs_size,
	arc_hdr_size(hdr), hdr);
	}

	if (HDR_HAS_RABD(hdr)) {
	(void) zfs_refcount_add_many(
	&new_state->arcs_size,
	HDR_GET_PSIZE(hdr), hdr);
	}
	}
	}

	if (update_old && old_state != arc_l2c_only) {
	ASSERT(HDR_HAS_L1HDR(hdr));
	if (GHOST_STATE(old_state)) {
	ASSERT0(bufcnt);
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	ASSERT(!HDR_HAS_RABD(hdr));

	/*
	* When moving a header off of a ghost state,
	* the header will not contain any arc buffers.
	* We use the arc header pointer for the reference
	* which is exactly what we did when we put the
	* header on the ghost state.
	*/

	(void) zfs_refcount_remove_many(&old_state->arcs_size,
	HDR_GET_LSIZE(hdr), hdr);
	} else {
	uint32_t buffers = 0;

	/*
	* Each individual buffer holds a unique reference,
	* thus we must remove each of these references one
	* at a time.
	*/
	for (arc_buf_t *buf = hdr->b_l1hdr.b_buf; buf != NULL;
	buf = buf->b_next) {
	ASSERT3U(bufcnt, !=, 0);
	buffers++;

	/*
	* When the arc_buf_t is sharing the data
	* block with the hdr, the owner of the
	* reference belongs to the hdr. Only
	* add to the refcount if the arc_buf_t is
	* not shared.
	*/
	if (arc_buf_is_shared(buf))
	continue;

	(void) zfs_refcount_remove_many(
	&old_state->arcs_size, arc_buf_size(buf),
	buf);
	}
	ASSERT3U(bufcnt, ==, buffers);
	ASSERT(hdr->b_l1hdr.b_pabd != NULL \|\|
	HDR_HAS_RABD(hdr));

	if (hdr->b_l1hdr.b_pabd != NULL) {
	(void) zfs_refcount_remove_many(
	&old_state->arcs_size, arc_hdr_size(hdr),
	hdr);
	}

	if (HDR_HAS_RABD(hdr)) {
	(void) zfs_refcount_remove_many(
	&old_state->arcs_size, HDR_GET_PSIZE(hdr),
	hdr);
	}
	}
	}

	if (HDR_HAS_L1HDR(hdr)) {
	hdr->b_l1hdr.b_state = new_state;

	if (HDR_HAS_L2HDR(hdr) && new_state != arc_l2c_only) {
	l2arc_hdr_arcstats_decrement_state(hdr);
	hdr->b_l2hdr.b_arcs_state = new_state->arcs_state;
	l2arc_hdr_arcstats_increment_state(hdr);
	}
	}
	}

	void
	arc_space_consume(uint64_t space, arc_space_type_t type)
	{
	ASSERT(type >= 0 && type < ARC_SPACE_NUMTYPES);

	switch (type) {
	default:
	break;
	case ARC_SPACE_DATA:
	ARCSTAT_INCR(arcstat_data_size, space);
	break;
	case ARC_SPACE_META:
	ARCSTAT_INCR(arcstat_metadata_size, space);
	break;
	case ARC_SPACE_BONUS:
	ARCSTAT_INCR(arcstat_bonus_size, space);
	break;
	case ARC_SPACE_DNODE:
	aggsum_add(&arc_sums.arcstat_dnode_size, space);
	break;
	case ARC_SPACE_DBUF:
	ARCSTAT_INCR(arcstat_dbuf_size, space);
	break;
	case ARC_SPACE_HDRS:
	ARCSTAT_INCR(arcstat_hdr_size, space);
	break;
	case ARC_SPACE_L2HDRS:
	aggsum_add(&arc_sums.arcstat_l2_hdr_size, space);
	break;
	case ARC_SPACE_ABD_CHUNK_WASTE:
	/*
	* Note: this includes space wasted by all scatter ABD's, not
	* just those allocated by the ARC. But the vast majority of
	* scatter ABD's come from the ARC, because other users are
	* very short-lived.
	*/
	ARCSTAT_INCR(arcstat_abd_chunk_waste_size, space);
	break;
	}

	if (type != ARC_SPACE_DATA && type != ARC_SPACE_ABD_CHUNK_WASTE)
	aggsum_add(&arc_sums.arcstat_meta_used, space);

	aggsum_add(&arc_sums.arcstat_size, space);
	}

	void
	arc_space_return(uint64_t space, arc_space_type_t type)
	{
	ASSERT(type >= 0 && type < ARC_SPACE_NUMTYPES);

	switch (type) {
	default:
	break;
	case ARC_SPACE_DATA:
	ARCSTAT_INCR(arcstat_data_size, -space);
	break;
	case ARC_SPACE_META:
	ARCSTAT_INCR(arcstat_metadata_size, -space);
	break;
	case ARC_SPACE_BONUS:
	ARCSTAT_INCR(arcstat_bonus_size, -space);
	break;
	case ARC_SPACE_DNODE:
	aggsum_add(&arc_sums.arcstat_dnode_size, -space);
	break;
	case ARC_SPACE_DBUF:
	ARCSTAT_INCR(arcstat_dbuf_size, -space);
	break;
	case ARC_SPACE_HDRS:
	ARCSTAT_INCR(arcstat_hdr_size, -space);
	break;
	case ARC_SPACE_L2HDRS:
	aggsum_add(&arc_sums.arcstat_l2_hdr_size, -space);
	break;
	case ARC_SPACE_ABD_CHUNK_WASTE:
	ARCSTAT_INCR(arcstat_abd_chunk_waste_size, -space);
	break;
	}

	if (type != ARC_SPACE_DATA && type != ARC_SPACE_ABD_CHUNK_WASTE) {
	ASSERT(aggsum_compare(&arc_sums.arcstat_meta_used,
	space) >= 0);
	ARCSTAT_MAX(arcstat_meta_max,
	aggsum_upper_bound(&arc_sums.arcstat_meta_used));
	aggsum_add(&arc_sums.arcstat_meta_used, -space);
	}

	ASSERT(aggsum_compare(&arc_sums.arcstat_size, space) >= 0);
	aggsum_add(&arc_sums.arcstat_size, -space);
	}

	/*
	* Given a hdr and a buf, returns whether that buf can share its b_data buffer
	* with the hdr's b_pabd.
	*/
	static boolean_t
	arc_can_share(arc_buf_hdr_t hdr, arc_buf_t buf)
	{
	/*
	* The criteria for sharing a hdr's data are:
	* 1. the buffer is not encrypted
	* 2. the hdr's compression matches the buf's compression
	* 3. the hdr doesn't need to be byteswapped
	* 4. the hdr isn't already being shared
	* 5. the buf is either compressed or it is the last buf in the hdr list
	*
	* Criterion #5 maintains the invariant that shared uncompressed
	* bufs must be the final buf in the hdr's b_buf list. Reading this, you
	* might ask, "if a compressed buf is allocated first, won't that be the
	* last thing in the list?", but in that case it's impossible to create
	* a shared uncompressed buf anyway (because the hdr must be compressed
	* to have the compressed buf). You might also think that #3 is
	* sufficient to make this guarantee, however it's possible
	* (specifically in the rare L2ARC write race mentioned in
	* arc_buf_alloc_impl()) there will be an existing uncompressed buf that
	* is shareable, but wasn't at the time of its allocation. Rather than
	* allow a new shared uncompressed buf to be created and then shuffle
	* the list around to make it the last element, this simply disallows
	* sharing if the new buf isn't the first to be added.
	*/
	ASSERT3P(buf->b_hdr, ==, hdr);
	boolean_t hdr_compressed =
	arc_hdr_get_compress(hdr) != ZIO_COMPRESS_OFF;
	boolean_t buf_compressed = ARC_BUF_COMPRESSED(buf) != 0;
	return (!ARC_BUF_ENCRYPTED(buf) &&
	buf_compressed == hdr_compressed &&
	hdr->b_l1hdr.b_byteswap == DMU_BSWAP_NUMFUNCS &&
	!HDR_SHARED_DATA(hdr) &&
	(ARC_BUF_LAST(buf) \|\| ARC_BUF_COMPRESSED(buf)));
	}

	/*
	* Allocate a buf for this hdr. If you care about the data that's in the hdr,
	* or if you want a compressed buffer, pass those flags in. Returns 0 if the
	* copy was made successfully, or an error code otherwise.
	*/
	static int
	arc_buf_alloc_impl(arc_buf_hdr_t hdr, spa_t spa, const zbookmark_phys_t *zb,
	void *tag, boolean_t encrypted, boolean_t compressed, boolean_t noauth,
	boolean_t fill, arc_buf_t **ret)
	{
	arc_buf_t *buf;
	arc_fill_flags_t flags = ARC_FILL_LOCKED;

	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT3U(HDR_GET_LSIZE(hdr), >, 0);
	VERIFY(hdr->b_type == ARC_BUFC_DATA \|\|
	hdr->b_type == ARC_BUFC_METADATA);
	ASSERT3P(ret, !=, NULL);
	ASSERT3P(*ret, ==, NULL);
	IMPLY(encrypted, compressed);

	buf = *ret = kmem_cache_alloc(buf_cache, KM_PUSHPAGE);
	buf->b_hdr = hdr;
	buf->b_data = NULL;
	buf->b_next = hdr->b_l1hdr.b_buf;
	buf->b_flags = 0;

	add_reference(hdr, tag);

	/*
	* We're about to change the hdr's b_flags. We must either
	* hold the hash_lock or be undiscoverable.
	*/
	ASSERT(HDR_EMPTY_OR_LOCKED(hdr));

	/*
	* Only honor requests for compressed bufs if the hdr is actually
	* compressed. This must be overridden if the buffer is encrypted since
	* encrypted buffers cannot be decompressed.
	*/
	if (encrypted) {
	buf->b_flags \|= ARC_BUF_FLAG_COMPRESSED;
	buf->b_flags \|= ARC_BUF_FLAG_ENCRYPTED;
	flags \|= ARC_FILL_COMPRESSED \| ARC_FILL_ENCRYPTED;
	} else if (compressed &&
	arc_hdr_get_compress(hdr) != ZIO_COMPRESS_OFF) {
	buf->b_flags \|= ARC_BUF_FLAG_COMPRESSED;
	flags \|= ARC_FILL_COMPRESSED;
	}

	if (noauth) {
	ASSERT0(encrypted);
	flags \|= ARC_FILL_NOAUTH;
	}

	/*
	* If the hdr's data can be shared then we share the data buffer and
	* set the appropriate bit in the hdr's b_flags to indicate the hdr is
	* sharing it's b_pabd with the arc_buf_t. Otherwise, we allocate a new
	* buffer to store the buf's data.
	*
	* There are two additional restrictions here because we're sharing
	* hdr -> buf instead of the usual buf -> hdr. First, the hdr can't be
	* actively involved in an L2ARC write, because if this buf is used by
	* an arc_write() then the hdr's data buffer will be released when the
	* write completes, even though the L2ARC write might still be using it.
	* Second, the hdr's ABD must be linear so that the buf's user doesn't
	* need to be ABD-aware. It must be allocated via
	* zio_[data_]buf_alloc(), not as a page, because we need to be able
	* to abd_release_ownership_of_buf(), which isn't allowed on "linear
	* page" buffers because the ABD code needs to handle freeing them
	* specially.
	*/
	boolean_t can_share = arc_can_share(hdr, buf) &&
	!HDR_L2_WRITING(hdr) &&
	hdr->b_l1hdr.b_pabd != NULL &&
	abd_is_linear(hdr->b_l1hdr.b_pabd) &&
	!abd_is_linear_page(hdr->b_l1hdr.b_pabd);

	/* Set up b_data and sharing */
	if (can_share) {
	buf->b_data = abd_to_buf(hdr->b_l1hdr.b_pabd);
	buf->b_flags \|= ARC_BUF_FLAG_SHARED;
	arc_hdr_set_flags(hdr, ARC_FLAG_SHARED_DATA);
	} else {
	buf->b_data =
	arc_get_data_buf(hdr, arc_buf_size(buf), buf);
	ARCSTAT_INCR(arcstat_overhead_size, arc_buf_size(buf));
	}
	VERIFY3P(buf->b_data, !=, NULL);

	hdr->b_l1hdr.b_buf = buf;
	hdr->b_l1hdr.b_bufcnt += 1;
	if (encrypted)
	hdr->b_crypt_hdr.b_ebufcnt += 1;

	/*
	* If the user wants the data from the hdr, we need to either copy or
	* decompress the data.
	*/
	if (fill) {
	ASSERT3P(zb, !=, NULL);
	return (arc_buf_fill(buf, spa, zb, flags));
	}

	return (0);
	}

	static char *arc_onloan_tag = "onloan";

	static inline void
	arc_loaned_bytes_update(int64_t delta)
	{
	atomic_add_64(&arc_loaned_bytes, delta);

	/* assert that it did not wrap around */
	ASSERT3S(atomic_add_64_nv(&arc_loaned_bytes, 0), >=, 0);
	}

	/*
	* Loan out an anonymous arc buffer. Loaned buffers are not counted as in
	* flight data by arc_tempreserve_space() until they are "returned". Loaned
	* buffers must be returned to the arc before they can be used by the DMU or
	* freed.
	*/
	arc_buf_t *
	arc_loan_buf(spa_t *spa, boolean_t is_metadata, int size)
	{
	arc_buf_t *buf = arc_alloc_buf(spa, arc_onloan_tag,
	is_metadata ? ARC_BUFC_METADATA : ARC_BUFC_DATA, size);

	arc_loaned_bytes_update(arc_buf_size(buf));

	return (buf);
	}

	arc_buf_t *
	arc_loan_compressed_buf(spa_t *spa, uint64_t psize, uint64_t lsize,
	enum zio_compress compression_type, uint8_t complevel)
	{
	arc_buf_t *buf = arc_alloc_compressed_buf(spa, arc_onloan_tag,
	psize, lsize, compression_type, complevel);

	arc_loaned_bytes_update(arc_buf_size(buf));

	return (buf);
	}

	arc_buf_t *
	arc_loan_raw_buf(spa_t *spa, uint64_t dsobj, boolean_t byteorder,
	const uint8_t salt, const uint8_t iv, const uint8_t *mac,
	dmu_object_type_t ot, uint64_t psize, uint64_t lsize,
	enum zio_compress compression_type, uint8_t complevel)
	{
	arc_buf_t *buf = arc_alloc_raw_buf(spa, arc_onloan_tag, dsobj,
	byteorder, salt, iv, mac, ot, psize, lsize, compression_type,
	complevel);

	atomic_add_64(&arc_loaned_bytes, psize);
	return (buf);
	}


	/*
	* Return a loaned arc buffer to the arc.
	*/
	void
	arc_return_buf(arc_buf_t buf, void tag)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	ASSERT3P(buf->b_data, !=, NULL);
	ASSERT(HDR_HAS_L1HDR(hdr));
	(void) zfs_refcount_add(&hdr->b_l1hdr.b_refcnt, tag);
	(void) zfs_refcount_remove(&hdr->b_l1hdr.b_refcnt, arc_onloan_tag);

	arc_loaned_bytes_update(-arc_buf_size(buf));
	}

	/* Detach an arc_buf from a dbuf (tag) */
	void
	arc_loan_inuse_buf(arc_buf_t buf, void tag)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	ASSERT3P(buf->b_data, !=, NULL);
	ASSERT(HDR_HAS_L1HDR(hdr));
	(void) zfs_refcount_add(&hdr->b_l1hdr.b_refcnt, arc_onloan_tag);
	(void) zfs_refcount_remove(&hdr->b_l1hdr.b_refcnt, tag);

	arc_loaned_bytes_update(arc_buf_size(buf));
	}

	static void
	l2arc_free_abd_on_write(abd_t *abd, size_t size, arc_buf_contents_t type)
	{
	l2arc_data_free_t df = kmem_alloc(sizeof (df), KM_SLEEP);

	df->l2df_abd = abd;
	df->l2df_size = size;
	df->l2df_type = type;
	mutex_enter(&l2arc_free_on_write_mtx);
	list_insert_head(l2arc_free_on_write, df);
	mutex_exit(&l2arc_free_on_write_mtx);
	}

	static void
	arc_hdr_free_on_write(arc_buf_hdr_t *hdr, boolean_t free_rdata)
	{
	arc_state_t *state = hdr->b_l1hdr.b_state;
	arc_buf_contents_t type = arc_buf_type(hdr);
	uint64_t size = (free_rdata) ? HDR_GET_PSIZE(hdr) : arc_hdr_size(hdr);

	/* protected by hash lock, if in the hash table */
	if (multilist_link_active(&hdr->b_l1hdr.b_arc_node)) {
	ASSERT(zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
	ASSERT(state != arc_anon && state != arc_l2c_only);

	(void) zfs_refcount_remove_many(&state->arcs_esize[type],
	size, hdr);
	}
	(void) zfs_refcount_remove_many(&state->arcs_size, size, hdr);
	if (type == ARC_BUFC_METADATA) {
	arc_space_return(size, ARC_SPACE_META);
	} else {
	ASSERT(type == ARC_BUFC_DATA);
	arc_space_return(size, ARC_SPACE_DATA);
	}

	if (free_rdata) {
	l2arc_free_abd_on_write(hdr->b_crypt_hdr.b_rabd, size, type);
	} else {
	l2arc_free_abd_on_write(hdr->b_l1hdr.b_pabd, size, type);
	}
	}

	/*
	* Share the arc_buf_t's data with the hdr. Whenever we are sharing the
	* data buffer, we transfer the refcount ownership to the hdr and update
	* the appropriate kstats.
	*/
	static void
	arc_share_buf(arc_buf_hdr_t hdr, arc_buf_t buf)
	{
	ASSERT(arc_can_share(hdr, buf));
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	ASSERT(!ARC_BUF_ENCRYPTED(buf));
	ASSERT(HDR_EMPTY_OR_LOCKED(hdr));

	/*
	* Start sharing the data buffer. We transfer the
	* refcount ownership to the hdr since it always owns
	* the refcount whenever an arc_buf_t is shared.
	*/
	zfs_refcount_transfer_ownership_many(&hdr->b_l1hdr.b_state->arcs_size,
	arc_hdr_size(hdr), buf, hdr);
	hdr->b_l1hdr.b_pabd = abd_get_from_buf(buf->b_data, arc_buf_size(buf));
	abd_take_ownership_of_buf(hdr->b_l1hdr.b_pabd,
	HDR_ISTYPE_METADATA(hdr));
	arc_hdr_set_flags(hdr, ARC_FLAG_SHARED_DATA);
	buf->b_flags \|= ARC_BUF_FLAG_SHARED;

	/*
	* Since we've transferred ownership to the hdr we need
	* to increment its compressed and uncompressed kstats and
	* decrement the overhead size.
	*/
	ARCSTAT_INCR(arcstat_compressed_size, arc_hdr_size(hdr));
	ARCSTAT_INCR(arcstat_uncompressed_size, HDR_GET_LSIZE(hdr));
	ARCSTAT_INCR(arcstat_overhead_size, -arc_buf_size(buf));
	}

	static void
	arc_unshare_buf(arc_buf_hdr_t hdr, arc_buf_t buf)
	{
	ASSERT(arc_buf_is_shared(buf));
	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
	ASSERT(HDR_EMPTY_OR_LOCKED(hdr));

	/*
	* We are no longer sharing this buffer so we need
	* to transfer its ownership to the rightful owner.
	*/
	zfs_refcount_transfer_ownership_many(&hdr->b_l1hdr.b_state->arcs_size,
	arc_hdr_size(hdr), hdr, buf);
	arc_hdr_clear_flags(hdr, ARC_FLAG_SHARED_DATA);
	abd_release_ownership_of_buf(hdr->b_l1hdr.b_pabd);
	abd_free(hdr->b_l1hdr.b_pabd);
	hdr->b_l1hdr.b_pabd = NULL;
	buf->b_flags &= ~ARC_BUF_FLAG_SHARED;

	/*
	* Since the buffer is no longer shared between
	* the arc buf and the hdr, count it as overhead.
	*/
	ARCSTAT_INCR(arcstat_compressed_size, -arc_hdr_size(hdr));
	ARCSTAT_INCR(arcstat_uncompressed_size, -HDR_GET_LSIZE(hdr));
	ARCSTAT_INCR(arcstat_overhead_size, arc_buf_size(buf));
	}

	/*
	* Remove an arc_buf_t from the hdr's buf list and return the last
	* arc_buf_t on the list. If no buffers remain on the list then return
	* NULL.
	*/
	static arc_buf_t *
	arc_buf_remove(arc_buf_hdr_t hdr, arc_buf_t buf)
	{
	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT(HDR_EMPTY_OR_LOCKED(hdr));

	arc_buf_t **bufp = &hdr->b_l1hdr.b_buf;
	arc_buf_t *lastbuf = NULL;

	/*
	* Remove the buf from the hdr list and locate the last
	* remaining buffer on the list.
	*/
	while (*bufp != NULL) {
	if (*bufp == buf)
	*bufp = buf->b_next;

	/*
	* If we've removed a buffer in the middle of
	* the list then update the lastbuf and update
	* bufp.
	*/
	if (*bufp != NULL) {
	lastbuf = *bufp;
	bufp = &(*bufp)->b_next;
	}
	}
	buf->b_next = NULL;
	ASSERT3P(lastbuf, !=, buf);
	IMPLY(hdr->b_l1hdr.b_bufcnt > 0, lastbuf != NULL);
	IMPLY(hdr->b_l1hdr.b_bufcnt > 0, hdr->b_l1hdr.b_buf != NULL);
	IMPLY(lastbuf != NULL, ARC_BUF_LAST(lastbuf));

	return (lastbuf);
	}

	/*
	* Free up buf->b_data and pull the arc_buf_t off of the arc_buf_hdr_t's
	* list and free it.
	*/
	static void
	arc_buf_destroy_impl(arc_buf_t *buf)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	/*
	* Free up the data associated with the buf but only if we're not
	* sharing this with the hdr. If we are sharing it with the hdr, the
	* hdr is responsible for doing the free.
	*/
	if (buf->b_data != NULL) {
	/*
	* We're about to change the hdr's b_flags. We must either
	* hold the hash_lock or be undiscoverable.
	*/
	ASSERT(HDR_EMPTY_OR_LOCKED(hdr));

	arc_cksum_verify(buf);
	arc_buf_unwatch(buf);

	if (arc_buf_is_shared(buf)) {
	arc_hdr_clear_flags(hdr, ARC_FLAG_SHARED_DATA);
	} else {
	uint64_t size = arc_buf_size(buf);
	arc_free_data_buf(hdr, buf->b_data, size, buf);
	ARCSTAT_INCR(arcstat_overhead_size, -size);
	}
	buf->b_data = NULL;

	ASSERT(hdr->b_l1hdr.b_bufcnt > 0);
	hdr->b_l1hdr.b_bufcnt -= 1;

	if (ARC_BUF_ENCRYPTED(buf)) {
	hdr->b_crypt_hdr.b_ebufcnt -= 1;

	/*
	* If we have no more encrypted buffers and we've
	* already gotten a copy of the decrypted data we can
	* free b_rabd to save some space.
	*/
	if (hdr->b_crypt_hdr.b_ebufcnt == 0 &&
	HDR_HAS_RABD(hdr) && hdr->b_l1hdr.b_pabd != NULL &&
	!HDR_IO_IN_PROGRESS(hdr)) {
	arc_hdr_free_abd(hdr, B_TRUE);
	}
	}
	}

	arc_buf_t *lastbuf = arc_buf_remove(hdr, buf);

	if (ARC_BUF_SHARED(buf) && !ARC_BUF_COMPRESSED(buf)) {
	/*
	* If the current arc_buf_t is sharing its data buffer with the
	* hdr, then reassign the hdr's b_pabd to share it with the new
	* buffer at the end of the list. The shared buffer is always
	* the last one on the hdr's buffer list.
	*
	* There is an equivalent case for compressed bufs, but since
	* they aren't guaranteed to be the last buf in the list and
	* that is an exceedingly rare case, we just allow that space be
	* wasted temporarily. We must also be careful not to share
	* encrypted buffers, since they cannot be shared.
	*/
	if (lastbuf != NULL && !ARC_BUF_ENCRYPTED(lastbuf)) {
	/* Only one buf can be shared at once */
	VERIFY(!arc_buf_is_shared(lastbuf));
	/* hdr is uncompressed so can't have compressed buf */
	VERIFY(!ARC_BUF_COMPRESSED(lastbuf));

	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
	arc_hdr_free_abd(hdr, B_FALSE);

	/*
	* We must setup a new shared block between the
	* last buffer and the hdr. The data would have
	* been allocated by the arc buf so we need to transfer
	* ownership to the hdr since it's now being shared.
	*/
	arc_share_buf(hdr, lastbuf);
	}
	} else if (HDR_SHARED_DATA(hdr)) {
	/*
	* Uncompressed shared buffers are always at the end
	* of the list. Compressed buffers don't have the
	* same requirements. This makes it hard to
	* simply assert that the lastbuf is shared so
	* we rely on the hdr's compression flags to determine
	* if we have a compressed, shared buffer.
	*/
	ASSERT3P(lastbuf, !=, NULL);
	ASSERT(arc_buf_is_shared(lastbuf) \|\|
	arc_hdr_get_compress(hdr) != ZIO_COMPRESS_OFF);
	}

	/*
	* Free the checksum if we're removing the last uncompressed buf from
	* this hdr.
	*/
	if (!arc_hdr_has_uncompressed_buf(hdr)) {
	arc_cksum_free(hdr);
	}

	/* clean up the buf */
	buf->b_hdr = NULL;
	kmem_cache_free(buf_cache, buf);
	}

	static void
	arc_hdr_alloc_abd(arc_buf_hdr_t *hdr, int alloc_flags)
	{
	uint64_t size;
	boolean_t alloc_rdata = ((alloc_flags & ARC_HDR_ALLOC_RDATA) != 0);

	ASSERT3U(HDR_GET_LSIZE(hdr), >, 0);
	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT(!HDR_SHARED_DATA(hdr) \|\| alloc_rdata);
	IMPLY(alloc_rdata, HDR_PROTECTED(hdr));

	if (alloc_rdata) {
	size = HDR_GET_PSIZE(hdr);
	ASSERT3P(hdr->b_crypt_hdr.b_rabd, ==, NULL);
	hdr->b_crypt_hdr.b_rabd = arc_get_data_abd(hdr, size, hdr,
	alloc_flags);
	ASSERT3P(hdr->b_crypt_hdr.b_rabd, !=, NULL);
	ARCSTAT_INCR(arcstat_raw_size, size);
	} else {
	size = arc_hdr_size(hdr);
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	hdr->b_l1hdr.b_pabd = arc_get_data_abd(hdr, size, hdr,
	alloc_flags);
	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
	}

	ARCSTAT_INCR(arcstat_compressed_size, size);
	ARCSTAT_INCR(arcstat_uncompressed_size, HDR_GET_LSIZE(hdr));
	}

	static void
	arc_hdr_free_abd(arc_buf_hdr_t *hdr, boolean_t free_rdata)
	{
	uint64_t size = (free_rdata) ? HDR_GET_PSIZE(hdr) : arc_hdr_size(hdr);

	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT(hdr->b_l1hdr.b_pabd != NULL \|\| HDR_HAS_RABD(hdr));
	IMPLY(free_rdata, HDR_HAS_RABD(hdr));

	/*
	* If the hdr is currently being written to the l2arc then
	* we defer freeing the data by adding it to the l2arc_free_on_write
	* list. The l2arc will free the data once it's finished
	* writing it to the l2arc device.
	*/
	if (HDR_L2_WRITING(hdr)) {
	arc_hdr_free_on_write(hdr, free_rdata);
	ARCSTAT_BUMP(arcstat_l2_free_on_write);
	} else if (free_rdata) {
	arc_free_data_abd(hdr, hdr->b_crypt_hdr.b_rabd, size, hdr);
	} else {
	arc_free_data_abd(hdr, hdr->b_l1hdr.b_pabd, size, hdr);
	}

	if (free_rdata) {
	hdr->b_crypt_hdr.b_rabd = NULL;
	ARCSTAT_INCR(arcstat_raw_size, -size);
	} else {
	hdr->b_l1hdr.b_pabd = NULL;
	}

	if (hdr->b_l1hdr.b_pabd == NULL && !HDR_HAS_RABD(hdr))
	hdr->b_l1hdr.b_byteswap = DMU_BSWAP_NUMFUNCS;

	ARCSTAT_INCR(arcstat_compressed_size, -size);
	ARCSTAT_INCR(arcstat_uncompressed_size, -HDR_GET_LSIZE(hdr));
	}

	/*
	* Allocate empty anonymous ARC header. The header will get its identity
	* assigned and buffers attached later as part of read or write operations.
	*
	* In case of read arc_read() assigns header its identify (b_dva + b_birth),
	* inserts it into ARC hash to become globally visible and allocates physical
	* (b_pabd) or raw (b_rabd) ABD buffer to read into from disk. On disk read
	* completion arc_read_done() allocates ARC buffer(s) as needed, potentially
	* sharing one of them with the physical ABD buffer.
	*
	* In case of write arc_alloc_buf() allocates ARC buffer to be filled with
	* data. Then after compression and/or encryption arc_write_ready() allocates
	* and fills (or potentially shares) physical (b_pabd) or raw (b_rabd) ABD
	* buffer. On disk write completion arc_write_done() assigns the header its
	* new identity (b_dva + b_birth) and inserts into ARC hash.
	*
	* In case of partial overwrite the old data is read first as described. Then
	* arc_release() either allocates new anonymous ARC header and moves the ARC
	* buffer to it, or reuses the old ARC header by discarding its identity and
	* removing it from ARC hash. After buffer modification normal write process
	* follows as described.
	*/
	static arc_buf_hdr_t *
	arc_hdr_alloc(uint64_t spa, int32_t psize, int32_t lsize,
	boolean_t protected, enum zio_compress compression_type, uint8_t complevel,
	arc_buf_contents_t type)
	{
	arc_buf_hdr_t *hdr;

	VERIFY(type == ARC_BUFC_DATA \|\| type == ARC_BUFC_METADATA);
	if (protected) {
	hdr = kmem_cache_alloc(hdr_full_crypt_cache, KM_PUSHPAGE);
	} else {
	hdr = kmem_cache_alloc(hdr_full_cache, KM_PUSHPAGE);
	}

	ASSERT(HDR_EMPTY(hdr));
	ASSERT3P(hdr->b_l1hdr.b_freeze_cksum, ==, NULL);
	HDR_SET_PSIZE(hdr, psize);
	HDR_SET_LSIZE(hdr, lsize);
	hdr->b_spa = spa;
	hdr->b_type = type;
	hdr->b_flags = 0;
	arc_hdr_set_flags(hdr, arc_bufc_to_flags(type) \| ARC_FLAG_HAS_L1HDR);
	arc_hdr_set_compress(hdr, compression_type);
	hdr->b_complevel = complevel;
	if (protected)
	arc_hdr_set_flags(hdr, ARC_FLAG_PROTECTED);

	hdr->b_l1hdr.b_state = arc_anon;
	hdr->b_l1hdr.b_arc_access = 0;
	hdr->b_l1hdr.b_mru_hits = 0;
	hdr->b_l1hdr.b_mru_ghost_hits = 0;
	hdr->b_l1hdr.b_mfu_hits = 0;
	hdr->b_l1hdr.b_mfu_ghost_hits = 0;
	hdr->b_l1hdr.b_bufcnt = 0;
	hdr->b_l1hdr.b_buf = NULL;

	ASSERT(zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt));

	return (hdr);
	}

	/*
	* Transition between the two allocation states for the arc_buf_hdr struct.
	* The arc_buf_hdr struct can be allocated with (hdr_full_cache) or without
	* (hdr_l2only_cache) the fields necessary for the L1 cache - the smaller
	* version is used when a cache buffer is only in the L2ARC in order to reduce
	* memory usage.
	*/
	static arc_buf_hdr_t *
	arc_hdr_realloc(arc_buf_hdr_t hdr, kmem_cache_t old, kmem_cache_t *new)
	{
	ASSERT(HDR_HAS_L2HDR(hdr));

	arc_buf_hdr_t *nhdr;
	l2arc_dev_t *dev = hdr->b_l2hdr.b_dev;

	ASSERT((old == hdr_full_cache && new == hdr_l2only_cache) \|\|
	(old == hdr_l2only_cache && new == hdr_full_cache));

	/*
	* if the caller wanted a new full header and the header is to be
	* encrypted we will actually allocate the header from the full crypt
	* cache instead. The same applies to freeing from the old cache.
	*/
	if (HDR_PROTECTED(hdr) && new == hdr_full_cache)
	new = hdr_full_crypt_cache;
	if (HDR_PROTECTED(hdr) && old == hdr_full_cache)
	old = hdr_full_crypt_cache;

	nhdr = kmem_cache_alloc(new, KM_PUSHPAGE);

	ASSERT(MUTEX_HELD(HDR_LOCK(hdr)));
	buf_hash_remove(hdr);

	bcopy(hdr, nhdr, HDR_L2ONLY_SIZE);

	if (new == hdr_full_cache \|\| new == hdr_full_crypt_cache) {
	arc_hdr_set_flags(nhdr, ARC_FLAG_HAS_L1HDR);
	/*
	* arc_access and arc_change_state need to be aware that a
	* header has just come out of L2ARC, so we set its state to
	* l2c_only even though it's about to change.
	*/
	nhdr->b_l1hdr.b_state = arc_l2c_only;

	/* Verify previous threads set to NULL before freeing */
	ASSERT3P(nhdr->b_l1hdr.b_pabd, ==, NULL);
	ASSERT(!HDR_HAS_RABD(hdr));
	} else {
	ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
	ASSERT0(hdr->b_l1hdr.b_bufcnt);
	ASSERT3P(hdr->b_l1hdr.b_freeze_cksum, ==, NULL);

	/*
	* If we've reached here, We must have been called from
	* arc_evict_hdr(), as such we should have already been
	* removed from any ghost list we were previously on
	* (which protects us from racing with arc_evict_state),
	* thus no locking is needed during this check.
	*/
	ASSERT(!multilist_link_active(&hdr->b_l1hdr.b_arc_node));

	/*
	* A buffer must not be moved into the arc_l2c_only
	* state if it's not finished being written out to the
	* l2arc device. Otherwise, the b_l1hdr.b_pabd field
	* might try to be accessed, even though it was removed.
	*/
	VERIFY(!HDR_L2_WRITING(hdr));
	VERIFY3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	ASSERT(!HDR_HAS_RABD(hdr));

	arc_hdr_clear_flags(nhdr, ARC_FLAG_HAS_L1HDR);
	}
	/*
	* The header has been reallocated so we need to re-insert it into any
	* lists it was on.
	*/
	(void) buf_hash_insert(nhdr, NULL);

	ASSERT(list_link_active(&hdr->b_l2hdr.b_l2node));

	mutex_enter(&dev->l2ad_mtx);

	/*
	* We must place the realloc'ed header back into the list at
	* the same spot. Otherwise, if it's placed earlier in the list,
	* l2arc_write_buffers() could find it during the function's
	* write phase, and try to write it out to the l2arc.
	*/
	list_insert_after(&dev->l2ad_buflist, hdr, nhdr);
	list_remove(&dev->l2ad_buflist, hdr);

	mutex_exit(&dev->l2ad_mtx);

	/*
	* Since we're using the pointer address as the tag when
	* incrementing and decrementing the l2ad_alloc refcount, we
	* must remove the old pointer (that we're about to destroy) and
	* add the new pointer to the refcount. Otherwise we'd remove
	* the wrong pointer address when calling arc_hdr_destroy() later.
	*/

	(void) zfs_refcount_remove_many(&dev->l2ad_alloc,
	arc_hdr_size(hdr), hdr);
	(void) zfs_refcount_add_many(&dev->l2ad_alloc,
	arc_hdr_size(nhdr), nhdr);

	buf_discard_identity(hdr);
	kmem_cache_free(old, hdr);

	return (nhdr);
	}

	/*
	* This function allows an L1 header to be reallocated as a crypt
	* header and vice versa. If we are going to a crypt header, the
	* new fields will be zeroed out.
	*/
	static arc_buf_hdr_t *
	arc_hdr_realloc_crypt(arc_buf_hdr_t *hdr, boolean_t need_crypt)
	{
	arc_buf_hdr_t *nhdr;
	arc_buf_t *buf;
	kmem_cache_t ncache, ocache;

	/*
	* This function requires that hdr is in the arc_anon state.
	* Therefore it won't have any L2ARC data for us to worry
	* about copying.
	*/
	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT(!HDR_HAS_L2HDR(hdr));
	ASSERT3U(!!HDR_PROTECTED(hdr), !=, need_crypt);
	ASSERT3P(hdr->b_l1hdr.b_state, ==, arc_anon);
	ASSERT(!multilist_link_active(&hdr->b_l1hdr.b_arc_node));
	ASSERT(!list_link_active(&hdr->b_l2hdr.b_l2node));
	ASSERT3P(hdr->b_hash_next, ==, NULL);

	if (need_crypt) {
	ncache = hdr_full_crypt_cache;
	ocache = hdr_full_cache;
	} else {
	ncache = hdr_full_cache;
	ocache = hdr_full_crypt_cache;
	}

	nhdr = kmem_cache_alloc(ncache, KM_PUSHPAGE);

	/*
	* Copy all members that aren't locks or condvars to the new header.
	* No lists are pointing to us (as we asserted above), so we don't
	* need to worry about the list nodes.
	*/
	nhdr->b_dva = hdr->b_dva;
	nhdr->b_birth = hdr->b_birth;
	nhdr->b_type = hdr->b_type;
	nhdr->b_flags = hdr->b_flags;
	nhdr->b_psize = hdr->b_psize;
	nhdr->b_lsize = hdr->b_lsize;
	nhdr->b_spa = hdr->b_spa;
	nhdr->b_l1hdr.b_freeze_cksum = hdr->b_l1hdr.b_freeze_cksum;
	nhdr->b_l1hdr.b_bufcnt = hdr->b_l1hdr.b_bufcnt;
	nhdr->b_l1hdr.b_byteswap = hdr->b_l1hdr.b_byteswap;
	nhdr->b_l1hdr.b_state = hdr->b_l1hdr.b_state;
	nhdr->b_l1hdr.b_arc_access = hdr->b_l1hdr.b_arc_access;
	nhdr->b_l1hdr.b_mru_hits = hdr->b_l1hdr.b_mru_hits;
	nhdr->b_l1hdr.b_mru_ghost_hits = hdr->b_l1hdr.b_mru_ghost_hits;
	nhdr->b_l1hdr.b_mfu_hits = hdr->b_l1hdr.b_mfu_hits;
	nhdr->b_l1hdr.b_mfu_ghost_hits = hdr->b_l1hdr.b_mfu_ghost_hits;
	nhdr->b_l1hdr.b_acb = hdr->b_l1hdr.b_acb;
	nhdr->b_l1hdr.b_pabd = hdr->b_l1hdr.b_pabd;

	/*
	* This zfs_refcount_add() exists only to ensure that the individual
	* arc buffers always point to a header that is referenced, avoiding
	* a small race condition that could trigger ASSERTs.
	*/
	(void) zfs_refcount_add(&nhdr->b_l1hdr.b_refcnt, FTAG);
	nhdr->b_l1hdr.b_buf = hdr->b_l1hdr.b_buf;
	for (buf = nhdr->b_l1hdr.b_buf; buf != NULL; buf = buf->b_next) {
	mutex_enter(&buf->b_evict_lock);
	buf->b_hdr = nhdr;
	mutex_exit(&buf->b_evict_lock);
	}

	zfs_refcount_transfer(&nhdr->b_l1hdr.b_refcnt, &hdr->b_l1hdr.b_refcnt);
	(void) zfs_refcount_remove(&nhdr->b_l1hdr.b_refcnt, FTAG);
	ASSERT0(zfs_refcount_count(&hdr->b_l1hdr.b_refcnt));

	if (need_crypt) {
	arc_hdr_set_flags(nhdr, ARC_FLAG_PROTECTED);
	} else {
	arc_hdr_clear_flags(nhdr, ARC_FLAG_PROTECTED);
	}

	/* unset all members of the original hdr */
	bzero(&hdr->b_dva, sizeof (dva_t));
	hdr->b_birth = 0;
	hdr->b_type = ARC_BUFC_INVALID;
	hdr->b_flags = 0;
	hdr->b_psize = 0;
	hdr->b_lsize = 0;
	hdr->b_spa = 0;
	hdr->b_l1hdr.b_freeze_cksum = NULL;
	hdr->b_l1hdr.b_buf = NULL;
	hdr->b_l1hdr.b_bufcnt = 0;
	hdr->b_l1hdr.b_byteswap = 0;
	hdr->b_l1hdr.b_state = NULL;
	hdr->b_l1hdr.b_arc_access = 0;
	hdr->b_l1hdr.b_mru_hits = 0;
	hdr->b_l1hdr.b_mru_ghost_hits = 0;
	hdr->b_l1hdr.b_mfu_hits = 0;
	hdr->b_l1hdr.b_mfu_ghost_hits = 0;
	hdr->b_l1hdr.b_acb = NULL;
	hdr->b_l1hdr.b_pabd = NULL;

	if (ocache == hdr_full_crypt_cache) {
	ASSERT(!HDR_HAS_RABD(hdr));
	hdr->b_crypt_hdr.b_ot = DMU_OT_NONE;
	hdr->b_crypt_hdr.b_ebufcnt = 0;
	hdr->b_crypt_hdr.b_dsobj = 0;
	bzero(hdr->b_crypt_hdr.b_salt, ZIO_DATA_SALT_LEN);
	bzero(hdr->b_crypt_hdr.b_iv, ZIO_DATA_IV_LEN);
	bzero(hdr->b_crypt_hdr.b_mac, ZIO_DATA_MAC_LEN);
	}

	buf_discard_identity(hdr);
	kmem_cache_free(ocache, hdr);

	return (nhdr);
	}

	/*
	* This function is used by the send / receive code to convert a newly
	* allocated arc_buf_t to one that is suitable for a raw encrypted write. It
	* is also used to allow the root objset block to be updated without altering
	* its embedded MACs. Both block types will always be uncompressed so we do not
	* have to worry about compression type or psize.
	*/
	void
	arc_convert_to_raw(arc_buf_t *buf, uint64_t dsobj, boolean_t byteorder,
	dmu_object_type_t ot, const uint8_t salt, const uint8_t iv,
	const uint8_t *mac)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	ASSERT(ot == DMU_OT_DNODE \|\| ot == DMU_OT_OBJSET);
	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT3P(hdr->b_l1hdr.b_state, ==, arc_anon);

	buf->b_flags \|= (ARC_BUF_FLAG_COMPRESSED \| ARC_BUF_FLAG_ENCRYPTED);
	if (!HDR_PROTECTED(hdr))
	hdr = arc_hdr_realloc_crypt(hdr, B_TRUE);
	hdr->b_crypt_hdr.b_dsobj = dsobj;
	hdr->b_crypt_hdr.b_ot = ot;
	hdr->b_l1hdr.b_byteswap = (byteorder == ZFS_HOST_BYTEORDER) ?
	DMU_BSWAP_NUMFUNCS : DMU_OT_BYTESWAP(ot);
	if (!arc_hdr_has_uncompressed_buf(hdr))
	arc_cksum_free(hdr);

	if (salt != NULL)
	bcopy(salt, hdr->b_crypt_hdr.b_salt, ZIO_DATA_SALT_LEN);
	if (iv != NULL)
	bcopy(iv, hdr->b_crypt_hdr.b_iv, ZIO_DATA_IV_LEN);
	if (mac != NULL)
	bcopy(mac, hdr->b_crypt_hdr.b_mac, ZIO_DATA_MAC_LEN);
	}

	/*
	* Allocate a new arc_buf_hdr_t and arc_buf_t and return the buf to the caller.
	* The buf is returned thawed since we expect the consumer to modify it.
	*/
	arc_buf_t *
	arc_alloc_buf(spa_t spa, void tag, arc_buf_contents_t type, int32_t size)
	{
	arc_buf_hdr_t *hdr = arc_hdr_alloc(spa_load_guid(spa), size, size,
	B_FALSE, ZIO_COMPRESS_OFF, 0, type);

	arc_buf_t *buf = NULL;
	VERIFY0(arc_buf_alloc_impl(hdr, spa, NULL, tag, B_FALSE, B_FALSE,
	B_FALSE, B_FALSE, &buf));
	arc_buf_thaw(buf);

	return (buf);
	}

	/*
	* Allocate a compressed buf in the same manner as arc_alloc_buf. Don't use this
	* for bufs containing metadata.
	*/
	arc_buf_t *
	arc_alloc_compressed_buf(spa_t spa, void tag, uint64_t psize, uint64_t lsize,
	enum zio_compress compression_type, uint8_t complevel)
	{
	ASSERT3U(lsize, >, 0);
	ASSERT3U(lsize, >=, psize);
	ASSERT3U(compression_type, >, ZIO_COMPRESS_OFF);
	ASSERT3U(compression_type, <, ZIO_COMPRESS_FUNCTIONS);

	arc_buf_hdr_t *hdr = arc_hdr_alloc(spa_load_guid(spa), psize, lsize,
	B_FALSE, compression_type, complevel, ARC_BUFC_DATA);

	arc_buf_t *buf = NULL;
	VERIFY0(arc_buf_alloc_impl(hdr, spa, NULL, tag, B_FALSE,
	B_TRUE, B_FALSE, B_FALSE, &buf));
	arc_buf_thaw(buf);
	ASSERT3P(hdr->b_l1hdr.b_freeze_cksum, ==, NULL);

	/*
	* To ensure that the hdr has the correct data in it if we call
	* arc_untransform() on this buf before it's been written to disk,
	* it's easiest if we just set up sharing between the buf and the hdr.
	*/
	arc_share_buf(hdr, buf);

	return (buf);
	}

	arc_buf_t *
	arc_alloc_raw_buf(spa_t spa, void tag, uint64_t dsobj, boolean_t byteorder,
	const uint8_t salt, const uint8_t iv, const uint8_t *mac,
	dmu_object_type_t ot, uint64_t psize, uint64_t lsize,
	enum zio_compress compression_type, uint8_t complevel)
	{
	arc_buf_hdr_t *hdr;
	arc_buf_t *buf;
	arc_buf_contents_t type = DMU_OT_IS_METADATA(ot) ?
	ARC_BUFC_METADATA : ARC_BUFC_DATA;

	ASSERT3U(lsize, >, 0);
	ASSERT3U(lsize, >=, psize);
	ASSERT3U(compression_type, >=, ZIO_COMPRESS_OFF);
	ASSERT3U(compression_type, <, ZIO_COMPRESS_FUNCTIONS);

	hdr = arc_hdr_alloc(spa_load_guid(spa), psize, lsize, B_TRUE,
	compression_type, complevel, type);

	hdr->b_crypt_hdr.b_dsobj = dsobj;
	hdr->b_crypt_hdr.b_ot = ot;
	hdr->b_l1hdr.b_byteswap = (byteorder == ZFS_HOST_BYTEORDER) ?
	DMU_BSWAP_NUMFUNCS : DMU_OT_BYTESWAP(ot);
	bcopy(salt, hdr->b_crypt_hdr.b_salt, ZIO_DATA_SALT_LEN);
	bcopy(iv, hdr->b_crypt_hdr.b_iv, ZIO_DATA_IV_LEN);
	bcopy(mac, hdr->b_crypt_hdr.b_mac, ZIO_DATA_MAC_LEN);

	/*
	* This buffer will be considered encrypted even if the ot is not an
	* encrypted type. It will become authenticated instead in
	* arc_write_ready().
	*/
	buf = NULL;
	VERIFY0(arc_buf_alloc_impl(hdr, spa, NULL, tag, B_TRUE, B_TRUE,
	B_FALSE, B_FALSE, &buf));
	arc_buf_thaw(buf);
	ASSERT3P(hdr->b_l1hdr.b_freeze_cksum, ==, NULL);

	return (buf);
	}

	static void
	l2arc_hdr_arcstats_update(arc_buf_hdr_t *hdr, boolean_t incr,
	boolean_t state_only)
	{
	l2arc_buf_hdr_t *l2hdr = &hdr->b_l2hdr;
	l2arc_dev_t *dev = l2hdr->b_dev;
	uint64_t lsize = HDR_GET_LSIZE(hdr);
	uint64_t psize = HDR_GET_PSIZE(hdr);
	uint64_t asize = vdev_psize_to_asize(dev->l2ad_vdev, psize);
	arc_buf_contents_t type = hdr->b_type;
	int64_t lsize_s;
	int64_t psize_s;
	int64_t asize_s;

	if (incr) {
	lsize_s = lsize;
	psize_s = psize;
	asize_s = asize;
	} else {
	lsize_s = -lsize;
	psize_s = -psize;
	asize_s = -asize;
	}

	/* If the buffer is a prefetch, count it as such. */
	if (HDR_PREFETCH(hdr)) {
	ARCSTAT_INCR(arcstat_l2_prefetch_asize, asize_s);
	} else {
	/*
	* We use the value stored in the L2 header upon initial
	* caching in L2ARC. This value will be updated in case
	* an MRU/MRU_ghost buffer transitions to MFU but the L2ARC
	* metadata (log entry) cannot currently be updated. Having
	* the ARC state in the L2 header solves the problem of a
	* possibly absent L1 header (apparent in buffers restored
	* from persistent L2ARC).
	*/
	switch (hdr->b_l2hdr.b_arcs_state) {
	case ARC_STATE_MRU_GHOST:
	case ARC_STATE_MRU:
	ARCSTAT_INCR(arcstat_l2_mru_asize, asize_s);
	break;
	case ARC_STATE_MFU_GHOST:
	case ARC_STATE_MFU:
	ARCSTAT_INCR(arcstat_l2_mfu_asize, asize_s);
	break;
	default:
	break;
	}
	}

	if (state_only)
	return;

	ARCSTAT_INCR(arcstat_l2_psize, psize_s);
	ARCSTAT_INCR(arcstat_l2_lsize, lsize_s);

	switch (type) {
	case ARC_BUFC_DATA:
	ARCSTAT_INCR(arcstat_l2_bufc_data_asize, asize_s);
	break;
	case ARC_BUFC_METADATA:
	ARCSTAT_INCR(arcstat_l2_bufc_metadata_asize, asize_s);
	break;
	default:
	break;
	}
	}


	static void
	arc_hdr_l2hdr_destroy(arc_buf_hdr_t *hdr)
	{
	l2arc_buf_hdr_t *l2hdr = &hdr->b_l2hdr;
	l2arc_dev_t *dev = l2hdr->b_dev;
	uint64_t psize = HDR_GET_PSIZE(hdr);
	uint64_t asize = vdev_psize_to_asize(dev->l2ad_vdev, psize);

	ASSERT(MUTEX_HELD(&dev->l2ad_mtx));
	ASSERT(HDR_HAS_L2HDR(hdr));

	list_remove(&dev->l2ad_buflist, hdr);

	l2arc_hdr_arcstats_decrement(hdr);
	vdev_space_update(dev->l2ad_vdev, -asize, 0, 0);

	(void) zfs_refcount_remove_many(&dev->l2ad_alloc, arc_hdr_size(hdr),
	hdr);
	arc_hdr_clear_flags(hdr, ARC_FLAG_HAS_L2HDR);
	}

	static void
	arc_hdr_destroy(arc_buf_hdr_t *hdr)
	{
	if (HDR_HAS_L1HDR(hdr)) {
	ASSERT(hdr->b_l1hdr.b_buf == NULL \|\|
	hdr->b_l1hdr.b_bufcnt > 0);
	ASSERT(zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
	ASSERT3P(hdr->b_l1hdr.b_state, ==, arc_anon);
	}
	ASSERT(!HDR_IO_IN_PROGRESS(hdr));
	ASSERT(!HDR_IN_HASH_TABLE(hdr));

	if (HDR_HAS_L2HDR(hdr)) {
	l2arc_dev_t *dev = hdr->b_l2hdr.b_dev;
	boolean_t buflist_held = MUTEX_HELD(&dev->l2ad_mtx);

	if (!buflist_held)
	mutex_enter(&dev->l2ad_mtx);

	/*
	* Even though we checked this conditional above, we
	* need to check this again now that we have the
	* l2ad_mtx. This is because we could be racing with
	* another thread calling l2arc_evict() which might have
	* destroyed this header's L2 portion as we were waiting
	* to acquire the l2ad_mtx. If that happens, we don't
	* want to re-destroy the header's L2 portion.
	*/
	if (HDR_HAS_L2HDR(hdr)) {

	if (!HDR_EMPTY(hdr))
	buf_discard_identity(hdr);

	arc_hdr_l2hdr_destroy(hdr);
	}

	if (!buflist_held)
	mutex_exit(&dev->l2ad_mtx);
	}

	/*
	* The header's identify can only be safely discarded once it is no
	* longer discoverable. This requires removing it from the hash table
	* and the l2arc header list. After this point the hash lock can not
	* be used to protect the header.
	*/
	if (!HDR_EMPTY(hdr))
	buf_discard_identity(hdr);

	if (HDR_HAS_L1HDR(hdr)) {
	arc_cksum_free(hdr);

	while (hdr->b_l1hdr.b_buf != NULL)
	arc_buf_destroy_impl(hdr->b_l1hdr.b_buf);

	if (hdr->b_l1hdr.b_pabd != NULL)
	arc_hdr_free_abd(hdr, B_FALSE);

	if (HDR_HAS_RABD(hdr))
	arc_hdr_free_abd(hdr, B_TRUE);
	}

	ASSERT3P(hdr->b_hash_next, ==, NULL);
	if (HDR_HAS_L1HDR(hdr)) {
	ASSERT(!multilist_link_active(&hdr->b_l1hdr.b_arc_node));
	ASSERT3P(hdr->b_l1hdr.b_acb, ==, NULL);

	if (!HDR_PROTECTED(hdr)) {
	kmem_cache_free(hdr_full_cache, hdr);
	} else {
	kmem_cache_free(hdr_full_crypt_cache, hdr);
	}
	} else {
	kmem_cache_free(hdr_l2only_cache, hdr);
	}
	}

	void
	arc_buf_destroy(arc_buf_t buf, void tag)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	if (hdr->b_l1hdr.b_state == arc_anon) {
	ASSERT3U(hdr->b_l1hdr.b_bufcnt, ==, 1);
	ASSERT(!HDR_IO_IN_PROGRESS(hdr));
	VERIFY0(remove_reference(hdr, NULL, tag));
	arc_hdr_destroy(hdr);
	return;
	}

	kmutex_t *hash_lock = HDR_LOCK(hdr);
	mutex_enter(hash_lock);

	ASSERT3P(hdr, ==, buf->b_hdr);
	ASSERT(hdr->b_l1hdr.b_bufcnt > 0);
	ASSERT3P(hash_lock, ==, HDR_LOCK(hdr));
	ASSERT3P(hdr->b_l1hdr.b_state, !=, arc_anon);
	ASSERT3P(buf->b_data, !=, NULL);

	(void) remove_reference(hdr, hash_lock, tag);
	arc_buf_destroy_impl(buf);
	mutex_exit(hash_lock);
	}

	/*
	* Evict the arc_buf_hdr that is provided as a parameter. The resultant
	* state of the header is dependent on its state prior to entering this
	* function. The following transitions are possible:
	*
	* - arc_mru -> arc_mru_ghost
	* - arc_mfu -> arc_mfu_ghost
	* - arc_mru_ghost -> arc_l2c_only
	* - arc_mru_ghost -> deleted
	* - arc_mfu_ghost -> arc_l2c_only
	* - arc_mfu_ghost -> deleted
	*
	* Return total size of evicted data buffers for eviction progress tracking.
	* When evicting from ghost states return logical buffer size to make eviction
	* progress at the same (or at least comparable) rate as from non-ghost states.
	*
	* Return *real_evicted for actual ARC size reduction to wake up threads
	* waiting for it. For non-ghost states it includes size of evicted data
	* buffers (the headers are not freed there). For ghost states it includes
	* only the evicted headers size.
	*/
	static int64_t
	arc_evict_hdr(arc_buf_hdr_t hdr, kmutex_t hash_lock, uint64_t *real_evicted)
	{
	arc_state_t evicted_state, state;
	int64_t bytes_evicted = 0;
	int min_lifetime = HDR_PRESCIENT_PREFETCH(hdr) ?
	arc_min_prescient_prefetch_ms : arc_min_prefetch_ms;

	ASSERT(MUTEX_HELD(hash_lock));
	ASSERT(HDR_HAS_L1HDR(hdr));

	*real_evicted = 0;
	state = hdr->b_l1hdr.b_state;
	if (GHOST_STATE(state)) {
	ASSERT(!HDR_IO_IN_PROGRESS(hdr));
	ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);

	/*
	* l2arc_write_buffers() relies on a header's L1 portion
	* (i.e. its b_pabd field) during it's write phase.
	* Thus, we cannot push a header onto the arc_l2c_only
	* state (removing its L1 piece) until the header is
	* done being written to the l2arc.
	*/
	if (HDR_HAS_L2HDR(hdr) && HDR_L2_WRITING(hdr)) {
	ARCSTAT_BUMP(arcstat_evict_l2_skip);
	return (bytes_evicted);
	}

	ARCSTAT_BUMP(arcstat_deleted);
	bytes_evicted += HDR_GET_LSIZE(hdr);

	DTRACE_PROBE1(arc__delete, arc_buf_hdr_t *, hdr);

	if (HDR_HAS_L2HDR(hdr)) {
	ASSERT(hdr->b_l1hdr.b_pabd == NULL);
	ASSERT(!HDR_HAS_RABD(hdr));
	/*
	* This buffer is cached on the 2nd Level ARC;
	* don't destroy the header.
	*/
	arc_change_state(arc_l2c_only, hdr, hash_lock);
	/*
	* dropping from L1+L2 cached to L2-only,
	* realloc to remove the L1 header.
	*/
	hdr = arc_hdr_realloc(hdr, hdr_full_cache,
	hdr_l2only_cache);
	*real_evicted += HDR_FULL_SIZE - HDR_L2ONLY_SIZE;
	} else {
	arc_change_state(arc_anon, hdr, hash_lock);
	arc_hdr_destroy(hdr);
	*real_evicted += HDR_FULL_SIZE;
	}
	return (bytes_evicted);
	}

	ASSERT(state == arc_mru \|\| state == arc_mfu);
	evicted_state = (state == arc_mru) ? arc_mru_ghost : arc_mfu_ghost;

	/* prefetch buffers have a minimum lifespan */
	if (HDR_IO_IN_PROGRESS(hdr) \|\|
	((hdr->b_flags & (ARC_FLAG_PREFETCH \| ARC_FLAG_INDIRECT)) &&
	ddi_get_lbolt() - hdr->b_l1hdr.b_arc_access <
	MSEC_TO_TICK(min_lifetime))) {
	ARCSTAT_BUMP(arcstat_evict_skip);
	return (bytes_evicted);
	}

	ASSERT0(zfs_refcount_count(&hdr->b_l1hdr.b_refcnt));
	while (hdr->b_l1hdr.b_buf) {
	arc_buf_t *buf = hdr->b_l1hdr.b_buf;
	if (!mutex_tryenter(&buf->b_evict_lock)) {
	ARCSTAT_BUMP(arcstat_mutex_miss);
	break;
	}
	if (buf->b_data != NULL) {
	bytes_evicted += HDR_GET_LSIZE(hdr);
	*real_evicted += HDR_GET_LSIZE(hdr);
	}
	mutex_exit(&buf->b_evict_lock);
	arc_buf_destroy_impl(buf);
	}

	if (HDR_HAS_L2HDR(hdr)) {
	ARCSTAT_INCR(arcstat_evict_l2_cached, HDR_GET_LSIZE(hdr));
	} else {
	if (l2arc_write_eligible(hdr->b_spa, hdr)) {
	ARCSTAT_INCR(arcstat_evict_l2_eligible,
	HDR_GET_LSIZE(hdr));

	switch (state->arcs_state) {
	case ARC_STATE_MRU:
	ARCSTAT_INCR(
	arcstat_evict_l2_eligible_mru,
	HDR_GET_LSIZE(hdr));
	break;
	case ARC_STATE_MFU:
	ARCSTAT_INCR(
	arcstat_evict_l2_eligible_mfu,
	HDR_GET_LSIZE(hdr));
	break;
	default:
	break;
	}
	} else {
	ARCSTAT_INCR(arcstat_evict_l2_ineligible,
	HDR_GET_LSIZE(hdr));
	}
	}

	if (hdr->b_l1hdr.b_bufcnt == 0) {
	arc_cksum_free(hdr);

	bytes_evicted += arc_hdr_size(hdr);
	*real_evicted += arc_hdr_size(hdr);

	/*
	* If this hdr is being evicted and has a compressed
	* buffer then we discard it here before we change states.
	* This ensures that the accounting is updated correctly
	* in arc_free_data_impl().
	*/
	if (hdr->b_l1hdr.b_pabd != NULL)
	arc_hdr_free_abd(hdr, B_FALSE);

	if (HDR_HAS_RABD(hdr))
	arc_hdr_free_abd(hdr, B_TRUE);

	arc_change_state(evicted_state, hdr, hash_lock);
	ASSERT(HDR_IN_HASH_TABLE(hdr));
	arc_hdr_set_flags(hdr, ARC_FLAG_IN_HASH_TABLE);
	DTRACE_PROBE1(arc__evict, arc_buf_hdr_t *, hdr);
	}

	return (bytes_evicted);
	}

	static void
	arc_set_need_free(void)
	{
	ASSERT(MUTEX_HELD(&arc_evict_lock));
	int64_t remaining = arc_free_memory() - arc_sys_free / 2;
	arc_evict_waiter_t *aw = list_tail(&arc_evict_waiters);
	if (aw == NULL) {
	arc_need_free = MAX(-remaining, 0);
	} else {
	arc_need_free =
	MAX(-remaining, (int64_t)(aw->aew_count - arc_evict_count));
	}
	}

	static uint64_t
	arc_evict_state_impl(multilist_t ml, int idx, arc_buf_hdr_t marker,
	uint64_t spa, uint64_t bytes)
	{
	multilist_sublist_t *mls;
	uint64_t bytes_evicted = 0, real_evicted = 0;
	arc_buf_hdr_t *hdr;
	kmutex_t *hash_lock;
	int evict_count = zfs_arc_evict_batch_limit;

	ASSERT3P(marker, !=, NULL);

	mls = multilist_sublist_lock(ml, idx);

	for (hdr = multilist_sublist_prev(mls, marker); likely(hdr != NULL);
	hdr = multilist_sublist_prev(mls, marker)) {
	if ((evict_count <= 0) \|\| (bytes_evicted >= bytes))
	break;

	/*
	* To keep our iteration location, move the marker
	* forward. Since we're not holding hdr's hash lock, we
	* must be very careful and not remove 'hdr' from the
	* sublist. Otherwise, other consumers might mistake the
	* 'hdr' as not being on a sublist when they call the
	* multilist_link_active() function (they all rely on
	* the hash lock protecting concurrent insertions and
	* removals). multilist_sublist_move_forward() was
	* specifically implemented to ensure this is the case
	* (only 'marker' will be removed and re-inserted).
	*/
	multilist_sublist_move_forward(mls, marker);

	/*
	* The only case where the b_spa field should ever be
	* zero, is the marker headers inserted by
	* arc_evict_state(). It's possible for multiple threads
	* to be calling arc_evict_state() concurrently (e.g.
	* dsl_pool_close() and zio_inject_fault()), so we must
	* skip any markers we see from these other threads.
	*/
	if (hdr->b_spa == 0)
	continue;

	/* we're only interested in evicting buffers of a certain spa */
	if (spa != 0 && hdr->b_spa != spa) {
	ARCSTAT_BUMP(arcstat_evict_skip);
	continue;
	}

	hash_lock = HDR_LOCK(hdr);

	/*
	* We aren't calling this function from any code path
	* that would already be holding a hash lock, so we're
	* asserting on this assumption to be defensive in case
	* this ever changes. Without this check, it would be
	* possible to incorrectly increment arcstat_mutex_miss
	* below (e.g. if the code changed such that we called
	* this function with a hash lock held).
	*/
	ASSERT(!MUTEX_HELD(hash_lock));

	if (mutex_tryenter(hash_lock)) {
	uint64_t revicted;
	uint64_t evicted = arc_evict_hdr(hdr, hash_lock,
	&revicted);
	mutex_exit(hash_lock);

	bytes_evicted += evicted;
	real_evicted += revicted;

	/*
	* If evicted is zero, arc_evict_hdr() must have
	* decided to skip this header, don't increment
	* evict_count in this case.
	*/
	if (evicted != 0)
	evict_count--;

	} else {
	ARCSTAT_BUMP(arcstat_mutex_miss);
	}
	}

	multilist_sublist_unlock(mls);

	/*
	* Increment the count of evicted bytes, and wake up any threads that
	* are waiting for the count to reach this value. Since the list is
	* ordered by ascending aew_count, we pop off the beginning of the
	* list until we reach the end, or a waiter that's past the current
	* "count". Doing this outside the loop reduces the number of times
	* we need to acquire the global arc_evict_lock.
	*
	* Only wake when there's sufficient free memory in the system
	* (specifically, arc_sys_free/2, which by default is a bit more than
	* 1/64th of RAM). See the comments in arc_wait_for_eviction().
	*/
	mutex_enter(&arc_evict_lock);
	arc_evict_count += real_evicted;

	if (arc_free_memory() > arc_sys_free / 2) {
	arc_evict_waiter_t *aw;
	while ((aw = list_head(&arc_evict_waiters)) != NULL &&
	aw->aew_count <= arc_evict_count) {
	list_remove(&arc_evict_waiters, aw);
	cv_broadcast(&aw->aew_cv);
	}
	}
	arc_set_need_free();
	mutex_exit(&arc_evict_lock);

	/*
	* If the ARC size is reduced from arc_c_max to arc_c_min (especially
	* if the average cached block is small), eviction can be on-CPU for
	* many seconds. To ensure that other threads that may be bound to
	* this CPU are able to make progress, make a voluntary preemption
	* call here.
	*/
	cond_resched();

	return (bytes_evicted);
	}

	/*
	* Evict buffers from the given arc state, until we've removed the
	* specified number of bytes. Move the removed buffers to the
	* appropriate evict state.
	*
	* This function makes a "best effort". It skips over any buffers
	* it can't get a hash_lock on, and so, may not catch all candidates.
	* It may also return without evicting as much space as requested.
	*
	* If bytes is specified using the special value ARC_EVICT_ALL, this
	* will evict all available (i.e. unlocked and evictable) buffers from
	* the given arc state; which is used by arc_flush().
	*/
	static uint64_t
	arc_evict_state(arc_state_t *state, uint64_t spa, uint64_t bytes,
	arc_buf_contents_t type)
	{
	uint64_t total_evicted = 0;
	multilist_t *ml = &state->arcs_list[type];
	int num_sublists;
	arc_buf_hdr_t **markers;

	num_sublists = multilist_get_num_sublists(ml);

	/*
	* If we've tried to evict from each sublist, made some
	* progress, but still have not hit the target number of bytes
	* to evict, we want to keep trying. The markers allow us to
	* pick up where we left off for each individual sublist, rather
	* than starting from the tail each time.
	*/
	markers = kmem_zalloc(sizeof (markers) num_sublists, KM_SLEEP);
	for (int i = 0; i < num_sublists; i++) {
	multilist_sublist_t *mls;

	markers[i] = kmem_cache_alloc(hdr_full_cache, KM_SLEEP);

	/*
	* A b_spa of 0 is used to indicate that this header is
	* a marker. This fact is used in arc_evict_type() and
	* arc_evict_state_impl().
	*/
	markers[i]->b_spa = 0;

	mls = multilist_sublist_lock(ml, i);
	multilist_sublist_insert_tail(mls, markers[i]);
	multilist_sublist_unlock(mls);
	}

	/*
	* While we haven't hit our target number of bytes to evict, or
	* we're evicting all available buffers.
	*/
	while (total_evicted < bytes) {
	int sublist_idx = multilist_get_random_index(ml);
	uint64_t scan_evicted = 0;

	/*
	* Try to reduce pinned dnodes with a floor of arc_dnode_limit.
	* Request that 10% of the LRUs be scanned by the superblock
	* shrinker.
	*/
	if (type == ARC_BUFC_DATA && aggsum_compare(
	&arc_sums.arcstat_dnode_size, arc_dnode_size_limit) > 0) {
	arc_prune_async((aggsum_upper_bound(
	&arc_sums.arcstat_dnode_size) -
	arc_dnode_size_limit) / sizeof (dnode_t) /
	zfs_arc_dnode_reduce_percent);
	}

	/*
	* Start eviction using a randomly selected sublist,
	* this is to try and evenly balance eviction across all
	* sublists. Always starting at the same sublist
	* (e.g. index 0) would cause evictions to favor certain
	* sublists over others.
	*/
	for (int i = 0; i < num_sublists; i++) {
	uint64_t bytes_remaining;
	uint64_t bytes_evicted;

	if (total_evicted < bytes)
	bytes_remaining = bytes - total_evicted;
	else
	break;

	bytes_evicted = arc_evict_state_impl(ml, sublist_idx,
	markers[sublist_idx], spa, bytes_remaining);

	scan_evicted += bytes_evicted;
	total_evicted += bytes_evicted;

	/* we've reached the end, wrap to the beginning */
	if (++sublist_idx >= num_sublists)
	sublist_idx = 0;
	}

	/*
	* If we didn't evict anything during this scan, we have
	* no reason to believe we'll evict more during another
	* scan, so break the loop.
	*/
	if (scan_evicted == 0) {
	/* This isn't possible, let's make that obvious */
	ASSERT3S(bytes, !=, 0);

	/*
	* When bytes is ARC_EVICT_ALL, the only way to
	* break the loop is when scan_evicted is zero.
	* In that case, we actually have evicted enough,
	* so we don't want to increment the kstat.
	*/
	if (bytes != ARC_EVICT_ALL) {
	ASSERT3S(total_evicted, <, bytes);
	ARCSTAT_BUMP(arcstat_evict_not_enough);
	}

	break;
	}
	}

	for (int i = 0; i < num_sublists; i++) {
	multilist_sublist_t *mls = multilist_sublist_lock(ml, i);
	multilist_sublist_remove(mls, markers[i]);
	multilist_sublist_unlock(mls);

	kmem_cache_free(hdr_full_cache, markers[i]);
	}
	kmem_free(markers, sizeof (markers) num_sublists);

	return (total_evicted);
	}

	/*
	* Flush all "evictable" data of the given type from the arc state
	* specified. This will not evict any "active" buffers (i.e. referenced).
	*
	* When 'retry' is set to B_FALSE, the function will make a single pass
	* over the state and evict any buffers that it can. Since it doesn't
	* continually retry the eviction, it might end up leaving some buffers
	* in the ARC due to lock misses.
	*
	* When 'retry' is set to B_TRUE, the function will continually retry the
	* eviction until all evictable buffers have been removed from the
	* state. As a result, if concurrent insertions into the state are
	* allowed (e.g. if the ARC isn't shutting down), this function might
	* wind up in an infinite loop, continually trying to evict buffers.
	*/
	static uint64_t
	arc_flush_state(arc_state_t *state, uint64_t spa, arc_buf_contents_t type,
	boolean_t retry)
	{
	uint64_t evicted = 0;

	while (zfs_refcount_count(&state->arcs_esize[type]) != 0) {
	evicted += arc_evict_state(state, spa, ARC_EVICT_ALL, type);

	if (!retry)
	break;
	}

	return (evicted);
	}

	/*
	* Evict the specified number of bytes from the state specified,
	* restricting eviction to the spa and type given. This function
	* prevents us from trying to evict more from a state's list than
	* is "evictable", and to skip evicting altogether when passed a
	* negative value for "bytes". In contrast, arc_evict_state() will
	* evict everything it can, when passed a negative value for "bytes".
	*/
	static uint64_t
	arc_evict_impl(arc_state_t *state, uint64_t spa, int64_t bytes,
	arc_buf_contents_t type)
	{
	uint64_t delta;

	if (bytes > 0 && zfs_refcount_count(&state->arcs_esize[type]) > 0) {
	delta = MIN(zfs_refcount_count(&state->arcs_esize[type]),
	bytes);
	return (arc_evict_state(state, spa, delta, type));
	}

	return (0);
	}

	/*
	* The goal of this function is to evict enough meta data buffers from the
	* ARC in order to enforce the arc_meta_limit. Achieving this is slightly
	* more complicated than it appears because it is common for data buffers
	* to have holds on meta data buffers. In addition, dnode meta data buffers
	* will be held by the dnodes in the block preventing them from being freed.
	* This means we can't simply traverse the ARC and expect to always find
	* enough unheld meta data buffer to release.
	*
	* Therefore, this function has been updated to make alternating passes
	* over the ARC releasing data buffers and then newly unheld meta data
	* buffers. This ensures forward progress is maintained and meta_used
	* will decrease. Normally this is sufficient, but if required the ARC
	* will call the registered prune callbacks causing dentry and inodes to
	* be dropped from the VFS cache. This will make dnode meta data buffers
	* available for reclaim.
	*/
	static uint64_t
	arc_evict_meta_balanced(uint64_t meta_used)
	{
	int64_t delta, prune = 0, adjustmnt;
	uint64_t total_evicted = 0;
	arc_buf_contents_t type = ARC_BUFC_DATA;
	int restarts = MAX(zfs_arc_meta_adjust_restarts, 0);

	restart:
	/*
	* This slightly differs than the way we evict from the mru in
	* arc_evict because we don't have a "target" value (i.e. no
	* "meta" arc_p). As a result, I think we can completely
	* cannibalize the metadata in the MRU before we evict the
	* metadata from the MFU. I think we probably need to implement a
	* "metadata arc_p" value to do this properly.
	*/
	adjustmnt = meta_used - arc_meta_limit;

	if (adjustmnt > 0 &&
	zfs_refcount_count(&arc_mru->arcs_esize[type]) > 0) {
	delta = MIN(zfs_refcount_count(&arc_mru->arcs_esize[type]),
	adjustmnt);
	total_evicted += arc_evict_impl(arc_mru, 0, delta, type);
	adjustmnt -= delta;
	}

	/*
	* We can't afford to recalculate adjustmnt here. If we do,
	* new metadata buffers can sneak into the MRU or ANON lists,
	* thus penalize the MFU metadata. Although the fudge factor is
	* small, it has been empirically shown to be significant for
	* certain workloads (e.g. creating many empty directories). As
	* such, we use the original calculation for adjustmnt, and
	* simply decrement the amount of data evicted from the MRU.
	*/

	if (adjustmnt > 0 &&
	zfs_refcount_count(&arc_mfu->arcs_esize[type]) > 0) {
	delta = MIN(zfs_refcount_count(&arc_mfu->arcs_esize[type]),
	adjustmnt);
	total_evicted += arc_evict_impl(arc_mfu, 0, delta, type);
	}

	adjustmnt = meta_used - arc_meta_limit;

	if (adjustmnt > 0 &&
	zfs_refcount_count(&arc_mru_ghost->arcs_esize[type]) > 0) {
	delta = MIN(adjustmnt,
	zfs_refcount_count(&arc_mru_ghost->arcs_esize[type]));
	total_evicted += arc_evict_impl(arc_mru_ghost, 0, delta, type);
	adjustmnt -= delta;
	}

	if (adjustmnt > 0 &&
	zfs_refcount_count(&arc_mfu_ghost->arcs_esize[type]) > 0) {
	delta = MIN(adjustmnt,
	zfs_refcount_count(&arc_mfu_ghost->arcs_esize[type]));
	total_evicted += arc_evict_impl(arc_mfu_ghost, 0, delta, type);
	}

	/*
	* If after attempting to make the requested adjustment to the ARC
	* the meta limit is still being exceeded then request that the
	* higher layers drop some cached objects which have holds on ARC
	* meta buffers. Requests to the upper layers will be made with
	* increasingly large scan sizes until the ARC is below the limit.
	*/
	if (meta_used > arc_meta_limit) {
	if (type == ARC_BUFC_DATA) {
	type = ARC_BUFC_METADATA;
	} else {
	type = ARC_BUFC_DATA;

	if (zfs_arc_meta_prune) {
	prune += zfs_arc_meta_prune;
	arc_prune_async(prune);
	}
	}

	if (restarts > 0) {
	restarts--;
	goto restart;
	}
	}
	return (total_evicted);
	}

	/*
	* Evict metadata buffers from the cache, such that arcstat_meta_used is
	* capped by the arc_meta_limit tunable.
	*/
	static uint64_t
	arc_evict_meta_only(uint64_t meta_used)
	{
	uint64_t total_evicted = 0;
	int64_t target;

	/*
	* If we're over the meta limit, we want to evict enough
	* metadata to get back under the meta limit. We don't want to
	* evict so much that we drop the MRU below arc_p, though. If
	* we're over the meta limit more than we're over arc_p, we
	* evict some from the MRU here, and some from the MFU below.
	*/
	target = MIN((int64_t)(meta_used - arc_meta_limit),
	(int64_t)(zfs_refcount_count(&arc_anon->arcs_size) +
	zfs_refcount_count(&arc_mru->arcs_size) - arc_p));

	total_evicted += arc_evict_impl(arc_mru, 0, target, ARC_BUFC_METADATA);

	/*
	* Similar to the above, we want to evict enough bytes to get us
	* below the meta limit, but not so much as to drop us below the
	* space allotted to the MFU (which is defined as arc_c - arc_p).
	*/
	target = MIN((int64_t)(meta_used - arc_meta_limit),
	(int64_t)(zfs_refcount_count(&arc_mfu->arcs_size) -
	(arc_c - arc_p)));

	total_evicted += arc_evict_impl(arc_mfu, 0, target, ARC_BUFC_METADATA);

	return (total_evicted);
	}

	static uint64_t
	arc_evict_meta(uint64_t meta_used)
	{
	if (zfs_arc_meta_strategy == ARC_STRATEGY_META_ONLY)
	return (arc_evict_meta_only(meta_used));
	else
	return (arc_evict_meta_balanced(meta_used));
	}

	/*
	* Return the type of the oldest buffer in the given arc state
	*
	* This function will select a random sublist of type ARC_BUFC_DATA and
	* a random sublist of type ARC_BUFC_METADATA. The tail of each sublist
	* is compared, and the type which contains the "older" buffer will be
	* returned.
	*/
	static arc_buf_contents_t
	arc_evict_type(arc_state_t *state)
	{
	multilist_t *data_ml = &state->arcs_list[ARC_BUFC_DATA];
	multilist_t *meta_ml = &state->arcs_list[ARC_BUFC_METADATA];
	int data_idx = multilist_get_random_index(data_ml);
	int meta_idx = multilist_get_random_index(meta_ml);
	multilist_sublist_t *data_mls;
	multilist_sublist_t *meta_mls;
	arc_buf_contents_t type;
	arc_buf_hdr_t *data_hdr;
	arc_buf_hdr_t *meta_hdr;

	/*
	* We keep the sublist lock until we're finished, to prevent
	* the headers from being destroyed via arc_evict_state().
	*/
	data_mls = multilist_sublist_lock(data_ml, data_idx);
	meta_mls = multilist_sublist_lock(meta_ml, meta_idx);

	/*
	* These two loops are to ensure we skip any markers that
	* might be at the tail of the lists due to arc_evict_state().
	*/

	for (data_hdr = multilist_sublist_tail(data_mls); data_hdr != NULL;
	data_hdr = multilist_sublist_prev(data_mls, data_hdr)) {
	if (data_hdr->b_spa != 0)
	break;
	}

	for (meta_hdr = multilist_sublist_tail(meta_mls); meta_hdr != NULL;
	meta_hdr = multilist_sublist_prev(meta_mls, meta_hdr)) {
	if (meta_hdr->b_spa != 0)
	break;
	}

	if (data_hdr == NULL && meta_hdr == NULL) {
	type = ARC_BUFC_DATA;
	} else if (data_hdr == NULL) {
	ASSERT3P(meta_hdr, !=, NULL);
	type = ARC_BUFC_METADATA;
	} else if (meta_hdr == NULL) {
	ASSERT3P(data_hdr, !=, NULL);
	type = ARC_BUFC_DATA;
	} else {
	ASSERT3P(data_hdr, !=, NULL);
	ASSERT3P(meta_hdr, !=, NULL);

	/* The headers can't be on the sublist without an L1 header */
	ASSERT(HDR_HAS_L1HDR(data_hdr));
	ASSERT(HDR_HAS_L1HDR(meta_hdr));

	if (data_hdr->b_l1hdr.b_arc_access <
	meta_hdr->b_l1hdr.b_arc_access) {
	type = ARC_BUFC_DATA;
	} else {
	type = ARC_BUFC_METADATA;
	}
	}

	multilist_sublist_unlock(meta_mls);
	multilist_sublist_unlock(data_mls);

	return (type);
	}

	/*
	* Evict buffers from the cache, such that arcstat_size is capped by arc_c.
	*/
	static uint64_t
	arc_evict(void)
	{
	uint64_t total_evicted = 0;
	uint64_t bytes;
	int64_t target;
	uint64_t asize = aggsum_value(&arc_sums.arcstat_size);
	uint64_t ameta = aggsum_value(&arc_sums.arcstat_meta_used);

	/*
	* If we're over arc_meta_limit, we want to correct that before
	* potentially evicting data buffers below.
	*/
	total_evicted += arc_evict_meta(ameta);

	/*
	* Adjust MRU size
	*
	* If we're over the target cache size, we want to evict enough
	* from the list to get back to our target size. We don't want
	* to evict too much from the MRU, such that it drops below
	* arc_p. So, if we're over our target cache size more than
	* the MRU is over arc_p, we'll evict enough to get back to
	* arc_p here, and then evict more from the MFU below.
	*/
	target = MIN((int64_t)(asize - arc_c),
	(int64_t)(zfs_refcount_count(&arc_anon->arcs_size) +
	zfs_refcount_count(&arc_mru->arcs_size) + ameta - arc_p));

	/*
	* If we're below arc_meta_min, always prefer to evict data.
	* Otherwise, try to satisfy the requested number of bytes to
	* evict from the type which contains older buffers; in an
	* effort to keep newer buffers in the cache regardless of their
	* type. If we cannot satisfy the number of bytes from this
	* type, spill over into the next type.
	*/
	if (arc_evict_type(arc_mru) == ARC_BUFC_METADATA &&
	ameta > arc_meta_min) {
	bytes = arc_evict_impl(arc_mru, 0, target, ARC_BUFC_METADATA);
	total_evicted += bytes;

	/*
	* If we couldn't evict our target number of bytes from
	* metadata, we try to get the rest from data.
	*/
	target -= bytes;

	total_evicted +=
	arc_evict_impl(arc_mru, 0, target, ARC_BUFC_DATA);
	} else {
	bytes = arc_evict_impl(arc_mru, 0, target, ARC_BUFC_DATA);
	total_evicted += bytes;

	/*
	* If we couldn't evict our target number of bytes from
	* data, we try to get the rest from metadata.
	*/
	target -= bytes;

	total_evicted +=
	arc_evict_impl(arc_mru, 0, target, ARC_BUFC_METADATA);
	}

	/*
	* Re-sum ARC stats after the first round of evictions.
	*/
	asize = aggsum_value(&arc_sums.arcstat_size);
	ameta = aggsum_value(&arc_sums.arcstat_meta_used);


	/*
	* Adjust MFU size
	*
	* Now that we've tried to evict enough from the MRU to get its
	* size back to arc_p, if we're still above the target cache
	* size, we evict the rest from the MFU.
	*/
	target = asize - arc_c;

	if (arc_evict_type(arc_mfu) == ARC_BUFC_METADATA &&
	ameta > arc_meta_min) {
	bytes = arc_evict_impl(arc_mfu, 0, target, ARC_BUFC_METADATA);
	total_evicted += bytes;

	/*
	* If we couldn't evict our target number of bytes from
	* metadata, we try to get the rest from data.
	*/
	target -= bytes;

	total_evicted +=
	arc_evict_impl(arc_mfu, 0, target, ARC_BUFC_DATA);
	} else {
	bytes = arc_evict_impl(arc_mfu, 0, target, ARC_BUFC_DATA);
	total_evicted += bytes;

	/*
	* If we couldn't evict our target number of bytes from
	* data, we try to get the rest from data.
	*/
	target -= bytes;

	total_evicted +=
	arc_evict_impl(arc_mfu, 0, target, ARC_BUFC_METADATA);
	}

	/*
	* Adjust ghost lists
	*
	* In addition to the above, the ARC also defines target values
	* for the ghost lists. The sum of the mru list and mru ghost
	* list should never exceed the target size of the cache, and
	* the sum of the mru list, mfu list, mru ghost list, and mfu
	* ghost list should never exceed twice the target size of the
	* cache. The following logic enforces these limits on the ghost
	* caches, and evicts from them as needed.
	*/
	target = zfs_refcount_count(&arc_mru->arcs_size) +
	zfs_refcount_count(&arc_mru_ghost->arcs_size) - arc_c;

	bytes = arc_evict_impl(arc_mru_ghost, 0, target, ARC_BUFC_DATA);
	total_evicted += bytes;

	target -= bytes;

	total_evicted +=
	arc_evict_impl(arc_mru_ghost, 0, target, ARC_BUFC_METADATA);

	/*
	* We assume the sum of the mru list and mfu list is less than
	* or equal to arc_c (we enforced this above), which means we
	* can use the simpler of the two equations below:
	*
	* mru + mfu + mru ghost + mfu ghost <= 2 * arc_c
	* mru ghost + mfu ghost <= arc_c
	*/
	target = zfs_refcount_count(&arc_mru_ghost->arcs_size) +
	zfs_refcount_count(&arc_mfu_ghost->arcs_size) - arc_c;

	bytes = arc_evict_impl(arc_mfu_ghost, 0, target, ARC_BUFC_DATA);
	total_evicted += bytes;

	target -= bytes;

	total_evicted +=
	arc_evict_impl(arc_mfu_ghost, 0, target, ARC_BUFC_METADATA);

	return (total_evicted);
	}

	void
	arc_flush(spa_t *spa, boolean_t retry)
	{
	uint64_t guid = 0;

	/*
	* If retry is B_TRUE, a spa must not be specified since we have
	* no good way to determine if all of a spa's buffers have been
	* evicted from an arc state.
	*/
	ASSERT(!retry \|\| spa == 0);

	if (spa != NULL)
	guid = spa_load_guid(spa);

	(void) arc_flush_state(arc_mru, guid, ARC_BUFC_DATA, retry);
	(void) arc_flush_state(arc_mru, guid, ARC_BUFC_METADATA, retry);

	(void) arc_flush_state(arc_mfu, guid, ARC_BUFC_DATA, retry);
	(void) arc_flush_state(arc_mfu, guid, ARC_BUFC_METADATA, retry);

	(void) arc_flush_state(arc_mru_ghost, guid, ARC_BUFC_DATA, retry);
	(void) arc_flush_state(arc_mru_ghost, guid, ARC_BUFC_METADATA, retry);

	(void) arc_flush_state(arc_mfu_ghost, guid, ARC_BUFC_DATA, retry);
	(void) arc_flush_state(arc_mfu_ghost, guid, ARC_BUFC_METADATA, retry);
	}

	void
	arc_reduce_target_size(int64_t to_free)
	{
	uint64_t asize = aggsum_value(&arc_sums.arcstat_size);

	/*
	* All callers want the ARC to actually evict (at least) this much
	* memory. Therefore we reduce from the lower of the current size and
	* the target size. This way, even if arc_c is much higher than
	* arc_size (as can be the case after many calls to arc_freed(), we will
	* immediately have arc_c < arc_size and therefore the arc_evict_zthr
	* will evict.
	*/
	uint64_t c = MIN(arc_c, asize);

	if (c > to_free && c - to_free > arc_c_min) {
	arc_c = c - to_free;
	atomic_add_64(&arc_p, -(arc_p >> arc_shrink_shift));
	if (arc_p > arc_c)
	arc_p = (arc_c >> 1);
	ASSERT(arc_c >= arc_c_min);
	ASSERT((int64_t)arc_p >= 0);
	} else {
	arc_c = arc_c_min;
	}

	if (asize > arc_c) {
	/* See comment in arc_evict_cb_check() on why lock+flag */
	mutex_enter(&arc_evict_lock);
	arc_evict_needed = B_TRUE;
	mutex_exit(&arc_evict_lock);
	zthr_wakeup(arc_evict_zthr);
	}
	}

	/*
	* Determine if the system is under memory pressure and is asking
	* to reclaim memory. A return value of B_TRUE indicates that the system
	* is under memory pressure and that the arc should adjust accordingly.
	*/
	boolean_t
	arc_reclaim_needed(void)
	{
	return (arc_available_memory() < 0);
	}

	void
	arc_kmem_reap_soon(void)
	{
	size_t i;
	kmem_cache_t *prev_cache = NULL;
	kmem_cache_t *prev_data_cache = NULL;
	extern kmem_cache_t *zio_buf_cache[];
	extern kmem_cache_t *zio_data_buf_cache[];

	#ifdef _KERNEL
	if ((aggsum_compare(&arc_sums.arcstat_meta_used,
	arc_meta_limit) >= 0) && zfs_arc_meta_prune) {
	/*
	* We are exceeding our meta-data cache limit.
	* Prune some entries to release holds on meta-data.
	*/
	arc_prune_async(zfs_arc_meta_prune);
	}
	#if defined(_ILP32)
	/*
	* Reclaim unused memory from all kmem caches.
	*/
	kmem_reap();
	#endif
	#endif

	for (i = 0; i < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; i++) {
	#if defined(_ILP32)
	/* reach upper limit of cache size on 32-bit */
	if (zio_buf_cache[i] == NULL)
	break;
	#endif
	if (zio_buf_cache[i] != prev_cache) {
	prev_cache = zio_buf_cache[i];
	kmem_cache_reap_now(zio_buf_cache[i]);
	}
	if (zio_data_buf_cache[i] != prev_data_cache) {
	prev_data_cache = zio_data_buf_cache[i];
	kmem_cache_reap_now(zio_data_buf_cache[i]);
	}
	}
	kmem_cache_reap_now(buf_cache);
	kmem_cache_reap_now(hdr_full_cache);
	kmem_cache_reap_now(hdr_l2only_cache);
	kmem_cache_reap_now(zfs_btree_leaf_cache);
	abd_cache_reap_now();
	}

	/* ARGSUSED */
	static boolean_t
	arc_evict_cb_check(void arg, zthr_t zthr)
	{
	#ifdef ZFS_DEBUG
	/*
	* This is necessary in order to keep the kstat information
	* up to date for tools that display kstat data such as the
	* mdb ::arc dcmd and the Linux crash utility. These tools
	* typically do not call kstat's update function, but simply
	* dump out stats from the most recent update. Without
	* this call, these commands may show stale stats for the
	* anon, mru, mru_ghost, mfu, and mfu_ghost lists. Even
	* with this call, the data might be out of date if the
	* evict thread hasn't been woken recently; but that should
	* suffice. The arc_state_t structures can be queried
	* directly if more accurate information is needed.
	*/
	if (arc_ksp != NULL)
	arc_ksp->ks_update(arc_ksp, KSTAT_READ);
	#endif

	/*
	* We have to rely on arc_wait_for_eviction() to tell us when to
	* evict, rather than checking if we are overflowing here, so that we
	* are sure to not leave arc_wait_for_eviction() waiting on aew_cv.
	* If we have become "not overflowing" since arc_wait_for_eviction()
	* checked, we need to wake it up. We could broadcast the CV here,
	* but arc_wait_for_eviction() may have not yet gone to sleep. We
	* would need to use a mutex to ensure that this function doesn't
	* broadcast until arc_wait_for_eviction() has gone to sleep (e.g.
	* the arc_evict_lock). However, the lock ordering of such a lock
	* would necessarily be incorrect with respect to the zthr_lock,
	* which is held before this function is called, and is held by
	* arc_wait_for_eviction() when it calls zthr_wakeup().
	*/
	return (arc_evict_needed);
	}

	/*
	* Keep arc_size under arc_c by running arc_evict which evicts data
	* from the ARC.
	*/
	/* ARGSUSED */
	static void
	arc_evict_cb(void arg, zthr_t zthr)
	{
	uint64_t evicted = 0;
	fstrans_cookie_t cookie = spl_fstrans_mark();

	/* Evict from cache */
	evicted = arc_evict();

	/*
	* If evicted is zero, we couldn't evict anything
	* via arc_evict(). This could be due to hash lock
	* collisions, but more likely due to the majority of
	* arc buffers being unevictable. Therefore, even if
	* arc_size is above arc_c, another pass is unlikely to
	* be helpful and could potentially cause us to enter an
	* infinite loop. Additionally, zthr_iscancelled() is
	* checked here so that if the arc is shutting down, the
	* broadcast will wake any remaining arc evict waiters.
	*/
	mutex_enter(&arc_evict_lock);
	arc_evict_needed = !zthr_iscancelled(arc_evict_zthr) &&
	evicted > 0 && aggsum_compare(&arc_sums.arcstat_size, arc_c) > 0;
	if (!arc_evict_needed) {
	/*
	* We're either no longer overflowing, or we
	* can't evict anything more, so we should wake
	* arc_get_data_impl() sooner.
	*/
	arc_evict_waiter_t *aw;
	while ((aw = list_remove_head(&arc_evict_waiters)) != NULL) {
	cv_broadcast(&aw->aew_cv);
	}
	arc_set_need_free();
	}
	mutex_exit(&arc_evict_lock);
	spl_fstrans_unmark(cookie);
	}

	/* ARGSUSED */
	static boolean_t
	arc_reap_cb_check(void arg, zthr_t zthr)
	{
	int64_t free_memory = arc_available_memory();
	static int reap_cb_check_counter = 0;

	/*
	* If a kmem reap is already active, don't schedule more. We must
	* check for this because kmem_cache_reap_soon() won't actually
	* block on the cache being reaped (this is to prevent callers from
	* becoming implicitly blocked by a system-wide kmem reap -- which,
	* on a system with many, many full magazines, can take minutes).
	*/
	if (!kmem_cache_reap_active() && free_memory < 0) {

	arc_no_grow = B_TRUE;
	arc_warm = B_TRUE;
	/*
	* Wait at least zfs_grow_retry (default 5) seconds
	* before considering growing.
	*/
	arc_growtime = gethrtime() + SEC2NSEC(arc_grow_retry);
	return (B_TRUE);
	} else if (free_memory < arc_c >> arc_no_grow_shift) {
	arc_no_grow = B_TRUE;
	} else if (gethrtime() >= arc_growtime) {
	arc_no_grow = B_FALSE;
	}

	/*
	* Called unconditionally every 60 seconds to reclaim unused
	* zstd compression and decompression context. This is done
	* here to avoid the need for an independent thread.
	*/
	if (!((reap_cb_check_counter++) % 60))
	zfs_zstd_cache_reap_now();

	return (B_FALSE);
	}

	/*
	* Keep enough free memory in the system by reaping the ARC's kmem
	* caches. To cause more slabs to be reapable, we may reduce the
	* target size of the cache (arc_c), causing the arc_evict_cb()
	* to free more buffers.
	*/
	/* ARGSUSED */
	static void
	arc_reap_cb(void arg, zthr_t zthr)
	{
	int64_t free_memory;
	fstrans_cookie_t cookie = spl_fstrans_mark();

	/*
	* Kick off asynchronous kmem_reap()'s of all our caches.
	*/
	arc_kmem_reap_soon();

	/*
	* Wait at least arc_kmem_cache_reap_retry_ms between
	* arc_kmem_reap_soon() calls. Without this check it is possible to
	* end up in a situation where we spend lots of time reaping
	* caches, while we're near arc_c_min. Waiting here also gives the
	* subsequent free memory check a chance of finding that the
	* asynchronous reap has already freed enough memory, and we don't
	* need to call arc_reduce_target_size().
	*/
	delay((hz * arc_kmem_cache_reap_retry_ms + 999) / 1000);

	/*
	* Reduce the target size as needed to maintain the amount of free
	* memory in the system at a fraction of the arc_size (1/128th by
	* default). If oversubscribed (free_memory < 0) then reduce the
	* target arc_size by the deficit amount plus the fractional
	* amount. If free memory is positive but less than the fractional
	* amount, reduce by what is needed to hit the fractional amount.
	*/
	free_memory = arc_available_memory();

	int64_t to_free =
	(arc_c >> arc_shrink_shift) - free_memory;
	if (to_free > 0) {
	arc_reduce_target_size(to_free);
	}
	spl_fstrans_unmark(cookie);
	}

	#ifdef _KERNEL
	/*
	* Determine the amount of memory eligible for eviction contained in the
	* ARC. All clean data reported by the ghost lists can always be safely
	* evicted. Due to arc_c_min, the same does not hold for all clean data
	* contained by the regular mru and mfu lists.
	*
	* In the case of the regular mru and mfu lists, we need to report as
	* much clean data as possible, such that evicting that same reported
	* data will not bring arc_size below arc_c_min. Thus, in certain
	* circumstances, the total amount of clean data in the mru and mfu
	* lists might not actually be evictable.
	*
	* The following two distinct cases are accounted for:
	*
	* 1. The sum of the amount of dirty data contained by both the mru and
	* mfu lists, plus the ARC's other accounting (e.g. the anon list),
	* is greater than or equal to arc_c_min.
	* (i.e. amount of dirty data >= arc_c_min)
	*
	* This is the easy case; all clean data contained by the mru and mfu
	* lists is evictable. Evicting all clean data can only drop arc_size
	* to the amount of dirty data, which is greater than arc_c_min.
	*
	* 2. The sum of the amount of dirty data contained by both the mru and
	* mfu lists, plus the ARC's other accounting (e.g. the anon list),
	* is less than arc_c_min.
	* (i.e. arc_c_min > amount of dirty data)
	*
	* 2.1. arc_size is greater than or equal arc_c_min.
	* (i.e. arc_size >= arc_c_min > amount of dirty data)
	*
	* In this case, not all clean data from the regular mru and mfu
	* lists is actually evictable; we must leave enough clean data
	* to keep arc_size above arc_c_min. Thus, the maximum amount of
	* evictable data from the two lists combined, is exactly the
	* difference between arc_size and arc_c_min.
	*
	* 2.2. arc_size is less than arc_c_min
	* (i.e. arc_c_min > arc_size > amount of dirty data)
	*
	* In this case, none of the data contained in the mru and mfu
	* lists is evictable, even if it's clean. Since arc_size is
	* already below arc_c_min, evicting any more would only
	* increase this negative difference.
	*/

	#endif /* _KERNEL */

	/*
	* Adapt arc info given the number of bytes we are trying to add and
	* the state that we are coming from. This function is only called
	* when we are adding new content to the cache.
	*/
	static void
	arc_adapt(int bytes, arc_state_t *state)
	{
	int mult;
	uint64_t arc_p_min = (arc_c >> arc_p_min_shift);
	int64_t mrug_size = zfs_refcount_count(&arc_mru_ghost->arcs_size);
	int64_t mfug_size = zfs_refcount_count(&arc_mfu_ghost->arcs_size);

	ASSERT(bytes > 0);
	/*
	* Adapt the target size of the MRU list:
	* - if we just hit in the MRU ghost list, then increase
	* the target size of the MRU list.
	* - if we just hit in the MFU ghost list, then increase
	* the target size of the MFU list by decreasing the
	* target size of the MRU list.
	*/
	if (state == arc_mru_ghost) {
	mult = (mrug_size >= mfug_size) ? 1 : (mfug_size / mrug_size);
	if (!zfs_arc_p_dampener_disable)
	mult = MIN(mult, 10); /* avoid wild arc_p adjustment */

	arc_p = MIN(arc_c - arc_p_min, arc_p + bytes * mult);
	} else if (state == arc_mfu_ghost) {
	uint64_t delta;

	mult = (mfug_size >= mrug_size) ? 1 : (mrug_size / mfug_size);
	if (!zfs_arc_p_dampener_disable)
	mult = MIN(mult, 10);

	delta = MIN(bytes * mult, arc_p);
	arc_p = MAX(arc_p_min, arc_p - delta);
	}
	ASSERT((int64_t)arc_p >= 0);

	/*
	* Wake reap thread if we do not have any available memory
	*/
	if (arc_reclaim_needed()) {
	zthr_wakeup(arc_reap_zthr);
	return;
	}

	if (arc_no_grow)
	return;

	if (arc_c >= arc_c_max)
	return;

	/*
	* If we're within (2 * maxblocksize) bytes of the target
	* cache size, increment the target cache size
	*/
	ASSERT3U(arc_c, >=, 2ULL << SPA_MAXBLOCKSHIFT);
	if (aggsum_upper_bound(&arc_sums.arcstat_size) >=
	arc_c - (2ULL << SPA_MAXBLOCKSHIFT)) {
	atomic_add_64(&arc_c, (int64_t)bytes);
	if (arc_c > arc_c_max)
	arc_c = arc_c_max;
	else if (state == arc_anon)
	atomic_add_64(&arc_p, (int64_t)bytes);
	if (arc_p > arc_c)
	arc_p = arc_c;
	}
	ASSERT((int64_t)arc_p >= 0);
	}

	/*
	* Check if arc_size has grown past our upper threshold, determined by
	* zfs_arc_overflow_shift.
	*/
	static arc_ovf_level_t
	arc_is_overflowing(boolean_t use_reserve)
	{
	/* Always allow at least one block of overflow */
	int64_t overflow = MAX(SPA_MAXBLOCKSIZE,
	arc_c >> zfs_arc_overflow_shift);

	/*
	* We just compare the lower bound here for performance reasons. Our
	* primary goals are to make sure that the arc never grows without
	* bound, and that it can reach its maximum size. This check
	* accomplishes both goals. The maximum amount we could run over by is
	* 2 * aggsum_borrow_multiplier * NUM_CPUS * the average size of a block
	* in the ARC. In practice, that's in the tens of MB, which is low
	* enough to be safe.
	*/
	int64_t over = aggsum_lower_bound(&arc_sums.arcstat_size) -
	arc_c - overflow / 2;
	if (!use_reserve)
	overflow /= 2;
	return (over < 0 ? ARC_OVF_NONE :
	over < overflow ? ARC_OVF_SOME : ARC_OVF_SEVERE);
	}

	static abd_t *
	arc_get_data_abd(arc_buf_hdr_t hdr, uint64_t size, void tag,
	int alloc_flags)
	{
	arc_buf_contents_t type = arc_buf_type(hdr);

	arc_get_data_impl(hdr, size, tag, alloc_flags);
	if (type == ARC_BUFC_METADATA) {
	return (abd_alloc(size, B_TRUE));
	} else {
	ASSERT(type == ARC_BUFC_DATA);
	return (abd_alloc(size, B_FALSE));
	}
	}

	static void *
	arc_get_data_buf(arc_buf_hdr_t hdr, uint64_t size, void tag)
	{
	arc_buf_contents_t type = arc_buf_type(hdr);

	arc_get_data_impl(hdr, size, tag, ARC_HDR_DO_ADAPT);
	if (type == ARC_BUFC_METADATA) {
	return (zio_buf_alloc(size));
	} else {
	ASSERT(type == ARC_BUFC_DATA);
	return (zio_data_buf_alloc(size));
	}
	}

	/*
	* Wait for the specified amount of data (in bytes) to be evicted from the
	* ARC, and for there to be sufficient free memory in the system. Waiting for
	* eviction ensures that the memory used by the ARC decreases. Waiting for
	* free memory ensures that the system won't run out of free pages, regardless
	* of ARC behavior and settings. See arc_lowmem_init().
	*/
	void
	arc_wait_for_eviction(uint64_t amount, boolean_t use_reserve)
	{
	switch (arc_is_overflowing(use_reserve)) {
	case ARC_OVF_NONE:
	return;
	case ARC_OVF_SOME:
	/*
	* This is a bit racy without taking arc_evict_lock, but the
	* worst that can happen is we either call zthr_wakeup() extra
	* time due to race with other thread here, or the set flag
	* get cleared by arc_evict_cb(), which is unlikely due to
	* big hysteresis, but also not important since at this level
	* of overflow the eviction is purely advisory. Same time
	* taking the global lock here every time without waiting for
	* the actual eviction creates a significant lock contention.
	*/
	if (!arc_evict_needed) {
	arc_evict_needed = B_TRUE;
	zthr_wakeup(arc_evict_zthr);
	}
	return;
	case ARC_OVF_SEVERE:
	default:
	{
	arc_evict_waiter_t aw;
	list_link_init(&aw.aew_node);
	cv_init(&aw.aew_cv, NULL, CV_DEFAULT, NULL);

	uint64_t last_count = 0;
	mutex_enter(&arc_evict_lock);
	if (!list_is_empty(&arc_evict_waiters)) {
	arc_evict_waiter_t *last =
	list_tail(&arc_evict_waiters);
	last_count = last->aew_count;
	} else if (!arc_evict_needed) {
	arc_evict_needed = B_TRUE;
	zthr_wakeup(arc_evict_zthr);
	}
	/*
	* Note, the last waiter's count may be less than
	* arc_evict_count if we are low on memory in which
	* case arc_evict_state_impl() may have deferred
	* wakeups (but still incremented arc_evict_count).
	*/
	aw.aew_count = MAX(last_count, arc_evict_count) + amount;

	list_insert_tail(&arc_evict_waiters, &aw);

	arc_set_need_free();

	DTRACE_PROBE3(arc__wait__for__eviction,
	uint64_t, amount,
	uint64_t, arc_evict_count,
	uint64_t, aw.aew_count);

	/*
	* We will be woken up either when arc_evict_count reaches
	* aew_count, or when the ARC is no longer overflowing and
	* eviction completes.
	* In case of "false" wakeup, we will still be on the list.
	*/
	do {
	cv_wait(&aw.aew_cv, &arc_evict_lock);
	} while (list_link_active(&aw.aew_node));
	mutex_exit(&arc_evict_lock);

	cv_destroy(&aw.aew_cv);
	}
	}
	}

	/*
	* Allocate a block and return it to the caller. If we are hitting the
	* hard limit for the cache size, we must sleep, waiting for the eviction
	* thread to catch up. If we're past the target size but below the hard
	* limit, we'll only signal the reclaim thread and continue on.
	*/
	static void
	arc_get_data_impl(arc_buf_hdr_t hdr, uint64_t size, void tag,
	int alloc_flags)
	{
	arc_state_t *state = hdr->b_l1hdr.b_state;
	arc_buf_contents_t type = arc_buf_type(hdr);

	if (alloc_flags & ARC_HDR_DO_ADAPT)
	arc_adapt(size, state);

	/*
	* If arc_size is currently overflowing, we must be adding data
	* faster than we are evicting. To ensure we don't compound the
	* problem by adding more data and forcing arc_size to grow even
	* further past it's target size, we wait for the eviction thread to
	* make some progress. We also wait for there to be sufficient free
	* memory in the system, as measured by arc_free_memory().
	*
	* Specifically, we wait for zfs_arc_eviction_pct percent of the
	* requested size to be evicted. This should be more than 100%, to
	* ensure that that progress is also made towards getting arc_size
	* under arc_c. See the comment above zfs_arc_eviction_pct.
	*/
	arc_wait_for_eviction(size * zfs_arc_eviction_pct / 100,
	alloc_flags & ARC_HDR_USE_RESERVE);

	VERIFY3U(hdr->b_type, ==, type);
	if (type == ARC_BUFC_METADATA) {
	arc_space_consume(size, ARC_SPACE_META);
	} else {
	arc_space_consume(size, ARC_SPACE_DATA);
	}

	/*
	* Update the state size. Note that ghost states have a
	* "ghost size" and so don't need to be updated.
	*/
	if (!GHOST_STATE(state)) {

	(void) zfs_refcount_add_many(&state->arcs_size, size, tag);

	/*
	* If this is reached via arc_read, the link is
	* protected by the hash lock. If reached via
	* arc_buf_alloc, the header should not be accessed by
	* any other thread. And, if reached via arc_read_done,
	* the hash lock will protect it if it's found in the
	* hash table; otherwise no other thread should be
	* trying to [add\|remove]_reference it.
	*/
	if (multilist_link_active(&hdr->b_l1hdr.b_arc_node)) {
	ASSERT(zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
	(void) zfs_refcount_add_many(&state->arcs_esize[type],
	size, tag);
	}

	/*
	* If we are growing the cache, and we are adding anonymous
	* data, and we have outgrown arc_p, update arc_p
	*/
	if (aggsum_upper_bound(&arc_sums.arcstat_size) < arc_c &&
	hdr->b_l1hdr.b_state == arc_anon &&
	(zfs_refcount_count(&arc_anon->arcs_size) +
	zfs_refcount_count(&arc_mru->arcs_size) > arc_p))
	arc_p = MIN(arc_c, arc_p + size);
	}
	}

	static void
	arc_free_data_abd(arc_buf_hdr_t hdr, abd_t abd, uint64_t size, void *tag)
	{
	arc_free_data_impl(hdr, size, tag);
	abd_free(abd);
	}

	static void
	arc_free_data_buf(arc_buf_hdr_t hdr, void buf, uint64_t size, void *tag)
	{
	arc_buf_contents_t type = arc_buf_type(hdr);

	arc_free_data_impl(hdr, size, tag);
	if (type == ARC_BUFC_METADATA) {
	zio_buf_free(buf, size);
	} else {
	ASSERT(type == ARC_BUFC_DATA);
	zio_data_buf_free(buf, size);
	}
	}

	/*
	* Free the arc data buffer.
	*/
	static void
	arc_free_data_impl(arc_buf_hdr_t hdr, uint64_t size, void tag)
	{
	arc_state_t *state = hdr->b_l1hdr.b_state;
	arc_buf_contents_t type = arc_buf_type(hdr);

	/* protected by hash lock, if in the hash table */
	if (multilist_link_active(&hdr->b_l1hdr.b_arc_node)) {
	ASSERT(zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
	ASSERT(state != arc_anon && state != arc_l2c_only);

	(void) zfs_refcount_remove_many(&state->arcs_esize[type],
	size, tag);
	}
	(void) zfs_refcount_remove_many(&state->arcs_size, size, tag);

	VERIFY3U(hdr->b_type, ==, type);
	if (type == ARC_BUFC_METADATA) {
	arc_space_return(size, ARC_SPACE_META);
	} else {
	ASSERT(type == ARC_BUFC_DATA);
	arc_space_return(size, ARC_SPACE_DATA);
	}
	}

	/*
	* This routine is called whenever a buffer is accessed.
	* NOTE: the hash lock is dropped in this function.
	*/
	static void
	arc_access(arc_buf_hdr_t hdr, kmutex_t hash_lock)
	{
	clock_t now;

	ASSERT(MUTEX_HELD(hash_lock));
	ASSERT(HDR_HAS_L1HDR(hdr));

	if (hdr->b_l1hdr.b_state == arc_anon) {
	/*
	* This buffer is not in the cache, and does not
	* appear in our "ghost" list. Add the new buffer
	* to the MRU state.
	*/

	ASSERT0(hdr->b_l1hdr.b_arc_access);
	hdr->b_l1hdr.b_arc_access = ddi_get_lbolt();
	DTRACE_PROBE1(new_state__mru, arc_buf_hdr_t *, hdr);
	arc_change_state(arc_mru, hdr, hash_lock);

	} else if (hdr->b_l1hdr.b_state == arc_mru) {
	now = ddi_get_lbolt();

	/*
	* If this buffer is here because of a prefetch, then either:
	* - clear the flag if this is a "referencing" read
	* (any subsequent access will bump this into the MFU state).
	* or
	* - move the buffer to the head of the list if this is
	* another prefetch (to make it less likely to be evicted).
	*/
	if (HDR_PREFETCH(hdr) \|\| HDR_PRESCIENT_PREFETCH(hdr)) {
	if (zfs_refcount_count(&hdr->b_l1hdr.b_refcnt) == 0) {
	/* link protected by hash lock */
	ASSERT(multilist_link_active(
	&hdr->b_l1hdr.b_arc_node));
	} else {
	if (HDR_HAS_L2HDR(hdr))
	l2arc_hdr_arcstats_decrement_state(hdr);
	arc_hdr_clear_flags(hdr,
	ARC_FLAG_PREFETCH \|
	ARC_FLAG_PRESCIENT_PREFETCH);
	hdr->b_l1hdr.b_mru_hits++;
	ARCSTAT_BUMP(arcstat_mru_hits);
	if (HDR_HAS_L2HDR(hdr))
	l2arc_hdr_arcstats_increment_state(hdr);
	}
	hdr->b_l1hdr.b_arc_access = now;
	return;
	}

	/*
	* This buffer has been "accessed" only once so far,
	* but it is still in the cache. Move it to the MFU
	* state.
	*/
	if (ddi_time_after(now, hdr->b_l1hdr.b_arc_access +
	ARC_MINTIME)) {
	/*
	* More than 125ms have passed since we
	* instantiated this buffer. Move it to the
	* most frequently used state.
	*/
	hdr->b_l1hdr.b_arc_access = now;
	DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, hdr);
	arc_change_state(arc_mfu, hdr, hash_lock);
	}
	hdr->b_l1hdr.b_mru_hits++;
	ARCSTAT_BUMP(arcstat_mru_hits);
	} else if (hdr->b_l1hdr.b_state == arc_mru_ghost) {
	arc_state_t *new_state;
	/*
	* This buffer has been "accessed" recently, but
	* was evicted from the cache. Move it to the
	* MFU state.
	*/
	if (HDR_PREFETCH(hdr) \|\| HDR_PRESCIENT_PREFETCH(hdr)) {
	new_state = arc_mru;
	if (zfs_refcount_count(&hdr->b_l1hdr.b_refcnt) > 0) {
	if (HDR_HAS_L2HDR(hdr))
	l2arc_hdr_arcstats_decrement_state(hdr);
	arc_hdr_clear_flags(hdr,
	ARC_FLAG_PREFETCH \|
	ARC_FLAG_PRESCIENT_PREFETCH);
	if (HDR_HAS_L2HDR(hdr))
	l2arc_hdr_arcstats_increment_state(hdr);
	}
	DTRACE_PROBE1(new_state__mru, arc_buf_hdr_t *, hdr);
	} else {
	new_state = arc_mfu;
	DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, hdr);
	}

	hdr->b_l1hdr.b_arc_access = ddi_get_lbolt();
	arc_change_state(new_state, hdr, hash_lock);

	hdr->b_l1hdr.b_mru_ghost_hits++;
	ARCSTAT_BUMP(arcstat_mru_ghost_hits);
	} else if (hdr->b_l1hdr.b_state == arc_mfu) {
	/*
	* This buffer has been accessed more than once and is
	* still in the cache. Keep it in the MFU state.
	*
	* NOTE: an add_reference() that occurred when we did
	* the arc_read() will have kicked this off the list.
	* If it was a prefetch, we will explicitly move it to
	* the head of the list now.
	*/

	hdr->b_l1hdr.b_mfu_hits++;
	ARCSTAT_BUMP(arcstat_mfu_hits);
	hdr->b_l1hdr.b_arc_access = ddi_get_lbolt();
	} else if (hdr->b_l1hdr.b_state == arc_mfu_ghost) {
	arc_state_t *new_state = arc_mfu;
	/*
	* This buffer has been accessed more than once but has
	* been evicted from the cache. Move it back to the
	* MFU state.
	*/

	if (HDR_PREFETCH(hdr) \|\| HDR_PRESCIENT_PREFETCH(hdr)) {
	/*
	* This is a prefetch access...
	* move this block back to the MRU state.
	*/
	new_state = arc_mru;
	}

	hdr->b_l1hdr.b_arc_access = ddi_get_lbolt();
	DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, hdr);
	arc_change_state(new_state, hdr, hash_lock);

	hdr->b_l1hdr.b_mfu_ghost_hits++;
	ARCSTAT_BUMP(arcstat_mfu_ghost_hits);
	} else if (hdr->b_l1hdr.b_state == arc_l2c_only) {
	/*
	* This buffer is on the 2nd Level ARC.
	*/

	hdr->b_l1hdr.b_arc_access = ddi_get_lbolt();
	DTRACE_PROBE1(new_state__mfu, arc_buf_hdr_t *, hdr);
	arc_change_state(arc_mfu, hdr, hash_lock);
	} else {
	cmn_err(CE_PANIC, "invalid arc state 0x%p",
	hdr->b_l1hdr.b_state);
	}
	}

	/*
	* This routine is called by dbuf_hold() to update the arc_access() state
	* which otherwise would be skipped for entries in the dbuf cache.
	*/
	void
	arc_buf_access(arc_buf_t *buf)
	{
	mutex_enter(&buf->b_evict_lock);
	arc_buf_hdr_t *hdr = buf->b_hdr;

	/*
	* Avoid taking the hash_lock when possible as an optimization.
	* The header must be checked again under the hash_lock in order
	* to handle the case where it is concurrently being released.
	*/
	if (hdr->b_l1hdr.b_state == arc_anon \|\| HDR_EMPTY(hdr)) {
	mutex_exit(&buf->b_evict_lock);
	return;
	}

	kmutex_t *hash_lock = HDR_LOCK(hdr);
	mutex_enter(hash_lock);

	if (hdr->b_l1hdr.b_state == arc_anon \|\| HDR_EMPTY(hdr)) {
	mutex_exit(hash_lock);
	mutex_exit(&buf->b_evict_lock);
	ARCSTAT_BUMP(arcstat_access_skip);
	return;
	}

	mutex_exit(&buf->b_evict_lock);

	ASSERT(hdr->b_l1hdr.b_state == arc_mru \|\|
	hdr->b_l1hdr.b_state == arc_mfu);

	DTRACE_PROBE1(arc__hit, arc_buf_hdr_t *, hdr);
	arc_access(hdr, hash_lock);
	mutex_exit(hash_lock);

	ARCSTAT_BUMP(arcstat_hits);
	ARCSTAT_CONDSTAT(!HDR_PREFETCH(hdr) && !HDR_PRESCIENT_PREFETCH(hdr),
	demand, prefetch, !HDR_ISTYPE_METADATA(hdr), data, metadata, hits);
	}

	/* a generic arc_read_done_func_t which you can use */
	/* ARGSUSED */
	void
	arc_bcopy_func(zio_t zio, const zbookmark_phys_t zb, const blkptr_t *bp,
	arc_buf_t buf, void arg)
	{
	if (buf == NULL)
	return;

	bcopy(buf->b_data, arg, arc_buf_size(buf));
	arc_buf_destroy(buf, arg);
	}

	/* a generic arc_read_done_func_t */
	/* ARGSUSED */
	void
	arc_getbuf_func(zio_t zio, const zbookmark_phys_t zb, const blkptr_t *bp,
	arc_buf_t buf, void arg)
	{
	arc_buf_t **bufp = arg;

	if (buf == NULL) {
	ASSERT(zio == NULL \|\| zio->io_error != 0);
	*bufp = NULL;
	} else {
	ASSERT(zio == NULL \|\| zio->io_error == 0);
	*bufp = buf;
	ASSERT(buf->b_data != NULL);
	}
	}

	static void
	arc_hdr_verify(arc_buf_hdr_t hdr, blkptr_t bp)
	{
	if (BP_IS_HOLE(bp) \|\| BP_IS_EMBEDDED(bp)) {
	ASSERT3U(HDR_GET_PSIZE(hdr), ==, 0);
	ASSERT3U(arc_hdr_get_compress(hdr), ==, ZIO_COMPRESS_OFF);
	} else {
	if (HDR_COMPRESSION_ENABLED(hdr)) {
	ASSERT3U(arc_hdr_get_compress(hdr), ==,
	BP_GET_COMPRESS(bp));
	}
	ASSERT3U(HDR_GET_LSIZE(hdr), ==, BP_GET_LSIZE(bp));
	ASSERT3U(HDR_GET_PSIZE(hdr), ==, BP_GET_PSIZE(bp));
	ASSERT3U(!!HDR_PROTECTED(hdr), ==, BP_IS_PROTECTED(bp));
	}
	}

	static void
	arc_read_done(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;
	arc_buf_hdr_t *hdr = zio->io_private;
	kmutex_t *hash_lock = NULL;
	arc_callback_t *callback_list;
	arc_callback_t *acb;
	boolean_t freeable = B_FALSE;

	/*
	* The hdr was inserted into hash-table and removed from lists
	* prior to starting I/O. We should find this header, since
	* it's in the hash table, and it should be legit since it's
	* not possible to evict it during the I/O. The only possible
	* reason for it not to be found is if we were freed during the
	* read.
	*/
	if (HDR_IN_HASH_TABLE(hdr)) {
	arc_buf_hdr_t *found;

	ASSERT3U(hdr->b_birth, ==, BP_PHYSICAL_BIRTH(zio->io_bp));
	ASSERT3U(hdr->b_dva.dva_word[0], ==,
	BP_IDENTITY(zio->io_bp)->dva_word[0]);
	ASSERT3U(hdr->b_dva.dva_word[1], ==,
	BP_IDENTITY(zio->io_bp)->dva_word[1]);

	found = buf_hash_find(hdr->b_spa, zio->io_bp, &hash_lock);

	ASSERT((found == hdr &&
	DVA_EQUAL(&hdr->b_dva, BP_IDENTITY(zio->io_bp))) \|\|
	(found == hdr && HDR_L2_READING(hdr)));
	ASSERT3P(hash_lock, !=, NULL);
	}

	if (BP_IS_PROTECTED(bp)) {
	hdr->b_crypt_hdr.b_ot = BP_GET_TYPE(bp);
	hdr->b_crypt_hdr.b_dsobj = zio->io_bookmark.zb_objset;
	zio_crypt_decode_params_bp(bp, hdr->b_crypt_hdr.b_salt,
	hdr->b_crypt_hdr.b_iv);

	- if (BP_GET_TYPE(bp) == DMU_OT_INTENT_LOG) {
	- void *tmpbuf;
	-
	- tmpbuf = abd_borrow_buf_copy(zio->io_abd,
	- sizeof (zil_chain_t));
	- zio_crypt_decode_mac_zil(tmpbuf,
	- hdr->b_crypt_hdr.b_mac);
	- abd_return_buf(zio->io_abd, tmpbuf,
	- sizeof (zil_chain_t));
	- } else {
	- zio_crypt_decode_mac_bp(bp, hdr->b_crypt_hdr.b_mac);
	+ if (zio->io_error == 0) {
	+ if (BP_GET_TYPE(bp) == DMU_OT_INTENT_LOG) {
	+ void *tmpbuf;
	+
	+ tmpbuf = abd_borrow_buf_copy(zio->io_abd,
	+ sizeof (zil_chain_t));
	+ zio_crypt_decode_mac_zil(tmpbuf,
	+ hdr->b_crypt_hdr.b_mac);
	+ abd_return_buf(zio->io_abd, tmpbuf,
	+ sizeof (zil_chain_t));
	+ } else {
	+ zio_crypt_decode_mac_bp(bp,
	+ hdr->b_crypt_hdr.b_mac);
	+ }
	}
	}

	if (zio->io_error == 0) {
	/* byteswap if necessary */
	if (BP_SHOULD_BYTESWAP(zio->io_bp)) {
	if (BP_GET_LEVEL(zio->io_bp) > 0) {
	hdr->b_l1hdr.b_byteswap = DMU_BSWAP_UINT64;
	} else {
	hdr->b_l1hdr.b_byteswap =
	DMU_OT_BYTESWAP(BP_GET_TYPE(zio->io_bp));
	}
	} else {
	hdr->b_l1hdr.b_byteswap = DMU_BSWAP_NUMFUNCS;
	}
	if (!HDR_L2_READING(hdr)) {
	hdr->b_complevel = zio->io_prop.zp_complevel;
	}
	}

	arc_hdr_clear_flags(hdr, ARC_FLAG_L2_EVICTED);
	if (l2arc_noprefetch && HDR_PREFETCH(hdr))
	arc_hdr_clear_flags(hdr, ARC_FLAG_L2CACHE);

	callback_list = hdr->b_l1hdr.b_acb;
	ASSERT3P(callback_list, !=, NULL);

	if (hash_lock && zio->io_error == 0 &&
	hdr->b_l1hdr.b_state == arc_anon) {
	/*
	* Only call arc_access on anonymous buffers. This is because
	* if we've issued an I/O for an evicted buffer, we've already
	* called arc_access (to prevent any simultaneous readers from
	* getting confused).
	*/
	arc_access(hdr, hash_lock);
	}

	/*
	* If a read request has a callback (i.e. acb_done is not NULL), then we
	* make a buf containing the data according to the parameters which were
	* passed in. The implementation of arc_buf_alloc_impl() ensures that we
	* aren't needlessly decompressing the data multiple times.
	*/
	int callback_cnt = 0;
	for (acb = callback_list; acb != NULL; acb = acb->acb_next) {
	if (!acb->acb_done \|\| acb->acb_nobuf)
	continue;

	callback_cnt++;

	if (zio->io_error != 0)
	continue;

	int error = arc_buf_alloc_impl(hdr, zio->io_spa,
	&acb->acb_zb, acb->acb_private, acb->acb_encrypted,
	acb->acb_compressed, acb->acb_noauth, B_TRUE,
	&acb->acb_buf);

	/*
	* Assert non-speculative zios didn't fail because an
	* encryption key wasn't loaded
	*/
	ASSERT((zio->io_flags & ZIO_FLAG_SPECULATIVE) \|\|
	error != EACCES);

	/*
	* If we failed to decrypt, report an error now (as the zio
	* layer would have done if it had done the transforms).
	*/
	if (error == ECKSUM) {
	ASSERT(BP_IS_PROTECTED(bp));
	error = SET_ERROR(EIO);
	if ((zio->io_flags & ZIO_FLAG_SPECULATIVE) == 0) {
	spa_log_error(zio->io_spa, &acb->acb_zb);
	(void) zfs_ereport_post(
	FM_EREPORT_ZFS_AUTHENTICATION,
	zio->io_spa, NULL, &acb->acb_zb, zio, 0);
	}
	}

	if (error != 0) {
	/*
	* Decompression or decryption failed. Set
	* io_error so that when we call acb_done
	* (below), we will indicate that the read
	* failed. Note that in the unusual case
	* where one callback is compressed and another
	* uncompressed, we will mark all of them
	* as failed, even though the uncompressed
	* one can't actually fail. In this case,
	* the hdr will not be anonymous, because
	* if there are multiple callbacks, it's
	* because multiple threads found the same
	* arc buf in the hash table.
	*/
	zio->io_error = error;
	}
	}

	/*
	* If there are multiple callbacks, we must have the hash lock,
	* because the only way for multiple threads to find this hdr is
	* in the hash table. This ensures that if there are multiple
	* callbacks, the hdr is not anonymous. If it were anonymous,
	* we couldn't use arc_buf_destroy() in the error case below.
	*/
	ASSERT(callback_cnt < 2 \|\| hash_lock != NULL);

	hdr->b_l1hdr.b_acb = NULL;
	arc_hdr_clear_flags(hdr, ARC_FLAG_IO_IN_PROGRESS);
	if (callback_cnt == 0)
	ASSERT(hdr->b_l1hdr.b_pabd != NULL \|\| HDR_HAS_RABD(hdr));

	ASSERT(zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt) \|\|
	callback_list != NULL);

	if (zio->io_error == 0) {
	arc_hdr_verify(hdr, zio->io_bp);
	} else {
	arc_hdr_set_flags(hdr, ARC_FLAG_IO_ERROR);
	if (hdr->b_l1hdr.b_state != arc_anon)
	arc_change_state(arc_anon, hdr, hash_lock);
	if (HDR_IN_HASH_TABLE(hdr))
	buf_hash_remove(hdr);
	freeable = zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt);
	}

	/*
	* Broadcast before we drop the hash_lock to avoid the possibility
	* that the hdr (and hence the cv) might be freed before we get to
	* the cv_broadcast().
	*/
	cv_broadcast(&hdr->b_l1hdr.b_cv);

	if (hash_lock != NULL) {
	mutex_exit(hash_lock);
	} else {
	/*
	* This block was freed while we waited for the read to
	* complete. It has been removed from the hash table and
	* moved to the anonymous state (so that it won't show up
	* in the cache).
	*/
	ASSERT3P(hdr->b_l1hdr.b_state, ==, arc_anon);
	freeable = zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt);
	}

	/* execute each callback and free its structure */
	while ((acb = callback_list) != NULL) {
	if (acb->acb_done != NULL) {
	if (zio->io_error != 0 && acb->acb_buf != NULL) {
	/*
	* If arc_buf_alloc_impl() fails during
	* decompression, the buf will still be
	* allocated, and needs to be freed here.
	*/
	arc_buf_destroy(acb->acb_buf,
	acb->acb_private);
	acb->acb_buf = NULL;
	}
	acb->acb_done(zio, &zio->io_bookmark, zio->io_bp,
	acb->acb_buf, acb->acb_private);
	}

	if (acb->acb_zio_dummy != NULL) {
	acb->acb_zio_dummy->io_error = zio->io_error;
	zio_nowait(acb->acb_zio_dummy);
	}

	callback_list = acb->acb_next;
	kmem_free(acb, sizeof (arc_callback_t));
	}

	if (freeable)
	arc_hdr_destroy(hdr);
	}

	/*
	* "Read" the block at the specified DVA (in bp) via the
	* cache. If the block is found in the cache, invoke the provided
	* callback immediately and return. Note that the `zio' parameter
	* in the callback will be NULL in this case, since no IO was
	* required. If the block is not in the cache pass the read request
	* on to the spa with a substitute callback function, so that the
	* requested block will be added to the cache.
	*
	* If a read request arrives for a block that has a read in-progress,
	* either wait for the in-progress read to complete (and return the
	* results); or, if this is a read with a "done" func, add a record
	* to the read to invoke the "done" func when the read completes,
	* and return; or just return.
	*
	* arc_read_done() will invoke all the requested "done" functions
	* for readers of this block.
	*/
	int
	arc_read(zio_t pio, spa_t spa, const blkptr_t *bp,
	arc_read_done_func_t done, void private, zio_priority_t priority,
	int zio_flags, arc_flags_t arc_flags, const zbookmark_phys_t zb)
	{
	arc_buf_hdr_t *hdr = NULL;
	kmutex_t *hash_lock = NULL;
	zio_t *rzio;
	uint64_t guid = spa_load_guid(spa);
	boolean_t compressed_read = (zio_flags & ZIO_FLAG_RAW_COMPRESS) != 0;
	boolean_t encrypted_read = BP_IS_ENCRYPTED(bp) &&
	(zio_flags & ZIO_FLAG_RAW_ENCRYPT) != 0;
	boolean_t noauth_read = BP_IS_AUTHENTICATED(bp) &&
	(zio_flags & ZIO_FLAG_RAW_ENCRYPT) != 0;
	boolean_t embedded_bp = !!BP_IS_EMBEDDED(bp);
	boolean_t no_buf = *arc_flags & ARC_FLAG_NO_BUF;
	int rc = 0;

	ASSERT(!embedded_bp \|\|
	BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA);
	ASSERT(!BP_IS_HOLE(bp));
	ASSERT(!BP_IS_REDACTED(bp));

	/*
	* Normally SPL_FSTRANS will already be set since kernel threads which
	* expect to call the DMU interfaces will set it when created. System
	* calls are similarly handled by setting/cleaning the bit in the
	* registered callback (module/os/.../zfs/zpl_*).
	*
	* External consumers such as Lustre which call the exported DMU
	* interfaces may not have set SPL_FSTRANS. To avoid a deadlock
	* on the hash_lock always set and clear the bit.
	*/
	fstrans_cookie_t cookie = spl_fstrans_mark();
	top:
	/*
	* Verify the block pointer contents are reasonable. This should
	* always be the case since the blkptr is protected by a checksum.
	* However, if there is damage it's desirable to detect this early
	* and treat it as a checksum error. This allows an alternate blkptr
	* to be tried when one is available (e.g. ditto blocks).
	*/
	if (!zfs_blkptr_verify(spa, bp, zio_flags & ZIO_FLAG_CONFIG_WRITER,
	BLK_VERIFY_LOG)) {
	rc = SET_ERROR(ECKSUM);
	goto out;
	}

	if (!embedded_bp) {
	/*
	* Embedded BP's have no DVA and require no I/O to "read".
	* Create an anonymous arc buf to back it.
	*/
	hdr = buf_hash_find(guid, bp, &hash_lock);
	}

	/*
	* Determine if we have an L1 cache hit or a cache miss. For simplicity
	* we maintain encrypted data separately from compressed / uncompressed
	* data. If the user is requesting raw encrypted data and we don't have
	* that in the header we will read from disk to guarantee that we can
	* get it even if the encryption keys aren't loaded.
	*/
	if (hdr != NULL && HDR_HAS_L1HDR(hdr) && (HDR_HAS_RABD(hdr) \|\|
	(hdr->b_l1hdr.b_pabd != NULL && !encrypted_read))) {
	arc_buf_t *buf = NULL;
	*arc_flags \|= ARC_FLAG_CACHED;

	if (HDR_IO_IN_PROGRESS(hdr)) {
	zio_t *head_zio = hdr->b_l1hdr.b_acb->acb_zio_head;

	if (*arc_flags & ARC_FLAG_CACHED_ONLY) {
	mutex_exit(hash_lock);
	ARCSTAT_BUMP(arcstat_cached_only_in_progress);
	rc = SET_ERROR(ENOENT);
	goto out;
	}

	ASSERT3P(head_zio, !=, NULL);
	if ((hdr->b_flags & ARC_FLAG_PRIO_ASYNC_READ) &&
	priority == ZIO_PRIORITY_SYNC_READ) {
	/*
	* This is a sync read that needs to wait for
	* an in-flight async read. Request that the
	* zio have its priority upgraded.
	*/
	zio_change_priority(head_zio, priority);
	DTRACE_PROBE1(arc__async__upgrade__sync,
	arc_buf_hdr_t *, hdr);
	ARCSTAT_BUMP(arcstat_async_upgrade_sync);
	}
	if (hdr->b_flags & ARC_FLAG_PREDICTIVE_PREFETCH) {
	arc_hdr_clear_flags(hdr,
	ARC_FLAG_PREDICTIVE_PREFETCH);
	}

	if (*arc_flags & ARC_FLAG_WAIT) {
	cv_wait(&hdr->b_l1hdr.b_cv, hash_lock);
	mutex_exit(hash_lock);
	goto top;
	}
	ASSERT(*arc_flags & ARC_FLAG_NOWAIT);

	if (done) {
	arc_callback_t *acb = NULL;

	acb = kmem_zalloc(sizeof (arc_callback_t),
	KM_SLEEP);
	acb->acb_done = done;
	acb->acb_private = private;
	acb->acb_compressed = compressed_read;
	acb->acb_encrypted = encrypted_read;
	acb->acb_noauth = noauth_read;
	acb->acb_nobuf = no_buf;
	acb->acb_zb = *zb;
	if (pio != NULL)
	acb->acb_zio_dummy = zio_null(pio,
	spa, NULL, NULL, NULL, zio_flags);

	ASSERT3P(acb->acb_done, !=, NULL);
	acb->acb_zio_head = head_zio;
	acb->acb_next = hdr->b_l1hdr.b_acb;
	hdr->b_l1hdr.b_acb = acb;
	}
	mutex_exit(hash_lock);
	goto out;
	}

	ASSERT(hdr->b_l1hdr.b_state == arc_mru \|\|
	hdr->b_l1hdr.b_state == arc_mfu);

	if (done && !no_buf) {
	if (hdr->b_flags & ARC_FLAG_PREDICTIVE_PREFETCH) {
	/*
	* This is a demand read which does not have to
	* wait for i/o because we did a predictive
	* prefetch i/o for it, which has completed.
	*/
	DTRACE_PROBE1(
	arc__demand__hit__predictive__prefetch,
	arc_buf_hdr_t *, hdr);
	ARCSTAT_BUMP(
	arcstat_demand_hit_predictive_prefetch);
	arc_hdr_clear_flags(hdr,
	ARC_FLAG_PREDICTIVE_PREFETCH);
	}

	if (hdr->b_flags & ARC_FLAG_PRESCIENT_PREFETCH) {
	ARCSTAT_BUMP(
	arcstat_demand_hit_prescient_prefetch);
	arc_hdr_clear_flags(hdr,
	ARC_FLAG_PRESCIENT_PREFETCH);
	}

	ASSERT(!embedded_bp \|\| !BP_IS_HOLE(bp));

	/* Get a buf with the desired data in it. */
	rc = arc_buf_alloc_impl(hdr, spa, zb, private,
	encrypted_read, compressed_read, noauth_read,
	B_TRUE, &buf);
	if (rc == ECKSUM) {
	/*
	* Convert authentication and decryption errors
	* to EIO (and generate an ereport if needed)
	* before leaving the ARC.
	*/
	rc = SET_ERROR(EIO);
	if ((zio_flags & ZIO_FLAG_SPECULATIVE) == 0) {
	spa_log_error(spa, zb);
	(void) zfs_ereport_post(
	FM_EREPORT_ZFS_AUTHENTICATION,
	spa, NULL, zb, NULL, 0);
	}
	}
	if (rc != 0) {
	(void) remove_reference(hdr, hash_lock,
	private);
	arc_buf_destroy_impl(buf);
	buf = NULL;
	}

	/* assert any errors weren't due to unloaded keys */
	ASSERT((zio_flags & ZIO_FLAG_SPECULATIVE) \|\|
	rc != EACCES);
	} else if (*arc_flags & ARC_FLAG_PREFETCH &&
	zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt)) {
	if (HDR_HAS_L2HDR(hdr))
	l2arc_hdr_arcstats_decrement_state(hdr);
	arc_hdr_set_flags(hdr, ARC_FLAG_PREFETCH);
	if (HDR_HAS_L2HDR(hdr))
	l2arc_hdr_arcstats_increment_state(hdr);
	}
	DTRACE_PROBE1(arc__hit, arc_buf_hdr_t *, hdr);
	arc_access(hdr, hash_lock);
	if (*arc_flags & ARC_FLAG_PRESCIENT_PREFETCH)
	arc_hdr_set_flags(hdr, ARC_FLAG_PRESCIENT_PREFETCH);
	if (*arc_flags & ARC_FLAG_L2CACHE)
	arc_hdr_set_flags(hdr, ARC_FLAG_L2CACHE);
	mutex_exit(hash_lock);
	ARCSTAT_BUMP(arcstat_hits);
	ARCSTAT_CONDSTAT(!HDR_PREFETCH(hdr),
	demand, prefetch, !HDR_ISTYPE_METADATA(hdr),
	data, metadata, hits);

	if (done)
	done(NULL, zb, bp, buf, private);
	} else {
	uint64_t lsize = BP_GET_LSIZE(bp);
	uint64_t psize = BP_GET_PSIZE(bp);
	arc_callback_t *acb;
	vdev_t *vd = NULL;
	uint64_t addr = 0;
	boolean_t devw = B_FALSE;
	uint64_t size;
	abd_t *hdr_abd;
	int alloc_flags = encrypted_read ? ARC_HDR_ALLOC_RDATA : 0;

	if (*arc_flags & ARC_FLAG_CACHED_ONLY) {
	rc = SET_ERROR(ENOENT);
	if (hash_lock != NULL)
	mutex_exit(hash_lock);
	goto out;
	}

	if (hdr == NULL) {
	/*
	* This block is not in the cache or it has
	* embedded data.
	*/
	arc_buf_hdr_t *exists = NULL;
	arc_buf_contents_t type = BP_GET_BUFC_TYPE(bp);
	hdr = arc_hdr_alloc(spa_load_guid(spa), psize, lsize,
	BP_IS_PROTECTED(bp), BP_GET_COMPRESS(bp), 0, type);

	if (!embedded_bp) {
	hdr->b_dva = *BP_IDENTITY(bp);
	hdr->b_birth = BP_PHYSICAL_BIRTH(bp);
	exists = buf_hash_insert(hdr, &hash_lock);
	}
	if (exists != NULL) {
	/* somebody beat us to the hash insert */
	mutex_exit(hash_lock);
	buf_discard_identity(hdr);
	arc_hdr_destroy(hdr);
	goto top; /* restart the IO request */
	}
	alloc_flags \|= ARC_HDR_DO_ADAPT;
	} else {
	/*
	* This block is in the ghost cache or encrypted data
	* was requested and we didn't have it. If it was
	* L2-only (and thus didn't have an L1 hdr),
	* we realloc the header to add an L1 hdr.
	*/
	if (!HDR_HAS_L1HDR(hdr)) {
	hdr = arc_hdr_realloc(hdr, hdr_l2only_cache,
	hdr_full_cache);
	}

	if (GHOST_STATE(hdr->b_l1hdr.b_state)) {
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	ASSERT(!HDR_HAS_RABD(hdr));
	ASSERT(!HDR_IO_IN_PROGRESS(hdr));
	ASSERT0(zfs_refcount_count(
	&hdr->b_l1hdr.b_refcnt));
	ASSERT3P(hdr->b_l1hdr.b_buf, ==, NULL);
	ASSERT3P(hdr->b_l1hdr.b_freeze_cksum, ==, NULL);
	} else if (HDR_IO_IN_PROGRESS(hdr)) {
	/*
	* If this header already had an IO in progress
	* and we are performing another IO to fetch
	* encrypted data we must wait until the first
	* IO completes so as not to confuse
	* arc_read_done(). This should be very rare
	* and so the performance impact shouldn't
	* matter.
	*/
	cv_wait(&hdr->b_l1hdr.b_cv, hash_lock);
	mutex_exit(hash_lock);
	goto top;
	}

	/*
	* This is a delicate dance that we play here.
	* This hdr might be in the ghost list so we access
	* it to move it out of the ghost list before we
	* initiate the read. If it's a prefetch then
	* it won't have a callback so we'll remove the
	* reference that arc_buf_alloc_impl() created. We
	* do this after we've called arc_access() to
	* avoid hitting an assert in remove_reference().
	*/
	arc_adapt(arc_hdr_size(hdr), hdr->b_l1hdr.b_state);
	arc_access(hdr, hash_lock);
	}

	arc_hdr_alloc_abd(hdr, alloc_flags);
	if (encrypted_read) {
	ASSERT(HDR_HAS_RABD(hdr));
	size = HDR_GET_PSIZE(hdr);
	hdr_abd = hdr->b_crypt_hdr.b_rabd;
	zio_flags \|= ZIO_FLAG_RAW;
	} else {
	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
	size = arc_hdr_size(hdr);
	hdr_abd = hdr->b_l1hdr.b_pabd;

	if (arc_hdr_get_compress(hdr) != ZIO_COMPRESS_OFF) {
	zio_flags \|= ZIO_FLAG_RAW_COMPRESS;
	}

	/*
	* For authenticated bp's, we do not ask the ZIO layer
	* to authenticate them since this will cause the entire
	* IO to fail if the key isn't loaded. Instead, we
	* defer authentication until arc_buf_fill(), which will
	* verify the data when the key is available.
	*/
	if (BP_IS_AUTHENTICATED(bp))
	zio_flags \|= ZIO_FLAG_RAW_ENCRYPT;
	}

	if (*arc_flags & ARC_FLAG_PREFETCH &&
	zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt)) {
	if (HDR_HAS_L2HDR(hdr))
	l2arc_hdr_arcstats_decrement_state(hdr);
	arc_hdr_set_flags(hdr, ARC_FLAG_PREFETCH);
	if (HDR_HAS_L2HDR(hdr))
	l2arc_hdr_arcstats_increment_state(hdr);
	}
	if (*arc_flags & ARC_FLAG_PRESCIENT_PREFETCH)
	arc_hdr_set_flags(hdr, ARC_FLAG_PRESCIENT_PREFETCH);
	if (*arc_flags & ARC_FLAG_L2CACHE)
	arc_hdr_set_flags(hdr, ARC_FLAG_L2CACHE);
	if (BP_IS_AUTHENTICATED(bp))
	arc_hdr_set_flags(hdr, ARC_FLAG_NOAUTH);
	if (BP_GET_LEVEL(bp) > 0)
	arc_hdr_set_flags(hdr, ARC_FLAG_INDIRECT);
	if (*arc_flags & ARC_FLAG_PREDICTIVE_PREFETCH)
	arc_hdr_set_flags(hdr, ARC_FLAG_PREDICTIVE_PREFETCH);
	ASSERT(!GHOST_STATE(hdr->b_l1hdr.b_state));

	acb = kmem_zalloc(sizeof (arc_callback_t), KM_SLEEP);
	acb->acb_done = done;
	acb->acb_private = private;
	acb->acb_compressed = compressed_read;
	acb->acb_encrypted = encrypted_read;
	acb->acb_noauth = noauth_read;
	acb->acb_zb = *zb;

	ASSERT3P(hdr->b_l1hdr.b_acb, ==, NULL);
	hdr->b_l1hdr.b_acb = acb;
	arc_hdr_set_flags(hdr, ARC_FLAG_IO_IN_PROGRESS);

	if (HDR_HAS_L2HDR(hdr) &&
	(vd = hdr->b_l2hdr.b_dev->l2ad_vdev) != NULL) {
	devw = hdr->b_l2hdr.b_dev->l2ad_writing;
	addr = hdr->b_l2hdr.b_daddr;
	/*
	* Lock out L2ARC device removal.
	*/
	if (vdev_is_dead(vd) \|\|
	!spa_config_tryenter(spa, SCL_L2ARC, vd, RW_READER))
	vd = NULL;
	}

	/*
	* We count both async reads and scrub IOs as asynchronous so
	* that both can be upgraded in the event of a cache hit while
	* the read IO is still in-flight.
	*/
	if (priority == ZIO_PRIORITY_ASYNC_READ \|\|
	priority == ZIO_PRIORITY_SCRUB)
	arc_hdr_set_flags(hdr, ARC_FLAG_PRIO_ASYNC_READ);
	else
	arc_hdr_clear_flags(hdr, ARC_FLAG_PRIO_ASYNC_READ);

	/*
	* At this point, we have a level 1 cache miss or a blkptr
	* with embedded data. Try again in L2ARC if possible.
	*/
	ASSERT3U(HDR_GET_LSIZE(hdr), ==, lsize);

	/*
	* Skip ARC stat bump for block pointers with embedded
	* data. The data are read from the blkptr itself via
	* decode_embedded_bp_compressed().
	*/
	if (!embedded_bp) {
	DTRACE_PROBE4(arc__miss, arc_buf_hdr_t *, hdr,
	blkptr_t *, bp, uint64_t, lsize,
	zbookmark_phys_t *, zb);
	ARCSTAT_BUMP(arcstat_misses);
	ARCSTAT_CONDSTAT(!HDR_PREFETCH(hdr),
	demand, prefetch, !HDR_ISTYPE_METADATA(hdr), data,
	metadata, misses);
	zfs_racct_read(size, 1);
	}

	/* Check if the spa even has l2 configured */
	const boolean_t spa_has_l2 = l2arc_ndev != 0 &&
	spa->spa_l2cache.sav_count > 0;

	if (vd != NULL && spa_has_l2 && !(l2arc_norw && devw)) {
	/*
	* Read from the L2ARC if the following are true:
	* 1. The L2ARC vdev was previously cached.
	* 2. This buffer still has L2ARC metadata.
	* 3. This buffer isn't currently writing to the L2ARC.
	* 4. The L2ARC entry wasn't evicted, which may
	* also have invalidated the vdev.
	* 5. This isn't prefetch or l2arc_noprefetch is 0.
	*/
	if (HDR_HAS_L2HDR(hdr) &&
	!HDR_L2_WRITING(hdr) && !HDR_L2_EVICTED(hdr) &&
	!(l2arc_noprefetch && HDR_PREFETCH(hdr))) {
	l2arc_read_callback_t *cb;
	abd_t *abd;
	uint64_t asize;

	DTRACE_PROBE1(l2arc__hit, arc_buf_hdr_t *, hdr);
	ARCSTAT_BUMP(arcstat_l2_hits);
	hdr->b_l2hdr.b_hits++;

	cb = kmem_zalloc(sizeof (l2arc_read_callback_t),
	KM_SLEEP);
	cb->l2rcb_hdr = hdr;
	cb->l2rcb_bp = *bp;
	cb->l2rcb_zb = *zb;
	cb->l2rcb_flags = zio_flags;

	/*
	* When Compressed ARC is disabled, but the
	* L2ARC block is compressed, arc_hdr_size()
	* will have returned LSIZE rather than PSIZE.
	*/
	if (HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF &&
	!HDR_COMPRESSION_ENABLED(hdr) &&
	HDR_GET_PSIZE(hdr) != 0) {
	size = HDR_GET_PSIZE(hdr);
	}

	asize = vdev_psize_to_asize(vd, size);
	if (asize != size) {
	abd = abd_alloc_for_io(asize,
	HDR_ISTYPE_METADATA(hdr));
	cb->l2rcb_abd = abd;
	} else {
	abd = hdr_abd;
	}

	ASSERT(addr >= VDEV_LABEL_START_SIZE &&
	addr + asize <= vd->vdev_psize -
	VDEV_LABEL_END_SIZE);

	/*
	* l2arc read. The SCL_L2ARC lock will be
	* released by l2arc_read_done().
	* Issue a null zio if the underlying buffer
	* was squashed to zero size by compression.
	*/
	ASSERT3U(arc_hdr_get_compress(hdr), !=,
	ZIO_COMPRESS_EMPTY);
	rzio = zio_read_phys(pio, vd, addr,
	asize, abd,
	ZIO_CHECKSUM_OFF,
	l2arc_read_done, cb, priority,
	zio_flags \| ZIO_FLAG_DONT_CACHE \|
	ZIO_FLAG_CANFAIL \|
	ZIO_FLAG_DONT_PROPAGATE \|
	ZIO_FLAG_DONT_RETRY, B_FALSE);
	acb->acb_zio_head = rzio;

	if (hash_lock != NULL)
	mutex_exit(hash_lock);

	DTRACE_PROBE2(l2arc__read, vdev_t *, vd,
	zio_t *, rzio);
	ARCSTAT_INCR(arcstat_l2_read_bytes,
	HDR_GET_PSIZE(hdr));

	if (*arc_flags & ARC_FLAG_NOWAIT) {
	zio_nowait(rzio);
	goto out;
	}

	ASSERT(*arc_flags & ARC_FLAG_WAIT);
	if (zio_wait(rzio) == 0)
	goto out;

	/* l2arc read error; goto zio_read() */
	if (hash_lock != NULL)
	mutex_enter(hash_lock);
	} else {
	DTRACE_PROBE1(l2arc__miss,
	arc_buf_hdr_t *, hdr);
	ARCSTAT_BUMP(arcstat_l2_misses);
	if (HDR_L2_WRITING(hdr))
	ARCSTAT_BUMP(arcstat_l2_rw_clash);
	spa_config_exit(spa, SCL_L2ARC, vd);
	}
	} else {
	if (vd != NULL)
	spa_config_exit(spa, SCL_L2ARC, vd);

	/*
	* Only a spa with l2 should contribute to l2
	* miss stats. (Including the case of having a
	* faulted cache device - that's also a miss.)
	*/
	if (spa_has_l2) {
	/*
	* Skip ARC stat bump for block pointers with
	* embedded data. The data are read from the
	* blkptr itself via
	* decode_embedded_bp_compressed().
	*/
	if (!embedded_bp) {
	DTRACE_PROBE1(l2arc__miss,
	arc_buf_hdr_t *, hdr);
	ARCSTAT_BUMP(arcstat_l2_misses);
	}
	}
	}

	rzio = zio_read(pio, spa, bp, hdr_abd, size,
	arc_read_done, hdr, priority, zio_flags, zb);
	acb->acb_zio_head = rzio;

	if (hash_lock != NULL)
	mutex_exit(hash_lock);

	if (*arc_flags & ARC_FLAG_WAIT) {
	rc = zio_wait(rzio);
	goto out;
	}

	ASSERT(*arc_flags & ARC_FLAG_NOWAIT);
	zio_nowait(rzio);
	}

	out:
	/* embedded bps don't actually go to disk */
	if (!embedded_bp)
	spa_read_history_add(spa, zb, *arc_flags);
	spl_fstrans_unmark(cookie);
	return (rc);
	}

	arc_prune_t *
	arc_add_prune_callback(arc_prune_func_t func, void private)
	{
	arc_prune_t *p;

	p = kmem_alloc(sizeof (*p), KM_SLEEP);
	p->p_pfunc = func;
	p->p_private = private;
	list_link_init(&p->p_node);
	zfs_refcount_create(&p->p_refcnt);

	mutex_enter(&arc_prune_mtx);
	zfs_refcount_add(&p->p_refcnt, &arc_prune_list);
	list_insert_head(&arc_prune_list, p);
	mutex_exit(&arc_prune_mtx);

	return (p);
	}

	void
	arc_remove_prune_callback(arc_prune_t *p)
	{
	boolean_t wait = B_FALSE;
	mutex_enter(&arc_prune_mtx);
	list_remove(&arc_prune_list, p);
	if (zfs_refcount_remove(&p->p_refcnt, &arc_prune_list) > 0)
	wait = B_TRUE;
	mutex_exit(&arc_prune_mtx);

	/* wait for arc_prune_task to finish */
	if (wait)
	taskq_wait_outstanding(arc_prune_taskq, 0);
	ASSERT0(zfs_refcount_count(&p->p_refcnt));
	zfs_refcount_destroy(&p->p_refcnt);
	kmem_free(p, sizeof (*p));
	}

	/*
	* Notify the arc that a block was freed, and thus will never be used again.
	*/
	void
	arc_freed(spa_t spa, const blkptr_t bp)
	{
	arc_buf_hdr_t *hdr;
	kmutex_t *hash_lock;
	uint64_t guid = spa_load_guid(spa);

	ASSERT(!BP_IS_EMBEDDED(bp));

	hdr = buf_hash_find(guid, bp, &hash_lock);
	if (hdr == NULL)
	return;

	/*
	* We might be trying to free a block that is still doing I/O
	* (i.e. prefetch) or has a reference (i.e. a dedup-ed,
	* dmu_sync-ed block). If this block is being prefetched, then it
	* would still have the ARC_FLAG_IO_IN_PROGRESS flag set on the hdr
	* until the I/O completes. A block may also have a reference if it is
	* part of a dedup-ed, dmu_synced write. The dmu_sync() function would
	* have written the new block to its final resting place on disk but
	* without the dedup flag set. This would have left the hdr in the MRU
	* state and discoverable. When the txg finally syncs it detects that
	* the block was overridden in open context and issues an override I/O.
	* Since this is a dedup block, the override I/O will determine if the
	* block is already in the DDT. If so, then it will replace the io_bp
	* with the bp from the DDT and allow the I/O to finish. When the I/O
	* reaches the done callback, dbuf_write_override_done, it will
	* check to see if the io_bp and io_bp_override are identical.
	* If they are not, then it indicates that the bp was replaced with
	* the bp in the DDT and the override bp is freed. This allows
	* us to arrive here with a reference on a block that is being
	* freed. So if we have an I/O in progress, or a reference to
	* this hdr, then we don't destroy the hdr.
	*/
	if (!HDR_HAS_L1HDR(hdr) \|\| (!HDR_IO_IN_PROGRESS(hdr) &&
	zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt))) {
	arc_change_state(arc_anon, hdr, hash_lock);
	arc_hdr_destroy(hdr);
	mutex_exit(hash_lock);
	} else {
	mutex_exit(hash_lock);
	}

	}

	/*
	* Release this buffer from the cache, making it an anonymous buffer. This
	* must be done after a read and prior to modifying the buffer contents.
	* If the buffer has more than one reference, we must make
	* a new hdr for the buffer.
	*/
	void
	arc_release(arc_buf_t buf, void tag)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;

	/*
	* It would be nice to assert that if its DMU metadata (level >
	* 0 \|\| it's the dnode file), then it must be syncing context.
	* But we don't know that information at this level.
	*/

	mutex_enter(&buf->b_evict_lock);

	ASSERT(HDR_HAS_L1HDR(hdr));

	/*
	* We don't grab the hash lock prior to this check, because if
	* the buffer's header is in the arc_anon state, it won't be
	* linked into the hash table.
	*/
	if (hdr->b_l1hdr.b_state == arc_anon) {
	mutex_exit(&buf->b_evict_lock);
	ASSERT(!HDR_IO_IN_PROGRESS(hdr));
	ASSERT(!HDR_IN_HASH_TABLE(hdr));
	ASSERT(!HDR_HAS_L2HDR(hdr));

	ASSERT3U(hdr->b_l1hdr.b_bufcnt, ==, 1);
	ASSERT3S(zfs_refcount_count(&hdr->b_l1hdr.b_refcnt), ==, 1);
	ASSERT(!list_link_active(&hdr->b_l1hdr.b_arc_node));

	hdr->b_l1hdr.b_arc_access = 0;

	/*
	* If the buf is being overridden then it may already
	* have a hdr that is not empty.
	*/
	buf_discard_identity(hdr);
	arc_buf_thaw(buf);

	return;
	}

	kmutex_t *hash_lock = HDR_LOCK(hdr);
	mutex_enter(hash_lock);

	/*
	* This assignment is only valid as long as the hash_lock is
	* held, we must be careful not to reference state or the
	* b_state field after dropping the lock.
	*/
	arc_state_t *state = hdr->b_l1hdr.b_state;
	ASSERT3P(hash_lock, ==, HDR_LOCK(hdr));
	ASSERT3P(state, !=, arc_anon);

	/* this buffer is not on any list */
	ASSERT3S(zfs_refcount_count(&hdr->b_l1hdr.b_refcnt), >, 0);

	if (HDR_HAS_L2HDR(hdr)) {
	mutex_enter(&hdr->b_l2hdr.b_dev->l2ad_mtx);

	/*
	* We have to recheck this conditional again now that
	* we're holding the l2ad_mtx to prevent a race with
	* another thread which might be concurrently calling
	* l2arc_evict(). In that case, l2arc_evict() might have
	* destroyed the header's L2 portion as we were waiting
	* to acquire the l2ad_mtx.
	*/
	if (HDR_HAS_L2HDR(hdr))
	arc_hdr_l2hdr_destroy(hdr);

	mutex_exit(&hdr->b_l2hdr.b_dev->l2ad_mtx);
	}

	/*
	* Do we have more than one buf?
	*/
	if (hdr->b_l1hdr.b_bufcnt > 1) {
	arc_buf_hdr_t *nhdr;
	uint64_t spa = hdr->b_spa;
	uint64_t psize = HDR_GET_PSIZE(hdr);
	uint64_t lsize = HDR_GET_LSIZE(hdr);
	boolean_t protected = HDR_PROTECTED(hdr);
	enum zio_compress compress = arc_hdr_get_compress(hdr);
	arc_buf_contents_t type = arc_buf_type(hdr);
	VERIFY3U(hdr->b_type, ==, type);

	ASSERT(hdr->b_l1hdr.b_buf != buf \|\| buf->b_next != NULL);
	(void) remove_reference(hdr, hash_lock, tag);

	if (arc_buf_is_shared(buf) && !ARC_BUF_COMPRESSED(buf)) {
	ASSERT3P(hdr->b_l1hdr.b_buf, !=, buf);
	ASSERT(ARC_BUF_LAST(buf));
	}

	/*
	* Pull the data off of this hdr and attach it to
	* a new anonymous hdr. Also find the last buffer
	* in the hdr's buffer list.
	*/
	arc_buf_t *lastbuf = arc_buf_remove(hdr, buf);
	ASSERT3P(lastbuf, !=, NULL);

	/*
	* If the current arc_buf_t and the hdr are sharing their data
	* buffer, then we must stop sharing that block.
	*/
	if (arc_buf_is_shared(buf)) {
	ASSERT3P(hdr->b_l1hdr.b_buf, !=, buf);
	VERIFY(!arc_buf_is_shared(lastbuf));

	/*
	* First, sever the block sharing relationship between
	* buf and the arc_buf_hdr_t.
	*/
	arc_unshare_buf(hdr, buf);

	/*
	* Now we need to recreate the hdr's b_pabd. Since we
	* have lastbuf handy, we try to share with it, but if
	* we can't then we allocate a new b_pabd and copy the
	* data from buf into it.
	*/
	if (arc_can_share(hdr, lastbuf)) {
	arc_share_buf(hdr, lastbuf);
	} else {
	arc_hdr_alloc_abd(hdr, ARC_HDR_DO_ADAPT);
	abd_copy_from_buf(hdr->b_l1hdr.b_pabd,
	buf->b_data, psize);
	}
	VERIFY3P(lastbuf->b_data, !=, NULL);
	} else if (HDR_SHARED_DATA(hdr)) {
	/*
	* Uncompressed shared buffers are always at the end
	* of the list. Compressed buffers don't have the
	* same requirements. This makes it hard to
	* simply assert that the lastbuf is shared so
	* we rely on the hdr's compression flags to determine
	* if we have a compressed, shared buffer.
	*/
	ASSERT(arc_buf_is_shared(lastbuf) \|\|
	arc_hdr_get_compress(hdr) != ZIO_COMPRESS_OFF);
	ASSERT(!ARC_BUF_SHARED(buf));
	}

	ASSERT(hdr->b_l1hdr.b_pabd != NULL \|\| HDR_HAS_RABD(hdr));
	ASSERT3P(state, !=, arc_l2c_only);

	(void) zfs_refcount_remove_many(&state->arcs_size,
	arc_buf_size(buf), buf);

	if (zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt)) {
	ASSERT3P(state, !=, arc_l2c_only);
	(void) zfs_refcount_remove_many(
	&state->arcs_esize[type],
	arc_buf_size(buf), buf);
	}

	hdr->b_l1hdr.b_bufcnt -= 1;
	if (ARC_BUF_ENCRYPTED(buf))
	hdr->b_crypt_hdr.b_ebufcnt -= 1;

	arc_cksum_verify(buf);
	arc_buf_unwatch(buf);

	/* if this is the last uncompressed buf free the checksum */
	if (!arc_hdr_has_uncompressed_buf(hdr))
	arc_cksum_free(hdr);

	mutex_exit(hash_lock);

	/*
	* Allocate a new hdr. The new hdr will contain a b_pabd
	* buffer which will be freed in arc_write().
	*/
	nhdr = arc_hdr_alloc(spa, psize, lsize, protected,
	compress, hdr->b_complevel, type);
	ASSERT3P(nhdr->b_l1hdr.b_buf, ==, NULL);
	ASSERT0(nhdr->b_l1hdr.b_bufcnt);
	ASSERT0(zfs_refcount_count(&nhdr->b_l1hdr.b_refcnt));
	VERIFY3U(nhdr->b_type, ==, type);
	ASSERT(!HDR_SHARED_DATA(nhdr));

	nhdr->b_l1hdr.b_buf = buf;
	nhdr->b_l1hdr.b_bufcnt = 1;
	if (ARC_BUF_ENCRYPTED(buf))
	nhdr->b_crypt_hdr.b_ebufcnt = 1;
	(void) zfs_refcount_add(&nhdr->b_l1hdr.b_refcnt, tag);
	buf->b_hdr = nhdr;

	mutex_exit(&buf->b_evict_lock);
	(void) zfs_refcount_add_many(&arc_anon->arcs_size,
	arc_buf_size(buf), buf);
	} else {
	mutex_exit(&buf->b_evict_lock);
	ASSERT(zfs_refcount_count(&hdr->b_l1hdr.b_refcnt) == 1);
	/* protected by hash lock, or hdr is on arc_anon */
	ASSERT(!multilist_link_active(&hdr->b_l1hdr.b_arc_node));
	ASSERT(!HDR_IO_IN_PROGRESS(hdr));
	hdr->b_l1hdr.b_mru_hits = 0;
	hdr->b_l1hdr.b_mru_ghost_hits = 0;
	hdr->b_l1hdr.b_mfu_hits = 0;
	hdr->b_l1hdr.b_mfu_ghost_hits = 0;
	arc_change_state(arc_anon, hdr, hash_lock);
	hdr->b_l1hdr.b_arc_access = 0;

	mutex_exit(hash_lock);
	buf_discard_identity(hdr);
	arc_buf_thaw(buf);
	}
	}

	int
	arc_released(arc_buf_t *buf)
	{
	int released;

	mutex_enter(&buf->b_evict_lock);
	released = (buf->b_data != NULL &&
	buf->b_hdr->b_l1hdr.b_state == arc_anon);
	mutex_exit(&buf->b_evict_lock);
	return (released);
	}

	#ifdef ZFS_DEBUG
	int
	arc_referenced(arc_buf_t *buf)
	{
	int referenced;

	mutex_enter(&buf->b_evict_lock);
	referenced = (zfs_refcount_count(&buf->b_hdr->b_l1hdr.b_refcnt));
	mutex_exit(&buf->b_evict_lock);
	return (referenced);
	}
	#endif

	static void
	arc_write_ready(zio_t *zio)
	{
	arc_write_callback_t *callback = zio->io_private;
	arc_buf_t *buf = callback->awcb_buf;
	arc_buf_hdr_t *hdr = buf->b_hdr;
	blkptr_t *bp = zio->io_bp;
	uint64_t psize = BP_IS_HOLE(bp) ? 0 : BP_GET_PSIZE(bp);
	fstrans_cookie_t cookie = spl_fstrans_mark();

	ASSERT(HDR_HAS_L1HDR(hdr));
	ASSERT(!zfs_refcount_is_zero(&buf->b_hdr->b_l1hdr.b_refcnt));
	ASSERT(hdr->b_l1hdr.b_bufcnt > 0);

	/*
	* If we're reexecuting this zio because the pool suspended, then
	* cleanup any state that was previously set the first time the
	* callback was invoked.
	*/
	if (zio->io_flags & ZIO_FLAG_REEXECUTED) {
	arc_cksum_free(hdr);
	arc_buf_unwatch(buf);
	if (hdr->b_l1hdr.b_pabd != NULL) {
	if (arc_buf_is_shared(buf)) {
	arc_unshare_buf(hdr, buf);
	} else {
	arc_hdr_free_abd(hdr, B_FALSE);
	}
	}

	if (HDR_HAS_RABD(hdr))
	arc_hdr_free_abd(hdr, B_TRUE);
	}
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);
	ASSERT(!HDR_HAS_RABD(hdr));
	ASSERT(!HDR_SHARED_DATA(hdr));
	ASSERT(!arc_buf_is_shared(buf));

	callback->awcb_ready(zio, buf, callback->awcb_private);

	if (HDR_IO_IN_PROGRESS(hdr))
	ASSERT(zio->io_flags & ZIO_FLAG_REEXECUTED);

	arc_hdr_set_flags(hdr, ARC_FLAG_IO_IN_PROGRESS);

	if (BP_IS_PROTECTED(bp) != !!HDR_PROTECTED(hdr))
	hdr = arc_hdr_realloc_crypt(hdr, BP_IS_PROTECTED(bp));

	if (BP_IS_PROTECTED(bp)) {
	/* ZIL blocks are written through zio_rewrite */
	ASSERT3U(BP_GET_TYPE(bp), !=, DMU_OT_INTENT_LOG);
	ASSERT(HDR_PROTECTED(hdr));

	if (BP_SHOULD_BYTESWAP(bp)) {
	if (BP_GET_LEVEL(bp) > 0) {
	hdr->b_l1hdr.b_byteswap = DMU_BSWAP_UINT64;
	} else {
	hdr->b_l1hdr.b_byteswap =
	DMU_OT_BYTESWAP(BP_GET_TYPE(bp));
	}
	} else {
	hdr->b_l1hdr.b_byteswap = DMU_BSWAP_NUMFUNCS;
	}

	hdr->b_crypt_hdr.b_ot = BP_GET_TYPE(bp);
	hdr->b_crypt_hdr.b_dsobj = zio->io_bookmark.zb_objset;
	zio_crypt_decode_params_bp(bp, hdr->b_crypt_hdr.b_salt,
	hdr->b_crypt_hdr.b_iv);
	zio_crypt_decode_mac_bp(bp, hdr->b_crypt_hdr.b_mac);
	}

	/*
	* If this block was written for raw encryption but the zio layer
	* ended up only authenticating it, adjust the buffer flags now.
	*/
	if (BP_IS_AUTHENTICATED(bp) && ARC_BUF_ENCRYPTED(buf)) {
	arc_hdr_set_flags(hdr, ARC_FLAG_NOAUTH);
	buf->b_flags &= ~ARC_BUF_FLAG_ENCRYPTED;
	if (BP_GET_COMPRESS(bp) == ZIO_COMPRESS_OFF)
	buf->b_flags &= ~ARC_BUF_FLAG_COMPRESSED;
	} else if (BP_IS_HOLE(bp) && ARC_BUF_ENCRYPTED(buf)) {
	buf->b_flags &= ~ARC_BUF_FLAG_ENCRYPTED;
	buf->b_flags &= ~ARC_BUF_FLAG_COMPRESSED;
	}

	/* this must be done after the buffer flags are adjusted */
	arc_cksum_compute(buf);

	enum zio_compress compress;
	if (BP_IS_HOLE(bp) \|\| BP_IS_EMBEDDED(bp)) {
	compress = ZIO_COMPRESS_OFF;
	} else {
	ASSERT3U(HDR_GET_LSIZE(hdr), ==, BP_GET_LSIZE(bp));
	compress = BP_GET_COMPRESS(bp);
	}
	HDR_SET_PSIZE(hdr, psize);
	arc_hdr_set_compress(hdr, compress);
	hdr->b_complevel = zio->io_prop.zp_complevel;

	if (zio->io_error != 0 \|\| psize == 0)
	goto out;

	/*
	* Fill the hdr with data. If the buffer is encrypted we have no choice
	* but to copy the data into b_radb. If the hdr is compressed, the data
	* we want is available from the zio, otherwise we can take it from
	* the buf.
	*
	* We might be able to share the buf's data with the hdr here. However,
	* doing so would cause the ARC to be full of linear ABDs if we write a
	* lot of shareable data. As a compromise, we check whether scattered
	* ABDs are allowed, and assume that if they are then the user wants
	* the ARC to be primarily filled with them regardless of the data being
	* written. Therefore, if they're allowed then we allocate one and copy
	* the data into it; otherwise, we share the data directly if we can.
	*/
	if (ARC_BUF_ENCRYPTED(buf)) {
	ASSERT3U(psize, >, 0);
	ASSERT(ARC_BUF_COMPRESSED(buf));
	arc_hdr_alloc_abd(hdr, ARC_HDR_DO_ADAPT \| ARC_HDR_ALLOC_RDATA \|
	ARC_HDR_USE_RESERVE);
	abd_copy(hdr->b_crypt_hdr.b_rabd, zio->io_abd, psize);
	} else if (!abd_size_alloc_linear(arc_buf_size(buf)) \|\|
	!arc_can_share(hdr, buf)) {
	/*
	* Ideally, we would always copy the io_abd into b_pabd, but the
	* user may have disabled compressed ARC, thus we must check the
	* hdr's compression setting rather than the io_bp's.
	*/
	if (BP_IS_ENCRYPTED(bp)) {
	ASSERT3U(psize, >, 0);
	arc_hdr_alloc_abd(hdr, ARC_HDR_DO_ADAPT \|
	ARC_HDR_ALLOC_RDATA \| ARC_HDR_USE_RESERVE);
	abd_copy(hdr->b_crypt_hdr.b_rabd, zio->io_abd, psize);
	} else if (arc_hdr_get_compress(hdr) != ZIO_COMPRESS_OFF &&
	!ARC_BUF_COMPRESSED(buf)) {
	ASSERT3U(psize, >, 0);
	arc_hdr_alloc_abd(hdr, ARC_HDR_DO_ADAPT \|
	ARC_HDR_USE_RESERVE);
	abd_copy(hdr->b_l1hdr.b_pabd, zio->io_abd, psize);
	} else {
	ASSERT3U(zio->io_orig_size, ==, arc_hdr_size(hdr));
	arc_hdr_alloc_abd(hdr, ARC_HDR_DO_ADAPT \|
	ARC_HDR_USE_RESERVE);
	abd_copy_from_buf(hdr->b_l1hdr.b_pabd, buf->b_data,
	arc_buf_size(buf));
	}
	} else {
	ASSERT3P(buf->b_data, ==, abd_to_buf(zio->io_orig_abd));
	ASSERT3U(zio->io_orig_size, ==, arc_buf_size(buf));
	ASSERT3U(hdr->b_l1hdr.b_bufcnt, ==, 1);

	arc_share_buf(hdr, buf);
	}

	out:
	arc_hdr_verify(hdr, bp);
	spl_fstrans_unmark(cookie);
	}

	static void
	arc_write_children_ready(zio_t *zio)
	{
	arc_write_callback_t *callback = zio->io_private;
	arc_buf_t *buf = callback->awcb_buf;

	callback->awcb_children_ready(zio, buf, callback->awcb_private);
	}

	/*
	* The SPA calls this callback for each physical write that happens on behalf
	* of a logical write. See the comment in dbuf_write_physdone() for details.
	*/
	static void
	arc_write_physdone(zio_t *zio)
	{
	arc_write_callback_t *cb = zio->io_private;
	if (cb->awcb_physdone != NULL)
	cb->awcb_physdone(zio, cb->awcb_buf, cb->awcb_private);
	}

	static void
	arc_write_done(zio_t *zio)
	{
	arc_write_callback_t *callback = zio->io_private;
	arc_buf_t *buf = callback->awcb_buf;
	arc_buf_hdr_t *hdr = buf->b_hdr;

	ASSERT3P(hdr->b_l1hdr.b_acb, ==, NULL);

	if (zio->io_error == 0) {
	arc_hdr_verify(hdr, zio->io_bp);

	if (BP_IS_HOLE(zio->io_bp) \|\| BP_IS_EMBEDDED(zio->io_bp)) {
	buf_discard_identity(hdr);
	} else {
	hdr->b_dva = *BP_IDENTITY(zio->io_bp);
	hdr->b_birth = BP_PHYSICAL_BIRTH(zio->io_bp);
	}
	} else {
	ASSERT(HDR_EMPTY(hdr));
	}

	/*
	* If the block to be written was all-zero or compressed enough to be
	* embedded in the BP, no write was performed so there will be no
	* dva/birth/checksum. The buffer must therefore remain anonymous
	* (and uncached).
	*/
	if (!HDR_EMPTY(hdr)) {
	arc_buf_hdr_t *exists;
	kmutex_t *hash_lock;

	ASSERT3U(zio->io_error, ==, 0);

	arc_cksum_verify(buf);

	exists = buf_hash_insert(hdr, &hash_lock);
	if (exists != NULL) {
	/*
	* This can only happen if we overwrite for
	* sync-to-convergence, because we remove
	* buffers from the hash table when we arc_free().
	*/
	if (zio->io_flags & ZIO_FLAG_IO_REWRITE) {
	if (!BP_EQUAL(&zio->io_bp_orig, zio->io_bp))
	panic("bad overwrite, hdr=%p exists=%p",
	(void )hdr, (void )exists);
	ASSERT(zfs_refcount_is_zero(
	&exists->b_l1hdr.b_refcnt));
	arc_change_state(arc_anon, exists, hash_lock);
	arc_hdr_destroy(exists);
	mutex_exit(hash_lock);
	exists = buf_hash_insert(hdr, &hash_lock);
	ASSERT3P(exists, ==, NULL);
	} else if (zio->io_flags & ZIO_FLAG_NOPWRITE) {
	/* nopwrite */
	ASSERT(zio->io_prop.zp_nopwrite);
	if (!BP_EQUAL(&zio->io_bp_orig, zio->io_bp))
	panic("bad nopwrite, hdr=%p exists=%p",
	(void )hdr, (void )exists);
	} else {
	/* Dedup */
	ASSERT(hdr->b_l1hdr.b_bufcnt == 1);
	ASSERT(hdr->b_l1hdr.b_state == arc_anon);
	ASSERT(BP_GET_DEDUP(zio->io_bp));
	ASSERT(BP_GET_LEVEL(zio->io_bp) == 0);
	}
	}
	arc_hdr_clear_flags(hdr, ARC_FLAG_IO_IN_PROGRESS);
	/* if it's not anon, we are doing a scrub */
	if (exists == NULL && hdr->b_l1hdr.b_state == arc_anon)
	arc_access(hdr, hash_lock);
	mutex_exit(hash_lock);
	} else {
	arc_hdr_clear_flags(hdr, ARC_FLAG_IO_IN_PROGRESS);
	}

	ASSERT(!zfs_refcount_is_zero(&hdr->b_l1hdr.b_refcnt));
	callback->awcb_done(zio, buf, callback->awcb_private);

	abd_free(zio->io_abd);
	kmem_free(callback, sizeof (arc_write_callback_t));
	}

	zio_t *
	arc_write(zio_t pio, spa_t spa, uint64_t txg,
	blkptr_t bp, arc_buf_t buf, boolean_t l2arc,
	const zio_prop_t zp, arc_write_done_func_t ready,
	arc_write_done_func_t children_ready, arc_write_done_func_t physdone,
	arc_write_done_func_t done, void private, zio_priority_t priority,
	int zio_flags, const zbookmark_phys_t *zb)
	{
	arc_buf_hdr_t *hdr = buf->b_hdr;
	arc_write_callback_t *callback;
	zio_t *zio;
	zio_prop_t localprop = *zp;

	ASSERT3P(ready, !=, NULL);
	ASSERT3P(done, !=, NULL);
	ASSERT(!HDR_IO_ERROR(hdr));
	ASSERT(!HDR_IO_IN_PROGRESS(hdr));
	ASSERT3P(hdr->b_l1hdr.b_acb, ==, NULL);
	ASSERT3U(hdr->b_l1hdr.b_bufcnt, >, 0);
	if (l2arc)
	arc_hdr_set_flags(hdr, ARC_FLAG_L2CACHE);

	if (ARC_BUF_ENCRYPTED(buf)) {
	ASSERT(ARC_BUF_COMPRESSED(buf));
	localprop.zp_encrypt = B_TRUE;
	localprop.zp_compress = HDR_GET_COMPRESS(hdr);
	localprop.zp_complevel = hdr->b_complevel;
	localprop.zp_byteorder =
	(hdr->b_l1hdr.b_byteswap == DMU_BSWAP_NUMFUNCS) ?
	ZFS_HOST_BYTEORDER : !ZFS_HOST_BYTEORDER;
	bcopy(hdr->b_crypt_hdr.b_salt, localprop.zp_salt,
	ZIO_DATA_SALT_LEN);
	bcopy(hdr->b_crypt_hdr.b_iv, localprop.zp_iv,
	ZIO_DATA_IV_LEN);
	bcopy(hdr->b_crypt_hdr.b_mac, localprop.zp_mac,
	ZIO_DATA_MAC_LEN);
	if (DMU_OT_IS_ENCRYPTED(localprop.zp_type)) {
	localprop.zp_nopwrite = B_FALSE;
	localprop.zp_copies =
	MIN(localprop.zp_copies, SPA_DVAS_PER_BP - 1);
	}
	zio_flags \|= ZIO_FLAG_RAW;
	} else if (ARC_BUF_COMPRESSED(buf)) {
	ASSERT3U(HDR_GET_LSIZE(hdr), !=, arc_buf_size(buf));
	localprop.zp_compress = HDR_GET_COMPRESS(hdr);
	localprop.zp_complevel = hdr->b_complevel;
	zio_flags \|= ZIO_FLAG_RAW_COMPRESS;
	}
	callback = kmem_zalloc(sizeof (arc_write_callback_t), KM_SLEEP);
	callback->awcb_ready = ready;
	callback->awcb_children_ready = children_ready;
	callback->awcb_physdone = physdone;
	callback->awcb_done = done;
	callback->awcb_private = private;
	callback->awcb_buf = buf;

	/*
	* The hdr's b_pabd is now stale, free it now. A new data block
	* will be allocated when the zio pipeline calls arc_write_ready().
	*/
	if (hdr->b_l1hdr.b_pabd != NULL) {
	/*
	* If the buf is currently sharing the data block with
	* the hdr then we need to break that relationship here.
	* The hdr will remain with a NULL data pointer and the
	* buf will take sole ownership of the block.
	*/
	if (arc_buf_is_shared(buf)) {
	arc_unshare_buf(hdr, buf);
	} else {
	arc_hdr_free_abd(hdr, B_FALSE);
	}
	VERIFY3P(buf->b_data, !=, NULL);
	}

	if (HDR_HAS_RABD(hdr))
	arc_hdr_free_abd(hdr, B_TRUE);

	if (!(zio_flags & ZIO_FLAG_RAW))
	arc_hdr_set_compress(hdr, ZIO_COMPRESS_OFF);

	ASSERT(!arc_buf_is_shared(buf));
	ASSERT3P(hdr->b_l1hdr.b_pabd, ==, NULL);

	zio = zio_write(pio, spa, txg, bp,
	abd_get_from_buf(buf->b_data, HDR_GET_LSIZE(hdr)),
	HDR_GET_LSIZE(hdr), arc_buf_size(buf), &localprop, arc_write_ready,
	(children_ready != NULL) ? arc_write_children_ready : NULL,
	arc_write_physdone, arc_write_done, callback,
	priority, zio_flags, zb);

	return (zio);
	}

	void
	arc_tempreserve_clear(uint64_t reserve)
	{
	atomic_add_64(&arc_tempreserve, -reserve);
	ASSERT((int64_t)arc_tempreserve >= 0);
	}

	int
	arc_tempreserve_space(spa_t *spa, uint64_t reserve, uint64_t txg)
	{
	int error;
	uint64_t anon_size;

	if (!arc_no_grow &&
	reserve > arc_c/4 &&
	reserve * 4 > (2ULL << SPA_MAXBLOCKSHIFT))
	arc_c = MIN(arc_c_max, reserve * 4);

	/*
	* Throttle when the calculated memory footprint for the TXG
	* exceeds the target ARC size.
	*/
	if (reserve > arc_c) {
	DMU_TX_STAT_BUMP(dmu_tx_memory_reserve);
	return (SET_ERROR(ERESTART));
	}

	/*
	* Don't count loaned bufs as in flight dirty data to prevent long
	* network delays from blocking transactions that are ready to be
	* assigned to a txg.
	*/

	/* assert that it has not wrapped around */
	ASSERT3S(atomic_add_64_nv(&arc_loaned_bytes, 0), >=, 0);

	anon_size = MAX((int64_t)(zfs_refcount_count(&arc_anon->arcs_size) -
	arc_loaned_bytes), 0);

	/*
	* Writes will, almost always, require additional memory allocations
	* in order to compress/encrypt/etc the data. We therefore need to
	* make sure that there is sufficient available memory for this.
	*/
	error = arc_memory_throttle(spa, reserve, txg);
	if (error != 0)
	return (error);

	/*
	* Throttle writes when the amount of dirty data in the cache
	* gets too large. We try to keep the cache less than half full
	* of dirty blocks so that our sync times don't grow too large.
	*
	* In the case of one pool being built on another pool, we want
	* to make sure we don't end up throttling the lower (backing)
	* pool when the upper pool is the majority contributor to dirty
	* data. To insure we make forward progress during throttling, we
	* also check the current pool's net dirty data and only throttle
	* if it exceeds zfs_arc_pool_dirty_percent of the anonymous dirty
	* data in the cache.
	*
	* Note: if two requests come in concurrently, we might let them
	* both succeed, when one of them should fail. Not a huge deal.
	*/
	uint64_t total_dirty = reserve + arc_tempreserve + anon_size;
	uint64_t spa_dirty_anon = spa_dirty_data(spa);
	uint64_t rarc_c = arc_warm ? arc_c : arc_c_max;
	if (total_dirty > rarc_c * zfs_arc_dirty_limit_percent / 100 &&
	anon_size > rarc_c * zfs_arc_anon_limit_percent / 100 &&
	spa_dirty_anon > anon_size * zfs_arc_pool_dirty_percent / 100) {
	#ifdef ZFS_DEBUG
	uint64_t meta_esize = zfs_refcount_count(
	&arc_anon->arcs_esize[ARC_BUFC_METADATA]);
	uint64_t data_esize =
	zfs_refcount_count(&arc_anon->arcs_esize[ARC_BUFC_DATA]);
	dprintf("failing, arc_tempreserve=%lluK anon_meta=%lluK "
	"anon_data=%lluK tempreserve=%lluK rarc_c=%lluK\n",
	(u_longlong_t)arc_tempreserve >> 10,
	(u_longlong_t)meta_esize >> 10,
	(u_longlong_t)data_esize >> 10,
	(u_longlong_t)reserve >> 10,
	(u_longlong_t)rarc_c >> 10);
	#endif
	DMU_TX_STAT_BUMP(dmu_tx_dirty_throttle);
	return (SET_ERROR(ERESTART));
	}
	atomic_add_64(&arc_tempreserve, reserve);
	return (0);
	}

	static void
	arc_kstat_update_state(arc_state_t state, kstat_named_t size,
	kstat_named_t evict_data, kstat_named_t evict_metadata)
	{
	size->value.ui64 = zfs_refcount_count(&state->arcs_size);
	evict_data->value.ui64 =
	zfs_refcount_count(&state->arcs_esize[ARC_BUFC_DATA]);
	evict_metadata->value.ui64 =
	zfs_refcount_count(&state->arcs_esize[ARC_BUFC_METADATA]);
	}

	static int
	arc_kstat_update(kstat_t *ksp, int rw)
	{
	arc_stats_t *as = ksp->ks_data;

	if (rw == KSTAT_WRITE)
	return (SET_ERROR(EACCES));

	as->arcstat_hits.value.ui64 =
	wmsum_value(&arc_sums.arcstat_hits);
	as->arcstat_misses.value.ui64 =
	wmsum_value(&arc_sums.arcstat_misses);
	as->arcstat_demand_data_hits.value.ui64 =
	wmsum_value(&arc_sums.arcstat_demand_data_hits);
	as->arcstat_demand_data_misses.value.ui64 =
	wmsum_value(&arc_sums.arcstat_demand_data_misses);
	as->arcstat_demand_metadata_hits.value.ui64 =
	wmsum_value(&arc_sums.arcstat_demand_metadata_hits);
	as->arcstat_demand_metadata_misses.value.ui64 =
	wmsum_value(&arc_sums.arcstat_demand_metadata_misses);
	as->arcstat_prefetch_data_hits.value.ui64 =
	wmsum_value(&arc_sums.arcstat_prefetch_data_hits);
	as->arcstat_prefetch_data_misses.value.ui64 =
	wmsum_value(&arc_sums.arcstat_prefetch_data_misses);
	as->arcstat_prefetch_metadata_hits.value.ui64 =
	wmsum_value(&arc_sums.arcstat_prefetch_metadata_hits);
	as->arcstat_prefetch_metadata_misses.value.ui64 =
	wmsum_value(&arc_sums.arcstat_prefetch_metadata_misses);
	as->arcstat_mru_hits.value.ui64 =
	wmsum_value(&arc_sums.arcstat_mru_hits);
	as->arcstat_mru_ghost_hits.value.ui64 =
	wmsum_value(&arc_sums.arcstat_mru_ghost_hits);
	as->arcstat_mfu_hits.value.ui64 =
	wmsum_value(&arc_sums.arcstat_mfu_hits);
	as->arcstat_mfu_ghost_hits.value.ui64 =
	wmsum_value(&arc_sums.arcstat_mfu_ghost_hits);
	as->arcstat_deleted.value.ui64 =
	wmsum_value(&arc_sums.arcstat_deleted);
	as->arcstat_mutex_miss.value.ui64 =
	wmsum_value(&arc_sums.arcstat_mutex_miss);
	as->arcstat_access_skip.value.ui64 =
	wmsum_value(&arc_sums.arcstat_access_skip);
	as->arcstat_evict_skip.value.ui64 =
	wmsum_value(&arc_sums.arcstat_evict_skip);
	as->arcstat_evict_not_enough.value.ui64 =
	wmsum_value(&arc_sums.arcstat_evict_not_enough);
	as->arcstat_evict_l2_cached.value.ui64 =
	wmsum_value(&arc_sums.arcstat_evict_l2_cached);
	as->arcstat_evict_l2_eligible.value.ui64 =
	wmsum_value(&arc_sums.arcstat_evict_l2_eligible);
	as->arcstat_evict_l2_eligible_mfu.value.ui64 =
	wmsum_value(&arc_sums.arcstat_evict_l2_eligible_mfu);
	as->arcstat_evict_l2_eligible_mru.value.ui64 =
	wmsum_value(&arc_sums.arcstat_evict_l2_eligible_mru);
	as->arcstat_evict_l2_ineligible.value.ui64 =
	wmsum_value(&arc_sums.arcstat_evict_l2_ineligible);
	as->arcstat_evict_l2_skip.value.ui64 =
	wmsum_value(&arc_sums.arcstat_evict_l2_skip);
	as->arcstat_hash_collisions.value.ui64 =
	wmsum_value(&arc_sums.arcstat_hash_collisions);
	as->arcstat_hash_chains.value.ui64 =
	wmsum_value(&arc_sums.arcstat_hash_chains);
	as->arcstat_size.value.ui64 =
	aggsum_value(&arc_sums.arcstat_size);
	as->arcstat_compressed_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_compressed_size);
	as->arcstat_uncompressed_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_uncompressed_size);
	as->arcstat_overhead_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_overhead_size);
	as->arcstat_hdr_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_hdr_size);
	as->arcstat_data_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_data_size);
	as->arcstat_metadata_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_metadata_size);
	as->arcstat_dbuf_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_dbuf_size);
	#if defined(COMPAT_FREEBSD11)
	as->arcstat_other_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_bonus_size) +
	aggsum_value(&arc_sums.arcstat_dnode_size) +
	wmsum_value(&arc_sums.arcstat_dbuf_size);
	#endif

	arc_kstat_update_state(arc_anon,
	&as->arcstat_anon_size,
	&as->arcstat_anon_evictable_data,
	&as->arcstat_anon_evictable_metadata);
	arc_kstat_update_state(arc_mru,
	&as->arcstat_mru_size,
	&as->arcstat_mru_evictable_data,
	&as->arcstat_mru_evictable_metadata);
	arc_kstat_update_state(arc_mru_ghost,
	&as->arcstat_mru_ghost_size,
	&as->arcstat_mru_ghost_evictable_data,
	&as->arcstat_mru_ghost_evictable_metadata);
	arc_kstat_update_state(arc_mfu,
	&as->arcstat_mfu_size,
	&as->arcstat_mfu_evictable_data,
	&as->arcstat_mfu_evictable_metadata);
	arc_kstat_update_state(arc_mfu_ghost,
	&as->arcstat_mfu_ghost_size,
	&as->arcstat_mfu_ghost_evictable_data,
	&as->arcstat_mfu_ghost_evictable_metadata);

	as->arcstat_dnode_size.value.ui64 =
	aggsum_value(&arc_sums.arcstat_dnode_size);
	as->arcstat_bonus_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_bonus_size);
	as->arcstat_l2_hits.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_hits);
	as->arcstat_l2_misses.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_misses);
	as->arcstat_l2_prefetch_asize.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_prefetch_asize);
	as->arcstat_l2_mru_asize.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_mru_asize);
	as->arcstat_l2_mfu_asize.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_mfu_asize);
	as->arcstat_l2_bufc_data_asize.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_bufc_data_asize);
	as->arcstat_l2_bufc_metadata_asize.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_bufc_metadata_asize);
	as->arcstat_l2_feeds.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_feeds);
	as->arcstat_l2_rw_clash.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rw_clash);
	as->arcstat_l2_read_bytes.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_read_bytes);
	as->arcstat_l2_write_bytes.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_write_bytes);
	as->arcstat_l2_writes_sent.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_writes_sent);
	as->arcstat_l2_writes_done.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_writes_done);
	as->arcstat_l2_writes_error.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_writes_error);
	as->arcstat_l2_writes_lock_retry.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_writes_lock_retry);
	as->arcstat_l2_evict_lock_retry.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_evict_lock_retry);
	as->arcstat_l2_evict_reading.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_evict_reading);
	as->arcstat_l2_evict_l1cached.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_evict_l1cached);
	as->arcstat_l2_free_on_write.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_free_on_write);
	as->arcstat_l2_abort_lowmem.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_abort_lowmem);
	as->arcstat_l2_cksum_bad.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_cksum_bad);
	as->arcstat_l2_io_error.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_io_error);
	as->arcstat_l2_lsize.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_lsize);
	as->arcstat_l2_psize.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_psize);
	as->arcstat_l2_hdr_size.value.ui64 =
	aggsum_value(&arc_sums.arcstat_l2_hdr_size);
	as->arcstat_l2_log_blk_writes.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_log_blk_writes);
	as->arcstat_l2_log_blk_asize.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_log_blk_asize);
	as->arcstat_l2_log_blk_count.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_log_blk_count);
	as->arcstat_l2_rebuild_success.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rebuild_success);
	as->arcstat_l2_rebuild_abort_unsupported.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rebuild_abort_unsupported);
	as->arcstat_l2_rebuild_abort_io_errors.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rebuild_abort_io_errors);
	as->arcstat_l2_rebuild_abort_dh_errors.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rebuild_abort_dh_errors);
	as->arcstat_l2_rebuild_abort_cksum_lb_errors.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rebuild_abort_cksum_lb_errors);
	as->arcstat_l2_rebuild_abort_lowmem.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rebuild_abort_lowmem);
	as->arcstat_l2_rebuild_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rebuild_size);
	as->arcstat_l2_rebuild_asize.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rebuild_asize);
	as->arcstat_l2_rebuild_bufs.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rebuild_bufs);
	as->arcstat_l2_rebuild_bufs_precached.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rebuild_bufs_precached);
	as->arcstat_l2_rebuild_log_blks.value.ui64 =
	wmsum_value(&arc_sums.arcstat_l2_rebuild_log_blks);
	as->arcstat_memory_throttle_count.value.ui64 =
	wmsum_value(&arc_sums.arcstat_memory_throttle_count);
	as->arcstat_memory_direct_count.value.ui64 =
	wmsum_value(&arc_sums.arcstat_memory_direct_count);
	as->arcstat_memory_indirect_count.value.ui64 =
	wmsum_value(&arc_sums.arcstat_memory_indirect_count);

	as->arcstat_memory_all_bytes.value.ui64 =
	arc_all_memory();
	as->arcstat_memory_free_bytes.value.ui64 =
	arc_free_memory();
	as->arcstat_memory_available_bytes.value.i64 =
	arc_available_memory();

	as->arcstat_prune.value.ui64 =
	wmsum_value(&arc_sums.arcstat_prune);
	as->arcstat_meta_used.value.ui64 =
	aggsum_value(&arc_sums.arcstat_meta_used);
	as->arcstat_async_upgrade_sync.value.ui64 =
	wmsum_value(&arc_sums.arcstat_async_upgrade_sync);
	as->arcstat_demand_hit_predictive_prefetch.value.ui64 =
	wmsum_value(&arc_sums.arcstat_demand_hit_predictive_prefetch);
	as->arcstat_demand_hit_prescient_prefetch.value.ui64 =
	wmsum_value(&arc_sums.arcstat_demand_hit_prescient_prefetch);
	as->arcstat_raw_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_raw_size);
	as->arcstat_cached_only_in_progress.value.ui64 =
	wmsum_value(&arc_sums.arcstat_cached_only_in_progress);
	as->arcstat_abd_chunk_waste_size.value.ui64 =
	wmsum_value(&arc_sums.arcstat_abd_chunk_waste_size);

	return (0);
	}

	/*
	* This function must return indices evenly distributed between all
	* sublists of the multilist. This is needed due to how the ARC eviction
	* code is laid out; arc_evict_state() assumes ARC buffers are evenly
	* distributed between all sublists and uses this assumption when
	* deciding which sublist to evict from and how much to evict from it.
	*/
	static unsigned int
	arc_state_multilist_index_func(multilist_t ml, void obj)
	{
	arc_buf_hdr_t *hdr = obj;

	/*
	* We rely on b_dva to generate evenly distributed index
	* numbers using buf_hash below. So, as an added precaution,
	* let's make sure we never add empty buffers to the arc lists.
	*/
	ASSERT(!HDR_EMPTY(hdr));

	/*
	* The assumption here, is the hash value for a given
	* arc_buf_hdr_t will remain constant throughout its lifetime
	* (i.e. its b_spa, b_dva, and b_birth fields don't change).
	* Thus, we don't need to store the header's sublist index
	* on insertion, as this index can be recalculated on removal.
	*
	* Also, the low order bits of the hash value are thought to be
	* distributed evenly. Otherwise, in the case that the multilist
	* has a power of two number of sublists, each sublists' usage
	* would not be evenly distributed. In this context full 64bit
	* division would be a waste of time, so limit it to 32 bits.
	*/
	return ((unsigned int)buf_hash(hdr->b_spa, &hdr->b_dva, hdr->b_birth) %
	multilist_get_num_sublists(ml));
	}

	static unsigned int
	arc_state_l2c_multilist_index_func(multilist_t ml, void obj)
	{
	panic("Header %p insert into arc_l2c_only %p", obj, ml);
	}

	#define WARN_IF_TUNING_IGNORED(tuning, value, do_warn) do { \
	if ((do_warn) && (tuning) && ((tuning) != (value))) { \
	cmn_err(CE_WARN, \
	"ignoring tunable %s (using %llu instead)", \
	(#tuning), (u_longlong_t)(value)); \
	} \
	} while (0)

	/*
	* Called during module initialization and periodically thereafter to
	* apply reasonable changes to the exposed performance tunings. Can also be
	* called explicitly by param_set_arc_*() functions when ARC tunables are
	* updated manually. Non-zero zfs_* values which differ from the currently set
	* values will be applied.
	*/
	void
	arc_tuning_update(boolean_t verbose)
	{
	uint64_t allmem = arc_all_memory();
	unsigned long limit;

	/* Valid range: 32M - <arc_c_max> */
	if ((zfs_arc_min) && (zfs_arc_min != arc_c_min) &&
	(zfs_arc_min >= 2ULL << SPA_MAXBLOCKSHIFT) &&
	(zfs_arc_min <= arc_c_max)) {
	arc_c_min = zfs_arc_min;
	arc_c = MAX(arc_c, arc_c_min);
	}
	WARN_IF_TUNING_IGNORED(zfs_arc_min, arc_c_min, verbose);

	/* Valid range: 64M - <all physical memory> */
	if ((zfs_arc_max) && (zfs_arc_max != arc_c_max) &&
	(zfs_arc_max >= MIN_ARC_MAX) && (zfs_arc_max < allmem) &&
	(zfs_arc_max > arc_c_min)) {
	arc_c_max = zfs_arc_max;
	arc_c = MIN(arc_c, arc_c_max);
	arc_p = (arc_c >> 1);
	if (arc_meta_limit > arc_c_max)
	arc_meta_limit = arc_c_max;
	if (arc_dnode_size_limit > arc_meta_limit)
	arc_dnode_size_limit = arc_meta_limit;
	}
	WARN_IF_TUNING_IGNORED(zfs_arc_max, arc_c_max, verbose);

	/* Valid range: 16M - <arc_c_max> */
	if ((zfs_arc_meta_min) && (zfs_arc_meta_min != arc_meta_min) &&
	(zfs_arc_meta_min >= 1ULL << SPA_MAXBLOCKSHIFT) &&
	(zfs_arc_meta_min <= arc_c_max)) {
	arc_meta_min = zfs_arc_meta_min;
	if (arc_meta_limit < arc_meta_min)
	arc_meta_limit = arc_meta_min;
	if (arc_dnode_size_limit < arc_meta_min)
	arc_dnode_size_limit = arc_meta_min;
	}
	WARN_IF_TUNING_IGNORED(zfs_arc_meta_min, arc_meta_min, verbose);

	/* Valid range: <arc_meta_min> - <arc_c_max> */
	limit = zfs_arc_meta_limit ? zfs_arc_meta_limit :
	MIN(zfs_arc_meta_limit_percent, 100) * arc_c_max / 100;
	if ((limit != arc_meta_limit) &&
	(limit >= arc_meta_min) &&
	(limit <= arc_c_max))
	arc_meta_limit = limit;
	WARN_IF_TUNING_IGNORED(zfs_arc_meta_limit, arc_meta_limit, verbose);

	/* Valid range: <arc_meta_min> - <arc_meta_limit> */
	limit = zfs_arc_dnode_limit ? zfs_arc_dnode_limit :
	MIN(zfs_arc_dnode_limit_percent, 100) * arc_meta_limit / 100;
	if ((limit != arc_dnode_size_limit) &&
	(limit >= arc_meta_min) &&
	(limit <= arc_meta_limit))
	arc_dnode_size_limit = limit;
	WARN_IF_TUNING_IGNORED(zfs_arc_dnode_limit, arc_dnode_size_limit,
	verbose);

	/* Valid range: 1 - N */
	if (zfs_arc_grow_retry)
	arc_grow_retry = zfs_arc_grow_retry;

	/* Valid range: 1 - N */
	if (zfs_arc_shrink_shift) {
	arc_shrink_shift = zfs_arc_shrink_shift;
	arc_no_grow_shift = MIN(arc_no_grow_shift, arc_shrink_shift -1);
	}

	/* Valid range: 1 - N */
	if (zfs_arc_p_min_shift)
	arc_p_min_shift = zfs_arc_p_min_shift;

	/* Valid range: 1 - N ms */
	if (zfs_arc_min_prefetch_ms)
	arc_min_prefetch_ms = zfs_arc_min_prefetch_ms;

	/* Valid range: 1 - N ms */
	if (zfs_arc_min_prescient_prefetch_ms) {
	arc_min_prescient_prefetch_ms =
	zfs_arc_min_prescient_prefetch_ms;
	}

	/* Valid range: 0 - 100 */
	if ((zfs_arc_lotsfree_percent >= 0) &&
	(zfs_arc_lotsfree_percent <= 100))
	arc_lotsfree_percent = zfs_arc_lotsfree_percent;
	WARN_IF_TUNING_IGNORED(zfs_arc_lotsfree_percent, arc_lotsfree_percent,
	verbose);

	/* Valid range: 0 - <all physical memory> */
	if ((zfs_arc_sys_free) && (zfs_arc_sys_free != arc_sys_free))
	arc_sys_free = MIN(MAX(zfs_arc_sys_free, 0), allmem);
	WARN_IF_TUNING_IGNORED(zfs_arc_sys_free, arc_sys_free, verbose);
	}

	static void
	arc_state_init(void)
	{
	multilist_create(&arc_mru->arcs_list[ARC_BUFC_METADATA],
	sizeof (arc_buf_hdr_t),
	offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	arc_state_multilist_index_func);
	multilist_create(&arc_mru->arcs_list[ARC_BUFC_DATA],
	sizeof (arc_buf_hdr_t),
	offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	arc_state_multilist_index_func);
	multilist_create(&arc_mru_ghost->arcs_list[ARC_BUFC_METADATA],
	sizeof (arc_buf_hdr_t),
	offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	arc_state_multilist_index_func);
	multilist_create(&arc_mru_ghost->arcs_list[ARC_BUFC_DATA],
	sizeof (arc_buf_hdr_t),
	offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	arc_state_multilist_index_func);
	multilist_create(&arc_mfu->arcs_list[ARC_BUFC_METADATA],
	sizeof (arc_buf_hdr_t),
	offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	arc_state_multilist_index_func);
	multilist_create(&arc_mfu->arcs_list[ARC_BUFC_DATA],
	sizeof (arc_buf_hdr_t),
	offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	arc_state_multilist_index_func);
	multilist_create(&arc_mfu_ghost->arcs_list[ARC_BUFC_METADATA],
	sizeof (arc_buf_hdr_t),
	offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	arc_state_multilist_index_func);
	multilist_create(&arc_mfu_ghost->arcs_list[ARC_BUFC_DATA],
	sizeof (arc_buf_hdr_t),
	offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	arc_state_multilist_index_func);
	/*
	* L2 headers should never be on the L2 state list since they don't
	* have L1 headers allocated. Special index function asserts that.
	*/
	multilist_create(&arc_l2c_only->arcs_list[ARC_BUFC_METADATA],
	sizeof (arc_buf_hdr_t),
	offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	arc_state_l2c_multilist_index_func);
	multilist_create(&arc_l2c_only->arcs_list[ARC_BUFC_DATA],
	sizeof (arc_buf_hdr_t),
	offsetof(arc_buf_hdr_t, b_l1hdr.b_arc_node),
	arc_state_l2c_multilist_index_func);

	zfs_refcount_create(&arc_anon->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_create(&arc_anon->arcs_esize[ARC_BUFC_DATA]);
	zfs_refcount_create(&arc_mru->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_create(&arc_mru->arcs_esize[ARC_BUFC_DATA]);
	zfs_refcount_create(&arc_mru_ghost->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_create(&arc_mru_ghost->arcs_esize[ARC_BUFC_DATA]);
	zfs_refcount_create(&arc_mfu->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_create(&arc_mfu->arcs_esize[ARC_BUFC_DATA]);
	zfs_refcount_create(&arc_mfu_ghost->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_create(&arc_mfu_ghost->arcs_esize[ARC_BUFC_DATA]);
	zfs_refcount_create(&arc_l2c_only->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_create(&arc_l2c_only->arcs_esize[ARC_BUFC_DATA]);

	zfs_refcount_create(&arc_anon->arcs_size);
	zfs_refcount_create(&arc_mru->arcs_size);
	zfs_refcount_create(&arc_mru_ghost->arcs_size);
	zfs_refcount_create(&arc_mfu->arcs_size);
	zfs_refcount_create(&arc_mfu_ghost->arcs_size);
	zfs_refcount_create(&arc_l2c_only->arcs_size);

	wmsum_init(&arc_sums.arcstat_hits, 0);
	wmsum_init(&arc_sums.arcstat_misses, 0);
	wmsum_init(&arc_sums.arcstat_demand_data_hits, 0);
	wmsum_init(&arc_sums.arcstat_demand_data_misses, 0);
	wmsum_init(&arc_sums.arcstat_demand_metadata_hits, 0);
	wmsum_init(&arc_sums.arcstat_demand_metadata_misses, 0);
	wmsum_init(&arc_sums.arcstat_prefetch_data_hits, 0);
	wmsum_init(&arc_sums.arcstat_prefetch_data_misses, 0);
	wmsum_init(&arc_sums.arcstat_prefetch_metadata_hits, 0);
	wmsum_init(&arc_sums.arcstat_prefetch_metadata_misses, 0);
	wmsum_init(&arc_sums.arcstat_mru_hits, 0);
	wmsum_init(&arc_sums.arcstat_mru_ghost_hits, 0);
	wmsum_init(&arc_sums.arcstat_mfu_hits, 0);
	wmsum_init(&arc_sums.arcstat_mfu_ghost_hits, 0);
	wmsum_init(&arc_sums.arcstat_deleted, 0);
	wmsum_init(&arc_sums.arcstat_mutex_miss, 0);
	wmsum_init(&arc_sums.arcstat_access_skip, 0);
	wmsum_init(&arc_sums.arcstat_evict_skip, 0);
	wmsum_init(&arc_sums.arcstat_evict_not_enough, 0);
	wmsum_init(&arc_sums.arcstat_evict_l2_cached, 0);
	wmsum_init(&arc_sums.arcstat_evict_l2_eligible, 0);
	wmsum_init(&arc_sums.arcstat_evict_l2_eligible_mfu, 0);
	wmsum_init(&arc_sums.arcstat_evict_l2_eligible_mru, 0);
	wmsum_init(&arc_sums.arcstat_evict_l2_ineligible, 0);
	wmsum_init(&arc_sums.arcstat_evict_l2_skip, 0);
	wmsum_init(&arc_sums.arcstat_hash_collisions, 0);
	wmsum_init(&arc_sums.arcstat_hash_chains, 0);
	aggsum_init(&arc_sums.arcstat_size, 0);
	wmsum_init(&arc_sums.arcstat_compressed_size, 0);
	wmsum_init(&arc_sums.arcstat_uncompressed_size, 0);
	wmsum_init(&arc_sums.arcstat_overhead_size, 0);
	wmsum_init(&arc_sums.arcstat_hdr_size, 0);
	wmsum_init(&arc_sums.arcstat_data_size, 0);
	wmsum_init(&arc_sums.arcstat_metadata_size, 0);
	wmsum_init(&arc_sums.arcstat_dbuf_size, 0);
	aggsum_init(&arc_sums.arcstat_dnode_size, 0);
	wmsum_init(&arc_sums.arcstat_bonus_size, 0);
	wmsum_init(&arc_sums.arcstat_l2_hits, 0);
	wmsum_init(&arc_sums.arcstat_l2_misses, 0);
	wmsum_init(&arc_sums.arcstat_l2_prefetch_asize, 0);
	wmsum_init(&arc_sums.arcstat_l2_mru_asize, 0);
	wmsum_init(&arc_sums.arcstat_l2_mfu_asize, 0);
	wmsum_init(&arc_sums.arcstat_l2_bufc_data_asize, 0);
	wmsum_init(&arc_sums.arcstat_l2_bufc_metadata_asize, 0);
	wmsum_init(&arc_sums.arcstat_l2_feeds, 0);
	wmsum_init(&arc_sums.arcstat_l2_rw_clash, 0);
	wmsum_init(&arc_sums.arcstat_l2_read_bytes, 0);
	wmsum_init(&arc_sums.arcstat_l2_write_bytes, 0);
	wmsum_init(&arc_sums.arcstat_l2_writes_sent, 0);
	wmsum_init(&arc_sums.arcstat_l2_writes_done, 0);
	wmsum_init(&arc_sums.arcstat_l2_writes_error, 0);
	wmsum_init(&arc_sums.arcstat_l2_writes_lock_retry, 0);
	wmsum_init(&arc_sums.arcstat_l2_evict_lock_retry, 0);
	wmsum_init(&arc_sums.arcstat_l2_evict_reading, 0);
	wmsum_init(&arc_sums.arcstat_l2_evict_l1cached, 0);
	wmsum_init(&arc_sums.arcstat_l2_free_on_write, 0);
	wmsum_init(&arc_sums.arcstat_l2_abort_lowmem, 0);
	wmsum_init(&arc_sums.arcstat_l2_cksum_bad, 0);
	wmsum_init(&arc_sums.arcstat_l2_io_error, 0);
	wmsum_init(&arc_sums.arcstat_l2_lsize, 0);
	wmsum_init(&arc_sums.arcstat_l2_psize, 0);
	aggsum_init(&arc_sums.arcstat_l2_hdr_size, 0);
	wmsum_init(&arc_sums.arcstat_l2_log_blk_writes, 0);
	wmsum_init(&arc_sums.arcstat_l2_log_blk_asize, 0);
	wmsum_init(&arc_sums.arcstat_l2_log_blk_count, 0);
	wmsum_init(&arc_sums.arcstat_l2_rebuild_success, 0);
	wmsum_init(&arc_sums.arcstat_l2_rebuild_abort_unsupported, 0);
	wmsum_init(&arc_sums.arcstat_l2_rebuild_abort_io_errors, 0);
	wmsum_init(&arc_sums.arcstat_l2_rebuild_abort_dh_errors, 0);
	wmsum_init(&arc_sums.arcstat_l2_rebuild_abort_cksum_lb_errors, 0);
	wmsum_init(&arc_sums.arcstat_l2_rebuild_abort_lowmem, 0);
	wmsum_init(&arc_sums.arcstat_l2_rebuild_size, 0);
	wmsum_init(&arc_sums.arcstat_l2_rebuild_asize, 0);
	wmsum_init(&arc_sums.arcstat_l2_rebuild_bufs, 0);
	wmsum_init(&arc_sums.arcstat_l2_rebuild_bufs_precached, 0);
	wmsum_init(&arc_sums.arcstat_l2_rebuild_log_blks, 0);
	wmsum_init(&arc_sums.arcstat_memory_throttle_count, 0);
	wmsum_init(&arc_sums.arcstat_memory_direct_count, 0);
	wmsum_init(&arc_sums.arcstat_memory_indirect_count, 0);
	wmsum_init(&arc_sums.arcstat_prune, 0);
	aggsum_init(&arc_sums.arcstat_meta_used, 0);
	wmsum_init(&arc_sums.arcstat_async_upgrade_sync, 0);
	wmsum_init(&arc_sums.arcstat_demand_hit_predictive_prefetch, 0);
	wmsum_init(&arc_sums.arcstat_demand_hit_prescient_prefetch, 0);
	wmsum_init(&arc_sums.arcstat_raw_size, 0);
	wmsum_init(&arc_sums.arcstat_cached_only_in_progress, 0);
	wmsum_init(&arc_sums.arcstat_abd_chunk_waste_size, 0);

	arc_anon->arcs_state = ARC_STATE_ANON;
	arc_mru->arcs_state = ARC_STATE_MRU;
	arc_mru_ghost->arcs_state = ARC_STATE_MRU_GHOST;
	arc_mfu->arcs_state = ARC_STATE_MFU;
	arc_mfu_ghost->arcs_state = ARC_STATE_MFU_GHOST;
	arc_l2c_only->arcs_state = ARC_STATE_L2C_ONLY;
	}

	static void
	arc_state_fini(void)
	{
	zfs_refcount_destroy(&arc_anon->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_destroy(&arc_anon->arcs_esize[ARC_BUFC_DATA]);
	zfs_refcount_destroy(&arc_mru->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_destroy(&arc_mru->arcs_esize[ARC_BUFC_DATA]);
	zfs_refcount_destroy(&arc_mru_ghost->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_destroy(&arc_mru_ghost->arcs_esize[ARC_BUFC_DATA]);
	zfs_refcount_destroy(&arc_mfu->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_destroy(&arc_mfu->arcs_esize[ARC_BUFC_DATA]);
	zfs_refcount_destroy(&arc_mfu_ghost->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_destroy(&arc_mfu_ghost->arcs_esize[ARC_BUFC_DATA]);
	zfs_refcount_destroy(&arc_l2c_only->arcs_esize[ARC_BUFC_METADATA]);
	zfs_refcount_destroy(&arc_l2c_only->arcs_esize[ARC_BUFC_DATA]);

	zfs_refcount_destroy(&arc_anon->arcs_size);
	zfs_refcount_destroy(&arc_mru->arcs_size);
	zfs_refcount_destroy(&arc_mru_ghost->arcs_size);
	zfs_refcount_destroy(&arc_mfu->arcs_size);
	zfs_refcount_destroy(&arc_mfu_ghost->arcs_size);
	zfs_refcount_destroy(&arc_l2c_only->arcs_size);

	multilist_destroy(&arc_mru->arcs_list[ARC_BUFC_METADATA]);
	multilist_destroy(&arc_mru_ghost->arcs_list[ARC_BUFC_METADATA]);
	multilist_destroy(&arc_mfu->arcs_list[ARC_BUFC_METADATA]);
	multilist_destroy(&arc_mfu_ghost->arcs_list[ARC_BUFC_METADATA]);
	multilist_destroy(&arc_mru->arcs_list[ARC_BUFC_DATA]);
	multilist_destroy(&arc_mru_ghost->arcs_list[ARC_BUFC_DATA]);
	multilist_destroy(&arc_mfu->arcs_list[ARC_BUFC_DATA]);
	multilist_destroy(&arc_mfu_ghost->arcs_list[ARC_BUFC_DATA]);
	multilist_destroy(&arc_l2c_only->arcs_list[ARC_BUFC_METADATA]);
	multilist_destroy(&arc_l2c_only->arcs_list[ARC_BUFC_DATA]);

	wmsum_fini(&arc_sums.arcstat_hits);
	wmsum_fini(&arc_sums.arcstat_misses);
	wmsum_fini(&arc_sums.arcstat_demand_data_hits);
	wmsum_fini(&arc_sums.arcstat_demand_data_misses);
	wmsum_fini(&arc_sums.arcstat_demand_metadata_hits);
	wmsum_fini(&arc_sums.arcstat_demand_metadata_misses);
	wmsum_fini(&arc_sums.arcstat_prefetch_data_hits);
	wmsum_fini(&arc_sums.arcstat_prefetch_data_misses);
	wmsum_fini(&arc_sums.arcstat_prefetch_metadata_hits);
	wmsum_fini(&arc_sums.arcstat_prefetch_metadata_misses);
	wmsum_fini(&arc_sums.arcstat_mru_hits);
	wmsum_fini(&arc_sums.arcstat_mru_ghost_hits);
	wmsum_fini(&arc_sums.arcstat_mfu_hits);
	wmsum_fini(&arc_sums.arcstat_mfu_ghost_hits);
	wmsum_fini(&arc_sums.arcstat_deleted);
	wmsum_fini(&arc_sums.arcstat_mutex_miss);
	wmsum_fini(&arc_sums.arcstat_access_skip);
	wmsum_fini(&arc_sums.arcstat_evict_skip);
	wmsum_fini(&arc_sums.arcstat_evict_not_enough);
	wmsum_fini(&arc_sums.arcstat_evict_l2_cached);
	wmsum_fini(&arc_sums.arcstat_evict_l2_eligible);
	wmsum_fini(&arc_sums.arcstat_evict_l2_eligible_mfu);
	wmsum_fini(&arc_sums.arcstat_evict_l2_eligible_mru);
	wmsum_fini(&arc_sums.arcstat_evict_l2_ineligible);
	wmsum_fini(&arc_sums.arcstat_evict_l2_skip);
	wmsum_fini(&arc_sums.arcstat_hash_collisions);
	wmsum_fini(&arc_sums.arcstat_hash_chains);
	aggsum_fini(&arc_sums.arcstat_size);
	wmsum_fini(&arc_sums.arcstat_compressed_size);
	wmsum_fini(&arc_sums.arcstat_uncompressed_size);
	wmsum_fini(&arc_sums.arcstat_overhead_size);
	wmsum_fini(&arc_sums.arcstat_hdr_size);
	wmsum_fini(&arc_sums.arcstat_data_size);
	wmsum_fini(&arc_sums.arcstat_metadata_size);
	wmsum_fini(&arc_sums.arcstat_dbuf_size);
	aggsum_fini(&arc_sums.arcstat_dnode_size);
	wmsum_fini(&arc_sums.arcstat_bonus_size);
	wmsum_fini(&arc_sums.arcstat_l2_hits);
	wmsum_fini(&arc_sums.arcstat_l2_misses);
	wmsum_fini(&arc_sums.arcstat_l2_prefetch_asize);
	wmsum_fini(&arc_sums.arcstat_l2_mru_asize);
	wmsum_fini(&arc_sums.arcstat_l2_mfu_asize);
	wmsum_fini(&arc_sums.arcstat_l2_bufc_data_asize);
	wmsum_fini(&arc_sums.arcstat_l2_bufc_metadata_asize);
	wmsum_fini(&arc_sums.arcstat_l2_feeds);
	wmsum_fini(&arc_sums.arcstat_l2_rw_clash);
	wmsum_fini(&arc_sums.arcstat_l2_read_bytes);
	wmsum_fini(&arc_sums.arcstat_l2_write_bytes);
	wmsum_fini(&arc_sums.arcstat_l2_writes_sent);
	wmsum_fini(&arc_sums.arcstat_l2_writes_done);
	wmsum_fini(&arc_sums.arcstat_l2_writes_error);
	wmsum_fini(&arc_sums.arcstat_l2_writes_lock_retry);
	wmsum_fini(&arc_sums.arcstat_l2_evict_lock_retry);
	wmsum_fini(&arc_sums.arcstat_l2_evict_reading);
	wmsum_fini(&arc_sums.arcstat_l2_evict_l1cached);
	wmsum_fini(&arc_sums.arcstat_l2_free_on_write);
	wmsum_fini(&arc_sums.arcstat_l2_abort_lowmem);
	wmsum_fini(&arc_sums.arcstat_l2_cksum_bad);
	wmsum_fini(&arc_sums.arcstat_l2_io_error);
	wmsum_fini(&arc_sums.arcstat_l2_lsize);
	wmsum_fini(&arc_sums.arcstat_l2_psize);
	aggsum_fini(&arc_sums.arcstat_l2_hdr_size);
	wmsum_fini(&arc_sums.arcstat_l2_log_blk_writes);
	wmsum_fini(&arc_sums.arcstat_l2_log_blk_asize);
	wmsum_fini(&arc_sums.arcstat_l2_log_blk_count);
	wmsum_fini(&arc_sums.arcstat_l2_rebuild_success);
	wmsum_fini(&arc_sums.arcstat_l2_rebuild_abort_unsupported);
	wmsum_fini(&arc_sums.arcstat_l2_rebuild_abort_io_errors);
	wmsum_fini(&arc_sums.arcstat_l2_rebuild_abort_dh_errors);
	wmsum_fini(&arc_sums.arcstat_l2_rebuild_abort_cksum_lb_errors);
	wmsum_fini(&arc_sums.arcstat_l2_rebuild_abort_lowmem);
	wmsum_fini(&arc_sums.arcstat_l2_rebuild_size);
	wmsum_fini(&arc_sums.arcstat_l2_rebuild_asize);
	wmsum_fini(&arc_sums.arcstat_l2_rebuild_bufs);
	wmsum_fini(&arc_sums.arcstat_l2_rebuild_bufs_precached);
	wmsum_fini(&arc_sums.arcstat_l2_rebuild_log_blks);
	wmsum_fini(&arc_sums.arcstat_memory_throttle_count);
	wmsum_fini(&arc_sums.arcstat_memory_direct_count);
	wmsum_fini(&arc_sums.arcstat_memory_indirect_count);
	wmsum_fini(&arc_sums.arcstat_prune);
	aggsum_fini(&arc_sums.arcstat_meta_used);
	wmsum_fini(&arc_sums.arcstat_async_upgrade_sync);
	wmsum_fini(&arc_sums.arcstat_demand_hit_predictive_prefetch);
	wmsum_fini(&arc_sums.arcstat_demand_hit_prescient_prefetch);
	wmsum_fini(&arc_sums.arcstat_raw_size);
	wmsum_fini(&arc_sums.arcstat_cached_only_in_progress);
	wmsum_fini(&arc_sums.arcstat_abd_chunk_waste_size);
	}

	uint64_t
	arc_target_bytes(void)
	{
	return (arc_c);
	}

	void
	arc_set_limits(uint64_t allmem)
	{
	/* Set min cache to 1/32 of all memory, or 32MB, whichever is more. */
	arc_c_min = MAX(allmem / 32, 2ULL << SPA_MAXBLOCKSHIFT);

	/* How to set default max varies by platform. */
	arc_c_max = arc_default_max(arc_c_min, allmem);
	}
	void
	arc_init(void)
	{
	uint64_t percent, allmem = arc_all_memory();
	mutex_init(&arc_evict_lock, NULL, MUTEX_DEFAULT, NULL);
	list_create(&arc_evict_waiters, sizeof (arc_evict_waiter_t),
	offsetof(arc_evict_waiter_t, aew_node));

	arc_min_prefetch_ms = 1000;
	arc_min_prescient_prefetch_ms = 6000;

	#if defined(_KERNEL)
	arc_lowmem_init();
	#endif

	arc_set_limits(allmem);

	#ifdef _KERNEL
	/*
	* If zfs_arc_max is non-zero at init, meaning it was set in the kernel
	* environment before the module was loaded, don't block setting the
	* maximum because it is less than arc_c_min, instead, reset arc_c_min
	* to a lower value.
	* zfs_arc_min will be handled by arc_tuning_update().
	*/
	if (zfs_arc_max != 0 && zfs_arc_max >= MIN_ARC_MAX &&
	zfs_arc_max < allmem) {
	arc_c_max = zfs_arc_max;
	if (arc_c_min >= arc_c_max) {
	arc_c_min = MAX(zfs_arc_max / 2,
	2ULL << SPA_MAXBLOCKSHIFT);
	}
	}
	#else
	/*
	* In userland, there's only the memory pressure that we artificially
	* create (see arc_available_memory()). Don't let arc_c get too
	* small, because it can cause transactions to be larger than
	* arc_c, causing arc_tempreserve_space() to fail.
	*/
	arc_c_min = MAX(arc_c_max / 2, 2ULL << SPA_MAXBLOCKSHIFT);
	#endif

	arc_c = arc_c_min;
	arc_p = (arc_c >> 1);

	/* Set min to 1/2 of arc_c_min */
	arc_meta_min = 1ULL << SPA_MAXBLOCKSHIFT;
	/*
	* Set arc_meta_limit to a percent of arc_c_max with a floor of
	* arc_meta_min, and a ceiling of arc_c_max.
	*/
	percent = MIN(zfs_arc_meta_limit_percent, 100);
	arc_meta_limit = MAX(arc_meta_min, (percent * arc_c_max) / 100);
	percent = MIN(zfs_arc_dnode_limit_percent, 100);
	arc_dnode_size_limit = (percent * arc_meta_limit) / 100;

	/* Apply user specified tunings */
	arc_tuning_update(B_TRUE);

	/* if kmem_flags are set, lets try to use less memory */
	if (kmem_debugging())
	arc_c = arc_c / 2;
	if (arc_c < arc_c_min)
	arc_c = arc_c_min;

	arc_register_hotplug();

	arc_state_init();

	buf_init();

	list_create(&arc_prune_list, sizeof (arc_prune_t),
	offsetof(arc_prune_t, p_node));
	mutex_init(&arc_prune_mtx, NULL, MUTEX_DEFAULT, NULL);

	arc_prune_taskq = taskq_create("arc_prune", 100, defclsyspri,
	boot_ncpus, INT_MAX, TASKQ_PREPOPULATE \| TASKQ_DYNAMIC \|
	TASKQ_THREADS_CPU_PCT);

	arc_ksp = kstat_create("zfs", 0, "arcstats", "misc", KSTAT_TYPE_NAMED,
	sizeof (arc_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL);

	if (arc_ksp != NULL) {
	arc_ksp->ks_data = &arc_stats;
	arc_ksp->ks_update = arc_kstat_update;
	kstat_install(arc_ksp);
	}

	arc_evict_zthr = zthr_create("arc_evict",
	arc_evict_cb_check, arc_evict_cb, NULL, defclsyspri);
	arc_reap_zthr = zthr_create_timer("arc_reap",
	arc_reap_cb_check, arc_reap_cb, NULL, SEC2NSEC(1), minclsyspri);

	arc_warm = B_FALSE;

	/*
	* Calculate maximum amount of dirty data per pool.
	*
	* If it has been set by a module parameter, take that.
	* Otherwise, use a percentage of physical memory defined by
	* zfs_dirty_data_max_percent (default 10%) with a cap at
	* zfs_dirty_data_max_max (default 4G or 25% of physical memory).
	*/
	#ifdef __LP64__
	if (zfs_dirty_data_max_max == 0)
	zfs_dirty_data_max_max = MIN(4ULL * 1024 * 1024 * 1024,
	allmem * zfs_dirty_data_max_max_percent / 100);
	#else
	if (zfs_dirty_data_max_max == 0)
	zfs_dirty_data_max_max = MIN(1ULL * 1024 * 1024 * 1024,
	allmem * zfs_dirty_data_max_max_percent / 100);
	#endif

	if (zfs_dirty_data_max == 0) {
	zfs_dirty_data_max = allmem *
	zfs_dirty_data_max_percent / 100;
	zfs_dirty_data_max = MIN(zfs_dirty_data_max,
	zfs_dirty_data_max_max);
	}

	if (zfs_wrlog_data_max == 0) {

	/*
	* dp_wrlog_total is reduced for each txg at the end of
	* spa_sync(). However, dp_dirty_total is reduced every time
	* a block is written out. Thus under normal operation,
	* dp_wrlog_total could grow 2 times as big as
	* zfs_dirty_data_max.
	*/
	zfs_wrlog_data_max = zfs_dirty_data_max * 2;
	}
	}

	void
	arc_fini(void)
	{
	arc_prune_t *p;

	#ifdef _KERNEL
	arc_lowmem_fini();
	#endif /* _KERNEL */

	/* Use B_TRUE to ensure all buffers are evicted */
	arc_flush(NULL, B_TRUE);

	if (arc_ksp != NULL) {
	kstat_delete(arc_ksp);
	arc_ksp = NULL;
	}

	taskq_wait(arc_prune_taskq);
	taskq_destroy(arc_prune_taskq);

	mutex_enter(&arc_prune_mtx);
	while ((p = list_head(&arc_prune_list)) != NULL) {
	list_remove(&arc_prune_list, p);
	zfs_refcount_remove(&p->p_refcnt, &arc_prune_list);
	zfs_refcount_destroy(&p->p_refcnt);
	kmem_free(p, sizeof (*p));
	}
	mutex_exit(&arc_prune_mtx);

	list_destroy(&arc_prune_list);
	mutex_destroy(&arc_prune_mtx);

	(void) zthr_cancel(arc_evict_zthr);
	(void) zthr_cancel(arc_reap_zthr);

	mutex_destroy(&arc_evict_lock);
	list_destroy(&arc_evict_waiters);

	/*
	* Free any buffers that were tagged for destruction. This needs
	* to occur before arc_state_fini() runs and destroys the aggsum
	* values which are updated when freeing scatter ABDs.
	*/
	l2arc_do_free_on_write();

	/*
	* buf_fini() must proceed arc_state_fini() because buf_fin() may
	* trigger the release of kmem magazines, which can callback to
	* arc_space_return() which accesses aggsums freed in act_state_fini().
	*/
	buf_fini();
	arc_state_fini();

	arc_unregister_hotplug();

	/*
	* We destroy the zthrs after all the ARC state has been
	* torn down to avoid the case of them receiving any
	* wakeup() signals after they are destroyed.
	*/
	zthr_destroy(arc_evict_zthr);
	zthr_destroy(arc_reap_zthr);

	ASSERT0(arc_loaned_bytes);
	}

	/*
	* Level 2 ARC
	*
	* The level 2 ARC (L2ARC) is a cache layer in-between main memory and disk.
	* It uses dedicated storage devices to hold cached data, which are populated
	* using large infrequent writes. The main role of this cache is to boost
	* the performance of random read workloads. The intended L2ARC devices
	* include short-stroked disks, solid state disks, and other media with
	* substantially faster read latency than disk.
	*
	* +-----------------------+
	* \| ARC \|
	* +-----------------------+
	* \| ^ ^
	* \| \| \|
	* l2arc_feed_thread() arc_read()
	* \| \| \|
	* \| l2arc read \|
	* V \| \|
	* +---------------+ \|
	* \| L2ARC \| \|
	* +---------------+ \|
	* \| ^ \|
	* l2arc_write() \| \|
	* \| \| \|
	* V \| \|
	* +-------+ +-------+
	* \| vdev \| \| vdev \|
	* \| cache \| \| cache \|
	* +-------+ +-------+
	* +=========+ .-----.
	* : L2ARC : \|-_____-\|
	* : devices : \| Disks \|
	* +=========+ `-_____-'
	*
	* Read requests are satisfied from the following sources, in order:
	*
	* 1) ARC
	* 2) vdev cache of L2ARC devices
	* 3) L2ARC devices
	* 4) vdev cache of disks
	* 5) disks
	*
	* Some L2ARC device types exhibit extremely slow write performance.
	* To accommodate for this there are some significant differences between
	* the L2ARC and traditional cache design:
	*
	* 1. There is no eviction path from the ARC to the L2ARC. Evictions from
	* the ARC behave as usual, freeing buffers and placing headers on ghost
	* lists. The ARC does not send buffers to the L2ARC during eviction as
	* this would add inflated write latencies for all ARC memory pressure.
	*
	* 2. The L2ARC attempts to cache data from the ARC before it is evicted.
	* It does this by periodically scanning buffers from the eviction-end of
	* the MFU and MRU ARC lists, copying them to the L2ARC devices if they are
	* not already there. It scans until a headroom of buffers is satisfied,
	* which itself is a buffer for ARC eviction. If a compressible buffer is
	* found during scanning and selected for writing to an L2ARC device, we
	* temporarily boost scanning headroom during the next scan cycle to make
	* sure we adapt to compression effects (which might significantly reduce
	* the data volume we write to L2ARC). The thread that does this is
	* l2arc_feed_thread(), illustrated below; example sizes are included to
	* provide a better sense of ratio than this diagram:
	*
	* head --> tail
	* +---------------------+----------+
	* ARC_mfu \|:::::#:::::::::::::::\|o#o###o###\|-->. # already on L2ARC
	* +---------------------+----------+ \| o L2ARC eligible
	* ARC_mru \|:#:::::::::::::::::::\|#o#ooo####\|-->\| : ARC buffer
	* +---------------------+----------+ \|
	* 15.9 Gbytes ^ 32 Mbytes \|
	* headroom \|
	* l2arc_feed_thread()
	* \|
	* l2arc write hand <--[oooo]--'
	* \| 8 Mbyte
	* \| write max
	* V
	* +==============================+
	* L2ARC dev \|####\|#\|###\|###\| \|####\| ... \|
	* +==============================+
	* 32 Gbytes
	*
	* 3. If an ARC buffer is copied to the L2ARC but then hit instead of
	* evicted, then the L2ARC has cached a buffer much sooner than it probably
	* needed to, potentially wasting L2ARC device bandwidth and storage. It is
	* safe to say that this is an uncommon case, since buffers at the end of
	* the ARC lists have moved there due to inactivity.
	*
	* 4. If the ARC evicts faster than the L2ARC can maintain a headroom,
	* then the L2ARC simply misses copying some buffers. This serves as a
	* pressure valve to prevent heavy read workloads from both stalling the ARC
	* with waits and clogging the L2ARC with writes. This also helps prevent
	* the potential for the L2ARC to churn if it attempts to cache content too
	* quickly, such as during backups of the entire pool.
	*
	* 5. After system boot and before the ARC has filled main memory, there are
	* no evictions from the ARC and so the tails of the ARC_mfu and ARC_mru
	* lists can remain mostly static. Instead of searching from tail of these
	* lists as pictured, the l2arc_feed_thread() will search from the list heads
	* for eligible buffers, greatly increasing its chance of finding them.
	*
	* The L2ARC device write speed is also boosted during this time so that
	* the L2ARC warms up faster. Since there have been no ARC evictions yet,
	* there are no L2ARC reads, and no fear of degrading read performance
	* through increased writes.
	*
	* 6. Writes to the L2ARC devices are grouped and sent in-sequence, so that
	* the vdev queue can aggregate them into larger and fewer writes. Each
	* device is written to in a rotor fashion, sweeping writes through
	* available space then repeating.
	*
	* 7. The L2ARC does not store dirty content. It never needs to flush
	* write buffers back to disk based storage.
	*
	* 8. If an ARC buffer is written (and dirtied) which also exists in the
	* L2ARC, the now stale L2ARC buffer is immediately dropped.
	*
	* The performance of the L2ARC can be tweaked by a number of tunables, which
	* may be necessary for different workloads:
	*
	* l2arc_write_max max write bytes per interval
	* l2arc_write_boost extra write bytes during device warmup
	* l2arc_noprefetch skip caching prefetched buffers
	* l2arc_headroom number of max device writes to precache
	* l2arc_headroom_boost when we find compressed buffers during ARC
	* scanning, we multiply headroom by this
	* percentage factor for the next scan cycle,
	* since more compressed buffers are likely to
	* be present
	* l2arc_feed_secs seconds between L2ARC writing
	*
	* Tunables may be removed or added as future performance improvements are
	* integrated, and also may become zpool properties.
	*
	* There are three key functions that control how the L2ARC warms up:
	*
	* l2arc_write_eligible() check if a buffer is eligible to cache
	* l2arc_write_size() calculate how much to write
	* l2arc_write_interval() calculate sleep delay between writes
	*
	* These three functions determine what to write, how much, and how quickly
	* to send writes.
	*
	* L2ARC persistence:
	*
	* When writing buffers to L2ARC, we periodically add some metadata to
	* make sure we can pick them up after reboot, thus dramatically reducing
	* the impact that any downtime has on the performance of storage systems
	* with large caches.
	*
	* The implementation works fairly simply by integrating the following two
	* modifications:
	*
	* *) When writing to the L2ARC, we occasionally write a "l2arc log block",
	* which is an additional piece of metadata which describes what's been
	* written. This allows us to rebuild the arc_buf_hdr_t structures of the
	* main ARC buffers. There are 2 linked-lists of log blocks headed by
	* dh_start_lbps[2]. We alternate which chain we append to, so they are
	* time-wise and offset-wise interleaved, but that is an optimization rather
	* than for correctness. The log block also includes a pointer to the
	* previous block in its chain.
	*
	* *) We reserve SPA_MINBLOCKSIZE of space at the start of each L2ARC device
	* for our header bookkeeping purposes. This contains a device header,
	* which contains our top-level reference structures. We update it each
	* time we write a new log block, so that we're able to locate it in the
	* L2ARC device. If this write results in an inconsistent device header
	* (e.g. due to power failure), we detect this by verifying the header's
	* checksum and simply fail to reconstruct the L2ARC after reboot.
	*
	* Implementation diagram:
	*
	* +=== L2ARC device (not to scale) ======================================+
	* \| ___two newest log block pointers__.__________ \|
	* \| / \dh_start_lbps[1] \|
	* \| / \ \dh_start_lbps[0]\|
	* \|.___/__. V V \|
	* \|\|L2 dev\|....\|lb \|bufs \|lb \|bufs \|lb \|bufs \|lb \|bufs \|lb \|---(empty)---\|
	* \|\| hdr\| ^ /^ /^ / / \|
	* \|+------+ ...--\-------/ \-----/--\------/ / \|
	* \| \--------------/ \--------------/ \|
	* +======================================================================+
	*
	* As can be seen on the diagram, rather than using a simple linked list,
	* we use a pair of linked lists with alternating elements. This is a
	* performance enhancement due to the fact that we only find out the
	* address of the next log block access once the current block has been
	* completely read in. Obviously, this hurts performance, because we'd be
	* keeping the device's I/O queue at only a 1 operation deep, thus
	* incurring a large amount of I/O round-trip latency. Having two lists
	* allows us to fetch two log blocks ahead of where we are currently
	* rebuilding L2ARC buffers.
	*
	* On-device data structures:
	*
	* L2ARC device header: l2arc_dev_hdr_phys_t
	* L2ARC log block: l2arc_log_blk_phys_t
	*
	* L2ARC reconstruction:
	*
	* When writing data, we simply write in the standard rotary fashion,
	* evicting buffers as we go and simply writing new data over them (writing
	* a new log block every now and then). This obviously means that once we
	* loop around the end of the device, we will start cutting into an already
	* committed log block (and its referenced data buffers), like so:
	*
	* current write head__ __old tail
	* \ /
	* V V
	* <--\|bufs \|lb \|bufs \|lb \| \|bufs \|lb \|bufs \|lb \|-->
	* ^ ^^^^^^^^^___________________________________
	* \| \
	* <<nextwrite>> may overwrite this blk and/or its bufs --'
	*
	* When importing the pool, we detect this situation and use it to stop
	* our scanning process (see l2arc_rebuild).
	*
	* There is one significant caveat to consider when rebuilding ARC contents
	* from an L2ARC device: what about invalidated buffers? Given the above
	* construction, we cannot update blocks which we've already written to amend
	* them to remove buffers which were invalidated. Thus, during reconstruction,
	* we might be populating the cache with buffers for data that's not on the
	* main pool anymore, or may have been overwritten!
	*
	* As it turns out, this isn't a problem. Every arc_read request includes
	* both the DVA and, crucially, the birth TXG of the BP the caller is
	* looking for. So even if the cache were populated by completely rotten
	* blocks for data that had been long deleted and/or overwritten, we'll
	* never actually return bad data from the cache, since the DVA with the
	* birth TXG uniquely identify a block in space and time - once created,
	* a block is immutable on disk. The worst thing we have done is wasted
	* some time and memory at l2arc rebuild to reconstruct outdated ARC
	* entries that will get dropped from the l2arc as it is being updated
	* with new blocks.
	*
	* L2ARC buffers that have been evicted by l2arc_evict() ahead of the write
	* hand are not restored. This is done by saving the offset (in bytes)
	* l2arc_evict() has evicted to in the L2ARC device header and taking it
	* into account when restoring buffers.
	*/

	static boolean_t
	l2arc_write_eligible(uint64_t spa_guid, arc_buf_hdr_t *hdr)
	{
	/*
	* A buffer is not eligible for the L2ARC if it:
	* 1. belongs to a different spa.
	* 2. is already cached on the L2ARC.
	* 3. has an I/O in progress (it may be an incomplete read).
	* 4. is flagged not eligible (zfs property).
	*/
	if (hdr->b_spa != spa_guid \|\| HDR_HAS_L2HDR(hdr) \|\|
	HDR_IO_IN_PROGRESS(hdr) \|\| !HDR_L2CACHE(hdr))
	return (B_FALSE);

	return (B_TRUE);
	}

	static uint64_t
	l2arc_write_size(l2arc_dev_t *dev)
	{
	uint64_t size, dev_size, tsize;

	/*
	* Make sure our globals have meaningful values in case the user
	* altered them.
	*/
	size = l2arc_write_max;
	if (size == 0) {
	cmn_err(CE_NOTE, "Bad value for l2arc_write_max, value must "
	"be greater than zero, resetting it to the default (%d)",
	L2ARC_WRITE_SIZE);
	size = l2arc_write_max = L2ARC_WRITE_SIZE;
	}

	if (arc_warm == B_FALSE)
	size += l2arc_write_boost;

	/*
	* Make sure the write size does not exceed the size of the cache
	* device. This is important in l2arc_evict(), otherwise infinite
	* iteration can occur.
	*/
	dev_size = dev->l2ad_end - dev->l2ad_start;
	tsize = size + l2arc_log_blk_overhead(size, dev);
	if (dev->l2ad_vdev->vdev_has_trim && l2arc_trim_ahead > 0)
	tsize += MAX(64 * 1024 * 1024,
	(tsize * l2arc_trim_ahead) / 100);

	if (tsize >= dev_size) {
	cmn_err(CE_NOTE, "l2arc_write_max or l2arc_write_boost "
	"plus the overhead of log blocks (persistent L2ARC, "
	"%llu bytes) exceeds the size of the cache device "
	"(guid %llu), resetting them to the default (%d)",
	(u_longlong_t)l2arc_log_blk_overhead(size, dev),
	(u_longlong_t)dev->l2ad_vdev->vdev_guid, L2ARC_WRITE_SIZE);
	size = l2arc_write_max = l2arc_write_boost = L2ARC_WRITE_SIZE;

	if (arc_warm == B_FALSE)
	size += l2arc_write_boost;
	}

	return (size);

	}

	static clock_t
	l2arc_write_interval(clock_t began, uint64_t wanted, uint64_t wrote)
	{
	clock_t interval, next, now;

	/*
	* If the ARC lists are busy, increase our write rate; if the
	* lists are stale, idle back. This is achieved by checking
	* how much we previously wrote - if it was more than half of
	* what we wanted, schedule the next write much sooner.
	*/
	if (l2arc_feed_again && wrote > (wanted / 2))
	interval = (hz * l2arc_feed_min_ms) / 1000;
	else
	interval = hz * l2arc_feed_secs;

	now = ddi_get_lbolt();
	next = MAX(now, MIN(now + interval, began + interval));

	return (next);
	}

	/*
	* Cycle through L2ARC devices. This is how L2ARC load balances.
	* If a device is returned, this also returns holding the spa config lock.
	*/
	static l2arc_dev_t *
	l2arc_dev_get_next(void)
	{
	l2arc_dev_t first, next = NULL;

	/*
	* Lock out the removal of spas (spa_namespace_lock), then removal
	* of cache devices (l2arc_dev_mtx). Once a device has been selected,
	* both locks will be dropped and a spa config lock held instead.
	*/
	mutex_enter(&spa_namespace_lock);
	mutex_enter(&l2arc_dev_mtx);

	/* if there are no vdevs, there is nothing to do */
	if (l2arc_ndev == 0)
	goto out;

	first = NULL;
	next = l2arc_dev_last;
	do {
	/* loop around the list looking for a non-faulted vdev */
	if (next == NULL) {
	next = list_head(l2arc_dev_list);
	} else {
	next = list_next(l2arc_dev_list, next);
	if (next == NULL)
	next = list_head(l2arc_dev_list);
	}

	/* if we have come back to the start, bail out */
	if (first == NULL)
	first = next;
	else if (next == first)
	break;

	} while (vdev_is_dead(next->l2ad_vdev) \|\| next->l2ad_rebuild \|\|
	next->l2ad_trim_all);

	/* if we were unable to find any usable vdevs, return NULL */
	if (vdev_is_dead(next->l2ad_vdev) \|\| next->l2ad_rebuild \|\|
	next->l2ad_trim_all)
	next = NULL;

	l2arc_dev_last = next;

	out:
	mutex_exit(&l2arc_dev_mtx);

	/*
	* Grab the config lock to prevent the 'next' device from being
	* removed while we are writing to it.
	*/
	if (next != NULL)
	spa_config_enter(next->l2ad_spa, SCL_L2ARC, next, RW_READER);
	mutex_exit(&spa_namespace_lock);

	return (next);
	}

	/*
	* Free buffers that were tagged for destruction.
	*/
	static void
	l2arc_do_free_on_write(void)
	{
	list_t *buflist;
	l2arc_data_free_t df, df_prev;

	mutex_enter(&l2arc_free_on_write_mtx);
	buflist = l2arc_free_on_write;

	for (df = list_tail(buflist); df; df = df_prev) {
	df_prev = list_prev(buflist, df);
	ASSERT3P(df->l2df_abd, !=, NULL);
	abd_free(df->l2df_abd);
	list_remove(buflist, df);
	kmem_free(df, sizeof (l2arc_data_free_t));
	}

	mutex_exit(&l2arc_free_on_write_mtx);
	}

	/*
	* A write to a cache device has completed. Update all headers to allow
	* reads from these buffers to begin.
	*/
	static void
	l2arc_write_done(zio_t *zio)
	{
	l2arc_write_callback_t *cb;
	l2arc_lb_abd_buf_t *abd_buf;
	l2arc_lb_ptr_buf_t *lb_ptr_buf;
	l2arc_dev_t *dev;
	l2arc_dev_hdr_phys_t *l2dhdr;
	list_t *buflist;
	arc_buf_hdr_t head, hdr, *hdr_prev;
	kmutex_t *hash_lock;
	int64_t bytes_dropped = 0;

	cb = zio->io_private;
	ASSERT3P(cb, !=, NULL);
	dev = cb->l2wcb_dev;
	l2dhdr = dev->l2ad_dev_hdr;
	ASSERT3P(dev, !=, NULL);
	head = cb->l2wcb_head;
	ASSERT3P(head, !=, NULL);
	buflist = &dev->l2ad_buflist;
	ASSERT3P(buflist, !=, NULL);
	DTRACE_PROBE2(l2arc__iodone, zio_t *, zio,
	l2arc_write_callback_t *, cb);

	/*
	* All writes completed, or an error was hit.
	*/
	top:
	mutex_enter(&dev->l2ad_mtx);
	for (hdr = list_prev(buflist, head); hdr; hdr = hdr_prev) {
	hdr_prev = list_prev(buflist, hdr);

	hash_lock = HDR_LOCK(hdr);

	/*
	* We cannot use mutex_enter or else we can deadlock
	* with l2arc_write_buffers (due to swapping the order
	* the hash lock and l2ad_mtx are taken).
	*/
	if (!mutex_tryenter(hash_lock)) {
	/*
	* Missed the hash lock. We must retry so we
	* don't leave the ARC_FLAG_L2_WRITING bit set.
	*/
	ARCSTAT_BUMP(arcstat_l2_writes_lock_retry);

	/*
	* We don't want to rescan the headers we've
	* already marked as having been written out, so
	* we reinsert the head node so we can pick up
	* where we left off.
	*/
	list_remove(buflist, head);
	list_insert_after(buflist, hdr, head);

	mutex_exit(&dev->l2ad_mtx);

	/*
	* We wait for the hash lock to become available
	* to try and prevent busy waiting, and increase
	* the chance we'll be able to acquire the lock
	* the next time around.
	*/
	mutex_enter(hash_lock);
	mutex_exit(hash_lock);
	goto top;
	}

	/*
	* We could not have been moved into the arc_l2c_only
	* state while in-flight due to our ARC_FLAG_L2_WRITING
	* bit being set. Let's just ensure that's being enforced.
	*/
	ASSERT(HDR_HAS_L1HDR(hdr));

	/*
	* Skipped - drop L2ARC entry and mark the header as no
	* longer L2 eligibile.
	*/
	if (zio->io_error != 0) {
	/*
	* Error - drop L2ARC entry.
	*/
	list_remove(buflist, hdr);
	arc_hdr_clear_flags(hdr, ARC_FLAG_HAS_L2HDR);

	uint64_t psize = HDR_GET_PSIZE(hdr);
	l2arc_hdr_arcstats_decrement(hdr);

	bytes_dropped +=
	vdev_psize_to_asize(dev->l2ad_vdev, psize);
	(void) zfs_refcount_remove_many(&dev->l2ad_alloc,
	arc_hdr_size(hdr), hdr);
	}

	/*
	* Allow ARC to begin reads and ghost list evictions to
	* this L2ARC entry.
	*/
	arc_hdr_clear_flags(hdr, ARC_FLAG_L2_WRITING);

	mutex_exit(hash_lock);
	}

	/*
	* Free the allocated abd buffers for writing the log blocks.
	* If the zio failed reclaim the allocated space and remove the
	* pointers to these log blocks from the log block pointer list
	* of the L2ARC device.
	*/
	while ((abd_buf = list_remove_tail(&cb->l2wcb_abd_list)) != NULL) {
	abd_free(abd_buf->abd);
	zio_buf_free(abd_buf, sizeof (*abd_buf));
	if (zio->io_error != 0) {
	lb_ptr_buf = list_remove_head(&dev->l2ad_lbptr_list);
	/*
	* L2BLK_GET_PSIZE returns aligned size for log
	* blocks.
	*/
	uint64_t asize =
	L2BLK_GET_PSIZE((lb_ptr_buf->lb_ptr)->lbp_prop);
	bytes_dropped += asize;
	ARCSTAT_INCR(arcstat_l2_log_blk_asize, -asize);
	ARCSTAT_BUMPDOWN(arcstat_l2_log_blk_count);
	zfs_refcount_remove_many(&dev->l2ad_lb_asize, asize,
	lb_ptr_buf);
	zfs_refcount_remove(&dev->l2ad_lb_count, lb_ptr_buf);
	kmem_free(lb_ptr_buf->lb_ptr,
	sizeof (l2arc_log_blkptr_t));
	kmem_free(lb_ptr_buf, sizeof (l2arc_lb_ptr_buf_t));
	}
	}
	list_destroy(&cb->l2wcb_abd_list);

	if (zio->io_error != 0) {
	ARCSTAT_BUMP(arcstat_l2_writes_error);

	/*
	* Restore the lbps array in the header to its previous state.
	* If the list of log block pointers is empty, zero out the
	* log block pointers in the device header.
	*/
	lb_ptr_buf = list_head(&dev->l2ad_lbptr_list);
	for (int i = 0; i < 2; i++) {
	if (lb_ptr_buf == NULL) {
	/*
	* If the list is empty zero out the device
	* header. Otherwise zero out the second log
	* block pointer in the header.
	*/
	if (i == 0) {
	bzero(l2dhdr, dev->l2ad_dev_hdr_asize);
	} else {
	bzero(&l2dhdr->dh_start_lbps[i],
	sizeof (l2arc_log_blkptr_t));
	}
	break;
	}
	bcopy(lb_ptr_buf->lb_ptr, &l2dhdr->dh_start_lbps[i],
	sizeof (l2arc_log_blkptr_t));
	lb_ptr_buf = list_next(&dev->l2ad_lbptr_list,
	lb_ptr_buf);
	}
	}

	ARCSTAT_BUMP(arcstat_l2_writes_done);
	list_remove(buflist, head);
	ASSERT(!HDR_HAS_L1HDR(head));
	kmem_cache_free(hdr_l2only_cache, head);
	mutex_exit(&dev->l2ad_mtx);

	ASSERT(dev->l2ad_vdev != NULL);
	vdev_space_update(dev->l2ad_vdev, -bytes_dropped, 0, 0);

	l2arc_do_free_on_write();

	kmem_free(cb, sizeof (l2arc_write_callback_t));
	}

	static int
	l2arc_untransform(zio_t zio, l2arc_read_callback_t cb)
	{
	int ret;
	spa_t *spa = zio->io_spa;
	arc_buf_hdr_t *hdr = cb->l2rcb_hdr;
	blkptr_t *bp = zio->io_bp;
	uint8_t salt[ZIO_DATA_SALT_LEN];
	uint8_t iv[ZIO_DATA_IV_LEN];
	uint8_t mac[ZIO_DATA_MAC_LEN];
	boolean_t no_crypt = B_FALSE;

	/*
	* ZIL data is never be written to the L2ARC, so we don't need
	* special handling for its unique MAC storage.
	*/
	ASSERT3U(BP_GET_TYPE(bp), !=, DMU_OT_INTENT_LOG);
	ASSERT(MUTEX_HELD(HDR_LOCK(hdr)));
	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);

	/*
	* If the data was encrypted, decrypt it now. Note that
	* we must check the bp here and not the hdr, since the
	* hdr does not have its encryption parameters updated
	* until arc_read_done().
	*/
	if (BP_IS_ENCRYPTED(bp)) {
	abd_t *eabd = arc_get_data_abd(hdr, arc_hdr_size(hdr), hdr,
	ARC_HDR_DO_ADAPT \| ARC_HDR_USE_RESERVE);

	zio_crypt_decode_params_bp(bp, salt, iv);
	zio_crypt_decode_mac_bp(bp, mac);

	ret = spa_do_crypt_abd(B_FALSE, spa, &cb->l2rcb_zb,
	BP_GET_TYPE(bp), BP_GET_DEDUP(bp), BP_SHOULD_BYTESWAP(bp),
	salt, iv, mac, HDR_GET_PSIZE(hdr), eabd,
	hdr->b_l1hdr.b_pabd, &no_crypt);
	if (ret != 0) {
	arc_free_data_abd(hdr, eabd, arc_hdr_size(hdr), hdr);
	goto error;
	}

	/*
	* If we actually performed decryption, replace b_pabd
	* with the decrypted data. Otherwise we can just throw
	* our decryption buffer away.
	*/
	if (!no_crypt) {
	arc_free_data_abd(hdr, hdr->b_l1hdr.b_pabd,
	arc_hdr_size(hdr), hdr);
	hdr->b_l1hdr.b_pabd = eabd;
	zio->io_abd = eabd;
	} else {
	arc_free_data_abd(hdr, eabd, arc_hdr_size(hdr), hdr);
	}
	}

	/*
	* If the L2ARC block was compressed, but ARC compression
	* is disabled we decompress the data into a new buffer and
	* replace the existing data.
	*/
	if (HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF &&
	!HDR_COMPRESSION_ENABLED(hdr)) {
	abd_t *cabd = arc_get_data_abd(hdr, arc_hdr_size(hdr), hdr,
	ARC_HDR_DO_ADAPT \| ARC_HDR_USE_RESERVE);
	void *tmp = abd_borrow_buf(cabd, arc_hdr_size(hdr));

	ret = zio_decompress_data(HDR_GET_COMPRESS(hdr),
	hdr->b_l1hdr.b_pabd, tmp, HDR_GET_PSIZE(hdr),
	HDR_GET_LSIZE(hdr), &hdr->b_complevel);
	if (ret != 0) {
	abd_return_buf_copy(cabd, tmp, arc_hdr_size(hdr));
	arc_free_data_abd(hdr, cabd, arc_hdr_size(hdr), hdr);
	goto error;
	}

	abd_return_buf_copy(cabd, tmp, arc_hdr_size(hdr));
	arc_free_data_abd(hdr, hdr->b_l1hdr.b_pabd,
	arc_hdr_size(hdr), hdr);
	hdr->b_l1hdr.b_pabd = cabd;
	zio->io_abd = cabd;
	zio->io_size = HDR_GET_LSIZE(hdr);
	}

	return (0);

	error:
	return (ret);
	}


	/*
	* A read to a cache device completed. Validate buffer contents before
	* handing over to the regular ARC routines.
	*/
	static void
	l2arc_read_done(zio_t *zio)
	{
	int tfm_error = 0;
	l2arc_read_callback_t *cb = zio->io_private;
	arc_buf_hdr_t *hdr;
	kmutex_t *hash_lock;
	boolean_t valid_cksum;
	boolean_t using_rdata = (BP_IS_ENCRYPTED(&cb->l2rcb_bp) &&
	(cb->l2rcb_flags & ZIO_FLAG_RAW_ENCRYPT));

	ASSERT3P(zio->io_vd, !=, NULL);
	ASSERT(zio->io_flags & ZIO_FLAG_DONT_PROPAGATE);

	spa_config_exit(zio->io_spa, SCL_L2ARC, zio->io_vd);

	ASSERT3P(cb, !=, NULL);
	hdr = cb->l2rcb_hdr;
	ASSERT3P(hdr, !=, NULL);

	hash_lock = HDR_LOCK(hdr);
	mutex_enter(hash_lock);
	ASSERT3P(hash_lock, ==, HDR_LOCK(hdr));

	/*
	* If the data was read into a temporary buffer,
	* move it and free the buffer.
	*/
	if (cb->l2rcb_abd != NULL) {
	ASSERT3U(arc_hdr_size(hdr), <, zio->io_size);
	if (zio->io_error == 0) {
	if (using_rdata) {
	abd_copy(hdr->b_crypt_hdr.b_rabd,
	cb->l2rcb_abd, arc_hdr_size(hdr));
	} else {
	abd_copy(hdr->b_l1hdr.b_pabd,
	cb->l2rcb_abd, arc_hdr_size(hdr));
	}
	}

	/*
	* The following must be done regardless of whether
	* there was an error:
	* - free the temporary buffer
	* - point zio to the real ARC buffer
	* - set zio size accordingly
	* These are required because zio is either re-used for
	* an I/O of the block in the case of the error
	* or the zio is passed to arc_read_done() and it
	* needs real data.
	*/
	abd_free(cb->l2rcb_abd);
	zio->io_size = zio->io_orig_size = arc_hdr_size(hdr);

	if (using_rdata) {
	ASSERT(HDR_HAS_RABD(hdr));
	zio->io_abd = zio->io_orig_abd =
	hdr->b_crypt_hdr.b_rabd;
	} else {
	ASSERT3P(hdr->b_l1hdr.b_pabd, !=, NULL);
	zio->io_abd = zio->io_orig_abd = hdr->b_l1hdr.b_pabd;
	}
	}

	ASSERT3P(zio->io_abd, !=, NULL);

	/*
	* Check this survived the L2ARC journey.
	*/
	ASSERT(zio->io_abd == hdr->b_l1hdr.b_pabd \|\|
	(HDR_HAS_RABD(hdr) && zio->io_abd == hdr->b_crypt_hdr.b_rabd));
	zio->io_bp_copy = cb->l2rcb_bp; /* XXX fix in L2ARC 2.0 */
	zio->io_bp = &zio->io_bp_copy; /* XXX fix in L2ARC 2.0 */
	zio->io_prop.zp_complevel = hdr->b_complevel;

	valid_cksum = arc_cksum_is_equal(hdr, zio);

	/*
	* b_rabd will always match the data as it exists on disk if it is
	* being used. Therefore if we are reading into b_rabd we do not
	* attempt to untransform the data.
	*/
	if (valid_cksum && !using_rdata)
	tfm_error = l2arc_untransform(zio, cb);

	if (valid_cksum && tfm_error == 0 && zio->io_error == 0 &&
	!HDR_L2_EVICTED(hdr)) {
	mutex_exit(hash_lock);
	zio->io_private = hdr;
	arc_read_done(zio);
	} else {
	/*
	* Buffer didn't survive caching. Increment stats and
	* reissue to the original storage device.
	*/
	if (zio->io_error != 0) {
	ARCSTAT_BUMP(arcstat_l2_io_error);
	} else {
	zio->io_error = SET_ERROR(EIO);
	}
	if (!valid_cksum \|\| tfm_error != 0)
	ARCSTAT_BUMP(arcstat_l2_cksum_bad);

	/*
	* If there's no waiter, issue an async i/o to the primary
	* storage now. If there is a waiter, the caller must
	* issue the i/o in a context where it's OK to block.
	*/
	if (zio->io_waiter == NULL) {
	zio_t *pio = zio_unique_parent(zio);
	void *abd = (using_rdata) ?
	hdr->b_crypt_hdr.b_rabd : hdr->b_l1hdr.b_pabd;

	ASSERT(!pio \|\| pio->io_child_type == ZIO_CHILD_LOGICAL);

	zio = zio_read(pio, zio->io_spa, zio->io_bp,
	abd, zio->io_size, arc_read_done,
	hdr, zio->io_priority, cb->l2rcb_flags,
	&cb->l2rcb_zb);

	/*
	* Original ZIO will be freed, so we need to update
	* ARC header with the new ZIO pointer to be used
	* by zio_change_priority() in arc_read().
	*/
	for (struct arc_callback *acb = hdr->b_l1hdr.b_acb;
	acb != NULL; acb = acb->acb_next)
	acb->acb_zio_head = zio;

	mutex_exit(hash_lock);
	zio_nowait(zio);
	} else {
	mutex_exit(hash_lock);
	}
	}

	kmem_free(cb, sizeof (l2arc_read_callback_t));
	}

	/*
	* This is the list priority from which the L2ARC will search for pages to
	* cache. This is used within loops (0..3) to cycle through lists in the
	* desired order. This order can have a significant effect on cache
	* performance.
	*
	* Currently the metadata lists are hit first, MFU then MRU, followed by
	* the data lists. This function returns a locked list, and also returns
	* the lock pointer.
	*/
	static multilist_sublist_t *
	l2arc_sublist_lock(int list_num)
	{
	multilist_t *ml = NULL;
	unsigned int idx;

	ASSERT(list_num >= 0 && list_num < L2ARC_FEED_TYPES);

	switch (list_num) {
	case 0:
	ml = &arc_mfu->arcs_list[ARC_BUFC_METADATA];
	break;
	case 1:
	ml = &arc_mru->arcs_list[ARC_BUFC_METADATA];
	break;
	case 2:
	ml = &arc_mfu->arcs_list[ARC_BUFC_DATA];
	break;
	case 3:
	ml = &arc_mru->arcs_list[ARC_BUFC_DATA];
	break;
	default:
	return (NULL);
	}

	/*
	* Return a randomly-selected sublist. This is acceptable
	* because the caller feeds only a little bit of data for each
	* call (8MB). Subsequent calls will result in different
	* sublists being selected.
	*/
	idx = multilist_get_random_index(ml);
	return (multilist_sublist_lock(ml, idx));
	}

	/*
	* Calculates the maximum overhead of L2ARC metadata log blocks for a given
	* L2ARC write size. l2arc_evict and l2arc_write_size need to include this
	* overhead in processing to make sure there is enough headroom available
	* when writing buffers.
	*/
	static inline uint64_t
	l2arc_log_blk_overhead(uint64_t write_sz, l2arc_dev_t *dev)
	{
	if (dev->l2ad_log_entries == 0) {
	return (0);
	} else {
	uint64_t log_entries = write_sz >> SPA_MINBLOCKSHIFT;

	uint64_t log_blocks = (log_entries +
	dev->l2ad_log_entries - 1) /
	dev->l2ad_log_entries;

	return (vdev_psize_to_asize(dev->l2ad_vdev,
	sizeof (l2arc_log_blk_phys_t)) * log_blocks);
	}
	}

	/*
	* Evict buffers from the device write hand to the distance specified in
	* bytes. This distance may span populated buffers, it may span nothing.
	* This is clearing a region on the L2ARC device ready for writing.
	* If the 'all' boolean is set, every buffer is evicted.
	*/
	static void
	l2arc_evict(l2arc_dev_t *dev, uint64_t distance, boolean_t all)
	{
	list_t *buflist;
	arc_buf_hdr_t hdr, hdr_prev;
	kmutex_t *hash_lock;
	uint64_t taddr;
	l2arc_lb_ptr_buf_t lb_ptr_buf, lb_ptr_buf_prev;
	vdev_t *vd = dev->l2ad_vdev;
	boolean_t rerun;

	buflist = &dev->l2ad_buflist;

	/*
	* We need to add in the worst case scenario of log block overhead.
	*/
	distance += l2arc_log_blk_overhead(distance, dev);
	if (vd->vdev_has_trim && l2arc_trim_ahead > 0) {
	/*
	* Trim ahead of the write size 64MB or (l2arc_trim_ahead/100)
	* times the write size, whichever is greater.
	*/
	distance += MAX(64 * 1024 * 1024,
	(distance * l2arc_trim_ahead) / 100);
	}

	top:
	rerun = B_FALSE;
	if (dev->l2ad_hand >= (dev->l2ad_end - distance)) {
	/*
	* When there is no space to accommodate upcoming writes,
	* evict to the end. Then bump the write and evict hands
	* to the start and iterate. This iteration does not
	* happen indefinitely as we make sure in
	* l2arc_write_size() that when the write hand is reset,
	* the write size does not exceed the end of the device.
	*/
	rerun = B_TRUE;
	taddr = dev->l2ad_end;
	} else {
	taddr = dev->l2ad_hand + distance;
	}
	DTRACE_PROBE4(l2arc__evict, l2arc_dev_t , dev, list_t , buflist,
	uint64_t, taddr, boolean_t, all);

	if (!all) {
	/*
	* This check has to be placed after deciding whether to
	* iterate (rerun).
	*/
	if (dev->l2ad_first) {
	/*
	* This is the first sweep through the device. There is
	* nothing to evict. We have already trimmmed the
	* whole device.
	*/
	goto out;
	} else {
	/*
	* Trim the space to be evicted.
	*/
	if (vd->vdev_has_trim && dev->l2ad_evict < taddr &&
	l2arc_trim_ahead > 0) {
	/*
	* We have to drop the spa_config lock because
	* vdev_trim_range() will acquire it.
	* l2ad_evict already accounts for the label
	* size. To prevent vdev_trim_ranges() from
	* adding it again, we subtract it from
	* l2ad_evict.
	*/
	spa_config_exit(dev->l2ad_spa, SCL_L2ARC, dev);
	vdev_trim_simple(vd,
	dev->l2ad_evict - VDEV_LABEL_START_SIZE,
	taddr - dev->l2ad_evict);
	spa_config_enter(dev->l2ad_spa, SCL_L2ARC, dev,
	RW_READER);
	}

	/*
	* When rebuilding L2ARC we retrieve the evict hand
	* from the header of the device. Of note, l2arc_evict()
	* does not actually delete buffers from the cache
	* device, but trimming may do so depending on the
	* hardware implementation. Thus keeping track of the
	* evict hand is useful.
	*/
	dev->l2ad_evict = MAX(dev->l2ad_evict, taddr);
	}
	}

	retry:
	mutex_enter(&dev->l2ad_mtx);
	/*
	* We have to account for evicted log blocks. Run vdev_space_update()
	* on log blocks whose offset (in bytes) is before the evicted offset
	* (in bytes) by searching in the list of pointers to log blocks
	* present in the L2ARC device.
	*/
	for (lb_ptr_buf = list_tail(&dev->l2ad_lbptr_list); lb_ptr_buf;
	lb_ptr_buf = lb_ptr_buf_prev) {

	lb_ptr_buf_prev = list_prev(&dev->l2ad_lbptr_list, lb_ptr_buf);

	/* L2BLK_GET_PSIZE returns aligned size for log blocks */
	uint64_t asize = L2BLK_GET_PSIZE(
	(lb_ptr_buf->lb_ptr)->lbp_prop);

	/*
	* We don't worry about log blocks left behind (ie
	* lbp_payload_start < l2ad_hand) because l2arc_write_buffers()
	* will never write more than l2arc_evict() evicts.
	*/
	if (!all && l2arc_log_blkptr_valid(dev, lb_ptr_buf->lb_ptr)) {
	break;
	} else {
	vdev_space_update(vd, -asize, 0, 0);
	ARCSTAT_INCR(arcstat_l2_log_blk_asize, -asize);
	ARCSTAT_BUMPDOWN(arcstat_l2_log_blk_count);
	zfs_refcount_remove_many(&dev->l2ad_lb_asize, asize,
	lb_ptr_buf);
	zfs_refcount_remove(&dev->l2ad_lb_count, lb_ptr_buf);
	list_remove(&dev->l2ad_lbptr_list, lb_ptr_buf);
	kmem_free(lb_ptr_buf->lb_ptr,
	sizeof (l2arc_log_blkptr_t));
	kmem_free(lb_ptr_buf, sizeof (l2arc_lb_ptr_buf_t));
	}
	}

	for (hdr = list_tail(buflist); hdr; hdr = hdr_prev) {
	hdr_prev = list_prev(buflist, hdr);

	ASSERT(!HDR_EMPTY(hdr));
	hash_lock = HDR_LOCK(hdr);

	/*
	* We cannot use mutex_enter or else we can deadlock
	* with l2arc_write_buffers (due to swapping the order
	* the hash lock and l2ad_mtx are taken).
	*/
	if (!mutex_tryenter(hash_lock)) {
	/*
	* Missed the hash lock. Retry.
	*/
	ARCSTAT_BUMP(arcstat_l2_evict_lock_retry);
	mutex_exit(&dev->l2ad_mtx);
	mutex_enter(hash_lock);
	mutex_exit(hash_lock);
	goto retry;
	}

	/*
	* A header can't be on this list if it doesn't have L2 header.
	*/
	ASSERT(HDR_HAS_L2HDR(hdr));

	/* Ensure this header has finished being written. */
	ASSERT(!HDR_L2_WRITING(hdr));
	ASSERT(!HDR_L2_WRITE_HEAD(hdr));

	if (!all && (hdr->b_l2hdr.b_daddr >= dev->l2ad_evict \|\|
	hdr->b_l2hdr.b_daddr < dev->l2ad_hand)) {
	/*
	* We've evicted to the target address,
	* or the end of the device.
	*/
	mutex_exit(hash_lock);
	break;
	}

	if (!HDR_HAS_L1HDR(hdr)) {
	ASSERT(!HDR_L2_READING(hdr));
	/*
	* This doesn't exist in the ARC. Destroy.
	* arc_hdr_destroy() will call list_remove()
	* and decrement arcstat_l2_lsize.
	*/
	arc_change_state(arc_anon, hdr, hash_lock);
	arc_hdr_destroy(hdr);
	} else {
	ASSERT(hdr->b_l1hdr.b_state != arc_l2c_only);
	ARCSTAT_BUMP(arcstat_l2_evict_l1cached);
	/*
	* Invalidate issued or about to be issued
	* reads, since we may be about to write
	* over this location.
	*/
	if (HDR_L2_READING(hdr)) {
	ARCSTAT_BUMP(arcstat_l2_evict_reading);
	arc_hdr_set_flags(hdr, ARC_FLAG_L2_EVICTED);
	}

	arc_hdr_l2hdr_destroy(hdr);
	}
	mutex_exit(hash_lock);
	}
	mutex_exit(&dev->l2ad_mtx);

	out:
	/*
	* We need to check if we evict all buffers, otherwise we may iterate
	* unnecessarily.
	*/
	if (!all && rerun) {
	/*
	* Bump device hand to the device start if it is approaching the
	* end. l2arc_evict() has already evicted ahead for this case.
	*/
	dev->l2ad_hand = dev->l2ad_start;
	dev->l2ad_evict = dev->l2ad_start;
	dev->l2ad_first = B_FALSE;
	goto top;
	}

	if (!all) {
	/*
	* In case of cache device removal (all) the following
	* assertions may be violated without functional consequences
	* as the device is about to be removed.
	*/
	ASSERT3U(dev->l2ad_hand + distance, <, dev->l2ad_end);
	if (!dev->l2ad_first)
	ASSERT3U(dev->l2ad_hand, <, dev->l2ad_evict);
	}
	}

	/*
	* Handle any abd transforms that might be required for writing to the L2ARC.
	* If successful, this function will always return an abd with the data
	* transformed as it is on disk in a new abd of asize bytes.
	*/
	static int
	l2arc_apply_transforms(spa_t spa, arc_buf_hdr_t hdr, uint64_t asize,
	abd_t **abd_out)
	{
	int ret;
	void *tmp = NULL;
	abd_t cabd = NULL, eabd = NULL, *to_write = hdr->b_l1hdr.b_pabd;
	enum zio_compress compress = HDR_GET_COMPRESS(hdr);
	uint64_t psize = HDR_GET_PSIZE(hdr);
	uint64_t size = arc_hdr_size(hdr);
	boolean_t ismd = HDR_ISTYPE_METADATA(hdr);
	boolean_t bswap = (hdr->b_l1hdr.b_byteswap != DMU_BSWAP_NUMFUNCS);
	dsl_crypto_key_t *dck = NULL;
	uint8_t mac[ZIO_DATA_MAC_LEN] = { 0 };
	boolean_t no_crypt = B_FALSE;

	ASSERT((HDR_GET_COMPRESS(hdr) != ZIO_COMPRESS_OFF &&
	!HDR_COMPRESSION_ENABLED(hdr)) \|\|
	HDR_ENCRYPTED(hdr) \|\| HDR_SHARED_DATA(hdr) \|\| psize != asize);
	ASSERT3U(psize, <=, asize);

	/*
	* If this data simply needs its own buffer, we simply allocate it
	* and copy the data. This may be done to eliminate a dependency on a
	* shared buffer or to reallocate the buffer to match asize.
	*/
	if (HDR_HAS_RABD(hdr) && asize != psize) {
	ASSERT3U(asize, >=, psize);
	to_write = abd_alloc_for_io(asize, ismd);
	abd_copy(to_write, hdr->b_crypt_hdr.b_rabd, psize);
	if (psize != asize)
	abd_zero_off(to_write, psize, asize - psize);
	goto out;
	}

	if ((compress == ZIO_COMPRESS_OFF \|\| HDR_COMPRESSION_ENABLED(hdr)) &&
	!HDR_ENCRYPTED(hdr)) {
	ASSERT3U(size, ==, psize);
	to_write = abd_alloc_for_io(asize, ismd);
	abd_copy(to_write, hdr->b_l1hdr.b_pabd, size);
	if (size != asize)
	abd_zero_off(to_write, size, asize - size);
	goto out;
	}

	if (compress != ZIO_COMPRESS_OFF && !HDR_COMPRESSION_ENABLED(hdr)) {
	cabd = abd_alloc_for_io(asize, ismd);
	tmp = abd_borrow_buf(cabd, asize);

	psize = zio_compress_data(compress, to_write, tmp, size,
	hdr->b_complevel);

	if (psize >= size) {
	abd_return_buf(cabd, tmp, asize);
	HDR_SET_COMPRESS(hdr, ZIO_COMPRESS_OFF);
	to_write = cabd;
	abd_copy(to_write, hdr->b_l1hdr.b_pabd, size);
	if (size != asize)
	abd_zero_off(to_write, size, asize - size);
	goto encrypt;
	}
	ASSERT3U(psize, <=, HDR_GET_PSIZE(hdr));
	if (psize < asize)
	bzero((char *)tmp + psize, asize - psize);
	psize = HDR_GET_PSIZE(hdr);
	abd_return_buf_copy(cabd, tmp, asize);
	to_write = cabd;
	}

	encrypt:
	if (HDR_ENCRYPTED(hdr)) {
	eabd = abd_alloc_for_io(asize, ismd);

	/*
	* If the dataset was disowned before the buffer
	* made it to this point, the key to re-encrypt
	* it won't be available. In this case we simply
	* won't write the buffer to the L2ARC.
	*/
	ret = spa_keystore_lookup_key(spa, hdr->b_crypt_hdr.b_dsobj,
	FTAG, &dck);
	if (ret != 0)
	goto error;

	ret = zio_do_crypt_abd(B_TRUE, &dck->dck_key,
	hdr->b_crypt_hdr.b_ot, bswap, hdr->b_crypt_hdr.b_salt,
	hdr->b_crypt_hdr.b_iv, mac, psize, to_write, eabd,
	&no_crypt);
	if (ret != 0)
	goto error;

	if (no_crypt)
	abd_copy(eabd, to_write, psize);

	if (psize != asize)
	abd_zero_off(eabd, psize, asize - psize);

	/* assert that the MAC we got here matches the one we saved */
	ASSERT0(bcmp(mac, hdr->b_crypt_hdr.b_mac, ZIO_DATA_MAC_LEN));
	spa_keystore_dsl_key_rele(spa, dck, FTAG);

	if (to_write == cabd)
	abd_free(cabd);

	to_write = eabd;
	}

	out:
	ASSERT3P(to_write, !=, hdr->b_l1hdr.b_pabd);
	*abd_out = to_write;
	return (0);

	error:
	if (dck != NULL)
	spa_keystore_dsl_key_rele(spa, dck, FTAG);
	if (cabd != NULL)
	abd_free(cabd);
	if (eabd != NULL)
	abd_free(eabd);

	*abd_out = NULL;
	return (ret);
	}

	static void
	l2arc_blk_fetch_done(zio_t *zio)
	{
	l2arc_read_callback_t *cb;

	cb = zio->io_private;
	if (cb->l2rcb_abd != NULL)
	abd_free(cb->l2rcb_abd);
	kmem_free(cb, sizeof (l2arc_read_callback_t));
	}

	/*
	* Find and write ARC buffers to the L2ARC device.
	*
	* An ARC_FLAG_L2_WRITING flag is set so that the L2ARC buffers are not valid
	* for reading until they have completed writing.
	* The headroom_boost is an in-out parameter used to maintain headroom boost
	* state between calls to this function.
	*
	* Returns the number of bytes actually written (which may be smaller than
	* the delta by which the device hand has changed due to alignment and the
	* writing of log blocks).
	*/
	static uint64_t
	l2arc_write_buffers(spa_t spa, l2arc_dev_t dev, uint64_t target_sz)
	{
	arc_buf_hdr_t hdr, hdr_prev, *head;
	uint64_t write_asize, write_psize, write_lsize, headroom;
	boolean_t full;
	l2arc_write_callback_t *cb = NULL;
	zio_t pio, wzio;
	uint64_t guid = spa_load_guid(spa);
	l2arc_dev_hdr_phys_t *l2dhdr = dev->l2ad_dev_hdr;

	ASSERT3P(dev->l2ad_vdev, !=, NULL);

	pio = NULL;
	write_lsize = write_asize = write_psize = 0;
	full = B_FALSE;
	head = kmem_cache_alloc(hdr_l2only_cache, KM_PUSHPAGE);
	arc_hdr_set_flags(head, ARC_FLAG_L2_WRITE_HEAD \| ARC_FLAG_HAS_L2HDR);

	/*
	* Copy buffers for L2ARC writing.
	*/
	for (int pass = 0; pass < L2ARC_FEED_TYPES; pass++) {
	/*
	* If pass == 1 or 3, we cache MRU metadata and data
	* respectively.
	*/
	if (l2arc_mfuonly) {
	if (pass == 1 \|\| pass == 3)
	continue;
	}

	multilist_sublist_t *mls = l2arc_sublist_lock(pass);
	uint64_t passed_sz = 0;

	VERIFY3P(mls, !=, NULL);

	/*
	* L2ARC fast warmup.
	*
	* Until the ARC is warm and starts to evict, read from the
	* head of the ARC lists rather than the tail.
	*/
	if (arc_warm == B_FALSE)
	hdr = multilist_sublist_head(mls);
	else
	hdr = multilist_sublist_tail(mls);

	headroom = target_sz * l2arc_headroom;
	if (zfs_compressed_arc_enabled)
	headroom = (headroom * l2arc_headroom_boost) / 100;

	for (; hdr; hdr = hdr_prev) {
	kmutex_t *hash_lock;
	abd_t *to_write = NULL;

	if (arc_warm == B_FALSE)
	hdr_prev = multilist_sublist_next(mls, hdr);
	else
	hdr_prev = multilist_sublist_prev(mls, hdr);

	hash_lock = HDR_LOCK(hdr);
	if (!mutex_tryenter(hash_lock)) {
	/*
	* Skip this buffer rather than waiting.
	*/
	continue;
	}

	passed_sz += HDR_GET_LSIZE(hdr);
	if (l2arc_headroom != 0 && passed_sz > headroom) {
	/*
	* Searched too far.
	*/
	mutex_exit(hash_lock);
	break;
	}

	if (!l2arc_write_eligible(guid, hdr)) {
	mutex_exit(hash_lock);
	continue;
	}

	/*
	* We rely on the L1 portion of the header below, so
	* it's invalid for this header to have been evicted out
	* of the ghost cache, prior to being written out. The
	* ARC_FLAG_L2_WRITING bit ensures this won't happen.
	*/
	ASSERT(HDR_HAS_L1HDR(hdr));

	ASSERT3U(HDR_GET_PSIZE(hdr), >, 0);
	ASSERT3U(arc_hdr_size(hdr), >, 0);
	ASSERT(hdr->b_l1hdr.b_pabd != NULL \|\|
	HDR_HAS_RABD(hdr));
	uint64_t psize = HDR_GET_PSIZE(hdr);
	uint64_t asize = vdev_psize_to_asize(dev->l2ad_vdev,
	psize);

	if ((write_asize + asize) > target_sz) {
	full = B_TRUE;
	mutex_exit(hash_lock);
	break;
	}

	/*
	* We rely on the L1 portion of the header below, so
	* it's invalid for this header to have been evicted out
	* of the ghost cache, prior to being written out. The
	* ARC_FLAG_L2_WRITING bit ensures this won't happen.
	*/
	arc_hdr_set_flags(hdr, ARC_FLAG_L2_WRITING);
	ASSERT(HDR_HAS_L1HDR(hdr));

	ASSERT3U(HDR_GET_PSIZE(hdr), >, 0);
	ASSERT(hdr->b_l1hdr.b_pabd != NULL \|\|
	HDR_HAS_RABD(hdr));
	ASSERT3U(arc_hdr_size(hdr), >, 0);

	/*
	* If this header has b_rabd, we can use this since it
	* must always match the data exactly as it exists on
	* disk. Otherwise, the L2ARC can normally use the
	* hdr's data, but if we're sharing data between the
	* hdr and one of its bufs, L2ARC needs its own copy of
	* the data so that the ZIO below can't race with the
	* buf consumer. To ensure that this copy will be
	* available for the lifetime of the ZIO and be cleaned
	* up afterwards, we add it to the l2arc_free_on_write
	* queue. If we need to apply any transforms to the
	* data (compression, encryption) we will also need the
	* extra buffer.
	*/
	if (HDR_HAS_RABD(hdr) && psize == asize) {
	to_write = hdr->b_crypt_hdr.b_rabd;
	} else if ((HDR_COMPRESSION_ENABLED(hdr) \|\|
	HDR_GET_COMPRESS(hdr) == ZIO_COMPRESS_OFF) &&
	!HDR_ENCRYPTED(hdr) && !HDR_SHARED_DATA(hdr) &&
	psize == asize) {
	to_write = hdr->b_l1hdr.b_pabd;
	} else {
	int ret;
	arc_buf_contents_t type = arc_buf_type(hdr);

	ret = l2arc_apply_transforms(spa, hdr, asize,
	&to_write);
	if (ret != 0) {
	arc_hdr_clear_flags(hdr,
	ARC_FLAG_L2_WRITING);
	mutex_exit(hash_lock);
	continue;
	}

	l2arc_free_abd_on_write(to_write, asize, type);
	}

	if (pio == NULL) {
	/*
	* Insert a dummy header on the buflist so
	* l2arc_write_done() can find where the
	* write buffers begin without searching.
	*/
	mutex_enter(&dev->l2ad_mtx);
	list_insert_head(&dev->l2ad_buflist, head);
	mutex_exit(&dev->l2ad_mtx);

	cb = kmem_alloc(
	sizeof (l2arc_write_callback_t), KM_SLEEP);
	cb->l2wcb_dev = dev;
	cb->l2wcb_head = head;
	/*
	* Create a list to save allocated abd buffers
	* for l2arc_log_blk_commit().
	*/
	list_create(&cb->l2wcb_abd_list,
	sizeof (l2arc_lb_abd_buf_t),
	offsetof(l2arc_lb_abd_buf_t, node));
	pio = zio_root(spa, l2arc_write_done, cb,
	ZIO_FLAG_CANFAIL);
	}

	hdr->b_l2hdr.b_dev = dev;
	hdr->b_l2hdr.b_hits = 0;

	hdr->b_l2hdr.b_daddr = dev->l2ad_hand;
	hdr->b_l2hdr.b_arcs_state =
	hdr->b_l1hdr.b_state->arcs_state;
	arc_hdr_set_flags(hdr, ARC_FLAG_HAS_L2HDR);

	mutex_enter(&dev->l2ad_mtx);
	list_insert_head(&dev->l2ad_buflist, hdr);
	mutex_exit(&dev->l2ad_mtx);

	(void) zfs_refcount_add_many(&dev->l2ad_alloc,
	arc_hdr_size(hdr), hdr);

	wzio = zio_write_phys(pio, dev->l2ad_vdev,
	hdr->b_l2hdr.b_daddr, asize, to_write,
	ZIO_CHECKSUM_OFF, NULL, hdr,
	ZIO_PRIORITY_ASYNC_WRITE,
	ZIO_FLAG_CANFAIL, B_FALSE);

	write_lsize += HDR_GET_LSIZE(hdr);
	DTRACE_PROBE2(l2arc__write, vdev_t *, dev->l2ad_vdev,
	zio_t *, wzio);

	write_psize += psize;
	write_asize += asize;
	dev->l2ad_hand += asize;
	l2arc_hdr_arcstats_increment(hdr);
	vdev_space_update(dev->l2ad_vdev, asize, 0, 0);

	mutex_exit(hash_lock);

	/*
	* Append buf info to current log and commit if full.
	* arcstat_l2_{size,asize} kstats are updated
	* internally.
	*/
	if (l2arc_log_blk_insert(dev, hdr))
	l2arc_log_blk_commit(dev, pio, cb);

	zio_nowait(wzio);
	}

	multilist_sublist_unlock(mls);

	if (full == B_TRUE)
	break;
	}

	/* No buffers selected for writing? */
	if (pio == NULL) {
	ASSERT0(write_lsize);
	ASSERT(!HDR_HAS_L1HDR(head));
	kmem_cache_free(hdr_l2only_cache, head);

	/*
	* Although we did not write any buffers l2ad_evict may
	* have advanced.
	*/
	if (dev->l2ad_evict != l2dhdr->dh_evict)
	l2arc_dev_hdr_update(dev);

	return (0);
	}

	if (!dev->l2ad_first)
	ASSERT3U(dev->l2ad_hand, <=, dev->l2ad_evict);

	ASSERT3U(write_asize, <=, target_sz);
	ARCSTAT_BUMP(arcstat_l2_writes_sent);
	ARCSTAT_INCR(arcstat_l2_write_bytes, write_psize);

	dev->l2ad_writing = B_TRUE;
	(void) zio_wait(pio);
	dev->l2ad_writing = B_FALSE;

	/*
	* Update the device header after the zio completes as
	* l2arc_write_done() may have updated the memory holding the log block
	* pointers in the device header.
	*/
	l2arc_dev_hdr_update(dev);

	return (write_asize);
	}

	static boolean_t
	l2arc_hdr_limit_reached(void)
	{
	int64_t s = aggsum_upper_bound(&arc_sums.arcstat_l2_hdr_size);

	return (arc_reclaim_needed() \|\| (s > arc_meta_limit * 3 / 4) \|\|
	(s > (arc_warm ? arc_c : arc_c_max) * l2arc_meta_percent / 100));
	}

	/*
	* This thread feeds the L2ARC at regular intervals. This is the beating
	* heart of the L2ARC.
	*/
	/* ARGSUSED */
	static void
	l2arc_feed_thread(void *unused)
	{
	callb_cpr_t cpr;
	l2arc_dev_t *dev;
	spa_t *spa;
	uint64_t size, wrote;
	clock_t begin, next = ddi_get_lbolt();
	fstrans_cookie_t cookie;

	CALLB_CPR_INIT(&cpr, &l2arc_feed_thr_lock, callb_generic_cpr, FTAG);

	mutex_enter(&l2arc_feed_thr_lock);

	cookie = spl_fstrans_mark();
	while (l2arc_thread_exit == 0) {
	CALLB_CPR_SAFE_BEGIN(&cpr);
	(void) cv_timedwait_idle(&l2arc_feed_thr_cv,
	&l2arc_feed_thr_lock, next);
	CALLB_CPR_SAFE_END(&cpr, &l2arc_feed_thr_lock);
	next = ddi_get_lbolt() + hz;

	/*
	* Quick check for L2ARC devices.
	*/
	mutex_enter(&l2arc_dev_mtx);
	if (l2arc_ndev == 0) {
	mutex_exit(&l2arc_dev_mtx);
	continue;
	}
	mutex_exit(&l2arc_dev_mtx);
	begin = ddi_get_lbolt();

	/*
	* This selects the next l2arc device to write to, and in
	* doing so the next spa to feed from: dev->l2ad_spa. This
	* will return NULL if there are now no l2arc devices or if
	* they are all faulted.
	*
	* If a device is returned, its spa's config lock is also
	* held to prevent device removal. l2arc_dev_get_next()
	* will grab and release l2arc_dev_mtx.
	*/
	if ((dev = l2arc_dev_get_next()) == NULL)
	continue;

	spa = dev->l2ad_spa;
	ASSERT3P(spa, !=, NULL);

	/*
	* If the pool is read-only then force the feed thread to
	* sleep a little longer.
	*/
	if (!spa_writeable(spa)) {
	next = ddi_get_lbolt() + 5 * l2arc_feed_secs * hz;
	spa_config_exit(spa, SCL_L2ARC, dev);
	continue;
	}

	/*
	* Avoid contributing to memory pressure.
	*/
	if (l2arc_hdr_limit_reached()) {
	ARCSTAT_BUMP(arcstat_l2_abort_lowmem);
	spa_config_exit(spa, SCL_L2ARC, dev);
	continue;
	}

	ARCSTAT_BUMP(arcstat_l2_feeds);

	size = l2arc_write_size(dev);

	/*
	* Evict L2ARC buffers that will be overwritten.
	*/
	l2arc_evict(dev, size, B_FALSE);

	/*
	* Write ARC buffers.
	*/
	wrote = l2arc_write_buffers(spa, dev, size);

	/*
	* Calculate interval between writes.
	*/
	next = l2arc_write_interval(begin, size, wrote);
	spa_config_exit(spa, SCL_L2ARC, dev);
	}
	spl_fstrans_unmark(cookie);

	l2arc_thread_exit = 0;
	cv_broadcast(&l2arc_feed_thr_cv);
	CALLB_CPR_EXIT(&cpr); /* drops l2arc_feed_thr_lock */
	thread_exit();
	}

	boolean_t
	l2arc_vdev_present(vdev_t *vd)
	{
	return (l2arc_vdev_get(vd) != NULL);
	}

	/*
	* Returns the l2arc_dev_t associated with a particular vdev_t or NULL if
	* the vdev_t isn't an L2ARC device.
	*/
	l2arc_dev_t *
	l2arc_vdev_get(vdev_t *vd)
	{
	l2arc_dev_t *dev;

	mutex_enter(&l2arc_dev_mtx);
	for (dev = list_head(l2arc_dev_list); dev != NULL;
	dev = list_next(l2arc_dev_list, dev)) {
	if (dev->l2ad_vdev == vd)
	break;
	}
	mutex_exit(&l2arc_dev_mtx);

	return (dev);
	}

	static void
	l2arc_rebuild_dev(l2arc_dev_t *dev, boolean_t reopen)
	{
	l2arc_dev_hdr_phys_t *l2dhdr = dev->l2ad_dev_hdr;
	uint64_t l2dhdr_asize = dev->l2ad_dev_hdr_asize;
	spa_t *spa = dev->l2ad_spa;

	/*
	* The L2ARC has to hold at least the payload of one log block for
	* them to be restored (persistent L2ARC). The payload of a log block
	* depends on the amount of its log entries. We always write log blocks
	* with 1022 entries. How many of them are committed or restored depends
	* on the size of the L2ARC device. Thus the maximum payload of
	* one log block is 1022 * SPA_MAXBLOCKSIZE = 16GB. If the L2ARC device
	* is less than that, we reduce the amount of committed and restored
	* log entries per block so as to enable persistence.
	*/
	if (dev->l2ad_end < l2arc_rebuild_blocks_min_l2size) {
	dev->l2ad_log_entries = 0;
	} else {
	dev->l2ad_log_entries = MIN((dev->l2ad_end -
	dev->l2ad_start) >> SPA_MAXBLOCKSHIFT,
	L2ARC_LOG_BLK_MAX_ENTRIES);
	}

	/*
	* Read the device header, if an error is returned do not rebuild L2ARC.
	*/
	if (l2arc_dev_hdr_read(dev) == 0 && dev->l2ad_log_entries > 0) {
	/*
	* If we are onlining a cache device (vdev_reopen) that was
	* still present (l2arc_vdev_present()) and rebuild is enabled,
	* we should evict all ARC buffers and pointers to log blocks
	* and reclaim their space before restoring its contents to
	* L2ARC.
	*/
	if (reopen) {
	if (!l2arc_rebuild_enabled) {
	return;
	} else {
	l2arc_evict(dev, 0, B_TRUE);
	/* start a new log block */
	dev->l2ad_log_ent_idx = 0;
	dev->l2ad_log_blk_payload_asize = 0;
	dev->l2ad_log_blk_payload_start = 0;
	}
	}
	/*
	* Just mark the device as pending for a rebuild. We won't
	* be starting a rebuild in line here as it would block pool
	* import. Instead spa_load_impl will hand that off to an
	* async task which will call l2arc_spa_rebuild_start.
	*/
	dev->l2ad_rebuild = B_TRUE;
	} else if (spa_writeable(spa)) {
	/*
	* In this case TRIM the whole device if l2arc_trim_ahead > 0,
	* otherwise create a new header. We zero out the memory holding
	* the header to reset dh_start_lbps. If we TRIM the whole
	* device the new header will be written by
	* vdev_trim_l2arc_thread() at the end of the TRIM to update the
	* trim_state in the header too. When reading the header, if
	* trim_state is not VDEV_TRIM_COMPLETE and l2arc_trim_ahead > 0
	* we opt to TRIM the whole device again.
	*/
	if (l2arc_trim_ahead > 0) {
	dev->l2ad_trim_all = B_TRUE;
	} else {
	bzero(l2dhdr, l2dhdr_asize);
	l2arc_dev_hdr_update(dev);
	}
	}
	}

	/*
	* Add a vdev for use by the L2ARC. By this point the spa has already
	* validated the vdev and opened it.
	*/
	void
	l2arc_add_vdev(spa_t spa, vdev_t vd)
	{
	l2arc_dev_t *adddev;
	uint64_t l2dhdr_asize;

	ASSERT(!l2arc_vdev_present(vd));

	/*
	* Create a new l2arc device entry.
	*/
	adddev = vmem_zalloc(sizeof (l2arc_dev_t), KM_SLEEP);
	adddev->l2ad_spa = spa;
	adddev->l2ad_vdev = vd;
	/* leave extra size for an l2arc device header */
	l2dhdr_asize = adddev->l2ad_dev_hdr_asize =
	MAX(sizeof (*adddev->l2ad_dev_hdr), 1 << vd->vdev_ashift);
	adddev->l2ad_start = VDEV_LABEL_START_SIZE + l2dhdr_asize;
	adddev->l2ad_end = VDEV_LABEL_START_SIZE + vdev_get_min_asize(vd);
	ASSERT3U(adddev->l2ad_start, <, adddev->l2ad_end);
	adddev->l2ad_hand = adddev->l2ad_start;
	adddev->l2ad_evict = adddev->l2ad_start;
	adddev->l2ad_first = B_TRUE;
	adddev->l2ad_writing = B_FALSE;
	adddev->l2ad_trim_all = B_FALSE;
	list_link_init(&adddev->l2ad_node);
	adddev->l2ad_dev_hdr = kmem_zalloc(l2dhdr_asize, KM_SLEEP);

	mutex_init(&adddev->l2ad_mtx, NULL, MUTEX_DEFAULT, NULL);
	/*
	* This is a list of all ARC buffers that are still valid on the
	* device.
	*/
	list_create(&adddev->l2ad_buflist, sizeof (arc_buf_hdr_t),
	offsetof(arc_buf_hdr_t, b_l2hdr.b_l2node));

	/*
	* This is a list of pointers to log blocks that are still present
	* on the device.
	*/
	list_create(&adddev->l2ad_lbptr_list, sizeof (l2arc_lb_ptr_buf_t),
	offsetof(l2arc_lb_ptr_buf_t, node));

	vdev_space_update(vd, 0, 0, adddev->l2ad_end - adddev->l2ad_hand);
	zfs_refcount_create(&adddev->l2ad_alloc);
	zfs_refcount_create(&adddev->l2ad_lb_asize);
	zfs_refcount_create(&adddev->l2ad_lb_count);

	/*
	* Decide if dev is eligible for L2ARC rebuild or whole device
	* trimming. This has to happen before the device is added in the
	* cache device list and l2arc_dev_mtx is released. Otherwise
	* l2arc_feed_thread() might already start writing on the
	* device.
	*/
	l2arc_rebuild_dev(adddev, B_FALSE);

	/*
	* Add device to global list
	*/
	mutex_enter(&l2arc_dev_mtx);
	list_insert_head(l2arc_dev_list, adddev);
	atomic_inc_64(&l2arc_ndev);
	mutex_exit(&l2arc_dev_mtx);
	}

	/*
	* Decide if a vdev is eligible for L2ARC rebuild, called from vdev_reopen()
	* in case of onlining a cache device.
	*/
	void
	l2arc_rebuild_vdev(vdev_t *vd, boolean_t reopen)
	{
	l2arc_dev_t *dev = NULL;

	dev = l2arc_vdev_get(vd);
	ASSERT3P(dev, !=, NULL);

	/*
	* In contrast to l2arc_add_vdev() we do not have to worry about
	* l2arc_feed_thread() invalidating previous content when onlining a
	* cache device. The device parameters (l2ad*) are not cleared when
	* offlining the device and writing new buffers will not invalidate
	* all previous content. In worst case only buffers that have not had
	* their log block written to the device will be lost.
	* When onlining the cache device (ie offline->online without exporting
	* the pool in between) this happens:
	* vdev_reopen() -> vdev_open() -> l2arc_rebuild_vdev()
	* \| \|
	* vdev_is_dead() = B_FALSE l2ad_rebuild = B_TRUE
	* During the time where vdev_is_dead = B_FALSE and until l2ad_rebuild
	* is set to B_TRUE we might write additional buffers to the device.
	*/
	l2arc_rebuild_dev(dev, reopen);
	}

	/*
	* Remove a vdev from the L2ARC.
	*/
	void
	l2arc_remove_vdev(vdev_t *vd)
	{
	l2arc_dev_t *remdev = NULL;

	/*
	* Find the device by vdev
	*/
	remdev = l2arc_vdev_get(vd);
	ASSERT3P(remdev, !=, NULL);

	/*
	* Cancel any ongoing or scheduled rebuild.
	*/
	mutex_enter(&l2arc_rebuild_thr_lock);
	if (remdev->l2ad_rebuild_began == B_TRUE) {
	remdev->l2ad_rebuild_cancel = B_TRUE;
	while (remdev->l2ad_rebuild == B_TRUE)
	cv_wait(&l2arc_rebuild_thr_cv, &l2arc_rebuild_thr_lock);
	}
	mutex_exit(&l2arc_rebuild_thr_lock);

	/*
	* Remove device from global list
	*/
	mutex_enter(&l2arc_dev_mtx);
	list_remove(l2arc_dev_list, remdev);
	l2arc_dev_last = NULL; /* may have been invalidated */
	atomic_dec_64(&l2arc_ndev);
	mutex_exit(&l2arc_dev_mtx);

	/*
	* Clear all buflists and ARC references. L2ARC device flush.
	*/
	l2arc_evict(remdev, 0, B_TRUE);
	list_destroy(&remdev->l2ad_buflist);
	ASSERT(list_is_empty(&remdev->l2ad_lbptr_list));
	list_destroy(&remdev->l2ad_lbptr_list);
	mutex_destroy(&remdev->l2ad_mtx);
	zfs_refcount_destroy(&remdev->l2ad_alloc);
	zfs_refcount_destroy(&remdev->l2ad_lb_asize);
	zfs_refcount_destroy(&remdev->l2ad_lb_count);
	kmem_free(remdev->l2ad_dev_hdr, remdev->l2ad_dev_hdr_asize);
	vmem_free(remdev, sizeof (l2arc_dev_t));
	}

	void
	l2arc_init(void)
	{
	l2arc_thread_exit = 0;
	l2arc_ndev = 0;

	mutex_init(&l2arc_feed_thr_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&l2arc_feed_thr_cv, NULL, CV_DEFAULT, NULL);
	mutex_init(&l2arc_rebuild_thr_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&l2arc_rebuild_thr_cv, NULL, CV_DEFAULT, NULL);
	mutex_init(&l2arc_dev_mtx, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&l2arc_free_on_write_mtx, NULL, MUTEX_DEFAULT, NULL);

	l2arc_dev_list = &L2ARC_dev_list;
	l2arc_free_on_write = &L2ARC_free_on_write;
	list_create(l2arc_dev_list, sizeof (l2arc_dev_t),
	offsetof(l2arc_dev_t, l2ad_node));
	list_create(l2arc_free_on_write, sizeof (l2arc_data_free_t),
	offsetof(l2arc_data_free_t, l2df_list_node));
	}

	void
	l2arc_fini(void)
	{
	mutex_destroy(&l2arc_feed_thr_lock);
	cv_destroy(&l2arc_feed_thr_cv);
	mutex_destroy(&l2arc_rebuild_thr_lock);
	cv_destroy(&l2arc_rebuild_thr_cv);
	mutex_destroy(&l2arc_dev_mtx);
	mutex_destroy(&l2arc_free_on_write_mtx);

	list_destroy(l2arc_dev_list);
	list_destroy(l2arc_free_on_write);
	}

	void
	l2arc_start(void)
	{
	if (!(spa_mode_global & SPA_MODE_WRITE))
	return;

	(void) thread_create(NULL, 0, l2arc_feed_thread, NULL, 0, &p0,
	TS_RUN, defclsyspri);
	}

	void
	l2arc_stop(void)
	{
	if (!(spa_mode_global & SPA_MODE_WRITE))
	return;

	mutex_enter(&l2arc_feed_thr_lock);
	cv_signal(&l2arc_feed_thr_cv); /* kick thread out of startup */
	l2arc_thread_exit = 1;
	while (l2arc_thread_exit != 0)
	cv_wait(&l2arc_feed_thr_cv, &l2arc_feed_thr_lock);
	mutex_exit(&l2arc_feed_thr_lock);
	}

	/*
	* Punches out rebuild threads for the L2ARC devices in a spa. This should
	* be called after pool import from the spa async thread, since starting
	* these threads directly from spa_import() will make them part of the
	* "zpool import" context and delay process exit (and thus pool import).
	*/
	void
	l2arc_spa_rebuild_start(spa_t *spa)
	{
	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	/*
	* Locate the spa's l2arc devices and kick off rebuild threads.
	*/
	for (int i = 0; i < spa->spa_l2cache.sav_count; i++) {
	l2arc_dev_t *dev =
	l2arc_vdev_get(spa->spa_l2cache.sav_vdevs[i]);
	if (dev == NULL) {
	/* Don't attempt a rebuild if the vdev is UNAVAIL */
	continue;
	}
	mutex_enter(&l2arc_rebuild_thr_lock);
	if (dev->l2ad_rebuild && !dev->l2ad_rebuild_cancel) {
	dev->l2ad_rebuild_began = B_TRUE;
	(void) thread_create(NULL, 0, l2arc_dev_rebuild_thread,
	dev, 0, &p0, TS_RUN, minclsyspri);
	}
	mutex_exit(&l2arc_rebuild_thr_lock);
	}
	}

	/*
	* Main entry point for L2ARC rebuilding.
	*/
	static void
	l2arc_dev_rebuild_thread(void *arg)
	{
	l2arc_dev_t *dev = arg;

	VERIFY(!dev->l2ad_rebuild_cancel);
	VERIFY(dev->l2ad_rebuild);
	(void) l2arc_rebuild(dev);
	mutex_enter(&l2arc_rebuild_thr_lock);
	dev->l2ad_rebuild_began = B_FALSE;
	dev->l2ad_rebuild = B_FALSE;
	mutex_exit(&l2arc_rebuild_thr_lock);

	thread_exit();
	}

	/*
	* This function implements the actual L2ARC metadata rebuild. It:
	* starts reading the log block chain and restores each block's contents
	* to memory (reconstructing arc_buf_hdr_t's).
	*
	* Operation stops under any of the following conditions:
	*
	* 1) We reach the end of the log block chain.
	* 2) We encounter any error condition (cksum errors, io errors)
	*/
	static int
	l2arc_rebuild(l2arc_dev_t *dev)
	{
	vdev_t *vd = dev->l2ad_vdev;
	spa_t *spa = vd->vdev_spa;
	int err = 0;
	l2arc_dev_hdr_phys_t *l2dhdr = dev->l2ad_dev_hdr;
	l2arc_log_blk_phys_t this_lb, next_lb;
	zio_t this_io = NULL, next_io = NULL;
	l2arc_log_blkptr_t lbps[2];
	l2arc_lb_ptr_buf_t *lb_ptr_buf;
	boolean_t lock_held;

	this_lb = vmem_zalloc(sizeof (*this_lb), KM_SLEEP);
	next_lb = vmem_zalloc(sizeof (*next_lb), KM_SLEEP);

	/*
	* We prevent device removal while issuing reads to the device,
	* then during the rebuilding phases we drop this lock again so
	* that a spa_unload or device remove can be initiated - this is
	* safe, because the spa will signal us to stop before removing
	* our device and wait for us to stop.
	*/
	spa_config_enter(spa, SCL_L2ARC, vd, RW_READER);
	lock_held = B_TRUE;

	/*
	* Retrieve the persistent L2ARC device state.
	* L2BLK_GET_PSIZE returns aligned size for log blocks.
	*/
	dev->l2ad_evict = MAX(l2dhdr->dh_evict, dev->l2ad_start);
	dev->l2ad_hand = MAX(l2dhdr->dh_start_lbps[0].lbp_daddr +
	L2BLK_GET_PSIZE((&l2dhdr->dh_start_lbps[0])->lbp_prop),
	dev->l2ad_start);
	dev->l2ad_first = !!(l2dhdr->dh_flags & L2ARC_DEV_HDR_EVICT_FIRST);

	vd->vdev_trim_action_time = l2dhdr->dh_trim_action_time;
	vd->vdev_trim_state = l2dhdr->dh_trim_state;

	/*
	* In case the zfs module parameter l2arc_rebuild_enabled is false
	* we do not start the rebuild process.
	*/
	if (!l2arc_rebuild_enabled)
	goto out;

	/* Prepare the rebuild process */
	bcopy(l2dhdr->dh_start_lbps, lbps, sizeof (lbps));

	/* Start the rebuild process */
	for (;;) {
	if (!l2arc_log_blkptr_valid(dev, &lbps[0]))
	break;

	if ((err = l2arc_log_blk_read(dev, &lbps[0], &lbps[1],
	this_lb, next_lb, this_io, &next_io)) != 0)
	goto out;

	/*
	* Our memory pressure valve. If the system is running low
	* on memory, rather than swamping memory with new ARC buf
	* hdrs, we opt not to rebuild the L2ARC. At this point,
	* however, we have already set up our L2ARC dev to chain in
	* new metadata log blocks, so the user may choose to offline/
	* online the L2ARC dev at a later time (or re-import the pool)
	* to reconstruct it (when there's less memory pressure).
	*/
	if (l2arc_hdr_limit_reached()) {
	ARCSTAT_BUMP(arcstat_l2_rebuild_abort_lowmem);
	cmn_err(CE_NOTE, "System running low on memory, "
	"aborting L2ARC rebuild.");
	err = SET_ERROR(ENOMEM);
	goto out;
	}

	spa_config_exit(spa, SCL_L2ARC, vd);
	lock_held = B_FALSE;

	/*
	* Now that we know that the next_lb checks out alright, we
	* can start reconstruction from this log block.
	* L2BLK_GET_PSIZE returns aligned size for log blocks.
	*/
	uint64_t asize = L2BLK_GET_PSIZE((&lbps[0])->lbp_prop);
	l2arc_log_blk_restore(dev, this_lb, asize);

	/*
	* log block restored, include its pointer in the list of
	* pointers to log blocks present in the L2ARC device.
	*/
	lb_ptr_buf = kmem_zalloc(sizeof (l2arc_lb_ptr_buf_t), KM_SLEEP);
	lb_ptr_buf->lb_ptr = kmem_zalloc(sizeof (l2arc_log_blkptr_t),
	KM_SLEEP);
	bcopy(&lbps[0], lb_ptr_buf->lb_ptr,
	sizeof (l2arc_log_blkptr_t));
	mutex_enter(&dev->l2ad_mtx);
	list_insert_tail(&dev->l2ad_lbptr_list, lb_ptr_buf);
	ARCSTAT_INCR(arcstat_l2_log_blk_asize, asize);
	ARCSTAT_BUMP(arcstat_l2_log_blk_count);
	zfs_refcount_add_many(&dev->l2ad_lb_asize, asize, lb_ptr_buf);
	zfs_refcount_add(&dev->l2ad_lb_count, lb_ptr_buf);
	mutex_exit(&dev->l2ad_mtx);
	vdev_space_update(vd, asize, 0, 0);

	/*
	* Protection against loops of log blocks:
	*
	* l2ad_hand l2ad_evict
	* V V
	* l2ad_start \|=======================================\| l2ad_end
	* -----\|\|\|----\|\|\|---\|\|\|----\|\|\|
	* (3) (2) (1) (0)
	* ---\|\|\|---\|\|\|----\|\|\|---\|\|\|
	* (7) (6) (5) (4)
	*
	* In this situation the pointer of log block (4) passes
	* l2arc_log_blkptr_valid() but the log block should not be
	* restored as it is overwritten by the payload of log block
	* (0). Only log blocks (0)-(3) should be restored. We check
	* whether l2ad_evict lies in between the payload starting
	* offset of the next log block (lbps[1].lbp_payload_start)
	* and the payload starting offset of the present log block
	* (lbps[0].lbp_payload_start). If true and this isn't the
	* first pass, we are looping from the beginning and we should
	* stop.
	*/
	if (l2arc_range_check_overlap(lbps[1].lbp_payload_start,
	lbps[0].lbp_payload_start, dev->l2ad_evict) &&
	!dev->l2ad_first)
	goto out;

	cond_resched();
	for (;;) {
	mutex_enter(&l2arc_rebuild_thr_lock);
	if (dev->l2ad_rebuild_cancel) {
	dev->l2ad_rebuild = B_FALSE;
	cv_signal(&l2arc_rebuild_thr_cv);
	mutex_exit(&l2arc_rebuild_thr_lock);
	err = SET_ERROR(ECANCELED);
	goto out;
	}
	mutex_exit(&l2arc_rebuild_thr_lock);
	if (spa_config_tryenter(spa, SCL_L2ARC, vd,
	RW_READER)) {
	lock_held = B_TRUE;
	break;
	}
	/*
	* L2ARC config lock held by somebody in writer,
	* possibly due to them trying to remove us. They'll
	* likely to want us to shut down, so after a little
	* delay, we check l2ad_rebuild_cancel and retry
	* the lock again.
	*/
	delay(1);
	}

	/*
	* Continue with the next log block.
	*/
	lbps[0] = lbps[1];
	lbps[1] = this_lb->lb_prev_lbp;
	PTR_SWAP(this_lb, next_lb);
	this_io = next_io;
	next_io = NULL;
	}

	if (this_io != NULL)
	l2arc_log_blk_fetch_abort(this_io);
	out:
	if (next_io != NULL)
	l2arc_log_blk_fetch_abort(next_io);
	vmem_free(this_lb, sizeof (*this_lb));
	vmem_free(next_lb, sizeof (*next_lb));

	if (!l2arc_rebuild_enabled) {
	spa_history_log_internal(spa, "L2ARC rebuild", NULL,
	"disabled");
	} else if (err == 0 && zfs_refcount_count(&dev->l2ad_lb_count) > 0) {
	ARCSTAT_BUMP(arcstat_l2_rebuild_success);
	spa_history_log_internal(spa, "L2ARC rebuild", NULL,
	"successful, restored %llu blocks",
	(u_longlong_t)zfs_refcount_count(&dev->l2ad_lb_count));
	} else if (err == 0 && zfs_refcount_count(&dev->l2ad_lb_count) == 0) {
	/*
	* No error but also nothing restored, meaning the lbps array
	* in the device header points to invalid/non-present log
	* blocks. Reset the header.
	*/
	spa_history_log_internal(spa, "L2ARC rebuild", NULL,
	"no valid log blocks");
	bzero(l2dhdr, dev->l2ad_dev_hdr_asize);
	l2arc_dev_hdr_update(dev);
	} else if (err == ECANCELED) {
	/*
	* In case the rebuild was canceled do not log to spa history
	* log as the pool may be in the process of being removed.
	*/
	zfs_dbgmsg("L2ARC rebuild aborted, restored %llu blocks",
	(u_longlong_t)zfs_refcount_count(&dev->l2ad_lb_count));
	} else if (err != 0) {
	spa_history_log_internal(spa, "L2ARC rebuild", NULL,
	"aborted, restored %llu blocks",
	(u_longlong_t)zfs_refcount_count(&dev->l2ad_lb_count));
	}

	if (lock_held)
	spa_config_exit(spa, SCL_L2ARC, vd);

	return (err);
	}

	/*
	* Attempts to read the device header on the provided L2ARC device and writes
	* it to `hdr'. On success, this function returns 0, otherwise the appropriate
	* error code is returned.
	*/
	static int
	l2arc_dev_hdr_read(l2arc_dev_t *dev)
	{
	int err;
	uint64_t guid;
	l2arc_dev_hdr_phys_t *l2dhdr = dev->l2ad_dev_hdr;
	const uint64_t l2dhdr_asize = dev->l2ad_dev_hdr_asize;
	abd_t *abd;

	guid = spa_guid(dev->l2ad_vdev->vdev_spa);

	abd = abd_get_from_buf(l2dhdr, l2dhdr_asize);

	err = zio_wait(zio_read_phys(NULL, dev->l2ad_vdev,
	VDEV_LABEL_START_SIZE, l2dhdr_asize, abd,
	ZIO_CHECKSUM_LABEL, NULL, NULL, ZIO_PRIORITY_SYNC_READ,
	ZIO_FLAG_DONT_CACHE \| ZIO_FLAG_CANFAIL \|
	ZIO_FLAG_DONT_PROPAGATE \| ZIO_FLAG_DONT_RETRY \|
	ZIO_FLAG_SPECULATIVE, B_FALSE));

	abd_free(abd);

	if (err != 0) {
	ARCSTAT_BUMP(arcstat_l2_rebuild_abort_dh_errors);
	zfs_dbgmsg("L2ARC IO error (%d) while reading device header, "
	"vdev guid: %llu", err,
	(u_longlong_t)dev->l2ad_vdev->vdev_guid);
	return (err);
	}

	if (l2dhdr->dh_magic == BSWAP_64(L2ARC_DEV_HDR_MAGIC))
	byteswap_uint64_array(l2dhdr, sizeof (*l2dhdr));

	if (l2dhdr->dh_magic != L2ARC_DEV_HDR_MAGIC \|\|
	l2dhdr->dh_spa_guid != guid \|\|
	l2dhdr->dh_vdev_guid != dev->l2ad_vdev->vdev_guid \|\|
	l2dhdr->dh_version != L2ARC_PERSISTENT_VERSION \|\|
	l2dhdr->dh_log_entries != dev->l2ad_log_entries \|\|
	l2dhdr->dh_end != dev->l2ad_end \|\|
	!l2arc_range_check_overlap(dev->l2ad_start, dev->l2ad_end,
	l2dhdr->dh_evict) \|\|
	(l2dhdr->dh_trim_state != VDEV_TRIM_COMPLETE &&
	l2arc_trim_ahead > 0)) {
	/*
	* Attempt to rebuild a device containing no actual dev hdr
	* or containing a header from some other pool or from another
	* version of persistent L2ARC.
	*/
	ARCSTAT_BUMP(arcstat_l2_rebuild_abort_unsupported);
	return (SET_ERROR(ENOTSUP));
	}

	return (0);
	}

	/*
	* Reads L2ARC log blocks from storage and validates their contents.
	*
	* This function implements a simple fetcher to make sure that while
	* we're processing one buffer the L2ARC is already fetching the next
	* one in the chain.
	*
	* The arguments this_lp and next_lp point to the current and next log block
	* address in the block chain. Similarly, this_lb and next_lb hold the
	* l2arc_log_blk_phys_t's of the current and next L2ARC blk.
	*
	* The `this_io' and `next_io' arguments are used for block fetching.
	* When issuing the first blk IO during rebuild, you should pass NULL for
	* `this_io'. This function will then issue a sync IO to read the block and
	* also issue an async IO to fetch the next block in the block chain. The
	* fetched IO is returned in `next_io'. On subsequent calls to this
	* function, pass the value returned in `next_io' from the previous call
	* as `this_io' and a fresh `next_io' pointer to hold the next fetch IO.
	* Prior to the call, you should initialize your `next_io' pointer to be
	* NULL. If no fetch IO was issued, the pointer is left set at NULL.
	*
	* On success, this function returns 0, otherwise it returns an appropriate
	* error code. On error the fetching IO is aborted and cleared before
	* returning from this function. Therefore, if we return `success', the
	* caller can assume that we have taken care of cleanup of fetch IOs.
	*/
	static int
	l2arc_log_blk_read(l2arc_dev_t *dev,
	const l2arc_log_blkptr_t this_lbp, const l2arc_log_blkptr_t next_lbp,
	l2arc_log_blk_phys_t this_lb, l2arc_log_blk_phys_t next_lb,
	zio_t this_io, zio_t *next_io)
	{
	int err = 0;
	zio_cksum_t cksum;
	abd_t *abd = NULL;
	uint64_t asize;

	ASSERT(this_lbp != NULL && next_lbp != NULL);
	ASSERT(this_lb != NULL && next_lb != NULL);
	ASSERT(next_io != NULL && *next_io == NULL);
	ASSERT(l2arc_log_blkptr_valid(dev, this_lbp));

	/*
	* Check to see if we have issued the IO for this log block in a
	* previous run. If not, this is the first call, so issue it now.
	*/
	if (this_io == NULL) {
	this_io = l2arc_log_blk_fetch(dev->l2ad_vdev, this_lbp,
	this_lb);
	}

	/*
	* Peek to see if we can start issuing the next IO immediately.
	*/
	if (l2arc_log_blkptr_valid(dev, next_lbp)) {
	/*
	* Start issuing IO for the next log block early - this
	* should help keep the L2ARC device busy while we
	* decompress and restore this log block.
	*/
	*next_io = l2arc_log_blk_fetch(dev->l2ad_vdev, next_lbp,
	next_lb);
	}

	/* Wait for the IO to read this log block to complete */
	if ((err = zio_wait(this_io)) != 0) {
	ARCSTAT_BUMP(arcstat_l2_rebuild_abort_io_errors);
	zfs_dbgmsg("L2ARC IO error (%d) while reading log block, "
	"offset: %llu, vdev guid: %llu", err,
	(u_longlong_t)this_lbp->lbp_daddr,
	(u_longlong_t)dev->l2ad_vdev->vdev_guid);
	goto cleanup;
	}

	/*
	* Make sure the buffer checks out.
	* L2BLK_GET_PSIZE returns aligned size for log blocks.
	*/
	asize = L2BLK_GET_PSIZE((this_lbp)->lbp_prop);
	fletcher_4_native(this_lb, asize, NULL, &cksum);
	if (!ZIO_CHECKSUM_EQUAL(cksum, this_lbp->lbp_cksum)) {
	ARCSTAT_BUMP(arcstat_l2_rebuild_abort_cksum_lb_errors);
	zfs_dbgmsg("L2ARC log block cksum failed, offset: %llu, "
	"vdev guid: %llu, l2ad_hand: %llu, l2ad_evict: %llu",
	(u_longlong_t)this_lbp->lbp_daddr,
	(u_longlong_t)dev->l2ad_vdev->vdev_guid,
	(u_longlong_t)dev->l2ad_hand,
	(u_longlong_t)dev->l2ad_evict);
	err = SET_ERROR(ECKSUM);
	goto cleanup;
	}

	/* Now we can take our time decoding this buffer */
	switch (L2BLK_GET_COMPRESS((this_lbp)->lbp_prop)) {
	case ZIO_COMPRESS_OFF:
	break;
	case ZIO_COMPRESS_LZ4:
	abd = abd_alloc_for_io(asize, B_TRUE);
	abd_copy_from_buf_off(abd, this_lb, 0, asize);
	if ((err = zio_decompress_data(
	L2BLK_GET_COMPRESS((this_lbp)->lbp_prop),
	abd, this_lb, asize, sizeof (*this_lb), NULL)) != 0) {
	err = SET_ERROR(EINVAL);
	goto cleanup;
	}
	break;
	default:
	err = SET_ERROR(EINVAL);
	goto cleanup;
	}
	if (this_lb->lb_magic == BSWAP_64(L2ARC_LOG_BLK_MAGIC))
	byteswap_uint64_array(this_lb, sizeof (*this_lb));
	if (this_lb->lb_magic != L2ARC_LOG_BLK_MAGIC) {
	err = SET_ERROR(EINVAL);
	goto cleanup;
	}
	cleanup:
	/* Abort an in-flight fetch I/O in case of error */
	if (err != 0 && *next_io != NULL) {
	l2arc_log_blk_fetch_abort(*next_io);
	*next_io = NULL;
	}
	if (abd != NULL)
	abd_free(abd);
	return (err);
	}

	/*
	* Restores the payload of a log block to ARC. This creates empty ARC hdr
	* entries which only contain an l2arc hdr, essentially restoring the
	* buffers to their L2ARC evicted state. This function also updates space
	* usage on the L2ARC vdev to make sure it tracks restored buffers.
	*/
	static void
	l2arc_log_blk_restore(l2arc_dev_t dev, const l2arc_log_blk_phys_t lb,
	uint64_t lb_asize)
	{
	uint64_t size = 0, asize = 0;
	uint64_t log_entries = dev->l2ad_log_entries;

	/*
	* Usually arc_adapt() is called only for data, not headers, but
	* since we may allocate significant amount of memory here, let ARC
	* grow its arc_c.
	*/
	arc_adapt(log_entries * HDR_L2ONLY_SIZE, arc_l2c_only);

	for (int i = log_entries - 1; i >= 0; i--) {
	/*
	* Restore goes in the reverse temporal direction to preserve
	* correct temporal ordering of buffers in the l2ad_buflist.
	* l2arc_hdr_restore also does a list_insert_tail instead of
	* list_insert_head on the l2ad_buflist:
	*
	* LIST l2ad_buflist LIST
	* HEAD <------ (time) ------ TAIL
	* direction +-----+-----+-----+-----+-----+ direction
	* of l2arc <== \| buf \| buf \| buf \| buf \| buf \| ===> of rebuild
	* fill +-----+-----+-----+-----+-----+
	* ^ ^
	* \| \|
	* \| \|
	* l2arc_feed_thread l2arc_rebuild
	* will place new bufs here restores bufs here
	*
	* During l2arc_rebuild() the device is not used by
	* l2arc_feed_thread() as dev->l2ad_rebuild is set to true.
	*/
	size += L2BLK_GET_LSIZE((&lb->lb_entries[i])->le_prop);
	asize += vdev_psize_to_asize(dev->l2ad_vdev,
	L2BLK_GET_PSIZE((&lb->lb_entries[i])->le_prop));
	l2arc_hdr_restore(&lb->lb_entries[i], dev);
	}

	/*
	* Record rebuild stats:
	* size Logical size of restored buffers in the L2ARC
	* asize Aligned size of restored buffers in the L2ARC
	*/
	ARCSTAT_INCR(arcstat_l2_rebuild_size, size);
	ARCSTAT_INCR(arcstat_l2_rebuild_asize, asize);
	ARCSTAT_INCR(arcstat_l2_rebuild_bufs, log_entries);
	ARCSTAT_F_AVG(arcstat_l2_log_blk_avg_asize, lb_asize);
	ARCSTAT_F_AVG(arcstat_l2_data_to_meta_ratio, asize / lb_asize);
	ARCSTAT_BUMP(arcstat_l2_rebuild_log_blks);
	}

	/*
	* Restores a single ARC buf hdr from a log entry. The ARC buffer is put
	* into a state indicating that it has been evicted to L2ARC.
	*/
	static void
	l2arc_hdr_restore(const l2arc_log_ent_phys_t le, l2arc_dev_t dev)
	{
	arc_buf_hdr_t hdr, exists;
	kmutex_t *hash_lock;
	arc_buf_contents_t type = L2BLK_GET_TYPE((le)->le_prop);
	uint64_t asize;

	/*
	* Do all the allocation before grabbing any locks, this lets us
	* sleep if memory is full and we don't have to deal with failed
	* allocations.
	*/
	hdr = arc_buf_alloc_l2only(L2BLK_GET_LSIZE((le)->le_prop), type,
	dev, le->le_dva, le->le_daddr,
	L2BLK_GET_PSIZE((le)->le_prop), le->le_birth,
	L2BLK_GET_COMPRESS((le)->le_prop), le->le_complevel,
	L2BLK_GET_PROTECTED((le)->le_prop),
	L2BLK_GET_PREFETCH((le)->le_prop),
	L2BLK_GET_STATE((le)->le_prop));
	asize = vdev_psize_to_asize(dev->l2ad_vdev,
	L2BLK_GET_PSIZE((le)->le_prop));

	/*
	* vdev_space_update() has to be called before arc_hdr_destroy() to
	* avoid underflow since the latter also calls vdev_space_update().
	*/
	l2arc_hdr_arcstats_increment(hdr);
	vdev_space_update(dev->l2ad_vdev, asize, 0, 0);

	mutex_enter(&dev->l2ad_mtx);
	list_insert_tail(&dev->l2ad_buflist, hdr);
	(void) zfs_refcount_add_many(&dev->l2ad_alloc, arc_hdr_size(hdr), hdr);
	mutex_exit(&dev->l2ad_mtx);

	exists = buf_hash_insert(hdr, &hash_lock);
	if (exists) {
	/* Buffer was already cached, no need to restore it. */
	arc_hdr_destroy(hdr);
	/*
	* If the buffer is already cached, check whether it has
	* L2ARC metadata. If not, enter them and update the flag.
	* This is important is case of onlining a cache device, since
	* we previously evicted all L2ARC metadata from ARC.
	*/
	if (!HDR_HAS_L2HDR(exists)) {
	arc_hdr_set_flags(exists, ARC_FLAG_HAS_L2HDR);
	exists->b_l2hdr.b_dev = dev;
	exists->b_l2hdr.b_daddr = le->le_daddr;
	exists->b_l2hdr.b_arcs_state =
	L2BLK_GET_STATE((le)->le_prop);
	mutex_enter(&dev->l2ad_mtx);
	list_insert_tail(&dev->l2ad_buflist, exists);
	(void) zfs_refcount_add_many(&dev->l2ad_alloc,
	arc_hdr_size(exists), exists);
	mutex_exit(&dev->l2ad_mtx);
	l2arc_hdr_arcstats_increment(exists);
	vdev_space_update(dev->l2ad_vdev, asize, 0, 0);
	}
	ARCSTAT_BUMP(arcstat_l2_rebuild_bufs_precached);
	}

	mutex_exit(hash_lock);
	}

	/*
	* Starts an asynchronous read IO to read a log block. This is used in log
	* block reconstruction to start reading the next block before we are done
	* decoding and reconstructing the current block, to keep the l2arc device
	* nice and hot with read IO to process.
	* The returned zio will contain a newly allocated memory buffers for the IO
	* data which should then be freed by the caller once the zio is no longer
	* needed (i.e. due to it having completed). If you wish to abort this
	* zio, you should do so using l2arc_log_blk_fetch_abort, which takes
	* care of disposing of the allocated buffers correctly.
	*/
	static zio_t *
	l2arc_log_blk_fetch(vdev_t vd, const l2arc_log_blkptr_t lbp,
	l2arc_log_blk_phys_t *lb)
	{
	uint32_t asize;
	zio_t *pio;
	l2arc_read_callback_t *cb;

	/* L2BLK_GET_PSIZE returns aligned size for log blocks */
	asize = L2BLK_GET_PSIZE((lbp)->lbp_prop);
	ASSERT(asize <= sizeof (l2arc_log_blk_phys_t));

	cb = kmem_zalloc(sizeof (l2arc_read_callback_t), KM_SLEEP);
	cb->l2rcb_abd = abd_get_from_buf(lb, asize);
	pio = zio_root(vd->vdev_spa, l2arc_blk_fetch_done, cb,
	ZIO_FLAG_DONT_CACHE \| ZIO_FLAG_CANFAIL \| ZIO_FLAG_DONT_PROPAGATE \|
	ZIO_FLAG_DONT_RETRY);
	(void) zio_nowait(zio_read_phys(pio, vd, lbp->lbp_daddr, asize,
	cb->l2rcb_abd, ZIO_CHECKSUM_OFF, NULL, NULL,
	ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_DONT_CACHE \| ZIO_FLAG_CANFAIL \|
	ZIO_FLAG_DONT_PROPAGATE \| ZIO_FLAG_DONT_RETRY, B_FALSE));

	return (pio);
	}

	/*
	* Aborts a zio returned from l2arc_log_blk_fetch and frees the data
	* buffers allocated for it.
	*/
	static void
	l2arc_log_blk_fetch_abort(zio_t *zio)
	{
	(void) zio_wait(zio);
	}

	/*
	* Creates a zio to update the device header on an l2arc device.
	*/
	void
	l2arc_dev_hdr_update(l2arc_dev_t *dev)
	{
	l2arc_dev_hdr_phys_t *l2dhdr = dev->l2ad_dev_hdr;
	const uint64_t l2dhdr_asize = dev->l2ad_dev_hdr_asize;
	abd_t *abd;
	int err;

	VERIFY(spa_config_held(dev->l2ad_spa, SCL_STATE_ALL, RW_READER));

	l2dhdr->dh_magic = L2ARC_DEV_HDR_MAGIC;
	l2dhdr->dh_version = L2ARC_PERSISTENT_VERSION;
	l2dhdr->dh_spa_guid = spa_guid(dev->l2ad_vdev->vdev_spa);
	l2dhdr->dh_vdev_guid = dev->l2ad_vdev->vdev_guid;
	l2dhdr->dh_log_entries = dev->l2ad_log_entries;
	l2dhdr->dh_evict = dev->l2ad_evict;
	l2dhdr->dh_start = dev->l2ad_start;
	l2dhdr->dh_end = dev->l2ad_end;
	l2dhdr->dh_lb_asize = zfs_refcount_count(&dev->l2ad_lb_asize);
	l2dhdr->dh_lb_count = zfs_refcount_count(&dev->l2ad_lb_count);
	l2dhdr->dh_flags = 0;
	l2dhdr->dh_trim_action_time = dev->l2ad_vdev->vdev_trim_action_time;
	l2dhdr->dh_trim_state = dev->l2ad_vdev->vdev_trim_state;
	if (dev->l2ad_first)
	l2dhdr->dh_flags \|= L2ARC_DEV_HDR_EVICT_FIRST;

	abd = abd_get_from_buf(l2dhdr, l2dhdr_asize);

	err = zio_wait(zio_write_phys(NULL, dev->l2ad_vdev,
	VDEV_LABEL_START_SIZE, l2dhdr_asize, abd, ZIO_CHECKSUM_LABEL, NULL,
	NULL, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_CANFAIL, B_FALSE));

	abd_free(abd);

	if (err != 0) {
	zfs_dbgmsg("L2ARC IO error (%d) while writing device header, "
	"vdev guid: %llu", err,
	(u_longlong_t)dev->l2ad_vdev->vdev_guid);
	}
	}

	/*
	* Commits a log block to the L2ARC device. This routine is invoked from
	* l2arc_write_buffers when the log block fills up.
	* This function allocates some memory to temporarily hold the serialized
	* buffer to be written. This is then released in l2arc_write_done.
	*/
	static void
	l2arc_log_blk_commit(l2arc_dev_t dev, zio_t pio, l2arc_write_callback_t *cb)
	{
	l2arc_log_blk_phys_t *lb = &dev->l2ad_log_blk;
	l2arc_dev_hdr_phys_t *l2dhdr = dev->l2ad_dev_hdr;
	uint64_t psize, asize;
	zio_t *wzio;
	l2arc_lb_abd_buf_t *abd_buf;
	uint8_t *tmpbuf;
	l2arc_lb_ptr_buf_t *lb_ptr_buf;

	VERIFY3S(dev->l2ad_log_ent_idx, ==, dev->l2ad_log_entries);

	tmpbuf = zio_buf_alloc(sizeof (*lb));
	abd_buf = zio_buf_alloc(sizeof (*abd_buf));
	abd_buf->abd = abd_get_from_buf(lb, sizeof (*lb));
	lb_ptr_buf = kmem_zalloc(sizeof (l2arc_lb_ptr_buf_t), KM_SLEEP);
	lb_ptr_buf->lb_ptr = kmem_zalloc(sizeof (l2arc_log_blkptr_t), KM_SLEEP);

	/* link the buffer into the block chain */
	lb->lb_prev_lbp = l2dhdr->dh_start_lbps[1];
	lb->lb_magic = L2ARC_LOG_BLK_MAGIC;

	/*
	* l2arc_log_blk_commit() may be called multiple times during a single
	* l2arc_write_buffers() call. Save the allocated abd buffers in a list
	* so we can free them in l2arc_write_done() later on.
	*/
	list_insert_tail(&cb->l2wcb_abd_list, abd_buf);

	/* try to compress the buffer */
	psize = zio_compress_data(ZIO_COMPRESS_LZ4,
	abd_buf->abd, tmpbuf, sizeof (*lb), 0);

	/* a log block is never entirely zero */
	ASSERT(psize != 0);
	asize = vdev_psize_to_asize(dev->l2ad_vdev, psize);
	ASSERT(asize <= sizeof (*lb));

	/*
	* Update the start log block pointer in the device header to point
	* to the log block we're about to write.
	*/
	l2dhdr->dh_start_lbps[1] = l2dhdr->dh_start_lbps[0];
	l2dhdr->dh_start_lbps[0].lbp_daddr = dev->l2ad_hand;
	l2dhdr->dh_start_lbps[0].lbp_payload_asize =
	dev->l2ad_log_blk_payload_asize;
	l2dhdr->dh_start_lbps[0].lbp_payload_start =
	dev->l2ad_log_blk_payload_start;
	L2BLK_SET_LSIZE(
	(&l2dhdr->dh_start_lbps[0])->lbp_prop, sizeof (*lb));
	L2BLK_SET_PSIZE(
	(&l2dhdr->dh_start_lbps[0])->lbp_prop, asize);
	L2BLK_SET_CHECKSUM(
	(&l2dhdr->dh_start_lbps[0])->lbp_prop,
	ZIO_CHECKSUM_FLETCHER_4);
	if (asize < sizeof (*lb)) {
	/* compression succeeded */
	bzero(tmpbuf + psize, asize - psize);
	L2BLK_SET_COMPRESS(
	(&l2dhdr->dh_start_lbps[0])->lbp_prop,
	ZIO_COMPRESS_LZ4);
	} else {
	/* compression failed */
	bcopy(lb, tmpbuf, sizeof (*lb));
	L2BLK_SET_COMPRESS(
	(&l2dhdr->dh_start_lbps[0])->lbp_prop,
	ZIO_COMPRESS_OFF);
	}

	/* checksum what we're about to write */
	fletcher_4_native(tmpbuf, asize, NULL,
	&l2dhdr->dh_start_lbps[0].lbp_cksum);

	abd_free(abd_buf->abd);

	/* perform the write itself */
	abd_buf->abd = abd_get_from_buf(tmpbuf, sizeof (*lb));
	abd_take_ownership_of_buf(abd_buf->abd, B_TRUE);
	wzio = zio_write_phys(pio, dev->l2ad_vdev, dev->l2ad_hand,
	asize, abd_buf->abd, ZIO_CHECKSUM_OFF, NULL, NULL,
	ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_CANFAIL, B_FALSE);
	DTRACE_PROBE2(l2arc__write, vdev_t , dev->l2ad_vdev, zio_t , wzio);
	(void) zio_nowait(wzio);

	dev->l2ad_hand += asize;
	/*
	* Include the committed log block's pointer in the list of pointers
	* to log blocks present in the L2ARC device.
	*/
	bcopy(&l2dhdr->dh_start_lbps[0], lb_ptr_buf->lb_ptr,
	sizeof (l2arc_log_blkptr_t));
	mutex_enter(&dev->l2ad_mtx);
	list_insert_head(&dev->l2ad_lbptr_list, lb_ptr_buf);
	ARCSTAT_INCR(arcstat_l2_log_blk_asize, asize);
	ARCSTAT_BUMP(arcstat_l2_log_blk_count);
	zfs_refcount_add_many(&dev->l2ad_lb_asize, asize, lb_ptr_buf);
	zfs_refcount_add(&dev->l2ad_lb_count, lb_ptr_buf);
	mutex_exit(&dev->l2ad_mtx);
	vdev_space_update(dev->l2ad_vdev, asize, 0, 0);

	/* bump the kstats */
	ARCSTAT_INCR(arcstat_l2_write_bytes, asize);
	ARCSTAT_BUMP(arcstat_l2_log_blk_writes);
	ARCSTAT_F_AVG(arcstat_l2_log_blk_avg_asize, asize);
	ARCSTAT_F_AVG(arcstat_l2_data_to_meta_ratio,
	dev->l2ad_log_blk_payload_asize / asize);

	/* start a new log block */
	dev->l2ad_log_ent_idx = 0;
	dev->l2ad_log_blk_payload_asize = 0;
	dev->l2ad_log_blk_payload_start = 0;
	}

	/*
	* Validates an L2ARC log block address to make sure that it can be read
	* from the provided L2ARC device.
	*/
	boolean_t
	l2arc_log_blkptr_valid(l2arc_dev_t dev, const l2arc_log_blkptr_t lbp)
	{
	/* L2BLK_GET_PSIZE returns aligned size for log blocks */
	uint64_t asize = L2BLK_GET_PSIZE((lbp)->lbp_prop);
	uint64_t end = lbp->lbp_daddr + asize - 1;
	uint64_t start = lbp->lbp_payload_start;
	boolean_t evicted = B_FALSE;

	/*
	* A log block is valid if all of the following conditions are true:
	* - it fits entirely (including its payload) between l2ad_start and
	* l2ad_end
	* - it has a valid size
	* - neither the log block itself nor part of its payload was evicted
	* by l2arc_evict():
	*
	* l2ad_hand l2ad_evict
	* \| \| lbp_daddr
	* \| start \| \| end
	* \| \| \| \| \|
	* V V V V V
	* l2ad_start ============================================ l2ad_end
	* --------------------------\|\|\|\|
	* ^ ^
	* \| log block
	* payload
	*/

	evicted =
	l2arc_range_check_overlap(start, end, dev->l2ad_hand) \|\|
	l2arc_range_check_overlap(start, end, dev->l2ad_evict) \|\|
	l2arc_range_check_overlap(dev->l2ad_hand, dev->l2ad_evict, start) \|\|
	l2arc_range_check_overlap(dev->l2ad_hand, dev->l2ad_evict, end);

	return (start >= dev->l2ad_start && end <= dev->l2ad_end &&
	asize > 0 && asize <= sizeof (l2arc_log_blk_phys_t) &&
	(!evicted \|\| dev->l2ad_first));
	}

	/*
	* Inserts ARC buffer header `hdr' into the current L2ARC log block on
	* the device. The buffer being inserted must be present in L2ARC.
	* Returns B_TRUE if the L2ARC log block is full and needs to be committed
	* to L2ARC, or B_FALSE if it still has room for more ARC buffers.
	*/
	static boolean_t
	l2arc_log_blk_insert(l2arc_dev_t dev, const arc_buf_hdr_t hdr)
	{
	l2arc_log_blk_phys_t *lb = &dev->l2ad_log_blk;
	l2arc_log_ent_phys_t *le;

	if (dev->l2ad_log_entries == 0)
	return (B_FALSE);

	int index = dev->l2ad_log_ent_idx++;

	ASSERT3S(index, <, dev->l2ad_log_entries);
	ASSERT(HDR_HAS_L2HDR(hdr));

	le = &lb->lb_entries[index];
	bzero(le, sizeof (*le));
	le->le_dva = hdr->b_dva;
	le->le_birth = hdr->b_birth;
	le->le_daddr = hdr->b_l2hdr.b_daddr;
	if (index == 0)
	dev->l2ad_log_blk_payload_start = le->le_daddr;
	L2BLK_SET_LSIZE((le)->le_prop, HDR_GET_LSIZE(hdr));
	L2BLK_SET_PSIZE((le)->le_prop, HDR_GET_PSIZE(hdr));
	L2BLK_SET_COMPRESS((le)->le_prop, HDR_GET_COMPRESS(hdr));
	le->le_complevel = hdr->b_complevel;
	L2BLK_SET_TYPE((le)->le_prop, hdr->b_type);
	L2BLK_SET_PROTECTED((le)->le_prop, !!(HDR_PROTECTED(hdr)));
	L2BLK_SET_PREFETCH((le)->le_prop, !!(HDR_PREFETCH(hdr)));
	L2BLK_SET_STATE((le)->le_prop, hdr->b_l1hdr.b_state->arcs_state);

	dev->l2ad_log_blk_payload_asize += vdev_psize_to_asize(dev->l2ad_vdev,
	HDR_GET_PSIZE(hdr));

	return (dev->l2ad_log_ent_idx == dev->l2ad_log_entries);
	}

	/*
	* Checks whether a given L2ARC device address sits in a time-sequential
	* range. The trick here is that the L2ARC is a rotary buffer, so we can't
	* just do a range comparison, we need to handle the situation in which the
	* range wraps around the end of the L2ARC device. Arguments:
	* bottom -- Lower end of the range to check (written to earlier).
	* top -- Upper end of the range to check (written to later).
	* check -- The address for which we want to determine if it sits in
	* between the top and bottom.
	*
	* The 3-way conditional below represents the following cases:
	*
	* bottom < top : Sequentially ordered case:
	* <check>--------+-------------------+
	* \| (overlap here?) \|
	* L2ARC dev V V
	* \|---------------<bottom>============<top>--------------\|
	*
	* bottom > top: Looped-around case:
	* <check>--------+------------------+
	* \| (overlap here?) \|
	* L2ARC dev V V
	* \|===============<top>---------------<bottom>===========\|
	* ^ ^
	* \| (or here?) \|
	* +---------------+---------<check>
	*
	* top == bottom : Just a single address comparison.
	*/
	boolean_t
	l2arc_range_check_overlap(uint64_t bottom, uint64_t top, uint64_t check)
	{
	if (bottom < top)
	return (bottom <= check && check <= top);
	else if (bottom > top)
	return (check <= top \|\| bottom <= check);
	else
	return (check == top);
	}

	EXPORT_SYMBOL(arc_buf_size);
	EXPORT_SYMBOL(arc_write);
	EXPORT_SYMBOL(arc_read);
	EXPORT_SYMBOL(arc_buf_info);
	EXPORT_SYMBOL(arc_getbuf_func);
	EXPORT_SYMBOL(arc_add_prune_callback);
	EXPORT_SYMBOL(arc_remove_prune_callback);

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, min, param_set_arc_min,
	param_get_long, ZMOD_RW, "Min arc size");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, max, param_set_arc_max,
	param_get_long, ZMOD_RW, "Max arc size");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, meta_limit, param_set_arc_long,
	param_get_long, ZMOD_RW, "Metadata limit for arc size");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, meta_limit_percent,
	param_set_arc_long, param_get_long, ZMOD_RW,
	"Percent of arc size for arc meta limit");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, meta_min, param_set_arc_long,
	param_get_long, ZMOD_RW, "Min arc metadata");

	ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, meta_prune, INT, ZMOD_RW,
	"Meta objects to scan for prune");

	ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, meta_adjust_restarts, INT, ZMOD_RW,
	"Limit number of restarts in arc_evict_meta");

	ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, meta_strategy, INT, ZMOD_RW,
	"Meta reclaim strategy");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, grow_retry, param_set_arc_int,
	param_get_int, ZMOD_RW, "Seconds before growing arc size");

	ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, p_dampener_disable, INT, ZMOD_RW,
	"Disable arc_p adapt dampener");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, shrink_shift, param_set_arc_int,
	param_get_int, ZMOD_RW, "log2(fraction of arc to reclaim)");

	ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, pc_percent, UINT, ZMOD_RW,
	"Percent of pagecache to reclaim arc to");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, p_min_shift, param_set_arc_int,
	param_get_int, ZMOD_RW, "arc_c shift to calc min/max arc_p");

	ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, average_blocksize, INT, ZMOD_RD,
	"Target average block size");

	ZFS_MODULE_PARAM(zfs, zfs_, compressed_arc_enabled, INT, ZMOD_RW,
	"Disable compressed arc buffers");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, min_prefetch_ms, param_set_arc_int,
	param_get_int, ZMOD_RW, "Min life of prefetch block in ms");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, min_prescient_prefetch_ms,
	param_set_arc_int, param_get_int, ZMOD_RW,
	"Min life of prescient prefetched block in ms");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, write_max, ULONG, ZMOD_RW,
	"Max write bytes per interval");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, write_boost, ULONG, ZMOD_RW,
	"Extra write bytes during device warmup");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, headroom, ULONG, ZMOD_RW,
	"Number of max device writes to precache");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, headroom_boost, ULONG, ZMOD_RW,
	"Compressed l2arc_headroom multiplier");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, trim_ahead, ULONG, ZMOD_RW,
	"TRIM ahead L2ARC write size multiplier");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, feed_secs, ULONG, ZMOD_RW,
	"Seconds between L2ARC writing");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, feed_min_ms, ULONG, ZMOD_RW,
	"Min feed interval in milliseconds");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, noprefetch, INT, ZMOD_RW,
	"Skip caching prefetched buffers");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, feed_again, INT, ZMOD_RW,
	"Turbo L2ARC warmup");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, norw, INT, ZMOD_RW,
	"No reads during writes");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, meta_percent, INT, ZMOD_RW,
	"Percent of ARC size allowed for L2ARC-only headers");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, rebuild_enabled, INT, ZMOD_RW,
	"Rebuild the L2ARC when importing a pool");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, rebuild_blocks_min_l2size, ULONG, ZMOD_RW,
	"Min size in bytes to write rebuild log blocks in L2ARC");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, mfuonly, INT, ZMOD_RW,
	"Cache only MFU data from ARC into L2ARC");

	ZFS_MODULE_PARAM(zfs_l2arc, l2arc_, exclude_special, INT, ZMOD_RW,
	"If set to 1 exclude dbufs on special vdevs from being cached to "
	"L2ARC.");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, lotsfree_percent, param_set_arc_int,
	param_get_int, ZMOD_RW, "System free memory I/O throttle in bytes");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, sys_free, param_set_arc_long,
	param_get_long, ZMOD_RW, "System free memory target size in bytes");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, dnode_limit, param_set_arc_long,
	param_get_long, ZMOD_RW, "Minimum bytes of dnodes in arc");

	ZFS_MODULE_PARAM_CALL(zfs_arc, zfs_arc_, dnode_limit_percent,
	param_set_arc_long, param_get_long, ZMOD_RW,
	"Percent of ARC meta buffers for dnodes");

	ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, dnode_reduce_percent, ULONG, ZMOD_RW,
	"Percentage of excess dnodes to try to unpin");

	ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, eviction_pct, INT, ZMOD_RW,
	"When full, ARC allocation waits for eviction of this % of alloc size");

	ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, evict_batch_limit, INT, ZMOD_RW,
	"The number of headers to evict per sublist before moving to the next");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/zfs/dbuf.c b/sys/contrib/openzfs/module/zfs/dbuf.c
	index fe54da425286..1a298deb1822 100644
	--- a/sys/contrib/openzfs/module/zfs/dbuf.c
	+++ b/sys/contrib/openzfs/module/zfs/dbuf.c
	@@ -1,5117 +1,5117 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2012, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
	* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
	* Copyright (c) 2019, Klara Inc.
	* Copyright (c) 2019, Allan Jude
	*/

	#include <sys/zfs_context.h>
	#include <sys/arc.h>
	#include <sys/dmu.h>
	#include <sys/dmu_send.h>
	#include <sys/dmu_impl.h>
	#include <sys/dbuf.h>
	#include <sys/dmu_objset.h>
	#include <sys/dsl_dataset.h>
	#include <sys/dsl_dir.h>
	#include <sys/dmu_tx.h>
	#include <sys/spa.h>
	#include <sys/zio.h>
	#include <sys/dmu_zfetch.h>
	#include <sys/sa.h>
	#include <sys/sa_impl.h>
	#include <sys/zfeature.h>
	#include <sys/blkptr.h>
	#include <sys/range_tree.h>
	#include <sys/trace_zfs.h>
	#include <sys/callb.h>
	#include <sys/abd.h>
	#include <sys/vdev.h>
	#include <cityhash.h>
	#include <sys/spa_impl.h>
	#include <sys/wmsum.h>
	#include <sys/vdev_impl.h>

	kstat_t *dbuf_ksp;

	typedef struct dbuf_stats {
	/*
	* Various statistics about the size of the dbuf cache.
	*/
	kstat_named_t cache_count;
	kstat_named_t cache_size_bytes;
	kstat_named_t cache_size_bytes_max;
	/*
	* Statistics regarding the bounds on the dbuf cache size.
	*/
	kstat_named_t cache_target_bytes;
	kstat_named_t cache_lowater_bytes;
	kstat_named_t cache_hiwater_bytes;
	/*
	* Total number of dbuf cache evictions that have occurred.
	*/
	kstat_named_t cache_total_evicts;
	/*
	* The distribution of dbuf levels in the dbuf cache and
	* the total size of all dbufs at each level.
	*/
	kstat_named_t cache_levels[DN_MAX_LEVELS];
	kstat_named_t cache_levels_bytes[DN_MAX_LEVELS];
	/*
	* Statistics about the dbuf hash table.
	*/
	kstat_named_t hash_hits;
	kstat_named_t hash_misses;
	kstat_named_t hash_collisions;
	kstat_named_t hash_elements;
	kstat_named_t hash_elements_max;
	/*
	* Number of sublists containing more than one dbuf in the dbuf
	* hash table. Keep track of the longest hash chain.
	*/
	kstat_named_t hash_chains;
	kstat_named_t hash_chain_max;
	/*
	* Number of times a dbuf_create() discovers that a dbuf was
	* already created and in the dbuf hash table.
	*/
	kstat_named_t hash_insert_race;
	/*
	* Statistics about the size of the metadata dbuf cache.
	*/
	kstat_named_t metadata_cache_count;
	kstat_named_t metadata_cache_size_bytes;
	kstat_named_t metadata_cache_size_bytes_max;
	/*
	* For diagnostic purposes, this is incremented whenever we can't add
	* something to the metadata cache because it's full, and instead put
	* the data in the regular dbuf cache.
	*/
	kstat_named_t metadata_cache_overflow;
	} dbuf_stats_t;

	dbuf_stats_t dbuf_stats = {
	{ "cache_count", KSTAT_DATA_UINT64 },
	{ "cache_size_bytes", KSTAT_DATA_UINT64 },
	{ "cache_size_bytes_max", KSTAT_DATA_UINT64 },
	{ "cache_target_bytes", KSTAT_DATA_UINT64 },
	{ "cache_lowater_bytes", KSTAT_DATA_UINT64 },
	{ "cache_hiwater_bytes", KSTAT_DATA_UINT64 },
	{ "cache_total_evicts", KSTAT_DATA_UINT64 },
	{ { "cache_levels_N", KSTAT_DATA_UINT64 } },
	{ { "cache_levels_bytes_N", KSTAT_DATA_UINT64 } },
	{ "hash_hits", KSTAT_DATA_UINT64 },
	{ "hash_misses", KSTAT_DATA_UINT64 },
	{ "hash_collisions", KSTAT_DATA_UINT64 },
	{ "hash_elements", KSTAT_DATA_UINT64 },
	{ "hash_elements_max", KSTAT_DATA_UINT64 },
	{ "hash_chains", KSTAT_DATA_UINT64 },
	{ "hash_chain_max", KSTAT_DATA_UINT64 },
	{ "hash_insert_race", KSTAT_DATA_UINT64 },
	{ "metadata_cache_count", KSTAT_DATA_UINT64 },
	{ "metadata_cache_size_bytes", KSTAT_DATA_UINT64 },
	{ "metadata_cache_size_bytes_max", KSTAT_DATA_UINT64 },
	{ "metadata_cache_overflow", KSTAT_DATA_UINT64 }
	};

	struct {
	wmsum_t cache_count;
	wmsum_t cache_total_evicts;
	wmsum_t cache_levels[DN_MAX_LEVELS];
	wmsum_t cache_levels_bytes[DN_MAX_LEVELS];
	wmsum_t hash_hits;
	wmsum_t hash_misses;
	wmsum_t hash_collisions;
	wmsum_t hash_chains;
	wmsum_t hash_insert_race;
	wmsum_t metadata_cache_count;
	wmsum_t metadata_cache_overflow;
	} dbuf_sums;

	#define DBUF_STAT_INCR(stat, val) \
	wmsum_add(&dbuf_sums.stat, val);
	#define DBUF_STAT_DECR(stat, val) \
	DBUF_STAT_INCR(stat, -(val));
	#define DBUF_STAT_BUMP(stat) \
	DBUF_STAT_INCR(stat, 1);
	#define DBUF_STAT_BUMPDOWN(stat) \
	DBUF_STAT_INCR(stat, -1);
	#define DBUF_STAT_MAX(stat, v) { \
	uint64_t _m; \
	while ((v) > (_m = dbuf_stats.stat.value.ui64) && \
	(_m != atomic_cas_64(&dbuf_stats.stat.value.ui64, _m, (v))))\
	continue; \
	}

	static boolean_t dbuf_undirty(dmu_buf_impl_t db, dmu_tx_t tx);
	static void dbuf_write(dbuf_dirty_record_t dr, arc_buf_t data, dmu_tx_t *tx);
	static void dbuf_sync_leaf_verify_bonus_dnode(dbuf_dirty_record_t *dr);
	static int dbuf_read_verify_dnode_crypt(dmu_buf_impl_t *db, uint32_t flags);

	extern inline void dmu_buf_init_user(dmu_buf_user_t *dbu,
	dmu_buf_evict_func_t *evict_func_sync,
	dmu_buf_evict_func_t *evict_func_async,
	dmu_buf_t **clear_on_evict_dbufp);

	/*
	* Global data structures and functions for the dbuf cache.
	*/
	static kmem_cache_t *dbuf_kmem_cache;
	static taskq_t *dbu_evict_taskq;

	static kthread_t *dbuf_cache_evict_thread;
	static kmutex_t dbuf_evict_lock;
	static kcondvar_t dbuf_evict_cv;
	static boolean_t dbuf_evict_thread_exit;

	/*
	* There are two dbuf caches; each dbuf can only be in one of them at a time.
	*
	* 1. Cache of metadata dbufs, to help make read-heavy administrative commands
	* from /sbin/zfs run faster. The "metadata cache" specifically stores dbufs
	* that represent the metadata that describes filesystems/snapshots/
	* bookmarks/properties/etc. We only evict from this cache when we export a
	* pool, to short-circuit as much I/O as possible for all administrative
	* commands that need the metadata. There is no eviction policy for this
	* cache, because we try to only include types in it which would occupy a
	* very small amount of space per object but create a large impact on the
	* performance of these commands. Instead, after it reaches a maximum size
	* (which should only happen on very small memory systems with a very large
	* number of filesystem objects), we stop taking new dbufs into the
	* metadata cache, instead putting them in the normal dbuf cache.
	*
	* 2. LRU cache of dbufs. The dbuf cache maintains a list of dbufs that
	* are not currently held but have been recently released. These dbufs
	* are not eligible for arc eviction until they are aged out of the cache.
	* Dbufs that are aged out of the cache will be immediately destroyed and
	* become eligible for arc eviction.
	*
	* Dbufs are added to these caches once the last hold is released. If a dbuf is
	* later accessed and still exists in the dbuf cache, then it will be removed
	* from the cache and later re-added to the head of the cache.
	*
	* If a given dbuf meets the requirements for the metadata cache, it will go
	* there, otherwise it will be considered for the generic LRU dbuf cache. The
	* caches and the refcounts tracking their sizes are stored in an array indexed
	* by those caches' matching enum values (from dbuf_cached_state_t).
	*/
	typedef struct dbuf_cache {
	multilist_t cache;
	zfs_refcount_t size ____cacheline_aligned;
	} dbuf_cache_t;
	dbuf_cache_t dbuf_caches[DB_CACHE_MAX];

	/* Size limits for the caches */
	unsigned long dbuf_cache_max_bytes = ULONG_MAX;
	unsigned long dbuf_metadata_cache_max_bytes = ULONG_MAX;

	/* Set the default sizes of the caches to log2 fraction of arc size */
	int dbuf_cache_shift = 5;
	int dbuf_metadata_cache_shift = 6;

	static unsigned long dbuf_cache_target_bytes(void);
	static unsigned long dbuf_metadata_cache_target_bytes(void);

	/*
	* The LRU dbuf cache uses a three-stage eviction policy:
	* - A low water marker designates when the dbuf eviction thread
	* should stop evicting from the dbuf cache.
	* - When we reach the maximum size (aka mid water mark), we
	* signal the eviction thread to run.
	* - The high water mark indicates when the eviction thread
	* is unable to keep up with the incoming load and eviction must
	* happen in the context of the calling thread.
	*
	* The dbuf cache:
	* (max size)
	* low water mid water hi water
	* +----------------------------------------+----------+----------+
	* \| \| \| \|
	* \| \| \| \|
	* \| \| \| \|
	* \| \| \| \|
	* +----------------------------------------+----------+----------+
	* stop signal evict
	* evicting eviction directly
	* thread
	*
	* The high and low water marks indicate the operating range for the eviction
	* thread. The low water mark is, by default, 90% of the total size of the
	* cache and the high water mark is at 110% (both of these percentages can be
	* changed by setting dbuf_cache_lowater_pct and dbuf_cache_hiwater_pct,
	* respectively). The eviction thread will try to ensure that the cache remains
	* within this range by waking up every second and checking if the cache is
	* above the low water mark. The thread can also be woken up by callers adding
	* elements into the cache if the cache is larger than the mid water (i.e max
	* cache size). Once the eviction thread is woken up and eviction is required,
	* it will continue evicting buffers until it's able to reduce the cache size
	* to the low water mark. If the cache size continues to grow and hits the high
	* water mark, then callers adding elements to the cache will begin to evict
	* directly from the cache until the cache is no longer above the high water
	* mark.
	*/

	/*
	* The percentage above and below the maximum cache size.
	*/
	uint_t dbuf_cache_hiwater_pct = 10;
	uint_t dbuf_cache_lowater_pct = 10;

	/* ARGSUSED */
	static int
	dbuf_cons(void vdb, void unused, int kmflag)
	{
	dmu_buf_impl_t *db = vdb;
	bzero(db, sizeof (dmu_buf_impl_t));

	mutex_init(&db->db_mtx, NULL, MUTEX_DEFAULT, NULL);
	rw_init(&db->db_rwlock, NULL, RW_DEFAULT, NULL);
	cv_init(&db->db_changed, NULL, CV_DEFAULT, NULL);
	multilist_link_init(&db->db_cache_link);
	zfs_refcount_create(&db->db_holds);

	return (0);
	}

	/* ARGSUSED */
	static void
	dbuf_dest(void vdb, void unused)
	{
	dmu_buf_impl_t *db = vdb;
	mutex_destroy(&db->db_mtx);
	rw_destroy(&db->db_rwlock);
	cv_destroy(&db->db_changed);
	ASSERT(!multilist_link_active(&db->db_cache_link));
	zfs_refcount_destroy(&db->db_holds);
	}

	/*
	* dbuf hash table routines
	*/
	static dbuf_hash_table_t dbuf_hash_table;

	/*
	* We use Cityhash for this. It's fast, and has good hash properties without
	* requiring any large static buffers.
	*/
	static uint64_t
	dbuf_hash(void *os, uint64_t obj, uint8_t lvl, uint64_t blkid)
	{
	return (cityhash4((uintptr_t)os, obj, (uint64_t)lvl, blkid));
	}

	#define DTRACE_SET_STATE(db, why) \
	DTRACE_PROBE2(dbuf__state_change, dmu_buf_impl_t *, db, \
	const char *, why)

	#define DBUF_EQUAL(dbuf, os, obj, level, blkid) \
	((dbuf)->db.db_object == (obj) && \
	(dbuf)->db_objset == (os) && \
	(dbuf)->db_level == (level) && \
	(dbuf)->db_blkid == (blkid))

	dmu_buf_impl_t *
	dbuf_find(objset_t *os, uint64_t obj, uint8_t level, uint64_t blkid)
	{
	dbuf_hash_table_t *h = &dbuf_hash_table;
	uint64_t hv;
	uint64_t idx;
	dmu_buf_impl_t *db;

	hv = dbuf_hash(os, obj, level, blkid);
	idx = hv & h->hash_table_mask;

	mutex_enter(DBUF_HASH_MUTEX(h, idx));
	for (db = h->hash_table[idx]; db != NULL; db = db->db_hash_next) {
	if (DBUF_EQUAL(db, os, obj, level, blkid)) {
	mutex_enter(&db->db_mtx);
	if (db->db_state != DB_EVICTING) {
	mutex_exit(DBUF_HASH_MUTEX(h, idx));
	return (db);
	}
	mutex_exit(&db->db_mtx);
	}
	}
	mutex_exit(DBUF_HASH_MUTEX(h, idx));
	return (NULL);
	}

	static dmu_buf_impl_t *
	dbuf_find_bonus(objset_t *os, uint64_t object)
	{
	dnode_t *dn;
	dmu_buf_impl_t *db = NULL;

	if (dnode_hold(os, object, FTAG, &dn) == 0) {
	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	if (dn->dn_bonus != NULL) {
	db = dn->dn_bonus;
	mutex_enter(&db->db_mtx);
	}
	rw_exit(&dn->dn_struct_rwlock);
	dnode_rele(dn, FTAG);
	}
	return (db);
	}

	/*
	* Insert an entry into the hash table. If there is already an element
	* equal to elem in the hash table, then the already existing element
	* will be returned and the new element will not be inserted.
	* Otherwise returns NULL.
	*/
	static dmu_buf_impl_t *
	dbuf_hash_insert(dmu_buf_impl_t *db)
	{
	dbuf_hash_table_t *h = &dbuf_hash_table;
	objset_t *os = db->db_objset;
	uint64_t obj = db->db.db_object;
	int level = db->db_level;
	uint64_t blkid, hv, idx;
	dmu_buf_impl_t *dbf;
	uint32_t i;

	blkid = db->db_blkid;
	hv = dbuf_hash(os, obj, level, blkid);
	idx = hv & h->hash_table_mask;

	mutex_enter(DBUF_HASH_MUTEX(h, idx));
	for (dbf = h->hash_table[idx], i = 0; dbf != NULL;
	dbf = dbf->db_hash_next, i++) {
	if (DBUF_EQUAL(dbf, os, obj, level, blkid)) {
	mutex_enter(&dbf->db_mtx);
	if (dbf->db_state != DB_EVICTING) {
	mutex_exit(DBUF_HASH_MUTEX(h, idx));
	return (dbf);
	}
	mutex_exit(&dbf->db_mtx);
	}
	}

	if (i > 0) {
	DBUF_STAT_BUMP(hash_collisions);
	if (i == 1)
	DBUF_STAT_BUMP(hash_chains);

	DBUF_STAT_MAX(hash_chain_max, i);
	}

	mutex_enter(&db->db_mtx);
	db->db_hash_next = h->hash_table[idx];
	h->hash_table[idx] = db;
	mutex_exit(DBUF_HASH_MUTEX(h, idx));
	uint64_t he = atomic_inc_64_nv(&dbuf_stats.hash_elements.value.ui64);
	DBUF_STAT_MAX(hash_elements_max, he);

	return (NULL);
	}

	/*
	* This returns whether this dbuf should be stored in the metadata cache, which
	* is based on whether it's from one of the dnode types that store data related
	* to traversing dataset hierarchies.
	*/
	static boolean_t
	dbuf_include_in_metadata_cache(dmu_buf_impl_t *db)
	{
	DB_DNODE_ENTER(db);
	dmu_object_type_t type = DB_DNODE(db)->dn_type;
	DB_DNODE_EXIT(db);

	/* Check if this dbuf is one of the types we care about */
	if (DMU_OT_IS_METADATA_CACHED(type)) {
	/* If we hit this, then we set something up wrong in dmu_ot */
	ASSERT(DMU_OT_IS_METADATA(type));

	/*
	* Sanity check for small-memory systems: don't allocate too
	* much memory for this purpose.
	*/
	if (zfs_refcount_count(
	&dbuf_caches[DB_DBUF_METADATA_CACHE].size) >
	dbuf_metadata_cache_target_bytes()) {
	DBUF_STAT_BUMP(metadata_cache_overflow);
	return (B_FALSE);
	}

	return (B_TRUE);
	}

	return (B_FALSE);
	}

	/*
	* Remove an entry from the hash table. It must be in the EVICTING state.
	*/
	static void
	dbuf_hash_remove(dmu_buf_impl_t *db)
	{
	dbuf_hash_table_t *h = &dbuf_hash_table;
	uint64_t hv, idx;
	dmu_buf_impl_t dbf, *dbp;

	hv = dbuf_hash(db->db_objset, db->db.db_object,
	db->db_level, db->db_blkid);
	idx = hv & h->hash_table_mask;

	/*
	* We mustn't hold db_mtx to maintain lock ordering:
	* DBUF_HASH_MUTEX > db_mtx.
	*/
	ASSERT(zfs_refcount_is_zero(&db->db_holds));
	ASSERT(db->db_state == DB_EVICTING);
	ASSERT(!MUTEX_HELD(&db->db_mtx));

	mutex_enter(DBUF_HASH_MUTEX(h, idx));
	dbp = &h->hash_table[idx];
	while ((dbf = *dbp) != db) {
	dbp = &dbf->db_hash_next;
	ASSERT(dbf != NULL);
	}
	*dbp = db->db_hash_next;
	db->db_hash_next = NULL;
	if (h->hash_table[idx] &&
	h->hash_table[idx]->db_hash_next == NULL)
	DBUF_STAT_BUMPDOWN(hash_chains);
	mutex_exit(DBUF_HASH_MUTEX(h, idx));
	atomic_dec_64(&dbuf_stats.hash_elements.value.ui64);
	}

	typedef enum {
	DBVU_EVICTING,
	DBVU_NOT_EVICTING
	} dbvu_verify_type_t;

	static void
	dbuf_verify_user(dmu_buf_impl_t *db, dbvu_verify_type_t verify_type)
	{
	#ifdef ZFS_DEBUG
	int64_t holds;

	if (db->db_user == NULL)
	return;

	/* Only data blocks support the attachment of user data. */
	ASSERT(db->db_level == 0);

	/* Clients must resolve a dbuf before attaching user data. */
	ASSERT(db->db.db_data != NULL);
	ASSERT3U(db->db_state, ==, DB_CACHED);

	holds = zfs_refcount_count(&db->db_holds);
	if (verify_type == DBVU_EVICTING) {
	/*
	* Immediate eviction occurs when holds == dirtycnt.
	* For normal eviction buffers, holds is zero on
	* eviction, except when dbuf_fix_old_data() calls
	* dbuf_clear_data(). However, the hold count can grow
	* during eviction even though db_mtx is held (see
	* dmu_bonus_hold() for an example), so we can only
	* test the generic invariant that holds >= dirtycnt.
	*/
	ASSERT3U(holds, >=, db->db_dirtycnt);
	} else {
	if (db->db_user_immediate_evict == TRUE)
	ASSERT3U(holds, >=, db->db_dirtycnt);
	else
	ASSERT3U(holds, >, 0);
	}
	#endif
	}

	static void
	dbuf_evict_user(dmu_buf_impl_t *db)
	{
	dmu_buf_user_t *dbu = db->db_user;

	ASSERT(MUTEX_HELD(&db->db_mtx));

	if (dbu == NULL)
	return;

	dbuf_verify_user(db, DBVU_EVICTING);
	db->db_user = NULL;

	#ifdef ZFS_DEBUG
	if (dbu->dbu_clear_on_evict_dbufp != NULL)
	*dbu->dbu_clear_on_evict_dbufp = NULL;
	#endif

	/*
	* There are two eviction callbacks - one that we call synchronously
	* and one that we invoke via a taskq. The async one is useful for
	* avoiding lock order reversals and limiting stack depth.
	*
	* Note that if we have a sync callback but no async callback,
	* it's likely that the sync callback will free the structure
	* containing the dbu. In that case we need to take care to not
	* dereference dbu after calling the sync evict func.
	*/
	boolean_t has_async = (dbu->dbu_evict_func_async != NULL);

	if (dbu->dbu_evict_func_sync != NULL)
	dbu->dbu_evict_func_sync(dbu);

	if (has_async) {
	taskq_dispatch_ent(dbu_evict_taskq, dbu->dbu_evict_func_async,
	dbu, 0, &dbu->dbu_tqent);
	}
	}

	boolean_t
	dbuf_is_metadata(dmu_buf_impl_t *db)
	{
	/*
	* Consider indirect blocks and spill blocks to be meta data.
	*/
	if (db->db_level > 0 \|\| db->db_blkid == DMU_SPILL_BLKID) {
	return (B_TRUE);
	} else {
	boolean_t is_metadata;

	DB_DNODE_ENTER(db);
	is_metadata = DMU_OT_IS_METADATA(DB_DNODE(db)->dn_type);
	DB_DNODE_EXIT(db);

	return (is_metadata);
	}
	}

	/*
	* We want to exclude buffers that are on a special allocation class from
	* L2ARC.
	*/
	boolean_t
	dbuf_is_l2cacheable(dmu_buf_impl_t *db)
	{
	vdev_t *vd = NULL;
	zfs_cache_type_t cache = db->db_objset->os_secondary_cache;
	blkptr_t *bp = db->db_blkptr;

	if (bp != NULL && !BP_IS_HOLE(bp)) {
	uint64_t vdev = DVA_GET_VDEV(bp->blk_dva);
	vdev_t *rvd = db->db_objset->os_spa->spa_root_vdev;

	if (vdev < rvd->vdev_children)
	vd = rvd->vdev_child[vdev];

	if (cache == ZFS_CACHE_ALL \|\|
	(dbuf_is_metadata(db) && cache == ZFS_CACHE_METADATA)) {
	if (vd == NULL)
	return (B_TRUE);

	if ((vd->vdev_alloc_bias != VDEV_BIAS_SPECIAL &&
	vd->vdev_alloc_bias != VDEV_BIAS_DEDUP) \|\|
	l2arc_exclude_special == 0)
	return (B_TRUE);
	}
	}

	return (B_FALSE);
	}

	static inline boolean_t
	dnode_level_is_l2cacheable(blkptr_t bp, dnode_t dn, int64_t level)
	{
	vdev_t *vd = NULL;
	zfs_cache_type_t cache = dn->dn_objset->os_secondary_cache;

	if (bp != NULL && !BP_IS_HOLE(bp)) {
	uint64_t vdev = DVA_GET_VDEV(bp->blk_dva);
	vdev_t *rvd = dn->dn_objset->os_spa->spa_root_vdev;

	if (vdev < rvd->vdev_children)
	vd = rvd->vdev_child[vdev];

	if (cache == ZFS_CACHE_ALL \|\| ((level > 0 \|\|
	DMU_OT_IS_METADATA(dn->dn_handle->dnh_dnode->dn_type)) &&
	cache == ZFS_CACHE_METADATA)) {
	if (vd == NULL)
	return (B_TRUE);

	if ((vd->vdev_alloc_bias != VDEV_BIAS_SPECIAL &&
	vd->vdev_alloc_bias != VDEV_BIAS_DEDUP) \|\|
	l2arc_exclude_special == 0)
	return (B_TRUE);
	}
	}

	return (B_FALSE);
	}


	/*
	* This function must return indices evenly distributed between all
	* sublists of the multilist. This is needed due to how the dbuf eviction
	* code is laid out; dbuf_evict_thread() assumes dbufs are evenly
	* distributed between all sublists and uses this assumption when
	* deciding which sublist to evict from and how much to evict from it.
	*/
	static unsigned int
	dbuf_cache_multilist_index_func(multilist_t ml, void obj)
	{
	dmu_buf_impl_t *db = obj;

	/*
	* The assumption here, is the hash value for a given
	* dmu_buf_impl_t will remain constant throughout it's lifetime
	* (i.e. it's objset, object, level and blkid fields don't change).
	* Thus, we don't need to store the dbuf's sublist index
	* on insertion, as this index can be recalculated on removal.
	*
	* Also, the low order bits of the hash value are thought to be
	* distributed evenly. Otherwise, in the case that the multilist
	* has a power of two number of sublists, each sublists' usage
	* would not be evenly distributed. In this context full 64bit
	* division would be a waste of time, so limit it to 32 bits.
	*/
	return ((unsigned int)dbuf_hash(db->db_objset, db->db.db_object,
	db->db_level, db->db_blkid) %
	multilist_get_num_sublists(ml));
	}

	/*
	* The target size of the dbuf cache can grow with the ARC target,
	* unless limited by the tunable dbuf_cache_max_bytes.
	*/
	static inline unsigned long
	dbuf_cache_target_bytes(void)
	{
	return (MIN(dbuf_cache_max_bytes,
	arc_target_bytes() >> dbuf_cache_shift));
	}

	/*
	* The target size of the dbuf metadata cache can grow with the ARC target,
	* unless limited by the tunable dbuf_metadata_cache_max_bytes.
	*/
	static inline unsigned long
	dbuf_metadata_cache_target_bytes(void)
	{
	return (MIN(dbuf_metadata_cache_max_bytes,
	arc_target_bytes() >> dbuf_metadata_cache_shift));
	}

	static inline uint64_t
	dbuf_cache_hiwater_bytes(void)
	{
	uint64_t dbuf_cache_target = dbuf_cache_target_bytes();
	return (dbuf_cache_target +
	(dbuf_cache_target * dbuf_cache_hiwater_pct) / 100);
	}

	static inline uint64_t
	dbuf_cache_lowater_bytes(void)
	{
	uint64_t dbuf_cache_target = dbuf_cache_target_bytes();
	return (dbuf_cache_target -
	(dbuf_cache_target * dbuf_cache_lowater_pct) / 100);
	}

	static inline boolean_t
	dbuf_cache_above_lowater(void)
	{
	return (zfs_refcount_count(&dbuf_caches[DB_DBUF_CACHE].size) >
	dbuf_cache_lowater_bytes());
	}

	/*
	* Evict the oldest eligible dbuf from the dbuf cache.
	*/
	static void
	dbuf_evict_one(void)
	{
	int idx = multilist_get_random_index(&dbuf_caches[DB_DBUF_CACHE].cache);
	multilist_sublist_t *mls = multilist_sublist_lock(
	&dbuf_caches[DB_DBUF_CACHE].cache, idx);

	ASSERT(!MUTEX_HELD(&dbuf_evict_lock));

	dmu_buf_impl_t *db = multilist_sublist_tail(mls);
	while (db != NULL && mutex_tryenter(&db->db_mtx) == 0) {
	db = multilist_sublist_prev(mls, db);
	}

	DTRACE_PROBE2(dbuf__evict__one, dmu_buf_impl_t *, db,
	multilist_sublist_t *, mls);

	if (db != NULL) {
	multilist_sublist_remove(mls, db);
	multilist_sublist_unlock(mls);
	(void) zfs_refcount_remove_many(
	&dbuf_caches[DB_DBUF_CACHE].size, db->db.db_size, db);
	DBUF_STAT_BUMPDOWN(cache_levels[db->db_level]);
	DBUF_STAT_BUMPDOWN(cache_count);
	DBUF_STAT_DECR(cache_levels_bytes[db->db_level],
	db->db.db_size);
	ASSERT3U(db->db_caching_status, ==, DB_DBUF_CACHE);
	db->db_caching_status = DB_NO_CACHE;
	dbuf_destroy(db);
	DBUF_STAT_BUMP(cache_total_evicts);
	} else {
	multilist_sublist_unlock(mls);
	}
	}

	/*
	* The dbuf evict thread is responsible for aging out dbufs from the
	* cache. Once the cache has reached it's maximum size, dbufs are removed
	* and destroyed. The eviction thread will continue running until the size
	* of the dbuf cache is at or below the maximum size. Once the dbuf is aged
	* out of the cache it is destroyed and becomes eligible for arc eviction.
	*/
	/* ARGSUSED */
	static void
	dbuf_evict_thread(void *unused)
	{
	callb_cpr_t cpr;

	CALLB_CPR_INIT(&cpr, &dbuf_evict_lock, callb_generic_cpr, FTAG);

	mutex_enter(&dbuf_evict_lock);
	while (!dbuf_evict_thread_exit) {
	while (!dbuf_cache_above_lowater() && !dbuf_evict_thread_exit) {
	CALLB_CPR_SAFE_BEGIN(&cpr);
	(void) cv_timedwait_idle_hires(&dbuf_evict_cv,
	&dbuf_evict_lock, SEC2NSEC(1), MSEC2NSEC(1), 0);
	CALLB_CPR_SAFE_END(&cpr, &dbuf_evict_lock);
	}
	mutex_exit(&dbuf_evict_lock);

	/*
	* Keep evicting as long as we're above the low water mark
	* for the cache. We do this without holding the locks to
	* minimize lock contention.
	*/
	while (dbuf_cache_above_lowater() && !dbuf_evict_thread_exit) {
	dbuf_evict_one();
	}

	mutex_enter(&dbuf_evict_lock);
	}

	dbuf_evict_thread_exit = B_FALSE;
	cv_broadcast(&dbuf_evict_cv);
	CALLB_CPR_EXIT(&cpr); /* drops dbuf_evict_lock */
	thread_exit();
	}

	/*
	* Wake up the dbuf eviction thread if the dbuf cache is at its max size.
	* If the dbuf cache is at its high water mark, then evict a dbuf from the
	* dbuf cache using the callers context.
	*/
	static void
	dbuf_evict_notify(uint64_t size)
	{
	/*
	* We check if we should evict without holding the dbuf_evict_lock,
	* because it's OK to occasionally make the wrong decision here,
	* and grabbing the lock results in massive lock contention.
	*/
	if (size > dbuf_cache_target_bytes()) {
	if (size > dbuf_cache_hiwater_bytes())
	dbuf_evict_one();
	cv_signal(&dbuf_evict_cv);
	}
	}

	static int
	dbuf_kstat_update(kstat_t *ksp, int rw)
	{
	dbuf_stats_t *ds = ksp->ks_data;

	if (rw == KSTAT_WRITE)
	return (SET_ERROR(EACCES));

	ds->cache_count.value.ui64 =
	wmsum_value(&dbuf_sums.cache_count);
	ds->cache_size_bytes.value.ui64 =
	zfs_refcount_count(&dbuf_caches[DB_DBUF_CACHE].size);
	ds->cache_target_bytes.value.ui64 = dbuf_cache_target_bytes();
	ds->cache_hiwater_bytes.value.ui64 = dbuf_cache_hiwater_bytes();
	ds->cache_lowater_bytes.value.ui64 = dbuf_cache_lowater_bytes();
	ds->cache_total_evicts.value.ui64 =
	wmsum_value(&dbuf_sums.cache_total_evicts);
	for (int i = 0; i < DN_MAX_LEVELS; i++) {
	ds->cache_levels[i].value.ui64 =
	wmsum_value(&dbuf_sums.cache_levels[i]);
	ds->cache_levels_bytes[i].value.ui64 =
	wmsum_value(&dbuf_sums.cache_levels_bytes[i]);
	}
	ds->hash_hits.value.ui64 =
	wmsum_value(&dbuf_sums.hash_hits);
	ds->hash_misses.value.ui64 =
	wmsum_value(&dbuf_sums.hash_misses);
	ds->hash_collisions.value.ui64 =
	wmsum_value(&dbuf_sums.hash_collisions);
	ds->hash_chains.value.ui64 =
	wmsum_value(&dbuf_sums.hash_chains);
	ds->hash_insert_race.value.ui64 =
	wmsum_value(&dbuf_sums.hash_insert_race);
	ds->metadata_cache_count.value.ui64 =
	wmsum_value(&dbuf_sums.metadata_cache_count);
	ds->metadata_cache_size_bytes.value.ui64 = zfs_refcount_count(
	&dbuf_caches[DB_DBUF_METADATA_CACHE].size);
	ds->metadata_cache_overflow.value.ui64 =
	wmsum_value(&dbuf_sums.metadata_cache_overflow);
	return (0);
	}

	void
	dbuf_init(void)
	{
	uint64_t hsize = 1ULL << 16;
	dbuf_hash_table_t *h = &dbuf_hash_table;
	int i;

	/*
	* The hash table is big enough to fill one eighth of physical memory
	* with an average block size of zfs_arc_average_blocksize (default 8K).
	* By default, the table will take up
	* totalmem * sizeof(void*) / 8K (1MB per GB with 8-byte pointers).
	*/
	while (hsize * zfs_arc_average_blocksize < arc_all_memory() / 8)
	hsize <<= 1;

	retry:
	h->hash_table_mask = hsize - 1;
	#if defined(_KERNEL)
	/*
	* Large allocations which do not require contiguous pages
	* should be using vmem_alloc() in the linux kernel
	*/
	h->hash_table = vmem_zalloc(hsize * sizeof (void *), KM_SLEEP);
	#else
	h->hash_table = kmem_zalloc(hsize * sizeof (void *), KM_NOSLEEP);
	#endif
	if (h->hash_table == NULL) {
	/* XXX - we should really return an error instead of assert */
	ASSERT(hsize > (1ULL << 10));
	hsize >>= 1;
	goto retry;
	}

	dbuf_kmem_cache = kmem_cache_create("dmu_buf_impl_t",
	sizeof (dmu_buf_impl_t),
	0, dbuf_cons, dbuf_dest, NULL, NULL, NULL, 0);

	for (i = 0; i < DBUF_MUTEXES; i++)
	mutex_init(&h->hash_mutexes[i], NULL, MUTEX_DEFAULT, NULL);

	dbuf_stats_init(h);

	/*
	* All entries are queued via taskq_dispatch_ent(), so min/maxalloc
	* configuration is not required.
	*/
	dbu_evict_taskq = taskq_create("dbu_evict", 1, defclsyspri, 0, 0, 0);

	for (dbuf_cached_state_t dcs = 0; dcs < DB_CACHE_MAX; dcs++) {
	multilist_create(&dbuf_caches[dcs].cache,
	sizeof (dmu_buf_impl_t),
	offsetof(dmu_buf_impl_t, db_cache_link),
	dbuf_cache_multilist_index_func);
	zfs_refcount_create(&dbuf_caches[dcs].size);
	}

	dbuf_evict_thread_exit = B_FALSE;
	mutex_init(&dbuf_evict_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&dbuf_evict_cv, NULL, CV_DEFAULT, NULL);
	dbuf_cache_evict_thread = thread_create(NULL, 0, dbuf_evict_thread,
	NULL, 0, &p0, TS_RUN, minclsyspri);

	wmsum_init(&dbuf_sums.cache_count, 0);
	wmsum_init(&dbuf_sums.cache_total_evicts, 0);
	for (i = 0; i < DN_MAX_LEVELS; i++) {
	wmsum_init(&dbuf_sums.cache_levels[i], 0);
	wmsum_init(&dbuf_sums.cache_levels_bytes[i], 0);
	}
	wmsum_init(&dbuf_sums.hash_hits, 0);
	wmsum_init(&dbuf_sums.hash_misses, 0);
	wmsum_init(&dbuf_sums.hash_collisions, 0);
	wmsum_init(&dbuf_sums.hash_chains, 0);
	wmsum_init(&dbuf_sums.hash_insert_race, 0);
	wmsum_init(&dbuf_sums.metadata_cache_count, 0);
	wmsum_init(&dbuf_sums.metadata_cache_overflow, 0);

	dbuf_ksp = kstat_create("zfs", 0, "dbufstats", "misc",
	KSTAT_TYPE_NAMED, sizeof (dbuf_stats) / sizeof (kstat_named_t),
	KSTAT_FLAG_VIRTUAL);
	if (dbuf_ksp != NULL) {
	for (i = 0; i < DN_MAX_LEVELS; i++) {
	snprintf(dbuf_stats.cache_levels[i].name,
	KSTAT_STRLEN, "cache_level_%d", i);
	dbuf_stats.cache_levels[i].data_type =
	KSTAT_DATA_UINT64;
	snprintf(dbuf_stats.cache_levels_bytes[i].name,
	KSTAT_STRLEN, "cache_level_%d_bytes", i);
	dbuf_stats.cache_levels_bytes[i].data_type =
	KSTAT_DATA_UINT64;
	}
	dbuf_ksp->ks_data = &dbuf_stats;
	dbuf_ksp->ks_update = dbuf_kstat_update;
	kstat_install(dbuf_ksp);
	}
	}

	void
	dbuf_fini(void)
	{
	dbuf_hash_table_t *h = &dbuf_hash_table;
	int i;

	dbuf_stats_destroy();

	for (i = 0; i < DBUF_MUTEXES; i++)
	mutex_destroy(&h->hash_mutexes[i]);
	#if defined(_KERNEL)
	/*
	* Large allocations which do not require contiguous pages
	* should be using vmem_free() in the linux kernel
	*/
	vmem_free(h->hash_table, (h->hash_table_mask + 1) * sizeof (void *));
	#else
	kmem_free(h->hash_table, (h->hash_table_mask + 1) * sizeof (void *));
	#endif
	kmem_cache_destroy(dbuf_kmem_cache);
	taskq_destroy(dbu_evict_taskq);

	mutex_enter(&dbuf_evict_lock);
	dbuf_evict_thread_exit = B_TRUE;
	while (dbuf_evict_thread_exit) {
	cv_signal(&dbuf_evict_cv);
	cv_wait(&dbuf_evict_cv, &dbuf_evict_lock);
	}
	mutex_exit(&dbuf_evict_lock);

	mutex_destroy(&dbuf_evict_lock);
	cv_destroy(&dbuf_evict_cv);

	for (dbuf_cached_state_t dcs = 0; dcs < DB_CACHE_MAX; dcs++) {
	zfs_refcount_destroy(&dbuf_caches[dcs].size);
	multilist_destroy(&dbuf_caches[dcs].cache);
	}

	if (dbuf_ksp != NULL) {
	kstat_delete(dbuf_ksp);
	dbuf_ksp = NULL;
	}

	wmsum_fini(&dbuf_sums.cache_count);
	wmsum_fini(&dbuf_sums.cache_total_evicts);
	for (i = 0; i < DN_MAX_LEVELS; i++) {
	wmsum_fini(&dbuf_sums.cache_levels[i]);
	wmsum_fini(&dbuf_sums.cache_levels_bytes[i]);
	}
	wmsum_fini(&dbuf_sums.hash_hits);
	wmsum_fini(&dbuf_sums.hash_misses);
	wmsum_fini(&dbuf_sums.hash_collisions);
	wmsum_fini(&dbuf_sums.hash_chains);
	wmsum_fini(&dbuf_sums.hash_insert_race);
	wmsum_fini(&dbuf_sums.metadata_cache_count);
	wmsum_fini(&dbuf_sums.metadata_cache_overflow);
	}

	/*
	* Other stuff.
	*/

	#ifdef ZFS_DEBUG
	static void
	dbuf_verify(dmu_buf_impl_t *db)
	{
	dnode_t *dn;
	dbuf_dirty_record_t *dr;
	uint32_t txg_prev;

	ASSERT(MUTEX_HELD(&db->db_mtx));

	if (!(zfs_flags & ZFS_DEBUG_DBUF_VERIFY))
	return;

	ASSERT(db->db_objset != NULL);
	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	if (dn == NULL) {
	ASSERT(db->db_parent == NULL);
	ASSERT(db->db_blkptr == NULL);
	} else {
	ASSERT3U(db->db.db_object, ==, dn->dn_object);
	ASSERT3P(db->db_objset, ==, dn->dn_objset);
	ASSERT3U(db->db_level, <, dn->dn_nlevels);
	ASSERT(db->db_blkid == DMU_BONUS_BLKID \|\|
	db->db_blkid == DMU_SPILL_BLKID \|\|
	!avl_is_empty(&dn->dn_dbufs));
	}
	if (db->db_blkid == DMU_BONUS_BLKID) {
	ASSERT(dn != NULL);
	ASSERT3U(db->db.db_size, >=, dn->dn_bonuslen);
	ASSERT3U(db->db.db_offset, ==, DMU_BONUS_BLKID);
	} else if (db->db_blkid == DMU_SPILL_BLKID) {
	ASSERT(dn != NULL);
	ASSERT0(db->db.db_offset);
	} else {
	ASSERT3U(db->db.db_offset, ==, db->db_blkid * db->db.db_size);
	}

	if ((dr = list_head(&db->db_dirty_records)) != NULL) {
	ASSERT(dr->dr_dbuf == db);
	txg_prev = dr->dr_txg;
	for (dr = list_next(&db->db_dirty_records, dr); dr != NULL;
	dr = list_next(&db->db_dirty_records, dr)) {
	ASSERT(dr->dr_dbuf == db);
	ASSERT(txg_prev > dr->dr_txg);
	txg_prev = dr->dr_txg;
	}
	}

	/*
	* We can't assert that db_size matches dn_datablksz because it
	* can be momentarily different when another thread is doing
	* dnode_set_blksz().
	*/
	if (db->db_level == 0 && db->db.db_object == DMU_META_DNODE_OBJECT) {
	dr = db->db_data_pending;
	/*
	* It should only be modified in syncing context, so
	* make sure we only have one copy of the data.
	*/
	ASSERT(dr == NULL \|\| dr->dt.dl.dr_data == db->db_buf);
	}

	/* verify db->db_blkptr */
	if (db->db_blkptr) {
	if (db->db_parent == dn->dn_dbuf) {
	/* db is pointed to by the dnode */
	/* ASSERT3U(db->db_blkid, <, dn->dn_nblkptr); */
	if (DMU_OBJECT_IS_SPECIAL(db->db.db_object))
	ASSERT(db->db_parent == NULL);
	else
	ASSERT(db->db_parent != NULL);
	if (db->db_blkid != DMU_SPILL_BLKID)
	ASSERT3P(db->db_blkptr, ==,
	&dn->dn_phys->dn_blkptr[db->db_blkid]);
	} else {
	/* db is pointed to by an indirect block */
	int epb __maybe_unused = db->db_parent->db.db_size >>
	SPA_BLKPTRSHIFT;
	ASSERT3U(db->db_parent->db_level, ==, db->db_level+1);
	ASSERT3U(db->db_parent->db.db_object, ==,
	db->db.db_object);
	/*
	* dnode_grow_indblksz() can make this fail if we don't
	* have the parent's rwlock. XXX indblksz no longer
	* grows. safe to do this now?
	*/
	if (RW_LOCK_HELD(&db->db_parent->db_rwlock)) {
	ASSERT3P(db->db_blkptr, ==,
	((blkptr_t *)db->db_parent->db.db_data +
	db->db_blkid % epb));
	}
	}
	}
	if ((db->db_blkptr == NULL \|\| BP_IS_HOLE(db->db_blkptr)) &&
	(db->db_buf == NULL \|\| db->db_buf->b_data) &&
	db->db.db_data && db->db_blkid != DMU_BONUS_BLKID &&
	db->db_state != DB_FILL && !dn->dn_free_txg) {
	/*
	* If the blkptr isn't set but they have nonzero data,
	* it had better be dirty, otherwise we'll lose that
	* data when we evict this buffer.
	*
	* There is an exception to this rule for indirect blocks; in
	* this case, if the indirect block is a hole, we fill in a few
	* fields on each of the child blocks (importantly, birth time)
	* to prevent hole birth times from being lost when you
	* partially fill in a hole.
	*/
	if (db->db_dirtycnt == 0) {
	if (db->db_level == 0) {
	uint64_t *buf = db->db.db_data;
	int i;

	for (i = 0; i < db->db.db_size >> 3; i++) {
	ASSERT(buf[i] == 0);
	}
	} else {
	blkptr_t *bps = db->db.db_data;
	ASSERT3U(1 << DB_DNODE(db)->dn_indblkshift, ==,
	db->db.db_size);
	/*
	* We want to verify that all the blkptrs in the
	* indirect block are holes, but we may have
	* automatically set up a few fields for them.
	* We iterate through each blkptr and verify
	* they only have those fields set.
	*/
	for (int i = 0;
	i < db->db.db_size / sizeof (blkptr_t);
	i++) {
	blkptr_t *bp = &bps[i];
	ASSERT(ZIO_CHECKSUM_IS_ZERO(
	&bp->blk_cksum));
	ASSERT(
	DVA_IS_EMPTY(&bp->blk_dva[0]) &&
	DVA_IS_EMPTY(&bp->blk_dva[1]) &&
	DVA_IS_EMPTY(&bp->blk_dva[2]));
	ASSERT0(bp->blk_fill);
	ASSERT0(bp->blk_pad[0]);
	ASSERT0(bp->blk_pad[1]);
	ASSERT(!BP_IS_EMBEDDED(bp));
	ASSERT(BP_IS_HOLE(bp));
	ASSERT0(bp->blk_phys_birth);
	}
	}
	}
	}
	DB_DNODE_EXIT(db);
	}
	#endif

	static void
	dbuf_clear_data(dmu_buf_impl_t *db)
	{
	ASSERT(MUTEX_HELD(&db->db_mtx));
	dbuf_evict_user(db);
	ASSERT3P(db->db_buf, ==, NULL);
	db->db.db_data = NULL;
	if (db->db_state != DB_NOFILL) {
	db->db_state = DB_UNCACHED;
	DTRACE_SET_STATE(db, "clear data");
	}
	}

	static void
	dbuf_set_data(dmu_buf_impl_t db, arc_buf_t buf)
	{
	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(buf != NULL);

	db->db_buf = buf;
	ASSERT(buf->b_data != NULL);
	db->db.db_data = buf->b_data;
	}

	static arc_buf_t *
	dbuf_alloc_arcbuf(dmu_buf_impl_t *db)
	{
	spa_t *spa = db->db_objset->os_spa;

	return (arc_alloc_buf(spa, db, DBUF_GET_BUFC_TYPE(db), db->db.db_size));
	}

	/*
	* Loan out an arc_buf for read. Return the loaned arc_buf.
	*/
	arc_buf_t *
	dbuf_loan_arcbuf(dmu_buf_impl_t *db)
	{
	arc_buf_t *abuf;

	ASSERT(db->db_blkid != DMU_BONUS_BLKID);
	mutex_enter(&db->db_mtx);
	if (arc_released(db->db_buf) \|\| zfs_refcount_count(&db->db_holds) > 1) {
	int blksz = db->db.db_size;
	spa_t *spa = db->db_objset->os_spa;

	mutex_exit(&db->db_mtx);
	abuf = arc_loan_buf(spa, B_FALSE, blksz);
	bcopy(db->db.db_data, abuf->b_data, blksz);
	} else {
	abuf = db->db_buf;
	arc_loan_inuse_buf(abuf, db);
	db->db_buf = NULL;
	dbuf_clear_data(db);
	mutex_exit(&db->db_mtx);
	}
	return (abuf);
	}

	/*
	* Calculate which level n block references the data at the level 0 offset
	* provided.
	*/
	uint64_t
	dbuf_whichblock(const dnode_t *dn, const int64_t level, const uint64_t offset)
	{
	if (dn->dn_datablkshift != 0 && dn->dn_indblkshift != 0) {
	/*
	* The level n blkid is equal to the level 0 blkid divided by
	* the number of level 0s in a level n block.
	*
	* The level 0 blkid is offset >> datablkshift =
	* offset / 2^datablkshift.
	*
	* The number of level 0s in a level n is the number of block
	* pointers in an indirect block, raised to the power of level.
	* This is 2^(indblkshift - SPA_BLKPTRSHIFT)^level =
	* 2^(level*(indblkshift - SPA_BLKPTRSHIFT)).
	*
	* Thus, the level n blkid is: offset /
	* ((2^datablkshift)(2^(level(indblkshift-SPA_BLKPTRSHIFT))))
	* = offset / 2^(datablkshift + level *
	* (indblkshift - SPA_BLKPTRSHIFT))
	* = offset >> (datablkshift + level *
	* (indblkshift - SPA_BLKPTRSHIFT))
	*/

	const unsigned exp = dn->dn_datablkshift +
	level * (dn->dn_indblkshift - SPA_BLKPTRSHIFT);

	if (exp >= 8 * sizeof (offset)) {
	/* This only happens on the highest indirection level */
	ASSERT3U(level, ==, dn->dn_nlevels - 1);
	return (0);
	}

	ASSERT3U(exp, <, 8 * sizeof (offset));

	return (offset >> exp);
	} else {
	ASSERT3U(offset, <, dn->dn_datablksz);
	return (0);
	}
	}

	/*
	* This function is used to lock the parent of the provided dbuf. This should be
	* used when modifying or reading db_blkptr.
	*/
	db_lock_type_t
	dmu_buf_lock_parent(dmu_buf_impl_t db, krw_t rw, void tag)
	{
	enum db_lock_type ret = DLT_NONE;
	if (db->db_parent != NULL) {
	rw_enter(&db->db_parent->db_rwlock, rw);
	ret = DLT_PARENT;
	} else if (dmu_objset_ds(db->db_objset) != NULL) {
	rrw_enter(&dmu_objset_ds(db->db_objset)->ds_bp_rwlock, rw,
	tag);
	ret = DLT_OBJSET;
	}
	/*
	* We only return a DLT_NONE lock when it's the top-most indirect block
	* of the meta-dnode of the MOS.
	*/
	return (ret);
	}

	/*
	* We need to pass the lock type in because it's possible that the block will
	* move from being the topmost indirect block in a dnode (and thus, have no
	* parent) to not the top-most via an indirection increase. This would cause a
	* panic if we didn't pass the lock type in.
	*/
	void
	dmu_buf_unlock_parent(dmu_buf_impl_t db, db_lock_type_t type, void tag)
	{
	if (type == DLT_PARENT)
	rw_exit(&db->db_parent->db_rwlock);
	else if (type == DLT_OBJSET)
	rrw_exit(&dmu_objset_ds(db->db_objset)->ds_bp_rwlock, tag);
	}

	static void
	dbuf_read_done(zio_t zio, const zbookmark_phys_t zb, const blkptr_t *bp,
	arc_buf_t buf, void vdb)
	{
	dmu_buf_impl_t *db = vdb;

	mutex_enter(&db->db_mtx);
	ASSERT3U(db->db_state, ==, DB_READ);
	/*
	* All reads are synchronous, so we must have a hold on the dbuf
	*/
	ASSERT(zfs_refcount_count(&db->db_holds) > 0);
	ASSERT(db->db_buf == NULL);
	ASSERT(db->db.db_data == NULL);
	if (buf == NULL) {
	/* i/o error */
	ASSERT(zio == NULL \|\| zio->io_error != 0);
	ASSERT(db->db_blkid != DMU_BONUS_BLKID);
	ASSERT3P(db->db_buf, ==, NULL);
	db->db_state = DB_UNCACHED;
	DTRACE_SET_STATE(db, "i/o error");
	} else if (db->db_level == 0 && db->db_freed_in_flight) {
	/* freed in flight */
	ASSERT(zio == NULL \|\| zio->io_error == 0);
	arc_release(buf, db);
	bzero(buf->b_data, db->db.db_size);
	arc_buf_freeze(buf);
	db->db_freed_in_flight = FALSE;
	dbuf_set_data(db, buf);
	db->db_state = DB_CACHED;
	DTRACE_SET_STATE(db, "freed in flight");
	} else {
	/* success */
	ASSERT(zio == NULL \|\| zio->io_error == 0);
	dbuf_set_data(db, buf);
	db->db_state = DB_CACHED;
	DTRACE_SET_STATE(db, "successful read");
	}
	cv_broadcast(&db->db_changed);
	dbuf_rele_and_unlock(db, NULL, B_FALSE);
	}

	/*
	* Shortcut for performing reads on bonus dbufs. Returns
	* an error if we fail to verify the dnode associated with
	* a decrypted block. Otherwise success.
	*/
	static int
	dbuf_read_bonus(dmu_buf_impl_t db, dnode_t dn, uint32_t flags)
	{
	int bonuslen, max_bonuslen, err;

	err = dbuf_read_verify_dnode_crypt(db, flags);
	if (err)
	return (err);

	bonuslen = MIN(dn->dn_bonuslen, dn->dn_phys->dn_bonuslen);
	max_bonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots);
	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(DB_DNODE_HELD(db));
	ASSERT3U(bonuslen, <=, db->db.db_size);
	db->db.db_data = kmem_alloc(max_bonuslen, KM_SLEEP);
	arc_space_consume(max_bonuslen, ARC_SPACE_BONUS);
	if (bonuslen < max_bonuslen)
	bzero(db->db.db_data, max_bonuslen);
	if (bonuslen)
	bcopy(DN_BONUS(dn->dn_phys), db->db.db_data, bonuslen);
	db->db_state = DB_CACHED;
	DTRACE_SET_STATE(db, "bonus buffer filled");
	return (0);
	}

	static void
	dbuf_handle_indirect_hole(dmu_buf_impl_t db, dnode_t dn)
	{
	blkptr_t *bps = db->db.db_data;
	uint32_t indbs = 1ULL << dn->dn_indblkshift;
	int n_bps = indbs >> SPA_BLKPTRSHIFT;

	for (int i = 0; i < n_bps; i++) {
	blkptr_t *bp = &bps[i];

	ASSERT3U(BP_GET_LSIZE(db->db_blkptr), ==, indbs);
	BP_SET_LSIZE(bp, BP_GET_LEVEL(db->db_blkptr) == 1 ?
	dn->dn_datablksz : BP_GET_LSIZE(db->db_blkptr));
	BP_SET_TYPE(bp, BP_GET_TYPE(db->db_blkptr));
	BP_SET_LEVEL(bp, BP_GET_LEVEL(db->db_blkptr) - 1);
	BP_SET_BIRTH(bp, db->db_blkptr->blk_birth, 0);
	}
	}

	/*
	* Handle reads on dbufs that are holes, if necessary. This function
	* requires that the dbuf's mutex is held. Returns success (0) if action
	* was taken, ENOENT if no action was taken.
	*/
	static int
	dbuf_read_hole(dmu_buf_impl_t db, dnode_t dn, uint32_t flags)
	{
	ASSERT(MUTEX_HELD(&db->db_mtx));

	int is_hole = db->db_blkptr == NULL \|\| BP_IS_HOLE(db->db_blkptr);
	/*
	* For level 0 blocks only, if the above check fails:
	* Recheck BP_IS_HOLE() after dnode_block_freed() in case dnode_sync()
	* processes the delete record and clears the bp while we are waiting
	* for the dn_mtx (resulting in a "no" from block_freed).
	*/
	if (!is_hole && db->db_level == 0) {
	is_hole = dnode_block_freed(dn, db->db_blkid) \|\|
	BP_IS_HOLE(db->db_blkptr);
	}

	if (is_hole) {
	dbuf_set_data(db, dbuf_alloc_arcbuf(db));
	bzero(db->db.db_data, db->db.db_size);

	if (db->db_blkptr != NULL && db->db_level > 0 &&
	BP_IS_HOLE(db->db_blkptr) &&
	db->db_blkptr->blk_birth != 0) {
	dbuf_handle_indirect_hole(db, dn);
	}
	db->db_state = DB_CACHED;
	DTRACE_SET_STATE(db, "hole read satisfied");
	return (0);
	}
	return (ENOENT);
	}

	/*
	* This function ensures that, when doing a decrypting read of a block,
	* we make sure we have decrypted the dnode associated with it. We must do
	* this so that we ensure we are fully authenticating the checksum-of-MACs
	* tree from the root of the objset down to this block. Indirect blocks are
	* always verified against their secure checksum-of-MACs assuming that the
	* dnode containing them is correct. Now that we are doing a decrypting read,
	* we can be sure that the key is loaded and verify that assumption. This is
	* especially important considering that we always read encrypted dnode
	* blocks as raw data (without verifying their MACs) to start, and
	* decrypt / authenticate them when we need to read an encrypted bonus buffer.
	*/
	static int
	dbuf_read_verify_dnode_crypt(dmu_buf_impl_t *db, uint32_t flags)
	{
	int err = 0;
	objset_t *os = db->db_objset;
	arc_buf_t *dnode_abuf;
	dnode_t *dn;
	zbookmark_phys_t zb;

	ASSERT(MUTEX_HELD(&db->db_mtx));

	if (!os->os_encrypted \|\| os->os_raw_receive \|\|
	(flags & DB_RF_NO_DECRYPT) != 0)
	return (0);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	dnode_abuf = (dn->dn_dbuf != NULL) ? dn->dn_dbuf->db_buf : NULL;

	if (dnode_abuf == NULL \|\| !arc_is_encrypted(dnode_abuf)) {
	DB_DNODE_EXIT(db);
	return (0);
	}

	SET_BOOKMARK(&zb, dmu_objset_id(os),
	DMU_META_DNODE_OBJECT, 0, dn->dn_dbuf->db_blkid);
	err = arc_untransform(dnode_abuf, os->os_spa, &zb, B_TRUE);

	/*
	* An error code of EACCES tells us that the key is still not
	* available. This is ok if we are only reading authenticated
	* (and therefore non-encrypted) blocks.
	*/
	if (err == EACCES && ((db->db_blkid != DMU_BONUS_BLKID &&
	!DMU_OT_IS_ENCRYPTED(dn->dn_type)) \|\|
	(db->db_blkid == DMU_BONUS_BLKID &&
	!DMU_OT_IS_ENCRYPTED(dn->dn_bonustype))))
	err = 0;

	DB_DNODE_EXIT(db);

	return (err);
	}

	/*
	* Drops db_mtx and the parent lock specified by dblt and tag before
	* returning.
	*/
	static int
	dbuf_read_impl(dmu_buf_impl_t db, zio_t zio, uint32_t flags,
	db_lock_type_t dblt, void *tag)
	{
	dnode_t *dn;
	zbookmark_phys_t zb;
	uint32_t aflags = ARC_FLAG_NOWAIT;
	int err, zio_flags;

	err = zio_flags = 0;
	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	ASSERT(!zfs_refcount_is_zero(&db->db_holds));
	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(db->db_state == DB_UNCACHED);
	ASSERT(db->db_buf == NULL);
	ASSERT(db->db_parent == NULL \|\|
	RW_LOCK_HELD(&db->db_parent->db_rwlock));

	if (db->db_blkid == DMU_BONUS_BLKID) {
	err = dbuf_read_bonus(db, dn, flags);
	goto early_unlock;
	}

	err = dbuf_read_hole(db, dn, flags);
	if (err == 0)
	goto early_unlock;

	/*
	* Any attempt to read a redacted block should result in an error. This
	* will never happen under normal conditions, but can be useful for
	* debugging purposes.
	*/
	if (BP_IS_REDACTED(db->db_blkptr)) {
	ASSERT(dsl_dataset_feature_is_active(
	db->db_objset->os_dsl_dataset,
	SPA_FEATURE_REDACTED_DATASETS));
	err = SET_ERROR(EIO);
	goto early_unlock;
	}

	SET_BOOKMARK(&zb, dmu_objset_id(db->db_objset),
	db->db.db_object, db->db_level, db->db_blkid);

	/*
	* All bps of an encrypted os should have the encryption bit set.
	* If this is not true it indicates tampering and we report an error.
	*/
	if (db->db_objset->os_encrypted && !BP_USES_CRYPT(db->db_blkptr)) {
	spa_log_error(db->db_objset->os_spa, &zb);
	zfs_panic_recover("unencrypted block in encrypted "
	"object set %llu", dmu_objset_id(db->db_objset));
	err = SET_ERROR(EIO);
	goto early_unlock;
	}

	err = dbuf_read_verify_dnode_crypt(db, flags);
	if (err != 0)
	goto early_unlock;

	DB_DNODE_EXIT(db);

	db->db_state = DB_READ;
	DTRACE_SET_STATE(db, "read issued");
	mutex_exit(&db->db_mtx);

	if (dbuf_is_l2cacheable(db))
	aflags \|= ARC_FLAG_L2CACHE;

	dbuf_add_ref(db, NULL);

	zio_flags = (flags & DB_RF_CANFAIL) ?
	ZIO_FLAG_CANFAIL : ZIO_FLAG_MUSTSUCCEED;

	if ((flags & DB_RF_NO_DECRYPT) && BP_IS_PROTECTED(db->db_blkptr))
	zio_flags \|= ZIO_FLAG_RAW;
	/*
	* The zio layer will copy the provided blkptr later, but we need to
	* do this now so that we can release the parent's rwlock. We have to
	* do that now so that if dbuf_read_done is called synchronously (on
	* an l1 cache hit) we don't acquire the db_mtx while holding the
	* parent's rwlock, which would be a lock ordering violation.
	*/
	blkptr_t bp = *db->db_blkptr;
	dmu_buf_unlock_parent(db, dblt, tag);
	(void) arc_read(zio, db->db_objset->os_spa, &bp,
	dbuf_read_done, db, ZIO_PRIORITY_SYNC_READ, zio_flags,
	&aflags, &zb);
	return (err);
	early_unlock:
	DB_DNODE_EXIT(db);
	mutex_exit(&db->db_mtx);
	dmu_buf_unlock_parent(db, dblt, tag);
	return (err);
	}

	/*
	* This is our just-in-time copy function. It makes a copy of buffers that
	* have been modified in a previous transaction group before we access them in
	* the current active group.
	*
	* This function is used in three places: when we are dirtying a buffer for the
	* first time in a txg, when we are freeing a range in a dnode that includes
	* this buffer, and when we are accessing a buffer which was received compressed
	* and later referenced in a WRITE_BYREF record.
	*
	* Note that when we are called from dbuf_free_range() we do not put a hold on
	* the buffer, we just traverse the active dbuf list for the dnode.
	*/
	static void
	dbuf_fix_old_data(dmu_buf_impl_t *db, uint64_t txg)
	{
	dbuf_dirty_record_t *dr = list_head(&db->db_dirty_records);

	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(db->db.db_data != NULL);
	ASSERT(db->db_level == 0);
	ASSERT(db->db.db_object != DMU_META_DNODE_OBJECT);

	if (dr == NULL \|\|
	(dr->dt.dl.dr_data !=
	((db->db_blkid == DMU_BONUS_BLKID) ? db->db.db_data : db->db_buf)))
	return;

	/*
	* If the last dirty record for this dbuf has not yet synced
	* and its referencing the dbuf data, either:
	* reset the reference to point to a new copy,
	* or (if there a no active holders)
	* just null out the current db_data pointer.
	*/
	ASSERT3U(dr->dr_txg, >=, txg - 2);
	if (db->db_blkid == DMU_BONUS_BLKID) {
	dnode_t *dn = DB_DNODE(db);
	int bonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots);
	dr->dt.dl.dr_data = kmem_alloc(bonuslen, KM_SLEEP);
	arc_space_consume(bonuslen, ARC_SPACE_BONUS);
	bcopy(db->db.db_data, dr->dt.dl.dr_data, bonuslen);
	} else if (zfs_refcount_count(&db->db_holds) > db->db_dirtycnt) {
	dnode_t *dn = DB_DNODE(db);
	int size = arc_buf_size(db->db_buf);
	arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
	spa_t *spa = db->db_objset->os_spa;
	enum zio_compress compress_type =
	arc_get_compression(db->db_buf);
	uint8_t complevel = arc_get_complevel(db->db_buf);

	if (arc_is_encrypted(db->db_buf)) {
	boolean_t byteorder;
	uint8_t salt[ZIO_DATA_SALT_LEN];
	uint8_t iv[ZIO_DATA_IV_LEN];
	uint8_t mac[ZIO_DATA_MAC_LEN];

	arc_get_raw_params(db->db_buf, &byteorder, salt,
	iv, mac);
	dr->dt.dl.dr_data = arc_alloc_raw_buf(spa, db,
	dmu_objset_id(dn->dn_objset), byteorder, salt, iv,
	mac, dn->dn_type, size, arc_buf_lsize(db->db_buf),
	compress_type, complevel);
	} else if (compress_type != ZIO_COMPRESS_OFF) {
	ASSERT3U(type, ==, ARC_BUFC_DATA);
	dr->dt.dl.dr_data = arc_alloc_compressed_buf(spa, db,
	size, arc_buf_lsize(db->db_buf), compress_type,
	complevel);
	} else {
	dr->dt.dl.dr_data = arc_alloc_buf(spa, db, type, size);
	}
	bcopy(db->db.db_data, dr->dt.dl.dr_data->b_data, size);
	} else {
	db->db_buf = NULL;
	dbuf_clear_data(db);
	}
	}

	int
	dbuf_read(dmu_buf_impl_t db, zio_t zio, uint32_t flags)
	{
	int err = 0;
	boolean_t prefetch;
	dnode_t *dn;

	/*
	* We don't have to hold the mutex to check db_state because it
	* can't be freed while we have a hold on the buffer.
	*/
	ASSERT(!zfs_refcount_is_zero(&db->db_holds));

	if (db->db_state == DB_NOFILL)
	return (SET_ERROR(EIO));

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);

	prefetch = db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID &&
	(flags & DB_RF_NOPREFETCH) == 0 && dn != NULL &&
	DBUF_IS_CACHEABLE(db);

	mutex_enter(&db->db_mtx);
	if (db->db_state == DB_CACHED) {
	spa_t *spa = dn->dn_objset->os_spa;

	/*
	* Ensure that this block's dnode has been decrypted if
	* the caller has requested decrypted data.
	*/
	err = dbuf_read_verify_dnode_crypt(db, flags);

	/*
	* If the arc buf is compressed or encrypted and the caller
	* requested uncompressed data, we need to untransform it
	* before returning. We also call arc_untransform() on any
	* unauthenticated blocks, which will verify their MAC if
	* the key is now available.
	*/
	if (err == 0 && db->db_buf != NULL &&
	(flags & DB_RF_NO_DECRYPT) == 0 &&
	(arc_is_encrypted(db->db_buf) \|\|
	arc_is_unauthenticated(db->db_buf) \|\|
	arc_get_compression(db->db_buf) != ZIO_COMPRESS_OFF)) {
	zbookmark_phys_t zb;

	SET_BOOKMARK(&zb, dmu_objset_id(db->db_objset),
	db->db.db_object, db->db_level, db->db_blkid);
	dbuf_fix_old_data(db, spa_syncing_txg(spa));
	err = arc_untransform(db->db_buf, spa, &zb, B_FALSE);
	dbuf_set_data(db, db->db_buf);
	}
	mutex_exit(&db->db_mtx);
	if (err == 0 && prefetch) {
	dmu_zfetch(&dn->dn_zfetch, db->db_blkid, 1, B_TRUE,
	B_FALSE, flags & DB_RF_HAVESTRUCT);
	}
	DB_DNODE_EXIT(db);
	DBUF_STAT_BUMP(hash_hits);
	} else if (db->db_state == DB_UNCACHED) {
	spa_t *spa = dn->dn_objset->os_spa;
	boolean_t need_wait = B_FALSE;

	db_lock_type_t dblt = dmu_buf_lock_parent(db, RW_READER, FTAG);

	if (zio == NULL &&
	db->db_blkptr != NULL && !BP_IS_HOLE(db->db_blkptr)) {
	zio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL);
	need_wait = B_TRUE;
	}
	err = dbuf_read_impl(db, zio, flags, dblt, FTAG);
	/*
	* dbuf_read_impl has dropped db_mtx and our parent's rwlock
	* for us
	*/
	if (!err && prefetch) {
	dmu_zfetch(&dn->dn_zfetch, db->db_blkid, 1, B_TRUE,
	db->db_state != DB_CACHED,
	flags & DB_RF_HAVESTRUCT);
	}

	DB_DNODE_EXIT(db);
	DBUF_STAT_BUMP(hash_misses);

	/*
	* If we created a zio_root we must execute it to avoid
	* leaking it, even if it isn't attached to any work due
	* to an error in dbuf_read_impl().
	*/
	if (need_wait) {
	if (err == 0)
	err = zio_wait(zio);
	else
	VERIFY0(zio_wait(zio));
	}
	} else {
	/*
	* Another reader came in while the dbuf was in flight
	* between UNCACHED and CACHED. Either a writer will finish
	* writing the buffer (sending the dbuf to CACHED) or the
	* first reader's request will reach the read_done callback
	* and send the dbuf to CACHED. Otherwise, a failure
	* occurred and the dbuf went to UNCACHED.
	*/
	mutex_exit(&db->db_mtx);
	if (prefetch) {
	dmu_zfetch(&dn->dn_zfetch, db->db_blkid, 1, B_TRUE,
	B_TRUE, flags & DB_RF_HAVESTRUCT);
	}
	DB_DNODE_EXIT(db);
	DBUF_STAT_BUMP(hash_misses);

	/* Skip the wait per the caller's request. */
	if ((flags & DB_RF_NEVERWAIT) == 0) {
	mutex_enter(&db->db_mtx);
	while (db->db_state == DB_READ \|\|
	db->db_state == DB_FILL) {
	ASSERT(db->db_state == DB_READ \|\|
	(flags & DB_RF_HAVESTRUCT) == 0);
	DTRACE_PROBE2(blocked__read, dmu_buf_impl_t *,
	db, zio_t *, zio);
	cv_wait(&db->db_changed, &db->db_mtx);
	}
	if (db->db_state == DB_UNCACHED)
	err = SET_ERROR(EIO);
	mutex_exit(&db->db_mtx);
	}
	}

	return (err);
	}

	static void
	dbuf_noread(dmu_buf_impl_t *db)
	{
	ASSERT(!zfs_refcount_is_zero(&db->db_holds));
	ASSERT(db->db_blkid != DMU_BONUS_BLKID);
	mutex_enter(&db->db_mtx);
	while (db->db_state == DB_READ \|\| db->db_state == DB_FILL)
	cv_wait(&db->db_changed, &db->db_mtx);
	if (db->db_state == DB_UNCACHED) {
	ASSERT(db->db_buf == NULL);
	ASSERT(db->db.db_data == NULL);
	dbuf_set_data(db, dbuf_alloc_arcbuf(db));
	db->db_state = DB_FILL;
	DTRACE_SET_STATE(db, "assigning filled buffer");
	} else if (db->db_state == DB_NOFILL) {
	dbuf_clear_data(db);
	} else {
	ASSERT3U(db->db_state, ==, DB_CACHED);
	}
	mutex_exit(&db->db_mtx);
	}

	void
	dbuf_unoverride(dbuf_dirty_record_t *dr)
	{
	dmu_buf_impl_t *db = dr->dr_dbuf;
	blkptr_t *bp = &dr->dt.dl.dr_overridden_by;
	uint64_t txg = dr->dr_txg;

	ASSERT(MUTEX_HELD(&db->db_mtx));
	/*
	* This assert is valid because dmu_sync() expects to be called by
	* a zilog's get_data while holding a range lock. This call only
	* comes from dbuf_dirty() callers who must also hold a range lock.
	*/
	ASSERT(dr->dt.dl.dr_override_state != DR_IN_DMU_SYNC);
	ASSERT(db->db_level == 0);

	if (db->db_blkid == DMU_BONUS_BLKID \|\|
	dr->dt.dl.dr_override_state == DR_NOT_OVERRIDDEN)
	return;

	ASSERT(db->db_data_pending != dr);

	/* free this block */
	if (!BP_IS_HOLE(bp) && !dr->dt.dl.dr_nopwrite)
	zio_free(db->db_objset->os_spa, txg, bp);

	dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN;
	dr->dt.dl.dr_nopwrite = B_FALSE;
	dr->dt.dl.dr_has_raw_params = B_FALSE;

	/*
	* Release the already-written buffer, so we leave it in
	* a consistent dirty state. Note that all callers are
	* modifying the buffer, so they will immediately do
	* another (redundant) arc_release(). Therefore, leave
	* the buf thawed to save the effort of freezing &
	* immediately re-thawing it.
	*/
	arc_release(dr->dt.dl.dr_data, db);
	}

	/*
	* Evict (if its unreferenced) or clear (if its referenced) any level-0
	* data blocks in the free range, so that any future readers will find
	* empty blocks.
	*/
	void
	dbuf_free_range(dnode_t *dn, uint64_t start_blkid, uint64_t end_blkid,
	dmu_tx_t *tx)
	{
	dmu_buf_impl_t *db_search;
	dmu_buf_impl_t db, db_next;
	uint64_t txg = tx->tx_txg;
	avl_index_t where;
	dbuf_dirty_record_t *dr;

	if (end_blkid > dn->dn_maxblkid &&
	!(start_blkid == DMU_SPILL_BLKID \|\| end_blkid == DMU_SPILL_BLKID))
	end_blkid = dn->dn_maxblkid;
	dprintf_dnode(dn, "start=%llu end=%llu\n", (u_longlong_t)start_blkid,
	(u_longlong_t)end_blkid);

	db_search = kmem_alloc(sizeof (dmu_buf_impl_t), KM_SLEEP);
	db_search->db_level = 0;
	db_search->db_blkid = start_blkid;
	db_search->db_state = DB_SEARCH;

	mutex_enter(&dn->dn_dbufs_mtx);
	db = avl_find(&dn->dn_dbufs, db_search, &where);
	ASSERT3P(db, ==, NULL);

	db = avl_nearest(&dn->dn_dbufs, where, AVL_AFTER);

	for (; db != NULL; db = db_next) {
	db_next = AVL_NEXT(&dn->dn_dbufs, db);
	ASSERT(db->db_blkid != DMU_BONUS_BLKID);

	if (db->db_level != 0 \|\| db->db_blkid > end_blkid) {
	break;
	}
	ASSERT3U(db->db_blkid, >=, start_blkid);

	/* found a level 0 buffer in the range */
	mutex_enter(&db->db_mtx);
	if (dbuf_undirty(db, tx)) {
	/* mutex has been dropped and dbuf destroyed */
	continue;
	}

	if (db->db_state == DB_UNCACHED \|\|
	db->db_state == DB_NOFILL \|\|
	db->db_state == DB_EVICTING) {
	ASSERT(db->db.db_data == NULL);
	mutex_exit(&db->db_mtx);
	continue;
	}
	if (db->db_state == DB_READ \|\| db->db_state == DB_FILL) {
	/* will be handled in dbuf_read_done or dbuf_rele */
	db->db_freed_in_flight = TRUE;
	mutex_exit(&db->db_mtx);
	continue;
	}
	if (zfs_refcount_count(&db->db_holds) == 0) {
	ASSERT(db->db_buf);
	dbuf_destroy(db);
	continue;
	}
	/* The dbuf is referenced */

	dr = list_head(&db->db_dirty_records);
	if (dr != NULL) {
	if (dr->dr_txg == txg) {
	/*
	* This buffer is "in-use", re-adjust the file
	* size to reflect that this buffer may
	* contain new data when we sync.
	*/
	if (db->db_blkid != DMU_SPILL_BLKID &&
	db->db_blkid > dn->dn_maxblkid)
	dn->dn_maxblkid = db->db_blkid;
	dbuf_unoverride(dr);
	} else {
	/*
	* This dbuf is not dirty in the open context.
	* Either uncache it (if its not referenced in
	* the open context) or reset its contents to
	* empty.
	*/
	dbuf_fix_old_data(db, txg);
	}
	}
	/* clear the contents if its cached */
	if (db->db_state == DB_CACHED) {
	ASSERT(db->db.db_data != NULL);
	arc_release(db->db_buf, db);
	rw_enter(&db->db_rwlock, RW_WRITER);
	bzero(db->db.db_data, db->db.db_size);
	rw_exit(&db->db_rwlock);
	arc_buf_freeze(db->db_buf);
	}

	mutex_exit(&db->db_mtx);
	}

	- kmem_free(db_search, sizeof (dmu_buf_impl_t));
	mutex_exit(&dn->dn_dbufs_mtx);
	+ kmem_free(db_search, sizeof (dmu_buf_impl_t));
	}

	void
	dbuf_new_size(dmu_buf_impl_t db, int size, dmu_tx_t tx)
	{
	arc_buf_t buf, old_buf;
	dbuf_dirty_record_t *dr;
	int osize = db->db.db_size;
	arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
	dnode_t *dn;

	ASSERT(db->db_blkid != DMU_BONUS_BLKID);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);

	/*
	* XXX we should be doing a dbuf_read, checking the return
	* value and returning that up to our callers
	*/
	dmu_buf_will_dirty(&db->db, tx);

	/* create the data buffer for the new block */
	buf = arc_alloc_buf(dn->dn_objset->os_spa, db, type, size);

	/* copy old block data to the new block */
	old_buf = db->db_buf;
	bcopy(old_buf->b_data, buf->b_data, MIN(osize, size));
	/* zero the remainder */
	if (size > osize)
	bzero((uint8_t *)buf->b_data + osize, size - osize);

	mutex_enter(&db->db_mtx);
	dbuf_set_data(db, buf);
	arc_buf_destroy(old_buf, db);
	db->db.db_size = size;

	dr = list_head(&db->db_dirty_records);
	/* dirty record added by dmu_buf_will_dirty() */
	VERIFY(dr != NULL);
	if (db->db_level == 0)
	dr->dt.dl.dr_data = buf;
	ASSERT3U(dr->dr_txg, ==, tx->tx_txg);
	ASSERT3U(dr->dr_accounted, ==, osize);
	dr->dr_accounted = size;
	mutex_exit(&db->db_mtx);

	dmu_objset_willuse_space(dn->dn_objset, size - osize, tx);
	DB_DNODE_EXIT(db);
	}

	void
	dbuf_release_bp(dmu_buf_impl_t *db)
	{
	objset_t *os __maybe_unused = db->db_objset;

	ASSERT(dsl_pool_sync_context(dmu_objset_pool(os)));
	ASSERT(arc_released(os->os_phys_buf) \|\|
	list_link_active(&os->os_dsl_dataset->ds_synced_link));
	ASSERT(db->db_parent == NULL \|\| arc_released(db->db_parent->db_buf));

	(void) arc_release(db->db_buf, db);
	}

	/*
	* We already have a dirty record for this TXG, and we are being
	* dirtied again.
	*/
	static void
	dbuf_redirty(dbuf_dirty_record_t *dr)
	{
	dmu_buf_impl_t *db = dr->dr_dbuf;

	ASSERT(MUTEX_HELD(&db->db_mtx));

	if (db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID) {
	/*
	* If this buffer has already been written out,
	* we now need to reset its state.
	*/
	dbuf_unoverride(dr);
	if (db->db.db_object != DMU_META_DNODE_OBJECT &&
	db->db_state != DB_NOFILL) {
	/* Already released on initial dirty, so just thaw. */
	ASSERT(arc_released(db->db_buf));
	arc_buf_thaw(db->db_buf);
	}
	}
	}

	dbuf_dirty_record_t *
	dbuf_dirty_lightweight(dnode_t dn, uint64_t blkid, dmu_tx_t tx)
	{
	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	IMPLY(dn->dn_objset->os_raw_receive, dn->dn_maxblkid >= blkid);
	dnode_new_blkid(dn, blkid, tx, B_TRUE, B_FALSE);
	ASSERT(dn->dn_maxblkid >= blkid);

	dbuf_dirty_record_t dr = kmem_zalloc(sizeof (dr), KM_SLEEP);
	list_link_init(&dr->dr_dirty_node);
	list_link_init(&dr->dr_dbuf_node);
	dr->dr_dnode = dn;
	dr->dr_txg = tx->tx_txg;
	dr->dt.dll.dr_blkid = blkid;
	dr->dr_accounted = dn->dn_datablksz;

	/*
	* There should not be any dbuf for the block that we're dirtying.
	* Otherwise the buffer contents could be inconsistent between the
	* dbuf and the lightweight dirty record.
	*/
	ASSERT3P(NULL, ==, dbuf_find(dn->dn_objset, dn->dn_object, 0, blkid));

	mutex_enter(&dn->dn_mtx);
	int txgoff = tx->tx_txg & TXG_MASK;
	if (dn->dn_free_ranges[txgoff] != NULL) {
	range_tree_clear(dn->dn_free_ranges[txgoff], blkid, 1);
	}

	if (dn->dn_nlevels == 1) {
	ASSERT3U(blkid, <, dn->dn_nblkptr);
	list_insert_tail(&dn->dn_dirty_records[txgoff], dr);
	mutex_exit(&dn->dn_mtx);
	rw_exit(&dn->dn_struct_rwlock);
	dnode_setdirty(dn, tx);
	} else {
	mutex_exit(&dn->dn_mtx);

	int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT;
	dmu_buf_impl_t *parent_db = dbuf_hold_level(dn,
	1, blkid >> epbs, FTAG);
	rw_exit(&dn->dn_struct_rwlock);
	if (parent_db == NULL) {
	kmem_free(dr, sizeof (*dr));
	return (NULL);
	}
	int err = dbuf_read(parent_db, NULL,
	(DB_RF_NOPREFETCH \| DB_RF_CANFAIL));
	if (err != 0) {
	dbuf_rele(parent_db, FTAG);
	kmem_free(dr, sizeof (*dr));
	return (NULL);
	}

	dbuf_dirty_record_t *parent_dr = dbuf_dirty(parent_db, tx);
	dbuf_rele(parent_db, FTAG);
	mutex_enter(&parent_dr->dt.di.dr_mtx);
	ASSERT3U(parent_dr->dr_txg, ==, tx->tx_txg);
	list_insert_tail(&parent_dr->dt.di.dr_children, dr);
	mutex_exit(&parent_dr->dt.di.dr_mtx);
	dr->dr_parent = parent_dr;
	}

	dmu_objset_willuse_space(dn->dn_objset, dr->dr_accounted, tx);

	return (dr);
	}

	dbuf_dirty_record_t *
	dbuf_dirty(dmu_buf_impl_t db, dmu_tx_t tx)
	{
	dnode_t *dn;
	objset_t *os;
	dbuf_dirty_record_t dr, dr_next, *dr_head;
	int txgoff = tx->tx_txg & TXG_MASK;
	boolean_t drop_struct_rwlock = B_FALSE;

	ASSERT(tx->tx_txg != 0);
	ASSERT(!zfs_refcount_is_zero(&db->db_holds));
	DMU_TX_DIRTY_BUF(tx, db);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	/*
	* Shouldn't dirty a regular buffer in syncing context. Private
	* objects may be dirtied in syncing context, but only if they
	* were already pre-dirtied in open context.
	*/
	#ifdef ZFS_DEBUG
	if (dn->dn_objset->os_dsl_dataset != NULL) {
	rrw_enter(&dn->dn_objset->os_dsl_dataset->ds_bp_rwlock,
	RW_READER, FTAG);
	}
	ASSERT(!dmu_tx_is_syncing(tx) \|\|
	BP_IS_HOLE(dn->dn_objset->os_rootbp) \|\|
	DMU_OBJECT_IS_SPECIAL(dn->dn_object) \|\|
	dn->dn_objset->os_dsl_dataset == NULL);
	if (dn->dn_objset->os_dsl_dataset != NULL)
	rrw_exit(&dn->dn_objset->os_dsl_dataset->ds_bp_rwlock, FTAG);
	#endif
	/*
	* We make this assert for private objects as well, but after we
	* check if we're already dirty. They are allowed to re-dirty
	* in syncing context.
	*/
	ASSERT(dn->dn_object == DMU_META_DNODE_OBJECT \|\|
	dn->dn_dirtyctx == DN_UNDIRTIED \|\| dn->dn_dirtyctx ==
	(dmu_tx_is_syncing(tx) ? DN_DIRTY_SYNC : DN_DIRTY_OPEN));

	mutex_enter(&db->db_mtx);
	/*
	* XXX make this true for indirects too? The problem is that
	* transactions created with dmu_tx_create_assigned() from
	* syncing context don't bother holding ahead.
	*/
	ASSERT(db->db_level != 0 \|\|
	db->db_state == DB_CACHED \|\| db->db_state == DB_FILL \|\|
	db->db_state == DB_NOFILL);

	mutex_enter(&dn->dn_mtx);
	dnode_set_dirtyctx(dn, tx, db);
	if (tx->tx_txg > dn->dn_dirty_txg)
	dn->dn_dirty_txg = tx->tx_txg;
	mutex_exit(&dn->dn_mtx);

	if (db->db_blkid == DMU_SPILL_BLKID)
	dn->dn_have_spill = B_TRUE;

	/*
	* If this buffer is already dirty, we're done.
	*/
	dr_head = list_head(&db->db_dirty_records);
	ASSERT(dr_head == NULL \|\| dr_head->dr_txg <= tx->tx_txg \|\|
	db->db.db_object == DMU_META_DNODE_OBJECT);
	dr_next = dbuf_find_dirty_lte(db, tx->tx_txg);
	if (dr_next && dr_next->dr_txg == tx->tx_txg) {
	DB_DNODE_EXIT(db);

	dbuf_redirty(dr_next);
	mutex_exit(&db->db_mtx);
	return (dr_next);
	}

	/*
	* Only valid if not already dirty.
	*/
	ASSERT(dn->dn_object == 0 \|\|
	dn->dn_dirtyctx == DN_UNDIRTIED \|\| dn->dn_dirtyctx ==
	(dmu_tx_is_syncing(tx) ? DN_DIRTY_SYNC : DN_DIRTY_OPEN));

	ASSERT3U(dn->dn_nlevels, >, db->db_level);

	/*
	* We should only be dirtying in syncing context if it's the
	* mos or we're initializing the os or it's a special object.
	* However, we are allowed to dirty in syncing context provided
	* we already dirtied it in open context. Hence we must make
	* this assertion only if we're not already dirty.
	*/
	os = dn->dn_objset;
	VERIFY3U(tx->tx_txg, <=, spa_final_dirty_txg(os->os_spa));
	#ifdef ZFS_DEBUG
	if (dn->dn_objset->os_dsl_dataset != NULL)
	rrw_enter(&os->os_dsl_dataset->ds_bp_rwlock, RW_READER, FTAG);
	ASSERT(!dmu_tx_is_syncing(tx) \|\| DMU_OBJECT_IS_SPECIAL(dn->dn_object) \|\|
	os->os_dsl_dataset == NULL \|\| BP_IS_HOLE(os->os_rootbp));
	if (dn->dn_objset->os_dsl_dataset != NULL)
	rrw_exit(&os->os_dsl_dataset->ds_bp_rwlock, FTAG);
	#endif
	ASSERT(db->db.db_size != 0);

	dprintf_dbuf(db, "size=%llx\n", (u_longlong_t)db->db.db_size);

	if (db->db_blkid != DMU_BONUS_BLKID) {
	dmu_objset_willuse_space(os, db->db.db_size, tx);
	}

	/*
	* If this buffer is dirty in an old transaction group we need
	* to make a copy of it so that the changes we make in this
	* transaction group won't leak out when we sync the older txg.
	*/
	dr = kmem_zalloc(sizeof (dbuf_dirty_record_t), KM_SLEEP);
	list_link_init(&dr->dr_dirty_node);
	list_link_init(&dr->dr_dbuf_node);
	dr->dr_dnode = dn;
	if (db->db_level == 0) {
	void *data_old = db->db_buf;

	if (db->db_state != DB_NOFILL) {
	if (db->db_blkid == DMU_BONUS_BLKID) {
	dbuf_fix_old_data(db, tx->tx_txg);
	data_old = db->db.db_data;
	} else if (db->db.db_object != DMU_META_DNODE_OBJECT) {
	/*
	* Release the data buffer from the cache so
	* that we can modify it without impacting
	* possible other users of this cached data
	* block. Note that indirect blocks and
	* private objects are not released until the
	* syncing state (since they are only modified
	* then).
	*/
	arc_release(db->db_buf, db);
	dbuf_fix_old_data(db, tx->tx_txg);
	data_old = db->db_buf;
	}
	ASSERT(data_old != NULL);
	}
	dr->dt.dl.dr_data = data_old;
	} else {
	mutex_init(&dr->dt.di.dr_mtx, NULL, MUTEX_NOLOCKDEP, NULL);
	list_create(&dr->dt.di.dr_children,
	sizeof (dbuf_dirty_record_t),
	offsetof(dbuf_dirty_record_t, dr_dirty_node));
	}
	if (db->db_blkid != DMU_BONUS_BLKID)
	dr->dr_accounted = db->db.db_size;
	dr->dr_dbuf = db;
	dr->dr_txg = tx->tx_txg;
	list_insert_before(&db->db_dirty_records, dr_next, dr);

	/*
	* We could have been freed_in_flight between the dbuf_noread
	* and dbuf_dirty. We win, as though the dbuf_noread() had
	* happened after the free.
	*/
	if (db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID &&
	db->db_blkid != DMU_SPILL_BLKID) {
	mutex_enter(&dn->dn_mtx);
	if (dn->dn_free_ranges[txgoff] != NULL) {
	range_tree_clear(dn->dn_free_ranges[txgoff],
	db->db_blkid, 1);
	}
	mutex_exit(&dn->dn_mtx);
	db->db_freed_in_flight = FALSE;
	}

	/*
	* This buffer is now part of this txg
	*/
	dbuf_add_ref(db, (void *)(uintptr_t)tx->tx_txg);
	db->db_dirtycnt += 1;
	ASSERT3U(db->db_dirtycnt, <=, 3);

	mutex_exit(&db->db_mtx);

	if (db->db_blkid == DMU_BONUS_BLKID \|\|
	db->db_blkid == DMU_SPILL_BLKID) {
	mutex_enter(&dn->dn_mtx);
	ASSERT(!list_link_active(&dr->dr_dirty_node));
	list_insert_tail(&dn->dn_dirty_records[txgoff], dr);
	mutex_exit(&dn->dn_mtx);
	dnode_setdirty(dn, tx);
	DB_DNODE_EXIT(db);
	return (dr);
	}

	if (!RW_WRITE_HELD(&dn->dn_struct_rwlock)) {
	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	drop_struct_rwlock = B_TRUE;
	}

	/*
	* If we are overwriting a dedup BP, then unless it is snapshotted,
	* when we get to syncing context we will need to decrement its
	* refcount in the DDT. Prefetch the relevant DDT block so that
	* syncing context won't have to wait for the i/o.
	*/
	if (db->db_blkptr != NULL) {
	db_lock_type_t dblt = dmu_buf_lock_parent(db, RW_READER, FTAG);
	ddt_prefetch(os->os_spa, db->db_blkptr);
	dmu_buf_unlock_parent(db, dblt, FTAG);
	}

	/*
	* We need to hold the dn_struct_rwlock to make this assertion,
	* because it protects dn_phys / dn_next_nlevels from changing.
	*/
	ASSERT((dn->dn_phys->dn_nlevels == 0 && db->db_level == 0) \|\|
	dn->dn_phys->dn_nlevels > db->db_level \|\|
	dn->dn_next_nlevels[txgoff] > db->db_level \|\|
	dn->dn_next_nlevels[(tx->tx_txg-1) & TXG_MASK] > db->db_level \|\|
	dn->dn_next_nlevels[(tx->tx_txg-2) & TXG_MASK] > db->db_level);


	if (db->db_level == 0) {
	ASSERT(!db->db_objset->os_raw_receive \|\|
	dn->dn_maxblkid >= db->db_blkid);
	dnode_new_blkid(dn, db->db_blkid, tx,
	drop_struct_rwlock, B_FALSE);
	ASSERT(dn->dn_maxblkid >= db->db_blkid);
	}

	if (db->db_level+1 < dn->dn_nlevels) {
	dmu_buf_impl_t *parent = db->db_parent;
	dbuf_dirty_record_t *di;
	int parent_held = FALSE;

	if (db->db_parent == NULL \|\| db->db_parent == dn->dn_dbuf) {
	int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT;
	parent = dbuf_hold_level(dn, db->db_level + 1,
	db->db_blkid >> epbs, FTAG);
	ASSERT(parent != NULL);
	parent_held = TRUE;
	}
	if (drop_struct_rwlock)
	rw_exit(&dn->dn_struct_rwlock);
	ASSERT3U(db->db_level + 1, ==, parent->db_level);
	di = dbuf_dirty(parent, tx);
	if (parent_held)
	dbuf_rele(parent, FTAG);

	mutex_enter(&db->db_mtx);
	/*
	* Since we've dropped the mutex, it's possible that
	* dbuf_undirty() might have changed this out from under us.
	*/
	if (list_head(&db->db_dirty_records) == dr \|\|
	dn->dn_object == DMU_META_DNODE_OBJECT) {
	mutex_enter(&di->dt.di.dr_mtx);
	ASSERT3U(di->dr_txg, ==, tx->tx_txg);
	ASSERT(!list_link_active(&dr->dr_dirty_node));
	list_insert_tail(&di->dt.di.dr_children, dr);
	mutex_exit(&di->dt.di.dr_mtx);
	dr->dr_parent = di;
	}
	mutex_exit(&db->db_mtx);
	} else {
	ASSERT(db->db_level + 1 == dn->dn_nlevels);
	ASSERT(db->db_blkid < dn->dn_nblkptr);
	ASSERT(db->db_parent == NULL \|\| db->db_parent == dn->dn_dbuf);
	mutex_enter(&dn->dn_mtx);
	ASSERT(!list_link_active(&dr->dr_dirty_node));
	list_insert_tail(&dn->dn_dirty_records[txgoff], dr);
	mutex_exit(&dn->dn_mtx);
	if (drop_struct_rwlock)
	rw_exit(&dn->dn_struct_rwlock);
	}

	dnode_setdirty(dn, tx);
	DB_DNODE_EXIT(db);
	return (dr);
	}

	static void
	dbuf_undirty_bonus(dbuf_dirty_record_t *dr)
	{
	dmu_buf_impl_t *db = dr->dr_dbuf;

	if (dr->dt.dl.dr_data != db->db.db_data) {
	struct dnode *dn = dr->dr_dnode;
	int max_bonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots);

	kmem_free(dr->dt.dl.dr_data, max_bonuslen);
	arc_space_return(max_bonuslen, ARC_SPACE_BONUS);
	}
	db->db_data_pending = NULL;
	ASSERT(list_next(&db->db_dirty_records, dr) == NULL);
	list_remove(&db->db_dirty_records, dr);
	if (dr->dr_dbuf->db_level != 0) {
	mutex_destroy(&dr->dt.di.dr_mtx);
	list_destroy(&dr->dt.di.dr_children);
	}
	kmem_free(dr, sizeof (dbuf_dirty_record_t));
	ASSERT3U(db->db_dirtycnt, >, 0);
	db->db_dirtycnt -= 1;
	}

	/*
	* Undirty a buffer in the transaction group referenced by the given
	* transaction. Return whether this evicted the dbuf.
	*/
	static boolean_t
	dbuf_undirty(dmu_buf_impl_t db, dmu_tx_t tx)
	{
	uint64_t txg = tx->tx_txg;

	ASSERT(txg != 0);

	/*
	* Due to our use of dn_nlevels below, this can only be called
	* in open context, unless we are operating on the MOS.
	* From syncing context, dn_nlevels may be different from the
	* dn_nlevels used when dbuf was dirtied.
	*/
	ASSERT(db->db_objset ==
	dmu_objset_pool(db->db_objset)->dp_meta_objset \|\|
	txg != spa_syncing_txg(dmu_objset_spa(db->db_objset)));
	ASSERT(db->db_blkid != DMU_BONUS_BLKID);
	ASSERT0(db->db_level);
	ASSERT(MUTEX_HELD(&db->db_mtx));

	/*
	* If this buffer is not dirty, we're done.
	*/
	dbuf_dirty_record_t *dr = dbuf_find_dirty_eq(db, txg);
	if (dr == NULL)
	return (B_FALSE);
	ASSERT(dr->dr_dbuf == db);

	dnode_t *dn = dr->dr_dnode;

	dprintf_dbuf(db, "size=%llx\n", (u_longlong_t)db->db.db_size);

	ASSERT(db->db.db_size != 0);

	dsl_pool_undirty_space(dmu_objset_pool(dn->dn_objset),
	dr->dr_accounted, txg);

	list_remove(&db->db_dirty_records, dr);

	/*
	* Note that there are three places in dbuf_dirty()
	* where this dirty record may be put on a list.
	* Make sure to do a list_remove corresponding to
	* every one of those list_insert calls.
	*/
	if (dr->dr_parent) {
	mutex_enter(&dr->dr_parent->dt.di.dr_mtx);
	list_remove(&dr->dr_parent->dt.di.dr_children, dr);
	mutex_exit(&dr->dr_parent->dt.di.dr_mtx);
	} else if (db->db_blkid == DMU_SPILL_BLKID \|\|
	db->db_level + 1 == dn->dn_nlevels) {
	ASSERT(db->db_blkptr == NULL \|\| db->db_parent == dn->dn_dbuf);
	mutex_enter(&dn->dn_mtx);
	list_remove(&dn->dn_dirty_records[txg & TXG_MASK], dr);
	mutex_exit(&dn->dn_mtx);
	}

	if (db->db_state != DB_NOFILL) {
	dbuf_unoverride(dr);

	ASSERT(db->db_buf != NULL);
	ASSERT(dr->dt.dl.dr_data != NULL);
	if (dr->dt.dl.dr_data != db->db_buf)
	arc_buf_destroy(dr->dt.dl.dr_data, db);
	}

	kmem_free(dr, sizeof (dbuf_dirty_record_t));

	ASSERT(db->db_dirtycnt > 0);
	db->db_dirtycnt -= 1;

	if (zfs_refcount_remove(&db->db_holds, (void *)(uintptr_t)txg) == 0) {
	ASSERT(db->db_state == DB_NOFILL \|\| arc_released(db->db_buf));
	dbuf_destroy(db);
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	static void
	dmu_buf_will_dirty_impl(dmu_buf_t db_fake, int flags, dmu_tx_t tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;

	ASSERT(tx->tx_txg != 0);
	ASSERT(!zfs_refcount_is_zero(&db->db_holds));

	/*
	* Quick check for dirtiness. For already dirty blocks, this
	* reduces runtime of this function by >90%, and overall performance
	* by 50% for some workloads (e.g. file deletion with indirect blocks
	* cached).
	*/
	mutex_enter(&db->db_mtx);

	if (db->db_state == DB_CACHED) {
	dbuf_dirty_record_t *dr = dbuf_find_dirty_eq(db, tx->tx_txg);
	/*
	* It's possible that it is already dirty but not cached,
	* because there are some calls to dbuf_dirty() that don't
	* go through dmu_buf_will_dirty().
	*/
	if (dr != NULL) {
	/* This dbuf is already dirty and cached. */
	dbuf_redirty(dr);
	mutex_exit(&db->db_mtx);
	return;
	}
	}
	mutex_exit(&db->db_mtx);

	DB_DNODE_ENTER(db);
	if (RW_WRITE_HELD(&DB_DNODE(db)->dn_struct_rwlock))
	flags \|= DB_RF_HAVESTRUCT;
	DB_DNODE_EXIT(db);
	(void) dbuf_read(db, NULL, flags);
	(void) dbuf_dirty(db, tx);
	}

	void
	dmu_buf_will_dirty(dmu_buf_t db_fake, dmu_tx_t tx)
	{
	dmu_buf_will_dirty_impl(db_fake,
	DB_RF_MUST_SUCCEED \| DB_RF_NOPREFETCH, tx);
	}

	boolean_t
	dmu_buf_is_dirty(dmu_buf_t db_fake, dmu_tx_t tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;
	dbuf_dirty_record_t *dr;

	mutex_enter(&db->db_mtx);
	dr = dbuf_find_dirty_eq(db, tx->tx_txg);
	mutex_exit(&db->db_mtx);
	return (dr != NULL);
	}

	void
	dmu_buf_will_not_fill(dmu_buf_t db_fake, dmu_tx_t tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;

	db->db_state = DB_NOFILL;
	DTRACE_SET_STATE(db, "allocating NOFILL buffer");
	dmu_buf_will_fill(db_fake, tx);
	}

	void
	dmu_buf_will_fill(dmu_buf_t db_fake, dmu_tx_t tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;

	ASSERT(db->db_blkid != DMU_BONUS_BLKID);
	ASSERT(tx->tx_txg != 0);
	ASSERT(db->db_level == 0);
	ASSERT(!zfs_refcount_is_zero(&db->db_holds));

	ASSERT(db->db.db_object != DMU_META_DNODE_OBJECT \|\|
	dmu_tx_private_ok(tx));

	dbuf_noread(db);
	(void) dbuf_dirty(db, tx);
	}

	/*
	* This function is effectively the same as dmu_buf_will_dirty(), but
	* indicates the caller expects raw encrypted data in the db, and provides
	* the crypt params (byteorder, salt, iv, mac) which should be stored in the
	* blkptr_t when this dbuf is written. This is only used for blocks of
	* dnodes, during raw receive.
	*/
	void
	dmu_buf_set_crypt_params(dmu_buf_t *db_fake, boolean_t byteorder,
	const uint8_t salt, const uint8_t iv, const uint8_t mac, dmu_tx_t tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;
	dbuf_dirty_record_t *dr;

	/*
	* dr_has_raw_params is only processed for blocks of dnodes
	* (see dbuf_sync_dnode_leaf_crypt()).
	*/
	ASSERT3U(db->db.db_object, ==, DMU_META_DNODE_OBJECT);
	ASSERT3U(db->db_level, ==, 0);
	ASSERT(db->db_objset->os_raw_receive);

	dmu_buf_will_dirty_impl(db_fake,
	DB_RF_MUST_SUCCEED \| DB_RF_NOPREFETCH \| DB_RF_NO_DECRYPT, tx);

	dr = dbuf_find_dirty_eq(db, tx->tx_txg);

	ASSERT3P(dr, !=, NULL);

	dr->dt.dl.dr_has_raw_params = B_TRUE;
	dr->dt.dl.dr_byteorder = byteorder;
	bcopy(salt, dr->dt.dl.dr_salt, ZIO_DATA_SALT_LEN);
	bcopy(iv, dr->dt.dl.dr_iv, ZIO_DATA_IV_LEN);
	bcopy(mac, dr->dt.dl.dr_mac, ZIO_DATA_MAC_LEN);
	}

	static void
	dbuf_override_impl(dmu_buf_impl_t db, const blkptr_t bp, dmu_tx_t *tx)
	{
	struct dirty_leaf *dl;
	dbuf_dirty_record_t *dr;

	dr = list_head(&db->db_dirty_records);
	ASSERT3U(dr->dr_txg, ==, tx->tx_txg);
	dl = &dr->dt.dl;
	dl->dr_overridden_by = *bp;
	dl->dr_override_state = DR_OVERRIDDEN;
	dl->dr_overridden_by.blk_birth = dr->dr_txg;
	}

	/* ARGSUSED */
	void
	dmu_buf_fill_done(dmu_buf_t dbuf, dmu_tx_t tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )dbuf;
	dbuf_states_t old_state;
	mutex_enter(&db->db_mtx);
	DBUF_VERIFY(db);

	old_state = db->db_state;
	db->db_state = DB_CACHED;
	if (old_state == DB_FILL) {
	if (db->db_level == 0 && db->db_freed_in_flight) {
	ASSERT(db->db_blkid != DMU_BONUS_BLKID);
	/* we were freed while filling */
	/* XXX dbuf_undirty? */
	bzero(db->db.db_data, db->db.db_size);
	db->db_freed_in_flight = FALSE;
	DTRACE_SET_STATE(db,
	"fill done handling freed in flight");
	} else {
	DTRACE_SET_STATE(db, "fill done");
	}
	cv_broadcast(&db->db_changed);
	}
	mutex_exit(&db->db_mtx);
	}

	void
	dmu_buf_write_embedded(dmu_buf_t dbuf, void data,
	bp_embedded_type_t etype, enum zio_compress comp,
	int uncompressed_size, int compressed_size, int byteorder,
	dmu_tx_t *tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )dbuf;
	struct dirty_leaf *dl;
	dmu_object_type_t type;
	dbuf_dirty_record_t *dr;

	if (etype == BP_EMBEDDED_TYPE_DATA) {
	ASSERT(spa_feature_is_active(dmu_objset_spa(db->db_objset),
	SPA_FEATURE_EMBEDDED_DATA));
	}

	DB_DNODE_ENTER(db);
	type = DB_DNODE(db)->dn_type;
	DB_DNODE_EXIT(db);

	ASSERT0(db->db_level);
	ASSERT(db->db_blkid != DMU_BONUS_BLKID);

	dmu_buf_will_not_fill(dbuf, tx);

	dr = list_head(&db->db_dirty_records);
	ASSERT3U(dr->dr_txg, ==, tx->tx_txg);
	dl = &dr->dt.dl;
	encode_embedded_bp_compressed(&dl->dr_overridden_by,
	data, comp, uncompressed_size, compressed_size);
	BPE_SET_ETYPE(&dl->dr_overridden_by, etype);
	BP_SET_TYPE(&dl->dr_overridden_by, type);
	BP_SET_LEVEL(&dl->dr_overridden_by, 0);
	BP_SET_BYTEORDER(&dl->dr_overridden_by, byteorder);

	dl->dr_override_state = DR_OVERRIDDEN;
	dl->dr_overridden_by.blk_birth = dr->dr_txg;
	}

	void
	dmu_buf_redact(dmu_buf_t dbuf, dmu_tx_t tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )dbuf;
	dmu_object_type_t type;
	ASSERT(dsl_dataset_feature_is_active(db->db_objset->os_dsl_dataset,
	SPA_FEATURE_REDACTED_DATASETS));

	DB_DNODE_ENTER(db);
	type = DB_DNODE(db)->dn_type;
	DB_DNODE_EXIT(db);

	ASSERT0(db->db_level);
	dmu_buf_will_not_fill(dbuf, tx);

	blkptr_t bp = { { { {0} } } };
	BP_SET_TYPE(&bp, type);
	BP_SET_LEVEL(&bp, 0);
	BP_SET_BIRTH(&bp, tx->tx_txg, 0);
	BP_SET_REDACTED(&bp);
	BPE_SET_LSIZE(&bp, dbuf->db_size);

	dbuf_override_impl(db, &bp, tx);
	}

	/*
	* Directly assign a provided arc buf to a given dbuf if it's not referenced
	* by anybody except our caller. Otherwise copy arcbuf's contents to dbuf.
	*/
	void
	dbuf_assign_arcbuf(dmu_buf_impl_t db, arc_buf_t buf, dmu_tx_t *tx)
	{
	ASSERT(!zfs_refcount_is_zero(&db->db_holds));
	ASSERT(db->db_blkid != DMU_BONUS_BLKID);
	ASSERT(db->db_level == 0);
	ASSERT3U(dbuf_is_metadata(db), ==, arc_is_metadata(buf));
	ASSERT(buf != NULL);
	ASSERT3U(arc_buf_lsize(buf), ==, db->db.db_size);
	ASSERT(tx->tx_txg != 0);

	arc_return_buf(buf, db);
	ASSERT(arc_released(buf));

	mutex_enter(&db->db_mtx);

	while (db->db_state == DB_READ \|\| db->db_state == DB_FILL)
	cv_wait(&db->db_changed, &db->db_mtx);

	ASSERT(db->db_state == DB_CACHED \|\| db->db_state == DB_UNCACHED);

	if (db->db_state == DB_CACHED &&
	zfs_refcount_count(&db->db_holds) - 1 > db->db_dirtycnt) {
	/*
	* In practice, we will never have a case where we have an
	* encrypted arc buffer while additional holds exist on the
	* dbuf. We don't handle this here so we simply assert that
	* fact instead.
	*/
	ASSERT(!arc_is_encrypted(buf));
	mutex_exit(&db->db_mtx);
	(void) dbuf_dirty(db, tx);
	bcopy(buf->b_data, db->db.db_data, db->db.db_size);
	arc_buf_destroy(buf, db);
	return;
	}

	if (db->db_state == DB_CACHED) {
	dbuf_dirty_record_t *dr = list_head(&db->db_dirty_records);

	ASSERT(db->db_buf != NULL);
	if (dr != NULL && dr->dr_txg == tx->tx_txg) {
	ASSERT(dr->dt.dl.dr_data == db->db_buf);

	if (!arc_released(db->db_buf)) {
	ASSERT(dr->dt.dl.dr_override_state ==
	DR_OVERRIDDEN);
	arc_release(db->db_buf, db);
	}
	dr->dt.dl.dr_data = buf;
	arc_buf_destroy(db->db_buf, db);
	} else if (dr == NULL \|\| dr->dt.dl.dr_data != db->db_buf) {
	arc_release(db->db_buf, db);
	arc_buf_destroy(db->db_buf, db);
	}
	db->db_buf = NULL;
	}
	ASSERT(db->db_buf == NULL);
	dbuf_set_data(db, buf);
	db->db_state = DB_FILL;
	DTRACE_SET_STATE(db, "filling assigned arcbuf");
	mutex_exit(&db->db_mtx);
	(void) dbuf_dirty(db, tx);
	dmu_buf_fill_done(&db->db, tx);
	}

	void
	dbuf_destroy(dmu_buf_impl_t *db)
	{
	dnode_t *dn;
	dmu_buf_impl_t *parent = db->db_parent;
	dmu_buf_impl_t *dndb;

	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(zfs_refcount_is_zero(&db->db_holds));

	if (db->db_buf != NULL) {
	arc_buf_destroy(db->db_buf, db);
	db->db_buf = NULL;
	}

	if (db->db_blkid == DMU_BONUS_BLKID) {
	int slots = DB_DNODE(db)->dn_num_slots;
	int bonuslen = DN_SLOTS_TO_BONUSLEN(slots);
	if (db->db.db_data != NULL) {
	kmem_free(db->db.db_data, bonuslen);
	arc_space_return(bonuslen, ARC_SPACE_BONUS);
	db->db_state = DB_UNCACHED;
	DTRACE_SET_STATE(db, "buffer cleared");
	}
	}

	dbuf_clear_data(db);

	if (multilist_link_active(&db->db_cache_link)) {
	ASSERT(db->db_caching_status == DB_DBUF_CACHE \|\|
	db->db_caching_status == DB_DBUF_METADATA_CACHE);

	multilist_remove(&dbuf_caches[db->db_caching_status].cache, db);
	(void) zfs_refcount_remove_many(
	&dbuf_caches[db->db_caching_status].size,
	db->db.db_size, db);

	if (db->db_caching_status == DB_DBUF_METADATA_CACHE) {
	DBUF_STAT_BUMPDOWN(metadata_cache_count);
	} else {
	DBUF_STAT_BUMPDOWN(cache_levels[db->db_level]);
	DBUF_STAT_BUMPDOWN(cache_count);
	DBUF_STAT_DECR(cache_levels_bytes[db->db_level],
	db->db.db_size);
	}
	db->db_caching_status = DB_NO_CACHE;
	}

	ASSERT(db->db_state == DB_UNCACHED \|\| db->db_state == DB_NOFILL);
	ASSERT(db->db_data_pending == NULL);
	ASSERT(list_is_empty(&db->db_dirty_records));

	db->db_state = DB_EVICTING;
	DTRACE_SET_STATE(db, "buffer eviction started");
	db->db_blkptr = NULL;

	/*
	* Now that db_state is DB_EVICTING, nobody else can find this via
	* the hash table. We can now drop db_mtx, which allows us to
	* acquire the dn_dbufs_mtx.
	*/
	mutex_exit(&db->db_mtx);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	dndb = dn->dn_dbuf;
	if (db->db_blkid != DMU_BONUS_BLKID) {
	boolean_t needlock = !MUTEX_HELD(&dn->dn_dbufs_mtx);
	if (needlock)
	mutex_enter_nested(&dn->dn_dbufs_mtx,
	NESTED_SINGLE);
	avl_remove(&dn->dn_dbufs, db);
	membar_producer();
	DB_DNODE_EXIT(db);
	if (needlock)
	mutex_exit(&dn->dn_dbufs_mtx);
	/*
	* Decrementing the dbuf count means that the hold corresponding
	* to the removed dbuf is no longer discounted in dnode_move(),
	* so the dnode cannot be moved until after we release the hold.
	* The membar_producer() ensures visibility of the decremented
	* value in dnode_move(), since DB_DNODE_EXIT doesn't actually
	* release any lock.
	*/
	mutex_enter(&dn->dn_mtx);
	dnode_rele_and_unlock(dn, db, B_TRUE);
	db->db_dnode_handle = NULL;

	dbuf_hash_remove(db);
	} else {
	DB_DNODE_EXIT(db);
	}

	ASSERT(zfs_refcount_is_zero(&db->db_holds));

	db->db_parent = NULL;

	ASSERT(db->db_buf == NULL);
	ASSERT(db->db.db_data == NULL);
	ASSERT(db->db_hash_next == NULL);
	ASSERT(db->db_blkptr == NULL);
	ASSERT(db->db_data_pending == NULL);
	ASSERT3U(db->db_caching_status, ==, DB_NO_CACHE);
	ASSERT(!multilist_link_active(&db->db_cache_link));

	kmem_cache_free(dbuf_kmem_cache, db);
	arc_space_return(sizeof (dmu_buf_impl_t), ARC_SPACE_DBUF);

	/*
	* If this dbuf is referenced from an indirect dbuf,
	* decrement the ref count on the indirect dbuf.
	*/
	if (parent && parent != dndb) {
	mutex_enter(&parent->db_mtx);
	dbuf_rele_and_unlock(parent, db, B_TRUE);
	}
	}

	/*
	* Note: While bpp will always be updated if the function returns success,
	* parentp will not be updated if the dnode does not have dn_dbuf filled in;
	* this happens when the dnode is the meta-dnode, or {user\|group\|project}used
	* object.
	*/
	__attribute__((always_inline))
	static inline int
	dbuf_findbp(dnode_t *dn, int level, uint64_t blkid, int fail_sparse,
	dmu_buf_impl_t parentp, blkptr_t bpp)
	{
	*parentp = NULL;
	*bpp = NULL;

	ASSERT(blkid != DMU_BONUS_BLKID);

	if (blkid == DMU_SPILL_BLKID) {
	mutex_enter(&dn->dn_mtx);
	if (dn->dn_have_spill &&
	(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR))
	*bpp = DN_SPILL_BLKPTR(dn->dn_phys);
	else
	*bpp = NULL;
	dbuf_add_ref(dn->dn_dbuf, NULL);
	*parentp = dn->dn_dbuf;
	mutex_exit(&dn->dn_mtx);
	return (0);
	}

	int nlevels =
	(dn->dn_phys->dn_nlevels == 0) ? 1 : dn->dn_phys->dn_nlevels;
	int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT;

	ASSERT3U(level * epbs, <, 64);
	ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));
	/*
	* This assertion shouldn't trip as long as the max indirect block size
	* is less than 1M. The reason for this is that up to that point,
	* the number of levels required to address an entire object with blocks
	* of size SPA_MINBLOCKSIZE satisfies nlevels * epbs + 1 <= 64. In
	* other words, if N * epbs + 1 > 64, then if (N-1) * epbs + 1 > 55
	* (i.e. we can address the entire object), objects will all use at most
	* N-1 levels and the assertion won't overflow. However, once epbs is
	* 13, 4 * 13 + 1 = 53, but 5 * 13 + 1 = 66. Then, 4 levels will not be
	* enough to address an entire object, so objects will have 5 levels,
	* but then this assertion will overflow.
	*
	* All this is to say that if we ever increase DN_MAX_INDBLKSHIFT, we
	* need to redo this logic to handle overflows.
	*/
	ASSERT(level >= nlevels \|\|
	((nlevels - level - 1) * epbs) +
	highbit64(dn->dn_phys->dn_nblkptr) <= 64);
	if (level >= nlevels \|\|
	blkid >= ((uint64_t)dn->dn_phys->dn_nblkptr <<
	((nlevels - level - 1) * epbs)) \|\|
	(fail_sparse &&
	blkid > (dn->dn_phys->dn_maxblkid >> (level * epbs)))) {
	/* the buffer has no parent yet */
	return (SET_ERROR(ENOENT));
	} else if (level < nlevels-1) {
	/* this block is referenced from an indirect block */
	int err;

	err = dbuf_hold_impl(dn, level + 1,
	blkid >> epbs, fail_sparse, FALSE, NULL, parentp);

	if (err)
	return (err);
	err = dbuf_read(*parentp, NULL,
	(DB_RF_HAVESTRUCT \| DB_RF_NOPREFETCH \| DB_RF_CANFAIL));
	if (err) {
	dbuf_rele(*parentp, NULL);
	*parentp = NULL;
	return (err);
	}
	rw_enter(&(*parentp)->db_rwlock, RW_READER);
	bpp = ((blkptr_t )(*parentp)->db.db_data) +
	(blkid & ((1ULL << epbs) - 1));
	if (blkid > (dn->dn_phys->dn_maxblkid >> (level * epbs)))
	ASSERT(BP_IS_HOLE(*bpp));
	rw_exit(&(*parentp)->db_rwlock);
	return (0);
	} else {
	/* the block is referenced from the dnode */
	ASSERT3U(level, ==, nlevels-1);
	ASSERT(dn->dn_phys->dn_nblkptr == 0 \|\|
	blkid < dn->dn_phys->dn_nblkptr);
	if (dn->dn_dbuf) {
	dbuf_add_ref(dn->dn_dbuf, NULL);
	*parentp = dn->dn_dbuf;
	}
	*bpp = &dn->dn_phys->dn_blkptr[blkid];
	return (0);
	}
	}

	static dmu_buf_impl_t *
	dbuf_create(dnode_t *dn, uint8_t level, uint64_t blkid,
	dmu_buf_impl_t parent, blkptr_t blkptr)
	{
	objset_t *os = dn->dn_objset;
	dmu_buf_impl_t db, odb;

	ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));
	ASSERT(dn->dn_type != DMU_OT_NONE);

	db = kmem_cache_alloc(dbuf_kmem_cache, KM_SLEEP);

	list_create(&db->db_dirty_records, sizeof (dbuf_dirty_record_t),
	offsetof(dbuf_dirty_record_t, dr_dbuf_node));

	db->db_objset = os;
	db->db.db_object = dn->dn_object;
	db->db_level = level;
	db->db_blkid = blkid;
	db->db_dirtycnt = 0;
	db->db_dnode_handle = dn->dn_handle;
	db->db_parent = parent;
	db->db_blkptr = blkptr;

	db->db_user = NULL;
	db->db_user_immediate_evict = FALSE;
	db->db_freed_in_flight = FALSE;
	db->db_pending_evict = FALSE;

	if (blkid == DMU_BONUS_BLKID) {
	ASSERT3P(parent, ==, dn->dn_dbuf);
	db->db.db_size = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots) -
	(dn->dn_nblkptr-1) * sizeof (blkptr_t);
	ASSERT3U(db->db.db_size, >=, dn->dn_bonuslen);
	db->db.db_offset = DMU_BONUS_BLKID;
	db->db_state = DB_UNCACHED;
	DTRACE_SET_STATE(db, "bonus buffer created");
	db->db_caching_status = DB_NO_CACHE;
	/* the bonus dbuf is not placed in the hash table */
	arc_space_consume(sizeof (dmu_buf_impl_t), ARC_SPACE_DBUF);
	return (db);
	} else if (blkid == DMU_SPILL_BLKID) {
	db->db.db_size = (blkptr != NULL) ?
	BP_GET_LSIZE(blkptr) : SPA_MINBLOCKSIZE;
	db->db.db_offset = 0;
	} else {
	int blocksize =
	db->db_level ? 1 << dn->dn_indblkshift : dn->dn_datablksz;
	db->db.db_size = blocksize;
	db->db.db_offset = db->db_blkid * blocksize;
	}

	/*
	* Hold the dn_dbufs_mtx while we get the new dbuf
	* in the hash table and added to the dbufs list.
	* This prevents a possible deadlock with someone
	* trying to look up this dbuf before it's added to the
	* dn_dbufs list.
	*/
	mutex_enter(&dn->dn_dbufs_mtx);
	db->db_state = DB_EVICTING; /* not worth logging this state change */
	if ((odb = dbuf_hash_insert(db)) != NULL) {
	/* someone else inserted it first */
	mutex_exit(&dn->dn_dbufs_mtx);
	kmem_cache_free(dbuf_kmem_cache, db);
	DBUF_STAT_BUMP(hash_insert_race);
	return (odb);
	}
	avl_add(&dn->dn_dbufs, db);

	db->db_state = DB_UNCACHED;
	DTRACE_SET_STATE(db, "regular buffer created");
	db->db_caching_status = DB_NO_CACHE;
	mutex_exit(&dn->dn_dbufs_mtx);
	arc_space_consume(sizeof (dmu_buf_impl_t), ARC_SPACE_DBUF);

	if (parent && parent != dn->dn_dbuf)
	dbuf_add_ref(parent, db);

	ASSERT(dn->dn_object == DMU_META_DNODE_OBJECT \|\|
	zfs_refcount_count(&dn->dn_holds) > 0);
	(void) zfs_refcount_add(&dn->dn_holds, db);

	dprintf_dbuf(db, "db=%p\n", db);

	return (db);
	}

	/*
	* This function returns a block pointer and information about the object,
	* given a dnode and a block. This is a publicly accessible version of
	* dbuf_findbp that only returns some information, rather than the
	* dbuf. Note that the dnode passed in must be held, and the dn_struct_rwlock
	* should be locked as (at least) a reader.
	*/
	int
	dbuf_dnode_findbp(dnode_t *dn, uint64_t level, uint64_t blkid,
	blkptr_t bp, uint16_t datablkszsec, uint8_t *indblkshift)
	{
	dmu_buf_impl_t *dbp = NULL;
	blkptr_t *bp2;
	int err = 0;
	ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));

	err = dbuf_findbp(dn, level, blkid, B_FALSE, &dbp, &bp2);
	if (err == 0) {
	bp = bp2;
	if (dbp != NULL)
	dbuf_rele(dbp, NULL);
	if (datablkszsec != NULL)
	*datablkszsec = dn->dn_phys->dn_datablkszsec;
	if (indblkshift != NULL)
	*indblkshift = dn->dn_phys->dn_indblkshift;
	}

	return (err);
	}

	typedef struct dbuf_prefetch_arg {
	spa_t dpa_spa; / The spa to issue the prefetch in. */
	zbookmark_phys_t dpa_zb; /* The target block to prefetch. */
	int dpa_epbs; /* Entries (blkptr_t's) Per Block Shift. */
	int dpa_curlevel; /* The current level that we're reading */
	dnode_t dpa_dnode; / The dnode associated with the prefetch */
	zio_priority_t dpa_prio; /* The priority I/Os should be issued at. */
	zio_t dpa_zio; / The parent zio_t for all prefetches. */
	arc_flags_t dpa_aflags; /* Flags to pass to the final prefetch. */
	dbuf_prefetch_fn dpa_cb; /* prefetch completion callback */
	void dpa_arg; / prefetch completion arg */
	} dbuf_prefetch_arg_t;

	static void
	dbuf_prefetch_fini(dbuf_prefetch_arg_t *dpa, boolean_t io_done)
	{
	if (dpa->dpa_cb != NULL)
	dpa->dpa_cb(dpa->dpa_arg, io_done);
	kmem_free(dpa, sizeof (*dpa));
	}

	static void
	dbuf_issue_final_prefetch_done(zio_t zio, const zbookmark_phys_t zb,
	const blkptr_t iobp, arc_buf_t abuf, void *private)
	{
	dbuf_prefetch_arg_t *dpa = private;

	dbuf_prefetch_fini(dpa, B_TRUE);
	if (abuf != NULL)
	arc_buf_destroy(abuf, private);
	}

	/*
	* Actually issue the prefetch read for the block given.
	*/
	static void
	dbuf_issue_final_prefetch(dbuf_prefetch_arg_t dpa, blkptr_t bp)
	{
	ASSERT(!BP_IS_REDACTED(bp) \|\|
	dsl_dataset_feature_is_active(
	dpa->dpa_dnode->dn_objset->os_dsl_dataset,
	SPA_FEATURE_REDACTED_DATASETS));

	if (BP_IS_HOLE(bp) \|\| BP_IS_EMBEDDED(bp) \|\| BP_IS_REDACTED(bp))
	return (dbuf_prefetch_fini(dpa, B_FALSE));

	int zio_flags = ZIO_FLAG_CANFAIL \| ZIO_FLAG_SPECULATIVE;
	arc_flags_t aflags =
	dpa->dpa_aflags \| ARC_FLAG_NOWAIT \| ARC_FLAG_PREFETCH \|
	ARC_FLAG_NO_BUF;

	/* dnodes are always read as raw and then converted later */
	if (BP_GET_TYPE(bp) == DMU_OT_DNODE && BP_IS_PROTECTED(bp) &&
	dpa->dpa_curlevel == 0)
	zio_flags \|= ZIO_FLAG_RAW;

	ASSERT3U(dpa->dpa_curlevel, ==, BP_GET_LEVEL(bp));
	ASSERT3U(dpa->dpa_curlevel, ==, dpa->dpa_zb.zb_level);
	ASSERT(dpa->dpa_zio != NULL);
	(void) arc_read(dpa->dpa_zio, dpa->dpa_spa, bp,
	dbuf_issue_final_prefetch_done, dpa,
	dpa->dpa_prio, zio_flags, &aflags, &dpa->dpa_zb);
	}

	/*
	* Called when an indirect block above our prefetch target is read in. This
	* will either read in the next indirect block down the tree or issue the actual
	* prefetch if the next block down is our target.
	*/
	static void
	dbuf_prefetch_indirect_done(zio_t zio, const zbookmark_phys_t zb,
	const blkptr_t iobp, arc_buf_t abuf, void *private)
	{
	dbuf_prefetch_arg_t *dpa = private;

	ASSERT3S(dpa->dpa_zb.zb_level, <, dpa->dpa_curlevel);
	ASSERT3S(dpa->dpa_curlevel, >, 0);

	if (abuf == NULL) {
	ASSERT(zio == NULL \|\| zio->io_error != 0);
	return (dbuf_prefetch_fini(dpa, B_TRUE));
	}
	ASSERT(zio == NULL \|\| zio->io_error == 0);

	/*
	* The dpa_dnode is only valid if we are called with a NULL
	* zio. This indicates that the arc_read() returned without
	* first calling zio_read() to issue a physical read. Once
	* a physical read is made the dpa_dnode must be invalidated
	* as the locks guarding it may have been dropped. If the
	* dpa_dnode is still valid, then we want to add it to the dbuf
	* cache. To do so, we must hold the dbuf associated with the block
	* we just prefetched, read its contents so that we associate it
	* with an arc_buf_t, and then release it.
	*/
	if (zio != NULL) {
	ASSERT3S(BP_GET_LEVEL(zio->io_bp), ==, dpa->dpa_curlevel);
	if (zio->io_flags & ZIO_FLAG_RAW_COMPRESS) {
	ASSERT3U(BP_GET_PSIZE(zio->io_bp), ==, zio->io_size);
	} else {
	ASSERT3U(BP_GET_LSIZE(zio->io_bp), ==, zio->io_size);
	}
	ASSERT3P(zio->io_spa, ==, dpa->dpa_spa);

	dpa->dpa_dnode = NULL;
	} else if (dpa->dpa_dnode != NULL) {
	uint64_t curblkid = dpa->dpa_zb.zb_blkid >>
	(dpa->dpa_epbs * (dpa->dpa_curlevel -
	dpa->dpa_zb.zb_level));
	dmu_buf_impl_t *db = dbuf_hold_level(dpa->dpa_dnode,
	dpa->dpa_curlevel, curblkid, FTAG);
	if (db == NULL) {
	arc_buf_destroy(abuf, private);
	return (dbuf_prefetch_fini(dpa, B_TRUE));
	}
	(void) dbuf_read(db, NULL,
	DB_RF_MUST_SUCCEED \| DB_RF_NOPREFETCH \| DB_RF_HAVESTRUCT);
	dbuf_rele(db, FTAG);
	}

	dpa->dpa_curlevel--;
	uint64_t nextblkid = dpa->dpa_zb.zb_blkid >>
	(dpa->dpa_epbs * (dpa->dpa_curlevel - dpa->dpa_zb.zb_level));
	blkptr_t bp = ((blkptr_t )abuf->b_data) +
	P2PHASE(nextblkid, 1ULL << dpa->dpa_epbs);

	ASSERT(!BP_IS_REDACTED(bp) \|\|
	dsl_dataset_feature_is_active(
	dpa->dpa_dnode->dn_objset->os_dsl_dataset,
	SPA_FEATURE_REDACTED_DATASETS));
	if (BP_IS_HOLE(bp) \|\| BP_IS_REDACTED(bp)) {
	dbuf_prefetch_fini(dpa, B_TRUE);
	} else if (dpa->dpa_curlevel == dpa->dpa_zb.zb_level) {
	ASSERT3U(nextblkid, ==, dpa->dpa_zb.zb_blkid);
	dbuf_issue_final_prefetch(dpa, bp);
	} else {
	arc_flags_t iter_aflags = ARC_FLAG_NOWAIT;
	zbookmark_phys_t zb;

	/* flag if L2ARC eligible, l2arc_noprefetch then decides */
	if (dpa->dpa_aflags & ARC_FLAG_L2CACHE)
	iter_aflags \|= ARC_FLAG_L2CACHE;

	ASSERT3U(dpa->dpa_curlevel, ==, BP_GET_LEVEL(bp));

	SET_BOOKMARK(&zb, dpa->dpa_zb.zb_objset,
	dpa->dpa_zb.zb_object, dpa->dpa_curlevel, nextblkid);

	(void) arc_read(dpa->dpa_zio, dpa->dpa_spa,
	bp, dbuf_prefetch_indirect_done, dpa, dpa->dpa_prio,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_SPECULATIVE,
	&iter_aflags, &zb);
	}

	arc_buf_destroy(abuf, private);
	}

	/*
	* Issue prefetch reads for the given block on the given level. If the indirect
	* blocks above that block are not in memory, we will read them in
	* asynchronously. As a result, this call never blocks waiting for a read to
	* complete. Note that the prefetch might fail if the dataset is encrypted and
	* the encryption key is unmapped before the IO completes.
	*/
	int
	dbuf_prefetch_impl(dnode_t *dn, int64_t level, uint64_t blkid,
	zio_priority_t prio, arc_flags_t aflags, dbuf_prefetch_fn cb,
	void *arg)
	{
	blkptr_t bp;
	int epbs, nlevels, curlevel;
	uint64_t curblkid;

	ASSERT(blkid != DMU_BONUS_BLKID);
	ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));

	if (blkid > dn->dn_maxblkid)
	goto no_issue;

	if (level == 0 && dnode_block_freed(dn, blkid))
	goto no_issue;

	/*
	* This dnode hasn't been written to disk yet, so there's nothing to
	* prefetch.
	*/
	nlevels = dn->dn_phys->dn_nlevels;
	if (level >= nlevels \|\| dn->dn_phys->dn_nblkptr == 0)
	goto no_issue;

	epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT;
	if (dn->dn_phys->dn_maxblkid < blkid << (epbs * level))
	goto no_issue;

	dmu_buf_impl_t *db = dbuf_find(dn->dn_objset, dn->dn_object,
	level, blkid);
	if (db != NULL) {
	mutex_exit(&db->db_mtx);
	/*
	* This dbuf already exists. It is either CACHED, or
	* (we assume) about to be read or filled.
	*/
	goto no_issue;
	}

	/*
	* Find the closest ancestor (indirect block) of the target block
	* that is present in the cache. In this indirect block, we will
	* find the bp that is at curlevel, curblkid.
	*/
	curlevel = level;
	curblkid = blkid;
	while (curlevel < nlevels - 1) {
	int parent_level = curlevel + 1;
	uint64_t parent_blkid = curblkid >> epbs;
	dmu_buf_impl_t *db;

	if (dbuf_hold_impl(dn, parent_level, parent_blkid,
	FALSE, TRUE, FTAG, &db) == 0) {
	blkptr_t *bpp = db->db_buf->b_data;
	bp = bpp[P2PHASE(curblkid, 1 << epbs)];
	dbuf_rele(db, FTAG);
	break;
	}

	curlevel = parent_level;
	curblkid = parent_blkid;
	}

	if (curlevel == nlevels - 1) {
	/* No cached indirect blocks found. */
	ASSERT3U(curblkid, <, dn->dn_phys->dn_nblkptr);
	bp = dn->dn_phys->dn_blkptr[curblkid];
	}
	ASSERT(!BP_IS_REDACTED(&bp) \|\|
	dsl_dataset_feature_is_active(dn->dn_objset->os_dsl_dataset,
	SPA_FEATURE_REDACTED_DATASETS));
	if (BP_IS_HOLE(&bp) \|\| BP_IS_REDACTED(&bp))
	goto no_issue;

	ASSERT3U(curlevel, ==, BP_GET_LEVEL(&bp));

	zio_t *pio = zio_root(dmu_objset_spa(dn->dn_objset), NULL, NULL,
	ZIO_FLAG_CANFAIL);

	dbuf_prefetch_arg_t dpa = kmem_zalloc(sizeof (dpa), KM_SLEEP);
	dsl_dataset_t *ds = dn->dn_objset->os_dsl_dataset;
	SET_BOOKMARK(&dpa->dpa_zb, ds != NULL ? ds->ds_object : DMU_META_OBJSET,
	dn->dn_object, level, blkid);
	dpa->dpa_curlevel = curlevel;
	dpa->dpa_prio = prio;
	dpa->dpa_aflags = aflags;
	dpa->dpa_spa = dn->dn_objset->os_spa;
	dpa->dpa_dnode = dn;
	dpa->dpa_epbs = epbs;
	dpa->dpa_zio = pio;
	dpa->dpa_cb = cb;
	dpa->dpa_arg = arg;

	/* flag if L2ARC eligible, l2arc_noprefetch then decides */
	if (dnode_level_is_l2cacheable(&bp, dn, level))
	dpa->dpa_aflags \|= ARC_FLAG_L2CACHE;

	/*
	* If we have the indirect just above us, no need to do the asynchronous
	* prefetch chain; we'll just run the last step ourselves. If we're at
	* a higher level, though, we want to issue the prefetches for all the
	* indirect blocks asynchronously, so we can go on with whatever we were
	* doing.
	*/
	if (curlevel == level) {
	ASSERT3U(curblkid, ==, blkid);
	dbuf_issue_final_prefetch(dpa, &bp);
	} else {
	arc_flags_t iter_aflags = ARC_FLAG_NOWAIT;
	zbookmark_phys_t zb;

	/* flag if L2ARC eligible, l2arc_noprefetch then decides */
	if (dnode_level_is_l2cacheable(&bp, dn, level))
	iter_aflags \|= ARC_FLAG_L2CACHE;

	SET_BOOKMARK(&zb, ds != NULL ? ds->ds_object : DMU_META_OBJSET,
	dn->dn_object, curlevel, curblkid);
	(void) arc_read(dpa->dpa_zio, dpa->dpa_spa,
	&bp, dbuf_prefetch_indirect_done, dpa, prio,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_SPECULATIVE,
	&iter_aflags, &zb);
	}
	/*
	* We use pio here instead of dpa_zio since it's possible that
	* dpa may have already been freed.
	*/
	zio_nowait(pio);
	return (1);
	no_issue:
	if (cb != NULL)
	cb(arg, B_FALSE);
	return (0);
	}

	int
	dbuf_prefetch(dnode_t *dn, int64_t level, uint64_t blkid, zio_priority_t prio,
	arc_flags_t aflags)
	{

	return (dbuf_prefetch_impl(dn, level, blkid, prio, aflags, NULL, NULL));
	}

	/*
	* Helper function for dbuf_hold_impl() to copy a buffer. Handles
	* the case of encrypted, compressed and uncompressed buffers by
	* allocating the new buffer, respectively, with arc_alloc_raw_buf(),
	* arc_alloc_compressed_buf() or arc_alloc_buf().*
	*
	* NOTE: Declared noinline to avoid stack bloat in dbuf_hold_impl().
	*/
	noinline static void
	dbuf_hold_copy(dnode_t dn, dmu_buf_impl_t db)
	{
	dbuf_dirty_record_t *dr = db->db_data_pending;
	arc_buf_t *data = dr->dt.dl.dr_data;
	enum zio_compress compress_type = arc_get_compression(data);
	uint8_t complevel = arc_get_complevel(data);

	if (arc_is_encrypted(data)) {
	boolean_t byteorder;
	uint8_t salt[ZIO_DATA_SALT_LEN];
	uint8_t iv[ZIO_DATA_IV_LEN];
	uint8_t mac[ZIO_DATA_MAC_LEN];

	arc_get_raw_params(data, &byteorder, salt, iv, mac);
	dbuf_set_data(db, arc_alloc_raw_buf(dn->dn_objset->os_spa, db,
	dmu_objset_id(dn->dn_objset), byteorder, salt, iv, mac,
	dn->dn_type, arc_buf_size(data), arc_buf_lsize(data),
	compress_type, complevel));
	} else if (compress_type != ZIO_COMPRESS_OFF) {
	dbuf_set_data(db, arc_alloc_compressed_buf(
	dn->dn_objset->os_spa, db, arc_buf_size(data),
	arc_buf_lsize(data), compress_type, complevel));
	} else {
	dbuf_set_data(db, arc_alloc_buf(dn->dn_objset->os_spa, db,
	DBUF_GET_BUFC_TYPE(db), db->db.db_size));
	}

	rw_enter(&db->db_rwlock, RW_WRITER);
	bcopy(data->b_data, db->db.db_data, arc_buf_size(data));
	rw_exit(&db->db_rwlock);
	}

	/*
	* Returns with db_holds incremented, and db_mtx not held.
	* Note: dn_struct_rwlock must be held.
	*/
	int
	dbuf_hold_impl(dnode_t *dn, uint8_t level, uint64_t blkid,
	boolean_t fail_sparse, boolean_t fail_uncached,
	void tag, dmu_buf_impl_t *dbp)
	{
	dmu_buf_impl_t db, parent = NULL;

	/* If the pool has been created, verify the tx_sync_lock is not held */
	spa_t *spa = dn->dn_objset->os_spa;
	dsl_pool_t *dp = spa->spa_dsl_pool;
	if (dp != NULL) {
	ASSERT(!MUTEX_HELD(&dp->dp_tx.tx_sync_lock));
	}

	ASSERT(blkid != DMU_BONUS_BLKID);
	ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock));
	ASSERT3U(dn->dn_nlevels, >, level);

	*dbp = NULL;

	/* dbuf_find() returns with db_mtx held */
	db = dbuf_find(dn->dn_objset, dn->dn_object, level, blkid);

	if (db == NULL) {
	blkptr_t *bp = NULL;
	int err;

	if (fail_uncached)
	return (SET_ERROR(ENOENT));

	ASSERT3P(parent, ==, NULL);
	err = dbuf_findbp(dn, level, blkid, fail_sparse, &parent, &bp);
	if (fail_sparse) {
	if (err == 0 && bp && BP_IS_HOLE(bp))
	err = SET_ERROR(ENOENT);
	if (err) {
	if (parent)
	dbuf_rele(parent, NULL);
	return (err);
	}
	}
	if (err && err != ENOENT)
	return (err);
	db = dbuf_create(dn, level, blkid, parent, bp);
	}

	if (fail_uncached && db->db_state != DB_CACHED) {
	mutex_exit(&db->db_mtx);
	return (SET_ERROR(ENOENT));
	}

	if (db->db_buf != NULL) {
	arc_buf_access(db->db_buf);
	ASSERT3P(db->db.db_data, ==, db->db_buf->b_data);
	}

	ASSERT(db->db_buf == NULL \|\| arc_referenced(db->db_buf));

	/*
	* If this buffer is currently syncing out, and we are
	* still referencing it from db_data, we need to make a copy
	* of it in case we decide we want to dirty it again in this txg.
	*/
	if (db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID &&
	dn->dn_object != DMU_META_DNODE_OBJECT &&
	db->db_state == DB_CACHED && db->db_data_pending) {
	dbuf_dirty_record_t *dr = db->db_data_pending;
	if (dr->dt.dl.dr_data == db->db_buf)
	dbuf_hold_copy(dn, db);
	}

	if (multilist_link_active(&db->db_cache_link)) {
	ASSERT(zfs_refcount_is_zero(&db->db_holds));
	ASSERT(db->db_caching_status == DB_DBUF_CACHE \|\|
	db->db_caching_status == DB_DBUF_METADATA_CACHE);

	multilist_remove(&dbuf_caches[db->db_caching_status].cache, db);
	(void) zfs_refcount_remove_many(
	&dbuf_caches[db->db_caching_status].size,
	db->db.db_size, db);

	if (db->db_caching_status == DB_DBUF_METADATA_CACHE) {
	DBUF_STAT_BUMPDOWN(metadata_cache_count);
	} else {
	DBUF_STAT_BUMPDOWN(cache_levels[db->db_level]);
	DBUF_STAT_BUMPDOWN(cache_count);
	DBUF_STAT_DECR(cache_levels_bytes[db->db_level],
	db->db.db_size);
	}
	db->db_caching_status = DB_NO_CACHE;
	}
	(void) zfs_refcount_add(&db->db_holds, tag);
	DBUF_VERIFY(db);
	mutex_exit(&db->db_mtx);

	/* NOTE: we can't rele the parent until after we drop the db_mtx */
	if (parent)
	dbuf_rele(parent, NULL);

	ASSERT3P(DB_DNODE(db), ==, dn);
	ASSERT3U(db->db_blkid, ==, blkid);
	ASSERT3U(db->db_level, ==, level);
	*dbp = db;

	return (0);
	}

	dmu_buf_impl_t *
	dbuf_hold(dnode_t dn, uint64_t blkid, void tag)
	{
	return (dbuf_hold_level(dn, 0, blkid, tag));
	}

	dmu_buf_impl_t *
	dbuf_hold_level(dnode_t dn, int level, uint64_t blkid, void tag)
	{
	dmu_buf_impl_t *db;
	int err = dbuf_hold_impl(dn, level, blkid, FALSE, FALSE, tag, &db);
	return (err ? NULL : db);
	}

	void
	dbuf_create_bonus(dnode_t *dn)
	{
	ASSERT(RW_WRITE_HELD(&dn->dn_struct_rwlock));

	ASSERT(dn->dn_bonus == NULL);
	dn->dn_bonus = dbuf_create(dn, 0, DMU_BONUS_BLKID, dn->dn_dbuf, NULL);
	}

	int
	dbuf_spill_set_blksz(dmu_buf_t db_fake, uint64_t blksz, dmu_tx_t tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;

	if (db->db_blkid != DMU_SPILL_BLKID)
	return (SET_ERROR(ENOTSUP));
	if (blksz == 0)
	blksz = SPA_MINBLOCKSIZE;
	ASSERT3U(blksz, <=, spa_maxblocksize(dmu_objset_spa(db->db_objset)));
	blksz = P2ROUNDUP(blksz, SPA_MINBLOCKSIZE);

	dbuf_new_size(db, blksz, tx);

	return (0);
	}

	void
	dbuf_rm_spill(dnode_t dn, dmu_tx_t tx)
	{
	dbuf_free_range(dn, DMU_SPILL_BLKID, DMU_SPILL_BLKID, tx);
	}

	#pragma weak dmu_buf_add_ref = dbuf_add_ref
	void
	dbuf_add_ref(dmu_buf_impl_t db, void tag)
	{
	int64_t holds = zfs_refcount_add(&db->db_holds, tag);
	VERIFY3S(holds, >, 1);
	}

	#pragma weak dmu_buf_try_add_ref = dbuf_try_add_ref
	boolean_t
	dbuf_try_add_ref(dmu_buf_t db_fake, objset_t os, uint64_t obj, uint64_t blkid,
	void *tag)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;
	dmu_buf_impl_t *found_db;
	boolean_t result = B_FALSE;

	if (blkid == DMU_BONUS_BLKID)
	found_db = dbuf_find_bonus(os, obj);
	else
	found_db = dbuf_find(os, obj, 0, blkid);

	if (found_db != NULL) {
	if (db == found_db && dbuf_refcount(db) > db->db_dirtycnt) {
	(void) zfs_refcount_add(&db->db_holds, tag);
	result = B_TRUE;
	}
	mutex_exit(&found_db->db_mtx);
	}
	return (result);
	}

	/*
	* If you call dbuf_rele() you had better not be referencing the dnode handle
	* unless you have some other direct or indirect hold on the dnode. (An indirect
	* hold is a hold on one of the dnode's dbufs, including the bonus buffer.)
	* Without that, the dbuf_rele() could lead to a dnode_rele() followed by the
	* dnode's parent dbuf evicting its dnode handles.
	*/
	void
	dbuf_rele(dmu_buf_impl_t db, void tag)
	{
	mutex_enter(&db->db_mtx);
	dbuf_rele_and_unlock(db, tag, B_FALSE);
	}

	void
	dmu_buf_rele(dmu_buf_t db, void tag)
	{
	dbuf_rele((dmu_buf_impl_t *)db, tag);
	}

	/*
	* dbuf_rele() for an already-locked dbuf. This is necessary to allow
	* db_dirtycnt and db_holds to be updated atomically. The 'evicting'
	* argument should be set if we are already in the dbuf-evicting code
	* path, in which case we don't want to recursively evict. This allows us to
	* avoid deeply nested stacks that would have a call flow similar to this:
	*
	* dbuf_rele()-->dbuf_rele_and_unlock()-->dbuf_evict_notify()
	* ^ \|
	* \| \|
	* +-----dbuf_destroy()<--dbuf_evict_one()<--------+
	*
	*/
	void
	dbuf_rele_and_unlock(dmu_buf_impl_t db, void tag, boolean_t evicting)
	{
	int64_t holds;
	uint64_t size;

	ASSERT(MUTEX_HELD(&db->db_mtx));
	DBUF_VERIFY(db);

	/*
	* Remove the reference to the dbuf before removing its hold on the
	* dnode so we can guarantee in dnode_move() that a referenced bonus
	* buffer has a corresponding dnode hold.
	*/
	holds = zfs_refcount_remove(&db->db_holds, tag);
	ASSERT(holds >= 0);

	/*
	* We can't freeze indirects if there is a possibility that they
	* may be modified in the current syncing context.
	*/
	if (db->db_buf != NULL &&
	holds == (db->db_level == 0 ? db->db_dirtycnt : 0)) {
	arc_buf_freeze(db->db_buf);
	}

	if (holds == db->db_dirtycnt &&
	db->db_level == 0 && db->db_user_immediate_evict)
	dbuf_evict_user(db);

	if (holds == 0) {
	if (db->db_blkid == DMU_BONUS_BLKID) {
	dnode_t *dn;
	boolean_t evict_dbuf = db->db_pending_evict;

	/*
	* If the dnode moves here, we cannot cross this
	* barrier until the move completes.
	*/
	DB_DNODE_ENTER(db);

	dn = DB_DNODE(db);
	atomic_dec_32(&dn->dn_dbufs_count);

	/*
	* Decrementing the dbuf count means that the bonus
	* buffer's dnode hold is no longer discounted in
	* dnode_move(). The dnode cannot move until after
	* the dnode_rele() below.
	*/
	DB_DNODE_EXIT(db);

	/*
	* Do not reference db after its lock is dropped.
	* Another thread may evict it.
	*/
	mutex_exit(&db->db_mtx);

	if (evict_dbuf)
	dnode_evict_bonus(dn);

	dnode_rele(dn, db);
	} else if (db->db_buf == NULL) {
	/*
	* This is a special case: we never associated this
	* dbuf with any data allocated from the ARC.
	*/
	ASSERT(db->db_state == DB_UNCACHED \|\|
	db->db_state == DB_NOFILL);
	dbuf_destroy(db);
	} else if (arc_released(db->db_buf)) {
	/*
	* This dbuf has anonymous data associated with it.
	*/
	dbuf_destroy(db);
	} else {
	boolean_t do_arc_evict = B_FALSE;
	blkptr_t bp;
	spa_t *spa = dmu_objset_spa(db->db_objset);

	if (!DBUF_IS_CACHEABLE(db) &&
	db->db_blkptr != NULL &&
	!BP_IS_HOLE(db->db_blkptr) &&
	!BP_IS_EMBEDDED(db->db_blkptr)) {
	do_arc_evict = B_TRUE;
	bp = *db->db_blkptr;
	}

	if (!DBUF_IS_CACHEABLE(db) \|\|
	db->db_pending_evict) {
	dbuf_destroy(db);
	} else if (!multilist_link_active(&db->db_cache_link)) {
	ASSERT3U(db->db_caching_status, ==,
	DB_NO_CACHE);

	dbuf_cached_state_t dcs =
	dbuf_include_in_metadata_cache(db) ?
	DB_DBUF_METADATA_CACHE : DB_DBUF_CACHE;
	db->db_caching_status = dcs;

	multilist_insert(&dbuf_caches[dcs].cache, db);
	uint64_t db_size = db->db.db_size;
	size = zfs_refcount_add_many(
	&dbuf_caches[dcs].size, db_size, db);
	uint8_t db_level = db->db_level;
	mutex_exit(&db->db_mtx);

	if (dcs == DB_DBUF_METADATA_CACHE) {
	DBUF_STAT_BUMP(metadata_cache_count);
	DBUF_STAT_MAX(
	metadata_cache_size_bytes_max,
	size);
	} else {
	DBUF_STAT_BUMP(cache_count);
	DBUF_STAT_MAX(cache_size_bytes_max,
	size);
	DBUF_STAT_BUMP(cache_levels[db_level]);
	DBUF_STAT_INCR(
	cache_levels_bytes[db_level],
	db_size);
	}

	if (dcs == DB_DBUF_CACHE && !evicting)
	dbuf_evict_notify(size);
	}

	if (do_arc_evict)
	arc_freed(spa, &bp);
	}
	} else {
	mutex_exit(&db->db_mtx);
	}

	}

	#pragma weak dmu_buf_refcount = dbuf_refcount
	uint64_t
	dbuf_refcount(dmu_buf_impl_t *db)
	{
	return (zfs_refcount_count(&db->db_holds));
	}

	uint64_t
	dmu_buf_user_refcount(dmu_buf_t *db_fake)
	{
	uint64_t holds;
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;

	mutex_enter(&db->db_mtx);
	ASSERT3U(zfs_refcount_count(&db->db_holds), >=, db->db_dirtycnt);
	holds = zfs_refcount_count(&db->db_holds) - db->db_dirtycnt;
	mutex_exit(&db->db_mtx);

	return (holds);
	}

	void *
	dmu_buf_replace_user(dmu_buf_t db_fake, dmu_buf_user_t old_user,
	dmu_buf_user_t *new_user)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;

	mutex_enter(&db->db_mtx);
	dbuf_verify_user(db, DBVU_NOT_EVICTING);
	if (db->db_user == old_user)
	db->db_user = new_user;
	else
	old_user = db->db_user;
	dbuf_verify_user(db, DBVU_NOT_EVICTING);
	mutex_exit(&db->db_mtx);

	return (old_user);
	}

	void *
	dmu_buf_set_user(dmu_buf_t db_fake, dmu_buf_user_t user)
	{
	return (dmu_buf_replace_user(db_fake, NULL, user));
	}

	void *
	dmu_buf_set_user_ie(dmu_buf_t db_fake, dmu_buf_user_t user)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;

	db->db_user_immediate_evict = TRUE;
	return (dmu_buf_set_user(db_fake, user));
	}

	void *
	dmu_buf_remove_user(dmu_buf_t db_fake, dmu_buf_user_t user)
	{
	return (dmu_buf_replace_user(db_fake, user, NULL));
	}

	void *
	dmu_buf_get_user(dmu_buf_t *db_fake)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;

	dbuf_verify_user(db, DBVU_NOT_EVICTING);
	return (db->db_user);
	}

	void
	dmu_buf_user_evict_wait()
	{
	taskq_wait(dbu_evict_taskq);
	}

	blkptr_t *
	dmu_buf_get_blkptr(dmu_buf_t *db)
	{
	dmu_buf_impl_t dbi = (dmu_buf_impl_t )db;
	return (dbi->db_blkptr);
	}

	objset_t *
	dmu_buf_get_objset(dmu_buf_t *db)
	{
	dmu_buf_impl_t dbi = (dmu_buf_impl_t )db;
	return (dbi->db_objset);
	}

	dnode_t *
	dmu_buf_dnode_enter(dmu_buf_t *db)
	{
	dmu_buf_impl_t dbi = (dmu_buf_impl_t )db;
	DB_DNODE_ENTER(dbi);
	return (DB_DNODE(dbi));
	}

	void
	dmu_buf_dnode_exit(dmu_buf_t *db)
	{
	dmu_buf_impl_t dbi = (dmu_buf_impl_t )db;
	DB_DNODE_EXIT(dbi);
	}

	static void
	dbuf_check_blkptr(dnode_t dn, dmu_buf_impl_t db)
	{
	/* ASSERT(dmu_tx_is_syncing(tx) */
	ASSERT(MUTEX_HELD(&db->db_mtx));

	if (db->db_blkptr != NULL)
	return;

	if (db->db_blkid == DMU_SPILL_BLKID) {
	db->db_blkptr = DN_SPILL_BLKPTR(dn->dn_phys);
	BP_ZERO(db->db_blkptr);
	return;
	}
	if (db->db_level == dn->dn_phys->dn_nlevels-1) {
	/*
	* This buffer was allocated at a time when there was
	* no available blkptrs from the dnode, or it was
	* inappropriate to hook it in (i.e., nlevels mismatch).
	*/
	ASSERT(db->db_blkid < dn->dn_phys->dn_nblkptr);
	ASSERT(db->db_parent == NULL);
	db->db_parent = dn->dn_dbuf;
	db->db_blkptr = &dn->dn_phys->dn_blkptr[db->db_blkid];
	DBUF_VERIFY(db);
	} else {
	dmu_buf_impl_t *parent = db->db_parent;
	int epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT;

	ASSERT(dn->dn_phys->dn_nlevels > 1);
	if (parent == NULL) {
	mutex_exit(&db->db_mtx);
	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	parent = dbuf_hold_level(dn, db->db_level + 1,
	db->db_blkid >> epbs, db);
	rw_exit(&dn->dn_struct_rwlock);
	mutex_enter(&db->db_mtx);
	db->db_parent = parent;
	}
	db->db_blkptr = (blkptr_t *)parent->db.db_data +
	(db->db_blkid & ((1ULL << epbs) - 1));
	DBUF_VERIFY(db);
	}
	}

	static void
	dbuf_sync_bonus(dbuf_dirty_record_t dr, dmu_tx_t tx)
	{
	dmu_buf_impl_t *db = dr->dr_dbuf;
	void *data = dr->dt.dl.dr_data;

	ASSERT0(db->db_level);
	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT(db->db_blkid == DMU_BONUS_BLKID);
	ASSERT(data != NULL);

	dnode_t *dn = dr->dr_dnode;
	ASSERT3U(DN_MAX_BONUS_LEN(dn->dn_phys), <=,
	DN_SLOTS_TO_BONUSLEN(dn->dn_phys->dn_extra_slots + 1));
	bcopy(data, DN_BONUS(dn->dn_phys), DN_MAX_BONUS_LEN(dn->dn_phys));

	dbuf_sync_leaf_verify_bonus_dnode(dr);

	dbuf_undirty_bonus(dr);
	dbuf_rele_and_unlock(db, (void *)(uintptr_t)tx->tx_txg, B_FALSE);
	}

	/*
	* When syncing out a blocks of dnodes, adjust the block to deal with
	* encryption. Normally, we make sure the block is decrypted before writing
	* it. If we have crypt params, then we are writing a raw (encrypted) block,
	* from a raw receive. In this case, set the ARC buf's crypt params so
	* that the BP will be filled with the correct byteorder, salt, iv, and mac.
	*/
	static void
	dbuf_prepare_encrypted_dnode_leaf(dbuf_dirty_record_t *dr)
	{
	int err;
	dmu_buf_impl_t *db = dr->dr_dbuf;

	ASSERT(MUTEX_HELD(&db->db_mtx));
	ASSERT3U(db->db.db_object, ==, DMU_META_DNODE_OBJECT);
	ASSERT3U(db->db_level, ==, 0);

	if (!db->db_objset->os_raw_receive && arc_is_encrypted(db->db_buf)) {
	zbookmark_phys_t zb;

	/*
	* Unfortunately, there is currently no mechanism for
	* syncing context to handle decryption errors. An error
	* here is only possible if an attacker maliciously
	* changed a dnode block and updated the associated
	* checksums going up the block tree.
	*/
	SET_BOOKMARK(&zb, dmu_objset_id(db->db_objset),
	db->db.db_object, db->db_level, db->db_blkid);
	err = arc_untransform(db->db_buf, db->db_objset->os_spa,
	&zb, B_TRUE);
	if (err)
	panic("Invalid dnode block MAC");
	} else if (dr->dt.dl.dr_has_raw_params) {
	(void) arc_release(dr->dt.dl.dr_data, db);
	arc_convert_to_raw(dr->dt.dl.dr_data,
	dmu_objset_id(db->db_objset),
	dr->dt.dl.dr_byteorder, DMU_OT_DNODE,
	dr->dt.dl.dr_salt, dr->dt.dl.dr_iv, dr->dt.dl.dr_mac);
	}
	}

	/*
	* dbuf_sync_indirect() is called recursively from dbuf_sync_list() so it
	* is critical the we not allow the compiler to inline this function in to
	* dbuf_sync_list() thereby drastically bloating the stack usage.
	*/
	noinline static void
	dbuf_sync_indirect(dbuf_dirty_record_t dr, dmu_tx_t tx)
	{
	dmu_buf_impl_t *db = dr->dr_dbuf;
	dnode_t *dn = dr->dr_dnode;

	ASSERT(dmu_tx_is_syncing(tx));

	dprintf_dbuf_bp(db, db->db_blkptr, "blkptr=%p", db->db_blkptr);

	mutex_enter(&db->db_mtx);

	ASSERT(db->db_level > 0);
	DBUF_VERIFY(db);

	/* Read the block if it hasn't been read yet. */
	if (db->db_buf == NULL) {
	mutex_exit(&db->db_mtx);
	(void) dbuf_read(db, NULL, DB_RF_MUST_SUCCEED);
	mutex_enter(&db->db_mtx);
	}
	ASSERT3U(db->db_state, ==, DB_CACHED);
	ASSERT(db->db_buf != NULL);

	/* Indirect block size must match what the dnode thinks it is. */
	ASSERT3U(db->db.db_size, ==, 1<<dn->dn_phys->dn_indblkshift);
	dbuf_check_blkptr(dn, db);

	/* Provide the pending dirty record to child dbufs */
	db->db_data_pending = dr;

	mutex_exit(&db->db_mtx);

	dbuf_write(dr, db->db_buf, tx);

	zio_t *zio = dr->dr_zio;
	mutex_enter(&dr->dt.di.dr_mtx);
	dbuf_sync_list(&dr->dt.di.dr_children, db->db_level - 1, tx);
	ASSERT(list_head(&dr->dt.di.dr_children) == NULL);
	mutex_exit(&dr->dt.di.dr_mtx);
	zio_nowait(zio);
	}

	/*
	* Verify that the size of the data in our bonus buffer does not exceed
	* its recorded size.
	*
	* The purpose of this verification is to catch any cases in development
	* where the size of a phys structure (i.e space_map_phys_t) grows and,
	* due to incorrect feature management, older pools expect to read more
	* data even though they didn't actually write it to begin with.
	*
	* For a example, this would catch an error in the feature logic where we
	* open an older pool and we expect to write the space map histogram of
	* a space map with size SPACE_MAP_SIZE_V0.
	*/
	static void
	dbuf_sync_leaf_verify_bonus_dnode(dbuf_dirty_record_t *dr)
	{
	#ifdef ZFS_DEBUG
	dnode_t *dn = dr->dr_dnode;

	/*
	* Encrypted bonus buffers can have data past their bonuslen.
	* Skip the verification of these blocks.
	*/
	if (DMU_OT_IS_ENCRYPTED(dn->dn_bonustype))
	return;

	uint16_t bonuslen = dn->dn_phys->dn_bonuslen;
	uint16_t maxbonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots);
	ASSERT3U(bonuslen, <=, maxbonuslen);

	arc_buf_t *datap = dr->dt.dl.dr_data;
	char datap_end = ((char )datap) + bonuslen;
	char datap_max = ((char )datap) + maxbonuslen;

	/* ensure that everything is zero after our data */
	for (; datap_end < datap_max; datap_end++)
	ASSERT(*datap_end == 0);
	#endif
	}

	static blkptr_t *
	dbuf_lightweight_bp(dbuf_dirty_record_t *dr)
	{
	/* This must be a lightweight dirty record. */
	ASSERT3P(dr->dr_dbuf, ==, NULL);
	dnode_t *dn = dr->dr_dnode;

	if (dn->dn_phys->dn_nlevels == 1) {
	VERIFY3U(dr->dt.dll.dr_blkid, <, dn->dn_phys->dn_nblkptr);
	return (&dn->dn_phys->dn_blkptr[dr->dt.dll.dr_blkid]);
	} else {
	dmu_buf_impl_t *parent_db = dr->dr_parent->dr_dbuf;
	int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT;
	VERIFY3U(parent_db->db_level, ==, 1);
	VERIFY3P(parent_db->db_dnode_handle->dnh_dnode, ==, dn);
	VERIFY3U(dr->dt.dll.dr_blkid >> epbs, ==, parent_db->db_blkid);
	blkptr_t *bp = parent_db->db.db_data;
	return (&bp[dr->dt.dll.dr_blkid & ((1 << epbs) - 1)]);
	}
	}

	static void
	dbuf_lightweight_ready(zio_t *zio)
	{
	dbuf_dirty_record_t *dr = zio->io_private;
	blkptr_t *bp = zio->io_bp;

	if (zio->io_error != 0)
	return;

	dnode_t *dn = dr->dr_dnode;

	blkptr_t *bp_orig = dbuf_lightweight_bp(dr);
	spa_t *spa = dmu_objset_spa(dn->dn_objset);
	int64_t delta = bp_get_dsize_sync(spa, bp) -
	bp_get_dsize_sync(spa, bp_orig);
	dnode_diduse_space(dn, delta);

	uint64_t blkid = dr->dt.dll.dr_blkid;
	mutex_enter(&dn->dn_mtx);
	if (blkid > dn->dn_phys->dn_maxblkid) {
	ASSERT0(dn->dn_objset->os_raw_receive);
	dn->dn_phys->dn_maxblkid = blkid;
	}
	mutex_exit(&dn->dn_mtx);

	if (!BP_IS_EMBEDDED(bp)) {
	uint64_t fill = BP_IS_HOLE(bp) ? 0 : 1;
	BP_SET_FILL(bp, fill);
	}

	dmu_buf_impl_t *parent_db;
	EQUIV(dr->dr_parent == NULL, dn->dn_phys->dn_nlevels == 1);
	if (dr->dr_parent == NULL) {
	parent_db = dn->dn_dbuf;
	} else {
	parent_db = dr->dr_parent->dr_dbuf;
	}
	rw_enter(&parent_db->db_rwlock, RW_WRITER);
	bp_orig = bp;
	rw_exit(&parent_db->db_rwlock);
	}

	static void
	dbuf_lightweight_physdone(zio_t *zio)
	{
	dbuf_dirty_record_t *dr = zio->io_private;
	dsl_pool_t *dp = spa_get_dsl(zio->io_spa);
	ASSERT3U(dr->dr_txg, ==, zio->io_txg);

	/*
	* The callback will be called io_phys_children times. Retire one
	* portion of our dirty space each time we are called. Any rounding
	* error will be cleaned up by dbuf_lightweight_done().
	*/
	int delta = dr->dr_accounted / zio->io_phys_children;
	dsl_pool_undirty_space(dp, delta, zio->io_txg);
	}

	static void
	dbuf_lightweight_done(zio_t *zio)
	{
	dbuf_dirty_record_t *dr = zio->io_private;

	VERIFY0(zio->io_error);

	objset_t *os = dr->dr_dnode->dn_objset;
	dmu_tx_t *tx = os->os_synctx;

	if (zio->io_flags & (ZIO_FLAG_IO_REWRITE \| ZIO_FLAG_NOPWRITE)) {
	ASSERT(BP_EQUAL(zio->io_bp, &zio->io_bp_orig));
	} else {
	dsl_dataset_t *ds = os->os_dsl_dataset;
	(void) dsl_dataset_block_kill(ds, &zio->io_bp_orig, tx, B_TRUE);
	dsl_dataset_block_born(ds, zio->io_bp, tx);
	}

	/*
	* See comment in dbuf_write_done().
	*/
	if (zio->io_phys_children == 0) {
	dsl_pool_undirty_space(dmu_objset_pool(os),
	dr->dr_accounted, zio->io_txg);
	} else {
	dsl_pool_undirty_space(dmu_objset_pool(os),
	dr->dr_accounted % zio->io_phys_children, zio->io_txg);
	}

	abd_free(dr->dt.dll.dr_abd);
	kmem_free(dr, sizeof (*dr));
	}

	noinline static void
	dbuf_sync_lightweight(dbuf_dirty_record_t dr, dmu_tx_t tx)
	{
	dnode_t *dn = dr->dr_dnode;
	zio_t *pio;
	if (dn->dn_phys->dn_nlevels == 1) {
	pio = dn->dn_zio;
	} else {
	pio = dr->dr_parent->dr_zio;
	}

	zbookmark_phys_t zb = {
	.zb_objset = dmu_objset_id(dn->dn_objset),
	.zb_object = dn->dn_object,
	.zb_level = 0,
	.zb_blkid = dr->dt.dll.dr_blkid,
	};

	/*
	* See comment in dbuf_write(). This is so that zio->io_bp_orig
	* will have the old BP in dbuf_lightweight_done().
	*/
	dr->dr_bp_copy = *dbuf_lightweight_bp(dr);

	dr->dr_zio = zio_write(pio, dmu_objset_spa(dn->dn_objset),
	dmu_tx_get_txg(tx), &dr->dr_bp_copy, dr->dt.dll.dr_abd,
	dn->dn_datablksz, abd_get_size(dr->dt.dll.dr_abd),
	&dr->dt.dll.dr_props, dbuf_lightweight_ready, NULL,
	dbuf_lightweight_physdone, dbuf_lightweight_done, dr,
	ZIO_PRIORITY_ASYNC_WRITE,
	ZIO_FLAG_MUSTSUCCEED \| dr->dt.dll.dr_flags, &zb);

	zio_nowait(dr->dr_zio);
	}

	/*
	* dbuf_sync_leaf() is called recursively from dbuf_sync_list() so it is
	* critical the we not allow the compiler to inline this function in to
	* dbuf_sync_list() thereby drastically bloating the stack usage.
	*/
	noinline static void
	dbuf_sync_leaf(dbuf_dirty_record_t dr, dmu_tx_t tx)
	{
	arc_buf_t **datap = &dr->dt.dl.dr_data;
	dmu_buf_impl_t *db = dr->dr_dbuf;
	dnode_t *dn = dr->dr_dnode;
	objset_t *os;
	uint64_t txg = tx->tx_txg;

	ASSERT(dmu_tx_is_syncing(tx));

	dprintf_dbuf_bp(db, db->db_blkptr, "blkptr=%p", db->db_blkptr);

	mutex_enter(&db->db_mtx);
	/*
	* To be synced, we must be dirtied. But we
	* might have been freed after the dirty.
	*/
	if (db->db_state == DB_UNCACHED) {
	/* This buffer has been freed since it was dirtied */
	ASSERT(db->db.db_data == NULL);
	} else if (db->db_state == DB_FILL) {
	/* This buffer was freed and is now being re-filled */
	ASSERT(db->db.db_data != dr->dt.dl.dr_data);
	} else {
	ASSERT(db->db_state == DB_CACHED \|\| db->db_state == DB_NOFILL);
	}
	DBUF_VERIFY(db);

	if (db->db_blkid == DMU_SPILL_BLKID) {
	mutex_enter(&dn->dn_mtx);
	if (!(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR)) {
	/*
	* In the previous transaction group, the bonus buffer
	* was entirely used to store the attributes for the
	* dnode which overrode the dn_spill field. However,
	* when adding more attributes to the file a spill
	* block was required to hold the extra attributes.
	*
	* Make sure to clear the garbage left in the dn_spill
	* field from the previous attributes in the bonus
	* buffer. Otherwise, after writing out the spill
	* block to the new allocated dva, it will free
	* the old block pointed to by the invalid dn_spill.
	*/
	db->db_blkptr = NULL;
	}
	dn->dn_phys->dn_flags \|= DNODE_FLAG_SPILL_BLKPTR;
	mutex_exit(&dn->dn_mtx);
	}

	/*
	* If this is a bonus buffer, simply copy the bonus data into the
	* dnode. It will be written out when the dnode is synced (and it
	* will be synced, since it must have been dirty for dbuf_sync to
	* be called).
	*/
	if (db->db_blkid == DMU_BONUS_BLKID) {
	ASSERT(dr->dr_dbuf == db);
	dbuf_sync_bonus(dr, tx);
	return;
	}

	os = dn->dn_objset;

	/*
	* This function may have dropped the db_mtx lock allowing a dmu_sync
	* operation to sneak in. As a result, we need to ensure that we
	* don't check the dr_override_state until we have returned from
	* dbuf_check_blkptr.
	*/
	dbuf_check_blkptr(dn, db);

	/*
	* If this buffer is in the middle of an immediate write,
	* wait for the synchronous IO to complete.
	*/
	while (dr->dt.dl.dr_override_state == DR_IN_DMU_SYNC) {
	ASSERT(dn->dn_object != DMU_META_DNODE_OBJECT);
	cv_wait(&db->db_changed, &db->db_mtx);
	ASSERT(dr->dt.dl.dr_override_state != DR_NOT_OVERRIDDEN);
	}

	/*
	* If this is a dnode block, ensure it is appropriately encrypted
	* or decrypted, depending on what we are writing to it this txg.
	*/
	if (os->os_encrypted && dn->dn_object == DMU_META_DNODE_OBJECT)
	dbuf_prepare_encrypted_dnode_leaf(dr);

	if (db->db_state != DB_NOFILL &&
	dn->dn_object != DMU_META_DNODE_OBJECT &&
	zfs_refcount_count(&db->db_holds) > 1 &&
	dr->dt.dl.dr_override_state != DR_OVERRIDDEN &&
	*datap == db->db_buf) {
	/*
	* If this buffer is currently "in use" (i.e., there
	* are active holds and db_data still references it),
	* then make a copy before we start the write so that
	* any modifications from the open txg will not leak
	* into this write.
	*
	* NOTE: this copy does not need to be made for
	* objects only modified in the syncing context (e.g.
	* DNONE_DNODE blocks).
	*/
	int psize = arc_buf_size(*datap);
	int lsize = arc_buf_lsize(*datap);
	arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db);
	enum zio_compress compress_type = arc_get_compression(*datap);
	uint8_t complevel = arc_get_complevel(*datap);

	if (arc_is_encrypted(*datap)) {
	boolean_t byteorder;
	uint8_t salt[ZIO_DATA_SALT_LEN];
	uint8_t iv[ZIO_DATA_IV_LEN];
	uint8_t mac[ZIO_DATA_MAC_LEN];

	arc_get_raw_params(*datap, &byteorder, salt, iv, mac);
	*datap = arc_alloc_raw_buf(os->os_spa, db,
	dmu_objset_id(os), byteorder, salt, iv, mac,
	dn->dn_type, psize, lsize, compress_type,
	complevel);
	} else if (compress_type != ZIO_COMPRESS_OFF) {
	ASSERT3U(type, ==, ARC_BUFC_DATA);
	*datap = arc_alloc_compressed_buf(os->os_spa, db,
	psize, lsize, compress_type, complevel);
	} else {
	*datap = arc_alloc_buf(os->os_spa, db, type, psize);
	}
	bcopy(db->db.db_data, (*datap)->b_data, psize);
	}
	db->db_data_pending = dr;

	mutex_exit(&db->db_mtx);

	dbuf_write(dr, *datap, tx);

	ASSERT(!list_link_active(&dr->dr_dirty_node));
	if (dn->dn_object == DMU_META_DNODE_OBJECT) {
	list_insert_tail(&dn->dn_dirty_records[txg & TXG_MASK], dr);
	} else {
	zio_nowait(dr->dr_zio);
	}
	}

	void
	dbuf_sync_list(list_t list, int level, dmu_tx_t tx)
	{
	dbuf_dirty_record_t *dr;

	while ((dr = list_head(list))) {
	if (dr->dr_zio != NULL) {
	/*
	* If we find an already initialized zio then we
	* are processing the meta-dnode, and we have finished.
	* The dbufs for all dnodes are put back on the list
	* during processing, so that we can zio_wait()
	* these IOs after initiating all child IOs.
	*/
	ASSERT3U(dr->dr_dbuf->db.db_object, ==,
	DMU_META_DNODE_OBJECT);
	break;
	}
	list_remove(list, dr);
	if (dr->dr_dbuf == NULL) {
	dbuf_sync_lightweight(dr, tx);
	} else {
	if (dr->dr_dbuf->db_blkid != DMU_BONUS_BLKID &&
	dr->dr_dbuf->db_blkid != DMU_SPILL_BLKID) {
	VERIFY3U(dr->dr_dbuf->db_level, ==, level);
	}
	if (dr->dr_dbuf->db_level > 0)
	dbuf_sync_indirect(dr, tx);
	else
	dbuf_sync_leaf(dr, tx);
	}
	}
	}

	/* ARGSUSED */
	static void
	dbuf_write_ready(zio_t zio, arc_buf_t buf, void *vdb)
	{
	dmu_buf_impl_t *db = vdb;
	dnode_t *dn;
	blkptr_t *bp = zio->io_bp;
	blkptr_t *bp_orig = &zio->io_bp_orig;
	spa_t *spa = zio->io_spa;
	int64_t delta;
	uint64_t fill = 0;
	int i;

	ASSERT3P(db->db_blkptr, !=, NULL);
	ASSERT3P(&db->db_data_pending->dr_bp_copy, ==, bp);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	delta = bp_get_dsize_sync(spa, bp) - bp_get_dsize_sync(spa, bp_orig);
	dnode_diduse_space(dn, delta - zio->io_prev_space_delta);
	zio->io_prev_space_delta = delta;

	if (bp->blk_birth != 0) {
	ASSERT((db->db_blkid != DMU_SPILL_BLKID &&
	BP_GET_TYPE(bp) == dn->dn_type) \|\|
	(db->db_blkid == DMU_SPILL_BLKID &&
	BP_GET_TYPE(bp) == dn->dn_bonustype) \|\|
	BP_IS_EMBEDDED(bp));
	ASSERT(BP_GET_LEVEL(bp) == db->db_level);
	}

	mutex_enter(&db->db_mtx);

	#ifdef ZFS_DEBUG
	if (db->db_blkid == DMU_SPILL_BLKID) {
	ASSERT(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR);
	ASSERT(!(BP_IS_HOLE(bp)) &&
	db->db_blkptr == DN_SPILL_BLKPTR(dn->dn_phys));
	}
	#endif

	if (db->db_level == 0) {
	mutex_enter(&dn->dn_mtx);
	if (db->db_blkid > dn->dn_phys->dn_maxblkid &&
	db->db_blkid != DMU_SPILL_BLKID) {
	ASSERT0(db->db_objset->os_raw_receive);
	dn->dn_phys->dn_maxblkid = db->db_blkid;
	}
	mutex_exit(&dn->dn_mtx);

	if (dn->dn_type == DMU_OT_DNODE) {
	i = 0;
	while (i < db->db.db_size) {
	dnode_phys_t *dnp =
	(void )(((char )db->db.db_data) + i);

	i += DNODE_MIN_SIZE;
	if (dnp->dn_type != DMU_OT_NONE) {
	fill++;
	i += dnp->dn_extra_slots *
	DNODE_MIN_SIZE;
	}
	}
	} else {
	if (BP_IS_HOLE(bp)) {
	fill = 0;
	} else {
	fill = 1;
	}
	}
	} else {
	blkptr_t *ibp = db->db.db_data;
	ASSERT3U(db->db.db_size, ==, 1<<dn->dn_phys->dn_indblkshift);
	for (i = db->db.db_size >> SPA_BLKPTRSHIFT; i > 0; i--, ibp++) {
	if (BP_IS_HOLE(ibp))
	continue;
	fill += BP_GET_FILL(ibp);
	}
	}
	DB_DNODE_EXIT(db);

	if (!BP_IS_EMBEDDED(bp))
	BP_SET_FILL(bp, fill);

	mutex_exit(&db->db_mtx);

	db_lock_type_t dblt = dmu_buf_lock_parent(db, RW_WRITER, FTAG);
	db->db_blkptr = bp;
	dmu_buf_unlock_parent(db, dblt, FTAG);
	}

	/* ARGSUSED */
	/*
	* This function gets called just prior to running through the compression
	* stage of the zio pipeline. If we're an indirect block comprised of only
	* holes, then we want this indirect to be compressed away to a hole. In
	* order to do that we must zero out any information about the holes that
	* this indirect points to prior to before we try to compress it.
	*/
	static void
	dbuf_write_children_ready(zio_t zio, arc_buf_t buf, void *vdb)
	{
	dmu_buf_impl_t *db = vdb;
	dnode_t *dn;
	blkptr_t *bp;
	unsigned int epbs, i;

	ASSERT3U(db->db_level, >, 0);
	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT;
	ASSERT3U(epbs, <, 31);

	/* Determine if all our children are holes */
	for (i = 0, bp = db->db.db_data; i < 1ULL << epbs; i++, bp++) {
	if (!BP_IS_HOLE(bp))
	break;
	}

	/*
	* If all the children are holes, then zero them all out so that
	* we may get compressed away.
	*/
	if (i == 1ULL << epbs) {
	/*
	* We only found holes. Grab the rwlock to prevent
	* anybody from reading the blocks we're about to
	* zero out.
	*/
	rw_enter(&db->db_rwlock, RW_WRITER);
	bzero(db->db.db_data, db->db.db_size);
	rw_exit(&db->db_rwlock);
	}
	DB_DNODE_EXIT(db);
	}

	/*
	* The SPA will call this callback several times for each zio - once
	* for every physical child i/o (zio->io_phys_children times). This
	* allows the DMU to monitor the progress of each logical i/o. For example,
	* there may be 2 copies of an indirect block, or many fragments of a RAID-Z
	* block. There may be a long delay before all copies/fragments are completed,
	* so this callback allows us to retire dirty space gradually, as the physical
	* i/os complete.
	*/
	/* ARGSUSED */
	static void
	dbuf_write_physdone(zio_t zio, arc_buf_t buf, void *arg)
	{
	dmu_buf_impl_t *db = arg;
	objset_t *os = db->db_objset;
	dsl_pool_t *dp = dmu_objset_pool(os);
	dbuf_dirty_record_t *dr;
	int delta = 0;

	dr = db->db_data_pending;
	ASSERT3U(dr->dr_txg, ==, zio->io_txg);

	/*
	* The callback will be called io_phys_children times. Retire one
	* portion of our dirty space each time we are called. Any rounding
	* error will be cleaned up by dbuf_write_done().
	*/
	delta = dr->dr_accounted / zio->io_phys_children;
	dsl_pool_undirty_space(dp, delta, zio->io_txg);
	}

	/* ARGSUSED */
	static void
	dbuf_write_done(zio_t zio, arc_buf_t buf, void *vdb)
	{
	dmu_buf_impl_t *db = vdb;
	blkptr_t *bp_orig = &zio->io_bp_orig;
	blkptr_t *bp = db->db_blkptr;
	objset_t *os = db->db_objset;
	dmu_tx_t *tx = os->os_synctx;

	ASSERT0(zio->io_error);
	ASSERT(db->db_blkptr == bp);

	/*
	* For nopwrites and rewrites we ensure that the bp matches our
	* original and bypass all the accounting.
	*/
	if (zio->io_flags & (ZIO_FLAG_IO_REWRITE \| ZIO_FLAG_NOPWRITE)) {
	ASSERT(BP_EQUAL(bp, bp_orig));
	} else {
	dsl_dataset_t *ds = os->os_dsl_dataset;
	(void) dsl_dataset_block_kill(ds, bp_orig, tx, B_TRUE);
	dsl_dataset_block_born(ds, bp, tx);
	}

	mutex_enter(&db->db_mtx);

	DBUF_VERIFY(db);

	dbuf_dirty_record_t *dr = db->db_data_pending;
	dnode_t *dn = dr->dr_dnode;
	ASSERT(!list_link_active(&dr->dr_dirty_node));
	ASSERT(dr->dr_dbuf == db);
	ASSERT(list_next(&db->db_dirty_records, dr) == NULL);
	list_remove(&db->db_dirty_records, dr);

	#ifdef ZFS_DEBUG
	if (db->db_blkid == DMU_SPILL_BLKID) {
	ASSERT(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR);
	ASSERT(!(BP_IS_HOLE(db->db_blkptr)) &&
	db->db_blkptr == DN_SPILL_BLKPTR(dn->dn_phys));
	}
	#endif

	if (db->db_level == 0) {
	ASSERT(db->db_blkid != DMU_BONUS_BLKID);
	ASSERT(dr->dt.dl.dr_override_state == DR_NOT_OVERRIDDEN);
	if (db->db_state != DB_NOFILL) {
	if (dr->dt.dl.dr_data != db->db_buf)
	arc_buf_destroy(dr->dt.dl.dr_data, db);
	}
	} else {
	ASSERT(list_head(&dr->dt.di.dr_children) == NULL);
	ASSERT3U(db->db.db_size, ==, 1 << dn->dn_phys->dn_indblkshift);
	if (!BP_IS_HOLE(db->db_blkptr)) {
	int epbs __maybe_unused = dn->dn_phys->dn_indblkshift -
	SPA_BLKPTRSHIFT;
	ASSERT3U(db->db_blkid, <=,
	dn->dn_phys->dn_maxblkid >> (db->db_level * epbs));
	ASSERT3U(BP_GET_LSIZE(db->db_blkptr), ==,
	db->db.db_size);
	}
	mutex_destroy(&dr->dt.di.dr_mtx);
	list_destroy(&dr->dt.di.dr_children);
	}

	cv_broadcast(&db->db_changed);
	ASSERT(db->db_dirtycnt > 0);
	db->db_dirtycnt -= 1;
	db->db_data_pending = NULL;
	dbuf_rele_and_unlock(db, (void *)(uintptr_t)tx->tx_txg, B_FALSE);

	/*
	* If we didn't do a physical write in this ZIO and we
	* still ended up here, it means that the space of the
	* dbuf that we just released (and undirtied) above hasn't
	* been marked as undirtied in the pool's accounting.
	*
	* Thus, we undirty that space in the pool's view of the
	* world here. For physical writes this type of update
	* happens in dbuf_write_physdone().
	*
	* If we did a physical write, cleanup any rounding errors
	* that came up due to writing multiple copies of a block
	* on disk [see dbuf_write_physdone()].
	*/
	if (zio->io_phys_children == 0) {
	dsl_pool_undirty_space(dmu_objset_pool(os),
	dr->dr_accounted, zio->io_txg);
	} else {
	dsl_pool_undirty_space(dmu_objset_pool(os),
	dr->dr_accounted % zio->io_phys_children, zio->io_txg);
	}

	kmem_free(dr, sizeof (dbuf_dirty_record_t));
	}

	static void
	dbuf_write_nofill_ready(zio_t *zio)
	{
	dbuf_write_ready(zio, NULL, zio->io_private);
	}

	static void
	dbuf_write_nofill_done(zio_t *zio)
	{
	dbuf_write_done(zio, NULL, zio->io_private);
	}

	static void
	dbuf_write_override_ready(zio_t *zio)
	{
	dbuf_dirty_record_t *dr = zio->io_private;
	dmu_buf_impl_t *db = dr->dr_dbuf;

	dbuf_write_ready(zio, NULL, db);
	}

	static void
	dbuf_write_override_done(zio_t *zio)
	{
	dbuf_dirty_record_t *dr = zio->io_private;
	dmu_buf_impl_t *db = dr->dr_dbuf;
	blkptr_t *obp = &dr->dt.dl.dr_overridden_by;

	mutex_enter(&db->db_mtx);
	if (!BP_EQUAL(zio->io_bp, obp)) {
	if (!BP_IS_HOLE(obp))
	dsl_free(spa_get_dsl(zio->io_spa), zio->io_txg, obp);
	arc_release(dr->dt.dl.dr_data, db);
	}
	mutex_exit(&db->db_mtx);

	dbuf_write_done(zio, NULL, db);

	if (zio->io_abd != NULL)
	abd_free(zio->io_abd);
	}

	typedef struct dbuf_remap_impl_callback_arg {
	objset_t *drica_os;
	uint64_t drica_blk_birth;
	dmu_tx_t *drica_tx;
	} dbuf_remap_impl_callback_arg_t;

	static void
	dbuf_remap_impl_callback(uint64_t vdev, uint64_t offset, uint64_t size,
	void *arg)
	{
	dbuf_remap_impl_callback_arg_t *drica = arg;
	objset_t *os = drica->drica_os;
	spa_t *spa = dmu_objset_spa(os);
	dmu_tx_t *tx = drica->drica_tx;

	ASSERT(dsl_pool_sync_context(spa_get_dsl(spa)));

	if (os == spa_meta_objset(spa)) {
	spa_vdev_indirect_mark_obsolete(spa, vdev, offset, size, tx);
	} else {
	dsl_dataset_block_remapped(dmu_objset_ds(os), vdev, offset,
	size, drica->drica_blk_birth, tx);
	}
	}

	static void
	dbuf_remap_impl(dnode_t dn, blkptr_t bp, krwlock_t rw, dmu_tx_t tx)
	{
	blkptr_t bp_copy = *bp;
	spa_t *spa = dmu_objset_spa(dn->dn_objset);
	dbuf_remap_impl_callback_arg_t drica;

	ASSERT(dsl_pool_sync_context(spa_get_dsl(spa)));

	drica.drica_os = dn->dn_objset;
	drica.drica_blk_birth = bp->blk_birth;
	drica.drica_tx = tx;
	if (spa_remap_blkptr(spa, &bp_copy, dbuf_remap_impl_callback,
	&drica)) {
	/*
	* If the blkptr being remapped is tracked by a livelist,
	* then we need to make sure the livelist reflects the update.
	* First, cancel out the old blkptr by appending a 'FREE'
	* entry. Next, add an 'ALLOC' to track the new version. This
	* way we avoid trying to free an inaccurate blkptr at delete.
	* Note that embedded blkptrs are not tracked in livelists.
	*/
	if (dn->dn_objset != spa_meta_objset(spa)) {
	dsl_dataset_t *ds = dmu_objset_ds(dn->dn_objset);
	if (dsl_deadlist_is_open(&ds->ds_dir->dd_livelist) &&
	bp->blk_birth > ds->ds_dir->dd_origin_txg) {
	ASSERT(!BP_IS_EMBEDDED(bp));
	ASSERT(dsl_dir_is_clone(ds->ds_dir));
	ASSERT(spa_feature_is_enabled(spa,
	SPA_FEATURE_LIVELIST));
	bplist_append(&ds->ds_dir->dd_pending_frees,
	bp);
	bplist_append(&ds->ds_dir->dd_pending_allocs,
	&bp_copy);
	}
	}

	/*
	* The db_rwlock prevents dbuf_read_impl() from
	* dereferencing the BP while we are changing it. To
	* avoid lock contention, only grab it when we are actually
	* changing the BP.
	*/
	if (rw != NULL)
	rw_enter(rw, RW_WRITER);
	*bp = bp_copy;
	if (rw != NULL)
	rw_exit(rw);
	}
	}

	/*
	* Remap any existing BP's to concrete vdevs, if possible.
	*/
	static void
	dbuf_remap(dnode_t dn, dmu_buf_impl_t db, dmu_tx_t *tx)
	{
	spa_t *spa = dmu_objset_spa(db->db_objset);
	ASSERT(dsl_pool_sync_context(spa_get_dsl(spa)));

	if (!spa_feature_is_active(spa, SPA_FEATURE_DEVICE_REMOVAL))
	return;

	if (db->db_level > 0) {
	blkptr_t *bp = db->db.db_data;
	for (int i = 0; i < db->db.db_size >> SPA_BLKPTRSHIFT; i++) {
	dbuf_remap_impl(dn, &bp[i], &db->db_rwlock, tx);
	}
	} else if (db->db.db_object == DMU_META_DNODE_OBJECT) {
	dnode_phys_t *dnp = db->db.db_data;
	ASSERT3U(db->db_dnode_handle->dnh_dnode->dn_type, ==,
	DMU_OT_DNODE);
	for (int i = 0; i < db->db.db_size >> DNODE_SHIFT;
	i += dnp[i].dn_extra_slots + 1) {
	for (int j = 0; j < dnp[i].dn_nblkptr; j++) {
	krwlock_t *lock = (dn->dn_dbuf == NULL ? NULL :
	&dn->dn_dbuf->db_rwlock);
	dbuf_remap_impl(dn, &dnp[i].dn_blkptr[j], lock,
	tx);
	}
	}
	}
	}


	/* Issue I/O to commit a dirty buffer to disk. */
	static void
	dbuf_write(dbuf_dirty_record_t dr, arc_buf_t data, dmu_tx_t *tx)
	{
	dmu_buf_impl_t *db = dr->dr_dbuf;
	dnode_t *dn = dr->dr_dnode;
	objset_t *os;
	dmu_buf_impl_t *parent = db->db_parent;
	uint64_t txg = tx->tx_txg;
	zbookmark_phys_t zb;
	zio_prop_t zp;
	zio_t pio; / parent I/O */
	int wp_flag = 0;

	ASSERT(dmu_tx_is_syncing(tx));

	os = dn->dn_objset;

	if (db->db_state != DB_NOFILL) {
	if (db->db_level > 0 \|\| dn->dn_type == DMU_OT_DNODE) {
	/*
	* Private object buffers are released here rather
	* than in dbuf_dirty() since they are only modified
	* in the syncing context and we don't want the
	* overhead of making multiple copies of the data.
	*/
	if (BP_IS_HOLE(db->db_blkptr)) {
	arc_buf_thaw(data);
	} else {
	dbuf_release_bp(db);
	}
	dbuf_remap(dn, db, tx);
	}
	}

	if (parent != dn->dn_dbuf) {
	/* Our parent is an indirect block. */
	/* We have a dirty parent that has been scheduled for write. */
	ASSERT(parent && parent->db_data_pending);
	/* Our parent's buffer is one level closer to the dnode. */
	ASSERT(db->db_level == parent->db_level-1);
	/*
	* We're about to modify our parent's db_data by modifying
	* our block pointer, so the parent must be released.
	*/
	ASSERT(arc_released(parent->db_buf));
	pio = parent->db_data_pending->dr_zio;
	} else {
	/* Our parent is the dnode itself. */
	ASSERT((db->db_level == dn->dn_phys->dn_nlevels-1 &&
	db->db_blkid != DMU_SPILL_BLKID) \|\|
	(db->db_blkid == DMU_SPILL_BLKID && db->db_level == 0));
	if (db->db_blkid != DMU_SPILL_BLKID)
	ASSERT3P(db->db_blkptr, ==,
	&dn->dn_phys->dn_blkptr[db->db_blkid]);
	pio = dn->dn_zio;
	}

	ASSERT(db->db_level == 0 \|\| data == db->db_buf);
	ASSERT3U(db->db_blkptr->blk_birth, <=, txg);
	ASSERT(pio);

	SET_BOOKMARK(&zb, os->os_dsl_dataset ?
	os->os_dsl_dataset->ds_object : DMU_META_OBJSET,
	db->db.db_object, db->db_level, db->db_blkid);

	if (db->db_blkid == DMU_SPILL_BLKID)
	wp_flag = WP_SPILL;
	wp_flag \|= (db->db_state == DB_NOFILL) ? WP_NOFILL : 0;

	dmu_write_policy(os, dn, db->db_level, wp_flag, &zp);

	/*
	* We copy the blkptr now (rather than when we instantiate the dirty
	* record), because its value can change between open context and
	* syncing context. We do not need to hold dn_struct_rwlock to read
	* db_blkptr because we are in syncing context.
	*/
	dr->dr_bp_copy = *db->db_blkptr;

	if (db->db_level == 0 &&
	dr->dt.dl.dr_override_state == DR_OVERRIDDEN) {
	/*
	* The BP for this block has been provided by open context
	* (by dmu_sync() or dmu_buf_write_embedded()).
	*/
	abd_t *contents = (data != NULL) ?
	abd_get_from_buf(data->b_data, arc_buf_size(data)) : NULL;

	dr->dr_zio = zio_write(pio, os->os_spa, txg, &dr->dr_bp_copy,
	contents, db->db.db_size, db->db.db_size, &zp,
	dbuf_write_override_ready, NULL, NULL,
	dbuf_write_override_done,
	dr, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb);
	mutex_enter(&db->db_mtx);
	dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN;
	zio_write_override(dr->dr_zio, &dr->dt.dl.dr_overridden_by,
	dr->dt.dl.dr_copies, dr->dt.dl.dr_nopwrite);
	mutex_exit(&db->db_mtx);
	} else if (db->db_state == DB_NOFILL) {
	ASSERT(zp.zp_checksum == ZIO_CHECKSUM_OFF \|\|
	zp.zp_checksum == ZIO_CHECKSUM_NOPARITY);
	dr->dr_zio = zio_write(pio, os->os_spa, txg,
	&dr->dr_bp_copy, NULL, db->db.db_size, db->db.db_size, &zp,
	dbuf_write_nofill_ready, NULL, NULL,
	dbuf_write_nofill_done, db,
	ZIO_PRIORITY_ASYNC_WRITE,
	ZIO_FLAG_MUSTSUCCEED \| ZIO_FLAG_NODATA, &zb);
	} else {
	ASSERT(arc_released(data));

	/*
	* For indirect blocks, we want to setup the children
	* ready callback so that we can properly handle an indirect
	* block that only contains holes.
	*/
	arc_write_done_func_t *children_ready_cb = NULL;
	if (db->db_level != 0)
	children_ready_cb = dbuf_write_children_ready;

	dr->dr_zio = arc_write(pio, os->os_spa, txg,
	&dr->dr_bp_copy, data, dbuf_is_l2cacheable(db),
	&zp, dbuf_write_ready,
	children_ready_cb, dbuf_write_physdone,
	dbuf_write_done, db, ZIO_PRIORITY_ASYNC_WRITE,
	ZIO_FLAG_MUSTSUCCEED, &zb);
	}
	}

	EXPORT_SYMBOL(dbuf_find);
	EXPORT_SYMBOL(dbuf_is_metadata);
	EXPORT_SYMBOL(dbuf_destroy);
	EXPORT_SYMBOL(dbuf_loan_arcbuf);
	EXPORT_SYMBOL(dbuf_whichblock);
	EXPORT_SYMBOL(dbuf_read);
	EXPORT_SYMBOL(dbuf_unoverride);
	EXPORT_SYMBOL(dbuf_free_range);
	EXPORT_SYMBOL(dbuf_new_size);
	EXPORT_SYMBOL(dbuf_release_bp);
	EXPORT_SYMBOL(dbuf_dirty);
	EXPORT_SYMBOL(dmu_buf_set_crypt_params);
	EXPORT_SYMBOL(dmu_buf_will_dirty);
	EXPORT_SYMBOL(dmu_buf_is_dirty);
	EXPORT_SYMBOL(dmu_buf_will_not_fill);
	EXPORT_SYMBOL(dmu_buf_will_fill);
	EXPORT_SYMBOL(dmu_buf_fill_done);
	EXPORT_SYMBOL(dmu_buf_rele);
	EXPORT_SYMBOL(dbuf_assign_arcbuf);
	EXPORT_SYMBOL(dbuf_prefetch);
	EXPORT_SYMBOL(dbuf_hold_impl);
	EXPORT_SYMBOL(dbuf_hold);
	EXPORT_SYMBOL(dbuf_hold_level);
	EXPORT_SYMBOL(dbuf_create_bonus);
	EXPORT_SYMBOL(dbuf_spill_set_blksz);
	EXPORT_SYMBOL(dbuf_rm_spill);
	EXPORT_SYMBOL(dbuf_add_ref);
	EXPORT_SYMBOL(dbuf_rele);
	EXPORT_SYMBOL(dbuf_rele_and_unlock);
	EXPORT_SYMBOL(dbuf_refcount);
	EXPORT_SYMBOL(dbuf_sync_list);
	EXPORT_SYMBOL(dmu_buf_set_user);
	EXPORT_SYMBOL(dmu_buf_set_user_ie);
	EXPORT_SYMBOL(dmu_buf_get_user);
	EXPORT_SYMBOL(dmu_buf_get_blkptr);

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM(zfs_dbuf_cache, dbuf_cache_, max_bytes, ULONG, ZMOD_RW,
	"Maximum size in bytes of the dbuf cache.");

	ZFS_MODULE_PARAM(zfs_dbuf_cache, dbuf_cache_, hiwater_pct, UINT, ZMOD_RW,
	"Percentage over dbuf_cache_max_bytes when dbufs must be evicted "
	"directly.");

	ZFS_MODULE_PARAM(zfs_dbuf_cache, dbuf_cache_, lowater_pct, UINT, ZMOD_RW,
	"Percentage below dbuf_cache_max_bytes when the evict thread stops "
	"evicting dbufs.");

	ZFS_MODULE_PARAM(zfs_dbuf, dbuf_, metadata_cache_max_bytes, ULONG, ZMOD_RW,
	"Maximum size in bytes of the dbuf metadata cache.");

	ZFS_MODULE_PARAM(zfs_dbuf, dbuf_, cache_shift, INT, ZMOD_RW,
	"Set the size of the dbuf cache to a log2 fraction of arc size.");

	ZFS_MODULE_PARAM(zfs_dbuf, dbuf_, metadata_cache_shift, INT, ZMOD_RW,
	"Set the size of the dbuf metadata cache to a log2 fraction of arc "
	"size.");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/zfs/dmu.c b/sys/contrib/openzfs/module/zfs/dmu.c
	index eee3e70bbc95..5ce11e86dae2 100644
	--- a/sys/contrib/openzfs/module/zfs/dmu.c
	+++ b/sys/contrib/openzfs/module/zfs/dmu.c
	@@ -1,2356 +1,2360 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
	* Copyright (c) 2013, Joyent, Inc. All rights reserved.
	* Copyright (c) 2016, Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2015 by Chunwei Chen. All rights reserved.
	* Copyright (c) 2019 Datto Inc.
	* Copyright (c) 2019, Klara Inc.
	* Copyright (c) 2019, Allan Jude
	*/

	#include <sys/dmu.h>
	#include <sys/dmu_impl.h>
	#include <sys/dmu_tx.h>
	#include <sys/dbuf.h>
	#include <sys/dnode.h>
	#include <sys/zfs_context.h>
	#include <sys/dmu_objset.h>
	#include <sys/dmu_traverse.h>
	#include <sys/dsl_dataset.h>
	#include <sys/dsl_dir.h>
	#include <sys/dsl_pool.h>
	#include <sys/dsl_synctask.h>
	#include <sys/dsl_prop.h>
	#include <sys/dmu_zfetch.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/zap.h>
	#include <sys/zio_checksum.h>
	#include <sys/zio_compress.h>
	#include <sys/sa.h>
	#include <sys/zfeature.h>
	#include <sys/abd.h>
	#include <sys/trace_zfs.h>
	#include <sys/zfs_racct.h>
	#include <sys/zfs_rlock.h>
	#ifdef _KERNEL
	#include <sys/vmsystm.h>
	#include <sys/zfs_znode.h>
	#endif

	/*
	* Enable/disable nopwrite feature.
	*/
	int zfs_nopwrite_enabled = 1;

	/*
	* Tunable to control percentage of dirtied L1 blocks from frees allowed into
	* one TXG. After this threshold is crossed, additional dirty blocks from frees
	* will wait until the next TXG.
	* A value of zero will disable this throttle.
	*/
	unsigned long zfs_per_txg_dirty_frees_percent = 5;

	/*
	- * Enable/disable forcing txg sync when dirty in dmu_offset_next.
	+ * Enable/disable forcing txg sync when dirty checking for holes with lseek().
	+ * By default this is enabled to ensure accurate hole reporting, it can result
	+ * in a significant performance penalty for lseek(SEEK_HOLE) heavy workloads.
	+ * Disabling this option will result in holes never being reported in dirty
	+ * files which is always safe.
	*/
	-int zfs_dmu_offset_next_sync = 0;
	+int zfs_dmu_offset_next_sync = 1;

	/*
	* Limit the amount we can prefetch with one call to this amount. This
	* helps to limit the amount of memory that can be used by prefetching.
	* Larger objects should be prefetched a bit at a time.
	*/
	int dmu_prefetch_max = 8 * SPA_MAXBLOCKSIZE;

	const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES] = {
	{DMU_BSWAP_UINT8, TRUE, FALSE, FALSE, "unallocated" },
	{DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "object directory" },
	{DMU_BSWAP_UINT64, TRUE, TRUE, FALSE, "object array" },
	{DMU_BSWAP_UINT8, TRUE, FALSE, FALSE, "packed nvlist" },
	{DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "packed nvlist size" },
	{DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "bpobj" },
	{DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "bpobj header" },
	{DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "SPA space map header" },
	{DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "SPA space map" },
	{DMU_BSWAP_UINT64, TRUE, FALSE, TRUE, "ZIL intent log" },
	{DMU_BSWAP_DNODE, TRUE, FALSE, TRUE, "DMU dnode" },
	{DMU_BSWAP_OBJSET, TRUE, TRUE, FALSE, "DMU objset" },
	{DMU_BSWAP_UINT64, TRUE, TRUE, FALSE, "DSL directory" },
	{DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL directory child map"},
	{DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL dataset snap map" },
	{DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL props" },
	{DMU_BSWAP_UINT64, TRUE, TRUE, FALSE, "DSL dataset" },
	{DMU_BSWAP_ZNODE, TRUE, FALSE, FALSE, "ZFS znode" },
	{DMU_BSWAP_OLDACL, TRUE, FALSE, TRUE, "ZFS V0 ACL" },
	{DMU_BSWAP_UINT8, FALSE, FALSE, TRUE, "ZFS plain file" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "ZFS directory" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "ZFS master node" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "ZFS delete queue" },
	{DMU_BSWAP_UINT8, FALSE, FALSE, TRUE, "zvol object" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "zvol prop" },
	{DMU_BSWAP_UINT8, FALSE, FALSE, TRUE, "other uint8[]" },
	{DMU_BSWAP_UINT64, FALSE, FALSE, TRUE, "other uint64[]" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "other ZAP" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "persistent error log" },
	{DMU_BSWAP_UINT8, TRUE, FALSE, FALSE, "SPA history" },
	{DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "SPA history offsets" },
	{DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "Pool properties" },
	{DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL permissions" },
	{DMU_BSWAP_ACL, TRUE, FALSE, TRUE, "ZFS ACL" },
	{DMU_BSWAP_UINT8, TRUE, FALSE, TRUE, "ZFS SYSACL" },
	{DMU_BSWAP_UINT8, TRUE, FALSE, TRUE, "FUID table" },
	{DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "FUID table size" },
	{DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL dataset next clones"},
	{DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "scan work queue" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "ZFS user/group/project used" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "ZFS user/group/project quota"},
	{DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "snapshot refcount tags"},
	{DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "DDT ZAP algorithm" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "DDT statistics" },
	{DMU_BSWAP_UINT8, TRUE, FALSE, TRUE, "System attributes" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "SA master node" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "SA attr registration" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, TRUE, "SA attr layouts" },
	{DMU_BSWAP_ZAP, TRUE, FALSE, FALSE, "scan translations" },
	{DMU_BSWAP_UINT8, FALSE, FALSE, TRUE, "deduplicated block" },
	{DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL deadlist map" },
	{DMU_BSWAP_UINT64, TRUE, TRUE, FALSE, "DSL deadlist map hdr" },
	{DMU_BSWAP_ZAP, TRUE, TRUE, FALSE, "DSL dir clones" },
	{DMU_BSWAP_UINT64, TRUE, FALSE, FALSE, "bpobj subobj" }
	};

	const dmu_object_byteswap_info_t dmu_ot_byteswap[DMU_BSWAP_NUMFUNCS] = {
	{ byteswap_uint8_array, "uint8" },
	{ byteswap_uint16_array, "uint16" },
	{ byteswap_uint32_array, "uint32" },
	{ byteswap_uint64_array, "uint64" },
	{ zap_byteswap, "zap" },
	{ dnode_buf_byteswap, "dnode" },
	{ dmu_objset_byteswap, "objset" },
	{ zfs_znode_byteswap, "znode" },
	{ zfs_oldacl_byteswap, "oldacl" },
	{ zfs_acl_byteswap, "acl" }
	};

	static int
	dmu_buf_hold_noread_by_dnode(dnode_t *dn, uint64_t offset,
	void tag, dmu_buf_t *dbp)
	{
	uint64_t blkid;
	dmu_buf_impl_t *db;

	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	blkid = dbuf_whichblock(dn, 0, offset);
	db = dbuf_hold(dn, blkid, tag);
	rw_exit(&dn->dn_struct_rwlock);

	if (db == NULL) {
	*dbp = NULL;
	return (SET_ERROR(EIO));
	}

	*dbp = &db->db;
	return (0);
	}
	int
	dmu_buf_hold_noread(objset_t *os, uint64_t object, uint64_t offset,
	void tag, dmu_buf_t *dbp)
	{
	dnode_t *dn;
	uint64_t blkid;
	dmu_buf_impl_t *db;
	int err;

	err = dnode_hold(os, object, FTAG, &dn);
	if (err)
	return (err);
	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	blkid = dbuf_whichblock(dn, 0, offset);
	db = dbuf_hold(dn, blkid, tag);
	rw_exit(&dn->dn_struct_rwlock);
	dnode_rele(dn, FTAG);

	if (db == NULL) {
	*dbp = NULL;
	return (SET_ERROR(EIO));
	}

	*dbp = &db->db;
	return (err);
	}

	int
	dmu_buf_hold_by_dnode(dnode_t *dn, uint64_t offset,
	void tag, dmu_buf_t *dbp, int flags)
	{
	int err;
	int db_flags = DB_RF_CANFAIL;

	if (flags & DMU_READ_NO_PREFETCH)
	db_flags \|= DB_RF_NOPREFETCH;
	if (flags & DMU_READ_NO_DECRYPT)
	db_flags \|= DB_RF_NO_DECRYPT;

	err = dmu_buf_hold_noread_by_dnode(dn, offset, tag, dbp);
	if (err == 0) {
	dmu_buf_impl_t db = (dmu_buf_impl_t )(*dbp);
	err = dbuf_read(db, NULL, db_flags);
	if (err != 0) {
	dbuf_rele(db, tag);
	*dbp = NULL;
	}
	}

	return (err);
	}

	int
	dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset,
	void tag, dmu_buf_t *dbp, int flags)
	{
	int err;
	int db_flags = DB_RF_CANFAIL;

	if (flags & DMU_READ_NO_PREFETCH)
	db_flags \|= DB_RF_NOPREFETCH;
	if (flags & DMU_READ_NO_DECRYPT)
	db_flags \|= DB_RF_NO_DECRYPT;

	err = dmu_buf_hold_noread(os, object, offset, tag, dbp);
	if (err == 0) {
	dmu_buf_impl_t db = (dmu_buf_impl_t )(*dbp);
	err = dbuf_read(db, NULL, db_flags);
	if (err != 0) {
	dbuf_rele(db, tag);
	*dbp = NULL;
	}
	}

	return (err);
	}

	int
	dmu_bonus_max(void)
	{
	return (DN_OLD_MAX_BONUSLEN);
	}

	int
	dmu_set_bonus(dmu_buf_t db_fake, int newsize, dmu_tx_t tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;
	dnode_t *dn;
	int error;

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);

	if (dn->dn_bonus != db) {
	error = SET_ERROR(EINVAL);
	} else if (newsize < 0 \|\| newsize > db_fake->db_size) {
	error = SET_ERROR(EINVAL);
	} else {
	dnode_setbonuslen(dn, newsize, tx);
	error = 0;
	}

	DB_DNODE_EXIT(db);
	return (error);
	}

	int
	dmu_set_bonustype(dmu_buf_t db_fake, dmu_object_type_t type, dmu_tx_t tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;
	dnode_t *dn;
	int error;

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);

	if (!DMU_OT_IS_VALID(type)) {
	error = SET_ERROR(EINVAL);
	} else if (dn->dn_bonus != db) {
	error = SET_ERROR(EINVAL);
	} else {
	dnode_setbonus_type(dn, type, tx);
	error = 0;
	}

	DB_DNODE_EXIT(db);
	return (error);
	}

	dmu_object_type_t
	dmu_get_bonustype(dmu_buf_t *db_fake)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;
	dnode_t *dn;
	dmu_object_type_t type;

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	type = dn->dn_bonustype;
	DB_DNODE_EXIT(db);

	return (type);
	}

	int
	dmu_rm_spill(objset_t os, uint64_t object, dmu_tx_t tx)
	{
	dnode_t *dn;
	int error;

	error = dnode_hold(os, object, FTAG, &dn);
	dbuf_rm_spill(dn, tx);
	rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
	dnode_rm_spill(dn, tx);
	rw_exit(&dn->dn_struct_rwlock);
	dnode_rele(dn, FTAG);
	return (error);
	}

	/*
	* Lookup and hold the bonus buffer for the provided dnode. If the dnode
	* has not yet been allocated a new bonus dbuf a will be allocated.
	* Returns ENOENT, EIO, or 0.
	*/
	int dmu_bonus_hold_by_dnode(dnode_t dn, void tag, dmu_buf_t **dbp,
	uint32_t flags)
	{
	dmu_buf_impl_t *db;
	int error;
	uint32_t db_flags = DB_RF_MUST_SUCCEED;

	if (flags & DMU_READ_NO_PREFETCH)
	db_flags \|= DB_RF_NOPREFETCH;
	if (flags & DMU_READ_NO_DECRYPT)
	db_flags \|= DB_RF_NO_DECRYPT;

	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	if (dn->dn_bonus == NULL) {
	rw_exit(&dn->dn_struct_rwlock);
	rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
	if (dn->dn_bonus == NULL)
	dbuf_create_bonus(dn);
	}
	db = dn->dn_bonus;

	/* as long as the bonus buf is held, the dnode will be held */
	if (zfs_refcount_add(&db->db_holds, tag) == 1) {
	VERIFY(dnode_add_ref(dn, db));
	atomic_inc_32(&dn->dn_dbufs_count);
	}

	/*
	* Wait to drop dn_struct_rwlock until after adding the bonus dbuf's
	* hold and incrementing the dbuf count to ensure that dnode_move() sees
	* a dnode hold for every dbuf.
	*/
	rw_exit(&dn->dn_struct_rwlock);

	error = dbuf_read(db, NULL, db_flags);
	if (error) {
	dnode_evict_bonus(dn);
	dbuf_rele(db, tag);
	*dbp = NULL;
	return (error);
	}

	*dbp = &db->db;
	return (0);
	}

	int
	dmu_bonus_hold(objset_t os, uint64_t object, void tag, dmu_buf_t **dbp)
	{
	dnode_t *dn;
	int error;

	error = dnode_hold(os, object, FTAG, &dn);
	if (error)
	return (error);

	error = dmu_bonus_hold_by_dnode(dn, tag, dbp, DMU_READ_NO_PREFETCH);
	dnode_rele(dn, FTAG);

	return (error);
	}

	/*
	* returns ENOENT, EIO, or 0.
	*
	* This interface will allocate a blank spill dbuf when a spill blk
	* doesn't already exist on the dnode.
	*
	* if you only want to find an already existing spill db, then
	* dmu_spill_hold_existing() should be used.
	*/
	int
	dmu_spill_hold_by_dnode(dnode_t dn, uint32_t flags, void tag, dmu_buf_t **dbp)
	{
	dmu_buf_impl_t *db = NULL;
	int err;

	if ((flags & DB_RF_HAVESTRUCT) == 0)
	rw_enter(&dn->dn_struct_rwlock, RW_READER);

	db = dbuf_hold(dn, DMU_SPILL_BLKID, tag);

	if ((flags & DB_RF_HAVESTRUCT) == 0)
	rw_exit(&dn->dn_struct_rwlock);

	if (db == NULL) {
	*dbp = NULL;
	return (SET_ERROR(EIO));
	}
	err = dbuf_read(db, NULL, flags);
	if (err == 0)
	*dbp = &db->db;
	else {
	dbuf_rele(db, tag);
	*dbp = NULL;
	}
	return (err);
	}

	int
	dmu_spill_hold_existing(dmu_buf_t bonus, void tag, dmu_buf_t **dbp)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )bonus;
	dnode_t *dn;
	int err;

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);

	if (spa_version(dn->dn_objset->os_spa) < SPA_VERSION_SA) {
	err = SET_ERROR(EINVAL);
	} else {
	rw_enter(&dn->dn_struct_rwlock, RW_READER);

	if (!dn->dn_have_spill) {
	err = SET_ERROR(ENOENT);
	} else {
	err = dmu_spill_hold_by_dnode(dn,
	DB_RF_HAVESTRUCT \| DB_RF_CANFAIL, tag, dbp);
	}

	rw_exit(&dn->dn_struct_rwlock);
	}

	DB_DNODE_EXIT(db);
	return (err);
	}

	int
	dmu_spill_hold_by_bonus(dmu_buf_t bonus, uint32_t flags, void tag,
	dmu_buf_t **dbp)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )bonus;
	dnode_t *dn;
	int err;
	uint32_t db_flags = DB_RF_CANFAIL;

	if (flags & DMU_READ_NO_DECRYPT)
	db_flags \|= DB_RF_NO_DECRYPT;

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	err = dmu_spill_hold_by_dnode(dn, db_flags, tag, dbp);
	DB_DNODE_EXIT(db);

	return (err);
	}

	/*
	* Note: longer-term, we should modify all of the dmu_buf_*() interfaces
	* to take a held dnode rather than <os, object> -- the lookup is wasteful,
	* and can induce severe lock contention when writing to several files
	* whose dnodes are in the same block.
	*/
	int
	dmu_buf_hold_array_by_dnode(dnode_t *dn, uint64_t offset, uint64_t length,
	boolean_t read, void tag, int numbufsp, dmu_buf_t ***dbpp, uint32_t flags)
	{
	dmu_buf_t **dbp;
	zstream_t *zs = NULL;
	uint64_t blkid, nblks, i;
	uint32_t dbuf_flags;
	int err;
	zio_t *zio = NULL;
	boolean_t missed = B_FALSE;

	ASSERT(length <= DMU_MAX_ACCESS);

	/*
	* Note: We directly notify the prefetch code of this read, so that
	* we can tell it about the multi-block read. dbuf_read() only knows
	* about the one block it is accessing.
	*/
	dbuf_flags = DB_RF_CANFAIL \| DB_RF_NEVERWAIT \| DB_RF_HAVESTRUCT \|
	DB_RF_NOPREFETCH;

	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	if (dn->dn_datablkshift) {
	int blkshift = dn->dn_datablkshift;
	nblks = (P2ROUNDUP(offset + length, 1ULL << blkshift) -
	P2ALIGN(offset, 1ULL << blkshift)) >> blkshift;
	} else {
	if (offset + length > dn->dn_datablksz) {
	zfs_panic_recover("zfs: accessing past end of object "
	"%llx/%llx (size=%u access=%llu+%llu)",
	(longlong_t)dn->dn_objset->
	os_dsl_dataset->ds_object,
	(longlong_t)dn->dn_object, dn->dn_datablksz,
	(longlong_t)offset, (longlong_t)length);
	rw_exit(&dn->dn_struct_rwlock);
	return (SET_ERROR(EIO));
	}
	nblks = 1;
	}
	dbp = kmem_zalloc(sizeof (dmu_buf_t ) nblks, KM_SLEEP);

	if (read)
	zio = zio_root(dn->dn_objset->os_spa, NULL, NULL,
	ZIO_FLAG_CANFAIL);
	blkid = dbuf_whichblock(dn, 0, offset);
	if ((flags & DMU_READ_NO_PREFETCH) == 0 &&
	DNODE_META_IS_CACHEABLE(dn) && length <= zfetch_array_rd_sz) {
	/*
	* Prepare the zfetch before initiating the demand reads, so
	* that if multiple threads block on same indirect block, we
	* base predictions on the original less racy request order.
	*/
	zs = dmu_zfetch_prepare(&dn->dn_zfetch, blkid, nblks,
	read && DNODE_IS_CACHEABLE(dn), B_TRUE);
	}
	for (i = 0; i < nblks; i++) {
	dmu_buf_impl_t *db = dbuf_hold(dn, blkid + i, tag);
	if (db == NULL) {
	if (zs)
	dmu_zfetch_run(zs, missed, B_TRUE);
	rw_exit(&dn->dn_struct_rwlock);
	dmu_buf_rele_array(dbp, nblks, tag);
	if (read)
	zio_nowait(zio);
	return (SET_ERROR(EIO));
	}

	/*
	* Initiate async demand data read.
	* We check the db_state after calling dbuf_read() because
	* (1) dbuf_read() may change the state to CACHED due to a
	* hit in the ARC, and (2) on a cache miss, a child will
	* have been added to "zio" but not yet completed, so the
	* state will not yet be CACHED.
	*/
	if (read) {
	(void) dbuf_read(db, zio, dbuf_flags);
	if (db->db_state != DB_CACHED)
	missed = B_TRUE;
	}
	dbp[i] = &db->db;
	}

	if (!read)
	zfs_racct_write(length, nblks);

	if (zs)
	dmu_zfetch_run(zs, missed, B_TRUE);
	rw_exit(&dn->dn_struct_rwlock);

	if (read) {
	/* wait for async read i/o */
	err = zio_wait(zio);
	if (err) {
	dmu_buf_rele_array(dbp, nblks, tag);
	return (err);
	}

	/* wait for other io to complete */
	for (i = 0; i < nblks; i++) {
	dmu_buf_impl_t db = (dmu_buf_impl_t )dbp[i];
	mutex_enter(&db->db_mtx);
	while (db->db_state == DB_READ \|\|
	db->db_state == DB_FILL)
	cv_wait(&db->db_changed, &db->db_mtx);
	if (db->db_state == DB_UNCACHED)
	err = SET_ERROR(EIO);
	mutex_exit(&db->db_mtx);
	if (err) {
	dmu_buf_rele_array(dbp, nblks, tag);
	return (err);
	}
	}
	}

	*numbufsp = nblks;
	*dbpp = dbp;
	return (0);
	}

	int
	dmu_buf_hold_array(objset_t *os, uint64_t object, uint64_t offset,
	uint64_t length, int read, void tag, int numbufsp, dmu_buf_t ***dbpp)
	{
	dnode_t *dn;
	int err;

	err = dnode_hold(os, object, FTAG, &dn);
	if (err)
	return (err);

	err = dmu_buf_hold_array_by_dnode(dn, offset, length, read, tag,
	numbufsp, dbpp, DMU_READ_PREFETCH);

	dnode_rele(dn, FTAG);

	return (err);
	}

	int
	dmu_buf_hold_array_by_bonus(dmu_buf_t *db_fake, uint64_t offset,
	uint64_t length, boolean_t read, void tag, int numbufsp,
	dmu_buf_t ***dbpp)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;
	dnode_t *dn;
	int err;

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	err = dmu_buf_hold_array_by_dnode(dn, offset, length, read, tag,
	numbufsp, dbpp, DMU_READ_PREFETCH);
	DB_DNODE_EXIT(db);

	return (err);
	}

	void
	dmu_buf_rele_array(dmu_buf_t *dbp_fake, int numbufs, void tag)
	{
	int i;
	dmu_buf_impl_t dbp = (dmu_buf_impl_t )dbp_fake;

	if (numbufs == 0)
	return;

	for (i = 0; i < numbufs; i++) {
	if (dbp[i])
	dbuf_rele(dbp[i], tag);
	}

	kmem_free(dbp, sizeof (dmu_buf_t ) numbufs);
	}

	/*
	* Issue prefetch i/os for the given blocks. If level is greater than 0, the
	* indirect blocks prefetched will be those that point to the blocks containing
	* the data starting at offset, and continuing to offset + len.
	*
	* Note that if the indirect blocks above the blocks being prefetched are not
	* in cache, they will be asynchronously read in.
	*/
	void
	dmu_prefetch(objset_t *os, uint64_t object, int64_t level, uint64_t offset,
	uint64_t len, zio_priority_t pri)
	{
	dnode_t *dn;
	uint64_t blkid;
	int nblks, err;

	if (len == 0) { /* they're interested in the bonus buffer */
	dn = DMU_META_DNODE(os);

	if (object == 0 \|\| object >= DN_MAX_OBJECT)
	return;

	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	blkid = dbuf_whichblock(dn, level,
	object * sizeof (dnode_phys_t));
	dbuf_prefetch(dn, level, blkid, pri, 0);
	rw_exit(&dn->dn_struct_rwlock);
	return;
	}

	/*
	* See comment before the definition of dmu_prefetch_max.
	*/
	len = MIN(len, dmu_prefetch_max);

	/*
	* XXX - Note, if the dnode for the requested object is not
	* already cached, we will do a synchronous read in the
	* dnode_hold() call. The same is true for any indirects.
	*/
	err = dnode_hold(os, object, FTAG, &dn);
	if (err != 0)
	return;

	/*
	* offset + len - 1 is the last byte we want to prefetch for, and offset
	* is the first. Then dbuf_whichblk(dn, level, off + len - 1) is the
	* last block we want to prefetch, and dbuf_whichblock(dn, level,
	* offset) is the first. Then the number we need to prefetch is the
	* last - first + 1.
	*/
	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	if (level > 0 \|\| dn->dn_datablkshift != 0) {
	nblks = dbuf_whichblock(dn, level, offset + len - 1) -
	dbuf_whichblock(dn, level, offset) + 1;
	} else {
	nblks = (offset < dn->dn_datablksz);
	}

	if (nblks != 0) {
	blkid = dbuf_whichblock(dn, level, offset);
	for (int i = 0; i < nblks; i++)
	dbuf_prefetch(dn, level, blkid + i, pri, 0);
	}
	rw_exit(&dn->dn_struct_rwlock);

	dnode_rele(dn, FTAG);
	}

	/*
	* Get the next "chunk" of file data to free. We traverse the file from
	* the end so that the file gets shorter over time (if we crashes in the
	* middle, this will leave us in a better state). We find allocated file
	* data by simply searching the allocated level 1 indirects.
	*
	* On input, *start should be the first offset that does not need to be
	* freed (e.g. "offset + length"). On return, *start will be the first
	* offset that should be freed and l1blks is set to the number of level 1
	* indirect blocks found within the chunk.
	*/
	static int
	get_next_chunk(dnode_t dn, uint64_t start, uint64_t minimum, uint64_t *l1blks)
	{
	uint64_t blks;
	uint64_t maxblks = DMU_MAX_ACCESS >> (dn->dn_indblkshift + 1);
	/* bytes of data covered by a level-1 indirect block */
	uint64_t iblkrange = (uint64_t)dn->dn_datablksz *
	EPB(dn->dn_indblkshift, SPA_BLKPTRSHIFT);

	ASSERT3U(minimum, <=, *start);

	/*
	* Check if we can free the entire range assuming that all of the
	* L1 blocks in this range have data. If we can, we use this
	* worst case value as an estimate so we can avoid having to look
	* at the object's actual data.
	*/
	uint64_t total_l1blks =
	(roundup(start, iblkrange) - (minimum / iblkrange iblkrange)) /
	iblkrange;
	if (total_l1blks <= maxblks) {
	*l1blks = total_l1blks;
	*start = minimum;
	return (0);
	}
	ASSERT(ISP2(iblkrange));

	for (blks = 0; *start > minimum && blks < maxblks; blks++) {
	int err;

	/*
	* dnode_next_offset(BACKWARDS) will find an allocated L1
	* indirect block at or before the input offset. We must
	* decrement *start so that it is at the end of the region
	* to search.
	*/
	(*start)--;

	err = dnode_next_offset(dn,
	DNODE_FIND_BACKWARDS, start, 2, 1, 0);

	/* if there are no indirect blocks before start, we are done */
	if (err == ESRCH) {
	*start = minimum;
	break;
	} else if (err != 0) {
	*l1blks = blks;
	return (err);
	}

	/* set start to the beginning of this L1 indirect */
	start = P2ALIGN(start, iblkrange);
	}
	if (*start < minimum)
	*start = minimum;
	*l1blks = blks;

	return (0);
	}

	/*
	* If this objset is of type OST_ZFS return true if vfs's unmounted flag is set,
	* otherwise return false.
	* Used below in dmu_free_long_range_impl() to enable abort when unmounting
	*/
	/ARGSUSED/
	static boolean_t
	dmu_objset_zfs_unmounting(objset_t *os)
	{
	#ifdef _KERNEL
	if (dmu_objset_type(os) == DMU_OST_ZFS)
	return (zfs_get_vfs_flag_unmounted(os));
	#endif
	return (B_FALSE);
	}

	static int
	dmu_free_long_range_impl(objset_t os, dnode_t dn, uint64_t offset,
	uint64_t length)
	{
	uint64_t object_size;
	int err;
	uint64_t dirty_frees_threshold;
	dsl_pool_t *dp = dmu_objset_pool(os);

	if (dn == NULL)
	return (SET_ERROR(EINVAL));

	object_size = (dn->dn_maxblkid + 1) * dn->dn_datablksz;
	if (offset >= object_size)
	return (0);

	if (zfs_per_txg_dirty_frees_percent <= 100)
	dirty_frees_threshold =
	zfs_per_txg_dirty_frees_percent * zfs_dirty_data_max / 100;
	else
	dirty_frees_threshold = zfs_dirty_data_max / 20;

	if (length == DMU_OBJECT_END \|\| offset + length > object_size)
	length = object_size - offset;

	while (length != 0) {
	uint64_t chunk_end, chunk_begin, chunk_len;
	uint64_t l1blks;
	dmu_tx_t *tx;

	if (dmu_objset_zfs_unmounting(dn->dn_objset))
	return (SET_ERROR(EINTR));

	chunk_end = chunk_begin = offset + length;

	/* move chunk_begin backwards to the beginning of this chunk */
	err = get_next_chunk(dn, &chunk_begin, offset, &l1blks);
	if (err)
	return (err);
	ASSERT3U(chunk_begin, >=, offset);
	ASSERT3U(chunk_begin, <=, chunk_end);

	chunk_len = chunk_end - chunk_begin;

	tx = dmu_tx_create(os);
	dmu_tx_hold_free(tx, dn->dn_object, chunk_begin, chunk_len);

	/*
	* Mark this transaction as typically resulting in a net
	* reduction in space used.
	*/
	dmu_tx_mark_netfree(tx);
	err = dmu_tx_assign(tx, TXG_WAIT);
	if (err) {
	dmu_tx_abort(tx);
	return (err);
	}

	uint64_t txg = dmu_tx_get_txg(tx);

	mutex_enter(&dp->dp_lock);
	uint64_t long_free_dirty =
	dp->dp_long_free_dirty_pertxg[txg & TXG_MASK];
	mutex_exit(&dp->dp_lock);

	/*
	* To avoid filling up a TXG with just frees, wait for
	* the next TXG to open before freeing more chunks if
	* we have reached the threshold of frees.
	*/
	if (dirty_frees_threshold != 0 &&
	long_free_dirty >= dirty_frees_threshold) {
	DMU_TX_STAT_BUMP(dmu_tx_dirty_frees_delay);
	dmu_tx_commit(tx);
	txg_wait_open(dp, 0, B_TRUE);
	continue;
	}

	/*
	* In order to prevent unnecessary write throttling, for each
	* TXG, we track the cumulative size of L1 blocks being dirtied
	* in dnode_free_range() below. We compare this number to a
	* tunable threshold, past which we prevent new L1 dirty freeing
	* blocks from being added into the open TXG. See
	* dmu_free_long_range_impl() for details. The threshold
	* prevents write throttle activation due to dirty freeing L1
	* blocks taking up a large percentage of zfs_dirty_data_max.
	*/
	mutex_enter(&dp->dp_lock);
	dp->dp_long_free_dirty_pertxg[txg & TXG_MASK] +=
	l1blks << dn->dn_indblkshift;
	mutex_exit(&dp->dp_lock);
	DTRACE_PROBE3(free__long__range,
	uint64_t, long_free_dirty, uint64_t, chunk_len,
	uint64_t, txg);
	dnode_free_range(dn, chunk_begin, chunk_len, tx);

	dmu_tx_commit(tx);

	length -= chunk_len;
	}
	return (0);
	}

	int
	dmu_free_long_range(objset_t *os, uint64_t object,
	uint64_t offset, uint64_t length)
	{
	dnode_t *dn;
	int err;

	err = dnode_hold(os, object, FTAG, &dn);
	if (err != 0)
	return (err);
	err = dmu_free_long_range_impl(os, dn, offset, length);

	/*
	* It is important to zero out the maxblkid when freeing the entire
	* file, so that (a) subsequent calls to dmu_free_long_range_impl()
	* will take the fast path, and (b) dnode_reallocate() can verify
	* that the entire file has been freed.
	*/
	if (err == 0 && offset == 0 && length == DMU_OBJECT_END)
	dn->dn_maxblkid = 0;

	dnode_rele(dn, FTAG);
	return (err);
	}

	int
	dmu_free_long_object(objset_t *os, uint64_t object)
	{
	dmu_tx_t *tx;
	int err;

	err = dmu_free_long_range(os, object, 0, DMU_OBJECT_END);
	if (err != 0)
	return (err);

	tx = dmu_tx_create(os);
	dmu_tx_hold_bonus(tx, object);
	dmu_tx_hold_free(tx, object, 0, DMU_OBJECT_END);
	dmu_tx_mark_netfree(tx);
	err = dmu_tx_assign(tx, TXG_WAIT);
	if (err == 0) {
	err = dmu_object_free(os, object, tx);
	dmu_tx_commit(tx);
	} else {
	dmu_tx_abort(tx);
	}

	return (err);
	}

	int
	dmu_free_range(objset_t *os, uint64_t object, uint64_t offset,
	uint64_t size, dmu_tx_t *tx)
	{
	dnode_t *dn;
	int err = dnode_hold(os, object, FTAG, &dn);
	if (err)
	return (err);
	ASSERT(offset < UINT64_MAX);
	ASSERT(size == DMU_OBJECT_END \|\| size <= UINT64_MAX - offset);
	dnode_free_range(dn, offset, size, tx);
	dnode_rele(dn, FTAG);
	return (0);
	}

	static int
	dmu_read_impl(dnode_t *dn, uint64_t offset, uint64_t size,
	void *buf, uint32_t flags)
	{
	dmu_buf_t **dbp;
	int numbufs, err = 0;

	/*
	* Deal with odd block sizes, where there can't be data past the first
	* block. If we ever do the tail block optimization, we will need to
	* handle that here as well.
	*/
	if (dn->dn_maxblkid == 0) {
	uint64_t newsz = offset > dn->dn_datablksz ? 0 :
	MIN(size, dn->dn_datablksz - offset);
	bzero((char *)buf + newsz, size - newsz);
	size = newsz;
	}

	while (size > 0) {
	uint64_t mylen = MIN(size, DMU_MAX_ACCESS / 2);
	int i;

	/*
	* NB: we could do this block-at-a-time, but it's nice
	* to be reading in parallel.
	*/
	err = dmu_buf_hold_array_by_dnode(dn, offset, mylen,
	TRUE, FTAG, &numbufs, &dbp, flags);
	if (err)
	break;

	for (i = 0; i < numbufs; i++) {
	uint64_t tocpy;
	int64_t bufoff;
	dmu_buf_t *db = dbp[i];

	ASSERT(size > 0);

	bufoff = offset - db->db_offset;
	tocpy = MIN(db->db_size - bufoff, size);

	(void) memcpy(buf, (char *)db->db_data + bufoff, tocpy);

	offset += tocpy;
	size -= tocpy;
	buf = (char *)buf + tocpy;
	}
	dmu_buf_rele_array(dbp, numbufs, FTAG);
	}
	return (err);
	}

	int
	dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size,
	void *buf, uint32_t flags)
	{
	dnode_t *dn;
	int err;

	err = dnode_hold(os, object, FTAG, &dn);
	if (err != 0)
	return (err);

	err = dmu_read_impl(dn, offset, size, buf, flags);
	dnode_rele(dn, FTAG);
	return (err);
	}

	int
	dmu_read_by_dnode(dnode_t dn, uint64_t offset, uint64_t size, void buf,
	uint32_t flags)
	{
	return (dmu_read_impl(dn, offset, size, buf, flags));
	}

	static void
	dmu_write_impl(dmu_buf_t **dbp, int numbufs, uint64_t offset, uint64_t size,
	const void buf, dmu_tx_t tx)
	{
	int i;

	for (i = 0; i < numbufs; i++) {
	uint64_t tocpy;
	int64_t bufoff;
	dmu_buf_t *db = dbp[i];

	ASSERT(size > 0);

	bufoff = offset - db->db_offset;
	tocpy = MIN(db->db_size - bufoff, size);

	ASSERT(i == 0 \|\| i == numbufs-1 \|\| tocpy == db->db_size);

	if (tocpy == db->db_size)
	dmu_buf_will_fill(db, tx);
	else
	dmu_buf_will_dirty(db, tx);

	(void) memcpy((char *)db->db_data + bufoff, buf, tocpy);

	if (tocpy == db->db_size)
	dmu_buf_fill_done(db, tx);

	offset += tocpy;
	size -= tocpy;
	buf = (char *)buf + tocpy;
	}
	}

	void
	dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size,
	const void buf, dmu_tx_t tx)
	{
	dmu_buf_t **dbp;
	int numbufs;

	if (size == 0)
	return;

	VERIFY0(dmu_buf_hold_array(os, object, offset, size,
	FALSE, FTAG, &numbufs, &dbp));
	dmu_write_impl(dbp, numbufs, offset, size, buf, tx);
	dmu_buf_rele_array(dbp, numbufs, FTAG);
	}

	/*
	* Note: Lustre is an external consumer of this interface.
	*/
	void
	dmu_write_by_dnode(dnode_t *dn, uint64_t offset, uint64_t size,
	const void buf, dmu_tx_t tx)
	{
	dmu_buf_t **dbp;
	int numbufs;

	if (size == 0)
	return;

	VERIFY0(dmu_buf_hold_array_by_dnode(dn, offset, size,
	FALSE, FTAG, &numbufs, &dbp, DMU_READ_PREFETCH));
	dmu_write_impl(dbp, numbufs, offset, size, buf, tx);
	dmu_buf_rele_array(dbp, numbufs, FTAG);
	}

	void
	dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size,
	dmu_tx_t *tx)
	{
	dmu_buf_t **dbp;
	int numbufs, i;

	if (size == 0)
	return;

	VERIFY(0 == dmu_buf_hold_array(os, object, offset, size,
	FALSE, FTAG, &numbufs, &dbp));

	for (i = 0; i < numbufs; i++) {
	dmu_buf_t *db = dbp[i];

	dmu_buf_will_not_fill(db, tx);
	}
	dmu_buf_rele_array(dbp, numbufs, FTAG);
	}

	void
	dmu_write_embedded(objset_t *os, uint64_t object, uint64_t offset,
	void *data, uint8_t etype, uint8_t comp, int uncompressed_size,
	int compressed_size, int byteorder, dmu_tx_t *tx)
	{
	dmu_buf_t *db;

	ASSERT3U(etype, <, NUM_BP_EMBEDDED_TYPES);
	ASSERT3U(comp, <, ZIO_COMPRESS_FUNCTIONS);
	VERIFY0(dmu_buf_hold_noread(os, object, offset,
	FTAG, &db));

	dmu_buf_write_embedded(db,
	data, (bp_embedded_type_t)etype, (enum zio_compress)comp,
	uncompressed_size, compressed_size, byteorder, tx);

	dmu_buf_rele(db, FTAG);
	}

	void
	dmu_redact(objset_t *os, uint64_t object, uint64_t offset, uint64_t size,
	dmu_tx_t *tx)
	{
	int numbufs, i;
	dmu_buf_t **dbp;

	VERIFY0(dmu_buf_hold_array(os, object, offset, size, FALSE, FTAG,
	&numbufs, &dbp));
	for (i = 0; i < numbufs; i++)
	dmu_buf_redact(dbp[i], tx);
	dmu_buf_rele_array(dbp, numbufs, FTAG);
	}

	#ifdef _KERNEL
	int
	dmu_read_uio_dnode(dnode_t dn, zfs_uio_t uio, uint64_t size)
	{
	dmu_buf_t **dbp;
	int numbufs, i, err;

	/*
	* NB: we could do this block-at-a-time, but it's nice
	* to be reading in parallel.
	*/
	err = dmu_buf_hold_array_by_dnode(dn, zfs_uio_offset(uio), size,
	TRUE, FTAG, &numbufs, &dbp, 0);
	if (err)
	return (err);

	for (i = 0; i < numbufs; i++) {
	uint64_t tocpy;
	int64_t bufoff;
	dmu_buf_t *db = dbp[i];

	ASSERT(size > 0);

	bufoff = zfs_uio_offset(uio) - db->db_offset;
	tocpy = MIN(db->db_size - bufoff, size);

	err = zfs_uio_fault_move((char *)db->db_data + bufoff, tocpy,
	UIO_READ, uio);

	if (err)
	break;

	size -= tocpy;
	}
	dmu_buf_rele_array(dbp, numbufs, FTAG);

	return (err);
	}

	/*
	* Read 'size' bytes into the uio buffer.
	* From object zdb->db_object.
	* Starting at zfs_uio_offset(uio).
	*
	* If the caller already has a dbuf in the target object
	* (e.g. its bonus buffer), this routine is faster than dmu_read_uio(),
	* because we don't have to find the dnode_t for the object.
	*/
	int
	dmu_read_uio_dbuf(dmu_buf_t zdb, zfs_uio_t uio, uint64_t size)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )zdb;
	dnode_t *dn;
	int err;

	if (size == 0)
	return (0);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	err = dmu_read_uio_dnode(dn, uio, size);
	DB_DNODE_EXIT(db);

	return (err);
	}

	/*
	* Read 'size' bytes into the uio buffer.
	* From the specified object
	* Starting at offset zfs_uio_offset(uio).
	*/
	int
	dmu_read_uio(objset_t os, uint64_t object, zfs_uio_t uio, uint64_t size)
	{
	dnode_t *dn;
	int err;

	if (size == 0)
	return (0);

	err = dnode_hold(os, object, FTAG, &dn);
	if (err)
	return (err);

	err = dmu_read_uio_dnode(dn, uio, size);

	dnode_rele(dn, FTAG);

	return (err);
	}

	int
	dmu_write_uio_dnode(dnode_t dn, zfs_uio_t uio, uint64_t size, dmu_tx_t *tx)
	{
	dmu_buf_t **dbp;
	int numbufs;
	int err = 0;
	int i;

	err = dmu_buf_hold_array_by_dnode(dn, zfs_uio_offset(uio), size,
	FALSE, FTAG, &numbufs, &dbp, DMU_READ_PREFETCH);
	if (err)
	return (err);

	for (i = 0; i < numbufs; i++) {
	uint64_t tocpy;
	int64_t bufoff;
	dmu_buf_t *db = dbp[i];

	ASSERT(size > 0);

	bufoff = zfs_uio_offset(uio) - db->db_offset;
	tocpy = MIN(db->db_size - bufoff, size);

	ASSERT(i == 0 \|\| i == numbufs-1 \|\| tocpy == db->db_size);

	if (tocpy == db->db_size)
	dmu_buf_will_fill(db, tx);
	else
	dmu_buf_will_dirty(db, tx);

	/*
	* XXX zfs_uiomove could block forever (eg.nfs-backed
	* pages). There needs to be a uiolockdown() function
	* to lock the pages in memory, so that zfs_uiomove won't
	* block.
	*/
	err = zfs_uio_fault_move((char *)db->db_data + bufoff,
	tocpy, UIO_WRITE, uio);

	if (tocpy == db->db_size)
	dmu_buf_fill_done(db, tx);

	if (err)
	break;

	size -= tocpy;
	}

	dmu_buf_rele_array(dbp, numbufs, FTAG);
	return (err);
	}

	/*
	* Write 'size' bytes from the uio buffer.
	* To object zdb->db_object.
	* Starting at offset zfs_uio_offset(uio).
	*
	* If the caller already has a dbuf in the target object
	* (e.g. its bonus buffer), this routine is faster than dmu_write_uio(),
	* because we don't have to find the dnode_t for the object.
	*/
	int
	dmu_write_uio_dbuf(dmu_buf_t zdb, zfs_uio_t uio, uint64_t size,
	dmu_tx_t *tx)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )zdb;
	dnode_t *dn;
	int err;

	if (size == 0)
	return (0);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	err = dmu_write_uio_dnode(dn, uio, size, tx);
	DB_DNODE_EXIT(db);

	return (err);
	}

	/*
	* Write 'size' bytes from the uio buffer.
	* To the specified object.
	* Starting at offset zfs_uio_offset(uio).
	*/
	int
	dmu_write_uio(objset_t os, uint64_t object, zfs_uio_t uio, uint64_t size,
	dmu_tx_t *tx)
	{
	dnode_t *dn;
	int err;

	if (size == 0)
	return (0);

	err = dnode_hold(os, object, FTAG, &dn);
	if (err)
	return (err);

	err = dmu_write_uio_dnode(dn, uio, size, tx);

	dnode_rele(dn, FTAG);

	return (err);
	}
	#endif /* _KERNEL */

	/*
	* Allocate a loaned anonymous arc buffer.
	*/
	arc_buf_t *
	dmu_request_arcbuf(dmu_buf_t *handle, int size)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )handle;

	return (arc_loan_buf(db->db_objset->os_spa, B_FALSE, size));
	}

	/*
	* Free a loaned arc buffer.
	*/
	void
	dmu_return_arcbuf(arc_buf_t *buf)
	{
	arc_return_buf(buf, FTAG);
	arc_buf_destroy(buf, FTAG);
	}

	/*
	* A "lightweight" write is faster than a regular write (e.g.
	* dmu_write_by_dnode() or dmu_assign_arcbuf_by_dnode()), because it avoids the
	* CPU cost of creating a dmu_buf_impl_t and arc_buf_[hdr_]_t. However, the
	* data can not be read or overwritten until the transaction's txg has been
	* synced. This makes it appropriate for workloads that are known to be
	* (temporarily) write-only, like "zfs receive".
	*
	* A single block is written, starting at the specified offset in bytes. If
	* the call is successful, it returns 0 and the provided abd has been
	* consumed (the caller should not free it).
	*/
	int
	dmu_lightweight_write_by_dnode(dnode_t dn, uint64_t offset, abd_t abd,
	const zio_prop_t zp, enum zio_flag flags, dmu_tx_t tx)
	{
	dbuf_dirty_record_t *dr =
	dbuf_dirty_lightweight(dn, dbuf_whichblock(dn, 0, offset), tx);
	if (dr == NULL)
	return (SET_ERROR(EIO));
	dr->dt.dll.dr_abd = abd;
	dr->dt.dll.dr_props = *zp;
	dr->dt.dll.dr_flags = flags;
	return (0);
	}

	/*
	* When possible directly assign passed loaned arc buffer to a dbuf.
	* If this is not possible copy the contents of passed arc buf via
	* dmu_write().
	*/
	int
	dmu_assign_arcbuf_by_dnode(dnode_t dn, uint64_t offset, arc_buf_t buf,
	dmu_tx_t *tx)
	{
	dmu_buf_impl_t *db;
	objset_t *os = dn->dn_objset;
	uint64_t object = dn->dn_object;
	uint32_t blksz = (uint32_t)arc_buf_lsize(buf);
	uint64_t blkid;

	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	blkid = dbuf_whichblock(dn, 0, offset);
	db = dbuf_hold(dn, blkid, FTAG);
	if (db == NULL)
	return (SET_ERROR(EIO));
	rw_exit(&dn->dn_struct_rwlock);

	/*
	* We can only assign if the offset is aligned and the arc buf is the
	* same size as the dbuf.
	*/
	if (offset == db->db.db_offset && blksz == db->db.db_size) {
	zfs_racct_write(blksz, 1);
	dbuf_assign_arcbuf(db, buf, tx);
	dbuf_rele(db, FTAG);
	} else {
	/* compressed bufs must always be assignable to their dbuf */
	ASSERT3U(arc_get_compression(buf), ==, ZIO_COMPRESS_OFF);
	ASSERT(!(buf->b_flags & ARC_BUF_FLAG_COMPRESSED));

	dbuf_rele(db, FTAG);
	dmu_write(os, object, offset, blksz, buf->b_data, tx);
	dmu_return_arcbuf(buf);
	}

	return (0);
	}

	int
	dmu_assign_arcbuf_by_dbuf(dmu_buf_t handle, uint64_t offset, arc_buf_t buf,
	dmu_tx_t *tx)
	{
	int err;
	dmu_buf_impl_t dbuf = (dmu_buf_impl_t )handle;

	DB_DNODE_ENTER(dbuf);
	err = dmu_assign_arcbuf_by_dnode(DB_DNODE(dbuf), offset, buf, tx);
	DB_DNODE_EXIT(dbuf);

	return (err);
	}

	typedef struct {
	dbuf_dirty_record_t *dsa_dr;
	dmu_sync_cb_t *dsa_done;
	zgd_t *dsa_zgd;
	dmu_tx_t *dsa_tx;
	} dmu_sync_arg_t;

	/* ARGSUSED */
	static void
	dmu_sync_ready(zio_t zio, arc_buf_t buf, void *varg)
	{
	dmu_sync_arg_t *dsa = varg;
	dmu_buf_t *db = dsa->dsa_zgd->zgd_db;
	blkptr_t *bp = zio->io_bp;

	if (zio->io_error == 0) {
	if (BP_IS_HOLE(bp)) {
	/*
	* A block of zeros may compress to a hole, but the
	* block size still needs to be known for replay.
	*/
	BP_SET_LSIZE(bp, db->db_size);
	} else if (!BP_IS_EMBEDDED(bp)) {
	ASSERT(BP_GET_LEVEL(bp) == 0);
	BP_SET_FILL(bp, 1);
	}
	}
	}

	static void
	dmu_sync_late_arrival_ready(zio_t *zio)
	{
	dmu_sync_ready(zio, NULL, zio->io_private);
	}

	/* ARGSUSED */
	static void
	dmu_sync_done(zio_t zio, arc_buf_t buf, void *varg)
	{
	dmu_sync_arg_t *dsa = varg;
	dbuf_dirty_record_t *dr = dsa->dsa_dr;
	dmu_buf_impl_t *db = dr->dr_dbuf;
	zgd_t *zgd = dsa->dsa_zgd;

	/*
	* Record the vdev(s) backing this blkptr so they can be flushed after
	* the writes for the lwb have completed.
	*/
	if (zio->io_error == 0) {
	zil_lwb_add_block(zgd->zgd_lwb, zgd->zgd_bp);
	}

	mutex_enter(&db->db_mtx);
	ASSERT(dr->dt.dl.dr_override_state == DR_IN_DMU_SYNC);
	if (zio->io_error == 0) {
	dr->dt.dl.dr_nopwrite = !!(zio->io_flags & ZIO_FLAG_NOPWRITE);
	if (dr->dt.dl.dr_nopwrite) {
	blkptr_t *bp = zio->io_bp;
	blkptr_t *bp_orig = &zio->io_bp_orig;
	uint8_t chksum = BP_GET_CHECKSUM(bp_orig);

	ASSERT(BP_EQUAL(bp, bp_orig));
	VERIFY(BP_EQUAL(bp, db->db_blkptr));
	ASSERT(zio->io_prop.zp_compress != ZIO_COMPRESS_OFF);
	VERIFY(zio_checksum_table[chksum].ci_flags &
	ZCHECKSUM_FLAG_NOPWRITE);
	}
	dr->dt.dl.dr_overridden_by = *zio->io_bp;
	dr->dt.dl.dr_override_state = DR_OVERRIDDEN;
	dr->dt.dl.dr_copies = zio->io_prop.zp_copies;

	/*
	* Old style holes are filled with all zeros, whereas
	* new-style holes maintain their lsize, type, level,
	* and birth time (see zio_write_compress). While we
	* need to reset the BP_SET_LSIZE() call that happened
	* in dmu_sync_ready for old style holes, we do not
	* want to wipe out the information contained in new
	* style holes. Thus, only zero out the block pointer if
	* it's an old style hole.
	*/
	if (BP_IS_HOLE(&dr->dt.dl.dr_overridden_by) &&
	dr->dt.dl.dr_overridden_by.blk_birth == 0)
	BP_ZERO(&dr->dt.dl.dr_overridden_by);
	} else {
	dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN;
	}
	cv_broadcast(&db->db_changed);
	mutex_exit(&db->db_mtx);

	dsa->dsa_done(dsa->dsa_zgd, zio->io_error);

	kmem_free(dsa, sizeof (*dsa));
	}

	static void
	dmu_sync_late_arrival_done(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;
	dmu_sync_arg_t *dsa = zio->io_private;
	zgd_t *zgd = dsa->dsa_zgd;

	if (zio->io_error == 0) {
	/*
	* Record the vdev(s) backing this blkptr so they can be
	* flushed after the writes for the lwb have completed.
	*/
	zil_lwb_add_block(zgd->zgd_lwb, zgd->zgd_bp);

	if (!BP_IS_HOLE(bp)) {
	blkptr_t *bp_orig __maybe_unused = &zio->io_bp_orig;
	ASSERT(!(zio->io_flags & ZIO_FLAG_NOPWRITE));
	ASSERT(BP_IS_HOLE(bp_orig) \|\| !BP_EQUAL(bp, bp_orig));
	ASSERT(zio->io_bp->blk_birth == zio->io_txg);
	ASSERT(zio->io_txg > spa_syncing_txg(zio->io_spa));
	zio_free(zio->io_spa, zio->io_txg, zio->io_bp);
	}
	}

	dmu_tx_commit(dsa->dsa_tx);

	dsa->dsa_done(dsa->dsa_zgd, zio->io_error);

	abd_free(zio->io_abd);
	kmem_free(dsa, sizeof (*dsa));
	}

	static int
	dmu_sync_late_arrival(zio_t pio, objset_t os, dmu_sync_cb_t done, zgd_t zgd,
	zio_prop_t zp, zbookmark_phys_t zb)
	{
	dmu_sync_arg_t *dsa;
	dmu_tx_t *tx;

	tx = dmu_tx_create(os);
	dmu_tx_hold_space(tx, zgd->zgd_db->db_size);
	if (dmu_tx_assign(tx, TXG_WAIT) != 0) {
	dmu_tx_abort(tx);
	/* Make zl_get_data do txg_waited_synced() */
	return (SET_ERROR(EIO));
	}

	/*
	* In order to prevent the zgd's lwb from being free'd prior to
	* dmu_sync_late_arrival_done() being called, we have to ensure
	* the lwb's "max txg" takes this tx's txg into account.
	*/
	zil_lwb_add_txg(zgd->zgd_lwb, dmu_tx_get_txg(tx));

	dsa = kmem_alloc(sizeof (dmu_sync_arg_t), KM_SLEEP);
	dsa->dsa_dr = NULL;
	dsa->dsa_done = done;
	dsa->dsa_zgd = zgd;
	dsa->dsa_tx = tx;

	/*
	* Since we are currently syncing this txg, it's nontrivial to
	* determine what BP to nopwrite against, so we disable nopwrite.
	*
	* When syncing, the db_blkptr is initially the BP of the previous
	* txg. We can not nopwrite against it because it will be changed
	* (this is similar to the non-late-arrival case where the dbuf is
	* dirty in a future txg).
	*
	* Then dbuf_write_ready() sets bp_blkptr to the location we will write.
	* We can not nopwrite against it because although the BP will not
	* (typically) be changed, the data has not yet been persisted to this
	* location.
	*
	* Finally, when dbuf_write_done() is called, it is theoretically
	* possible to always nopwrite, because the data that was written in
	* this txg is the same data that we are trying to write. However we
	* would need to check that this dbuf is not dirty in any future
	* txg's (as we do in the normal dmu_sync() path). For simplicity, we
	* don't nopwrite in this case.
	*/
	zp->zp_nopwrite = B_FALSE;

	zio_nowait(zio_write(pio, os->os_spa, dmu_tx_get_txg(tx), zgd->zgd_bp,
	abd_get_from_buf(zgd->zgd_db->db_data, zgd->zgd_db->db_size),
	zgd->zgd_db->db_size, zgd->zgd_db->db_size, zp,
	dmu_sync_late_arrival_ready, NULL, NULL, dmu_sync_late_arrival_done,
	dsa, ZIO_PRIORITY_SYNC_WRITE, ZIO_FLAG_CANFAIL, zb));

	return (0);
	}

	/*
	* Intent log support: sync the block associated with db to disk.
	* N.B. and XXX: the caller is responsible for making sure that the
	* data isn't changing while dmu_sync() is writing it.
	*
	* Return values:
	*
	* EEXIST: this txg has already been synced, so there's nothing to do.
	* The caller should not log the write.
	*
	* ENOENT: the block was dbuf_free_range()'d, so there's nothing to do.
	* The caller should not log the write.
	*
	* EALREADY: this block is already in the process of being synced.
	* The caller should track its progress (somehow).
	*
	* EIO: could not do the I/O.
	* The caller should do a txg_wait_synced().
	*
	* 0: the I/O has been initiated.
	* The caller should log this blkptr in the done callback.
	* It is possible that the I/O will fail, in which case
	* the error will be reported to the done callback and
	* propagated to pio from zio_done().
	*/
	int
	dmu_sync(zio_t pio, uint64_t txg, dmu_sync_cb_t done, zgd_t *zgd)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )zgd->zgd_db;
	objset_t *os = db->db_objset;
	dsl_dataset_t *ds = os->os_dsl_dataset;
	dbuf_dirty_record_t dr, dr_next;
	dmu_sync_arg_t *dsa;
	zbookmark_phys_t zb;
	zio_prop_t zp;
	dnode_t *dn;

	ASSERT(pio != NULL);
	ASSERT(txg != 0);

	SET_BOOKMARK(&zb, ds->ds_object,
	db->db.db_object, db->db_level, db->db_blkid);

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	dmu_write_policy(os, dn, db->db_level, WP_DMU_SYNC, &zp);
	DB_DNODE_EXIT(db);

	/*
	* If we're frozen (running ziltest), we always need to generate a bp.
	*/
	if (txg > spa_freeze_txg(os->os_spa))
	return (dmu_sync_late_arrival(pio, os, done, zgd, &zp, &zb));

	/*
	* Grabbing db_mtx now provides a barrier between dbuf_sync_leaf()
	* and us. If we determine that this txg is not yet syncing,
	* but it begins to sync a moment later, that's OK because the
	* sync thread will block in dbuf_sync_leaf() until we drop db_mtx.
	*/
	mutex_enter(&db->db_mtx);

	if (txg <= spa_last_synced_txg(os->os_spa)) {
	/*
	* This txg has already synced. There's nothing to do.
	*/
	mutex_exit(&db->db_mtx);
	return (SET_ERROR(EEXIST));
	}

	if (txg <= spa_syncing_txg(os->os_spa)) {
	/*
	* This txg is currently syncing, so we can't mess with
	* the dirty record anymore; just write a new log block.
	*/
	mutex_exit(&db->db_mtx);
	return (dmu_sync_late_arrival(pio, os, done, zgd, &zp, &zb));
	}

	dr = dbuf_find_dirty_eq(db, txg);

	if (dr == NULL) {
	/*
	* There's no dr for this dbuf, so it must have been freed.
	* There's no need to log writes to freed blocks, so we're done.
	*/
	mutex_exit(&db->db_mtx);
	return (SET_ERROR(ENOENT));
	}

	dr_next = list_next(&db->db_dirty_records, dr);
	ASSERT(dr_next == NULL \|\| dr_next->dr_txg < txg);

	if (db->db_blkptr != NULL) {
	/*
	* We need to fill in zgd_bp with the current blkptr so that
	* the nopwrite code can check if we're writing the same
	* data that's already on disk. We can only nopwrite if we
	* are sure that after making the copy, db_blkptr will not
	* change until our i/o completes. We ensure this by
	* holding the db_mtx, and only allowing nopwrite if the
	* block is not already dirty (see below). This is verified
	* by dmu_sync_done(), which VERIFYs that the db_blkptr has
	* not changed.
	*/
	zgd->zgd_bp = db->db_blkptr;
	}

	/*
	* Assume the on-disk data is X, the current syncing data (in
	* txg - 1) is Y, and the current in-memory data is Z (currently
	* in dmu_sync).
	*
	* We usually want to perform a nopwrite if X and Z are the
	* same. However, if Y is different (i.e. the BP is going to
	* change before this write takes effect), then a nopwrite will
	* be incorrect - we would override with X, which could have
	* been freed when Y was written.
	*
	* (Note that this is not a concern when we are nop-writing from
	* syncing context, because X and Y must be identical, because
	* all previous txgs have been synced.)
	*
	* Therefore, we disable nopwrite if the current BP could change
	* before this TXG. There are two ways it could change: by
	* being dirty (dr_next is non-NULL), or by being freed
	* (dnode_block_freed()). This behavior is verified by
	* zio_done(), which VERIFYs that the override BP is identical
	* to the on-disk BP.
	*/
	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	if (dr_next != NULL \|\| dnode_block_freed(dn, db->db_blkid))
	zp.zp_nopwrite = B_FALSE;
	DB_DNODE_EXIT(db);

	ASSERT(dr->dr_txg == txg);
	if (dr->dt.dl.dr_override_state == DR_IN_DMU_SYNC \|\|
	dr->dt.dl.dr_override_state == DR_OVERRIDDEN) {
	/*
	* We have already issued a sync write for this buffer,
	* or this buffer has already been synced. It could not
	* have been dirtied since, or we would have cleared the state.
	*/
	mutex_exit(&db->db_mtx);
	return (SET_ERROR(EALREADY));
	}

	ASSERT(dr->dt.dl.dr_override_state == DR_NOT_OVERRIDDEN);
	dr->dt.dl.dr_override_state = DR_IN_DMU_SYNC;
	mutex_exit(&db->db_mtx);

	dsa = kmem_alloc(sizeof (dmu_sync_arg_t), KM_SLEEP);
	dsa->dsa_dr = dr;
	dsa->dsa_done = done;
	dsa->dsa_zgd = zgd;
	dsa->dsa_tx = NULL;

	zio_nowait(arc_write(pio, os->os_spa, txg,
	zgd->zgd_bp, dr->dt.dl.dr_data, dbuf_is_l2cacheable(db),
	&zp, dmu_sync_ready, NULL, NULL, dmu_sync_done, dsa,
	ZIO_PRIORITY_SYNC_WRITE, ZIO_FLAG_CANFAIL, &zb));

	return (0);
	}

	int
	dmu_object_set_nlevels(objset_t os, uint64_t object, int nlevels, dmu_tx_t tx)
	{
	dnode_t *dn;
	int err;

	err = dnode_hold(os, object, FTAG, &dn);
	if (err)
	return (err);
	err = dnode_set_nlevels(dn, nlevels, tx);
	dnode_rele(dn, FTAG);
	return (err);
	}

	int
	dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, int ibs,
	dmu_tx_t *tx)
	{
	dnode_t *dn;
	int err;

	err = dnode_hold(os, object, FTAG, &dn);
	if (err)
	return (err);
	err = dnode_set_blksz(dn, size, ibs, tx);
	dnode_rele(dn, FTAG);
	return (err);
	}

	int
	dmu_object_set_maxblkid(objset_t *os, uint64_t object, uint64_t maxblkid,
	dmu_tx_t *tx)
	{
	dnode_t *dn;
	int err;

	err = dnode_hold(os, object, FTAG, &dn);
	if (err)
	return (err);
	rw_enter(&dn->dn_struct_rwlock, RW_WRITER);
	dnode_new_blkid(dn, maxblkid, tx, B_FALSE, B_TRUE);
	rw_exit(&dn->dn_struct_rwlock);
	dnode_rele(dn, FTAG);
	return (0);
	}

	void
	dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum,
	dmu_tx_t *tx)
	{
	dnode_t *dn;

	/*
	* Send streams include each object's checksum function. This
	* check ensures that the receiving system can understand the
	* checksum function transmitted.
	*/
	ASSERT3U(checksum, <, ZIO_CHECKSUM_LEGACY_FUNCTIONS);

	VERIFY0(dnode_hold(os, object, FTAG, &dn));
	ASSERT3U(checksum, <, ZIO_CHECKSUM_FUNCTIONS);
	dn->dn_checksum = checksum;
	dnode_setdirty(dn, tx);
	dnode_rele(dn, FTAG);
	}

	void
	dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress,
	dmu_tx_t *tx)
	{
	dnode_t *dn;

	/*
	* Send streams include each object's compression function. This
	* check ensures that the receiving system can understand the
	* compression function transmitted.
	*/
	ASSERT3U(compress, <, ZIO_COMPRESS_LEGACY_FUNCTIONS);

	VERIFY0(dnode_hold(os, object, FTAG, &dn));
	dn->dn_compress = compress;
	dnode_setdirty(dn, tx);
	dnode_rele(dn, FTAG);
	}

	/*
	* When the "redundant_metadata" property is set to "most", only indirect
	* blocks of this level and higher will have an additional ditto block.
	*/
	int zfs_redundant_metadata_most_ditto_level = 2;

	void
	dmu_write_policy(objset_t os, dnode_t dn, int level, int wp, zio_prop_t *zp)
	{
	dmu_object_type_t type = dn ? dn->dn_type : DMU_OT_OBJSET;
	boolean_t ismd = (level > 0 \|\| DMU_OT_IS_METADATA(type) \|\|
	(wp & WP_SPILL));
	enum zio_checksum checksum = os->os_checksum;
	enum zio_compress compress = os->os_compress;
	uint8_t complevel = os->os_complevel;
	enum zio_checksum dedup_checksum = os->os_dedup_checksum;
	boolean_t dedup = B_FALSE;
	boolean_t nopwrite = B_FALSE;
	boolean_t dedup_verify = os->os_dedup_verify;
	boolean_t encrypt = B_FALSE;
	int copies = os->os_copies;

	/*
	* We maintain different write policies for each of the following
	* types of data:
	* 1. metadata
	* 2. preallocated blocks (i.e. level-0 blocks of a dump device)
	* 3. all other level 0 blocks
	*/
	if (ismd) {
	/*
	* XXX -- we should design a compression algorithm
	* that specializes in arrays of bps.
	*/
	compress = zio_compress_select(os->os_spa,
	ZIO_COMPRESS_ON, ZIO_COMPRESS_ON);

	/*
	* Metadata always gets checksummed. If the data
	* checksum is multi-bit correctable, and it's not a
	* ZBT-style checksum, then it's suitable for metadata
	* as well. Otherwise, the metadata checksum defaults
	* to fletcher4.
	*/
	if (!(zio_checksum_table[checksum].ci_flags &
	ZCHECKSUM_FLAG_METADATA) \|\|
	(zio_checksum_table[checksum].ci_flags &
	ZCHECKSUM_FLAG_EMBEDDED))
	checksum = ZIO_CHECKSUM_FLETCHER_4;

	if (os->os_redundant_metadata == ZFS_REDUNDANT_METADATA_ALL \|\|
	(os->os_redundant_metadata ==
	ZFS_REDUNDANT_METADATA_MOST &&
	(level >= zfs_redundant_metadata_most_ditto_level \|\|
	DMU_OT_IS_METADATA(type) \|\| (wp & WP_SPILL))))
	copies++;
	} else if (wp & WP_NOFILL) {
	ASSERT(level == 0);

	/*
	* If we're writing preallocated blocks, we aren't actually
	* writing them so don't set any policy properties. These
	* blocks are currently only used by an external subsystem
	* outside of zfs (i.e. dump) and not written by the zio
	* pipeline.
	*/
	compress = ZIO_COMPRESS_OFF;
	checksum = ZIO_CHECKSUM_OFF;
	} else {
	compress = zio_compress_select(os->os_spa, dn->dn_compress,
	compress);
	complevel = zio_complevel_select(os->os_spa, compress,
	complevel, complevel);

	checksum = (dedup_checksum == ZIO_CHECKSUM_OFF) ?
	zio_checksum_select(dn->dn_checksum, checksum) :
	dedup_checksum;

	/*
	* Determine dedup setting. If we are in dmu_sync(),
	* we won't actually dedup now because that's all
	* done in syncing context; but we do want to use the
	* dedup checksum. If the checksum is not strong
	* enough to ensure unique signatures, force
	* dedup_verify.
	*/
	if (dedup_checksum != ZIO_CHECKSUM_OFF) {
	dedup = (wp & WP_DMU_SYNC) ? B_FALSE : B_TRUE;
	if (!(zio_checksum_table[checksum].ci_flags &
	ZCHECKSUM_FLAG_DEDUP))
	dedup_verify = B_TRUE;
	}

	/*
	* Enable nopwrite if we have secure enough checksum
	* algorithm (see comment in zio_nop_write) and
	* compression is enabled. We don't enable nopwrite if
	* dedup is enabled as the two features are mutually
	* exclusive.
	*/
	nopwrite = (!dedup && (zio_checksum_table[checksum].ci_flags &
	ZCHECKSUM_FLAG_NOPWRITE) &&
	compress != ZIO_COMPRESS_OFF && zfs_nopwrite_enabled);
	}

	/*
	* All objects in an encrypted objset are protected from modification
	* via a MAC. Encrypted objects store their IV and salt in the last DVA
	* in the bp, so we cannot use all copies. Encrypted objects are also
	* not subject to nopwrite since writing the same data will still
	* result in a new ciphertext. Only encrypted blocks can be dedup'd
	* to avoid ambiguity in the dedup code since the DDT does not store
	* object types.
	*/
	if (os->os_encrypted && (wp & WP_NOFILL) == 0) {
	encrypt = B_TRUE;

	if (DMU_OT_IS_ENCRYPTED(type)) {
	copies = MIN(copies, SPA_DVAS_PER_BP - 1);
	nopwrite = B_FALSE;
	} else {
	dedup = B_FALSE;
	}

	if (level <= 0 &&
	(type == DMU_OT_DNODE \|\| type == DMU_OT_OBJSET)) {
	compress = ZIO_COMPRESS_EMPTY;
	}
	}

	zp->zp_compress = compress;
	zp->zp_complevel = complevel;
	zp->zp_checksum = checksum;
	zp->zp_type = (wp & WP_SPILL) ? dn->dn_bonustype : type;
	zp->zp_level = level;
	zp->zp_copies = MIN(copies, spa_max_replication(os->os_spa));
	zp->zp_dedup = dedup;
	zp->zp_dedup_verify = dedup && dedup_verify;
	zp->zp_nopwrite = nopwrite;
	zp->zp_encrypt = encrypt;
	zp->zp_byteorder = ZFS_HOST_BYTEORDER;
	bzero(zp->zp_salt, ZIO_DATA_SALT_LEN);
	bzero(zp->zp_iv, ZIO_DATA_IV_LEN);
	bzero(zp->zp_mac, ZIO_DATA_MAC_LEN);
	zp->zp_zpl_smallblk = DMU_OT_IS_FILE(zp->zp_type) ?
	os->os_zpl_special_smallblock : 0;

	ASSERT3U(zp->zp_compress, !=, ZIO_COMPRESS_INHERIT);
	}

	/*
	* This function is only called from zfs_holey_common() for zpl_llseek()
	* in order to determine the location of holes. In order to accurately
	* report holes all dirty data must be synced to disk. This causes extremely
	* poor performance when seeking for holes in a dirty file. As a compromise,
	* only provide hole data when the dnode is clean. When a dnode is dirty
	* report the dnode as having no holes which is always a safe thing to do.
	*/
	int
	dmu_offset_next(objset_t os, uint64_t object, boolean_t hole, uint64_t off)
	{
	dnode_t *dn;
	int err;

	restart:
	err = dnode_hold(os, object, FTAG, &dn);
	if (err)
	return (err);

	rw_enter(&dn->dn_struct_rwlock, RW_READER);

	if (dnode_is_dirty(dn)) {
	/*
	* If the zfs_dmu_offset_next_sync module option is enabled
	* then strict hole reporting has been requested. Dirty
	* dnodes must be synced to disk to accurately report all
	- * holes. When disabled (the default) dirty dnodes are
	- * reported to not have any holes which is always safe.
	+ * holes. When disabled dirty dnodes are reported to not
	+ * have any holes which is always safe.
	*
	* When called by zfs_holey_common() the zp->z_rangelock
	* is held to prevent zfs_write() and mmap writeback from
	* re-dirtying the dnode after txg_wait_synced().
	*/
	if (zfs_dmu_offset_next_sync) {
	rw_exit(&dn->dn_struct_rwlock);
	dnode_rele(dn, FTAG);
	txg_wait_synced(dmu_objset_pool(os), 0);
	goto restart;
	}

	err = SET_ERROR(EBUSY);
	} else {
	err = dnode_next_offset(dn, DNODE_FIND_HAVELOCK \|
	(hole ? DNODE_FIND_HOLE : 0), off, 1, 1, 0);
	}

	rw_exit(&dn->dn_struct_rwlock);
	dnode_rele(dn, FTAG);

	return (err);
	}

	void
	__dmu_object_info_from_dnode(dnode_t dn, dmu_object_info_t doi)
	{
	dnode_phys_t *dnp = dn->dn_phys;

	doi->doi_data_block_size = dn->dn_datablksz;
	doi->doi_metadata_block_size = dn->dn_indblkshift ?
	1ULL << dn->dn_indblkshift : 0;
	doi->doi_type = dn->dn_type;
	doi->doi_bonus_type = dn->dn_bonustype;
	doi->doi_bonus_size = dn->dn_bonuslen;
	doi->doi_dnodesize = dn->dn_num_slots << DNODE_SHIFT;
	doi->doi_indirection = dn->dn_nlevels;
	doi->doi_checksum = dn->dn_checksum;
	doi->doi_compress = dn->dn_compress;
	doi->doi_nblkptr = dn->dn_nblkptr;
	doi->doi_physical_blocks_512 = (DN_USED_BYTES(dnp) + 256) >> 9;
	doi->doi_max_offset = (dn->dn_maxblkid + 1) * dn->dn_datablksz;
	doi->doi_fill_count = 0;
	for (int i = 0; i < dnp->dn_nblkptr; i++)
	doi->doi_fill_count += BP_GET_FILL(&dnp->dn_blkptr[i]);
	}

	void
	dmu_object_info_from_dnode(dnode_t dn, dmu_object_info_t doi)
	{
	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	mutex_enter(&dn->dn_mtx);

	__dmu_object_info_from_dnode(dn, doi);

	mutex_exit(&dn->dn_mtx);
	rw_exit(&dn->dn_struct_rwlock);
	}

	/*
	* Get information on a DMU object.
	* If doi is NULL, just indicates whether the object exists.
	*/
	int
	dmu_object_info(objset_t os, uint64_t object, dmu_object_info_t doi)
	{
	dnode_t *dn;
	int err = dnode_hold(os, object, FTAG, &dn);

	if (err)
	return (err);

	if (doi != NULL)
	dmu_object_info_from_dnode(dn, doi);

	dnode_rele(dn, FTAG);
	return (0);
	}

	/*
	* As above, but faster; can be used when you have a held dbuf in hand.
	*/
	void
	dmu_object_info_from_db(dmu_buf_t db_fake, dmu_object_info_t doi)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;

	DB_DNODE_ENTER(db);
	dmu_object_info_from_dnode(DB_DNODE(db), doi);
	DB_DNODE_EXIT(db);
	}

	/*
	* Faster still when you only care about the size.
	*/
	void
	dmu_object_size_from_db(dmu_buf_t db_fake, uint32_t blksize,
	u_longlong_t *nblk512)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;
	dnode_t *dn;

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);

	*blksize = dn->dn_datablksz;
	/* add in number of slots used for the dnode itself */
	*nblk512 = ((DN_USED_BYTES(dn->dn_phys) + SPA_MINBLOCKSIZE/2) >>
	SPA_MINBLOCKSHIFT) + dn->dn_num_slots;
	DB_DNODE_EXIT(db);
	}

	void
	dmu_object_dnsize_from_db(dmu_buf_t db_fake, int dnsize)
	{
	dmu_buf_impl_t db = (dmu_buf_impl_t )db_fake;
	dnode_t *dn;

	DB_DNODE_ENTER(db);
	dn = DB_DNODE(db);
	*dnsize = dn->dn_num_slots << DNODE_SHIFT;
	DB_DNODE_EXIT(db);
	}

	void
	byteswap_uint64_array(void *vbuf, size_t size)
	{
	uint64_t *buf = vbuf;
	size_t count = size >> 3;
	int i;

	ASSERT((size & 7) == 0);

	for (i = 0; i < count; i++)
	buf[i] = BSWAP_64(buf[i]);
	}

	void
	byteswap_uint32_array(void *vbuf, size_t size)
	{
	uint32_t *buf = vbuf;
	size_t count = size >> 2;
	int i;

	ASSERT((size & 3) == 0);

	for (i = 0; i < count; i++)
	buf[i] = BSWAP_32(buf[i]);
	}

	void
	byteswap_uint16_array(void *vbuf, size_t size)
	{
	uint16_t *buf = vbuf;
	size_t count = size >> 1;
	int i;

	ASSERT((size & 1) == 0);

	for (i = 0; i < count; i++)
	buf[i] = BSWAP_16(buf[i]);
	}

	/* ARGSUSED */
	void
	byteswap_uint8_array(void *vbuf, size_t size)
	{
	}

	void
	dmu_init(void)
	{
	abd_init();
	zfs_dbgmsg_init();
	sa_cache_init();
	dmu_objset_init();
	dnode_init();
	zfetch_init();
	dmu_tx_init();
	l2arc_init();
	arc_init();
	dbuf_init();
	}

	void
	dmu_fini(void)
	{
	arc_fini(); /* arc depends on l2arc, so arc must go first */
	l2arc_fini();
	dmu_tx_fini();
	zfetch_fini();
	dbuf_fini();
	dnode_fini();
	dmu_objset_fini();
	sa_cache_fini();
	zfs_dbgmsg_fini();
	abd_fini();
	}

	EXPORT_SYMBOL(dmu_bonus_hold);
	EXPORT_SYMBOL(dmu_bonus_hold_by_dnode);
	EXPORT_SYMBOL(dmu_buf_hold_array_by_bonus);
	EXPORT_SYMBOL(dmu_buf_rele_array);
	EXPORT_SYMBOL(dmu_prefetch);
	EXPORT_SYMBOL(dmu_free_range);
	EXPORT_SYMBOL(dmu_free_long_range);
	EXPORT_SYMBOL(dmu_free_long_object);
	EXPORT_SYMBOL(dmu_read);
	EXPORT_SYMBOL(dmu_read_by_dnode);
	EXPORT_SYMBOL(dmu_write);
	EXPORT_SYMBOL(dmu_write_by_dnode);
	EXPORT_SYMBOL(dmu_prealloc);
	EXPORT_SYMBOL(dmu_object_info);
	EXPORT_SYMBOL(dmu_object_info_from_dnode);
	EXPORT_SYMBOL(dmu_object_info_from_db);
	EXPORT_SYMBOL(dmu_object_size_from_db);
	EXPORT_SYMBOL(dmu_object_dnsize_from_db);
	EXPORT_SYMBOL(dmu_object_set_nlevels);
	EXPORT_SYMBOL(dmu_object_set_blocksize);
	EXPORT_SYMBOL(dmu_object_set_maxblkid);
	EXPORT_SYMBOL(dmu_object_set_checksum);
	EXPORT_SYMBOL(dmu_object_set_compress);
	EXPORT_SYMBOL(dmu_offset_next);
	EXPORT_SYMBOL(dmu_write_policy);
	EXPORT_SYMBOL(dmu_sync);
	EXPORT_SYMBOL(dmu_request_arcbuf);
	EXPORT_SYMBOL(dmu_return_arcbuf);
	EXPORT_SYMBOL(dmu_assign_arcbuf_by_dnode);
	EXPORT_SYMBOL(dmu_assign_arcbuf_by_dbuf);
	EXPORT_SYMBOL(dmu_buf_hold);
	EXPORT_SYMBOL(dmu_ot);

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM(zfs, zfs_, nopwrite_enabled, INT, ZMOD_RW,
	"Enable NOP writes");

	ZFS_MODULE_PARAM(zfs, zfs_, per_txg_dirty_frees_percent, ULONG, ZMOD_RW,
	"Percentage of dirtied blocks from frees in one TXG");

	ZFS_MODULE_PARAM(zfs, zfs_, dmu_offset_next_sync, INT, ZMOD_RW,
	"Enable forcing txg sync to find holes");

	ZFS_MODULE_PARAM(zfs, , dmu_prefetch_max, INT, ZMOD_RW,
	"Limit one prefetch call to this size");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/zfs/dmu_send.c b/sys/contrib/openzfs/module/zfs/dmu_send.c
	index 0658e13c2d25..2f2fd4c3d6c8 100644
	--- a/sys/contrib/openzfs/module/zfs/dmu_send.c
	+++ b/sys/contrib/openzfs/module/zfs/dmu_send.c
	@@ -1,3096 +1,3107 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2011, 2018 by Delphix. All rights reserved.
	* Copyright (c) 2014, Joyent, Inc. All rights reserved.
	* Copyright 2014 HybridCluster. All rights reserved.
	* Copyright 2016 RackTop Systems.
	* Copyright (c) 2016 Actifio, Inc. All rights reserved.
	* Copyright (c) 2019, Klara Inc.
	* Copyright (c) 2019, Allan Jude
	*/

	#include <sys/dmu.h>
	#include <sys/dmu_impl.h>
	#include <sys/dmu_tx.h>
	#include <sys/dbuf.h>
	#include <sys/dnode.h>
	#include <sys/zfs_context.h>
	#include <sys/dmu_objset.h>
	#include <sys/dmu_traverse.h>
	#include <sys/dsl_dataset.h>
	#include <sys/dsl_dir.h>
	#include <sys/dsl_prop.h>
	#include <sys/dsl_pool.h>
	#include <sys/dsl_synctask.h>
	#include <sys/spa_impl.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/zap.h>
	#include <sys/zio_checksum.h>
	#include <sys/zfs_znode.h>
	#include <zfs_fletcher.h>
	#include <sys/avl.h>
	#include <sys/ddt.h>
	#include <sys/zfs_onexit.h>
	#include <sys/dmu_send.h>
	#include <sys/dmu_recv.h>
	#include <sys/dsl_destroy.h>
	#include <sys/blkptr.h>
	#include <sys/dsl_bookmark.h>
	#include <sys/zfeature.h>
	#include <sys/bqueue.h>
	#include <sys/zvol.h>
	#include <sys/policy.h>
	#include <sys/objlist.h>
	#ifdef _KERNEL
	#include <sys/zfs_vfsops.h>
	#endif

	/* Set this tunable to TRUE to replace corrupt data with 0x2f5baddb10c */
	int zfs_send_corrupt_data = B_FALSE;
	/*
	* This tunable controls the amount of data (measured in bytes) that will be
	* prefetched by zfs send. If the main thread is blocking on reads that haven't
	* completed, this variable might need to be increased. If instead the main
	* thread is issuing new reads because the prefetches have fallen out of the
	* cache, this may need to be decreased.
	*/
	int zfs_send_queue_length = SPA_MAXBLOCKSIZE;
	/*
	* This tunable controls the length of the queues that zfs send worker threads
	* use to communicate. If the send_main_thread is blocking on these queues,
	* this variable may need to be increased. If there is a significant slowdown
	* at the start of a send as these threads consume all the available IO
	* resources, this variable may need to be decreased.
	*/
	int zfs_send_no_prefetch_queue_length = 1024 * 1024;
	/*
	* These tunables control the fill fraction of the queues by zfs send. The fill
	* fraction controls the frequency with which threads have to be cv_signaled.
	* If a lot of cpu time is being spent on cv_signal, then these should be tuned
	* down. If the queues empty before the signalled thread can catch up, then
	* these should be tuned up.
	*/
	int zfs_send_queue_ff = 20;
	int zfs_send_no_prefetch_queue_ff = 20;

	/*
	* Use this to override the recordsize calculation for fast zfs send estimates.
	*/
	int zfs_override_estimate_recordsize = 0;

	/* Set this tunable to FALSE to disable setting of DRR_FLAG_FREERECORDS */
	int zfs_send_set_freerecords_bit = B_TRUE;

	/* Set this tunable to FALSE is disable sending unmodified spill blocks. */
	int zfs_send_unmodified_spill_blocks = B_TRUE;

	static inline boolean_t
	overflow_multiply(uint64_t a, uint64_t b, uint64_t *c)
	{
	uint64_t temp = a * b;
	if (b != 0 && temp / b != a)
	return (B_FALSE);
	*c = temp;
	return (B_TRUE);
	}

	struct send_thread_arg {
	bqueue_t q;
	objset_t os; / Objset to traverse */
	uint64_t fromtxg; /* Traverse from this txg */
	int flags; /* flags to pass to traverse_dataset */
	int error_code;
	boolean_t cancel;
	zbookmark_phys_t resume;
	uint64_t *num_blocks_visited;
	};

	struct redact_list_thread_arg {
	boolean_t cancel;
	bqueue_t q;
	zbookmark_phys_t resume;
	redaction_list_t *rl;
	boolean_t mark_redact;
	int error_code;
	uint64_t *num_blocks_visited;
	};

	struct send_merge_thread_arg {
	bqueue_t q;
	objset_t *os;
	struct redact_list_thread_arg *from_arg;
	struct send_thread_arg *to_arg;
	struct redact_list_thread_arg *redact_arg;
	int error;
	boolean_t cancel;
	};

	struct send_range {
	boolean_t eos_marker; /* Marks the end of the stream */
	uint64_t object;
	uint64_t start_blkid;
	uint64_t end_blkid;
	bqueue_node_t ln;
	enum type {DATA, HOLE, OBJECT, OBJECT_RANGE, REDACT,
	PREVIOUSLY_REDACTED} type;
	union {
	struct srd {
	dmu_object_type_t obj_type;
	uint32_t datablksz; // logical size
	uint32_t datasz; // payload size
	blkptr_t bp;
	arc_buf_t *abuf;
	abd_t *abd;
	kmutex_t lock;
	kcondvar_t cv;
	boolean_t io_outstanding;
	+ boolean_t io_compressed;
	int io_err;
	} data;
	struct srh {
	uint32_t datablksz;
	} hole;
	struct sro {
	/*
	* This is a pointer because embedding it in the
	* struct causes these structures to be massively larger
	* for all range types; this makes the code much less
	* memory efficient.
	*/
	dnode_phys_t *dnp;
	blkptr_t bp;
	} object;
	struct srr {
	uint32_t datablksz;
	} redact;
	struct sror {
	blkptr_t bp;
	} object_range;
	} sru;
	};

	/*
	* The list of data whose inclusion in a send stream can be pending from
	* one call to backup_cb to another. Multiple calls to dump_free(),
	* dump_freeobjects(), and dump_redact() can be aggregated into a single
	* DRR_FREE, DRR_FREEOBJECTS, or DRR_REDACT replay record.
	*/
	typedef enum {
	PENDING_NONE,
	PENDING_FREE,
	PENDING_FREEOBJECTS,
	PENDING_REDACT
	} dmu_pendop_t;

	typedef struct dmu_send_cookie {
	dmu_replay_record_t *dsc_drr;
	dmu_send_outparams_t *dsc_dso;
	offset_t *dsc_off;
	objset_t *dsc_os;
	zio_cksum_t dsc_zc;
	uint64_t dsc_toguid;
	uint64_t dsc_fromtxg;
	int dsc_err;
	dmu_pendop_t dsc_pending_op;
	uint64_t dsc_featureflags;
	uint64_t dsc_last_data_object;
	uint64_t dsc_last_data_offset;
	uint64_t dsc_resume_object;
	uint64_t dsc_resume_offset;
	boolean_t dsc_sent_begin;
	boolean_t dsc_sent_end;
	} dmu_send_cookie_t;

	static int do_dump(dmu_send_cookie_t dscp, struct send_range range);

	static void
	range_free(struct send_range *range)
	{
	if (range->type == OBJECT) {
	size_t size = sizeof (dnode_phys_t) *
	(range->sru.object.dnp->dn_extra_slots + 1);
	kmem_free(range->sru.object.dnp, size);
	} else if (range->type == DATA) {
	mutex_enter(&range->sru.data.lock);
	while (range->sru.data.io_outstanding)
	cv_wait(&range->sru.data.cv, &range->sru.data.lock);
	if (range->sru.data.abd != NULL)
	abd_free(range->sru.data.abd);
	if (range->sru.data.abuf != NULL) {
	arc_buf_destroy(range->sru.data.abuf,
	&range->sru.data.abuf);
	}
	mutex_exit(&range->sru.data.lock);

	cv_destroy(&range->sru.data.cv);
	mutex_destroy(&range->sru.data.lock);
	}
	kmem_free(range, sizeof (*range));
	}

	/*
	* For all record types except BEGIN, fill in the checksum (overlaid in
	* drr_u.drr_checksum.drr_checksum). The checksum verifies everything
	* up to the start of the checksum itself.
	*/
	static int
	dump_record(dmu_send_cookie_t dscp, void payload, int payload_len)
	{
	dmu_send_outparams_t *dso = dscp->dsc_dso;
	ASSERT3U(offsetof(dmu_replay_record_t, drr_u.drr_checksum.drr_checksum),
	==, sizeof (dmu_replay_record_t) - sizeof (zio_cksum_t));
	(void) fletcher_4_incremental_native(dscp->dsc_drr,
	offsetof(dmu_replay_record_t, drr_u.drr_checksum.drr_checksum),
	&dscp->dsc_zc);
	if (dscp->dsc_drr->drr_type == DRR_BEGIN) {
	dscp->dsc_sent_begin = B_TRUE;
	} else {
	ASSERT(ZIO_CHECKSUM_IS_ZERO(&dscp->dsc_drr->drr_u.
	drr_checksum.drr_checksum));
	dscp->dsc_drr->drr_u.drr_checksum.drr_checksum = dscp->dsc_zc;
	}
	if (dscp->dsc_drr->drr_type == DRR_END) {
	dscp->dsc_sent_end = B_TRUE;
	}
	(void) fletcher_4_incremental_native(&dscp->dsc_drr->
	drr_u.drr_checksum.drr_checksum,
	sizeof (zio_cksum_t), &dscp->dsc_zc);
	*dscp->dsc_off += sizeof (dmu_replay_record_t);
	dscp->dsc_err = dso->dso_outfunc(dscp->dsc_os, dscp->dsc_drr,
	sizeof (dmu_replay_record_t), dso->dso_arg);
	if (dscp->dsc_err != 0)
	return (SET_ERROR(EINTR));
	if (payload_len != 0) {
	*dscp->dsc_off += payload_len;
	/*
	* payload is null when dso_dryrun == B_TRUE (i.e. when we're
	* doing a send size calculation)
	*/
	if (payload != NULL) {
	(void) fletcher_4_incremental_native(
	payload, payload_len, &dscp->dsc_zc);
	}

	/*
	* The code does not rely on this (len being a multiple of 8).
	* We keep this assertion because of the corresponding assertion
	* in receive_read(). Keeping this assertion ensures that we do
	* not inadvertently break backwards compatibility (causing the
	* assertion in receive_read() to trigger on old software).
	*
	* Raw sends cannot be received on old software, and so can
	* bypass this assertion.
	*/

	ASSERT((payload_len % 8 == 0) \|\|
	(dscp->dsc_featureflags & DMU_BACKUP_FEATURE_RAW));

	dscp->dsc_err = dso->dso_outfunc(dscp->dsc_os, payload,
	payload_len, dso->dso_arg);
	if (dscp->dsc_err != 0)
	return (SET_ERROR(EINTR));
	}
	return (0);
	}

	/*
	* Fill in the drr_free struct, or perform aggregation if the previous record is
	* also a free record, and the two are adjacent.
	*
	* Note that we send free records even for a full send, because we want to be
	* able to receive a full send as a clone, which requires a list of all the free
	* and freeobject records that were generated on the source.
	*/
	static int
	dump_free(dmu_send_cookie_t *dscp, uint64_t object, uint64_t offset,
	uint64_t length)
	{
	struct drr_free *drrf = &(dscp->dsc_drr->drr_u.drr_free);

	/*
	* When we receive a free record, dbuf_free_range() assumes
	* that the receiving system doesn't have any dbufs in the range
	* being freed. This is always true because there is a one-record
	* constraint: we only send one WRITE record for any given
	* object,offset. We know that the one-record constraint is
	* true because we always send data in increasing order by
	* object,offset.
	*
	* If the increasing-order constraint ever changes, we should find
	* another way to assert that the one-record constraint is still
	* satisfied.
	*/
	ASSERT(object > dscp->dsc_last_data_object \|\|
	(object == dscp->dsc_last_data_object &&
	offset > dscp->dsc_last_data_offset));

	/*
	* If there is a pending op, but it's not PENDING_FREE, push it out,
	* since free block aggregation can only be done for blocks of the
	* same type (i.e., DRR_FREE records can only be aggregated with
	* other DRR_FREE records. DRR_FREEOBJECTS records can only be
	* aggregated with other DRR_FREEOBJECTS records).
	*/
	if (dscp->dsc_pending_op != PENDING_NONE &&
	dscp->dsc_pending_op != PENDING_FREE) {
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	dscp->dsc_pending_op = PENDING_NONE;
	}

	if (dscp->dsc_pending_op == PENDING_FREE) {
	/*
	* Check to see whether this free block can be aggregated
	* with pending one.
	*/
	if (drrf->drr_object == object && drrf->drr_offset +
	drrf->drr_length == offset) {
	if (offset + length < offset \|\| length == UINT64_MAX)
	drrf->drr_length = UINT64_MAX;
	else
	drrf->drr_length += length;
	return (0);
	} else {
	/* not a continuation. Push out pending record */
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	dscp->dsc_pending_op = PENDING_NONE;
	}
	}
	/* create a FREE record and make it pending */
	bzero(dscp->dsc_drr, sizeof (dmu_replay_record_t));
	dscp->dsc_drr->drr_type = DRR_FREE;
	drrf->drr_object = object;
	drrf->drr_offset = offset;
	if (offset + length < offset)
	drrf->drr_length = DMU_OBJECT_END;
	else
	drrf->drr_length = length;
	drrf->drr_toguid = dscp->dsc_toguid;
	if (length == DMU_OBJECT_END) {
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	} else {
	dscp->dsc_pending_op = PENDING_FREE;
	}

	return (0);
	}

	/*
	* Fill in the drr_redact struct, or perform aggregation if the previous record
	* is also a redaction record, and the two are adjacent.
	*/
	static int
	dump_redact(dmu_send_cookie_t *dscp, uint64_t object, uint64_t offset,
	uint64_t length)
	{
	struct drr_redact *drrr = &dscp->dsc_drr->drr_u.drr_redact;

	/*
	* If there is a pending op, but it's not PENDING_REDACT, push it out,
	* since free block aggregation can only be done for blocks of the
	* same type (i.e., DRR_REDACT records can only be aggregated with
	* other DRR_REDACT records).
	*/
	if (dscp->dsc_pending_op != PENDING_NONE &&
	dscp->dsc_pending_op != PENDING_REDACT) {
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	dscp->dsc_pending_op = PENDING_NONE;
	}

	if (dscp->dsc_pending_op == PENDING_REDACT) {
	/*
	* Check to see whether this redacted block can be aggregated
	* with pending one.
	*/
	if (drrr->drr_object == object && drrr->drr_offset +
	drrr->drr_length == offset) {
	drrr->drr_length += length;
	return (0);
	} else {
	/* not a continuation. Push out pending record */
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	dscp->dsc_pending_op = PENDING_NONE;
	}
	}
	/* create a REDACT record and make it pending */
	bzero(dscp->dsc_drr, sizeof (dmu_replay_record_t));
	dscp->dsc_drr->drr_type = DRR_REDACT;
	drrr->drr_object = object;
	drrr->drr_offset = offset;
	drrr->drr_length = length;
	drrr->drr_toguid = dscp->dsc_toguid;
	dscp->dsc_pending_op = PENDING_REDACT;

	return (0);
	}

	static int
	dmu_dump_write(dmu_send_cookie_t *dscp, dmu_object_type_t type, uint64_t object,
	- uint64_t offset, int lsize, int psize, const blkptr_t bp, void data)
	+ uint64_t offset, int lsize, int psize, const blkptr_t *bp,
	+ boolean_t io_compressed, void *data)
	{
	uint64_t payload_size;
	boolean_t raw = (dscp->dsc_featureflags & DMU_BACKUP_FEATURE_RAW);
	struct drr_write *drrw = &(dscp->dsc_drr->drr_u.drr_write);

	/*
	* We send data in increasing object, offset order.
	* See comment in dump_free() for details.
	*/
	ASSERT(object > dscp->dsc_last_data_object \|\|
	(object == dscp->dsc_last_data_object &&
	offset > dscp->dsc_last_data_offset));
	dscp->dsc_last_data_object = object;
	dscp->dsc_last_data_offset = offset + lsize - 1;

	/*
	* If there is any kind of pending aggregation (currently either
	* a grouping of free objects or free blocks), push it out to
	* the stream, since aggregation can't be done across operations
	* of different types.
	*/
	if (dscp->dsc_pending_op != PENDING_NONE) {
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	dscp->dsc_pending_op = PENDING_NONE;
	}
	/* write a WRITE record */
	bzero(dscp->dsc_drr, sizeof (dmu_replay_record_t));
	dscp->dsc_drr->drr_type = DRR_WRITE;
	drrw->drr_object = object;
	drrw->drr_type = type;
	drrw->drr_offset = offset;
	drrw->drr_toguid = dscp->dsc_toguid;
	drrw->drr_logical_size = lsize;

	/* only set the compression fields if the buf is compressed or raw */
	- if (raw \|\| lsize != psize) {
	+ boolean_t compressed =
	+ (bp != NULL ? BP_GET_COMPRESS(bp) != ZIO_COMPRESS_OFF &&
	+ io_compressed : lsize != psize);
	+ if (raw \|\| compressed) {
	ASSERT(raw \|\| dscp->dsc_featureflags &
	DMU_BACKUP_FEATURE_COMPRESSED);
	ASSERT(!BP_IS_EMBEDDED(bp));
	ASSERT3S(psize, >, 0);

	if (raw) {
	ASSERT(BP_IS_PROTECTED(bp));

	/*
	* This is a raw protected block so we need to pass
	* along everything the receiving side will need to
	* interpret this block, including the byteswap, salt,
	* IV, and MAC.
	*/
	if (BP_SHOULD_BYTESWAP(bp))
	drrw->drr_flags \|= DRR_RAW_BYTESWAP;
	zio_crypt_decode_params_bp(bp, drrw->drr_salt,
	drrw->drr_iv);
	zio_crypt_decode_mac_bp(bp, drrw->drr_mac);
	} else {
	/* this is a compressed block */
	ASSERT(dscp->dsc_featureflags &
	DMU_BACKUP_FEATURE_COMPRESSED);
	ASSERT(!BP_SHOULD_BYTESWAP(bp));
	ASSERT(!DMU_OT_IS_METADATA(BP_GET_TYPE(bp)));
	ASSERT3U(BP_GET_COMPRESS(bp), !=, ZIO_COMPRESS_OFF);
	ASSERT3S(lsize, >=, psize);
	}

	/* set fields common to compressed and raw sends */
	drrw->drr_compressiontype = BP_GET_COMPRESS(bp);
	drrw->drr_compressed_size = psize;
	payload_size = drrw->drr_compressed_size;
	} else {
	payload_size = drrw->drr_logical_size;
	}

	if (bp == NULL \|\| BP_IS_EMBEDDED(bp) \|\| (BP_IS_PROTECTED(bp) && !raw)) {
	/*
	* There's no pre-computed checksum for partial-block writes,
	* embedded BP's, or encrypted BP's that are being sent as
	* plaintext, so (like fletcher4-checksummed blocks) userland
	* will have to compute a dedup-capable checksum itself.
	*/
	drrw->drr_checksumtype = ZIO_CHECKSUM_OFF;
	} else {
	drrw->drr_checksumtype = BP_GET_CHECKSUM(bp);
	if (zio_checksum_table[drrw->drr_checksumtype].ci_flags &
	ZCHECKSUM_FLAG_DEDUP)
	drrw->drr_flags \|= DRR_CHECKSUM_DEDUP;
	DDK_SET_LSIZE(&drrw->drr_key, BP_GET_LSIZE(bp));
	DDK_SET_PSIZE(&drrw->drr_key, BP_GET_PSIZE(bp));
	DDK_SET_COMPRESS(&drrw->drr_key, BP_GET_COMPRESS(bp));
	DDK_SET_CRYPT(&drrw->drr_key, BP_IS_PROTECTED(bp));
	drrw->drr_key.ddk_cksum = bp->blk_cksum;
	}

	if (dump_record(dscp, data, payload_size) != 0)
	return (SET_ERROR(EINTR));
	return (0);
	}

	static int
	dump_write_embedded(dmu_send_cookie_t *dscp, uint64_t object, uint64_t offset,
	int blksz, const blkptr_t *bp)
	{
	char buf[BPE_PAYLOAD_SIZE];
	struct drr_write_embedded *drrw =
	&(dscp->dsc_drr->drr_u.drr_write_embedded);

	if (dscp->dsc_pending_op != PENDING_NONE) {
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	dscp->dsc_pending_op = PENDING_NONE;
	}

	ASSERT(BP_IS_EMBEDDED(bp));

	bzero(dscp->dsc_drr, sizeof (dmu_replay_record_t));
	dscp->dsc_drr->drr_type = DRR_WRITE_EMBEDDED;
	drrw->drr_object = object;
	drrw->drr_offset = offset;
	drrw->drr_length = blksz;
	drrw->drr_toguid = dscp->dsc_toguid;
	drrw->drr_compression = BP_GET_COMPRESS(bp);
	drrw->drr_etype = BPE_GET_ETYPE(bp);
	drrw->drr_lsize = BPE_GET_LSIZE(bp);
	drrw->drr_psize = BPE_GET_PSIZE(bp);

	decode_embedded_bp_compressed(bp, buf);

	if (dump_record(dscp, buf, P2ROUNDUP(drrw->drr_psize, 8)) != 0)
	return (SET_ERROR(EINTR));
	return (0);
	}

	static int
	dump_spill(dmu_send_cookie_t dscp, const blkptr_t bp, uint64_t object,
	void *data)
	{
	struct drr_spill *drrs = &(dscp->dsc_drr->drr_u.drr_spill);
	uint64_t blksz = BP_GET_LSIZE(bp);
	uint64_t payload_size = blksz;

	if (dscp->dsc_pending_op != PENDING_NONE) {
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	dscp->dsc_pending_op = PENDING_NONE;
	}

	/* write a SPILL record */
	bzero(dscp->dsc_drr, sizeof (dmu_replay_record_t));
	dscp->dsc_drr->drr_type = DRR_SPILL;
	drrs->drr_object = object;
	drrs->drr_length = blksz;
	drrs->drr_toguid = dscp->dsc_toguid;

	/* See comment in dump_dnode() for full details */
	if (zfs_send_unmodified_spill_blocks &&
	(bp->blk_birth <= dscp->dsc_fromtxg)) {
	drrs->drr_flags \|= DRR_SPILL_UNMODIFIED;
	}

	/* handle raw send fields */
	if (dscp->dsc_featureflags & DMU_BACKUP_FEATURE_RAW) {
	ASSERT(BP_IS_PROTECTED(bp));

	if (BP_SHOULD_BYTESWAP(bp))
	drrs->drr_flags \|= DRR_RAW_BYTESWAP;
	drrs->drr_compressiontype = BP_GET_COMPRESS(bp);
	drrs->drr_compressed_size = BP_GET_PSIZE(bp);
	zio_crypt_decode_params_bp(bp, drrs->drr_salt, drrs->drr_iv);
	zio_crypt_decode_mac_bp(bp, drrs->drr_mac);
	payload_size = drrs->drr_compressed_size;
	}

	if (dump_record(dscp, data, payload_size) != 0)
	return (SET_ERROR(EINTR));
	return (0);
	}

	static int
	dump_freeobjects(dmu_send_cookie_t *dscp, uint64_t firstobj, uint64_t numobjs)
	{
	struct drr_freeobjects *drrfo = &(dscp->dsc_drr->drr_u.drr_freeobjects);
	uint64_t maxobj = DNODES_PER_BLOCK *
	(DMU_META_DNODE(dscp->dsc_os)->dn_maxblkid + 1);

	/*
	* ZoL < 0.7 does not handle large FREEOBJECTS records correctly,
	* leading to zfs recv never completing. to avoid this issue, don't
	* send FREEOBJECTS records for object IDs which cannot exist on the
	* receiving side.
	*/
	if (maxobj > 0) {
	if (maxobj <= firstobj)
	return (0);

	if (maxobj < firstobj + numobjs)
	numobjs = maxobj - firstobj;
	}

	/*
	* If there is a pending op, but it's not PENDING_FREEOBJECTS,
	* push it out, since free block aggregation can only be done for
	* blocks of the same type (i.e., DRR_FREE records can only be
	* aggregated with other DRR_FREE records. DRR_FREEOBJECTS records
	* can only be aggregated with other DRR_FREEOBJECTS records).
	*/
	if (dscp->dsc_pending_op != PENDING_NONE &&
	dscp->dsc_pending_op != PENDING_FREEOBJECTS) {
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	dscp->dsc_pending_op = PENDING_NONE;
	}

	if (dscp->dsc_pending_op == PENDING_FREEOBJECTS) {
	/*
	* See whether this free object array can be aggregated
	* with pending one
	*/
	if (drrfo->drr_firstobj + drrfo->drr_numobjs == firstobj) {
	drrfo->drr_numobjs += numobjs;
	return (0);
	} else {
	/* can't be aggregated. Push out pending record */
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	dscp->dsc_pending_op = PENDING_NONE;
	}
	}

	/* write a FREEOBJECTS record */
	bzero(dscp->dsc_drr, sizeof (dmu_replay_record_t));
	dscp->dsc_drr->drr_type = DRR_FREEOBJECTS;
	drrfo->drr_firstobj = firstobj;
	drrfo->drr_numobjs = numobjs;
	drrfo->drr_toguid = dscp->dsc_toguid;

	dscp->dsc_pending_op = PENDING_FREEOBJECTS;

	return (0);
	}

	static int
	dump_dnode(dmu_send_cookie_t dscp, const blkptr_t bp, uint64_t object,
	dnode_phys_t *dnp)
	{
	struct drr_object *drro = &(dscp->dsc_drr->drr_u.drr_object);
	int bonuslen;

	if (object < dscp->dsc_resume_object) {
	/*
	* Note: when resuming, we will visit all the dnodes in
	* the block of dnodes that we are resuming from. In
	* this case it's unnecessary to send the dnodes prior to
	* the one we are resuming from. We should be at most one
	* block's worth of dnodes behind the resume point.
	*/
	ASSERT3U(dscp->dsc_resume_object - object, <,
	1 << (DNODE_BLOCK_SHIFT - DNODE_SHIFT));
	return (0);
	}

	if (dnp == NULL \|\| dnp->dn_type == DMU_OT_NONE)
	return (dump_freeobjects(dscp, object, 1));

	if (dscp->dsc_pending_op != PENDING_NONE) {
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	dscp->dsc_pending_op = PENDING_NONE;
	}

	/* write an OBJECT record */
	bzero(dscp->dsc_drr, sizeof (dmu_replay_record_t));
	dscp->dsc_drr->drr_type = DRR_OBJECT;
	drro->drr_object = object;
	drro->drr_type = dnp->dn_type;
	drro->drr_bonustype = dnp->dn_bonustype;
	drro->drr_blksz = dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT;
	drro->drr_bonuslen = dnp->dn_bonuslen;
	drro->drr_dn_slots = dnp->dn_extra_slots + 1;
	drro->drr_checksumtype = dnp->dn_checksum;
	drro->drr_compress = dnp->dn_compress;
	drro->drr_toguid = dscp->dsc_toguid;

	if (!(dscp->dsc_featureflags & DMU_BACKUP_FEATURE_LARGE_BLOCKS) &&
	drro->drr_blksz > SPA_OLD_MAXBLOCKSIZE)
	drro->drr_blksz = SPA_OLD_MAXBLOCKSIZE;

	bonuslen = P2ROUNDUP(dnp->dn_bonuslen, 8);

	if ((dscp->dsc_featureflags & DMU_BACKUP_FEATURE_RAW)) {
	ASSERT(BP_IS_ENCRYPTED(bp));

	if (BP_SHOULD_BYTESWAP(bp))
	drro->drr_flags \|= DRR_RAW_BYTESWAP;

	/* needed for reconstructing dnp on recv side */
	drro->drr_maxblkid = dnp->dn_maxblkid;
	drro->drr_indblkshift = dnp->dn_indblkshift;
	drro->drr_nlevels = dnp->dn_nlevels;
	drro->drr_nblkptr = dnp->dn_nblkptr;

	/*
	* Since we encrypt the entire bonus area, the (raw) part
	* beyond the bonuslen is actually nonzero, so we need
	* to send it.
	*/
	if (bonuslen != 0) {
	drro->drr_raw_bonuslen = DN_MAX_BONUS_LEN(dnp);
	bonuslen = drro->drr_raw_bonuslen;
	}
	}

	/*
	* DRR_OBJECT_SPILL is set for every dnode which references a
	* spill block. This allows the receiving pool to definitively
	* determine when a spill block should be kept or freed.
	*/
	if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR)
	drro->drr_flags \|= DRR_OBJECT_SPILL;

	if (dump_record(dscp, DN_BONUS(dnp), bonuslen) != 0)
	return (SET_ERROR(EINTR));

	/* Free anything past the end of the file. */
	if (dump_free(dscp, object, (dnp->dn_maxblkid + 1) *
	(dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT), DMU_OBJECT_END) != 0)
	return (SET_ERROR(EINTR));

	/*
	* Send DRR_SPILL records for unmodified spill blocks. This is useful
	* because changing certain attributes of the object (e.g. blocksize)
	* can cause old versions of ZFS to incorrectly remove a spill block.
	* Including these records in the stream forces an up to date version
	* to always be written ensuring they're never lost. Current versions
	* of the code which understand the DRR_FLAG_SPILL_BLOCK feature can
	* ignore these unmodified spill blocks.
	*/
	if (zfs_send_unmodified_spill_blocks &&
	(dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) &&
	(DN_SPILL_BLKPTR(dnp)->blk_birth <= dscp->dsc_fromtxg)) {
	struct send_range record;
	blkptr_t *bp = DN_SPILL_BLKPTR(dnp);

	bzero(&record, sizeof (struct send_range));
	record.type = DATA;
	record.object = object;
	record.eos_marker = B_FALSE;
	record.start_blkid = DMU_SPILL_BLKID;
	record.end_blkid = record.start_blkid + 1;
	record.sru.data.bp = *bp;
	record.sru.data.obj_type = dnp->dn_type;
	record.sru.data.datablksz = BP_GET_LSIZE(bp);

	if (do_dump(dscp, &record) != 0)
	return (SET_ERROR(EINTR));
	}

	if (dscp->dsc_err != 0)
	return (SET_ERROR(EINTR));

	return (0);
	}

	static int
	dump_object_range(dmu_send_cookie_t dscp, const blkptr_t bp,
	uint64_t firstobj, uint64_t numslots)
	{
	struct drr_object_range *drror =
	&(dscp->dsc_drr->drr_u.drr_object_range);

	/* we only use this record type for raw sends */
	ASSERT(BP_IS_PROTECTED(bp));
	ASSERT(dscp->dsc_featureflags & DMU_BACKUP_FEATURE_RAW);
	ASSERT3U(BP_GET_COMPRESS(bp), ==, ZIO_COMPRESS_OFF);
	ASSERT3U(BP_GET_TYPE(bp), ==, DMU_OT_DNODE);
	ASSERT0(BP_GET_LEVEL(bp));

	if (dscp->dsc_pending_op != PENDING_NONE) {
	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	dscp->dsc_pending_op = PENDING_NONE;
	}

	bzero(dscp->dsc_drr, sizeof (dmu_replay_record_t));
	dscp->dsc_drr->drr_type = DRR_OBJECT_RANGE;
	drror->drr_firstobj = firstobj;
	drror->drr_numslots = numslots;
	drror->drr_toguid = dscp->dsc_toguid;
	if (BP_SHOULD_BYTESWAP(bp))
	drror->drr_flags \|= DRR_RAW_BYTESWAP;
	zio_crypt_decode_params_bp(bp, drror->drr_salt, drror->drr_iv);
	zio_crypt_decode_mac_bp(bp, drror->drr_mac);

	if (dump_record(dscp, NULL, 0) != 0)
	return (SET_ERROR(EINTR));
	return (0);
	}

	static boolean_t
	send_do_embed(const blkptr_t *bp, uint64_t featureflags)
	{
	if (!BP_IS_EMBEDDED(bp))
	return (B_FALSE);

	/*
	* Compression function must be legacy, or explicitly enabled.
	*/
	if ((BP_GET_COMPRESS(bp) >= ZIO_COMPRESS_LEGACY_FUNCTIONS &&
	!(featureflags & DMU_BACKUP_FEATURE_LZ4)))
	return (B_FALSE);

	/*
	* If we have not set the ZSTD feature flag, we can't send ZSTD
	* compressed embedded blocks, as the receiver may not support them.
	*/
	if ((BP_GET_COMPRESS(bp) == ZIO_COMPRESS_ZSTD &&
	!(featureflags & DMU_BACKUP_FEATURE_ZSTD)))
	return (B_FALSE);

	/*
	* Embed type must be explicitly enabled.
	*/
	switch (BPE_GET_ETYPE(bp)) {
	case BP_EMBEDDED_TYPE_DATA:
	if (featureflags & DMU_BACKUP_FEATURE_EMBED_DATA)
	return (B_TRUE);
	break;
	default:
	return (B_FALSE);
	}
	return (B_FALSE);
	}

	/*
	* This function actually handles figuring out what kind of record needs to be
	* dumped, and calling the appropriate helper function. In most cases,
	* the data has already been read by send_reader_thread().
	*/
	static int
	do_dump(dmu_send_cookie_t dscp, struct send_range range)
	{
	int err = 0;
	switch (range->type) {
	case OBJECT:
	err = dump_dnode(dscp, &range->sru.object.bp, range->object,
	range->sru.object.dnp);
	return (err);
	case OBJECT_RANGE: {
	ASSERT3U(range->start_blkid + 1, ==, range->end_blkid);
	if (!(dscp->dsc_featureflags & DMU_BACKUP_FEATURE_RAW)) {
	return (0);
	}
	uint64_t epb = BP_GET_LSIZE(&range->sru.object_range.bp) >>
	DNODE_SHIFT;
	uint64_t firstobj = range->start_blkid * epb;
	err = dump_object_range(dscp, &range->sru.object_range.bp,
	firstobj, epb);
	break;
	}
	case REDACT: {
	struct srr *srrp = &range->sru.redact;
	err = dump_redact(dscp, range->object, range->start_blkid *
	srrp->datablksz, (range->end_blkid - range->start_blkid) *
	srrp->datablksz);
	return (err);
	}
	case DATA: {
	struct srd *srdp = &range->sru.data;
	blkptr_t *bp = &srdp->bp;
	spa_t *spa =
	dmu_objset_spa(dscp->dsc_os);

	ASSERT3U(srdp->datablksz, ==, BP_GET_LSIZE(bp));
	ASSERT3U(range->start_blkid + 1, ==, range->end_blkid);
	if (BP_GET_TYPE(bp) == DMU_OT_SA) {
	arc_flags_t aflags = ARC_FLAG_WAIT;
	enum zio_flag zioflags = ZIO_FLAG_CANFAIL;

	if (dscp->dsc_featureflags & DMU_BACKUP_FEATURE_RAW) {
	ASSERT(BP_IS_PROTECTED(bp));
	zioflags \|= ZIO_FLAG_RAW;
	}

	zbookmark_phys_t zb;
	ASSERT3U(range->start_blkid, ==, DMU_SPILL_BLKID);
	zb.zb_objset = dmu_objset_id(dscp->dsc_os);
	zb.zb_object = range->object;
	zb.zb_level = 0;
	zb.zb_blkid = range->start_blkid;

	arc_buf_t *abuf = NULL;
	if (!dscp->dsc_dso->dso_dryrun && arc_read(NULL, spa,
	bp, arc_getbuf_func, &abuf, ZIO_PRIORITY_ASYNC_READ,
	zioflags, &aflags, &zb) != 0)
	return (SET_ERROR(EIO));

	err = dump_spill(dscp, bp, zb.zb_object,
	(abuf == NULL ? NULL : abuf->b_data));
	if (abuf != NULL)
	arc_buf_destroy(abuf, &abuf);
	return (err);
	}
	if (send_do_embed(bp, dscp->dsc_featureflags)) {
	err = dump_write_embedded(dscp, range->object,
	range->start_blkid * srdp->datablksz,
	srdp->datablksz, bp);
	return (err);
	}
	ASSERT(range->object > dscp->dsc_resume_object \|\|
	(range->object == dscp->dsc_resume_object &&
	range->start_blkid * srdp->datablksz >=
	dscp->dsc_resume_offset));
	/* it's a level-0 block of a regular object */

	mutex_enter(&srdp->lock);
	while (srdp->io_outstanding)
	cv_wait(&srdp->cv, &srdp->lock);
	err = srdp->io_err;
	mutex_exit(&srdp->lock);

	if (err != 0) {
	if (zfs_send_corrupt_data &&
	!dscp->dsc_dso->dso_dryrun) {
	/*
	* Send a block filled with 0x"zfs badd bloc"
	*/
	srdp->abuf = arc_alloc_buf(spa, &srdp->abuf,
	ARC_BUFC_DATA, srdp->datablksz);
	uint64_t *ptr;
	for (ptr = srdp->abuf->b_data;
	(char )ptr < (char )srdp->abuf->b_data +
	srdp->datablksz; ptr++)
	*ptr = 0x2f5baddb10cULL;
	} else {
	return (SET_ERROR(EIO));
	}
	}

	ASSERT(dscp->dsc_dso->dso_dryrun \|\|
	srdp->abuf != NULL \|\| srdp->abd != NULL);

	uint64_t offset = range->start_blkid * srdp->datablksz;

	char *data = NULL;
	if (srdp->abd != NULL) {
	data = abd_to_buf(srdp->abd);
	ASSERT3P(srdp->abuf, ==, NULL);
	} else if (srdp->abuf != NULL) {
	data = srdp->abuf->b_data;
	}

	/*
	* If we have large blocks stored on disk but the send flags
	* don't allow us to send large blocks, we split the data from
	* the arc buf into chunks.
	*/
	if (srdp->datablksz > SPA_OLD_MAXBLOCKSIZE &&
	!(dscp->dsc_featureflags &
	DMU_BACKUP_FEATURE_LARGE_BLOCKS)) {
	while (srdp->datablksz > 0 && err == 0) {
	int n = MIN(srdp->datablksz,
	SPA_OLD_MAXBLOCKSIZE);
	err = dmu_dump_write(dscp, srdp->obj_type,
	- range->object, offset, n, n, NULL, data);
	+ range->object, offset, n, n, NULL, B_FALSE,
	+ data);
	offset += n;
	/*
	* When doing dry run, data==NULL is used as a
	* sentinel value by
	* dmu_dump_write()->dump_record().
	*/
	if (data != NULL)
	data += n;
	srdp->datablksz -= n;
	}
	} else {
	err = dmu_dump_write(dscp, srdp->obj_type,
	range->object, offset,
	- srdp->datablksz, srdp->datasz, bp, data);
	+ srdp->datablksz, srdp->datasz, bp,
	+ srdp->io_compressed, data);
	}
	return (err);
	}
	case HOLE: {
	struct srh *srhp = &range->sru.hole;
	if (range->object == DMU_META_DNODE_OBJECT) {
	uint32_t span = srhp->datablksz >> DNODE_SHIFT;
	uint64_t first_obj = range->start_blkid * span;
	uint64_t numobj = range->end_blkid * span - first_obj;
	return (dump_freeobjects(dscp, first_obj, numobj));
	}
	uint64_t offset = 0;

	/*
	* If this multiply overflows, we don't need to send this block.
	* Even if it has a birth time, it can never not be a hole, so
	* we don't need to send records for it.
	*/
	if (!overflow_multiply(range->start_blkid, srhp->datablksz,
	&offset)) {
	return (0);
	}
	uint64_t len = 0;

	if (!overflow_multiply(range->end_blkid, srhp->datablksz, &len))
	len = UINT64_MAX;
	len = len - offset;
	return (dump_free(dscp, range->object, offset, len));
	}
	default:
	panic("Invalid range type in do_dump: %d", range->type);
	}
	return (err);
	}

	static struct send_range *
	range_alloc(enum type type, uint64_t object, uint64_t start_blkid,
	uint64_t end_blkid, boolean_t eos)
	{
	struct send_range range = kmem_alloc(sizeof (range), KM_SLEEP);
	range->type = type;
	range->object = object;
	range->start_blkid = start_blkid;
	range->end_blkid = end_blkid;
	range->eos_marker = eos;
	if (type == DATA) {
	range->sru.data.abd = NULL;
	range->sru.data.abuf = NULL;
	mutex_init(&range->sru.data.lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&range->sru.data.cv, NULL, CV_DEFAULT, NULL);
	range->sru.data.io_outstanding = 0;
	range->sru.data.io_err = 0;
	+ range->sru.data.io_compressed = B_FALSE;
	}
	return (range);
	}

	/*
	* This is the callback function to traverse_dataset that acts as a worker
	* thread for dmu_send_impl.
	*/
	/ARGSUSED/
	static int
	send_cb(spa_t spa, zilog_t zilog, const blkptr_t *bp,
	const zbookmark_phys_t zb, const struct dnode_phys dnp, void *arg)
	{
	struct send_thread_arg *sta = arg;
	struct send_range *record;

	ASSERT(zb->zb_object == DMU_META_DNODE_OBJECT \|\|
	zb->zb_object >= sta->resume.zb_object);

	/*
	* All bps of an encrypted os should have the encryption bit set.
	* If this is not true it indicates tampering and we report an error.
	*/
	if (sta->os->os_encrypted &&
	!BP_IS_HOLE(bp) && !BP_USES_CRYPT(bp)) {
	spa_log_error(spa, zb);
	zfs_panic_recover("unencrypted block in encrypted "
	"object set %llu", dmu_objset_id(sta->os));
	return (SET_ERROR(EIO));
	}

	if (sta->cancel)
	return (SET_ERROR(EINTR));
	if (zb->zb_object != DMU_META_DNODE_OBJECT &&
	DMU_OBJECT_IS_SPECIAL(zb->zb_object))
	return (0);
	atomic_inc_64(sta->num_blocks_visited);

	if (zb->zb_level == ZB_DNODE_LEVEL) {
	if (zb->zb_object == DMU_META_DNODE_OBJECT)
	return (0);
	record = range_alloc(OBJECT, zb->zb_object, 0, 0, B_FALSE);
	record->sru.object.bp = *bp;
	size_t size = sizeof (dnp) (dnp->dn_extra_slots + 1);
	record->sru.object.dnp = kmem_alloc(size, KM_SLEEP);
	bcopy(dnp, record->sru.object.dnp, size);
	bqueue_enqueue(&sta->q, record, sizeof (*record));
	return (0);
	}
	if (zb->zb_level == 0 && zb->zb_object == DMU_META_DNODE_OBJECT &&
	!BP_IS_HOLE(bp)) {
	record = range_alloc(OBJECT_RANGE, 0, zb->zb_blkid,
	zb->zb_blkid + 1, B_FALSE);
	record->sru.object_range.bp = *bp;
	bqueue_enqueue(&sta->q, record, sizeof (*record));
	return (0);
	}
	if (zb->zb_level < 0 \|\| (zb->zb_level > 0 && !BP_IS_HOLE(bp)))
	return (0);
	if (zb->zb_object == DMU_META_DNODE_OBJECT && !BP_IS_HOLE(bp))
	return (0);

	uint64_t span = bp_span_in_blocks(dnp->dn_indblkshift, zb->zb_level);
	uint64_t start;

	/*
	* If this multiply overflows, we don't need to send this block.
	* Even if it has a birth time, it can never not be a hole, so
	* we don't need to send records for it.
	*/
	if (!overflow_multiply(span, zb->zb_blkid, &start) \|\| (!(zb->zb_blkid ==
	DMU_SPILL_BLKID \|\| DMU_OT_IS_METADATA(dnp->dn_type)) &&
	span * zb->zb_blkid > dnp->dn_maxblkid)) {
	ASSERT(BP_IS_HOLE(bp));
	return (0);
	}

	if (zb->zb_blkid == DMU_SPILL_BLKID)
	ASSERT3U(BP_GET_TYPE(bp), ==, DMU_OT_SA);

	enum type record_type = DATA;
	if (BP_IS_HOLE(bp))
	record_type = HOLE;
	else if (BP_IS_REDACTED(bp))
	record_type = REDACT;
	else
	record_type = DATA;

	record = range_alloc(record_type, zb->zb_object, start,
	(start + span < start ? 0 : start + span), B_FALSE);

	uint64_t datablksz = (zb->zb_blkid == DMU_SPILL_BLKID ?
	BP_GET_LSIZE(bp) : dnp->dn_datablkszsec << SPA_MINBLOCKSHIFT);

	if (BP_IS_HOLE(bp)) {
	record->sru.hole.datablksz = datablksz;
	} else if (BP_IS_REDACTED(bp)) {
	record->sru.redact.datablksz = datablksz;
	} else {
	record->sru.data.datablksz = datablksz;
	record->sru.data.obj_type = dnp->dn_type;
	record->sru.data.bp = *bp;
	}

	bqueue_enqueue(&sta->q, record, sizeof (*record));
	return (0);
	}

	struct redact_list_cb_arg {
	uint64_t *num_blocks_visited;
	bqueue_t *q;
	boolean_t *cancel;
	boolean_t mark_redact;
	};

	static int
	redact_list_cb(redact_block_phys_t rb, void arg)
	{
	struct redact_list_cb_arg *rlcap = arg;

	atomic_inc_64(rlcap->num_blocks_visited);
	if (*rlcap->cancel)
	return (-1);

	struct send_range *data = range_alloc(REDACT, rb->rbp_object,
	rb->rbp_blkid, rb->rbp_blkid + redact_block_get_count(rb), B_FALSE);
	ASSERT3U(data->end_blkid, >, rb->rbp_blkid);
	if (rlcap->mark_redact) {
	data->type = REDACT;
	data->sru.redact.datablksz = redact_block_get_size(rb);
	} else {
	data->type = PREVIOUSLY_REDACTED;
	}
	bqueue_enqueue(rlcap->q, data, sizeof (*data));

	return (0);
	}

	/*
	* This function kicks off the traverse_dataset. It also handles setting the
	* error code of the thread in case something goes wrong, and pushes the End of
	* Stream record when the traverse_dataset call has finished.
	*/
	static void
	send_traverse_thread(void *arg)
	{
	struct send_thread_arg *st_arg = arg;
	int err = 0;
	struct send_range *data;
	fstrans_cookie_t cookie = spl_fstrans_mark();

	err = traverse_dataset_resume(st_arg->os->os_dsl_dataset,
	st_arg->fromtxg, &st_arg->resume,
	st_arg->flags, send_cb, st_arg);

	if (err != EINTR)
	st_arg->error_code = err;
	data = range_alloc(DATA, 0, 0, 0, B_TRUE);
	bqueue_enqueue_flush(&st_arg->q, data, sizeof (*data));
	spl_fstrans_unmark(cookie);
	thread_exit();
	}

	/*
	* Utility function that causes End of Stream records to compare after of all
	* others, so that other threads' comparison logic can stay simple.
	*/
	static int __attribute__((unused))
	send_range_after(const struct send_range from, const struct send_range to)
	{
	if (from->eos_marker == B_TRUE)
	return (1);
	if (to->eos_marker == B_TRUE)
	return (-1);

	uint64_t from_obj = from->object;
	uint64_t from_end_obj = from->object + 1;
	uint64_t to_obj = to->object;
	uint64_t to_end_obj = to->object + 1;
	if (from_obj == 0) {
	ASSERT(from->type == HOLE \|\| from->type == OBJECT_RANGE);
	from_obj = from->start_blkid << DNODES_PER_BLOCK_SHIFT;
	from_end_obj = from->end_blkid << DNODES_PER_BLOCK_SHIFT;
	}
	if (to_obj == 0) {
	ASSERT(to->type == HOLE \|\| to->type == OBJECT_RANGE);
	to_obj = to->start_blkid << DNODES_PER_BLOCK_SHIFT;
	to_end_obj = to->end_blkid << DNODES_PER_BLOCK_SHIFT;
	}

	if (from_end_obj <= to_obj)
	return (-1);
	if (from_obj >= to_end_obj)
	return (1);
	int64_t cmp = TREE_CMP(to->type == OBJECT_RANGE, from->type ==
	OBJECT_RANGE);
	if (unlikely(cmp))
	return (cmp);
	cmp = TREE_CMP(to->type == OBJECT, from->type == OBJECT);
	if (unlikely(cmp))
	return (cmp);
	if (from->end_blkid <= to->start_blkid)
	return (-1);
	if (from->start_blkid >= to->end_blkid)
	return (1);
	return (0);
	}

	/*
	* Pop the new data off the queue, check that the records we receive are in
	* the right order, but do not free the old data. This is used so that the
	* records can be sent on to the main thread without copying the data.
	*/
	static struct send_range *
	get_next_range_nofree(bqueue_t bq, struct send_range prev)
	{
	struct send_range *next = bqueue_dequeue(bq);
	ASSERT3S(send_range_after(prev, next), ==, -1);
	return (next);
	}

	/*
	* Pop the new data off the queue, check that the records we receive are in
	* the right order, and free the old data.
	*/
	static struct send_range *
	get_next_range(bqueue_t bq, struct send_range prev)
	{
	struct send_range *next = get_next_range_nofree(bq, prev);
	range_free(prev);
	return (next);
	}

	static void
	redact_list_thread(void *arg)
	{
	struct redact_list_thread_arg *rlt_arg = arg;
	struct send_range *record;
	fstrans_cookie_t cookie = spl_fstrans_mark();
	if (rlt_arg->rl != NULL) {
	struct redact_list_cb_arg rlcba = {0};
	rlcba.cancel = &rlt_arg->cancel;
	rlcba.q = &rlt_arg->q;
	rlcba.num_blocks_visited = rlt_arg->num_blocks_visited;
	rlcba.mark_redact = rlt_arg->mark_redact;
	int err = dsl_redaction_list_traverse(rlt_arg->rl,
	&rlt_arg->resume, redact_list_cb, &rlcba);
	if (err != EINTR)
	rlt_arg->error_code = err;
	}
	record = range_alloc(DATA, 0, 0, 0, B_TRUE);
	bqueue_enqueue_flush(&rlt_arg->q, record, sizeof (*record));
	spl_fstrans_unmark(cookie);

	thread_exit();
	}

	/*
	* Compare the start point of the two provided ranges. End of stream ranges
	* compare last, objects compare before any data or hole inside that object and
	* multi-object holes that start at the same object.
	*/
	static int
	send_range_start_compare(struct send_range r1, struct send_range r2)
	{
	uint64_t r1_objequiv = r1->object;
	uint64_t r1_l0equiv = r1->start_blkid;
	uint64_t r2_objequiv = r2->object;
	uint64_t r2_l0equiv = r2->start_blkid;
	int64_t cmp = TREE_CMP(r1->eos_marker, r2->eos_marker);
	if (unlikely(cmp))
	return (cmp);
	if (r1->object == 0) {
	r1_objequiv = r1->start_blkid * DNODES_PER_BLOCK;
	r1_l0equiv = 0;
	}
	if (r2->object == 0) {
	r2_objequiv = r2->start_blkid * DNODES_PER_BLOCK;
	r2_l0equiv = 0;
	}

	cmp = TREE_CMP(r1_objequiv, r2_objequiv);
	if (likely(cmp))
	return (cmp);
	cmp = TREE_CMP(r2->type == OBJECT_RANGE, r1->type == OBJECT_RANGE);
	if (unlikely(cmp))
	return (cmp);
	cmp = TREE_CMP(r2->type == OBJECT, r1->type == OBJECT);
	if (unlikely(cmp))
	return (cmp);

	return (TREE_CMP(r1_l0equiv, r2_l0equiv));
	}

	enum q_idx {
	REDACT_IDX = 0,
	TO_IDX,
	FROM_IDX,
	NUM_THREADS
	};

	/*
	* This function returns the next range the send_merge_thread should operate on.
	* The inputs are two arrays; the first one stores the range at the front of the
	* queues stored in the second one. The ranges are sorted in descending
	* priority order; the metadata from earlier ranges overrules metadata from
	* later ranges. out_mask is used to return which threads the ranges came from;
	* bit i is set if ranges[i] started at the same place as the returned range.
	*
	* This code is not hardcoded to compare a specific number of threads; it could
	* be used with any number, just by changing the q_idx enum.
	*
	* The "next range" is the one with the earliest start; if two starts are equal,
	* the highest-priority range is the next to operate on. If a higher-priority
	* range starts in the middle of the first range, then the first range will be
	* truncated to end where the higher-priority range starts, and we will operate
	* on that one next time. In this way, we make sure that each block covered by
	* some range gets covered by a returned range, and each block covered is
	* returned using the metadata of the highest-priority range it appears in.
	*
	* For example, if the three ranges at the front of the queues were [2,4),
	* [3,5), and [1,3), then the ranges returned would be [1,2) with the metadata
	* from the third range, [2,4) with the metadata from the first range, and then
	* [4,5) with the metadata from the second.
	*/
	static struct send_range *
	find_next_range(struct send_range ranges, bqueue_t qs, uint64_t *out_mask)
	{
	int idx = 0; // index of the range with the earliest start
	int i;
	uint64_t bmask = 0;
	for (i = 1; i < NUM_THREADS; i++) {
	if (send_range_start_compare(ranges[i], ranges[idx]) < 0)
	idx = i;
	}
	if (ranges[idx]->eos_marker) {
	struct send_range *ret = range_alloc(DATA, 0, 0, 0, B_TRUE);
	*out_mask = 0;
	return (ret);
	}
	/*
	* Find all the ranges that start at that same point.
	*/
	for (i = 0; i < NUM_THREADS; i++) {
	if (send_range_start_compare(ranges[i], ranges[idx]) == 0)
	bmask \|= 1 << i;
	}
	*out_mask = bmask;
	/*
	* OBJECT_RANGE records only come from the TO thread, and should always
	* be treated as overlapping with nothing and sent on immediately. They
	* are only used in raw sends, and are never redacted.
	*/
	if (ranges[idx]->type == OBJECT_RANGE) {
	ASSERT3U(idx, ==, TO_IDX);
	ASSERT3U(*out_mask, ==, 1 << TO_IDX);
	struct send_range *ret = ranges[idx];
	ranges[idx] = get_next_range_nofree(qs[idx], ranges[idx]);
	return (ret);
	}
	/*
	* Find the first start or end point after the start of the first range.
	*/
	uint64_t first_change = ranges[idx]->end_blkid;
	for (i = 0; i < NUM_THREADS; i++) {
	if (i == idx \|\| ranges[i]->eos_marker \|\|
	ranges[i]->object > ranges[idx]->object \|\|
	ranges[i]->object == DMU_META_DNODE_OBJECT)
	continue;
	ASSERT3U(ranges[i]->object, ==, ranges[idx]->object);
	if (first_change > ranges[i]->start_blkid &&
	(bmask & (1 << i)) == 0)
	first_change = ranges[i]->start_blkid;
	else if (first_change > ranges[i]->end_blkid)
	first_change = ranges[i]->end_blkid;
	}
	/*
	* Update all ranges to no longer overlap with the range we're
	* returning. All such ranges must start at the same place as the range
	* being returned, and end at or after first_change. Thus we update
	* their start to first_change. If that makes them size 0, then free
	* them and pull a new range from that thread.
	*/
	for (i = 0; i < NUM_THREADS; i++) {
	if (i == idx \|\| (bmask & (1 << i)) == 0)
	continue;
	ASSERT3U(first_change, >, ranges[i]->start_blkid);
	ranges[i]->start_blkid = first_change;
	ASSERT3U(ranges[i]->start_blkid, <=, ranges[i]->end_blkid);
	if (ranges[i]->start_blkid == ranges[i]->end_blkid)
	ranges[i] = get_next_range(qs[i], ranges[i]);
	}
	/*
	* Short-circuit the simple case; if the range doesn't overlap with
	* anything else, or it only overlaps with things that start at the same
	* place and are longer, send it on.
	*/
	if (first_change == ranges[idx]->end_blkid) {
	struct send_range *ret = ranges[idx];
	ranges[idx] = get_next_range_nofree(qs[idx], ranges[idx]);
	return (ret);
	}

	/*
	* Otherwise, return a truncated copy of ranges[idx] and move the start
	* of ranges[idx] back to first_change.
	*/
	struct send_range ret = kmem_alloc(sizeof (ret), KM_SLEEP);
	ret = ranges[idx];
	ret->end_blkid = first_change;
	ranges[idx]->start_blkid = first_change;
	return (ret);
	}

	#define FROM_AND_REDACT_BITS ((1 << REDACT_IDX) \| (1 << FROM_IDX))

	/*
	* Merge the results from the from thread and the to thread, and then hand the
	* records off to send_prefetch_thread to prefetch them. If this is not a
	* send from a redaction bookmark, the from thread will push an end of stream
	* record and stop, and we'll just send everything that was changed in the
	* to_ds since the ancestor's creation txg. If it is, then since
	* traverse_dataset has a canonical order, we can compare each change as
	* they're pulled off the queues. That will give us a stream that is
	* appropriately sorted, and covers all records. In addition, we pull the
	* data from the redact_list_thread and use that to determine which blocks
	* should be redacted.
	*/
	static void
	send_merge_thread(void *arg)
	{
	struct send_merge_thread_arg *smt_arg = arg;
	struct send_range *front_ranges[NUM_THREADS];
	bqueue_t *queues[NUM_THREADS];
	int err = 0;
	fstrans_cookie_t cookie = spl_fstrans_mark();

	if (smt_arg->redact_arg == NULL) {
	front_ranges[REDACT_IDX] =
	kmem_zalloc(sizeof (struct send_range), KM_SLEEP);
	front_ranges[REDACT_IDX]->eos_marker = B_TRUE;
	front_ranges[REDACT_IDX]->type = REDACT;
	queues[REDACT_IDX] = NULL;
	} else {
	front_ranges[REDACT_IDX] =
	bqueue_dequeue(&smt_arg->redact_arg->q);
	queues[REDACT_IDX] = &smt_arg->redact_arg->q;
	}
	front_ranges[TO_IDX] = bqueue_dequeue(&smt_arg->to_arg->q);
	queues[TO_IDX] = &smt_arg->to_arg->q;
	front_ranges[FROM_IDX] = bqueue_dequeue(&smt_arg->from_arg->q);
	queues[FROM_IDX] = &smt_arg->from_arg->q;
	uint64_t mask = 0;
	struct send_range *range;
	for (range = find_next_range(front_ranges, queues, &mask);
	!range->eos_marker && err == 0 && !smt_arg->cancel;
	range = find_next_range(front_ranges, queues, &mask)) {
	/*
	* If the range in question was in both the from redact bookmark
	* and the bookmark we're using to redact, then don't send it.
	* It's already redacted on the receiving system, so a redaction
	* record would be redundant.
	*/
	if ((mask & FROM_AND_REDACT_BITS) == FROM_AND_REDACT_BITS) {
	ASSERT3U(range->type, ==, REDACT);
	range_free(range);
	continue;
	}
	bqueue_enqueue(&smt_arg->q, range, sizeof (*range));

	if (smt_arg->to_arg->error_code != 0) {
	err = smt_arg->to_arg->error_code;
	} else if (smt_arg->from_arg->error_code != 0) {
	err = smt_arg->from_arg->error_code;
	} else if (smt_arg->redact_arg != NULL &&
	smt_arg->redact_arg->error_code != 0) {
	err = smt_arg->redact_arg->error_code;
	}
	}
	if (smt_arg->cancel && err == 0)
	err = SET_ERROR(EINTR);
	smt_arg->error = err;
	if (smt_arg->error != 0) {
	smt_arg->to_arg->cancel = B_TRUE;
	smt_arg->from_arg->cancel = B_TRUE;
	if (smt_arg->redact_arg != NULL)
	smt_arg->redact_arg->cancel = B_TRUE;
	}
	for (int i = 0; i < NUM_THREADS; i++) {
	while (!front_ranges[i]->eos_marker) {
	front_ranges[i] = get_next_range(queues[i],
	front_ranges[i]);
	}
	range_free(front_ranges[i]);
	}
	if (range == NULL)
	range = kmem_zalloc(sizeof (*range), KM_SLEEP);
	range->eos_marker = B_TRUE;
	bqueue_enqueue_flush(&smt_arg->q, range, 1);
	spl_fstrans_unmark(cookie);
	thread_exit();
	}

	struct send_reader_thread_arg {
	struct send_merge_thread_arg *smta;
	bqueue_t q;
	boolean_t cancel;
	boolean_t issue_reads;
	uint64_t featureflags;
	int error;
	};

	static void
	dmu_send_read_done(zio_t *zio)
	{
	struct send_range *range = zio->io_private;

	mutex_enter(&range->sru.data.lock);
	if (zio->io_error != 0) {
	abd_free(range->sru.data.abd);
	range->sru.data.abd = NULL;
	range->sru.data.io_err = zio->io_error;
	}

	ASSERT(range->sru.data.io_outstanding);
	range->sru.data.io_outstanding = B_FALSE;
	cv_broadcast(&range->sru.data.cv);
	mutex_exit(&range->sru.data.lock);
	}

	static void
	issue_data_read(struct send_reader_thread_arg srta, struct send_range range)
	{
	struct srd *srdp = &range->sru.data;
	blkptr_t *bp = &srdp->bp;
	objset_t *os = srta->smta->os;

	ASSERT3U(range->type, ==, DATA);
	ASSERT3U(range->start_blkid + 1, ==, range->end_blkid);
	/*
	* If we have large blocks stored on disk but
	* the send flags don't allow us to send large
	* blocks, we split the data from the arc buf
	* into chunks.
	*/
	boolean_t split_large_blocks =
	srdp->datablksz > SPA_OLD_MAXBLOCKSIZE &&
	!(srta->featureflags & DMU_BACKUP_FEATURE_LARGE_BLOCKS);
	/*
	* We should only request compressed data from the ARC if all
	* the following are true:
	* - stream compression was requested
	* - we aren't splitting large blocks into smaller chunks
	* - the data won't need to be byteswapped before sending
	* - this isn't an embedded block
	* - this isn't metadata (if receiving on a different endian
	* system it can be byteswapped more easily)
	*/
	boolean_t request_compressed =
	(srta->featureflags & DMU_BACKUP_FEATURE_COMPRESSED) &&
	!split_large_blocks && !BP_SHOULD_BYTESWAP(bp) &&
	!BP_IS_EMBEDDED(bp) && !DMU_OT_IS_METADATA(BP_GET_TYPE(bp));

	enum zio_flag zioflags = ZIO_FLAG_CANFAIL;

	- if (srta->featureflags & DMU_BACKUP_FEATURE_RAW)
	+ if (srta->featureflags & DMU_BACKUP_FEATURE_RAW) {
	zioflags \|= ZIO_FLAG_RAW;
	- else if (request_compressed)
	+ srdp->io_compressed = B_TRUE;
	+ } else if (request_compressed) {
	zioflags \|= ZIO_FLAG_RAW_COMPRESS;
	+ srdp->io_compressed = B_TRUE;
	+ }

	srdp->datasz = (zioflags & ZIO_FLAG_RAW_COMPRESS) ?
	BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp);

	if (!srta->issue_reads)
	return;
	if (BP_IS_REDACTED(bp))
	return;
	if (send_do_embed(bp, srta->featureflags))
	return;

	zbookmark_phys_t zb = {
	.zb_objset = dmu_objset_id(os),
	.zb_object = range->object,
	.zb_level = 0,
	.zb_blkid = range->start_blkid,
	};

	arc_flags_t aflags = ARC_FLAG_CACHED_ONLY;

	int arc_err = arc_read(NULL, os->os_spa, bp,
	arc_getbuf_func, &srdp->abuf, ZIO_PRIORITY_ASYNC_READ,
	zioflags, &aflags, &zb);
	/*
	* If the data is not already cached in the ARC, we read directly
	* from zio. This avoids the performance overhead of adding a new
	* entry to the ARC, and we also avoid polluting the ARC cache with
	* data that is not likely to be used in the future.
	*/
	if (arc_err != 0) {
	srdp->abd = abd_alloc_linear(srdp->datasz, B_FALSE);
	srdp->io_outstanding = B_TRUE;
	zio_nowait(zio_read(NULL, os->os_spa, bp, srdp->abd,
	srdp->datasz, dmu_send_read_done, range,
	ZIO_PRIORITY_ASYNC_READ, zioflags, &zb));
	}
	}

	/*
	* Create a new record with the given values.
	*/
	static void
	enqueue_range(struct send_reader_thread_arg srta, bqueue_t q, dnode_t *dn,
	uint64_t blkid, uint64_t count, const blkptr_t *bp, uint32_t datablksz)
	{
	enum type range_type = (bp == NULL \|\| BP_IS_HOLE(bp) ? HOLE :
	(BP_IS_REDACTED(bp) ? REDACT : DATA));

	struct send_range *range = range_alloc(range_type, dn->dn_object,
	blkid, blkid + count, B_FALSE);

	if (blkid == DMU_SPILL_BLKID)
	ASSERT3U(BP_GET_TYPE(bp), ==, DMU_OT_SA);

	switch (range_type) {
	case HOLE:
	range->sru.hole.datablksz = datablksz;
	break;
	case DATA:
	ASSERT3U(count, ==, 1);
	range->sru.data.datablksz = datablksz;
	range->sru.data.obj_type = dn->dn_type;
	range->sru.data.bp = *bp;
	issue_data_read(srta, range);
	break;
	case REDACT:
	range->sru.redact.datablksz = datablksz;
	break;
	default:
	break;
	}
	bqueue_enqueue(q, range, datablksz);
	}

	/*
	* This thread is responsible for two things: First, it retrieves the correct
	* blkptr in the to ds if we need to send the data because of something from
	* the from thread. As a result of this, we're the first ones to discover that
	* some indirect blocks can be discarded because they're not holes. Second,
	* it issues prefetches for the data we need to send.
	*/
	static void
	send_reader_thread(void *arg)
	{
	struct send_reader_thread_arg *srta = arg;
	struct send_merge_thread_arg *smta = srta->smta;
	bqueue_t *inq = &smta->q;
	bqueue_t *outq = &srta->q;
	objset_t *os = smta->os;
	fstrans_cookie_t cookie = spl_fstrans_mark();
	struct send_range *range = bqueue_dequeue(inq);
	int err = 0;

	/*
	* If the record we're analyzing is from a redaction bookmark from the
	* fromds, then we need to know whether or not it exists in the tods so
	* we know whether to create records for it or not. If it does, we need
	* the datablksz so we can generate an appropriate record for it.
	* Finally, if it isn't redacted, we need the blkptr so that we can send
	* a WRITE record containing the actual data.
	*/
	uint64_t last_obj = UINT64_MAX;
	uint64_t last_obj_exists = B_TRUE;
	while (!range->eos_marker && !srta->cancel && smta->error == 0 &&
	err == 0) {
	switch (range->type) {
	case DATA:
	issue_data_read(srta, range);
	bqueue_enqueue(outq, range, range->sru.data.datablksz);
	range = get_next_range_nofree(inq, range);
	break;
	case HOLE:
	case OBJECT:
	case OBJECT_RANGE:
	case REDACT: // Redacted blocks must exist
	bqueue_enqueue(outq, range, sizeof (*range));
	range = get_next_range_nofree(inq, range);
	break;
	case PREVIOUSLY_REDACTED: {
	/*
	* This entry came from the "from bookmark" when
	* sending from a bookmark that has a redaction
	* list. We need to check if this object/blkid
	* exists in the target ("to") dataset, and if
	* not then we drop this entry. We also need
	* to fill in the block pointer so that we know
	* what to prefetch.
	*
	* To accomplish the above, we first cache whether or
	* not the last object we examined exists. If it
	* doesn't, we can drop this record. If it does, we hold
	* the dnode and use it to call dbuf_dnode_findbp. We do
	* this instead of dbuf_bookmark_findbp because we will
	* often operate on large ranges, and holding the dnode
	* once is more efficient.
	*/
	boolean_t object_exists = B_TRUE;
	/*
	* If the data is redacted, we only care if it exists,
	* so that we don't send records for objects that have
	* been deleted.
	*/
	dnode_t *dn;
	if (range->object == last_obj && !last_obj_exists) {
	/*
	* If we're still examining the same object as
	* previously, and it doesn't exist, we don't
	* need to call dbuf_bookmark_findbp.
	*/
	object_exists = B_FALSE;
	} else {
	err = dnode_hold(os, range->object, FTAG, &dn);
	if (err == ENOENT) {
	object_exists = B_FALSE;
	err = 0;
	}
	last_obj = range->object;
	last_obj_exists = object_exists;
	}

	if (err != 0) {
	break;
	} else if (!object_exists) {
	/*
	* The block was modified, but doesn't
	* exist in the to dataset; if it was
	* deleted in the to dataset, then we'll
	* visit the hole bp for it at some point.
	*/
	range = get_next_range(inq, range);
	continue;
	}
	uint64_t file_max =
	(dn->dn_maxblkid < range->end_blkid ?
	dn->dn_maxblkid : range->end_blkid);
	/*
	* The object exists, so we need to try to find the
	* blkptr for each block in the range we're processing.
	*/
	rw_enter(&dn->dn_struct_rwlock, RW_READER);
	for (uint64_t blkid = range->start_blkid;
	blkid < file_max; blkid++) {
	blkptr_t bp;
	uint32_t datablksz =
	dn->dn_phys->dn_datablkszsec <<
	SPA_MINBLOCKSHIFT;
	uint64_t offset = blkid * datablksz;
	/*
	* This call finds the next non-hole block in
	* the object. This is to prevent a
	* performance problem where we're unredacting
	* a large hole. Using dnode_next_offset to
	* skip over the large hole avoids iterating
	* over every block in it.
	*/
	err = dnode_next_offset(dn, DNODE_FIND_HAVELOCK,
	&offset, 1, 1, 0);
	if (err == ESRCH) {
	offset = UINT64_MAX;
	err = 0;
	} else if (err != 0) {
	break;
	}
	if (offset != blkid * datablksz) {
	/*
	* if there is a hole from here
	* (blkid) to offset
	*/
	offset = MIN(offset, file_max *
	datablksz);
	uint64_t nblks = (offset / datablksz) -
	blkid;
	enqueue_range(srta, outq, dn, blkid,
	nblks, NULL, datablksz);
	blkid += nblks;
	}
	if (blkid >= file_max)
	break;
	err = dbuf_dnode_findbp(dn, 0, blkid, &bp,
	NULL, NULL);
	if (err != 0)
	break;
	ASSERT(!BP_IS_HOLE(&bp));
	enqueue_range(srta, outq, dn, blkid, 1, &bp,
	datablksz);
	}
	rw_exit(&dn->dn_struct_rwlock);
	dnode_rele(dn, FTAG);
	range = get_next_range(inq, range);
	}
	}
	}
	if (srta->cancel \|\| err != 0) {
	smta->cancel = B_TRUE;
	srta->error = err;
	} else if (smta->error != 0) {
	srta->error = smta->error;
	}
	while (!range->eos_marker)
	range = get_next_range(inq, range);

	bqueue_enqueue_flush(outq, range, 1);
	spl_fstrans_unmark(cookie);
	thread_exit();
	}

	#define NUM_SNAPS_NOT_REDACTED UINT64_MAX

	struct dmu_send_params {
	/* Pool args */
	void *tag; // Tag that dp was held with, will be used to release dp.
	dsl_pool_t *dp;
	/* To snapshot args */
	const char *tosnap;
	dsl_dataset_t *to_ds;
	/* From snapshot args */
	zfs_bookmark_phys_t ancestor_zb;
	uint64_t *fromredactsnaps;
	/* NUM_SNAPS_NOT_REDACTED if not sending from redaction bookmark */
	uint64_t numfromredactsnaps;
	/* Stream params */
	boolean_t is_clone;
	boolean_t embedok;
	boolean_t large_block_ok;
	boolean_t compressok;
	boolean_t rawok;
	boolean_t savedok;
	uint64_t resumeobj;
	uint64_t resumeoff;
	uint64_t saved_guid;
	zfs_bookmark_phys_t *redactbook;
	/* Stream output params */
	dmu_send_outparams_t *dso;

	/* Stream progress params */
	offset_t *off;
	int outfd;
	char saved_toname[MAXNAMELEN];
	};

	static int
	setup_featureflags(struct dmu_send_params dspp, objset_t os,
	uint64_t *featureflags)
	{
	dsl_dataset_t *to_ds = dspp->to_ds;
	dsl_pool_t *dp = dspp->dp;
	#ifdef _KERNEL
	if (dmu_objset_type(os) == DMU_OST_ZFS) {
	uint64_t version;
	if (zfs_get_zplprop(os, ZFS_PROP_VERSION, &version) != 0)
	return (SET_ERROR(EINVAL));

	if (version >= ZPL_VERSION_SA)
	*featureflags \|= DMU_BACKUP_FEATURE_SA_SPILL;
	}
	#endif

	/* raw sends imply large_block_ok */
	if ((dspp->rawok \|\| dspp->large_block_ok) &&
	dsl_dataset_feature_is_active(to_ds, SPA_FEATURE_LARGE_BLOCKS)) {
	*featureflags \|= DMU_BACKUP_FEATURE_LARGE_BLOCKS;
	}

	/* encrypted datasets will not have embedded blocks */
	if ((dspp->embedok \|\| dspp->rawok) && !os->os_encrypted &&
	spa_feature_is_active(dp->dp_spa, SPA_FEATURE_EMBEDDED_DATA)) {
	*featureflags \|= DMU_BACKUP_FEATURE_EMBED_DATA;
	}

	/* raw send implies compressok */
	if (dspp->compressok \|\| dspp->rawok)
	*featureflags \|= DMU_BACKUP_FEATURE_COMPRESSED;

	if (dspp->rawok && os->os_encrypted)
	*featureflags \|= DMU_BACKUP_FEATURE_RAW;

	if ((*featureflags &
	(DMU_BACKUP_FEATURE_EMBED_DATA \| DMU_BACKUP_FEATURE_COMPRESSED \|
	DMU_BACKUP_FEATURE_RAW)) != 0 &&
	spa_feature_is_active(dp->dp_spa, SPA_FEATURE_LZ4_COMPRESS)) {
	*featureflags \|= DMU_BACKUP_FEATURE_LZ4;
	}

	/*
	* We specifically do not include DMU_BACKUP_FEATURE_EMBED_DATA here to
	* allow sending ZSTD compressed datasets to a receiver that does not
	* support ZSTD
	*/
	if ((*featureflags &
	(DMU_BACKUP_FEATURE_COMPRESSED \| DMU_BACKUP_FEATURE_RAW)) != 0 &&
	dsl_dataset_feature_is_active(to_ds, SPA_FEATURE_ZSTD_COMPRESS)) {
	*featureflags \|= DMU_BACKUP_FEATURE_ZSTD;
	}

	if (dspp->resumeobj != 0 \|\| dspp->resumeoff != 0) {
	*featureflags \|= DMU_BACKUP_FEATURE_RESUMING;
	}

	if (dspp->redactbook != NULL) {
	*featureflags \|= DMU_BACKUP_FEATURE_REDACTED;
	}

	if (dsl_dataset_feature_is_active(to_ds, SPA_FEATURE_LARGE_DNODE)) {
	*featureflags \|= DMU_BACKUP_FEATURE_LARGE_DNODE;
	}
	return (0);
	}

	static dmu_replay_record_t *
	create_begin_record(struct dmu_send_params dspp, objset_t os,
	uint64_t featureflags)
	{
	dmu_replay_record_t *drr = kmem_zalloc(sizeof (dmu_replay_record_t),
	KM_SLEEP);
	drr->drr_type = DRR_BEGIN;

	struct drr_begin *drrb = &drr->drr_u.drr_begin;
	dsl_dataset_t *to_ds = dspp->to_ds;

	drrb->drr_magic = DMU_BACKUP_MAGIC;
	drrb->drr_creation_time = dsl_dataset_phys(to_ds)->ds_creation_time;
	drrb->drr_type = dmu_objset_type(os);
	drrb->drr_toguid = dsl_dataset_phys(to_ds)->ds_guid;
	drrb->drr_fromguid = dspp->ancestor_zb.zbm_guid;

	DMU_SET_STREAM_HDRTYPE(drrb->drr_versioninfo, DMU_SUBSTREAM);
	DMU_SET_FEATUREFLAGS(drrb->drr_versioninfo, featureflags);

	if (dspp->is_clone)
	drrb->drr_flags \|= DRR_FLAG_CLONE;
	if (dsl_dataset_phys(dspp->to_ds)->ds_flags & DS_FLAG_CI_DATASET)
	drrb->drr_flags \|= DRR_FLAG_CI_DATA;
	if (zfs_send_set_freerecords_bit)
	drrb->drr_flags \|= DRR_FLAG_FREERECORDS;
	drr->drr_u.drr_begin.drr_flags \|= DRR_FLAG_SPILL_BLOCK;

	if (dspp->savedok) {
	drrb->drr_toguid = dspp->saved_guid;
	strlcpy(drrb->drr_toname, dspp->saved_toname,
	sizeof (drrb->drr_toname));
	} else {
	dsl_dataset_name(to_ds, drrb->drr_toname);
	if (!to_ds->ds_is_snapshot) {
	(void) strlcat(drrb->drr_toname, "@--head--",
	sizeof (drrb->drr_toname));
	}
	}
	return (drr);
	}

	static void
	setup_to_thread(struct send_thread_arg to_arg, objset_t to_os,
	dmu_sendstatus_t *dssp, uint64_t fromtxg, boolean_t rawok)
	{
	VERIFY0(bqueue_init(&to_arg->q, zfs_send_no_prefetch_queue_ff,
	MAX(zfs_send_no_prefetch_queue_length, 2 * zfs_max_recordsize),
	offsetof(struct send_range, ln)));
	to_arg->error_code = 0;
	to_arg->cancel = B_FALSE;
	to_arg->os = to_os;
	to_arg->fromtxg = fromtxg;
	to_arg->flags = TRAVERSE_PRE \| TRAVERSE_PREFETCH_METADATA;
	if (rawok)
	to_arg->flags \|= TRAVERSE_NO_DECRYPT;
	if (zfs_send_corrupt_data)
	to_arg->flags \|= TRAVERSE_HARD;
	to_arg->num_blocks_visited = &dssp->dss_blocks;
	(void) thread_create(NULL, 0, send_traverse_thread, to_arg, 0,
	curproc, TS_RUN, minclsyspri);
	}

	static void
	setup_from_thread(struct redact_list_thread_arg *from_arg,
	redaction_list_t from_rl, dmu_sendstatus_t dssp)
	{
	VERIFY0(bqueue_init(&from_arg->q, zfs_send_no_prefetch_queue_ff,
	MAX(zfs_send_no_prefetch_queue_length, 2 * zfs_max_recordsize),
	offsetof(struct send_range, ln)));
	from_arg->error_code = 0;
	from_arg->cancel = B_FALSE;
	from_arg->rl = from_rl;
	from_arg->mark_redact = B_FALSE;
	from_arg->num_blocks_visited = &dssp->dss_blocks;
	/*
	* If from_ds is null, send_traverse_thread just returns success and
	* enqueues an eos marker.
	*/
	(void) thread_create(NULL, 0, redact_list_thread, from_arg, 0,
	curproc, TS_RUN, minclsyspri);
	}

	static void
	setup_redact_list_thread(struct redact_list_thread_arg *rlt_arg,
	struct dmu_send_params dspp, redaction_list_t rl, dmu_sendstatus_t *dssp)
	{
	if (dspp->redactbook == NULL)
	return;

	rlt_arg->cancel = B_FALSE;
	VERIFY0(bqueue_init(&rlt_arg->q, zfs_send_no_prefetch_queue_ff,
	MAX(zfs_send_no_prefetch_queue_length, 2 * zfs_max_recordsize),
	offsetof(struct send_range, ln)));
	rlt_arg->error_code = 0;
	rlt_arg->mark_redact = B_TRUE;
	rlt_arg->rl = rl;
	rlt_arg->num_blocks_visited = &dssp->dss_blocks;

	(void) thread_create(NULL, 0, redact_list_thread, rlt_arg, 0,
	curproc, TS_RUN, minclsyspri);
	}

	static void
	setup_merge_thread(struct send_merge_thread_arg *smt_arg,
	struct dmu_send_params dspp, struct redact_list_thread_arg from_arg,
	struct send_thread_arg to_arg, struct redact_list_thread_arg rlt_arg,
	objset_t *os)
	{
	VERIFY0(bqueue_init(&smt_arg->q, zfs_send_no_prefetch_queue_ff,
	MAX(zfs_send_no_prefetch_queue_length, 2 * zfs_max_recordsize),
	offsetof(struct send_range, ln)));
	smt_arg->cancel = B_FALSE;
	smt_arg->error = 0;
	smt_arg->from_arg = from_arg;
	smt_arg->to_arg = to_arg;
	if (dspp->redactbook != NULL)
	smt_arg->redact_arg = rlt_arg;

	smt_arg->os = os;
	(void) thread_create(NULL, 0, send_merge_thread, smt_arg, 0, curproc,
	TS_RUN, minclsyspri);
	}

	static void
	setup_reader_thread(struct send_reader_thread_arg *srt_arg,
	struct dmu_send_params dspp, struct send_merge_thread_arg smt_arg,
	uint64_t featureflags)
	{
	VERIFY0(bqueue_init(&srt_arg->q, zfs_send_queue_ff,
	MAX(zfs_send_queue_length, 2 * zfs_max_recordsize),
	offsetof(struct send_range, ln)));
	srt_arg->smta = smt_arg;
	srt_arg->issue_reads = !dspp->dso->dso_dryrun;
	srt_arg->featureflags = featureflags;
	(void) thread_create(NULL, 0, send_reader_thread, srt_arg, 0,
	curproc, TS_RUN, minclsyspri);
	}

	static int
	setup_resume_points(struct dmu_send_params *dspp,
	struct send_thread_arg to_arg, struct redact_list_thread_arg from_arg,
	struct redact_list_thread_arg *rlt_arg,
	struct send_merge_thread_arg smt_arg, boolean_t resuming, objset_t os,
	redaction_list_t redact_rl, nvlist_t nvl)
	{
	dsl_dataset_t *to_ds = dspp->to_ds;
	int err = 0;

	uint64_t obj = 0;
	uint64_t blkid = 0;
	if (resuming) {
	obj = dspp->resumeobj;
	dmu_object_info_t to_doi;
	err = dmu_object_info(os, obj, &to_doi);
	if (err != 0)
	return (err);

	blkid = dspp->resumeoff / to_doi.doi_data_block_size;
	}
	/*
	* If we're resuming a redacted send, we can skip to the appropriate
	* point in the redaction bookmark by binary searching through it.
	*/
	if (redact_rl != NULL) {
	SET_BOOKMARK(&rlt_arg->resume, to_ds->ds_object, obj, 0, blkid);
	}

	SET_BOOKMARK(&to_arg->resume, to_ds->ds_object, obj, 0, blkid);
	if (nvlist_exists(nvl, BEGINNV_REDACT_FROM_SNAPS)) {
	uint64_t objset = dspp->ancestor_zb.zbm_redaction_obj;
	/*
	* Note: If the resume point is in an object whose
	* blocksize is different in the from vs to snapshots,
	* we will have divided by the "wrong" blocksize.
	* However, in this case fromsnap's send_cb() will
	* detect that the blocksize has changed and therefore
	* ignore this object.
	*
	* If we're resuming a send from a redaction bookmark,
	* we still cannot accidentally suggest blocks behind
	* the to_ds. In addition, we know that any blocks in
	* the object in the to_ds will have to be sent, since
	* the size changed. Therefore, we can't cause any harm
	* this way either.
	*/
	SET_BOOKMARK(&from_arg->resume, objset, obj, 0, blkid);
	}
	if (resuming) {
	fnvlist_add_uint64(nvl, BEGINNV_RESUME_OBJECT, dspp->resumeobj);
	fnvlist_add_uint64(nvl, BEGINNV_RESUME_OFFSET, dspp->resumeoff);
	}
	return (0);
	}

	static dmu_sendstatus_t *
	setup_send_progress(struct dmu_send_params *dspp)
	{
	dmu_sendstatus_t dssp = kmem_zalloc(sizeof (dssp), KM_SLEEP);
	dssp->dss_outfd = dspp->outfd;
	dssp->dss_off = dspp->off;
	dssp->dss_proc = curproc;
	mutex_enter(&dspp->to_ds->ds_sendstream_lock);
	list_insert_head(&dspp->to_ds->ds_sendstreams, dssp);
	mutex_exit(&dspp->to_ds->ds_sendstream_lock);
	return (dssp);
	}

	/*
	* Actually do the bulk of the work in a zfs send.
	*
	* The idea is that we want to do a send from ancestor_zb to to_ds. We also
	* want to not send any data that has been modified by all the datasets in
	* redactsnaparr, and store the list of blocks that are redacted in this way in
	* a bookmark named redactbook, created on the to_ds. We do this by creating
	* several worker threads, whose function is described below.
	*
	* There are three cases.
	* The first case is a redacted zfs send. In this case there are 5 threads.
	* The first thread is the to_ds traversal thread: it calls dataset_traverse on
	* the to_ds and finds all the blocks that have changed since ancestor_zb (if
	* it's a full send, that's all blocks in the dataset). It then sends those
	* blocks on to the send merge thread. The redact list thread takes the data
	* from the redaction bookmark and sends those blocks on to the send merge
	* thread. The send merge thread takes the data from the to_ds traversal
	* thread, and combines it with the redaction records from the redact list
	* thread. If a block appears in both the to_ds's data and the redaction data,
	* the send merge thread will mark it as redacted and send it on to the prefetch
	* thread. Otherwise, the send merge thread will send the block on to the
	* prefetch thread unchanged. The prefetch thread will issue prefetch reads for
	* any data that isn't redacted, and then send the data on to the main thread.
	* The main thread behaves the same as in a normal send case, issuing demand
	* reads for data blocks and sending out records over the network
	*
	* The graphic below diagrams the flow of data in the case of a redacted zfs
	* send. Each box represents a thread, and each line represents the flow of
	* data.
	*
	* Records from the \|
	* redaction bookmark \|
	* +--------------------+ \| +---------------------------+
	* \| \| v \| Send Merge Thread \|
	* \| Redact List Thread +----------> Apply redaction marks to \|
	* \| \| \| records as specified by \|
	* +--------------------+ \| redaction ranges \|
	* +----^---------------+------+
	* \| \| Merged data
	* \| \|
	* \| +------------v--------+
	* \| \| Prefetch Thread \|
	* +--------------------+ \| \| Issues prefetch \|
	* \| to_ds Traversal \| \| \| reads of data blocks\|
	* \| Thread (finds +---------------+ +------------+--------+
	* \| candidate blocks) \| Blocks modified \| Prefetched data
	* +--------------------+ by to_ds since \|
	* ancestor_zb +------------v----+
	* \| Main Thread \| File Descriptor
	* \| Sends data over +->(to zfs receive)
	* \| wire \|
	* +-----------------+
	*
	* The second case is an incremental send from a redaction bookmark. The to_ds
	* traversal thread and the main thread behave the same as in the redacted
	* send case. The new thread is the from bookmark traversal thread. It
	* iterates over the redaction list in the redaction bookmark, and enqueues
	* records for each block that was redacted in the original send. The send
	* merge thread now has to merge the data from the two threads. For details
	* about that process, see the header comment of send_merge_thread(). Any data
	* it decides to send on will be prefetched by the prefetch thread. Note that
	* you can perform a redacted send from a redaction bookmark; in that case,
	* the data flow behaves very similarly to the flow in the redacted send case,
	* except with the addition of the bookmark traversal thread iterating over the
	* redaction bookmark. The send_merge_thread also has to take on the
	* responsibility of merging the redact list thread's records, the bookmark
	* traversal thread's records, and the to_ds records.
	*
	* +---------------------+
	* \| \|
	* \| Redact List Thread +--------------+
	* \| \| \|
	* +---------------------+ \|
	* Blocks in redaction list \| Ranges modified by every secure snap
	* of from bookmark \| (or EOS if not readcted)
	* \|
	* +---------------------+ \| +----v----------------------+
	* \| bookmark Traversal \| v \| Send Merge Thread \|
	* \| Thread (finds +---------> Merges bookmark, rlt, and \|
	* \| candidate blocks) \| \| to_ds send records \|
	* +---------------------+ +----^---------------+------+
	* \| \| Merged data
	* \| +------------v--------+
	* \| \| Prefetch Thread \|
	* +--------------------+ \| \| Issues prefetch \|
	* \| to_ds Traversal \| \| \| reads of data blocks\|
	* \| Thread (finds +---------------+ +------------+--------+
	* \| candidate blocks) \| Blocks modified \| Prefetched data
	* +--------------------+ by to_ds since +------------v----+
	* ancestor_zb \| Main Thread \| File Descriptor
	* \| Sends data over +->(to zfs receive)
	* \| wire \|
	* +-----------------+
	*
	* The final case is a simple zfs full or incremental send. The to_ds traversal
	* thread behaves the same as always. The redact list thread is never started.
	* The send merge thread takes all the blocks that the to_ds traversal thread
	* sends it, prefetches the data, and sends the blocks on to the main thread.
	* The main thread sends the data over the wire.
	*
	* To keep performance acceptable, we want to prefetch the data in the worker
	* threads. While the to_ds thread could simply use the TRAVERSE_PREFETCH
	* feature built into traverse_dataset, the combining and deletion of records
	* due to redaction and sends from redaction bookmarks mean that we could
	* issue many unnecessary prefetches. As a result, we only prefetch data
	* after we've determined that the record is not going to be redacted. To
	* prevent the prefetching from getting too far ahead of the main thread, the
	* blocking queues that are used for communication are capped not by the
	* number of entries in the queue, but by the sum of the size of the
	* prefetches associated with them. The limit on the amount of data that the
	* thread can prefetch beyond what the main thread has reached is controlled
	* by the global variable zfs_send_queue_length. In addition, to prevent poor
	* performance in the beginning of a send, we also limit the distance ahead
	* that the traversal threads can be. That distance is controlled by the
	* zfs_send_no_prefetch_queue_length tunable.
	*
	* Note: Releases dp using the specified tag.
	*/
	static int
	dmu_send_impl(struct dmu_send_params *dspp)
	{
	objset_t *os;
	dmu_replay_record_t *drr;
	dmu_sendstatus_t *dssp;
	dmu_send_cookie_t dsc = {0};
	int err;
	uint64_t fromtxg = dspp->ancestor_zb.zbm_creation_txg;
	uint64_t featureflags = 0;
	struct redact_list_thread_arg *from_arg;
	struct send_thread_arg *to_arg;
	struct redact_list_thread_arg *rlt_arg;
	struct send_merge_thread_arg *smt_arg;
	struct send_reader_thread_arg *srt_arg;
	struct send_range *range;
	redaction_list_t *from_rl = NULL;
	redaction_list_t *redact_rl = NULL;
	boolean_t resuming = (dspp->resumeobj != 0 \|\| dspp->resumeoff != 0);
	boolean_t book_resuming = resuming;

	dsl_dataset_t *to_ds = dspp->to_ds;
	zfs_bookmark_phys_t *ancestor_zb = &dspp->ancestor_zb;
	dsl_pool_t *dp = dspp->dp;
	void *tag = dspp->tag;

	err = dmu_objset_from_ds(to_ds, &os);
	if (err != 0) {
	dsl_pool_rele(dp, tag);
	return (err);
	}

	/*
	* If this is a non-raw send of an encrypted ds, we can ensure that
	* the objset_phys_t is authenticated. This is safe because this is
	* either a snapshot or we have owned the dataset, ensuring that
	* it can't be modified.
	*/
	if (!dspp->rawok && os->os_encrypted &&
	arc_is_unauthenticated(os->os_phys_buf)) {
	zbookmark_phys_t zb;

	SET_BOOKMARK(&zb, to_ds->ds_object, ZB_ROOT_OBJECT,
	ZB_ROOT_LEVEL, ZB_ROOT_BLKID);
	err = arc_untransform(os->os_phys_buf, os->os_spa,
	&zb, B_FALSE);
	if (err != 0) {
	dsl_pool_rele(dp, tag);
	return (err);
	}

	ASSERT0(arc_is_unauthenticated(os->os_phys_buf));
	}

	if ((err = setup_featureflags(dspp, os, &featureflags)) != 0) {
	dsl_pool_rele(dp, tag);
	return (err);
	}

	/*
	* If we're doing a redacted send, hold the bookmark's redaction list.
	*/
	if (dspp->redactbook != NULL) {
	err = dsl_redaction_list_hold_obj(dp,
	dspp->redactbook->zbm_redaction_obj, FTAG,
	&redact_rl);
	if (err != 0) {
	dsl_pool_rele(dp, tag);
	return (SET_ERROR(EINVAL));
	}
	dsl_redaction_list_long_hold(dp, redact_rl, FTAG);
	}

	/*
	* If we're sending from a redaction bookmark, hold the redaction list
	* so that we can consider sending the redacted blocks.
	*/
	if (ancestor_zb->zbm_redaction_obj != 0) {
	err = dsl_redaction_list_hold_obj(dp,
	ancestor_zb->zbm_redaction_obj, FTAG, &from_rl);
	if (err != 0) {
	if (redact_rl != NULL) {
	dsl_redaction_list_long_rele(redact_rl, FTAG);
	dsl_redaction_list_rele(redact_rl, FTAG);
	}
	dsl_pool_rele(dp, tag);
	return (SET_ERROR(EINVAL));
	}
	dsl_redaction_list_long_hold(dp, from_rl, FTAG);
	}

	dsl_dataset_long_hold(to_ds, FTAG);

	from_arg = kmem_zalloc(sizeof (*from_arg), KM_SLEEP);
	to_arg = kmem_zalloc(sizeof (*to_arg), KM_SLEEP);
	rlt_arg = kmem_zalloc(sizeof (*rlt_arg), KM_SLEEP);
	smt_arg = kmem_zalloc(sizeof (*smt_arg), KM_SLEEP);
	srt_arg = kmem_zalloc(sizeof (*srt_arg), KM_SLEEP);

	drr = create_begin_record(dspp, os, featureflags);
	dssp = setup_send_progress(dspp);

	dsc.dsc_drr = drr;
	dsc.dsc_dso = dspp->dso;
	dsc.dsc_os = os;
	dsc.dsc_off = dspp->off;
	dsc.dsc_toguid = dsl_dataset_phys(to_ds)->ds_guid;
	dsc.dsc_fromtxg = fromtxg;
	dsc.dsc_pending_op = PENDING_NONE;
	dsc.dsc_featureflags = featureflags;
	dsc.dsc_resume_object = dspp->resumeobj;
	dsc.dsc_resume_offset = dspp->resumeoff;

	dsl_pool_rele(dp, tag);

	void *payload = NULL;
	size_t payload_len = 0;
	nvlist_t *nvl = fnvlist_alloc();

	/*
	* If we're doing a redacted send, we include the snapshots we're
	* redacted with respect to so that the target system knows what send
	* streams can be correctly received on top of this dataset. If we're
	* instead sending a redacted dataset, we include the snapshots that the
	* dataset was created with respect to.
	*/
	if (dspp->redactbook != NULL) {
	fnvlist_add_uint64_array(nvl, BEGINNV_REDACT_SNAPS,
	redact_rl->rl_phys->rlp_snaps,
	redact_rl->rl_phys->rlp_num_snaps);
	} else if (dsl_dataset_feature_is_active(to_ds,
	SPA_FEATURE_REDACTED_DATASETS)) {
	uint64_t *tods_guids;
	uint64_t length;
	VERIFY(dsl_dataset_get_uint64_array_feature(to_ds,
	SPA_FEATURE_REDACTED_DATASETS, &length, &tods_guids));
	fnvlist_add_uint64_array(nvl, BEGINNV_REDACT_SNAPS, tods_guids,
	length);
	}

	/*
	* If we're sending from a redaction bookmark, then we should retrieve
	* the guids of that bookmark so we can send them over the wire.
	*/
	if (from_rl != NULL) {
	fnvlist_add_uint64_array(nvl, BEGINNV_REDACT_FROM_SNAPS,
	from_rl->rl_phys->rlp_snaps,
	from_rl->rl_phys->rlp_num_snaps);
	}

	/*
	* If the snapshot we're sending from is redacted, include the redaction
	* list in the stream.
	*/
	if (dspp->numfromredactsnaps != NUM_SNAPS_NOT_REDACTED) {
	ASSERT3P(from_rl, ==, NULL);
	fnvlist_add_uint64_array(nvl, BEGINNV_REDACT_FROM_SNAPS,
	dspp->fromredactsnaps, (uint_t)dspp->numfromredactsnaps);
	if (dspp->numfromredactsnaps > 0) {
	kmem_free(dspp->fromredactsnaps,
	dspp->numfromredactsnaps * sizeof (uint64_t));
	dspp->fromredactsnaps = NULL;
	}
	}

	if (resuming \|\| book_resuming) {
	err = setup_resume_points(dspp, to_arg, from_arg,
	rlt_arg, smt_arg, resuming, os, redact_rl, nvl);
	if (err != 0)
	goto out;
	}

	if (featureflags & DMU_BACKUP_FEATURE_RAW) {
	uint64_t ivset_guid = (ancestor_zb != NULL) ?
	ancestor_zb->zbm_ivset_guid : 0;
	nvlist_t *keynvl = NULL;
	ASSERT(os->os_encrypted);

	err = dsl_crypto_populate_key_nvlist(os, ivset_guid,
	&keynvl);
	if (err != 0) {
	fnvlist_free(nvl);
	goto out;
	}

	fnvlist_add_nvlist(nvl, "crypt_keydata", keynvl);
	fnvlist_free(keynvl);
	}

	if (!nvlist_empty(nvl)) {
	payload = fnvlist_pack(nvl, &payload_len);
	drr->drr_payloadlen = payload_len;
	}

	fnvlist_free(nvl);
	err = dump_record(&dsc, payload, payload_len);
	fnvlist_pack_free(payload, payload_len);
	if (err != 0) {
	err = dsc.dsc_err;
	goto out;
	}

	setup_to_thread(to_arg, os, dssp, fromtxg, dspp->rawok);
	setup_from_thread(from_arg, from_rl, dssp);
	setup_redact_list_thread(rlt_arg, dspp, redact_rl, dssp);
	setup_merge_thread(smt_arg, dspp, from_arg, to_arg, rlt_arg, os);
	setup_reader_thread(srt_arg, dspp, smt_arg, featureflags);

	range = bqueue_dequeue(&srt_arg->q);
	while (err == 0 && !range->eos_marker) {
	err = do_dump(&dsc, range);
	range = get_next_range(&srt_arg->q, range);
	if (issig(JUSTLOOKING) && issig(FORREAL))
	err = SET_ERROR(EINTR);
	}

	/*
	* If we hit an error or are interrupted, cancel our worker threads and
	* clear the queue of any pending records. The threads will pass the
	* cancel up the tree of worker threads, and each one will clean up any
	* pending records before exiting.
	*/
	if (err != 0) {
	srt_arg->cancel = B_TRUE;
	while (!range->eos_marker) {
	range = get_next_range(&srt_arg->q, range);
	}
	}
	range_free(range);

	bqueue_destroy(&srt_arg->q);
	bqueue_destroy(&smt_arg->q);
	if (dspp->redactbook != NULL)
	bqueue_destroy(&rlt_arg->q);
	bqueue_destroy(&to_arg->q);
	bqueue_destroy(&from_arg->q);

	if (err == 0 && srt_arg->error != 0)
	err = srt_arg->error;

	if (err != 0)
	goto out;

	if (dsc.dsc_pending_op != PENDING_NONE)
	if (dump_record(&dsc, NULL, 0) != 0)
	err = SET_ERROR(EINTR);

	if (err != 0) {
	if (err == EINTR && dsc.dsc_err != 0)
	err = dsc.dsc_err;
	goto out;
	}

	/*
	* Send the DRR_END record if this is not a saved stream.
	* Otherwise, the omitted DRR_END record will signal to
	* the receive side that the stream is incomplete.
	*/
	if (!dspp->savedok) {
	bzero(drr, sizeof (dmu_replay_record_t));
	drr->drr_type = DRR_END;
	drr->drr_u.drr_end.drr_checksum = dsc.dsc_zc;
	drr->drr_u.drr_end.drr_toguid = dsc.dsc_toguid;

	if (dump_record(&dsc, NULL, 0) != 0)
	err = dsc.dsc_err;
	}
	out:
	mutex_enter(&to_ds->ds_sendstream_lock);
	list_remove(&to_ds->ds_sendstreams, dssp);
	mutex_exit(&to_ds->ds_sendstream_lock);

	VERIFY(err != 0 \|\| (dsc.dsc_sent_begin &&
	(dsc.dsc_sent_end \|\| dspp->savedok)));

	kmem_free(drr, sizeof (dmu_replay_record_t));
	kmem_free(dssp, sizeof (dmu_sendstatus_t));
	kmem_free(from_arg, sizeof (*from_arg));
	kmem_free(to_arg, sizeof (*to_arg));
	kmem_free(rlt_arg, sizeof (*rlt_arg));
	kmem_free(smt_arg, sizeof (*smt_arg));
	kmem_free(srt_arg, sizeof (*srt_arg));

	dsl_dataset_long_rele(to_ds, FTAG);
	if (from_rl != NULL) {
	dsl_redaction_list_long_rele(from_rl, FTAG);
	dsl_redaction_list_rele(from_rl, FTAG);
	}
	if (redact_rl != NULL) {
	dsl_redaction_list_long_rele(redact_rl, FTAG);
	dsl_redaction_list_rele(redact_rl, FTAG);
	}

	return (err);
	}

	int
	dmu_send_obj(const char *pool, uint64_t tosnap, uint64_t fromsnap,
	boolean_t embedok, boolean_t large_block_ok, boolean_t compressok,
	boolean_t rawok, boolean_t savedok, int outfd, offset_t *off,
	dmu_send_outparams_t *dsop)
	{
	int err;
	dsl_dataset_t *fromds;
	ds_hold_flags_t dsflags;
	struct dmu_send_params dspp = {0};
	dspp.embedok = embedok;
	dspp.large_block_ok = large_block_ok;
	dspp.compressok = compressok;
	dspp.outfd = outfd;
	dspp.off = off;
	dspp.dso = dsop;
	dspp.tag = FTAG;
	dspp.rawok = rawok;
	dspp.savedok = savedok;

	dsflags = (rawok) ? DS_HOLD_FLAG_NONE : DS_HOLD_FLAG_DECRYPT;
	err = dsl_pool_hold(pool, FTAG, &dspp.dp);
	if (err != 0)
	return (err);

	err = dsl_dataset_hold_obj_flags(dspp.dp, tosnap, dsflags, FTAG,
	&dspp.to_ds);
	if (err != 0) {
	dsl_pool_rele(dspp.dp, FTAG);
	return (err);
	}

	if (fromsnap != 0) {
	err = dsl_dataset_hold_obj_flags(dspp.dp, fromsnap, dsflags,
	FTAG, &fromds);
	if (err != 0) {
	dsl_dataset_rele_flags(dspp.to_ds, dsflags, FTAG);
	dsl_pool_rele(dspp.dp, FTAG);
	return (err);
	}
	dspp.ancestor_zb.zbm_guid = dsl_dataset_phys(fromds)->ds_guid;
	dspp.ancestor_zb.zbm_creation_txg =
	dsl_dataset_phys(fromds)->ds_creation_txg;
	dspp.ancestor_zb.zbm_creation_time =
	dsl_dataset_phys(fromds)->ds_creation_time;

	if (dsl_dataset_is_zapified(fromds)) {
	(void) zap_lookup(dspp.dp->dp_meta_objset,
	fromds->ds_object, DS_FIELD_IVSET_GUID, 8, 1,
	&dspp.ancestor_zb.zbm_ivset_guid);
	}

	/* See dmu_send for the reasons behind this. */
	uint64_t *fromredact;

	if (!dsl_dataset_get_uint64_array_feature(fromds,
	SPA_FEATURE_REDACTED_DATASETS,
	&dspp.numfromredactsnaps,
	&fromredact)) {
	dspp.numfromredactsnaps = NUM_SNAPS_NOT_REDACTED;
	} else if (dspp.numfromredactsnaps > 0) {
	uint64_t size = dspp.numfromredactsnaps *
	sizeof (uint64_t);
	dspp.fromredactsnaps = kmem_zalloc(size, KM_SLEEP);
	bcopy(fromredact, dspp.fromredactsnaps, size);
	}

	boolean_t is_before =
	dsl_dataset_is_before(dspp.to_ds, fromds, 0);
	dspp.is_clone = (dspp.to_ds->ds_dir !=
	fromds->ds_dir);
	dsl_dataset_rele(fromds, FTAG);
	if (!is_before) {
	dsl_pool_rele(dspp.dp, FTAG);
	err = SET_ERROR(EXDEV);
	} else {
	err = dmu_send_impl(&dspp);
	}
	} else {
	dspp.numfromredactsnaps = NUM_SNAPS_NOT_REDACTED;
	err = dmu_send_impl(&dspp);
	}
	dsl_dataset_rele(dspp.to_ds, FTAG);
	return (err);
	}

	int
	dmu_send(const char tosnap, const char fromsnap, boolean_t embedok,
	boolean_t large_block_ok, boolean_t compressok, boolean_t rawok,
	boolean_t savedok, uint64_t resumeobj, uint64_t resumeoff,
	const char redactbook, int outfd, offset_t off,
	dmu_send_outparams_t *dsop)
	{
	int err = 0;
	ds_hold_flags_t dsflags;
	boolean_t owned = B_FALSE;
	dsl_dataset_t *fromds = NULL;
	zfs_bookmark_phys_t book = {0};
	struct dmu_send_params dspp = {0};

	dsflags = (rawok) ? DS_HOLD_FLAG_NONE : DS_HOLD_FLAG_DECRYPT;
	dspp.tosnap = tosnap;
	dspp.embedok = embedok;
	dspp.large_block_ok = large_block_ok;
	dspp.compressok = compressok;
	dspp.outfd = outfd;
	dspp.off = off;
	dspp.dso = dsop;
	dspp.tag = FTAG;
	dspp.resumeobj = resumeobj;
	dspp.resumeoff = resumeoff;
	dspp.rawok = rawok;
	dspp.savedok = savedok;

	if (fromsnap != NULL && strpbrk(fromsnap, "@#") == NULL)
	return (SET_ERROR(EINVAL));

	err = dsl_pool_hold(tosnap, FTAG, &dspp.dp);
	if (err != 0)
	return (err);

	if (strchr(tosnap, '@') == NULL && spa_writeable(dspp.dp->dp_spa)) {
	/*
	* We are sending a filesystem or volume. Ensure
	* that it doesn't change by owning the dataset.
	*/

	if (savedok) {
	/*
	* We are looking for the dataset that represents the
	* partially received send stream. If this stream was
	* received as a new snapshot of an existing dataset,
	* this will be saved in a hidden clone named
	* "<pool>/<dataset>/%recv". Otherwise, the stream
	* will be saved in the live dataset itself. In
	* either case we need to use dsl_dataset_own_force()
	* because the stream is marked as inconsistent,
	* which would normally make it unavailable to be
	* owned.
	*/
	char *name = kmem_asprintf("%s/%s", tosnap,
	recv_clone_name);
	err = dsl_dataset_own_force(dspp.dp, name, dsflags,
	FTAG, &dspp.to_ds);
	if (err == ENOENT) {
	err = dsl_dataset_own_force(dspp.dp, tosnap,
	dsflags, FTAG, &dspp.to_ds);
	}

	if (err == 0) {
	err = zap_lookup(dspp.dp->dp_meta_objset,
	dspp.to_ds->ds_object,
	DS_FIELD_RESUME_TOGUID, 8, 1,
	&dspp.saved_guid);
	}

	if (err == 0) {
	err = zap_lookup(dspp.dp->dp_meta_objset,
	dspp.to_ds->ds_object,
	DS_FIELD_RESUME_TONAME, 1,
	sizeof (dspp.saved_toname),
	dspp.saved_toname);
	}
	if (err != 0)
	dsl_dataset_disown(dspp.to_ds, dsflags, FTAG);

	kmem_strfree(name);
	} else {
	err = dsl_dataset_own(dspp.dp, tosnap, dsflags,
	FTAG, &dspp.to_ds);
	}
	owned = B_TRUE;
	} else {
	err = dsl_dataset_hold_flags(dspp.dp, tosnap, dsflags, FTAG,
	&dspp.to_ds);
	}

	if (err != 0) {
	dsl_pool_rele(dspp.dp, FTAG);
	return (err);
	}

	if (redactbook != NULL) {
	char path[ZFS_MAX_DATASET_NAME_LEN];
	(void) strlcpy(path, tosnap, sizeof (path));
	char *at = strchr(path, '@');
	if (at == NULL) {
	err = EINVAL;
	} else {
	(void) snprintf(at, sizeof (path) - (at - path), "#%s",
	redactbook);
	err = dsl_bookmark_lookup(dspp.dp, path,
	NULL, &book);
	dspp.redactbook = &book;
	}
	}

	if (err != 0) {
	dsl_pool_rele(dspp.dp, FTAG);
	if (owned)
	dsl_dataset_disown(dspp.to_ds, dsflags, FTAG);
	else
	dsl_dataset_rele_flags(dspp.to_ds, dsflags, FTAG);
	return (err);
	}

	if (fromsnap != NULL) {
	zfs_bookmark_phys_t *zb = &dspp.ancestor_zb;
	int fsnamelen;
	if (strpbrk(tosnap, "@#") != NULL)
	fsnamelen = strpbrk(tosnap, "@#") - tosnap;
	else
	fsnamelen = strlen(tosnap);

	/*
	* If the fromsnap is in a different filesystem, then
	* mark the send stream as a clone.
	*/
	if (strncmp(tosnap, fromsnap, fsnamelen) != 0 \|\|
	(fromsnap[fsnamelen] != '@' &&
	fromsnap[fsnamelen] != '#')) {
	dspp.is_clone = B_TRUE;
	}

	if (strchr(fromsnap, '@') != NULL) {
	err = dsl_dataset_hold(dspp.dp, fromsnap, FTAG,
	&fromds);

	if (err != 0) {
	ASSERT3P(fromds, ==, NULL);
	} else {
	/*
	* We need to make a deep copy of the redact
	* snapshots of the from snapshot, because the
	* array will be freed when we evict from_ds.
	*/
	uint64_t *fromredact;
	if (!dsl_dataset_get_uint64_array_feature(
	fromds, SPA_FEATURE_REDACTED_DATASETS,
	&dspp.numfromredactsnaps,
	&fromredact)) {
	dspp.numfromredactsnaps =
	NUM_SNAPS_NOT_REDACTED;
	} else if (dspp.numfromredactsnaps > 0) {
	uint64_t size =
	dspp.numfromredactsnaps *
	sizeof (uint64_t);
	dspp.fromredactsnaps = kmem_zalloc(size,
	KM_SLEEP);
	bcopy(fromredact, dspp.fromredactsnaps,
	size);
	}
	if (!dsl_dataset_is_before(dspp.to_ds, fromds,
	0)) {
	err = SET_ERROR(EXDEV);
	} else {
	zb->zbm_creation_txg =
	dsl_dataset_phys(fromds)->
	ds_creation_txg;
	zb->zbm_creation_time =
	dsl_dataset_phys(fromds)->
	ds_creation_time;
	zb->zbm_guid =
	dsl_dataset_phys(fromds)->ds_guid;
	zb->zbm_redaction_obj = 0;

	if (dsl_dataset_is_zapified(fromds)) {
	(void) zap_lookup(
	dspp.dp->dp_meta_objset,
	fromds->ds_object,
	DS_FIELD_IVSET_GUID, 8, 1,
	&zb->zbm_ivset_guid);
	}
	}
	dsl_dataset_rele(fromds, FTAG);
	}
	} else {
	dspp.numfromredactsnaps = NUM_SNAPS_NOT_REDACTED;
	err = dsl_bookmark_lookup(dspp.dp, fromsnap, dspp.to_ds,
	zb);
	if (err == EXDEV && zb->zbm_redaction_obj != 0 &&
	zb->zbm_guid ==
	dsl_dataset_phys(dspp.to_ds)->ds_guid)
	err = 0;
	}

	if (err == 0) {
	/* dmu_send_impl will call dsl_pool_rele for us. */
	err = dmu_send_impl(&dspp);
	} else {
	dsl_pool_rele(dspp.dp, FTAG);
	}
	} else {
	dspp.numfromredactsnaps = NUM_SNAPS_NOT_REDACTED;
	err = dmu_send_impl(&dspp);
	}
	if (owned)
	dsl_dataset_disown(dspp.to_ds, dsflags, FTAG);
	else
	dsl_dataset_rele_flags(dspp.to_ds, dsflags, FTAG);
	return (err);
	}

	static int
	dmu_adjust_send_estimate_for_indirects(dsl_dataset_t *ds, uint64_t uncompressed,
	uint64_t compressed, boolean_t stream_compressed, uint64_t *sizep)
	{
	int err = 0;
	uint64_t size;
	/*
	* Assume that space (both on-disk and in-stream) is dominated by
	* data. We will adjust for indirect blocks and the copies property,
	* but ignore per-object space used (eg, dnodes and DRR_OBJECT records).
	*/

	uint64_t recordsize;
	uint64_t record_count;
	objset_t *os;
	VERIFY0(dmu_objset_from_ds(ds, &os));

	/* Assume all (uncompressed) blocks are recordsize. */
	if (zfs_override_estimate_recordsize != 0) {
	recordsize = zfs_override_estimate_recordsize;
	} else if (os->os_phys->os_type == DMU_OST_ZVOL) {
	err = dsl_prop_get_int_ds(ds,
	zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), &recordsize);
	} else {
	err = dsl_prop_get_int_ds(ds,
	zfs_prop_to_name(ZFS_PROP_RECORDSIZE), &recordsize);
	}
	if (err != 0)
	return (err);
	record_count = uncompressed / recordsize;

	/*
	* If we're estimating a send size for a compressed stream, use the
	* compressed data size to estimate the stream size. Otherwise, use the
	* uncompressed data size.
	*/
	size = stream_compressed ? compressed : uncompressed;

	/*
	* Subtract out approximate space used by indirect blocks.
	* Assume most space is used by data blocks (non-indirect, non-dnode).
	* Assume no ditto blocks or internal fragmentation.
	*
	* Therefore, space used by indirect blocks is sizeof(blkptr_t) per
	* block.
	*/
	size -= record_count * sizeof (blkptr_t);

	/* Add in the space for the record associated with each block. */
	size += record_count * sizeof (dmu_replay_record_t);

	*sizep = size;

	return (0);
	}

	int
	dmu_send_estimate_fast(dsl_dataset_t origds, dsl_dataset_t fromds,
	zfs_bookmark_phys_t *frombook, boolean_t stream_compressed,
	boolean_t saved, uint64_t *sizep)
	{
	int err;
	dsl_dataset_t *ds = origds;
	uint64_t uncomp, comp;

	ASSERT(dsl_pool_config_held(origds->ds_dir->dd_pool));
	ASSERT(fromds == NULL \|\| frombook == NULL);

	/*
	* If this is a saved send we may actually be sending
	* from the %recv clone used for resuming.
	*/
	if (saved) {
	objset_t *mos = origds->ds_dir->dd_pool->dp_meta_objset;
	uint64_t guid;
	char dsname[ZFS_MAX_DATASET_NAME_LEN + 6];

	dsl_dataset_name(origds, dsname);
	(void) strcat(dsname, "/");
	(void) strcat(dsname, recv_clone_name);

	err = dsl_dataset_hold(origds->ds_dir->dd_pool,
	dsname, FTAG, &ds);
	if (err != ENOENT && err != 0) {
	return (err);
	} else if (err == ENOENT) {
	ds = origds;
	}

	/* check that this dataset has partially received data */
	err = zap_lookup(mos, ds->ds_object,
	DS_FIELD_RESUME_TOGUID, 8, 1, &guid);
	if (err != 0) {
	err = SET_ERROR(err == ENOENT ? EINVAL : err);
	goto out;
	}

	err = zap_lookup(mos, ds->ds_object,
	DS_FIELD_RESUME_TONAME, 1, sizeof (dsname), dsname);
	if (err != 0) {
	err = SET_ERROR(err == ENOENT ? EINVAL : err);
	goto out;
	}
	}

	/* tosnap must be a snapshot or the target of a saved send */
	if (!ds->ds_is_snapshot && ds == origds)
	return (SET_ERROR(EINVAL));

	if (fromds != NULL) {
	uint64_t used;
	if (!fromds->ds_is_snapshot) {
	err = SET_ERROR(EINVAL);
	goto out;
	}

	if (!dsl_dataset_is_before(ds, fromds, 0)) {
	err = SET_ERROR(EXDEV);
	goto out;
	}

	err = dsl_dataset_space_written(fromds, ds, &used, &comp,
	&uncomp);
	if (err != 0)
	goto out;
	} else if (frombook != NULL) {
	uint64_t used;
	err = dsl_dataset_space_written_bookmark(frombook, ds, &used,
	&comp, &uncomp);
	if (err != 0)
	goto out;
	} else {
	uncomp = dsl_dataset_phys(ds)->ds_uncompressed_bytes;
	comp = dsl_dataset_phys(ds)->ds_compressed_bytes;
	}

	err = dmu_adjust_send_estimate_for_indirects(ds, uncomp, comp,
	stream_compressed, sizep);
	/*
	* Add the size of the BEGIN and END records to the estimate.
	*/
	sizep += 2 sizeof (dmu_replay_record_t);

	out:
	if (ds != origds)
	dsl_dataset_rele(ds, FTAG);
	return (err);
	}

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM(zfs_send, zfs_send_, corrupt_data, INT, ZMOD_RW,
	"Allow sending corrupt data");

	ZFS_MODULE_PARAM(zfs_send, zfs_send_, queue_length, INT, ZMOD_RW,
	"Maximum send queue length");

	ZFS_MODULE_PARAM(zfs_send, zfs_send_, unmodified_spill_blocks, INT, ZMOD_RW,
	"Send unmodified spill blocks");

	ZFS_MODULE_PARAM(zfs_send, zfs_send_, no_prefetch_queue_length, INT, ZMOD_RW,
	"Maximum send queue length for non-prefetch queues");

	ZFS_MODULE_PARAM(zfs_send, zfs_send_, queue_ff, INT, ZMOD_RW,
	"Send queue fill fraction");

	ZFS_MODULE_PARAM(zfs_send, zfs_send_, no_prefetch_queue_ff, INT, ZMOD_RW,
	"Send queue fill fraction for non-prefetch queues");

	ZFS_MODULE_PARAM(zfs_send, zfs_, override_estimate_recordsize, INT, ZMOD_RW,
	"Override block size estimate with fixed size");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/zfs/fm.c b/sys/contrib/openzfs/module/zfs/fm.c
	index b8a1c7c8a5ca..cdf0fcf71556 100644
	--- a/sys/contrib/openzfs/module/zfs/fm.c
	+++ b/sys/contrib/openzfs/module/zfs/fm.c
	@@ -1,1372 +1,1374 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2004, 2010, Oracle and/or its affiliates. All rights reserved.
	*/

	/*
	* Fault Management Architecture (FMA) Resource and Protocol Support
	*
	* The routines contained herein provide services to support kernel subsystems
	* in publishing fault management telemetry (see PSARC 2002/412 and 2003/089).
	*
	* Name-Value Pair Lists
	*
	* The embodiment of an FMA protocol element (event, fmri or authority) is a
	* name-value pair list (nvlist_t). FMA-specific nvlist constructor and
	* destructor functions, fm_nvlist_create() and fm_nvlist_destroy(), are used
	* to create an nvpair list using custom allocators. Callers may choose to
	* allocate either from the kernel memory allocator, or from a preallocated
	* buffer, useful in constrained contexts like high-level interrupt routines.
	*
	* Protocol Event and FMRI Construction
	*
	* Convenience routines are provided to construct nvlist events according to
	* the FMA Event Protocol and Naming Schema specification for ereports and
	* FMRIs for the dev, cpu, hc, mem, legacy hc and de schemes.
	*
	* ENA Manipulation
	*
	* Routines to generate ENA formats 0, 1 and 2 are available as well as
	* routines to increment formats 1 and 2. Individual fields within the
	* ENA are extractable via fm_ena_time_get(), fm_ena_id_get(),
	* fm_ena_format_get() and fm_ena_gen_get().
	*/

	#include <sys/types.h>
	#include <sys/time.h>
	#include <sys/list.h>
	#include <sys/nvpair.h>
	#include <sys/cmn_err.h>
	#include <sys/sysmacros.h>
	#include <sys/sunddi.h>
	#include <sys/systeminfo.h>
	#include <sys/fm/util.h>
	#include <sys/fm/protocol.h>
	#include <sys/kstat.h>
	#include <sys/zfs_context.h>
	#ifdef _KERNEL
	#include <sys/atomic.h>
	#include <sys/condvar.h>
	#include <sys/zfs_ioctl.h>

	int zfs_zevent_len_max = 512;

	static int zevent_len_cur = 0;
	static int zevent_waiters = 0;
	static int zevent_flags = 0;

	/* Num events rate limited since the last time zfs_zevent_next() was called */
	static uint64_t ratelimit_dropped = 0;

	/*
	* The EID (Event IDentifier) is used to uniquely tag a zevent when it is
	* posted. The posted EIDs are monotonically increasing but not persistent.
	* They will be reset to the initial value (1) each time the kernel module is
	* loaded.
	*/
	static uint64_t zevent_eid = 0;

	static kmutex_t zevent_lock;
	static list_t zevent_list;
	static kcondvar_t zevent_cv;
	#endif /* _KERNEL */


	/*
	* Common fault management kstats to record event generation failures
	*/

	struct erpt_kstat {
	kstat_named_t erpt_dropped; /* num erpts dropped on post */
	kstat_named_t erpt_set_failed; /* num erpt set failures */
	kstat_named_t fmri_set_failed; /* num fmri set failures */
	kstat_named_t payload_set_failed; /* num payload set failures */
	kstat_named_t erpt_duplicates; /* num duplicate erpts */
	};

	static struct erpt_kstat erpt_kstat_data = {
	{ "erpt-dropped", KSTAT_DATA_UINT64 },
	{ "erpt-set-failed", KSTAT_DATA_UINT64 },
	{ "fmri-set-failed", KSTAT_DATA_UINT64 },
	{ "payload-set-failed", KSTAT_DATA_UINT64 },
	{ "erpt-duplicates", KSTAT_DATA_UINT64 }
	};

	kstat_t *fm_ksp;

	#ifdef _KERNEL

	static zevent_t *
	zfs_zevent_alloc(void)
	{
	zevent_t *ev;

	ev = kmem_zalloc(sizeof (zevent_t), KM_SLEEP);

	list_create(&ev->ev_ze_list, sizeof (zfs_zevent_t),
	offsetof(zfs_zevent_t, ze_node));
	list_link_init(&ev->ev_node);

	return (ev);
	}

	static void
	zfs_zevent_free(zevent_t *ev)
	{
	/* Run provided cleanup callback */
	ev->ev_cb(ev->ev_nvl, ev->ev_detector);

	list_destroy(&ev->ev_ze_list);
	kmem_free(ev, sizeof (zevent_t));
	}

	static void
	zfs_zevent_drain(zevent_t *ev)
	{
	zfs_zevent_t *ze;

	ASSERT(MUTEX_HELD(&zevent_lock));
	list_remove(&zevent_list, ev);

	/* Remove references to this event in all private file data */
	while ((ze = list_head(&ev->ev_ze_list)) != NULL) {
	list_remove(&ev->ev_ze_list, ze);
	ze->ze_zevent = NULL;
	ze->ze_dropped++;
	}

	zfs_zevent_free(ev);
	}

	void
	zfs_zevent_drain_all(int *count)
	{
	zevent_t *ev;

	mutex_enter(&zevent_lock);
	while ((ev = list_head(&zevent_list)) != NULL)
	zfs_zevent_drain(ev);

	*count = zevent_len_cur;
	zevent_len_cur = 0;
	mutex_exit(&zevent_lock);
	}

	/*
	* New zevents are inserted at the head. If the maximum queue
	* length is exceeded a zevent will be drained from the tail.
	* As part of this any user space processes which currently have
	* a reference to this zevent_t in their private data will have
	* this reference set to NULL.
	*/
	static void
	zfs_zevent_insert(zevent_t *ev)
	{
	ASSERT(MUTEX_HELD(&zevent_lock));
	list_insert_head(&zevent_list, ev);

	if (zevent_len_cur >= zfs_zevent_len_max)
	zfs_zevent_drain(list_tail(&zevent_list));
	else
	zevent_len_cur++;
	}

	/*
	* Post a zevent. The cb will be called when nvl and detector are no longer
	* needed, i.e.:
	* - An error happened and a zevent can't be posted. In this case, cb is called
	* before zfs_zevent_post() returns.
	* - The event is being drained and freed.
	*/
	int
	zfs_zevent_post(nvlist_t nvl, nvlist_t detector, zevent_cb_t *cb)
	{
	inode_timespec_t tv;
	int64_t tv_array[2];
	uint64_t eid;
	size_t nvl_size = 0;
	zevent_t *ev;
	int error;

	ASSERT(cb != NULL);

	gethrestime(&tv);
	tv_array[0] = tv.tv_sec;
	tv_array[1] = tv.tv_nsec;

	error = nvlist_add_int64_array(nvl, FM_EREPORT_TIME, tv_array, 2);
	if (error) {
	atomic_inc_64(&erpt_kstat_data.erpt_set_failed.value.ui64);
	goto out;
	}

	eid = atomic_inc_64_nv(&zevent_eid);
	error = nvlist_add_uint64(nvl, FM_EREPORT_EID, eid);
	if (error) {
	atomic_inc_64(&erpt_kstat_data.erpt_set_failed.value.ui64);
	goto out;
	}

	error = nvlist_size(nvl, &nvl_size, NV_ENCODE_NATIVE);
	if (error) {
	atomic_inc_64(&erpt_kstat_data.erpt_dropped.value.ui64);
	goto out;
	}

	if (nvl_size > ERPT_DATA_SZ \|\| nvl_size == 0) {
	atomic_inc_64(&erpt_kstat_data.erpt_dropped.value.ui64);
	error = EOVERFLOW;
	goto out;
	}

	ev = zfs_zevent_alloc();
	if (ev == NULL) {
	atomic_inc_64(&erpt_kstat_data.erpt_dropped.value.ui64);
	error = ENOMEM;
	goto out;
	}

	ev->ev_nvl = nvl;
	ev->ev_detector = detector;
	ev->ev_cb = cb;
	ev->ev_eid = eid;

	mutex_enter(&zevent_lock);
	zfs_zevent_insert(ev);
	cv_broadcast(&zevent_cv);
	mutex_exit(&zevent_lock);

	out:
	if (error)
	cb(nvl, detector);

	return (error);
	}

	void
	zfs_zevent_track_duplicate(void)
	{
	atomic_inc_64(&erpt_kstat_data.erpt_duplicates.value.ui64);
	}

	static int
	zfs_zevent_minor_to_state(minor_t minor, zfs_zevent_t **ze)
	{
	*ze = zfsdev_get_state(minor, ZST_ZEVENT);
	if (*ze == NULL)
	return (SET_ERROR(EBADF));

	return (0);
	}

	zfs_file_t *
	zfs_zevent_fd_hold(int fd, minor_t minorp, zfs_zevent_t *ze)
	{
	zfs_file_t *fp = zfs_file_get(fd);
	if (fp == NULL)
	return (NULL);

	int error = zfsdev_getminor(fp, minorp);
	if (error == 0)
	error = zfs_zevent_minor_to_state(*minorp, ze);

	if (error) {
	zfs_zevent_fd_rele(fp);
	fp = NULL;
	}

	return (fp);
	}

	void
	zfs_zevent_fd_rele(zfs_file_t *fp)
	{
	zfs_file_put(fp);
	}

	/*
	* Get the next zevent in the stream and place a copy in 'event'. This
	* may fail with ENOMEM if the encoded nvlist size exceeds the passed
	* 'event_size'. In this case the stream pointer is not advanced and
	* and 'event_size' is set to the minimum required buffer size.
	*/
	int
	zfs_zevent_next(zfs_zevent_t ze, nvlist_t event, uint64_t event_size,
	uint64_t *dropped)
	{
	zevent_t *ev;
	size_t size;
	int error = 0;

	mutex_enter(&zevent_lock);
	if (ze->ze_zevent == NULL) {
	/* New stream start at the beginning/tail */
	ev = list_tail(&zevent_list);
	if (ev == NULL) {
	error = ENOENT;
	goto out;
	}
	} else {
	/*
	* Existing stream continue with the next element and remove
	* ourselves from the wait queue for the previous element
	*/
	ev = list_prev(&zevent_list, ze->ze_zevent);
	if (ev == NULL) {
	error = ENOENT;
	goto out;
	}
	}

	VERIFY(nvlist_size(ev->ev_nvl, &size, NV_ENCODE_NATIVE) == 0);
	if (size > *event_size) {
	*event_size = size;
	error = ENOMEM;
	goto out;
	}

	if (ze->ze_zevent)
	list_remove(&ze->ze_zevent->ev_ze_list, ze);

	ze->ze_zevent = ev;
	list_insert_head(&ev->ev_ze_list, ze);
	(void) nvlist_dup(ev->ev_nvl, event, KM_SLEEP);
	*dropped = ze->ze_dropped;

	#ifdef _KERNEL
	/* Include events dropped due to rate limiting */
	*dropped += atomic_swap_64(&ratelimit_dropped, 0);
	#endif
	ze->ze_dropped = 0;
	out:
	mutex_exit(&zevent_lock);

	return (error);
	}

	/*
	* Wait in an interruptible state for any new events.
	*/
	int
	zfs_zevent_wait(zfs_zevent_t *ze)
	{
	int error = EAGAIN;

	mutex_enter(&zevent_lock);
	zevent_waiters++;

	while (error == EAGAIN) {
	if (zevent_flags & ZEVENT_SHUTDOWN) {
	error = SET_ERROR(ESHUTDOWN);
	break;
	}

	error = cv_wait_sig(&zevent_cv, &zevent_lock);
	if (signal_pending(current)) {
	error = SET_ERROR(EINTR);
	break;
	} else if (!list_is_empty(&zevent_list)) {
	error = 0;
	continue;
	} else {
	error = EAGAIN;
	}
	}

	zevent_waiters--;
	mutex_exit(&zevent_lock);

	return (error);
	}

	/*
	* The caller may seek to a specific EID by passing that EID. If the EID
	* is still available in the posted list of events the cursor is positioned
	* there. Otherwise ENOENT is returned and the cursor is not moved.
	*
	* There are two reserved EIDs which may be passed and will never fail.
	* ZEVENT_SEEK_START positions the cursor at the start of the list, and
	* ZEVENT_SEEK_END positions the cursor at the end of the list.
	*/
	int
	zfs_zevent_seek(zfs_zevent_t *ze, uint64_t eid)
	{
	zevent_t *ev;
	int error = 0;

	mutex_enter(&zevent_lock);

	if (eid == ZEVENT_SEEK_START) {
	if (ze->ze_zevent)
	list_remove(&ze->ze_zevent->ev_ze_list, ze);

	ze->ze_zevent = NULL;
	goto out;
	}

	if (eid == ZEVENT_SEEK_END) {
	if (ze->ze_zevent)
	list_remove(&ze->ze_zevent->ev_ze_list, ze);

	ev = list_head(&zevent_list);
	if (ev) {
	ze->ze_zevent = ev;
	list_insert_head(&ev->ev_ze_list, ze);
	} else {
	ze->ze_zevent = NULL;
	}

	goto out;
	}

	for (ev = list_tail(&zevent_list); ev != NULL;
	ev = list_prev(&zevent_list, ev)) {
	if (ev->ev_eid == eid) {
	if (ze->ze_zevent)
	list_remove(&ze->ze_zevent->ev_ze_list, ze);

	ze->ze_zevent = ev;
	list_insert_head(&ev->ev_ze_list, ze);
	break;
	}
	}

	if (ev == NULL)
	error = ENOENT;

	out:
	mutex_exit(&zevent_lock);

	return (error);
	}

	void
	zfs_zevent_init(zfs_zevent_t **zep)
	{
	zfs_zevent_t *ze;

	ze = *zep = kmem_zalloc(sizeof (zfs_zevent_t), KM_SLEEP);
	list_link_init(&ze->ze_node);
	}

	void
	zfs_zevent_destroy(zfs_zevent_t *ze)
	{
	mutex_enter(&zevent_lock);
	if (ze->ze_zevent)
	list_remove(&ze->ze_zevent->ev_ze_list, ze);
	mutex_exit(&zevent_lock);

	kmem_free(ze, sizeof (zfs_zevent_t));
	}
	#endif /* _KERNEL */

	/*
	* Wrappers for FM nvlist allocators
	*/
	/* ARGSUSED */
	static void *
	i_fm_alloc(nv_alloc_t *nva, size_t size)
	{
	return (kmem_zalloc(size, KM_SLEEP));
	}

	/* ARGSUSED */
	static void
	i_fm_free(nv_alloc_t nva, void buf, size_t size)
	{
	kmem_free(buf, size);
	}

	const nv_alloc_ops_t fm_mem_alloc_ops = {
	.nv_ao_init = NULL,
	.nv_ao_fini = NULL,
	.nv_ao_alloc = i_fm_alloc,
	.nv_ao_free = i_fm_free,
	.nv_ao_reset = NULL
	};

	/*
	* Create and initialize a new nv_alloc_t for a fixed buffer, buf. A pointer
	* to the newly allocated nv_alloc_t structure is returned upon success or NULL
	* is returned to indicate that the nv_alloc structure could not be created.
	*/
	nv_alloc_t *
	fm_nva_xcreate(char *buf, size_t bufsz)
	{
	nv_alloc_t *nvhdl = kmem_zalloc(sizeof (nv_alloc_t), KM_SLEEP);

	if (bufsz == 0 \|\| nv_alloc_init(nvhdl, nv_fixed_ops, buf, bufsz) != 0) {
	kmem_free(nvhdl, sizeof (nv_alloc_t));
	return (NULL);
	}

	return (nvhdl);
	}

	/*
	* Destroy a previously allocated nv_alloc structure. The fixed buffer
	* associated with nva must be freed by the caller.
	*/
	void
	fm_nva_xdestroy(nv_alloc_t *nva)
	{
	nv_alloc_fini(nva);
	kmem_free(nva, sizeof (nv_alloc_t));
	}

	/*
	* Create a new nv list. A pointer to a new nv list structure is returned
	* upon success or NULL is returned to indicate that the structure could
	* not be created. The newly created nv list is created and managed by the
	* operations installed in nva. If nva is NULL, the default FMA nva
	* operations are installed and used.
	*
	* When called from the kernel and nva == NULL, this function must be called
	* from passive kernel context with no locks held that can prevent a
	* sleeping memory allocation from occurring. Otherwise, this function may
	* be called from other kernel contexts as long a valid nva created via
	* fm_nva_create() is supplied.
	*/
	nvlist_t *
	fm_nvlist_create(nv_alloc_t *nva)
	{
	int hdl_alloced = 0;
	nvlist_t *nvl;
	nv_alloc_t *nvhdl;

	if (nva == NULL) {
	nvhdl = kmem_zalloc(sizeof (nv_alloc_t), KM_SLEEP);

	if (nv_alloc_init(nvhdl, &fm_mem_alloc_ops, NULL, 0) != 0) {
	kmem_free(nvhdl, sizeof (nv_alloc_t));
	return (NULL);
	}
	hdl_alloced = 1;
	} else {
	nvhdl = nva;
	}

	if (nvlist_xalloc(&nvl, NV_UNIQUE_NAME, nvhdl) != 0) {
	if (hdl_alloced) {
	nv_alloc_fini(nvhdl);
	kmem_free(nvhdl, sizeof (nv_alloc_t));
	}
	return (NULL);
	}

	return (nvl);
	}

	/*
	* Destroy a previously allocated nvlist structure. flag indicates whether
	* or not the associated nva structure should be freed (FM_NVA_FREE) or
	* retained (FM_NVA_RETAIN). Retaining the nv alloc structure allows
	* it to be re-used for future nvlist creation operations.
	*/
	void
	fm_nvlist_destroy(nvlist_t *nvl, int flag)
	{
	nv_alloc_t *nva = nvlist_lookup_nv_alloc(nvl);

	nvlist_free(nvl);

	if (nva != NULL) {
	if (flag == FM_NVA_FREE)
	fm_nva_xdestroy(nva);
	}
	}

	int
	i_fm_payload_set(nvlist_t payload, const char name, va_list ap)
	{
	int nelem, ret = 0;
	data_type_t type;

	while (ret == 0 && name != NULL) {
	type = va_arg(ap, data_type_t);
	switch (type) {
	case DATA_TYPE_BYTE:
	ret = nvlist_add_byte(payload, name,
	va_arg(ap, uint_t));
	break;
	case DATA_TYPE_BYTE_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_byte_array(payload, name,
	va_arg(ap, uchar_t *), nelem);
	break;
	case DATA_TYPE_BOOLEAN_VALUE:
	ret = nvlist_add_boolean_value(payload, name,
	va_arg(ap, boolean_t));
	break;
	case DATA_TYPE_BOOLEAN_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_boolean_array(payload, name,
	va_arg(ap, boolean_t *), nelem);
	break;
	case DATA_TYPE_INT8:
	ret = nvlist_add_int8(payload, name,
	va_arg(ap, int));
	break;
	case DATA_TYPE_INT8_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_int8_array(payload, name,
	va_arg(ap, int8_t *), nelem);
	break;
	case DATA_TYPE_UINT8:
	ret = nvlist_add_uint8(payload, name,
	va_arg(ap, uint_t));
	break;
	case DATA_TYPE_UINT8_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_uint8_array(payload, name,
	va_arg(ap, uint8_t *), nelem);
	break;
	case DATA_TYPE_INT16:
	ret = nvlist_add_int16(payload, name,
	va_arg(ap, int));
	break;
	case DATA_TYPE_INT16_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_int16_array(payload, name,
	va_arg(ap, int16_t *), nelem);
	break;
	case DATA_TYPE_UINT16:
	ret = nvlist_add_uint16(payload, name,
	va_arg(ap, uint_t));
	break;
	case DATA_TYPE_UINT16_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_uint16_array(payload, name,
	va_arg(ap, uint16_t *), nelem);
	break;
	case DATA_TYPE_INT32:
	ret = nvlist_add_int32(payload, name,
	va_arg(ap, int32_t));
	break;
	case DATA_TYPE_INT32_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_int32_array(payload, name,
	va_arg(ap, int32_t *), nelem);
	break;
	case DATA_TYPE_UINT32:
	ret = nvlist_add_uint32(payload, name,
	va_arg(ap, uint32_t));
	break;
	case DATA_TYPE_UINT32_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_uint32_array(payload, name,
	va_arg(ap, uint32_t *), nelem);
	break;
	case DATA_TYPE_INT64:
	ret = nvlist_add_int64(payload, name,
	va_arg(ap, int64_t));
	break;
	case DATA_TYPE_INT64_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_int64_array(payload, name,
	va_arg(ap, int64_t *), nelem);
	break;
	case DATA_TYPE_UINT64:
	ret = nvlist_add_uint64(payload, name,
	va_arg(ap, uint64_t));
	break;
	case DATA_TYPE_UINT64_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_uint64_array(payload, name,
	va_arg(ap, uint64_t *), nelem);
	break;
	case DATA_TYPE_STRING:
	ret = nvlist_add_string(payload, name,
	va_arg(ap, char *));
	break;
	case DATA_TYPE_STRING_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_string_array(payload, name,
	- va_arg(ap, char **), nelem);
	+ va_arg(ap, const char **), nelem);
	break;
	case DATA_TYPE_NVLIST:
	ret = nvlist_add_nvlist(payload, name,
	va_arg(ap, nvlist_t *));
	break;
	case DATA_TYPE_NVLIST_ARRAY:
	nelem = va_arg(ap, int);
	ret = nvlist_add_nvlist_array(payload, name,
	- va_arg(ap, nvlist_t **), nelem);
	+ va_arg(ap, const nvlist_t **), nelem);
	break;
	default:
	ret = EINVAL;
	}

	name = va_arg(ap, char *);
	}
	return (ret);
	}

	void
	fm_payload_set(nvlist_t *payload, ...)
	{
	int ret;
	const char *name;
	va_list ap;

	va_start(ap, payload);
	name = va_arg(ap, char *);
	ret = i_fm_payload_set(payload, name, ap);
	va_end(ap);

	if (ret)
	atomic_inc_64(&erpt_kstat_data.payload_set_failed.value.ui64);
	}

	/*
	* Set-up and validate the members of an ereport event according to:
	*
	* Member name Type Value
	* ====================================================
	* class string ereport
	* version uint8_t 0
	* ena uint64_t <ena>
	* detector nvlist_t <detector>
	* ereport-payload nvlist_t <var args>
	*
	* We don't actually add a 'version' member to the payload. Really,
	* the version quoted to us by our caller is that of the category 1
	* "ereport" event class (and we require FM_EREPORT_VERS0) but
	* the payload version of the actual leaf class event under construction
	* may be something else. Callers should supply a version in the varargs,
	* or (better) we could take two version arguments - one for the
	* ereport category 1 classification (expect FM_EREPORT_VERS0) and one
	* for the leaf class.
	*/
	void
	fm_ereport_set(nvlist_t ereport, int version, const char erpt_class,
	uint64_t ena, const nvlist_t *detector, ...)
	{
	char ereport_class[FM_MAX_CLASS];
	const char *name;
	va_list ap;
	int ret;

	if (version != FM_EREPORT_VERS0) {
	atomic_inc_64(&erpt_kstat_data.erpt_set_failed.value.ui64);
	return;
	}

	(void) snprintf(ereport_class, FM_MAX_CLASS, "%s.%s",
	FM_EREPORT_CLASS, erpt_class);
	if (nvlist_add_string(ereport, FM_CLASS, ereport_class) != 0) {
	atomic_inc_64(&erpt_kstat_data.erpt_set_failed.value.ui64);
	return;
	}

	if (nvlist_add_uint64(ereport, FM_EREPORT_ENA, ena)) {
	atomic_inc_64(&erpt_kstat_data.erpt_set_failed.value.ui64);
	}

	if (nvlist_add_nvlist(ereport, FM_EREPORT_DETECTOR,
	(nvlist_t *)detector) != 0) {
	atomic_inc_64(&erpt_kstat_data.erpt_set_failed.value.ui64);
	}

	va_start(ap, detector);
	name = va_arg(ap, const char *);
	ret = i_fm_payload_set(ereport, name, ap);
	va_end(ap);

	if (ret)
	atomic_inc_64(&erpt_kstat_data.erpt_set_failed.value.ui64);
	}

	/*
	* Set-up and validate the members of an hc fmri according to;
	*
	* Member name Type Value
	* ===================================================
	* version uint8_t 0
	* auth nvlist_t <auth>
	* hc-name string <name>
	* hc-id string <id>
	*
	* Note that auth and hc-id are optional members.
	*/

	#define HC_MAXPAIRS 20
	#define HC_MAXNAMELEN 50

	static int
	fm_fmri_hc_set_common(nvlist_t fmri, int version, const nvlist_t auth)
	{
	if (version != FM_HC_SCHEME_VERSION) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return (0);
	}

	if (nvlist_add_uint8(fmri, FM_VERSION, version) != 0 \|\|
	nvlist_add_string(fmri, FM_FMRI_SCHEME, FM_FMRI_SCHEME_HC) != 0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return (0);
	}

	if (auth != NULL && nvlist_add_nvlist(fmri, FM_FMRI_AUTHORITY,
	(nvlist_t *)auth) != 0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return (0);
	}

	return (1);
	}

	void
	fm_fmri_hc_set(nvlist_t fmri, int version, const nvlist_t auth,
	nvlist_t *snvl, int npairs, ...)
	{
	nv_alloc_t *nva = nvlist_lookup_nv_alloc(fmri);
	nvlist_t *pairs[HC_MAXPAIRS];
	va_list ap;
	int i;

	if (!fm_fmri_hc_set_common(fmri, version, auth))
	return;

	npairs = MIN(npairs, HC_MAXPAIRS);

	va_start(ap, npairs);
	for (i = 0; i < npairs; i++) {
	const char name = va_arg(ap, const char );
	uint32_t id = va_arg(ap, uint32_t);
	char idstr[11];

	(void) snprintf(idstr, sizeof (idstr), "%u", id);

	pairs[i] = fm_nvlist_create(nva);
	if (nvlist_add_string(pairs[i], FM_FMRI_HC_NAME, name) != 0 \|\|
	nvlist_add_string(pairs[i], FM_FMRI_HC_ID, idstr) != 0) {
	atomic_inc_64(
	&erpt_kstat_data.fmri_set_failed.value.ui64);
	}
	}
	va_end(ap);

	- if (nvlist_add_nvlist_array(fmri, FM_FMRI_HC_LIST, pairs, npairs) != 0)
	+ if (nvlist_add_nvlist_array(fmri, FM_FMRI_HC_LIST,
	+ (const nvlist_t **)pairs, npairs) != 0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	+ }

	for (i = 0; i < npairs; i++)
	fm_nvlist_destroy(pairs[i], FM_NVA_RETAIN);

	if (snvl != NULL) {
	if (nvlist_add_nvlist(fmri, FM_FMRI_HC_SPECIFIC, snvl) != 0) {
	atomic_inc_64(
	&erpt_kstat_data.fmri_set_failed.value.ui64);
	}
	}
	}

	void
	fm_fmri_hc_create(nvlist_t fmri, int version, const nvlist_t auth,
	nvlist_t snvl, nvlist_t bboard, int npairs, ...)
	{
	nv_alloc_t *nva = nvlist_lookup_nv_alloc(fmri);
	nvlist_t *pairs[HC_MAXPAIRS];
	nvlist_t **hcl;
	uint_t n;
	int i, j;
	va_list ap;
	char hcname, hcid;

	if (!fm_fmri_hc_set_common(fmri, version, auth))
	return;

	/*
	* copy the bboard nvpairs to the pairs array
	*/
	if (nvlist_lookup_nvlist_array(bboard, FM_FMRI_HC_LIST, &hcl, &n)
	!= 0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}

	for (i = 0; i < n; i++) {
	if (nvlist_lookup_string(hcl[i], FM_FMRI_HC_NAME,
	&hcname) != 0) {
	atomic_inc_64(
	&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}
	if (nvlist_lookup_string(hcl[i], FM_FMRI_HC_ID, &hcid) != 0) {
	atomic_inc_64(
	&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}

	pairs[i] = fm_nvlist_create(nva);
	if (nvlist_add_string(pairs[i], FM_FMRI_HC_NAME, hcname) != 0 \|\|
	nvlist_add_string(pairs[i], FM_FMRI_HC_ID, hcid) != 0) {
	for (j = 0; j <= i; j++) {
	if (pairs[j] != NULL)
	fm_nvlist_destroy(pairs[j],
	FM_NVA_RETAIN);
	}
	atomic_inc_64(
	&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}
	}

	/*
	* create the pairs from passed in pairs
	*/
	npairs = MIN(npairs, HC_MAXPAIRS);

	va_start(ap, npairs);
	for (i = n; i < npairs + n; i++) {
	const char name = va_arg(ap, const char );
	uint32_t id = va_arg(ap, uint32_t);
	char idstr[11];
	(void) snprintf(idstr, sizeof (idstr), "%u", id);
	pairs[i] = fm_nvlist_create(nva);
	if (nvlist_add_string(pairs[i], FM_FMRI_HC_NAME, name) != 0 \|\|
	nvlist_add_string(pairs[i], FM_FMRI_HC_ID, idstr) != 0) {
	for (j = 0; j <= i; j++) {
	if (pairs[j] != NULL)
	fm_nvlist_destroy(pairs[j],
	FM_NVA_RETAIN);
	}
	atomic_inc_64(
	&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}
	}
	va_end(ap);

	/*
	* Create the fmri hc list
	*/
	- if (nvlist_add_nvlist_array(fmri, FM_FMRI_HC_LIST, pairs,
	- npairs + n) != 0) {
	+ if (nvlist_add_nvlist_array(fmri, FM_FMRI_HC_LIST,
	+ (const nvlist_t **)pairs, npairs + n) != 0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}

	for (i = 0; i < npairs + n; i++) {
	fm_nvlist_destroy(pairs[i], FM_NVA_RETAIN);
	}

	if (snvl != NULL) {
	if (nvlist_add_nvlist(fmri, FM_FMRI_HC_SPECIFIC, snvl) != 0) {
	atomic_inc_64(
	&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}
	}
	}

	/*
	* Set-up and validate the members of an dev fmri according to:
	*
	* Member name Type Value
	* ====================================================
	* version uint8_t 0
	* auth nvlist_t <auth>
	* devpath string <devpath>
	* [devid] string <devid>
	* [target-port-l0id] string <target-port-lun0-id>
	*
	* Note that auth and devid are optional members.
	*/
	void
	fm_fmri_dev_set(nvlist_t fmri_dev, int version, const nvlist_t auth,
	const char devpath, const char devid, const char *tpl0)
	{
	int err = 0;

	if (version != DEV_SCHEME_VERSION0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}

	err \|= nvlist_add_uint8(fmri_dev, FM_VERSION, version);
	err \|= nvlist_add_string(fmri_dev, FM_FMRI_SCHEME, FM_FMRI_SCHEME_DEV);

	if (auth != NULL) {
	err \|= nvlist_add_nvlist(fmri_dev, FM_FMRI_AUTHORITY,
	(nvlist_t *)auth);
	}

	err \|= nvlist_add_string(fmri_dev, FM_FMRI_DEV_PATH, devpath);

	if (devid != NULL)
	err \|= nvlist_add_string(fmri_dev, FM_FMRI_DEV_ID, devid);

	if (tpl0 != NULL)
	err \|= nvlist_add_string(fmri_dev, FM_FMRI_DEV_TGTPTLUN0, tpl0);

	if (err)
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);

	}

	/*
	* Set-up and validate the members of an cpu fmri according to:
	*
	* Member name Type Value
	* ====================================================
	* version uint8_t 0
	* auth nvlist_t <auth>
	* cpuid uint32_t <cpu_id>
	* cpumask uint8_t <cpu_mask>
	* serial uint64_t <serial_id>
	*
	* Note that auth, cpumask, serial are optional members.
	*
	*/
	void
	fm_fmri_cpu_set(nvlist_t fmri_cpu, int version, const nvlist_t auth,
	uint32_t cpu_id, uint8_t cpu_maskp, const char serial_idp)
	{
	uint64_t *failedp = &erpt_kstat_data.fmri_set_failed.value.ui64;

	if (version < CPU_SCHEME_VERSION1) {
	atomic_inc_64(failedp);
	return;
	}

	if (nvlist_add_uint8(fmri_cpu, FM_VERSION, version) != 0) {
	atomic_inc_64(failedp);
	return;
	}

	if (nvlist_add_string(fmri_cpu, FM_FMRI_SCHEME,
	FM_FMRI_SCHEME_CPU) != 0) {
	atomic_inc_64(failedp);
	return;
	}

	if (auth != NULL && nvlist_add_nvlist(fmri_cpu, FM_FMRI_AUTHORITY,
	(nvlist_t *)auth) != 0)
	atomic_inc_64(failedp);

	if (nvlist_add_uint32(fmri_cpu, FM_FMRI_CPU_ID, cpu_id) != 0)
	atomic_inc_64(failedp);

	if (cpu_maskp != NULL && nvlist_add_uint8(fmri_cpu, FM_FMRI_CPU_MASK,
	*cpu_maskp) != 0)
	atomic_inc_64(failedp);

	if (serial_idp == NULL \|\| nvlist_add_string(fmri_cpu,
	FM_FMRI_CPU_SERIAL_ID, (char *)serial_idp) != 0)
	atomic_inc_64(failedp);
	}

	/*
	* Set-up and validate the members of a mem according to:
	*
	* Member name Type Value
	* ====================================================
	* version uint8_t 0
	* auth nvlist_t <auth> [optional]
	* unum string <unum>
	* serial string <serial> [optional*]
	* offset uint64_t <offset> [optional]
	*
	* * serial is required if offset is present
	*/
	void
	fm_fmri_mem_set(nvlist_t fmri, int version, const nvlist_t auth,
	const char unum, const char serial, uint64_t offset)
	{
	if (version != MEM_SCHEME_VERSION0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}

	if (!serial && (offset != (uint64_t)-1)) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}

	if (nvlist_add_uint8(fmri, FM_VERSION, version) != 0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}

	if (nvlist_add_string(fmri, FM_FMRI_SCHEME, FM_FMRI_SCHEME_MEM) != 0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}

	if (auth != NULL) {
	if (nvlist_add_nvlist(fmri, FM_FMRI_AUTHORITY,
	(nvlist_t *)auth) != 0) {
	atomic_inc_64(
	&erpt_kstat_data.fmri_set_failed.value.ui64);
	}
	}

	if (nvlist_add_string(fmri, FM_FMRI_MEM_UNUM, unum) != 0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	}

	if (serial != NULL) {
	if (nvlist_add_string_array(fmri, FM_FMRI_MEM_SERIAL_ID,
	- (char **)&serial, 1) != 0) {
	+ (const char **)&serial, 1) != 0) {
	atomic_inc_64(
	&erpt_kstat_data.fmri_set_failed.value.ui64);
	}
	if (offset != (uint64_t)-1 && nvlist_add_uint64(fmri,
	FM_FMRI_MEM_OFFSET, offset) != 0) {
	atomic_inc_64(
	&erpt_kstat_data.fmri_set_failed.value.ui64);
	}
	}
	}

	void
	fm_fmri_zfs_set(nvlist_t *fmri, int version, uint64_t pool_guid,
	uint64_t vdev_guid)
	{
	if (version != ZFS_SCHEME_VERSION0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}

	if (nvlist_add_uint8(fmri, FM_VERSION, version) != 0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}

	if (nvlist_add_string(fmri, FM_FMRI_SCHEME, FM_FMRI_SCHEME_ZFS) != 0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	return;
	}

	if (nvlist_add_uint64(fmri, FM_FMRI_ZFS_POOL, pool_guid) != 0) {
	atomic_inc_64(&erpt_kstat_data.fmri_set_failed.value.ui64);
	}

	if (vdev_guid != 0) {
	if (nvlist_add_uint64(fmri, FM_FMRI_ZFS_VDEV, vdev_guid) != 0) {
	atomic_inc_64(
	&erpt_kstat_data.fmri_set_failed.value.ui64);
	}
	}
	}

	uint64_t
	fm_ena_increment(uint64_t ena)
	{
	uint64_t new_ena;

	switch (ENA_FORMAT(ena)) {
	case FM_ENA_FMT1:
	new_ena = ena + (1 << ENA_FMT1_GEN_SHFT);
	break;
	case FM_ENA_FMT2:
	new_ena = ena + (1 << ENA_FMT2_GEN_SHFT);
	break;
	default:
	new_ena = 0;
	}

	return (new_ena);
	}

	uint64_t
	fm_ena_generate_cpu(uint64_t timestamp, processorid_t cpuid, uchar_t format)
	{
	uint64_t ena = 0;

	switch (format) {
	case FM_ENA_FMT1:
	if (timestamp) {
	ena = (uint64_t)((format & ENA_FORMAT_MASK) \|
	((cpuid << ENA_FMT1_CPUID_SHFT) &
	ENA_FMT1_CPUID_MASK) \|
	((timestamp << ENA_FMT1_TIME_SHFT) &
	ENA_FMT1_TIME_MASK));
	} else {
	ena = (uint64_t)((format & ENA_FORMAT_MASK) \|
	((cpuid << ENA_FMT1_CPUID_SHFT) &
	ENA_FMT1_CPUID_MASK) \|
	((gethrtime() << ENA_FMT1_TIME_SHFT) &
	ENA_FMT1_TIME_MASK));
	}
	break;
	case FM_ENA_FMT2:
	ena = (uint64_t)((format & ENA_FORMAT_MASK) \|
	((timestamp << ENA_FMT2_TIME_SHFT) & ENA_FMT2_TIME_MASK));
	break;
	default:
	break;
	}

	return (ena);
	}

	uint64_t
	fm_ena_generate(uint64_t timestamp, uchar_t format)
	{
	uint64_t ena;

	kpreempt_disable();
	ena = fm_ena_generate_cpu(timestamp, getcpuid(), format);
	kpreempt_enable();

	return (ena);
	}

	uint64_t
	fm_ena_generation_get(uint64_t ena)
	{
	uint64_t gen;

	switch (ENA_FORMAT(ena)) {
	case FM_ENA_FMT1:
	gen = (ena & ENA_FMT1_GEN_MASK) >> ENA_FMT1_GEN_SHFT;
	break;
	case FM_ENA_FMT2:
	gen = (ena & ENA_FMT2_GEN_MASK) >> ENA_FMT2_GEN_SHFT;
	break;
	default:
	gen = 0;
	break;
	}

	return (gen);
	}

	uchar_t
	fm_ena_format_get(uint64_t ena)
	{

	return (ENA_FORMAT(ena));
	}

	uint64_t
	fm_ena_id_get(uint64_t ena)
	{
	uint64_t id;

	switch (ENA_FORMAT(ena)) {
	case FM_ENA_FMT1:
	id = (ena & ENA_FMT1_ID_MASK) >> ENA_FMT1_ID_SHFT;
	break;
	case FM_ENA_FMT2:
	id = (ena & ENA_FMT2_ID_MASK) >> ENA_FMT2_ID_SHFT;
	break;
	default:
	id = 0;
	}

	return (id);
	}

	uint64_t
	fm_ena_time_get(uint64_t ena)
	{
	uint64_t time;

	switch (ENA_FORMAT(ena)) {
	case FM_ENA_FMT1:
	time = (ena & ENA_FMT1_TIME_MASK) >> ENA_FMT1_TIME_SHFT;
	break;
	case FM_ENA_FMT2:
	time = (ena & ENA_FMT2_TIME_MASK) >> ENA_FMT2_TIME_SHFT;
	break;
	default:
	time = 0;
	}

	return (time);
	}

	#ifdef _KERNEL
	/*
	* Helper function to increment ereport dropped count. Used by the event
	* rate limiting code to give feedback to the user about how many events were
	* rate limited by including them in the 'dropped' count.
	*/
	void
	fm_erpt_dropped_increment(void)
	{
	atomic_inc_64(&ratelimit_dropped);
	}

	void
	fm_init(void)
	{
	zevent_len_cur = 0;
	zevent_flags = 0;

	/* Initialize zevent allocation and generation kstats */
	fm_ksp = kstat_create("zfs", 0, "fm", "misc", KSTAT_TYPE_NAMED,
	sizeof (struct erpt_kstat) / sizeof (kstat_named_t),
	KSTAT_FLAG_VIRTUAL);

	if (fm_ksp != NULL) {
	fm_ksp->ks_data = &erpt_kstat_data;
	kstat_install(fm_ksp);
	} else {
	cmn_err(CE_NOTE, "failed to create fm/misc kstat\n");
	}

	mutex_init(&zevent_lock, NULL, MUTEX_DEFAULT, NULL);
	list_create(&zevent_list, sizeof (zevent_t),
	offsetof(zevent_t, ev_node));
	cv_init(&zevent_cv, NULL, CV_DEFAULT, NULL);

	zfs_ereport_init();
	}

	void
	fm_fini(void)
	{
	int count;

	zfs_ereport_fini();

	zfs_zevent_drain_all(&count);

	mutex_enter(&zevent_lock);
	cv_broadcast(&zevent_cv);

	zevent_flags \|= ZEVENT_SHUTDOWN;
	while (zevent_waiters > 0) {
	mutex_exit(&zevent_lock);
	schedule();
	mutex_enter(&zevent_lock);
	}
	mutex_exit(&zevent_lock);

	cv_destroy(&zevent_cv);
	list_destroy(&zevent_list);
	mutex_destroy(&zevent_lock);

	if (fm_ksp != NULL) {
	kstat_delete(fm_ksp);
	fm_ksp = NULL;
	}
	}
	#endif /* _KERNEL */

	ZFS_MODULE_PARAM(zfs_zevent, zfs_zevent_, len_max, INT, ZMOD_RW,
	"Max event queue length");
	diff --git a/sys/contrib/openzfs/module/zfs/spa.c b/sys/contrib/openzfs/module/zfs/spa.c
	index 7546e3e414f1..e5c41dc2b069 100644
	--- a/sys/contrib/openzfs/module/zfs/spa.c
	+++ b/sys/contrib/openzfs/module/zfs/spa.c
	@@ -1,9942 +1,9951 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2018, Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
	* Copyright 2013 Saso Kiselkov. All rights reserved.
	* Copyright (c) 2014 Integros [integros.com]
	* Copyright 2016 Toomas Soome <tsoome@me.com>
	* Copyright (c) 2016 Actifio, Inc. All rights reserved.
	* Copyright 2018 Joyent, Inc.
	* Copyright (c) 2017, 2019, Datto Inc. All rights reserved.
	* Copyright 2017 Joyent, Inc.
	* Copyright (c) 2017, Intel Corporation.
	* Copyright (c) 2021, Colm Buckley <colm@tuatha.org>
	*/

	/*
	* SPA: Storage Pool Allocator
	*
	* This file contains all the routines used when modifying on-disk SPA state.
	* This includes opening, importing, destroying, exporting a pool, and syncing a
	* pool.
	*/

	#include <sys/zfs_context.h>
	#include <sys/fm/fs/zfs.h>
	#include <sys/spa_impl.h>
	#include <sys/zio.h>
	#include <sys/zio_checksum.h>
	#include <sys/dmu.h>
	#include <sys/dmu_tx.h>
	#include <sys/zap.h>
	#include <sys/zil.h>
	#include <sys/ddt.h>
	#include <sys/vdev_impl.h>
	#include <sys/vdev_removal.h>
	#include <sys/vdev_indirect_mapping.h>
	#include <sys/vdev_indirect_births.h>
	#include <sys/vdev_initialize.h>
	#include <sys/vdev_rebuild.h>
	#include <sys/vdev_trim.h>
	#include <sys/vdev_disk.h>
	#include <sys/vdev_draid.h>
	#include <sys/metaslab.h>
	#include <sys/metaslab_impl.h>
	#include <sys/mmp.h>
	#include <sys/uberblock_impl.h>
	#include <sys/txg.h>
	#include <sys/avl.h>
	#include <sys/bpobj.h>
	#include <sys/dmu_traverse.h>
	#include <sys/dmu_objset.h>
	#include <sys/unique.h>
	#include <sys/dsl_pool.h>
	#include <sys/dsl_dataset.h>
	#include <sys/dsl_dir.h>
	#include <sys/dsl_prop.h>
	#include <sys/dsl_synctask.h>
	#include <sys/fs/zfs.h>
	#include <sys/arc.h>
	#include <sys/callb.h>
	#include <sys/systeminfo.h>
	#include <sys/spa_boot.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/dsl_scan.h>
	#include <sys/zfeature.h>
	#include <sys/dsl_destroy.h>
	#include <sys/zvol.h>

	#ifdef _KERNEL
	#include <sys/fm/protocol.h>
	#include <sys/fm/util.h>
	#include <sys/callb.h>
	#include <sys/zone.h>
	#include <sys/vmsystm.h>
	#endif /* _KERNEL */

	#include "zfs_prop.h"
	#include "zfs_comutil.h"

	/*
	* The interval, in seconds, at which failed configuration cache file writes
	* should be retried.
	*/
	int zfs_ccw_retry_interval = 300;

	typedef enum zti_modes {
	ZTI_MODE_FIXED, /* value is # of threads (min 1) */
	ZTI_MODE_BATCH, /* cpu-intensive; value is ignored */
	ZTI_MODE_SCALE, /* Taskqs scale with CPUs. */
	ZTI_MODE_NULL, /* don't create a taskq */
	ZTI_NMODES
	} zti_modes_t;

	#define ZTI_P(n, q) { ZTI_MODE_FIXED, (n), (q) }
	#define ZTI_PCT(n) { ZTI_MODE_ONLINE_PERCENT, (n), 1 }
	#define ZTI_BATCH { ZTI_MODE_BATCH, 0, 1 }
	#define ZTI_SCALE { ZTI_MODE_SCALE, 0, 1 }
	#define ZTI_NULL { ZTI_MODE_NULL, 0, 0 }

	#define ZTI_N(n) ZTI_P(n, 1)
	#define ZTI_ONE ZTI_N(1)

	typedef struct zio_taskq_info {
	zti_modes_t zti_mode;
	uint_t zti_value;
	uint_t zti_count;
	} zio_taskq_info_t;

	static const char *const zio_taskq_types[ZIO_TASKQ_TYPES] = {
	"iss", "iss_h", "int", "int_h"
	};

	/*
	* This table defines the taskq settings for each ZFS I/O type. When
	* initializing a pool, we use this table to create an appropriately sized
	* taskq. Some operations are low volume and therefore have a small, static
	* number of threads assigned to their taskqs using the ZTI_N(#) or ZTI_ONE
	* macros. Other operations process a large amount of data; the ZTI_BATCH
	* macro causes us to create a taskq oriented for throughput. Some operations
	* are so high frequency and short-lived that the taskq itself can become a
	* point of lock contention. The ZTI_P(#, #) macro indicates that we need an
	* additional degree of parallelism specified by the number of threads per-
	* taskq and the number of taskqs; when dispatching an event in this case, the
	* particular taskq is chosen at random. ZTI_SCALE is similar to ZTI_BATCH,
	* but with number of taskqs also scaling with number of CPUs.
	*
	* The different taskq priorities are to handle the different contexts (issue
	* and interrupt) and then to reserve threads for ZIO_PRIORITY_NOW I/Os that
	* need to be handled with minimum delay.
	*/
	const zio_taskq_info_t zio_taskqs[ZIO_TYPES][ZIO_TASKQ_TYPES] = {
	/* ISSUE ISSUE_HIGH INTR INTR_HIGH */
	{ ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* NULL */
	{ ZTI_N(8), ZTI_NULL, ZTI_SCALE, ZTI_NULL }, /* READ */
	{ ZTI_BATCH, ZTI_N(5), ZTI_SCALE, ZTI_N(5) }, /* WRITE */
	{ ZTI_SCALE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* FREE */
	{ ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* CLAIM */
	{ ZTI_ONE, ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* IOCTL */
	{ ZTI_N(4), ZTI_NULL, ZTI_ONE, ZTI_NULL }, /* TRIM */
	};

	static void spa_sync_version(void arg, dmu_tx_t tx);
	static void spa_sync_props(void arg, dmu_tx_t tx);
	static boolean_t spa_has_active_shared_spare(spa_t *spa);
	static int spa_load_impl(spa_t spa, spa_import_type_t type, char *ereport);
	static void spa_vdev_resilver_done(spa_t *spa);

	uint_t zio_taskq_batch_pct = 80; /* 1 thread per cpu in pset */
	uint_t zio_taskq_batch_tpq; /* threads per taskq */
	boolean_t zio_taskq_sysdc = B_TRUE; /* use SDC scheduling class */
	uint_t zio_taskq_basedc = 80; /* base duty cycle */

	boolean_t spa_create_process = B_TRUE; /* no process ==> no sysdc */

	/*
	* Report any spa_load_verify errors found, but do not fail spa_load.
	* This is used by zdb to analyze non-idle pools.
	*/
	boolean_t spa_load_verify_dryrun = B_FALSE;

	/*
	* Allow read spacemaps in case of readonly import (spa_mode == SPA_MODE_READ).
	* This is used by zdb for spacemaps verification.
	*/
	boolean_t spa_mode_readable_spacemaps = B_FALSE;

	/*
	* This (illegal) pool name is used when temporarily importing a spa_t in order
	* to get the vdev stats associated with the imported devices.
	*/
	#define TRYIMPORT_NAME "$import"

	/*
	* For debugging purposes: print out vdev tree during pool import.
	*/
	int spa_load_print_vdev_tree = B_FALSE;

	/*
	* A non-zero value for zfs_max_missing_tvds means that we allow importing
	* pools with missing top-level vdevs. This is strictly intended for advanced
	* pool recovery cases since missing data is almost inevitable. Pools with
	* missing devices can only be imported read-only for safety reasons, and their
	* fail-mode will be automatically set to "continue".
	*
	* With 1 missing vdev we should be able to import the pool and mount all
	* datasets. User data that was not modified after the missing device has been
	* added should be recoverable. This means that snapshots created prior to the
	* addition of that device should be completely intact.
	*
	* With 2 missing vdevs, some datasets may fail to mount since there are
	* dataset statistics that are stored as regular metadata. Some data might be
	* recoverable if those vdevs were added recently.
	*
	* With 3 or more missing vdevs, the pool is severely damaged and MOS entries
	* may be missing entirely. Chances of data recovery are very low. Note that
	* there are also risks of performing an inadvertent rewind as we might be
	* missing all the vdevs with the latest uberblocks.
	*/
	unsigned long zfs_max_missing_tvds = 0;

	/*
	* The parameters below are similar to zfs_max_missing_tvds but are only
	* intended for a preliminary open of the pool with an untrusted config which
	* might be incomplete or out-dated.
	*
	* We are more tolerant for pools opened from a cachefile since we could have
	* an out-dated cachefile where a device removal was not registered.
	* We could have set the limit arbitrarily high but in the case where devices
	* are really missing we would want to return the proper error codes; we chose
	* SPA_DVAS_PER_BP - 1 so that some copies of the MOS would still be available
	* and we get a chance to retrieve the trusted config.
	*/
	uint64_t zfs_max_missing_tvds_cachefile = SPA_DVAS_PER_BP - 1;

	/*
	* In the case where config was assembled by scanning device paths (/dev/dsks
	* by default) we are less tolerant since all the existing devices should have
	* been detected and we want spa_load to return the right error codes.
	*/
	uint64_t zfs_max_missing_tvds_scan = 0;

	/*
	* Debugging aid that pauses spa_sync() towards the end.
	*/
	boolean_t zfs_pause_spa_sync = B_FALSE;

	/*
	* Variables to indicate the livelist condense zthr func should wait at certain
	* points for the livelist to be removed - used to test condense/destroy races
	*/
	int zfs_livelist_condense_zthr_pause = 0;
	int zfs_livelist_condense_sync_pause = 0;

	/*
	* Variables to track whether or not condense cancellation has been
	* triggered in testing.
	*/
	int zfs_livelist_condense_sync_cancel = 0;
	int zfs_livelist_condense_zthr_cancel = 0;

	/*
	* Variable to track whether or not extra ALLOC blkptrs were added to a
	* livelist entry while it was being condensed (caused by the way we track
	* remapped blkptrs in dbuf_remap_impl)
	*/
	int zfs_livelist_condense_new_alloc = 0;

	/*
	* ==========================================================================
	* SPA properties routines
	* ==========================================================================
	*/

	/*
	* Add a (source=src, propname=propval) list to an nvlist.
	*/
	static void
	spa_prop_add_list(nvlist_t nvl, zpool_prop_t prop, char strval,
	uint64_t intval, zprop_source_t src)
	{
	const char *propname = zpool_prop_to_name(prop);
	nvlist_t *propval;

	propval = fnvlist_alloc();
	fnvlist_add_uint64(propval, ZPROP_SOURCE, src);

	if (strval != NULL)
	fnvlist_add_string(propval, ZPROP_VALUE, strval);
	else
	fnvlist_add_uint64(propval, ZPROP_VALUE, intval);

	fnvlist_add_nvlist(nvl, propname, propval);
	nvlist_free(propval);
	}

	/*
	* Get property values from the spa configuration.
	*/
	static void
	spa_prop_get_config(spa_t spa, nvlist_t *nvp)
	{
	vdev_t *rvd = spa->spa_root_vdev;
	dsl_pool_t *pool = spa->spa_dsl_pool;
	uint64_t size, alloc, cap, version;
	const zprop_source_t src = ZPROP_SRC_NONE;
	spa_config_dirent_t *dp;
	metaslab_class_t *mc = spa_normal_class(spa);

	ASSERT(MUTEX_HELD(&spa->spa_props_lock));

	if (rvd != NULL) {
	alloc = metaslab_class_get_alloc(mc);
	alloc += metaslab_class_get_alloc(spa_special_class(spa));
	alloc += metaslab_class_get_alloc(spa_dedup_class(spa));
	alloc += metaslab_class_get_alloc(spa_embedded_log_class(spa));

	size = metaslab_class_get_space(mc);
	size += metaslab_class_get_space(spa_special_class(spa));
	size += metaslab_class_get_space(spa_dedup_class(spa));
	size += metaslab_class_get_space(spa_embedded_log_class(spa));

	spa_prop_add_list(*nvp, ZPOOL_PROP_NAME, spa_name(spa), 0, src);
	spa_prop_add_list(*nvp, ZPOOL_PROP_SIZE, NULL, size, src);
	spa_prop_add_list(*nvp, ZPOOL_PROP_ALLOCATED, NULL, alloc, src);
	spa_prop_add_list(*nvp, ZPOOL_PROP_FREE, NULL,
	size - alloc, src);
	spa_prop_add_list(*nvp, ZPOOL_PROP_CHECKPOINT, NULL,
	spa->spa_checkpoint_info.sci_dspace, src);

	spa_prop_add_list(*nvp, ZPOOL_PROP_FRAGMENTATION, NULL,
	metaslab_class_fragmentation(mc), src);
	spa_prop_add_list(*nvp, ZPOOL_PROP_EXPANDSZ, NULL,
	metaslab_class_expandable_space(mc), src);
	spa_prop_add_list(*nvp, ZPOOL_PROP_READONLY, NULL,
	(spa_mode(spa) == SPA_MODE_READ), src);

	cap = (size == 0) ? 0 : (alloc * 100 / size);
	spa_prop_add_list(*nvp, ZPOOL_PROP_CAPACITY, NULL, cap, src);

	spa_prop_add_list(*nvp, ZPOOL_PROP_DEDUPRATIO, NULL,
	ddt_get_pool_dedup_ratio(spa), src);

	spa_prop_add_list(*nvp, ZPOOL_PROP_HEALTH, NULL,
	rvd->vdev_state, src);

	version = spa_version(spa);
	if (version == zpool_prop_default_numeric(ZPOOL_PROP_VERSION)) {
	spa_prop_add_list(*nvp, ZPOOL_PROP_VERSION, NULL,
	version, ZPROP_SRC_DEFAULT);
	} else {
	spa_prop_add_list(*nvp, ZPOOL_PROP_VERSION, NULL,
	version, ZPROP_SRC_LOCAL);
	}
	spa_prop_add_list(*nvp, ZPOOL_PROP_LOAD_GUID,
	NULL, spa_load_guid(spa), src);
	}

	if (pool != NULL) {
	/*
	* The $FREE directory was introduced in SPA_VERSION_DEADLISTS,
	* when opening pools before this version freedir will be NULL.
	*/
	if (pool->dp_free_dir != NULL) {
	spa_prop_add_list(*nvp, ZPOOL_PROP_FREEING, NULL,
	dsl_dir_phys(pool->dp_free_dir)->dd_used_bytes,
	src);
	} else {
	spa_prop_add_list(*nvp, ZPOOL_PROP_FREEING,
	NULL, 0, src);
	}

	if (pool->dp_leak_dir != NULL) {
	spa_prop_add_list(*nvp, ZPOOL_PROP_LEAKED, NULL,
	dsl_dir_phys(pool->dp_leak_dir)->dd_used_bytes,
	src);
	} else {
	spa_prop_add_list(*nvp, ZPOOL_PROP_LEAKED,
	NULL, 0, src);
	}
	}

	spa_prop_add_list(*nvp, ZPOOL_PROP_GUID, NULL, spa_guid(spa), src);

	if (spa->spa_comment != NULL) {
	spa_prop_add_list(*nvp, ZPOOL_PROP_COMMENT, spa->spa_comment,
	0, ZPROP_SRC_LOCAL);
	}

	if (spa->spa_compatibility != NULL) {
	spa_prop_add_list(*nvp, ZPOOL_PROP_COMPATIBILITY,
	spa->spa_compatibility, 0, ZPROP_SRC_LOCAL);
	}

	if (spa->spa_root != NULL)
	spa_prop_add_list(*nvp, ZPOOL_PROP_ALTROOT, spa->spa_root,
	0, ZPROP_SRC_LOCAL);

	if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_BLOCKS)) {
	spa_prop_add_list(*nvp, ZPOOL_PROP_MAXBLOCKSIZE, NULL,
	MIN(zfs_max_recordsize, SPA_MAXBLOCKSIZE), ZPROP_SRC_NONE);
	} else {
	spa_prop_add_list(*nvp, ZPOOL_PROP_MAXBLOCKSIZE, NULL,
	SPA_OLD_MAXBLOCKSIZE, ZPROP_SRC_NONE);
	}

	if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_DNODE)) {
	spa_prop_add_list(*nvp, ZPOOL_PROP_MAXDNODESIZE, NULL,
	DNODE_MAX_SIZE, ZPROP_SRC_NONE);
	} else {
	spa_prop_add_list(*nvp, ZPOOL_PROP_MAXDNODESIZE, NULL,
	DNODE_MIN_SIZE, ZPROP_SRC_NONE);
	}

	if ((dp = list_head(&spa->spa_config_list)) != NULL) {
	if (dp->scd_path == NULL) {
	spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE,
	"none", 0, ZPROP_SRC_LOCAL);
	} else if (strcmp(dp->scd_path, spa_config_path) != 0) {
	spa_prop_add_list(*nvp, ZPOOL_PROP_CACHEFILE,
	dp->scd_path, 0, ZPROP_SRC_LOCAL);
	}
	}
	}

	/*
	* Get zpool property values.
	*/
	int
	spa_prop_get(spa_t spa, nvlist_t *nvp)
	{
	objset_t *mos = spa->spa_meta_objset;
	zap_cursor_t zc;
	zap_attribute_t za;
	dsl_pool_t *dp;
	int err;

	err = nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_SLEEP);
	if (err)
	return (err);

	dp = spa_get_dsl(spa);
	dsl_pool_config_enter(dp, FTAG);
	mutex_enter(&spa->spa_props_lock);

	/*
	* Get properties from the spa config.
	*/
	spa_prop_get_config(spa, nvp);

	/* If no pool property object, no more prop to get. */
	if (mos == NULL \|\| spa->spa_pool_props_object == 0)
	goto out;

	/*
	* Get properties from the MOS pool property object.
	*/
	for (zap_cursor_init(&zc, mos, spa->spa_pool_props_object);
	(err = zap_cursor_retrieve(&zc, &za)) == 0;
	zap_cursor_advance(&zc)) {
	uint64_t intval = 0;
	char *strval = NULL;
	zprop_source_t src = ZPROP_SRC_DEFAULT;
	zpool_prop_t prop;

	if ((prop = zpool_name_to_prop(za.za_name)) == ZPOOL_PROP_INVAL)
	continue;

	switch (za.za_integer_length) {
	case 8:
	/* integer property */
	if (za.za_first_integer !=
	zpool_prop_default_numeric(prop))
	src = ZPROP_SRC_LOCAL;

	if (prop == ZPOOL_PROP_BOOTFS) {
	dsl_dataset_t *ds = NULL;

	err = dsl_dataset_hold_obj(dp,
	za.za_first_integer, FTAG, &ds);
	if (err != 0)
	break;

	strval = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN,
	KM_SLEEP);
	dsl_dataset_name(ds, strval);
	dsl_dataset_rele(ds, FTAG);
	} else {
	strval = NULL;
	intval = za.za_first_integer;
	}

	spa_prop_add_list(*nvp, prop, strval, intval, src);

	if (strval != NULL)
	kmem_free(strval, ZFS_MAX_DATASET_NAME_LEN);

	break;

	case 1:
	/* string property */
	strval = kmem_alloc(za.za_num_integers, KM_SLEEP);
	err = zap_lookup(mos, spa->spa_pool_props_object,
	za.za_name, 1, za.za_num_integers, strval);
	if (err) {
	kmem_free(strval, za.za_num_integers);
	break;
	}
	spa_prop_add_list(*nvp, prop, strval, 0, src);
	kmem_free(strval, za.za_num_integers);
	break;

	default:
	break;
	}
	}
	zap_cursor_fini(&zc);
	out:
	mutex_exit(&spa->spa_props_lock);
	dsl_pool_config_exit(dp, FTAG);
	if (err && err != ENOENT) {
	nvlist_free(*nvp);
	*nvp = NULL;
	return (err);
	}

	return (0);
	}

	/*
	* Validate the given pool properties nvlist and modify the list
	* for the property values to be set.
	*/
	static int
	spa_prop_validate(spa_t spa, nvlist_t props)
	{
	nvpair_t *elem;
	int error = 0, reset_bootfs = 0;
	uint64_t objnum = 0;
	boolean_t has_feature = B_FALSE;

	elem = NULL;
	while ((elem = nvlist_next_nvpair(props, elem)) != NULL) {
	uint64_t intval;
	char strval, slash, check, fname;
	const char *propname = nvpair_name(elem);
	zpool_prop_t prop = zpool_name_to_prop(propname);

	switch (prop) {
	case ZPOOL_PROP_INVAL:
	if (!zpool_prop_feature(propname)) {
	error = SET_ERROR(EINVAL);
	break;
	}

	/*
	* Sanitize the input.
	*/
	if (nvpair_type(elem) != DATA_TYPE_UINT64) {
	error = SET_ERROR(EINVAL);
	break;
	}

	if (nvpair_value_uint64(elem, &intval) != 0) {
	error = SET_ERROR(EINVAL);
	break;
	}

	if (intval != 0) {
	error = SET_ERROR(EINVAL);
	break;
	}

	fname = strchr(propname, '@') + 1;
	if (zfeature_lookup_name(fname, NULL) != 0) {
	error = SET_ERROR(EINVAL);
	break;
	}

	has_feature = B_TRUE;
	break;

	case ZPOOL_PROP_VERSION:
	error = nvpair_value_uint64(elem, &intval);
	if (!error &&
	(intval < spa_version(spa) \|\|
	intval > SPA_VERSION_BEFORE_FEATURES \|\|
	has_feature))
	error = SET_ERROR(EINVAL);
	break;

	case ZPOOL_PROP_DELEGATION:
	case ZPOOL_PROP_AUTOREPLACE:
	case ZPOOL_PROP_LISTSNAPS:
	case ZPOOL_PROP_AUTOEXPAND:
	case ZPOOL_PROP_AUTOTRIM:
	error = nvpair_value_uint64(elem, &intval);
	if (!error && intval > 1)
	error = SET_ERROR(EINVAL);
	break;

	case ZPOOL_PROP_MULTIHOST:
	error = nvpair_value_uint64(elem, &intval);
	if (!error && intval > 1)
	error = SET_ERROR(EINVAL);

	if (!error) {
	uint32_t hostid = zone_get_hostid(NULL);
	if (hostid)
	spa->spa_hostid = hostid;
	else
	error = SET_ERROR(ENOTSUP);
	}

	break;

	case ZPOOL_PROP_BOOTFS:
	/*
	* If the pool version is less than SPA_VERSION_BOOTFS,
	* or the pool is still being created (version == 0),
	* the bootfs property cannot be set.
	*/
	if (spa_version(spa) < SPA_VERSION_BOOTFS) {
	error = SET_ERROR(ENOTSUP);
	break;
	}

	/*
	* Make sure the vdev config is bootable
	*/
	if (!vdev_is_bootable(spa->spa_root_vdev)) {
	error = SET_ERROR(ENOTSUP);
	break;
	}

	reset_bootfs = 1;

	error = nvpair_value_string(elem, &strval);

	if (!error) {
	objset_t *os;

	if (strval == NULL \|\| strval[0] == '\0') {
	objnum = zpool_prop_default_numeric(
	ZPOOL_PROP_BOOTFS);
	break;
	}

	error = dmu_objset_hold(strval, FTAG, &os);
	if (error != 0)
	break;

	/* Must be ZPL. */
	if (dmu_objset_type(os) != DMU_OST_ZFS) {
	error = SET_ERROR(ENOTSUP);
	} else {
	objnum = dmu_objset_id(os);
	}
	dmu_objset_rele(os, FTAG);
	}
	break;

	case ZPOOL_PROP_FAILUREMODE:
	error = nvpair_value_uint64(elem, &intval);
	if (!error && intval > ZIO_FAILURE_MODE_PANIC)
	error = SET_ERROR(EINVAL);

	/*
	* This is a special case which only occurs when
	* the pool has completely failed. This allows
	* the user to change the in-core failmode property
	* without syncing it out to disk (I/Os might
	* currently be blocked). We do this by returning
	* EIO to the caller (spa_prop_set) to trick it
	* into thinking we encountered a property validation
	* error.
	*/
	if (!error && spa_suspended(spa)) {
	spa->spa_failmode = intval;
	error = SET_ERROR(EIO);
	}
	break;

	case ZPOOL_PROP_CACHEFILE:
	if ((error = nvpair_value_string(elem, &strval)) != 0)
	break;

	if (strval[0] == '\0')
	break;

	if (strcmp(strval, "none") == 0)
	break;

	if (strval[0] != '/') {
	error = SET_ERROR(EINVAL);
	break;
	}

	slash = strrchr(strval, '/');
	ASSERT(slash != NULL);

	if (slash[1] == '\0' \|\| strcmp(slash, "/.") == 0 \|\|
	strcmp(slash, "/..") == 0)
	error = SET_ERROR(EINVAL);
	break;

	case ZPOOL_PROP_COMMENT:
	if ((error = nvpair_value_string(elem, &strval)) != 0)
	break;
	for (check = strval; *check != '\0'; check++) {
	if (!isprint(*check)) {
	error = SET_ERROR(EINVAL);
	break;
	}
	}
	if (strlen(strval) > ZPROP_MAX_COMMENT)
	error = SET_ERROR(E2BIG);
	break;

	default:
	break;
	}

	if (error)
	break;
	}

	(void) nvlist_remove_all(props,
	zpool_prop_to_name(ZPOOL_PROP_DEDUPDITTO));

	if (!error && reset_bootfs) {
	error = nvlist_remove(props,
	zpool_prop_to_name(ZPOOL_PROP_BOOTFS), DATA_TYPE_STRING);

	if (!error) {
	error = nvlist_add_uint64(props,
	zpool_prop_to_name(ZPOOL_PROP_BOOTFS), objnum);
	}
	}

	return (error);
	}

	void
	spa_configfile_set(spa_t spa, nvlist_t nvp, boolean_t need_sync)
	{
	char *cachefile;
	spa_config_dirent_t *dp;

	if (nvlist_lookup_string(nvp, zpool_prop_to_name(ZPOOL_PROP_CACHEFILE),
	&cachefile) != 0)
	return;

	dp = kmem_alloc(sizeof (spa_config_dirent_t),
	KM_SLEEP);

	if (cachefile[0] == '\0')
	dp->scd_path = spa_strdup(spa_config_path);
	else if (strcmp(cachefile, "none") == 0)
	dp->scd_path = NULL;
	else
	dp->scd_path = spa_strdup(cachefile);

	list_insert_head(&spa->spa_config_list, dp);
	if (need_sync)
	spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE);
	}

	int
	spa_prop_set(spa_t spa, nvlist_t nvp)
	{
	int error;
	nvpair_t *elem = NULL;
	boolean_t need_sync = B_FALSE;

	if ((error = spa_prop_validate(spa, nvp)) != 0)
	return (error);

	while ((elem = nvlist_next_nvpair(nvp, elem)) != NULL) {
	zpool_prop_t prop = zpool_name_to_prop(nvpair_name(elem));

	if (prop == ZPOOL_PROP_CACHEFILE \|\|
	prop == ZPOOL_PROP_ALTROOT \|\|
	prop == ZPOOL_PROP_READONLY)
	continue;

	if (prop == ZPOOL_PROP_VERSION \|\| prop == ZPOOL_PROP_INVAL) {
	- uint64_t ver;
	+ uint64_t ver = 0;

	if (prop == ZPOOL_PROP_VERSION) {
	VERIFY(nvpair_value_uint64(elem, &ver) == 0);
	} else {
	ASSERT(zpool_prop_feature(nvpair_name(elem)));
	ver = SPA_VERSION_FEATURES;
	need_sync = B_TRUE;
	}

	/* Save time if the version is already set. */
	if (ver == spa_version(spa))
	continue;

	/*
	* In addition to the pool directory object, we might
	* create the pool properties object, the features for
	* read object, the features for write object, or the
	* feature descriptions object.
	*/
	error = dsl_sync_task(spa->spa_name, NULL,
	spa_sync_version, &ver,
	6, ZFS_SPACE_CHECK_RESERVED);
	if (error)
	return (error);
	continue;
	}

	need_sync = B_TRUE;
	break;
	}

	if (need_sync) {
	return (dsl_sync_task(spa->spa_name, NULL, spa_sync_props,
	nvp, 6, ZFS_SPACE_CHECK_RESERVED));
	}

	return (0);
	}

	/*
	* If the bootfs property value is dsobj, clear it.
	*/
	void
	spa_prop_clear_bootfs(spa_t spa, uint64_t dsobj, dmu_tx_t tx)
	{
	if (spa->spa_bootfs == dsobj && spa->spa_pool_props_object != 0) {
	VERIFY(zap_remove(spa->spa_meta_objset,
	spa->spa_pool_props_object,
	zpool_prop_to_name(ZPOOL_PROP_BOOTFS), tx) == 0);
	spa->spa_bootfs = 0;
	}
	}

	/ARGSUSED/
	static int
	spa_change_guid_check(void arg, dmu_tx_t tx)
	{
	uint64_t *newguid __maybe_unused = arg;
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;
	vdev_t *rvd = spa->spa_root_vdev;
	uint64_t vdev_state;

	if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) {
	int error = (spa_has_checkpoint(spa)) ?
	ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT;
	return (SET_ERROR(error));
	}

	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
	vdev_state = rvd->vdev_state;
	spa_config_exit(spa, SCL_STATE, FTAG);

	if (vdev_state != VDEV_STATE_HEALTHY)
	return (SET_ERROR(ENXIO));

	ASSERT3U(spa_guid(spa), !=, *newguid);

	return (0);
	}

	static void
	spa_change_guid_sync(void arg, dmu_tx_t tx)
	{
	uint64_t *newguid = arg;
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;
	uint64_t oldguid;
	vdev_t *rvd = spa->spa_root_vdev;

	oldguid = spa_guid(spa);

	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
	rvd->vdev_guid = *newguid;
	rvd->vdev_guid_sum += (*newguid - oldguid);
	vdev_config_dirty(rvd);
	spa_config_exit(spa, SCL_STATE, FTAG);

	spa_history_log_internal(spa, "guid change", tx, "old=%llu new=%llu",
	(u_longlong_t)oldguid, (u_longlong_t)*newguid);
	}

	/*
	* Change the GUID for the pool. This is done so that we can later
	* re-import a pool built from a clone of our own vdevs. We will modify
	* the root vdev's guid, our own pool guid, and then mark all of our
	* vdevs dirty. Note that we must make sure that all our vdevs are
	* online when we do this, or else any vdevs that weren't present
	* would be orphaned from our pool. We are also going to issue a
	* sysevent to update any watchers.
	*/
	int
	spa_change_guid(spa_t *spa)
	{
	int error;
	uint64_t guid;

	mutex_enter(&spa->spa_vdev_top_lock);
	mutex_enter(&spa_namespace_lock);
	guid = spa_generate_guid(NULL);

	error = dsl_sync_task(spa->spa_name, spa_change_guid_check,
	spa_change_guid_sync, &guid, 5, ZFS_SPACE_CHECK_RESERVED);

	if (error == 0) {
	spa_write_cachefile(spa, B_FALSE, B_TRUE);
	spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_REGUID);
	}

	mutex_exit(&spa_namespace_lock);
	mutex_exit(&spa->spa_vdev_top_lock);

	return (error);
	}

	/*
	* ==========================================================================
	* SPA state manipulation (open/create/destroy/import/export)
	* ==========================================================================
	*/

	static int
	spa_error_entry_compare(const void a, const void b)
	{
	const spa_error_entry_t sa = (const spa_error_entry_t )a;
	const spa_error_entry_t sb = (const spa_error_entry_t )b;
	int ret;

	ret = memcmp(&sa->se_bookmark, &sb->se_bookmark,
	sizeof (zbookmark_phys_t));

	return (TREE_ISIGN(ret));
	}

	/*
	* Utility function which retrieves copies of the current logs and
	* re-initializes them in the process.
	*/
	void
	spa_get_errlists(spa_t spa, avl_tree_t last, avl_tree_t *scrub)
	{
	ASSERT(MUTEX_HELD(&spa->spa_errlist_lock));

	bcopy(&spa->spa_errlist_last, last, sizeof (avl_tree_t));
	bcopy(&spa->spa_errlist_scrub, scrub, sizeof (avl_tree_t));

	avl_create(&spa->spa_errlist_scrub,
	spa_error_entry_compare, sizeof (spa_error_entry_t),
	offsetof(spa_error_entry_t, se_avl));
	avl_create(&spa->spa_errlist_last,
	spa_error_entry_compare, sizeof (spa_error_entry_t),
	offsetof(spa_error_entry_t, se_avl));
	}

	static void
	spa_taskqs_init(spa_t *spa, zio_type_t t, zio_taskq_type_t q)
	{
	const zio_taskq_info_t *ztip = &zio_taskqs[t][q];
	enum zti_modes mode = ztip->zti_mode;
	uint_t value = ztip->zti_value;
	uint_t count = ztip->zti_count;
	spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];
	uint_t cpus, flags = TASKQ_DYNAMIC;
	boolean_t batch = B_FALSE;

	switch (mode) {
	case ZTI_MODE_FIXED:
	ASSERT3U(value, >, 0);
	break;

	case ZTI_MODE_BATCH:
	batch = B_TRUE;
	flags \|= TASKQ_THREADS_CPU_PCT;
	value = MIN(zio_taskq_batch_pct, 100);
	break;

	case ZTI_MODE_SCALE:
	flags \|= TASKQ_THREADS_CPU_PCT;
	/*
	* We want more taskqs to reduce lock contention, but we want
	* less for better request ordering and CPU utilization.
	*/
	cpus = MAX(1, boot_ncpus * zio_taskq_batch_pct / 100);
	if (zio_taskq_batch_tpq > 0) {
	count = MAX(1, (cpus + zio_taskq_batch_tpq / 2) /
	zio_taskq_batch_tpq);
	} else {
	/*
	* Prefer 6 threads per taskq, but no more taskqs
	* than threads in them on large systems. For 80%:
	*
	* taskq taskq total
	* cpus taskqs percent threads threads
	* ------- ------- ------- ------- -------
	* 1 1 80% 1 1
	* 2 1 80% 1 1
	* 4 1 80% 3 3
	* 8 2 40% 3 6
	* 16 3 27% 4 12
	* 32 5 16% 5 25
	* 64 7 11% 7 49
	* 128 10 8% 10 100
	* 256 14 6% 15 210
	*/
	count = 1 + cpus / 6;
	while (count * count > cpus)
	count--;
	}
	/* Limit each taskq within 100% to not trigger assertion. */
	count = MAX(count, (zio_taskq_batch_pct + 99) / 100);
	value = (zio_taskq_batch_pct + count / 2) / count;
	break;

	case ZTI_MODE_NULL:
	tqs->stqs_count = 0;
	tqs->stqs_taskq = NULL;
	return;

	default:
	panic("unrecognized mode for %s_%s taskq (%u:%u) in "
	"spa_activate()",
	zio_type_name[t], zio_taskq_types[q], mode, value);
	break;
	}

	ASSERT3U(count, >, 0);
	tqs->stqs_count = count;
	tqs->stqs_taskq = kmem_alloc(count * sizeof (taskq_t *), KM_SLEEP);

	for (uint_t i = 0; i < count; i++) {
	taskq_t *tq;
	char name[32];

	if (count > 1)
	(void) snprintf(name, sizeof (name), "%s_%s_%u",
	zio_type_name[t], zio_taskq_types[q], i);
	else
	(void) snprintf(name, sizeof (name), "%s_%s",
	zio_type_name[t], zio_taskq_types[q]);

	if (zio_taskq_sysdc && spa->spa_proc != &p0) {
	if (batch)
	flags \|= TASKQ_DC_BATCH;

	tq = taskq_create_sysdc(name, value, 50, INT_MAX,
	spa->spa_proc, zio_taskq_basedc, flags);
	} else {
	pri_t pri = maxclsyspri;
	/*
	* The write issue taskq can be extremely CPU
	* intensive. Run it at slightly less important
	* priority than the other taskqs.
	*
	* Under Linux and FreeBSD this means incrementing
	* the priority value as opposed to platforms like
	* illumos where it should be decremented.
	*
	* On FreeBSD, if priorities divided by four (RQ_PPQ)
	* are equal then a difference between them is
	* insignificant.
	*/
	if (t == ZIO_TYPE_WRITE && q == ZIO_TASKQ_ISSUE) {
	#if defined(__linux__)
	pri++;
	#elif defined(__FreeBSD__)
	pri += 4;
	#else
	#error "unknown OS"
	#endif
	}
	tq = taskq_create_proc(name, value, pri, 50,
	INT_MAX, spa->spa_proc, flags);
	}

	tqs->stqs_taskq[i] = tq;
	}
	}

	static void
	spa_taskqs_fini(spa_t *spa, zio_type_t t, zio_taskq_type_t q)
	{
	spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];

	if (tqs->stqs_taskq == NULL) {
	ASSERT3U(tqs->stqs_count, ==, 0);
	return;
	}

	for (uint_t i = 0; i < tqs->stqs_count; i++) {
	ASSERT3P(tqs->stqs_taskq[i], !=, NULL);
	taskq_destroy(tqs->stqs_taskq[i]);
	}

	kmem_free(tqs->stqs_taskq, tqs->stqs_count * sizeof (taskq_t *));
	tqs->stqs_taskq = NULL;
	}

	/*
	* Dispatch a task to the appropriate taskq for the ZFS I/O type and priority.
	* Note that a type may have multiple discrete taskqs to avoid lock contention
	* on the taskq itself. In that case we choose which taskq at random by using
	* the low bits of gethrtime().
	*/
	void
	spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
	task_func_t func, void arg, uint_t flags, taskq_ent_t *ent)
	{
	spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];
	taskq_t *tq;

	ASSERT3P(tqs->stqs_taskq, !=, NULL);
	ASSERT3U(tqs->stqs_count, !=, 0);

	if (tqs->stqs_count == 1) {
	tq = tqs->stqs_taskq[0];
	} else {
	tq = tqs->stqs_taskq[((uint64_t)gethrtime()) % tqs->stqs_count];
	}

	taskq_dispatch_ent(tq, func, arg, flags, ent);
	}

	/*
	* Same as spa_taskq_dispatch_ent() but block on the task until completion.
	*/
	void
	spa_taskq_dispatch_sync(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
	task_func_t func, void arg, uint_t flags)
	{
	spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];
	taskq_t *tq;
	taskqid_t id;

	ASSERT3P(tqs->stqs_taskq, !=, NULL);
	ASSERT3U(tqs->stqs_count, !=, 0);

	if (tqs->stqs_count == 1) {
	tq = tqs->stqs_taskq[0];
	} else {
	tq = tqs->stqs_taskq[((uint64_t)gethrtime()) % tqs->stqs_count];
	}

	id = taskq_dispatch(tq, func, arg, flags);
	if (id)
	taskq_wait_id(tq, id);
	}

	static void
	spa_create_zio_taskqs(spa_t *spa)
	{
	for (int t = 0; t < ZIO_TYPES; t++) {
	for (int q = 0; q < ZIO_TASKQ_TYPES; q++) {
	spa_taskqs_init(spa, t, q);
	}
	}
	}

	/*
	* Disabled until spa_thread() can be adapted for Linux.
	*/
	#undef HAVE_SPA_THREAD

	#if defined(_KERNEL) && defined(HAVE_SPA_THREAD)
	static void
	spa_thread(void *arg)
	{
	psetid_t zio_taskq_psrset_bind = PS_NONE;
	callb_cpr_t cprinfo;

	spa_t *spa = arg;
	user_t *pu = PTOU(curproc);

	CALLB_CPR_INIT(&cprinfo, &spa->spa_proc_lock, callb_generic_cpr,
	spa->spa_name);

	ASSERT(curproc != &p0);
	(void) snprintf(pu->u_psargs, sizeof (pu->u_psargs),
	"zpool-%s", spa->spa_name);
	(void) strlcpy(pu->u_comm, pu->u_psargs, sizeof (pu->u_comm));

	/* bind this thread to the requested psrset */
	if (zio_taskq_psrset_bind != PS_NONE) {
	pool_lock();
	mutex_enter(&cpu_lock);
	mutex_enter(&pidlock);
	mutex_enter(&curproc->p_lock);

	if (cpupart_bind_thread(curthread, zio_taskq_psrset_bind,
	0, NULL, NULL) == 0) {
	curthread->t_bind_pset = zio_taskq_psrset_bind;
	} else {
	cmn_err(CE_WARN,
	"Couldn't bind process for zfs pool \"%s\" to "
	"pset %d\n", spa->spa_name, zio_taskq_psrset_bind);
	}

	mutex_exit(&curproc->p_lock);
	mutex_exit(&pidlock);
	mutex_exit(&cpu_lock);
	pool_unlock();
	}

	if (zio_taskq_sysdc) {
	sysdc_thread_enter(curthread, 100, 0);
	}

	spa->spa_proc = curproc;
	spa->spa_did = curthread->t_did;

	spa_create_zio_taskqs(spa);

	mutex_enter(&spa->spa_proc_lock);
	ASSERT(spa->spa_proc_state == SPA_PROC_CREATED);

	spa->spa_proc_state = SPA_PROC_ACTIVE;
	cv_broadcast(&spa->spa_proc_cv);

	CALLB_CPR_SAFE_BEGIN(&cprinfo);
	while (spa->spa_proc_state == SPA_PROC_ACTIVE)
	cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock);
	CALLB_CPR_SAFE_END(&cprinfo, &spa->spa_proc_lock);

	ASSERT(spa->spa_proc_state == SPA_PROC_DEACTIVATE);
	spa->spa_proc_state = SPA_PROC_GONE;
	spa->spa_proc = &p0;
	cv_broadcast(&spa->spa_proc_cv);
	CALLB_CPR_EXIT(&cprinfo); /* drops spa_proc_lock */

	mutex_enter(&curproc->p_lock);
	lwp_exit();
	}
	#endif

	/*
	* Activate an uninitialized pool.
	*/
	static void
	spa_activate(spa_t *spa, spa_mode_t mode)
	{
	ASSERT(spa->spa_state == POOL_STATE_UNINITIALIZED);

	spa->spa_state = POOL_STATE_ACTIVE;
	spa->spa_mode = mode;
	spa->spa_read_spacemaps = spa_mode_readable_spacemaps;

	spa->spa_normal_class = metaslab_class_create(spa, zfs_metaslab_ops);
	spa->spa_log_class = metaslab_class_create(spa, zfs_metaslab_ops);
	spa->spa_embedded_log_class =
	metaslab_class_create(spa, zfs_metaslab_ops);
	spa->spa_special_class = metaslab_class_create(spa, zfs_metaslab_ops);
	spa->spa_dedup_class = metaslab_class_create(spa, zfs_metaslab_ops);

	/* Try to create a covering process */
	mutex_enter(&spa->spa_proc_lock);
	ASSERT(spa->spa_proc_state == SPA_PROC_NONE);
	ASSERT(spa->spa_proc == &p0);
	spa->spa_did = 0;

	#ifdef HAVE_SPA_THREAD
	/* Only create a process if we're going to be around a while. */
	if (spa_create_process && strcmp(spa->spa_name, TRYIMPORT_NAME) != 0) {
	if (newproc(spa_thread, (caddr_t)spa, syscid, maxclsyspri,
	NULL, 0) == 0) {
	spa->spa_proc_state = SPA_PROC_CREATED;
	while (spa->spa_proc_state == SPA_PROC_CREATED) {
	cv_wait(&spa->spa_proc_cv,
	&spa->spa_proc_lock);
	}
	ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE);
	ASSERT(spa->spa_proc != &p0);
	ASSERT(spa->spa_did != 0);
	} else {
	#ifdef _KERNEL
	cmn_err(CE_WARN,
	"Couldn't create process for zfs pool \"%s\"\n",
	spa->spa_name);
	#endif
	}
	}
	#endif /* HAVE_SPA_THREAD */
	mutex_exit(&spa->spa_proc_lock);

	/* If we didn't create a process, we need to create our taskqs. */
	if (spa->spa_proc == &p0) {
	spa_create_zio_taskqs(spa);
	}

	for (size_t i = 0; i < TXG_SIZE; i++) {
	spa->spa_txg_zio[i] = zio_root(spa, NULL, NULL,
	ZIO_FLAG_CANFAIL);
	}

	list_create(&spa->spa_config_dirty_list, sizeof (vdev_t),
	offsetof(vdev_t, vdev_config_dirty_node));
	list_create(&spa->spa_evicting_os_list, sizeof (objset_t),
	offsetof(objset_t, os_evicting_node));
	list_create(&spa->spa_state_dirty_list, sizeof (vdev_t),
	offsetof(vdev_t, vdev_state_dirty_node));

	txg_list_create(&spa->spa_vdev_txg_list, spa,
	offsetof(struct vdev, vdev_txg_node));

	avl_create(&spa->spa_errlist_scrub,
	spa_error_entry_compare, sizeof (spa_error_entry_t),
	offsetof(spa_error_entry_t, se_avl));
	avl_create(&spa->spa_errlist_last,
	spa_error_entry_compare, sizeof (spa_error_entry_t),
	offsetof(spa_error_entry_t, se_avl));

	spa_keystore_init(&spa->spa_keystore);

	/*
	* This taskq is used to perform zvol-minor-related tasks
	* asynchronously. This has several advantages, including easy
	* resolution of various deadlocks.
	*
	* The taskq must be single threaded to ensure tasks are always
	* processed in the order in which they were dispatched.
	*
	* A taskq per pool allows one to keep the pools independent.
	* This way if one pool is suspended, it will not impact another.
	*
	* The preferred location to dispatch a zvol minor task is a sync
	* task. In this context, there is easy access to the spa_t and minimal
	* error handling is required because the sync task must succeed.
	*/
	spa->spa_zvol_taskq = taskq_create("z_zvol", 1, defclsyspri,
	1, INT_MAX, 0);

	/*
	* Taskq dedicated to prefetcher threads: this is used to prevent the
	* pool traverse code from monopolizing the global (and limited)
	* system_taskq by inappropriately scheduling long running tasks on it.
	*/
	spa->spa_prefetch_taskq = taskq_create("z_prefetch", 100,
	defclsyspri, 1, INT_MAX, TASKQ_DYNAMIC \| TASKQ_THREADS_CPU_PCT);

	/*
	* The taskq to upgrade datasets in this pool. Currently used by
	* feature SPA_FEATURE_USEROBJ_ACCOUNTING/SPA_FEATURE_PROJECT_QUOTA.
	*/
	spa->spa_upgrade_taskq = taskq_create("z_upgrade", 100,
	defclsyspri, 1, INT_MAX, TASKQ_DYNAMIC \| TASKQ_THREADS_CPU_PCT);
	}

	/*
	* Opposite of spa_activate().
	*/
	static void
	spa_deactivate(spa_t *spa)
	{
	ASSERT(spa->spa_sync_on == B_FALSE);
	ASSERT(spa->spa_dsl_pool == NULL);
	ASSERT(spa->spa_root_vdev == NULL);
	ASSERT(spa->spa_async_zio_root == NULL);
	ASSERT(spa->spa_state != POOL_STATE_UNINITIALIZED);

	spa_evicting_os_wait(spa);

	if (spa->spa_zvol_taskq) {
	taskq_destroy(spa->spa_zvol_taskq);
	spa->spa_zvol_taskq = NULL;
	}

	if (spa->spa_prefetch_taskq) {
	taskq_destroy(spa->spa_prefetch_taskq);
	spa->spa_prefetch_taskq = NULL;
	}

	if (spa->spa_upgrade_taskq) {
	taskq_destroy(spa->spa_upgrade_taskq);
	spa->spa_upgrade_taskq = NULL;
	}

	txg_list_destroy(&spa->spa_vdev_txg_list);

	list_destroy(&spa->spa_config_dirty_list);
	list_destroy(&spa->spa_evicting_os_list);
	list_destroy(&spa->spa_state_dirty_list);

	taskq_cancel_id(system_delay_taskq, spa->spa_deadman_tqid);

	for (int t = 0; t < ZIO_TYPES; t++) {
	for (int q = 0; q < ZIO_TASKQ_TYPES; q++) {
	spa_taskqs_fini(spa, t, q);
	}
	}

	for (size_t i = 0; i < TXG_SIZE; i++) {
	ASSERT3P(spa->spa_txg_zio[i], !=, NULL);
	VERIFY0(zio_wait(spa->spa_txg_zio[i]));
	spa->spa_txg_zio[i] = NULL;
	}

	metaslab_class_destroy(spa->spa_normal_class);
	spa->spa_normal_class = NULL;

	metaslab_class_destroy(spa->spa_log_class);
	spa->spa_log_class = NULL;

	metaslab_class_destroy(spa->spa_embedded_log_class);
	spa->spa_embedded_log_class = NULL;

	metaslab_class_destroy(spa->spa_special_class);
	spa->spa_special_class = NULL;

	metaslab_class_destroy(spa->spa_dedup_class);
	spa->spa_dedup_class = NULL;

	/*
	* If this was part of an import or the open otherwise failed, we may
	* still have errors left in the queues. Empty them just in case.
	*/
	spa_errlog_drain(spa);
	avl_destroy(&spa->spa_errlist_scrub);
	avl_destroy(&spa->spa_errlist_last);

	spa_keystore_fini(&spa->spa_keystore);

	spa->spa_state = POOL_STATE_UNINITIALIZED;

	mutex_enter(&spa->spa_proc_lock);
	if (spa->spa_proc_state != SPA_PROC_NONE) {
	ASSERT(spa->spa_proc_state == SPA_PROC_ACTIVE);
	spa->spa_proc_state = SPA_PROC_DEACTIVATE;
	cv_broadcast(&spa->spa_proc_cv);
	while (spa->spa_proc_state == SPA_PROC_DEACTIVATE) {
	ASSERT(spa->spa_proc != &p0);
	cv_wait(&spa->spa_proc_cv, &spa->spa_proc_lock);
	}
	ASSERT(spa->spa_proc_state == SPA_PROC_GONE);
	spa->spa_proc_state = SPA_PROC_NONE;
	}
	ASSERT(spa->spa_proc == &p0);
	mutex_exit(&spa->spa_proc_lock);

	/*
	* We want to make sure spa_thread() has actually exited the ZFS
	* module, so that the module can't be unloaded out from underneath
	* it.
	*/
	if (spa->spa_did != 0) {
	thread_join(spa->spa_did);
	spa->spa_did = 0;
	}
	}

	/*
	* Verify a pool configuration, and construct the vdev tree appropriately. This
	* will create all the necessary vdevs in the appropriate layout, with each vdev
	* in the CLOSED state. This will prep the pool before open/creation/import.
	* All vdev validation is done by the vdev_alloc() routine.
	*/
	int
	spa_config_parse(spa_t spa, vdev_t vdp, nvlist_t nv, vdev_t *parent,
	uint_t id, int atype)
	{
	nvlist_t **child;
	uint_t children;
	int error;

	if ((error = vdev_alloc(spa, vdp, nv, parent, id, atype)) != 0)
	return (error);

	if ((*vdp)->vdev_ops->vdev_op_leaf)
	return (0);

	error = nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children);

	if (error == ENOENT)
	return (0);

	if (error) {
	vdev_free(*vdp);
	*vdp = NULL;
	return (SET_ERROR(EINVAL));
	}

	for (int c = 0; c < children; c++) {
	vdev_t *vd;
	if ((error = spa_config_parse(spa, &vd, child[c], *vdp, c,
	atype)) != 0) {
	vdev_free(*vdp);
	*vdp = NULL;
	return (error);
	}
	}

	ASSERT(*vdp != NULL);

	return (0);
	}

	static boolean_t
	spa_should_flush_logs_on_unload(spa_t *spa)
	{
	if (!spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP))
	return (B_FALSE);

	if (!spa_writeable(spa))
	return (B_FALSE);

	if (!spa->spa_sync_on)
	return (B_FALSE);

	if (spa_state(spa) != POOL_STATE_EXPORTED)
	return (B_FALSE);

	if (zfs_keep_log_spacemaps_at_export)
	return (B_FALSE);

	return (B_TRUE);
	}

	/*
	* Opens a transaction that will set the flag that will instruct
	* spa_sync to attempt to flush all the metaslabs for that txg.
	*/
	static void
	spa_unload_log_sm_flush_all(spa_t *spa)
	{
	dmu_tx_t *tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir);
	VERIFY0(dmu_tx_assign(tx, TXG_WAIT));

	ASSERT3U(spa->spa_log_flushall_txg, ==, 0);
	spa->spa_log_flushall_txg = dmu_tx_get_txg(tx);

	dmu_tx_commit(tx);
	txg_wait_synced(spa_get_dsl(spa), spa->spa_log_flushall_txg);
	}

	static void
	spa_unload_log_sm_metadata(spa_t *spa)
	{
	void *cookie = NULL;
	spa_log_sm_t *sls;
	while ((sls = avl_destroy_nodes(&spa->spa_sm_logs_by_txg,
	&cookie)) != NULL) {
	VERIFY0(sls->sls_mscount);
	kmem_free(sls, sizeof (spa_log_sm_t));
	}

	for (log_summary_entry_t *e = list_head(&spa->spa_log_summary);
	e != NULL; e = list_head(&spa->spa_log_summary)) {
	VERIFY0(e->lse_mscount);
	list_remove(&spa->spa_log_summary, e);
	kmem_free(e, sizeof (log_summary_entry_t));
	}

	spa->spa_unflushed_stats.sus_nblocks = 0;
	spa->spa_unflushed_stats.sus_memused = 0;
	spa->spa_unflushed_stats.sus_blocklimit = 0;
	}

	static void
	spa_destroy_aux_threads(spa_t *spa)
	{
	if (spa->spa_condense_zthr != NULL) {
	zthr_destroy(spa->spa_condense_zthr);
	spa->spa_condense_zthr = NULL;
	}
	if (spa->spa_checkpoint_discard_zthr != NULL) {
	zthr_destroy(spa->spa_checkpoint_discard_zthr);
	spa->spa_checkpoint_discard_zthr = NULL;
	}
	if (spa->spa_livelist_delete_zthr != NULL) {
	zthr_destroy(spa->spa_livelist_delete_zthr);
	spa->spa_livelist_delete_zthr = NULL;
	}
	if (spa->spa_livelist_condense_zthr != NULL) {
	zthr_destroy(spa->spa_livelist_condense_zthr);
	spa->spa_livelist_condense_zthr = NULL;
	}
	}

	/*
	* Opposite of spa_load().
	*/
	static void
	spa_unload(spa_t *spa)
	{
	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	ASSERT(spa_state(spa) != POOL_STATE_UNINITIALIZED);

	spa_import_progress_remove(spa_guid(spa));
	spa_load_note(spa, "UNLOADING");

	spa_wake_waiters(spa);

	/*
	* If the log space map feature is enabled and the pool is getting
	* exported (but not destroyed), we want to spend some time flushing
	* as many metaslabs as we can in an attempt to destroy log space
	* maps and save import time.
	*/
	if (spa_should_flush_logs_on_unload(spa))
	spa_unload_log_sm_flush_all(spa);

	/*
	* Stop async tasks.
	*/
	spa_async_suspend(spa);

	if (spa->spa_root_vdev) {
	vdev_t *root_vdev = spa->spa_root_vdev;
	vdev_initialize_stop_all(root_vdev, VDEV_INITIALIZE_ACTIVE);
	vdev_trim_stop_all(root_vdev, VDEV_TRIM_ACTIVE);
	vdev_autotrim_stop_all(spa);
	vdev_rebuild_stop_all(spa);
	}

	/*
	* Stop syncing.
	*/
	if (spa->spa_sync_on) {
	txg_sync_stop(spa->spa_dsl_pool);
	spa->spa_sync_on = B_FALSE;
	}

	/*
	* This ensures that there is no async metaslab prefetching
	* while we attempt to unload the spa.
	*/
	if (spa->spa_root_vdev != NULL) {
	for (int c = 0; c < spa->spa_root_vdev->vdev_children; c++) {
	vdev_t *vc = spa->spa_root_vdev->vdev_child[c];
	if (vc->vdev_mg != NULL)
	taskq_wait(vc->vdev_mg->mg_taskq);
	}
	}

	if (spa->spa_mmp.mmp_thread)
	mmp_thread_stop(spa);

	/*
	* Wait for any outstanding async I/O to complete.
	*/
	if (spa->spa_async_zio_root != NULL) {
	for (int i = 0; i < max_ncpus; i++)
	(void) zio_wait(spa->spa_async_zio_root[i]);
	kmem_free(spa->spa_async_zio_root, max_ncpus * sizeof (void *));
	spa->spa_async_zio_root = NULL;
	}

	if (spa->spa_vdev_removal != NULL) {
	spa_vdev_removal_destroy(spa->spa_vdev_removal);
	spa->spa_vdev_removal = NULL;
	}

	spa_destroy_aux_threads(spa);

	spa_condense_fini(spa);

	bpobj_close(&spa->spa_deferred_bpobj);

	spa_config_enter(spa, SCL_ALL, spa, RW_WRITER);

	/*
	* Close all vdevs.
	*/
	if (spa->spa_root_vdev)
	vdev_free(spa->spa_root_vdev);
	ASSERT(spa->spa_root_vdev == NULL);

	/*
	* Close the dsl pool.
	*/
	if (spa->spa_dsl_pool) {
	dsl_pool_close(spa->spa_dsl_pool);
	spa->spa_dsl_pool = NULL;
	spa->spa_meta_objset = NULL;
	}

	ddt_unload(spa);
	spa_unload_log_sm_metadata(spa);

	/*
	* Drop and purge level 2 cache
	*/
	spa_l2cache_drop(spa);

	for (int i = 0; i < spa->spa_spares.sav_count; i++)
	vdev_free(spa->spa_spares.sav_vdevs[i]);
	if (spa->spa_spares.sav_vdevs) {
	kmem_free(spa->spa_spares.sav_vdevs,
	spa->spa_spares.sav_count * sizeof (void *));
	spa->spa_spares.sav_vdevs = NULL;
	}
	if (spa->spa_spares.sav_config) {
	nvlist_free(spa->spa_spares.sav_config);
	spa->spa_spares.sav_config = NULL;
	}
	spa->spa_spares.sav_count = 0;

	for (int i = 0; i < spa->spa_l2cache.sav_count; i++) {
	vdev_clear_stats(spa->spa_l2cache.sav_vdevs[i]);
	vdev_free(spa->spa_l2cache.sav_vdevs[i]);
	}
	if (spa->spa_l2cache.sav_vdevs) {
	kmem_free(spa->spa_l2cache.sav_vdevs,
	spa->spa_l2cache.sav_count * sizeof (void *));
	spa->spa_l2cache.sav_vdevs = NULL;
	}
	if (spa->spa_l2cache.sav_config) {
	nvlist_free(spa->spa_l2cache.sav_config);
	spa->spa_l2cache.sav_config = NULL;
	}
	spa->spa_l2cache.sav_count = 0;

	spa->spa_async_suspended = 0;

	spa->spa_indirect_vdevs_loaded = B_FALSE;

	if (spa->spa_comment != NULL) {
	spa_strfree(spa->spa_comment);
	spa->spa_comment = NULL;
	}
	if (spa->spa_compatibility != NULL) {
	spa_strfree(spa->spa_compatibility);
	spa->spa_compatibility = NULL;
	}

	spa_config_exit(spa, SCL_ALL, spa);
	}

	/*
	* Load (or re-load) the current list of vdevs describing the active spares for
	* this pool. When this is called, we have some form of basic information in
	* 'spa_spares.sav_config'. We parse this into vdevs, try to open them, and
	* then re-generate a more complete list including status information.
	*/
	void
	spa_load_spares(spa_t *spa)
	{
	nvlist_t **spares;
	uint_t nspares;
	int i;
	vdev_t vd, tvd;

	#ifndef _KERNEL
	/*
	* zdb opens both the current state of the pool and the
	* checkpointed state (if present), with a different spa_t.
	*
	* As spare vdevs are shared among open pools, we skip loading
	* them when we load the checkpointed state of the pool.
	*/
	if (!spa_writeable(spa))
	return;
	#endif

	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	/*
	* First, close and free any existing spare vdevs.
	*/
	for (i = 0; i < spa->spa_spares.sav_count; i++) {
	vd = spa->spa_spares.sav_vdevs[i];

	/* Undo the call to spa_activate() below */
	if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid,
	B_FALSE)) != NULL && tvd->vdev_isspare)
	spa_spare_remove(tvd);
	vdev_close(vd);
	vdev_free(vd);
	}

	if (spa->spa_spares.sav_vdevs)
	kmem_free(spa->spa_spares.sav_vdevs,
	spa->spa_spares.sav_count * sizeof (void *));

	if (spa->spa_spares.sav_config == NULL)
	nspares = 0;
	else
	VERIFY0(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config,
	ZPOOL_CONFIG_SPARES, &spares, &nspares));

	spa->spa_spares.sav_count = (int)nspares;
	spa->spa_spares.sav_vdevs = NULL;

	if (nspares == 0)
	return;

	/*
	* Construct the array of vdevs, opening them to get status in the
	* process. For each spare, there is potentially two different vdev_t
	* structures associated with it: one in the list of spares (used only
	* for basic validation purposes) and one in the active vdev
	* configuration (if it's spared in). During this phase we open and
	* validate each vdev on the spare list. If the vdev also exists in the
	* active configuration, then we also mark this vdev as an active spare.
	*/
	spa->spa_spares.sav_vdevs = kmem_zalloc(nspares * sizeof (void *),
	KM_SLEEP);
	for (i = 0; i < spa->spa_spares.sav_count; i++) {
	VERIFY(spa_config_parse(spa, &vd, spares[i], NULL, 0,
	VDEV_ALLOC_SPARE) == 0);
	ASSERT(vd != NULL);

	spa->spa_spares.sav_vdevs[i] = vd;

	if ((tvd = spa_lookup_by_guid(spa, vd->vdev_guid,
	B_FALSE)) != NULL) {
	if (!tvd->vdev_isspare)
	spa_spare_add(tvd);

	/*
	* We only mark the spare active if we were successfully
	* able to load the vdev. Otherwise, importing a pool
	* with a bad active spare would result in strange
	* behavior, because multiple pool would think the spare
	* is actively in use.
	*
	* There is a vulnerability here to an equally bizarre
	* circumstance, where a dead active spare is later
	* brought back to life (onlined or otherwise). Given
	* the rarity of this scenario, and the extra complexity
	* it adds, we ignore the possibility.
	*/
	if (!vdev_is_dead(tvd))
	spa_spare_activate(tvd);
	}

	vd->vdev_top = vd;
	vd->vdev_aux = &spa->spa_spares;

	if (vdev_open(vd) != 0)
	continue;

	if (vdev_validate_aux(vd) == 0)
	spa_spare_add(vd);
	}

	/*
	* Recompute the stashed list of spares, with status information
	* this time.
	*/
	fnvlist_remove(spa->spa_spares.sav_config, ZPOOL_CONFIG_SPARES);

	spares = kmem_alloc(spa->spa_spares.sav_count * sizeof (void *),
	KM_SLEEP);
	for (i = 0; i < spa->spa_spares.sav_count; i++)
	spares[i] = vdev_config_generate(spa,
	spa->spa_spares.sav_vdevs[i], B_TRUE, VDEV_CONFIG_SPARE);
	fnvlist_add_nvlist_array(spa->spa_spares.sav_config,
	- ZPOOL_CONFIG_SPARES, spares, spa->spa_spares.sav_count);
	+ ZPOOL_CONFIG_SPARES, (const nvlist_t * const *)spares,
	+ spa->spa_spares.sav_count);
	for (i = 0; i < spa->spa_spares.sav_count; i++)
	nvlist_free(spares[i]);
	kmem_free(spares, spa->spa_spares.sav_count * sizeof (void *));
	}

	/*
	* Load (or re-load) the current list of vdevs describing the active l2cache for
	* this pool. When this is called, we have some form of basic information in
	* 'spa_l2cache.sav_config'. We parse this into vdevs, try to open them, and
	* then re-generate a more complete list including status information.
	* Devices which are already active have their details maintained, and are
	* not re-opened.
	*/
	void
	spa_load_l2cache(spa_t *spa)
	{
	nvlist_t **l2cache = NULL;
	uint_t nl2cache;
	int i, j, oldnvdevs;
	uint64_t guid;
	vdev_t vd, oldvdevs, *newvdevs;
	spa_aux_vdev_t *sav = &spa->spa_l2cache;

	#ifndef _KERNEL
	/*
	* zdb opens both the current state of the pool and the
	* checkpointed state (if present), with a different spa_t.
	*
	* As L2 caches are part of the ARC which is shared among open
	* pools, we skip loading them when we load the checkpointed
	* state of the pool.
	*/
	if (!spa_writeable(spa))
	return;
	#endif

	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	oldvdevs = sav->sav_vdevs;
	oldnvdevs = sav->sav_count;
	sav->sav_vdevs = NULL;
	sav->sav_count = 0;

	if (sav->sav_config == NULL) {
	nl2cache = 0;
	newvdevs = NULL;
	goto out;
	}

	VERIFY0(nvlist_lookup_nvlist_array(sav->sav_config,
	ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache));
	newvdevs = kmem_alloc(nl2cache * sizeof (void *), KM_SLEEP);

	/*
	* Process new nvlist of vdevs.
	*/
	for (i = 0; i < nl2cache; i++) {
	guid = fnvlist_lookup_uint64(l2cache[i], ZPOOL_CONFIG_GUID);

	newvdevs[i] = NULL;
	for (j = 0; j < oldnvdevs; j++) {
	vd = oldvdevs[j];
	if (vd != NULL && guid == vd->vdev_guid) {
	/*
	* Retain previous vdev for add/remove ops.
	*/
	newvdevs[i] = vd;
	oldvdevs[j] = NULL;
	break;
	}
	}

	if (newvdevs[i] == NULL) {
	/*
	* Create new vdev
	*/
	VERIFY(spa_config_parse(spa, &vd, l2cache[i], NULL, 0,
	VDEV_ALLOC_L2CACHE) == 0);
	ASSERT(vd != NULL);
	newvdevs[i] = vd;

	/*
	* Commit this vdev as an l2cache device,
	* even if it fails to open.
	*/
	spa_l2cache_add(vd);

	vd->vdev_top = vd;
	vd->vdev_aux = sav;

	spa_l2cache_activate(vd);

	if (vdev_open(vd) != 0)
	continue;

	(void) vdev_validate_aux(vd);

	if (!vdev_is_dead(vd))
	l2arc_add_vdev(spa, vd);

	/*
	* Upon cache device addition to a pool or pool
	* creation with a cache device or if the header
	* of the device is invalid we issue an async
	* TRIM command for the whole device which will
	* execute if l2arc_trim_ahead > 0.
	*/
	spa_async_request(spa, SPA_ASYNC_L2CACHE_TRIM);
	}
	}

	sav->sav_vdevs = newvdevs;
	sav->sav_count = (int)nl2cache;

	/*
	* Recompute the stashed list of l2cache devices, with status
	* information this time.
	*/
	fnvlist_remove(sav->sav_config, ZPOOL_CONFIG_L2CACHE);

	if (sav->sav_count > 0)
	l2cache = kmem_alloc(sav->sav_count * sizeof (void *),
	KM_SLEEP);
	for (i = 0; i < sav->sav_count; i++)
	l2cache[i] = vdev_config_generate(spa,
	sav->sav_vdevs[i], B_TRUE, VDEV_CONFIG_L2CACHE);
	- fnvlist_add_nvlist_array(sav->sav_config, ZPOOL_CONFIG_L2CACHE, l2cache,
	- sav->sav_count);
	+ fnvlist_add_nvlist_array(sav->sav_config, ZPOOL_CONFIG_L2CACHE,
	+ (const nvlist_t * const *)l2cache, sav->sav_count);

	out:
	/*
	* Purge vdevs that were dropped
	*/
	for (i = 0; i < oldnvdevs; i++) {
	uint64_t pool;

	vd = oldvdevs[i];
	if (vd != NULL) {
	ASSERT(vd->vdev_isl2cache);

	if (spa_l2cache_exists(vd->vdev_guid, &pool) &&
	pool != 0ULL && l2arc_vdev_present(vd))
	l2arc_remove_vdev(vd);
	vdev_clear_stats(vd);
	vdev_free(vd);
	}
	}

	if (oldvdevs)
	kmem_free(oldvdevs, oldnvdevs * sizeof (void *));

	for (i = 0; i < sav->sav_count; i++)
	nvlist_free(l2cache[i]);
	if (sav->sav_count)
	kmem_free(l2cache, sav->sav_count * sizeof (void *));
	}

	static int
	load_nvlist(spa_t spa, uint64_t obj, nvlist_t *value)
	{
	dmu_buf_t *db;
	char *packed = NULL;
	size_t nvsize = 0;
	int error;
	*value = NULL;

	error = dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db);
	if (error)
	return (error);

	nvsize = (uint64_t )db->db_data;
	dmu_buf_rele(db, FTAG);

	packed = vmem_alloc(nvsize, KM_SLEEP);
	error = dmu_read(spa->spa_meta_objset, obj, 0, nvsize, packed,
	DMU_READ_PREFETCH);
	if (error == 0)
	error = nvlist_unpack(packed, nvsize, value, 0);
	vmem_free(packed, nvsize);

	return (error);
	}

	/*
	* Concrete top-level vdevs that are not missing and are not logs. At every
	* spa_sync we write new uberblocks to at least SPA_SYNC_MIN_VDEVS core tvds.
	*/
	static uint64_t
	spa_healthy_core_tvds(spa_t *spa)
	{
	vdev_t *rvd = spa->spa_root_vdev;
	uint64_t tvds = 0;

	for (uint64_t i = 0; i < rvd->vdev_children; i++) {
	vdev_t *vd = rvd->vdev_child[i];
	if (vd->vdev_islog)
	continue;
	if (vdev_is_concrete(vd) && !vdev_is_dead(vd))
	tvds++;
	}

	return (tvds);
	}

	/*
	* Checks to see if the given vdev could not be opened, in which case we post a
	* sysevent to notify the autoreplace code that the device has been removed.
	*/
	static void
	spa_check_removed(vdev_t *vd)
	{
	for (uint64_t c = 0; c < vd->vdev_children; c++)
	spa_check_removed(vd->vdev_child[c]);

	if (vd->vdev_ops->vdev_op_leaf && vdev_is_dead(vd) &&
	vdev_is_concrete(vd)) {
	zfs_post_autoreplace(vd->vdev_spa, vd);
	spa_event_notify(vd->vdev_spa, vd, NULL, ESC_ZFS_VDEV_CHECK);
	}
	}

	static int
	spa_check_for_missing_logs(spa_t *spa)
	{
	vdev_t *rvd = spa->spa_root_vdev;

	/*
	* If we're doing a normal import, then build up any additional
	* diagnostic information about missing log devices.
	* We'll pass this up to the user for further processing.
	*/
	if (!(spa->spa_import_flags & ZFS_IMPORT_MISSING_LOG)) {
	nvlist_t *child, nv;
	uint64_t idx = 0;

	child = kmem_alloc(rvd->vdev_children * sizeof (nvlist_t *),
	KM_SLEEP);
	nv = fnvlist_alloc();

	for (uint64_t c = 0; c < rvd->vdev_children; c++) {
	vdev_t *tvd = rvd->vdev_child[c];

	/*
	* We consider a device as missing only if it failed
	* to open (i.e. offline or faulted is not considered
	* as missing).
	*/
	if (tvd->vdev_islog &&
	tvd->vdev_state == VDEV_STATE_CANT_OPEN) {
	child[idx++] = vdev_config_generate(spa, tvd,
	B_FALSE, VDEV_CONFIG_MISSING);
	}
	}

	if (idx > 0) {
	- fnvlist_add_nvlist_array(nv,
	- ZPOOL_CONFIG_CHILDREN, child, idx);
	+ fnvlist_add_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	+ (const nvlist_t * const *)child, idx);
	fnvlist_add_nvlist(spa->spa_load_info,
	ZPOOL_CONFIG_MISSING_DEVICES, nv);

	for (uint64_t i = 0; i < idx; i++)
	nvlist_free(child[i]);
	}
	nvlist_free(nv);
	kmem_free(child, rvd->vdev_children * sizeof (char **));

	if (idx > 0) {
	spa_load_failed(spa, "some log devices are missing");
	vdev_dbgmsg_print_tree(rvd, 2);
	return (SET_ERROR(ENXIO));
	}
	} else {
	for (uint64_t c = 0; c < rvd->vdev_children; c++) {
	vdev_t *tvd = rvd->vdev_child[c];

	if (tvd->vdev_islog &&
	tvd->vdev_state == VDEV_STATE_CANT_OPEN) {
	spa_set_log_state(spa, SPA_LOG_CLEAR);
	spa_load_note(spa, "some log devices are "
	"missing, ZIL is dropped.");
	vdev_dbgmsg_print_tree(rvd, 2);
	break;
	}
	}
	}

	return (0);
	}

	/*
	* Check for missing log devices
	*/
	static boolean_t
	spa_check_logs(spa_t *spa)
	{
	boolean_t rv = B_FALSE;
	dsl_pool_t *dp = spa_get_dsl(spa);

	switch (spa->spa_log_state) {
	default:
	break;
	case SPA_LOG_MISSING:
	/* need to recheck in case slog has been restored */
	case SPA_LOG_UNKNOWN:
	rv = (dmu_objset_find_dp(dp, dp->dp_root_dir_obj,
	zil_check_log_chain, NULL, DS_FIND_CHILDREN) != 0);
	if (rv)
	spa_set_log_state(spa, SPA_LOG_MISSING);
	break;
	}
	return (rv);
	}

	/*
	* Passivate any log vdevs (note, does not apply to embedded log metaslabs).
	*/
	static boolean_t
	spa_passivate_log(spa_t *spa)
	{
	vdev_t *rvd = spa->spa_root_vdev;
	boolean_t slog_found = B_FALSE;

	ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER));

	for (int c = 0; c < rvd->vdev_children; c++) {
	vdev_t *tvd = rvd->vdev_child[c];

	if (tvd->vdev_islog) {
	ASSERT3P(tvd->vdev_log_mg, ==, NULL);
	metaslab_group_passivate(tvd->vdev_mg);
	slog_found = B_TRUE;
	}
	}

	return (slog_found);
	}

	/*
	* Activate any log vdevs (note, does not apply to embedded log metaslabs).
	*/
	static void
	spa_activate_log(spa_t *spa)
	{
	vdev_t *rvd = spa->spa_root_vdev;

	ASSERT(spa_config_held(spa, SCL_ALLOC, RW_WRITER));

	for (int c = 0; c < rvd->vdev_children; c++) {
	vdev_t *tvd = rvd->vdev_child[c];

	if (tvd->vdev_islog) {
	ASSERT3P(tvd->vdev_log_mg, ==, NULL);
	metaslab_group_activate(tvd->vdev_mg);
	}
	}
	}

	int
	spa_reset_logs(spa_t *spa)
	{
	int error;

	error = dmu_objset_find(spa_name(spa), zil_reset,
	NULL, DS_FIND_CHILDREN);
	if (error == 0) {
	/*
	* We successfully offlined the log device, sync out the
	* current txg so that the "stubby" block can be removed
	* by zil_sync().
	*/
	txg_wait_synced(spa->spa_dsl_pool, 0);
	}
	return (error);
	}

	static void
	spa_aux_check_removed(spa_aux_vdev_t *sav)
	{
	for (int i = 0; i < sav->sav_count; i++)
	spa_check_removed(sav->sav_vdevs[i]);
	}

	void
	spa_claim_notify(zio_t *zio)
	{
	spa_t *spa = zio->io_spa;

	if (zio->io_error)
	return;

	mutex_enter(&spa->spa_props_lock); /* any mutex will do */
	if (spa->spa_claim_max_txg < zio->io_bp->blk_birth)
	spa->spa_claim_max_txg = zio->io_bp->blk_birth;
	mutex_exit(&spa->spa_props_lock);
	}

	typedef struct spa_load_error {
	uint64_t sle_meta_count;
	uint64_t sle_data_count;
	} spa_load_error_t;

	static void
	spa_load_verify_done(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;
	spa_load_error_t *sle = zio->io_private;
	dmu_object_type_t type = BP_GET_TYPE(bp);
	int error = zio->io_error;
	spa_t *spa = zio->io_spa;

	abd_free(zio->io_abd);
	if (error) {
	if ((BP_GET_LEVEL(bp) != 0 \|\| DMU_OT_IS_METADATA(type)) &&
	type != DMU_OT_INTENT_LOG)
	atomic_inc_64(&sle->sle_meta_count);
	else
	atomic_inc_64(&sle->sle_data_count);
	}

	mutex_enter(&spa->spa_scrub_lock);
	spa->spa_load_verify_bytes -= BP_GET_PSIZE(bp);
	cv_broadcast(&spa->spa_scrub_io_cv);
	mutex_exit(&spa->spa_scrub_lock);
	}

	/*
	* Maximum number of inflight bytes is the log2 fraction of the arc size.
	* By default, we set it to 1/16th of the arc.
	*/
	int spa_load_verify_shift = 4;
	int spa_load_verify_metadata = B_TRUE;
	int spa_load_verify_data = B_TRUE;

	/ARGSUSED/
	static int
	spa_load_verify_cb(spa_t spa, zilog_t zilog, const blkptr_t *bp,
	const zbookmark_phys_t zb, const dnode_phys_t dnp, void *arg)
	{
	if (zb->zb_level == ZB_DNODE_LEVEL \|\| BP_IS_HOLE(bp) \|\|
	BP_IS_EMBEDDED(bp) \|\| BP_IS_REDACTED(bp))
	return (0);
	/*
	* Note: normally this routine will not be called if
	* spa_load_verify_metadata is not set. However, it may be useful
	* to manually set the flag after the traversal has begun.
	*/
	if (!spa_load_verify_metadata)
	return (0);
	if (!BP_IS_METADATA(bp) && !spa_load_verify_data)
	return (0);

	uint64_t maxinflight_bytes =
	arc_target_bytes() >> spa_load_verify_shift;
	zio_t *rio = arg;
	size_t size = BP_GET_PSIZE(bp);

	mutex_enter(&spa->spa_scrub_lock);
	while (spa->spa_load_verify_bytes >= maxinflight_bytes)
	cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock);
	spa->spa_load_verify_bytes += size;
	mutex_exit(&spa->spa_scrub_lock);

	zio_nowait(zio_read(rio, spa, bp, abd_alloc_for_io(size, B_FALSE), size,
	spa_load_verify_done, rio->io_private, ZIO_PRIORITY_SCRUB,
	ZIO_FLAG_SPECULATIVE \| ZIO_FLAG_CANFAIL \|
	ZIO_FLAG_SCRUB \| ZIO_FLAG_RAW, zb));
	return (0);
	}

	/* ARGSUSED */
	static int
	verify_dataset_name_len(dsl_pool_t dp, dsl_dataset_t ds, void *arg)
	{
	if (dsl_dataset_namelen(ds) >= ZFS_MAX_DATASET_NAME_LEN)
	return (SET_ERROR(ENAMETOOLONG));

	return (0);
	}

	static int
	spa_load_verify(spa_t *spa)
	{
	zio_t *rio;
	spa_load_error_t sle = { 0 };
	zpool_load_policy_t policy;
	boolean_t verify_ok = B_FALSE;
	int error = 0;

	zpool_get_load_policy(spa->spa_config, &policy);

	if (policy.zlp_rewind & ZPOOL_NEVER_REWIND)
	return (0);

	dsl_pool_config_enter(spa->spa_dsl_pool, FTAG);
	error = dmu_objset_find_dp(spa->spa_dsl_pool,
	spa->spa_dsl_pool->dp_root_dir_obj, verify_dataset_name_len, NULL,
	DS_FIND_CHILDREN);
	dsl_pool_config_exit(spa->spa_dsl_pool, FTAG);
	if (error != 0)
	return (error);

	rio = zio_root(spa, NULL, &sle,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_SPECULATIVE);

	if (spa_load_verify_metadata) {
	if (spa->spa_extreme_rewind) {
	spa_load_note(spa, "performing a complete scan of the "
	"pool since extreme rewind is on. This may take "
	"a very long time.\n (spa_load_verify_data=%u, "
	"spa_load_verify_metadata=%u)",
	spa_load_verify_data, spa_load_verify_metadata);
	}

	error = traverse_pool(spa, spa->spa_verify_min_txg,
	TRAVERSE_PRE \| TRAVERSE_PREFETCH_METADATA \|
	TRAVERSE_NO_DECRYPT, spa_load_verify_cb, rio);
	}

	(void) zio_wait(rio);
	ASSERT0(spa->spa_load_verify_bytes);

	spa->spa_load_meta_errors = sle.sle_meta_count;
	spa->spa_load_data_errors = sle.sle_data_count;

	if (sle.sle_meta_count != 0 \|\| sle.sle_data_count != 0) {
	spa_load_note(spa, "spa_load_verify found %llu metadata errors "
	"and %llu data errors", (u_longlong_t)sle.sle_meta_count,
	(u_longlong_t)sle.sle_data_count);
	}

	if (spa_load_verify_dryrun \|\|
	(!error && sle.sle_meta_count <= policy.zlp_maxmeta &&
	sle.sle_data_count <= policy.zlp_maxdata)) {
	int64_t loss = 0;

	verify_ok = B_TRUE;
	spa->spa_load_txg = spa->spa_uberblock.ub_txg;
	spa->spa_load_txg_ts = spa->spa_uberblock.ub_timestamp;

	loss = spa->spa_last_ubsync_txg_ts - spa->spa_load_txg_ts;
	fnvlist_add_uint64(spa->spa_load_info, ZPOOL_CONFIG_LOAD_TIME,
	spa->spa_load_txg_ts);
	fnvlist_add_int64(spa->spa_load_info, ZPOOL_CONFIG_REWIND_TIME,
	loss);
	fnvlist_add_uint64(spa->spa_load_info,
	ZPOOL_CONFIG_LOAD_DATA_ERRORS, sle.sle_data_count);
	} else {
	spa->spa_load_max_txg = spa->spa_uberblock.ub_txg;
	}

	if (spa_load_verify_dryrun)
	return (0);

	if (error) {
	if (error != ENXIO && error != EIO)
	error = SET_ERROR(EIO);
	return (error);
	}

	return (verify_ok ? 0 : EIO);
	}

	/*
	* Find a value in the pool props object.
	*/
	static void
	spa_prop_find(spa_t spa, zpool_prop_t prop, uint64_t val)
	{
	(void) zap_lookup(spa->spa_meta_objset, spa->spa_pool_props_object,
	zpool_prop_to_name(prop), sizeof (uint64_t), 1, val);
	}

	/*
	* Find a value in the pool directory object.
	*/
	static int
	spa_dir_prop(spa_t spa, const char name, uint64_t *val, boolean_t log_enoent)
	{
	int error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
	name, sizeof (uint64_t), 1, val);

	if (error != 0 && (error != ENOENT \|\| log_enoent)) {
	spa_load_failed(spa, "couldn't get '%s' value in MOS directory "
	"[error=%d]", name, error);
	}

	return (error);
	}

	static int
	spa_vdev_err(vdev_t *vdev, vdev_aux_t aux, int err)
	{
	vdev_set_state(vdev, B_TRUE, VDEV_STATE_CANT_OPEN, aux);
	return (SET_ERROR(err));
	}

	boolean_t
	spa_livelist_delete_check(spa_t *spa)
	{
	return (spa->spa_livelists_to_delete != 0);
	}

	/* ARGSUSED */
	static boolean_t
	spa_livelist_delete_cb_check(void arg, zthr_t z)
	{
	spa_t *spa = arg;
	return (spa_livelist_delete_check(spa));
	}

	static int
	delete_blkptr_cb(void arg, const blkptr_t bp, dmu_tx_t *tx)
	{
	spa_t *spa = arg;
	zio_free(spa, tx->tx_txg, bp);
	dsl_dir_diduse_space(tx->tx_pool->dp_free_dir, DD_USED_HEAD,
	-bp_get_dsize_sync(spa, bp),
	-BP_GET_PSIZE(bp), -BP_GET_UCSIZE(bp), tx);
	return (0);
	}

	static int
	dsl_get_next_livelist_obj(objset_t os, uint64_t zap_obj, uint64_t llp)
	{
	int err;
	zap_cursor_t zc;
	zap_attribute_t za;
	zap_cursor_init(&zc, os, zap_obj);
	err = zap_cursor_retrieve(&zc, &za);
	zap_cursor_fini(&zc);
	if (err == 0)
	*llp = za.za_first_integer;
	return (err);
	}

	/*
	* Components of livelist deletion that must be performed in syncing
	* context: freeing block pointers and updating the pool-wide data
	* structures to indicate how much work is left to do
	*/
	typedef struct sublist_delete_arg {
	spa_t *spa;
	dsl_deadlist_t *ll;
	uint64_t key;
	bplist_t *to_free;
	} sublist_delete_arg_t;

	static void
	sublist_delete_sync(void arg, dmu_tx_t tx)
	{
	sublist_delete_arg_t *sda = arg;
	spa_t *spa = sda->spa;
	dsl_deadlist_t *ll = sda->ll;
	uint64_t key = sda->key;
	bplist_t *to_free = sda->to_free;

	bplist_iterate(to_free, delete_blkptr_cb, spa, tx);
	dsl_deadlist_remove_entry(ll, key, tx);
	}

	typedef struct livelist_delete_arg {
	spa_t *spa;
	uint64_t ll_obj;
	uint64_t zap_obj;
	} livelist_delete_arg_t;

	static void
	livelist_delete_sync(void arg, dmu_tx_t tx)
	{
	livelist_delete_arg_t *lda = arg;
	spa_t *spa = lda->spa;
	uint64_t ll_obj = lda->ll_obj;
	uint64_t zap_obj = lda->zap_obj;
	objset_t *mos = spa->spa_meta_objset;
	uint64_t count;

	/* free the livelist and decrement the feature count */
	VERIFY0(zap_remove_int(mos, zap_obj, ll_obj, tx));
	dsl_deadlist_free(mos, ll_obj, tx);
	spa_feature_decr(spa, SPA_FEATURE_LIVELIST, tx);
	VERIFY0(zap_count(mos, zap_obj, &count));
	if (count == 0) {
	/* no more livelists to delete */
	VERIFY0(zap_remove(mos, DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_DELETED_CLONES, tx));
	VERIFY0(zap_destroy(mos, zap_obj, tx));
	spa->spa_livelists_to_delete = 0;
	spa_notify_waiters(spa);
	}
	}

	/*
	* Load in the value for the livelist to be removed and open it. Then,
	* load its first sublist and determine which block pointers should actually
	* be freed. Then, call a synctask which performs the actual frees and updates
	* the pool-wide livelist data.
	*/
	/* ARGSUSED */
	static void
	spa_livelist_delete_cb(void arg, zthr_t z)
	{
	spa_t *spa = arg;
	uint64_t ll_obj = 0, count;
	objset_t *mos = spa->spa_meta_objset;
	uint64_t zap_obj = spa->spa_livelists_to_delete;
	/*
	* Determine the next livelist to delete. This function should only
	* be called if there is at least one deleted clone.
	*/
	VERIFY0(dsl_get_next_livelist_obj(mos, zap_obj, &ll_obj));
	VERIFY0(zap_count(mos, ll_obj, &count));
	if (count > 0) {
	dsl_deadlist_t *ll;
	dsl_deadlist_entry_t *dle;
	bplist_t to_free;
	ll = kmem_zalloc(sizeof (dsl_deadlist_t), KM_SLEEP);
	dsl_deadlist_open(ll, mos, ll_obj);
	dle = dsl_deadlist_first(ll);
	ASSERT3P(dle, !=, NULL);
	bplist_create(&to_free);
	int err = dsl_process_sub_livelist(&dle->dle_bpobj, &to_free,
	z, NULL);
	if (err == 0) {
	sublist_delete_arg_t sync_arg = {
	.spa = spa,
	.ll = ll,
	.key = dle->dle_mintxg,
	.to_free = &to_free
	};
	zfs_dbgmsg("deleting sublist (id %llu) from"
	" livelist %llu, %lld remaining",
	(u_longlong_t)dle->dle_bpobj.bpo_object,
	(u_longlong_t)ll_obj, (longlong_t)count - 1);
	VERIFY0(dsl_sync_task(spa_name(spa), NULL,
	sublist_delete_sync, &sync_arg, 0,
	ZFS_SPACE_CHECK_DESTROY));
	} else {
	VERIFY3U(err, ==, EINTR);
	}
	bplist_clear(&to_free);
	bplist_destroy(&to_free);
	dsl_deadlist_close(ll);
	kmem_free(ll, sizeof (dsl_deadlist_t));
	} else {
	livelist_delete_arg_t sync_arg = {
	.spa = spa,
	.ll_obj = ll_obj,
	.zap_obj = zap_obj
	};
	zfs_dbgmsg("deletion of livelist %llu completed",
	(u_longlong_t)ll_obj);
	VERIFY0(dsl_sync_task(spa_name(spa), NULL, livelist_delete_sync,
	&sync_arg, 0, ZFS_SPACE_CHECK_DESTROY));
	}
	}

	static void
	spa_start_livelist_destroy_thread(spa_t *spa)
	{
	ASSERT3P(spa->spa_livelist_delete_zthr, ==, NULL);
	spa->spa_livelist_delete_zthr =
	zthr_create("z_livelist_destroy",
	spa_livelist_delete_cb_check, spa_livelist_delete_cb, spa,
	minclsyspri);
	}

	typedef struct livelist_new_arg {
	bplist_t *allocs;
	bplist_t *frees;
	} livelist_new_arg_t;

	static int
	livelist_track_new_cb(void arg, const blkptr_t bp, boolean_t bp_freed,
	dmu_tx_t *tx)
	{
	ASSERT(tx == NULL);
	livelist_new_arg_t *lna = arg;
	if (bp_freed) {
	bplist_append(lna->frees, bp);
	} else {
	bplist_append(lna->allocs, bp);
	zfs_livelist_condense_new_alloc++;
	}
	return (0);
	}

	typedef struct livelist_condense_arg {
	spa_t *spa;
	bplist_t to_keep;
	uint64_t first_size;
	uint64_t next_size;
	} livelist_condense_arg_t;

	static void
	spa_livelist_condense_sync(void arg, dmu_tx_t tx)
	{
	livelist_condense_arg_t *lca = arg;
	spa_t *spa = lca->spa;
	bplist_t new_frees;
	dsl_dataset_t *ds = spa->spa_to_condense.ds;

	/* Have we been cancelled? */
	if (spa->spa_to_condense.cancelled) {
	zfs_livelist_condense_sync_cancel++;
	goto out;
	}

	dsl_deadlist_entry_t *first = spa->spa_to_condense.first;
	dsl_deadlist_entry_t *next = spa->spa_to_condense.next;
	dsl_deadlist_t *ll = &ds->ds_dir->dd_livelist;

	/*
	* It's possible that the livelist was changed while the zthr was
	* running. Therefore, we need to check for new blkptrs in the two
	* entries being condensed and continue to track them in the livelist.
	* Because of the way we handle remapped blkptrs (see dbuf_remap_impl),
	* it's possible that the newly added blkptrs are FREEs or ALLOCs so
	* we need to sort them into two different bplists.
	*/
	uint64_t first_obj = first->dle_bpobj.bpo_object;
	uint64_t next_obj = next->dle_bpobj.bpo_object;
	uint64_t cur_first_size = first->dle_bpobj.bpo_phys->bpo_num_blkptrs;
	uint64_t cur_next_size = next->dle_bpobj.bpo_phys->bpo_num_blkptrs;

	bplist_create(&new_frees);
	livelist_new_arg_t new_bps = {
	.allocs = &lca->to_keep,
	.frees = &new_frees,
	};

	if (cur_first_size > lca->first_size) {
	VERIFY0(livelist_bpobj_iterate_from_nofree(&first->dle_bpobj,
	livelist_track_new_cb, &new_bps, lca->first_size));
	}
	if (cur_next_size > lca->next_size) {
	VERIFY0(livelist_bpobj_iterate_from_nofree(&next->dle_bpobj,
	livelist_track_new_cb, &new_bps, lca->next_size));
	}

	dsl_deadlist_clear_entry(first, ll, tx);
	ASSERT(bpobj_is_empty(&first->dle_bpobj));
	dsl_deadlist_remove_entry(ll, next->dle_mintxg, tx);

	bplist_iterate(&lca->to_keep, dsl_deadlist_insert_alloc_cb, ll, tx);
	bplist_iterate(&new_frees, dsl_deadlist_insert_free_cb, ll, tx);
	bplist_destroy(&new_frees);

	char dsname[ZFS_MAX_DATASET_NAME_LEN];
	dsl_dataset_name(ds, dsname);
	zfs_dbgmsg("txg %llu condensing livelist of %s (id %llu), bpobj %llu "
	"(%llu blkptrs) and bpobj %llu (%llu blkptrs) -> bpobj %llu "
	"(%llu blkptrs)", (u_longlong_t)tx->tx_txg, dsname,
	(u_longlong_t)ds->ds_object, (u_longlong_t)first_obj,
	(u_longlong_t)cur_first_size, (u_longlong_t)next_obj,
	(u_longlong_t)cur_next_size,
	(u_longlong_t)first->dle_bpobj.bpo_object,
	(u_longlong_t)first->dle_bpobj.bpo_phys->bpo_num_blkptrs);
	out:
	dmu_buf_rele(ds->ds_dbuf, spa);
	spa->spa_to_condense.ds = NULL;
	bplist_clear(&lca->to_keep);
	bplist_destroy(&lca->to_keep);
	kmem_free(lca, sizeof (livelist_condense_arg_t));
	spa->spa_to_condense.syncing = B_FALSE;
	}

	static void
	spa_livelist_condense_cb(void arg, zthr_t t)
	{
	while (zfs_livelist_condense_zthr_pause &&
	!(zthr_has_waiters(t) \|\| zthr_iscancelled(t)))
	delay(1);

	spa_t *spa = arg;
	dsl_deadlist_entry_t *first = spa->spa_to_condense.first;
	dsl_deadlist_entry_t *next = spa->spa_to_condense.next;
	uint64_t first_size, next_size;

	livelist_condense_arg_t *lca =
	kmem_alloc(sizeof (livelist_condense_arg_t), KM_SLEEP);
	bplist_create(&lca->to_keep);

	/*
	* Process the livelists (matching FREEs and ALLOCs) in open context
	* so we have minimal work in syncing context to condense.
	*
	* We save bpobj sizes (first_size and next_size) to use later in
	* syncing context to determine if entries were added to these sublists
	* while in open context. This is possible because the clone is still
	* active and open for normal writes and we want to make sure the new,
	* unprocessed blockpointers are inserted into the livelist normally.
	*
	* Note that dsl_process_sub_livelist() both stores the size number of
	* blockpointers and iterates over them while the bpobj's lock held, so
	* the sizes returned to us are consistent which what was actually
	* processed.
	*/
	int err = dsl_process_sub_livelist(&first->dle_bpobj, &lca->to_keep, t,
	&first_size);
	if (err == 0)
	err = dsl_process_sub_livelist(&next->dle_bpobj, &lca->to_keep,
	t, &next_size);

	if (err == 0) {
	while (zfs_livelist_condense_sync_pause &&
	!(zthr_has_waiters(t) \|\| zthr_iscancelled(t)))
	delay(1);

	dmu_tx_t *tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir);
	dmu_tx_mark_netfree(tx);
	dmu_tx_hold_space(tx, 1);
	err = dmu_tx_assign(tx, TXG_NOWAIT \| TXG_NOTHROTTLE);
	if (err == 0) {
	/*
	* Prevent the condense zthr restarting before
	* the synctask completes.
	*/
	spa->spa_to_condense.syncing = B_TRUE;
	lca->spa = spa;
	lca->first_size = first_size;
	lca->next_size = next_size;
	dsl_sync_task_nowait(spa_get_dsl(spa),
	spa_livelist_condense_sync, lca, tx);
	dmu_tx_commit(tx);
	return;
	}
	}
	/*
	* Condensing can not continue: either it was externally stopped or
	* we were unable to assign to a tx because the pool has run out of
	* space. In the second case, we'll just end up trying to condense
	* again in a later txg.
	*/
	ASSERT(err != 0);
	bplist_clear(&lca->to_keep);
	bplist_destroy(&lca->to_keep);
	kmem_free(lca, sizeof (livelist_condense_arg_t));
	dmu_buf_rele(spa->spa_to_condense.ds->ds_dbuf, spa);
	spa->spa_to_condense.ds = NULL;
	if (err == EINTR)
	zfs_livelist_condense_zthr_cancel++;
	}

	/* ARGSUSED */
	/*
	* Check that there is something to condense but that a condense is not
	* already in progress and that condensing has not been cancelled.
	*/
	static boolean_t
	spa_livelist_condense_cb_check(void arg, zthr_t z)
	{
	spa_t *spa = arg;
	if ((spa->spa_to_condense.ds != NULL) &&
	(spa->spa_to_condense.syncing == B_FALSE) &&
	(spa->spa_to_condense.cancelled == B_FALSE)) {
	return (B_TRUE);
	}
	return (B_FALSE);
	}

	static void
	spa_start_livelist_condensing_thread(spa_t *spa)
	{
	spa->spa_to_condense.ds = NULL;
	spa->spa_to_condense.first = NULL;
	spa->spa_to_condense.next = NULL;
	spa->spa_to_condense.syncing = B_FALSE;
	spa->spa_to_condense.cancelled = B_FALSE;

	ASSERT3P(spa->spa_livelist_condense_zthr, ==, NULL);
	spa->spa_livelist_condense_zthr =
	zthr_create("z_livelist_condense",
	spa_livelist_condense_cb_check,
	spa_livelist_condense_cb, spa, minclsyspri);
	}

	static void
	spa_spawn_aux_threads(spa_t *spa)
	{
	ASSERT(spa_writeable(spa));

	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	spa_start_indirect_condensing_thread(spa);
	spa_start_livelist_destroy_thread(spa);
	spa_start_livelist_condensing_thread(spa);

	ASSERT3P(spa->spa_checkpoint_discard_zthr, ==, NULL);
	spa->spa_checkpoint_discard_zthr =
	zthr_create("z_checkpoint_discard",
	spa_checkpoint_discard_thread_check,
	spa_checkpoint_discard_thread, spa, minclsyspri);
	}

	/*
	* Fix up config after a partly-completed split. This is done with the
	* ZPOOL_CONFIG_SPLIT nvlist. Both the splitting pool and the split-off
	* pool have that entry in their config, but only the splitting one contains
	* a list of all the guids of the vdevs that are being split off.
	*
	* This function determines what to do with that list: either rejoin
	* all the disks to the pool, or complete the splitting process. To attempt
	* the rejoin, each disk that is offlined is marked online again, and
	* we do a reopen() call. If the vdev label for every disk that was
	* marked online indicates it was successfully split off (VDEV_AUX_SPLIT_POOL)
	* then we call vdev_split() on each disk, and complete the split.
	*
	* Otherwise we leave the config alone, with all the vdevs in place in
	* the original pool.
	*/
	static void
	spa_try_repair(spa_t spa, nvlist_t config)
	{
	uint_t extracted;
	uint64_t *glist;
	uint_t i, gcount;
	nvlist_t *nvl;
	vdev_t **vd;
	boolean_t attempt_reopen;

	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) != 0)
	return;

	/* check that the config is complete */
	if (nvlist_lookup_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST,
	&glist, &gcount) != 0)
	return;

	vd = kmem_zalloc(gcount * sizeof (vdev_t *), KM_SLEEP);

	/* attempt to online all the vdevs & validate */
	attempt_reopen = B_TRUE;
	for (i = 0; i < gcount; i++) {
	if (glist[i] == 0) /* vdev is hole */
	continue;

	vd[i] = spa_lookup_by_guid(spa, glist[i], B_FALSE);
	if (vd[i] == NULL) {
	/*
	* Don't bother attempting to reopen the disks;
	* just do the split.
	*/
	attempt_reopen = B_FALSE;
	} else {
	/* attempt to re-online it */
	vd[i]->vdev_offline = B_FALSE;
	}
	}

	if (attempt_reopen) {
	vdev_reopen(spa->spa_root_vdev);

	/* check each device to see what state it's in */
	for (extracted = 0, i = 0; i < gcount; i++) {
	if (vd[i] != NULL &&
	vd[i]->vdev_stat.vs_aux != VDEV_AUX_SPLIT_POOL)
	break;
	++extracted;
	}
	}

	/*
	* If every disk has been moved to the new pool, or if we never
	* even attempted to look at them, then we split them off for
	* good.
	*/
	if (!attempt_reopen \|\| gcount == extracted) {
	for (i = 0; i < gcount; i++)
	if (vd[i] != NULL)
	vdev_split(vd[i]);
	vdev_reopen(spa->spa_root_vdev);
	}

	kmem_free(vd, gcount * sizeof (vdev_t *));
	}

	static int
	spa_load(spa_t *spa, spa_load_state_t state, spa_import_type_t type)
	{
	char *ereport = FM_EREPORT_ZFS_POOL;
	int error;

	spa->spa_load_state = state;
	(void) spa_import_progress_set_state(spa_guid(spa),
	spa_load_state(spa));

	gethrestime(&spa->spa_loaded_ts);
	error = spa_load_impl(spa, type, &ereport);

	/*
	* Don't count references from objsets that are already closed
	* and are making their way through the eviction process.
	*/
	spa_evicting_os_wait(spa);
	spa->spa_minref = zfs_refcount_count(&spa->spa_refcount);
	if (error) {
	if (error != EEXIST) {
	spa->spa_loaded_ts.tv_sec = 0;
	spa->spa_loaded_ts.tv_nsec = 0;
	}
	if (error != EBADF) {
	(void) zfs_ereport_post(ereport, spa,
	NULL, NULL, NULL, 0);
	}
	}
	spa->spa_load_state = error ? SPA_LOAD_ERROR : SPA_LOAD_NONE;
	spa->spa_ena = 0;

	(void) spa_import_progress_set_state(spa_guid(spa),
	spa_load_state(spa));

	return (error);
	}

	#ifdef ZFS_DEBUG
	/*
	* Count the number of per-vdev ZAPs associated with all of the vdevs in the
	* vdev tree rooted in the given vd, and ensure that each ZAP is present in the
	* spa's per-vdev ZAP list.
	*/
	static uint64_t
	vdev_count_verify_zaps(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;
	uint64_t total = 0;

	if (vd->vdev_top_zap != 0) {
	total++;
	ASSERT0(zap_lookup_int(spa->spa_meta_objset,
	spa->spa_all_vdev_zaps, vd->vdev_top_zap));
	}
	if (vd->vdev_leaf_zap != 0) {
	total++;
	ASSERT0(zap_lookup_int(spa->spa_meta_objset,
	spa->spa_all_vdev_zaps, vd->vdev_leaf_zap));
	}

	for (uint64_t i = 0; i < vd->vdev_children; i++) {
	total += vdev_count_verify_zaps(vd->vdev_child[i]);
	}

	return (total);
	}
	#endif

	/*
	* Determine whether the activity check is required.
	*/
	static boolean_t
	spa_activity_check_required(spa_t spa, uberblock_t ub, nvlist_t *label,
	nvlist_t *config)
	{
	uint64_t state = 0;
	uint64_t hostid = 0;
	uint64_t tryconfig_txg = 0;
	uint64_t tryconfig_timestamp = 0;
	uint16_t tryconfig_mmp_seq = 0;
	nvlist_t *nvinfo;

	if (nvlist_exists(config, ZPOOL_CONFIG_LOAD_INFO)) {
	nvinfo = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_LOAD_INFO);
	(void) nvlist_lookup_uint64(nvinfo, ZPOOL_CONFIG_MMP_TXG,
	&tryconfig_txg);
	(void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_TIMESTAMP,
	&tryconfig_timestamp);
	(void) nvlist_lookup_uint16(nvinfo, ZPOOL_CONFIG_MMP_SEQ,
	&tryconfig_mmp_seq);
	}

	(void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_STATE, &state);

	/*
	* Disable the MMP activity check - This is used by zdb which
	* is intended to be used on potentially active pools.
	*/
	if (spa->spa_import_flags & ZFS_IMPORT_SKIP_MMP)
	return (B_FALSE);

	/*
	* Skip the activity check when the MMP feature is disabled.
	*/
	if (ub->ub_mmp_magic == MMP_MAGIC && ub->ub_mmp_delay == 0)
	return (B_FALSE);

	/*
	* If the tryconfig_ values are nonzero, they are the results of an
	* earlier tryimport. If they all match the uberblock we just found,
	* then the pool has not changed and we return false so we do not test
	* a second time.
	*/
	if (tryconfig_txg && tryconfig_txg == ub->ub_txg &&
	tryconfig_timestamp && tryconfig_timestamp == ub->ub_timestamp &&
	tryconfig_mmp_seq && tryconfig_mmp_seq ==
	(MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0))
	return (B_FALSE);

	/*
	* Allow the activity check to be skipped when importing the pool
	* on the same host which last imported it. Since the hostid from
	* configuration may be stale use the one read from the label.
	*/
	if (nvlist_exists(label, ZPOOL_CONFIG_HOSTID))
	hostid = fnvlist_lookup_uint64(label, ZPOOL_CONFIG_HOSTID);

	if (hostid == spa_get_hostid(spa))
	return (B_FALSE);

	/*
	* Skip the activity test when the pool was cleanly exported.
	*/
	if (state != POOL_STATE_ACTIVE)
	return (B_FALSE);

	return (B_TRUE);
	}

	/*
	* Nanoseconds the activity check must watch for changes on-disk.
	*/
	static uint64_t
	spa_activity_check_duration(spa_t spa, uberblock_t ub)
	{
	uint64_t import_intervals = MAX(zfs_multihost_import_intervals, 1);
	uint64_t multihost_interval = MSEC2NSEC(
	MMP_INTERVAL_OK(zfs_multihost_interval));
	uint64_t import_delay = MAX(NANOSEC, import_intervals *
	multihost_interval);

	/*
	* Local tunables determine a minimum duration except for the case
	* where we know when the remote host will suspend the pool if MMP
	* writes do not land.
	*
	* See Big Theory comment at the top of mmp.c for the reasoning behind
	* these cases and times.
	*/

	ASSERT(MMP_IMPORT_SAFETY_FACTOR >= 100);

	if (MMP_INTERVAL_VALID(ub) && MMP_FAIL_INT_VALID(ub) &&
	MMP_FAIL_INT(ub) > 0) {

	/* MMP on remote host will suspend pool after failed writes */
	import_delay = MMP_FAIL_INT(ub) * MSEC2NSEC(MMP_INTERVAL(ub)) *
	MMP_IMPORT_SAFETY_FACTOR / 100;

	zfs_dbgmsg("fail_intvals>0 import_delay=%llu ub_mmp "
	"mmp_fails=%llu ub_mmp mmp_interval=%llu "
	"import_intervals=%llu", (u_longlong_t)import_delay,
	(u_longlong_t)MMP_FAIL_INT(ub),
	(u_longlong_t)MMP_INTERVAL(ub),
	(u_longlong_t)import_intervals);

	} else if (MMP_INTERVAL_VALID(ub) && MMP_FAIL_INT_VALID(ub) &&
	MMP_FAIL_INT(ub) == 0) {

	/* MMP on remote host will never suspend pool */
	import_delay = MAX(import_delay, (MSEC2NSEC(MMP_INTERVAL(ub)) +
	ub->ub_mmp_delay) * import_intervals);

	zfs_dbgmsg("fail_intvals=0 import_delay=%llu ub_mmp "
	"mmp_interval=%llu ub_mmp_delay=%llu "
	"import_intervals=%llu", (u_longlong_t)import_delay,
	(u_longlong_t)MMP_INTERVAL(ub),
	(u_longlong_t)ub->ub_mmp_delay,
	(u_longlong_t)import_intervals);

	} else if (MMP_VALID(ub)) {
	/*
	* zfs-0.7 compatibility case
	*/

	import_delay = MAX(import_delay, (multihost_interval +
	ub->ub_mmp_delay) * import_intervals);

	zfs_dbgmsg("import_delay=%llu ub_mmp_delay=%llu "
	"import_intervals=%llu leaves=%u",
	(u_longlong_t)import_delay,
	(u_longlong_t)ub->ub_mmp_delay,
	(u_longlong_t)import_intervals,
	vdev_count_leaves(spa));
	} else {
	/* Using local tunings is the only reasonable option */
	zfs_dbgmsg("pool last imported on non-MMP aware "
	"host using import_delay=%llu multihost_interval=%llu "
	"import_intervals=%llu", (u_longlong_t)import_delay,
	(u_longlong_t)multihost_interval,
	(u_longlong_t)import_intervals);
	}

	return (import_delay);
	}

	/*
	* Perform the import activity check. If the user canceled the import or
	* we detected activity then fail.
	*/
	static int
	spa_activity_check(spa_t spa, uberblock_t ub, nvlist_t *config)
	{
	uint64_t txg = ub->ub_txg;
	uint64_t timestamp = ub->ub_timestamp;
	uint64_t mmp_config = ub->ub_mmp_config;
	uint16_t mmp_seq = MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0;
	uint64_t import_delay;
	hrtime_t import_expire;
	nvlist_t *mmp_label = NULL;
	vdev_t *rvd = spa->spa_root_vdev;
	kcondvar_t cv;
	kmutex_t mtx;
	int error = 0;

	cv_init(&cv, NULL, CV_DEFAULT, NULL);
	mutex_init(&mtx, NULL, MUTEX_DEFAULT, NULL);
	mutex_enter(&mtx);

	/*
	* If ZPOOL_CONFIG_MMP_TXG is present an activity check was performed
	* during the earlier tryimport. If the txg recorded there is 0 then
	* the pool is known to be active on another host.
	*
	* Otherwise, the pool might be in use on another host. Check for
	* changes in the uberblocks on disk if necessary.
	*/
	if (nvlist_exists(config, ZPOOL_CONFIG_LOAD_INFO)) {
	nvlist_t *nvinfo = fnvlist_lookup_nvlist(config,
	ZPOOL_CONFIG_LOAD_INFO);

	if (nvlist_exists(nvinfo, ZPOOL_CONFIG_MMP_TXG) &&
	fnvlist_lookup_uint64(nvinfo, ZPOOL_CONFIG_MMP_TXG) == 0) {
	vdev_uberblock_load(rvd, ub, &mmp_label);
	error = SET_ERROR(EREMOTEIO);
	goto out;
	}
	}

	import_delay = spa_activity_check_duration(spa, ub);

	/* Add a small random factor in case of simultaneous imports (0-25%) */
	import_delay += import_delay * random_in_range(250) / 1000;

	import_expire = gethrtime() + import_delay;

	while (gethrtime() < import_expire) {
	(void) spa_import_progress_set_mmp_check(spa_guid(spa),
	NSEC2SEC(import_expire - gethrtime()));

	vdev_uberblock_load(rvd, ub, &mmp_label);

	if (txg != ub->ub_txg \|\| timestamp != ub->ub_timestamp \|\|
	mmp_seq != (MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0)) {
	zfs_dbgmsg("multihost activity detected "
	"txg %llu ub_txg %llu "
	"timestamp %llu ub_timestamp %llu "
	"mmp_config %#llx ub_mmp_config %#llx",
	(u_longlong_t)txg, (u_longlong_t)ub->ub_txg,
	(u_longlong_t)timestamp,
	(u_longlong_t)ub->ub_timestamp,
	(u_longlong_t)mmp_config,
	(u_longlong_t)ub->ub_mmp_config);

	error = SET_ERROR(EREMOTEIO);
	break;
	}

	if (mmp_label) {
	nvlist_free(mmp_label);
	mmp_label = NULL;
	}

	error = cv_timedwait_sig(&cv, &mtx, ddi_get_lbolt() + hz);
	if (error != -1) {
	error = SET_ERROR(EINTR);
	break;
	}
	error = 0;
	}

	out:
	mutex_exit(&mtx);
	mutex_destroy(&mtx);
	cv_destroy(&cv);

	/*
	* If the pool is determined to be active store the status in the
	* spa->spa_load_info nvlist. If the remote hostname or hostid are
	* available from configuration read from disk store them as well.
	* This allows 'zpool import' to generate a more useful message.
	*
	* ZPOOL_CONFIG_MMP_STATE - observed pool status (mandatory)
	* ZPOOL_CONFIG_MMP_HOSTNAME - hostname from the active pool
	* ZPOOL_CONFIG_MMP_HOSTID - hostid from the active pool
	*/
	if (error == EREMOTEIO) {
	char *hostname = "<unknown>";
	uint64_t hostid = 0;

	if (mmp_label) {
	if (nvlist_exists(mmp_label, ZPOOL_CONFIG_HOSTNAME)) {
	hostname = fnvlist_lookup_string(mmp_label,
	ZPOOL_CONFIG_HOSTNAME);
	fnvlist_add_string(spa->spa_load_info,
	ZPOOL_CONFIG_MMP_HOSTNAME, hostname);
	}

	if (nvlist_exists(mmp_label, ZPOOL_CONFIG_HOSTID)) {
	hostid = fnvlist_lookup_uint64(mmp_label,
	ZPOOL_CONFIG_HOSTID);
	fnvlist_add_uint64(spa->spa_load_info,
	ZPOOL_CONFIG_MMP_HOSTID, hostid);
	}
	}

	fnvlist_add_uint64(spa->spa_load_info,
	ZPOOL_CONFIG_MMP_STATE, MMP_STATE_ACTIVE);
	fnvlist_add_uint64(spa->spa_load_info,
	ZPOOL_CONFIG_MMP_TXG, 0);

	error = spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO);
	}

	if (mmp_label)
	nvlist_free(mmp_label);

	return (error);
	}

	static int
	spa_verify_host(spa_t spa, nvlist_t mos_config)
	{
	uint64_t hostid;
	char *hostname;
	uint64_t myhostid = 0;

	if (!spa_is_root(spa) && nvlist_lookup_uint64(mos_config,
	ZPOOL_CONFIG_HOSTID, &hostid) == 0) {
	hostname = fnvlist_lookup_string(mos_config,
	ZPOOL_CONFIG_HOSTNAME);

	myhostid = zone_get_hostid(NULL);

	if (hostid != 0 && myhostid != 0 && hostid != myhostid) {
	cmn_err(CE_WARN, "pool '%s' could not be "
	"loaded as it was last accessed by "
	"another system (host: %s hostid: 0x%llx). "
	"See: https://openzfs.github.io/openzfs-docs/msg/"
	"ZFS-8000-EY",
	spa_name(spa), hostname, (u_longlong_t)hostid);
	spa_load_failed(spa, "hostid verification failed: pool "
	"last accessed by host: %s (hostid: 0x%llx)",
	hostname, (u_longlong_t)hostid);
	return (SET_ERROR(EBADF));
	}
	}

	return (0);
	}

	static int
	spa_ld_parse_config(spa_t *spa, spa_import_type_t type)
	{
	int error = 0;
	nvlist_t nvtree, nvl, *config = spa->spa_config;
	int parse;
	vdev_t *rvd;
	uint64_t pool_guid;
	char *comment;
	char *compatibility;

	/*
	* Versioning wasn't explicitly added to the label until later, so if
	* it's not present treat it as the initial version.
	*/
	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_VERSION,
	&spa->spa_ubsync.ub_version) != 0)
	spa->spa_ubsync.ub_version = SPA_VERSION_INITIAL;

	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &pool_guid)) {
	spa_load_failed(spa, "invalid config provided: '%s' missing",
	ZPOOL_CONFIG_POOL_GUID);
	return (SET_ERROR(EINVAL));
	}

	/*
	* If we are doing an import, ensure that the pool is not already
	* imported by checking if its pool guid already exists in the
	* spa namespace.
	*
	* The only case that we allow an already imported pool to be
	* imported again, is when the pool is checkpointed and we want to
	* look at its checkpointed state from userland tools like zdb.
	*/
	#ifdef _KERNEL
	if ((spa->spa_load_state == SPA_LOAD_IMPORT \|\|
	spa->spa_load_state == SPA_LOAD_TRYIMPORT) &&
	spa_guid_exists(pool_guid, 0)) {
	#else
	if ((spa->spa_load_state == SPA_LOAD_IMPORT \|\|
	spa->spa_load_state == SPA_LOAD_TRYIMPORT) &&
	spa_guid_exists(pool_guid, 0) &&
	!spa_importing_readonly_checkpoint(spa)) {
	#endif
	spa_load_failed(spa, "a pool with guid %llu is already open",
	(u_longlong_t)pool_guid);
	return (SET_ERROR(EEXIST));
	}

	spa->spa_config_guid = pool_guid;

	nvlist_free(spa->spa_load_info);
	spa->spa_load_info = fnvlist_alloc();

	ASSERT(spa->spa_comment == NULL);
	if (nvlist_lookup_string(config, ZPOOL_CONFIG_COMMENT, &comment) == 0)
	spa->spa_comment = spa_strdup(comment);

	ASSERT(spa->spa_compatibility == NULL);
	if (nvlist_lookup_string(config, ZPOOL_CONFIG_COMPATIBILITY,
	&compatibility) == 0)
	spa->spa_compatibility = spa_strdup(compatibility);

	(void) nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_TXG,
	&spa->spa_config_txg);

	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_SPLIT, &nvl) == 0)
	spa->spa_config_splitting = fnvlist_dup(nvl);

	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvtree)) {
	spa_load_failed(spa, "invalid config provided: '%s' missing",
	ZPOOL_CONFIG_VDEV_TREE);
	return (SET_ERROR(EINVAL));
	}

	/*
	* Create "The Godfather" zio to hold all async IOs
	*/
	spa->spa_async_zio_root = kmem_alloc(max_ncpus * sizeof (void *),
	KM_SLEEP);
	for (int i = 0; i < max_ncpus; i++) {
	spa->spa_async_zio_root[i] = zio_root(spa, NULL, NULL,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_SPECULATIVE \|
	ZIO_FLAG_GODFATHER);
	}

	/*
	* Parse the configuration into a vdev tree. We explicitly set the
	* value that will be returned by spa_version() since parsing the
	* configuration requires knowing the version number.
	*/
	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	parse = (type == SPA_IMPORT_EXISTING ?
	VDEV_ALLOC_LOAD : VDEV_ALLOC_SPLIT);
	error = spa_config_parse(spa, &rvd, nvtree, NULL, 0, parse);
	spa_config_exit(spa, SCL_ALL, FTAG);

	if (error != 0) {
	spa_load_failed(spa, "unable to parse config [error=%d]",
	error);
	return (error);
	}

	ASSERT(spa->spa_root_vdev == rvd);
	ASSERT3U(spa->spa_min_ashift, >=, SPA_MINBLOCKSHIFT);
	ASSERT3U(spa->spa_max_ashift, <=, SPA_MAXBLOCKSHIFT);

	if (type != SPA_IMPORT_ASSEMBLE) {
	ASSERT(spa_guid(spa) == pool_guid);
	}

	return (0);
	}

	/*
	* Recursively open all vdevs in the vdev tree. This function is called twice:
	* first with the untrusted config, then with the trusted config.
	*/
	static int
	spa_ld_open_vdevs(spa_t *spa)
	{
	int error = 0;

	/*
	* spa_missing_tvds_allowed defines how many top-level vdevs can be
	* missing/unopenable for the root vdev to be still considered openable.
	*/
	if (spa->spa_trust_config) {
	spa->spa_missing_tvds_allowed = zfs_max_missing_tvds;
	} else if (spa->spa_config_source == SPA_CONFIG_SRC_CACHEFILE) {
	spa->spa_missing_tvds_allowed = zfs_max_missing_tvds_cachefile;
	} else if (spa->spa_config_source == SPA_CONFIG_SRC_SCAN) {
	spa->spa_missing_tvds_allowed = zfs_max_missing_tvds_scan;
	} else {
	spa->spa_missing_tvds_allowed = 0;
	}

	spa->spa_missing_tvds_allowed =
	MAX(zfs_max_missing_tvds, spa->spa_missing_tvds_allowed);

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	error = vdev_open(spa->spa_root_vdev);
	spa_config_exit(spa, SCL_ALL, FTAG);

	if (spa->spa_missing_tvds != 0) {
	spa_load_note(spa, "vdev tree has %lld missing top-level "
	"vdevs.", (u_longlong_t)spa->spa_missing_tvds);
	if (spa->spa_trust_config && (spa->spa_mode & SPA_MODE_WRITE)) {
	/*
	* Although theoretically we could allow users to open
	* incomplete pools in RW mode, we'd need to add a lot
	* of extra logic (e.g. adjust pool space to account
	* for missing vdevs).
	* This limitation also prevents users from accidentally
	* opening the pool in RW mode during data recovery and
	* damaging it further.
	*/
	spa_load_note(spa, "pools with missing top-level "
	"vdevs can only be opened in read-only mode.");
	error = SET_ERROR(ENXIO);
	} else {
	spa_load_note(spa, "current settings allow for maximum "
	"%lld missing top-level vdevs at this stage.",
	(u_longlong_t)spa->spa_missing_tvds_allowed);
	}
	}
	if (error != 0) {
	spa_load_failed(spa, "unable to open vdev tree [error=%d]",
	error);
	}
	if (spa->spa_missing_tvds != 0 \|\| error != 0)
	vdev_dbgmsg_print_tree(spa->spa_root_vdev, 2);

	return (error);
	}

	/*
	* We need to validate the vdev labels against the configuration that
	* we have in hand. This function is called twice: first with an untrusted
	* config, then with a trusted config. The validation is more strict when the
	* config is trusted.
	*/
	static int
	spa_ld_validate_vdevs(spa_t *spa)
	{
	int error = 0;
	vdev_t *rvd = spa->spa_root_vdev;

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	error = vdev_validate(rvd);
	spa_config_exit(spa, SCL_ALL, FTAG);

	if (error != 0) {
	spa_load_failed(spa, "vdev_validate failed [error=%d]", error);
	return (error);
	}

	if (rvd->vdev_state <= VDEV_STATE_CANT_OPEN) {
	spa_load_failed(spa, "cannot open vdev tree after invalidating "
	"some vdevs");
	vdev_dbgmsg_print_tree(rvd, 2);
	return (SET_ERROR(ENXIO));
	}

	return (0);
	}

	static void
	spa_ld_select_uberblock_done(spa_t spa, uberblock_t ub)
	{
	spa->spa_state = POOL_STATE_ACTIVE;
	spa->spa_ubsync = spa->spa_uberblock;
	spa->spa_verify_min_txg = spa->spa_extreme_rewind ?
	TXG_INITIAL - 1 : spa_last_synced_txg(spa) - TXG_DEFER_SIZE - 1;
	spa->spa_first_txg = spa->spa_last_ubsync_txg ?
	spa->spa_last_ubsync_txg : spa_last_synced_txg(spa) + 1;
	spa->spa_claim_max_txg = spa->spa_first_txg;
	spa->spa_prev_software_version = ub->ub_software_version;
	}

	static int
	spa_ld_select_uberblock(spa_t *spa, spa_import_type_t type)
	{
	vdev_t *rvd = spa->spa_root_vdev;
	nvlist_t *label;
	uberblock_t *ub = &spa->spa_uberblock;
	boolean_t activity_check = B_FALSE;

	/*
	* If we are opening the checkpointed state of the pool by
	* rewinding to it, at this point we will have written the
	* checkpointed uberblock to the vdev labels, so searching
	* the labels will find the right uberblock. However, if
	* we are opening the checkpointed state read-only, we have
	* not modified the labels. Therefore, we must ignore the
	* labels and continue using the spa_uberblock that was set
	* by spa_ld_checkpoint_rewind.
	*
	* Note that it would be fine to ignore the labels when
	* rewinding (opening writeable) as well. However, if we
	* crash just after writing the labels, we will end up
	* searching the labels. Doing so in the common case means
	* that this code path gets exercised normally, rather than
	* just in the edge case.
	*/
	if (ub->ub_checkpoint_txg != 0 &&
	spa_importing_readonly_checkpoint(spa)) {
	spa_ld_select_uberblock_done(spa, ub);
	return (0);
	}

	/*
	* Find the best uberblock.
	*/
	vdev_uberblock_load(rvd, ub, &label);

	/*
	* If we weren't able to find a single valid uberblock, return failure.
	*/
	if (ub->ub_txg == 0) {
	nvlist_free(label);
	spa_load_failed(spa, "no valid uberblock found");
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, ENXIO));
	}

	if (spa->spa_load_max_txg != UINT64_MAX) {
	(void) spa_import_progress_set_max_txg(spa_guid(spa),
	(u_longlong_t)spa->spa_load_max_txg);
	}
	spa_load_note(spa, "using uberblock with txg=%llu",
	(u_longlong_t)ub->ub_txg);


	/*
	* For pools which have the multihost property on determine if the
	* pool is truly inactive and can be safely imported. Prevent
	* hosts which don't have a hostid set from importing the pool.
	*/
	activity_check = spa_activity_check_required(spa, ub, label,
	spa->spa_config);
	if (activity_check) {
	if (ub->ub_mmp_magic == MMP_MAGIC && ub->ub_mmp_delay &&
	spa_get_hostid(spa) == 0) {
	nvlist_free(label);
	fnvlist_add_uint64(spa->spa_load_info,
	ZPOOL_CONFIG_MMP_STATE, MMP_STATE_NO_HOSTID);
	return (spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO));
	}

	int error = spa_activity_check(spa, ub, spa->spa_config);
	if (error) {
	nvlist_free(label);
	return (error);
	}

	fnvlist_add_uint64(spa->spa_load_info,
	ZPOOL_CONFIG_MMP_STATE, MMP_STATE_INACTIVE);
	fnvlist_add_uint64(spa->spa_load_info,
	ZPOOL_CONFIG_MMP_TXG, ub->ub_txg);
	fnvlist_add_uint16(spa->spa_load_info,
	ZPOOL_CONFIG_MMP_SEQ,
	(MMP_SEQ_VALID(ub) ? MMP_SEQ(ub) : 0));
	}

	/*
	* If the pool has an unsupported version we can't open it.
	*/
	if (!SPA_VERSION_IS_SUPPORTED(ub->ub_version)) {
	nvlist_free(label);
	spa_load_failed(spa, "version %llu is not supported",
	(u_longlong_t)ub->ub_version);
	return (spa_vdev_err(rvd, VDEV_AUX_VERSION_NEWER, ENOTSUP));
	}

	if (ub->ub_version >= SPA_VERSION_FEATURES) {
	nvlist_t *features;

	/*
	* If we weren't able to find what's necessary for reading the
	* MOS in the label, return failure.
	*/
	if (label == NULL) {
	spa_load_failed(spa, "label config unavailable");
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA,
	ENXIO));
	}

	if (nvlist_lookup_nvlist(label, ZPOOL_CONFIG_FEATURES_FOR_READ,
	&features) != 0) {
	nvlist_free(label);
	spa_load_failed(spa, "invalid label: '%s' missing",
	ZPOOL_CONFIG_FEATURES_FOR_READ);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA,
	ENXIO));
	}

	/*
	* Update our in-core representation with the definitive values
	* from the label.
	*/
	nvlist_free(spa->spa_label_features);
	spa->spa_label_features = fnvlist_dup(features);
	}

	nvlist_free(label);

	/*
	* Look through entries in the label nvlist's features_for_read. If
	* there is a feature listed there which we don't understand then we
	* cannot open a pool.
	*/
	if (ub->ub_version >= SPA_VERSION_FEATURES) {
	nvlist_t *unsup_feat;

	unsup_feat = fnvlist_alloc();

	for (nvpair_t *nvp = nvlist_next_nvpair(spa->spa_label_features,
	NULL); nvp != NULL;
	nvp = nvlist_next_nvpair(spa->spa_label_features, nvp)) {
	if (!zfeature_is_supported(nvpair_name(nvp))) {
	fnvlist_add_string(unsup_feat,
	nvpair_name(nvp), "");
	}
	}

	if (!nvlist_empty(unsup_feat)) {
	fnvlist_add_nvlist(spa->spa_load_info,
	ZPOOL_CONFIG_UNSUP_FEAT, unsup_feat);
	nvlist_free(unsup_feat);
	spa_load_failed(spa, "some features are unsupported");
	return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT,
	ENOTSUP));
	}

	nvlist_free(unsup_feat);
	}

	if (type != SPA_IMPORT_ASSEMBLE && spa->spa_config_splitting) {
	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	spa_try_repair(spa, spa->spa_config);
	spa_config_exit(spa, SCL_ALL, FTAG);
	nvlist_free(spa->spa_config_splitting);
	spa->spa_config_splitting = NULL;
	}

	/*
	* Initialize internal SPA structures.
	*/
	spa_ld_select_uberblock_done(spa, ub);

	return (0);
	}

	static int
	spa_ld_open_rootbp(spa_t *spa)
	{
	int error = 0;
	vdev_t *rvd = spa->spa_root_vdev;

	error = dsl_pool_init(spa, spa->spa_first_txg, &spa->spa_dsl_pool);
	if (error != 0) {
	spa_load_failed(spa, "unable to open rootbp in dsl_pool_init "
	"[error=%d]", error);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}
	spa->spa_meta_objset = spa->spa_dsl_pool->dp_meta_objset;

	return (0);
	}

	static int
	spa_ld_trusted_config(spa_t *spa, spa_import_type_t type,
	boolean_t reloading)
	{
	vdev_t mrvd, rvd = spa->spa_root_vdev;
	nvlist_t nv, mos_config, *policy;
	int error = 0, copy_error;
	uint64_t healthy_tvds, healthy_tvds_mos;
	uint64_t mos_config_txg;

	if (spa_dir_prop(spa, DMU_POOL_CONFIG, &spa->spa_config_object, B_TRUE)
	!= 0)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));

	/*
	* If we're assembling a pool from a split, the config provided is
	* already trusted so there is nothing to do.
	*/
	if (type == SPA_IMPORT_ASSEMBLE)
	return (0);

	healthy_tvds = spa_healthy_core_tvds(spa);

	if (load_nvlist(spa, spa->spa_config_object, &mos_config)
	!= 0) {
	spa_load_failed(spa, "unable to retrieve MOS config");
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	/*
	* If we are doing an open, pool owner wasn't verified yet, thus do
	* the verification here.
	*/
	if (spa->spa_load_state == SPA_LOAD_OPEN) {
	error = spa_verify_host(spa, mos_config);
	if (error != 0) {
	nvlist_free(mos_config);
	return (error);
	}
	}

	nv = fnvlist_lookup_nvlist(mos_config, ZPOOL_CONFIG_VDEV_TREE);

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);

	/*
	* Build a new vdev tree from the trusted config
	*/
	error = spa_config_parse(spa, &mrvd, nv, NULL, 0, VDEV_ALLOC_LOAD);
	if (error != 0) {
	nvlist_free(mos_config);
	spa_config_exit(spa, SCL_ALL, FTAG);
	spa_load_failed(spa, "spa_config_parse failed [error=%d]",
	error);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error));
	}

	/*
	* Vdev paths in the MOS may be obsolete. If the untrusted config was
	* obtained by scanning /dev/dsk, then it will have the right vdev
	* paths. We update the trusted MOS config with this information.
	* We first try to copy the paths with vdev_copy_path_strict, which
	* succeeds only when both configs have exactly the same vdev tree.
	* If that fails, we fall back to a more flexible method that has a
	* best effort policy.
	*/
	copy_error = vdev_copy_path_strict(rvd, mrvd);
	if (copy_error != 0 \|\| spa_load_print_vdev_tree) {
	spa_load_note(spa, "provided vdev tree:");
	vdev_dbgmsg_print_tree(rvd, 2);
	spa_load_note(spa, "MOS vdev tree:");
	vdev_dbgmsg_print_tree(mrvd, 2);
	}
	if (copy_error != 0) {
	spa_load_note(spa, "vdev_copy_path_strict failed, falling "
	"back to vdev_copy_path_relaxed");
	vdev_copy_path_relaxed(rvd, mrvd);
	}

	vdev_close(rvd);
	vdev_free(rvd);
	spa->spa_root_vdev = mrvd;
	rvd = mrvd;
	spa_config_exit(spa, SCL_ALL, FTAG);

	/*
	* We will use spa_config if we decide to reload the spa or if spa_load
	* fails and we rewind. We must thus regenerate the config using the
	* MOS information with the updated paths. ZPOOL_LOAD_POLICY is used to
	* pass settings on how to load the pool and is not stored in the MOS.
	* We copy it over to our new, trusted config.
	*/
	mos_config_txg = fnvlist_lookup_uint64(mos_config,
	ZPOOL_CONFIG_POOL_TXG);
	nvlist_free(mos_config);
	mos_config = spa_config_generate(spa, NULL, mos_config_txg, B_FALSE);
	if (nvlist_lookup_nvlist(spa->spa_config, ZPOOL_LOAD_POLICY,
	&policy) == 0)
	fnvlist_add_nvlist(mos_config, ZPOOL_LOAD_POLICY, policy);
	spa_config_set(spa, mos_config);
	spa->spa_config_source = SPA_CONFIG_SRC_MOS;

	/*
	* Now that we got the config from the MOS, we should be more strict
	* in checking blkptrs and can make assumptions about the consistency
	* of the vdev tree. spa_trust_config must be set to true before opening
	* vdevs in order for them to be writeable.
	*/
	spa->spa_trust_config = B_TRUE;

	/*
	* Open and validate the new vdev tree
	*/
	error = spa_ld_open_vdevs(spa);
	if (error != 0)
	return (error);

	error = spa_ld_validate_vdevs(spa);
	if (error != 0)
	return (error);

	if (copy_error != 0 \|\| spa_load_print_vdev_tree) {
	spa_load_note(spa, "final vdev tree:");
	vdev_dbgmsg_print_tree(rvd, 2);
	}

	if (spa->spa_load_state != SPA_LOAD_TRYIMPORT &&
	!spa->spa_extreme_rewind && zfs_max_missing_tvds == 0) {
	/*
	* Sanity check to make sure that we are indeed loading the
	* latest uberblock. If we missed SPA_SYNC_MIN_VDEVS tvds
	* in the config provided and they happened to be the only ones
	* to have the latest uberblock, we could involuntarily perform
	* an extreme rewind.
	*/
	healthy_tvds_mos = spa_healthy_core_tvds(spa);
	if (healthy_tvds_mos - healthy_tvds >=
	SPA_SYNC_MIN_VDEVS) {
	spa_load_note(spa, "config provided misses too many "
	"top-level vdevs compared to MOS (%lld vs %lld). ",
	(u_longlong_t)healthy_tvds,
	(u_longlong_t)healthy_tvds_mos);
	spa_load_note(spa, "vdev tree:");
	vdev_dbgmsg_print_tree(rvd, 2);
	if (reloading) {
	spa_load_failed(spa, "config was already "
	"provided from MOS. Aborting.");
	return (spa_vdev_err(rvd,
	VDEV_AUX_CORRUPT_DATA, EIO));
	}
	spa_load_note(spa, "spa must be reloaded using MOS "
	"config");
	return (SET_ERROR(EAGAIN));
	}
	}

	error = spa_check_for_missing_logs(spa);
	if (error != 0)
	return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM, ENXIO));

	if (rvd->vdev_guid_sum != spa->spa_uberblock.ub_guid_sum) {
	spa_load_failed(spa, "uberblock guid sum doesn't match MOS "
	"guid sum (%llu != %llu)",
	(u_longlong_t)spa->spa_uberblock.ub_guid_sum,
	(u_longlong_t)rvd->vdev_guid_sum);
	return (spa_vdev_err(rvd, VDEV_AUX_BAD_GUID_SUM,
	ENXIO));
	}

	return (0);
	}

	static int
	spa_ld_open_indirect_vdev_metadata(spa_t *spa)
	{
	int error = 0;
	vdev_t *rvd = spa->spa_root_vdev;

	/*
	* Everything that we read before spa_remove_init() must be stored
	* on concreted vdevs. Therefore we do this as early as possible.
	*/
	error = spa_remove_init(spa);
	if (error != 0) {
	spa_load_failed(spa, "spa_remove_init failed [error=%d]",
	error);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	/*
	* Retrieve information needed to condense indirect vdev mappings.
	*/
	error = spa_condense_init(spa);
	if (error != 0) {
	spa_load_failed(spa, "spa_condense_init failed [error=%d]",
	error);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error));
	}

	return (0);
	}

	static int
	spa_ld_check_features(spa_t spa, boolean_t missing_feat_writep)
	{
	int error = 0;
	vdev_t *rvd = spa->spa_root_vdev;

	if (spa_version(spa) >= SPA_VERSION_FEATURES) {
	boolean_t missing_feat_read = B_FALSE;
	nvlist_t unsup_feat, enabled_feat;

	if (spa_dir_prop(spa, DMU_POOL_FEATURES_FOR_READ,
	&spa->spa_feat_for_read_obj, B_TRUE) != 0) {
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	if (spa_dir_prop(spa, DMU_POOL_FEATURES_FOR_WRITE,
	&spa->spa_feat_for_write_obj, B_TRUE) != 0) {
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	if (spa_dir_prop(spa, DMU_POOL_FEATURE_DESCRIPTIONS,
	&spa->spa_feat_desc_obj, B_TRUE) != 0) {
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	enabled_feat = fnvlist_alloc();
	unsup_feat = fnvlist_alloc();

	if (!spa_features_check(spa, B_FALSE,
	unsup_feat, enabled_feat))
	missing_feat_read = B_TRUE;

	if (spa_writeable(spa) \|\|
	spa->spa_load_state == SPA_LOAD_TRYIMPORT) {
	if (!spa_features_check(spa, B_TRUE,
	unsup_feat, enabled_feat)) {
	*missing_feat_writep = B_TRUE;
	}
	}

	fnvlist_add_nvlist(spa->spa_load_info,
	ZPOOL_CONFIG_ENABLED_FEAT, enabled_feat);

	if (!nvlist_empty(unsup_feat)) {
	fnvlist_add_nvlist(spa->spa_load_info,
	ZPOOL_CONFIG_UNSUP_FEAT, unsup_feat);
	}

	fnvlist_free(enabled_feat);
	fnvlist_free(unsup_feat);

	if (!missing_feat_read) {
	fnvlist_add_boolean(spa->spa_load_info,
	ZPOOL_CONFIG_CAN_RDONLY);
	}

	/*
	* If the state is SPA_LOAD_TRYIMPORT, our objective is
	* twofold: to determine whether the pool is available for
	* import in read-write mode and (if it is not) whether the
	* pool is available for import in read-only mode. If the pool
	* is available for import in read-write mode, it is displayed
	* as available in userland; if it is not available for import
	* in read-only mode, it is displayed as unavailable in
	* userland. If the pool is available for import in read-only
	* mode but not read-write mode, it is displayed as unavailable
	* in userland with a special note that the pool is actually
	* available for open in read-only mode.
	*
	* As a result, if the state is SPA_LOAD_TRYIMPORT and we are
	* missing a feature for write, we must first determine whether
	* the pool can be opened read-only before returning to
	* userland in order to know whether to display the
	* abovementioned note.
	*/
	if (missing_feat_read \|\| (*missing_feat_writep &&
	spa_writeable(spa))) {
	spa_load_failed(spa, "pool uses unsupported features");
	return (spa_vdev_err(rvd, VDEV_AUX_UNSUP_FEAT,
	ENOTSUP));
	}

	/*
	* Load refcounts for ZFS features from disk into an in-memory
	* cache during SPA initialization.
	*/
	for (spa_feature_t i = 0; i < SPA_FEATURES; i++) {
	uint64_t refcount;

	error = feature_get_refcount_from_disk(spa,
	&spa_feature_table[i], &refcount);
	if (error == 0) {
	spa->spa_feat_refcount_cache[i] = refcount;
	} else if (error == ENOTSUP) {
	spa->spa_feat_refcount_cache[i] =
	SPA_FEATURE_DISABLED;
	} else {
	spa_load_failed(spa, "error getting refcount "
	"for feature %s [error=%d]",
	spa_feature_table[i].fi_guid, error);
	return (spa_vdev_err(rvd,
	VDEV_AUX_CORRUPT_DATA, EIO));
	}
	}
	}

	if (spa_feature_is_active(spa, SPA_FEATURE_ENABLED_TXG)) {
	if (spa_dir_prop(spa, DMU_POOL_FEATURE_ENABLED_TXG,
	&spa->spa_feat_enabled_txg_obj, B_TRUE) != 0)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	/*
	* Encryption was added before bookmark_v2, even though bookmark_v2
	* is now a dependency. If this pool has encryption enabled without
	* bookmark_v2, trigger an errata message.
	*/
	if (spa_feature_is_enabled(spa, SPA_FEATURE_ENCRYPTION) &&
	!spa_feature_is_enabled(spa, SPA_FEATURE_BOOKMARK_V2)) {
	spa->spa_errata = ZPOOL_ERRATA_ZOL_8308_ENCRYPTION;
	}

	return (0);
	}

	static int
	spa_ld_load_special_directories(spa_t *spa)
	{
	int error = 0;
	vdev_t *rvd = spa->spa_root_vdev;

	spa->spa_is_initializing = B_TRUE;
	error = dsl_pool_open(spa->spa_dsl_pool);
	spa->spa_is_initializing = B_FALSE;
	if (error != 0) {
	spa_load_failed(spa, "dsl_pool_open failed [error=%d]", error);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	return (0);
	}

	static int
	spa_ld_get_props(spa_t *spa)
	{
	int error = 0;
	uint64_t obj;
	vdev_t *rvd = spa->spa_root_vdev;

	/* Grab the checksum salt from the MOS. */
	error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_CHECKSUM_SALT, 1,
	sizeof (spa->spa_cksum_salt.zcs_bytes),
	spa->spa_cksum_salt.zcs_bytes);
	if (error == ENOENT) {
	/* Generate a new salt for subsequent use */
	(void) random_get_pseudo_bytes(spa->spa_cksum_salt.zcs_bytes,
	sizeof (spa->spa_cksum_salt.zcs_bytes));
	} else if (error != 0) {
	spa_load_failed(spa, "unable to retrieve checksum salt from "
	"MOS [error=%d]", error);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	if (spa_dir_prop(spa, DMU_POOL_SYNC_BPOBJ, &obj, B_TRUE) != 0)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	error = bpobj_open(&spa->spa_deferred_bpobj, spa->spa_meta_objset, obj);
	if (error != 0) {
	spa_load_failed(spa, "error opening deferred-frees bpobj "
	"[error=%d]", error);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	/*
	* Load the bit that tells us to use the new accounting function
	* (raid-z deflation). If we have an older pool, this will not
	* be present.
	*/
	error = spa_dir_prop(spa, DMU_POOL_DEFLATE, &spa->spa_deflate, B_FALSE);
	if (error != 0 && error != ENOENT)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));

	error = spa_dir_prop(spa, DMU_POOL_CREATION_VERSION,
	&spa->spa_creation_version, B_FALSE);
	if (error != 0 && error != ENOENT)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));

	/*
	* Load the persistent error log. If we have an older pool, this will
	* not be present.
	*/
	error = spa_dir_prop(spa, DMU_POOL_ERRLOG_LAST, &spa->spa_errlog_last,
	B_FALSE);
	if (error != 0 && error != ENOENT)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));

	error = spa_dir_prop(spa, DMU_POOL_ERRLOG_SCRUB,
	&spa->spa_errlog_scrub, B_FALSE);
	if (error != 0 && error != ENOENT)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));

	/*
	* Load the livelist deletion field. If a livelist is queued for
	* deletion, indicate that in the spa
	*/
	error = spa_dir_prop(spa, DMU_POOL_DELETED_CLONES,
	&spa->spa_livelists_to_delete, B_FALSE);
	if (error != 0 && error != ENOENT)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));

	/*
	* Load the history object. If we have an older pool, this
	* will not be present.
	*/
	error = spa_dir_prop(spa, DMU_POOL_HISTORY, &spa->spa_history, B_FALSE);
	if (error != 0 && error != ENOENT)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));

	/*
	* Load the per-vdev ZAP map. If we have an older pool, this will not
	* be present; in this case, defer its creation to a later time to
	* avoid dirtying the MOS this early / out of sync context. See
	* spa_sync_config_object.
	*/

	/* The sentinel is only available in the MOS config. */
	nvlist_t *mos_config;
	if (load_nvlist(spa, spa->spa_config_object, &mos_config) != 0) {
	spa_load_failed(spa, "unable to retrieve MOS config");
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	error = spa_dir_prop(spa, DMU_POOL_VDEV_ZAP_MAP,
	&spa->spa_all_vdev_zaps, B_FALSE);

	if (error == ENOENT) {
	VERIFY(!nvlist_exists(mos_config,
	ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS));
	spa->spa_avz_action = AVZ_ACTION_INITIALIZE;
	ASSERT0(vdev_count_verify_zaps(spa->spa_root_vdev));
	} else if (error != 0) {
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	} else if (!nvlist_exists(mos_config, ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS)) {
	/*
	* An older version of ZFS overwrote the sentinel value, so
	* we have orphaned per-vdev ZAPs in the MOS. Defer their
	* destruction to later; see spa_sync_config_object.
	*/
	spa->spa_avz_action = AVZ_ACTION_DESTROY;
	/*
	* We're assuming that no vdevs have had their ZAPs created
	* before this. Better be sure of it.
	*/
	ASSERT0(vdev_count_verify_zaps(spa->spa_root_vdev));
	}
	nvlist_free(mos_config);

	spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION);

	error = spa_dir_prop(spa, DMU_POOL_PROPS, &spa->spa_pool_props_object,
	B_FALSE);
	if (error && error != ENOENT)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));

	if (error == 0) {
	uint64_t autoreplace = 0;

	spa_prop_find(spa, ZPOOL_PROP_BOOTFS, &spa->spa_bootfs);
	spa_prop_find(spa, ZPOOL_PROP_AUTOREPLACE, &autoreplace);
	spa_prop_find(spa, ZPOOL_PROP_DELEGATION, &spa->spa_delegation);
	spa_prop_find(spa, ZPOOL_PROP_FAILUREMODE, &spa->spa_failmode);
	spa_prop_find(spa, ZPOOL_PROP_AUTOEXPAND, &spa->spa_autoexpand);
	spa_prop_find(spa, ZPOOL_PROP_MULTIHOST, &spa->spa_multihost);
	spa_prop_find(spa, ZPOOL_PROP_AUTOTRIM, &spa->spa_autotrim);
	spa->spa_autoreplace = (autoreplace != 0);
	}

	/*
	* If we are importing a pool with missing top-level vdevs,
	* we enforce that the pool doesn't panic or get suspended on
	* error since the likelihood of missing data is extremely high.
	*/
	if (spa->spa_missing_tvds > 0 &&
	spa->spa_failmode != ZIO_FAILURE_MODE_CONTINUE &&
	spa->spa_load_state != SPA_LOAD_TRYIMPORT) {
	spa_load_note(spa, "forcing failmode to 'continue' "
	"as some top level vdevs are missing");
	spa->spa_failmode = ZIO_FAILURE_MODE_CONTINUE;
	}

	return (0);
	}

	static int
	spa_ld_open_aux_vdevs(spa_t *spa, spa_import_type_t type)
	{
	int error = 0;
	vdev_t *rvd = spa->spa_root_vdev;

	/*
	* If we're assembling the pool from the split-off vdevs of
	* an existing pool, we don't want to attach the spares & cache
	* devices.
	*/

	/*
	* Load any hot spares for this pool.
	*/
	error = spa_dir_prop(spa, DMU_POOL_SPARES, &spa->spa_spares.sav_object,
	B_FALSE);
	if (error != 0 && error != ENOENT)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	if (error == 0 && type != SPA_IMPORT_ASSEMBLE) {
	ASSERT(spa_version(spa) >= SPA_VERSION_SPARES);
	if (load_nvlist(spa, spa->spa_spares.sav_object,
	&spa->spa_spares.sav_config) != 0) {
	spa_load_failed(spa, "error loading spares nvlist");
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	spa_load_spares(spa);
	spa_config_exit(spa, SCL_ALL, FTAG);
	} else if (error == 0) {
	spa->spa_spares.sav_sync = B_TRUE;
	}

	/*
	* Load any level 2 ARC devices for this pool.
	*/
	error = spa_dir_prop(spa, DMU_POOL_L2CACHE,
	&spa->spa_l2cache.sav_object, B_FALSE);
	if (error != 0 && error != ENOENT)
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	if (error == 0 && type != SPA_IMPORT_ASSEMBLE) {
	ASSERT(spa_version(spa) >= SPA_VERSION_L2CACHE);
	if (load_nvlist(spa, spa->spa_l2cache.sav_object,
	&spa->spa_l2cache.sav_config) != 0) {
	spa_load_failed(spa, "error loading l2cache nvlist");
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	spa_load_l2cache(spa);
	spa_config_exit(spa, SCL_ALL, FTAG);
	} else if (error == 0) {
	spa->spa_l2cache.sav_sync = B_TRUE;
	}

	return (0);
	}

	static int
	spa_ld_load_vdev_metadata(spa_t *spa)
	{
	int error = 0;
	vdev_t *rvd = spa->spa_root_vdev;

	/*
	* If the 'multihost' property is set, then never allow a pool to
	* be imported when the system hostid is zero. The exception to
	* this rule is zdb which is always allowed to access pools.
	*/
	if (spa_multihost(spa) && spa_get_hostid(spa) == 0 &&
	(spa->spa_import_flags & ZFS_IMPORT_SKIP_MMP) == 0) {
	fnvlist_add_uint64(spa->spa_load_info,
	ZPOOL_CONFIG_MMP_STATE, MMP_STATE_NO_HOSTID);
	return (spa_vdev_err(rvd, VDEV_AUX_ACTIVE, EREMOTEIO));
	}

	/*
	* If the 'autoreplace' property is set, then post a resource notifying
	* the ZFS DE that it should not issue any faults for unopenable
	* devices. We also iterate over the vdevs, and post a sysevent for any
	* unopenable vdevs so that the normal autoreplace handler can take
	* over.
	*/
	if (spa->spa_autoreplace && spa->spa_load_state != SPA_LOAD_TRYIMPORT) {
	spa_check_removed(spa->spa_root_vdev);
	/*
	* For the import case, this is done in spa_import(), because
	* at this point we're using the spare definitions from
	* the MOS config, not necessarily from the userland config.
	*/
	if (spa->spa_load_state != SPA_LOAD_IMPORT) {
	spa_aux_check_removed(&spa->spa_spares);
	spa_aux_check_removed(&spa->spa_l2cache);
	}
	}

	/*
	* Load the vdev metadata such as metaslabs, DTLs, spacemap object, etc.
	*/
	error = vdev_load(rvd);
	if (error != 0) {
	spa_load_failed(spa, "vdev_load failed [error=%d]", error);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error));
	}

	error = spa_ld_log_spacemaps(spa);
	if (error != 0) {
	spa_load_failed(spa, "spa_ld_log_sm_data failed [error=%d]",
	error);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, error));
	}

	/*
	* Propagate the leaf DTLs we just loaded all the way up the vdev tree.
	*/
	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	vdev_dtl_reassess(rvd, 0, 0, B_FALSE, B_FALSE);
	spa_config_exit(spa, SCL_ALL, FTAG);

	return (0);
	}

	static int
	spa_ld_load_dedup_tables(spa_t *spa)
	{
	int error = 0;
	vdev_t *rvd = spa->spa_root_vdev;

	error = ddt_load(spa);
	if (error != 0) {
	spa_load_failed(spa, "ddt_load failed [error=%d]", error);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA, EIO));
	}

	return (0);
	}

	static int
	spa_ld_verify_logs(spa_t spa, spa_import_type_t type, char *ereport)
	{
	vdev_t *rvd = spa->spa_root_vdev;

	if (type != SPA_IMPORT_ASSEMBLE && spa_writeable(spa)) {
	boolean_t missing = spa_check_logs(spa);
	if (missing) {
	if (spa->spa_missing_tvds != 0) {
	spa_load_note(spa, "spa_check_logs failed "
	"so dropping the logs");
	} else {
	*ereport = FM_EREPORT_ZFS_LOG_REPLAY;
	spa_load_failed(spa, "spa_check_logs failed");
	return (spa_vdev_err(rvd, VDEV_AUX_BAD_LOG,
	ENXIO));
	}
	}
	}

	return (0);
	}

	static int
	spa_ld_verify_pool_data(spa_t *spa)
	{
	int error = 0;
	vdev_t *rvd = spa->spa_root_vdev;

	/*
	* We've successfully opened the pool, verify that we're ready
	* to start pushing transactions.
	*/
	if (spa->spa_load_state != SPA_LOAD_TRYIMPORT) {
	error = spa_load_verify(spa);
	if (error != 0) {
	spa_load_failed(spa, "spa_load_verify failed "
	"[error=%d]", error);
	return (spa_vdev_err(rvd, VDEV_AUX_CORRUPT_DATA,
	error));
	}
	}

	return (0);
	}

	static void
	spa_ld_claim_log_blocks(spa_t *spa)
	{
	dmu_tx_t *tx;
	dsl_pool_t *dp = spa_get_dsl(spa);

	/*
	* Claim log blocks that haven't been committed yet.
	* This must all happen in a single txg.
	* Note: spa_claim_max_txg is updated by spa_claim_notify(),
	* invoked from zil_claim_log_block()'s i/o done callback.
	* Price of rollback is that we abandon the log.
	*/
	spa->spa_claiming = B_TRUE;

	tx = dmu_tx_create_assigned(dp, spa_first_txg(spa));
	(void) dmu_objset_find_dp(dp, dp->dp_root_dir_obj,
	zil_claim, tx, DS_FIND_CHILDREN);
	dmu_tx_commit(tx);

	spa->spa_claiming = B_FALSE;

	spa_set_log_state(spa, SPA_LOG_GOOD);
	}

	static void
	spa_ld_check_for_config_update(spa_t *spa, uint64_t config_cache_txg,
	boolean_t update_config_cache)
	{
	vdev_t *rvd = spa->spa_root_vdev;
	int need_update = B_FALSE;

	/*
	* If the config cache is stale, or we have uninitialized
	* metaslabs (see spa_vdev_add()), then update the config.
	*
	* If this is a verbatim import, trust the current
	* in-core spa_config and update the disk labels.
	*/
	if (update_config_cache \|\| config_cache_txg != spa->spa_config_txg \|\|
	spa->spa_load_state == SPA_LOAD_IMPORT \|\|
	spa->spa_load_state == SPA_LOAD_RECOVER \|\|
	(spa->spa_import_flags & ZFS_IMPORT_VERBATIM))
	need_update = B_TRUE;

	for (int c = 0; c < rvd->vdev_children; c++)
	if (rvd->vdev_child[c]->vdev_ms_array == 0)
	need_update = B_TRUE;

	/*
	* Update the config cache asynchronously in case we're the
	* root pool, in which case the config cache isn't writable yet.
	*/
	if (need_update)
	spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE);
	}

	static void
	spa_ld_prepare_for_reload(spa_t *spa)
	{
	spa_mode_t mode = spa->spa_mode;
	int async_suspended = spa->spa_async_suspended;

	spa_unload(spa);
	spa_deactivate(spa);
	spa_activate(spa, mode);

	/*
	* We save the value of spa_async_suspended as it gets reset to 0 by
	* spa_unload(). We want to restore it back to the original value before
	* returning as we might be calling spa_async_resume() later.
	*/
	spa->spa_async_suspended = async_suspended;
	}

	static int
	spa_ld_read_checkpoint_txg(spa_t *spa)
	{
	uberblock_t checkpoint;
	int error = 0;

	ASSERT0(spa->spa_checkpoint_txg);
	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_ZPOOL_CHECKPOINT, sizeof (uint64_t),
	sizeof (uberblock_t) / sizeof (uint64_t), &checkpoint);

	if (error == ENOENT)
	return (0);

	if (error != 0)
	return (error);

	ASSERT3U(checkpoint.ub_txg, !=, 0);
	ASSERT3U(checkpoint.ub_checkpoint_txg, !=, 0);
	ASSERT3U(checkpoint.ub_timestamp, !=, 0);
	spa->spa_checkpoint_txg = checkpoint.ub_txg;
	spa->spa_checkpoint_info.sci_timestamp = checkpoint.ub_timestamp;

	return (0);
	}

	static int
	spa_ld_mos_init(spa_t *spa, spa_import_type_t type)
	{
	int error = 0;

	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	ASSERT(spa->spa_config_source != SPA_CONFIG_SRC_NONE);

	/*
	* Never trust the config that is provided unless we are assembling
	* a pool following a split.
	* This means don't trust blkptrs and the vdev tree in general. This
	* also effectively puts the spa in read-only mode since
	* spa_writeable() checks for spa_trust_config to be true.
	* We will later load a trusted config from the MOS.
	*/
	if (type != SPA_IMPORT_ASSEMBLE)
	spa->spa_trust_config = B_FALSE;

	/*
	* Parse the config provided to create a vdev tree.
	*/
	error = spa_ld_parse_config(spa, type);
	if (error != 0)
	return (error);

	spa_import_progress_add(spa);

	/*
	* Now that we have the vdev tree, try to open each vdev. This involves
	* opening the underlying physical device, retrieving its geometry and
	* probing the vdev with a dummy I/O. The state of each vdev will be set
	* based on the success of those operations. After this we'll be ready
	* to read from the vdevs.
	*/
	error = spa_ld_open_vdevs(spa);
	if (error != 0)
	return (error);

	/*
	* Read the label of each vdev and make sure that the GUIDs stored
	* there match the GUIDs in the config provided.
	* If we're assembling a new pool that's been split off from an
	* existing pool, the labels haven't yet been updated so we skip
	* validation for now.
	*/
	if (type != SPA_IMPORT_ASSEMBLE) {
	error = spa_ld_validate_vdevs(spa);
	if (error != 0)
	return (error);
	}

	/*
	* Read all vdev labels to find the best uberblock (i.e. latest,
	* unless spa_load_max_txg is set) and store it in spa_uberblock. We
	* get the list of features required to read blkptrs in the MOS from
	* the vdev label with the best uberblock and verify that our version
	* of zfs supports them all.
	*/
	error = spa_ld_select_uberblock(spa, type);
	if (error != 0)
	return (error);

	/*
	* Pass that uberblock to the dsl_pool layer which will open the root
	* blkptr. This blkptr points to the latest version of the MOS and will
	* allow us to read its contents.
	*/
	error = spa_ld_open_rootbp(spa);
	if (error != 0)
	return (error);

	return (0);
	}

	static int
	spa_ld_checkpoint_rewind(spa_t *spa)
	{
	uberblock_t checkpoint;
	int error = 0;

	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	ASSERT(spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT);

	error = zap_lookup(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_ZPOOL_CHECKPOINT, sizeof (uint64_t),
	sizeof (uberblock_t) / sizeof (uint64_t), &checkpoint);

	if (error != 0) {
	spa_load_failed(spa, "unable to retrieve checkpointed "
	"uberblock from the MOS config [error=%d]", error);

	if (error == ENOENT)
	error = ZFS_ERR_NO_CHECKPOINT;

	return (error);
	}

	ASSERT3U(checkpoint.ub_txg, <, spa->spa_uberblock.ub_txg);
	ASSERT3U(checkpoint.ub_txg, ==, checkpoint.ub_checkpoint_txg);

	/*
	* We need to update the txg and timestamp of the checkpointed
	* uberblock to be higher than the latest one. This ensures that
	* the checkpointed uberblock is selected if we were to close and
	* reopen the pool right after we've written it in the vdev labels.
	* (also see block comment in vdev_uberblock_compare)
	*/
	checkpoint.ub_txg = spa->spa_uberblock.ub_txg + 1;
	checkpoint.ub_timestamp = gethrestime_sec();

	/*
	* Set current uberblock to be the checkpointed uberblock.
	*/
	spa->spa_uberblock = checkpoint;

	/*
	* If we are doing a normal rewind, then the pool is open for
	* writing and we sync the "updated" checkpointed uberblock to
	* disk. Once this is done, we've basically rewound the whole
	* pool and there is no way back.
	*
	* There are cases when we don't want to attempt and sync the
	* checkpointed uberblock to disk because we are opening a
	* pool as read-only. Specifically, verifying the checkpointed
	* state with zdb, and importing the checkpointed state to get
	* a "preview" of its content.
	*/
	if (spa_writeable(spa)) {
	vdev_t *rvd = spa->spa_root_vdev;

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	vdev_t *svd[SPA_SYNC_MIN_VDEVS] = { NULL };
	int svdcount = 0;
	int children = rvd->vdev_children;
	int c0 = random_in_range(children);

	for (int c = 0; c < children; c++) {
	vdev_t *vd = rvd->vdev_child[(c0 + c) % children];

	/* Stop when revisiting the first vdev */
	if (c > 0 && svd[0] == vd)
	break;

	if (vd->vdev_ms_array == 0 \|\| vd->vdev_islog \|\|
	!vdev_is_concrete(vd))
	continue;

	svd[svdcount++] = vd;
	if (svdcount == SPA_SYNC_MIN_VDEVS)
	break;
	}
	error = vdev_config_sync(svd, svdcount, spa->spa_first_txg);
	if (error == 0)
	spa->spa_last_synced_guid = rvd->vdev_guid;
	spa_config_exit(spa, SCL_ALL, FTAG);

	if (error != 0) {
	spa_load_failed(spa, "failed to write checkpointed "
	"uberblock to the vdev labels [error=%d]", error);
	return (error);
	}
	}

	return (0);
	}

	static int
	spa_ld_mos_with_trusted_config(spa_t *spa, spa_import_type_t type,
	boolean_t *update_config_cache)
	{
	int error;

	/*
	* Parse the config for pool, open and validate vdevs,
	* select an uberblock, and use that uberblock to open
	* the MOS.
	*/
	error = spa_ld_mos_init(spa, type);
	if (error != 0)
	return (error);

	/*
	* Retrieve the trusted config stored in the MOS and use it to create
	* a new, exact version of the vdev tree, then reopen all vdevs.
	*/
	error = spa_ld_trusted_config(spa, type, B_FALSE);
	if (error == EAGAIN) {
	if (update_config_cache != NULL)
	*update_config_cache = B_TRUE;

	/*
	* Redo the loading process with the trusted config if it is
	* too different from the untrusted config.
	*/
	spa_ld_prepare_for_reload(spa);
	spa_load_note(spa, "RELOADING");
	error = spa_ld_mos_init(spa, type);
	if (error != 0)
	return (error);

	error = spa_ld_trusted_config(spa, type, B_TRUE);
	if (error != 0)
	return (error);

	} else if (error != 0) {
	return (error);
	}

	return (0);
	}

	/*
	* Load an existing storage pool, using the config provided. This config
	* describes which vdevs are part of the pool and is later validated against
	* partial configs present in each vdev's label and an entire copy of the
	* config stored in the MOS.
	*/
	static int
	spa_load_impl(spa_t spa, spa_import_type_t type, char *ereport)
	{
	int error = 0;
	boolean_t missing_feat_write = B_FALSE;
	boolean_t checkpoint_rewind =
	(spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT);
	boolean_t update_config_cache = B_FALSE;

	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	ASSERT(spa->spa_config_source != SPA_CONFIG_SRC_NONE);

	spa_load_note(spa, "LOADING");

	error = spa_ld_mos_with_trusted_config(spa, type, &update_config_cache);
	if (error != 0)
	return (error);

	/*
	* If we are rewinding to the checkpoint then we need to repeat
	* everything we've done so far in this function but this time
	* selecting the checkpointed uberblock and using that to open
	* the MOS.
	*/
	if (checkpoint_rewind) {
	/*
	* If we are rewinding to the checkpoint update config cache
	* anyway.
	*/
	update_config_cache = B_TRUE;

	/*
	* Extract the checkpointed uberblock from the current MOS
	* and use this as the pool's uberblock from now on. If the
	* pool is imported as writeable we also write the checkpoint
	* uberblock to the labels, making the rewind permanent.
	*/
	error = spa_ld_checkpoint_rewind(spa);
	if (error != 0)
	return (error);

	/*
	* Redo the loading process again with the
	* checkpointed uberblock.
	*/
	spa_ld_prepare_for_reload(spa);
	spa_load_note(spa, "LOADING checkpointed uberblock");
	error = spa_ld_mos_with_trusted_config(spa, type, NULL);
	if (error != 0)
	return (error);
	}

	/*
	* Retrieve the checkpoint txg if the pool has a checkpoint.
	*/
	error = spa_ld_read_checkpoint_txg(spa);
	if (error != 0)
	return (error);

	/*
	* Retrieve the mapping of indirect vdevs. Those vdevs were removed
	* from the pool and their contents were re-mapped to other vdevs. Note
	* that everything that we read before this step must have been
	* rewritten on concrete vdevs after the last device removal was
	* initiated. Otherwise we could be reading from indirect vdevs before
	* we have loaded their mappings.
	*/
	error = spa_ld_open_indirect_vdev_metadata(spa);
	if (error != 0)
	return (error);

	/*
	* Retrieve the full list of active features from the MOS and check if
	* they are all supported.
	*/
	error = spa_ld_check_features(spa, &missing_feat_write);
	if (error != 0)
	return (error);

	/*
	* Load several special directories from the MOS needed by the dsl_pool
	* layer.
	*/
	error = spa_ld_load_special_directories(spa);
	if (error != 0)
	return (error);

	/*
	* Retrieve pool properties from the MOS.
	*/
	error = spa_ld_get_props(spa);
	if (error != 0)
	return (error);

	/*
	* Retrieve the list of auxiliary devices - cache devices and spares -
	* and open them.
	*/
	error = spa_ld_open_aux_vdevs(spa, type);
	if (error != 0)
	return (error);

	/*
	* Load the metadata for all vdevs. Also check if unopenable devices
	* should be autoreplaced.
	*/
	error = spa_ld_load_vdev_metadata(spa);
	if (error != 0)
	return (error);

	error = spa_ld_load_dedup_tables(spa);
	if (error != 0)
	return (error);

	/*
	* Verify the logs now to make sure we don't have any unexpected errors
	* when we claim log blocks later.
	*/
	error = spa_ld_verify_logs(spa, type, ereport);
	if (error != 0)
	return (error);

	if (missing_feat_write) {
	ASSERT(spa->spa_load_state == SPA_LOAD_TRYIMPORT);

	/*
	* At this point, we know that we can open the pool in
	* read-only mode but not read-write mode. We now have enough
	* information and can return to userland.
	*/
	return (spa_vdev_err(spa->spa_root_vdev, VDEV_AUX_UNSUP_FEAT,
	ENOTSUP));
	}

	/*
	* Traverse the last txgs to make sure the pool was left off in a safe
	* state. When performing an extreme rewind, we verify the whole pool,
	* which can take a very long time.
	*/
	error = spa_ld_verify_pool_data(spa);
	if (error != 0)
	return (error);

	/*
	* Calculate the deflated space for the pool. This must be done before
	* we write anything to the pool because we'd need to update the space
	* accounting using the deflated sizes.
	*/
	spa_update_dspace(spa);

	/*
	* We have now retrieved all the information we needed to open the
	* pool. If we are importing the pool in read-write mode, a few
	* additional steps must be performed to finish the import.
	*/
	if (spa_writeable(spa) && (spa->spa_load_state == SPA_LOAD_RECOVER \|\|
	spa->spa_load_max_txg == UINT64_MAX)) {
	uint64_t config_cache_txg = spa->spa_config_txg;

	ASSERT(spa->spa_load_state != SPA_LOAD_TRYIMPORT);

	/*
	* In case of a checkpoint rewind, log the original txg
	* of the checkpointed uberblock.
	*/
	if (checkpoint_rewind) {
	spa_history_log_internal(spa, "checkpoint rewind",
	NULL, "rewound state to txg=%llu",
	(u_longlong_t)spa->spa_uberblock.ub_checkpoint_txg);
	}

	/*
	* Traverse the ZIL and claim all blocks.
	*/
	spa_ld_claim_log_blocks(spa);

	/*
	* Kick-off the syncing thread.
	*/
	spa->spa_sync_on = B_TRUE;
	txg_sync_start(spa->spa_dsl_pool);
	mmp_thread_start(spa);

	/*
	* Wait for all claims to sync. We sync up to the highest
	* claimed log block birth time so that claimed log blocks
	* don't appear to be from the future. spa_claim_max_txg
	* will have been set for us by ZIL traversal operations
	* performed above.
	*/
	txg_wait_synced(spa->spa_dsl_pool, spa->spa_claim_max_txg);

	/*
	* Check if we need to request an update of the config. On the
	* next sync, we would update the config stored in vdev labels
	* and the cachefile (by default /etc/zfs/zpool.cache).
	*/
	spa_ld_check_for_config_update(spa, config_cache_txg,
	update_config_cache);

	/*
	* Check if a rebuild was in progress and if so resume it.
	* Then check all DTLs to see if anything needs resilvering.
	* The resilver will be deferred if a rebuild was started.
	*/
	if (vdev_rebuild_active(spa->spa_root_vdev)) {
	vdev_rebuild_restart(spa);
	} else if (!dsl_scan_resilvering(spa->spa_dsl_pool) &&
	vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL)) {
	spa_async_request(spa, SPA_ASYNC_RESILVER);
	}

	/*
	* Log the fact that we booted up (so that we can detect if
	* we rebooted in the middle of an operation).
	*/
	spa_history_log_version(spa, "open", NULL);

	spa_restart_removal(spa);
	spa_spawn_aux_threads(spa);

	/*
	* Delete any inconsistent datasets.
	*
	* Note:
	* Since we may be issuing deletes for clones here,
	* we make sure to do so after we've spawned all the
	* auxiliary threads above (from which the livelist
	* deletion zthr is part of).
	*/
	(void) dmu_objset_find(spa_name(spa),
	dsl_destroy_inconsistent, NULL, DS_FIND_CHILDREN);

	/*
	* Clean up any stale temporary dataset userrefs.
	*/
	dsl_pool_clean_tmp_userrefs(spa->spa_dsl_pool);

	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	vdev_initialize_restart(spa->spa_root_vdev);
	vdev_trim_restart(spa->spa_root_vdev);
	vdev_autotrim_restart(spa);
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	}

	spa_import_progress_remove(spa_guid(spa));
	spa_async_request(spa, SPA_ASYNC_L2CACHE_REBUILD);

	spa_load_note(spa, "LOADED");

	return (0);
	}

	static int
	spa_load_retry(spa_t *spa, spa_load_state_t state)
	{
	spa_mode_t mode = spa->spa_mode;

	spa_unload(spa);
	spa_deactivate(spa);

	spa->spa_load_max_txg = spa->spa_uberblock.ub_txg - 1;

	spa_activate(spa, mode);
	spa_async_suspend(spa);

	spa_load_note(spa, "spa_load_retry: rewind, max txg: %llu",
	(u_longlong_t)spa->spa_load_max_txg);

	return (spa_load(spa, state, SPA_IMPORT_EXISTING));
	}

	/*
	* If spa_load() fails this function will try loading prior txg's. If
	* 'state' is SPA_LOAD_RECOVER and one of these loads succeeds the pool
	* will be rewound to that txg. If 'state' is not SPA_LOAD_RECOVER this
	* function will not rewind the pool and will return the same error as
	* spa_load().
	*/
	static int
	spa_load_best(spa_t *spa, spa_load_state_t state, uint64_t max_request,
	int rewind_flags)
	{
	nvlist_t *loadinfo = NULL;
	nvlist_t *config = NULL;
	int load_error, rewind_error;
	uint64_t safe_rewind_txg;
	uint64_t min_txg;

	if (spa->spa_load_txg && state == SPA_LOAD_RECOVER) {
	spa->spa_load_max_txg = spa->spa_load_txg;
	spa_set_log_state(spa, SPA_LOG_CLEAR);
	} else {
	spa->spa_load_max_txg = max_request;
	if (max_request != UINT64_MAX)
	spa->spa_extreme_rewind = B_TRUE;
	}

	load_error = rewind_error = spa_load(spa, state, SPA_IMPORT_EXISTING);
	if (load_error == 0)
	return (0);
	if (load_error == ZFS_ERR_NO_CHECKPOINT) {
	/*
	* When attempting checkpoint-rewind on a pool with no
	* checkpoint, we should not attempt to load uberblocks
	* from previous txgs when spa_load fails.
	*/
	ASSERT(spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT);
	spa_import_progress_remove(spa_guid(spa));
	return (load_error);
	}

	if (spa->spa_root_vdev != NULL)
	config = spa_config_generate(spa, NULL, -1ULL, B_TRUE);

	spa->spa_last_ubsync_txg = spa->spa_uberblock.ub_txg;
	spa->spa_last_ubsync_txg_ts = spa->spa_uberblock.ub_timestamp;

	if (rewind_flags & ZPOOL_NEVER_REWIND) {
	nvlist_free(config);
	spa_import_progress_remove(spa_guid(spa));
	return (load_error);
	}

	if (state == SPA_LOAD_RECOVER) {
	/* Price of rolling back is discarding txgs, including log */
	spa_set_log_state(spa, SPA_LOG_CLEAR);
	} else {
	/*
	* If we aren't rolling back save the load info from our first
	* import attempt so that we can restore it after attempting
	* to rewind.
	*/
	loadinfo = spa->spa_load_info;
	spa->spa_load_info = fnvlist_alloc();
	}

	spa->spa_load_max_txg = spa->spa_last_ubsync_txg;
	safe_rewind_txg = spa->spa_last_ubsync_txg - TXG_DEFER_SIZE;
	min_txg = (rewind_flags & ZPOOL_EXTREME_REWIND) ?
	TXG_INITIAL : safe_rewind_txg;

	/*
	* Continue as long as we're finding errors, we're still within
	* the acceptable rewind range, and we're still finding uberblocks
	*/
	while (rewind_error && spa->spa_uberblock.ub_txg >= min_txg &&
	spa->spa_uberblock.ub_txg <= spa->spa_load_max_txg) {
	if (spa->spa_load_max_txg < safe_rewind_txg)
	spa->spa_extreme_rewind = B_TRUE;
	rewind_error = spa_load_retry(spa, state);
	}

	spa->spa_extreme_rewind = B_FALSE;
	spa->spa_load_max_txg = UINT64_MAX;

	if (config && (rewind_error \|\| state != SPA_LOAD_RECOVER))
	spa_config_set(spa, config);
	else
	nvlist_free(config);

	if (state == SPA_LOAD_RECOVER) {
	ASSERT3P(loadinfo, ==, NULL);
	spa_import_progress_remove(spa_guid(spa));
	return (rewind_error);
	} else {
	/* Store the rewind info as part of the initial load info */
	fnvlist_add_nvlist(loadinfo, ZPOOL_CONFIG_REWIND_INFO,
	spa->spa_load_info);

	/* Restore the initial load info */
	fnvlist_free(spa->spa_load_info);
	spa->spa_load_info = loadinfo;

	spa_import_progress_remove(spa_guid(spa));
	return (load_error);
	}
	}

	/*
	* Pool Open/Import
	*
	* The import case is identical to an open except that the configuration is sent
	* down from userland, instead of grabbed from the configuration cache. For the
	* case of an open, the pool configuration will exist in the
	* POOL_STATE_UNINITIALIZED state.
	*
	* The stats information (gen/count/ustats) is used to gather vdev statistics at
	* the same time open the pool, without having to keep around the spa_t in some
	* ambiguous state.
	*/
	static int
	spa_open_common(const char pool, spa_t spapp, void tag, nvlist_t *nvpolicy,
	nvlist_t **config)
	{
	spa_t *spa;
	spa_load_state_t state = SPA_LOAD_OPEN;
	int error;
	int locked = B_FALSE;
	int firstopen = B_FALSE;

	*spapp = NULL;

	/*
	* As disgusting as this is, we need to support recursive calls to this
	* function because dsl_dir_open() is called during spa_load(), and ends
	* up calling spa_open() again. The real fix is to figure out how to
	* avoid dsl_dir_open() calling this in the first place.
	*/
	if (MUTEX_NOT_HELD(&spa_namespace_lock)) {
	mutex_enter(&spa_namespace_lock);
	locked = B_TRUE;
	}

	if ((spa = spa_lookup(pool)) == NULL) {
	if (locked)
	mutex_exit(&spa_namespace_lock);
	return (SET_ERROR(ENOENT));
	}

	if (spa->spa_state == POOL_STATE_UNINITIALIZED) {
	zpool_load_policy_t policy;

	firstopen = B_TRUE;

	zpool_get_load_policy(nvpolicy ? nvpolicy : spa->spa_config,
	&policy);
	if (policy.zlp_rewind & ZPOOL_DO_REWIND)
	state = SPA_LOAD_RECOVER;

	spa_activate(spa, spa_mode_global);

	if (state != SPA_LOAD_RECOVER)
	spa->spa_last_ubsync_txg = spa->spa_load_txg = 0;
	spa->spa_config_source = SPA_CONFIG_SRC_CACHEFILE;

	zfs_dbgmsg("spa_open_common: opening %s", pool);
	error = spa_load_best(spa, state, policy.zlp_txg,
	policy.zlp_rewind);

	if (error == EBADF) {
	/*
	* If vdev_validate() returns failure (indicated by
	* EBADF), it indicates that one of the vdevs indicates
	* that the pool has been exported or destroyed. If
	* this is the case, the config cache is out of sync and
	* we should remove the pool from the namespace.
	*/
	spa_unload(spa);
	spa_deactivate(spa);
	spa_write_cachefile(spa, B_TRUE, B_TRUE);
	spa_remove(spa);
	if (locked)
	mutex_exit(&spa_namespace_lock);
	return (SET_ERROR(ENOENT));
	}

	if (error) {
	/*
	* We can't open the pool, but we still have useful
	* information: the state of each vdev after the
	* attempted vdev_open(). Return this to the user.
	*/
	if (config != NULL && spa->spa_config) {
	*config = fnvlist_dup(spa->spa_config);
	fnvlist_add_nvlist(*config,
	ZPOOL_CONFIG_LOAD_INFO,
	spa->spa_load_info);
	}
	spa_unload(spa);
	spa_deactivate(spa);
	spa->spa_last_open_failed = error;
	if (locked)
	mutex_exit(&spa_namespace_lock);
	*spapp = NULL;
	return (error);
	}
	}

	spa_open_ref(spa, tag);

	if (config != NULL)
	*config = spa_config_generate(spa, NULL, -1ULL, B_TRUE);

	/*
	* If we've recovered the pool, pass back any information we
	* gathered while doing the load.
	*/
	if (state == SPA_LOAD_RECOVER) {
	fnvlist_add_nvlist(*config, ZPOOL_CONFIG_LOAD_INFO,
	spa->spa_load_info);
	}

	if (locked) {
	spa->spa_last_open_failed = 0;
	spa->spa_last_ubsync_txg = 0;
	spa->spa_load_txg = 0;
	mutex_exit(&spa_namespace_lock);
	}

	if (firstopen)
	zvol_create_minors_recursive(spa_name(spa));

	*spapp = spa;

	return (0);
	}

	int
	spa_open_rewind(const char name, spa_t spapp, void tag, nvlist_t *policy,
	nvlist_t **config)
	{
	return (spa_open_common(name, spapp, tag, policy, config));
	}

	int
	spa_open(const char name, spa_t spapp, void tag)
	{
	return (spa_open_common(name, spapp, tag, NULL, NULL));
	}

	/*
	* Lookup the given spa_t, incrementing the inject count in the process,
	* preventing it from being exported or destroyed.
	*/
	spa_t *
	spa_inject_addref(char *name)
	{
	spa_t *spa;

	mutex_enter(&spa_namespace_lock);
	if ((spa = spa_lookup(name)) == NULL) {
	mutex_exit(&spa_namespace_lock);
	return (NULL);
	}
	spa->spa_inject_ref++;
	mutex_exit(&spa_namespace_lock);

	return (spa);
	}

	void
	spa_inject_delref(spa_t *spa)
	{
	mutex_enter(&spa_namespace_lock);
	spa->spa_inject_ref--;
	mutex_exit(&spa_namespace_lock);
	}

	/*
	* Add spares device information to the nvlist.
	*/
	static void
	spa_add_spares(spa_t spa, nvlist_t config)
	{
	nvlist_t **spares;
	uint_t i, nspares;
	nvlist_t *nvroot;
	uint64_t guid;
	vdev_stat_t *vs;
	uint_t vsc;
	uint64_t pool;

	ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER));

	if (spa->spa_spares.sav_count == 0)
	return;

	nvroot = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE);
	VERIFY0(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config,
	ZPOOL_CONFIG_SPARES, &spares, &nspares));
	if (nspares != 0) {
	- fnvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, spares,
	- nspares);
	+ fnvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	+ (const nvlist_t * const *)spares, nspares);
	VERIFY0(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	&spares, &nspares));

	/*
	* Go through and find any spares which have since been
	* repurposed as an active spare. If this is the case, update
	* their status appropriately.
	*/
	for (i = 0; i < nspares; i++) {
	guid = fnvlist_lookup_uint64(spares[i],
	ZPOOL_CONFIG_GUID);
	if (spa_spare_exists(guid, &pool, NULL) &&
	pool != 0ULL) {
	VERIFY0(nvlist_lookup_uint64_array(spares[i],
	ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs,
	&vsc));
	vs->vs_state = VDEV_STATE_CANT_OPEN;
	vs->vs_aux = VDEV_AUX_SPARED;
	}
	}
	}
	}

	/*
	* Add l2cache device information to the nvlist, including vdev stats.
	*/
	static void
	spa_add_l2cache(spa_t spa, nvlist_t config)
	{
	nvlist_t **l2cache;
	uint_t i, j, nl2cache;
	nvlist_t *nvroot;
	uint64_t guid;
	vdev_t *vd;
	vdev_stat_t *vs;
	uint_t vsc;

	ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER));

	if (spa->spa_l2cache.sav_count == 0)
	return;

	nvroot = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE);
	VERIFY0(nvlist_lookup_nvlist_array(spa->spa_l2cache.sav_config,
	ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache));
	if (nl2cache != 0) {
	- fnvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, l2cache,
	- nl2cache);
	+ fnvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
	+ (const nvlist_t * const *)l2cache, nl2cache);
	VERIFY0(nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
	&l2cache, &nl2cache));

	/*
	* Update level 2 cache device stats.
	*/

	for (i = 0; i < nl2cache; i++) {
	guid = fnvlist_lookup_uint64(l2cache[i],
	ZPOOL_CONFIG_GUID);

	vd = NULL;
	for (j = 0; j < spa->spa_l2cache.sav_count; j++) {
	if (guid ==
	spa->spa_l2cache.sav_vdevs[j]->vdev_guid) {
	vd = spa->spa_l2cache.sav_vdevs[j];
	break;
	}
	}
	ASSERT(vd != NULL);

	VERIFY0(nvlist_lookup_uint64_array(l2cache[i],
	ZPOOL_CONFIG_VDEV_STATS, (uint64_t **)&vs, &vsc));
	vdev_get_stats(vd, vs);
	vdev_config_generate_stats(vd, l2cache[i]);

	}
	}
	}

	static void
	spa_feature_stats_from_disk(spa_t spa, nvlist_t features)
	{
	zap_cursor_t zc;
	zap_attribute_t za;

	if (spa->spa_feat_for_read_obj != 0) {
	for (zap_cursor_init(&zc, spa->spa_meta_objset,
	spa->spa_feat_for_read_obj);
	zap_cursor_retrieve(&zc, &za) == 0;
	zap_cursor_advance(&zc)) {
	ASSERT(za.za_integer_length == sizeof (uint64_t) &&
	za.za_num_integers == 1);
	VERIFY0(nvlist_add_uint64(features, za.za_name,
	za.za_first_integer));
	}
	zap_cursor_fini(&zc);
	}

	if (spa->spa_feat_for_write_obj != 0) {
	for (zap_cursor_init(&zc, spa->spa_meta_objset,
	spa->spa_feat_for_write_obj);
	zap_cursor_retrieve(&zc, &za) == 0;
	zap_cursor_advance(&zc)) {
	ASSERT(za.za_integer_length == sizeof (uint64_t) &&
	za.za_num_integers == 1);
	VERIFY0(nvlist_add_uint64(features, za.za_name,
	za.za_first_integer));
	}
	zap_cursor_fini(&zc);
	}
	}

	static void
	spa_feature_stats_from_cache(spa_t spa, nvlist_t features)
	{
	int i;

	for (i = 0; i < SPA_FEATURES; i++) {
	zfeature_info_t feature = spa_feature_table[i];
	uint64_t refcount;

	if (feature_get_refcount(spa, &feature, &refcount) != 0)
	continue;

	VERIFY0(nvlist_add_uint64(features, feature.fi_guid, refcount));
	}
	}

	/*
	* Store a list of pool features and their reference counts in the
	* config.
	*
	* The first time this is called on a spa, allocate a new nvlist, fetch
	* the pool features and reference counts from disk, then save the list
	* in the spa. In subsequent calls on the same spa use the saved nvlist
	* and refresh its values from the cached reference counts. This
	* ensures we don't block here on I/O on a suspended pool so 'zpool
	* clear' can resume the pool.
	*/
	static void
	spa_add_feature_stats(spa_t spa, nvlist_t config)
	{
	nvlist_t *features;

	ASSERT(spa_config_held(spa, SCL_CONFIG, RW_READER));

	mutex_enter(&spa->spa_feat_stats_lock);
	features = spa->spa_feat_stats;

	if (features != NULL) {
	spa_feature_stats_from_cache(spa, features);
	} else {
	VERIFY0(nvlist_alloc(&features, NV_UNIQUE_NAME, KM_SLEEP));
	spa->spa_feat_stats = features;
	spa_feature_stats_from_disk(spa, features);
	}

	VERIFY0(nvlist_add_nvlist(config, ZPOOL_CONFIG_FEATURE_STATS,
	features));

	mutex_exit(&spa->spa_feat_stats_lock);
	}

	int
	spa_get_stats(const char name, nvlist_t *config,
	char *altroot, size_t buflen)
	{
	int error;
	spa_t *spa;

	*config = NULL;
	error = spa_open_common(name, &spa, FTAG, NULL, config);

	if (spa != NULL) {
	/*
	* This still leaves a window of inconsistency where the spares
	* or l2cache devices could change and the config would be
	* self-inconsistent.
	*/
	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);

	if (*config != NULL) {
	uint64_t loadtimes[2];

	loadtimes[0] = spa->spa_loaded_ts.tv_sec;
	loadtimes[1] = spa->spa_loaded_ts.tv_nsec;
	fnvlist_add_uint64_array(*config,
	ZPOOL_CONFIG_LOADED_TIME, loadtimes, 2);

	fnvlist_add_uint64(*config,
	ZPOOL_CONFIG_ERRCOUNT,
	spa_get_errlog_size(spa));

	if (spa_suspended(spa)) {
	fnvlist_add_uint64(*config,
	ZPOOL_CONFIG_SUSPENDED,
	spa->spa_failmode);
	fnvlist_add_uint64(*config,
	ZPOOL_CONFIG_SUSPENDED_REASON,
	spa->spa_suspended);
	}

	spa_add_spares(spa, *config);
	spa_add_l2cache(spa, *config);
	spa_add_feature_stats(spa, *config);
	}
	}

	/*
	* We want to get the alternate root even for faulted pools, so we cheat
	* and call spa_lookup() directly.
	*/
	if (altroot) {
	if (spa == NULL) {
	mutex_enter(&spa_namespace_lock);
	spa = spa_lookup(name);
	if (spa)
	spa_altroot(spa, altroot, buflen);
	else
	altroot[0] = '\0';
	spa = NULL;
	mutex_exit(&spa_namespace_lock);
	} else {
	spa_altroot(spa, altroot, buflen);
	}
	}

	if (spa != NULL) {
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	spa_close(spa, FTAG);
	}

	return (error);
	}

	/*
	* Validate that the auxiliary device array is well formed. We must have an
	* array of nvlists, each which describes a valid leaf vdev. If this is an
	* import (mode is VDEV_ALLOC_SPARE), then we allow corrupted spares to be
	* specified, as long as they are well-formed.
	*/
	static int
	spa_validate_aux_devs(spa_t spa, nvlist_t nvroot, uint64_t crtxg, int mode,
	spa_aux_vdev_t sav, const char config, uint64_t version,
	vdev_labeltype_t label)
	{
	nvlist_t **dev;
	uint_t i, ndev;
	vdev_t *vd;
	int error;

	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	/*
	* It's acceptable to have no devs specified.
	*/
	if (nvlist_lookup_nvlist_array(nvroot, config, &dev, &ndev) != 0)
	return (0);

	if (ndev == 0)
	return (SET_ERROR(EINVAL));

	/*
	* Make sure the pool is formatted with a version that supports this
	* device type.
	*/
	if (spa_version(spa) < version)
	return (SET_ERROR(ENOTSUP));

	/*
	* Set the pending device list so we correctly handle device in-use
	* checking.
	*/
	sav->sav_pending = dev;
	sav->sav_npending = ndev;

	for (i = 0; i < ndev; i++) {
	if ((error = spa_config_parse(spa, &vd, dev[i], NULL, 0,
	mode)) != 0)
	goto out;

	if (!vd->vdev_ops->vdev_op_leaf) {
	vdev_free(vd);
	error = SET_ERROR(EINVAL);
	goto out;
	}

	vd->vdev_top = vd;

	if ((error = vdev_open(vd)) == 0 &&
	(error = vdev_label_init(vd, crtxg, label)) == 0) {
	fnvlist_add_uint64(dev[i], ZPOOL_CONFIG_GUID,
	vd->vdev_guid);
	}

	vdev_free(vd);

	if (error &&
	(mode != VDEV_ALLOC_SPARE && mode != VDEV_ALLOC_L2CACHE))
	goto out;
	else
	error = 0;
	}

	out:
	sav->sav_pending = NULL;
	sav->sav_npending = 0;
	return (error);
	}

	static int
	spa_validate_aux(spa_t spa, nvlist_t nvroot, uint64_t crtxg, int mode)
	{
	int error;

	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	if ((error = spa_validate_aux_devs(spa, nvroot, crtxg, mode,
	&spa->spa_spares, ZPOOL_CONFIG_SPARES, SPA_VERSION_SPARES,
	VDEV_LABEL_SPARE)) != 0) {
	return (error);
	}

	return (spa_validate_aux_devs(spa, nvroot, crtxg, mode,
	&spa->spa_l2cache, ZPOOL_CONFIG_L2CACHE, SPA_VERSION_L2CACHE,
	VDEV_LABEL_L2CACHE));
	}

	static void
	spa_set_aux_vdevs(spa_aux_vdev_t sav, nvlist_t *devs, int ndevs,
	const char *config)
	{
	int i;

	if (sav->sav_config != NULL) {
	nvlist_t **olddevs;
	uint_t oldndevs;
	nvlist_t **newdevs;

	/*
	* Generate new dev list by concatenating with the
	* current dev list.
	*/
	VERIFY0(nvlist_lookup_nvlist_array(sav->sav_config, config,
	&olddevs, &oldndevs));

	newdevs = kmem_alloc(sizeof (void )
	(ndevs + oldndevs), KM_SLEEP);
	for (i = 0; i < oldndevs; i++)
	newdevs[i] = fnvlist_dup(olddevs[i]);
	for (i = 0; i < ndevs; i++)
	newdevs[i + oldndevs] = fnvlist_dup(devs[i]);

	fnvlist_remove(sav->sav_config, config);

	- fnvlist_add_nvlist_array(sav->sav_config, config, newdevs,
	- ndevs + oldndevs);
	+ fnvlist_add_nvlist_array(sav->sav_config, config,
	+ (const nvlist_t * const *)newdevs, ndevs + oldndevs);
	for (i = 0; i < oldndevs + ndevs; i++)
	nvlist_free(newdevs[i]);
	kmem_free(newdevs, (oldndevs + ndevs) * sizeof (void *));
	} else {
	/*
	* Generate a new dev list.
	*/
	sav->sav_config = fnvlist_alloc();
	- fnvlist_add_nvlist_array(sav->sav_config, config, devs, ndevs);
	+ fnvlist_add_nvlist_array(sav->sav_config, config,
	+ (const nvlist_t * const *)devs, ndevs);
	}
	}

	/*
	* Stop and drop level 2 ARC devices
	*/
	void
	spa_l2cache_drop(spa_t *spa)
	{
	vdev_t *vd;
	int i;
	spa_aux_vdev_t *sav = &spa->spa_l2cache;

	for (i = 0; i < sav->sav_count; i++) {
	uint64_t pool;

	vd = sav->sav_vdevs[i];
	ASSERT(vd != NULL);

	if (spa_l2cache_exists(vd->vdev_guid, &pool) &&
	pool != 0ULL && l2arc_vdev_present(vd))
	l2arc_remove_vdev(vd);
	}
	}

	/*
	* Verify encryption parameters for spa creation. If we are encrypting, we must
	* have the encryption feature flag enabled.
	*/
	static int
	spa_create_check_encryption_params(dsl_crypto_params_t *dcp,
	boolean_t has_encryption)
	{
	if (dcp->cp_crypt != ZIO_CRYPT_OFF &&
	dcp->cp_crypt != ZIO_CRYPT_INHERIT &&
	!has_encryption)
	return (SET_ERROR(ENOTSUP));

	return (dmu_objset_create_crypt_check(NULL, dcp, NULL));
	}

	/*
	* Pool Creation
	*/
	int
	spa_create(const char pool, nvlist_t nvroot, nvlist_t *props,
	nvlist_t zplprops, dsl_crypto_params_t dcp)
	{
	spa_t *spa;
	char *altroot = NULL;
	vdev_t *rvd;
	dsl_pool_t *dp;
	dmu_tx_t *tx;
	int error = 0;
	uint64_t txg = TXG_INITIAL;
	nvlist_t spares, l2cache;
	uint_t nspares, nl2cache;
	uint64_t version, obj, ndraid = 0;
	boolean_t has_features;
	boolean_t has_encryption;
	boolean_t has_allocclass;
	spa_feature_t feat;
	char *feat_name;
	char *poolname;
	nvlist_t *nvl;

	if (props == NULL \|\|
	nvlist_lookup_string(props, "tname", &poolname) != 0)
	poolname = (char *)pool;

	/*
	* If this pool already exists, return failure.
	*/
	mutex_enter(&spa_namespace_lock);
	if (spa_lookup(poolname) != NULL) {
	mutex_exit(&spa_namespace_lock);
	return (SET_ERROR(EEXIST));
	}

	/*
	* Allocate a new spa_t structure.
	*/
	nvl = fnvlist_alloc();
	fnvlist_add_string(nvl, ZPOOL_CONFIG_POOL_NAME, pool);
	(void) nvlist_lookup_string(props,
	zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot);
	spa = spa_add(poolname, nvl, altroot);
	fnvlist_free(nvl);
	spa_activate(spa, spa_mode_global);

	if (props && (error = spa_prop_validate(spa, props))) {
	spa_deactivate(spa);
	spa_remove(spa);
	mutex_exit(&spa_namespace_lock);
	return (error);
	}

	/*
	* Temporary pool names should never be written to disk.
	*/
	if (poolname != pool)
	spa->spa_import_flags \|= ZFS_IMPORT_TEMP_NAME;

	has_features = B_FALSE;
	has_encryption = B_FALSE;
	has_allocclass = B_FALSE;
	for (nvpair_t *elem = nvlist_next_nvpair(props, NULL);
	elem != NULL; elem = nvlist_next_nvpair(props, elem)) {
	if (zpool_prop_feature(nvpair_name(elem))) {
	has_features = B_TRUE;

	feat_name = strchr(nvpair_name(elem), '@') + 1;
	VERIFY0(zfeature_lookup_name(feat_name, &feat));
	if (feat == SPA_FEATURE_ENCRYPTION)
	has_encryption = B_TRUE;
	if (feat == SPA_FEATURE_ALLOCATION_CLASSES)
	has_allocclass = B_TRUE;
	}
	}

	/* verify encryption params, if they were provided */
	if (dcp != NULL) {
	error = spa_create_check_encryption_params(dcp, has_encryption);
	if (error != 0) {
	spa_deactivate(spa);
	spa_remove(spa);
	mutex_exit(&spa_namespace_lock);
	return (error);
	}
	}
	if (!has_allocclass && zfs_special_devs(nvroot, NULL)) {
	spa_deactivate(spa);
	spa_remove(spa);
	mutex_exit(&spa_namespace_lock);
	return (ENOTSUP);
	}

	if (has_features \|\| nvlist_lookup_uint64(props,
	zpool_prop_to_name(ZPOOL_PROP_VERSION), &version) != 0) {
	version = SPA_VERSION;
	}
	ASSERT(SPA_VERSION_IS_SUPPORTED(version));

	spa->spa_first_txg = txg;
	spa->spa_uberblock.ub_txg = txg - 1;
	spa->spa_uberblock.ub_version = version;
	spa->spa_ubsync = spa->spa_uberblock;
	spa->spa_load_state = SPA_LOAD_CREATE;
	spa->spa_removing_phys.sr_state = DSS_NONE;
	spa->spa_removing_phys.sr_removing_vdev = -1;
	spa->spa_removing_phys.sr_prev_indirect_vdev = -1;
	spa->spa_indirect_vdevs_loaded = B_TRUE;

	/*
	* Create "The Godfather" zio to hold all async IOs
	*/
	spa->spa_async_zio_root = kmem_alloc(max_ncpus * sizeof (void *),
	KM_SLEEP);
	for (int i = 0; i < max_ncpus; i++) {
	spa->spa_async_zio_root[i] = zio_root(spa, NULL, NULL,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_SPECULATIVE \|
	ZIO_FLAG_GODFATHER);
	}

	/*
	* Create the root vdev.
	*/
	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);

	error = spa_config_parse(spa, &rvd, nvroot, NULL, 0, VDEV_ALLOC_ADD);

	ASSERT(error != 0 \|\| rvd != NULL);
	ASSERT(error != 0 \|\| spa->spa_root_vdev == rvd);

	if (error == 0 && !zfs_allocatable_devs(nvroot))
	error = SET_ERROR(EINVAL);

	if (error == 0 &&
	(error = vdev_create(rvd, txg, B_FALSE)) == 0 &&
	(error = vdev_draid_spare_create(nvroot, rvd, &ndraid, 0)) == 0 &&
	(error = spa_validate_aux(spa, nvroot, txg, VDEV_ALLOC_ADD)) == 0) {
	/*
	* instantiate the metaslab groups (this will dirty the vdevs)
	* we can no longer error exit past this point
	*/
	for (int c = 0; error == 0 && c < rvd->vdev_children; c++) {
	vdev_t *vd = rvd->vdev_child[c];

	vdev_metaslab_set_size(vd);
	vdev_expand(vd, txg);
	}
	}

	spa_config_exit(spa, SCL_ALL, FTAG);

	if (error != 0) {
	spa_unload(spa);
	spa_deactivate(spa);
	spa_remove(spa);
	mutex_exit(&spa_namespace_lock);
	return (error);
	}

	/*
	* Get the list of spares, if specified.
	*/
	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	&spares, &nspares) == 0) {
	spa->spa_spares.sav_config = fnvlist_alloc();
	fnvlist_add_nvlist_array(spa->spa_spares.sav_config,
	- ZPOOL_CONFIG_SPARES, spares, nspares);
	+ ZPOOL_CONFIG_SPARES, (const nvlist_t * const *)spares,
	+ nspares);
	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	spa_load_spares(spa);
	spa_config_exit(spa, SCL_ALL, FTAG);
	spa->spa_spares.sav_sync = B_TRUE;
	}

	/*
	* Get the list of level 2 cache devices, if specified.
	*/
	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
	&l2cache, &nl2cache) == 0) {
	- spa->spa_l2cache.sav_config = fnvlist_alloc();
	+ VERIFY0(nvlist_alloc(&spa->spa_l2cache.sav_config,
	+ NV_UNIQUE_NAME, KM_SLEEP));
	fnvlist_add_nvlist_array(spa->spa_l2cache.sav_config,
	- ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache);
	+ ZPOOL_CONFIG_L2CACHE, (const nvlist_t * const *)l2cache,
	+ nl2cache);
	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	spa_load_l2cache(spa);
	spa_config_exit(spa, SCL_ALL, FTAG);
	spa->spa_l2cache.sav_sync = B_TRUE;
	}

	spa->spa_is_initializing = B_TRUE;
	spa->spa_dsl_pool = dp = dsl_pool_create(spa, zplprops, dcp, txg);
	spa->spa_is_initializing = B_FALSE;

	/*
	* Create DDTs (dedup tables).
	*/
	ddt_create(spa);

	spa_update_dspace(spa);

	tx = dmu_tx_create_assigned(dp, txg);

	/*
	* Create the pool's history object.
	*/
	if (version >= SPA_VERSION_ZPOOL_HISTORY && !spa->spa_history)
	spa_history_create_obj(spa, tx);

	spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_CREATE);
	spa_history_log_version(spa, "create", tx);

	/*
	* Create the pool config object.
	*/
	spa->spa_config_object = dmu_object_alloc(spa->spa_meta_objset,
	DMU_OT_PACKED_NVLIST, SPA_CONFIG_BLOCKSIZE,
	DMU_OT_PACKED_NVLIST_SIZE, sizeof (uint64_t), tx);

	if (zap_add(spa->spa_meta_objset,
	DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CONFIG,
	sizeof (uint64_t), 1, &spa->spa_config_object, tx) != 0) {
	cmn_err(CE_PANIC, "failed to add pool config");
	}

	if (zap_add(spa->spa_meta_objset,
	DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CREATION_VERSION,
	sizeof (uint64_t), 1, &version, tx) != 0) {
	cmn_err(CE_PANIC, "failed to add pool version");
	}

	/* Newly created pools with the right version are always deflated. */
	if (version >= SPA_VERSION_RAIDZ_DEFLATE) {
	spa->spa_deflate = TRUE;
	if (zap_add(spa->spa_meta_objset,
	DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DEFLATE,
	sizeof (uint64_t), 1, &spa->spa_deflate, tx) != 0) {
	cmn_err(CE_PANIC, "failed to add deflate");
	}
	}

	/*
	* Create the deferred-free bpobj. Turn off compression
	* because sync-to-convergence takes longer if the blocksize
	* keeps changing.
	*/
	obj = bpobj_alloc(spa->spa_meta_objset, 1 << 14, tx);
	dmu_object_set_compress(spa->spa_meta_objset, obj,
	ZIO_COMPRESS_OFF, tx);
	if (zap_add(spa->spa_meta_objset,
	DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SYNC_BPOBJ,
	sizeof (uint64_t), 1, &obj, tx) != 0) {
	cmn_err(CE_PANIC, "failed to add bpobj");
	}
	VERIFY3U(0, ==, bpobj_open(&spa->spa_deferred_bpobj,
	spa->spa_meta_objset, obj));

	/*
	* Generate some random noise for salted checksums to operate on.
	*/
	(void) random_get_pseudo_bytes(spa->spa_cksum_salt.zcs_bytes,
	sizeof (spa->spa_cksum_salt.zcs_bytes));

	/*
	* Set pool properties.
	*/
	spa->spa_bootfs = zpool_prop_default_numeric(ZPOOL_PROP_BOOTFS);
	spa->spa_delegation = zpool_prop_default_numeric(ZPOOL_PROP_DELEGATION);
	spa->spa_failmode = zpool_prop_default_numeric(ZPOOL_PROP_FAILUREMODE);
	spa->spa_autoexpand = zpool_prop_default_numeric(ZPOOL_PROP_AUTOEXPAND);
	spa->spa_multihost = zpool_prop_default_numeric(ZPOOL_PROP_MULTIHOST);
	spa->spa_autotrim = zpool_prop_default_numeric(ZPOOL_PROP_AUTOTRIM);

	if (props != NULL) {
	spa_configfile_set(spa, props, B_FALSE);
	spa_sync_props(props, tx);
	}

	for (int i = 0; i < ndraid; i++)
	spa_feature_incr(spa, SPA_FEATURE_DRAID, tx);

	dmu_tx_commit(tx);

	spa->spa_sync_on = B_TRUE;
	txg_sync_start(dp);
	mmp_thread_start(spa);
	txg_wait_synced(dp, txg);

	spa_spawn_aux_threads(spa);

	spa_write_cachefile(spa, B_FALSE, B_TRUE);

	/*
	* Don't count references from objsets that are already closed
	* and are making their way through the eviction process.
	*/
	spa_evicting_os_wait(spa);
	spa->spa_minref = zfs_refcount_count(&spa->spa_refcount);
	spa->spa_load_state = SPA_LOAD_NONE;

	mutex_exit(&spa_namespace_lock);

	return (0);
	}

	/*
	* Import a non-root pool into the system.
	*/
	int
	spa_import(char pool, nvlist_t config, nvlist_t *props, uint64_t flags)
	{
	spa_t *spa;
	char *altroot = NULL;
	spa_load_state_t state = SPA_LOAD_IMPORT;
	zpool_load_policy_t policy;
	spa_mode_t mode = spa_mode_global;
	uint64_t readonly = B_FALSE;
	int error;
	nvlist_t *nvroot;
	nvlist_t spares, l2cache;
	uint_t nspares, nl2cache;

	/*
	* If a pool with this name exists, return failure.
	*/
	mutex_enter(&spa_namespace_lock);
	if (spa_lookup(pool) != NULL) {
	mutex_exit(&spa_namespace_lock);
	return (SET_ERROR(EEXIST));
	}

	/*
	* Create and initialize the spa structure.
	*/
	(void) nvlist_lookup_string(props,
	zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot);
	(void) nvlist_lookup_uint64(props,
	zpool_prop_to_name(ZPOOL_PROP_READONLY), &readonly);
	if (readonly)
	mode = SPA_MODE_READ;
	spa = spa_add(pool, config, altroot);
	spa->spa_import_flags = flags;

	/*
	* Verbatim import - Take a pool and insert it into the namespace
	* as if it had been loaded at boot.
	*/
	if (spa->spa_import_flags & ZFS_IMPORT_VERBATIM) {
	if (props != NULL)
	spa_configfile_set(spa, props, B_FALSE);

	spa_write_cachefile(spa, B_FALSE, B_TRUE);
	spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_IMPORT);
	zfs_dbgmsg("spa_import: verbatim import of %s", pool);
	mutex_exit(&spa_namespace_lock);
	return (0);
	}

	spa_activate(spa, mode);

	/*
	* Don't start async tasks until we know everything is healthy.
	*/
	spa_async_suspend(spa);

	zpool_get_load_policy(config, &policy);
	if (policy.zlp_rewind & ZPOOL_DO_REWIND)
	state = SPA_LOAD_RECOVER;

	spa->spa_config_source = SPA_CONFIG_SRC_TRYIMPORT;

	if (state != SPA_LOAD_RECOVER) {
	spa->spa_last_ubsync_txg = spa->spa_load_txg = 0;
	zfs_dbgmsg("spa_import: importing %s", pool);
	} else {
	zfs_dbgmsg("spa_import: importing %s, max_txg=%lld "
	"(RECOVERY MODE)", pool, (longlong_t)policy.zlp_txg);
	}
	error = spa_load_best(spa, state, policy.zlp_txg, policy.zlp_rewind);

	/*
	* Propagate anything learned while loading the pool and pass it
	* back to caller (i.e. rewind info, missing devices, etc).
	*/
	fnvlist_add_nvlist(config, ZPOOL_CONFIG_LOAD_INFO, spa->spa_load_info);

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	/*
	* Toss any existing sparelist, as it doesn't have any validity
	* anymore, and conflicts with spa_has_spare().
	*/
	if (spa->spa_spares.sav_config) {
	nvlist_free(spa->spa_spares.sav_config);
	spa->spa_spares.sav_config = NULL;
	spa_load_spares(spa);
	}
	if (spa->spa_l2cache.sav_config) {
	nvlist_free(spa->spa_l2cache.sav_config);
	spa->spa_l2cache.sav_config = NULL;
	spa_load_l2cache(spa);
	}

	nvroot = fnvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE);
	spa_config_exit(spa, SCL_ALL, FTAG);

	if (props != NULL)
	spa_configfile_set(spa, props, B_FALSE);

	if (error != 0 \|\| (props && spa_writeable(spa) &&
	(error = spa_prop_set(spa, props)))) {
	spa_unload(spa);
	spa_deactivate(spa);
	spa_remove(spa);
	mutex_exit(&spa_namespace_lock);
	return (error);
	}

	spa_async_resume(spa);

	/*
	* Override any spares and level 2 cache devices as specified by
	* the user, as these may have correct device names/devids, etc.
	*/
	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	&spares, &nspares) == 0) {
	if (spa->spa_spares.sav_config)
	fnvlist_remove(spa->spa_spares.sav_config,
	ZPOOL_CONFIG_SPARES);
	else
	spa->spa_spares.sav_config = fnvlist_alloc();
	fnvlist_add_nvlist_array(spa->spa_spares.sav_config,
	- ZPOOL_CONFIG_SPARES, spares, nspares);
	+ ZPOOL_CONFIG_SPARES, (const nvlist_t * const *)spares,
	+ nspares);
	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	spa_load_spares(spa);
	spa_config_exit(spa, SCL_ALL, FTAG);
	spa->spa_spares.sav_sync = B_TRUE;
	}
	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE,
	&l2cache, &nl2cache) == 0) {
	if (spa->spa_l2cache.sav_config)
	fnvlist_remove(spa->spa_l2cache.sav_config,
	ZPOOL_CONFIG_L2CACHE);
	else
	spa->spa_l2cache.sav_config = fnvlist_alloc();
	fnvlist_add_nvlist_array(spa->spa_l2cache.sav_config,
	- ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache);
	+ ZPOOL_CONFIG_L2CACHE, (const nvlist_t * const *)l2cache,
	+ nl2cache);
	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	spa_load_l2cache(spa);
	spa_config_exit(spa, SCL_ALL, FTAG);
	spa->spa_l2cache.sav_sync = B_TRUE;
	}

	/*
	* Check for any removed devices.
	*/
	if (spa->spa_autoreplace) {
	spa_aux_check_removed(&spa->spa_spares);
	spa_aux_check_removed(&spa->spa_l2cache);
	}

	if (spa_writeable(spa)) {
	/*
	* Update the config cache to include the newly-imported pool.
	*/
	spa_config_update(spa, SPA_CONFIG_UPDATE_POOL);
	}

	/*
	* It's possible that the pool was expanded while it was exported.
	* We kick off an async task to handle this for us.
	*/
	spa_async_request(spa, SPA_ASYNC_AUTOEXPAND);

	spa_history_log_version(spa, "import", NULL);

	spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_IMPORT);

	mutex_exit(&spa_namespace_lock);

	zvol_create_minors_recursive(pool);

	return (0);
	}

	nvlist_t *
	spa_tryimport(nvlist_t *tryconfig)
	{
	nvlist_t *config = NULL;
	char poolname, cachefile;
	spa_t *spa;
	uint64_t state;
	int error;
	zpool_load_policy_t policy;

	if (nvlist_lookup_string(tryconfig, ZPOOL_CONFIG_POOL_NAME, &poolname))
	return (NULL);

	if (nvlist_lookup_uint64(tryconfig, ZPOOL_CONFIG_POOL_STATE, &state))
	return (NULL);

	/*
	* Create and initialize the spa structure.
	*/
	mutex_enter(&spa_namespace_lock);
	spa = spa_add(TRYIMPORT_NAME, tryconfig, NULL);
	spa_activate(spa, SPA_MODE_READ);

	/*
	* Rewind pool if a max txg was provided.
	*/
	zpool_get_load_policy(spa->spa_config, &policy);
	if (policy.zlp_txg != UINT64_MAX) {
	spa->spa_load_max_txg = policy.zlp_txg;
	spa->spa_extreme_rewind = B_TRUE;
	zfs_dbgmsg("spa_tryimport: importing %s, max_txg=%lld",
	poolname, (longlong_t)policy.zlp_txg);
	} else {
	zfs_dbgmsg("spa_tryimport: importing %s", poolname);
	}

	if (nvlist_lookup_string(tryconfig, ZPOOL_CONFIG_CACHEFILE, &cachefile)
	== 0) {
	zfs_dbgmsg("spa_tryimport: using cachefile '%s'", cachefile);
	spa->spa_config_source = SPA_CONFIG_SRC_CACHEFILE;
	} else {
	spa->spa_config_source = SPA_CONFIG_SRC_SCAN;
	}

	error = spa_load(spa, SPA_LOAD_TRYIMPORT, SPA_IMPORT_EXISTING);

	/*
	* If 'tryconfig' was at least parsable, return the current config.
	*/
	if (spa->spa_root_vdev != NULL) {
	config = spa_config_generate(spa, NULL, -1ULL, B_TRUE);
	fnvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, poolname);
	fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE, state);
	fnvlist_add_uint64(config, ZPOOL_CONFIG_TIMESTAMP,
	spa->spa_uberblock.ub_timestamp);
	fnvlist_add_nvlist(config, ZPOOL_CONFIG_LOAD_INFO,
	spa->spa_load_info);
	fnvlist_add_uint64(config, ZPOOL_CONFIG_ERRATA,
	spa->spa_errata);

	/*
	* If the bootfs property exists on this pool then we
	* copy it out so that external consumers can tell which
	* pools are bootable.
	*/
	if ((!error \|\| error == EEXIST) && spa->spa_bootfs) {
	char *tmpname = kmem_alloc(MAXPATHLEN, KM_SLEEP);

	/*
	* We have to play games with the name since the
	* pool was opened as TRYIMPORT_NAME.
	*/
	if (dsl_dsobj_to_dsname(spa_name(spa),
	spa->spa_bootfs, tmpname) == 0) {
	char *cp;
	char *dsname;

	dsname = kmem_alloc(MAXPATHLEN, KM_SLEEP);

	cp = strchr(tmpname, '/');
	if (cp == NULL) {
	(void) strlcpy(dsname, tmpname,
	MAXPATHLEN);
	} else {
	(void) snprintf(dsname, MAXPATHLEN,
	"%s/%s", poolname, ++cp);
	}
	fnvlist_add_string(config, ZPOOL_CONFIG_BOOTFS,
	dsname);
	kmem_free(dsname, MAXPATHLEN);
	}
	kmem_free(tmpname, MAXPATHLEN);
	}

	/*
	* Add the list of hot spares and level 2 cache devices.
	*/
	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	spa_add_spares(spa, config);
	spa_add_l2cache(spa, config);
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	}

	spa_unload(spa);
	spa_deactivate(spa);
	spa_remove(spa);
	mutex_exit(&spa_namespace_lock);

	return (config);
	}

	/*
	* Pool export/destroy
	*
	* The act of destroying or exporting a pool is very simple. We make sure there
	* is no more pending I/O and any references to the pool are gone. Then, we
	* update the pool state and sync all the labels to disk, removing the
	* configuration from the cache afterwards. If the 'hardforce' flag is set, then
	* we don't sync the labels or remove the configuration cache.
	*/
	static int
	spa_export_common(const char pool, int new_state, nvlist_t *oldconfig,
	boolean_t force, boolean_t hardforce)
	{
	int error;
	spa_t *spa;

	if (oldconfig)
	*oldconfig = NULL;

	if (!(spa_mode_global & SPA_MODE_WRITE))
	return (SET_ERROR(EROFS));

	mutex_enter(&spa_namespace_lock);
	if ((spa = spa_lookup(pool)) == NULL) {
	mutex_exit(&spa_namespace_lock);
	return (SET_ERROR(ENOENT));
	}

	if (spa->spa_is_exporting) {
	/* the pool is being exported by another thread */
	mutex_exit(&spa_namespace_lock);
	return (SET_ERROR(ZFS_ERR_EXPORT_IN_PROGRESS));
	}
	spa->spa_is_exporting = B_TRUE;

	/*
	* Put a hold on the pool, drop the namespace lock, stop async tasks,
	* reacquire the namespace lock, and see if we can export.
	*/
	spa_open_ref(spa, FTAG);
	mutex_exit(&spa_namespace_lock);
	spa_async_suspend(spa);
	if (spa->spa_zvol_taskq) {
	zvol_remove_minors(spa, spa_name(spa), B_TRUE);
	taskq_wait(spa->spa_zvol_taskq);
	}
	mutex_enter(&spa_namespace_lock);
	spa_close(spa, FTAG);

	if (spa->spa_state == POOL_STATE_UNINITIALIZED)
	goto export_spa;
	/*
	* The pool will be in core if it's openable, in which case we can
	* modify its state. Objsets may be open only because they're dirty,
	* so we have to force it to sync before checking spa_refcnt.
	*/
	if (spa->spa_sync_on) {
	txg_wait_synced(spa->spa_dsl_pool, 0);
	spa_evicting_os_wait(spa);
	}

	/*
	* A pool cannot be exported or destroyed if there are active
	* references. If we are resetting a pool, allow references by
	* fault injection handlers.
	*/
	if (!spa_refcount_zero(spa) \|\| (spa->spa_inject_ref != 0)) {
	error = SET_ERROR(EBUSY);
	goto fail;
	}

	if (spa->spa_sync_on) {
	/*
	* A pool cannot be exported if it has an active shared spare.
	* This is to prevent other pools stealing the active spare
	* from an exported pool. At user's own will, such pool can
	* be forcedly exported.
	*/
	if (!force && new_state == POOL_STATE_EXPORTED &&
	spa_has_active_shared_spare(spa)) {
	error = SET_ERROR(EXDEV);
	goto fail;
	}

	/*
	* We're about to export or destroy this pool. Make sure
	* we stop all initialization and trim activity here before
	* we set the spa_final_txg. This will ensure that all
	* dirty data resulting from the initialization is
	* committed to disk before we unload the pool.
	*/
	if (spa->spa_root_vdev != NULL) {
	vdev_t *rvd = spa->spa_root_vdev;
	vdev_initialize_stop_all(rvd, VDEV_INITIALIZE_ACTIVE);
	vdev_trim_stop_all(rvd, VDEV_TRIM_ACTIVE);
	vdev_autotrim_stop_all(spa);
	vdev_rebuild_stop_all(spa);
	}

	/*
	* We want this to be reflected on every label,
	* so mark them all dirty. spa_unload() will do the
	* final sync that pushes these changes out.
	*/
	if (new_state != POOL_STATE_UNINITIALIZED && !hardforce) {
	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	spa->spa_state = new_state;
	spa->spa_final_txg = spa_last_synced_txg(spa) +
	TXG_DEFER_SIZE + 1;
	vdev_config_dirty(spa->spa_root_vdev);
	spa_config_exit(spa, SCL_ALL, FTAG);
	}
	}

	export_spa:
	if (new_state == POOL_STATE_DESTROYED)
	spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_DESTROY);
	else if (new_state == POOL_STATE_EXPORTED)
	spa_event_notify(spa, NULL, NULL, ESC_ZFS_POOL_EXPORT);

	if (spa->spa_state != POOL_STATE_UNINITIALIZED) {
	spa_unload(spa);
	spa_deactivate(spa);
	}

	if (oldconfig && spa->spa_config)
	*oldconfig = fnvlist_dup(spa->spa_config);

	if (new_state != POOL_STATE_UNINITIALIZED) {
	if (!hardforce)
	spa_write_cachefile(spa, B_TRUE, B_TRUE);
	spa_remove(spa);
	} else {
	/*
	* If spa_remove() is not called for this spa_t and
	* there is any possibility that it can be reused,
	* we make sure to reset the exporting flag.
	*/
	spa->spa_is_exporting = B_FALSE;
	}

	mutex_exit(&spa_namespace_lock);
	return (0);

	fail:
	spa->spa_is_exporting = B_FALSE;
	spa_async_resume(spa);
	mutex_exit(&spa_namespace_lock);
	return (error);
	}

	/*
	* Destroy a storage pool.
	*/
	int
	spa_destroy(const char *pool)
	{
	return (spa_export_common(pool, POOL_STATE_DESTROYED, NULL,
	B_FALSE, B_FALSE));
	}

	/*
	* Export a storage pool.
	*/
	int
	spa_export(const char pool, nvlist_t *oldconfig, boolean_t force,
	boolean_t hardforce)
	{
	return (spa_export_common(pool, POOL_STATE_EXPORTED, oldconfig,
	force, hardforce));
	}

	/*
	* Similar to spa_export(), this unloads the spa_t without actually removing it
	* from the namespace in any way.
	*/
	int
	spa_reset(const char *pool)
	{
	return (spa_export_common(pool, POOL_STATE_UNINITIALIZED, NULL,
	B_FALSE, B_FALSE));
	}

	/*
	* ==========================================================================
	* Device manipulation
	* ==========================================================================
	*/

	/*
	* This is called as a synctask to increment the draid feature flag
	*/
	static void
	spa_draid_feature_incr(void arg, dmu_tx_t tx)
	{
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;
	int draid = (int)(uintptr_t)arg;

	for (int c = 0; c < draid; c++)
	spa_feature_incr(spa, SPA_FEATURE_DRAID, tx);
	}

	/*
	* Add a device to a storage pool.
	*/
	int
	spa_vdev_add(spa_t spa, nvlist_t nvroot)
	{
	uint64_t txg, ndraid = 0;
	int error;
	vdev_t *rvd = spa->spa_root_vdev;
	vdev_t vd, tvd;
	nvlist_t spares, l2cache;
	uint_t nspares, nl2cache;

	ASSERT(spa_writeable(spa));

	txg = spa_vdev_enter(spa);

	if ((error = spa_config_parse(spa, &vd, nvroot, NULL, 0,
	VDEV_ALLOC_ADD)) != 0)
	return (spa_vdev_exit(spa, NULL, txg, error));

	spa->spa_pending_vdev = vd; /* spa_vdev_exit() will clear this */

	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES, &spares,
	&nspares) != 0)
	nspares = 0;

	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_L2CACHE, &l2cache,
	&nl2cache) != 0)
	nl2cache = 0;

	if (vd->vdev_children == 0 && nspares == 0 && nl2cache == 0)
	return (spa_vdev_exit(spa, vd, txg, EINVAL));

	if (vd->vdev_children != 0 &&
	(error = vdev_create(vd, txg, B_FALSE)) != 0) {
	return (spa_vdev_exit(spa, vd, txg, error));
	}

	/*
	* The virtual dRAID spares must be added after vdev tree is created
	* and the vdev guids are generated. The guid of their associated
	* dRAID is stored in the config and used when opening the spare.
	*/
	if ((error = vdev_draid_spare_create(nvroot, vd, &ndraid,
	rvd->vdev_children)) == 0) {
	if (ndraid > 0 && nvlist_lookup_nvlist_array(nvroot,
	ZPOOL_CONFIG_SPARES, &spares, &nspares) != 0)
	nspares = 0;
	} else {
	return (spa_vdev_exit(spa, vd, txg, error));
	}

	/*
	* We must validate the spares and l2cache devices after checking the
	* children. Otherwise, vdev_inuse() will blindly overwrite the spare.
	*/
	if ((error = spa_validate_aux(spa, nvroot, txg, VDEV_ALLOC_ADD)) != 0)
	return (spa_vdev_exit(spa, vd, txg, error));

	/*
	* If we are in the middle of a device removal, we can only add
	* devices which match the existing devices in the pool.
	* If we are in the middle of a removal, or have some indirect
	* vdevs, we can not add raidz or dRAID top levels.
	*/
	if (spa->spa_vdev_removal != NULL \|\|
	spa->spa_removing_phys.sr_prev_indirect_vdev != -1) {
	for (int c = 0; c < vd->vdev_children; c++) {
	tvd = vd->vdev_child[c];
	if (spa->spa_vdev_removal != NULL &&
	tvd->vdev_ashift != spa->spa_max_ashift) {
	return (spa_vdev_exit(spa, vd, txg, EINVAL));
	}
	/* Fail if top level vdev is raidz or a dRAID */
	if (vdev_get_nparity(tvd) != 0)
	return (spa_vdev_exit(spa, vd, txg, EINVAL));

	/*
	* Need the top level mirror to be
	* a mirror of leaf vdevs only
	*/
	if (tvd->vdev_ops == &vdev_mirror_ops) {
	for (uint64_t cid = 0;
	cid < tvd->vdev_children; cid++) {
	vdev_t *cvd = tvd->vdev_child[cid];
	if (!cvd->vdev_ops->vdev_op_leaf) {
	return (spa_vdev_exit(spa, vd,
	txg, EINVAL));
	}
	}
	}
	}
	}

	for (int c = 0; c < vd->vdev_children; c++) {
	tvd = vd->vdev_child[c];
	vdev_remove_child(vd, tvd);
	tvd->vdev_id = rvd->vdev_children;
	vdev_add_child(rvd, tvd);
	vdev_config_dirty(tvd);
	}

	if (nspares != 0) {
	spa_set_aux_vdevs(&spa->spa_spares, spares, nspares,
	ZPOOL_CONFIG_SPARES);
	spa_load_spares(spa);
	spa->spa_spares.sav_sync = B_TRUE;
	}

	if (nl2cache != 0) {
	spa_set_aux_vdevs(&spa->spa_l2cache, l2cache, nl2cache,
	ZPOOL_CONFIG_L2CACHE);
	spa_load_l2cache(spa);
	spa->spa_l2cache.sav_sync = B_TRUE;
	}

	/*
	* We can't increment a feature while holding spa_vdev so we
	* have to do it in a synctask.
	*/
	if (ndraid != 0) {
	dmu_tx_t *tx;

	tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg);
	dsl_sync_task_nowait(spa->spa_dsl_pool, spa_draid_feature_incr,
	(void *)(uintptr_t)ndraid, tx);
	dmu_tx_commit(tx);
	}

	/*
	* We have to be careful when adding new vdevs to an existing pool.
	* If other threads start allocating from these vdevs before we
	* sync the config cache, and we lose power, then upon reboot we may
	* fail to open the pool because there are DVAs that the config cache
	* can't translate. Therefore, we first add the vdevs without
	* initializing metaslabs; sync the config cache (via spa_vdev_exit());
	* and then let spa_config_update() initialize the new metaslabs.
	*
	* spa_load() checks for added-but-not-initialized vdevs, so that
	* if we lose power at any point in this sequence, the remaining
	* steps will be completed the next time we load the pool.
	*/
	(void) spa_vdev_exit(spa, vd, txg, 0);

	mutex_enter(&spa_namespace_lock);
	spa_config_update(spa, SPA_CONFIG_UPDATE_POOL);
	spa_event_notify(spa, NULL, NULL, ESC_ZFS_VDEV_ADD);
	mutex_exit(&spa_namespace_lock);

	return (0);
	}

	/*
	* Attach a device to a mirror. The arguments are the path to any device
	* in the mirror, and the nvroot for the new device. If the path specifies
	* a device that is not mirrored, we automatically insert the mirror vdev.
	*
	* If 'replacing' is specified, the new device is intended to replace the
	* existing device; in this case the two devices are made into their own
	* mirror using the 'replacing' vdev, which is functionally identical to
	* the mirror vdev (it actually reuses all the same ops) but has a few
	* extra rules: you can't attach to it after it's been created, and upon
	* completion of resilvering, the first disk (the one being replaced)
	* is automatically detached.
	*
	* If 'rebuild' is specified, then sequential reconstruction (a.ka. rebuild)
	* should be performed instead of traditional healing reconstruction. From
	* an administrators perspective these are both resilver operations.
	*/
	int
	spa_vdev_attach(spa_t spa, uint64_t guid, nvlist_t nvroot, int replacing,
	int rebuild)
	{
	uint64_t txg, dtl_max_txg;
	vdev_t *rvd = spa->spa_root_vdev;
	vdev_t oldvd, newvd, newrootvd, pvd, *tvd;
	vdev_ops_t *pvops;
	char oldvdpath, newvdpath;
	int newvd_isspare;
	int error;

	ASSERT(spa_writeable(spa));

	txg = spa_vdev_enter(spa);

	oldvd = spa_lookup_by_guid(spa, guid, B_FALSE);

	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) {
	error = (spa_has_checkpoint(spa)) ?
	ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT;
	return (spa_vdev_exit(spa, NULL, txg, error));
	}

	if (rebuild) {
	if (!spa_feature_is_enabled(spa, SPA_FEATURE_DEVICE_REBUILD))
	return (spa_vdev_exit(spa, NULL, txg, ENOTSUP));

	if (dsl_scan_resilvering(spa_get_dsl(spa)))
	return (spa_vdev_exit(spa, NULL, txg,
	ZFS_ERR_RESILVER_IN_PROGRESS));
	} else {
	if (vdev_rebuild_active(rvd))
	return (spa_vdev_exit(spa, NULL, txg,
	ZFS_ERR_REBUILD_IN_PROGRESS));
	}

	if (spa->spa_vdev_removal != NULL)
	return (spa_vdev_exit(spa, NULL, txg, EBUSY));

	if (oldvd == NULL)
	return (spa_vdev_exit(spa, NULL, txg, ENODEV));

	if (!oldvd->vdev_ops->vdev_op_leaf)
	return (spa_vdev_exit(spa, NULL, txg, ENOTSUP));

	pvd = oldvd->vdev_parent;

	if ((error = spa_config_parse(spa, &newrootvd, nvroot, NULL, 0,
	VDEV_ALLOC_ATTACH)) != 0)
	return (spa_vdev_exit(spa, NULL, txg, EINVAL));

	if (newrootvd->vdev_children != 1)
	return (spa_vdev_exit(spa, newrootvd, txg, EINVAL));

	newvd = newrootvd->vdev_child[0];

	if (!newvd->vdev_ops->vdev_op_leaf)
	return (spa_vdev_exit(spa, newrootvd, txg, EINVAL));

	if ((error = vdev_create(newrootvd, txg, replacing)) != 0)
	return (spa_vdev_exit(spa, newrootvd, txg, error));

	/*
	* Spares can't replace logs
	*/
	if (oldvd->vdev_top->vdev_islog && newvd->vdev_isspare)
	return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));

	/*
	* A dRAID spare can only replace a child of its parent dRAID vdev.
	*/
	if (newvd->vdev_ops == &vdev_draid_spare_ops &&
	oldvd->vdev_top != vdev_draid_spare_get_parent(newvd)) {
	return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));
	}

	if (rebuild) {
	/*
	* For rebuilds, the top vdev must support reconstruction
	* using only space maps. This means the only allowable
	* vdevs types are the root vdev, a mirror, or dRAID.
	*/
	tvd = pvd;
	if (pvd->vdev_top != NULL)
	tvd = pvd->vdev_top;

	if (tvd->vdev_ops != &vdev_mirror_ops &&
	tvd->vdev_ops != &vdev_root_ops &&
	tvd->vdev_ops != &vdev_draid_ops) {
	return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));
	}
	}

	if (!replacing) {
	/*
	* For attach, the only allowable parent is a mirror or the root
	* vdev.
	*/
	if (pvd->vdev_ops != &vdev_mirror_ops &&
	pvd->vdev_ops != &vdev_root_ops)
	return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));

	pvops = &vdev_mirror_ops;
	} else {
	/*
	* Active hot spares can only be replaced by inactive hot
	* spares.
	*/
	if (pvd->vdev_ops == &vdev_spare_ops &&
	oldvd->vdev_isspare &&
	!spa_has_spare(spa, newvd->vdev_guid))
	return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));

	/*
	* If the source is a hot spare, and the parent isn't already a
	* spare, then we want to create a new hot spare. Otherwise, we
	* want to create a replacing vdev. The user is not allowed to
	* attach to a spared vdev child unless the 'isspare' state is
	* the same (spare replaces spare, non-spare replaces
	* non-spare).
	*/
	if (pvd->vdev_ops == &vdev_replacing_ops &&
	spa_version(spa) < SPA_VERSION_MULTI_REPLACE) {
	return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));
	} else if (pvd->vdev_ops == &vdev_spare_ops &&
	newvd->vdev_isspare != oldvd->vdev_isspare) {
	return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));
	}

	if (newvd->vdev_isspare)
	pvops = &vdev_spare_ops;
	else
	pvops = &vdev_replacing_ops;
	}

	/*
	* Make sure the new device is big enough.
	*/
	if (newvd->vdev_asize < vdev_get_min_asize(oldvd))
	return (spa_vdev_exit(spa, newrootvd, txg, EOVERFLOW));

	/*
	* The new device cannot have a higher alignment requirement
	* than the top-level vdev.
	*/
	if (newvd->vdev_ashift > oldvd->vdev_top->vdev_ashift)
	return (spa_vdev_exit(spa, newrootvd, txg, ENOTSUP));

	/*
	* If this is an in-place replacement, update oldvd's path and devid
	* to make it distinguishable from newvd, and unopenable from now on.
	*/
	if (strcmp(oldvd->vdev_path, newvd->vdev_path) == 0) {
	spa_strfree(oldvd->vdev_path);
	oldvd->vdev_path = kmem_alloc(strlen(newvd->vdev_path) + 5,
	KM_SLEEP);
	(void) snprintf(oldvd->vdev_path, strlen(newvd->vdev_path) + 5,
	"%s/%s", newvd->vdev_path, "old");
	if (oldvd->vdev_devid != NULL) {
	spa_strfree(oldvd->vdev_devid);
	oldvd->vdev_devid = NULL;
	}
	}

	/*
	* If the parent is not a mirror, or if we're replacing, insert the new
	* mirror/replacing/spare vdev above oldvd.
	*/
	if (pvd->vdev_ops != pvops)
	pvd = vdev_add_parent(oldvd, pvops);

	ASSERT(pvd->vdev_top->vdev_parent == rvd);
	ASSERT(pvd->vdev_ops == pvops);
	ASSERT(oldvd->vdev_parent == pvd);

	/*
	* Extract the new device from its root and add it to pvd.
	*/
	vdev_remove_child(newrootvd, newvd);
	newvd->vdev_id = pvd->vdev_children;
	newvd->vdev_crtxg = oldvd->vdev_crtxg;
	vdev_add_child(pvd, newvd);

	/*
	* Reevaluate the parent vdev state.
	*/
	vdev_propagate_state(pvd);

	tvd = newvd->vdev_top;
	ASSERT(pvd->vdev_top == tvd);
	ASSERT(tvd->vdev_parent == rvd);

	vdev_config_dirty(tvd);

	/*
	* Set newvd's DTL to [TXG_INITIAL, dtl_max_txg) so that we account
	* for any dmu_sync-ed blocks. It will propagate upward when
	* spa_vdev_exit() calls vdev_dtl_reassess().
	*/
	dtl_max_txg = txg + TXG_CONCURRENT_STATES;

	vdev_dtl_dirty(newvd, DTL_MISSING,
	TXG_INITIAL, dtl_max_txg - TXG_INITIAL);

	if (newvd->vdev_isspare) {
	spa_spare_activate(newvd);
	spa_event_notify(spa, newvd, NULL, ESC_ZFS_VDEV_SPARE);
	}

	oldvdpath = spa_strdup(oldvd->vdev_path);
	newvdpath = spa_strdup(newvd->vdev_path);
	newvd_isspare = newvd->vdev_isspare;

	/*
	* Mark newvd's DTL dirty in this txg.
	*/
	vdev_dirty(tvd, VDD_DTL, newvd, txg);

	/*
	* Schedule the resilver or rebuild to restart in the future. We do
	* this to ensure that dmu_sync-ed blocks have been stitched into the
	* respective datasets.
	*/
	if (rebuild) {
	newvd->vdev_rebuild_txg = txg;

	vdev_rebuild(tvd);
	} else {
	newvd->vdev_resilver_txg = txg;

	if (dsl_scan_resilvering(spa_get_dsl(spa)) &&
	spa_feature_is_enabled(spa, SPA_FEATURE_RESILVER_DEFER)) {
	vdev_defer_resilver(newvd);
	} else {
	dsl_scan_restart_resilver(spa->spa_dsl_pool,
	dtl_max_txg);
	}
	}

	if (spa->spa_bootfs)
	spa_event_notify(spa, newvd, NULL, ESC_ZFS_BOOTFS_VDEV_ATTACH);

	spa_event_notify(spa, newvd, NULL, ESC_ZFS_VDEV_ATTACH);

	/*
	* Commit the config
	*/
	(void) spa_vdev_exit(spa, newrootvd, dtl_max_txg, 0);

	spa_history_log_internal(spa, "vdev attach", NULL,
	"%s vdev=%s %s vdev=%s",
	replacing && newvd_isspare ? "spare in" :
	replacing ? "replace" : "attach", newvdpath,
	replacing ? "for" : "to", oldvdpath);

	spa_strfree(oldvdpath);
	spa_strfree(newvdpath);

	return (0);
	}

	/*
	* Detach a device from a mirror or replacing vdev.
	*
	* If 'replace_done' is specified, only detach if the parent
	* is a replacing vdev.
	*/
	int
	spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done)
	{
	uint64_t txg;
	int error;
	vdev_t *rvd __maybe_unused = spa->spa_root_vdev;
	vdev_t vd, pvd, cvd, tvd;
	boolean_t unspare = B_FALSE;
	uint64_t unspare_guid = 0;
	char *vdpath;

	ASSERT(spa_writeable(spa));

	txg = spa_vdev_detach_enter(spa, guid);

	vd = spa_lookup_by_guid(spa, guid, B_FALSE);

	/*
	* Besides being called directly from the userland through the
	* ioctl interface, spa_vdev_detach() can be potentially called
	* at the end of spa_vdev_resilver_done().
	*
	* In the regular case, when we have a checkpoint this shouldn't
	* happen as we never empty the DTLs of a vdev during the scrub
	* [see comment in dsl_scan_done()]. Thus spa_vdev_resilvering_done()
	* should never get here when we have a checkpoint.
	*
	* That said, even in a case when we checkpoint the pool exactly
	* as spa_vdev_resilver_done() calls this function everything
	* should be fine as the resilver will return right away.
	*/
	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) {
	error = (spa_has_checkpoint(spa)) ?
	ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT;
	return (spa_vdev_exit(spa, NULL, txg, error));
	}

	if (vd == NULL)
	return (spa_vdev_exit(spa, NULL, txg, ENODEV));

	if (!vd->vdev_ops->vdev_op_leaf)
	return (spa_vdev_exit(spa, NULL, txg, ENOTSUP));

	pvd = vd->vdev_parent;

	/*
	* If the parent/child relationship is not as expected, don't do it.
	* Consider M(A,R(B,C)) -- that is, a mirror of A with a replacing
	* vdev that's replacing B with C. The user's intent in replacing
	* is to go from M(A,B) to M(A,C). If the user decides to cancel
	* the replace by detaching C, the expected behavior is to end up
	* M(A,B). But suppose that right after deciding to detach C,
	* the replacement of B completes. We would have M(A,C), and then
	* ask to detach C, which would leave us with just A -- not what
	* the user wanted. To prevent this, we make sure that the
	* parent/child relationship hasn't changed -- in this example,
	* that C's parent is still the replacing vdev R.
	*/
	if (pvd->vdev_guid != pguid && pguid != 0)
	return (spa_vdev_exit(spa, NULL, txg, EBUSY));

	/*
	* Only 'replacing' or 'spare' vdevs can be replaced.
	*/
	if (replace_done && pvd->vdev_ops != &vdev_replacing_ops &&
	pvd->vdev_ops != &vdev_spare_ops)
	return (spa_vdev_exit(spa, NULL, txg, ENOTSUP));

	ASSERT(pvd->vdev_ops != &vdev_spare_ops \|\|
	spa_version(spa) >= SPA_VERSION_SPARES);

	/*
	* Only mirror, replacing, and spare vdevs support detach.
	*/
	if (pvd->vdev_ops != &vdev_replacing_ops &&
	pvd->vdev_ops != &vdev_mirror_ops &&
	pvd->vdev_ops != &vdev_spare_ops)
	return (spa_vdev_exit(spa, NULL, txg, ENOTSUP));

	/*
	* If this device has the only valid copy of some data,
	* we cannot safely detach it.
	*/
	if (vdev_dtl_required(vd))
	return (spa_vdev_exit(spa, NULL, txg, EBUSY));

	ASSERT(pvd->vdev_children >= 2);

	/*
	* If we are detaching the second disk from a replacing vdev, then
	* check to see if we changed the original vdev's path to have "/old"
	* at the end in spa_vdev_attach(). If so, undo that change now.
	*/
	if (pvd->vdev_ops == &vdev_replacing_ops && vd->vdev_id > 0 &&
	vd->vdev_path != NULL) {
	size_t len = strlen(vd->vdev_path);

	for (int c = 0; c < pvd->vdev_children; c++) {
	cvd = pvd->vdev_child[c];

	if (cvd == vd \|\| cvd->vdev_path == NULL)
	continue;

	if (strncmp(cvd->vdev_path, vd->vdev_path, len) == 0 &&
	strcmp(cvd->vdev_path + len, "/old") == 0) {
	spa_strfree(cvd->vdev_path);
	cvd->vdev_path = spa_strdup(vd->vdev_path);
	break;
	}
	}
	}

	/*
	* If we are detaching the original disk from a normal spare, then it
	* implies that the spare should become a real disk, and be removed
	* from the active spare list for the pool. dRAID spares on the
	* other hand are coupled to the pool and thus should never be removed
	* from the spares list.
	*/
	if (pvd->vdev_ops == &vdev_spare_ops && vd->vdev_id == 0) {
	vdev_t *last_cvd = pvd->vdev_child[pvd->vdev_children - 1];

	if (last_cvd->vdev_isspare &&
	last_cvd->vdev_ops != &vdev_draid_spare_ops) {
	unspare = B_TRUE;
	}
	}

	/*
	* Erase the disk labels so the disk can be used for other things.
	* This must be done after all other error cases are handled,
	* but before we disembowel vd (so we can still do I/O to it).
	* But if we can't do it, don't treat the error as fatal --
	* it may be that the unwritability of the disk is the reason
	* it's being detached!
	*/
	error = vdev_label_init(vd, 0, VDEV_LABEL_REMOVE);

	/*
	* Remove vd from its parent and compact the parent's children.
	*/
	vdev_remove_child(pvd, vd);
	vdev_compact_children(pvd);

	/*
	* Remember one of the remaining children so we can get tvd below.
	*/
	cvd = pvd->vdev_child[pvd->vdev_children - 1];

	/*
	* If we need to remove the remaining child from the list of hot spares,
	* do it now, marking the vdev as no longer a spare in the process.
	* We must do this before vdev_remove_parent(), because that can
	* change the GUID if it creates a new toplevel GUID. For a similar
	* reason, we must remove the spare now, in the same txg as the detach;
	* otherwise someone could attach a new sibling, change the GUID, and
	* the subsequent attempt to spa_vdev_remove(unspare_guid) would fail.
	*/
	if (unspare) {
	ASSERT(cvd->vdev_isspare);
	spa_spare_remove(cvd);
	unspare_guid = cvd->vdev_guid;
	(void) spa_vdev_remove(spa, unspare_guid, B_TRUE);
	cvd->vdev_unspare = B_TRUE;
	}

	/*
	* If the parent mirror/replacing vdev only has one child,
	* the parent is no longer needed. Remove it from the tree.
	*/
	if (pvd->vdev_children == 1) {
	if (pvd->vdev_ops == &vdev_spare_ops)
	cvd->vdev_unspare = B_FALSE;
	vdev_remove_parent(cvd);
	}

	/*
	* We don't set tvd until now because the parent we just removed
	* may have been the previous top-level vdev.
	*/
	tvd = cvd->vdev_top;
	ASSERT(tvd->vdev_parent == rvd);

	/*
	* Reevaluate the parent vdev state.
	*/
	vdev_propagate_state(cvd);

	/*
	* If the 'autoexpand' property is set on the pool then automatically
	* try to expand the size of the pool. For example if the device we
	* just detached was smaller than the others, it may be possible to
	* add metaslabs (i.e. grow the pool). We need to reopen the vdev
	* first so that we can obtain the updated sizes of the leaf vdevs.
	*/
	if (spa->spa_autoexpand) {
	vdev_reopen(tvd);
	vdev_expand(tvd, txg);
	}

	vdev_config_dirty(tvd);

	/*
	* Mark vd's DTL as dirty in this txg. vdev_dtl_sync() will see that
	* vd->vdev_detached is set and free vd's DTL object in syncing context.
	* But first make sure we're not on any other txg's DTL list, to
	* prevent vd from being accessed after it's freed.
	*/
	vdpath = spa_strdup(vd->vdev_path ? vd->vdev_path : "none");
	for (int t = 0; t < TXG_SIZE; t++)
	(void) txg_list_remove_this(&tvd->vdev_dtl_list, vd, t);
	vd->vdev_detached = B_TRUE;
	vdev_dirty(tvd, VDD_DTL, vd, txg);

	spa_event_notify(spa, vd, NULL, ESC_ZFS_VDEV_REMOVE);
	spa_notify_waiters(spa);

	/* hang on to the spa before we release the lock */
	spa_open_ref(spa, FTAG);

	error = spa_vdev_exit(spa, vd, txg, 0);

	spa_history_log_internal(spa, "detach", NULL,
	"vdev=%s", vdpath);
	spa_strfree(vdpath);

	/*
	* If this was the removal of the original device in a hot spare vdev,
	* then we want to go through and remove the device from the hot spare
	* list of every other pool.
	*/
	if (unspare) {
	spa_t *altspa = NULL;

	mutex_enter(&spa_namespace_lock);
	while ((altspa = spa_next(altspa)) != NULL) {
	if (altspa->spa_state != POOL_STATE_ACTIVE \|\|
	altspa == spa)
	continue;

	spa_open_ref(altspa, FTAG);
	mutex_exit(&spa_namespace_lock);
	(void) spa_vdev_remove(altspa, unspare_guid, B_TRUE);
	mutex_enter(&spa_namespace_lock);
	spa_close(altspa, FTAG);
	}
	mutex_exit(&spa_namespace_lock);

	/* search the rest of the vdevs for spares to remove */
	spa_vdev_resilver_done(spa);
	}

	/* all done with the spa; OK to release */
	mutex_enter(&spa_namespace_lock);
	spa_close(spa, FTAG);
	mutex_exit(&spa_namespace_lock);

	return (error);
	}

	static int
	spa_vdev_initialize_impl(spa_t *spa, uint64_t guid, uint64_t cmd_type,
	list_t *vd_list)
	{
	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	spa_config_enter(spa, SCL_CONFIG \| SCL_STATE, FTAG, RW_READER);

	/* Look up vdev and ensure it's a leaf. */
	vdev_t *vd = spa_lookup_by_guid(spa, guid, B_FALSE);
	if (vd == NULL \|\| vd->vdev_detached) {
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	return (SET_ERROR(ENODEV));
	} else if (!vd->vdev_ops->vdev_op_leaf \|\| !vdev_is_concrete(vd)) {
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	return (SET_ERROR(EINVAL));
	} else if (!vdev_writeable(vd)) {
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	return (SET_ERROR(EROFS));
	}
	mutex_enter(&vd->vdev_initialize_lock);
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);

	/*
	* When we activate an initialize action we check to see
	* if the vdev_initialize_thread is NULL. We do this instead
	* of using the vdev_initialize_state since there might be
	* a previous initialization process which has completed but
	* the thread is not exited.
	*/
	if (cmd_type == POOL_INITIALIZE_START &&
	(vd->vdev_initialize_thread != NULL \|\|
	vd->vdev_top->vdev_removing)) {
	mutex_exit(&vd->vdev_initialize_lock);
	return (SET_ERROR(EBUSY));
	} else if (cmd_type == POOL_INITIALIZE_CANCEL &&
	(vd->vdev_initialize_state != VDEV_INITIALIZE_ACTIVE &&
	vd->vdev_initialize_state != VDEV_INITIALIZE_SUSPENDED)) {
	mutex_exit(&vd->vdev_initialize_lock);
	return (SET_ERROR(ESRCH));
	} else if (cmd_type == POOL_INITIALIZE_SUSPEND &&
	vd->vdev_initialize_state != VDEV_INITIALIZE_ACTIVE) {
	mutex_exit(&vd->vdev_initialize_lock);
	return (SET_ERROR(ESRCH));
	}

	switch (cmd_type) {
	case POOL_INITIALIZE_START:
	vdev_initialize(vd);
	break;
	case POOL_INITIALIZE_CANCEL:
	vdev_initialize_stop(vd, VDEV_INITIALIZE_CANCELED, vd_list);
	break;
	case POOL_INITIALIZE_SUSPEND:
	vdev_initialize_stop(vd, VDEV_INITIALIZE_SUSPENDED, vd_list);
	break;
	default:
	panic("invalid cmd_type %llu", (unsigned long long)cmd_type);
	}
	mutex_exit(&vd->vdev_initialize_lock);

	return (0);
	}

	int
	spa_vdev_initialize(spa_t spa, nvlist_t nv, uint64_t cmd_type,
	nvlist_t *vdev_errlist)
	{
	int total_errors = 0;
	list_t vd_list;

	list_create(&vd_list, sizeof (vdev_t),
	offsetof(vdev_t, vdev_initialize_node));

	/*
	* We hold the namespace lock through the whole function
	* to prevent any changes to the pool while we're starting or
	* stopping initialization. The config and state locks are held so that
	* we can properly assess the vdev state before we commit to
	* the initializing operation.
	*/
	mutex_enter(&spa_namespace_lock);

	for (nvpair_t *pair = nvlist_next_nvpair(nv, NULL);
	pair != NULL; pair = nvlist_next_nvpair(nv, pair)) {
	uint64_t vdev_guid = fnvpair_value_uint64(pair);

	int error = spa_vdev_initialize_impl(spa, vdev_guid, cmd_type,
	&vd_list);
	if (error != 0) {
	char guid_as_str[MAXNAMELEN];

	(void) snprintf(guid_as_str, sizeof (guid_as_str),
	"%llu", (unsigned long long)vdev_guid);
	fnvlist_add_int64(vdev_errlist, guid_as_str, error);
	total_errors++;
	}
	}

	/* Wait for all initialize threads to stop. */
	vdev_initialize_stop_wait(spa, &vd_list);

	/* Sync out the initializing state */
	txg_wait_synced(spa->spa_dsl_pool, 0);
	mutex_exit(&spa_namespace_lock);

	list_destroy(&vd_list);

	return (total_errors);
	}

	static int
	spa_vdev_trim_impl(spa_t *spa, uint64_t guid, uint64_t cmd_type,
	uint64_t rate, boolean_t partial, boolean_t secure, list_t *vd_list)
	{
	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	spa_config_enter(spa, SCL_CONFIG \| SCL_STATE, FTAG, RW_READER);

	/* Look up vdev and ensure it's a leaf. */
	vdev_t *vd = spa_lookup_by_guid(spa, guid, B_FALSE);
	if (vd == NULL \|\| vd->vdev_detached) {
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	return (SET_ERROR(ENODEV));
	} else if (!vd->vdev_ops->vdev_op_leaf \|\| !vdev_is_concrete(vd)) {
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	return (SET_ERROR(EINVAL));
	} else if (!vdev_writeable(vd)) {
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	return (SET_ERROR(EROFS));
	} else if (!vd->vdev_has_trim) {
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	return (SET_ERROR(EOPNOTSUPP));
	} else if (secure && !vd->vdev_has_securetrim) {
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	return (SET_ERROR(EOPNOTSUPP));
	}
	mutex_enter(&vd->vdev_trim_lock);
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);

	/*
	* When we activate a TRIM action we check to see if the
	* vdev_trim_thread is NULL. We do this instead of using the
	* vdev_trim_state since there might be a previous TRIM process
	* which has completed but the thread is not exited.
	*/
	if (cmd_type == POOL_TRIM_START &&
	(vd->vdev_trim_thread != NULL \|\| vd->vdev_top->vdev_removing)) {
	mutex_exit(&vd->vdev_trim_lock);
	return (SET_ERROR(EBUSY));
	} else if (cmd_type == POOL_TRIM_CANCEL &&
	(vd->vdev_trim_state != VDEV_TRIM_ACTIVE &&
	vd->vdev_trim_state != VDEV_TRIM_SUSPENDED)) {
	mutex_exit(&vd->vdev_trim_lock);
	return (SET_ERROR(ESRCH));
	} else if (cmd_type == POOL_TRIM_SUSPEND &&
	vd->vdev_trim_state != VDEV_TRIM_ACTIVE) {
	mutex_exit(&vd->vdev_trim_lock);
	return (SET_ERROR(ESRCH));
	}

	switch (cmd_type) {
	case POOL_TRIM_START:
	vdev_trim(vd, rate, partial, secure);
	break;
	case POOL_TRIM_CANCEL:
	vdev_trim_stop(vd, VDEV_TRIM_CANCELED, vd_list);
	break;
	case POOL_TRIM_SUSPEND:
	vdev_trim_stop(vd, VDEV_TRIM_SUSPENDED, vd_list);
	break;
	default:
	panic("invalid cmd_type %llu", (unsigned long long)cmd_type);
	}
	mutex_exit(&vd->vdev_trim_lock);

	return (0);
	}

	/*
	* Initiates a manual TRIM for the requested vdevs. This kicks off individual
	* TRIM threads for each child vdev. These threads pass over all of the free
	* space in the vdev's metaslabs and issues TRIM commands for that space.
	*/
	int
	spa_vdev_trim(spa_t spa, nvlist_t nv, uint64_t cmd_type, uint64_t rate,
	boolean_t partial, boolean_t secure, nvlist_t *vdev_errlist)
	{
	int total_errors = 0;
	list_t vd_list;

	list_create(&vd_list, sizeof (vdev_t),
	offsetof(vdev_t, vdev_trim_node));

	/*
	* We hold the namespace lock through the whole function
	* to prevent any changes to the pool while we're starting or
	* stopping TRIM. The config and state locks are held so that
	* we can properly assess the vdev state before we commit to
	* the TRIM operation.
	*/
	mutex_enter(&spa_namespace_lock);

	for (nvpair_t *pair = nvlist_next_nvpair(nv, NULL);
	pair != NULL; pair = nvlist_next_nvpair(nv, pair)) {
	uint64_t vdev_guid = fnvpair_value_uint64(pair);

	int error = spa_vdev_trim_impl(spa, vdev_guid, cmd_type,
	rate, partial, secure, &vd_list);
	if (error != 0) {
	char guid_as_str[MAXNAMELEN];

	(void) snprintf(guid_as_str, sizeof (guid_as_str),
	"%llu", (unsigned long long)vdev_guid);
	fnvlist_add_int64(vdev_errlist, guid_as_str, error);
	total_errors++;
	}
	}

	/* Wait for all TRIM threads to stop. */
	vdev_trim_stop_wait(spa, &vd_list);

	/* Sync out the TRIM state */
	txg_wait_synced(spa->spa_dsl_pool, 0);
	mutex_exit(&spa_namespace_lock);

	list_destroy(&vd_list);

	return (total_errors);
	}

	/*
	* Split a set of devices from their mirrors, and create a new pool from them.
	*/
	int
	spa_vdev_split_mirror(spa_t spa, char newname, nvlist_t *config,
	nvlist_t *props, boolean_t exp)
	{
	int error = 0;
	uint64_t txg, *glist;
	spa_t *newspa;
	uint_t c, children, lastlog;
	nvlist_t *child, nvl, *tmp;
	dmu_tx_t *tx;
	char *altroot = NULL;
	vdev_t rvd, vml = NULL; / vdev modify list */
	boolean_t activate_slog;

	ASSERT(spa_writeable(spa));

	txg = spa_vdev_enter(spa);

	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) {
	error = (spa_has_checkpoint(spa)) ?
	ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT;
	return (spa_vdev_exit(spa, NULL, txg, error));
	}

	/* clear the log and flush everything up to now */
	activate_slog = spa_passivate_log(spa);
	(void) spa_vdev_config_exit(spa, NULL, txg, 0, FTAG);
	error = spa_reset_logs(spa);
	txg = spa_vdev_config_enter(spa);

	if (activate_slog)
	spa_activate_log(spa);

	if (error != 0)
	return (spa_vdev_exit(spa, NULL, txg, error));

	/* check new spa name before going any further */
	if (spa_lookup(newname) != NULL)
	return (spa_vdev_exit(spa, NULL, txg, EEXIST));

	/*
	* scan through all the children to ensure they're all mirrors
	*/
	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, &nvl) != 0 \|\|
	nvlist_lookup_nvlist_array(nvl, ZPOOL_CONFIG_CHILDREN, &child,
	&children) != 0)
	return (spa_vdev_exit(spa, NULL, txg, EINVAL));

	/* first, check to ensure we've got the right child count */
	rvd = spa->spa_root_vdev;
	lastlog = 0;
	for (c = 0; c < rvd->vdev_children; c++) {
	vdev_t *vd = rvd->vdev_child[c];

	/* don't count the holes & logs as children */
	if (vd->vdev_islog \|\| (vd->vdev_ops != &vdev_indirect_ops &&
	!vdev_is_concrete(vd))) {
	if (lastlog == 0)
	lastlog = c;
	continue;
	}

	lastlog = 0;
	}
	if (children != (lastlog != 0 ? lastlog : rvd->vdev_children))
	return (spa_vdev_exit(spa, NULL, txg, EINVAL));

	/* next, ensure no spare or cache devices are part of the split */
	if (nvlist_lookup_nvlist(nvl, ZPOOL_CONFIG_SPARES, &tmp) == 0 \|\|
	nvlist_lookup_nvlist(nvl, ZPOOL_CONFIG_L2CACHE, &tmp) == 0)
	return (spa_vdev_exit(spa, NULL, txg, EINVAL));

	vml = kmem_zalloc(children * sizeof (vdev_t *), KM_SLEEP);
	glist = kmem_zalloc(children * sizeof (uint64_t), KM_SLEEP);

	/* then, loop over each vdev and validate it */
	for (c = 0; c < children; c++) {
	uint64_t is_hole = 0;

	(void) nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_IS_HOLE,
	&is_hole);

	if (is_hole != 0) {
	if (spa->spa_root_vdev->vdev_child[c]->vdev_ishole \|\|
	spa->spa_root_vdev->vdev_child[c]->vdev_islog) {
	continue;
	} else {
	error = SET_ERROR(EINVAL);
	break;
	}
	}

	/* deal with indirect vdevs */
	if (spa->spa_root_vdev->vdev_child[c]->vdev_ops ==
	&vdev_indirect_ops)
	continue;

	/* which disk is going to be split? */
	if (nvlist_lookup_uint64(child[c], ZPOOL_CONFIG_GUID,
	&glist[c]) != 0) {
	error = SET_ERROR(EINVAL);
	break;
	}

	/* look it up in the spa */
	vml[c] = spa_lookup_by_guid(spa, glist[c], B_FALSE);
	if (vml[c] == NULL) {
	error = SET_ERROR(ENODEV);
	break;
	}

	/* make sure there's nothing stopping the split */
	if (vml[c]->vdev_parent->vdev_ops != &vdev_mirror_ops \|\|
	vml[c]->vdev_islog \|\|
	!vdev_is_concrete(vml[c]) \|\|
	vml[c]->vdev_isspare \|\|
	vml[c]->vdev_isl2cache \|\|
	!vdev_writeable(vml[c]) \|\|
	vml[c]->vdev_children != 0 \|\|
	vml[c]->vdev_state != VDEV_STATE_HEALTHY \|\|
	c != spa->spa_root_vdev->vdev_child[c]->vdev_id) {
	error = SET_ERROR(EINVAL);
	break;
	}

	if (vdev_dtl_required(vml[c]) \|\|
	vdev_resilver_needed(vml[c], NULL, NULL)) {
	error = SET_ERROR(EBUSY);
	break;
	}

	/* we need certain info from the top level */
	fnvlist_add_uint64(child[c], ZPOOL_CONFIG_METASLAB_ARRAY,
	vml[c]->vdev_top->vdev_ms_array);
	fnvlist_add_uint64(child[c], ZPOOL_CONFIG_METASLAB_SHIFT,
	vml[c]->vdev_top->vdev_ms_shift);
	fnvlist_add_uint64(child[c], ZPOOL_CONFIG_ASIZE,
	vml[c]->vdev_top->vdev_asize);
	fnvlist_add_uint64(child[c], ZPOOL_CONFIG_ASHIFT,
	vml[c]->vdev_top->vdev_ashift);

	/* transfer per-vdev ZAPs */
	ASSERT3U(vml[c]->vdev_leaf_zap, !=, 0);
	VERIFY0(nvlist_add_uint64(child[c],
	ZPOOL_CONFIG_VDEV_LEAF_ZAP, vml[c]->vdev_leaf_zap));

	ASSERT3U(vml[c]->vdev_top->vdev_top_zap, !=, 0);
	VERIFY0(nvlist_add_uint64(child[c],
	ZPOOL_CONFIG_VDEV_TOP_ZAP,
	vml[c]->vdev_parent->vdev_top_zap));
	}

	if (error != 0) {
	kmem_free(vml, children * sizeof (vdev_t *));
	kmem_free(glist, children * sizeof (uint64_t));
	return (spa_vdev_exit(spa, NULL, txg, error));
	}

	/* stop writers from using the disks */
	for (c = 0; c < children; c++) {
	if (vml[c] != NULL)
	vml[c]->vdev_offline = B_TRUE;
	}
	vdev_reopen(spa->spa_root_vdev);

	/*
	* Temporarily record the splitting vdevs in the spa config. This
	* will disappear once the config is regenerated.
	*/
	nvl = fnvlist_alloc();
	fnvlist_add_uint64_array(nvl, ZPOOL_CONFIG_SPLIT_LIST, glist, children);
	kmem_free(glist, children * sizeof (uint64_t));

	mutex_enter(&spa->spa_props_lock);
	fnvlist_add_nvlist(spa->spa_config, ZPOOL_CONFIG_SPLIT, nvl);
	mutex_exit(&spa->spa_props_lock);
	spa->spa_config_splitting = nvl;
	vdev_config_dirty(spa->spa_root_vdev);

	/* configure and create the new pool */
	fnvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, newname);
	fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE,
	exp ? POOL_STATE_EXPORTED : POOL_STATE_ACTIVE);
	fnvlist_add_uint64(config, ZPOOL_CONFIG_VERSION, spa_version(spa));
	fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG, spa->spa_config_txg);
	fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID,
	spa_generate_guid(NULL));
	VERIFY0(nvlist_add_boolean(config, ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS));
	(void) nvlist_lookup_string(props,
	zpool_prop_to_name(ZPOOL_PROP_ALTROOT), &altroot);

	/* add the new pool to the namespace */
	newspa = spa_add(newname, config, altroot);
	newspa->spa_avz_action = AVZ_ACTION_REBUILD;
	newspa->spa_config_txg = spa->spa_config_txg;
	spa_set_log_state(newspa, SPA_LOG_CLEAR);

	/* release the spa config lock, retaining the namespace lock */
	spa_vdev_config_exit(spa, NULL, txg, 0, FTAG);

	if (zio_injection_enabled)
	zio_handle_panic_injection(spa, FTAG, 1);

	spa_activate(newspa, spa_mode_global);
	spa_async_suspend(newspa);

	/*
	* Temporarily stop the initializing and TRIM activity. We set the
	* state to ACTIVE so that we know to resume initializing or TRIM
	* once the split has completed.
	*/
	list_t vd_initialize_list;
	list_create(&vd_initialize_list, sizeof (vdev_t),
	offsetof(vdev_t, vdev_initialize_node));

	list_t vd_trim_list;
	list_create(&vd_trim_list, sizeof (vdev_t),
	offsetof(vdev_t, vdev_trim_node));

	for (c = 0; c < children; c++) {
	if (vml[c] != NULL && vml[c]->vdev_ops != &vdev_indirect_ops) {
	mutex_enter(&vml[c]->vdev_initialize_lock);
	vdev_initialize_stop(vml[c],
	VDEV_INITIALIZE_ACTIVE, &vd_initialize_list);
	mutex_exit(&vml[c]->vdev_initialize_lock);

	mutex_enter(&vml[c]->vdev_trim_lock);
	vdev_trim_stop(vml[c], VDEV_TRIM_ACTIVE, &vd_trim_list);
	mutex_exit(&vml[c]->vdev_trim_lock);
	}
	}

	vdev_initialize_stop_wait(spa, &vd_initialize_list);
	vdev_trim_stop_wait(spa, &vd_trim_list);

	list_destroy(&vd_initialize_list);
	list_destroy(&vd_trim_list);

	newspa->spa_config_source = SPA_CONFIG_SRC_SPLIT;
	newspa->spa_is_splitting = B_TRUE;

	/* create the new pool from the disks of the original pool */
	error = spa_load(newspa, SPA_LOAD_IMPORT, SPA_IMPORT_ASSEMBLE);
	if (error)
	goto out;

	/* if that worked, generate a real config for the new pool */
	if (newspa->spa_root_vdev != NULL) {
	newspa->spa_config_splitting = fnvlist_alloc();
	fnvlist_add_uint64(newspa->spa_config_splitting,
	ZPOOL_CONFIG_SPLIT_GUID, spa_guid(spa));
	spa_config_set(newspa, spa_config_generate(newspa, NULL, -1ULL,
	B_TRUE));
	}

	/* set the props */
	if (props != NULL) {
	spa_configfile_set(newspa, props, B_FALSE);
	error = spa_prop_set(newspa, props);
	if (error)
	goto out;
	}

	/* flush everything */
	txg = spa_vdev_config_enter(newspa);
	vdev_config_dirty(newspa->spa_root_vdev);
	(void) spa_vdev_config_exit(newspa, NULL, txg, 0, FTAG);

	if (zio_injection_enabled)
	zio_handle_panic_injection(spa, FTAG, 2);

	spa_async_resume(newspa);

	/* finally, update the original pool's config */
	txg = spa_vdev_config_enter(spa);
	tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir);
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error != 0)
	dmu_tx_abort(tx);
	for (c = 0; c < children; c++) {
	if (vml[c] != NULL && vml[c]->vdev_ops != &vdev_indirect_ops) {
	vdev_t *tvd = vml[c]->vdev_top;

	/*
	* Need to be sure the detachable VDEV is not
	* on any other txg's DTL list to prevent it
	* from being accessed after it's freed.
	*/
	for (int t = 0; t < TXG_SIZE; t++) {
	(void) txg_list_remove_this(
	&tvd->vdev_dtl_list, vml[c], t);
	}

	vdev_split(vml[c]);
	if (error == 0)
	spa_history_log_internal(spa, "detach", tx,
	"vdev=%s", vml[c]->vdev_path);

	vdev_free(vml[c]);
	}
	}
	spa->spa_avz_action = AVZ_ACTION_REBUILD;
	vdev_config_dirty(spa->spa_root_vdev);
	spa->spa_config_splitting = NULL;
	nvlist_free(nvl);
	if (error == 0)
	dmu_tx_commit(tx);
	(void) spa_vdev_exit(spa, NULL, txg, 0);

	if (zio_injection_enabled)
	zio_handle_panic_injection(spa, FTAG, 3);

	/* split is complete; log a history record */
	spa_history_log_internal(newspa, "split", NULL,
	"from pool %s", spa_name(spa));

	newspa->spa_is_splitting = B_FALSE;
	kmem_free(vml, children * sizeof (vdev_t *));

	/* if we're not going to mount the filesystems in userland, export */
	if (exp)
	error = spa_export_common(newname, POOL_STATE_EXPORTED, NULL,
	B_FALSE, B_FALSE);

	return (error);

	out:
	spa_unload(newspa);
	spa_deactivate(newspa);
	spa_remove(newspa);

	txg = spa_vdev_config_enter(spa);

	/* re-online all offlined disks */
	for (c = 0; c < children; c++) {
	if (vml[c] != NULL)
	vml[c]->vdev_offline = B_FALSE;
	}

	/* restart initializing or trimming disks as necessary */
	spa_async_request(spa, SPA_ASYNC_INITIALIZE_RESTART);
	spa_async_request(spa, SPA_ASYNC_TRIM_RESTART);
	spa_async_request(spa, SPA_ASYNC_AUTOTRIM_RESTART);

	vdev_reopen(spa->spa_root_vdev);

	nvlist_free(spa->spa_config_splitting);
	spa->spa_config_splitting = NULL;
	(void) spa_vdev_exit(spa, NULL, txg, error);

	kmem_free(vml, children * sizeof (vdev_t *));
	return (error);
	}

	/*
	* Find any device that's done replacing, or a vdev marked 'unspare' that's
	* currently spared, so we can detach it.
	*/
	static vdev_t *
	spa_vdev_resilver_done_hunt(vdev_t *vd)
	{
	vdev_t newvd, oldvd;

	for (int c = 0; c < vd->vdev_children; c++) {
	oldvd = spa_vdev_resilver_done_hunt(vd->vdev_child[c]);
	if (oldvd != NULL)
	return (oldvd);
	}

	/*
	* Check for a completed replacement. We always consider the first
	* vdev in the list to be the oldest vdev, and the last one to be
	* the newest (see spa_vdev_attach() for how that works). In
	* the case where the newest vdev is faulted, we will not automatically
	* remove it after a resilver completes. This is OK as it will require
	* user intervention to determine which disk the admin wishes to keep.
	*/
	if (vd->vdev_ops == &vdev_replacing_ops) {
	ASSERT(vd->vdev_children > 1);

	newvd = vd->vdev_child[vd->vdev_children - 1];
	oldvd = vd->vdev_child[0];

	if (vdev_dtl_empty(newvd, DTL_MISSING) &&
	vdev_dtl_empty(newvd, DTL_OUTAGE) &&
	!vdev_dtl_required(oldvd))
	return (oldvd);
	}

	/*
	* Check for a completed resilver with the 'unspare' flag set.
	* Also potentially update faulted state.
	*/
	if (vd->vdev_ops == &vdev_spare_ops) {
	vdev_t *first = vd->vdev_child[0];
	vdev_t *last = vd->vdev_child[vd->vdev_children - 1];

	if (last->vdev_unspare) {
	oldvd = first;
	newvd = last;
	} else if (first->vdev_unspare) {
	oldvd = last;
	newvd = first;
	} else {
	oldvd = NULL;
	}

	if (oldvd != NULL &&
	vdev_dtl_empty(newvd, DTL_MISSING) &&
	vdev_dtl_empty(newvd, DTL_OUTAGE) &&
	!vdev_dtl_required(oldvd))
	return (oldvd);

	vdev_propagate_state(vd);

	/*
	* If there are more than two spares attached to a disk,
	* and those spares are not required, then we want to
	* attempt to free them up now so that they can be used
	* by other pools. Once we're back down to a single
	* disk+spare, we stop removing them.
	*/
	if (vd->vdev_children > 2) {
	newvd = vd->vdev_child[1];

	if (newvd->vdev_isspare && last->vdev_isspare &&
	vdev_dtl_empty(last, DTL_MISSING) &&
	vdev_dtl_empty(last, DTL_OUTAGE) &&
	!vdev_dtl_required(newvd))
	return (newvd);
	}
	}

	return (NULL);
	}

	static void
	spa_vdev_resilver_done(spa_t *spa)
	{
	vdev_t vd, pvd, *ppvd;
	uint64_t guid, sguid, pguid, ppguid;

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);

	while ((vd = spa_vdev_resilver_done_hunt(spa->spa_root_vdev)) != NULL) {
	pvd = vd->vdev_parent;
	ppvd = pvd->vdev_parent;
	guid = vd->vdev_guid;
	pguid = pvd->vdev_guid;
	ppguid = ppvd->vdev_guid;
	sguid = 0;
	/*
	* If we have just finished replacing a hot spared device, then
	* we need to detach the parent's first child (the original hot
	* spare) as well.
	*/
	if (ppvd->vdev_ops == &vdev_spare_ops && pvd->vdev_id == 0 &&
	ppvd->vdev_children == 2) {
	ASSERT(pvd->vdev_ops == &vdev_replacing_ops);
	sguid = ppvd->vdev_child[1]->vdev_guid;
	}
	ASSERT(vd->vdev_resilver_txg == 0 \|\| !vdev_dtl_required(vd));

	spa_config_exit(spa, SCL_ALL, FTAG);
	if (spa_vdev_detach(spa, guid, pguid, B_TRUE) != 0)
	return;
	if (sguid && spa_vdev_detach(spa, sguid, ppguid, B_TRUE) != 0)
	return;
	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	}

	spa_config_exit(spa, SCL_ALL, FTAG);

	/*
	* If a detach was not performed above replace waiters will not have
	* been notified. In which case we must do so now.
	*/
	spa_notify_waiters(spa);
	}

	/*
	* Update the stored path or FRU for this vdev.
	*/
	static int
	spa_vdev_set_common(spa_t spa, uint64_t guid, const char value,
	boolean_t ispath)
	{
	vdev_t *vd;
	boolean_t sync = B_FALSE;

	ASSERT(spa_writeable(spa));

	spa_vdev_state_enter(spa, SCL_ALL);

	if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL)
	return (spa_vdev_state_exit(spa, NULL, ENOENT));

	if (!vd->vdev_ops->vdev_op_leaf)
	return (spa_vdev_state_exit(spa, NULL, ENOTSUP));

	if (ispath) {
	if (strcmp(value, vd->vdev_path) != 0) {
	spa_strfree(vd->vdev_path);
	vd->vdev_path = spa_strdup(value);
	sync = B_TRUE;
	}
	} else {
	if (vd->vdev_fru == NULL) {
	vd->vdev_fru = spa_strdup(value);
	sync = B_TRUE;
	} else if (strcmp(value, vd->vdev_fru) != 0) {
	spa_strfree(vd->vdev_fru);
	vd->vdev_fru = spa_strdup(value);
	sync = B_TRUE;
	}
	}

	return (spa_vdev_state_exit(spa, sync ? vd : NULL, 0));
	}

	int
	spa_vdev_setpath(spa_t spa, uint64_t guid, const char newpath)
	{
	return (spa_vdev_set_common(spa, guid, newpath, B_TRUE));
	}

	int
	spa_vdev_setfru(spa_t spa, uint64_t guid, const char newfru)
	{
	return (spa_vdev_set_common(spa, guid, newfru, B_FALSE));
	}

	/*
	* ==========================================================================
	* SPA Scanning
	* ==========================================================================
	*/
	int
	spa_scrub_pause_resume(spa_t *spa, pool_scrub_cmd_t cmd)
	{
	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0);

	if (dsl_scan_resilvering(spa->spa_dsl_pool))
	return (SET_ERROR(EBUSY));

	return (dsl_scrub_set_pause_resume(spa->spa_dsl_pool, cmd));
	}

	int
	spa_scan_stop(spa_t *spa)
	{
	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0);
	if (dsl_scan_resilvering(spa->spa_dsl_pool))
	return (SET_ERROR(EBUSY));
	return (dsl_scan_cancel(spa->spa_dsl_pool));
	}

	int
	spa_scan(spa_t *spa, pool_scan_func_t func)
	{
	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == 0);

	if (func >= POOL_SCAN_FUNCS \|\| func == POOL_SCAN_NONE)
	return (SET_ERROR(ENOTSUP));

	if (func == POOL_SCAN_RESILVER &&
	!spa_feature_is_enabled(spa, SPA_FEATURE_RESILVER_DEFER))
	return (SET_ERROR(ENOTSUP));

	/*
	* If a resilver was requested, but there is no DTL on a
	* writeable leaf device, we have nothing to do.
	*/
	if (func == POOL_SCAN_RESILVER &&
	!vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL)) {
	spa_async_request(spa, SPA_ASYNC_RESILVER_DONE);
	return (0);
	}

	return (dsl_scan(spa->spa_dsl_pool, func));
	}

	/*
	* ==========================================================================
	* SPA async task processing
	* ==========================================================================
	*/

	static void
	spa_async_remove(spa_t spa, vdev_t vd)
	{
	if (vd->vdev_remove_wanted) {
	vd->vdev_remove_wanted = B_FALSE;
	vd->vdev_delayed_close = B_FALSE;
	vdev_set_state(vd, B_FALSE, VDEV_STATE_REMOVED, VDEV_AUX_NONE);

	/*
	* We want to clear the stats, but we don't want to do a full
	* vdev_clear() as that will cause us to throw away
	* degraded/faulted state as well as attempt to reopen the
	* device, all of which is a waste.
	*/
	vd->vdev_stat.vs_read_errors = 0;
	vd->vdev_stat.vs_write_errors = 0;
	vd->vdev_stat.vs_checksum_errors = 0;

	vdev_state_dirty(vd->vdev_top);

	/* Tell userspace that the vdev is gone. */
	zfs_post_remove(spa, vd);
	}

	for (int c = 0; c < vd->vdev_children; c++)
	spa_async_remove(spa, vd->vdev_child[c]);
	}

	static void
	spa_async_probe(spa_t spa, vdev_t vd)
	{
	if (vd->vdev_probe_wanted) {
	vd->vdev_probe_wanted = B_FALSE;
	vdev_reopen(vd); /* vdev_open() does the actual probe */
	}

	for (int c = 0; c < vd->vdev_children; c++)
	spa_async_probe(spa, vd->vdev_child[c]);
	}

	static void
	spa_async_autoexpand(spa_t spa, vdev_t vd)
	{
	if (!spa->spa_autoexpand)
	return;

	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];
	spa_async_autoexpand(spa, cvd);
	}

	if (!vd->vdev_ops->vdev_op_leaf \|\| vd->vdev_physpath == NULL)
	return;

	spa_event_notify(vd->vdev_spa, vd, NULL, ESC_ZFS_VDEV_AUTOEXPAND);
	}

	static void
	spa_async_thread(void *arg)
	{
	spa_t spa = (spa_t )arg;
	dsl_pool_t *dp = spa->spa_dsl_pool;
	int tasks;

	ASSERT(spa->spa_sync_on);

	mutex_enter(&spa->spa_async_lock);
	tasks = spa->spa_async_tasks;
	spa->spa_async_tasks = 0;
	mutex_exit(&spa->spa_async_lock);

	/*
	* See if the config needs to be updated.
	*/
	if (tasks & SPA_ASYNC_CONFIG_UPDATE) {
	uint64_t old_space, new_space;

	mutex_enter(&spa_namespace_lock);
	old_space = metaslab_class_get_space(spa_normal_class(spa));
	old_space += metaslab_class_get_space(spa_special_class(spa));
	old_space += metaslab_class_get_space(spa_dedup_class(spa));
	old_space += metaslab_class_get_space(
	spa_embedded_log_class(spa));

	spa_config_update(spa, SPA_CONFIG_UPDATE_POOL);

	new_space = metaslab_class_get_space(spa_normal_class(spa));
	new_space += metaslab_class_get_space(spa_special_class(spa));
	new_space += metaslab_class_get_space(spa_dedup_class(spa));
	new_space += metaslab_class_get_space(
	spa_embedded_log_class(spa));
	mutex_exit(&spa_namespace_lock);

	/*
	* If the pool grew as a result of the config update,
	* then log an internal history event.
	*/
	if (new_space != old_space) {
	spa_history_log_internal(spa, "vdev online", NULL,
	"pool '%s' size: %llu(+%llu)",
	spa_name(spa), (u_longlong_t)new_space,
	(u_longlong_t)(new_space - old_space));
	}
	}

	/*
	* See if any devices need to be marked REMOVED.
	*/
	if (tasks & SPA_ASYNC_REMOVE) {
	spa_vdev_state_enter(spa, SCL_NONE);
	spa_async_remove(spa, spa->spa_root_vdev);
	for (int i = 0; i < spa->spa_l2cache.sav_count; i++)
	spa_async_remove(spa, spa->spa_l2cache.sav_vdevs[i]);
	for (int i = 0; i < spa->spa_spares.sav_count; i++)
	spa_async_remove(spa, spa->spa_spares.sav_vdevs[i]);
	(void) spa_vdev_state_exit(spa, NULL, 0);
	}

	if ((tasks & SPA_ASYNC_AUTOEXPAND) && !spa_suspended(spa)) {
	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	spa_async_autoexpand(spa, spa->spa_root_vdev);
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	}

	/*
	* See if any devices need to be probed.
	*/
	if (tasks & SPA_ASYNC_PROBE) {
	spa_vdev_state_enter(spa, SCL_NONE);
	spa_async_probe(spa, spa->spa_root_vdev);
	(void) spa_vdev_state_exit(spa, NULL, 0);
	}

	/*
	* If any devices are done replacing, detach them.
	*/
	if (tasks & SPA_ASYNC_RESILVER_DONE \|\|
	tasks & SPA_ASYNC_REBUILD_DONE) {
	spa_vdev_resilver_done(spa);
	}

	/*
	* Kick off a resilver.
	*/
	if (tasks & SPA_ASYNC_RESILVER &&
	!vdev_rebuild_active(spa->spa_root_vdev) &&
	(!dsl_scan_resilvering(dp) \|\|
	!spa_feature_is_enabled(dp->dp_spa, SPA_FEATURE_RESILVER_DEFER)))
	dsl_scan_restart_resilver(dp, 0);

	if (tasks & SPA_ASYNC_INITIALIZE_RESTART) {
	mutex_enter(&spa_namespace_lock);
	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	vdev_initialize_restart(spa->spa_root_vdev);
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	mutex_exit(&spa_namespace_lock);
	}

	if (tasks & SPA_ASYNC_TRIM_RESTART) {
	mutex_enter(&spa_namespace_lock);
	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	vdev_trim_restart(spa->spa_root_vdev);
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	mutex_exit(&spa_namespace_lock);
	}

	if (tasks & SPA_ASYNC_AUTOTRIM_RESTART) {
	mutex_enter(&spa_namespace_lock);
	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	vdev_autotrim_restart(spa);
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	mutex_exit(&spa_namespace_lock);
	}

	/*
	* Kick off L2 cache whole device TRIM.
	*/
	if (tasks & SPA_ASYNC_L2CACHE_TRIM) {
	mutex_enter(&spa_namespace_lock);
	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	vdev_trim_l2arc(spa);
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	mutex_exit(&spa_namespace_lock);
	}

	/*
	* Kick off L2 cache rebuilding.
	*/
	if (tasks & SPA_ASYNC_L2CACHE_REBUILD) {
	mutex_enter(&spa_namespace_lock);
	spa_config_enter(spa, SCL_L2ARC, FTAG, RW_READER);
	l2arc_spa_rebuild_start(spa);
	spa_config_exit(spa, SCL_L2ARC, FTAG);
	mutex_exit(&spa_namespace_lock);
	}

	/*
	* Let the world know that we're done.
	*/
	mutex_enter(&spa->spa_async_lock);
	spa->spa_async_thread = NULL;
	cv_broadcast(&spa->spa_async_cv);
	mutex_exit(&spa->spa_async_lock);
	thread_exit();
	}

	void
	spa_async_suspend(spa_t *spa)
	{
	mutex_enter(&spa->spa_async_lock);
	spa->spa_async_suspended++;
	while (spa->spa_async_thread != NULL)
	cv_wait(&spa->spa_async_cv, &spa->spa_async_lock);
	mutex_exit(&spa->spa_async_lock);

	spa_vdev_remove_suspend(spa);

	zthr_t *condense_thread = spa->spa_condense_zthr;
	if (condense_thread != NULL)
	zthr_cancel(condense_thread);

	zthr_t *discard_thread = spa->spa_checkpoint_discard_zthr;
	if (discard_thread != NULL)
	zthr_cancel(discard_thread);

	zthr_t *ll_delete_thread = spa->spa_livelist_delete_zthr;
	if (ll_delete_thread != NULL)
	zthr_cancel(ll_delete_thread);

	zthr_t *ll_condense_thread = spa->spa_livelist_condense_zthr;
	if (ll_condense_thread != NULL)
	zthr_cancel(ll_condense_thread);
	}

	void
	spa_async_resume(spa_t *spa)
	{
	mutex_enter(&spa->spa_async_lock);
	ASSERT(spa->spa_async_suspended != 0);
	spa->spa_async_suspended--;
	mutex_exit(&spa->spa_async_lock);
	spa_restart_removal(spa);

	zthr_t *condense_thread = spa->spa_condense_zthr;
	if (condense_thread != NULL)
	zthr_resume(condense_thread);

	zthr_t *discard_thread = spa->spa_checkpoint_discard_zthr;
	if (discard_thread != NULL)
	zthr_resume(discard_thread);

	zthr_t *ll_delete_thread = spa->spa_livelist_delete_zthr;
	if (ll_delete_thread != NULL)
	zthr_resume(ll_delete_thread);

	zthr_t *ll_condense_thread = spa->spa_livelist_condense_zthr;
	if (ll_condense_thread != NULL)
	zthr_resume(ll_condense_thread);
	}

	static boolean_t
	spa_async_tasks_pending(spa_t *spa)
	{
	uint_t non_config_tasks;
	uint_t config_task;
	boolean_t config_task_suspended;

	non_config_tasks = spa->spa_async_tasks & ~SPA_ASYNC_CONFIG_UPDATE;
	config_task = spa->spa_async_tasks & SPA_ASYNC_CONFIG_UPDATE;
	if (spa->spa_ccw_fail_time == 0) {
	config_task_suspended = B_FALSE;
	} else {
	config_task_suspended =
	(gethrtime() - spa->spa_ccw_fail_time) <
	((hrtime_t)zfs_ccw_retry_interval * NANOSEC);
	}

	return (non_config_tasks \|\| (config_task && !config_task_suspended));
	}

	static void
	spa_async_dispatch(spa_t *spa)
	{
	mutex_enter(&spa->spa_async_lock);
	if (spa_async_tasks_pending(spa) &&
	!spa->spa_async_suspended &&
	spa->spa_async_thread == NULL)
	spa->spa_async_thread = thread_create(NULL, 0,
	spa_async_thread, spa, 0, &p0, TS_RUN, maxclsyspri);
	mutex_exit(&spa->spa_async_lock);
	}

	void
	spa_async_request(spa_t *spa, int task)
	{
	zfs_dbgmsg("spa=%s async request task=%u", spa->spa_name, task);
	mutex_enter(&spa->spa_async_lock);
	spa->spa_async_tasks \|= task;
	mutex_exit(&spa->spa_async_lock);
	}

	int
	spa_async_tasks(spa_t *spa)
	{
	return (spa->spa_async_tasks);
	}

	/*
	* ==========================================================================
	* SPA syncing routines
	* ==========================================================================
	*/


	static int
	bpobj_enqueue_cb(void arg, const blkptr_t bp, boolean_t bp_freed,
	dmu_tx_t *tx)
	{
	bpobj_t *bpo = arg;
	bpobj_enqueue(bpo, bp, bp_freed, tx);
	return (0);
	}

	int
	bpobj_enqueue_alloc_cb(void arg, const blkptr_t bp, dmu_tx_t *tx)
	{
	return (bpobj_enqueue_cb(arg, bp, B_FALSE, tx));
	}

	int
	bpobj_enqueue_free_cb(void arg, const blkptr_t bp, dmu_tx_t *tx)
	{
	return (bpobj_enqueue_cb(arg, bp, B_TRUE, tx));
	}

	static int
	spa_free_sync_cb(void arg, const blkptr_t bp, dmu_tx_t *tx)
	{
	zio_t *pio = arg;

	zio_nowait(zio_free_sync(pio, pio->io_spa, dmu_tx_get_txg(tx), bp,
	pio->io_flags));
	return (0);
	}

	static int
	bpobj_spa_free_sync_cb(void arg, const blkptr_t bp, boolean_t bp_freed,
	dmu_tx_t *tx)
	{
	ASSERT(!bp_freed);
	return (spa_free_sync_cb(arg, bp, tx));
	}

	/*
	* Note: this simple function is not inlined to make it easier to dtrace the
	* amount of time spent syncing frees.
	*/
	static void
	spa_sync_frees(spa_t spa, bplist_t bpl, dmu_tx_t *tx)
	{
	zio_t *zio = zio_root(spa, NULL, NULL, 0);
	bplist_iterate(bpl, spa_free_sync_cb, zio, tx);
	VERIFY(zio_wait(zio) == 0);
	}

	/*
	* Note: this simple function is not inlined to make it easier to dtrace the
	* amount of time spent syncing deferred frees.
	*/
	static void
	spa_sync_deferred_frees(spa_t spa, dmu_tx_t tx)
	{
	if (spa_sync_pass(spa) != 1)
	return;

	/*
	* Note:
	* If the log space map feature is active, we stop deferring
	* frees to the next TXG and therefore running this function
	* would be considered a no-op as spa_deferred_bpobj should
	* not have any entries.
	*
	* That said we run this function anyway (instead of returning
	* immediately) for the edge-case scenario where we just
	* activated the log space map feature in this TXG but we have
	* deferred frees from the previous TXG.
	*/
	zio_t *zio = zio_root(spa, NULL, NULL, 0);
	VERIFY3U(bpobj_iterate(&spa->spa_deferred_bpobj,
	bpobj_spa_free_sync_cb, zio, tx), ==, 0);
	VERIFY0(zio_wait(zio));
	}

	static void
	spa_sync_nvlist(spa_t spa, uint64_t obj, nvlist_t nv, dmu_tx_t *tx)
	{
	char *packed = NULL;
	size_t bufsize;
	size_t nvsize = 0;
	dmu_buf_t *db;

	VERIFY(nvlist_size(nv, &nvsize, NV_ENCODE_XDR) == 0);

	/*
	* Write full (SPA_CONFIG_BLOCKSIZE) blocks of configuration
	* information. This avoids the dmu_buf_will_dirty() path and
	* saves us a pre-read to get data we don't actually care about.
	*/
	bufsize = P2ROUNDUP((uint64_t)nvsize, SPA_CONFIG_BLOCKSIZE);
	packed = vmem_alloc(bufsize, KM_SLEEP);

	VERIFY(nvlist_pack(nv, &packed, &nvsize, NV_ENCODE_XDR,
	KM_SLEEP) == 0);
	bzero(packed + nvsize, bufsize - nvsize);

	dmu_write(spa->spa_meta_objset, obj, 0, bufsize, packed, tx);

	vmem_free(packed, bufsize);

	VERIFY(0 == dmu_bonus_hold(spa->spa_meta_objset, obj, FTAG, &db));
	dmu_buf_will_dirty(db, tx);
	(uint64_t )db->db_data = nvsize;
	dmu_buf_rele(db, FTAG);
	}

	static void
	spa_sync_aux_dev(spa_t spa, spa_aux_vdev_t sav, dmu_tx_t *tx,
	const char config, const char entry)
	{
	nvlist_t *nvroot;
	nvlist_t **list;
	int i;

	if (!sav->sav_sync)
	return;

	/*
	* Update the MOS nvlist describing the list of available devices.
	* spa_validate_aux() will have already made sure this nvlist is
	* valid and the vdevs are labeled appropriately.
	*/
	if (sav->sav_object == 0) {
	sav->sav_object = dmu_object_alloc(spa->spa_meta_objset,
	DMU_OT_PACKED_NVLIST, 1 << 14, DMU_OT_PACKED_NVLIST_SIZE,
	sizeof (uint64_t), tx);
	VERIFY(zap_update(spa->spa_meta_objset,
	DMU_POOL_DIRECTORY_OBJECT, entry, sizeof (uint64_t), 1,
	&sav->sav_object, tx) == 0);
	}

	nvroot = fnvlist_alloc();
	if (sav->sav_count == 0) {
	- fnvlist_add_nvlist_array(nvroot, config, NULL, 0);
	+ fnvlist_add_nvlist_array(nvroot, config,
	+ (const nvlist_t * const *)NULL, 0);
	} else {
	list = kmem_alloc(sav->sav_countsizeof (void ), KM_SLEEP);
	for (i = 0; i < sav->sav_count; i++)
	list[i] = vdev_config_generate(spa, sav->sav_vdevs[i],
	B_FALSE, VDEV_CONFIG_L2CACHE);
	- fnvlist_add_nvlist_array(nvroot, config, list, sav->sav_count);
	+ fnvlist_add_nvlist_array(nvroot, config,
	+ (const nvlist_t * const *)list, sav->sav_count);
	for (i = 0; i < sav->sav_count; i++)
	nvlist_free(list[i]);
	kmem_free(list, sav->sav_count * sizeof (void *));
	}

	spa_sync_nvlist(spa, sav->sav_object, nvroot, tx);
	nvlist_free(nvroot);

	sav->sav_sync = B_FALSE;
	}

	/*
	* Rebuild spa's all-vdev ZAP from the vdev ZAPs indicated in each vdev_t.
	* The all-vdev ZAP must be empty.
	*/
	static void
	spa_avz_build(vdev_t vd, uint64_t avz, dmu_tx_t tx)
	{
	spa_t *spa = vd->vdev_spa;

	if (vd->vdev_top_zap != 0) {
	VERIFY0(zap_add_int(spa->spa_meta_objset, avz,
	vd->vdev_top_zap, tx));
	}
	if (vd->vdev_leaf_zap != 0) {
	VERIFY0(zap_add_int(spa->spa_meta_objset, avz,
	vd->vdev_leaf_zap, tx));
	}
	for (uint64_t i = 0; i < vd->vdev_children; i++) {
	spa_avz_build(vd->vdev_child[i], avz, tx);
	}
	}

	static void
	spa_sync_config_object(spa_t spa, dmu_tx_t tx)
	{
	nvlist_t *config;

	/*
	* If the pool is being imported from a pre-per-vdev-ZAP version of ZFS,
	* its config may not be dirty but we still need to build per-vdev ZAPs.
	* Similarly, if the pool is being assembled (e.g. after a split), we
	* need to rebuild the AVZ although the config may not be dirty.
	*/
	if (list_is_empty(&spa->spa_config_dirty_list) &&
	spa->spa_avz_action == AVZ_ACTION_NONE)
	return;

	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);

	ASSERT(spa->spa_avz_action == AVZ_ACTION_NONE \|\|
	spa->spa_avz_action == AVZ_ACTION_INITIALIZE \|\|
	spa->spa_all_vdev_zaps != 0);

	if (spa->spa_avz_action == AVZ_ACTION_REBUILD) {
	/* Make and build the new AVZ */
	uint64_t new_avz = zap_create(spa->spa_meta_objset,
	DMU_OTN_ZAP_METADATA, DMU_OT_NONE, 0, tx);
	spa_avz_build(spa->spa_root_vdev, new_avz, tx);

	/* Diff old AVZ with new one */
	zap_cursor_t zc;
	zap_attribute_t za;

	for (zap_cursor_init(&zc, spa->spa_meta_objset,
	spa->spa_all_vdev_zaps);
	zap_cursor_retrieve(&zc, &za) == 0;
	zap_cursor_advance(&zc)) {
	uint64_t vdzap = za.za_first_integer;
	if (zap_lookup_int(spa->spa_meta_objset, new_avz,
	vdzap) == ENOENT) {
	/*
	* ZAP is listed in old AVZ but not in new one;
	* destroy it
	*/
	VERIFY0(zap_destroy(spa->spa_meta_objset, vdzap,
	tx));
	}
	}

	zap_cursor_fini(&zc);

	/* Destroy the old AVZ */
	VERIFY0(zap_destroy(spa->spa_meta_objset,
	spa->spa_all_vdev_zaps, tx));

	/* Replace the old AVZ in the dir obj with the new one */
	VERIFY0(zap_update(spa->spa_meta_objset,
	DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_VDEV_ZAP_MAP,
	sizeof (new_avz), 1, &new_avz, tx));

	spa->spa_all_vdev_zaps = new_avz;
	} else if (spa->spa_avz_action == AVZ_ACTION_DESTROY) {
	zap_cursor_t zc;
	zap_attribute_t za;

	/* Walk through the AVZ and destroy all listed ZAPs */
	for (zap_cursor_init(&zc, spa->spa_meta_objset,
	spa->spa_all_vdev_zaps);
	zap_cursor_retrieve(&zc, &za) == 0;
	zap_cursor_advance(&zc)) {
	uint64_t zap = za.za_first_integer;
	VERIFY0(zap_destroy(spa->spa_meta_objset, zap, tx));
	}

	zap_cursor_fini(&zc);

	/* Destroy and unlink the AVZ itself */
	VERIFY0(zap_destroy(spa->spa_meta_objset,
	spa->spa_all_vdev_zaps, tx));
	VERIFY0(zap_remove(spa->spa_meta_objset,
	DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_VDEV_ZAP_MAP, tx));
	spa->spa_all_vdev_zaps = 0;
	}

	if (spa->spa_all_vdev_zaps == 0) {
	spa->spa_all_vdev_zaps = zap_create_link(spa->spa_meta_objset,
	DMU_OTN_ZAP_METADATA, DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_VDEV_ZAP_MAP, tx);
	}
	spa->spa_avz_action = AVZ_ACTION_NONE;

	/* Create ZAPs for vdevs that don't have them. */
	vdev_construct_zaps(spa->spa_root_vdev, tx);

	config = spa_config_generate(spa, spa->spa_root_vdev,
	dmu_tx_get_txg(tx), B_FALSE);

	/*
	* If we're upgrading the spa version then make sure that
	* the config object gets updated with the correct version.
	*/
	if (spa->spa_ubsync.ub_version < spa->spa_uberblock.ub_version)
	fnvlist_add_uint64(config, ZPOOL_CONFIG_VERSION,
	spa->spa_uberblock.ub_version);

	spa_config_exit(spa, SCL_STATE, FTAG);

	nvlist_free(spa->spa_config_syncing);
	spa->spa_config_syncing = config;

	spa_sync_nvlist(spa, spa->spa_config_object, config, tx);
	}

	static void
	spa_sync_version(void arg, dmu_tx_t tx)
	{
	uint64_t *versionp = arg;
	uint64_t version = *versionp;
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;

	/*
	* Setting the version is special cased when first creating the pool.
	*/
	ASSERT(tx->tx_txg != TXG_INITIAL);

	ASSERT(SPA_VERSION_IS_SUPPORTED(version));
	ASSERT(version >= spa_version(spa));

	spa->spa_uberblock.ub_version = version;
	vdev_config_dirty(spa->spa_root_vdev);
	spa_history_log_internal(spa, "set", tx, "version=%lld",
	(longlong_t)version);
	}

	/*
	* Set zpool properties.
	*/
	static void
	spa_sync_props(void arg, dmu_tx_t tx)
	{
	nvlist_t *nvp = arg;
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;
	objset_t *mos = spa->spa_meta_objset;
	nvpair_t *elem = NULL;

	mutex_enter(&spa->spa_props_lock);

	while ((elem = nvlist_next_nvpair(nvp, elem))) {
	uint64_t intval;
	char strval, fname;
	zpool_prop_t prop;
	const char *propname;
	zprop_type_t proptype;
	spa_feature_t fid;

	switch (prop = zpool_name_to_prop(nvpair_name(elem))) {
	case ZPOOL_PROP_INVAL:
	/*
	* We checked this earlier in spa_prop_validate().
	*/
	ASSERT(zpool_prop_feature(nvpair_name(elem)));

	fname = strchr(nvpair_name(elem), '@') + 1;
	VERIFY0(zfeature_lookup_name(fname, &fid));

	spa_feature_enable(spa, fid, tx);
	spa_history_log_internal(spa, "set", tx,
	"%s=enabled", nvpair_name(elem));
	break;

	case ZPOOL_PROP_VERSION:
	intval = fnvpair_value_uint64(elem);
	/*
	* The version is synced separately before other
	* properties and should be correct by now.
	*/
	ASSERT3U(spa_version(spa), >=, intval);
	break;

	case ZPOOL_PROP_ALTROOT:
	/*
	* 'altroot' is a non-persistent property. It should
	* have been set temporarily at creation or import time.
	*/
	ASSERT(spa->spa_root != NULL);
	break;

	case ZPOOL_PROP_READONLY:
	case ZPOOL_PROP_CACHEFILE:
	/*
	* 'readonly' and 'cachefile' are also non-persistent
	* properties.
	*/
	break;
	case ZPOOL_PROP_COMMENT:
	strval = fnvpair_value_string(elem);
	if (spa->spa_comment != NULL)
	spa_strfree(spa->spa_comment);
	spa->spa_comment = spa_strdup(strval);
	/*
	* We need to dirty the configuration on all the vdevs
	* so that their labels get updated. We also need to
	* update the cache file to keep it in sync with the
	* MOS version. It's unnecessary to do this for pool
	* creation since the vdev's configuration has already
	* been dirtied.
	*/
	if (tx->tx_txg != TXG_INITIAL) {
	vdev_config_dirty(spa->spa_root_vdev);
	spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE);
	}
	spa_history_log_internal(spa, "set", tx,
	"%s=%s", nvpair_name(elem), strval);
	break;
	case ZPOOL_PROP_COMPATIBILITY:
	strval = fnvpair_value_string(elem);
	if (spa->spa_compatibility != NULL)
	spa_strfree(spa->spa_compatibility);
	spa->spa_compatibility = spa_strdup(strval);
	/*
	* Dirty the configuration on vdevs as above.
	*/
	if (tx->tx_txg != TXG_INITIAL) {
	vdev_config_dirty(spa->spa_root_vdev);
	spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE);
	}

	spa_history_log_internal(spa, "set", tx,
	"%s=%s", nvpair_name(elem), strval);
	break;

	default:
	/*
	* Set pool property values in the poolprops mos object.
	*/
	if (spa->spa_pool_props_object == 0) {
	spa->spa_pool_props_object =
	zap_create_link(mos, DMU_OT_POOL_PROPS,
	DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_PROPS,
	tx);
	}

	/* normalize the property name */
	propname = zpool_prop_to_name(prop);
	proptype = zpool_prop_get_type(prop);

	if (nvpair_type(elem) == DATA_TYPE_STRING) {
	ASSERT(proptype == PROP_TYPE_STRING);
	strval = fnvpair_value_string(elem);
	VERIFY0(zap_update(mos,
	spa->spa_pool_props_object, propname,
	1, strlen(strval) + 1, strval, tx));
	spa_history_log_internal(spa, "set", tx,
	"%s=%s", nvpair_name(elem), strval);
	} else if (nvpair_type(elem) == DATA_TYPE_UINT64) {
	intval = fnvpair_value_uint64(elem);

	if (proptype == PROP_TYPE_INDEX) {
	const char *unused;
	VERIFY0(zpool_prop_index_to_string(
	prop, intval, &unused));
	}
	VERIFY0(zap_update(mos,
	spa->spa_pool_props_object, propname,
	8, 1, &intval, tx));
	spa_history_log_internal(spa, "set", tx,
	"%s=%lld", nvpair_name(elem),
	(longlong_t)intval);
	} else {
	ASSERT(0); /* not allowed */
	}

	switch (prop) {
	case ZPOOL_PROP_DELEGATION:
	spa->spa_delegation = intval;
	break;
	case ZPOOL_PROP_BOOTFS:
	spa->spa_bootfs = intval;
	break;
	case ZPOOL_PROP_FAILUREMODE:
	spa->spa_failmode = intval;
	break;
	case ZPOOL_PROP_AUTOTRIM:
	spa->spa_autotrim = intval;
	spa_async_request(spa,
	SPA_ASYNC_AUTOTRIM_RESTART);
	break;
	case ZPOOL_PROP_AUTOEXPAND:
	spa->spa_autoexpand = intval;
	if (tx->tx_txg != TXG_INITIAL)
	spa_async_request(spa,
	SPA_ASYNC_AUTOEXPAND);
	break;
	case ZPOOL_PROP_MULTIHOST:
	spa->spa_multihost = intval;
	break;
	default:
	break;
	}
	}

	}

	mutex_exit(&spa->spa_props_lock);
	}

	/*
	* Perform one-time upgrade on-disk changes. spa_version() does not
	* reflect the new version this txg, so there must be no changes this
	* txg to anything that the upgrade code depends on after it executes.
	* Therefore this must be called after dsl_pool_sync() does the sync
	* tasks.
	*/
	static void
	spa_sync_upgrades(spa_t spa, dmu_tx_t tx)
	{
	if (spa_sync_pass(spa) != 1)
	return;

	dsl_pool_t *dp = spa->spa_dsl_pool;
	rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG);

	if (spa->spa_ubsync.ub_version < SPA_VERSION_ORIGIN &&
	spa->spa_uberblock.ub_version >= SPA_VERSION_ORIGIN) {
	dsl_pool_create_origin(dp, tx);

	/* Keeping the origin open increases spa_minref */
	spa->spa_minref += 3;
	}

	if (spa->spa_ubsync.ub_version < SPA_VERSION_NEXT_CLONES &&
	spa->spa_uberblock.ub_version >= SPA_VERSION_NEXT_CLONES) {
	dsl_pool_upgrade_clones(dp, tx);
	}

	if (spa->spa_ubsync.ub_version < SPA_VERSION_DIR_CLONES &&
	spa->spa_uberblock.ub_version >= SPA_VERSION_DIR_CLONES) {
	dsl_pool_upgrade_dir_clones(dp, tx);

	/* Keeping the freedir open increases spa_minref */
	spa->spa_minref += 3;
	}

	if (spa->spa_ubsync.ub_version < SPA_VERSION_FEATURES &&
	spa->spa_uberblock.ub_version >= SPA_VERSION_FEATURES) {
	spa_feature_create_zap_objects(spa, tx);
	}

	/*
	* LZ4_COMPRESS feature's behaviour was changed to activate_on_enable
	* when possibility to use lz4 compression for metadata was added
	* Old pools that have this feature enabled must be upgraded to have
	* this feature active
	*/
	if (spa->spa_uberblock.ub_version >= SPA_VERSION_FEATURES) {
	boolean_t lz4_en = spa_feature_is_enabled(spa,
	SPA_FEATURE_LZ4_COMPRESS);
	boolean_t lz4_ac = spa_feature_is_active(spa,
	SPA_FEATURE_LZ4_COMPRESS);

	if (lz4_en && !lz4_ac)
	spa_feature_incr(spa, SPA_FEATURE_LZ4_COMPRESS, tx);
	}

	/*
	* If we haven't written the salt, do so now. Note that the
	* feature may not be activated yet, but that's fine since
	* the presence of this ZAP entry is backwards compatible.
	*/
	if (zap_contains(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_CHECKSUM_SALT) == ENOENT) {
	VERIFY0(zap_add(spa->spa_meta_objset,
	DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_CHECKSUM_SALT, 1,
	sizeof (spa->spa_cksum_salt.zcs_bytes),
	spa->spa_cksum_salt.zcs_bytes, tx));
	}

	rrw_exit(&dp->dp_config_rwlock, FTAG);
	}

	static void
	vdev_indirect_state_sync_verify(vdev_t *vd)
	{
	vdev_indirect_mapping_t *vim __maybe_unused = vd->vdev_indirect_mapping;
	vdev_indirect_births_t *vib __maybe_unused = vd->vdev_indirect_births;

	if (vd->vdev_ops == &vdev_indirect_ops) {
	ASSERT(vim != NULL);
	ASSERT(vib != NULL);
	}

	uint64_t obsolete_sm_object = 0;
	ASSERT0(vdev_obsolete_sm_object(vd, &obsolete_sm_object));
	if (obsolete_sm_object != 0) {
	ASSERT(vd->vdev_obsolete_sm != NULL);
	ASSERT(vd->vdev_removing \|\|
	vd->vdev_ops == &vdev_indirect_ops);
	ASSERT(vdev_indirect_mapping_num_entries(vim) > 0);
	ASSERT(vdev_indirect_mapping_bytes_mapped(vim) > 0);
	ASSERT3U(obsolete_sm_object, ==,
	space_map_object(vd->vdev_obsolete_sm));
	ASSERT3U(vdev_indirect_mapping_bytes_mapped(vim), >=,
	space_map_allocated(vd->vdev_obsolete_sm));
	}
	ASSERT(vd->vdev_obsolete_segments != NULL);

	/*
	* Since frees / remaps to an indirect vdev can only
	* happen in syncing context, the obsolete segments
	* tree must be empty when we start syncing.
	*/
	ASSERT0(range_tree_space(vd->vdev_obsolete_segments));
	}

	/*
	* Set the top-level vdev's max queue depth. Evaluate each top-level's
	* async write queue depth in case it changed. The max queue depth will
	* not change in the middle of syncing out this txg.
	*/
	static void
	spa_sync_adjust_vdev_max_queue_depth(spa_t *spa)
	{
	ASSERT(spa_writeable(spa));

	vdev_t *rvd = spa->spa_root_vdev;
	uint32_t max_queue_depth = zfs_vdev_async_write_max_active *
	zfs_vdev_queue_depth_pct / 100;
	metaslab_class_t *normal = spa_normal_class(spa);
	metaslab_class_t *special = spa_special_class(spa);
	metaslab_class_t *dedup = spa_dedup_class(spa);

	uint64_t slots_per_allocator = 0;
	for (int c = 0; c < rvd->vdev_children; c++) {
	vdev_t *tvd = rvd->vdev_child[c];

	metaslab_group_t *mg = tvd->vdev_mg;
	if (mg == NULL \|\| !metaslab_group_initialized(mg))
	continue;

	metaslab_class_t *mc = mg->mg_class;
	if (mc != normal && mc != special && mc != dedup)
	continue;

	/*
	* It is safe to do a lock-free check here because only async
	* allocations look at mg_max_alloc_queue_depth, and async
	* allocations all happen from spa_sync().
	*/
	for (int i = 0; i < mg->mg_allocators; i++) {
	ASSERT0(zfs_refcount_count(
	&(mg->mg_allocator[i].mga_alloc_queue_depth)));
	}
	mg->mg_max_alloc_queue_depth = max_queue_depth;

	for (int i = 0; i < mg->mg_allocators; i++) {
	mg->mg_allocator[i].mga_cur_max_alloc_queue_depth =
	zfs_vdev_def_queue_depth;
	}
	slots_per_allocator += zfs_vdev_def_queue_depth;
	}

	for (int i = 0; i < spa->spa_alloc_count; i++) {
	ASSERT0(zfs_refcount_count(&normal->mc_allocator[i].
	mca_alloc_slots));
	ASSERT0(zfs_refcount_count(&special->mc_allocator[i].
	mca_alloc_slots));
	ASSERT0(zfs_refcount_count(&dedup->mc_allocator[i].
	mca_alloc_slots));
	normal->mc_allocator[i].mca_alloc_max_slots =
	slots_per_allocator;
	special->mc_allocator[i].mca_alloc_max_slots =
	slots_per_allocator;
	dedup->mc_allocator[i].mca_alloc_max_slots =
	slots_per_allocator;
	}
	normal->mc_alloc_throttle_enabled = zio_dva_throttle_enabled;
	special->mc_alloc_throttle_enabled = zio_dva_throttle_enabled;
	dedup->mc_alloc_throttle_enabled = zio_dva_throttle_enabled;
	}

	static void
	spa_sync_condense_indirect(spa_t spa, dmu_tx_t tx)
	{
	ASSERT(spa_writeable(spa));

	vdev_t *rvd = spa->spa_root_vdev;
	for (int c = 0; c < rvd->vdev_children; c++) {
	vdev_t *vd = rvd->vdev_child[c];
	vdev_indirect_state_sync_verify(vd);

	if (vdev_indirect_should_condense(vd)) {
	spa_condense_indirect_start_sync(vd, tx);
	break;
	}
	}
	}

	static void
	spa_sync_iterate_to_convergence(spa_t spa, dmu_tx_t tx)
	{
	objset_t *mos = spa->spa_meta_objset;
	dsl_pool_t *dp = spa->spa_dsl_pool;
	uint64_t txg = tx->tx_txg;
	bplist_t *free_bpl = &spa->spa_free_bplist[txg & TXG_MASK];

	do {
	int pass = ++spa->spa_sync_pass;

	spa_sync_config_object(spa, tx);
	spa_sync_aux_dev(spa, &spa->spa_spares, tx,
	ZPOOL_CONFIG_SPARES, DMU_POOL_SPARES);
	spa_sync_aux_dev(spa, &spa->spa_l2cache, tx,
	ZPOOL_CONFIG_L2CACHE, DMU_POOL_L2CACHE);
	spa_errlog_sync(spa, txg);
	dsl_pool_sync(dp, txg);

	if (pass < zfs_sync_pass_deferred_free \|\|
	spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP)) {
	/*
	* If the log space map feature is active we don't
	* care about deferred frees and the deferred bpobj
	* as the log space map should effectively have the
	* same results (i.e. appending only to one object).
	*/
	spa_sync_frees(spa, free_bpl, tx);
	} else {
	/*
	* We can not defer frees in pass 1, because
	* we sync the deferred frees later in pass 1.
	*/
	ASSERT3U(pass, >, 1);
	bplist_iterate(free_bpl, bpobj_enqueue_alloc_cb,
	&spa->spa_deferred_bpobj, tx);
	}

	ddt_sync(spa, txg);
	dsl_scan_sync(dp, tx);
	svr_sync(spa, tx);
	spa_sync_upgrades(spa, tx);

	spa_flush_metaslabs(spa, tx);

	vdev_t *vd = NULL;
	while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, txg))
	!= NULL)
	vdev_sync(vd, txg);

	/*
	* Note: We need to check if the MOS is dirty because we could
	* have marked the MOS dirty without updating the uberblock
	* (e.g. if we have sync tasks but no dirty user data). We need
	* to check the uberblock's rootbp because it is updated if we
	* have synced out dirty data (though in this case the MOS will
	* most likely also be dirty due to second order effects, we
	* don't want to rely on that here).
	*/
	if (pass == 1 &&
	spa->spa_uberblock.ub_rootbp.blk_birth < txg &&
	!dmu_objset_is_dirty(mos, txg)) {
	/*
	* Nothing changed on the first pass, therefore this
	* TXG is a no-op. Avoid syncing deferred frees, so
	* that we can keep this TXG as a no-op.
	*/
	ASSERT(txg_list_empty(&dp->dp_dirty_datasets, txg));
	ASSERT(txg_list_empty(&dp->dp_dirty_dirs, txg));
	ASSERT(txg_list_empty(&dp->dp_sync_tasks, txg));
	ASSERT(txg_list_empty(&dp->dp_early_sync_tasks, txg));
	break;
	}

	spa_sync_deferred_frees(spa, tx);
	} while (dmu_objset_is_dirty(mos, txg));
	}

	/*
	* Rewrite the vdev configuration (which includes the uberblock) to
	* commit the transaction group.
	*
	* If there are no dirty vdevs, we sync the uberblock to a few random
	* top-level vdevs that are known to be visible in the config cache
	* (see spa_vdev_add() for a complete description). If there are dirty
	* vdevs, sync the uberblock to all vdevs.
	*/
	static void
	spa_sync_rewrite_vdev_config(spa_t spa, dmu_tx_t tx)
	{
	vdev_t *rvd = spa->spa_root_vdev;
	uint64_t txg = tx->tx_txg;

	for (;;) {
	int error = 0;

	/*
	* We hold SCL_STATE to prevent vdev open/close/etc.
	* while we're attempting to write the vdev labels.
	*/
	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);

	if (list_is_empty(&spa->spa_config_dirty_list)) {
	vdev_t *svd[SPA_SYNC_MIN_VDEVS] = { NULL };
	int svdcount = 0;
	int children = rvd->vdev_children;
	int c0 = random_in_range(children);

	for (int c = 0; c < children; c++) {
	vdev_t *vd =
	rvd->vdev_child[(c0 + c) % children];

	/* Stop when revisiting the first vdev */
	if (c > 0 && svd[0] == vd)
	break;

	if (vd->vdev_ms_array == 0 \|\|
	vd->vdev_islog \|\|
	!vdev_is_concrete(vd))
	continue;

	svd[svdcount++] = vd;
	if (svdcount == SPA_SYNC_MIN_VDEVS)
	break;
	}
	error = vdev_config_sync(svd, svdcount, txg);
	} else {
	error = vdev_config_sync(rvd->vdev_child,
	rvd->vdev_children, txg);
	}

	if (error == 0)
	spa->spa_last_synced_guid = rvd->vdev_guid;

	spa_config_exit(spa, SCL_STATE, FTAG);

	if (error == 0)
	break;
	zio_suspend(spa, NULL, ZIO_SUSPEND_IOERR);
	zio_resume_wait(spa);
	}
	}

	/*
	* Sync the specified transaction group. New blocks may be dirtied as
	* part of the process, so we iterate until it converges.
	*/
	void
	spa_sync(spa_t *spa, uint64_t txg)
	{
	vdev_t *vd = NULL;

	VERIFY(spa_writeable(spa));

	/*
	* Wait for i/os issued in open context that need to complete
	* before this txg syncs.
	*/
	(void) zio_wait(spa->spa_txg_zio[txg & TXG_MASK]);
	spa->spa_txg_zio[txg & TXG_MASK] = zio_root(spa, NULL, NULL,
	ZIO_FLAG_CANFAIL);

	/*
	* Lock out configuration changes.
	*/
	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);

	spa->spa_syncing_txg = txg;
	spa->spa_sync_pass = 0;

	for (int i = 0; i < spa->spa_alloc_count; i++) {
	mutex_enter(&spa->spa_allocs[i].spaa_lock);
	VERIFY0(avl_numnodes(&spa->spa_allocs[i].spaa_tree));
	mutex_exit(&spa->spa_allocs[i].spaa_lock);
	}

	/*
	* If there are any pending vdev state changes, convert them
	* into config changes that go out with this transaction group.
	*/
	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
	while (list_head(&spa->spa_state_dirty_list) != NULL) {
	/*
	* We need the write lock here because, for aux vdevs,
	* calling vdev_config_dirty() modifies sav_config.
	* This is ugly and will become unnecessary when we
	* eliminate the aux vdev wart by integrating all vdevs
	* into the root vdev tree.
	*/
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	spa_config_enter(spa, SCL_CONFIG \| SCL_STATE, FTAG, RW_WRITER);
	while ((vd = list_head(&spa->spa_state_dirty_list)) != NULL) {
	vdev_state_clean(vd);
	vdev_config_dirty(vd);
	}
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	spa_config_enter(spa, SCL_CONFIG \| SCL_STATE, FTAG, RW_READER);
	}
	spa_config_exit(spa, SCL_STATE, FTAG);

	dsl_pool_t *dp = spa->spa_dsl_pool;
	dmu_tx_t *tx = dmu_tx_create_assigned(dp, txg);

	spa->spa_sync_starttime = gethrtime();
	taskq_cancel_id(system_delay_taskq, spa->spa_deadman_tqid);
	spa->spa_deadman_tqid = taskq_dispatch_delay(system_delay_taskq,
	spa_deadman, spa, TQ_SLEEP, ddi_get_lbolt() +
	NSEC_TO_TICK(spa->spa_deadman_synctime));

	/*
	* If we are upgrading to SPA_VERSION_RAIDZ_DEFLATE this txg,
	* set spa_deflate if we have no raid-z vdevs.
	*/
	if (spa->spa_ubsync.ub_version < SPA_VERSION_RAIDZ_DEFLATE &&
	spa->spa_uberblock.ub_version >= SPA_VERSION_RAIDZ_DEFLATE) {
	vdev_t *rvd = spa->spa_root_vdev;

	int i;
	for (i = 0; i < rvd->vdev_children; i++) {
	vd = rvd->vdev_child[i];
	if (vd->vdev_deflate_ratio != SPA_MINBLOCKSIZE)
	break;
	}
	if (i == rvd->vdev_children) {
	spa->spa_deflate = TRUE;
	VERIFY0(zap_add(spa->spa_meta_objset,
	DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_DEFLATE,
	sizeof (uint64_t), 1, &spa->spa_deflate, tx));
	}
	}

	spa_sync_adjust_vdev_max_queue_depth(spa);

	spa_sync_condense_indirect(spa, tx);

	spa_sync_iterate_to_convergence(spa, tx);

	#ifdef ZFS_DEBUG
	if (!list_is_empty(&spa->spa_config_dirty_list)) {
	/*
	* Make sure that the number of ZAPs for all the vdevs matches
	* the number of ZAPs in the per-vdev ZAP list. This only gets
	* called if the config is dirty; otherwise there may be
	* outstanding AVZ operations that weren't completed in
	* spa_sync_config_object.
	*/
	uint64_t all_vdev_zap_entry_count;
	ASSERT0(zap_count(spa->spa_meta_objset,
	spa->spa_all_vdev_zaps, &all_vdev_zap_entry_count));
	ASSERT3U(vdev_count_verify_zaps(spa->spa_root_vdev), ==,
	all_vdev_zap_entry_count);
	}
	#endif

	if (spa->spa_vdev_removal != NULL) {
	ASSERT0(spa->spa_vdev_removal->svr_bytes_done[txg & TXG_MASK]);
	}

	spa_sync_rewrite_vdev_config(spa, tx);
	dmu_tx_commit(tx);

	taskq_cancel_id(system_delay_taskq, spa->spa_deadman_tqid);
	spa->spa_deadman_tqid = 0;

	/*
	* Clear the dirty config list.
	*/
	while ((vd = list_head(&spa->spa_config_dirty_list)) != NULL)
	vdev_config_clean(vd);

	/*
	* Now that the new config has synced transactionally,
	* let it become visible to the config cache.
	*/
	if (spa->spa_config_syncing != NULL) {
	spa_config_set(spa, spa->spa_config_syncing);
	spa->spa_config_txg = txg;
	spa->spa_config_syncing = NULL;
	}

	dsl_pool_sync_done(dp, txg);

	for (int i = 0; i < spa->spa_alloc_count; i++) {
	mutex_enter(&spa->spa_allocs[i].spaa_lock);
	VERIFY0(avl_numnodes(&spa->spa_allocs[i].spaa_tree));
	mutex_exit(&spa->spa_allocs[i].spaa_lock);
	}

	/*
	* Update usable space statistics.
	*/
	while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, TXG_CLEAN(txg)))
	!= NULL)
	vdev_sync_done(vd, txg);

	metaslab_class_evict_old(spa->spa_normal_class, txg);
	metaslab_class_evict_old(spa->spa_log_class, txg);

	spa_sync_close_syncing_log_sm(spa);

	spa_update_dspace(spa);

	/*
	* It had better be the case that we didn't dirty anything
	* since vdev_config_sync().
	*/
	ASSERT(txg_list_empty(&dp->dp_dirty_datasets, txg));
	ASSERT(txg_list_empty(&dp->dp_dirty_dirs, txg));
	ASSERT(txg_list_empty(&spa->spa_vdev_txg_list, txg));

	while (zfs_pause_spa_sync)
	delay(1);

	spa->spa_sync_pass = 0;

	/*
	* Update the last synced uberblock here. We want to do this at
	* the end of spa_sync() so that consumers of spa_last_synced_txg()
	* will be guaranteed that all the processing associated with
	* that txg has been completed.
	*/
	spa->spa_ubsync = spa->spa_uberblock;
	spa_config_exit(spa, SCL_CONFIG, FTAG);

	spa_handle_ignored_writes(spa);

	/*
	* If any async tasks have been requested, kick them off.
	*/
	spa_async_dispatch(spa);
	}

	/*
	* Sync all pools. We don't want to hold the namespace lock across these
	* operations, so we take a reference on the spa_t and drop the lock during the
	* sync.
	*/
	void
	spa_sync_allpools(void)
	{
	spa_t *spa = NULL;
	mutex_enter(&spa_namespace_lock);
	while ((spa = spa_next(spa)) != NULL) {
	if (spa_state(spa) != POOL_STATE_ACTIVE \|\|
	!spa_writeable(spa) \|\| spa_suspended(spa))
	continue;
	spa_open_ref(spa, FTAG);
	mutex_exit(&spa_namespace_lock);
	txg_wait_synced(spa_get_dsl(spa), 0);
	mutex_enter(&spa_namespace_lock);
	spa_close(spa, FTAG);
	}
	mutex_exit(&spa_namespace_lock);
	}

	/*
	* ==========================================================================
	* Miscellaneous routines
	* ==========================================================================
	*/

	/*
	* Remove all pools in the system.
	*/
	void
	spa_evict_all(void)
	{
	spa_t *spa;

	/*
	* Remove all cached state. All pools should be closed now,
	* so every spa in the AVL tree should be unreferenced.
	*/
	mutex_enter(&spa_namespace_lock);
	while ((spa = spa_next(NULL)) != NULL) {
	/*
	* Stop async tasks. The async thread may need to detach
	* a device that's been replaced, which requires grabbing
	* spa_namespace_lock, so we must drop it here.
	*/
	spa_open_ref(spa, FTAG);
	mutex_exit(&spa_namespace_lock);
	spa_async_suspend(spa);
	mutex_enter(&spa_namespace_lock);
	spa_close(spa, FTAG);

	if (spa->spa_state != POOL_STATE_UNINITIALIZED) {
	spa_unload(spa);
	spa_deactivate(spa);
	}
	spa_remove(spa);
	}
	mutex_exit(&spa_namespace_lock);
	}

	vdev_t *
	spa_lookup_by_guid(spa_t *spa, uint64_t guid, boolean_t aux)
	{
	vdev_t *vd;
	int i;

	if ((vd = vdev_lookup_by_guid(spa->spa_root_vdev, guid)) != NULL)
	return (vd);

	if (aux) {
	for (i = 0; i < spa->spa_l2cache.sav_count; i++) {
	vd = spa->spa_l2cache.sav_vdevs[i];
	if (vd->vdev_guid == guid)
	return (vd);
	}

	for (i = 0; i < spa->spa_spares.sav_count; i++) {
	vd = spa->spa_spares.sav_vdevs[i];
	if (vd->vdev_guid == guid)
	return (vd);
	}
	}

	return (NULL);
	}

	void
	spa_upgrade(spa_t *spa, uint64_t version)
	{
	ASSERT(spa_writeable(spa));

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);

	/*
	* This should only be called for a non-faulted pool, and since a
	* future version would result in an unopenable pool, this shouldn't be
	* possible.
	*/
	ASSERT(SPA_VERSION_IS_SUPPORTED(spa->spa_uberblock.ub_version));
	ASSERT3U(version, >=, spa->spa_uberblock.ub_version);

	spa->spa_uberblock.ub_version = version;
	vdev_config_dirty(spa->spa_root_vdev);

	spa_config_exit(spa, SCL_ALL, FTAG);

	txg_wait_synced(spa_get_dsl(spa), 0);
	}

	static boolean_t
	spa_has_aux_vdev(spa_t spa, uint64_t guid, spa_aux_vdev_t sav)
	{
	int i;
	uint64_t vdev_guid;

	for (i = 0; i < sav->sav_count; i++)
	if (sav->sav_vdevs[i]->vdev_guid == guid)
	return (B_TRUE);

	for (i = 0; i < sav->sav_npending; i++) {
	if (nvlist_lookup_uint64(sav->sav_pending[i], ZPOOL_CONFIG_GUID,
	&vdev_guid) == 0 && vdev_guid == guid)
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	boolean_t
	spa_has_l2cache(spa_t *spa, uint64_t guid)
	{
	return (spa_has_aux_vdev(spa, guid, &spa->spa_l2cache));
	}

	boolean_t
	spa_has_spare(spa_t *spa, uint64_t guid)
	{
	return (spa_has_aux_vdev(spa, guid, &spa->spa_spares));
	}

	/*
	* Check if a pool has an active shared spare device.
	* Note: reference count of an active spare is 2, as a spare and as a replace
	*/
	static boolean_t
	spa_has_active_shared_spare(spa_t *spa)
	{
	int i, refcnt;
	uint64_t pool;
	spa_aux_vdev_t *sav = &spa->spa_spares;

	for (i = 0; i < sav->sav_count; i++) {
	if (spa_spare_exists(sav->sav_vdevs[i]->vdev_guid, &pool,
	&refcnt) && pool != 0ULL && pool == spa_guid(spa) &&
	refcnt > 2)
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	uint64_t
	spa_total_metaslabs(spa_t *spa)
	{
	vdev_t *rvd = spa->spa_root_vdev;

	uint64_t m = 0;
	for (uint64_t c = 0; c < rvd->vdev_children; c++) {
	vdev_t *vd = rvd->vdev_child[c];
	if (!vdev_is_concrete(vd))
	continue;
	m += vd->vdev_ms_count;
	}
	return (m);
	}

	/*
	* Notify any waiting threads that some activity has switched from being in-
	* progress to not-in-progress so that the thread can wake up and determine
	* whether it is finished waiting.
	*/
	void
	spa_notify_waiters(spa_t *spa)
	{
	/*
	* Acquiring spa_activities_lock here prevents the cv_broadcast from
	* happening between the waiting thread's check and cv_wait.
	*/
	mutex_enter(&spa->spa_activities_lock);
	cv_broadcast(&spa->spa_activities_cv);
	mutex_exit(&spa->spa_activities_lock);
	}

	/*
	* Notify any waiting threads that the pool is exporting, and then block until
	* they are finished using the spa_t.
	*/
	void
	spa_wake_waiters(spa_t *spa)
	{
	mutex_enter(&spa->spa_activities_lock);
	spa->spa_waiters_cancel = B_TRUE;
	cv_broadcast(&spa->spa_activities_cv);
	while (spa->spa_waiters != 0)
	cv_wait(&spa->spa_waiters_cv, &spa->spa_activities_lock);
	spa->spa_waiters_cancel = B_FALSE;
	mutex_exit(&spa->spa_activities_lock);
	}

	/* Whether the vdev or any of its descendants are being initialized/trimmed. */
	static boolean_t
	spa_vdev_activity_in_progress_impl(vdev_t *vd, zpool_wait_activity_t activity)
	{
	spa_t *spa = vd->vdev_spa;

	ASSERT(spa_config_held(spa, SCL_CONFIG \| SCL_STATE, RW_READER));
	ASSERT(MUTEX_HELD(&spa->spa_activities_lock));
	ASSERT(activity == ZPOOL_WAIT_INITIALIZE \|\|
	activity == ZPOOL_WAIT_TRIM);

	kmutex_t *lock = activity == ZPOOL_WAIT_INITIALIZE ?
	&vd->vdev_initialize_lock : &vd->vdev_trim_lock;

	mutex_exit(&spa->spa_activities_lock);
	mutex_enter(lock);
	mutex_enter(&spa->spa_activities_lock);

	boolean_t in_progress = (activity == ZPOOL_WAIT_INITIALIZE) ?
	(vd->vdev_initialize_state == VDEV_INITIALIZE_ACTIVE) :
	(vd->vdev_trim_state == VDEV_TRIM_ACTIVE);
	mutex_exit(lock);

	if (in_progress)
	return (B_TRUE);

	for (int i = 0; i < vd->vdev_children; i++) {
	if (spa_vdev_activity_in_progress_impl(vd->vdev_child[i],
	activity))
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	/*
	* If use_guid is true, this checks whether the vdev specified by guid is
	* being initialized/trimmed. Otherwise, it checks whether any vdev in the pool
	* is being initialized/trimmed. The caller must hold the config lock and
	* spa_activities_lock.
	*/
	static int
	spa_vdev_activity_in_progress(spa_t *spa, boolean_t use_guid, uint64_t guid,
	zpool_wait_activity_t activity, boolean_t *in_progress)
	{
	mutex_exit(&spa->spa_activities_lock);
	spa_config_enter(spa, SCL_CONFIG \| SCL_STATE, FTAG, RW_READER);
	mutex_enter(&spa->spa_activities_lock);

	vdev_t *vd;
	if (use_guid) {
	vd = spa_lookup_by_guid(spa, guid, B_FALSE);
	if (vd == NULL \|\| !vd->vdev_ops->vdev_op_leaf) {
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	return (EINVAL);
	}
	} else {
	vd = spa->spa_root_vdev;
	}

	*in_progress = spa_vdev_activity_in_progress_impl(vd, activity);

	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	return (0);
	}

	/*
	* Locking for waiting threads
	* ---------------------------
	*
	* Waiting threads need a way to check whether a given activity is in progress,
	* and then, if it is, wait for it to complete. Each activity will have some
	* in-memory representation of the relevant on-disk state which can be used to
	* determine whether or not the activity is in progress. The in-memory state and
	* the locking used to protect it will be different for each activity, and may
	* not be suitable for use with a cvar (e.g., some state is protected by the
	* config lock). To allow waiting threads to wait without any races, another
	* lock, spa_activities_lock, is used.
	*
	* When the state is checked, both the activity-specific lock (if there is one)
	* and spa_activities_lock are held. In some cases, the activity-specific lock
	* is acquired explicitly (e.g. the config lock). In others, the locking is
	* internal to some check (e.g. bpobj_is_empty). After checking, the waiting
	* thread releases the activity-specific lock and, if the activity is in
	* progress, then cv_waits using spa_activities_lock.
	*
	* The waiting thread is woken when another thread, one completing some
	* activity, updates the state of the activity and then calls
	* spa_notify_waiters, which will cv_broadcast. This 'completing' thread only
	* needs to hold its activity-specific lock when updating the state, and this
	* lock can (but doesn't have to) be dropped before calling spa_notify_waiters.
	*
	* Because spa_notify_waiters acquires spa_activities_lock before broadcasting,
	* and because it is held when the waiting thread checks the state of the
	* activity, it can never be the case that the completing thread both updates
	* the activity state and cv_broadcasts in between the waiting thread's check
	* and cv_wait. Thus, a waiting thread can never miss a wakeup.
	*
	* In order to prevent deadlock, when the waiting thread does its check, in some
	* cases it will temporarily drop spa_activities_lock in order to acquire the
	* activity-specific lock. The order in which spa_activities_lock and the
	* activity specific lock are acquired in the waiting thread is determined by
	* the order in which they are acquired in the completing thread; if the
	* completing thread calls spa_notify_waiters with the activity-specific lock
	* held, then the waiting thread must also acquire the activity-specific lock
	* first.
	*/

	static int
	spa_activity_in_progress(spa_t *spa, zpool_wait_activity_t activity,
	boolean_t use_tag, uint64_t tag, boolean_t *in_progress)
	{
	int error = 0;

	ASSERT(MUTEX_HELD(&spa->spa_activities_lock));

	switch (activity) {
	case ZPOOL_WAIT_CKPT_DISCARD:
	*in_progress =
	(spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT) &&
	zap_contains(spa_meta_objset(spa),
	DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_ZPOOL_CHECKPOINT) ==
	ENOENT);
	break;
	case ZPOOL_WAIT_FREE:
	*in_progress = ((spa_version(spa) >= SPA_VERSION_DEADLISTS &&
	!bpobj_is_empty(&spa->spa_dsl_pool->dp_free_bpobj)) \|\|
	spa_feature_is_active(spa, SPA_FEATURE_ASYNC_DESTROY) \|\|
	spa_livelist_delete_check(spa));
	break;
	case ZPOOL_WAIT_INITIALIZE:
	case ZPOOL_WAIT_TRIM:
	error = spa_vdev_activity_in_progress(spa, use_tag, tag,
	activity, in_progress);
	break;
	case ZPOOL_WAIT_REPLACE:
	mutex_exit(&spa->spa_activities_lock);
	spa_config_enter(spa, SCL_CONFIG \| SCL_STATE, FTAG, RW_READER);
	mutex_enter(&spa->spa_activities_lock);

	*in_progress = vdev_replace_in_progress(spa->spa_root_vdev);
	spa_config_exit(spa, SCL_CONFIG \| SCL_STATE, FTAG);
	break;
	case ZPOOL_WAIT_REMOVE:
	*in_progress = (spa->spa_removing_phys.sr_state ==
	DSS_SCANNING);
	break;
	case ZPOOL_WAIT_RESILVER:
	if ((*in_progress = vdev_rebuild_active(spa->spa_root_vdev)))
	break;
	fallthrough;
	case ZPOOL_WAIT_SCRUB:
	{
	boolean_t scanning, paused, is_scrub;
	dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan;

	is_scrub = (scn->scn_phys.scn_func == POOL_SCAN_SCRUB);
	scanning = (scn->scn_phys.scn_state == DSS_SCANNING);
	paused = dsl_scan_is_paused_scrub(scn);
	*in_progress = (scanning && !paused &&
	is_scrub == (activity == ZPOOL_WAIT_SCRUB));
	break;
	}
	default:
	panic("unrecognized value for activity %d", activity);
	}

	return (error);
	}

	static int
	spa_wait_common(const char *pool, zpool_wait_activity_t activity,
	boolean_t use_tag, uint64_t tag, boolean_t *waited)
	{
	/*
	* The tag is used to distinguish between instances of an activity.
	* 'initialize' and 'trim' are the only activities that we use this for.
	* The other activities can only have a single instance in progress in a
	* pool at one time, making the tag unnecessary.
	*
	* There can be multiple devices being replaced at once, but since they
	* all finish once resilvering finishes, we don't bother keeping track
	* of them individually, we just wait for them all to finish.
	*/
	if (use_tag && activity != ZPOOL_WAIT_INITIALIZE &&
	activity != ZPOOL_WAIT_TRIM)
	return (EINVAL);

	if (activity < 0 \|\| activity >= ZPOOL_WAIT_NUM_ACTIVITIES)
	return (EINVAL);

	spa_t *spa;
	int error = spa_open(pool, &spa, FTAG);
	if (error != 0)
	return (error);

	/*
	* Increment the spa's waiter count so that we can call spa_close and
	* still ensure that the spa_t doesn't get freed before this thread is
	* finished with it when the pool is exported. We want to call spa_close
	* before we start waiting because otherwise the additional ref would
	* prevent the pool from being exported or destroyed throughout the
	* potentially long wait.
	*/
	mutex_enter(&spa->spa_activities_lock);
	spa->spa_waiters++;
	spa_close(spa, FTAG);

	*waited = B_FALSE;
	for (;;) {
	boolean_t in_progress;
	error = spa_activity_in_progress(spa, activity, use_tag, tag,
	&in_progress);

	if (error \|\| !in_progress \|\| spa->spa_waiters_cancel)
	break;

	*waited = B_TRUE;

	if (cv_wait_sig(&spa->spa_activities_cv,
	&spa->spa_activities_lock) == 0) {
	error = EINTR;
	break;
	}
	}

	spa->spa_waiters--;
	cv_signal(&spa->spa_waiters_cv);
	mutex_exit(&spa->spa_activities_lock);

	return (error);
	}

	/*
	* Wait for a particular instance of the specified activity to complete, where
	* the instance is identified by 'tag'
	*/
	int
	spa_wait_tag(const char *pool, zpool_wait_activity_t activity, uint64_t tag,
	boolean_t *waited)
	{
	return (spa_wait_common(pool, activity, B_TRUE, tag, waited));
	}

	/*
	* Wait for all instances of the specified activity complete
	*/
	int
	spa_wait(const char pool, zpool_wait_activity_t activity, boolean_t waited)
	{

	return (spa_wait_common(pool, activity, B_FALSE, 0, waited));
	}

	sysevent_t *
	spa_event_create(spa_t spa, vdev_t vd, nvlist_t hist_nvl, const char name)
	{
	sysevent_t *ev = NULL;
	#ifdef _KERNEL
	nvlist_t *resource;

	resource = zfs_event_create(spa, vd, FM_SYSEVENT_CLASS, name, hist_nvl);
	if (resource) {
	ev = kmem_alloc(sizeof (sysevent_t), KM_SLEEP);
	ev->resource = resource;
	}
	#endif
	return (ev);
	}

	void
	spa_event_post(sysevent_t *ev)
	{
	#ifdef _KERNEL
	if (ev) {
	zfs_zevent_post(ev->resource, NULL, zfs_zevent_post_cb);
	kmem_free(ev, sizeof (*ev));
	}
	#endif
	}

	/*
	* Post a zevent corresponding to the given sysevent. The 'name' must be one
	* of the event definitions in sys/sysevent/eventdefs.h. The payload will be
	* filled in from the spa and (optionally) the vdev. This doesn't do anything
	* in the userland libzpool, as we don't want consumers to misinterpret ztest
	* or zdb as real changes.
	*/
	void
	spa_event_notify(spa_t spa, vdev_t vd, nvlist_t hist_nvl, const char name)
	{
	spa_event_post(spa_event_create(spa, vd, hist_nvl, name));
	}

	/* state manipulation functions */
	EXPORT_SYMBOL(spa_open);
	EXPORT_SYMBOL(spa_open_rewind);
	EXPORT_SYMBOL(spa_get_stats);
	EXPORT_SYMBOL(spa_create);
	EXPORT_SYMBOL(spa_import);
	EXPORT_SYMBOL(spa_tryimport);
	EXPORT_SYMBOL(spa_destroy);
	EXPORT_SYMBOL(spa_export);
	EXPORT_SYMBOL(spa_reset);
	EXPORT_SYMBOL(spa_async_request);
	EXPORT_SYMBOL(spa_async_suspend);
	EXPORT_SYMBOL(spa_async_resume);
	EXPORT_SYMBOL(spa_inject_addref);
	EXPORT_SYMBOL(spa_inject_delref);
	EXPORT_SYMBOL(spa_scan_stat_init);
	EXPORT_SYMBOL(spa_scan_get_stats);

	/* device manipulation */
	EXPORT_SYMBOL(spa_vdev_add);
	EXPORT_SYMBOL(spa_vdev_attach);
	EXPORT_SYMBOL(spa_vdev_detach);
	EXPORT_SYMBOL(spa_vdev_setpath);
	EXPORT_SYMBOL(spa_vdev_setfru);
	EXPORT_SYMBOL(spa_vdev_split_mirror);

	/* spare statech is global across all pools) */
	EXPORT_SYMBOL(spa_spare_add);
	EXPORT_SYMBOL(spa_spare_remove);
	EXPORT_SYMBOL(spa_spare_exists);
	EXPORT_SYMBOL(spa_spare_activate);

	/* L2ARC statech is global across all pools) */
	EXPORT_SYMBOL(spa_l2cache_add);
	EXPORT_SYMBOL(spa_l2cache_remove);
	EXPORT_SYMBOL(spa_l2cache_exists);
	EXPORT_SYMBOL(spa_l2cache_activate);
	EXPORT_SYMBOL(spa_l2cache_drop);

	/* scanning */
	EXPORT_SYMBOL(spa_scan);
	EXPORT_SYMBOL(spa_scan_stop);

	/* spa syncing */
	EXPORT_SYMBOL(spa_sync); /* only for DMU use */
	EXPORT_SYMBOL(spa_sync_allpools);

	/* properties */
	EXPORT_SYMBOL(spa_prop_set);
	EXPORT_SYMBOL(spa_prop_get);
	EXPORT_SYMBOL(spa_prop_clear_bootfs);

	/* asynchronous event notification */
	EXPORT_SYMBOL(spa_event_notify);

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM(zfs_spa, spa_, load_verify_shift, INT, ZMOD_RW,
	"log2 fraction of arc that can be used by inflight I/Os when "
	"verifying pool during import");

	ZFS_MODULE_PARAM(zfs_spa, spa_, load_verify_metadata, INT, ZMOD_RW,
	"Set to traverse metadata on pool import");

	ZFS_MODULE_PARAM(zfs_spa, spa_, load_verify_data, INT, ZMOD_RW,
	"Set to traverse data on pool import");

	ZFS_MODULE_PARAM(zfs_spa, spa_, load_print_vdev_tree, INT, ZMOD_RW,
	"Print vdev tree to zfs_dbgmsg during pool import");

	ZFS_MODULE_PARAM(zfs_zio, zio_, taskq_batch_pct, UINT, ZMOD_RD,
	"Percentage of CPUs to run an IO worker thread");

	ZFS_MODULE_PARAM(zfs_zio, zio_, taskq_batch_tpq, UINT, ZMOD_RD,
	"Number of threads per IO worker taskqueue");

	ZFS_MODULE_PARAM(zfs, zfs_, max_missing_tvds, ULONG, ZMOD_RW,
	"Allow importing pool with up to this number of missing top-level "
	"vdevs (in read-only mode)");

	ZFS_MODULE_PARAM(zfs_livelist_condense, zfs_livelist_condense_, zthr_pause, INT, ZMOD_RW,
	"Set the livelist condense zthr to pause");

	ZFS_MODULE_PARAM(zfs_livelist_condense, zfs_livelist_condense_, sync_pause, INT, ZMOD_RW,
	"Set the livelist condense synctask to pause");

	ZFS_MODULE_PARAM(zfs_livelist_condense, zfs_livelist_condense_, sync_cancel, INT, ZMOD_RW,
	"Whether livelist condensing was canceled in the synctask");

	ZFS_MODULE_PARAM(zfs_livelist_condense, zfs_livelist_condense_, zthr_cancel, INT, ZMOD_RW,
	"Whether livelist condensing was canceled in the zthr function");

	ZFS_MODULE_PARAM(zfs_livelist_condense, zfs_livelist_condense_, new_alloc, INT, ZMOD_RW,
	"Whether extra ALLOC blkptrs were added to a livelist entry while it "
	"was being condensed");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/zfs/spa_misc.c b/sys/contrib/openzfs/module/zfs/spa_misc.c
	index 1ecd2294dba0..3f74631983c4 100644
	--- a/sys/contrib/openzfs/module/zfs/spa_misc.c
	+++ b/sys/contrib/openzfs/module/zfs/spa_misc.c
	@@ -1,2963 +1,2955 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2019 by Delphix. All rights reserved.
	* Copyright 2015 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
	* Copyright 2013 Saso Kiselkov. All rights reserved.
	* Copyright (c) 2017 Datto Inc.
	* Copyright (c) 2017, Intel Corporation.
	* Copyright (c) 2019, loli10K <ezomori.nozomu@gmail.com>. All rights reserved.
	*/

	#include <sys/zfs_context.h>
	#include <sys/spa_impl.h>
	#include <sys/zio.h>
	#include <sys/zio_checksum.h>
	#include <sys/zio_compress.h>
	#include <sys/dmu.h>
	#include <sys/dmu_tx.h>
	#include <sys/zap.h>
	#include <sys/zil.h>
	#include <sys/vdev_impl.h>
	#include <sys/vdev_initialize.h>
	#include <sys/vdev_trim.h>
	#include <sys/vdev_file.h>
	#include <sys/vdev_raidz.h>
	#include <sys/metaslab.h>
	#include <sys/uberblock_impl.h>
	#include <sys/txg.h>
	#include <sys/avl.h>
	#include <sys/unique.h>
	#include <sys/dsl_pool.h>
	#include <sys/dsl_dir.h>
	#include <sys/dsl_prop.h>
	#include <sys/fm/util.h>
	#include <sys/dsl_scan.h>
	#include <sys/fs/zfs.h>
	#include <sys/metaslab_impl.h>
	#include <sys/arc.h>
	#include <sys/ddt.h>
	#include <sys/kstat.h>
	#include "zfs_prop.h"
	#include <sys/btree.h>
	#include <sys/zfeature.h>
	#include <sys/qat.h>
	#include <sys/zstd/zstd.h>

	/*
	* SPA locking
	*
	* There are three basic locks for managing spa_t structures:
	*
	* spa_namespace_lock (global mutex)
	*
	* This lock must be acquired to do any of the following:
	*
	* - Lookup a spa_t by name
	* - Add or remove a spa_t from the namespace
	* - Increase spa_refcount from non-zero
	* - Check if spa_refcount is zero
	* - Rename a spa_t
	* - add/remove/attach/detach devices
	* - Held for the duration of create/destroy/import/export
	*
	* It does not need to handle recursion. A create or destroy may
	* reference objects (files or zvols) in other pools, but by
	* definition they must have an existing reference, and will never need
	* to lookup a spa_t by name.
	*
	* spa_refcount (per-spa zfs_refcount_t protected by mutex)
	*
	* This reference count keep track of any active users of the spa_t. The
	* spa_t cannot be destroyed or freed while this is non-zero. Internally,
	* the refcount is never really 'zero' - opening a pool implicitly keeps
	* some references in the DMU. Internally we check against spa_minref, but
	* present the image of a zero/non-zero value to consumers.
	*
	* spa_config_lock[] (per-spa array of rwlocks)
	*
	* This protects the spa_t from config changes, and must be held in
	* the following circumstances:
	*
	* - RW_READER to perform I/O to the spa
	* - RW_WRITER to change the vdev config
	*
	* The locking order is fairly straightforward:
	*
	* spa_namespace_lock -> spa_refcount
	*
	* The namespace lock must be acquired to increase the refcount from 0
	* or to check if it is zero.
	*
	* spa_refcount -> spa_config_lock[]
	*
	* There must be at least one valid reference on the spa_t to acquire
	* the config lock.
	*
	* spa_namespace_lock -> spa_config_lock[]
	*
	* The namespace lock must always be taken before the config lock.
	*
	*
	* The spa_namespace_lock can be acquired directly and is globally visible.
	*
	* The namespace is manipulated using the following functions, all of which
	* require the spa_namespace_lock to be held.
	*
	* spa_lookup() Lookup a spa_t by name.
	*
	* spa_add() Create a new spa_t in the namespace.
	*
	* spa_remove() Remove a spa_t from the namespace. This also
	* frees up any memory associated with the spa_t.
	*
	* spa_next() Returns the next spa_t in the system, or the
	* first if NULL is passed.
	*
	* spa_evict_all() Shutdown and remove all spa_t structures in
	* the system.
	*
	* spa_guid_exists() Determine whether a pool/device guid exists.
	*
	* The spa_refcount is manipulated using the following functions:
	*
	* spa_open_ref() Adds a reference to the given spa_t. Must be
	* called with spa_namespace_lock held if the
	* refcount is currently zero.
	*
	* spa_close() Remove a reference from the spa_t. This will
	* not free the spa_t or remove it from the
	* namespace. No locking is required.
	*
	* spa_refcount_zero() Returns true if the refcount is currently
	* zero. Must be called with spa_namespace_lock
	* held.
	*
	* The spa_config_lock[] is an array of rwlocks, ordered as follows:
	* SCL_CONFIG > SCL_STATE > SCL_ALLOC > SCL_ZIO > SCL_FREE > SCL_VDEV.
	* spa_config_lock[] is manipulated with spa_config_{enter,exit,held}().
	*
	* To read the configuration, it suffices to hold one of these locks as reader.
	* To modify the configuration, you must hold all locks as writer. To modify
	* vdev state without altering the vdev tree's topology (e.g. online/offline),
	* you must hold SCL_STATE and SCL_ZIO as writer.
	*
	* We use these distinct config locks to avoid recursive lock entry.
	* For example, spa_sync() (which holds SCL_CONFIG as reader) induces
	* block allocations (SCL_ALLOC), which may require reading space maps
	* from disk (dmu_read() -> zio_read() -> SCL_ZIO).
	*
	* The spa config locks cannot be normal rwlocks because we need the
	* ability to hand off ownership. For example, SCL_ZIO is acquired
	* by the issuing thread and later released by an interrupt thread.
	* They do, however, obey the usual write-wanted semantics to prevent
	* writer (i.e. system administrator) starvation.
	*
	* The lock acquisition rules are as follows:
	*
	* SCL_CONFIG
	* Protects changes to the vdev tree topology, such as vdev
	* add/remove/attach/detach. Protects the dirty config list
	* (spa_config_dirty_list) and the set of spares and l2arc devices.
	*
	* SCL_STATE
	* Protects changes to pool state and vdev state, such as vdev
	* online/offline/fault/degrade/clear. Protects the dirty state list
	* (spa_state_dirty_list) and global pool state (spa_state).
	*
	* SCL_ALLOC
	* Protects changes to metaslab groups and classes.
	* Held as reader by metaslab_alloc() and metaslab_claim().
	*
	* SCL_ZIO
	* Held by bp-level zios (those which have no io_vd upon entry)
	* to prevent changes to the vdev tree. The bp-level zio implicitly
	* protects all of its vdev child zios, which do not hold SCL_ZIO.
	*
	* SCL_FREE
	* Protects changes to metaslab groups and classes.
	* Held as reader by metaslab_free(). SCL_FREE is distinct from
	* SCL_ALLOC, and lower than SCL_ZIO, so that we can safely free
	* blocks in zio_done() while another i/o that holds either
	* SCL_ALLOC or SCL_ZIO is waiting for this i/o to complete.
	*
	* SCL_VDEV
	* Held as reader to prevent changes to the vdev tree during trivial
	* inquiries such as bp_get_dsize(). SCL_VDEV is distinct from the
	* other locks, and lower than all of them, to ensure that it's safe
	* to acquire regardless of caller context.
	*
	* In addition, the following rules apply:
	*
	* (a) spa_props_lock protects pool properties, spa_config and spa_config_list.
	* The lock ordering is SCL_CONFIG > spa_props_lock.
	*
	* (b) I/O operations on leaf vdevs. For any zio operation that takes
	* an explicit vdev_t argument -- such as zio_ioctl(), zio_read_phys(),
	* or zio_write_phys() -- the caller must ensure that the config cannot
	* cannot change in the interim, and that the vdev cannot be reopened.
	* SCL_STATE as reader suffices for both.
	*
	* The vdev configuration is protected by spa_vdev_enter() / spa_vdev_exit().
	*
	* spa_vdev_enter() Acquire the namespace lock and the config lock
	* for writing.
	*
	* spa_vdev_exit() Release the config lock, wait for all I/O
	* to complete, sync the updated configs to the
	* cache, and release the namespace lock.
	*
	* vdev state is protected by spa_vdev_state_enter() / spa_vdev_state_exit().
	* Like spa_vdev_enter/exit, these are convenience wrappers -- the actual
	* locking is, always, based on spa_namespace_lock and spa_config_lock[].
	*/

	static avl_tree_t spa_namespace_avl;
	kmutex_t spa_namespace_lock;
	static kcondvar_t spa_namespace_cv;
	int spa_max_replication_override = SPA_DVAS_PER_BP;

	static kmutex_t spa_spare_lock;
	static avl_tree_t spa_spare_avl;
	static kmutex_t spa_l2cache_lock;
	static avl_tree_t spa_l2cache_avl;

	kmem_cache_t *spa_buffer_pool;
	spa_mode_t spa_mode_global = SPA_MODE_UNINIT;

	#ifdef ZFS_DEBUG
	/*
	* Everything except dprintf, set_error, spa, and indirect_remap is on
	* by default in debug builds.
	*/
	int zfs_flags = ~(ZFS_DEBUG_DPRINTF \| ZFS_DEBUG_SET_ERROR \|
	ZFS_DEBUG_INDIRECT_REMAP);
	#else
	int zfs_flags = 0;
	#endif

	/*
	* zfs_recover can be set to nonzero to attempt to recover from
	* otherwise-fatal errors, typically caused by on-disk corruption. When
	* set, calls to zfs_panic_recover() will turn into warning messages.
	* This should only be used as a last resort, as it typically results
	* in leaked space, or worse.
	*/
	int zfs_recover = B_FALSE;

	/*
	* If destroy encounters an EIO while reading metadata (e.g. indirect
	* blocks), space referenced by the missing metadata can not be freed.
	* Normally this causes the background destroy to become "stalled", as
	* it is unable to make forward progress. While in this stalled state,
	* all remaining space to free from the error-encountering filesystem is
	* "temporarily leaked". Set this flag to cause it to ignore the EIO,
	* permanently leak the space from indirect blocks that can not be read,
	* and continue to free everything else that it can.
	*
	* The default, "stalling" behavior is useful if the storage partially
	* fails (i.e. some but not all i/os fail), and then later recovers. In
	* this case, we will be able to continue pool operations while it is
	* partially failed, and when it recovers, we can continue to free the
	* space, with no leaks. However, note that this case is actually
	* fairly rare.
	*
	* Typically pools either (a) fail completely (but perhaps temporarily,
	* e.g. a top-level vdev going offline), or (b) have localized,
	* permanent errors (e.g. disk returns the wrong data due to bit flip or
	* firmware bug). In case (a), this setting does not matter because the
	* pool will be suspended and the sync thread will not be able to make
	* forward progress regardless. In case (b), because the error is
	* permanent, the best we can do is leak the minimum amount of space,
	* which is what setting this flag will do. Therefore, it is reasonable
	* for this flag to normally be set, but we chose the more conservative
	* approach of not setting it, so that there is no possibility of
	* leaking space in the "partial temporary" failure case.
	*/
	int zfs_free_leak_on_eio = B_FALSE;

	/*
	* Expiration time in milliseconds. This value has two meanings. First it is
	* used to determine when the spa_deadman() logic should fire. By default the
	* spa_deadman() will fire if spa_sync() has not completed in 600 seconds.
	* Secondly, the value determines if an I/O is considered "hung". Any I/O that
	* has not completed in zfs_deadman_synctime_ms is considered "hung" resulting
	* in one of three behaviors controlled by zfs_deadman_failmode.
	*/
	unsigned long zfs_deadman_synctime_ms = 600000UL;

	/*
	* This value controls the maximum amount of time zio_wait() will block for an
	* outstanding IO. By default this is 300 seconds at which point the "hung"
	* behavior will be applied as described for zfs_deadman_synctime_ms.
	*/
	unsigned long zfs_deadman_ziotime_ms = 300000UL;

	/*
	* Check time in milliseconds. This defines the frequency at which we check
	* for hung I/O.
	*/
	unsigned long zfs_deadman_checktime_ms = 60000UL;

	/*
	* By default the deadman is enabled.
	*/
	int zfs_deadman_enabled = 1;

	/*
	* Controls the behavior of the deadman when it detects a "hung" I/O.
	* Valid values are zfs_deadman_failmode=<wait\|continue\|panic>.
	*
	* wait - Wait for the "hung" I/O (default)
	* continue - Attempt to recover from a "hung" I/O
	* panic - Panic the system
	*/
	char *zfs_deadman_failmode = "wait";

	/*
	* The worst case is single-sector max-parity RAID-Z blocks, in which
	* case the space requirement is exactly (VDEV_RAIDZ_MAXPARITY + 1)
	* times the size; so just assume that. Add to this the fact that
	* we can have up to 3 DVAs per bp, and one more factor of 2 because
	* the block may be dittoed with up to 3 DVAs by ddt_sync(). All together,
	* the worst case is:
	* (VDEV_RAIDZ_MAXPARITY + 1) * SPA_DVAS_PER_BP * 2 == 24
	*/
	int spa_asize_inflation = 24;

	/*
	* Normally, we don't allow the last 3.2% (1/(2^spa_slop_shift)) of space in
	* the pool to be consumed (bounded by spa_max_slop). This ensures that we
	* don't run the pool completely out of space, due to unaccounted changes (e.g.
	* to the MOS). It also limits the worst-case time to allocate space. If we
	* have less than this amount of free space, most ZPL operations (e.g. write,
	* create) will return ENOSPC. The ZIL metaslabs (spa_embedded_log_class) are
	* also part of this 3.2% of space which can't be consumed by normal writes;
	* the slop space "proper" (spa_get_slop_space()) is decreased by the embedded
	* log space.
	*
	* Certain operations (e.g. file removal, most administrative actions) can
	* use half the slop space. They will only return ENOSPC if less than half
	* the slop space is free. Typically, once the pool has less than the slop
	* space free, the user will use these operations to free up space in the pool.
	* These are the operations that call dsl_pool_adjustedsize() with the netfree
	* argument set to TRUE.
	*
	* Operations that are almost guaranteed to free up space in the absence of
	* a pool checkpoint can use up to three quarters of the slop space
	* (e.g zfs destroy).
	*
	* A very restricted set of operations are always permitted, regardless of
	* the amount of free space. These are the operations that call
	* dsl_sync_task(ZFS_SPACE_CHECK_NONE). If these operations result in a net
	* increase in the amount of space used, it is possible to run the pool
	* completely out of space, causing it to be permanently read-only.
	*
	* Note that on very small pools, the slop space will be larger than
	* 3.2%, in an effort to have it be at least spa_min_slop (128MB),
	* but we never allow it to be more than half the pool size.
	*
	* Further, on very large pools, the slop space will be smaller than
	* 3.2%, to avoid reserving much more space than we actually need; bounded
	* by spa_max_slop (128GB).
	*
	* See also the comments in zfs_space_check_t.
	*/
	int spa_slop_shift = 5;
	uint64_t spa_min_slop = 128ULL * 1024 * 1024;
	uint64_t spa_max_slop = 128ULL * 1024 * 1024 * 1024;
	int spa_allocators = 4;


	void
	spa_load_failed(spa_t spa, const char fmt, ...)
	{
	va_list adx;
	char buf[256];

	va_start(adx, fmt);
	(void) vsnprintf(buf, sizeof (buf), fmt, adx);
	va_end(adx);

	zfs_dbgmsg("spa_load(%s, config %s): FAILED: %s", spa->spa_name,
	spa->spa_trust_config ? "trusted" : "untrusted", buf);
	}

	void
	spa_load_note(spa_t spa, const char fmt, ...)
	{
	va_list adx;
	char buf[256];

	va_start(adx, fmt);
	(void) vsnprintf(buf, sizeof (buf), fmt, adx);
	va_end(adx);

	zfs_dbgmsg("spa_load(%s, config %s): %s", spa->spa_name,
	spa->spa_trust_config ? "trusted" : "untrusted", buf);
	}

	/*
	* By default dedup and user data indirects land in the special class
	*/
	int zfs_ddt_data_is_special = B_TRUE;
	int zfs_user_indirect_is_special = B_TRUE;

	/*
	* The percentage of special class final space reserved for metadata only.
	* Once we allocate 100 - zfs_special_class_metadata_reserve_pct we only
	* let metadata into the class.
	*/
	int zfs_special_class_metadata_reserve_pct = 25;

	/*
	* ==========================================================================
	* SPA config locking
	* ==========================================================================
	*/
	static void
	spa_config_lock_init(spa_t *spa)
	{
	for (int i = 0; i < SCL_LOCKS; i++) {
	spa_config_lock_t *scl = &spa->spa_config_lock[i];
	mutex_init(&scl->scl_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&scl->scl_cv, NULL, CV_DEFAULT, NULL);
	scl->scl_writer = NULL;
	scl->scl_write_wanted = 0;
	scl->scl_count = 0;
	}
	}

	static void
	spa_config_lock_destroy(spa_t *spa)
	{
	for (int i = 0; i < SCL_LOCKS; i++) {
	spa_config_lock_t *scl = &spa->spa_config_lock[i];
	mutex_destroy(&scl->scl_lock);
	cv_destroy(&scl->scl_cv);
	ASSERT(scl->scl_writer == NULL);
	ASSERT(scl->scl_write_wanted == 0);
	ASSERT(scl->scl_count == 0);
	}
	}

	int
	spa_config_tryenter(spa_t spa, int locks, void tag, krw_t rw)
	{
	for (int i = 0; i < SCL_LOCKS; i++) {
	spa_config_lock_t *scl = &spa->spa_config_lock[i];
	if (!(locks & (1 << i)))
	continue;
	mutex_enter(&scl->scl_lock);
	if (rw == RW_READER) {
	if (scl->scl_writer \|\| scl->scl_write_wanted) {
	mutex_exit(&scl->scl_lock);
	spa_config_exit(spa, locks & ((1 << i) - 1),
	tag);
	return (0);
	}
	} else {
	ASSERT(scl->scl_writer != curthread);
	if (scl->scl_count != 0) {
	mutex_exit(&scl->scl_lock);
	spa_config_exit(spa, locks & ((1 << i) - 1),
	tag);
	return (0);
	}
	scl->scl_writer = curthread;
	}
	scl->scl_count++;
	mutex_exit(&scl->scl_lock);
	}
	return (1);
	}

	void
	spa_config_enter(spa_t spa, int locks, const void tag, krw_t rw)
	{
	int wlocks_held = 0;

	ASSERT3U(SCL_LOCKS, <, sizeof (wlocks_held) * NBBY);

	for (int i = 0; i < SCL_LOCKS; i++) {
	spa_config_lock_t *scl = &spa->spa_config_lock[i];
	if (scl->scl_writer == curthread)
	wlocks_held \|= (1 << i);
	if (!(locks & (1 << i)))
	continue;
	mutex_enter(&scl->scl_lock);
	if (rw == RW_READER) {
	while (scl->scl_writer \|\| scl->scl_write_wanted) {
	cv_wait(&scl->scl_cv, &scl->scl_lock);
	}
	} else {
	ASSERT(scl->scl_writer != curthread);
	while (scl->scl_count != 0) {
	scl->scl_write_wanted++;
	cv_wait(&scl->scl_cv, &scl->scl_lock);
	scl->scl_write_wanted--;
	}
	scl->scl_writer = curthread;
	}
	scl->scl_count++;
	mutex_exit(&scl->scl_lock);
	}
	ASSERT3U(wlocks_held, <=, locks);
	}

	void
	spa_config_exit(spa_t spa, int locks, const void tag)
	{
	for (int i = SCL_LOCKS - 1; i >= 0; i--) {
	spa_config_lock_t *scl = &spa->spa_config_lock[i];
	if (!(locks & (1 << i)))
	continue;
	mutex_enter(&scl->scl_lock);
	ASSERT(scl->scl_count > 0);
	if (--scl->scl_count == 0) {
	ASSERT(scl->scl_writer == NULL \|\|
	scl->scl_writer == curthread);
	scl->scl_writer = NULL; /* OK in either case */
	cv_broadcast(&scl->scl_cv);
	}
	mutex_exit(&scl->scl_lock);
	}
	}

	int
	spa_config_held(spa_t *spa, int locks, krw_t rw)
	{
	int locks_held = 0;

	for (int i = 0; i < SCL_LOCKS; i++) {
	spa_config_lock_t *scl = &spa->spa_config_lock[i];
	if (!(locks & (1 << i)))
	continue;
	if ((rw == RW_READER && scl->scl_count != 0) \|\|
	(rw == RW_WRITER && scl->scl_writer == curthread))
	locks_held \|= 1 << i;
	}

	return (locks_held);
	}

	/*
	* ==========================================================================
	* SPA namespace functions
	* ==========================================================================
	*/

	/*
	* Lookup the named spa_t in the AVL tree. The spa_namespace_lock must be held.
	* Returns NULL if no matching spa_t is found.
	*/
	spa_t *
	spa_lookup(const char *name)
	{
	static spa_t search; /* spa_t is large; don't allocate on stack */
	spa_t *spa;
	avl_index_t where;
	char *cp;

	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	(void) strlcpy(search.spa_name, name, sizeof (search.spa_name));

	/*
	* If it's a full dataset name, figure out the pool name and
	* just use that.
	*/
	cp = strpbrk(search.spa_name, "/@#");
	if (cp != NULL)
	*cp = '\0';

	spa = avl_find(&spa_namespace_avl, &search, &where);

	return (spa);
	}

	/*
	* Fires when spa_sync has not completed within zfs_deadman_synctime_ms.
	* If the zfs_deadman_enabled flag is set then it inspects all vdev queues
	* looking for potentially hung I/Os.
	*/
	void
	spa_deadman(void *arg)
	{
	spa_t *spa = arg;

	/* Disable the deadman if the pool is suspended. */
	if (spa_suspended(spa))
	return;

	zfs_dbgmsg("slow spa_sync: started %llu seconds ago, calls %llu",
	(gethrtime() - spa->spa_sync_starttime) / NANOSEC,
	(u_longlong_t)++spa->spa_deadman_calls);
	if (zfs_deadman_enabled)
	vdev_deadman(spa->spa_root_vdev, FTAG);

	spa->spa_deadman_tqid = taskq_dispatch_delay(system_delay_taskq,
	spa_deadman, spa, TQ_SLEEP, ddi_get_lbolt() +
	MSEC_TO_TICK(zfs_deadman_checktime_ms));
	}

	static int
	spa_log_sm_sort_by_txg(const void va, const void vb)
	{
	const spa_log_sm_t *a = va;
	const spa_log_sm_t *b = vb;

	return (TREE_CMP(a->sls_txg, b->sls_txg));
	}

	/*
	* Create an uninitialized spa_t with the given name. Requires
	* spa_namespace_lock. The caller must ensure that the spa_t doesn't already
	* exist by calling spa_lookup() first.
	*/
	spa_t *
	spa_add(const char name, nvlist_t config, const char *altroot)
	{
	spa_t *spa;
	spa_config_dirent_t *dp;

	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	spa = kmem_zalloc(sizeof (spa_t), KM_SLEEP);

	mutex_init(&spa->spa_async_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_errlist_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_errlog_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_evicting_os_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_history_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_proc_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_props_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_cksum_tmpls_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_scrub_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_suspend_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_vdev_top_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_feat_stats_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_flushed_ms_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa->spa_activities_lock, NULL, MUTEX_DEFAULT, NULL);

	cv_init(&spa->spa_async_cv, NULL, CV_DEFAULT, NULL);
	cv_init(&spa->spa_evicting_os_cv, NULL, CV_DEFAULT, NULL);
	cv_init(&spa->spa_proc_cv, NULL, CV_DEFAULT, NULL);
	cv_init(&spa->spa_scrub_io_cv, NULL, CV_DEFAULT, NULL);
	cv_init(&spa->spa_suspend_cv, NULL, CV_DEFAULT, NULL);
	cv_init(&spa->spa_activities_cv, NULL, CV_DEFAULT, NULL);
	cv_init(&spa->spa_waiters_cv, NULL, CV_DEFAULT, NULL);

	for (int t = 0; t < TXG_SIZE; t++)
	bplist_create(&spa->spa_free_bplist[t]);

	(void) strlcpy(spa->spa_name, name, sizeof (spa->spa_name));
	spa->spa_state = POOL_STATE_UNINITIALIZED;
	spa->spa_freeze_txg = UINT64_MAX;
	spa->spa_final_txg = UINT64_MAX;
	spa->spa_load_max_txg = UINT64_MAX;
	spa->spa_proc = &p0;
	spa->spa_proc_state = SPA_PROC_NONE;
	spa->spa_trust_config = B_TRUE;
	spa->spa_hostid = zone_get_hostid(NULL);

	spa->spa_deadman_synctime = MSEC2NSEC(zfs_deadman_synctime_ms);
	spa->spa_deadman_ziotime = MSEC2NSEC(zfs_deadman_ziotime_ms);
	spa_set_deadman_failmode(spa, zfs_deadman_failmode);

	zfs_refcount_create(&spa->spa_refcount);
	spa_config_lock_init(spa);
	spa_stats_init(spa);

	avl_add(&spa_namespace_avl, spa);

	/*
	* Set the alternate root, if there is one.
	*/
	if (altroot)
	spa->spa_root = spa_strdup(altroot);

	spa->spa_alloc_count = spa_allocators;
	spa->spa_allocs = kmem_zalloc(spa->spa_alloc_count *
	sizeof (spa_alloc_t), KM_SLEEP);
	for (int i = 0; i < spa->spa_alloc_count; i++) {
	mutex_init(&spa->spa_allocs[i].spaa_lock, NULL, MUTEX_DEFAULT,
	NULL);
	avl_create(&spa->spa_allocs[i].spaa_tree, zio_bookmark_compare,
	sizeof (zio_t), offsetof(zio_t, io_alloc_node));
	}
	avl_create(&spa->spa_metaslabs_by_flushed, metaslab_sort_by_flushed,
	sizeof (metaslab_t), offsetof(metaslab_t, ms_spa_txg_node));
	avl_create(&spa->spa_sm_logs_by_txg, spa_log_sm_sort_by_txg,
	sizeof (spa_log_sm_t), offsetof(spa_log_sm_t, sls_node));
	list_create(&spa->spa_log_summary, sizeof (log_summary_entry_t),
	offsetof(log_summary_entry_t, lse_node));

	/*
	* Every pool starts with the default cachefile
	*/
	list_create(&spa->spa_config_list, sizeof (spa_config_dirent_t),
	offsetof(spa_config_dirent_t, scd_link));

	dp = kmem_zalloc(sizeof (spa_config_dirent_t), KM_SLEEP);
	dp->scd_path = altroot ? NULL : spa_strdup(spa_config_path);
	list_insert_head(&spa->spa_config_list, dp);

	VERIFY(nvlist_alloc(&spa->spa_load_info, NV_UNIQUE_NAME,
	KM_SLEEP) == 0);

	if (config != NULL) {
	nvlist_t *features;

	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ,
	&features) == 0) {
	VERIFY(nvlist_dup(features, &spa->spa_label_features,
	0) == 0);
	}

	VERIFY(nvlist_dup(config, &spa->spa_config, 0) == 0);
	}

	if (spa->spa_label_features == NULL) {
	VERIFY(nvlist_alloc(&spa->spa_label_features, NV_UNIQUE_NAME,
	KM_SLEEP) == 0);
	}

	spa->spa_min_ashift = INT_MAX;
	spa->spa_max_ashift = 0;
	spa->spa_min_alloc = INT_MAX;

	/* Reset cached value */
	spa->spa_dedup_dspace = ~0ULL;

	/*
	* As a pool is being created, treat all features as disabled by
	* setting SPA_FEATURE_DISABLED for all entries in the feature
	* refcount cache.
	*/
	for (int i = 0; i < SPA_FEATURES; i++) {
	spa->spa_feat_refcount_cache[i] = SPA_FEATURE_DISABLED;
	}

	list_create(&spa->spa_leaf_list, sizeof (vdev_t),
	offsetof(vdev_t, vdev_leaf_node));

	return (spa);
	}

	/*
	* Removes a spa_t from the namespace, freeing up any memory used. Requires
	* spa_namespace_lock. This is called only after the spa_t has been closed and
	* deactivated.
	*/
	void
	spa_remove(spa_t *spa)
	{
	spa_config_dirent_t *dp;

	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	ASSERT(spa_state(spa) == POOL_STATE_UNINITIALIZED);
	ASSERT3U(zfs_refcount_count(&spa->spa_refcount), ==, 0);
	ASSERT0(spa->spa_waiters);

	nvlist_free(spa->spa_config_splitting);

	avl_remove(&spa_namespace_avl, spa);
	cv_broadcast(&spa_namespace_cv);

	if (spa->spa_root)
	spa_strfree(spa->spa_root);

	while ((dp = list_head(&spa->spa_config_list)) != NULL) {
	list_remove(&spa->spa_config_list, dp);
	if (dp->scd_path != NULL)
	spa_strfree(dp->scd_path);
	kmem_free(dp, sizeof (spa_config_dirent_t));
	}

	for (int i = 0; i < spa->spa_alloc_count; i++) {
	avl_destroy(&spa->spa_allocs[i].spaa_tree);
	mutex_destroy(&spa->spa_allocs[i].spaa_lock);
	}
	kmem_free(spa->spa_allocs, spa->spa_alloc_count *
	sizeof (spa_alloc_t));

	avl_destroy(&spa->spa_metaslabs_by_flushed);
	avl_destroy(&spa->spa_sm_logs_by_txg);
	list_destroy(&spa->spa_log_summary);
	list_destroy(&spa->spa_config_list);
	list_destroy(&spa->spa_leaf_list);

	nvlist_free(spa->spa_label_features);
	nvlist_free(spa->spa_load_info);
	nvlist_free(spa->spa_feat_stats);
	spa_config_set(spa, NULL);

	zfs_refcount_destroy(&spa->spa_refcount);

	spa_stats_destroy(spa);
	spa_config_lock_destroy(spa);

	for (int t = 0; t < TXG_SIZE; t++)
	bplist_destroy(&spa->spa_free_bplist[t]);

	zio_checksum_templates_free(spa);

	cv_destroy(&spa->spa_async_cv);
	cv_destroy(&spa->spa_evicting_os_cv);
	cv_destroy(&spa->spa_proc_cv);
	cv_destroy(&spa->spa_scrub_io_cv);
	cv_destroy(&spa->spa_suspend_cv);
	cv_destroy(&spa->spa_activities_cv);
	cv_destroy(&spa->spa_waiters_cv);

	mutex_destroy(&spa->spa_flushed_ms_lock);
	mutex_destroy(&spa->spa_async_lock);
	mutex_destroy(&spa->spa_errlist_lock);
	mutex_destroy(&spa->spa_errlog_lock);
	mutex_destroy(&spa->spa_evicting_os_lock);
	mutex_destroy(&spa->spa_history_lock);
	mutex_destroy(&spa->spa_proc_lock);
	mutex_destroy(&spa->spa_props_lock);
	mutex_destroy(&spa->spa_cksum_tmpls_lock);
	mutex_destroy(&spa->spa_scrub_lock);
	mutex_destroy(&spa->spa_suspend_lock);
	mutex_destroy(&spa->spa_vdev_top_lock);
	mutex_destroy(&spa->spa_feat_stats_lock);
	mutex_destroy(&spa->spa_activities_lock);

	kmem_free(spa, sizeof (spa_t));
	}

	/*
	* Given a pool, return the next pool in the namespace, or NULL if there is
	* none. If 'prev' is NULL, return the first pool.
	*/
	spa_t *
	spa_next(spa_t *prev)
	{
	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	if (prev)
	return (AVL_NEXT(&spa_namespace_avl, prev));
	else
	return (avl_first(&spa_namespace_avl));
	}

	/*
	* ==========================================================================
	* SPA refcount functions
	* ==========================================================================
	*/

	/*
	* Add a reference to the given spa_t. Must have at least one reference, or
	* have the namespace lock held.
	*/
	void
	spa_open_ref(spa_t spa, void tag)
	{
	ASSERT(zfs_refcount_count(&spa->spa_refcount) >= spa->spa_minref \|\|
	MUTEX_HELD(&spa_namespace_lock));
	(void) zfs_refcount_add(&spa->spa_refcount, tag);
	}

	/*
	* Remove a reference to the given spa_t. Must have at least one reference, or
	* have the namespace lock held.
	*/
	void
	spa_close(spa_t spa, void tag)
	{
	ASSERT(zfs_refcount_count(&spa->spa_refcount) > spa->spa_minref \|\|
	MUTEX_HELD(&spa_namespace_lock));
	(void) zfs_refcount_remove(&spa->spa_refcount, tag);
	}

	/*
	* Remove a reference to the given spa_t held by a dsl dir that is
	* being asynchronously released. Async releases occur from a taskq
	* performing eviction of dsl datasets and dirs. The namespace lock
	* isn't held and the hold by the object being evicted may contribute to
	* spa_minref (e.g. dataset or directory released during pool export),
	* so the asserts in spa_close() do not apply.
	*/
	void
	spa_async_close(spa_t spa, void tag)
	{
	(void) zfs_refcount_remove(&spa->spa_refcount, tag);
	}

	/*
	* Check to see if the spa refcount is zero. Must be called with
	* spa_namespace_lock held. We really compare against spa_minref, which is the
	* number of references acquired when opening a pool
	*/
	boolean_t
	spa_refcount_zero(spa_t *spa)
	{
	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	return (zfs_refcount_count(&spa->spa_refcount) == spa->spa_minref);
	}

	/*
	* ==========================================================================
	* SPA spare and l2cache tracking
	* ==========================================================================
	*/

	/*
	* Hot spares and cache devices are tracked using the same code below,
	* for 'auxiliary' devices.
	*/

	typedef struct spa_aux {
	uint64_t aux_guid;
	uint64_t aux_pool;
	avl_node_t aux_avl;
	int aux_count;
	} spa_aux_t;

	static inline int
	spa_aux_compare(const void a, const void b)
	{
	const spa_aux_t sa = (const spa_aux_t )a;
	const spa_aux_t sb = (const spa_aux_t )b;

	return (TREE_CMP(sa->aux_guid, sb->aux_guid));
	}

	static void
	spa_aux_add(vdev_t vd, avl_tree_t avl)
	{
	avl_index_t where;
	spa_aux_t search;
	spa_aux_t *aux;

	search.aux_guid = vd->vdev_guid;
	if ((aux = avl_find(avl, &search, &where)) != NULL) {
	aux->aux_count++;
	} else {
	aux = kmem_zalloc(sizeof (spa_aux_t), KM_SLEEP);
	aux->aux_guid = vd->vdev_guid;
	aux->aux_count = 1;
	avl_insert(avl, aux, where);
	}
	}

	static void
	spa_aux_remove(vdev_t vd, avl_tree_t avl)
	{
	spa_aux_t search;
	spa_aux_t *aux;
	avl_index_t where;

	search.aux_guid = vd->vdev_guid;
	aux = avl_find(avl, &search, &where);

	ASSERT(aux != NULL);

	if (--aux->aux_count == 0) {
	avl_remove(avl, aux);
	kmem_free(aux, sizeof (spa_aux_t));
	} else if (aux->aux_pool == spa_guid(vd->vdev_spa)) {
	aux->aux_pool = 0ULL;
	}
	}

	static boolean_t
	spa_aux_exists(uint64_t guid, uint64_t pool, int refcnt, avl_tree_t *avl)
	{
	spa_aux_t search, *found;

	search.aux_guid = guid;
	found = avl_find(avl, &search, NULL);

	if (pool) {
	if (found)
	*pool = found->aux_pool;
	else
	*pool = 0ULL;
	}

	if (refcnt) {
	if (found)
	*refcnt = found->aux_count;
	else
	*refcnt = 0;
	}

	return (found != NULL);
	}

	static void
	spa_aux_activate(vdev_t vd, avl_tree_t avl)
	{
	spa_aux_t search, *found;
	avl_index_t where;

	search.aux_guid = vd->vdev_guid;
	found = avl_find(avl, &search, &where);
	ASSERT(found != NULL);
	ASSERT(found->aux_pool == 0ULL);

	found->aux_pool = spa_guid(vd->vdev_spa);
	}

	/*
	* Spares are tracked globally due to the following constraints:
	*
	* - A spare may be part of multiple pools.
	* - A spare may be added to a pool even if it's actively in use within
	* another pool.
	* - A spare in use in any pool can only be the source of a replacement if
	* the target is a spare in the same pool.
	*
	* We keep track of all spares on the system through the use of a reference
	* counted AVL tree. When a vdev is added as a spare, or used as a replacement
	* spare, then we bump the reference count in the AVL tree. In addition, we set
	* the 'vdev_isspare' member to indicate that the device is a spare (active or
	* inactive). When a spare is made active (used to replace a device in the
	* pool), we also keep track of which pool its been made a part of.
	*
	* The 'spa_spare_lock' protects the AVL tree. These functions are normally
	* called under the spa_namespace lock as part of vdev reconfiguration. The
	* separate spare lock exists for the status query path, which does not need to
	* be completely consistent with respect to other vdev configuration changes.
	*/

	static int
	spa_spare_compare(const void a, const void b)
	{
	return (spa_aux_compare(a, b));
	}

	void
	spa_spare_add(vdev_t *vd)
	{
	mutex_enter(&spa_spare_lock);
	ASSERT(!vd->vdev_isspare);
	spa_aux_add(vd, &spa_spare_avl);
	vd->vdev_isspare = B_TRUE;
	mutex_exit(&spa_spare_lock);
	}

	void
	spa_spare_remove(vdev_t *vd)
	{
	mutex_enter(&spa_spare_lock);
	ASSERT(vd->vdev_isspare);
	spa_aux_remove(vd, &spa_spare_avl);
	vd->vdev_isspare = B_FALSE;
	mutex_exit(&spa_spare_lock);
	}

	boolean_t
	spa_spare_exists(uint64_t guid, uint64_t pool, int refcnt)
	{
	boolean_t found;

	mutex_enter(&spa_spare_lock);
	found = spa_aux_exists(guid, pool, refcnt, &spa_spare_avl);
	mutex_exit(&spa_spare_lock);

	return (found);
	}

	void
	spa_spare_activate(vdev_t *vd)
	{
	mutex_enter(&spa_spare_lock);
	ASSERT(vd->vdev_isspare);
	spa_aux_activate(vd, &spa_spare_avl);
	mutex_exit(&spa_spare_lock);
	}

	/*
	* Level 2 ARC devices are tracked globally for the same reasons as spares.
	* Cache devices currently only support one pool per cache device, and so
	* for these devices the aux reference count is currently unused beyond 1.
	*/

	static int
	spa_l2cache_compare(const void a, const void b)
	{
	return (spa_aux_compare(a, b));
	}

	void
	spa_l2cache_add(vdev_t *vd)
	{
	mutex_enter(&spa_l2cache_lock);
	ASSERT(!vd->vdev_isl2cache);
	spa_aux_add(vd, &spa_l2cache_avl);
	vd->vdev_isl2cache = B_TRUE;
	mutex_exit(&spa_l2cache_lock);
	}

	void
	spa_l2cache_remove(vdev_t *vd)
	{
	mutex_enter(&spa_l2cache_lock);
	ASSERT(vd->vdev_isl2cache);
	spa_aux_remove(vd, &spa_l2cache_avl);
	vd->vdev_isl2cache = B_FALSE;
	mutex_exit(&spa_l2cache_lock);
	}

	boolean_t
	spa_l2cache_exists(uint64_t guid, uint64_t *pool)
	{
	boolean_t found;

	mutex_enter(&spa_l2cache_lock);
	found = spa_aux_exists(guid, pool, NULL, &spa_l2cache_avl);
	mutex_exit(&spa_l2cache_lock);

	return (found);
	}

	void
	spa_l2cache_activate(vdev_t *vd)
	{
	mutex_enter(&spa_l2cache_lock);
	ASSERT(vd->vdev_isl2cache);
	spa_aux_activate(vd, &spa_l2cache_avl);
	mutex_exit(&spa_l2cache_lock);
	}

	/*
	* ==========================================================================
	* SPA vdev locking
	* ==========================================================================
	*/

	/*
	* Lock the given spa_t for the purpose of adding or removing a vdev.
	* Grabs the global spa_namespace_lock plus the spa config lock for writing.
	* It returns the next transaction group for the spa_t.
	*/
	uint64_t
	spa_vdev_enter(spa_t *spa)
	{
	mutex_enter(&spa->spa_vdev_top_lock);
	mutex_enter(&spa_namespace_lock);

	vdev_autotrim_stop_all(spa);

	return (spa_vdev_config_enter(spa));
	}

	/*
	* The same as spa_vdev_enter() above but additionally takes the guid of
	* the vdev being detached. When there is a rebuild in process it will be
	* suspended while the vdev tree is modified then resumed by spa_vdev_exit().
	* The rebuild is canceled if only a single child remains after the detach.
	*/
	uint64_t
	spa_vdev_detach_enter(spa_t *spa, uint64_t guid)
	{
	mutex_enter(&spa->spa_vdev_top_lock);
	mutex_enter(&spa_namespace_lock);

	vdev_autotrim_stop_all(spa);

	if (guid != 0) {
	vdev_t *vd = spa_lookup_by_guid(spa, guid, B_FALSE);
	if (vd) {
	vdev_rebuild_stop_wait(vd->vdev_top);
	}
	}

	return (spa_vdev_config_enter(spa));
	}

	/*
	* Internal implementation for spa_vdev_enter(). Used when a vdev
	* operation requires multiple syncs (i.e. removing a device) while
	* keeping the spa_namespace_lock held.
	*/
	uint64_t
	spa_vdev_config_enter(spa_t *spa)
	{
	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	spa_config_enter(spa, SCL_ALL, spa, RW_WRITER);

	return (spa_last_synced_txg(spa) + 1);
	}

	/*
	* Used in combination with spa_vdev_config_enter() to allow the syncing
	* of multiple transactions without releasing the spa_namespace_lock.
	*/
	void
	spa_vdev_config_exit(spa_t spa, vdev_t vd, uint64_t txg, int error, char *tag)
	{
	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	int config_changed = B_FALSE;

	ASSERT(txg > spa_last_synced_txg(spa));

	spa->spa_pending_vdev = NULL;

	/*
	* Reassess the DTLs.
	*/
	vdev_dtl_reassess(spa->spa_root_vdev, 0, 0, B_FALSE, B_FALSE);

	if (error == 0 && !list_is_empty(&spa->spa_config_dirty_list)) {
	config_changed = B_TRUE;
	spa->spa_config_generation++;
	}

	/*
	* Verify the metaslab classes.
	*/
	ASSERT(metaslab_class_validate(spa_normal_class(spa)) == 0);
	ASSERT(metaslab_class_validate(spa_log_class(spa)) == 0);
	ASSERT(metaslab_class_validate(spa_embedded_log_class(spa)) == 0);
	ASSERT(metaslab_class_validate(spa_special_class(spa)) == 0);
	ASSERT(metaslab_class_validate(spa_dedup_class(spa)) == 0);

	spa_config_exit(spa, SCL_ALL, spa);

	/*
	* Panic the system if the specified tag requires it. This
	* is useful for ensuring that configurations are updated
	* transactionally.
	*/
	if (zio_injection_enabled)
	zio_handle_panic_injection(spa, tag, 0);

	/*
	* Note: this txg_wait_synced() is important because it ensures
	* that there won't be more than one config change per txg.
	* This allows us to use the txg as the generation number.
	*/
	if (error == 0)
	txg_wait_synced(spa->spa_dsl_pool, txg);

	if (vd != NULL) {
	ASSERT(!vd->vdev_detached \|\| vd->vdev_dtl_sm == NULL);
	if (vd->vdev_ops->vdev_op_leaf) {
	mutex_enter(&vd->vdev_initialize_lock);
	vdev_initialize_stop(vd, VDEV_INITIALIZE_CANCELED,
	NULL);
	mutex_exit(&vd->vdev_initialize_lock);

	mutex_enter(&vd->vdev_trim_lock);
	vdev_trim_stop(vd, VDEV_TRIM_CANCELED, NULL);
	mutex_exit(&vd->vdev_trim_lock);
	}

	/*
	* The vdev may be both a leaf and top-level device.
	*/
	vdev_autotrim_stop_wait(vd);

	spa_config_enter(spa, SCL_STATE_ALL, spa, RW_WRITER);
	vdev_free(vd);
	spa_config_exit(spa, SCL_STATE_ALL, spa);
	}

	/*
	* If the config changed, update the config cache.
	*/
	if (config_changed)
	spa_write_cachefile(spa, B_FALSE, B_TRUE);
	}

	/*
	* Unlock the spa_t after adding or removing a vdev. Besides undoing the
	* locking of spa_vdev_enter(), we also want make sure the transactions have
	* synced to disk, and then update the global configuration cache with the new
	* information.
	*/
	int
	spa_vdev_exit(spa_t spa, vdev_t vd, uint64_t txg, int error)
	{
	vdev_autotrim_restart(spa);
	vdev_rebuild_restart(spa);

	spa_vdev_config_exit(spa, vd, txg, error, FTAG);
	mutex_exit(&spa_namespace_lock);
	mutex_exit(&spa->spa_vdev_top_lock);

	return (error);
	}

	/*
	* Lock the given spa_t for the purpose of changing vdev state.
	*/
	void
	spa_vdev_state_enter(spa_t *spa, int oplocks)
	{
	int locks = SCL_STATE_ALL \| oplocks;

	/*
	* Root pools may need to read of the underlying devfs filesystem
	* when opening up a vdev. Unfortunately if we're holding the
	* SCL_ZIO lock it will result in a deadlock when we try to issue
	* the read from the root filesystem. Instead we "prefetch"
	* the associated vnodes that we need prior to opening the
	* underlying devices and cache them so that we can prevent
	* any I/O when we are doing the actual open.
	*/
	if (spa_is_root(spa)) {
	int low = locks & ~(SCL_ZIO - 1);
	int high = locks & ~low;

	spa_config_enter(spa, high, spa, RW_WRITER);
	vdev_hold(spa->spa_root_vdev);
	spa_config_enter(spa, low, spa, RW_WRITER);
	} else {
	spa_config_enter(spa, locks, spa, RW_WRITER);
	}
	spa->spa_vdev_locks = locks;
	}

	int
	spa_vdev_state_exit(spa_t spa, vdev_t vd, int error)
	{
	boolean_t config_changed = B_FALSE;
	vdev_t *vdev_top;

	if (vd == NULL \|\| vd == spa->spa_root_vdev) {
	vdev_top = spa->spa_root_vdev;
	} else {
	vdev_top = vd->vdev_top;
	}

	if (vd != NULL \|\| error == 0)
	vdev_dtl_reassess(vdev_top, 0, 0, B_FALSE, B_FALSE);

	if (vd != NULL) {
	if (vd != spa->spa_root_vdev)
	vdev_state_dirty(vdev_top);

	config_changed = B_TRUE;
	spa->spa_config_generation++;
	}

	if (spa_is_root(spa))
	vdev_rele(spa->spa_root_vdev);

	ASSERT3U(spa->spa_vdev_locks, >=, SCL_STATE_ALL);
	spa_config_exit(spa, spa->spa_vdev_locks, spa);

	/*
	* If anything changed, wait for it to sync. This ensures that,
	* from the system administrator's perspective, zpool(8) commands
	* are synchronous. This is important for things like zpool offline:
	* when the command completes, you expect no further I/O from ZFS.
	*/
	if (vd != NULL)
	txg_wait_synced(spa->spa_dsl_pool, 0);

	/*
	* If the config changed, update the config cache.
	*/
	if (config_changed) {
	mutex_enter(&spa_namespace_lock);
	spa_write_cachefile(spa, B_FALSE, B_TRUE);
	mutex_exit(&spa_namespace_lock);
	}

	return (error);
	}

	/*
	* ==========================================================================
	* Miscellaneous functions
	* ==========================================================================
	*/

	void
	spa_activate_mos_feature(spa_t spa, const char feature, dmu_tx_t *tx)
	{
	if (!nvlist_exists(spa->spa_label_features, feature)) {
	fnvlist_add_boolean(spa->spa_label_features, feature);
	/*
	* When we are creating the pool (tx_txg==TXG_INITIAL), we can't
	* dirty the vdev config because lock SCL_CONFIG is not held.
	* Thankfully, in this case we don't need to dirty the config
	* because it will be written out anyway when we finish
	* creating the pool.
	*/
	if (tx->tx_txg != TXG_INITIAL)
	vdev_config_dirty(spa->spa_root_vdev);
	}
	}

	void
	spa_deactivate_mos_feature(spa_t spa, const char feature)
	{
	if (nvlist_remove_all(spa->spa_label_features, feature) == 0)
	vdev_config_dirty(spa->spa_root_vdev);
	}

	/*
	* Return the spa_t associated with given pool_guid, if it exists. If
	* device_guid is non-zero, determine whether the pool exists and contains
	* a device with the specified device_guid.
	*/
	spa_t *
	spa_by_guid(uint64_t pool_guid, uint64_t device_guid)
	{
	spa_t *spa;
	avl_tree_t *t = &spa_namespace_avl;

	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	for (spa = avl_first(t); spa != NULL; spa = AVL_NEXT(t, spa)) {
	if (spa->spa_state == POOL_STATE_UNINITIALIZED)
	continue;
	if (spa->spa_root_vdev == NULL)
	continue;
	if (spa_guid(spa) == pool_guid) {
	if (device_guid == 0)
	break;

	if (vdev_lookup_by_guid(spa->spa_root_vdev,
	device_guid) != NULL)
	break;

	/*
	* Check any devices we may be in the process of adding.
	*/
	if (spa->spa_pending_vdev) {
	if (vdev_lookup_by_guid(spa->spa_pending_vdev,
	device_guid) != NULL)
	break;
	}
	}
	}

	return (spa);
	}

	/*
	* Determine whether a pool with the given pool_guid exists.
	*/
	boolean_t
	spa_guid_exists(uint64_t pool_guid, uint64_t device_guid)
	{
	return (spa_by_guid(pool_guid, device_guid) != NULL);
	}

	char *
	spa_strdup(const char *s)
	{
	size_t len;
	char *new;

	len = strlen(s);
	new = kmem_alloc(len + 1, KM_SLEEP);
	bcopy(s, new, len);
	new[len] = '\0';

	return (new);
	}

	void
	spa_strfree(char *s)
	{
	kmem_free(s, strlen(s) + 1);
	}

	uint64_t
	spa_generate_guid(spa_t *spa)
	{
	uint64_t guid;

	if (spa != NULL) {
	do {
	(void) random_get_pseudo_bytes((void *)&guid,
	sizeof (guid));
	} while (guid == 0 \|\| spa_guid_exists(spa_guid(spa), guid));
	} else {
	do {
	(void) random_get_pseudo_bytes((void *)&guid,
	sizeof (guid));
	} while (guid == 0 \|\| spa_guid_exists(guid, 0));
	}

	return (guid);
	}

	void
	snprintf_blkptr(char buf, size_t buflen, const blkptr_t bp)
	{
	char type[256];
	char *checksum = NULL;
	char *compress = NULL;

	if (bp != NULL) {
	if (BP_GET_TYPE(bp) & DMU_OT_NEWTYPE) {
	dmu_object_byteswap_t bswap =
	DMU_OT_BYTESWAP(BP_GET_TYPE(bp));
	(void) snprintf(type, sizeof (type), "bswap %s %s",
	DMU_OT_IS_METADATA(BP_GET_TYPE(bp)) ?
	"metadata" : "data",
	dmu_ot_byteswap[bswap].ob_name);
	} else {
	(void) strlcpy(type, dmu_ot[BP_GET_TYPE(bp)].ot_name,
	sizeof (type));
	}
	if (!BP_IS_EMBEDDED(bp)) {
	checksum =
	zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_name;
	}
	compress = zio_compress_table[BP_GET_COMPRESS(bp)].ci_name;
	}

	SNPRINTF_BLKPTR(snprintf, ' ', buf, buflen, bp, type, checksum,
	compress);
	}

	void
	spa_freeze(spa_t *spa)
	{
	uint64_t freeze_txg = 0;

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	if (spa->spa_freeze_txg == UINT64_MAX) {
	freeze_txg = spa_last_synced_txg(spa) + TXG_SIZE;
	spa->spa_freeze_txg = freeze_txg;
	}
	spa_config_exit(spa, SCL_ALL, FTAG);
	if (freeze_txg != 0)
	txg_wait_synced(spa_get_dsl(spa), freeze_txg);
	}

	void
	zfs_panic_recover(const char *fmt, ...)
	{
	va_list adx;

	va_start(adx, fmt);
	vcmn_err(zfs_recover ? CE_WARN : CE_PANIC, fmt, adx);
	va_end(adx);
	}

	/*
	* This is a stripped-down version of strtoull, suitable only for converting
	* lowercase hexadecimal numbers that don't overflow.
	*/
	uint64_t
	zfs_strtonum(const char str, char *nptr)
	{
	uint64_t val = 0;
	char c;
	int digit;

	while ((c = *str) != '\0') {
	if (c >= '0' && c <= '9')
	digit = c - '0';
	else if (c >= 'a' && c <= 'f')
	digit = 10 + c - 'a';
	else
	break;

	val *= 16;
	val += digit;

	str++;
	}

	if (nptr)
	nptr = (char )str;

	return (val);
	}

	void
	spa_activate_allocation_classes(spa_t spa, dmu_tx_t tx)
	{
	/*
	* We bump the feature refcount for each special vdev added to the pool
	*/
	ASSERT(spa_feature_is_enabled(spa, SPA_FEATURE_ALLOCATION_CLASSES));
	spa_feature_incr(spa, SPA_FEATURE_ALLOCATION_CLASSES, tx);
	}

	/*
	* ==========================================================================
	* Accessor functions
	* ==========================================================================
	*/

	boolean_t
	spa_shutting_down(spa_t *spa)
	{
	return (spa->spa_async_suspended);
	}

	dsl_pool_t *
	spa_get_dsl(spa_t *spa)
	{
	return (spa->spa_dsl_pool);
	}

	boolean_t
	spa_is_initializing(spa_t *spa)
	{
	return (spa->spa_is_initializing);
	}

	boolean_t
	spa_indirect_vdevs_loaded(spa_t *spa)
	{
	return (spa->spa_indirect_vdevs_loaded);
	}

	blkptr_t *
	spa_get_rootblkptr(spa_t *spa)
	{
	return (&spa->spa_ubsync.ub_rootbp);
	}

	void
	spa_set_rootblkptr(spa_t spa, const blkptr_t bp)
	{
	spa->spa_uberblock.ub_rootbp = *bp;
	}

	void
	spa_altroot(spa_t spa, char buf, size_t buflen)
	{
	if (spa->spa_root == NULL)
	buf[0] = '\0';
	else
	(void) strncpy(buf, spa->spa_root, buflen);
	}

	int
	spa_sync_pass(spa_t *spa)
	{
	return (spa->spa_sync_pass);
	}

	char *
	spa_name(spa_t *spa)
	{
	return (spa->spa_name);
	}

	uint64_t
	spa_guid(spa_t *spa)
	{
	dsl_pool_t *dp = spa_get_dsl(spa);
	uint64_t guid;

	/*
	* If we fail to parse the config during spa_load(), we can go through
	* the error path (which posts an ereport) and end up here with no root
	* vdev. We stash the original pool guid in 'spa_config_guid' to handle
	* this case.
	*/
	if (spa->spa_root_vdev == NULL)
	return (spa->spa_config_guid);

	guid = spa->spa_last_synced_guid != 0 ?
	spa->spa_last_synced_guid : spa->spa_root_vdev->vdev_guid;

	/*
	* Return the most recently synced out guid unless we're
	* in syncing context.
	*/
	if (dp && dsl_pool_sync_context(dp))
	return (spa->spa_root_vdev->vdev_guid);
	else
	return (guid);
	}

	uint64_t
	spa_load_guid(spa_t *spa)
	{
	/*
	* This is a GUID that exists solely as a reference for the
	* purposes of the arc. It is generated at load time, and
	* is never written to persistent storage.
	*/
	return (spa->spa_load_guid);
	}

	uint64_t
	spa_last_synced_txg(spa_t *spa)
	{
	return (spa->spa_ubsync.ub_txg);
	}

	uint64_t
	spa_first_txg(spa_t *spa)
	{
	return (spa->spa_first_txg);
	}

	uint64_t
	spa_syncing_txg(spa_t *spa)
	{
	return (spa->spa_syncing_txg);
	}

	/*
	* Return the last txg where data can be dirtied. The final txgs
	* will be used to just clear out any deferred frees that remain.
	*/
	uint64_t
	spa_final_dirty_txg(spa_t *spa)
	{
	return (spa->spa_final_txg - TXG_DEFER_SIZE);
	}

	pool_state_t
	spa_state(spa_t *spa)
	{
	return (spa->spa_state);
	}

	spa_load_state_t
	spa_load_state(spa_t *spa)
	{
	return (spa->spa_load_state);
	}

	uint64_t
	spa_freeze_txg(spa_t *spa)
	{
	return (spa->spa_freeze_txg);
	}

	/*
	* Return the inflated asize for a logical write in bytes. This is used by the
	* DMU to calculate the space a logical write will require on disk.
	* If lsize is smaller than the largest physical block size allocatable on this
	* pool we use its value instead, since the write will end up using the whole
	* block anyway.
	*/
	uint64_t
	spa_get_worst_case_asize(spa_t *spa, uint64_t lsize)
	{
	if (lsize == 0)
	return (0); /* No inflation needed */
	return (MAX(lsize, 1 << spa->spa_max_ashift) * spa_asize_inflation);
	}

	/*
	* Return the amount of slop space in bytes. It is typically 1/32 of the pool
	* (3.2%), minus the embedded log space. On very small pools, it may be
	* slightly larger than this. On very large pools, it will be capped to
	* the value of spa_max_slop. The embedded log space is not included in
	* spa_dspace. By subtracting it, the usable space (per "zfs list") is a
	* constant 97% of the total space, regardless of metaslab size (assuming the
	* default spa_slop_shift=5 and a non-tiny pool).
	*
	* See the comment above spa_slop_shift for more details.
	*/
	uint64_t
	spa_get_slop_space(spa_t *spa)
	{
	uint64_t space = 0;
	uint64_t slop = 0;

	/*
	* Make sure spa_dedup_dspace has been set.
	*/
	if (spa->spa_dedup_dspace == ~0ULL)
	spa_update_dspace(spa);

	/*
	* spa_get_dspace() includes the space only logically "used" by
	* deduplicated data, so since it's not useful to reserve more
	* space with more deduplicated data, we subtract that out here.
	*/
	space = spa_get_dspace(spa) - spa->spa_dedup_dspace;
	slop = MIN(space >> spa_slop_shift, spa_max_slop);

	/*
	* Subtract the embedded log space, but no more than half the (3.2%)
	* unusable space. Note, the "no more than half" is only relevant if
	* zfs_embedded_slog_min_ms >> spa_slop_shift < 2, which is not true by
	* default.
	*/
	uint64_t embedded_log =
	metaslab_class_get_dspace(spa_embedded_log_class(spa));
	slop -= MIN(embedded_log, slop >> 1);

	/*
	* Slop space should be at least spa_min_slop, but no more than half
	* the entire pool.
	*/
	slop = MAX(slop, MIN(space >> 1, spa_min_slop));
	return (slop);
	}

	uint64_t
	spa_get_dspace(spa_t *spa)
	{
	return (spa->spa_dspace);
	}

	uint64_t
	spa_get_checkpoint_space(spa_t *spa)
	{
	return (spa->spa_checkpoint_info.sci_dspace);
	}

	void
	spa_update_dspace(spa_t *spa)
	{
	spa->spa_dspace = metaslab_class_get_dspace(spa_normal_class(spa)) +
	ddt_get_dedup_dspace(spa);
	- if (spa->spa_vdev_removal != NULL) {
	+ if (spa->spa_nonallocating_dspace > 0) {
	/*
	- * We can't allocate from the removing device, so subtract
	- * its size if it was included in dspace (i.e. if this is a
	- * normal-class vdev, not special/dedup). This prevents the
	- * DMU/DSL from filling up the (now smaller) pool while we
	- * are in the middle of removing the device.
	+ * Subtract the space provided by all non-allocating vdevs that
	+ * contribute to dspace. If a file is overwritten, its old
	+ * blocks are freed and new blocks are allocated. If there are
	+ * no snapshots of the file, the available space should remain
	+ * the same. The old blocks could be freed from the
	+ * non-allocating vdev, but the new blocks must be allocated on
	+ * other (allocating) vdevs. By reserving the entire size of
	+ * the non-allocating vdevs (including allocated space), we
	+ * ensure that there will be enough space on the allocating
	+ * vdevs for this file overwrite to succeed.
	*
	* Note that the DMU/DSL doesn't actually know or care
	* how much space is allocated (it does its own tracking
	* of how much space has been logically used). So it
	* doesn't matter that the data we are moving may be
	- * allocated twice (on the old device and the new
	- * device).
	+ * allocated twice (on the old device and the new device).
	*/
	- spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);
	- vdev_t *vd =
	- vdev_lookup_top(spa, spa->spa_vdev_removal->svr_vdev_id);
	- /*
	- * If the stars align, we can wind up here after
	- * vdev_remove_complete() has cleared vd->vdev_mg but before
	- * spa->spa_vdev_removal gets cleared, so we must check before
	- * we dereference.
	- */
	- if (vd->vdev_mg &&
	- vd->vdev_mg->mg_class == spa_normal_class(spa)) {
	- spa->spa_dspace -= spa_deflate(spa) ?
	- vd->vdev_stat.vs_dspace : vd->vdev_stat.vs_space;
	- }
	- spa_config_exit(spa, SCL_VDEV, FTAG);
	+ ASSERT3U(spa->spa_dspace, >=, spa->spa_nonallocating_dspace);
	+ spa->spa_dspace -= spa->spa_nonallocating_dspace;
	}
	}

	/*
	* Return the failure mode that has been set to this pool. The default
	* behavior will be to block all I/Os when a complete failure occurs.
	*/
	uint64_t
	spa_get_failmode(spa_t *spa)
	{
	return (spa->spa_failmode);
	}

	boolean_t
	spa_suspended(spa_t *spa)
	{
	return (spa->spa_suspended != ZIO_SUSPEND_NONE);
	}

	uint64_t
	spa_version(spa_t *spa)
	{
	return (spa->spa_ubsync.ub_version);
	}

	boolean_t
	spa_deflate(spa_t *spa)
	{
	return (spa->spa_deflate);
	}

	metaslab_class_t *
	spa_normal_class(spa_t *spa)
	{
	return (spa->spa_normal_class);
	}

	metaslab_class_t *
	spa_log_class(spa_t *spa)
	{
	return (spa->spa_log_class);
	}

	metaslab_class_t *
	spa_embedded_log_class(spa_t *spa)
	{
	return (spa->spa_embedded_log_class);
	}

	metaslab_class_t *
	spa_special_class(spa_t *spa)
	{
	return (spa->spa_special_class);
	}

	metaslab_class_t *
	spa_dedup_class(spa_t *spa)
	{
	return (spa->spa_dedup_class);
	}

	/*
	* Locate an appropriate allocation class
	*/
	metaslab_class_t *
	spa_preferred_class(spa_t *spa, uint64_t size, dmu_object_type_t objtype,
	uint_t level, uint_t special_smallblk)
	{
	/*
	* ZIL allocations determine their class in zio_alloc_zil().
	*/
	ASSERT(objtype != DMU_OT_INTENT_LOG);

	boolean_t has_special_class = spa->spa_special_class->mc_groups != 0;

	if (DMU_OT_IS_DDT(objtype)) {
	if (spa->spa_dedup_class->mc_groups != 0)
	return (spa_dedup_class(spa));
	else if (has_special_class && zfs_ddt_data_is_special)
	return (spa_special_class(spa));
	else
	return (spa_normal_class(spa));
	}

	/* Indirect blocks for user data can land in special if allowed */
	if (level > 0 && (DMU_OT_IS_FILE(objtype) \|\| objtype == DMU_OT_ZVOL)) {
	if (has_special_class && zfs_user_indirect_is_special)
	return (spa_special_class(spa));
	else
	return (spa_normal_class(spa));
	}

	if (DMU_OT_IS_METADATA(objtype) \|\| level > 0) {
	if (has_special_class)
	return (spa_special_class(spa));
	else
	return (spa_normal_class(spa));
	}

	/*
	* Allow small file blocks in special class in some cases (like
	* for the dRAID vdev feature). But always leave a reserve of
	* zfs_special_class_metadata_reserve_pct exclusively for metadata.
	*/
	if (DMU_OT_IS_FILE(objtype) &&
	has_special_class && size <= special_smallblk) {
	metaslab_class_t *special = spa_special_class(spa);
	uint64_t alloc = metaslab_class_get_alloc(special);
	uint64_t space = metaslab_class_get_space(special);
	uint64_t limit =
	(space * (100 - zfs_special_class_metadata_reserve_pct))
	/ 100;

	if (alloc < limit)
	return (special);
	}

	return (spa_normal_class(spa));
	}

	void
	spa_evicting_os_register(spa_t spa, objset_t os)
	{
	mutex_enter(&spa->spa_evicting_os_lock);
	list_insert_head(&spa->spa_evicting_os_list, os);
	mutex_exit(&spa->spa_evicting_os_lock);
	}

	void
	spa_evicting_os_deregister(spa_t spa, objset_t os)
	{
	mutex_enter(&spa->spa_evicting_os_lock);
	list_remove(&spa->spa_evicting_os_list, os);
	cv_broadcast(&spa->spa_evicting_os_cv);
	mutex_exit(&spa->spa_evicting_os_lock);
	}

	void
	spa_evicting_os_wait(spa_t *spa)
	{
	mutex_enter(&spa->spa_evicting_os_lock);
	while (!list_is_empty(&spa->spa_evicting_os_list))
	cv_wait(&spa->spa_evicting_os_cv, &spa->spa_evicting_os_lock);
	mutex_exit(&spa->spa_evicting_os_lock);

	dmu_buf_user_evict_wait();
	}

	int
	spa_max_replication(spa_t *spa)
	{
	/*
	* As of SPA_VERSION == SPA_VERSION_DITTO_BLOCKS, we are able to
	* handle BPs with more than one DVA allocated. Set our max
	* replication level accordingly.
	*/
	if (spa_version(spa) < SPA_VERSION_DITTO_BLOCKS)
	return (1);
	return (MIN(SPA_DVAS_PER_BP, spa_max_replication_override));
	}

	int
	spa_prev_software_version(spa_t *spa)
	{
	return (spa->spa_prev_software_version);
	}

	uint64_t
	spa_deadman_synctime(spa_t *spa)
	{
	return (spa->spa_deadman_synctime);
	}

	spa_autotrim_t
	spa_get_autotrim(spa_t *spa)
	{
	return (spa->spa_autotrim);
	}

	uint64_t
	spa_deadman_ziotime(spa_t *spa)
	{
	return (spa->spa_deadman_ziotime);
	}

	uint64_t
	spa_get_deadman_failmode(spa_t *spa)
	{
	return (spa->spa_deadman_failmode);
	}

	void
	spa_set_deadman_failmode(spa_t spa, const char failmode)
	{
	if (strcmp(failmode, "wait") == 0)
	spa->spa_deadman_failmode = ZIO_FAILURE_MODE_WAIT;
	else if (strcmp(failmode, "continue") == 0)
	spa->spa_deadman_failmode = ZIO_FAILURE_MODE_CONTINUE;
	else if (strcmp(failmode, "panic") == 0)
	spa->spa_deadman_failmode = ZIO_FAILURE_MODE_PANIC;
	else
	spa->spa_deadman_failmode = ZIO_FAILURE_MODE_WAIT;
	}

	void
	spa_set_deadman_ziotime(hrtime_t ns)
	{
	spa_t *spa = NULL;

	if (spa_mode_global != SPA_MODE_UNINIT) {
	mutex_enter(&spa_namespace_lock);
	while ((spa = spa_next(spa)) != NULL)
	spa->spa_deadman_ziotime = ns;
	mutex_exit(&spa_namespace_lock);
	}
	}

	void
	spa_set_deadman_synctime(hrtime_t ns)
	{
	spa_t *spa = NULL;

	if (spa_mode_global != SPA_MODE_UNINIT) {
	mutex_enter(&spa_namespace_lock);
	while ((spa = spa_next(spa)) != NULL)
	spa->spa_deadman_synctime = ns;
	mutex_exit(&spa_namespace_lock);
	}
	}

	uint64_t
	dva_get_dsize_sync(spa_t spa, const dva_t dva)
	{
	uint64_t asize = DVA_GET_ASIZE(dva);
	uint64_t dsize = asize;

	ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0);

	if (asize != 0 && spa->spa_deflate) {
	vdev_t *vd = vdev_lookup_top(spa, DVA_GET_VDEV(dva));
	if (vd != NULL)
	dsize = (asize >> SPA_MINBLOCKSHIFT) *
	vd->vdev_deflate_ratio;
	}

	return (dsize);
	}

	uint64_t
	bp_get_dsize_sync(spa_t spa, const blkptr_t bp)
	{
	uint64_t dsize = 0;

	for (int d = 0; d < BP_GET_NDVAS(bp); d++)
	dsize += dva_get_dsize_sync(spa, &bp->blk_dva[d]);

	return (dsize);
	}

	uint64_t
	bp_get_dsize(spa_t spa, const blkptr_t bp)
	{
	uint64_t dsize = 0;

	spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);

	for (int d = 0; d < BP_GET_NDVAS(bp); d++)
	dsize += dva_get_dsize_sync(spa, &bp->blk_dva[d]);

	spa_config_exit(spa, SCL_VDEV, FTAG);

	return (dsize);
	}

	uint64_t
	spa_dirty_data(spa_t *spa)
	{
	return (spa->spa_dsl_pool->dp_dirty_total);
	}

	/*
	* ==========================================================================
	* SPA Import Progress Routines
	* ==========================================================================
	*/

	typedef struct spa_import_progress {
	uint64_t pool_guid; /* unique id for updates */
	char *pool_name;
	spa_load_state_t spa_load_state;
	uint64_t mmp_sec_remaining; /* MMP activity check */
	uint64_t spa_load_max_txg; /* rewind txg */
	procfs_list_node_t smh_node;
	} spa_import_progress_t;

	spa_history_list_t *spa_import_progress_list = NULL;

	static int
	spa_import_progress_show_header(struct seq_file *f)
	{
	seq_printf(f, "%-20s %-14s %-14s %-12s %s\n", "pool_guid",
	"load_state", "multihost_secs", "max_txg",
	"pool_name");
	return (0);
	}

	static int
	spa_import_progress_show(struct seq_file f, void data)
	{
	spa_import_progress_t sip = (spa_import_progress_t )data;

	seq_printf(f, "%-20llu %-14llu %-14llu %-12llu %s\n",
	(u_longlong_t)sip->pool_guid, (u_longlong_t)sip->spa_load_state,
	(u_longlong_t)sip->mmp_sec_remaining,
	(u_longlong_t)sip->spa_load_max_txg,
	(sip->pool_name ? sip->pool_name : "-"));

	return (0);
	}

	/* Remove oldest elements from list until there are no more than 'size' left */
	static void
	spa_import_progress_truncate(spa_history_list_t *shl, unsigned int size)
	{
	spa_import_progress_t *sip;
	while (shl->size > size) {
	sip = list_remove_head(&shl->procfs_list.pl_list);
	if (sip->pool_name)
	spa_strfree(sip->pool_name);
	kmem_free(sip, sizeof (spa_import_progress_t));
	shl->size--;
	}

	IMPLY(size == 0, list_is_empty(&shl->procfs_list.pl_list));
	}

	static void
	spa_import_progress_init(void)
	{
	spa_import_progress_list = kmem_zalloc(sizeof (spa_history_list_t),
	KM_SLEEP);

	spa_import_progress_list->size = 0;

	spa_import_progress_list->procfs_list.pl_private =
	spa_import_progress_list;

	procfs_list_install("zfs",
	NULL,
	"import_progress",
	0644,
	&spa_import_progress_list->procfs_list,
	spa_import_progress_show,
	spa_import_progress_show_header,
	NULL,
	offsetof(spa_import_progress_t, smh_node));
	}

	static void
	spa_import_progress_destroy(void)
	{
	spa_history_list_t *shl = spa_import_progress_list;
	procfs_list_uninstall(&shl->procfs_list);
	spa_import_progress_truncate(shl, 0);
	procfs_list_destroy(&shl->procfs_list);
	kmem_free(shl, sizeof (spa_history_list_t));
	}

	int
	spa_import_progress_set_state(uint64_t pool_guid,
	spa_load_state_t load_state)
	{
	spa_history_list_t *shl = spa_import_progress_list;
	spa_import_progress_t *sip;
	int error = ENOENT;

	if (shl->size == 0)
	return (0);

	mutex_enter(&shl->procfs_list.pl_lock);
	for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL;
	sip = list_prev(&shl->procfs_list.pl_list, sip)) {
	if (sip->pool_guid == pool_guid) {
	sip->spa_load_state = load_state;
	error = 0;
	break;
	}
	}
	mutex_exit(&shl->procfs_list.pl_lock);

	return (error);
	}

	int
	spa_import_progress_set_max_txg(uint64_t pool_guid, uint64_t load_max_txg)
	{
	spa_history_list_t *shl = spa_import_progress_list;
	spa_import_progress_t *sip;
	int error = ENOENT;

	if (shl->size == 0)
	return (0);

	mutex_enter(&shl->procfs_list.pl_lock);
	for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL;
	sip = list_prev(&shl->procfs_list.pl_list, sip)) {
	if (sip->pool_guid == pool_guid) {
	sip->spa_load_max_txg = load_max_txg;
	error = 0;
	break;
	}
	}
	mutex_exit(&shl->procfs_list.pl_lock);

	return (error);
	}

	int
	spa_import_progress_set_mmp_check(uint64_t pool_guid,
	uint64_t mmp_sec_remaining)
	{
	spa_history_list_t *shl = spa_import_progress_list;
	spa_import_progress_t *sip;
	int error = ENOENT;

	if (shl->size == 0)
	return (0);

	mutex_enter(&shl->procfs_list.pl_lock);
	for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL;
	sip = list_prev(&shl->procfs_list.pl_list, sip)) {
	if (sip->pool_guid == pool_guid) {
	sip->mmp_sec_remaining = mmp_sec_remaining;
	error = 0;
	break;
	}
	}
	mutex_exit(&shl->procfs_list.pl_lock);

	return (error);
	}

	/*
	* A new import is in progress, add an entry.
	*/
	void
	spa_import_progress_add(spa_t *spa)
	{
	spa_history_list_t *shl = spa_import_progress_list;
	spa_import_progress_t *sip;
	char *poolname = NULL;

	sip = kmem_zalloc(sizeof (spa_import_progress_t), KM_SLEEP);
	sip->pool_guid = spa_guid(spa);

	(void) nvlist_lookup_string(spa->spa_config, ZPOOL_CONFIG_POOL_NAME,
	&poolname);
	if (poolname == NULL)
	poolname = spa_name(spa);
	sip->pool_name = spa_strdup(poolname);
	sip->spa_load_state = spa_load_state(spa);

	mutex_enter(&shl->procfs_list.pl_lock);
	procfs_list_add(&shl->procfs_list, sip);
	shl->size++;
	mutex_exit(&shl->procfs_list.pl_lock);
	}

	void
	spa_import_progress_remove(uint64_t pool_guid)
	{
	spa_history_list_t *shl = spa_import_progress_list;
	spa_import_progress_t *sip;

	mutex_enter(&shl->procfs_list.pl_lock);
	for (sip = list_tail(&shl->procfs_list.pl_list); sip != NULL;
	sip = list_prev(&shl->procfs_list.pl_list, sip)) {
	if (sip->pool_guid == pool_guid) {
	if (sip->pool_name)
	spa_strfree(sip->pool_name);
	list_remove(&shl->procfs_list.pl_list, sip);
	shl->size--;
	kmem_free(sip, sizeof (spa_import_progress_t));
	break;
	}
	}
	mutex_exit(&shl->procfs_list.pl_lock);
	}

	/*
	* ==========================================================================
	* Initialization and Termination
	* ==========================================================================
	*/

	static int
	spa_name_compare(const void a1, const void a2)
	{
	const spa_t *s1 = a1;
	const spa_t *s2 = a2;
	int s;

	s = strcmp(s1->spa_name, s2->spa_name);

	return (TREE_ISIGN(s));
	}

	void
	spa_boot_init(void)
	{
	spa_config_load();
	}

	void
	spa_init(spa_mode_t mode)
	{
	mutex_init(&spa_namespace_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa_spare_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&spa_l2cache_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&spa_namespace_cv, NULL, CV_DEFAULT, NULL);

	avl_create(&spa_namespace_avl, spa_name_compare, sizeof (spa_t),
	offsetof(spa_t, spa_avl));

	avl_create(&spa_spare_avl, spa_spare_compare, sizeof (spa_aux_t),
	offsetof(spa_aux_t, aux_avl));

	avl_create(&spa_l2cache_avl, spa_l2cache_compare, sizeof (spa_aux_t),
	offsetof(spa_aux_t, aux_avl));

	spa_mode_global = mode;

	#ifndef _KERNEL
	if (spa_mode_global != SPA_MODE_READ && dprintf_find_string("watch")) {
	struct sigaction sa;

	sa.sa_flags = SA_SIGINFO;
	sigemptyset(&sa.sa_mask);
	sa.sa_sigaction = arc_buf_sigsegv;

	if (sigaction(SIGSEGV, &sa, NULL) == -1) {
	perror("could not enable watchpoints: "
	"sigaction(SIGSEGV, ...) = ");
	} else {
	arc_watch = B_TRUE;
	}
	}
	#endif

	fm_init();
	zfs_refcount_init();
	unique_init();
	zfs_btree_init();
	metaslab_stat_init();
	ddt_init();
	zio_init();
	dmu_init();
	zil_init();
	vdev_cache_stat_init();
	vdev_mirror_stat_init();
	vdev_raidz_math_init();
	vdev_file_init();
	zfs_prop_init();
	zpool_prop_init();
	zpool_feature_init();
	spa_config_load();
	+ vdev_prop_init();
	l2arc_start();
	scan_init();
	qat_init();
	spa_import_progress_init();
	}

	void
	spa_fini(void)
	{
	l2arc_stop();

	spa_evict_all();

	vdev_file_fini();
	vdev_cache_stat_fini();
	vdev_mirror_stat_fini();
	vdev_raidz_math_fini();
	zil_fini();
	dmu_fini();
	zio_fini();
	ddt_fini();
	metaslab_stat_fini();
	zfs_btree_fini();
	unique_fini();
	zfs_refcount_fini();
	fm_fini();
	scan_fini();
	qat_fini();
	spa_import_progress_destroy();

	avl_destroy(&spa_namespace_avl);
	avl_destroy(&spa_spare_avl);
	avl_destroy(&spa_l2cache_avl);

	cv_destroy(&spa_namespace_cv);
	mutex_destroy(&spa_namespace_lock);
	mutex_destroy(&spa_spare_lock);
	mutex_destroy(&spa_l2cache_lock);
	}

	/*
	* Return whether this pool has a dedicated slog device. No locking needed.
	* It's not a problem if the wrong answer is returned as it's only for
	* performance and not correctness.
	*/
	boolean_t
	spa_has_slogs(spa_t *spa)
	{
	return (spa->spa_log_class->mc_groups != 0);
	}

	spa_log_state_t
	spa_get_log_state(spa_t *spa)
	{
	return (spa->spa_log_state);
	}

	void
	spa_set_log_state(spa_t *spa, spa_log_state_t state)
	{
	spa->spa_log_state = state;
	}

	boolean_t
	spa_is_root(spa_t *spa)
	{
	return (spa->spa_is_root);
	}

	boolean_t
	spa_writeable(spa_t *spa)
	{
	return (!!(spa->spa_mode & SPA_MODE_WRITE) && spa->spa_trust_config);
	}

	/*
	* Returns true if there is a pending sync task in any of the current
	* syncing txg, the current quiescing txg, or the current open txg.
	*/
	boolean_t
	spa_has_pending_synctask(spa_t *spa)
	{
	return (!txg_all_lists_empty(&spa->spa_dsl_pool->dp_sync_tasks) \|\|
	!txg_all_lists_empty(&spa->spa_dsl_pool->dp_early_sync_tasks));
	}

	spa_mode_t
	spa_mode(spa_t *spa)
	{
	return (spa->spa_mode);
	}

	uint64_t
	spa_bootfs(spa_t *spa)
	{
	return (spa->spa_bootfs);
	}

	uint64_t
	spa_delegation(spa_t *spa)
	{
	return (spa->spa_delegation);
	}

	objset_t *
	spa_meta_objset(spa_t *spa)
	{
	return (spa->spa_meta_objset);
	}

	enum zio_checksum
	spa_dedup_checksum(spa_t *spa)
	{
	return (spa->spa_dedup_checksum);
	}

	/*
	* Reset pool scan stat per scan pass (or reboot).
	*/
	void
	spa_scan_stat_init(spa_t *spa)
	{
	/* data not stored on disk */
	spa->spa_scan_pass_start = gethrestime_sec();
	if (dsl_scan_is_paused_scrub(spa->spa_dsl_pool->dp_scan))
	spa->spa_scan_pass_scrub_pause = spa->spa_scan_pass_start;
	else
	spa->spa_scan_pass_scrub_pause = 0;
	spa->spa_scan_pass_scrub_spent_paused = 0;
	spa->spa_scan_pass_exam = 0;
	spa->spa_scan_pass_issued = 0;
	vdev_scan_stat_init(spa->spa_root_vdev);
	}

	/*
	* Get scan stats for zpool status reports
	*/
	int
	spa_scan_get_stats(spa_t spa, pool_scan_stat_t ps)
	{
	dsl_scan_t *scn = spa->spa_dsl_pool ? spa->spa_dsl_pool->dp_scan : NULL;

	if (scn == NULL \|\| scn->scn_phys.scn_func == POOL_SCAN_NONE)
	return (SET_ERROR(ENOENT));
	bzero(ps, sizeof (pool_scan_stat_t));

	/* data stored on disk */
	ps->pss_func = scn->scn_phys.scn_func;
	ps->pss_state = scn->scn_phys.scn_state;
	ps->pss_start_time = scn->scn_phys.scn_start_time;
	ps->pss_end_time = scn->scn_phys.scn_end_time;
	ps->pss_to_examine = scn->scn_phys.scn_to_examine;
	ps->pss_examined = scn->scn_phys.scn_examined;
	ps->pss_to_process = scn->scn_phys.scn_to_process;
	ps->pss_processed = scn->scn_phys.scn_processed;
	ps->pss_errors = scn->scn_phys.scn_errors;

	/* data not stored on disk */
	ps->pss_pass_exam = spa->spa_scan_pass_exam;
	ps->pss_pass_start = spa->spa_scan_pass_start;
	ps->pss_pass_scrub_pause = spa->spa_scan_pass_scrub_pause;
	ps->pss_pass_scrub_spent_paused = spa->spa_scan_pass_scrub_spent_paused;
	ps->pss_pass_issued = spa->spa_scan_pass_issued;
	ps->pss_issued =
	scn->scn_issued_before_pass + spa->spa_scan_pass_issued;

	return (0);
	}

	int
	spa_maxblocksize(spa_t *spa)
	{
	if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_BLOCKS))
	return (SPA_MAXBLOCKSIZE);
	else
	return (SPA_OLD_MAXBLOCKSIZE);
	}


	/*
	* Returns the txg that the last device removal completed. No indirect mappings
	* have been added since this txg.
	*/
	uint64_t
	spa_get_last_removal_txg(spa_t *spa)
	{
	uint64_t vdevid;
	uint64_t ret = -1ULL;

	spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);
	/*
	* sr_prev_indirect_vdev is only modified while holding all the
	* config locks, so it is sufficient to hold SCL_VDEV as reader when
	* examining it.
	*/
	vdevid = spa->spa_removing_phys.sr_prev_indirect_vdev;

	while (vdevid != -1ULL) {
	vdev_t *vd = vdev_lookup_top(spa, vdevid);
	vdev_indirect_births_t *vib = vd->vdev_indirect_births;

	ASSERT3P(vd->vdev_ops, ==, &vdev_indirect_ops);

	/*
	* If the removal did not remap any data, we don't care.
	*/
	if (vdev_indirect_births_count(vib) != 0) {
	ret = vdev_indirect_births_last_entry_txg(vib);
	break;
	}

	vdevid = vd->vdev_indirect_config.vic_prev_indirect_vdev;
	}
	spa_config_exit(spa, SCL_VDEV, FTAG);

	IMPLY(ret != -1ULL,
	spa_feature_is_active(spa, SPA_FEATURE_DEVICE_REMOVAL));

	return (ret);
	}

	int
	spa_maxdnodesize(spa_t *spa)
	{
	if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_DNODE))
	return (DNODE_MAX_SIZE);
	else
	return (DNODE_MIN_SIZE);
	}

	boolean_t
	spa_multihost(spa_t *spa)
	{
	return (spa->spa_multihost ? B_TRUE : B_FALSE);
	}

	uint32_t
	spa_get_hostid(spa_t *spa)
	{
	return (spa->spa_hostid);
	}

	boolean_t
	spa_trust_config(spa_t *spa)
	{
	return (spa->spa_trust_config);
	}

	uint64_t
	spa_missing_tvds_allowed(spa_t *spa)
	{
	return (spa->spa_missing_tvds_allowed);
	}

	space_map_t *
	spa_syncing_log_sm(spa_t *spa)
	{
	return (spa->spa_syncing_log_sm);
	}

	void
	spa_set_missing_tvds(spa_t *spa, uint64_t missing)
	{
	spa->spa_missing_tvds = missing;
	}

	/*
	* Return the pool state string ("ONLINE", "DEGRADED", "SUSPENDED", etc).
	*/
	const char *
	spa_state_to_name(spa_t *spa)
	{
	ASSERT3P(spa, !=, NULL);

	/*
	* it is possible for the spa to exist, without root vdev
	* as the spa transitions during import/export
	*/
	vdev_t *rvd = spa->spa_root_vdev;
	if (rvd == NULL) {
	return ("TRANSITIONING");
	}
	vdev_state_t state = rvd->vdev_state;
	vdev_aux_t aux = rvd->vdev_stat.vs_aux;

	if (spa_suspended(spa) &&
	(spa_get_failmode(spa) != ZIO_FAILURE_MODE_CONTINUE))
	return ("SUSPENDED");

	switch (state) {
	case VDEV_STATE_CLOSED:
	case VDEV_STATE_OFFLINE:
	return ("OFFLINE");
	case VDEV_STATE_REMOVED:
	return ("REMOVED");
	case VDEV_STATE_CANT_OPEN:
	if (aux == VDEV_AUX_CORRUPT_DATA \|\| aux == VDEV_AUX_BAD_LOG)
	return ("FAULTED");
	else if (aux == VDEV_AUX_SPLIT_POOL)
	return ("SPLIT");
	else
	return ("UNAVAIL");
	case VDEV_STATE_FAULTED:
	return ("FAULTED");
	case VDEV_STATE_DEGRADED:
	return ("DEGRADED");
	case VDEV_STATE_HEALTHY:
	return ("ONLINE");
	default:
	break;
	}

	return ("UNKNOWN");
	}

	boolean_t
	spa_top_vdevs_spacemap_addressable(spa_t *spa)
	{
	vdev_t *rvd = spa->spa_root_vdev;
	for (uint64_t c = 0; c < rvd->vdev_children; c++) {
	if (!vdev_is_spacemap_addressable(rvd->vdev_child[c]))
	return (B_FALSE);
	}
	return (B_TRUE);
	}

	boolean_t
	spa_has_checkpoint(spa_t *spa)
	{
	return (spa->spa_checkpoint_txg != 0);
	}

	boolean_t
	spa_importing_readonly_checkpoint(spa_t *spa)
	{
	return ((spa->spa_import_flags & ZFS_IMPORT_CHECKPOINT) &&
	spa->spa_mode == SPA_MODE_READ);
	}

	uint64_t
	spa_min_claim_txg(spa_t *spa)
	{
	uint64_t checkpoint_txg = spa->spa_uberblock.ub_checkpoint_txg;

	if (checkpoint_txg != 0)
	return (checkpoint_txg + 1);

	return (spa->spa_first_txg);
	}

	/*
	* If there is a checkpoint, async destroys may consume more space from
	* the pool instead of freeing it. In an attempt to save the pool from
	* getting suspended when it is about to run out of space, we stop
	* processing async destroys.
	*/
	boolean_t
	spa_suspend_async_destroy(spa_t *spa)
	{
	dsl_pool_t *dp = spa_get_dsl(spa);

	uint64_t unreserved = dsl_pool_unreserved_space(dp,
	ZFS_SPACE_CHECK_EXTRA_RESERVED);
	uint64_t used = dsl_dir_phys(dp->dp_root_dir)->dd_used_bytes;
	uint64_t avail = (unreserved > used) ? (unreserved - used) : 0;

	if (spa_has_checkpoint(spa) && avail == 0)
	return (B_TRUE);

	return (B_FALSE);
	}

	#if defined(_KERNEL)

	int
	param_set_deadman_failmode_common(const char *val)
	{
	spa_t *spa = NULL;
	char *p;

	if (val == NULL)
	return (SET_ERROR(EINVAL));

	if ((p = strchr(val, '\n')) != NULL)
	*p = '\0';

	if (strcmp(val, "wait") != 0 && strcmp(val, "continue") != 0 &&
	strcmp(val, "panic"))
	return (SET_ERROR(EINVAL));

	if (spa_mode_global != SPA_MODE_UNINIT) {
	mutex_enter(&spa_namespace_lock);
	while ((spa = spa_next(spa)) != NULL)
	spa_set_deadman_failmode(spa, val);
	mutex_exit(&spa_namespace_lock);
	}

	return (0);
	}
	#endif

	/* Namespace manipulation */
	EXPORT_SYMBOL(spa_lookup);
	EXPORT_SYMBOL(spa_add);
	EXPORT_SYMBOL(spa_remove);
	EXPORT_SYMBOL(spa_next);

	/* Refcount functions */
	EXPORT_SYMBOL(spa_open_ref);
	EXPORT_SYMBOL(spa_close);
	EXPORT_SYMBOL(spa_refcount_zero);

	/* Pool configuration lock */
	EXPORT_SYMBOL(spa_config_tryenter);
	EXPORT_SYMBOL(spa_config_enter);
	EXPORT_SYMBOL(spa_config_exit);
	EXPORT_SYMBOL(spa_config_held);

	/* Pool vdev add/remove lock */
	EXPORT_SYMBOL(spa_vdev_enter);
	EXPORT_SYMBOL(spa_vdev_exit);

	/* Pool vdev state change lock */
	EXPORT_SYMBOL(spa_vdev_state_enter);
	EXPORT_SYMBOL(spa_vdev_state_exit);

	/* Accessor functions */
	EXPORT_SYMBOL(spa_shutting_down);
	EXPORT_SYMBOL(spa_get_dsl);
	EXPORT_SYMBOL(spa_get_rootblkptr);
	EXPORT_SYMBOL(spa_set_rootblkptr);
	EXPORT_SYMBOL(spa_altroot);
	EXPORT_SYMBOL(spa_sync_pass);
	EXPORT_SYMBOL(spa_name);
	EXPORT_SYMBOL(spa_guid);
	EXPORT_SYMBOL(spa_last_synced_txg);
	EXPORT_SYMBOL(spa_first_txg);
	EXPORT_SYMBOL(spa_syncing_txg);
	EXPORT_SYMBOL(spa_version);
	EXPORT_SYMBOL(spa_state);
	EXPORT_SYMBOL(spa_load_state);
	EXPORT_SYMBOL(spa_freeze_txg);
	EXPORT_SYMBOL(spa_get_dspace);
	EXPORT_SYMBOL(spa_update_dspace);
	EXPORT_SYMBOL(spa_deflate);
	EXPORT_SYMBOL(spa_normal_class);
	EXPORT_SYMBOL(spa_log_class);
	EXPORT_SYMBOL(spa_special_class);
	EXPORT_SYMBOL(spa_preferred_class);
	EXPORT_SYMBOL(spa_max_replication);
	EXPORT_SYMBOL(spa_prev_software_version);
	EXPORT_SYMBOL(spa_get_failmode);
	EXPORT_SYMBOL(spa_suspended);
	EXPORT_SYMBOL(spa_bootfs);
	EXPORT_SYMBOL(spa_delegation);
	EXPORT_SYMBOL(spa_meta_objset);
	EXPORT_SYMBOL(spa_maxblocksize);
	EXPORT_SYMBOL(spa_maxdnodesize);

	/* Miscellaneous support routines */
	EXPORT_SYMBOL(spa_guid_exists);
	EXPORT_SYMBOL(spa_strdup);
	EXPORT_SYMBOL(spa_strfree);
	EXPORT_SYMBOL(spa_generate_guid);
	EXPORT_SYMBOL(snprintf_blkptr);
	EXPORT_SYMBOL(spa_freeze);
	EXPORT_SYMBOL(spa_upgrade);
	EXPORT_SYMBOL(spa_evict_all);
	EXPORT_SYMBOL(spa_lookup_by_guid);
	EXPORT_SYMBOL(spa_has_spare);
	EXPORT_SYMBOL(dva_get_dsize_sync);
	EXPORT_SYMBOL(bp_get_dsize_sync);
	EXPORT_SYMBOL(bp_get_dsize);
	EXPORT_SYMBOL(spa_has_slogs);
	EXPORT_SYMBOL(spa_is_root);
	EXPORT_SYMBOL(spa_writeable);
	EXPORT_SYMBOL(spa_mode);
	EXPORT_SYMBOL(spa_namespace_lock);
	EXPORT_SYMBOL(spa_trust_config);
	EXPORT_SYMBOL(spa_missing_tvds_allowed);
	EXPORT_SYMBOL(spa_set_missing_tvds);
	EXPORT_SYMBOL(spa_state_to_name);
	EXPORT_SYMBOL(spa_importing_readonly_checkpoint);
	EXPORT_SYMBOL(spa_min_claim_txg);
	EXPORT_SYMBOL(spa_suspend_async_destroy);
	EXPORT_SYMBOL(spa_has_checkpoint);
	EXPORT_SYMBOL(spa_top_vdevs_spacemap_addressable);

	ZFS_MODULE_PARAM(zfs, zfs_, flags, UINT, ZMOD_RW,
	"Set additional debugging flags");

	ZFS_MODULE_PARAM(zfs, zfs_, recover, INT, ZMOD_RW,
	"Set to attempt to recover from fatal errors");

	ZFS_MODULE_PARAM(zfs, zfs_, free_leak_on_eio, INT, ZMOD_RW,
	"Set to ignore IO errors during free and permanently leak the space");

	ZFS_MODULE_PARAM(zfs_deadman, zfs_deadman_, checktime_ms, ULONG, ZMOD_RW,
	"Dead I/O check interval in milliseconds");

	ZFS_MODULE_PARAM(zfs_deadman, zfs_deadman_, enabled, INT, ZMOD_RW,
	"Enable deadman timer");

	ZFS_MODULE_PARAM(zfs_spa, spa_, asize_inflation, INT, ZMOD_RW,
	"SPA size estimate multiplication factor");

	ZFS_MODULE_PARAM(zfs, zfs_, ddt_data_is_special, INT, ZMOD_RW,
	"Place DDT data into the special class");

	ZFS_MODULE_PARAM(zfs, zfs_, user_indirect_is_special, INT, ZMOD_RW,
	"Place user data indirect blocks into the special class");

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM_CALL(zfs_deadman, zfs_deadman_, failmode,
	param_set_deadman_failmode, param_get_charp, ZMOD_RW,
	"Failmode for deadman timer");

	ZFS_MODULE_PARAM_CALL(zfs_deadman, zfs_deadman_, synctime_ms,
	param_set_deadman_synctime, param_get_ulong, ZMOD_RW,
	"Pool sync expiration time in milliseconds");

	ZFS_MODULE_PARAM_CALL(zfs_deadman, zfs_deadman_, ziotime_ms,
	param_set_deadman_ziotime, param_get_ulong, ZMOD_RW,
	"IO expiration time in milliseconds");

	ZFS_MODULE_PARAM(zfs, zfs_, special_class_metadata_reserve_pct, INT, ZMOD_RW,
	"Small file blocks in special vdevs depends on this much "
	"free space available");
	/* END CSTYLED */

	ZFS_MODULE_PARAM_CALL(zfs_spa, spa_, slop_shift, param_set_slop_shift,
	param_get_int, ZMOD_RW, "Reserved free space in pool");
	diff --git a/sys/contrib/openzfs/module/zfs/vdev.c b/sys/contrib/openzfs/module/zfs/vdev.c
	index 4a67ba85f58a..5784bd2a09c3 100644
	--- a/sys/contrib/openzfs/module/zfs/vdev.c
	+++ b/sys/contrib/openzfs/module/zfs/vdev.c
	@@ -1,5453 +1,6071 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2021 by Delphix. All rights reserved.
	* Copyright 2017 Nexenta Systems, Inc.
	* Copyright (c) 2014 Integros [integros.com]
	* Copyright 2016 Toomas Soome <tsoome@me.com>
	* Copyright 2017 Joyent, Inc.
	* Copyright (c) 2017, Intel Corporation.
	* Copyright (c) 2019, Datto Inc. All rights reserved.
	+ * Copyright (c) 2021, Klara Inc.
	* Copyright [2021] Hewlett Packard Enterprise Development LP
	*/

	#include <sys/zfs_context.h>
	#include <sys/fm/fs/zfs.h>
	#include <sys/spa.h>
	#include <sys/spa_impl.h>
	#include <sys/bpobj.h>
	#include <sys/dmu.h>
	#include <sys/dmu_tx.h>
	#include <sys/dsl_dir.h>
	#include <sys/vdev_impl.h>
	#include <sys/vdev_rebuild.h>
	#include <sys/vdev_draid.h>
	#include <sys/uberblock_impl.h>
	#include <sys/metaslab.h>
	#include <sys/metaslab_impl.h>
	#include <sys/space_map.h>
	#include <sys/space_reftree.h>
	#include <sys/zio.h>
	#include <sys/zap.h>
	#include <sys/fs/zfs.h>
	#include <sys/arc.h>
	#include <sys/zil.h>
	#include <sys/dsl_scan.h>
	#include <sys/vdev_raidz.h>
	#include <sys/abd.h>
	#include <sys/vdev_initialize.h>
	#include <sys/vdev_trim.h>
	#include <sys/zvol.h>
	#include <sys/zfs_ratelimit.h>
	+#include "zfs_prop.h"

	/*
	* One metaslab from each (normal-class) vdev is used by the ZIL. These are
	* called "embedded slog metaslabs", are referenced by vdev_log_mg, and are
	* part of the spa_embedded_log_class. The metaslab with the most free space
	* in each vdev is selected for this purpose when the pool is opened (or a
	* vdev is added). See vdev_metaslab_init().
	*
	* Log blocks can be allocated from the following locations. Each one is tried
	* in order until the allocation succeeds:
	* 1. dedicated log vdevs, aka "slog" (spa_log_class)
	* 2. embedded slog metaslabs (spa_embedded_log_class)
	* 3. other metaslabs in normal vdevs (spa_normal_class)
	*
	* zfs_embedded_slog_min_ms disables the embedded slog if there are fewer
	* than this number of metaslabs in the vdev. This ensures that we don't set
	* aside an unreasonable amount of space for the ZIL. If set to less than
	* 1 << (spa_slop_shift + 1), on small pools the usable space may be reduced
	* (by more than 1<<spa_slop_shift) due to the embedded slog metaslab.
	*/
	int zfs_embedded_slog_min_ms = 64;

	/* default target for number of metaslabs per top-level vdev */
	int zfs_vdev_default_ms_count = 200;

	/* minimum number of metaslabs per top-level vdev */
	int zfs_vdev_min_ms_count = 16;

	/* practical upper limit of total metaslabs per top-level vdev */
	int zfs_vdev_ms_count_limit = 1ULL << 17;

	/* lower limit for metaslab size (512M) */
	int zfs_vdev_default_ms_shift = 29;

	/* upper limit for metaslab size (16G) */
	int zfs_vdev_max_ms_shift = 34;

	int vdev_validate_skip = B_FALSE;

	/*
	* Since the DTL space map of a vdev is not expected to have a lot of
	* entries, we default its block size to 4K.
	*/
	int zfs_vdev_dtl_sm_blksz = (1 << 12);

	/*
	* Rate limit slow IO (delay) events to this many per second.
	*/
	unsigned int zfs_slow_io_events_per_second = 20;

	/*
	* Rate limit checksum events after this many checksum errors per second.
	*/
	unsigned int zfs_checksum_events_per_second = 20;

	/*
	* Ignore errors during scrub/resilver. Allows to work around resilver
	* upon import when there are pool errors.
	*/
	int zfs_scan_ignore_errors = 0;

	/*
	* vdev-wide space maps that have lots of entries written to them at
	* the end of each transaction can benefit from a higher I/O bandwidth
	* (e.g. vdev_obsolete_sm), thus we default their block size to 128K.
	*/
	int zfs_vdev_standard_sm_blksz = (1 << 17);

	/*
	* Tunable parameter for debugging or performance analysis. Setting this
	* will cause pool corruption on power loss if a volatile out-of-order
	* write cache is enabled.
	*/
	int zfs_nocacheflush = 0;

	uint64_t zfs_vdev_max_auto_ashift = ASHIFT_MAX;
	uint64_t zfs_vdev_min_auto_ashift = ASHIFT_MIN;

	void
	vdev_dbgmsg(vdev_t vd, const char fmt, ...)
	{
	va_list adx;
	char buf[256];

	va_start(adx, fmt);
	(void) vsnprintf(buf, sizeof (buf), fmt, adx);
	va_end(adx);

	if (vd->vdev_path != NULL) {
	zfs_dbgmsg("%s vdev '%s': %s", vd->vdev_ops->vdev_op_type,
	vd->vdev_path, buf);
	} else {
	zfs_dbgmsg("%s-%llu vdev (guid %llu): %s",
	vd->vdev_ops->vdev_op_type,
	(u_longlong_t)vd->vdev_id,
	(u_longlong_t)vd->vdev_guid, buf);
	}
	}

	void
	vdev_dbgmsg_print_tree(vdev_t *vd, int indent)
	{
	char state[20];

	if (vd->vdev_ishole \|\| vd->vdev_ops == &vdev_missing_ops) {
	zfs_dbgmsg("%*svdev %llu: %s", indent, "",
	(u_longlong_t)vd->vdev_id,
	vd->vdev_ops->vdev_op_type);
	return;
	}

	switch (vd->vdev_state) {
	case VDEV_STATE_UNKNOWN:
	(void) snprintf(state, sizeof (state), "unknown");
	break;
	case VDEV_STATE_CLOSED:
	(void) snprintf(state, sizeof (state), "closed");
	break;
	case VDEV_STATE_OFFLINE:
	(void) snprintf(state, sizeof (state), "offline");
	break;
	case VDEV_STATE_REMOVED:
	(void) snprintf(state, sizeof (state), "removed");
	break;
	case VDEV_STATE_CANT_OPEN:
	(void) snprintf(state, sizeof (state), "can't open");
	break;
	case VDEV_STATE_FAULTED:
	(void) snprintf(state, sizeof (state), "faulted");
	break;
	case VDEV_STATE_DEGRADED:
	(void) snprintf(state, sizeof (state), "degraded");
	break;
	case VDEV_STATE_HEALTHY:
	(void) snprintf(state, sizeof (state), "healthy");
	break;
	default:
	(void) snprintf(state, sizeof (state), "<state %u>",
	(uint_t)vd->vdev_state);
	}

	zfs_dbgmsg("%*svdev %u: %s%s, guid: %llu, path: %s, %s", indent,
	"", (int)vd->vdev_id, vd->vdev_ops->vdev_op_type,
	vd->vdev_islog ? " (log)" : "",
	(u_longlong_t)vd->vdev_guid,
	vd->vdev_path ? vd->vdev_path : "N/A", state);

	for (uint64_t i = 0; i < vd->vdev_children; i++)
	vdev_dbgmsg_print_tree(vd->vdev_child[i], indent + 2);
	}

	/*
	* Virtual device management.
	*/

	static vdev_ops_t *vdev_ops_table[] = {
	&vdev_root_ops,
	&vdev_raidz_ops,
	&vdev_draid_ops,
	&vdev_draid_spare_ops,
	&vdev_mirror_ops,
	&vdev_replacing_ops,
	&vdev_spare_ops,
	&vdev_disk_ops,
	&vdev_file_ops,
	&vdev_missing_ops,
	&vdev_hole_ops,
	&vdev_indirect_ops,
	NULL
	};

	/*
	* Given a vdev type, return the appropriate ops vector.
	*/
	static vdev_ops_t *
	vdev_getops(const char *type)
	{
	vdev_ops_t ops, *opspp;

	for (opspp = vdev_ops_table; (ops = *opspp) != NULL; opspp++)
	if (strcmp(ops->vdev_op_type, type) == 0)
	break;

	return (ops);
	}

	/*
	* Given a vdev and a metaslab class, find which metaslab group we're
	* interested in. All vdevs may belong to two different metaslab classes.
	* Dedicated slog devices use only the primary metaslab group, rather than a
	* separate log group. For embedded slogs, the vdev_log_mg will be non-NULL.
	*/
	metaslab_group_t *
	vdev_get_mg(vdev_t vd, metaslab_class_t mc)
	{
	if (mc == spa_embedded_log_class(vd->vdev_spa) &&
	vd->vdev_log_mg != NULL)
	return (vd->vdev_log_mg);
	else
	return (vd->vdev_mg);
	}

	/* ARGSUSED */
	void
	vdev_default_xlate(vdev_t vd, const range_seg64_t logical_rs,
	range_seg64_t physical_rs, range_seg64_t remain_rs)
	{
	physical_rs->rs_start = logical_rs->rs_start;
	physical_rs->rs_end = logical_rs->rs_end;
	}

	/*
	* Derive the enumerated allocation bias from string input.
	* String origin is either the per-vdev zap or zpool(8).
	*/
	static vdev_alloc_bias_t
	vdev_derive_alloc_bias(const char *bias)
	{
	vdev_alloc_bias_t alloc_bias = VDEV_BIAS_NONE;

	if (strcmp(bias, VDEV_ALLOC_BIAS_LOG) == 0)
	alloc_bias = VDEV_BIAS_LOG;
	else if (strcmp(bias, VDEV_ALLOC_BIAS_SPECIAL) == 0)
	alloc_bias = VDEV_BIAS_SPECIAL;
	else if (strcmp(bias, VDEV_ALLOC_BIAS_DEDUP) == 0)
	alloc_bias = VDEV_BIAS_DEDUP;

	return (alloc_bias);
	}

	/*
	* Default asize function: return the MAX of psize with the asize of
	* all children. This is what's used by anything other than RAID-Z.
	*/
	uint64_t
	vdev_default_asize(vdev_t *vd, uint64_t psize)
	{
	uint64_t asize = P2ROUNDUP(psize, 1ULL << vd->vdev_top->vdev_ashift);
	uint64_t csize;

	for (int c = 0; c < vd->vdev_children; c++) {
	csize = vdev_psize_to_asize(vd->vdev_child[c], psize);
	asize = MAX(asize, csize);
	}

	return (asize);
	}

	uint64_t
	vdev_default_min_asize(vdev_t *vd)
	{
	return (vd->vdev_min_asize);
	}

	/*
	* Get the minimum allocatable size. We define the allocatable size as
	* the vdev's asize rounded to the nearest metaslab. This allows us to
	* replace or attach devices which don't have the same physical size but
	* can still satisfy the same number of allocations.
	*/
	uint64_t
	vdev_get_min_asize(vdev_t *vd)
	{
	vdev_t *pvd = vd->vdev_parent;

	/*
	* If our parent is NULL (inactive spare or cache) or is the root,
	* just return our own asize.
	*/
	if (pvd == NULL)
	return (vd->vdev_asize);

	/*
	* The top-level vdev just returns the allocatable size rounded
	* to the nearest metaslab.
	*/
	if (vd == vd->vdev_top)
	return (P2ALIGN(vd->vdev_asize, 1ULL << vd->vdev_ms_shift));

	return (pvd->vdev_ops->vdev_op_min_asize(pvd));
	}

	void
	vdev_set_min_asize(vdev_t *vd)
	{
	vd->vdev_min_asize = vdev_get_min_asize(vd);

	for (int c = 0; c < vd->vdev_children; c++)
	vdev_set_min_asize(vd->vdev_child[c]);
	}

	/*
	* Get the minimal allocation size for the top-level vdev.
	*/
	uint64_t
	vdev_get_min_alloc(vdev_t *vd)
	{
	uint64_t min_alloc = 1ULL << vd->vdev_ashift;

	if (vd->vdev_ops->vdev_op_min_alloc != NULL)
	min_alloc = vd->vdev_ops->vdev_op_min_alloc(vd);

	return (min_alloc);
	}

	/*
	* Get the parity level for a top-level vdev.
	*/
	uint64_t
	vdev_get_nparity(vdev_t *vd)
	{
	uint64_t nparity = 0;

	if (vd->vdev_ops->vdev_op_nparity != NULL)
	nparity = vd->vdev_ops->vdev_op_nparity(vd);

	return (nparity);
	}

	/*
	* Get the number of data disks for a top-level vdev.
	*/
	uint64_t
	vdev_get_ndisks(vdev_t *vd)
	{
	uint64_t ndisks = 1;

	if (vd->vdev_ops->vdev_op_ndisks != NULL)
	ndisks = vd->vdev_ops->vdev_op_ndisks(vd);

	return (ndisks);
	}

	vdev_t *
	vdev_lookup_top(spa_t *spa, uint64_t vdev)
	{
	vdev_t *rvd = spa->spa_root_vdev;

	ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0);

	if (vdev < rvd->vdev_children) {
	ASSERT(rvd->vdev_child[vdev] != NULL);
	return (rvd->vdev_child[vdev]);
	}

	return (NULL);
	}

	vdev_t *
	vdev_lookup_by_guid(vdev_t *vd, uint64_t guid)
	{
	vdev_t *mvd;

	if (vd->vdev_guid == guid)
	return (vd);

	for (int c = 0; c < vd->vdev_children; c++)
	if ((mvd = vdev_lookup_by_guid(vd->vdev_child[c], guid)) !=
	NULL)
	return (mvd);

	return (NULL);
	}

	static int
	vdev_count_leaves_impl(vdev_t *vd)
	{
	int n = 0;

	if (vd->vdev_ops->vdev_op_leaf)
	return (1);

	for (int c = 0; c < vd->vdev_children; c++)
	n += vdev_count_leaves_impl(vd->vdev_child[c]);

	return (n);
	}

	int
	vdev_count_leaves(spa_t *spa)
	{
	int rc;

	spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);
	rc = vdev_count_leaves_impl(spa->spa_root_vdev);
	spa_config_exit(spa, SCL_VDEV, FTAG);

	return (rc);
	}

	void
	vdev_add_child(vdev_t pvd, vdev_t cvd)
	{
	size_t oldsize, newsize;
	uint64_t id = cvd->vdev_id;
	vdev_t **newchild;

	ASSERT(spa_config_held(cvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL);
	ASSERT(cvd->vdev_parent == NULL);

	cvd->vdev_parent = pvd;

	if (pvd == NULL)
	return;

	ASSERT(id >= pvd->vdev_children \|\| pvd->vdev_child[id] == NULL);

	oldsize = pvd->vdev_children * sizeof (vdev_t *);
	pvd->vdev_children = MAX(pvd->vdev_children, id + 1);
	newsize = pvd->vdev_children * sizeof (vdev_t *);

	newchild = kmem_alloc(newsize, KM_SLEEP);
	if (pvd->vdev_child != NULL) {
	bcopy(pvd->vdev_child, newchild, oldsize);
	kmem_free(pvd->vdev_child, oldsize);
	}

	pvd->vdev_child = newchild;
	pvd->vdev_child[id] = cvd;

	cvd->vdev_top = (pvd->vdev_top ? pvd->vdev_top: cvd);
	ASSERT(cvd->vdev_top->vdev_parent->vdev_parent == NULL);

	/*
	* Walk up all ancestors to update guid sum.
	*/
	for (; pvd != NULL; pvd = pvd->vdev_parent)
	pvd->vdev_guid_sum += cvd->vdev_guid_sum;

	if (cvd->vdev_ops->vdev_op_leaf) {
	list_insert_head(&cvd->vdev_spa->spa_leaf_list, cvd);
	cvd->vdev_spa->spa_leaf_list_gen++;
	}
	}

	void
	vdev_remove_child(vdev_t pvd, vdev_t cvd)
	{
	int c;
	uint_t id = cvd->vdev_id;

	ASSERT(cvd->vdev_parent == pvd);

	if (pvd == NULL)
	return;

	ASSERT(id < pvd->vdev_children);
	ASSERT(pvd->vdev_child[id] == cvd);

	pvd->vdev_child[id] = NULL;
	cvd->vdev_parent = NULL;

	for (c = 0; c < pvd->vdev_children; c++)
	if (pvd->vdev_child[c])
	break;

	if (c == pvd->vdev_children) {
	kmem_free(pvd->vdev_child, c * sizeof (vdev_t *));
	pvd->vdev_child = NULL;
	pvd->vdev_children = 0;
	}

	if (cvd->vdev_ops->vdev_op_leaf) {
	spa_t *spa = cvd->vdev_spa;
	list_remove(&spa->spa_leaf_list, cvd);
	spa->spa_leaf_list_gen++;
	}

	/*
	* Walk up all ancestors to update guid sum.
	*/
	for (; pvd != NULL; pvd = pvd->vdev_parent)
	pvd->vdev_guid_sum -= cvd->vdev_guid_sum;
	}

	/*
	* Remove any holes in the child array.
	*/
	void
	vdev_compact_children(vdev_t *pvd)
	{
	vdev_t *newchild, cvd;
	int oldc = pvd->vdev_children;
	int newc;

	ASSERT(spa_config_held(pvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	if (oldc == 0)
	return;

	for (int c = newc = 0; c < oldc; c++)
	if (pvd->vdev_child[c])
	newc++;

	if (newc > 0) {
	newchild = kmem_zalloc(newc * sizeof (vdev_t *), KM_SLEEP);

	for (int c = newc = 0; c < oldc; c++) {
	if ((cvd = pvd->vdev_child[c]) != NULL) {
	newchild[newc] = cvd;
	cvd->vdev_id = newc++;
	}
	}
	} else {
	newchild = NULL;
	}

	kmem_free(pvd->vdev_child, oldc * sizeof (vdev_t *));
	pvd->vdev_child = newchild;
	pvd->vdev_children = newc;
	}

	/*
	* Allocate and minimally initialize a vdev_t.
	*/
	vdev_t *
	vdev_alloc_common(spa_t spa, uint_t id, uint64_t guid, vdev_ops_t ops)
	{
	vdev_t *vd;
	vdev_indirect_config_t *vic;

	vd = kmem_zalloc(sizeof (vdev_t), KM_SLEEP);
	vic = &vd->vdev_indirect_config;

	if (spa->spa_root_vdev == NULL) {
	ASSERT(ops == &vdev_root_ops);
	spa->spa_root_vdev = vd;
	spa->spa_load_guid = spa_generate_guid(NULL);
	}

	if (guid == 0 && ops != &vdev_hole_ops) {
	if (spa->spa_root_vdev == vd) {
	/*
	* The root vdev's guid will also be the pool guid,
	* which must be unique among all pools.
	*/
	guid = spa_generate_guid(NULL);
	} else {
	/*
	* Any other vdev's guid must be unique within the pool.
	*/
	guid = spa_generate_guid(spa);
	}
	ASSERT(!spa_guid_exists(spa_guid(spa), guid));
	}

	vd->vdev_spa = spa;
	vd->vdev_id = id;
	vd->vdev_guid = guid;
	vd->vdev_guid_sum = guid;
	vd->vdev_ops = ops;
	vd->vdev_state = VDEV_STATE_CLOSED;
	vd->vdev_ishole = (ops == &vdev_hole_ops);
	vic->vic_prev_indirect_vdev = UINT64_MAX;

	rw_init(&vd->vdev_indirect_rwlock, NULL, RW_DEFAULT, NULL);
	mutex_init(&vd->vdev_obsolete_lock, NULL, MUTEX_DEFAULT, NULL);
	vd->vdev_obsolete_segments = range_tree_create(NULL, RANGE_SEG64, NULL,
	0, 0);

	/*
	* Initialize rate limit structs for events. We rate limit ZIO delay
	* and checksum events so that we don't overwhelm ZED with thousands
	* of events when a disk is acting up.
	*/
	zfs_ratelimit_init(&vd->vdev_delay_rl, &zfs_slow_io_events_per_second,
	1);
	zfs_ratelimit_init(&vd->vdev_deadman_rl, &zfs_slow_io_events_per_second,
	1);
	zfs_ratelimit_init(&vd->vdev_checksum_rl,
	&zfs_checksum_events_per_second, 1);

	list_link_init(&vd->vdev_config_dirty_node);
	list_link_init(&vd->vdev_state_dirty_node);
	list_link_init(&vd->vdev_initialize_node);
	list_link_init(&vd->vdev_leaf_node);
	list_link_init(&vd->vdev_trim_node);

	mutex_init(&vd->vdev_dtl_lock, NULL, MUTEX_NOLOCKDEP, NULL);
	mutex_init(&vd->vdev_stat_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&vd->vdev_probe_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&vd->vdev_scan_io_queue_lock, NULL, MUTEX_DEFAULT, NULL);

	mutex_init(&vd->vdev_initialize_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&vd->vdev_initialize_io_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&vd->vdev_initialize_cv, NULL, CV_DEFAULT, NULL);
	cv_init(&vd->vdev_initialize_io_cv, NULL, CV_DEFAULT, NULL);

	mutex_init(&vd->vdev_trim_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&vd->vdev_autotrim_lock, NULL, MUTEX_DEFAULT, NULL);
	mutex_init(&vd->vdev_trim_io_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&vd->vdev_trim_cv, NULL, CV_DEFAULT, NULL);
	cv_init(&vd->vdev_autotrim_cv, NULL, CV_DEFAULT, NULL);
	cv_init(&vd->vdev_trim_io_cv, NULL, CV_DEFAULT, NULL);

	mutex_init(&vd->vdev_rebuild_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&vd->vdev_rebuild_cv, NULL, CV_DEFAULT, NULL);

	for (int t = 0; t < DTL_TYPES; t++) {
	vd->vdev_dtl[t] = range_tree_create(NULL, RANGE_SEG64, NULL, 0,
	0);
	}

	txg_list_create(&vd->vdev_ms_list, spa,
	offsetof(struct metaslab, ms_txg_node));
	txg_list_create(&vd->vdev_dtl_list, spa,
	offsetof(struct vdev, vdev_dtl_node));
	vd->vdev_stat.vs_timestamp = gethrtime();
	vdev_queue_init(vd);
	vdev_cache_init(vd);

	return (vd);
	}

	/*
	* Allocate a new vdev. The 'alloctype' is used to control whether we are
	* creating a new vdev or loading an existing one - the behavior is slightly
	* different for each case.
	*/
	int
	vdev_alloc(spa_t spa, vdev_t vdp, nvlist_t nv, vdev_t *parent, uint_t id,
	int alloctype)
	{
	vdev_ops_t *ops;
	char *type;
	uint64_t guid = 0, islog;
	vdev_t *vd;
	vdev_indirect_config_t *vic;
	char *tmp = NULL;
	int rc;
	vdev_alloc_bias_t alloc_bias = VDEV_BIAS_NONE;
	boolean_t top_level = (parent && !parent->vdev_parent);

	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type) != 0)
	return (SET_ERROR(EINVAL));

	if ((ops = vdev_getops(type)) == NULL)
	return (SET_ERROR(EINVAL));

	/*
	* If this is a load, get the vdev guid from the nvlist.
	* Otherwise, vdev_alloc_common() will generate one for us.
	*/
	if (alloctype == VDEV_ALLOC_LOAD) {
	uint64_t label_id;

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ID, &label_id) \|\|
	label_id != id)
	return (SET_ERROR(EINVAL));

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0)
	return (SET_ERROR(EINVAL));
	} else if (alloctype == VDEV_ALLOC_SPARE) {
	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0)
	return (SET_ERROR(EINVAL));
	} else if (alloctype == VDEV_ALLOC_L2CACHE) {
	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0)
	return (SET_ERROR(EINVAL));
	} else if (alloctype == VDEV_ALLOC_ROOTPOOL) {
	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_GUID, &guid) != 0)
	return (SET_ERROR(EINVAL));
	}

	/*
	* The first allocated vdev must be of type 'root'.
	*/
	if (ops != &vdev_root_ops && spa->spa_root_vdev == NULL)
	return (SET_ERROR(EINVAL));

	/*
	* Determine whether we're a log vdev.
	*/
	islog = 0;
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &islog);
	if (islog && spa_version(spa) < SPA_VERSION_SLOGS)
	return (SET_ERROR(ENOTSUP));

	if (ops == &vdev_hole_ops && spa_version(spa) < SPA_VERSION_HOLES)
	return (SET_ERROR(ENOTSUP));

	if (top_level && alloctype == VDEV_ALLOC_ADD) {
	char *bias;

	/*
	* If creating a top-level vdev, check for allocation
	* classes input.
	*/
	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_ALLOCATION_BIAS,
	&bias) == 0) {
	alloc_bias = vdev_derive_alloc_bias(bias);

	/* spa_vdev_add() expects feature to be enabled */
	if (spa->spa_load_state != SPA_LOAD_CREATE &&
	!spa_feature_is_enabled(spa,
	SPA_FEATURE_ALLOCATION_CLASSES)) {
	return (SET_ERROR(ENOTSUP));
	}
	}

	/* spa_vdev_add() expects feature to be enabled */
	if (ops == &vdev_draid_ops &&
	spa->spa_load_state != SPA_LOAD_CREATE &&
	!spa_feature_is_enabled(spa, SPA_FEATURE_DRAID)) {
	return (SET_ERROR(ENOTSUP));
	}
	}

	/*
	* Initialize the vdev specific data. This is done before calling
	* vdev_alloc_common() since it may fail and this simplifies the
	* error reporting and cleanup code paths.
	*/
	void *tsd = NULL;
	if (ops->vdev_op_init != NULL) {
	rc = ops->vdev_op_init(spa, nv, &tsd);
	if (rc != 0) {
	return (rc);
	}
	}

	vd = vdev_alloc_common(spa, id, guid, ops);
	vd->vdev_tsd = tsd;
	vd->vdev_islog = islog;

	if (top_level && alloc_bias != VDEV_BIAS_NONE)
	vd->vdev_alloc_bias = alloc_bias;

	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &vd->vdev_path) == 0)
	vd->vdev_path = spa_strdup(vd->vdev_path);

	/*
	* ZPOOL_CONFIG_AUX_STATE = "external" means we previously forced a
	* fault on a vdev and want it to persist across imports (like with
	* zpool offline -f).
	*/
	rc = nvlist_lookup_string(nv, ZPOOL_CONFIG_AUX_STATE, &tmp);
	if (rc == 0 && tmp != NULL && strcmp(tmp, "external") == 0) {
	vd->vdev_stat.vs_aux = VDEV_AUX_EXTERNAL;
	vd->vdev_faulted = 1;
	vd->vdev_label_aux = VDEV_AUX_EXTERNAL;
	}

	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_DEVID, &vd->vdev_devid) == 0)
	vd->vdev_devid = spa_strdup(vd->vdev_devid);
	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_PHYS_PATH,
	&vd->vdev_physpath) == 0)
	vd->vdev_physpath = spa_strdup(vd->vdev_physpath);

	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_VDEV_ENC_SYSFS_PATH,
	&vd->vdev_enc_sysfs_path) == 0)
	vd->vdev_enc_sysfs_path = spa_strdup(vd->vdev_enc_sysfs_path);

	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_FRU, &vd->vdev_fru) == 0)
	vd->vdev_fru = spa_strdup(vd->vdev_fru);

	/*
	* Set the whole_disk property. If it's not specified, leave the value
	* as -1.
	*/
	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK,
	&vd->vdev_wholedisk) != 0)
	vd->vdev_wholedisk = -1ULL;

	vic = &vd->vdev_indirect_config;

	ASSERT0(vic->vic_mapping_object);
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_INDIRECT_OBJECT,
	&vic->vic_mapping_object);
	ASSERT0(vic->vic_births_object);
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_INDIRECT_BIRTHS,
	&vic->vic_births_object);
	ASSERT3U(vic->vic_prev_indirect_vdev, ==, UINT64_MAX);
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_PREV_INDIRECT_VDEV,
	&vic->vic_prev_indirect_vdev);

	/*
	* Look for the 'not present' flag. This will only be set if the device
	* was not present at the time of import.
	*/
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT,
	&vd->vdev_not_present);

	/*
	* Get the alignment requirement.
	*/
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ASHIFT, &vd->vdev_ashift);

	/*
	* Retrieve the vdev creation time.
	*/
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_CREATE_TXG,
	&vd->vdev_crtxg);

	/*
	* If we're a top-level vdev, try to load the allocation parameters.
	*/
	if (top_level &&
	(alloctype == VDEV_ALLOC_LOAD \|\| alloctype == VDEV_ALLOC_SPLIT)) {
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_METASLAB_ARRAY,
	&vd->vdev_ms_array);
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_METASLAB_SHIFT,
	&vd->vdev_ms_shift);
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_ASIZE,
	&vd->vdev_asize);
	+ (void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NONALLOCATING,
	+ &vd->vdev_noalloc);
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REMOVING,
	&vd->vdev_removing);
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_VDEV_TOP_ZAP,
	&vd->vdev_top_zap);
	} else {
	ASSERT0(vd->vdev_top_zap);
	}

	if (top_level && alloctype != VDEV_ALLOC_ATTACH) {
	ASSERT(alloctype == VDEV_ALLOC_LOAD \|\|
	alloctype == VDEV_ALLOC_ADD \|\|
	alloctype == VDEV_ALLOC_SPLIT \|\|
	alloctype == VDEV_ALLOC_ROOTPOOL);
	/* Note: metaslab_group_create() is now deferred */
	}

	if (vd->vdev_ops->vdev_op_leaf &&
	(alloctype == VDEV_ALLOC_LOAD \|\| alloctype == VDEV_ALLOC_SPLIT)) {
	(void) nvlist_lookup_uint64(nv,
	ZPOOL_CONFIG_VDEV_LEAF_ZAP, &vd->vdev_leaf_zap);
	} else {
	ASSERT0(vd->vdev_leaf_zap);
	}

	/*
	* If we're a leaf vdev, try to load the DTL object and other state.
	*/

	if (vd->vdev_ops->vdev_op_leaf &&
	(alloctype == VDEV_ALLOC_LOAD \|\| alloctype == VDEV_ALLOC_L2CACHE \|\|
	alloctype == VDEV_ALLOC_ROOTPOOL)) {
	if (alloctype == VDEV_ALLOC_LOAD) {
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DTL,
	&vd->vdev_dtl_object);
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_UNSPARE,
	&vd->vdev_unspare);
	}

	if (alloctype == VDEV_ALLOC_ROOTPOOL) {
	uint64_t spare = 0;

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_SPARE,
	&spare) == 0 && spare)
	spa_spare_add(vd);
	}

	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_OFFLINE,
	&vd->vdev_offline);

	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_RESILVER_TXG,
	&vd->vdev_resilver_txg);

	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REBUILD_TXG,
	&vd->vdev_rebuild_txg);

	if (nvlist_exists(nv, ZPOOL_CONFIG_RESILVER_DEFER))
	vdev_defer_resilver(vd);

	/*
	* In general, when importing a pool we want to ignore the
	* persistent fault state, as the diagnosis made on another
	* system may not be valid in the current context. The only
	* exception is if we forced a vdev to a persistently faulted
	* state with 'zpool offline -f'. The persistent fault will
	* remain across imports until cleared.
	*
	* Local vdevs will remain in the faulted state.
	*/
	if (spa_load_state(spa) == SPA_LOAD_OPEN \|\|
	spa_load_state(spa) == SPA_LOAD_IMPORT) {
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_FAULTED,
	&vd->vdev_faulted);
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DEGRADED,
	&vd->vdev_degraded);
	(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_REMOVED,
	&vd->vdev_removed);

	if (vd->vdev_faulted \|\| vd->vdev_degraded) {
	char *aux;

	vd->vdev_label_aux =
	VDEV_AUX_ERR_EXCEEDED;
	if (nvlist_lookup_string(nv,
	ZPOOL_CONFIG_AUX_STATE, &aux) == 0 &&
	strcmp(aux, "external") == 0)
	vd->vdev_label_aux = VDEV_AUX_EXTERNAL;
	else
	vd->vdev_faulted = 0ULL;
	}
	}
	}

	/*
	* Add ourselves to the parent's list of children.
	*/
	vdev_add_child(parent, vd);

	*vdp = vd;

	return (0);
	}

	void
	vdev_free(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;

	ASSERT3P(vd->vdev_initialize_thread, ==, NULL);
	ASSERT3P(vd->vdev_trim_thread, ==, NULL);
	ASSERT3P(vd->vdev_autotrim_thread, ==, NULL);
	ASSERT3P(vd->vdev_rebuild_thread, ==, NULL);

	/*
	* Scan queues are normally destroyed at the end of a scan. If the
	* queue exists here, that implies the vdev is being removed while
	* the scan is still running.
	*/
	if (vd->vdev_scan_io_queue != NULL) {
	mutex_enter(&vd->vdev_scan_io_queue_lock);
	dsl_scan_io_queue_destroy(vd->vdev_scan_io_queue);
	vd->vdev_scan_io_queue = NULL;
	mutex_exit(&vd->vdev_scan_io_queue_lock);
	}

	/*
	* vdev_free() implies closing the vdev first. This is simpler than
	* trying to ensure complicated semantics for all callers.
	*/
	vdev_close(vd);

	ASSERT(!list_link_active(&vd->vdev_config_dirty_node));
	ASSERT(!list_link_active(&vd->vdev_state_dirty_node));

	/*
	* Free all children.
	*/
	for (int c = 0; c < vd->vdev_children; c++)
	vdev_free(vd->vdev_child[c]);

	ASSERT(vd->vdev_child == NULL);
	ASSERT(vd->vdev_guid_sum == vd->vdev_guid);

	if (vd->vdev_ops->vdev_op_fini != NULL)
	vd->vdev_ops->vdev_op_fini(vd);

	/*
	* Discard allocation state.
	*/
	if (vd->vdev_mg != NULL) {
	vdev_metaslab_fini(vd);
	metaslab_group_destroy(vd->vdev_mg);
	vd->vdev_mg = NULL;
	}
	if (vd->vdev_log_mg != NULL) {
	ASSERT0(vd->vdev_ms_count);
	metaslab_group_destroy(vd->vdev_log_mg);
	vd->vdev_log_mg = NULL;
	}

	ASSERT0(vd->vdev_stat.vs_space);
	ASSERT0(vd->vdev_stat.vs_dspace);
	ASSERT0(vd->vdev_stat.vs_alloc);

	/*
	* Remove this vdev from its parent's child list.
	*/
	vdev_remove_child(vd->vdev_parent, vd);

	ASSERT(vd->vdev_parent == NULL);
	ASSERT(!list_link_active(&vd->vdev_leaf_node));

	/*
	* Clean up vdev structure.
	*/
	vdev_queue_fini(vd);
	vdev_cache_fini(vd);

	if (vd->vdev_path)
	spa_strfree(vd->vdev_path);
	if (vd->vdev_devid)
	spa_strfree(vd->vdev_devid);
	if (vd->vdev_physpath)
	spa_strfree(vd->vdev_physpath);

	if (vd->vdev_enc_sysfs_path)
	spa_strfree(vd->vdev_enc_sysfs_path);

	if (vd->vdev_fru)
	spa_strfree(vd->vdev_fru);

	if (vd->vdev_isspare)
	spa_spare_remove(vd);
	if (vd->vdev_isl2cache)
	spa_l2cache_remove(vd);

	txg_list_destroy(&vd->vdev_ms_list);
	txg_list_destroy(&vd->vdev_dtl_list);

	mutex_enter(&vd->vdev_dtl_lock);
	space_map_close(vd->vdev_dtl_sm);
	for (int t = 0; t < DTL_TYPES; t++) {
	range_tree_vacate(vd->vdev_dtl[t], NULL, NULL);
	range_tree_destroy(vd->vdev_dtl[t]);
	}
	mutex_exit(&vd->vdev_dtl_lock);

	EQUIV(vd->vdev_indirect_births != NULL,
	vd->vdev_indirect_mapping != NULL);
	if (vd->vdev_indirect_births != NULL) {
	vdev_indirect_mapping_close(vd->vdev_indirect_mapping);
	vdev_indirect_births_close(vd->vdev_indirect_births);
	}

	if (vd->vdev_obsolete_sm != NULL) {
	ASSERT(vd->vdev_removing \|\|
	vd->vdev_ops == &vdev_indirect_ops);
	space_map_close(vd->vdev_obsolete_sm);
	vd->vdev_obsolete_sm = NULL;
	}
	range_tree_destroy(vd->vdev_obsolete_segments);
	rw_destroy(&vd->vdev_indirect_rwlock);
	mutex_destroy(&vd->vdev_obsolete_lock);

	mutex_destroy(&vd->vdev_dtl_lock);
	mutex_destroy(&vd->vdev_stat_lock);
	mutex_destroy(&vd->vdev_probe_lock);
	mutex_destroy(&vd->vdev_scan_io_queue_lock);

	mutex_destroy(&vd->vdev_initialize_lock);
	mutex_destroy(&vd->vdev_initialize_io_lock);
	cv_destroy(&vd->vdev_initialize_io_cv);
	cv_destroy(&vd->vdev_initialize_cv);

	mutex_destroy(&vd->vdev_trim_lock);
	mutex_destroy(&vd->vdev_autotrim_lock);
	mutex_destroy(&vd->vdev_trim_io_lock);
	cv_destroy(&vd->vdev_trim_cv);
	cv_destroy(&vd->vdev_autotrim_cv);
	cv_destroy(&vd->vdev_trim_io_cv);

	mutex_destroy(&vd->vdev_rebuild_lock);
	cv_destroy(&vd->vdev_rebuild_cv);

	zfs_ratelimit_fini(&vd->vdev_delay_rl);
	zfs_ratelimit_fini(&vd->vdev_deadman_rl);
	zfs_ratelimit_fini(&vd->vdev_checksum_rl);

	if (vd == spa->spa_root_vdev)
	spa->spa_root_vdev = NULL;

	kmem_free(vd, sizeof (vdev_t));
	}

	/*
	* Transfer top-level vdev state from svd to tvd.
	*/
	static void
	vdev_top_transfer(vdev_t svd, vdev_t tvd)
	{
	spa_t *spa = svd->vdev_spa;
	metaslab_t *msp;
	vdev_t *vd;
	int t;

	ASSERT(tvd == tvd->vdev_top);

	tvd->vdev_pending_fastwrite = svd->vdev_pending_fastwrite;
	tvd->vdev_ms_array = svd->vdev_ms_array;
	tvd->vdev_ms_shift = svd->vdev_ms_shift;
	tvd->vdev_ms_count = svd->vdev_ms_count;
	tvd->vdev_top_zap = svd->vdev_top_zap;

	svd->vdev_ms_array = 0;
	svd->vdev_ms_shift = 0;
	svd->vdev_ms_count = 0;
	svd->vdev_top_zap = 0;

	if (tvd->vdev_mg)
	ASSERT3P(tvd->vdev_mg, ==, svd->vdev_mg);
	if (tvd->vdev_log_mg)
	ASSERT3P(tvd->vdev_log_mg, ==, svd->vdev_log_mg);
	tvd->vdev_mg = svd->vdev_mg;
	tvd->vdev_log_mg = svd->vdev_log_mg;
	tvd->vdev_ms = svd->vdev_ms;

	svd->vdev_mg = NULL;
	svd->vdev_log_mg = NULL;
	svd->vdev_ms = NULL;

	if (tvd->vdev_mg != NULL)
	tvd->vdev_mg->mg_vd = tvd;
	if (tvd->vdev_log_mg != NULL)
	tvd->vdev_log_mg->mg_vd = tvd;

	tvd->vdev_checkpoint_sm = svd->vdev_checkpoint_sm;
	svd->vdev_checkpoint_sm = NULL;

	tvd->vdev_alloc_bias = svd->vdev_alloc_bias;
	svd->vdev_alloc_bias = VDEV_BIAS_NONE;

	tvd->vdev_stat.vs_alloc = svd->vdev_stat.vs_alloc;
	tvd->vdev_stat.vs_space = svd->vdev_stat.vs_space;
	tvd->vdev_stat.vs_dspace = svd->vdev_stat.vs_dspace;

	svd->vdev_stat.vs_alloc = 0;
	svd->vdev_stat.vs_space = 0;
	svd->vdev_stat.vs_dspace = 0;

	/*
	* State which may be set on a top-level vdev that's in the
	* process of being removed.
	*/
	ASSERT0(tvd->vdev_indirect_config.vic_births_object);
	ASSERT0(tvd->vdev_indirect_config.vic_mapping_object);
	ASSERT3U(tvd->vdev_indirect_config.vic_prev_indirect_vdev, ==, -1ULL);
	ASSERT3P(tvd->vdev_indirect_mapping, ==, NULL);
	ASSERT3P(tvd->vdev_indirect_births, ==, NULL);
	ASSERT3P(tvd->vdev_obsolete_sm, ==, NULL);
	+ ASSERT0(tvd->vdev_noalloc);
	ASSERT0(tvd->vdev_removing);
	ASSERT0(tvd->vdev_rebuilding);
	+ tvd->vdev_noalloc = svd->vdev_noalloc;
	tvd->vdev_removing = svd->vdev_removing;
	tvd->vdev_rebuilding = svd->vdev_rebuilding;
	tvd->vdev_rebuild_config = svd->vdev_rebuild_config;
	tvd->vdev_indirect_config = svd->vdev_indirect_config;
	tvd->vdev_indirect_mapping = svd->vdev_indirect_mapping;
	tvd->vdev_indirect_births = svd->vdev_indirect_births;
	range_tree_swap(&svd->vdev_obsolete_segments,
	&tvd->vdev_obsolete_segments);
	tvd->vdev_obsolete_sm = svd->vdev_obsolete_sm;
	svd->vdev_indirect_config.vic_mapping_object = 0;
	svd->vdev_indirect_config.vic_births_object = 0;
	svd->vdev_indirect_config.vic_prev_indirect_vdev = -1ULL;
	svd->vdev_indirect_mapping = NULL;
	svd->vdev_indirect_births = NULL;
	svd->vdev_obsolete_sm = NULL;
	+ svd->vdev_noalloc = 0;
	svd->vdev_removing = 0;
	svd->vdev_rebuilding = 0;

	for (t = 0; t < TXG_SIZE; t++) {
	while ((msp = txg_list_remove(&svd->vdev_ms_list, t)) != NULL)
	(void) txg_list_add(&tvd->vdev_ms_list, msp, t);
	while ((vd = txg_list_remove(&svd->vdev_dtl_list, t)) != NULL)
	(void) txg_list_add(&tvd->vdev_dtl_list, vd, t);
	if (txg_list_remove_this(&spa->spa_vdev_txg_list, svd, t))
	(void) txg_list_add(&spa->spa_vdev_txg_list, tvd, t);
	}

	if (list_link_active(&svd->vdev_config_dirty_node)) {
	vdev_config_clean(svd);
	vdev_config_dirty(tvd);
	}

	if (list_link_active(&svd->vdev_state_dirty_node)) {
	vdev_state_clean(svd);
	vdev_state_dirty(tvd);
	}

	tvd->vdev_deflate_ratio = svd->vdev_deflate_ratio;
	svd->vdev_deflate_ratio = 0;

	tvd->vdev_islog = svd->vdev_islog;
	svd->vdev_islog = 0;

	dsl_scan_io_queue_vdev_xfer(svd, tvd);
	}

	static void
	vdev_top_update(vdev_t tvd, vdev_t vd)
	{
	if (vd == NULL)
	return;

	vd->vdev_top = tvd;

	for (int c = 0; c < vd->vdev_children; c++)
	vdev_top_update(tvd, vd->vdev_child[c]);
	}

	/*
	* Add a mirror/replacing vdev above an existing vdev. There is no need to
	* call .vdev_op_init() since mirror/replacing vdevs do not have private state.
	*/
	vdev_t *
	vdev_add_parent(vdev_t cvd, vdev_ops_t ops)
	{
	spa_t *spa = cvd->vdev_spa;
	vdev_t *pvd = cvd->vdev_parent;
	vdev_t *mvd;

	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	mvd = vdev_alloc_common(spa, cvd->vdev_id, 0, ops);

	mvd->vdev_asize = cvd->vdev_asize;
	mvd->vdev_min_asize = cvd->vdev_min_asize;
	mvd->vdev_max_asize = cvd->vdev_max_asize;
	mvd->vdev_psize = cvd->vdev_psize;
	mvd->vdev_ashift = cvd->vdev_ashift;
	mvd->vdev_logical_ashift = cvd->vdev_logical_ashift;
	mvd->vdev_physical_ashift = cvd->vdev_physical_ashift;
	mvd->vdev_state = cvd->vdev_state;
	mvd->vdev_crtxg = cvd->vdev_crtxg;

	vdev_remove_child(pvd, cvd);
	vdev_add_child(pvd, mvd);
	cvd->vdev_id = mvd->vdev_children;
	vdev_add_child(mvd, cvd);
	vdev_top_update(cvd->vdev_top, cvd->vdev_top);

	if (mvd == mvd->vdev_top)
	vdev_top_transfer(cvd, mvd);

	return (mvd);
	}

	/*
	* Remove a 1-way mirror/replacing vdev from the tree.
	*/
	void
	vdev_remove_parent(vdev_t *cvd)
	{
	vdev_t *mvd = cvd->vdev_parent;
	vdev_t *pvd = mvd->vdev_parent;

	ASSERT(spa_config_held(cvd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	ASSERT(mvd->vdev_children == 1);
	ASSERT(mvd->vdev_ops == &vdev_mirror_ops \|\|
	mvd->vdev_ops == &vdev_replacing_ops \|\|
	mvd->vdev_ops == &vdev_spare_ops);
	cvd->vdev_ashift = mvd->vdev_ashift;
	cvd->vdev_logical_ashift = mvd->vdev_logical_ashift;
	cvd->vdev_physical_ashift = mvd->vdev_physical_ashift;
	vdev_remove_child(mvd, cvd);
	vdev_remove_child(pvd, mvd);

	/*
	* If cvd will replace mvd as a top-level vdev, preserve mvd's guid.
	* Otherwise, we could have detached an offline device, and when we
	* go to import the pool we'll think we have two top-level vdevs,
	* instead of a different version of the same top-level vdev.
	*/
	if (mvd->vdev_top == mvd) {
	uint64_t guid_delta = mvd->vdev_guid - cvd->vdev_guid;
	cvd->vdev_orig_guid = cvd->vdev_guid;
	cvd->vdev_guid += guid_delta;
	cvd->vdev_guid_sum += guid_delta;

	/*
	* If pool not set for autoexpand, we need to also preserve
	* mvd's asize to prevent automatic expansion of cvd.
	* Otherwise if we are adjusting the mirror by attaching and
	* detaching children of non-uniform sizes, the mirror could
	* autoexpand, unexpectedly requiring larger devices to
	* re-establish the mirror.
	*/
	if (!cvd->vdev_spa->spa_autoexpand)
	cvd->vdev_asize = mvd->vdev_asize;
	}
	cvd->vdev_id = mvd->vdev_id;
	vdev_add_child(pvd, cvd);
	vdev_top_update(cvd->vdev_top, cvd->vdev_top);

	if (cvd == cvd->vdev_top)
	vdev_top_transfer(mvd, cvd);

	ASSERT(mvd->vdev_children == 0);
	vdev_free(mvd);
	}

	void
	vdev_metaslab_group_create(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;

	/*
	* metaslab_group_create was delayed until allocation bias was available
	*/
	if (vd->vdev_mg == NULL) {
	metaslab_class_t *mc;

	if (vd->vdev_islog && vd->vdev_alloc_bias == VDEV_BIAS_NONE)
	vd->vdev_alloc_bias = VDEV_BIAS_LOG;

	ASSERT3U(vd->vdev_islog, ==,
	(vd->vdev_alloc_bias == VDEV_BIAS_LOG));

	switch (vd->vdev_alloc_bias) {
	case VDEV_BIAS_LOG:
	mc = spa_log_class(spa);
	break;
	case VDEV_BIAS_SPECIAL:
	mc = spa_special_class(spa);
	break;
	case VDEV_BIAS_DEDUP:
	mc = spa_dedup_class(spa);
	break;
	default:
	mc = spa_normal_class(spa);
	}

	vd->vdev_mg = metaslab_group_create(mc, vd,
	spa->spa_alloc_count);

	if (!vd->vdev_islog) {
	vd->vdev_log_mg = metaslab_group_create(
	spa_embedded_log_class(spa), vd, 1);
	}

	/*
	* The spa ashift min/max only apply for the normal metaslab
	* class. Class destination is late binding so ashift boundary
	* setting had to wait until now.
	*/
	if (vd->vdev_top == vd && vd->vdev_ashift != 0 &&
	mc == spa_normal_class(spa) && vd->vdev_aux == NULL) {
	if (vd->vdev_ashift > spa->spa_max_ashift)
	spa->spa_max_ashift = vd->vdev_ashift;
	if (vd->vdev_ashift < spa->spa_min_ashift)
	spa->spa_min_ashift = vd->vdev_ashift;

	uint64_t min_alloc = vdev_get_min_alloc(vd);
	if (min_alloc < spa->spa_min_alloc)
	spa->spa_min_alloc = min_alloc;
	}
	}
	}

	int
	vdev_metaslab_init(vdev_t *vd, uint64_t txg)
	{
	spa_t *spa = vd->vdev_spa;
	uint64_t oldc = vd->vdev_ms_count;
	uint64_t newc = vd->vdev_asize >> vd->vdev_ms_shift;
	metaslab_t **mspp;
	int error;
	boolean_t expanding = (oldc != 0);

	ASSERT(txg == 0 \|\| spa_config_held(spa, SCL_ALLOC, RW_WRITER));

	/*
	* This vdev is not being allocated from yet or is a hole.
	*/
	if (vd->vdev_ms_shift == 0)
	return (0);

	ASSERT(!vd->vdev_ishole);

	ASSERT(oldc <= newc);

	mspp = vmem_zalloc(newc * sizeof (*mspp), KM_SLEEP);

	if (expanding) {
	bcopy(vd->vdev_ms, mspp, oldc * sizeof (*mspp));
	vmem_free(vd->vdev_ms, oldc * sizeof (*mspp));
	}

	vd->vdev_ms = mspp;
	vd->vdev_ms_count = newc;

	for (uint64_t m = oldc; m < newc; m++) {
	uint64_t object = 0;
	/*
	* vdev_ms_array may be 0 if we are creating the "fake"
	* metaslabs for an indirect vdev for zdb's leak detection.
	* See zdb_leak_init().
	*/
	if (txg == 0 && vd->vdev_ms_array != 0) {
	error = dmu_read(spa->spa_meta_objset,
	vd->vdev_ms_array,
	m * sizeof (uint64_t), sizeof (uint64_t), &object,
	DMU_READ_PREFETCH);
	if (error != 0) {
	vdev_dbgmsg(vd, "unable to read the metaslab "
	"array [error=%d]", error);
	return (error);
	}
	}

	error = metaslab_init(vd->vdev_mg, m, object, txg,
	&(vd->vdev_ms[m]));
	if (error != 0) {
	vdev_dbgmsg(vd, "metaslab_init failed [error=%d]",
	error);
	return (error);
	}
	}

	/*
	* Find the emptiest metaslab on the vdev and mark it for use for
	* embedded slog by moving it from the regular to the log metaslab
	* group.
	*/
	if (vd->vdev_mg->mg_class == spa_normal_class(spa) &&
	vd->vdev_ms_count > zfs_embedded_slog_min_ms &&
	avl_is_empty(&vd->vdev_log_mg->mg_metaslab_tree)) {
	uint64_t slog_msid = 0;
	uint64_t smallest = UINT64_MAX;

	/*
	* Note, we only search the new metaslabs, because the old
	* (pre-existing) ones may be active (e.g. have non-empty
	* range_tree's), and we don't move them to the new
	* metaslab_t.
	*/
	for (uint64_t m = oldc; m < newc; m++) {
	uint64_t alloc =
	space_map_allocated(vd->vdev_ms[m]->ms_sm);
	if (alloc < smallest) {
	slog_msid = m;
	smallest = alloc;
	}
	}
	metaslab_t *slog_ms = vd->vdev_ms[slog_msid];
	/*
	* The metaslab was marked as dirty at the end of
	* metaslab_init(). Remove it from the dirty list so that we
	* can uninitialize and reinitialize it to the new class.
	*/
	if (txg != 0) {
	(void) txg_list_remove_this(&vd->vdev_ms_list,
	slog_ms, txg);
	}
	uint64_t sm_obj = space_map_object(slog_ms->ms_sm);
	metaslab_fini(slog_ms);
	VERIFY0(metaslab_init(vd->vdev_log_mg, slog_msid, sm_obj, txg,
	&vd->vdev_ms[slog_msid]));
	}

	if (txg == 0)
	spa_config_enter(spa, SCL_ALLOC, FTAG, RW_WRITER);

	/*
	- * If the vdev is being removed we don't activate
	- * the metaslabs since we want to ensure that no new
	- * allocations are performed on this device.
	+ * If the vdev is marked as non-allocating then don't
	+ * activate the metaslabs since we want to ensure that
	+ * no allocations are performed on this device.
	*/
	- if (!expanding && !vd->vdev_removing) {
	+ if (vd->vdev_noalloc) {
	+ /* track non-allocating vdev space */
	+ spa->spa_nonallocating_dspace += spa_deflate(spa) ?
	+ vd->vdev_stat.vs_dspace : vd->vdev_stat.vs_space;
	+ } else if (!expanding) {
	metaslab_group_activate(vd->vdev_mg);
	if (vd->vdev_log_mg != NULL)
	metaslab_group_activate(vd->vdev_log_mg);
	}

	if (txg == 0)
	spa_config_exit(spa, SCL_ALLOC, FTAG);

	/*
	* Regardless whether this vdev was just added or it is being
	* expanded, the metaslab count has changed. Recalculate the
	* block limit.
	*/
	spa_log_sm_set_blocklimit(spa);

	return (0);
	}

	void
	vdev_metaslab_fini(vdev_t *vd)
	{
	if (vd->vdev_checkpoint_sm != NULL) {
	ASSERT(spa_feature_is_active(vd->vdev_spa,
	SPA_FEATURE_POOL_CHECKPOINT));
	space_map_close(vd->vdev_checkpoint_sm);
	/*
	* Even though we close the space map, we need to set its
	* pointer to NULL. The reason is that vdev_metaslab_fini()
	* may be called multiple times for certain operations
	* (i.e. when destroying a pool) so we need to ensure that
	* this clause never executes twice. This logic is similar
	* to the one used for the vdev_ms clause below.
	*/
	vd->vdev_checkpoint_sm = NULL;
	}

	if (vd->vdev_ms != NULL) {
	metaslab_group_t *mg = vd->vdev_mg;

	metaslab_group_passivate(mg);
	if (vd->vdev_log_mg != NULL) {
	ASSERT(!vd->vdev_islog);
	metaslab_group_passivate(vd->vdev_log_mg);
	}

	uint64_t count = vd->vdev_ms_count;
	for (uint64_t m = 0; m < count; m++) {
	metaslab_t *msp = vd->vdev_ms[m];
	if (msp != NULL)
	metaslab_fini(msp);
	}
	vmem_free(vd->vdev_ms, count * sizeof (metaslab_t *));
	vd->vdev_ms = NULL;
	vd->vdev_ms_count = 0;

	for (int i = 0; i < RANGE_TREE_HISTOGRAM_SIZE; i++) {
	ASSERT0(mg->mg_histogram[i]);
	if (vd->vdev_log_mg != NULL)
	ASSERT0(vd->vdev_log_mg->mg_histogram[i]);
	}
	}
	ASSERT0(vd->vdev_ms_count);
	ASSERT3U(vd->vdev_pending_fastwrite, ==, 0);
	}

	typedef struct vdev_probe_stats {
	boolean_t vps_readable;
	boolean_t vps_writeable;
	int vps_flags;
	} vdev_probe_stats_t;

	static void
	vdev_probe_done(zio_t *zio)
	{
	spa_t *spa = zio->io_spa;
	vdev_t *vd = zio->io_vd;
	vdev_probe_stats_t *vps = zio->io_private;

	ASSERT(vd->vdev_probe_zio != NULL);

	if (zio->io_type == ZIO_TYPE_READ) {
	if (zio->io_error == 0)
	vps->vps_readable = 1;
	if (zio->io_error == 0 && spa_writeable(spa)) {
	zio_nowait(zio_write_phys(vd->vdev_probe_zio, vd,
	zio->io_offset, zio->io_size, zio->io_abd,
	ZIO_CHECKSUM_OFF, vdev_probe_done, vps,
	ZIO_PRIORITY_SYNC_WRITE, vps->vps_flags, B_TRUE));
	} else {
	abd_free(zio->io_abd);
	}
	} else if (zio->io_type == ZIO_TYPE_WRITE) {
	if (zio->io_error == 0)
	vps->vps_writeable = 1;
	abd_free(zio->io_abd);
	} else if (zio->io_type == ZIO_TYPE_NULL) {
	zio_t *pio;
	zio_link_t *zl;

	vd->vdev_cant_read \|= !vps->vps_readable;
	vd->vdev_cant_write \|= !vps->vps_writeable;

	if (vdev_readable(vd) &&
	(vdev_writeable(vd) \|\| !spa_writeable(spa))) {
	zio->io_error = 0;
	} else {
	ASSERT(zio->io_error != 0);
	vdev_dbgmsg(vd, "failed probe");
	(void) zfs_ereport_post(FM_EREPORT_ZFS_PROBE_FAILURE,
	spa, vd, NULL, NULL, 0);
	zio->io_error = SET_ERROR(ENXIO);
	}

	mutex_enter(&vd->vdev_probe_lock);
	ASSERT(vd->vdev_probe_zio == zio);
	vd->vdev_probe_zio = NULL;
	mutex_exit(&vd->vdev_probe_lock);

	zl = NULL;
	while ((pio = zio_walk_parents(zio, &zl)) != NULL)
	if (!vdev_accessible(vd, pio))
	pio->io_error = SET_ERROR(ENXIO);

	kmem_free(vps, sizeof (*vps));
	}
	}

	/*
	* Determine whether this device is accessible.
	*
	* Read and write to several known locations: the pad regions of each
	* vdev label but the first, which we leave alone in case it contains
	* a VTOC.
	*/
	zio_t *
	vdev_probe(vdev_t vd, zio_t zio)
	{
	spa_t *spa = vd->vdev_spa;
	vdev_probe_stats_t *vps = NULL;
	zio_t *pio;

	ASSERT(vd->vdev_ops->vdev_op_leaf);

	/*
	* Don't probe the probe.
	*/
	if (zio && (zio->io_flags & ZIO_FLAG_PROBE))
	return (NULL);

	/*
	* To prevent 'probe storms' when a device fails, we create
	* just one probe i/o at a time. All zios that want to probe
	* this vdev will become parents of the probe io.
	*/
	mutex_enter(&vd->vdev_probe_lock);

	if ((pio = vd->vdev_probe_zio) == NULL) {
	vps = kmem_zalloc(sizeof (*vps), KM_SLEEP);

	vps->vps_flags = ZIO_FLAG_CANFAIL \| ZIO_FLAG_PROBE \|
	ZIO_FLAG_DONT_CACHE \| ZIO_FLAG_DONT_AGGREGATE \|
	ZIO_FLAG_TRYHARD;

	if (spa_config_held(spa, SCL_ZIO, RW_WRITER)) {
	/*
	* vdev_cant_read and vdev_cant_write can only
	* transition from TRUE to FALSE when we have the
	* SCL_ZIO lock as writer; otherwise they can only
	* transition from FALSE to TRUE. This ensures that
	* any zio looking at these values can assume that
	* failures persist for the life of the I/O. That's
	* important because when a device has intermittent
	* connectivity problems, we want to ensure that
	* they're ascribed to the device (ENXIO) and not
	* the zio (EIO).
	*
	* Since we hold SCL_ZIO as writer here, clear both
	* values so the probe can reevaluate from first
	* principles.
	*/
	vps->vps_flags \|= ZIO_FLAG_CONFIG_WRITER;
	vd->vdev_cant_read = B_FALSE;
	vd->vdev_cant_write = B_FALSE;
	}

	vd->vdev_probe_zio = pio = zio_null(NULL, spa, vd,
	vdev_probe_done, vps,
	vps->vps_flags \| ZIO_FLAG_DONT_PROPAGATE);

	/*
	* We can't change the vdev state in this context, so we
	* kick off an async task to do it on our behalf.
	*/
	if (zio != NULL) {
	vd->vdev_probe_wanted = B_TRUE;
	spa_async_request(spa, SPA_ASYNC_PROBE);
	}
	}

	if (zio != NULL)
	zio_add_child(zio, pio);

	mutex_exit(&vd->vdev_probe_lock);

	if (vps == NULL) {
	ASSERT(zio != NULL);
	return (NULL);
	}

	for (int l = 1; l < VDEV_LABELS; l++) {
	zio_nowait(zio_read_phys(pio, vd,
	vdev_label_offset(vd->vdev_psize, l,
	offsetof(vdev_label_t, vl_be)), VDEV_PAD_SIZE,
	abd_alloc_for_io(VDEV_PAD_SIZE, B_TRUE),
	ZIO_CHECKSUM_OFF, vdev_probe_done, vps,
	ZIO_PRIORITY_SYNC_READ, vps->vps_flags, B_TRUE));
	}

	if (zio == NULL)
	return (pio);

	zio_nowait(pio);
	return (NULL);
	}

	static void
	vdev_load_child(void *arg)
	{
	vdev_t *vd = arg;

	vd->vdev_load_error = vdev_load(vd);
	}

	static void
	vdev_open_child(void *arg)
	{
	vdev_t *vd = arg;

	vd->vdev_open_thread = curthread;
	vd->vdev_open_error = vdev_open(vd);
	vd->vdev_open_thread = NULL;
	}

	static boolean_t
	vdev_uses_zvols(vdev_t *vd)
	{
	#ifdef _KERNEL
	if (zvol_is_zvol(vd->vdev_path))
	return (B_TRUE);
	#endif

	for (int c = 0; c < vd->vdev_children; c++)
	if (vdev_uses_zvols(vd->vdev_child[c]))
	return (B_TRUE);

	return (B_FALSE);
	}

	/*
	* Returns B_TRUE if the passed child should be opened.
	*/
	static boolean_t
	vdev_default_open_children_func(vdev_t *vd)
	{
	return (B_TRUE);
	}

	/*
	* Open the requested child vdevs. If any of the leaf vdevs are using
	* a ZFS volume then do the opens in a single thread. This avoids a
	* deadlock when the current thread is holding the spa_namespace_lock.
	*/
	static void
	vdev_open_children_impl(vdev_t vd, vdev_open_children_func_t open_func)
	{
	int children = vd->vdev_children;

	taskq_t *tq = taskq_create("vdev_open", children, minclsyspri,
	children, children, TASKQ_PREPOPULATE);
	vd->vdev_nonrot = B_TRUE;

	for (int c = 0; c < children; c++) {
	vdev_t *cvd = vd->vdev_child[c];

	if (open_func(cvd) == B_FALSE)
	continue;

	if (tq == NULL \|\| vdev_uses_zvols(vd)) {
	cvd->vdev_open_error = vdev_open(cvd);
	} else {
	VERIFY(taskq_dispatch(tq, vdev_open_child,
	cvd, TQ_SLEEP) != TASKQID_INVALID);
	}

	vd->vdev_nonrot &= cvd->vdev_nonrot;
	}

	if (tq != NULL) {
	taskq_wait(tq);
	taskq_destroy(tq);
	}
	}

	/*
	* Open all child vdevs.
	*/
	void
	vdev_open_children(vdev_t *vd)
	{
	vdev_open_children_impl(vd, vdev_default_open_children_func);
	}

	/*
	* Conditionally open a subset of child vdevs.
	*/
	void
	vdev_open_children_subset(vdev_t vd, vdev_open_children_func_t open_func)
	{
	vdev_open_children_impl(vd, open_func);
	}

	/*
	* Compute the raidz-deflation ratio. Note, we hard-code
	* in 128k (1 << 17) because it is the "typical" blocksize.
	* Even though SPA_MAXBLOCKSIZE changed, this algorithm can not change,
	* otherwise it would inconsistently account for existing bp's.
	*/
	static void
	vdev_set_deflate_ratio(vdev_t *vd)
	{
	if (vd == vd->vdev_top && !vd->vdev_ishole && vd->vdev_ashift != 0) {
	vd->vdev_deflate_ratio = (1 << 17) /
	(vdev_psize_to_asize(vd, 1 << 17) >> SPA_MINBLOCKSHIFT);
	}
	}

	/*
	* Maximize performance by inflating the configured ashift for top level
	* vdevs to be as close to the physical ashift as possible while maintaining
	* administrator defined limits and ensuring it doesn't go below the
	* logical ashift.
	*/
	static void
	vdev_ashift_optimize(vdev_t *vd)
	{
	ASSERT(vd == vd->vdev_top);

	if (vd->vdev_ashift < vd->vdev_physical_ashift) {
	vd->vdev_ashift = MIN(
	MAX(zfs_vdev_max_auto_ashift, vd->vdev_ashift),
	MAX(zfs_vdev_min_auto_ashift,
	vd->vdev_physical_ashift));
	} else {
	/*
	* If the logical and physical ashifts are the same, then
	* we ensure that the top-level vdev's ashift is not smaller
	* than our minimum ashift value. For the unusual case
	* where logical ashift > physical ashift, we can't cap
	* the calculated ashift based on max ashift as that
	* would cause failures.
	* We still check if we need to increase it to match
	* the min ashift.
	*/
	vd->vdev_ashift = MAX(zfs_vdev_min_auto_ashift,
	vd->vdev_ashift);
	}
	}

	/*
	* Prepare a virtual device for access.
	*/
	int
	vdev_open(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;
	int error;
	uint64_t osize = 0;
	uint64_t max_osize = 0;
	uint64_t asize, max_asize, psize;
	uint64_t logical_ashift = 0;
	uint64_t physical_ashift = 0;

	ASSERT(vd->vdev_open_thread == curthread \|\|
	spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL);
	ASSERT(vd->vdev_state == VDEV_STATE_CLOSED \|\|
	vd->vdev_state == VDEV_STATE_CANT_OPEN \|\|
	vd->vdev_state == VDEV_STATE_OFFLINE);

	vd->vdev_stat.vs_aux = VDEV_AUX_NONE;
	vd->vdev_cant_read = B_FALSE;
	vd->vdev_cant_write = B_FALSE;
	vd->vdev_min_asize = vdev_get_min_asize(vd);

	/*
	* If this vdev is not removed, check its fault status. If it's
	* faulted, bail out of the open.
	*/
	if (!vd->vdev_removed && vd->vdev_faulted) {
	ASSERT(vd->vdev_children == 0);
	ASSERT(vd->vdev_label_aux == VDEV_AUX_ERR_EXCEEDED \|\|
	vd->vdev_label_aux == VDEV_AUX_EXTERNAL);
	vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED,
	vd->vdev_label_aux);
	return (SET_ERROR(ENXIO));
	} else if (vd->vdev_offline) {
	ASSERT(vd->vdev_children == 0);
	vdev_set_state(vd, B_TRUE, VDEV_STATE_OFFLINE, VDEV_AUX_NONE);
	return (SET_ERROR(ENXIO));
	}

	error = vd->vdev_ops->vdev_op_open(vd, &osize, &max_osize,
	&logical_ashift, &physical_ashift);
	/*
	* Physical volume size should never be larger than its max size, unless
	* the disk has shrunk while we were reading it or the device is buggy
	* or damaged: either way it's not safe for use, bail out of the open.
	*/
	if (osize > max_osize) {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_OPEN_FAILED);
	return (SET_ERROR(ENXIO));
	}

	/*
	* Reset the vdev_reopening flag so that we actually close
	* the vdev on error.
	*/
	vd->vdev_reopening = B_FALSE;
	if (zio_injection_enabled && error == 0)
	error = zio_handle_device_injection(vd, NULL, SET_ERROR(ENXIO));

	if (error) {
	if (vd->vdev_removed &&
	vd->vdev_stat.vs_aux != VDEV_AUX_OPEN_FAILED)
	vd->vdev_removed = B_FALSE;

	if (vd->vdev_stat.vs_aux == VDEV_AUX_CHILDREN_OFFLINE) {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_OFFLINE,
	vd->vdev_stat.vs_aux);
	} else {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN,
	vd->vdev_stat.vs_aux);
	}
	return (error);
	}

	vd->vdev_removed = B_FALSE;

	/*
	* Recheck the faulted flag now that we have confirmed that
	* the vdev is accessible. If we're faulted, bail.
	*/
	if (vd->vdev_faulted) {
	ASSERT(vd->vdev_children == 0);
	ASSERT(vd->vdev_label_aux == VDEV_AUX_ERR_EXCEEDED \|\|
	vd->vdev_label_aux == VDEV_AUX_EXTERNAL);
	vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED,
	vd->vdev_label_aux);
	return (SET_ERROR(ENXIO));
	}

	if (vd->vdev_degraded) {
	ASSERT(vd->vdev_children == 0);
	vdev_set_state(vd, B_TRUE, VDEV_STATE_DEGRADED,
	VDEV_AUX_ERR_EXCEEDED);
	} else {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_HEALTHY, 0);
	}

	/*
	* For hole or missing vdevs we just return success.
	*/
	if (vd->vdev_ishole \|\| vd->vdev_ops == &vdev_missing_ops)
	return (0);

	for (int c = 0; c < vd->vdev_children; c++) {
	if (vd->vdev_child[c]->vdev_state != VDEV_STATE_HEALTHY) {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_DEGRADED,
	VDEV_AUX_NONE);
	break;
	}
	}

	osize = P2ALIGN(osize, (uint64_t)sizeof (vdev_label_t));
	max_osize = P2ALIGN(max_osize, (uint64_t)sizeof (vdev_label_t));

	if (vd->vdev_children == 0) {
	if (osize < SPA_MINDEVSIZE) {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_TOO_SMALL);
	return (SET_ERROR(EOVERFLOW));
	}
	psize = osize;
	asize = osize - (VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE);
	max_asize = max_osize - (VDEV_LABEL_START_SIZE +
	VDEV_LABEL_END_SIZE);
	} else {
	if (vd->vdev_parent != NULL && osize < SPA_MINDEVSIZE -
	(VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE)) {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_TOO_SMALL);
	return (SET_ERROR(EOVERFLOW));
	}
	psize = 0;
	asize = osize;
	max_asize = max_osize;
	}

	/*
	* If the vdev was expanded, record this so that we can re-create the
	* uberblock rings in labels {2,3}, during the next sync.
	*/
	if ((psize > vd->vdev_psize) && (vd->vdev_psize != 0))
	vd->vdev_copy_uberblocks = B_TRUE;

	vd->vdev_psize = psize;

	/*
	* Make sure the allocatable size hasn't shrunk too much.
	*/
	if (asize < vd->vdev_min_asize) {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_BAD_LABEL);
	return (SET_ERROR(EINVAL));
	}

	/*
	* We can always set the logical/physical ashift members since
	* their values are only used to calculate the vdev_ashift when
	* the device is first added to the config. These values should
	* not be used for anything else since they may change whenever
	* the device is reopened and we don't store them in the label.
	*/
	vd->vdev_physical_ashift =
	MAX(physical_ashift, vd->vdev_physical_ashift);
	vd->vdev_logical_ashift = MAX(logical_ashift,
	vd->vdev_logical_ashift);

	if (vd->vdev_asize == 0) {
	/*
	* This is the first-ever open, so use the computed values.
	* For compatibility, a different ashift can be requested.
	*/
	vd->vdev_asize = asize;
	vd->vdev_max_asize = max_asize;

	/*
	* If the vdev_ashift was not overridden at creation time,
	* then set it the logical ashift and optimize the ashift.
	*/
	if (vd->vdev_ashift == 0) {
	vd->vdev_ashift = vd->vdev_logical_ashift;

	if (vd->vdev_logical_ashift > ASHIFT_MAX) {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_ASHIFT_TOO_BIG);
	return (SET_ERROR(EDOM));
	}

	if (vd->vdev_top == vd) {
	vdev_ashift_optimize(vd);
	}
	}
	if (vd->vdev_ashift != 0 && (vd->vdev_ashift < ASHIFT_MIN \|\|
	vd->vdev_ashift > ASHIFT_MAX)) {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_BAD_ASHIFT);
	return (SET_ERROR(EDOM));
	}
	} else {
	/*
	* Make sure the alignment required hasn't increased.
	*/
	if (vd->vdev_ashift > vd->vdev_top->vdev_ashift &&
	vd->vdev_ops->vdev_op_leaf) {
	(void) zfs_ereport_post(
	FM_EREPORT_ZFS_DEVICE_BAD_ASHIFT,
	spa, vd, NULL, NULL, 0);
	vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_BAD_LABEL);
	return (SET_ERROR(EDOM));
	}
	vd->vdev_max_asize = max_asize;
	}

	/*
	* If all children are healthy we update asize if either:
	* The asize has increased, due to a device expansion caused by dynamic
	* LUN growth or vdev replacement, and automatic expansion is enabled;
	* making the additional space available.
	*
	* The asize has decreased, due to a device shrink usually caused by a
	* vdev replace with a smaller device. This ensures that calculations
	* based of max_asize and asize e.g. esize are always valid. It's safe
	* to do this as we've already validated that asize is greater than
	* vdev_min_asize.
	*/
	if (vd->vdev_state == VDEV_STATE_HEALTHY &&
	((asize > vd->vdev_asize &&
	(vd->vdev_expanding \|\| spa->spa_autoexpand)) \|\|
	(asize < vd->vdev_asize)))
	vd->vdev_asize = asize;

	vdev_set_min_asize(vd);

	/*
	* Ensure we can issue some IO before declaring the
	* vdev open for business.
	*/
	if (vd->vdev_ops->vdev_op_leaf &&
	(error = zio_wait(vdev_probe(vd, NULL))) != 0) {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_FAULTED,
	VDEV_AUX_ERR_EXCEEDED);
	return (error);
	}

	/*
	* Track the minimum allocation size.
	*/
	if (vd->vdev_top == vd && vd->vdev_ashift != 0 &&
	vd->vdev_islog == 0 && vd->vdev_aux == NULL) {
	uint64_t min_alloc = vdev_get_min_alloc(vd);
	if (min_alloc < spa->spa_min_alloc)
	spa->spa_min_alloc = min_alloc;
	}

	/*
	* If this is a leaf vdev, assess whether a resilver is needed.
	* But don't do this if we are doing a reopen for a scrub, since
	* this would just restart the scrub we are already doing.
	*/
	if (vd->vdev_ops->vdev_op_leaf && !spa->spa_scrub_reopen)
	dsl_scan_assess_vdev(spa->spa_dsl_pool, vd);

	return (0);
	}

	static void
	vdev_validate_child(void *arg)
	{
	vdev_t *vd = arg;

	vd->vdev_validate_thread = curthread;
	vd->vdev_validate_error = vdev_validate(vd);
	vd->vdev_validate_thread = NULL;
	}

	/*
	* Called once the vdevs are all opened, this routine validates the label
	* contents. This needs to be done before vdev_load() so that we don't
	* inadvertently do repair I/Os to the wrong device.
	*
	* This function will only return failure if one of the vdevs indicates that it
	* has since been destroyed or exported. This is only possible if
	* /etc/zfs/zpool.cache was readonly at the time. Otherwise, the vdev state
	* will be updated but the function will return 0.
	*/
	int
	vdev_validate(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;
	taskq_t *tq = NULL;
	nvlist_t *label;
	uint64_t guid = 0, aux_guid = 0, top_guid;
	uint64_t state;
	nvlist_t *nvl;
	uint64_t txg;
	int children = vd->vdev_children;

	if (vdev_validate_skip)
	return (0);

	if (children > 0) {
	tq = taskq_create("vdev_validate", children, minclsyspri,
	children, children, TASKQ_PREPOPULATE);
	}

	for (uint64_t c = 0; c < children; c++) {
	vdev_t *cvd = vd->vdev_child[c];

	if (tq == NULL \|\| vdev_uses_zvols(cvd)) {
	vdev_validate_child(cvd);
	} else {
	VERIFY(taskq_dispatch(tq, vdev_validate_child, cvd,
	TQ_SLEEP) != TASKQID_INVALID);
	}
	}
	if (tq != NULL) {
	taskq_wait(tq);
	taskq_destroy(tq);
	}
	for (int c = 0; c < children; c++) {
	int error = vd->vdev_child[c]->vdev_validate_error;

	if (error != 0)
	return (SET_ERROR(EBADF));
	}


	/*
	* If the device has already failed, or was marked offline, don't do
	* any further validation. Otherwise, label I/O will fail and we will
	* overwrite the previous state.
	*/
	if (!vd->vdev_ops->vdev_op_leaf \|\| !vdev_readable(vd))
	return (0);

	/*
	* If we are performing an extreme rewind, we allow for a label that
	* was modified at a point after the current txg.
	* If config lock is not held do not check for the txg. spa_sync could
	* be updating the vdev's label before updating spa_last_synced_txg.
	*/
	if (spa->spa_extreme_rewind \|\| spa_last_synced_txg(spa) == 0 \|\|
	spa_config_held(spa, SCL_CONFIG, RW_WRITER) != SCL_CONFIG)
	txg = UINT64_MAX;
	else
	txg = spa_last_synced_txg(spa);

	if ((label = vdev_label_read_config(vd, txg)) == NULL) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_BAD_LABEL);
	vdev_dbgmsg(vd, "vdev_validate: failed reading config for "
	"txg %llu", (u_longlong_t)txg);
	return (0);
	}

	/*
	* Determine if this vdev has been split off into another
	* pool. If so, then refuse to open it.
	*/
	if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_SPLIT_GUID,
	&aux_guid) == 0 && aux_guid == spa_guid(spa)) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_SPLIT_POOL);
	nvlist_free(label);
	vdev_dbgmsg(vd, "vdev_validate: vdev split into other pool");
	return (0);
	}

	if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_GUID, &guid) != 0) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	nvlist_free(label);
	vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label",
	ZPOOL_CONFIG_POOL_GUID);
	return (0);
	}

	/*
	* If config is not trusted then ignore the spa guid check. This is
	* necessary because if the machine crashed during a re-guid the new
	* guid might have been written to all of the vdev labels, but not the
	* cached config. The check will be performed again once we have the
	* trusted config from the MOS.
	*/
	if (spa->spa_trust_config && guid != spa_guid(spa)) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	nvlist_free(label);
	vdev_dbgmsg(vd, "vdev_validate: vdev label pool_guid doesn't "
	"match config (%llu != %llu)", (u_longlong_t)guid,
	(u_longlong_t)spa_guid(spa));
	return (0);
	}

	if (nvlist_lookup_nvlist(label, ZPOOL_CONFIG_VDEV_TREE, &nvl)
	!= 0 \|\| nvlist_lookup_uint64(nvl, ZPOOL_CONFIG_ORIG_GUID,
	&aux_guid) != 0)
	aux_guid = 0;

	if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &guid) != 0) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	nvlist_free(label);
	vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label",
	ZPOOL_CONFIG_GUID);
	return (0);
	}

	if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_TOP_GUID, &top_guid)
	!= 0) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	nvlist_free(label);
	vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label",
	ZPOOL_CONFIG_TOP_GUID);
	return (0);
	}

	/*
	* If this vdev just became a top-level vdev because its sibling was
	* detached, it will have adopted the parent's vdev guid -- but the
	* label may or may not be on disk yet. Fortunately, either version
	* of the label will have the same top guid, so if we're a top-level
	* vdev, we can safely compare to that instead.
	* However, if the config comes from a cachefile that failed to update
	* after the detach, a top-level vdev will appear as a non top-level
	* vdev in the config. Also relax the constraints if we perform an
	* extreme rewind.
	*
	* If we split this vdev off instead, then we also check the
	* original pool's guid. We don't want to consider the vdev
	* corrupt if it is partway through a split operation.
	*/
	if (vd->vdev_guid != guid && vd->vdev_guid != aux_guid) {
	boolean_t mismatch = B_FALSE;
	if (spa->spa_trust_config && !spa->spa_extreme_rewind) {
	if (vd != vd->vdev_top \|\| vd->vdev_guid != top_guid)
	mismatch = B_TRUE;
	} else {
	if (vd->vdev_guid != top_guid &&
	vd->vdev_top->vdev_guid != guid)
	mismatch = B_TRUE;
	}

	if (mismatch) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	nvlist_free(label);
	vdev_dbgmsg(vd, "vdev_validate: config guid "
	"doesn't match label guid");
	vdev_dbgmsg(vd, "CONFIG: guid %llu, top_guid %llu",
	(u_longlong_t)vd->vdev_guid,
	(u_longlong_t)vd->vdev_top->vdev_guid);
	vdev_dbgmsg(vd, "LABEL: guid %llu, top_guid %llu, "
	"aux_guid %llu", (u_longlong_t)guid,
	(u_longlong_t)top_guid, (u_longlong_t)aux_guid);
	return (0);
	}
	}

	if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_STATE,
	&state) != 0) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	nvlist_free(label);
	vdev_dbgmsg(vd, "vdev_validate: '%s' missing from label",
	ZPOOL_CONFIG_POOL_STATE);
	return (0);
	}

	nvlist_free(label);

	/*
	* If this is a verbatim import, no need to check the
	* state of the pool.
	*/
	if (!(spa->spa_import_flags & ZFS_IMPORT_VERBATIM) &&
	spa_load_state(spa) == SPA_LOAD_OPEN &&
	state != POOL_STATE_ACTIVE) {
	vdev_dbgmsg(vd, "vdev_validate: invalid pool state (%llu) "
	"for spa %s", (u_longlong_t)state, spa->spa_name);
	return (SET_ERROR(EBADF));
	}

	/*
	* If we were able to open and validate a vdev that was
	* previously marked permanently unavailable, clear that state
	* now.
	*/
	if (vd->vdev_not_present)
	vd->vdev_not_present = 0;

	return (0);
	}

	static void
	vdev_copy_path_impl(vdev_t svd, vdev_t dvd)
	{
	char old, new;
	if (svd->vdev_path != NULL && dvd->vdev_path != NULL) {
	if (strcmp(svd->vdev_path, dvd->vdev_path) != 0) {
	zfs_dbgmsg("vdev_copy_path: vdev %llu: path changed "
	"from '%s' to '%s'", (u_longlong_t)dvd->vdev_guid,
	dvd->vdev_path, svd->vdev_path);
	spa_strfree(dvd->vdev_path);
	dvd->vdev_path = spa_strdup(svd->vdev_path);
	}
	} else if (svd->vdev_path != NULL) {
	dvd->vdev_path = spa_strdup(svd->vdev_path);
	zfs_dbgmsg("vdev_copy_path: vdev %llu: path set to '%s'",
	(u_longlong_t)dvd->vdev_guid, dvd->vdev_path);
	}

	/*
	* Our enclosure sysfs path may have changed between imports
	*/
	old = dvd->vdev_enc_sysfs_path;
	new = svd->vdev_enc_sysfs_path;
	if ((old != NULL && new == NULL) \|\|
	(old == NULL && new != NULL) \|\|
	((old != NULL && new != NULL) && strcmp(new, old) != 0)) {
	zfs_dbgmsg("vdev_copy_path: vdev %llu: vdev_enc_sysfs_path "
	"changed from '%s' to '%s'", (u_longlong_t)dvd->vdev_guid,
	old, new);

	if (dvd->vdev_enc_sysfs_path)
	spa_strfree(dvd->vdev_enc_sysfs_path);

	if (svd->vdev_enc_sysfs_path) {
	dvd->vdev_enc_sysfs_path = spa_strdup(
	svd->vdev_enc_sysfs_path);
	} else {
	dvd->vdev_enc_sysfs_path = NULL;
	}
	}
	}

	/*
	* Recursively copy vdev paths from one vdev to another. Source and destination
	* vdev trees must have same geometry otherwise return error. Intended to copy
	* paths from userland config into MOS config.
	*/
	int
	vdev_copy_path_strict(vdev_t svd, vdev_t dvd)
	{
	if ((svd->vdev_ops == &vdev_missing_ops) \|\|
	(svd->vdev_ishole && dvd->vdev_ishole) \|\|
	(dvd->vdev_ops == &vdev_indirect_ops))
	return (0);

	if (svd->vdev_ops != dvd->vdev_ops) {
	vdev_dbgmsg(svd, "vdev_copy_path: vdev type mismatch: %s != %s",
	svd->vdev_ops->vdev_op_type, dvd->vdev_ops->vdev_op_type);
	return (SET_ERROR(EINVAL));
	}

	if (svd->vdev_guid != dvd->vdev_guid) {
	vdev_dbgmsg(svd, "vdev_copy_path: guids mismatch (%llu != "
	"%llu)", (u_longlong_t)svd->vdev_guid,
	(u_longlong_t)dvd->vdev_guid);
	return (SET_ERROR(EINVAL));
	}

	if (svd->vdev_children != dvd->vdev_children) {
	vdev_dbgmsg(svd, "vdev_copy_path: children count mismatch: "
	"%llu != %llu", (u_longlong_t)svd->vdev_children,
	(u_longlong_t)dvd->vdev_children);
	return (SET_ERROR(EINVAL));
	}

	for (uint64_t i = 0; i < svd->vdev_children; i++) {
	int error = vdev_copy_path_strict(svd->vdev_child[i],
	dvd->vdev_child[i]);
	if (error != 0)
	return (error);
	}

	if (svd->vdev_ops->vdev_op_leaf)
	vdev_copy_path_impl(svd, dvd);

	return (0);
	}

	static void
	vdev_copy_path_search(vdev_t stvd, vdev_t dvd)
	{
	ASSERT(stvd->vdev_top == stvd);
	ASSERT3U(stvd->vdev_id, ==, dvd->vdev_top->vdev_id);

	for (uint64_t i = 0; i < dvd->vdev_children; i++) {
	vdev_copy_path_search(stvd, dvd->vdev_child[i]);
	}

	if (!dvd->vdev_ops->vdev_op_leaf \|\| !vdev_is_concrete(dvd))
	return;

	/*
	* The idea here is that while a vdev can shift positions within
	* a top vdev (when replacing, attaching mirror, etc.) it cannot
	* step outside of it.
	*/
	vdev_t *vd = vdev_lookup_by_guid(stvd, dvd->vdev_guid);

	if (vd == NULL \|\| vd->vdev_ops != dvd->vdev_ops)
	return;

	ASSERT(vd->vdev_ops->vdev_op_leaf);

	vdev_copy_path_impl(vd, dvd);
	}

	/*
	* Recursively copy vdev paths from one root vdev to another. Source and
	* destination vdev trees may differ in geometry. For each destination leaf
	* vdev, search a vdev with the same guid and top vdev id in the source.
	* Intended to copy paths from userland config into MOS config.
	*/
	void
	vdev_copy_path_relaxed(vdev_t srvd, vdev_t drvd)
	{
	uint64_t children = MIN(srvd->vdev_children, drvd->vdev_children);
	ASSERT(srvd->vdev_ops == &vdev_root_ops);
	ASSERT(drvd->vdev_ops == &vdev_root_ops);

	for (uint64_t i = 0; i < children; i++) {
	vdev_copy_path_search(srvd->vdev_child[i],
	drvd->vdev_child[i]);
	}
	}

	/*
	* Close a virtual device.
	*/
	void
	vdev_close(vdev_t *vd)
	{
	vdev_t *pvd = vd->vdev_parent;
	spa_t *spa __maybe_unused = vd->vdev_spa;

	ASSERT(vd != NULL);
	ASSERT(vd->vdev_open_thread == curthread \|\|
	spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL);

	/*
	* If our parent is reopening, then we are as well, unless we are
	* going offline.
	*/
	if (pvd != NULL && pvd->vdev_reopening)
	vd->vdev_reopening = (pvd->vdev_reopening && !vd->vdev_offline);

	vd->vdev_ops->vdev_op_close(vd);

	vdev_cache_purge(vd);

	/*
	* We record the previous state before we close it, so that if we are
	* doing a reopen(), we don't generate FMA ereports if we notice that
	* it's still faulted.
	*/
	vd->vdev_prevstate = vd->vdev_state;

	if (vd->vdev_offline)
	vd->vdev_state = VDEV_STATE_OFFLINE;
	else
	vd->vdev_state = VDEV_STATE_CLOSED;
	vd->vdev_stat.vs_aux = VDEV_AUX_NONE;
	}

	void
	vdev_hold(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;

	ASSERT(spa_is_root(spa));
	if (spa->spa_state == POOL_STATE_UNINITIALIZED)
	return;

	for (int c = 0; c < vd->vdev_children; c++)
	vdev_hold(vd->vdev_child[c]);

	if (vd->vdev_ops->vdev_op_leaf && vd->vdev_ops->vdev_op_hold != NULL)
	vd->vdev_ops->vdev_op_hold(vd);
	}

	void
	vdev_rele(vdev_t *vd)
	{
	ASSERT(spa_is_root(vd->vdev_spa));
	for (int c = 0; c < vd->vdev_children; c++)
	vdev_rele(vd->vdev_child[c]);

	if (vd->vdev_ops->vdev_op_leaf && vd->vdev_ops->vdev_op_rele != NULL)
	vd->vdev_ops->vdev_op_rele(vd);
	}

	/*
	* Reopen all interior vdevs and any unopened leaves. We don't actually
	* reopen leaf vdevs which had previously been opened as they might deadlock
	* on the spa_config_lock. Instead we only obtain the leaf's physical size.
	* If the leaf has never been opened then open it, as usual.
	*/
	void
	vdev_reopen(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;

	ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL);

	/* set the reopening flag unless we're taking the vdev offline */
	vd->vdev_reopening = !vd->vdev_offline;
	vdev_close(vd);
	(void) vdev_open(vd);

	/*
	* Call vdev_validate() here to make sure we have the same device.
	* Otherwise, a device with an invalid label could be successfully
	* opened in response to vdev_reopen().
	*/
	if (vd->vdev_aux) {
	(void) vdev_validate_aux(vd);
	if (vdev_readable(vd) && vdev_writeable(vd) &&
	vd->vdev_aux == &spa->spa_l2cache) {
	/*
	* In case the vdev is present we should evict all ARC
	* buffers and pointers to log blocks and reclaim their
	* space before restoring its contents to L2ARC.
	*/
	if (l2arc_vdev_present(vd)) {
	l2arc_rebuild_vdev(vd, B_TRUE);
	} else {
	l2arc_add_vdev(spa, vd);
	}
	spa_async_request(spa, SPA_ASYNC_L2CACHE_REBUILD);
	spa_async_request(spa, SPA_ASYNC_L2CACHE_TRIM);
	}
	} else {
	(void) vdev_validate(vd);
	}

	/*
	* Reassess parent vdev's health.
	*/
	vdev_propagate_state(vd);
	}

	int
	vdev_create(vdev_t *vd, uint64_t txg, boolean_t isreplacing)
	{
	int error;

	/*
	* Normally, partial opens (e.g. of a mirror) are allowed.
	* For a create, however, we want to fail the request if
	* there are any components we can't open.
	*/
	error = vdev_open(vd);

	if (error \|\| vd->vdev_state != VDEV_STATE_HEALTHY) {
	vdev_close(vd);
	return (error ? error : SET_ERROR(ENXIO));
	}

	/*
	* Recursively load DTLs and initialize all labels.
	*/
	if ((error = vdev_dtl_load(vd)) != 0 \|\|
	(error = vdev_label_init(vd, txg, isreplacing ?
	VDEV_LABEL_REPLACE : VDEV_LABEL_CREATE)) != 0) {
	vdev_close(vd);
	return (error);
	}

	return (0);
	}

	void
	vdev_metaslab_set_size(vdev_t *vd)
	{
	uint64_t asize = vd->vdev_asize;
	uint64_t ms_count = asize >> zfs_vdev_default_ms_shift;
	uint64_t ms_shift;

	/*
	* There are two dimensions to the metaslab sizing calculation:
	* the size of the metaslab and the count of metaslabs per vdev.
	*
	* The default values used below are a good balance between memory
	* usage (larger metaslab size means more memory needed for loaded
	* metaslabs; more metaslabs means more memory needed for the
	* metaslab_t structs), metaslab load time (larger metaslabs take
	* longer to load), and metaslab sync time (more metaslabs means
	* more time spent syncing all of them).
	*
	* In general, we aim for zfs_vdev_default_ms_count (200) metaslabs.
	* The range of the dimensions are as follows:
	*
	* 2^29 <= ms_size <= 2^34
	* 16 <= ms_count <= 131,072
	*
	* On the lower end of vdev sizes, we aim for metaslabs sizes of
	* at least 512MB (2^29) to minimize fragmentation effects when
	* testing with smaller devices. However, the count constraint
	* of at least 16 metaslabs will override this minimum size goal.
	*
	* On the upper end of vdev sizes, we aim for a maximum metaslab
	* size of 16GB. However, we will cap the total count to 2^17
	* metaslabs to keep our memory footprint in check and let the
	* metaslab size grow from there if that limit is hit.
	*
	* The net effect of applying above constrains is summarized below.
	*
	* vdev size metaslab count
	* --------------\|-----------------
	* < 8GB ~16
	* 8GB - 100GB one per 512MB
	* 100GB - 3TB ~200
	* 3TB - 2PB one per 16GB
	* > 2PB ~131,072
	* --------------------------------
	*
	* Finally, note that all of the above calculate the initial
	* number of metaslabs. Expanding a top-level vdev will result
	* in additional metaslabs being allocated making it possible
	* to exceed the zfs_vdev_ms_count_limit.
	*/

	if (ms_count < zfs_vdev_min_ms_count)
	ms_shift = highbit64(asize / zfs_vdev_min_ms_count);
	else if (ms_count > zfs_vdev_default_ms_count)
	ms_shift = highbit64(asize / zfs_vdev_default_ms_count);
	else
	ms_shift = zfs_vdev_default_ms_shift;

	if (ms_shift < SPA_MAXBLOCKSHIFT) {
	ms_shift = SPA_MAXBLOCKSHIFT;
	} else if (ms_shift > zfs_vdev_max_ms_shift) {
	ms_shift = zfs_vdev_max_ms_shift;
	/* cap the total count to constrain memory footprint */
	if ((asize >> ms_shift) > zfs_vdev_ms_count_limit)
	ms_shift = highbit64(asize / zfs_vdev_ms_count_limit);
	}

	vd->vdev_ms_shift = ms_shift;
	ASSERT3U(vd->vdev_ms_shift, >=, SPA_MAXBLOCKSHIFT);
	}

	void
	vdev_dirty(vdev_t vd, int flags, void arg, uint64_t txg)
	{
	ASSERT(vd == vd->vdev_top);
	/* indirect vdevs don't have metaslabs or dtls */
	ASSERT(vdev_is_concrete(vd) \|\| flags == 0);
	ASSERT(ISP2(flags));
	ASSERT(spa_writeable(vd->vdev_spa));

	if (flags & VDD_METASLAB)
	(void) txg_list_add(&vd->vdev_ms_list, arg, txg);

	if (flags & VDD_DTL)
	(void) txg_list_add(&vd->vdev_dtl_list, arg, txg);

	(void) txg_list_add(&vd->vdev_spa->spa_vdev_txg_list, vd, txg);
	}

	void
	vdev_dirty_leaves(vdev_t *vd, int flags, uint64_t txg)
	{
	for (int c = 0; c < vd->vdev_children; c++)
	vdev_dirty_leaves(vd->vdev_child[c], flags, txg);

	if (vd->vdev_ops->vdev_op_leaf)
	vdev_dirty(vd->vdev_top, flags, vd, txg);
	}

	/*
	* DTLs.
	*
	* A vdev's DTL (dirty time log) is the set of transaction groups for which
	* the vdev has less than perfect replication. There are four kinds of DTL:
	*
	* DTL_MISSING: txgs for which the vdev has no valid copies of the data
	*
	* DTL_PARTIAL: txgs for which data is available, but not fully replicated
	*
	* DTL_SCRUB: the txgs that could not be repaired by the last scrub; upon
	* scrub completion, DTL_SCRUB replaces DTL_MISSING in the range of
	* txgs that was scrubbed.
	*
	* DTL_OUTAGE: txgs which cannot currently be read, whether due to
	* persistent errors or just some device being offline.
	* Unlike the other three, the DTL_OUTAGE map is not generally
	* maintained; it's only computed when needed, typically to
	* determine whether a device can be detached.
	*
	* For leaf vdevs, DTL_MISSING and DTL_PARTIAL are identical: the device
	* either has the data or it doesn't.
	*
	* For interior vdevs such as mirror and RAID-Z the picture is more complex.
	* A vdev's DTL_PARTIAL is the union of its children's DTL_PARTIALs, because
	* if any child is less than fully replicated, then so is its parent.
	* A vdev's DTL_MISSING is a modified union of its children's DTL_MISSINGs,
	* comprising only those txgs which appear in 'maxfaults' or more children;
	* those are the txgs we don't have enough replication to read. For example,
	* double-parity RAID-Z can tolerate up to two missing devices (maxfaults == 2);
	* thus, its DTL_MISSING consists of the set of txgs that appear in more than
	* two child DTL_MISSING maps.
	*
	* It should be clear from the above that to compute the DTLs and outage maps
	* for all vdevs, it suffices to know just the leaf vdevs' DTL_MISSING maps.
	* Therefore, that is all we keep on disk. When loading the pool, or after
	* a configuration change, we generate all other DTLs from first principles.
	*/
	void
	vdev_dtl_dirty(vdev_t *vd, vdev_dtl_type_t t, uint64_t txg, uint64_t size)
	{
	range_tree_t *rt = vd->vdev_dtl[t];

	ASSERT(t < DTL_TYPES);
	ASSERT(vd != vd->vdev_spa->spa_root_vdev);
	ASSERT(spa_writeable(vd->vdev_spa));

	mutex_enter(&vd->vdev_dtl_lock);
	if (!range_tree_contains(rt, txg, size))
	range_tree_add(rt, txg, size);
	mutex_exit(&vd->vdev_dtl_lock);
	}

	boolean_t
	vdev_dtl_contains(vdev_t *vd, vdev_dtl_type_t t, uint64_t txg, uint64_t size)
	{
	range_tree_t *rt = vd->vdev_dtl[t];
	boolean_t dirty = B_FALSE;

	ASSERT(t < DTL_TYPES);
	ASSERT(vd != vd->vdev_spa->spa_root_vdev);

	/*
	* While we are loading the pool, the DTLs have not been loaded yet.
	* This isn't a problem but it can result in devices being tried
	* which are known to not have the data. In which case, the import
	* is relying on the checksum to ensure that we get the right data.
	* Note that while importing we are only reading the MOS, which is
	* always checksummed.
	*/
	mutex_enter(&vd->vdev_dtl_lock);
	if (!range_tree_is_empty(rt))
	dirty = range_tree_contains(rt, txg, size);
	mutex_exit(&vd->vdev_dtl_lock);

	return (dirty);
	}

	boolean_t
	vdev_dtl_empty(vdev_t *vd, vdev_dtl_type_t t)
	{
	range_tree_t *rt = vd->vdev_dtl[t];
	boolean_t empty;

	mutex_enter(&vd->vdev_dtl_lock);
	empty = range_tree_is_empty(rt);
	mutex_exit(&vd->vdev_dtl_lock);

	return (empty);
	}

	/*
	* Check if the txg falls within the range which must be
	* resilvered. DVAs outside this range can always be skipped.
	*/
	boolean_t
	vdev_default_need_resilver(vdev_t vd, const dva_t dva, size_t psize,
	uint64_t phys_birth)
	{
	/* Set by sequential resilver. */
	if (phys_birth == TXG_UNKNOWN)
	return (B_TRUE);

	return (vdev_dtl_contains(vd, DTL_PARTIAL, phys_birth, 1));
	}

	/*
	* Returns B_TRUE if the vdev determines the DVA needs to be resilvered.
	*/
	boolean_t
	vdev_dtl_need_resilver(vdev_t vd, const dva_t dva, size_t psize,
	uint64_t phys_birth)
	{
	ASSERT(vd != vd->vdev_spa->spa_root_vdev);

	if (vd->vdev_ops->vdev_op_need_resilver == NULL \|\|
	vd->vdev_ops->vdev_op_leaf)
	return (B_TRUE);

	return (vd->vdev_ops->vdev_op_need_resilver(vd, dva, psize,
	phys_birth));
	}

	/*
	* Returns the lowest txg in the DTL range.
	*/
	static uint64_t
	vdev_dtl_min(vdev_t *vd)
	{
	ASSERT(MUTEX_HELD(&vd->vdev_dtl_lock));
	ASSERT3U(range_tree_space(vd->vdev_dtl[DTL_MISSING]), !=, 0);
	ASSERT0(vd->vdev_children);

	return (range_tree_min(vd->vdev_dtl[DTL_MISSING]) - 1);
	}

	/*
	* Returns the highest txg in the DTL.
	*/
	static uint64_t
	vdev_dtl_max(vdev_t *vd)
	{
	ASSERT(MUTEX_HELD(&vd->vdev_dtl_lock));
	ASSERT3U(range_tree_space(vd->vdev_dtl[DTL_MISSING]), !=, 0);
	ASSERT0(vd->vdev_children);

	return (range_tree_max(vd->vdev_dtl[DTL_MISSING]));
	}

	/*
	* Determine if a resilvering vdev should remove any DTL entries from
	* its range. If the vdev was resilvering for the entire duration of the
	* scan then it should excise that range from its DTLs. Otherwise, this
	* vdev is considered partially resilvered and should leave its DTL
	* entries intact. The comment in vdev_dtl_reassess() describes how we
	* excise the DTLs.
	*/
	static boolean_t
	vdev_dtl_should_excise(vdev_t *vd, boolean_t rebuild_done)
	{
	ASSERT0(vd->vdev_children);

	if (vd->vdev_state < VDEV_STATE_DEGRADED)
	return (B_FALSE);

	if (vd->vdev_resilver_deferred)
	return (B_FALSE);

	if (range_tree_is_empty(vd->vdev_dtl[DTL_MISSING]))
	return (B_TRUE);

	if (rebuild_done) {
	vdev_rebuild_t *vr = &vd->vdev_top->vdev_rebuild_config;
	vdev_rebuild_phys_t *vrp = &vr->vr_rebuild_phys;

	/* Rebuild not initiated by attach */
	if (vd->vdev_rebuild_txg == 0)
	return (B_TRUE);

	/*
	* When a rebuild completes without error then all missing data
	* up to the rebuild max txg has been reconstructed and the DTL
	* is eligible for excision.
	*/
	if (vrp->vrp_rebuild_state == VDEV_REBUILD_COMPLETE &&
	vdev_dtl_max(vd) <= vrp->vrp_max_txg) {
	ASSERT3U(vrp->vrp_min_txg, <=, vdev_dtl_min(vd));
	ASSERT3U(vrp->vrp_min_txg, <, vd->vdev_rebuild_txg);
	ASSERT3U(vd->vdev_rebuild_txg, <=, vrp->vrp_max_txg);
	return (B_TRUE);
	}
	} else {
	dsl_scan_t *scn = vd->vdev_spa->spa_dsl_pool->dp_scan;
	dsl_scan_phys_t *scnp __maybe_unused = &scn->scn_phys;

	/* Resilver not initiated by attach */
	if (vd->vdev_resilver_txg == 0)
	return (B_TRUE);

	/*
	* When a resilver is initiated the scan will assign the
	* scn_max_txg value to the highest txg value that exists
	* in all DTLs. If this device's max DTL is not part of this
	* scan (i.e. it is not in the range (scn_min_txg, scn_max_txg]
	* then it is not eligible for excision.
	*/
	if (vdev_dtl_max(vd) <= scn->scn_phys.scn_max_txg) {
	ASSERT3U(scnp->scn_min_txg, <=, vdev_dtl_min(vd));
	ASSERT3U(scnp->scn_min_txg, <, vd->vdev_resilver_txg);
	ASSERT3U(vd->vdev_resilver_txg, <=, scnp->scn_max_txg);
	return (B_TRUE);
	}
	}

	return (B_FALSE);
	}

	/*
	* Reassess DTLs after a config change or scrub completion. If txg == 0 no
	* write operations will be issued to the pool.
	*/
	void
	vdev_dtl_reassess(vdev_t *vd, uint64_t txg, uint64_t scrub_txg,
	boolean_t scrub_done, boolean_t rebuild_done)
	{
	spa_t *spa = vd->vdev_spa;
	avl_tree_t reftree;
	int minref;

	ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0);

	for (int c = 0; c < vd->vdev_children; c++)
	vdev_dtl_reassess(vd->vdev_child[c], txg,
	scrub_txg, scrub_done, rebuild_done);

	if (vd == spa->spa_root_vdev \|\| !vdev_is_concrete(vd) \|\| vd->vdev_aux)
	return;

	if (vd->vdev_ops->vdev_op_leaf) {
	dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan;
	vdev_rebuild_t *vr = &vd->vdev_top->vdev_rebuild_config;
	boolean_t check_excise = B_FALSE;
	boolean_t wasempty = B_TRUE;

	mutex_enter(&vd->vdev_dtl_lock);

	/*
	* If requested, pretend the scan or rebuild completed cleanly.
	*/
	if (zfs_scan_ignore_errors) {
	if (scn != NULL)
	scn->scn_phys.scn_errors = 0;
	if (vr != NULL)
	vr->vr_rebuild_phys.vrp_errors = 0;
	}

	if (scrub_txg != 0 &&
	!range_tree_is_empty(vd->vdev_dtl[DTL_MISSING])) {
	wasempty = B_FALSE;
	zfs_dbgmsg("guid:%llu txg:%llu scrub:%llu started:%d "
	"dtl:%llu/%llu errors:%llu",
	(u_longlong_t)vd->vdev_guid, (u_longlong_t)txg,
	(u_longlong_t)scrub_txg, spa->spa_scrub_started,
	(u_longlong_t)vdev_dtl_min(vd),
	(u_longlong_t)vdev_dtl_max(vd),
	(u_longlong_t)(scn ? scn->scn_phys.scn_errors : 0));
	}

	/*
	* If we've completed a scrub/resilver or a rebuild cleanly
	* then determine if this vdev should remove any DTLs. We
	* only want to excise regions on vdevs that were available
	* during the entire duration of this scan.
	*/
	if (rebuild_done &&
	vr != NULL && vr->vr_rebuild_phys.vrp_errors == 0) {
	check_excise = B_TRUE;
	} else {
	if (spa->spa_scrub_started \|\|
	(scn != NULL && scn->scn_phys.scn_errors == 0)) {
	check_excise = B_TRUE;
	}
	}

	if (scrub_txg && check_excise &&
	vdev_dtl_should_excise(vd, rebuild_done)) {
	/*
	* We completed a scrub, resilver or rebuild up to
	* scrub_txg. If we did it without rebooting, then
	* the scrub dtl will be valid, so excise the old
	* region and fold in the scrub dtl. Otherwise,
	* leave the dtl as-is if there was an error.
	*
	* There's little trick here: to excise the beginning
	* of the DTL_MISSING map, we put it into a reference
	* tree and then add a segment with refcnt -1 that
	* covers the range [0, scrub_txg). This means
	* that each txg in that range has refcnt -1 or 0.
	* We then add DTL_SCRUB with a refcnt of 2, so that
	* entries in the range [0, scrub_txg) will have a
	* positive refcnt -- either 1 or 2. We then convert
	* the reference tree into the new DTL_MISSING map.
	*/
	space_reftree_create(&reftree);
	space_reftree_add_map(&reftree,
	vd->vdev_dtl[DTL_MISSING], 1);
	space_reftree_add_seg(&reftree, 0, scrub_txg, -1);
	space_reftree_add_map(&reftree,
	vd->vdev_dtl[DTL_SCRUB], 2);
	space_reftree_generate_map(&reftree,
	vd->vdev_dtl[DTL_MISSING], 1);
	space_reftree_destroy(&reftree);

	if (!range_tree_is_empty(vd->vdev_dtl[DTL_MISSING])) {
	zfs_dbgmsg("update DTL_MISSING:%llu/%llu",
	(u_longlong_t)vdev_dtl_min(vd),
	(u_longlong_t)vdev_dtl_max(vd));
	} else if (!wasempty) {
	zfs_dbgmsg("DTL_MISSING is now empty");
	}
	}
	range_tree_vacate(vd->vdev_dtl[DTL_PARTIAL], NULL, NULL);
	range_tree_walk(vd->vdev_dtl[DTL_MISSING],
	range_tree_add, vd->vdev_dtl[DTL_PARTIAL]);
	if (scrub_done)
	range_tree_vacate(vd->vdev_dtl[DTL_SCRUB], NULL, NULL);
	range_tree_vacate(vd->vdev_dtl[DTL_OUTAGE], NULL, NULL);
	if (!vdev_readable(vd))
	range_tree_add(vd->vdev_dtl[DTL_OUTAGE], 0, -1ULL);
	else
	range_tree_walk(vd->vdev_dtl[DTL_MISSING],
	range_tree_add, vd->vdev_dtl[DTL_OUTAGE]);

	/*
	* If the vdev was resilvering or rebuilding and no longer
	* has any DTLs then reset the appropriate flag and dirty
	* the top level so that we persist the change.
	*/
	if (txg != 0 &&
	range_tree_is_empty(vd->vdev_dtl[DTL_MISSING]) &&
	range_tree_is_empty(vd->vdev_dtl[DTL_OUTAGE])) {
	if (vd->vdev_rebuild_txg != 0) {
	vd->vdev_rebuild_txg = 0;
	vdev_config_dirty(vd->vdev_top);
	} else if (vd->vdev_resilver_txg != 0) {
	vd->vdev_resilver_txg = 0;
	vdev_config_dirty(vd->vdev_top);
	}
	}

	mutex_exit(&vd->vdev_dtl_lock);

	if (txg != 0)
	vdev_dirty(vd->vdev_top, VDD_DTL, vd, txg);
	return;
	}

	mutex_enter(&vd->vdev_dtl_lock);
	for (int t = 0; t < DTL_TYPES; t++) {
	/* account for child's outage in parent's missing map */
	int s = (t == DTL_MISSING) ? DTL_OUTAGE: t;
	if (t == DTL_SCRUB)
	continue; /* leaf vdevs only */
	if (t == DTL_PARTIAL)
	minref = 1; /* i.e. non-zero */
	else if (vdev_get_nparity(vd) != 0)
	minref = vdev_get_nparity(vd) + 1; /* RAID-Z, dRAID */
	else
	minref = vd->vdev_children; /* any kind of mirror */
	space_reftree_create(&reftree);
	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];
	mutex_enter(&cvd->vdev_dtl_lock);
	space_reftree_add_map(&reftree, cvd->vdev_dtl[s], 1);
	mutex_exit(&cvd->vdev_dtl_lock);
	}
	space_reftree_generate_map(&reftree, vd->vdev_dtl[t], minref);
	space_reftree_destroy(&reftree);
	}
	mutex_exit(&vd->vdev_dtl_lock);
	}

	int
	vdev_dtl_load(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;
	objset_t *mos = spa->spa_meta_objset;
	range_tree_t *rt;
	int error = 0;

	if (vd->vdev_ops->vdev_op_leaf && vd->vdev_dtl_object != 0) {
	ASSERT(vdev_is_concrete(vd));

	/*
	* If the dtl cannot be sync'd there is no need to open it.
	*/
	if (spa->spa_mode == SPA_MODE_READ && !spa->spa_read_spacemaps)
	return (0);

	error = space_map_open(&vd->vdev_dtl_sm, mos,
	vd->vdev_dtl_object, 0, -1ULL, 0);
	if (error)
	return (error);
	ASSERT(vd->vdev_dtl_sm != NULL);

	rt = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0);
	error = space_map_load(vd->vdev_dtl_sm, rt, SM_ALLOC);
	if (error == 0) {
	mutex_enter(&vd->vdev_dtl_lock);
	range_tree_walk(rt, range_tree_add,
	vd->vdev_dtl[DTL_MISSING]);
	mutex_exit(&vd->vdev_dtl_lock);
	}

	range_tree_vacate(rt, NULL, NULL);
	range_tree_destroy(rt);

	return (error);
	}

	for (int c = 0; c < vd->vdev_children; c++) {
	error = vdev_dtl_load(vd->vdev_child[c]);
	if (error != 0)
	break;
	}

	return (error);
	}

	static void
	vdev_zap_allocation_data(vdev_t vd, dmu_tx_t tx)
	{
	spa_t *spa = vd->vdev_spa;
	objset_t *mos = spa->spa_meta_objset;
	vdev_alloc_bias_t alloc_bias = vd->vdev_alloc_bias;
	const char *string;

	ASSERT(alloc_bias != VDEV_BIAS_NONE);

	string =
	(alloc_bias == VDEV_BIAS_LOG) ? VDEV_ALLOC_BIAS_LOG :
	(alloc_bias == VDEV_BIAS_SPECIAL) ? VDEV_ALLOC_BIAS_SPECIAL :
	(alloc_bias == VDEV_BIAS_DEDUP) ? VDEV_ALLOC_BIAS_DEDUP : NULL;

	ASSERT(string != NULL);
	VERIFY0(zap_add(mos, vd->vdev_top_zap, VDEV_TOP_ZAP_ALLOCATION_BIAS,
	1, strlen(string) + 1, string, tx));

	if (alloc_bias == VDEV_BIAS_SPECIAL \|\| alloc_bias == VDEV_BIAS_DEDUP) {
	spa_activate_allocation_classes(spa, tx);
	}
	}

	void
	vdev_destroy_unlink_zap(vdev_t vd, uint64_t zapobj, dmu_tx_t tx)
	{
	spa_t *spa = vd->vdev_spa;

	VERIFY0(zap_destroy(spa->spa_meta_objset, zapobj, tx));
	VERIFY0(zap_remove_int(spa->spa_meta_objset, spa->spa_all_vdev_zaps,
	zapobj, tx));
	}

	uint64_t
	vdev_create_link_zap(vdev_t vd, dmu_tx_t tx)
	{
	spa_t *spa = vd->vdev_spa;
	uint64_t zap = zap_create(spa->spa_meta_objset, DMU_OTN_ZAP_METADATA,
	DMU_OT_NONE, 0, tx);

	ASSERT(zap != 0);
	VERIFY0(zap_add_int(spa->spa_meta_objset, spa->spa_all_vdev_zaps,
	zap, tx));

	return (zap);
	}

	void
	vdev_construct_zaps(vdev_t vd, dmu_tx_t tx)
	{
	if (vd->vdev_ops != &vdev_hole_ops &&
	vd->vdev_ops != &vdev_missing_ops &&
	vd->vdev_ops != &vdev_root_ops &&
	!vd->vdev_top->vdev_removing) {
	if (vd->vdev_ops->vdev_op_leaf && vd->vdev_leaf_zap == 0) {
	vd->vdev_leaf_zap = vdev_create_link_zap(vd, tx);
	}
	if (vd == vd->vdev_top && vd->vdev_top_zap == 0) {
	vd->vdev_top_zap = vdev_create_link_zap(vd, tx);
	if (vd->vdev_alloc_bias != VDEV_BIAS_NONE)
	vdev_zap_allocation_data(vd, tx);
	}
	}

	for (uint64_t i = 0; i < vd->vdev_children; i++) {
	vdev_construct_zaps(vd->vdev_child[i], tx);
	}
	}

	static void
	vdev_dtl_sync(vdev_t *vd, uint64_t txg)
	{
	spa_t *spa = vd->vdev_spa;
	range_tree_t *rt = vd->vdev_dtl[DTL_MISSING];
	objset_t *mos = spa->spa_meta_objset;
	range_tree_t *rtsync;
	dmu_tx_t *tx;
	uint64_t object = space_map_object(vd->vdev_dtl_sm);

	ASSERT(vdev_is_concrete(vd));
	ASSERT(vd->vdev_ops->vdev_op_leaf);

	tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg);

	if (vd->vdev_detached \|\| vd->vdev_top->vdev_removing) {
	mutex_enter(&vd->vdev_dtl_lock);
	space_map_free(vd->vdev_dtl_sm, tx);
	space_map_close(vd->vdev_dtl_sm);
	vd->vdev_dtl_sm = NULL;
	mutex_exit(&vd->vdev_dtl_lock);

	/*
	* We only destroy the leaf ZAP for detached leaves or for
	* removed log devices. Removed data devices handle leaf ZAP
	* cleanup later, once cancellation is no longer possible.
	*/
	if (vd->vdev_leaf_zap != 0 && (vd->vdev_detached \|\|
	vd->vdev_top->vdev_islog)) {
	vdev_destroy_unlink_zap(vd, vd->vdev_leaf_zap, tx);
	vd->vdev_leaf_zap = 0;
	}

	dmu_tx_commit(tx);
	return;
	}

	if (vd->vdev_dtl_sm == NULL) {
	uint64_t new_object;

	new_object = space_map_alloc(mos, zfs_vdev_dtl_sm_blksz, tx);
	VERIFY3U(new_object, !=, 0);

	VERIFY0(space_map_open(&vd->vdev_dtl_sm, mos, new_object,
	0, -1ULL, 0));
	ASSERT(vd->vdev_dtl_sm != NULL);
	}

	rtsync = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0);

	mutex_enter(&vd->vdev_dtl_lock);
	range_tree_walk(rt, range_tree_add, rtsync);
	mutex_exit(&vd->vdev_dtl_lock);

	space_map_truncate(vd->vdev_dtl_sm, zfs_vdev_dtl_sm_blksz, tx);
	space_map_write(vd->vdev_dtl_sm, rtsync, SM_ALLOC, SM_NO_VDEVID, tx);
	range_tree_vacate(rtsync, NULL, NULL);

	range_tree_destroy(rtsync);

	/*
	* If the object for the space map has changed then dirty
	* the top level so that we update the config.
	*/
	if (object != space_map_object(vd->vdev_dtl_sm)) {
	vdev_dbgmsg(vd, "txg %llu, spa %s, DTL old object %llu, "
	"new object %llu", (u_longlong_t)txg, spa_name(spa),
	(u_longlong_t)object,
	(u_longlong_t)space_map_object(vd->vdev_dtl_sm));
	vdev_config_dirty(vd->vdev_top);
	}

	dmu_tx_commit(tx);
	}

	/*
	* Determine whether the specified vdev can be offlined/detached/removed
	* without losing data.
	*/
	boolean_t
	vdev_dtl_required(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;
	vdev_t *tvd = vd->vdev_top;
	uint8_t cant_read = vd->vdev_cant_read;
	boolean_t required;

	ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL);

	if (vd == spa->spa_root_vdev \|\| vd == tvd)
	return (B_TRUE);

	/*
	* Temporarily mark the device as unreadable, and then determine
	* whether this results in any DTL outages in the top-level vdev.
	* If not, we can safely offline/detach/remove the device.
	*/
	vd->vdev_cant_read = B_TRUE;
	vdev_dtl_reassess(tvd, 0, 0, B_FALSE, B_FALSE);
	required = !vdev_dtl_empty(tvd, DTL_OUTAGE);
	vd->vdev_cant_read = cant_read;
	vdev_dtl_reassess(tvd, 0, 0, B_FALSE, B_FALSE);

	if (!required && zio_injection_enabled) {
	required = !!zio_handle_device_injection(vd, NULL,
	SET_ERROR(ECHILD));
	}

	return (required);
	}

	/*
	* Determine if resilver is needed, and if so the txg range.
	*/
	boolean_t
	vdev_resilver_needed(vdev_t vd, uint64_t minp, uint64_t *maxp)
	{
	boolean_t needed = B_FALSE;
	uint64_t thismin = UINT64_MAX;
	uint64_t thismax = 0;

	if (vd->vdev_children == 0) {
	mutex_enter(&vd->vdev_dtl_lock);
	if (!range_tree_is_empty(vd->vdev_dtl[DTL_MISSING]) &&
	vdev_writeable(vd)) {

	thismin = vdev_dtl_min(vd);
	thismax = vdev_dtl_max(vd);
	needed = B_TRUE;
	}
	mutex_exit(&vd->vdev_dtl_lock);
	} else {
	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];
	uint64_t cmin, cmax;

	if (vdev_resilver_needed(cvd, &cmin, &cmax)) {
	thismin = MIN(thismin, cmin);
	thismax = MAX(thismax, cmax);
	needed = B_TRUE;
	}
	}
	}

	if (needed && minp) {
	*minp = thismin;
	*maxp = thismax;
	}
	return (needed);
	}

	/*
	* Gets the checkpoint space map object from the vdev's ZAP. On success sm_obj
	* will contain either the checkpoint spacemap object or zero if none exists.
	* All other errors are returned to the caller.
	*/
	int
	vdev_checkpoint_sm_object(vdev_t vd, uint64_t sm_obj)
	{
	ASSERT0(spa_config_held(vd->vdev_spa, SCL_ALL, RW_WRITER));

	if (vd->vdev_top_zap == 0) {
	*sm_obj = 0;
	return (0);
	}

	int error = zap_lookup(spa_meta_objset(vd->vdev_spa), vd->vdev_top_zap,
	VDEV_TOP_ZAP_POOL_CHECKPOINT_SM, sizeof (uint64_t), 1, sm_obj);
	if (error == ENOENT) {
	*sm_obj = 0;
	error = 0;
	}

	return (error);
	}

	int
	vdev_load(vdev_t *vd)
	{
	int children = vd->vdev_children;
	int error = 0;
	taskq_t *tq = NULL;

	/*
	* It's only worthwhile to use the taskq for the root vdev, because the
	* slow part is metaslab_init, and that only happens for top-level
	* vdevs.
	*/
	if (vd->vdev_ops == &vdev_root_ops && vd->vdev_children > 0) {
	tq = taskq_create("vdev_load", children, minclsyspri,
	children, children, TASKQ_PREPOPULATE);
	}

	/*
	* Recursively load all children.
	*/
	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];

	if (tq == NULL \|\| vdev_uses_zvols(cvd)) {
	cvd->vdev_load_error = vdev_load(cvd);
	} else {
	VERIFY(taskq_dispatch(tq, vdev_load_child,
	cvd, TQ_SLEEP) != TASKQID_INVALID);
	}
	}

	if (tq != NULL) {
	taskq_wait(tq);
	taskq_destroy(tq);
	}

	for (int c = 0; c < vd->vdev_children; c++) {
	int error = vd->vdev_child[c]->vdev_load_error;

	if (error != 0)
	return (error);
	}

	vdev_set_deflate_ratio(vd);

	/*
	* On spa_load path, grab the allocation bias from our zap
	*/
	if (vd == vd->vdev_top && vd->vdev_top_zap != 0) {
	spa_t *spa = vd->vdev_spa;
	char bias_str[64];

	error = zap_lookup(spa->spa_meta_objset, vd->vdev_top_zap,
	VDEV_TOP_ZAP_ALLOCATION_BIAS, 1, sizeof (bias_str),
	bias_str);
	if (error == 0) {
	ASSERT(vd->vdev_alloc_bias == VDEV_BIAS_NONE);
	vd->vdev_alloc_bias = vdev_derive_alloc_bias(bias_str);
	} else if (error != ENOENT) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	vdev_dbgmsg(vd, "vdev_load: zap_lookup(top_zap=%llu) "
	"failed [error=%d]",
	(u_longlong_t)vd->vdev_top_zap, error);
	return (error);
	}
	}

	/*
	* Load any rebuild state from the top-level vdev zap.
	*/
	if (vd == vd->vdev_top && vd->vdev_top_zap != 0) {
	error = vdev_rebuild_load(vd);
	if (error && error != ENOTSUP) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	vdev_dbgmsg(vd, "vdev_load: vdev_rebuild_load "
	"failed [error=%d]", error);
	return (error);
	}
	}

	/*
	* If this is a top-level vdev, initialize its metaslabs.
	*/
	if (vd == vd->vdev_top && vdev_is_concrete(vd)) {
	vdev_metaslab_group_create(vd);

	if (vd->vdev_ashift == 0 \|\| vd->vdev_asize == 0) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	vdev_dbgmsg(vd, "vdev_load: invalid size. ashift=%llu, "
	"asize=%llu", (u_longlong_t)vd->vdev_ashift,
	(u_longlong_t)vd->vdev_asize);
	return (SET_ERROR(ENXIO));
	}

	error = vdev_metaslab_init(vd, 0);
	if (error != 0) {
	vdev_dbgmsg(vd, "vdev_load: metaslab_init failed "
	"[error=%d]", error);
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	return (error);
	}

	uint64_t checkpoint_sm_obj;
	error = vdev_checkpoint_sm_object(vd, &checkpoint_sm_obj);
	if (error == 0 && checkpoint_sm_obj != 0) {
	objset_t *mos = spa_meta_objset(vd->vdev_spa);
	ASSERT(vd->vdev_asize != 0);
	ASSERT3P(vd->vdev_checkpoint_sm, ==, NULL);

	error = space_map_open(&vd->vdev_checkpoint_sm,
	mos, checkpoint_sm_obj, 0, vd->vdev_asize,
	vd->vdev_ashift);
	if (error != 0) {
	vdev_dbgmsg(vd, "vdev_load: space_map_open "
	"failed for checkpoint spacemap (obj %llu) "
	"[error=%d]",
	(u_longlong_t)checkpoint_sm_obj, error);
	return (error);
	}
	ASSERT3P(vd->vdev_checkpoint_sm, !=, NULL);

	/*
	* Since the checkpoint_sm contains free entries
	* exclusively we can use space_map_allocated() to
	* indicate the cumulative checkpointed space that
	* has been freed.
	*/
	vd->vdev_stat.vs_checkpoint_space =
	-space_map_allocated(vd->vdev_checkpoint_sm);
	vd->vdev_spa->spa_checkpoint_info.sci_dspace +=
	vd->vdev_stat.vs_checkpoint_space;
	} else if (error != 0) {
	vdev_dbgmsg(vd, "vdev_load: failed to retrieve "
	"checkpoint space map object from vdev ZAP "
	"[error=%d]", error);
	return (error);
	}
	}

	/*
	* If this is a leaf vdev, load its DTL.
	*/
	if (vd->vdev_ops->vdev_op_leaf && (error = vdev_dtl_load(vd)) != 0) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	vdev_dbgmsg(vd, "vdev_load: vdev_dtl_load failed "
	"[error=%d]", error);
	return (error);
	}

	uint64_t obsolete_sm_object;
	error = vdev_obsolete_sm_object(vd, &obsolete_sm_object);
	if (error == 0 && obsolete_sm_object != 0) {
	objset_t *mos = vd->vdev_spa->spa_meta_objset;
	ASSERT(vd->vdev_asize != 0);
	ASSERT3P(vd->vdev_obsolete_sm, ==, NULL);

	if ((error = space_map_open(&vd->vdev_obsolete_sm, mos,
	obsolete_sm_object, 0, vd->vdev_asize, 0))) {
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	vdev_dbgmsg(vd, "vdev_load: space_map_open failed for "
	"obsolete spacemap (obj %llu) [error=%d]",
	(u_longlong_t)obsolete_sm_object, error);
	return (error);
	}
	} else if (error != 0) {
	vdev_dbgmsg(vd, "vdev_load: failed to retrieve obsolete "
	"space map object from vdev ZAP [error=%d]", error);
	return (error);
	}

	return (0);
	}

	/*
	* The special vdev case is used for hot spares and l2cache devices. Its
	* sole purpose it to set the vdev state for the associated vdev. To do this,
	* we make sure that we can open the underlying device, then try to read the
	* label, and make sure that the label is sane and that it hasn't been
	* repurposed to another pool.
	*/
	int
	vdev_validate_aux(vdev_t *vd)
	{
	nvlist_t *label;
	uint64_t guid, version;
	uint64_t state;

	if (!vdev_readable(vd))
	return (0);

	if ((label = vdev_label_read_config(vd, -1ULL)) == NULL) {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	return (-1);
	}

	if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_VERSION, &version) != 0 \|\|
	!SPA_VERSION_IS_SUPPORTED(version) \|\|
	nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID, &guid) != 0 \|\|
	guid != vd->vdev_guid \|\|
	nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_STATE, &state) != 0) {
	vdev_set_state(vd, B_TRUE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	nvlist_free(label);
	return (-1);
	}

	/*
	* We don't actually check the pool state here. If it's in fact in
	* use by another pool, we update this fact on the fly when requested.
	*/
	nvlist_free(label);
	return (0);
	}

	static void
	vdev_destroy_ms_flush_data(vdev_t vd, dmu_tx_t tx)
	{
	objset_t *mos = spa_meta_objset(vd->vdev_spa);

	if (vd->vdev_top_zap == 0)
	return;

	uint64_t object = 0;
	int err = zap_lookup(mos, vd->vdev_top_zap,
	VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS, sizeof (uint64_t), 1, &object);
	if (err == ENOENT)
	return;
	VERIFY0(err);

	VERIFY0(dmu_object_free(mos, object, tx));
	VERIFY0(zap_remove(mos, vd->vdev_top_zap,
	VDEV_TOP_ZAP_MS_UNFLUSHED_PHYS_TXGS, tx));
	}

	/*
	* Free the objects used to store this vdev's spacemaps, and the array
	* that points to them.
	*/
	void
	vdev_destroy_spacemaps(vdev_t vd, dmu_tx_t tx)
	{
	if (vd->vdev_ms_array == 0)
	return;

	objset_t *mos = vd->vdev_spa->spa_meta_objset;
	uint64_t array_count = vd->vdev_asize >> vd->vdev_ms_shift;
	size_t array_bytes = array_count * sizeof (uint64_t);
	uint64_t *smobj_array = kmem_alloc(array_bytes, KM_SLEEP);
	VERIFY0(dmu_read(mos, vd->vdev_ms_array, 0,
	array_bytes, smobj_array, 0));

	for (uint64_t i = 0; i < array_count; i++) {
	uint64_t smobj = smobj_array[i];
	if (smobj == 0)
	continue;

	space_map_free_obj(mos, smobj, tx);
	}

	kmem_free(smobj_array, array_bytes);
	VERIFY0(dmu_object_free(mos, vd->vdev_ms_array, tx));
	vdev_destroy_ms_flush_data(vd, tx);
	vd->vdev_ms_array = 0;
	}

	static void
	vdev_remove_empty_log(vdev_t *vd, uint64_t txg)
	{
	spa_t *spa = vd->vdev_spa;

	ASSERT(vd->vdev_islog);
	ASSERT(vd == vd->vdev_top);
	ASSERT3U(txg, ==, spa_syncing_txg(spa));

	dmu_tx_t *tx = dmu_tx_create_assigned(spa_get_dsl(spa), txg);

	vdev_destroy_spacemaps(vd, tx);
	if (vd->vdev_top_zap != 0) {
	vdev_destroy_unlink_zap(vd, vd->vdev_top_zap, tx);
	vd->vdev_top_zap = 0;
	}

	dmu_tx_commit(tx);
	}

	void
	vdev_sync_done(vdev_t *vd, uint64_t txg)
	{
	metaslab_t *msp;
	boolean_t reassess = !txg_list_empty(&vd->vdev_ms_list, TXG_CLEAN(txg));

	ASSERT(vdev_is_concrete(vd));

	while ((msp = txg_list_remove(&vd->vdev_ms_list, TXG_CLEAN(txg)))
	!= NULL)
	metaslab_sync_done(msp, txg);

	if (reassess) {
	metaslab_sync_reassess(vd->vdev_mg);
	if (vd->vdev_log_mg != NULL)
	metaslab_sync_reassess(vd->vdev_log_mg);
	}
	}

	void
	vdev_sync(vdev_t *vd, uint64_t txg)
	{
	spa_t *spa = vd->vdev_spa;
	vdev_t *lvd;
	metaslab_t *msp;

	ASSERT3U(txg, ==, spa->spa_syncing_txg);
	dmu_tx_t *tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg);
	if (range_tree_space(vd->vdev_obsolete_segments) > 0) {
	ASSERT(vd->vdev_removing \|\|
	vd->vdev_ops == &vdev_indirect_ops);

	vdev_indirect_sync_obsolete(vd, tx);

	/*
	* If the vdev is indirect, it can't have dirty
	* metaslabs or DTLs.
	*/
	if (vd->vdev_ops == &vdev_indirect_ops) {
	ASSERT(txg_list_empty(&vd->vdev_ms_list, txg));
	ASSERT(txg_list_empty(&vd->vdev_dtl_list, txg));
	dmu_tx_commit(tx);
	return;
	}
	}

	ASSERT(vdev_is_concrete(vd));

	if (vd->vdev_ms_array == 0 && vd->vdev_ms_shift != 0 &&
	!vd->vdev_removing) {
	ASSERT(vd == vd->vdev_top);
	ASSERT0(vd->vdev_indirect_config.vic_mapping_object);
	vd->vdev_ms_array = dmu_object_alloc(spa->spa_meta_objset,
	DMU_OT_OBJECT_ARRAY, 0, DMU_OT_NONE, 0, tx);
	ASSERT(vd->vdev_ms_array != 0);
	vdev_config_dirty(vd);
	}

	while ((msp = txg_list_remove(&vd->vdev_ms_list, txg)) != NULL) {
	metaslab_sync(msp, txg);
	(void) txg_list_add(&vd->vdev_ms_list, msp, TXG_CLEAN(txg));
	}

	while ((lvd = txg_list_remove(&vd->vdev_dtl_list, txg)) != NULL)
	vdev_dtl_sync(lvd, txg);

	/*
	* If this is an empty log device being removed, destroy the
	* metadata associated with it.
	*/
	if (vd->vdev_islog && vd->vdev_stat.vs_alloc == 0 && vd->vdev_removing)
	vdev_remove_empty_log(vd, txg);

	(void) txg_list_add(&spa->spa_vdev_txg_list, vd, TXG_CLEAN(txg));
	dmu_tx_commit(tx);
	}

	uint64_t
	vdev_psize_to_asize(vdev_t *vd, uint64_t psize)
	{
	return (vd->vdev_ops->vdev_op_asize(vd, psize));
	}

	/*
	* Mark the given vdev faulted. A faulted vdev behaves as if the device could
	* not be opened, and no I/O is attempted.
	*/
	int
	vdev_fault(spa_t *spa, uint64_t guid, vdev_aux_t aux)
	{
	vdev_t vd, tvd;

	spa_vdev_state_enter(spa, SCL_NONE);

	if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL)
	return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV)));

	if (!vd->vdev_ops->vdev_op_leaf)
	return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP)));

	tvd = vd->vdev_top;

	/*
	* If user did a 'zpool offline -f' then make the fault persist across
	* reboots.
	*/
	if (aux == VDEV_AUX_EXTERNAL_PERSIST) {
	/*
	* There are two kinds of forced faults: temporary and
	* persistent. Temporary faults go away at pool import, while
	* persistent faults stay set. Both types of faults can be
	* cleared with a zpool clear.
	*
	* We tell if a vdev is persistently faulted by looking at the
	* ZPOOL_CONFIG_AUX_STATE nvpair. If it's set to "external" at
	* import then it's a persistent fault. Otherwise, it's
	* temporary. We get ZPOOL_CONFIG_AUX_STATE set to "external"
	* by setting vd.vdev_stat.vs_aux to VDEV_AUX_EXTERNAL. This
	* tells vdev_config_generate() (which gets run later) to set
	* ZPOOL_CONFIG_AUX_STATE to "external" in the nvlist.
	*/
	vd->vdev_stat.vs_aux = VDEV_AUX_EXTERNAL;
	vd->vdev_tmpoffline = B_FALSE;
	aux = VDEV_AUX_EXTERNAL;
	} else {
	vd->vdev_tmpoffline = B_TRUE;
	}

	/*
	* We don't directly use the aux state here, but if we do a
	* vdev_reopen(), we need this value to be present to remember why we
	* were faulted.
	*/
	vd->vdev_label_aux = aux;

	/*
	* Faulted state takes precedence over degraded.
	*/
	vd->vdev_delayed_close = B_FALSE;
	vd->vdev_faulted = 1ULL;
	vd->vdev_degraded = 0ULL;
	vdev_set_state(vd, B_FALSE, VDEV_STATE_FAULTED, aux);

	/*
	* If this device has the only valid copy of the data, then
	* back off and simply mark the vdev as degraded instead.
	*/
	if (!tvd->vdev_islog && vd->vdev_aux == NULL && vdev_dtl_required(vd)) {
	vd->vdev_degraded = 1ULL;
	vd->vdev_faulted = 0ULL;

	/*
	* If we reopen the device and it's not dead, only then do we
	* mark it degraded.
	*/
	vdev_reopen(tvd);

	if (vdev_readable(vd))
	vdev_set_state(vd, B_FALSE, VDEV_STATE_DEGRADED, aux);
	}

	return (spa_vdev_state_exit(spa, vd, 0));
	}

	/*
	* Mark the given vdev degraded. A degraded vdev is purely an indication to the
	* user that something is wrong. The vdev continues to operate as normal as far
	* as I/O is concerned.
	*/
	int
	vdev_degrade(spa_t *spa, uint64_t guid, vdev_aux_t aux)
	{
	vdev_t *vd;

	spa_vdev_state_enter(spa, SCL_NONE);

	if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL)
	return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV)));

	if (!vd->vdev_ops->vdev_op_leaf)
	return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP)));

	/*
	* If the vdev is already faulted, then don't do anything.
	*/
	if (vd->vdev_faulted \|\| vd->vdev_degraded)
	return (spa_vdev_state_exit(spa, NULL, 0));

	vd->vdev_degraded = 1ULL;
	if (!vdev_is_dead(vd))
	vdev_set_state(vd, B_FALSE, VDEV_STATE_DEGRADED,
	aux);

	return (spa_vdev_state_exit(spa, vd, 0));
	}

	/*
	* Online the given vdev.
	*
	* If 'ZFS_ONLINE_UNSPARE' is set, it implies two things. First, any attached
	* spare device should be detached when the device finishes resilvering.
	* Second, the online should be treated like a 'test' online case, so no FMA
	* events are generated if the device fails to open.
	*/
	int
	vdev_online(spa_t spa, uint64_t guid, uint64_t flags, vdev_state_t newstate)
	{
	vdev_t vd, tvd, pvd, rvd = spa->spa_root_vdev;
	boolean_t wasoffline;
	vdev_state_t oldstate;

	spa_vdev_state_enter(spa, SCL_NONE);

	if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL)
	return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV)));

	if (!vd->vdev_ops->vdev_op_leaf)
	return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP)));

	wasoffline = (vd->vdev_offline \|\| vd->vdev_tmpoffline);
	oldstate = vd->vdev_state;

	tvd = vd->vdev_top;
	vd->vdev_offline = B_FALSE;
	vd->vdev_tmpoffline = B_FALSE;
	vd->vdev_checkremove = !!(flags & ZFS_ONLINE_CHECKREMOVE);
	vd->vdev_forcefault = !!(flags & ZFS_ONLINE_FORCEFAULT);

	/* XXX - L2ARC 1.0 does not support expansion */
	if (!vd->vdev_aux) {
	for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent)
	pvd->vdev_expanding = !!((flags & ZFS_ONLINE_EXPAND) \|\|
	spa->spa_autoexpand);
	vd->vdev_expansion_time = gethrestime_sec();
	}

	vdev_reopen(tvd);
	vd->vdev_checkremove = vd->vdev_forcefault = B_FALSE;

	if (!vd->vdev_aux) {
	for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent)
	pvd->vdev_expanding = B_FALSE;
	}

	if (newstate)
	*newstate = vd->vdev_state;
	if ((flags & ZFS_ONLINE_UNSPARE) &&
	!vdev_is_dead(vd) && vd->vdev_parent &&
	vd->vdev_parent->vdev_ops == &vdev_spare_ops &&
	vd->vdev_parent->vdev_child[0] == vd)
	vd->vdev_unspare = B_TRUE;

	if ((flags & ZFS_ONLINE_EXPAND) \|\| spa->spa_autoexpand) {

	/* XXX - L2ARC 1.0 does not support expansion */
	if (vd->vdev_aux)
	return (spa_vdev_state_exit(spa, vd, ENOTSUP));
	spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE);
	}

	/* Restart initializing if necessary */
	mutex_enter(&vd->vdev_initialize_lock);
	if (vdev_writeable(vd) &&
	vd->vdev_initialize_thread == NULL &&
	vd->vdev_initialize_state == VDEV_INITIALIZE_ACTIVE) {
	(void) vdev_initialize(vd);
	}
	mutex_exit(&vd->vdev_initialize_lock);

	/*
	* Restart trimming if necessary. We do not restart trimming for cache
	* devices here. This is triggered by l2arc_rebuild_vdev()
	* asynchronously for the whole device or in l2arc_evict() as it evicts
	* space for upcoming writes.
	*/
	mutex_enter(&vd->vdev_trim_lock);
	if (vdev_writeable(vd) && !vd->vdev_isl2cache &&
	vd->vdev_trim_thread == NULL &&
	vd->vdev_trim_state == VDEV_TRIM_ACTIVE) {
	(void) vdev_trim(vd, vd->vdev_trim_rate, vd->vdev_trim_partial,
	vd->vdev_trim_secure);
	}
	mutex_exit(&vd->vdev_trim_lock);

	if (wasoffline \|\|
	(oldstate < VDEV_STATE_DEGRADED &&
	vd->vdev_state >= VDEV_STATE_DEGRADED))
	spa_event_notify(spa, vd, NULL, ESC_ZFS_VDEV_ONLINE);

	return (spa_vdev_state_exit(spa, vd, 0));
	}

	static int
	vdev_offline_locked(spa_t *spa, uint64_t guid, uint64_t flags)
	{
	vdev_t vd, tvd;
	int error = 0;
	uint64_t generation;
	metaslab_group_t *mg;

	top:
	spa_vdev_state_enter(spa, SCL_ALLOC);

	if ((vd = spa_lookup_by_guid(spa, guid, B_TRUE)) == NULL)
	return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENODEV)));

	if (!vd->vdev_ops->vdev_op_leaf)
	return (spa_vdev_state_exit(spa, NULL, SET_ERROR(ENOTSUP)));

	if (vd->vdev_ops == &vdev_draid_spare_ops)
	return (spa_vdev_state_exit(spa, NULL, ENOTSUP));

	tvd = vd->vdev_top;
	mg = tvd->vdev_mg;
	generation = spa->spa_config_generation + 1;

	/*
	* If the device isn't already offline, try to offline it.
	*/
	if (!vd->vdev_offline) {
	/*
	* If this device has the only valid copy of some data,
	* don't allow it to be offlined. Log devices are always
	* expendable.
	*/
	if (!tvd->vdev_islog && vd->vdev_aux == NULL &&
	vdev_dtl_required(vd))
	return (spa_vdev_state_exit(spa, NULL,
	SET_ERROR(EBUSY)));

	/*
	* If the top-level is a slog and it has had allocations
	* then proceed. We check that the vdev's metaslab group
	* is not NULL since it's possible that we may have just
	* added this vdev but not yet initialized its metaslabs.
	*/
	if (tvd->vdev_islog && mg != NULL) {
	/*
	* Prevent any future allocations.
	*/
	ASSERT3P(tvd->vdev_log_mg, ==, NULL);
	metaslab_group_passivate(mg);
	(void) spa_vdev_state_exit(spa, vd, 0);

	error = spa_reset_logs(spa);

	/*
	* If the log device was successfully reset but has
	* checkpointed data, do not offline it.
	*/
	if (error == 0 &&
	tvd->vdev_checkpoint_sm != NULL) {
	ASSERT3U(space_map_allocated(
	tvd->vdev_checkpoint_sm), !=, 0);
	error = ZFS_ERR_CHECKPOINT_EXISTS;
	}

	spa_vdev_state_enter(spa, SCL_ALLOC);

	/*
	* Check to see if the config has changed.
	*/
	if (error \|\| generation != spa->spa_config_generation) {
	metaslab_group_activate(mg);
	if (error)
	return (spa_vdev_state_exit(spa,
	vd, error));
	(void) spa_vdev_state_exit(spa, vd, 0);
	goto top;
	}
	ASSERT0(tvd->vdev_stat.vs_alloc);
	}

	/*
	* Offline this device and reopen its top-level vdev.
	* If the top-level vdev is a log device then just offline
	* it. Otherwise, if this action results in the top-level
	* vdev becoming unusable, undo it and fail the request.
	*/
	vd->vdev_offline = B_TRUE;
	vdev_reopen(tvd);

	if (!tvd->vdev_islog && vd->vdev_aux == NULL &&
	vdev_is_dead(tvd)) {
	vd->vdev_offline = B_FALSE;
	vdev_reopen(tvd);
	return (spa_vdev_state_exit(spa, NULL,
	SET_ERROR(EBUSY)));
	}

	/*
	* Add the device back into the metaslab rotor so that
	* once we online the device it's open for business.
	*/
	if (tvd->vdev_islog && mg != NULL)
	metaslab_group_activate(mg);
	}

	vd->vdev_tmpoffline = !!(flags & ZFS_OFFLINE_TEMPORARY);

	return (spa_vdev_state_exit(spa, vd, 0));
	}

	int
	vdev_offline(spa_t *spa, uint64_t guid, uint64_t flags)
	{
	int error;

	mutex_enter(&spa->spa_vdev_top_lock);
	error = vdev_offline_locked(spa, guid, flags);
	mutex_exit(&spa->spa_vdev_top_lock);

	return (error);
	}

	/*
	* Clear the error counts associated with this vdev. Unlike vdev_online() and
	* vdev_offline(), we assume the spa config is locked. We also clear all
	* children. If 'vd' is NULL, then the user wants to clear all vdevs.
	*/
	void
	vdev_clear(spa_t spa, vdev_t vd)
	{
	vdev_t *rvd = spa->spa_root_vdev;

	ASSERT(spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL);

	if (vd == NULL)
	vd = rvd;

	vd->vdev_stat.vs_read_errors = 0;
	vd->vdev_stat.vs_write_errors = 0;
	vd->vdev_stat.vs_checksum_errors = 0;
	vd->vdev_stat.vs_slow_ios = 0;

	for (int c = 0; c < vd->vdev_children; c++)
	vdev_clear(spa, vd->vdev_child[c]);

	/*
	* It makes no sense to "clear" an indirect vdev.
	*/
	if (!vdev_is_concrete(vd))
	return;

	/*
	* If we're in the FAULTED state or have experienced failed I/O, then
	* clear the persistent state and attempt to reopen the device. We
	* also mark the vdev config dirty, so that the new faulted state is
	* written out to disk.
	*/
	if (vd->vdev_faulted \|\| vd->vdev_degraded \|\|
	!vdev_readable(vd) \|\| !vdev_writeable(vd)) {
	/*
	* When reopening in response to a clear event, it may be due to
	* a fmadm repair request. In this case, if the device is
	* still broken, we want to still post the ereport again.
	*/
	vd->vdev_forcefault = B_TRUE;

	vd->vdev_faulted = vd->vdev_degraded = 0ULL;
	vd->vdev_cant_read = B_FALSE;
	vd->vdev_cant_write = B_FALSE;
	vd->vdev_stat.vs_aux = 0;

	vdev_reopen(vd == rvd ? rvd : vd->vdev_top);

	vd->vdev_forcefault = B_FALSE;

	if (vd != rvd && vdev_writeable(vd->vdev_top))
	vdev_state_dirty(vd->vdev_top);

	/* If a resilver isn't required, check if vdevs can be culled */
	if (vd->vdev_aux == NULL && !vdev_is_dead(vd) &&
	!dsl_scan_resilvering(spa->spa_dsl_pool) &&
	!dsl_scan_resilver_scheduled(spa->spa_dsl_pool))
	spa_async_request(spa, SPA_ASYNC_RESILVER_DONE);

	spa_event_notify(spa, vd, NULL, ESC_ZFS_VDEV_CLEAR);
	}

	/*
	* When clearing a FMA-diagnosed fault, we always want to
	* unspare the device, as we assume that the original spare was
	* done in response to the FMA fault.
	*/
	if (!vdev_is_dead(vd) && vd->vdev_parent != NULL &&
	vd->vdev_parent->vdev_ops == &vdev_spare_ops &&
	vd->vdev_parent->vdev_child[0] == vd)
	vd->vdev_unspare = B_TRUE;

	/* Clear recent error events cache (i.e. duplicate events tracking) */
	zfs_ereport_clear(spa, vd);
	}

	boolean_t
	vdev_is_dead(vdev_t *vd)
	{
	/*
	* Holes and missing devices are always considered "dead".
	* This simplifies the code since we don't have to check for
	* these types of devices in the various code paths.
	* Instead we rely on the fact that we skip over dead devices
	* before issuing I/O to them.
	*/
	return (vd->vdev_state < VDEV_STATE_DEGRADED \|\|
	vd->vdev_ops == &vdev_hole_ops \|\|
	vd->vdev_ops == &vdev_missing_ops);
	}

	boolean_t
	vdev_readable(vdev_t *vd)
	{
	return (!vdev_is_dead(vd) && !vd->vdev_cant_read);
	}

	boolean_t
	vdev_writeable(vdev_t *vd)
	{
	return (!vdev_is_dead(vd) && !vd->vdev_cant_write &&
	vdev_is_concrete(vd));
	}

	boolean_t
	vdev_allocatable(vdev_t *vd)
	{
	uint64_t state = vd->vdev_state;

	/*
	* We currently allow allocations from vdevs which may be in the
	* process of reopening (i.e. VDEV_STATE_CLOSED). If the device
	* fails to reopen then we'll catch it later when we're holding
	* the proper locks. Note that we have to get the vdev state
	* in a local variable because although it changes atomically,
	* we're asking two separate questions about it.
	*/
	return (!(state < VDEV_STATE_DEGRADED && state != VDEV_STATE_CLOSED) &&
	!vd->vdev_cant_write && vdev_is_concrete(vd) &&
	vd->vdev_mg->mg_initialized);
	}

	boolean_t
	vdev_accessible(vdev_t vd, zio_t zio)
	{
	ASSERT(zio->io_vd == vd);

	if (vdev_is_dead(vd) \|\| vd->vdev_remove_wanted)
	return (B_FALSE);

	if (zio->io_type == ZIO_TYPE_READ)
	return (!vd->vdev_cant_read);

	if (zio->io_type == ZIO_TYPE_WRITE)
	return (!vd->vdev_cant_write);

	return (B_TRUE);
	}

	static void
	vdev_get_child_stat(vdev_t cvd, vdev_stat_t vs, vdev_stat_t *cvs)
	{
	/*
	* Exclude the dRAID spare when aggregating to avoid double counting
	* the ops and bytes. These IOs are counted by the physical leaves.
	*/
	if (cvd->vdev_ops == &vdev_draid_spare_ops)
	return;

	for (int t = 0; t < VS_ZIO_TYPES; t++) {
	vs->vs_ops[t] += cvs->vs_ops[t];
	vs->vs_bytes[t] += cvs->vs_bytes[t];
	}

	cvs->vs_scan_removing = cvd->vdev_removing;
	}

	/*
	* Get extended stats
	*/
	static void
	vdev_get_child_stat_ex(vdev_t cvd, vdev_stat_ex_t vsx, vdev_stat_ex_t *cvsx)
	{
	int t, b;
	for (t = 0; t < ZIO_TYPES; t++) {
	for (b = 0; b < ARRAY_SIZE(vsx->vsx_disk_histo[0]); b++)
	vsx->vsx_disk_histo[t][b] += cvsx->vsx_disk_histo[t][b];

	for (b = 0; b < ARRAY_SIZE(vsx->vsx_total_histo[0]); b++) {
	vsx->vsx_total_histo[t][b] +=
	cvsx->vsx_total_histo[t][b];
	}
	}

	for (t = 0; t < ZIO_PRIORITY_NUM_QUEUEABLE; t++) {
	for (b = 0; b < ARRAY_SIZE(vsx->vsx_queue_histo[0]); b++) {
	vsx->vsx_queue_histo[t][b] +=
	cvsx->vsx_queue_histo[t][b];
	}
	vsx->vsx_active_queue[t] += cvsx->vsx_active_queue[t];
	vsx->vsx_pend_queue[t] += cvsx->vsx_pend_queue[t];

	for (b = 0; b < ARRAY_SIZE(vsx->vsx_ind_histo[0]); b++)
	vsx->vsx_ind_histo[t][b] += cvsx->vsx_ind_histo[t][b];

	for (b = 0; b < ARRAY_SIZE(vsx->vsx_agg_histo[0]); b++)
	vsx->vsx_agg_histo[t][b] += cvsx->vsx_agg_histo[t][b];
	}

	}

	boolean_t
	vdev_is_spacemap_addressable(vdev_t *vd)
	{
	if (spa_feature_is_active(vd->vdev_spa, SPA_FEATURE_SPACEMAP_V2))
	return (B_TRUE);

	/*
	* If double-word space map entries are not enabled we assume
	* 47 bits of the space map entry are dedicated to the entry's
	* offset (see SM_OFFSET_BITS in space_map.h). We then use that
	* to calculate the maximum address that can be described by a
	* space map entry for the given device.
	*/
	uint64_t shift = vd->vdev_ashift + SM_OFFSET_BITS;

	if (shift >= 63) /* detect potential overflow */
	return (B_TRUE);

	return (vd->vdev_asize < (1ULL << shift));
	}

	/*
	* Get statistics for the given vdev.
	*/
	static void
	vdev_get_stats_ex_impl(vdev_t vd, vdev_stat_t vs, vdev_stat_ex_t *vsx)
	{
	int t;
	/*
	* If we're getting stats on the root vdev, aggregate the I/O counts
	* over all top-level vdevs (i.e. the direct children of the root).
	*/
	if (!vd->vdev_ops->vdev_op_leaf) {
	if (vs) {
	memset(vs->vs_ops, 0, sizeof (vs->vs_ops));
	memset(vs->vs_bytes, 0, sizeof (vs->vs_bytes));
	}
	if (vsx)
	memset(vsx, 0, sizeof (*vsx));

	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];
	vdev_stat_t *cvs = &cvd->vdev_stat;
	vdev_stat_ex_t *cvsx = &cvd->vdev_stat_ex;

	vdev_get_stats_ex_impl(cvd, cvs, cvsx);
	if (vs)
	vdev_get_child_stat(cvd, vs, cvs);
	if (vsx)
	vdev_get_child_stat_ex(cvd, vsx, cvsx);
	}
	} else {
	/*
	* We're a leaf. Just copy our ZIO active queue stats in. The
	* other leaf stats are updated in vdev_stat_update().
	*/
	if (!vsx)
	return;

	memcpy(vsx, &vd->vdev_stat_ex, sizeof (vd->vdev_stat_ex));

	for (t = 0; t < ARRAY_SIZE(vd->vdev_queue.vq_class); t++) {
	vsx->vsx_active_queue[t] =
	vd->vdev_queue.vq_class[t].vqc_active;
	vsx->vsx_pend_queue[t] = avl_numnodes(
	&vd->vdev_queue.vq_class[t].vqc_queued_tree);
	}
	}
	}

	void
	vdev_get_stats_ex(vdev_t vd, vdev_stat_t vs, vdev_stat_ex_t *vsx)
	{
	vdev_t *tvd = vd->vdev_top;
	mutex_enter(&vd->vdev_stat_lock);
	if (vs) {
	bcopy(&vd->vdev_stat, vs, sizeof (*vs));
	vs->vs_timestamp = gethrtime() - vs->vs_timestamp;
	vs->vs_state = vd->vdev_state;
	vs->vs_rsize = vdev_get_min_asize(vd);

	if (vd->vdev_ops->vdev_op_leaf) {
	vs->vs_rsize += VDEV_LABEL_START_SIZE +
	VDEV_LABEL_END_SIZE;
	/*
	* Report initializing progress. Since we don't
	* have the initializing locks held, this is only
	* an estimate (although a fairly accurate one).
	*/
	vs->vs_initialize_bytes_done =
	vd->vdev_initialize_bytes_done;
	vs->vs_initialize_bytes_est =
	vd->vdev_initialize_bytes_est;
	vs->vs_initialize_state = vd->vdev_initialize_state;
	vs->vs_initialize_action_time =
	vd->vdev_initialize_action_time;

	/*
	* Report manual TRIM progress. Since we don't have
	* the manual TRIM locks held, this is only an
	* estimate (although fairly accurate one).
	*/
	vs->vs_trim_notsup = !vd->vdev_has_trim;
	vs->vs_trim_bytes_done = vd->vdev_trim_bytes_done;
	vs->vs_trim_bytes_est = vd->vdev_trim_bytes_est;
	vs->vs_trim_state = vd->vdev_trim_state;
	vs->vs_trim_action_time = vd->vdev_trim_action_time;

	/* Set when there is a deferred resilver. */
	vs->vs_resilver_deferred = vd->vdev_resilver_deferred;
	}

	/*
	* Report expandable space on top-level, non-auxiliary devices
	* only. The expandable space is reported in terms of metaslab
	* sized units since that determines how much space the pool
	* can expand.
	*/
	if (vd->vdev_aux == NULL && tvd != NULL) {
	vs->vs_esize = P2ALIGN(
	vd->vdev_max_asize - vd->vdev_asize,
	1ULL << tvd->vdev_ms_shift);
	}

	vs->vs_configured_ashift = vd->vdev_top != NULL
	? vd->vdev_top->vdev_ashift : vd->vdev_ashift;
	vs->vs_logical_ashift = vd->vdev_logical_ashift;
	vs->vs_physical_ashift = vd->vdev_physical_ashift;

	/*
	* Report fragmentation and rebuild progress for top-level,
	* non-auxiliary, concrete devices.
	*/
	if (vd->vdev_aux == NULL && vd == vd->vdev_top &&
	vdev_is_concrete(vd)) {
	/*
	* The vdev fragmentation rating doesn't take into
	* account the embedded slog metaslab (vdev_log_mg).
	* Since it's only one metaslab, it would have a tiny
	* impact on the overall fragmentation.
	*/
	vs->vs_fragmentation = (vd->vdev_mg != NULL) ?
	vd->vdev_mg->mg_fragmentation : 0;
	}
	+ vs->vs_noalloc = MAX(vd->vdev_noalloc,
	+ tvd ? tvd->vdev_noalloc : 0);
	}

	vdev_get_stats_ex_impl(vd, vs, vsx);
	mutex_exit(&vd->vdev_stat_lock);
	}

	void
	vdev_get_stats(vdev_t vd, vdev_stat_t vs)
	{
	return (vdev_get_stats_ex(vd, vs, NULL));
	}

	void
	vdev_clear_stats(vdev_t *vd)
	{
	mutex_enter(&vd->vdev_stat_lock);
	vd->vdev_stat.vs_space = 0;
	vd->vdev_stat.vs_dspace = 0;
	vd->vdev_stat.vs_alloc = 0;
	mutex_exit(&vd->vdev_stat_lock);
	}

	void
	vdev_scan_stat_init(vdev_t *vd)
	{
	vdev_stat_t *vs = &vd->vdev_stat;

	for (int c = 0; c < vd->vdev_children; c++)
	vdev_scan_stat_init(vd->vdev_child[c]);

	mutex_enter(&vd->vdev_stat_lock);
	vs->vs_scan_processed = 0;
	mutex_exit(&vd->vdev_stat_lock);
	}

	void
	vdev_stat_update(zio_t *zio, uint64_t psize)
	{
	spa_t *spa = zio->io_spa;
	vdev_t *rvd = spa->spa_root_vdev;
	vdev_t *vd = zio->io_vd ? zio->io_vd : rvd;
	vdev_t *pvd;
	uint64_t txg = zio->io_txg;
	vdev_stat_t *vs = &vd->vdev_stat;
	vdev_stat_ex_t *vsx = &vd->vdev_stat_ex;
	zio_type_t type = zio->io_type;
	int flags = zio->io_flags;

	/*
	* If this i/o is a gang leader, it didn't do any actual work.
	*/
	if (zio->io_gang_tree)
	return;

	if (zio->io_error == 0) {
	/*
	* If this is a root i/o, don't count it -- we've already
	* counted the top-level vdevs, and vdev_get_stats() will
	* aggregate them when asked. This reduces contention on
	* the root vdev_stat_lock and implicitly handles blocks
	* that compress away to holes, for which there is no i/o.
	* (Holes never create vdev children, so all the counters
	* remain zero, which is what we want.)
	*
	* Note: this only applies to successful i/o (io_error == 0)
	* because unlike i/o counts, errors are not additive.
	* When reading a ditto block, for example, failure of
	* one top-level vdev does not imply a root-level error.
	*/
	if (vd == rvd)
	return;

	ASSERT(vd == zio->io_vd);

	if (flags & ZIO_FLAG_IO_BYPASS)
	return;

	mutex_enter(&vd->vdev_stat_lock);

	if (flags & ZIO_FLAG_IO_REPAIR) {
	/*
	* Repair is the result of a resilver issued by the
	* scan thread (spa_sync).
	*/
	if (flags & ZIO_FLAG_SCAN_THREAD) {
	dsl_scan_t *scn = spa->spa_dsl_pool->dp_scan;
	dsl_scan_phys_t *scn_phys = &scn->scn_phys;
	uint64_t *processed = &scn_phys->scn_processed;

	if (vd->vdev_ops->vdev_op_leaf)
	atomic_add_64(processed, psize);
	vs->vs_scan_processed += psize;
	}

	/*
	* Repair is the result of a rebuild issued by the
	* rebuild thread (vdev_rebuild_thread). To avoid
	* double counting repaired bytes the virtual dRAID
	* spare vdev is excluded from the processed bytes.
	*/
	if (zio->io_priority == ZIO_PRIORITY_REBUILD) {
	vdev_t *tvd = vd->vdev_top;
	vdev_rebuild_t *vr = &tvd->vdev_rebuild_config;
	vdev_rebuild_phys_t *vrp = &vr->vr_rebuild_phys;
	uint64_t *rebuilt = &vrp->vrp_bytes_rebuilt;

	if (vd->vdev_ops->vdev_op_leaf &&
	vd->vdev_ops != &vdev_draid_spare_ops) {
	atomic_add_64(rebuilt, psize);
	}
	vs->vs_rebuild_processed += psize;
	}

	if (flags & ZIO_FLAG_SELF_HEAL)
	vs->vs_self_healed += psize;
	}

	/*
	* The bytes/ops/histograms are recorded at the leaf level and
	* aggregated into the higher level vdevs in vdev_get_stats().
	*/
	if (vd->vdev_ops->vdev_op_leaf &&
	(zio->io_priority < ZIO_PRIORITY_NUM_QUEUEABLE)) {
	zio_type_t vs_type = type;
	zio_priority_t priority = zio->io_priority;

	/*
	* TRIM ops and bytes are reported to user space as
	* ZIO_TYPE_IOCTL. This is done to preserve the
	* vdev_stat_t structure layout for user space.
	*/
	if (type == ZIO_TYPE_TRIM)
	vs_type = ZIO_TYPE_IOCTL;

	/*
	* Solely for the purposes of 'zpool iostat -lqrw'
	* reporting use the priority to categorize the IO.
	* Only the following are reported to user space:
	*
	* ZIO_PRIORITY_SYNC_READ,
	* ZIO_PRIORITY_SYNC_WRITE,
	* ZIO_PRIORITY_ASYNC_READ,
	* ZIO_PRIORITY_ASYNC_WRITE,
	* ZIO_PRIORITY_SCRUB,
	* ZIO_PRIORITY_TRIM,
	* ZIO_PRIORITY_REBUILD.
	*/
	if (priority == ZIO_PRIORITY_INITIALIZING) {
	ASSERT3U(type, ==, ZIO_TYPE_WRITE);
	priority = ZIO_PRIORITY_ASYNC_WRITE;
	} else if (priority == ZIO_PRIORITY_REMOVAL) {
	priority = ((type == ZIO_TYPE_WRITE) ?
	ZIO_PRIORITY_ASYNC_WRITE :
	ZIO_PRIORITY_ASYNC_READ);
	}

	vs->vs_ops[vs_type]++;
	vs->vs_bytes[vs_type] += psize;

	if (flags & ZIO_FLAG_DELEGATED) {
	vsx->vsx_agg_histo[priority]
	[RQ_HISTO(zio->io_size)]++;
	} else {
	vsx->vsx_ind_histo[priority]
	[RQ_HISTO(zio->io_size)]++;
	}

	if (zio->io_delta && zio->io_delay) {
	vsx->vsx_queue_histo[priority]
	[L_HISTO(zio->io_delta - zio->io_delay)]++;
	vsx->vsx_disk_histo[type]
	[L_HISTO(zio->io_delay)]++;
	vsx->vsx_total_histo[type]
	[L_HISTO(zio->io_delta)]++;
	}
	}

	mutex_exit(&vd->vdev_stat_lock);
	return;
	}

	if (flags & ZIO_FLAG_SPECULATIVE)
	return;

	/*
	* If this is an I/O error that is going to be retried, then ignore the
	* error. Otherwise, the user may interpret B_FAILFAST I/O errors as
	* hard errors, when in reality they can happen for any number of
	* innocuous reasons (bus resets, MPxIO link failure, etc).
	*/
	if (zio->io_error == EIO &&
	!(zio->io_flags & ZIO_FLAG_IO_RETRY))
	return;

	/*
	* Intent logs writes won't propagate their error to the root
	* I/O so don't mark these types of failures as pool-level
	* errors.
	*/
	if (zio->io_vd == NULL && (zio->io_flags & ZIO_FLAG_DONT_PROPAGATE))
	return;

	if (type == ZIO_TYPE_WRITE && txg != 0 &&
	(!(flags & ZIO_FLAG_IO_REPAIR) \|\|
	(flags & ZIO_FLAG_SCAN_THREAD) \|\|
	spa->spa_claiming)) {
	/*
	* This is either a normal write (not a repair), or it's
	* a repair induced by the scrub thread, or it's a repair
	* made by zil_claim() during spa_load() in the first txg.
	* In the normal case, we commit the DTL change in the same
	* txg as the block was born. In the scrub-induced repair
	* case, we know that scrubs run in first-pass syncing context,
	* so we commit the DTL change in spa_syncing_txg(spa).
	* In the zil_claim() case, we commit in spa_first_txg(spa).
	*
	* We currently do not make DTL entries for failed spontaneous
	* self-healing writes triggered by normal (non-scrubbing)
	* reads, because we have no transactional context in which to
	* do so -- and it's not clear that it'd be desirable anyway.
	*/
	if (vd->vdev_ops->vdev_op_leaf) {
	uint64_t commit_txg = txg;
	if (flags & ZIO_FLAG_SCAN_THREAD) {
	ASSERT(flags & ZIO_FLAG_IO_REPAIR);
	ASSERT(spa_sync_pass(spa) == 1);
	vdev_dtl_dirty(vd, DTL_SCRUB, txg, 1);
	commit_txg = spa_syncing_txg(spa);
	} else if (spa->spa_claiming) {
	ASSERT(flags & ZIO_FLAG_IO_REPAIR);
	commit_txg = spa_first_txg(spa);
	}
	ASSERT(commit_txg >= spa_syncing_txg(spa));
	if (vdev_dtl_contains(vd, DTL_MISSING, txg, 1))
	return;
	for (pvd = vd; pvd != rvd; pvd = pvd->vdev_parent)
	vdev_dtl_dirty(pvd, DTL_PARTIAL, txg, 1);
	vdev_dirty(vd->vdev_top, VDD_DTL, vd, commit_txg);
	}
	if (vd != rvd)
	vdev_dtl_dirty(vd, DTL_MISSING, txg, 1);
	}
	}

	int64_t
	vdev_deflated_space(vdev_t *vd, int64_t space)
	{
	ASSERT((space & (SPA_MINBLOCKSIZE-1)) == 0);
	ASSERT(vd->vdev_deflate_ratio != 0 \|\| vd->vdev_isl2cache);

	return ((space >> SPA_MINBLOCKSHIFT) * vd->vdev_deflate_ratio);
	}

	/*
	* Update the in-core space usage stats for this vdev, its metaslab class,
	* and the root vdev.
	*/
	void
	vdev_space_update(vdev_t *vd, int64_t alloc_delta, int64_t defer_delta,
	int64_t space_delta)
	{
	int64_t dspace_delta;
	spa_t *spa = vd->vdev_spa;
	vdev_t *rvd = spa->spa_root_vdev;

	ASSERT(vd == vd->vdev_top);

	/*
	* Apply the inverse of the psize-to-asize (ie. RAID-Z) space-expansion
	* factor. We must calculate this here and not at the root vdev
	* because the root vdev's psize-to-asize is simply the max of its
	* children's, thus not accurate enough for us.
	*/
	dspace_delta = vdev_deflated_space(vd, space_delta);

	mutex_enter(&vd->vdev_stat_lock);
	/* ensure we won't underflow */
	if (alloc_delta < 0) {
	ASSERT3U(vd->vdev_stat.vs_alloc, >=, -alloc_delta);
	}

	vd->vdev_stat.vs_alloc += alloc_delta;
	vd->vdev_stat.vs_space += space_delta;
	vd->vdev_stat.vs_dspace += dspace_delta;
	mutex_exit(&vd->vdev_stat_lock);

	/* every class but log contributes to root space stats */
	if (vd->vdev_mg != NULL && !vd->vdev_islog) {
	ASSERT(!vd->vdev_isl2cache);
	mutex_enter(&rvd->vdev_stat_lock);
	rvd->vdev_stat.vs_alloc += alloc_delta;
	rvd->vdev_stat.vs_space += space_delta;
	rvd->vdev_stat.vs_dspace += dspace_delta;
	mutex_exit(&rvd->vdev_stat_lock);
	}
	/* Note: metaslab_class_space_update moved to metaslab_space_update */
	}

	/*
	* Mark a top-level vdev's config as dirty, placing it on the dirty list
	* so that it will be written out next time the vdev configuration is synced.
	* If the root vdev is specified (vdev_top == NULL), dirty all top-level vdevs.
	*/
	void
	vdev_config_dirty(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;
	vdev_t *rvd = spa->spa_root_vdev;
	int c;

	ASSERT(spa_writeable(spa));

	/*
	* If this is an aux vdev (as with l2cache and spare devices), then we
	* update the vdev config manually and set the sync flag.
	*/
	if (vd->vdev_aux != NULL) {
	spa_aux_vdev_t *sav = vd->vdev_aux;
	nvlist_t **aux;
	uint_t naux;

	for (c = 0; c < sav->sav_count; c++) {
	if (sav->sav_vdevs[c] == vd)
	break;
	}

	if (c == sav->sav_count) {
	/*
	* We're being removed. There's nothing more to do.
	*/
	ASSERT(sav->sav_sync == B_TRUE);
	return;
	}

	sav->sav_sync = B_TRUE;

	if (nvlist_lookup_nvlist_array(sav->sav_config,
	ZPOOL_CONFIG_L2CACHE, &aux, &naux) != 0) {
	VERIFY(nvlist_lookup_nvlist_array(sav->sav_config,
	ZPOOL_CONFIG_SPARES, &aux, &naux) == 0);
	}

	ASSERT(c < naux);

	/*
	* Setting the nvlist in the middle if the array is a little
	* sketchy, but it will work.
	*/
	nvlist_free(aux[c]);
	aux[c] = vdev_config_generate(spa, vd, B_TRUE, 0);

	return;
	}

	/*
	* The dirty list is protected by the SCL_CONFIG lock. The caller
	* must either hold SCL_CONFIG as writer, or must be the sync thread
	* (which holds SCL_CONFIG as reader). There's only one sync thread,
	* so this is sufficient to ensure mutual exclusion.
	*/
	ASSERT(spa_config_held(spa, SCL_CONFIG, RW_WRITER) \|\|
	(dsl_pool_sync_context(spa_get_dsl(spa)) &&
	spa_config_held(spa, SCL_CONFIG, RW_READER)));

	if (vd == rvd) {
	for (c = 0; c < rvd->vdev_children; c++)
	vdev_config_dirty(rvd->vdev_child[c]);
	} else {
	ASSERT(vd == vd->vdev_top);

	if (!list_link_active(&vd->vdev_config_dirty_node) &&
	vdev_is_concrete(vd)) {
	list_insert_head(&spa->spa_config_dirty_list, vd);
	}
	}
	}

	void
	vdev_config_clean(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;

	ASSERT(spa_config_held(spa, SCL_CONFIG, RW_WRITER) \|\|
	(dsl_pool_sync_context(spa_get_dsl(spa)) &&
	spa_config_held(spa, SCL_CONFIG, RW_READER)));

	ASSERT(list_link_active(&vd->vdev_config_dirty_node));
	list_remove(&spa->spa_config_dirty_list, vd);
	}

	/*
	* Mark a top-level vdev's state as dirty, so that the next pass of
	* spa_sync() can convert this into vdev_config_dirty(). We distinguish
	* the state changes from larger config changes because they require
	* much less locking, and are often needed for administrative actions.
	*/
	void
	vdev_state_dirty(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;

	ASSERT(spa_writeable(spa));
	ASSERT(vd == vd->vdev_top);

	/*
	* The state list is protected by the SCL_STATE lock. The caller
	* must either hold SCL_STATE as writer, or must be the sync thread
	* (which holds SCL_STATE as reader). There's only one sync thread,
	* so this is sufficient to ensure mutual exclusion.
	*/
	ASSERT(spa_config_held(spa, SCL_STATE, RW_WRITER) \|\|
	(dsl_pool_sync_context(spa_get_dsl(spa)) &&
	spa_config_held(spa, SCL_STATE, RW_READER)));

	if (!list_link_active(&vd->vdev_state_dirty_node) &&
	vdev_is_concrete(vd))
	list_insert_head(&spa->spa_state_dirty_list, vd);
	}

	void
	vdev_state_clean(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;

	ASSERT(spa_config_held(spa, SCL_STATE, RW_WRITER) \|\|
	(dsl_pool_sync_context(spa_get_dsl(spa)) &&
	spa_config_held(spa, SCL_STATE, RW_READER)));

	ASSERT(list_link_active(&vd->vdev_state_dirty_node));
	list_remove(&spa->spa_state_dirty_list, vd);
	}

	/*
	* Propagate vdev state up from children to parent.
	*/
	void
	vdev_propagate_state(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;
	vdev_t *rvd = spa->spa_root_vdev;
	int degraded = 0, faulted = 0;
	int corrupted = 0;
	vdev_t *child;

	if (vd->vdev_children > 0) {
	for (int c = 0; c < vd->vdev_children; c++) {
	child = vd->vdev_child[c];

	/*
	* Don't factor holes or indirect vdevs into the
	* decision.
	*/
	if (!vdev_is_concrete(child))
	continue;

	if (!vdev_readable(child) \|\|
	(!vdev_writeable(child) && spa_writeable(spa))) {
	/*
	* Root special: if there is a top-level log
	* device, treat the root vdev as if it were
	* degraded.
	*/
	if (child->vdev_islog && vd == rvd)
	degraded++;
	else
	faulted++;
	} else if (child->vdev_state <= VDEV_STATE_DEGRADED) {
	degraded++;
	}

	if (child->vdev_stat.vs_aux == VDEV_AUX_CORRUPT_DATA)
	corrupted++;
	}

	vd->vdev_ops->vdev_op_state_change(vd, faulted, degraded);

	/*
	* Root special: if there is a top-level vdev that cannot be
	* opened due to corrupted metadata, then propagate the root
	* vdev's aux state as 'corrupt' rather than 'insufficient
	* replicas'.
	*/
	if (corrupted && vd == rvd &&
	rvd->vdev_state == VDEV_STATE_CANT_OPEN)
	vdev_set_state(rvd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_CORRUPT_DATA);
	}

	if (vd->vdev_parent)
	vdev_propagate_state(vd->vdev_parent);
	}

	/*
	* Set a vdev's state. If this is during an open, we don't update the parent
	* state, because we're in the process of opening children depth-first.
	* Otherwise, we propagate the change to the parent.
	*
	* If this routine places a device in a faulted state, an appropriate ereport is
	* generated.
	*/
	void
	vdev_set_state(vdev_t *vd, boolean_t isopen, vdev_state_t state, vdev_aux_t aux)
	{
	uint64_t save_state;
	spa_t *spa = vd->vdev_spa;

	if (state == vd->vdev_state) {
	/*
	* Since vdev_offline() code path is already in an offline
	* state we can miss a statechange event to OFFLINE. Check
	* the previous state to catch this condition.
	*/
	if (vd->vdev_ops->vdev_op_leaf &&
	(state == VDEV_STATE_OFFLINE) &&
	(vd->vdev_prevstate >= VDEV_STATE_FAULTED)) {
	/* post an offline state change */
	zfs_post_state_change(spa, vd, vd->vdev_prevstate);
	}
	vd->vdev_stat.vs_aux = aux;
	return;
	}

	save_state = vd->vdev_state;

	vd->vdev_state = state;
	vd->vdev_stat.vs_aux = aux;

	/*
	* If we are setting the vdev state to anything but an open state, then
	* always close the underlying device unless the device has requested
	* a delayed close (i.e. we're about to remove or fault the device).
	* Otherwise, we keep accessible but invalid devices open forever.
	* We don't call vdev_close() itself, because that implies some extra
	* checks (offline, etc) that we don't want here. This is limited to
	* leaf devices, because otherwise closing the device will affect other
	* children.
	*/
	if (!vd->vdev_delayed_close && vdev_is_dead(vd) &&
	vd->vdev_ops->vdev_op_leaf)
	vd->vdev_ops->vdev_op_close(vd);

	if (vd->vdev_removed &&
	state == VDEV_STATE_CANT_OPEN &&
	(aux == VDEV_AUX_OPEN_FAILED \|\| vd->vdev_checkremove)) {
	/*
	* If the previous state is set to VDEV_STATE_REMOVED, then this
	* device was previously marked removed and someone attempted to
	* reopen it. If this failed due to a nonexistent device, then
	* keep the device in the REMOVED state. We also let this be if
	* it is one of our special test online cases, which is only
	* attempting to online the device and shouldn't generate an FMA
	* fault.
	*/
	vd->vdev_state = VDEV_STATE_REMOVED;
	vd->vdev_stat.vs_aux = VDEV_AUX_NONE;
	} else if (state == VDEV_STATE_REMOVED) {
	vd->vdev_removed = B_TRUE;
	} else if (state == VDEV_STATE_CANT_OPEN) {
	/*
	* If we fail to open a vdev during an import or recovery, we
	* mark it as "not available", which signifies that it was
	* never there to begin with. Failure to open such a device
	* is not considered an error.
	*/
	if ((spa_load_state(spa) == SPA_LOAD_IMPORT \|\|
	spa_load_state(spa) == SPA_LOAD_RECOVER) &&
	vd->vdev_ops->vdev_op_leaf)
	vd->vdev_not_present = 1;

	/*
	* Post the appropriate ereport. If the 'prevstate' field is
	* set to something other than VDEV_STATE_UNKNOWN, it indicates
	* that this is part of a vdev_reopen(). In this case, we don't
	* want to post the ereport if the device was already in the
	* CANT_OPEN state beforehand.
	*
	* If the 'checkremove' flag is set, then this is an attempt to
	* online the device in response to an insertion event. If we
	* hit this case, then we have detected an insertion event for a
	* faulted or offline device that wasn't in the removed state.
	* In this scenario, we don't post an ereport because we are
	* about to replace the device, or attempt an online with
	* vdev_forcefault, which will generate the fault for us.
	*/
	if ((vd->vdev_prevstate != state \|\| vd->vdev_forcefault) &&
	!vd->vdev_not_present && !vd->vdev_checkremove &&
	vd != spa->spa_root_vdev) {
	const char *class;

	switch (aux) {
	case VDEV_AUX_OPEN_FAILED:
	class = FM_EREPORT_ZFS_DEVICE_OPEN_FAILED;
	break;
	case VDEV_AUX_CORRUPT_DATA:
	class = FM_EREPORT_ZFS_DEVICE_CORRUPT_DATA;
	break;
	case VDEV_AUX_NO_REPLICAS:
	class = FM_EREPORT_ZFS_DEVICE_NO_REPLICAS;
	break;
	case VDEV_AUX_BAD_GUID_SUM:
	class = FM_EREPORT_ZFS_DEVICE_BAD_GUID_SUM;
	break;
	case VDEV_AUX_TOO_SMALL:
	class = FM_EREPORT_ZFS_DEVICE_TOO_SMALL;
	break;
	case VDEV_AUX_BAD_LABEL:
	class = FM_EREPORT_ZFS_DEVICE_BAD_LABEL;
	break;
	case VDEV_AUX_BAD_ASHIFT:
	class = FM_EREPORT_ZFS_DEVICE_BAD_ASHIFT;
	break;
	default:
	class = FM_EREPORT_ZFS_DEVICE_UNKNOWN;
	}

	(void) zfs_ereport_post(class, spa, vd, NULL, NULL,
	save_state);
	}

	/* Erase any notion of persistent removed state */
	vd->vdev_removed = B_FALSE;
	} else {
	vd->vdev_removed = B_FALSE;
	}

	/*
	* Notify ZED of any significant state-change on a leaf vdev.
	*
	*/
	if (vd->vdev_ops->vdev_op_leaf) {
	/* preserve original state from a vdev_reopen() */
	if ((vd->vdev_prevstate != VDEV_STATE_UNKNOWN) &&
	(vd->vdev_prevstate != vd->vdev_state) &&
	(save_state <= VDEV_STATE_CLOSED))
	save_state = vd->vdev_prevstate;

	/* filter out state change due to initial vdev_open */
	if (save_state > VDEV_STATE_CLOSED)
	zfs_post_state_change(spa, vd, save_state);
	}

	if (!isopen && vd->vdev_parent)
	vdev_propagate_state(vd->vdev_parent);
	}

	boolean_t
	vdev_children_are_offline(vdev_t *vd)
	{
	ASSERT(!vd->vdev_ops->vdev_op_leaf);

	for (uint64_t i = 0; i < vd->vdev_children; i++) {
	if (vd->vdev_child[i]->vdev_state != VDEV_STATE_OFFLINE)
	return (B_FALSE);
	}

	return (B_TRUE);
	}

	/*
	* Check the vdev configuration to ensure that it's capable of supporting
	* a root pool. We do not support partial configuration.
	*/
	boolean_t
	vdev_is_bootable(vdev_t *vd)
	{
	if (!vd->vdev_ops->vdev_op_leaf) {
	const char *vdev_type = vd->vdev_ops->vdev_op_type;

	if (strcmp(vdev_type, VDEV_TYPE_MISSING) == 0)
	return (B_FALSE);
	}

	for (int c = 0; c < vd->vdev_children; c++) {
	if (!vdev_is_bootable(vd->vdev_child[c]))
	return (B_FALSE);
	}
	return (B_TRUE);
	}

	boolean_t
	vdev_is_concrete(vdev_t *vd)
	{
	vdev_ops_t *ops = vd->vdev_ops;
	if (ops == &vdev_indirect_ops \|\| ops == &vdev_hole_ops \|\|
	ops == &vdev_missing_ops \|\| ops == &vdev_root_ops) {
	return (B_FALSE);
	} else {
	return (B_TRUE);
	}
	}

	/*
	* Determine if a log device has valid content. If the vdev was
	* removed or faulted in the MOS config then we know that
	* the content on the log device has already been written to the pool.
	*/
	boolean_t
	vdev_log_state_valid(vdev_t *vd)
	{
	if (vd->vdev_ops->vdev_op_leaf && !vd->vdev_faulted &&
	!vd->vdev_removed)
	return (B_TRUE);

	for (int c = 0; c < vd->vdev_children; c++)
	if (vdev_log_state_valid(vd->vdev_child[c]))
	return (B_TRUE);

	return (B_FALSE);
	}

	/*
	* Expand a vdev if possible.
	*/
	void
	vdev_expand(vdev_t *vd, uint64_t txg)
	{
	ASSERT(vd->vdev_top == vd);
	ASSERT(spa_config_held(vd->vdev_spa, SCL_ALL, RW_WRITER) == SCL_ALL);
	ASSERT(vdev_is_concrete(vd));

	vdev_set_deflate_ratio(vd);

	if ((vd->vdev_asize >> vd->vdev_ms_shift) > vd->vdev_ms_count &&
	vdev_is_concrete(vd)) {
	vdev_metaslab_group_create(vd);
	VERIFY(vdev_metaslab_init(vd, txg) == 0);
	vdev_config_dirty(vd);
	}
	}

	/*
	* Split a vdev.
	*/
	void
	vdev_split(vdev_t *vd)
	{
	vdev_t cvd, pvd = vd->vdev_parent;

	vdev_remove_child(pvd, vd);
	vdev_compact_children(pvd);

	cvd = pvd->vdev_child[0];
	if (pvd->vdev_children == 1) {
	vdev_remove_parent(cvd);
	cvd->vdev_splitting = B_TRUE;
	}
	vdev_propagate_state(cvd);
	}

	void
	vdev_deadman(vdev_t vd, char tag)
	{
	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];

	vdev_deadman(cvd, tag);
	}

	if (vd->vdev_ops->vdev_op_leaf) {
	vdev_queue_t *vq = &vd->vdev_queue;

	mutex_enter(&vq->vq_lock);
	if (avl_numnodes(&vq->vq_active_tree) > 0) {
	spa_t *spa = vd->vdev_spa;
	zio_t *fio;
	uint64_t delta;

	zfs_dbgmsg("slow vdev: %s has %lu active IOs",
	vd->vdev_path, avl_numnodes(&vq->vq_active_tree));

	/*
	* Look at the head of all the pending queues,
	* if any I/O has been outstanding for longer than
	* the spa_deadman_synctime invoke the deadman logic.
	*/
	fio = avl_first(&vq->vq_active_tree);
	delta = gethrtime() - fio->io_timestamp;
	if (delta > spa_deadman_synctime(spa))
	zio_deadman(fio, tag);
	}
	mutex_exit(&vq->vq_lock);
	}
	}

	void
	vdev_defer_resilver(vdev_t *vd)
	{
	ASSERT(vd->vdev_ops->vdev_op_leaf);

	vd->vdev_resilver_deferred = B_TRUE;
	vd->vdev_spa->spa_resilver_deferred = B_TRUE;
	}

	/*
	* Clears the resilver deferred flag on all leaf devs under vd. Returns
	* B_TRUE if we have devices that need to be resilvered and are available to
	* accept resilver I/Os.
	*/
	boolean_t
	vdev_clear_resilver_deferred(vdev_t vd, dmu_tx_t tx)
	{
	boolean_t resilver_needed = B_FALSE;
	spa_t *spa = vd->vdev_spa;

	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];
	resilver_needed \|= vdev_clear_resilver_deferred(cvd, tx);
	}

	if (vd == spa->spa_root_vdev &&
	spa_feature_is_active(spa, SPA_FEATURE_RESILVER_DEFER)) {
	spa_feature_decr(spa, SPA_FEATURE_RESILVER_DEFER, tx);
	vdev_config_dirty(vd);
	spa->spa_resilver_deferred = B_FALSE;
	return (resilver_needed);
	}

	if (!vdev_is_concrete(vd) \|\| vd->vdev_aux \|\|
	!vd->vdev_ops->vdev_op_leaf)
	return (resilver_needed);

	vd->vdev_resilver_deferred = B_FALSE;

	return (!vdev_is_dead(vd) && !vd->vdev_offline &&
	vdev_resilver_needed(vd, NULL, NULL));
	}

	boolean_t
	vdev_xlate_is_empty(range_seg64_t *rs)
	{
	return (rs->rs_start == rs->rs_end);
	}

	/*
	* Translate a logical range to the first contiguous physical range for the
	* specified vdev_t. This function is initially called with a leaf vdev and
	* will walk each parent vdev until it reaches a top-level vdev. Once the
	* top-level is reached the physical range is initialized and the recursive
	* function begins to unwind. As it unwinds it calls the parent's vdev
	* specific translation function to do the real conversion.
	*/
	void
	vdev_xlate(vdev_t vd, const range_seg64_t logical_rs,
	range_seg64_t physical_rs, range_seg64_t remain_rs)
	{
	/*
	* Walk up the vdev tree
	*/
	if (vd != vd->vdev_top) {
	vdev_xlate(vd->vdev_parent, logical_rs, physical_rs,
	remain_rs);
	} else {
	/*
	* We've reached the top-level vdev, initialize the physical
	* range to the logical range and set an empty remaining
	* range then start to unwind.
	*/
	physical_rs->rs_start = logical_rs->rs_start;
	physical_rs->rs_end = logical_rs->rs_end;

	remain_rs->rs_start = logical_rs->rs_start;
	remain_rs->rs_end = logical_rs->rs_start;

	return;
	}

	vdev_t *pvd = vd->vdev_parent;
	ASSERT3P(pvd, !=, NULL);
	ASSERT3P(pvd->vdev_ops->vdev_op_xlate, !=, NULL);

	/*
	* As this recursive function unwinds, translate the logical
	* range into its physical and any remaining components by calling
	* the vdev specific translate function.
	*/
	range_seg64_t intermediate = { 0 };
	pvd->vdev_ops->vdev_op_xlate(vd, physical_rs, &intermediate, remain_rs);

	physical_rs->rs_start = intermediate.rs_start;
	physical_rs->rs_end = intermediate.rs_end;
	}

	void
	vdev_xlate_walk(vdev_t vd, const range_seg64_t logical_rs,
	vdev_xlate_func_t func, void arg)
	{
	range_seg64_t iter_rs = *logical_rs;
	range_seg64_t physical_rs;
	range_seg64_t remain_rs;

	while (!vdev_xlate_is_empty(&iter_rs)) {

	vdev_xlate(vd, &iter_rs, &physical_rs, &remain_rs);

	/*
	* With raidz and dRAID, it's possible that the logical range
	* does not live on this leaf vdev. Only when there is a non-
	* zero physical size call the provided function.
	*/
	if (!vdev_xlate_is_empty(&physical_rs))
	func(arg, &physical_rs);

	iter_rs = remain_rs;
	}
	}

	+static char *
	+vdev_name(vdev_t vd, char buf, int buflen)
	+{
	+ if (vd->vdev_path == NULL) {
	+ if (strcmp(vd->vdev_ops->vdev_op_type, "root") == 0) {
	+ strlcpy(buf, vd->vdev_spa->spa_name, buflen);
	+ } else if (!vd->vdev_ops->vdev_op_leaf) {
	+ snprintf(buf, buflen, "%s-%llu",
	+ vd->vdev_ops->vdev_op_type,
	+ (u_longlong_t)vd->vdev_id);
	+ }
	+ } else {
	+ strlcpy(buf, vd->vdev_path, buflen);
	+ }
	+ return (buf);
	+}
	+
	/*
	* Look at the vdev tree and determine whether any devices are currently being
	* replaced.
	*/
	boolean_t
	vdev_replace_in_progress(vdev_t *vdev)
	{
	ASSERT(spa_config_held(vdev->vdev_spa, SCL_ALL, RW_READER) != 0);

	if (vdev->vdev_ops == &vdev_replacing_ops)
	return (B_TRUE);

	/*
	* A 'spare' vdev indicates that we have a replace in progress, unless
	* it has exactly two children, and the second, the hot spare, has
	* finished being resilvered.
	*/
	if (vdev->vdev_ops == &vdev_spare_ops && (vdev->vdev_children > 2 \|\|
	!vdev_dtl_empty(vdev->vdev_child[1], DTL_MISSING)))
	return (B_TRUE);

	for (int i = 0; i < vdev->vdev_children; i++) {
	if (vdev_replace_in_progress(vdev->vdev_child[i]))
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	+/*
	+ * Add a (source=src, propname=propval) list to an nvlist.
	+ */
	+static void
	+vdev_prop_add_list(nvlist_t nvl, const char propname, char *strval,
	+ uint64_t intval, zprop_source_t src)
	+{
	+ nvlist_t *propval;
	+
	+ propval = fnvlist_alloc();
	+ fnvlist_add_uint64(propval, ZPROP_SOURCE, src);
	+
	+ if (strval != NULL)
	+ fnvlist_add_string(propval, ZPROP_VALUE, strval);
	+ else
	+ fnvlist_add_uint64(propval, ZPROP_VALUE, intval);
	+
	+ fnvlist_add_nvlist(nvl, propname, propval);
	+ nvlist_free(propval);
	+}
	+
	+static void
	+vdev_props_set_sync(void arg, dmu_tx_t tx)
	+{
	+ vdev_t *vd;
	+ nvlist_t *nvp = arg;
	+ spa_t *spa = dmu_tx_pool(tx)->dp_spa;
	+ objset_t *mos = spa->spa_meta_objset;
	+ nvpair_t *elem = NULL;
	+ uint64_t vdev_guid;
	+ nvlist_t *nvprops;
	+
	+ vdev_guid = fnvlist_lookup_uint64(nvp, ZPOOL_VDEV_PROPS_SET_VDEV);
	+ nvprops = fnvlist_lookup_nvlist(nvp, ZPOOL_VDEV_PROPS_SET_PROPS);
	+ vd = spa_lookup_by_guid(spa, vdev_guid, B_TRUE);
	+ VERIFY(vd != NULL);
	+
	+ mutex_enter(&spa->spa_props_lock);
	+
	+ while ((elem = nvlist_next_nvpair(nvprops, elem)) != NULL) {
	+ uint64_t intval, objid = 0;
	+ char *strval;
	+ vdev_prop_t prop;
	+ const char *propname = nvpair_name(elem);
	+ zprop_type_t proptype;
	+
	+ /*
	+ * Set vdev property values in the vdev props mos object.
	+ */
	+ if (vd->vdev_top_zap != 0) {
	+ objid = vd->vdev_top_zap;
	+ } else if (vd->vdev_leaf_zap != 0) {
	+ objid = vd->vdev_leaf_zap;
	+ } else {
	+ panic("vdev not top or leaf");
	+ }
	+
	+ switch (prop = vdev_name_to_prop(propname)) {
	+ case VDEV_PROP_USER:
	+ if (vdev_prop_user(propname)) {
	+ strval = fnvpair_value_string(elem);
	+ if (strlen(strval) == 0) {
	+ /* remove the property if value == "" */
	+ (void) zap_remove(mos, objid, propname,
	+ tx);
	+ } else {
	+ VERIFY0(zap_update(mos, objid, propname,
	+ 1, strlen(strval) + 1, strval, tx));
	+ }
	+ spa_history_log_internal(spa, "vdev set", tx,
	+ "vdev_guid=%llu: %s=%s",
	+ (u_longlong_t)vdev_guid, nvpair_name(elem),
	+ strval);
	+ }
	+ break;
	+ default:
	+ /* normalize the property name */
	+ propname = vdev_prop_to_name(prop);
	+ proptype = vdev_prop_get_type(prop);
	+
	+ if (nvpair_type(elem) == DATA_TYPE_STRING) {
	+ ASSERT(proptype == PROP_TYPE_STRING);
	+ strval = fnvpair_value_string(elem);
	+ VERIFY0(zap_update(mos, objid, propname,
	+ 1, strlen(strval) + 1, strval, tx));
	+ spa_history_log_internal(spa, "vdev set", tx,
	+ "vdev_guid=%llu: %s=%s",
	+ (u_longlong_t)vdev_guid, nvpair_name(elem),
	+ strval);
	+ } else if (nvpair_type(elem) == DATA_TYPE_UINT64) {
	+ intval = fnvpair_value_uint64(elem);
	+
	+ if (proptype == PROP_TYPE_INDEX) {
	+ const char *unused;
	+ VERIFY0(vdev_prop_index_to_string(
	+ prop, intval, &unused));
	+ }
	+ VERIFY0(zap_update(mos, objid, propname,
	+ sizeof (uint64_t), 1, &intval, tx));
	+ spa_history_log_internal(spa, "vdev set", tx,
	+ "vdev_guid=%llu: %s=%lld",
	+ (u_longlong_t)vdev_guid,
	+ nvpair_name(elem), (longlong_t)intval);
	+ } else {
	+ panic("invalid vdev property type %u",
	+ nvpair_type(elem));
	+ }
	+ }
	+
	+ }
	+
	+ mutex_exit(&spa->spa_props_lock);
	+}
	+
	+int
	+vdev_prop_set(vdev_t vd, nvlist_t innvl, nvlist_t *outnvl)
	+{
	+ spa_t *spa = vd->vdev_spa;
	+ nvpair_t *elem = NULL;
	+ uint64_t vdev_guid;
	+ nvlist_t *nvprops;
	+ int error;
	+
	+ ASSERT(vd != NULL);
	+
	+ if (nvlist_lookup_uint64(innvl, ZPOOL_VDEV_PROPS_SET_VDEV,
	+ &vdev_guid) != 0)
	+ return (SET_ERROR(EINVAL));
	+
	+ if (nvlist_lookup_nvlist(innvl, ZPOOL_VDEV_PROPS_SET_PROPS,
	+ &nvprops) != 0)
	+ return (SET_ERROR(EINVAL));
	+
	+ if ((vd = spa_lookup_by_guid(spa, vdev_guid, B_TRUE)) == NULL)
	+ return (SET_ERROR(EINVAL));
	+
	+ while ((elem = nvlist_next_nvpair(nvprops, elem)) != NULL) {
	+ char *propname = nvpair_name(elem);
	+ vdev_prop_t prop = vdev_name_to_prop(propname);
	+ uint64_t intval = 0;
	+ char *strval = NULL;
	+
	+ if (prop == VDEV_PROP_USER && !vdev_prop_user(propname)) {
	+ error = EINVAL;
	+ goto end;
	+ }
	+
	+ if (vdev_prop_readonly(prop)) {
	+ error = EROFS;
	+ goto end;
	+ }
	+
	+ /* Special Processing */
	+ switch (prop) {
	+ case VDEV_PROP_PATH:
	+ if (vd->vdev_path == NULL) {
	+ error = EROFS;
	+ break;
	+ }
	+ if (nvpair_value_string(elem, &strval) != 0) {
	+ error = EINVAL;
	+ break;
	+ }
	+ /* New path must start with /dev/ */
	+ if (strncmp(strval, "/dev/", 5)) {
	+ error = EINVAL;
	+ break;
	+ }
	+ error = spa_vdev_setpath(spa, vdev_guid, strval);
	+ break;
	+ case VDEV_PROP_ALLOCATING:
	+ if (nvpair_value_uint64(elem, &intval) != 0) {
	+ error = EINVAL;
	+ break;
	+ }
	+ if (intval != vd->vdev_noalloc)
	+ break;
	+ if (intval == 0)
	+ error = spa_vdev_noalloc(spa, vdev_guid);
	+ else
	+ error = spa_vdev_alloc(spa, vdev_guid);
	+ break;
	+ default:
	+ /* Most processing is done in vdev_props_set_sync */
	+ break;
	+ }
	+end:
	+ if (error != 0) {
	+ intval = error;
	+ vdev_prop_add_list(outnvl, propname, strval, intval, 0);
	+ return (error);
	+ }
	+ }
	+
	+ return (dsl_sync_task(spa->spa_name, NULL, vdev_props_set_sync,
	+ innvl, 6, ZFS_SPACE_CHECK_EXTRA_RESERVED));
	+}
	+
	+int
	+vdev_prop_get(vdev_t vd, nvlist_t innvl, nvlist_t *outnvl)
	+{
	+ spa_t *spa = vd->vdev_spa;
	+ objset_t *mos = spa->spa_meta_objset;
	+ int err = 0;
	+ uint64_t objid;
	+ uint64_t vdev_guid;
	+ nvpair_t *elem = NULL;
	+ nvlist_t *nvprops = NULL;
	+ uint64_t intval = 0;
	+ char *strval = NULL;
	+ const char *propname = NULL;
	+ vdev_prop_t prop;
	+
	+ ASSERT(vd != NULL);
	+ ASSERT(mos != NULL);
	+
	+ if (nvlist_lookup_uint64(innvl, ZPOOL_VDEV_PROPS_GET_VDEV,
	+ &vdev_guid) != 0)
	+ return (SET_ERROR(EINVAL));
	+
	+ nvlist_lookup_nvlist(innvl, ZPOOL_VDEV_PROPS_GET_PROPS, &nvprops);
	+
	+ if (vd->vdev_top_zap != 0) {
	+ objid = vd->vdev_top_zap;
	+ } else if (vd->vdev_leaf_zap != 0) {
	+ objid = vd->vdev_leaf_zap;
	+ } else {
	+ return (SET_ERROR(EINVAL));
	+ }
	+ ASSERT(objid != 0);
	+
	+ mutex_enter(&spa->spa_props_lock);
	+
	+ if (nvprops != NULL) {
	+ char namebuf[64] = { 0 };
	+
	+ while ((elem = nvlist_next_nvpair(nvprops, elem)) != NULL) {
	+ intval = 0;
	+ strval = NULL;
	+ propname = nvpair_name(elem);
	+ prop = vdev_name_to_prop(propname);
	+ zprop_source_t src = ZPROP_SRC_DEFAULT;
	+ uint64_t integer_size, num_integers;
	+
	+ switch (prop) {
	+ /* Special Read-only Properties */
	+ case VDEV_PROP_NAME:
	+ strval = vdev_name(vd, namebuf,
	+ sizeof (namebuf));
	+ if (strval == NULL)
	+ continue;
	+ vdev_prop_add_list(outnvl, propname, strval, 0,
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_CAPACITY:
	+ /* percent used */
	+ intval = (vd->vdev_stat.vs_dspace == 0) ? 0 :
	+ (vd->vdev_stat.vs_alloc * 100 /
	+ vd->vdev_stat.vs_dspace);
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ intval, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_STATE:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_state, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_GUID:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_guid, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_ASIZE:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_asize, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_PSIZE:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_psize, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_ASHIFT:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_ashift, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_SIZE:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_dspace, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_FREE:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_dspace -
	+ vd->vdev_stat.vs_alloc, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_ALLOCATED:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_alloc, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_EXPANDSZ:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_esize, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_FRAGMENTATION:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_fragmentation,
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_PARITY:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vdev_get_nparity(vd), ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_PATH:
	+ if (vd->vdev_path == NULL)
	+ continue;
	+ vdev_prop_add_list(outnvl, propname,
	+ vd->vdev_path, 0, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_DEVID:
	+ if (vd->vdev_devid == NULL)
	+ continue;
	+ vdev_prop_add_list(outnvl, propname,
	+ vd->vdev_devid, 0, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_PHYS_PATH:
	+ if (vd->vdev_physpath == NULL)
	+ continue;
	+ vdev_prop_add_list(outnvl, propname,
	+ vd->vdev_physpath, 0, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_ENC_PATH:
	+ if (vd->vdev_enc_sysfs_path == NULL)
	+ continue;
	+ vdev_prop_add_list(outnvl, propname,
	+ vd->vdev_enc_sysfs_path, 0, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_FRU:
	+ if (vd->vdev_fru == NULL)
	+ continue;
	+ vdev_prop_add_list(outnvl, propname,
	+ vd->vdev_fru, 0, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_PARENT:
	+ if (vd->vdev_parent != NULL) {
	+ strval = vdev_name(vd->vdev_parent,
	+ namebuf, sizeof (namebuf));
	+ vdev_prop_add_list(outnvl, propname,
	+ strval, 0, ZPROP_SRC_NONE);
	+ }
	+ continue;
	+ case VDEV_PROP_CHILDREN:
	+ if (vd->vdev_children > 0)
	+ strval = kmem_zalloc(ZAP_MAXVALUELEN,
	+ KM_SLEEP);
	+ for (uint64_t i = 0; i < vd->vdev_children;
	+ i++) {
	+ char *vname;
	+
	+ vname = vdev_name(vd->vdev_child[i],
	+ namebuf, sizeof (namebuf));
	+ if (vname == NULL)
	+ vname = "(unknown)";
	+ if (strlen(strval) > 0)
	+ strlcat(strval, ",",
	+ ZAP_MAXVALUELEN);
	+ strlcat(strval, vname, ZAP_MAXVALUELEN);
	+ }
	+ if (strval != NULL) {
	+ vdev_prop_add_list(outnvl, propname,
	+ strval, 0, ZPROP_SRC_NONE);
	+ kmem_free(strval, ZAP_MAXVALUELEN);
	+ }
	+ continue;
	+ case VDEV_PROP_NUMCHILDREN:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_children, ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_READ_ERRORS:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_read_errors,
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_WRITE_ERRORS:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_write_errors,
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_CHECKSUM_ERRORS:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_checksum_errors,
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_INITIALIZE_ERRORS:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_initialize_errors,
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_OPS_NULL:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_ops[ZIO_TYPE_NULL],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_OPS_READ:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_ops[ZIO_TYPE_READ],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_OPS_WRITE:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_ops[ZIO_TYPE_WRITE],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_OPS_FREE:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_ops[ZIO_TYPE_FREE],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_OPS_CLAIM:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_ops[ZIO_TYPE_CLAIM],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_OPS_TRIM:
	+ /*
	+ * TRIM ops and bytes are reported to user
	+ * space as ZIO_TYPE_IOCTL. This is done to
	+ * preserve the vdev_stat_t structure layout
	+ * for user space.
	+ */
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_ops[ZIO_TYPE_IOCTL],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_BYTES_NULL:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_bytes[ZIO_TYPE_NULL],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_BYTES_READ:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_bytes[ZIO_TYPE_READ],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_BYTES_WRITE:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_bytes[ZIO_TYPE_WRITE],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_BYTES_FREE:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_bytes[ZIO_TYPE_FREE],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_BYTES_CLAIM:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_bytes[ZIO_TYPE_CLAIM],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_BYTES_TRIM:
	+ /*
	+ * TRIM ops and bytes are reported to user
	+ * space as ZIO_TYPE_IOCTL. This is done to
	+ * preserve the vdev_stat_t structure layout
	+ * for user space.
	+ */
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_stat.vs_bytes[ZIO_TYPE_IOCTL],
	+ ZPROP_SRC_NONE);
	+ continue;
	+ case VDEV_PROP_REMOVING:
	+ vdev_prop_add_list(outnvl, propname, NULL,
	+ vd->vdev_removing, ZPROP_SRC_NONE);
	+ continue;
	+ /* Numeric Properites */
	+ case VDEV_PROP_ALLOCATING:
	+ src = ZPROP_SRC_LOCAL;
	+ strval = NULL;
	+
	+ err = zap_lookup(mos, objid, nvpair_name(elem),
	+ sizeof (uint64_t), 1, &intval);
	+ if (err == ENOENT) {
	+ intval =
	+ vdev_prop_default_numeric(prop);
	+ err = 0;
	+ } else if (err)
	+ break;
	+ if (intval == vdev_prop_default_numeric(prop))
	+ src = ZPROP_SRC_DEFAULT;
	+
	+ /* Leaf vdevs cannot have this property */
	+ if (vd->vdev_mg == NULL &&
	+ vd->vdev_top != NULL) {
	+ src = ZPROP_SRC_NONE;
	+ intval = ZPROP_BOOLEAN_NA;
	+ }
	+
	+ vdev_prop_add_list(outnvl, propname, strval,
	+ intval, src);
	+ break;
	+ /* Text Properties */
	+ case VDEV_PROP_COMMENT:
	+ /* Exists in the ZAP below */
	+ /* FALLTHRU */
	+ case VDEV_PROP_USER:
	+ /* User Properites */
	+ src = ZPROP_SRC_LOCAL;
	+
	+ err = zap_length(mos, objid, nvpair_name(elem),
	+ &integer_size, &num_integers);
	+ if (err)
	+ break;
	+
	+ switch (integer_size) {
	+ case 8:
	+ /* User properties cannot be integers */
	+ err = EINVAL;
	+ break;
	+ case 1:
	+ /* string property */
	+ strval = kmem_alloc(num_integers,
	+ KM_SLEEP);
	+ err = zap_lookup(mos, objid,
	+ nvpair_name(elem), 1,
	+ num_integers, strval);
	+ if (err) {
	+ kmem_free(strval,
	+ num_integers);
	+ break;
	+ }
	+ vdev_prop_add_list(outnvl, propname,
	+ strval, 0, src);
	+ kmem_free(strval, num_integers);
	+ break;
	+ }
	+ break;
	+ default:
	+ err = ENOENT;
	+ break;
	+ }
	+ if (err)
	+ break;
	+ }
	+ } else {
	+ /*
	+ * Get all properties from the MOS vdev property object.
	+ */
	+ zap_cursor_t zc;
	+ zap_attribute_t za;
	+ for (zap_cursor_init(&zc, mos, objid);
	+ (err = zap_cursor_retrieve(&zc, &za)) == 0;
	+ zap_cursor_advance(&zc)) {
	+ intval = 0;
	+ strval = NULL;
	+ zprop_source_t src = ZPROP_SRC_DEFAULT;
	+ propname = za.za_name;
	+ prop = vdev_name_to_prop(propname);
	+
	+ switch (za.za_integer_length) {
	+ case 8:
	+ /* We do not allow integer user properties */
	+ /* This is likely an internal value */
	+ break;
	+ case 1:
	+ /* string property */
	+ strval = kmem_alloc(za.za_num_integers,
	+ KM_SLEEP);
	+ err = zap_lookup(mos, objid, za.za_name, 1,
	+ za.za_num_integers, strval);
	+ if (err) {
	+ kmem_free(strval, za.za_num_integers);
	+ break;
	+ }
	+ vdev_prop_add_list(outnvl, propname, strval, 0,
	+ src);
	+ kmem_free(strval, za.za_num_integers);
	+ break;
	+
	+ default:
	+ break;
	+ }
	+ }
	+ zap_cursor_fini(&zc);
	+ }
	+
	+ mutex_exit(&spa->spa_props_lock);
	+ if (err && err != ENOENT) {
	+ return (err);
	+ }
	+
	+ return (0);
	+}
	+
	EXPORT_SYMBOL(vdev_fault);
	EXPORT_SYMBOL(vdev_degrade);
	EXPORT_SYMBOL(vdev_online);
	EXPORT_SYMBOL(vdev_offline);
	EXPORT_SYMBOL(vdev_clear);

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, default_ms_count, INT, ZMOD_RW,
	"Target number of metaslabs per top-level vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, default_ms_shift, INT, ZMOD_RW,
	"Default limit for metaslab size");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, min_ms_count, INT, ZMOD_RW,
	"Minimum number of metaslabs per top-level vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, ms_count_limit, INT, ZMOD_RW,
	"Practical upper limit of total metaslabs per top-level vdev");

	ZFS_MODULE_PARAM(zfs, zfs_, slow_io_events_per_second, UINT, ZMOD_RW,
	"Rate limit slow IO (delay) events to this many per second");

	ZFS_MODULE_PARAM(zfs, zfs_, checksum_events_per_second, UINT, ZMOD_RW,
	"Rate limit checksum events to this many checksum errors per second "
	"(do not set below zed threshold).");

	ZFS_MODULE_PARAM(zfs, zfs_, scan_ignore_errors, INT, ZMOD_RW,
	"Ignore errors during resilver/scrub");

	ZFS_MODULE_PARAM(zfs_vdev, vdev_, validate_skip, INT, ZMOD_RW,
	"Bypass vdev_validate()");

	ZFS_MODULE_PARAM(zfs, zfs_, nocacheflush, INT, ZMOD_RW,
	"Disable cache flushes");

	ZFS_MODULE_PARAM(zfs, zfs_, embedded_slog_min_ms, INT, ZMOD_RW,
	"Minimum number of metaslabs required to dedicate one for log blocks");

	ZFS_MODULE_PARAM_CALL(zfs_vdev, zfs_vdev_, min_auto_ashift,
	param_set_min_auto_ashift, param_get_ulong, ZMOD_RW,
	"Minimum ashift used when creating new top-level vdevs");

	ZFS_MODULE_PARAM_CALL(zfs_vdev, zfs_vdev_, max_auto_ashift,
	param_set_max_auto_ashift, param_get_ulong, ZMOD_RW,
	"Maximum ashift used when optimizing for logical -> physical sector "
	"size on new top-level vdevs");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/zfs/vdev_cache.c b/sys/contrib/openzfs/module/zfs/vdev_cache.c
	index 6e82184b800d..35ed1a3352fe 100644
	--- a/sys/contrib/openzfs/module/zfs/vdev_cache.c
	+++ b/sys/contrib/openzfs/module/zfs/vdev_cache.c
	@@ -1,437 +1,438 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright 2009 Sun Microsystems, Inc. All rights reserved.
	* Use is subject to license terms.
	*/
	/*
	* Copyright (c) 2013, 2016 by Delphix. All rights reserved.
	*/

	#include <sys/zfs_context.h>
	#include <sys/spa.h>
	#include <sys/vdev_impl.h>
	#include <sys/zio.h>
	#include <sys/kstat.h>
	#include <sys/abd.h>

	/*
	* Virtual device read-ahead caching.
	*
	* This file implements a simple LRU read-ahead cache. When the DMU reads
	* a given block, it will often want other, nearby blocks soon thereafter.
	* We take advantage of this by reading a larger disk region and caching
	* the result. In the best case, this can turn 128 back-to-back 512-byte
	* reads into a single 64k read followed by 127 cache hits; this reduces
	* latency dramatically. In the worst case, it can turn an isolated 512-byte
	* read into a 64k read, which doesn't affect latency all that much but is
	* terribly wasteful of bandwidth. A more intelligent version of the cache
	* could keep track of access patterns and not do read-ahead unless it sees
	* at least two temporally close I/Os to the same region. Currently, only
	* metadata I/O is inflated. A further enhancement could take advantage of
	* more semantic information about the I/O. And it could use something
	* faster than an AVL tree; that was chosen solely for convenience.
	*
	* There are five cache operations: allocate, fill, read, write, evict.
	*
	* (1) Allocate. This reserves a cache entry for the specified region.
	* We separate the allocate and fill operations so that multiple threads
	* don't generate I/O for the same cache miss.
	*
	* (2) Fill. When the I/O for a cache miss completes, the fill routine
	* places the data in the previously allocated cache entry.
	*
	* (3) Read. Read data from the cache.
	*
	* (4) Write. Update cache contents after write completion.
	*
	* (5) Evict. When allocating a new entry, we evict the oldest (LRU) entry
	* if the total cache size exceeds zfs_vdev_cache_size.
	*/

	/*
	* These tunables are for performance analysis.
	*/
	/*
	* All i/os smaller than zfs_vdev_cache_max will be turned into
	* 1<<zfs_vdev_cache_bshift byte reads by the vdev_cache (aka software
	* track buffer). At most zfs_vdev_cache_size bytes will be kept in each
	* vdev's vdev_cache.
	*
	* TODO: Note that with the current ZFS code, it turns out that the
	* vdev cache is not helpful, and in some cases actually harmful. It
	* is better if we disable this. Once some time has passed, we should
	* actually remove this to simplify the code. For now we just disable
	* it by setting the zfs_vdev_cache_size to zero. Note that Solaris 11
	* has made these same changes.
	*/
	int zfs_vdev_cache_max = 1<<14; /* 16KB */
	int zfs_vdev_cache_size = 0;
	int zfs_vdev_cache_bshift = 16;

	#define VCBS (1 << zfs_vdev_cache_bshift) /* 64KB */

	kstat_t *vdc_ksp = NULL;

	typedef struct vdc_stats {
	kstat_named_t vdc_stat_delegations;
	kstat_named_t vdc_stat_hits;
	kstat_named_t vdc_stat_misses;
	} vdc_stats_t;

	static vdc_stats_t vdc_stats = {
	{ "delegations", KSTAT_DATA_UINT64 },
	{ "hits", KSTAT_DATA_UINT64 },
	{ "misses", KSTAT_DATA_UINT64 }
	};

	#define VDCSTAT_BUMP(stat) atomic_inc_64(&vdc_stats.stat.value.ui64);

	static inline int
	vdev_cache_offset_compare(const void a1, const void a2)
	{
	const vdev_cache_entry_t ve1 = (const vdev_cache_entry_t )a1;
	const vdev_cache_entry_t ve2 = (const vdev_cache_entry_t )a2;

	return (TREE_CMP(ve1->ve_offset, ve2->ve_offset));
	}

	static int
	vdev_cache_lastused_compare(const void a1, const void a2)
	{
	const vdev_cache_entry_t ve1 = (const vdev_cache_entry_t )a1;
	const vdev_cache_entry_t ve2 = (const vdev_cache_entry_t )a2;

	int cmp = TREE_CMP(ve1->ve_lastused, ve2->ve_lastused);
	if (likely(cmp))
	return (cmp);

	/*
	* Among equally old entries, sort by offset to ensure uniqueness.
	*/
	return (vdev_cache_offset_compare(a1, a2));
	}

	/*
	* Evict the specified entry from the cache.
	*/
	static void
	vdev_cache_evict(vdev_cache_t vc, vdev_cache_entry_t ve)
	{
	ASSERT(MUTEX_HELD(&vc->vc_lock));
	ASSERT3P(ve->ve_fill_io, ==, NULL);
	ASSERT3P(ve->ve_abd, !=, NULL);

	avl_remove(&vc->vc_lastused_tree, ve);
	avl_remove(&vc->vc_offset_tree, ve);
	abd_free(ve->ve_abd);
	kmem_free(ve, sizeof (vdev_cache_entry_t));
	}

	/*
	* Allocate an entry in the cache. At the point we don't have the data,
	* we're just creating a placeholder so that multiple threads don't all
	* go off and read the same blocks.
	*/
	static vdev_cache_entry_t *
	vdev_cache_allocate(zio_t *zio)
	{
	vdev_cache_t *vc = &zio->io_vd->vdev_cache;
	uint64_t offset = P2ALIGN(zio->io_offset, VCBS);
	vdev_cache_entry_t *ve;

	ASSERT(MUTEX_HELD(&vc->vc_lock));

	if (zfs_vdev_cache_size == 0)
	return (NULL);

	/*
	* If adding a new entry would exceed the cache size,
	* evict the oldest entry (LRU).
	*/
	if ((avl_numnodes(&vc->vc_lastused_tree) << zfs_vdev_cache_bshift) >
	zfs_vdev_cache_size) {
	ve = avl_first(&vc->vc_lastused_tree);
	if (ve->ve_fill_io != NULL)
	return (NULL);
	ASSERT3U(ve->ve_hits, !=, 0);
	vdev_cache_evict(vc, ve);
	}

	ve = kmem_zalloc(sizeof (vdev_cache_entry_t), KM_SLEEP);
	ve->ve_offset = offset;
	ve->ve_lastused = ddi_get_lbolt();
	ve->ve_abd = abd_alloc_for_io(VCBS, B_TRUE);

	avl_add(&vc->vc_offset_tree, ve);
	avl_add(&vc->vc_lastused_tree, ve);

	return (ve);
	}

	static void
	vdev_cache_hit(vdev_cache_t vc, vdev_cache_entry_t ve, zio_t *zio)
	{
	uint64_t cache_phase = P2PHASE(zio->io_offset, VCBS);

	ASSERT(MUTEX_HELD(&vc->vc_lock));
	ASSERT3P(ve->ve_fill_io, ==, NULL);

	if (ve->ve_lastused != ddi_get_lbolt()) {
	avl_remove(&vc->vc_lastused_tree, ve);
	ve->ve_lastused = ddi_get_lbolt();
	avl_add(&vc->vc_lastused_tree, ve);
	}

	ve->ve_hits++;
	abd_copy_off(zio->io_abd, ve->ve_abd, 0, cache_phase, zio->io_size);
	}

	/*
	* Fill a previously allocated cache entry with data.
	*/
	static void
	vdev_cache_fill(zio_t *fio)
	{
	vdev_t *vd = fio->io_vd;
	vdev_cache_t *vc = &vd->vdev_cache;
	vdev_cache_entry_t *ve = fio->io_private;
	zio_t *pio;

	ASSERT3U(fio->io_size, ==, VCBS);

	/*
	* Add data to the cache.
	*/
	mutex_enter(&vc->vc_lock);

	ASSERT3P(ve->ve_fill_io, ==, fio);
	ASSERT3U(ve->ve_offset, ==, fio->io_offset);
	ASSERT3P(ve->ve_abd, ==, fio->io_abd);

	ve->ve_fill_io = NULL;

	/*
	* Even if this cache line was invalidated by a missed write update,
	* any reads that were queued up before the missed update are still
	* valid, so we can satisfy them from this line before we evict it.
	*/
	zio_link_t *zl = NULL;
	while ((pio = zio_walk_parents(fio, &zl)) != NULL)
	vdev_cache_hit(vc, ve, pio);

	if (fio->io_error \|\| ve->ve_missed_update)
	vdev_cache_evict(vc, ve);

	mutex_exit(&vc->vc_lock);
	}

	/*
	* Read data from the cache. Returns B_TRUE cache hit, B_FALSE on miss.
	*/
	boolean_t
	vdev_cache_read(zio_t *zio)
	{
	vdev_cache_t *vc = &zio->io_vd->vdev_cache;
	- vdev_cache_entry_t ve, ve_search;
	+ vdev_cache_entry_t *ve, ve_search;
	uint64_t cache_offset = P2ALIGN(zio->io_offset, VCBS);
	zio_t *fio;
	uint64_t cache_phase __maybe_unused = P2PHASE(zio->io_offset, VCBS);

	ASSERT3U(zio->io_type, ==, ZIO_TYPE_READ);

	+ if (zfs_vdev_cache_size == 0)
	+ return (B_FALSE);
	+
	if (zio->io_flags & ZIO_FLAG_DONT_CACHE)
	return (B_FALSE);

	if (zio->io_size > zfs_vdev_cache_max)
	return (B_FALSE);

	/*
	* If the I/O straddles two or more cache blocks, don't cache it.
	*/
	if (P2BOUNDARY(zio->io_offset, zio->io_size, VCBS))
	return (B_FALSE);

	ASSERT3U(cache_phase + zio->io_size, <=, VCBS);

	mutex_enter(&vc->vc_lock);

	- ve_search = kmem_alloc(sizeof (vdev_cache_entry_t), KM_SLEEP);
	- ve_search->ve_offset = cache_offset;
	- ve = avl_find(&vc->vc_offset_tree, ve_search, NULL);
	- kmem_free(ve_search, sizeof (vdev_cache_entry_t));
	+ ve_search.ve_offset = cache_offset;
	+ ve = avl_find(&vc->vc_offset_tree, &ve_search, NULL);

	if (ve != NULL) {
	if (ve->ve_missed_update) {
	mutex_exit(&vc->vc_lock);
	return (B_FALSE);
	}

	if ((fio = ve->ve_fill_io) != NULL) {
	zio_vdev_io_bypass(zio);
	zio_add_child(zio, fio);
	mutex_exit(&vc->vc_lock);
	VDCSTAT_BUMP(vdc_stat_delegations);
	return (B_TRUE);
	}

	vdev_cache_hit(vc, ve, zio);
	zio_vdev_io_bypass(zio);

	mutex_exit(&vc->vc_lock);
	VDCSTAT_BUMP(vdc_stat_hits);
	return (B_TRUE);
	}

	ve = vdev_cache_allocate(zio);

	if (ve == NULL) {
	mutex_exit(&vc->vc_lock);
	return (B_FALSE);
	}

	fio = zio_vdev_delegated_io(zio->io_vd, cache_offset,
	ve->ve_abd, VCBS, ZIO_TYPE_READ, ZIO_PRIORITY_NOW,
	ZIO_FLAG_DONT_CACHE, vdev_cache_fill, ve);

	ve->ve_fill_io = fio;
	zio_vdev_io_bypass(zio);
	zio_add_child(zio, fio);

	mutex_exit(&vc->vc_lock);
	zio_nowait(fio);
	VDCSTAT_BUMP(vdc_stat_misses);

	return (B_TRUE);
	}

	/*
	* Update cache contents upon write completion.
	*/
	void
	vdev_cache_write(zio_t *zio)
	{
	vdev_cache_t *vc = &zio->io_vd->vdev_cache;
	vdev_cache_entry_t *ve, ve_search;
	uint64_t io_start = zio->io_offset;
	uint64_t io_end = io_start + zio->io_size;
	uint64_t min_offset = P2ALIGN(io_start, VCBS);
	uint64_t max_offset = P2ROUNDUP(io_end, VCBS);
	avl_index_t where;

	ASSERT3U(zio->io_type, ==, ZIO_TYPE_WRITE);

	mutex_enter(&vc->vc_lock);

	ve_search.ve_offset = min_offset;
	ve = avl_find(&vc->vc_offset_tree, &ve_search, &where);

	if (ve == NULL)
	ve = avl_nearest(&vc->vc_offset_tree, where, AVL_AFTER);

	while (ve != NULL && ve->ve_offset < max_offset) {
	uint64_t start = MAX(ve->ve_offset, io_start);
	uint64_t end = MIN(ve->ve_offset + VCBS, io_end);

	if (ve->ve_fill_io != NULL) {
	ve->ve_missed_update = 1;
	} else {
	abd_copy_off(ve->ve_abd, zio->io_abd,
	start - ve->ve_offset, start - io_start,
	end - start);
	}
	ve = AVL_NEXT(&vc->vc_offset_tree, ve);
	}
	mutex_exit(&vc->vc_lock);
	}

	void
	vdev_cache_purge(vdev_t *vd)
	{
	vdev_cache_t *vc = &vd->vdev_cache;
	vdev_cache_entry_t *ve;

	mutex_enter(&vc->vc_lock);
	while ((ve = avl_first(&vc->vc_offset_tree)) != NULL)
	vdev_cache_evict(vc, ve);
	mutex_exit(&vc->vc_lock);
	}

	void
	vdev_cache_init(vdev_t *vd)
	{
	vdev_cache_t *vc = &vd->vdev_cache;

	mutex_init(&vc->vc_lock, NULL, MUTEX_DEFAULT, NULL);

	avl_create(&vc->vc_offset_tree, vdev_cache_offset_compare,
	sizeof (vdev_cache_entry_t),
	offsetof(struct vdev_cache_entry, ve_offset_node));

	avl_create(&vc->vc_lastused_tree, vdev_cache_lastused_compare,
	sizeof (vdev_cache_entry_t),
	offsetof(struct vdev_cache_entry, ve_lastused_node));
	}

	void
	vdev_cache_fini(vdev_t *vd)
	{
	vdev_cache_t *vc = &vd->vdev_cache;

	vdev_cache_purge(vd);

	avl_destroy(&vc->vc_offset_tree);
	avl_destroy(&vc->vc_lastused_tree);

	mutex_destroy(&vc->vc_lock);
	}

	void
	vdev_cache_stat_init(void)
	{
	vdc_ksp = kstat_create("zfs", 0, "vdev_cache_stats", "misc",
	KSTAT_TYPE_NAMED, sizeof (vdc_stats) / sizeof (kstat_named_t),
	KSTAT_FLAG_VIRTUAL);
	if (vdc_ksp != NULL) {
	vdc_ksp->ks_data = &vdc_stats;
	kstat_install(vdc_ksp);
	}
	}

	void
	vdev_cache_stat_fini(void)
	{
	if (vdc_ksp != NULL) {
	kstat_delete(vdc_ksp);
	vdc_ksp = NULL;
	}
	}

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, cache_max, INT, ZMOD_RW,
	"Inflate reads small than max");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, cache_size, INT, ZMOD_RD,
	"Total size of the per-disk cache");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, cache_bshift, INT, ZMOD_RW,
	"Shift size to inflate reads too");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/zfs/vdev_draid.c b/sys/contrib/openzfs/module/zfs/vdev_draid.c
	index b8f82d52e8f0..9c222feb9f06 100644
	--- a/sys/contrib/openzfs/module/zfs/vdev_draid.c
	+++ b/sys/contrib/openzfs/module/zfs/vdev_draid.c
	@@ -1,2784 +1,2784 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2018 Intel Corporation.
	* Copyright (c) 2020 by Lawrence Livermore National Security, LLC.
	*/

	#include <sys/zfs_context.h>
	#include <sys/spa.h>
	#include <sys/spa_impl.h>
	#include <sys/vdev_impl.h>
	#include <sys/vdev_draid.h>
	#include <sys/vdev_raidz.h>
	#include <sys/vdev_rebuild.h>
	#include <sys/abd.h>
	#include <sys/zio.h>
	#include <sys/nvpair.h>
	#include <sys/zio_checksum.h>
	#include <sys/fs/zfs.h>
	#include <sys/fm/fs/zfs.h>
	#include <zfs_fletcher.h>

	#ifdef ZFS_DEBUG
	#include <sys/vdev.h> /* For vdev_xlate() in vdev_draid_io_verify() */
	#endif

	/*
	* dRAID is a distributed spare implementation for ZFS. A dRAID vdev is
	* comprised of multiple raidz redundancy groups which are spread over the
	* dRAID children. To ensure an even distribution, and avoid hot spots, a
	* permutation mapping is applied to the order of the dRAID children.
	* This mixing effectively distributes the parity columns evenly over all
	* of the disks in the dRAID.
	*
	* This is beneficial because it means when resilvering all of the disks
	* can participate thereby increasing the available IOPs and bandwidth.
	* Furthermore, by reserving a small fraction of each child's total capacity
	* virtual distributed spare disks can be created. These spares similarly
	* benefit from the performance gains of spanning all of the children. The
	* consequence of which is that resilvering to a distributed spare can
	* substantially reduce the time required to restore full parity to pool
	* with a failed disks.
	*
	* === dRAID group layout ===
	*
	* First, let's define a "row" in the configuration to be a 16M chunk from
	* each physical drive at the same offset. This is the minimum allowable
	* size since it must be possible to store a full 16M block when there is
	* only a single data column. Next, we define a "group" to be a set of
	* sequential disks containing both the parity and data columns. We allow
	* groups to span multiple rows in order to align any group size to any
	* number of physical drives. Finally, a "slice" is comprised of the rows
	* which contain the target number of groups. The permutation mappings
	* are applied in a round robin fashion to each slice.
	*
	* Given D+P drives in a group (including parity drives) and C-S physical
	* drives (not including the spare drives), we can distribute the groups
	* across R rows without remainder by selecting the least common multiple
	* of D+P and C-S as the number of groups; i.e. ngroups = LCM(D+P, C-S).
	*
	* In the example below, there are C=14 physical drives in the configuration
	* with S=2 drives worth of spare capacity. Each group has a width of 9
	* which includes D=8 data and P=1 parity drive. There are 4 groups and
	* 3 rows per slice. Each group has a size of 144M (16M * 9) and a slice
	* size is 576M (144M * 4). When allocating from a dRAID each group is
	* filled before moving on to the next as show in slice0 below.
	*
	* data disks (8 data + 1 parity) spares (2)
	* +===+===+===+===+===+===+===+===+===+===+===+===+===+===+
	* ^ \| 2 \| 6 \| 1 \| 11\| 4 \| 0 \| 7 \| 10\| 8 \| 9 \| 13\| 5 \| 12\| 3 \| device map 0
	* \| +===+===+===+===+===+===+===+===+===+===+===+===+===+===+
	* \| \| group 0 \| group 1..\| \|
	* \| +-----------------------------------+-----------+-------\|
	* \| \| 0 1 2 3 4 5 6 7 8 \| 36 37 38\| \| r
	* \| \| 9 10 11 12 13 14 15 16 17\| 45 46 47\| \| o
	* \| \| 18 19 20 21 22 23 24 25 26\| 54 55 56\| \| w
	* \| 27 28 29 30 31 32 33 34 35\| 63 64 65\| \| 0
	* s +-----------------------+-----------------------+-------+
	* l \| ..group 1 \| group 2.. \| \|
	* i +-----------------------+-----------------------+-------+
	* c \| 39 40 41 42 43 44\| 72 73 74 75 76 77\| \| r
	* e \| 48 49 50 51 52 53\| 81 82 83 84 85 86\| \| o
	* 0 \| 57 58 59 60 61 62\| 90 91 92 93 94 95\| \| w
	* \| 66 67 68 69 70 71\| 99 100 101 102 103 104\| \| 1
	* \| +-----------+-----------+-----------------------+-------+
	* \| \|..group 2 \| group 3 \| \|
	* \| +-----------+-----------+-----------------------+-------+
	* \| \| 78 79 80\|108 109 110 111 112 113 114 115 116\| \| r
	* \| \| 87 88 89\|117 118 119 120 121 122 123 124 125\| \| o
	* \| \| 96 97 98\|126 127 128 129 130 131 132 133 134\| \| w
	* v \|105 106 107\|135 136 137 138 139 140 141 142 143\| \| 2
	* +===+===+===+===+===+===+===+===+===+===+===+===+===+===+
	* \| 9 \| 11\| 12\| 2 \| 4 \| 1 \| 3 \| 0 \| 10\| 13\| 8 \| 5 \| 6 \| 7 \| device map 1
	* s +===+===+===+===+===+===+===+===+===+===+===+===+===+===+
	* l \| group 4 \| group 5..\| \| row 3
	* i +-----------------------+-----------+-----------+-------\|
	* c \| ..group 5 \| group 6.. \| \| row 4
	* e +-----------+-----------+-----------------------+-------+
	* 1 \|..group 6 \| group 7 \| \| row 5
	* +===+===+===+===+===+===+===+===+===+===+===+===+===+===+
	* \| 3 \| 5 \| 10\| 8 \| 6 \| 11\| 12\| 0 \| 2 \| 4 \| 7 \| 1 \| 9 \| 13\| device map 2
	* s +===+===+===+===+===+===+===+===+===+===+===+===+===+===+
	* l \| group 8 \| group 9..\| \| row 6
	* i +-----------------------------------------------+-------\|
	* c \| ..group 9 \| group 10.. \| \| row 7
	* e +-----------------------+-----------------------+-------+
	* 2 \|..group 10 \| group 11 \| \| row 8
	* +-----------+-----------------------------------+-------+
	*
	* This layout has several advantages over requiring that each row contain
	* a whole number of groups.
	*
	* 1. The group count is not a relevant parameter when defining a dRAID
	* layout. Only the group width is needed, and all groups will have
	* the desired size.
	*
	* 2. All possible group widths (<= physical disk count) can be supported.
	*
	* 3. The logic within vdev_draid.c is simplified when the group width is
	* the same for all groups (although some of the logic around computing
	* permutation numbers and drive offsets is more complicated).
	*
	* N.B. The following array describes all valid dRAID permutation maps.
	* Each row is used to generate a permutation map for a different number
	* of children from a unique seed. The seeds were generated and carefully
	* evaluated by the 'draid' utility in order to provide balanced mappings.
	* In addition to the seed a checksum of the in-memory mapping is stored
	* for verification.
	*
	* The imbalance ratio of a given failure (e.g. 5 disks wide, child 3 failed,
	* with a given permutation map) is the ratio of the amounts of I/O that will
	* be sent to the least and most busy disks when resilvering. The average
	* imbalance ratio (of a given number of disks and permutation map) is the
	* average of the ratios of all possible single and double disk failures.
	*
	* In order to achieve a low imbalance ratio the number of permutations in
	* the mapping must be significantly larger than the number of children.
	* For dRAID the number of permutations has been limited to 512 to minimize
	* the map size. This does result in a gradually increasing imbalance ratio
	* as seen in the table below. Increasing the number of permutations for
	* larger child counts would reduce the imbalance ratio. However, in practice
	* when there are a large number of children each child is responsible for
	* fewer total IOs so it's less of a concern.
	*
	* Note these values are hard coded and must never be changed. Existing
	* pools depend on the same mapping always being generated in order to
	* read and write from the correct locations. Any change would make
	* existing pools completely inaccessible.
	*/
	static const draid_map_t draid_maps[VDEV_DRAID_MAX_MAPS] = {
	{ 2, 256, 0x89ef3dabbcc7de37, 0x00000000433d433d }, /* 1.000 */
	{ 3, 256, 0x89a57f3de98121b4, 0x00000000bcd8b7b5 }, /* 1.000 */
	{ 4, 256, 0xc9ea9ec82340c885, 0x00000001819d7c69 }, /* 1.000 */
	{ 5, 256, 0xf46733b7f4d47dfd, 0x00000002a1648d74 }, /* 1.010 */
	{ 6, 256, 0x88c3c62d8585b362, 0x00000003d3b0c2c4 }, /* 1.031 */
	{ 7, 256, 0x3a65d809b4d1b9d5, 0x000000055c4183ee }, /* 1.043 */
	{ 8, 256, 0xe98930e3c5d2e90a, 0x00000006edfb0329 }, /* 1.059 */
	{ 9, 256, 0x5a5430036b982ccb, 0x00000008ceaf6934 }, /* 1.056 */
	{ 10, 256, 0x92bf389e9eadac74, 0x0000000b26668c09 }, /* 1.072 */
	{ 11, 256, 0x74ccebf1dcf3ae80, 0x0000000dd691358c }, /* 1.083 */
	{ 12, 256, 0x8847e41a1a9f5671, 0x00000010a0c63c8e }, /* 1.097 */
	{ 13, 256, 0x7481b56debf0e637, 0x0000001424121fe4 }, /* 1.100 */
	{ 14, 256, 0x559b8c44065f8967, 0x00000016ab2ff079 }, /* 1.121 */
	{ 15, 256, 0x34c49545a2ee7f01, 0x0000001a6028efd6 }, /* 1.103 */
	{ 16, 256, 0xb85f4fa81a7698f7, 0x0000001e95ff5e66 }, /* 1.111 */
	{ 17, 256, 0x6353e47b7e47aba0, 0x00000021a81fa0fe }, /* 1.133 */
	{ 18, 256, 0xaa549746b1cbb81c, 0x00000026f02494c9 }, /* 1.131 */
	{ 19, 256, 0x892e343f2f31d690, 0x00000029eb392835 }, /* 1.130 */
	{ 20, 256, 0x76914824db98cc3f, 0x0000003004f31a7c }, /* 1.141 */
	{ 21, 256, 0x4b3cbabf9cfb1d0f, 0x00000036363a2408 }, /* 1.139 */
	{ 22, 256, 0xf45c77abb4f035d4, 0x00000038dd0f3e84 }, /* 1.150 */
	{ 23, 256, 0x5e18bd7f3fd4baf4, 0x0000003f0660391f }, /* 1.174 */
	{ 24, 256, 0xa7b3a4d285d6503b, 0x000000443dfc9ff6 }, /* 1.168 */
	{ 25, 256, 0x56ac7dd967521f5a, 0x0000004b03a87eb7 }, /* 1.180 */
	{ 26, 256, 0x3a42dfda4eb880f7, 0x000000522c719bba }, /* 1.226 */
	{ 27, 256, 0xd200d2fc6b54bf60, 0x0000005760b4fdf5 }, /* 1.228 */
	{ 28, 256, 0xc52605bbd486c546, 0x0000005e00d8f74c }, /* 1.217 */
	{ 29, 256, 0xc761779e63cd762f, 0x00000067be3cd85c }, /* 1.239 */
	{ 30, 256, 0xca577b1e07f85ca5, 0x0000006f5517f3e4 }, /* 1.238 */
	{ 31, 256, 0xfd50a593c518b3d4, 0x0000007370e7778f }, /* 1.273 */
	{ 32, 512, 0xc6c87ba5b042650b, 0x000000f7eb08a156 }, /* 1.191 */
	{ 33, 512, 0xc3880d0c9d458304, 0x0000010734b5d160 }, /* 1.199 */
	{ 34, 512, 0xe920927e4d8b2c97, 0x00000118c1edbce0 }, /* 1.195 */
	{ 35, 512, 0x8da7fcda87bde316, 0x0000012a3e9f9110 }, /* 1.201 */
	{ 36, 512, 0xcf09937491514a29, 0x0000013bd6a24bef }, /* 1.194 */
	{ 37, 512, 0x9b5abbf345cbd7cc, 0x0000014b9d90fac3 }, /* 1.237 */
	{ 38, 512, 0x506312a44668d6a9, 0x0000015e1b5f6148 }, /* 1.242 */
	{ 39, 512, 0x71659ede62b4755f, 0x00000173ef029bcd }, /* 1.231 */
	{ 40, 512, 0xa7fde73fb74cf2d7, 0x000001866fb72748 }, /* 1.233 */
	{ 41, 512, 0x19e8b461a1dea1d3, 0x000001a046f76b23 }, /* 1.271 */
	{ 42, 512, 0x031c9b868cc3e976, 0x000001afa64c49d3 }, /* 1.263 */
	{ 43, 512, 0xbaa5125faa781854, 0x000001c76789e278 }, /* 1.270 */
	{ 44, 512, 0x4ed55052550d721b, 0x000001d800ccd8eb }, /* 1.281 */
	{ 45, 512, 0x0fd63ddbdff90677, 0x000001f08ad59ed2 }, /* 1.282 */
	{ 46, 512, 0x36d66546de7fdd6f, 0x000002016f09574b }, /* 1.286 */
	{ 47, 512, 0x99f997e7eafb69d7, 0x0000021e42e47cb6 }, /* 1.329 */
	{ 48, 512, 0xbecd9c2571312c5d, 0x000002320fe2872b }, /* 1.286 */
	{ 49, 512, 0xd97371329e488a32, 0x0000024cd73f2ca7 }, /* 1.322 */
	{ 50, 512, 0x30e9b136670749ee, 0x000002681c83b0e0 }, /* 1.335 */
	{ 51, 512, 0x11ad6bc8f47aaeb4, 0x0000027e9261b5d5 }, /* 1.305 */
	{ 52, 512, 0x68e445300af432c1, 0x0000029aa0eb7dbf }, /* 1.330 */
	{ 53, 512, 0x910fb561657ea98c, 0x000002b3dca04853 }, /* 1.365 */
	{ 54, 512, 0xd619693d8ce5e7a5, 0x000002cc280e9c97 }, /* 1.334 */
	{ 55, 512, 0x24e281f564dbb60a, 0x000002e9fa842713 }, /* 1.364 */
	{ 56, 512, 0x947a7d3bdaab44c5, 0x000003046680f72e }, /* 1.374 */
	{ 57, 512, 0x2d44fec9c093e0de, 0x00000324198ba810 }, /* 1.363 */
	{ 58, 512, 0x87743c272d29bb4c, 0x0000033ec48c9ac9 }, /* 1.401 */
	{ 59, 512, 0x96aa3b6f67f5d923, 0x0000034faead902c }, /* 1.392 */
	{ 60, 512, 0x94a4f1faf520b0d3, 0x0000037d713ab005 }, /* 1.360 */
	{ 61, 512, 0xb13ed3a272f711a2, 0x00000397368f3cbd }, /* 1.396 */
	{ 62, 512, 0x3b1b11805fa4a64a, 0x000003b8a5e2840c }, /* 1.453 */
	{ 63, 512, 0x4c74caad9172ba71, 0x000003d4be280290 }, /* 1.437 */
	{ 64, 512, 0x035ff643923dd29e, 0x000003fad6c355e1 }, /* 1.402 */
	{ 65, 512, 0x768e9171b11abd3c, 0x0000040eb07fed20 }, /* 1.459 */
	{ 66, 512, 0x75880e6f78a13ddd, 0x000004433d6acf14 }, /* 1.423 */
	{ 67, 512, 0x910b9714f698a877, 0x00000451ea65d5db }, /* 1.447 */
	{ 68, 512, 0x87f5db6f9fdcf5c7, 0x000004732169e3f7 }, /* 1.450 */
	{ 69, 512, 0x836d4968fbaa3706, 0x000004954068a380 }, /* 1.455 */
	{ 70, 512, 0xc567d73a036421ab, 0x000004bd7cb7bd3d }, /* 1.463 */
	{ 71, 512, 0x619df40f240b8fed, 0x000004e376c2e972 }, /* 1.463 */
	{ 72, 512, 0x42763a680d5bed8e, 0x000005084275c680 }, /* 1.452 */
	{ 73, 512, 0x5866f064b3230431, 0x0000052906f2c9ab }, /* 1.498 */
	{ 74, 512, 0x9fa08548b1621a44, 0x0000054708019247 }, /* 1.526 */
	{ 75, 512, 0xb6053078ce0fc303, 0x00000572cc5c72b0 }, /* 1.491 */
	{ 76, 512, 0x4a7aad7bf3890923, 0x0000058e987bc8e9 }, /* 1.470 */
	{ 77, 512, 0xe165613fd75b5a53, 0x000005c20473a211 }, /* 1.527 */
	{ 78, 512, 0x3ff154ac878163a6, 0x000005d659194bf3 }, /* 1.509 */
	{ 79, 512, 0x24b93ade0aa8a532, 0x0000060a201c4f8e }, /* 1.569 */
	{ 80, 512, 0xc18e2d14cd9bb554, 0x0000062c55cfe48c }, /* 1.555 */
	{ 81, 512, 0x98cc78302feb58b6, 0x0000066656a07194 }, /* 1.509 */
	{ 82, 512, 0xc6c5fd5a2abc0543, 0x0000067cff94fbf8 }, /* 1.596 */
	{ 83, 512, 0xa7962f514acbba21, 0x000006ab7b5afa2e }, /* 1.568 */
	{ 84, 512, 0xba02545069ddc6dc, 0x000006d19861364f }, /* 1.541 */
	{ 85, 512, 0x447c73192c35073e, 0x000006fce315ce35 }, /* 1.623 */
	{ 86, 512, 0x48beef9e2d42b0c2, 0x00000720a8e38b6b }, /* 1.620 */
	{ 87, 512, 0x4874cf98541a35e0, 0x00000758382a2273 }, /* 1.597 */
	{ 88, 512, 0xad4cf8333a31127a, 0x00000781e1651b1b }, /* 1.575 */
	{ 89, 512, 0x47ae4859d57888c1, 0x000007b27edbe5bc }, /* 1.627 */
	{ 90, 512, 0x06f7723cfe5d1891, 0x000007dc2a96d8eb }, /* 1.596 */
	{ 91, 512, 0xd4e44218d660576d, 0x0000080ac46f02d5 }, /* 1.622 */
	{ 92, 512, 0x7066702b0d5be1f2, 0x00000832c96d154e }, /* 1.695 */
	{ 93, 512, 0x011209b4f9e11fb9, 0x0000085eefda104c }, /* 1.605 */
	{ 94, 512, 0x47ffba30a0b35708, 0x00000899badc32dc }, /* 1.625 */
	{ 95, 512, 0x1a95a6ac4538aaa8, 0x000008b6b69a42b2 }, /* 1.687 */
	{ 96, 512, 0xbda2b239bb2008eb, 0x000008f22d2de38a }, /* 1.621 */
	{ 97, 512, 0x7ffa0bea90355c6c, 0x0000092e5b23b816 }, /* 1.699 */
	{ 98, 512, 0x1d56ba34be426795, 0x0000094f482e5d1b }, /* 1.688 */
	{ 99, 512, 0x0aa89d45c502e93d, 0x00000977d94a98ce }, /* 1.642 */
	{ 100, 512, 0x54369449f6857774, 0x000009c06c9b34cc }, /* 1.683 */
	{ 101, 512, 0xf7d4dd8445b46765, 0x000009e5dc542259 }, /* 1.755 */
	{ 102, 512, 0xfa8866312f169469, 0x00000a16b54eae93 }, /* 1.692 */
	{ 103, 512, 0xd8a5aea08aef3ff9, 0x00000a381d2cbfe7 }, /* 1.747 */
	{ 104, 512, 0x66bcd2c3d5f9ef0e, 0x00000a8191817be7 }, /* 1.751 */
	{ 105, 512, 0x3fb13a47a012ec81, 0x00000ab562b9a254 }, /* 1.751 */
	{ 106, 512, 0x43100f01c9e5e3ca, 0x00000aeee84c185f }, /* 1.726 */
	{ 107, 512, 0xca09c50ccee2d054, 0x00000b1c359c047d }, /* 1.788 */
	{ 108, 512, 0xd7176732ac503f9b, 0x00000b578bc52a73 }, /* 1.740 */
	{ 109, 512, 0xed206e51f8d9422d, 0x00000b8083e0d960 }, /* 1.780 */
	{ 110, 512, 0x17ead5dc6ba0dcd6, 0x00000bcfb1a32ca8 }, /* 1.836 */
	{ 111, 512, 0x5f1dc21e38a969eb, 0x00000c0171becdd6 }, /* 1.778 */
	{ 112, 512, 0xddaa973de33ec528, 0x00000c3edaba4b95 }, /* 1.831 */
	{ 113, 512, 0x2a5eccd7735a3630, 0x00000c630664e7df }, /* 1.825 */
	{ 114, 512, 0xafcccee5c0b71446, 0x00000cb65392f6e4 }, /* 1.826 */
	{ 115, 512, 0x8fa30c5e7b147e27, 0x00000cd4db391e55 }, /* 1.843 */
	{ 116, 512, 0x5afe0711fdfafd82, 0x00000d08cb4ec35d }, /* 1.826 */
	{ 117, 512, 0x533a6090238afd4c, 0x00000d336f115d1b }, /* 1.803 */
	{ 118, 512, 0x90cf11b595e39a84, 0x00000d8e041c2048 }, /* 1.857 */
	{ 119, 512, 0x0d61a3b809444009, 0x00000dcb798afe35 }, /* 1.877 */
	{ 120, 512, 0x7f34da0f54b0d114, 0x00000df3922664e1 }, /* 1.849 */
	{ 121, 512, 0xa52258d5b72f6551, 0x00000e4d37a9872d }, /* 1.867 */
	{ 122, 512, 0xc1de54d7672878db, 0x00000e6583a94cf6 }, /* 1.978 */
	{ 123, 512, 0x1d03354316a414ab, 0x00000ebffc50308d }, /* 1.947 */
	{ 124, 512, 0xcebdcc377665412c, 0x00000edee1997cea }, /* 1.865 */
	{ 125, 512, 0x4ddd4c04b1a12344, 0x00000f21d64b373f }, /* 1.881 */
	{ 126, 512, 0x64fc8f94e3973658, 0x00000f8f87a8896b }, /* 1.882 */
	{ 127, 512, 0x68765f78034a334e, 0x00000fb8fe62197e }, /* 1.867 */
	{ 128, 512, 0xaf36b871a303e816, 0x00000fec6f3afb1e }, /* 1.972 */
	{ 129, 512, 0x2a4cbf73866c3a28, 0x00001027febfe4e5 }, /* 1.896 */
	{ 130, 512, 0x9cb128aacdcd3b2f, 0x0000106aa8ac569d }, /* 1.965 */
	{ 131, 512, 0x5511d41c55869124, 0x000010bbd755ddf1 }, /* 1.963 */
	{ 132, 512, 0x42f92461937f284a, 0x000010fb8bceb3b5 }, /* 1.925 */
	{ 133, 512, 0xe2d89a1cf6f1f287, 0x0000114cf5331e34 }, /* 1.862 */
	{ 134, 512, 0xdc631a038956200e, 0x0000116428d2adc5 }, /* 2.042 */
	{ 135, 512, 0xb2e5ac222cd236be, 0x000011ca88e4d4d2 }, /* 1.935 */
	{ 136, 512, 0xbc7d8236655d88e7, 0x000011e39cb94e66 }, /* 2.005 */
	{ 137, 512, 0x073e02d88d2d8e75, 0x0000123136c7933c }, /* 2.041 */
	{ 138, 512, 0x3ddb9c3873166be0, 0x00001280e4ec6d52 }, /* 1.997 */
	{ 139, 512, 0x7d3b1a845420e1b5, 0x000012c2e7cd6a44 }, /* 1.996 */
	{ 140, 512, 0x60102308aa7b2a6c, 0x000012fc490e6c7d }, /* 2.053 */
	{ 141, 512, 0xdb22bb2f9eb894aa, 0x00001343f5a85a1a }, /* 1.971 */
	{ 142, 512, 0xd853f879a13b1606, 0x000013bb7d5f9048 }, /* 2.018 */
	{ 143, 512, 0x001620a03f804b1d, 0x000013e74cc794fd }, /* 1.961 */
	{ 144, 512, 0xfdb52dda76fbf667, 0x00001442d2f22480 }, /* 2.046 */
	{ 145, 512, 0xa9160110f66e24ff, 0x0000144b899f9dbb }, /* 1.968 */
	{ 146, 512, 0x77306a30379ae03b, 0x000014cb98eb1f81 }, /* 2.143 */
	{ 147, 512, 0x14f5985d2752319d, 0x000014feab821fc9 }, /* 2.064 */
	{ 148, 512, 0xa4b8ff11de7863f8, 0x0000154a0e60b9c9 }, /* 2.023 */
	{ 149, 512, 0x44b345426455c1b3, 0x000015999c3c569c }, /* 2.136 */
	{ 150, 512, 0x272677826049b46c, 0x000015c9697f4b92 }, /* 2.063 */
	{ 151, 512, 0x2f9216e2cd74fe40, 0x0000162b1f7bbd39 }, /* 1.974 */
	{ 152, 512, 0x706ae3e763ad8771, 0x00001661371c55e1 }, /* 2.210 */
	{ 153, 512, 0xf7fd345307c2480e, 0x000016e251f28b6a }, /* 2.006 */
	{ 154, 512, 0x6e94e3d26b3139eb, 0x000016f2429bb8c6 }, /* 2.193 */
	{ 155, 512, 0x5458bbfbb781fcba, 0x0000173efdeca1b9 }, /* 2.163 */
	{ 156, 512, 0xa80e2afeccd93b33, 0x000017bfdcb78adc }, /* 2.046 */
	{ 157, 512, 0x1e4ccbb22796cf9d, 0x00001826fdcc39c9 }, /* 2.084 */
	{ 158, 512, 0x8fba4b676aaa3663, 0x00001841a1379480 }, /* 2.264 */
	{ 159, 512, 0xf82b843814b315fa, 0x000018886e19b8a3 }, /* 2.074 */
	{ 160, 512, 0x7f21e920ecf753a3, 0x0000191812ca0ea7 }, /* 2.282 */
	{ 161, 512, 0x48bb8ea2c4caa620, 0x0000192f310faccf }, /* 2.148 */
	{ 162, 512, 0x5cdb652b4952c91b, 0x0000199e1d7437c7 }, /* 2.355 */
	{ 163, 512, 0x6ac1ba6f78c06cd4, 0x000019cd11f82c70 }, /* 2.164 */
	{ 164, 512, 0x9faf5f9ca2669a56, 0x00001a18d5431f6a }, /* 2.393 */
	{ 165, 512, 0xaa57e9383eb01194, 0x00001a9e7d253d85 }, /* 2.178 */
	{ 166, 512, 0x896967bf495c34d2, 0x00001afb8319b9fc }, /* 2.334 */
	{ 167, 512, 0xdfad5f05de225f1b, 0x00001b3a59c3093b }, /* 2.266 */
	{ 168, 512, 0xfd299a99f9f2abdd, 0x00001bb6f1a10799 }, /* 2.304 */
	{ 169, 512, 0xdda239e798fe9fd4, 0x00001bfae0c9692d }, /* 2.218 */
	{ 170, 512, 0x5fca670414a32c3e, 0x00001c22129dbcff }, /* 2.377 */
	{ 171, 512, 0x1bb8934314b087de, 0x00001c955db36cd0 }, /* 2.155 */
	{ 172, 512, 0xd96394b4b082200d, 0x00001cfc8619b7e6 }, /* 2.404 */
	{ 173, 512, 0xb612a7735b1c8cbc, 0x00001d303acdd585 }, /* 2.205 */
	{ 174, 512, 0x28e7430fe5875fe1, 0x00001d7ed5b3697d }, /* 2.359 */
	{ 175, 512, 0x5038e89efdd981b9, 0x00001dc40ec35c59 }, /* 2.158 */
	{ 176, 512, 0x075fd78f1d14db7c, 0x00001e31c83b4a2b }, /* 2.614 */
	{ 177, 512, 0xc50fafdb5021be15, 0x00001e7cdac82fbc }, /* 2.239 */
	{ 178, 512, 0xe6dc7572ce7b91c7, 0x00001edd8bb454fc }, /* 2.493 */
	{ 179, 512, 0x21f7843e7beda537, 0x00001f3a8e019d6c }, /* 2.327 */
	{ 180, 512, 0xc83385e20b43ec82, 0x00001f70735ec137 }, /* 2.231 */
	{ 181, 512, 0xca818217dddb21fd, 0x0000201ca44c5a3c }, /* 2.237 */
	{ 182, 512, 0xe6035defea48f933, 0x00002038e3346658 }, /* 2.691 */
	{ 183, 512, 0x47262a4f953dac5a, 0x000020c2e554314e }, /* 2.170 */
	{ 184, 512, 0xe24c7246260873ea, 0x000021197e618d64 }, /* 2.600 */
	{ 185, 512, 0xeef6b57c9b58e9e1, 0x0000217ea48ecddc }, /* 2.391 */
	{ 186, 512, 0x2becd3346e386142, 0x000021c496d4a5f9 }, /* 2.677 */
	{ 187, 512, 0x63c6207bdf3b40a3, 0x0000220e0f2eec0c }, /* 2.410 */
	{ 188, 512, 0x3056ce8989767d4b, 0x0000228eb76cd137 }, /* 2.776 */
	{ 189, 512, 0x91af61c307cee780, 0x000022e17e2ea501 }, /* 2.266 */
	{ 190, 512, 0xda359da225f6d54f, 0x00002358a2debc19 }, /* 2.717 */
	{ 191, 512, 0x0a5f7a2a55607ba0, 0x0000238a79dac18c }, /* 2.474 */
	{ 192, 512, 0x27bb75bf5224638a, 0x00002403a58e2351 }, /* 2.673 */
	{ 193, 512, 0x1ebfdb94630f5d0f, 0x00002492a10cb339 }, /* 2.420 */
	{ 194, 512, 0x6eae5e51d9c5f6fb, 0x000024ce4bf98715 }, /* 2.898 */
	{ 195, 512, 0x08d903b4daedc2e0, 0x0000250d1e15886c }, /* 2.363 */
	{ 196, 512, 0xc722a2f7fa7cd686, 0x0000258a99ed0c9e }, /* 2.747 */
	{ 197, 512, 0x8f71faf0e54e361d, 0x000025dee11976f5 }, /* 2.531 */
	{ 198, 512, 0x87f64695c91a54e7, 0x0000264e00a43da0 }, /* 2.707 */
	{ 199, 512, 0xc719cbac2c336b92, 0x000026d327277ac1 }, /* 2.315 */
	{ 200, 512, 0xe7e647afaf771ade, 0x000027523a5c44bf }, /* 3.012 */
	{ 201, 512, 0x12d4b5c38ce8c946, 0x0000273898432545 }, /* 2.378 */
	{ 202, 512, 0xf2e0cd4067bdc94a, 0x000027e47bb2c935 }, /* 2.969 */
	{ 203, 512, 0x21b79f14d6d947d3, 0x0000281e64977f0d }, /* 2.594 */
	{ 204, 512, 0x515093f952f18cd6, 0x0000289691a473fd }, /* 2.763 */
	{ 205, 512, 0xd47b160a1b1022c8, 0x00002903e8b52411 }, /* 2.457 */
	{ 206, 512, 0xc02fc96684715a16, 0x0000297515608601 }, /* 3.057 */
	{ 207, 512, 0xef51e68efba72ed0, 0x000029ef73604804 }, /* 2.590 */
	{ 208, 512, 0x9e3be6e5448b4f33, 0x00002a2846ed074b }, /* 3.047 */
	{ 209, 512, 0x81d446c6d5fec063, 0x00002a92ca693455 }, /* 2.676 */
	{ 210, 512, 0xff215de8224e57d5, 0x00002b2271fe3729 }, /* 2.993 */
	{ 211, 512, 0xe2524d9ba8f69796, 0x00002b64b99c3ba2 }, /* 2.457 */
	{ 212, 512, 0xf6b28e26097b7e4b, 0x00002bd768b6e068 }, /* 3.182 */
	{ 213, 512, 0x893a487f30ce1644, 0x00002c67f722b4b2 }, /* 2.563 */
	{ 214, 512, 0x386566c3fc9871df, 0x00002cc1cf8b4037 }, /* 3.025 */
	{ 215, 512, 0x1e0ed78edf1f558a, 0x00002d3948d36c7f }, /* 2.730 */
	{ 216, 512, 0xe3bc20c31e61f113, 0x00002d6d6b12e025 }, /* 3.036 */
	{ 217, 512, 0xd6c3ad2e23021882, 0x00002deff7572241 }, /* 2.722 */
	{ 218, 512, 0xb4a9f95cf0f69c5a, 0x00002e67d537aa36 }, /* 3.356 */
	{ 219, 512, 0x6e98ed6f6c38e82f, 0x00002e9720626789 }, /* 2.697 */
	{ 220, 512, 0x2e01edba33fddac7, 0x00002f407c6b0198 }, /* 2.979 */
	{ 221, 512, 0x559d02e1f5f57ccc, 0x00002fb6a5ab4f24 }, /* 2.858 */
	{ 222, 512, 0xac18f5a916adcd8e, 0x0000304ae1c5c57e }, /* 3.258 */
	{ 223, 512, 0x15789fbaddb86f4b, 0x0000306f6e019c78 }, /* 2.693 */
	{ 224, 512, 0xf4a9c36d5bc4c408, 0x000030da40434213 }, /* 3.259 */
	{ 225, 512, 0xf640f90fd2727f44, 0x00003189ed37b90c }, /* 2.733 */
	{ 226, 512, 0xb5313d390d61884a, 0x000031e152616b37 }, /* 3.235 */
	{ 227, 512, 0x4bae6b3ce9160939, 0x0000321f40aeac42 }, /* 2.983 */
	{ 228, 512, 0x838c34480f1a66a1, 0x000032f389c0f78e }, /* 3.308 */
	{ 229, 512, 0xb1c4a52c8e3d6060, 0x0000330062a40284 }, /* 2.715 */
	{ 230, 512, 0xe0f1110c6d0ed822, 0x0000338be435644f }, /* 3.540 */
	{ 231, 512, 0x9f1a8ccdcea68d4b, 0x000034045a4e97e1 }, /* 2.779 */
	{ 232, 512, 0x3261ed62223f3099, 0x000034702cfc401c }, /* 3.084 */
	{ 233, 512, 0xf2191e2311022d65, 0x00003509dd19c9fc }, /* 2.987 */
	{ 234, 512, 0xf102a395c2033abc, 0x000035654dc96fae }, /* 3.341 */
	{ 235, 512, 0x11fe378f027906b6, 0x000035b5193b0264 }, /* 2.793 */
	{ 236, 512, 0xf777f2c026b337aa, 0x000036704f5d9297 }, /* 3.518 */
	{ 237, 512, 0x1b04e9c2ee143f32, 0x000036dfbb7af218 }, /* 2.962 */
	{ 238, 512, 0x2fcec95266f9352c, 0x00003785c8df24a9 }, /* 3.196 */
	{ 239, 512, 0xfe2b0e47e427dd85, 0x000037cbdf5da729 }, /* 2.914 */
	{ 240, 512, 0x72b49bf2225f6c6d, 0x0000382227c15855 }, /* 3.408 */
	{ 241, 512, 0x50486b43df7df9c7, 0x0000389b88be6453 }, /* 2.903 */
	{ 242, 512, 0x5192a3e53181c8ab, 0x000038ddf3d67263 }, /* 3.778 */
	{ 243, 512, 0xe9f5d8365296fd5e, 0x0000399f1c6c9e9c }, /* 3.026 */
	{ 244, 512, 0xc740263f0301efa8, 0x00003a147146512d }, /* 3.347 */
	{ 245, 512, 0x23cd0f2b5671e67d, 0x00003ab10bcc0d9d }, /* 3.212 */
	{ 246, 512, 0x002ccc7e5cd41390, 0x00003ad6cd14a6c0 }, /* 3.482 */
	{ 247, 512, 0x9aafb3c02544b31b, 0x00003b8cb8779fb0 }, /* 3.146 */
	{ 248, 512, 0x72ba07a78b121999, 0x00003c24142a5a3f }, /* 3.626 */
	{ 249, 512, 0x3d784aa58edfc7b4, 0x00003cd084817d99 }, /* 2.952 */
	{ 250, 512, 0xaab750424d8004af, 0x00003d506a8e098e }, /* 3.463 */
	{ 251, 512, 0x84403fcf8e6b5ca2, 0x00003d4c54c2aec4 }, /* 3.131 */
	{ 252, 512, 0x71eb7455ec98e207, 0x00003e655715cf2c }, /* 3.538 */
	{ 253, 512, 0xd752b4f19301595b, 0x00003ecd7b2ca5ac }, /* 2.974 */
	{ 254, 512, 0xc4674129750499de, 0x00003e99e86d3e95 }, /* 3.843 */
	{ 255, 512, 0x9772baff5cd12ef5, 0x00003f895c019841 }, /* 3.088 */
	};

	/*
	* Verify the map is valid. Each device index must appear exactly
	* once in every row, and the permutation array checksum must match.
	*/
	static int
	verify_perms(uint8_t *perms, uint64_t children, uint64_t nperms,
	uint64_t checksum)
	{
	int countssz = sizeof (uint16_t) * children;
	uint16_t *counts = kmem_zalloc(countssz, KM_SLEEP);

	for (int i = 0; i < nperms; i++) {
	for (int j = 0; j < children; j++) {
	uint8_t val = perms[(i * children) + j];

	if (val >= children \|\| counts[val] != i) {
	kmem_free(counts, countssz);
	return (EINVAL);
	}

	counts[val]++;
	}
	}

	if (checksum != 0) {
	int permssz = sizeof (uint8_t) * children * nperms;
	zio_cksum_t cksum;

	fletcher_4_native_varsize(perms, permssz, &cksum);

	if (checksum != cksum.zc_word[0]) {
	kmem_free(counts, countssz);
	return (ECKSUM);
	}
	}

	kmem_free(counts, countssz);

	return (0);
	}

	/*
	* Generate the permutation array for the draid_map_t. These maps control
	* the placement of all data in a dRAID. Therefore it's critical that the
	* seed always generates the same mapping. We provide our own pseudo-random
	* number generator for this purpose.
	*/
	int
	vdev_draid_generate_perms(const draid_map_t map, uint8_t *permsp)
	{
	VERIFY3U(map->dm_children, >=, VDEV_DRAID_MIN_CHILDREN);
	VERIFY3U(map->dm_children, <=, VDEV_DRAID_MAX_CHILDREN);
	VERIFY3U(map->dm_seed, !=, 0);
	VERIFY3U(map->dm_nperms, !=, 0);
	VERIFY3P(map->dm_perms, ==, NULL);

	#ifdef _KERNEL
	/*
	* The kernel code always provides both a map_seed and checksum.
	* Only the tests/zfs-tests/cmd/draid/draid.c utility will provide
	* a zero checksum when generating new candidate maps.
	*/
	VERIFY3U(map->dm_checksum, !=, 0);
	#endif
	uint64_t children = map->dm_children;
	uint64_t nperms = map->dm_nperms;
	int rowsz = sizeof (uint8_t) * children;
	int permssz = rowsz * nperms;
	uint8_t *perms;

	/* Allocate the permutation array */
	perms = vmem_alloc(permssz, KM_SLEEP);

	/* Setup an initial row with a known pattern */
	uint8_t *initial_row = kmem_alloc(rowsz, KM_SLEEP);
	for (int i = 0; i < children; i++)
	initial_row[i] = i;

	uint64_t draid_seed[2] = { VDEV_DRAID_SEED, map->dm_seed };
	uint8_t current_row, previous_row = initial_row;

	/*
	* Perform a Fisher-Yates shuffle of each row using the previous
	* row as the starting point. An initial_row with known pattern
	* is used as the input for the first row.
	*/
	for (int i = 0; i < nperms; i++) {
	current_row = &perms[i * children];
	memcpy(current_row, previous_row, rowsz);

	for (int j = children - 1; j > 0; j--) {
	uint64_t k = vdev_draid_rand(draid_seed) % (j + 1);
	uint8_t val = current_row[j];
	current_row[j] = current_row[k];
	current_row[k] = val;
	}

	previous_row = current_row;
	}

	kmem_free(initial_row, rowsz);

	int error = verify_perms(perms, children, nperms, map->dm_checksum);
	if (error) {
	vmem_free(perms, permssz);
	return (error);
	}

	*permsp = perms;

	return (0);
	}

	/*
	* Lookup the fixed draid_map_t for the requested number of children.
	*/
	int
	vdev_draid_lookup_map(uint64_t children, const draid_map_t **mapp)
	{
	for (int i = 0; i <= VDEV_DRAID_MAX_MAPS; i++) {
	if (draid_maps[i].dm_children == children) {
	*mapp = &draid_maps[i];
	return (0);
	}
	}

	return (ENOENT);
	}

	/*
	* Lookup the permutation array and iteration id for the provided offset.
	*/
	static void
	vdev_draid_get_perm(vdev_draid_config_t *vdc, uint64_t pindex,
	uint8_t *base, uint64_t iter)
	{
	uint64_t ncols = vdc->vdc_children;
	uint64_t poff = pindex % (vdc->vdc_nperms * ncols);

	base = vdc->vdc_perms + (poff / ncols) ncols;
	*iter = poff % ncols;
	}

	static inline uint64_t
	vdev_draid_permute_id(vdev_draid_config_t *vdc,
	uint8_t *base, uint64_t iter, uint64_t index)
	{
	return ((base[index] + iter) % vdc->vdc_children);
	}

	/*
	* Return the asize which is the psize rounded up to a full group width.
	* i.e. vdev_draid_psize_to_asize().
	*/
	static uint64_t
	vdev_draid_asize(vdev_t *vd, uint64_t psize)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;
	uint64_t ashift = vd->vdev_ashift;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);

	uint64_t rows = ((psize - 1) / (vdc->vdc_ndata << ashift)) + 1;
	uint64_t asize = (rows * vdc->vdc_groupwidth) << ashift;

	ASSERT3U(asize, !=, 0);
	ASSERT3U(asize % (vdc->vdc_groupwidth), ==, 0);

	return (asize);
	}

	/*
	* Deflate the asize to the psize, this includes stripping parity.
	*/
	uint64_t
	vdev_draid_asize_to_psize(vdev_t *vd, uint64_t asize)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	ASSERT0(asize % vdc->vdc_groupwidth);

	return ((asize / vdc->vdc_groupwidth) * vdc->vdc_ndata);
	}

	/*
	* Convert a logical offset to the corresponding group number.
	*/
	static uint64_t
	vdev_draid_offset_to_group(vdev_t *vd, uint64_t offset)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);

	return (offset / vdc->vdc_groupsz);
	}

	/*
	* Convert a group number to the logical starting offset for that group.
	*/
	static uint64_t
	vdev_draid_group_to_offset(vdev_t *vd, uint64_t group)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);

	return (group * vdc->vdc_groupsz);
	}

	/*
	* Full stripe writes. When writing, all columns (D+P) are required. Parity
	* is calculated over all the columns, including empty zero filled sectors,
	* and each is written to disk. While only the data columns are needed for
	* a normal read, all of the columns are required for reconstruction when
	* performing a sequential resilver.
	*
	* For "big columns" it's sufficient to map the correct range of the zio ABD.
	* Partial columns require allocating a gang ABD in order to zero fill the
	* empty sectors. When the column is empty a zero filled sector must be
	* mapped. In all cases the data ABDs must be the same size as the parity
	* ABDs (e.g. rc->rc_size == parity_size).
	*/
	static void
	vdev_draid_map_alloc_write(zio_t zio, uint64_t abd_offset, raidz_row_t rr)
	{
	uint64_t skip_size = 1ULL << zio->io_vd->vdev_top->vdev_ashift;
	uint64_t parity_size = rr->rr_col[0].rc_size;
	uint64_t abd_off = abd_offset;

	ASSERT3U(zio->io_type, ==, ZIO_TYPE_WRITE);
	ASSERT3U(parity_size, ==, abd_get_size(rr->rr_col[0].rc_abd));

	for (uint64_t c = rr->rr_firstdatacol; c < rr->rr_cols; c++) {
	raidz_col_t *rc = &rr->rr_col[c];

	if (rc->rc_size == 0) {
	/* empty data column (small write), add a skip sector */
	ASSERT3U(skip_size, ==, parity_size);
	rc->rc_abd = abd_get_zeros(skip_size);
	} else if (rc->rc_size == parity_size) {
	/* this is a "big column" */
	rc->rc_abd = abd_get_offset_struct(&rc->rc_abdstruct,
	zio->io_abd, abd_off, rc->rc_size);
	} else {
	/* short data column, add a skip sector */
	ASSERT3U(rc->rc_size + skip_size, ==, parity_size);
	rc->rc_abd = abd_alloc_gang();
	abd_gang_add(rc->rc_abd, abd_get_offset_size(
	zio->io_abd, abd_off, rc->rc_size), B_TRUE);
	abd_gang_add(rc->rc_abd, abd_get_zeros(skip_size),
	B_TRUE);
	}

	ASSERT3U(abd_get_size(rc->rc_abd), ==, parity_size);

	abd_off += rc->rc_size;
	rc->rc_size = parity_size;
	}

	IMPLY(abd_offset != 0, abd_off == zio->io_size);
	}

	/*
	* Scrub/resilver reads. In order to store the contents of the skip sectors
	* an additional ABD is allocated. The columns are handled in the same way
	* as a full stripe write except instead of using the zero ABD the newly
	* allocated skip ABD is used to back the skip sectors. In all cases the
	* data ABD must be the same size as the parity ABDs.
	*/
	static void
	vdev_draid_map_alloc_scrub(zio_t zio, uint64_t abd_offset, raidz_row_t rr)
	{
	uint64_t skip_size = 1ULL << zio->io_vd->vdev_top->vdev_ashift;
	uint64_t parity_size = rr->rr_col[0].rc_size;
	uint64_t abd_off = abd_offset;
	uint64_t skip_off = 0;

	ASSERT3U(zio->io_type, ==, ZIO_TYPE_READ);
	ASSERT3P(rr->rr_abd_empty, ==, NULL);

	if (rr->rr_nempty > 0) {
	rr->rr_abd_empty = abd_alloc_linear(rr->rr_nempty * skip_size,
	B_FALSE);
	}

	for (uint64_t c = rr->rr_firstdatacol; c < rr->rr_cols; c++) {
	raidz_col_t *rc = &rr->rr_col[c];

	if (rc->rc_size == 0) {
	/* empty data column (small read), add a skip sector */
	ASSERT3U(skip_size, ==, parity_size);
	ASSERT3U(rr->rr_nempty, !=, 0);
	rc->rc_abd = abd_get_offset_size(rr->rr_abd_empty,
	skip_off, skip_size);
	skip_off += skip_size;
	} else if (rc->rc_size == parity_size) {
	/* this is a "big column" */
	rc->rc_abd = abd_get_offset_struct(&rc->rc_abdstruct,
	zio->io_abd, abd_off, rc->rc_size);
	} else {
	/* short data column, add a skip sector */
	ASSERT3U(rc->rc_size + skip_size, ==, parity_size);
	ASSERT3U(rr->rr_nempty, !=, 0);
	rc->rc_abd = abd_alloc_gang();
	abd_gang_add(rc->rc_abd, abd_get_offset_size(
	zio->io_abd, abd_off, rc->rc_size), B_TRUE);
	abd_gang_add(rc->rc_abd, abd_get_offset_size(
	rr->rr_abd_empty, skip_off, skip_size), B_TRUE);
	skip_off += skip_size;
	}

	uint64_t abd_size = abd_get_size(rc->rc_abd);
	ASSERT3U(abd_size, ==, abd_get_size(rr->rr_col[0].rc_abd));

	/*
	* Increase rc_size so the skip ABD is included in subsequent
	* parity calculations.
	*/
	abd_off += rc->rc_size;
	rc->rc_size = abd_size;
	}

	IMPLY(abd_offset != 0, abd_off == zio->io_size);
	ASSERT3U(skip_off, ==, rr->rr_nempty * skip_size);
	}

	/*
	* Normal reads. In this common case only the columns containing data
	* are read in to the zio ABDs. Neither the parity columns or empty skip
	* sectors are read unless the checksum fails verification. In which case
	* vdev_raidz_read_all() will call vdev_draid_map_alloc_empty() to expand
	* the raid map in order to allow reconstruction using the parity data and
	* skip sectors.
	*/
	static void
	vdev_draid_map_alloc_read(zio_t zio, uint64_t abd_offset, raidz_row_t rr)
	{
	uint64_t abd_off = abd_offset;

	ASSERT3U(zio->io_type, ==, ZIO_TYPE_READ);

	for (uint64_t c = rr->rr_firstdatacol; c < rr->rr_cols; c++) {
	raidz_col_t *rc = &rr->rr_col[c];

	if (rc->rc_size > 0) {
	rc->rc_abd = abd_get_offset_struct(&rc->rc_abdstruct,
	zio->io_abd, abd_off, rc->rc_size);
	abd_off += rc->rc_size;
	}
	}

	IMPLY(abd_offset != 0, abd_off == zio->io_size);
	}

	/*
	* Converts a normal "read" raidz_row_t to a "scrub" raidz_row_t. The key
	* difference is that an ABD is allocated to back skip sectors so they may
	* be read in to memory, verified, and repaired if needed.
	*/
	void
	vdev_draid_map_alloc_empty(zio_t zio, raidz_row_t rr)
	{
	uint64_t skip_size = 1ULL << zio->io_vd->vdev_top->vdev_ashift;
	uint64_t parity_size = rr->rr_col[0].rc_size;
	uint64_t skip_off = 0;

	ASSERT3U(zio->io_type, ==, ZIO_TYPE_READ);
	ASSERT3P(rr->rr_abd_empty, ==, NULL);

	if (rr->rr_nempty > 0) {
	rr->rr_abd_empty = abd_alloc_linear(rr->rr_nempty * skip_size,
	B_FALSE);
	}

	for (uint64_t c = rr->rr_firstdatacol; c < rr->rr_cols; c++) {
	raidz_col_t *rc = &rr->rr_col[c];

	if (rc->rc_size == 0) {
	/* empty data column (small read), add a skip sector */
	ASSERT3U(skip_size, ==, parity_size);
	ASSERT3U(rr->rr_nempty, !=, 0);
	ASSERT3P(rc->rc_abd, ==, NULL);
	rc->rc_abd = abd_get_offset_size(rr->rr_abd_empty,
	skip_off, skip_size);
	skip_off += skip_size;
	} else if (rc->rc_size == parity_size) {
	/* this is a "big column", nothing to add */
	ASSERT3P(rc->rc_abd, !=, NULL);
	} else {
	/*
	* short data column, add a skip sector and clear
	* rc_tried to force the entire column to be re-read
	* thereby including the missing skip sector data
	* which is needed for reconstruction.
	*/
	ASSERT3U(rc->rc_size + skip_size, ==, parity_size);
	ASSERT3U(rr->rr_nempty, !=, 0);
	ASSERT3P(rc->rc_abd, !=, NULL);
	ASSERT(!abd_is_gang(rc->rc_abd));
	abd_t *read_abd = rc->rc_abd;
	rc->rc_abd = abd_alloc_gang();
	abd_gang_add(rc->rc_abd, read_abd, B_TRUE);
	abd_gang_add(rc->rc_abd, abd_get_offset_size(
	rr->rr_abd_empty, skip_off, skip_size), B_TRUE);
	skip_off += skip_size;
	rc->rc_tried = 0;
	}

	/*
	* Increase rc_size so the empty ABD is included in subsequent
	* parity calculations.
	*/
	rc->rc_size = parity_size;
	}

	ASSERT3U(skip_off, ==, rr->rr_nempty * skip_size);
	}

	/*
	* Given a logical address within a dRAID configuration, return the physical
	* address on the first drive in the group that this address maps to
	* (at position 'start' in permutation number 'perm').
	*/
	static uint64_t
	vdev_draid_logical_to_physical(vdev_t *vd, uint64_t logical_offset,
	uint64_t perm, uint64_t start)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	/* b is the dRAID (parent) sector offset. */
	uint64_t ashift = vd->vdev_top->vdev_ashift;
	uint64_t b_offset = logical_offset >> ashift;

	/*
	* The height of a row in units of the vdev's minimum sector size.
	* This is the amount of data written to each disk of each group
	* in a given permutation.
	*/
	uint64_t rowheight_sectors = VDEV_DRAID_ROWHEIGHT >> ashift;

	/*
	* We cycle through a disk permutation every groupsz * ngroups chunk
	* of address space. Note that ngroups * groupsz must be a multiple
	* of the number of data drives (ndisks) in order to guarantee
	* alignment. So, for example, if our row height is 16MB, our group
	* size is 10, and there are 13 data drives in the draid, then ngroups
	* will be 13, we will change permutation every 2.08GB and each
	* disk will have 160MB of data per chunk.
	*/
	uint64_t groupwidth = vdc->vdc_groupwidth;
	uint64_t ngroups = vdc->vdc_ngroups;
	uint64_t ndisks = vdc->vdc_ndisks;

	/*
	* groupstart is where the group this IO will land in "starts" in
	* the permutation array.
	*/
	uint64_t group = logical_offset / vdc->vdc_groupsz;
	uint64_t groupstart = (group * groupwidth) % ndisks;
	ASSERT3U(groupstart + groupwidth, <=, ndisks + groupstart);
	*start = groupstart;

	/* b_offset is the sector offset within a group chunk */
	b_offset = b_offset % (rowheight_sectors * groupwidth);
	ASSERT0(b_offset % groupwidth);

	/*
	* Find the starting byte offset on each child vdev:
	* - within a permutation there are ngroups groups spread over the
	* rows, where each row covers a slice portion of the disk
	* - each permutation has (groupwidth * ngroups) / ndisks rows
	* - so each permutation covers rows * slice portion of the disk
	* - so we need to find the row where this IO group target begins
	*/
	*perm = group / ngroups;
	uint64_t row = (perm ((groupwidth * ngroups) / ndisks)) +
	(((group % ngroups) * groupwidth) / ndisks);

	return (((rowheight_sectors * row) +
	(b_offset / groupwidth)) << ashift);
	}

	static uint64_t
	vdev_draid_map_alloc_row(zio_t zio, raidz_row_t *rrp, uint64_t io_offset,
	uint64_t abd_offset, uint64_t abd_size)
	{
	vdev_t *vd = zio->io_vd;
	vdev_draid_config_t *vdc = vd->vdev_tsd;
	uint64_t ashift = vd->vdev_top->vdev_ashift;
	uint64_t io_size = abd_size;
	uint64_t io_asize = vdev_draid_asize(vd, io_size);
	uint64_t group = vdev_draid_offset_to_group(vd, io_offset);
	uint64_t start_offset = vdev_draid_group_to_offset(vd, group + 1);

	/*
	* Limit the io_size to the space remaining in the group. A second
	* row in the raidz_map_t is created for the remainder.
	*/
	if (io_offset + io_asize > start_offset) {
	io_size = vdev_draid_asize_to_psize(vd,
	start_offset - io_offset);
	}

	/*
	* At most a block may span the logical end of one group and the start
	* of the next group. Therefore, at the end of a group the io_size must
	* span the group width evenly and the remainder must be aligned to the
	* start of the next group.
	*/
	IMPLY(abd_offset == 0 && io_size < zio->io_size,
	(io_asize >> ashift) % vdc->vdc_groupwidth == 0);
	IMPLY(abd_offset != 0,
	vdev_draid_group_to_offset(vd, group) == io_offset);

	/* Lookup starting byte offset on each child vdev */
	uint64_t groupstart, perm;
	uint64_t physical_offset = vdev_draid_logical_to_physical(vd,
	io_offset, &perm, &groupstart);

	/*
	* If there is less than groupwidth drives available after the group
	* start, the group is going to wrap onto the next row. 'wrap' is the
	* group disk number that starts on the next row.
	*/
	uint64_t ndisks = vdc->vdc_ndisks;
	uint64_t groupwidth = vdc->vdc_groupwidth;
	uint64_t wrap = groupwidth;

	if (groupstart + groupwidth > ndisks)
	wrap = ndisks - groupstart;

	/* The io size in units of the vdev's minimum sector size. */
	const uint64_t psize = io_size >> ashift;

	/*
	* "Quotient": The number of data sectors for this stripe on all but
	* the "big column" child vdevs that also contain "remainder" data.
	*/
	uint64_t q = psize / vdc->vdc_ndata;

	/*
	* "Remainder": The number of partial stripe data sectors in this I/O.
	* This will add a sector to some, but not all, child vdevs.
	*/
	uint64_t r = psize - q * vdc->vdc_ndata;

	/* The number of "big columns" - those which contain remainder data. */
	uint64_t bc = (r == 0 ? 0 : r + vdc->vdc_nparity);
	ASSERT3U(bc, <, groupwidth);

	/* The total number of data and parity sectors for this I/O. */
	uint64_t tot = psize + (vdc->vdc_nparity * (q + (r == 0 ? 0 : 1)));

	raidz_row_t *rr;
	rr = kmem_alloc(offsetof(raidz_row_t, rr_col[groupwidth]), KM_SLEEP);
	rr->rr_cols = groupwidth;
	rr->rr_scols = groupwidth;
	rr->rr_bigcols = bc;
	rr->rr_missingdata = 0;
	rr->rr_missingparity = 0;
	rr->rr_firstdatacol = vdc->vdc_nparity;
	rr->rr_abd_empty = NULL;
	#ifdef ZFS_DEBUG
	rr->rr_offset = io_offset;
	rr->rr_size = io_size;
	#endif
	*rrp = rr;

	uint8_t *base;
	uint64_t iter, asize = 0;
	vdev_draid_get_perm(vdc, perm, &base, &iter);
	for (uint64_t i = 0; i < groupwidth; i++) {
	raidz_col_t *rc = &rr->rr_col[i];
	uint64_t c = (groupstart + i) % ndisks;

	/* increment the offset if we wrap to the next row */
	if (i == wrap)
	physical_offset += VDEV_DRAID_ROWHEIGHT;

	rc->rc_devidx = vdev_draid_permute_id(vdc, base, iter, c);
	rc->rc_offset = physical_offset;
	rc->rc_abd = NULL;
	rc->rc_orig_data = NULL;
	rc->rc_error = 0;
	rc->rc_tried = 0;
	rc->rc_skipped = 0;
	rc->rc_force_repair = 0;
	rc->rc_allow_repair = 1;
	rc->rc_need_orig_restore = B_FALSE;

	if (q == 0 && i >= bc)
	rc->rc_size = 0;
	else if (i < bc)
	rc->rc_size = (q + 1) << ashift;
	else
	rc->rc_size = q << ashift;

	asize += rc->rc_size;
	}

	ASSERT3U(asize, ==, tot << ashift);
	rr->rr_nempty = roundup(tot, groupwidth) - tot;
	IMPLY(bc > 0, rr->rr_nempty == groupwidth - bc);

	/* Allocate buffers for the parity columns */
	for (uint64_t c = 0; c < rr->rr_firstdatacol; c++) {
	raidz_col_t *rc = &rr->rr_col[c];
	rc->rc_abd = abd_alloc_linear(rc->rc_size, B_FALSE);
	}

	/*
	* Map buffers for data columns and allocate/map buffers for skip
	* sectors. There are three distinct cases for dRAID which are
	* required to support sequential rebuild.
	*/
	if (zio->io_type == ZIO_TYPE_WRITE) {
	vdev_draid_map_alloc_write(zio, abd_offset, rr);
	} else if ((rr->rr_nempty > 0) &&
	(zio->io_flags & (ZIO_FLAG_SCRUB \| ZIO_FLAG_RESILVER))) {
	vdev_draid_map_alloc_scrub(zio, abd_offset, rr);
	} else {
	ASSERT3U(zio->io_type, ==, ZIO_TYPE_READ);
	vdev_draid_map_alloc_read(zio, abd_offset, rr);
	}

	return (io_size);
	}

	/*
	* Allocate the raidz mapping to be applied to the dRAID I/O. The parity
	* calculations for dRAID are identical to raidz however there are a few
	* differences in the layout.
	*
	* - dRAID always allocates a full stripe width. Any extra sectors due
	* this padding are zero filled and written to disk. They will be read
	* back during a scrub or repair operation since they are included in
	* the parity calculation. This property enables sequential resilvering.
	*
	* - When the block at the logical offset spans redundancy groups then two
	* rows are allocated in the raidz_map_t. One row resides at the end of
	* the first group and the other at the start of the following group.
	*/
	static raidz_map_t *
	vdev_draid_map_alloc(zio_t *zio)
	{
	raidz_row_t *rr[2];
	uint64_t abd_offset = 0;
	uint64_t abd_size = zio->io_size;
	uint64_t io_offset = zio->io_offset;
	uint64_t size;
	int nrows = 1;

	size = vdev_draid_map_alloc_row(zio, &rr[0], io_offset,
	abd_offset, abd_size);
	if (size < abd_size) {
	vdev_t *vd = zio->io_vd;

	io_offset += vdev_draid_asize(vd, size);
	abd_offset += size;
	abd_size -= size;
	nrows++;

	ASSERT3U(io_offset, ==, vdev_draid_group_to_offset(
	vd, vdev_draid_offset_to_group(vd, io_offset)));
	ASSERT3U(abd_offset, <, zio->io_size);
	ASSERT3U(abd_size, !=, 0);

	size = vdev_draid_map_alloc_row(zio, &rr[1],
	io_offset, abd_offset, abd_size);
	VERIFY3U(size, ==, abd_size);
	}

	raidz_map_t *rm;
	rm = kmem_zalloc(offsetof(raidz_map_t, rm_row[nrows]), KM_SLEEP);
	rm->rm_ops = vdev_raidz_math_get_ops();
	rm->rm_nrows = nrows;
	rm->rm_row[0] = rr[0];
	if (nrows == 2)
	rm->rm_row[1] = rr[1];

	return (rm);
	}

	/*
	* Given an offset into a dRAID return the next group width aligned offset
	* which can be used to start an allocation.
	*/
	static uint64_t
	vdev_draid_get_astart(vdev_t *vd, const uint64_t start)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);

	return (roundup(start, vdc->vdc_groupwidth << vd->vdev_ashift));
	}

	/*
	* Allocatable space for dRAID is (children - nspares) * sizeof(smallest child)
	* rounded down to the last full slice. So each child must provide at least
	* 1 / (children - nspares) of its asize.
	*/
	static uint64_t
	vdev_draid_min_asize(vdev_t *vd)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);

	return (VDEV_DRAID_REFLOW_RESERVE +
	(vd->vdev_min_asize + vdc->vdc_ndisks - 1) / (vdc->vdc_ndisks));
	}

	/*
	* When using dRAID the minimum allocation size is determined by the number
	* of data disks in the redundancy group. Full stripes are always used.
	*/
	static uint64_t
	vdev_draid_min_alloc(vdev_t *vd)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);

	return (vdc->vdc_ndata << vd->vdev_ashift);
	}

	/*
	* Returns true if the txg range does not exist on any leaf vdev.
	*
	* A dRAID spare does not fit into the DTL model. While it has child vdevs
	* there is no redundancy among them, and the effective child vdev is
	* determined by offset. Essentially we do a vdev_dtl_reassess() on the
	* fly by replacing a dRAID spare with the child vdev under the offset.
	* Note that it is a recursive process because the child vdev can be
	* another dRAID spare and so on.
	*/
	boolean_t
	vdev_draid_missing(vdev_t *vd, uint64_t physical_offset, uint64_t txg,
	uint64_t size)
	{
	if (vd->vdev_ops == &vdev_spare_ops \|\|
	vd->vdev_ops == &vdev_replacing_ops) {
	/*
	* Check all of the readable children, if any child
	* contains the txg range the data it is not missing.
	*/
	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];

	if (!vdev_readable(cvd))
	continue;

	if (!vdev_draid_missing(cvd, physical_offset,
	txg, size))
	return (B_FALSE);
	}

	return (B_TRUE);
	}

	if (vd->vdev_ops == &vdev_draid_spare_ops) {
	/*
	* When sequentially resilvering we don't have a proper
	* txg range so instead we must presume all txgs are
	* missing on this vdev until the resilver completes.
	*/
	if (vd->vdev_rebuild_txg != 0)
	return (B_TRUE);

	/*
	* DTL_MISSING is set for all prior txgs when a resilver
	* is started in spa_vdev_attach().
	*/
	if (vdev_dtl_contains(vd, DTL_MISSING, txg, size))
	return (B_TRUE);

	/*
	* Consult the DTL on the relevant vdev. Either a vdev
	* leaf or spare/replace mirror child may be returned so
	* we must recursively call vdev_draid_missing_impl().
	*/
	vd = vdev_draid_spare_get_child(vd, physical_offset);
	if (vd == NULL)
	return (B_TRUE);

	return (vdev_draid_missing(vd, physical_offset,
	txg, size));
	}

	return (vdev_dtl_contains(vd, DTL_MISSING, txg, size));
	}

	/*
	* Returns true if the txg is only partially replicated on the leaf vdevs.
	*/
	static boolean_t
	vdev_draid_partial(vdev_t *vd, uint64_t physical_offset, uint64_t txg,
	uint64_t size)
	{
	if (vd->vdev_ops == &vdev_spare_ops \|\|
	vd->vdev_ops == &vdev_replacing_ops) {
	/*
	* Check all of the readable children, if any child is
	* missing the txg range then it is partially replicated.
	*/
	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];

	if (!vdev_readable(cvd))
	continue;

	if (vdev_draid_partial(cvd, physical_offset, txg, size))
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	if (vd->vdev_ops == &vdev_draid_spare_ops) {
	/*
	* When sequentially resilvering we don't have a proper
	* txg range so instead we must presume all txgs are
	* missing on this vdev until the resilver completes.
	*/
	if (vd->vdev_rebuild_txg != 0)
	return (B_TRUE);

	/*
	* DTL_MISSING is set for all prior txgs when a resilver
	* is started in spa_vdev_attach().
	*/
	if (vdev_dtl_contains(vd, DTL_MISSING, txg, size))
	return (B_TRUE);

	/*
	* Consult the DTL on the relevant vdev. Either a vdev
	* leaf or spare/replace mirror child may be returned so
	* we must recursively call vdev_draid_missing_impl().
	*/
	vd = vdev_draid_spare_get_child(vd, physical_offset);
	if (vd == NULL)
	return (B_TRUE);

	return (vdev_draid_partial(vd, physical_offset, txg, size));
	}

	return (vdev_dtl_contains(vd, DTL_MISSING, txg, size));
	}

	/*
	* Determine if the vdev is readable at the given offset.
	*/
	boolean_t
	vdev_draid_readable(vdev_t *vd, uint64_t physical_offset)
	{
	if (vd->vdev_ops == &vdev_draid_spare_ops) {
	vd = vdev_draid_spare_get_child(vd, physical_offset);
	if (vd == NULL)
	return (B_FALSE);
	}

	if (vd->vdev_ops == &vdev_spare_ops \|\|
	vd->vdev_ops == &vdev_replacing_ops) {

	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];

	if (!vdev_readable(cvd))
	continue;

	if (vdev_draid_readable(cvd, physical_offset))
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	return (vdev_readable(vd));
	}

	/*
	* Returns the first distributed spare found under the provided vdev tree.
	*/
	static vdev_t *
	vdev_draid_find_spare(vdev_t *vd)
	{
	if (vd->vdev_ops == &vdev_draid_spare_ops)
	return (vd);

	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *svd = vdev_draid_find_spare(vd->vdev_child[c]);
	if (svd != NULL)
	return (svd);
	}

	return (NULL);
	}

	/*
	* Returns B_TRUE if the passed in vdev is currently "faulted".
	* Faulted, in this context, means that the vdev represents a
	* replacing or sparing vdev tree.
	*/
	static boolean_t
	vdev_draid_faulted(vdev_t *vd, uint64_t physical_offset)
	{
	if (vd->vdev_ops == &vdev_draid_spare_ops) {
	vd = vdev_draid_spare_get_child(vd, physical_offset);
	if (vd == NULL)
	return (B_FALSE);

	/*
	* After resolving the distributed spare to a leaf vdev
	* check the parent to determine if it's "faulted".
	*/
	vd = vd->vdev_parent;
	}

	return (vd->vdev_ops == &vdev_replacing_ops \|\|
	vd->vdev_ops == &vdev_spare_ops);
	}

	/*
	* Determine if the dRAID block at the logical offset is degraded.
	* Used by sequential resilver.
	*/
	static boolean_t
	vdev_draid_group_degraded(vdev_t *vd, uint64_t offset)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);
	ASSERT3U(vdev_draid_get_astart(vd, offset), ==, offset);

	uint64_t groupstart, perm;
	uint64_t physical_offset = vdev_draid_logical_to_physical(vd,
	offset, &perm, &groupstart);

	uint8_t *base;
	uint64_t iter;
	vdev_draid_get_perm(vdc, perm, &base, &iter);

	for (uint64_t i = 0; i < vdc->vdc_groupwidth; i++) {
	uint64_t c = (groupstart + i) % vdc->vdc_ndisks;
	uint64_t cid = vdev_draid_permute_id(vdc, base, iter, c);
	vdev_t *cvd = vd->vdev_child[cid];

	/* Group contains a faulted vdev. */
	if (vdev_draid_faulted(cvd, physical_offset))
	return (B_TRUE);

	/*
	* Always check groups with active distributed spares
	* because any vdev failure in the pool will affect them.
	*/
	if (vdev_draid_find_spare(cvd) != NULL)
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	/*
	* Determine if the txg is missing. Used by healing resilver.
	*/
	static boolean_t
	vdev_draid_group_missing(vdev_t *vd, uint64_t offset, uint64_t txg,
	uint64_t size)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);
	ASSERT3U(vdev_draid_get_astart(vd, offset), ==, offset);

	uint64_t groupstart, perm;
	uint64_t physical_offset = vdev_draid_logical_to_physical(vd,
	offset, &perm, &groupstart);

	uint8_t *base;
	uint64_t iter;
	vdev_draid_get_perm(vdc, perm, &base, &iter);

	for (uint64_t i = 0; i < vdc->vdc_groupwidth; i++) {
	uint64_t c = (groupstart + i) % vdc->vdc_ndisks;
	uint64_t cid = vdev_draid_permute_id(vdc, base, iter, c);
	vdev_t *cvd = vd->vdev_child[cid];

	/* Transaction group is known to be partially replicated. */
	if (vdev_draid_partial(cvd, physical_offset, txg, size))
	return (B_TRUE);

	/*
	* Always check groups with active distributed spares
	* because any vdev failure in the pool will affect them.
	*/
	if (vdev_draid_find_spare(cvd) != NULL)
	return (B_TRUE);
	}

	return (B_FALSE);
	}

	/*
	* Find the smallest child asize and largest sector size to calculate the
	* available capacity. Distributed spares are ignored since their capacity
	* is also based of the minimum child size in the top-level dRAID.
	*/
	static void
	vdev_draid_calculate_asize(vdev_t vd, uint64_t asizep, uint64_t *max_asizep,
	uint64_t logical_ashiftp, uint64_t physical_ashiftp)
	{
	uint64_t logical_ashift = 0, physical_ashift = 0;
	uint64_t asize = 0, max_asize = 0;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);

	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];

	if (cvd->vdev_ops == &vdev_draid_spare_ops)
	continue;

	asize = MIN(asize - 1, cvd->vdev_asize - 1) + 1;
	max_asize = MIN(max_asize - 1, cvd->vdev_max_asize - 1) + 1;
	logical_ashift = MAX(logical_ashift, cvd->vdev_ashift);
	physical_ashift = MAX(physical_ashift,
	cvd->vdev_physical_ashift);
	}

	*asizep = asize;
	*max_asizep = max_asize;
	*logical_ashiftp = logical_ashift;
	*physical_ashiftp = physical_ashift;
	}

	/*
	* Open spare vdevs.
	*/
	static boolean_t
	vdev_draid_open_spares(vdev_t *vd)
	{
	return (vd->vdev_ops == &vdev_draid_spare_ops \|\|
	vd->vdev_ops == &vdev_replacing_ops \|\|
	vd->vdev_ops == &vdev_spare_ops);
	}

	/*
	* Open all children, excluding spares.
	*/
	static boolean_t
	vdev_draid_open_children(vdev_t *vd)
	{
	return (!vdev_draid_open_spares(vd));
	}

	/*
	* Open a top-level dRAID vdev.
	*/
	static int
	vdev_draid_open(vdev_t vd, uint64_t asize, uint64_t *max_asize,
	uint64_t logical_ashift, uint64_t physical_ashift)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;
	uint64_t nparity = vdc->vdc_nparity;
	int open_errors = 0;

	if (nparity > VDEV_DRAID_MAXPARITY \|\|
	vd->vdev_children < nparity + 1) {
	vd->vdev_stat.vs_aux = VDEV_AUX_BAD_LABEL;
	return (SET_ERROR(EINVAL));
	}

	/*
	* First open the normal children then the distributed spares. This
	* ordering is important to ensure the distributed spares calculate
	* the correct psize in the event that the dRAID vdevs were expanded.
	*/
	vdev_open_children_subset(vd, vdev_draid_open_children);
	vdev_open_children_subset(vd, vdev_draid_open_spares);

	/* Verify enough of the children are available to continue. */
	for (int c = 0; c < vd->vdev_children; c++) {
	if (vd->vdev_child[c]->vdev_open_error != 0) {
	if ((++open_errors) > nparity) {
	vd->vdev_stat.vs_aux = VDEV_AUX_NO_REPLICAS;
	return (SET_ERROR(ENXIO));
	}
	}
	}

	/*
	* Allocatable capacity is the sum of the space on all children less
	* the number of distributed spares rounded down to last full row
	* and then to the last full group. An additional 32MB of scratch
	* space is reserved at the end of each child for use by the dRAID
	* expansion feature.
	*/
	uint64_t child_asize, child_max_asize;
	vdev_draid_calculate_asize(vd, &child_asize, &child_max_asize,
	logical_ashift, physical_ashift);

	/*
	* Should be unreachable since the minimum child size is 64MB, but
	* we want to make sure an underflow absolutely cannot occur here.
	*/
	if (child_asize < VDEV_DRAID_REFLOW_RESERVE \|\|
	child_max_asize < VDEV_DRAID_REFLOW_RESERVE) {
	return (SET_ERROR(ENXIO));
	}

	child_asize = ((child_asize - VDEV_DRAID_REFLOW_RESERVE) /
	VDEV_DRAID_ROWHEIGHT) * VDEV_DRAID_ROWHEIGHT;
	child_max_asize = ((child_max_asize - VDEV_DRAID_REFLOW_RESERVE) /
	VDEV_DRAID_ROWHEIGHT) * VDEV_DRAID_ROWHEIGHT;

	asize = (((child_asize vdc->vdc_ndisks) / vdc->vdc_groupsz) *
	vdc->vdc_groupsz);
	max_asize = (((child_max_asize vdc->vdc_ndisks) / vdc->vdc_groupsz) *
	vdc->vdc_groupsz);

	return (0);
	}

	/*
	* Close a top-level dRAID vdev.
	*/
	static void
	vdev_draid_close(vdev_t *vd)
	{
	for (int c = 0; c < vd->vdev_children; c++) {
	if (vd->vdev_child[c] != NULL)
	vdev_close(vd->vdev_child[c]);
	}
	}

	/*
	* Return the maximum asize for a rebuild zio in the provided range
	* given the following constraints. A dRAID chunks may not:
	*
	* - Exceed the maximum allowed block size (SPA_MAXBLOCKSIZE), or
	* - Span dRAID redundancy groups.
	*/
	static uint64_t
	vdev_draid_rebuild_asize(vdev_t *vd, uint64_t start, uint64_t asize,
	uint64_t max_segment)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);

	uint64_t ashift = vd->vdev_ashift;
	uint64_t ndata = vdc->vdc_ndata;
	uint64_t psize = MIN(P2ROUNDUP(max_segment * ndata, 1 << ashift),
	SPA_MAXBLOCKSIZE);

	ASSERT3U(vdev_draid_get_astart(vd, start), ==, start);
	ASSERT3U(asize % (vdc->vdc_groupwidth << ashift), ==, 0);

	/* Chunks must evenly span all data columns in the group. */
	psize = (((psize >> ashift) / ndata) * ndata) << ashift;
	uint64_t chunk_size = MIN(asize, vdev_psize_to_asize(vd, psize));

	/* Reduce the chunk size to the group space remaining. */
	uint64_t group = vdev_draid_offset_to_group(vd, start);
	uint64_t left = vdev_draid_group_to_offset(vd, group + 1) - start;
	chunk_size = MIN(chunk_size, left);

	ASSERT3U(chunk_size % (vdc->vdc_groupwidth << ashift), ==, 0);
	ASSERT3U(vdev_draid_offset_to_group(vd, start), ==,
	vdev_draid_offset_to_group(vd, start + chunk_size - 1));

	return (chunk_size);
	}

	/*
	* Align the start of the metaslab to the group width and slightly reduce
	* its size to a multiple of the group width. Since full stripe writes are
	* required by dRAID this space is unallocable. Furthermore, aligning the
	* metaslab start is important for vdev initialize and TRIM which both operate
	* on metaslab boundaries which vdev_xlate() expects to be aligned.
	*/
	static void
	vdev_draid_metaslab_init(vdev_t vd, uint64_t ms_start, uint64_t *ms_size)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);

	uint64_t sz = vdc->vdc_groupwidth << vd->vdev_ashift;
	uint64_t astart = vdev_draid_get_astart(vd, *ms_start);
	uint64_t asize = ((ms_size - (astart - ms_start)) / sz) * sz;

	*ms_start = astart;
	*ms_size = asize;

	ASSERT0(*ms_start % sz);
	ASSERT0(*ms_size % sz);
	}

	/*
	* Add virtual dRAID spares to the list of valid spares. In order to accomplish
	* this the existing array must be freed and reallocated with the additional
	* entries.
	*/
	int
	vdev_draid_spare_create(nvlist_t nvroot, vdev_t vd, uint64_t *ndraidp,
	uint64_t next_vdev_id)
	{
	uint64_t draid_nspares = 0;
	uint64_t ndraid = 0;
	int error;

	for (uint64_t i = 0; i < vd->vdev_children; i++) {
	vdev_t *cvd = vd->vdev_child[i];

	if (cvd->vdev_ops == &vdev_draid_ops) {
	vdev_draid_config_t *vdc = cvd->vdev_tsd;
	draid_nspares += vdc->vdc_nspares;
	ndraid++;
	}
	}

	if (draid_nspares == 0) {
	*ndraidp = ndraid;
	return (0);
	}

	nvlist_t old_spares, new_spares;
	uint_t old_nspares;
	error = nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	&old_spares, &old_nspares);
	if (error)
	old_nspares = 0;

	/* Allocate memory and copy of the existing spares. */
	new_spares = kmem_alloc(sizeof (nvlist_t )
	(draid_nspares + old_nspares), KM_SLEEP);
	for (uint_t i = 0; i < old_nspares; i++)
	new_spares[i] = fnvlist_dup(old_spares[i]);

	/* Add new distributed spares to ZPOOL_CONFIG_SPARES. */
	uint64_t n = old_nspares;
	for (uint64_t vdev_id = 0; vdev_id < vd->vdev_children; vdev_id++) {
	vdev_t *cvd = vd->vdev_child[vdev_id];
	char path[64];

	if (cvd->vdev_ops != &vdev_draid_ops)
	continue;

	vdev_draid_config_t *vdc = cvd->vdev_tsd;
	uint64_t nspares = vdc->vdc_nspares;
	uint64_t nparity = vdc->vdc_nparity;

	for (uint64_t spare_id = 0; spare_id < nspares; spare_id++) {
	bzero(path, sizeof (path));
	(void) snprintf(path, sizeof (path) - 1,
	"%s%llu-%llu-%llu", VDEV_TYPE_DRAID,
	(u_longlong_t)nparity,
	(u_longlong_t)next_vdev_id + vdev_id,
	(u_longlong_t)spare_id);

	nvlist_t *spare = fnvlist_alloc();
	fnvlist_add_string(spare, ZPOOL_CONFIG_PATH, path);
	fnvlist_add_string(spare, ZPOOL_CONFIG_TYPE,
	VDEV_TYPE_DRAID_SPARE);
	fnvlist_add_uint64(spare, ZPOOL_CONFIG_TOP_GUID,
	cvd->vdev_guid);
	fnvlist_add_uint64(spare, ZPOOL_CONFIG_SPARE_ID,
	spare_id);
	fnvlist_add_uint64(spare, ZPOOL_CONFIG_IS_LOG, 0);
	fnvlist_add_uint64(spare, ZPOOL_CONFIG_IS_SPARE, 1);
	fnvlist_add_uint64(spare, ZPOOL_CONFIG_WHOLE_DISK, 1);
	fnvlist_add_uint64(spare, ZPOOL_CONFIG_ASHIFT,
	cvd->vdev_ashift);

	new_spares[n] = spare;
	n++;
	}
	}

	if (n > 0) {
	(void) nvlist_remove_all(nvroot, ZPOOL_CONFIG_SPARES);
	fnvlist_add_nvlist_array(nvroot, ZPOOL_CONFIG_SPARES,
	- new_spares, n);
	+ (const nvlist_t **)new_spares, n);
	}

	for (int i = 0; i < n; i++)
	nvlist_free(new_spares[i]);

	kmem_free(new_spares, sizeof (new_spares) n);
	*ndraidp = ndraid;

	return (0);
	}

	/*
	* Determine if any portion of the provided block resides on a child vdev
	* with a dirty DTL and therefore needs to be resilvered.
	*/
	static boolean_t
	vdev_draid_need_resilver(vdev_t vd, const dva_t dva, size_t psize,
	uint64_t phys_birth)
	{
	uint64_t offset = DVA_GET_OFFSET(dva);
	uint64_t asize = vdev_draid_asize(vd, psize);

	if (phys_birth == TXG_UNKNOWN) {
	/*
	* Sequential resilver. There is no meaningful phys_birth
	* for this block, we can only determine if block resides
	* in a degraded group in which case it must be resilvered.
	*/
	ASSERT3U(vdev_draid_offset_to_group(vd, offset), ==,
	vdev_draid_offset_to_group(vd, offset + asize - 1));

	return (vdev_draid_group_degraded(vd, offset));
	} else {
	/*
	* Healing resilver. TXGs not in DTL_PARTIAL are intact,
	* as are blocks in non-degraded groups.
	*/
	if (!vdev_dtl_contains(vd, DTL_PARTIAL, phys_birth, 1))
	return (B_FALSE);

	if (vdev_draid_group_missing(vd, offset, phys_birth, 1))
	return (B_TRUE);

	/* The block may span groups in which case check both. */
	if (vdev_draid_offset_to_group(vd, offset) !=
	vdev_draid_offset_to_group(vd, offset + asize - 1)) {
	if (vdev_draid_group_missing(vd,
	offset + asize, phys_birth, 1))
	return (B_TRUE);
	}

	return (B_FALSE);
	}
	}

	static boolean_t
	vdev_draid_rebuilding(vdev_t *vd)
	{
	if (vd->vdev_ops->vdev_op_leaf && vd->vdev_rebuild_txg)
	return (B_TRUE);

	for (int i = 0; i < vd->vdev_children; i++) {
	if (vdev_draid_rebuilding(vd->vdev_child[i])) {
	return (B_TRUE);
	}
	}

	return (B_FALSE);
	}

	static void
	vdev_draid_io_verify(vdev_t vd, raidz_row_t rr, int col)
	{
	#ifdef ZFS_DEBUG
	range_seg64_t logical_rs, physical_rs, remain_rs;
	logical_rs.rs_start = rr->rr_offset;
	logical_rs.rs_end = logical_rs.rs_start +
	vdev_draid_asize(vd, rr->rr_size);

	raidz_col_t *rc = &rr->rr_col[col];
	vdev_t *cvd = vd->vdev_child[rc->rc_devidx];

	vdev_xlate(cvd, &logical_rs, &physical_rs, &remain_rs);
	ASSERT(vdev_xlate_is_empty(&remain_rs));
	ASSERT3U(rc->rc_offset, ==, physical_rs.rs_start);
	ASSERT3U(rc->rc_offset, <, physical_rs.rs_end);
	ASSERT3U(rc->rc_offset + rc->rc_size, ==, physical_rs.rs_end);
	#endif
	}

	/*
	* For write operations:
	* 1. Generate the parity data
	* 2. Create child zio write operations to each column's vdev, for both
	* data and parity. A gang ABD is allocated by vdev_draid_map_alloc()
	* if a skip sector needs to be added to a column.
	*/
	static void
	vdev_draid_io_start_write(zio_t zio, raidz_row_t rr)
	{
	vdev_t *vd = zio->io_vd;
	raidz_map_t *rm = zio->io_vsd;

	vdev_raidz_generate_parity_row(rm, rr);

	for (int c = 0; c < rr->rr_cols; c++) {
	raidz_col_t *rc = &rr->rr_col[c];

	/*
	* Empty columns are zero filled and included in the parity
	* calculation and therefore must be written.
	*/
	ASSERT3U(rc->rc_size, !=, 0);

	/* Verify physical to logical translation */
	vdev_draid_io_verify(vd, rr, c);

	zio_nowait(zio_vdev_child_io(zio, NULL,
	vd->vdev_child[rc->rc_devidx], rc->rc_offset,
	rc->rc_abd, rc->rc_size, zio->io_type, zio->io_priority,
	0, vdev_raidz_child_done, rc));
	}
	}

	/*
	* For read operations:
	* 1. The vdev_draid_map_alloc() function will create a minimal raidz
	* mapping for the read based on the zio->io_flags. There are two
	* possible mappings either 1) a normal read, or 2) a scrub/resilver.
	* 2. Create the zio read operations. This will include all parity
	* columns and skip sectors for a scrub/resilver.
	*/
	static void
	vdev_draid_io_start_read(zio_t zio, raidz_row_t rr)
	{
	vdev_t *vd = zio->io_vd;

	/* Sequential rebuild must do IO at redundancy group boundary. */
	IMPLY(zio->io_priority == ZIO_PRIORITY_REBUILD, rr->rr_nempty == 0);

	/*
	* Iterate over the columns in reverse order so that we hit the parity
	* last. Any errors along the way will force us to read the parity.
	* For scrub/resilver IOs which verify skip sectors, a gang ABD will
	* have been allocated to store them and rc->rc_size is increased.
	*/
	for (int c = rr->rr_cols - 1; c >= 0; c--) {
	raidz_col_t *rc = &rr->rr_col[c];
	vdev_t *cvd = vd->vdev_child[rc->rc_devidx];

	if (!vdev_draid_readable(cvd, rc->rc_offset)) {
	if (c >= rr->rr_firstdatacol)
	rr->rr_missingdata++;
	else
	rr->rr_missingparity++;
	rc->rc_error = SET_ERROR(ENXIO);
	rc->rc_tried = 1;
	rc->rc_skipped = 1;
	continue;
	}

	if (vdev_draid_missing(cvd, rc->rc_offset, zio->io_txg, 1)) {
	if (c >= rr->rr_firstdatacol)
	rr->rr_missingdata++;
	else
	rr->rr_missingparity++;
	rc->rc_error = SET_ERROR(ESTALE);
	rc->rc_skipped = 1;
	continue;
	}

	/*
	* Empty columns may be read during vdev_draid_io_done().
	* Only skip them after the readable and missing checks
	* verify they are available.
	*/
	if (rc->rc_size == 0) {
	rc->rc_skipped = 1;
	continue;
	}

	if (zio->io_flags & ZIO_FLAG_RESILVER) {
	vdev_t *svd;

	/*
	* Sequential rebuilds need to always consider the data
	* on the child being rebuilt to be stale. This is
	* important when all columns are available to aid
	* known reconstruction in identifing which columns
	* contain incorrect data.
	*
	* Furthermore, all repairs need to be constrained to
	* the devices being rebuilt because without a checksum
	* we cannot verify the data is actually correct and
	* performing an incorrect repair could result in
	* locking in damage and making the data unrecoverable.
	*/
	if (zio->io_priority == ZIO_PRIORITY_REBUILD) {
	if (vdev_draid_rebuilding(cvd)) {
	if (c >= rr->rr_firstdatacol)
	rr->rr_missingdata++;
	else
	rr->rr_missingparity++;
	rc->rc_error = SET_ERROR(ESTALE);
	rc->rc_skipped = 1;
	rc->rc_allow_repair = 1;
	continue;
	} else {
	rc->rc_allow_repair = 0;
	}
	} else {
	rc->rc_allow_repair = 1;
	}

	/*
	* If this child is a distributed spare then the
	* offset might reside on the vdev being replaced.
	* In which case this data must be written to the
	* new device. Failure to do so would result in
	* checksum errors when the old device is detached
	* and the pool is scrubbed.
	*/
	if ((svd = vdev_draid_find_spare(cvd)) != NULL) {
	svd = vdev_draid_spare_get_child(svd,
	rc->rc_offset);
	if (svd && (svd->vdev_ops == &vdev_spare_ops \|\|
	svd->vdev_ops == &vdev_replacing_ops)) {
	rc->rc_force_repair = 1;

	if (vdev_draid_rebuilding(svd))
	rc->rc_allow_repair = 1;
	}
	}

	/*
	* Always issue a repair IO to this child when its
	* a spare or replacing vdev with an active rebuild.
	*/
	if ((cvd->vdev_ops == &vdev_spare_ops \|\|
	cvd->vdev_ops == &vdev_replacing_ops) &&
	vdev_draid_rebuilding(cvd)) {
	rc->rc_force_repair = 1;
	rc->rc_allow_repair = 1;
	}
	}
	}

	/*
	* Either a parity or data column is missing this means a repair
	* may be attempted by vdev_draid_io_done(). Expand the raid map
	* to read in empty columns which are needed along with the parity
	* during reconstruction.
	*/
	if ((rr->rr_missingdata > 0 \|\| rr->rr_missingparity > 0) &&
	rr->rr_nempty > 0 && rr->rr_abd_empty == NULL) {
	vdev_draid_map_alloc_empty(zio, rr);
	}

	for (int c = rr->rr_cols - 1; c >= 0; c--) {
	raidz_col_t *rc = &rr->rr_col[c];
	vdev_t *cvd = vd->vdev_child[rc->rc_devidx];

	if (rc->rc_error \|\| rc->rc_size == 0)
	continue;

	if (c >= rr->rr_firstdatacol \|\| rr->rr_missingdata > 0 \|\|
	(zio->io_flags & (ZIO_FLAG_SCRUB \| ZIO_FLAG_RESILVER))) {
	zio_nowait(zio_vdev_child_io(zio, NULL, cvd,
	rc->rc_offset, rc->rc_abd, rc->rc_size,
	zio->io_type, zio->io_priority, 0,
	vdev_raidz_child_done, rc));
	}
	}
	}

	/*
	* Start an IO operation to a dRAID vdev.
	*/
	static void
	vdev_draid_io_start(zio_t *zio)
	{
	vdev_t *vd __maybe_unused = zio->io_vd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);
	ASSERT3U(zio->io_offset, ==, vdev_draid_get_astart(vd, zio->io_offset));

	raidz_map_t *rm = vdev_draid_map_alloc(zio);
	zio->io_vsd = rm;
	zio->io_vsd_ops = &vdev_raidz_vsd_ops;

	if (zio->io_type == ZIO_TYPE_WRITE) {
	for (int i = 0; i < rm->rm_nrows; i++) {
	vdev_draid_io_start_write(zio, rm->rm_row[i]);
	}
	} else {
	ASSERT(zio->io_type == ZIO_TYPE_READ);

	for (int i = 0; i < rm->rm_nrows; i++) {
	vdev_draid_io_start_read(zio, rm->rm_row[i]);
	}
	}

	zio_execute(zio);
	}

	/*
	* Complete an IO operation on a dRAID vdev. The raidz logic can be applied
	* to dRAID since the layout is fully described by the raidz_map_t.
	*/
	static void
	vdev_draid_io_done(zio_t *zio)
	{
	vdev_raidz_io_done(zio);
	}

	static void
	vdev_draid_state_change(vdev_t *vd, int faulted, int degraded)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;
	ASSERT(vd->vdev_ops == &vdev_draid_ops);

	if (faulted > vdc->vdc_nparity)
	vdev_set_state(vd, B_FALSE, VDEV_STATE_CANT_OPEN,
	VDEV_AUX_NO_REPLICAS);
	else if (degraded + faulted != 0)
	vdev_set_state(vd, B_FALSE, VDEV_STATE_DEGRADED, VDEV_AUX_NONE);
	else
	vdev_set_state(vd, B_FALSE, VDEV_STATE_HEALTHY, VDEV_AUX_NONE);
	}

	static void
	vdev_draid_xlate(vdev_t cvd, const range_seg64_t logical_rs,
	range_seg64_t physical_rs, range_seg64_t remain_rs)
	{
	vdev_t *raidvd = cvd->vdev_parent;
	ASSERT(raidvd->vdev_ops == &vdev_draid_ops);

	vdev_draid_config_t *vdc = raidvd->vdev_tsd;
	uint64_t ashift = raidvd->vdev_top->vdev_ashift;

	/* Make sure the offsets are block-aligned */
	ASSERT0(logical_rs->rs_start % (1 << ashift));
	ASSERT0(logical_rs->rs_end % (1 << ashift));

	uint64_t logical_start = logical_rs->rs_start;
	uint64_t logical_end = logical_rs->rs_end;

	/*
	* Unaligned ranges must be skipped. All metaslabs are correctly
	* aligned so this should not happen, but this case is handled in
	* case it's needed by future callers.
	*/
	uint64_t astart = vdev_draid_get_astart(raidvd, logical_start);
	if (astart != logical_start) {
	physical_rs->rs_start = logical_start;
	physical_rs->rs_end = logical_start;
	remain_rs->rs_start = MIN(astart, logical_end);
	remain_rs->rs_end = logical_end;
	return;
	}

	/*
	* Unlike with mirrors and raidz a dRAID logical range can map
	* to multiple non-contiguous physical ranges. This is handled by
	* limiting the size of the logical range to a single group and
	* setting the remain argument such that it describes the remaining
	* unmapped logical range. This is stricter than absolutely
	* necessary but helps simplify the logic below.
	*/
	uint64_t group = vdev_draid_offset_to_group(raidvd, logical_start);
	uint64_t nextstart = vdev_draid_group_to_offset(raidvd, group + 1);
	if (logical_end > nextstart)
	logical_end = nextstart;

	/* Find the starting offset for each vdev in the group */
	uint64_t perm, groupstart;
	uint64_t start = vdev_draid_logical_to_physical(raidvd,
	logical_start, &perm, &groupstart);
	uint64_t end = start;

	uint8_t *base;
	uint64_t iter, id;
	vdev_draid_get_perm(vdc, perm, &base, &iter);

	/*
	* Check if the passed child falls within the group. If it does
	* update the start and end to reflect the physical range.
	* Otherwise, leave them unmodified which will result in an empty
	* (zero-length) physical range being returned.
	*/
	for (uint64_t i = 0; i < vdc->vdc_groupwidth; i++) {
	uint64_t c = (groupstart + i) % vdc->vdc_ndisks;

	if (c == 0 && i != 0) {
	/* the group wrapped, increment the start */
	start += VDEV_DRAID_ROWHEIGHT;
	end = start;
	}

	id = vdev_draid_permute_id(vdc, base, iter, c);
	if (id == cvd->vdev_id) {
	uint64_t b_size = (logical_end >> ashift) -
	(logical_start >> ashift);
	ASSERT3U(b_size, >, 0);
	end = start + ((((b_size - 1) /
	vdc->vdc_groupwidth) + 1) << ashift);
	break;
	}
	}
	physical_rs->rs_start = start;
	physical_rs->rs_end = end;

	/*
	* Only top-level vdevs are allowed to set remain_rs because
	* when .vdev_op_xlate() is called for their children the full
	* logical range is not provided by vdev_xlate().
	*/
	remain_rs->rs_start = logical_end;
	remain_rs->rs_end = logical_rs->rs_end;

	ASSERT3U(physical_rs->rs_start, <=, logical_start);
	ASSERT3U(physical_rs->rs_end - physical_rs->rs_start, <=,
	logical_end - logical_start);
	}

	/*
	* Add dRAID specific fields to the config nvlist.
	*/
	static void
	vdev_draid_config_generate(vdev_t vd, nvlist_t nv)
	{
	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_ops);
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	fnvlist_add_uint64(nv, ZPOOL_CONFIG_NPARITY, vdc->vdc_nparity);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_DRAID_NDATA, vdc->vdc_ndata);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_DRAID_NSPARES, vdc->vdc_nspares);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_DRAID_NGROUPS, vdc->vdc_ngroups);
	}

	/*
	* Initialize private dRAID specific fields from the nvlist.
	*/
	static int
	vdev_draid_init(spa_t spa, nvlist_t nv, void **tsd)
	{
	uint64_t ndata, nparity, nspares, ngroups;
	int error;

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DRAID_NDATA, &ndata))
	return (SET_ERROR(EINVAL));

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_NPARITY, &nparity) \|\|
	nparity == 0 \|\| nparity > VDEV_DRAID_MAXPARITY) {
	return (SET_ERROR(EINVAL));
	}

	uint_t children;
	nvlist_t **child;
	if (nvlist_lookup_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	&child, &children) != 0 \|\| children == 0 \|\|
	children > VDEV_DRAID_MAX_CHILDREN) {
	return (SET_ERROR(EINVAL));
	}

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DRAID_NSPARES, &nspares) \|\|
	nspares > 100 \|\| nspares > (children - (ndata + nparity))) {
	return (SET_ERROR(EINVAL));
	}

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_DRAID_NGROUPS, &ngroups) \|\|
	ngroups == 0 \|\| ngroups > VDEV_DRAID_MAX_CHILDREN) {
	return (SET_ERROR(EINVAL));
	}

	/*
	* Validate the minimum number of children exist per group for the
	* specified parity level (draid1 >= 2, draid2 >= 3, draid3 >= 4).
	*/
	if (children < (ndata + nparity + nspares))
	return (SET_ERROR(EINVAL));

	/*
	* Create the dRAID configuration using the pool nvlist configuration
	* and the fixed mapping for the correct number of children.
	*/
	vdev_draid_config_t *vdc;
	const draid_map_t *map;

	error = vdev_draid_lookup_map(children, &map);
	if (error)
	return (SET_ERROR(EINVAL));

	vdc = kmem_zalloc(sizeof (*vdc), KM_SLEEP);
	vdc->vdc_ndata = ndata;
	vdc->vdc_nparity = nparity;
	vdc->vdc_nspares = nspares;
	vdc->vdc_children = children;
	vdc->vdc_ngroups = ngroups;
	vdc->vdc_nperms = map->dm_nperms;

	error = vdev_draid_generate_perms(map, &vdc->vdc_perms);
	if (error) {
	kmem_free(vdc, sizeof (*vdc));
	return (SET_ERROR(EINVAL));
	}

	/*
	* Derived constants.
	*/
	vdc->vdc_groupwidth = vdc->vdc_ndata + vdc->vdc_nparity;
	vdc->vdc_ndisks = vdc->vdc_children - vdc->vdc_nspares;
	vdc->vdc_groupsz = vdc->vdc_groupwidth * VDEV_DRAID_ROWHEIGHT;
	vdc->vdc_devslicesz = (vdc->vdc_groupsz * vdc->vdc_ngroups) /
	vdc->vdc_ndisks;

	ASSERT3U(vdc->vdc_groupwidth, >=, 2);
	ASSERT3U(vdc->vdc_groupwidth, <=, vdc->vdc_ndisks);
	ASSERT3U(vdc->vdc_groupsz, >=, 2 * VDEV_DRAID_ROWHEIGHT);
	ASSERT3U(vdc->vdc_devslicesz, >=, VDEV_DRAID_ROWHEIGHT);
	ASSERT3U(vdc->vdc_devslicesz % VDEV_DRAID_ROWHEIGHT, ==, 0);
	ASSERT3U((vdc->vdc_groupwidth * vdc->vdc_ngroups) %
	vdc->vdc_ndisks, ==, 0);

	*tsd = vdc;

	return (0);
	}

	static void
	vdev_draid_fini(vdev_t *vd)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	vmem_free(vdc->vdc_perms, sizeof (uint8_t) *
	vdc->vdc_children * vdc->vdc_nperms);
	kmem_free(vdc, sizeof (*vdc));
	}

	static uint64_t
	vdev_draid_nparity(vdev_t *vd)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	return (vdc->vdc_nparity);
	}

	static uint64_t
	vdev_draid_ndisks(vdev_t *vd)
	{
	vdev_draid_config_t *vdc = vd->vdev_tsd;

	return (vdc->vdc_ndisks);
	}

	vdev_ops_t vdev_draid_ops = {
	.vdev_op_init = vdev_draid_init,
	.vdev_op_fini = vdev_draid_fini,
	.vdev_op_open = vdev_draid_open,
	.vdev_op_close = vdev_draid_close,
	.vdev_op_asize = vdev_draid_asize,
	.vdev_op_min_asize = vdev_draid_min_asize,
	.vdev_op_min_alloc = vdev_draid_min_alloc,
	.vdev_op_io_start = vdev_draid_io_start,
	.vdev_op_io_done = vdev_draid_io_done,
	.vdev_op_state_change = vdev_draid_state_change,
	.vdev_op_need_resilver = vdev_draid_need_resilver,
	.vdev_op_hold = NULL,
	.vdev_op_rele = NULL,
	.vdev_op_remap = NULL,
	.vdev_op_xlate = vdev_draid_xlate,
	.vdev_op_rebuild_asize = vdev_draid_rebuild_asize,
	.vdev_op_metaslab_init = vdev_draid_metaslab_init,
	.vdev_op_config_generate = vdev_draid_config_generate,
	.vdev_op_nparity = vdev_draid_nparity,
	.vdev_op_ndisks = vdev_draid_ndisks,
	.vdev_op_type = VDEV_TYPE_DRAID,
	.vdev_op_leaf = B_FALSE,
	};


	/*
	* A dRAID distributed spare is a virtual leaf vdev which is included in the
	* parent dRAID configuration. The last N columns of the dRAID permutation
	* table are used to determine on which dRAID children a specific offset
	* should be written. These spare leaf vdevs can only be used to replace
	* faulted children in the same dRAID configuration.
	*/

	/*
	* Distributed spare state. All fields are set when the distributed spare is
	* first opened and are immutable.
	*/
	typedef struct {
	vdev_t vds_draid_vdev; / top-level parent dRAID vdev */
	uint64_t vds_top_guid; /* top-level parent dRAID guid */
	uint64_t vds_spare_id; /* spare id (0 - vdc->vdc_nspares-1) */
	} vdev_draid_spare_t;

	/*
	* Returns the parent dRAID vdev to which the distributed spare belongs.
	* This may be safely called even when the vdev is not open.
	*/
	vdev_t *
	vdev_draid_spare_get_parent(vdev_t *vd)
	{
	vdev_draid_spare_t *vds = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_spare_ops);

	if (vds->vds_draid_vdev != NULL)
	return (vds->vds_draid_vdev);

	return (vdev_lookup_by_guid(vd->vdev_spa->spa_root_vdev,
	vds->vds_top_guid));
	}

	/*
	* A dRAID space is active when it's the child of a vdev using the
	* vdev_spare_ops, vdev_replacing_ops or vdev_draid_ops.
	*/
	static boolean_t
	vdev_draid_spare_is_active(vdev_t *vd)
	{
	vdev_t *pvd = vd->vdev_parent;

	if (pvd != NULL && (pvd->vdev_ops == &vdev_spare_ops \|\|
	pvd->vdev_ops == &vdev_replacing_ops \|\|
	pvd->vdev_ops == &vdev_draid_ops)) {
	return (B_TRUE);
	} else {
	return (B_FALSE);
	}
	}

	/*
	* Given a dRAID distribute spare vdev, returns the physical child vdev
	* on which the provided offset resides. This may involve recursing through
	* multiple layers of distributed spares. Note that offset is relative to
	* this vdev.
	*/
	vdev_t *
	vdev_draid_spare_get_child(vdev_t *vd, uint64_t physical_offset)
	{
	vdev_draid_spare_t *vds = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_spare_ops);

	/* The vdev is closed */
	if (vds->vds_draid_vdev == NULL)
	return (NULL);

	vdev_t *tvd = vds->vds_draid_vdev;
	vdev_draid_config_t *vdc = tvd->vdev_tsd;

	ASSERT3P(tvd->vdev_ops, ==, &vdev_draid_ops);
	ASSERT3U(vds->vds_spare_id, <, vdc->vdc_nspares);

	uint8_t *base;
	uint64_t iter;
	uint64_t perm = physical_offset / vdc->vdc_devslicesz;

	vdev_draid_get_perm(vdc, perm, &base, &iter);

	uint64_t cid = vdev_draid_permute_id(vdc, base, iter,
	(tvd->vdev_children - 1) - vds->vds_spare_id);
	vdev_t *cvd = tvd->vdev_child[cid];

	if (cvd->vdev_ops == &vdev_draid_spare_ops)
	return (vdev_draid_spare_get_child(cvd, physical_offset));

	return (cvd);
	}

	/* ARGSUSED */
	static void
	vdev_draid_spare_close(vdev_t *vd)
	{
	vdev_draid_spare_t *vds = vd->vdev_tsd;
	vds->vds_draid_vdev = NULL;
	}

	/*
	* Opening a dRAID spare device is done by looking up the associated dRAID
	* top-level vdev guid from the spare configuration.
	*/
	static int
	vdev_draid_spare_open(vdev_t vd, uint64_t psize, uint64_t *max_psize,
	uint64_t logical_ashift, uint64_t physical_ashift)
	{
	vdev_draid_spare_t *vds = vd->vdev_tsd;
	vdev_t *rvd = vd->vdev_spa->spa_root_vdev;
	uint64_t asize, max_asize;

	vdev_t *tvd = vdev_lookup_by_guid(rvd, vds->vds_top_guid);
	if (tvd == NULL) {
	/*
	* When spa_vdev_add() is labeling new spares the
	* associated dRAID is not attached to the root vdev
	* nor does this spare have a parent. Simulate a valid
	* device in order to allow the label to be initialized
	* and the distributed spare added to the configuration.
	*/
	if (vd->vdev_parent == NULL) {
	psize = max_psize = SPA_MINDEVSIZE;
	logical_ashift = physical_ashift = ASHIFT_MIN;
	return (0);
	}

	return (SET_ERROR(EINVAL));
	}

	vdev_draid_config_t *vdc = tvd->vdev_tsd;
	if (tvd->vdev_ops != &vdev_draid_ops \|\| vdc == NULL)
	return (SET_ERROR(EINVAL));

	if (vds->vds_spare_id >= vdc->vdc_nspares)
	return (SET_ERROR(EINVAL));

	/*
	* Neither tvd->vdev_asize or tvd->vdev_max_asize can be used here
	* because the caller may be vdev_draid_open() in which case the
	* values are stale as they haven't yet been updated by vdev_open().
	* To avoid this always recalculate the dRAID asize and max_asize.
	*/
	vdev_draid_calculate_asize(tvd, &asize, &max_asize,
	logical_ashift, physical_ashift);

	*psize = asize + VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE;
	*max_psize = max_asize + VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE;

	vds->vds_draid_vdev = tvd;

	return (0);
	}

	/*
	* Completed distributed spare IO. Store the result in the parent zio
	* as if it had performed the operation itself. Only the first error is
	* preserved if there are multiple errors.
	*/
	static void
	vdev_draid_spare_child_done(zio_t *zio)
	{
	zio_t *pio = zio->io_private;

	/*
	* IOs are issued to non-writable vdevs in order to keep their
	* DTLs accurate. However, we don't want to propagate the
	* error in to the distributed spare's DTL. When resilvering
	* vdev_draid_need_resilver() will consult the relevant DTL
	* to determine if the data is missing and must be repaired.
	*/
	if (!vdev_writeable(zio->io_vd))
	return;

	if (pio->io_error == 0)
	pio->io_error = zio->io_error;
	}

	/*
	* Returns a valid label nvlist for the distributed spare vdev. This is
	* used to bypass the IO pipeline to avoid the complexity of constructing
	* a complete label with valid checksum to return when read.
	*/
	nvlist_t *
	vdev_draid_read_config_spare(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;
	spa_aux_vdev_t *sav = &spa->spa_spares;
	uint64_t guid = vd->vdev_guid;

	nvlist_t *nv = fnvlist_alloc();
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_IS_SPARE, 1);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_CREATE_TXG, vd->vdev_crtxg);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_VERSION, spa_version(spa));
	fnvlist_add_string(nv, ZPOOL_CONFIG_POOL_NAME, spa_name(spa));
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_POOL_GUID, spa_guid(spa));
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_POOL_TXG, spa->spa_config_txg);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_TOP_GUID, vd->vdev_top->vdev_guid);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_POOL_STATE,
	vdev_draid_spare_is_active(vd) ?
	POOL_STATE_ACTIVE : POOL_STATE_SPARE);

	/* Set the vdev guid based on the vdev list in sav_count. */
	for (int i = 0; i < sav->sav_count; i++) {
	if (sav->sav_vdevs[i]->vdev_ops == &vdev_draid_spare_ops &&
	strcmp(sav->sav_vdevs[i]->vdev_path, vd->vdev_path) == 0) {
	guid = sav->sav_vdevs[i]->vdev_guid;
	break;
	}
	}

	fnvlist_add_uint64(nv, ZPOOL_CONFIG_GUID, guid);

	return (nv);
	}

	/*
	* Handle any ioctl requested of the distributed spare. Only flushes
	* are supported in which case all children must be flushed.
	*/
	static int
	vdev_draid_spare_ioctl(zio_t *zio)
	{
	vdev_t *vd = zio->io_vd;
	int error = 0;

	if (zio->io_cmd == DKIOCFLUSHWRITECACHE) {
	for (int c = 0; c < vd->vdev_children; c++) {
	zio_nowait(zio_vdev_child_io(zio, NULL,
	vd->vdev_child[c], zio->io_offset, zio->io_abd,
	zio->io_size, zio->io_type, zio->io_priority, 0,
	vdev_draid_spare_child_done, zio));
	}
	} else {
	error = SET_ERROR(ENOTSUP);
	}

	return (error);
	}

	/*
	* Initiate an IO to the distributed spare. For normal IOs this entails using
	* the zio->io_offset and permutation table to calculate which child dRAID vdev
	* is responsible for the data. Then passing along the zio to that child to
	* perform the actual IO. The label ranges are not stored on disk and require
	* some special handling which is described below.
	*/
	static void
	vdev_draid_spare_io_start(zio_t *zio)
	{
	vdev_t cvd = NULL, vd = zio->io_vd;
	vdev_draid_spare_t *vds = vd->vdev_tsd;
	uint64_t offset = zio->io_offset - VDEV_LABEL_START_SIZE;

	/*
	* If the vdev is closed, it's likely in the REMOVED or FAULTED state.
	* Nothing to be done here but return failure.
	*/
	if (vds == NULL) {
	zio->io_error = ENXIO;
	zio_interrupt(zio);
	return;
	}

	switch (zio->io_type) {
	case ZIO_TYPE_IOCTL:
	zio->io_error = vdev_draid_spare_ioctl(zio);
	break;

	case ZIO_TYPE_WRITE:
	if (VDEV_OFFSET_IS_LABEL(vd, zio->io_offset)) {
	/*
	* Accept probe IOs and config writers to simulate the
	* existence of an on disk label. vdev_label_sync(),
	* vdev_uberblock_sync() and vdev_copy_uberblocks()
	* skip the distributed spares. This only leaves
	* vdev_label_init() which is allowed to succeed to
	* avoid adding special cases the function.
	*/
	if (zio->io_flags & ZIO_FLAG_PROBE \|\|
	zio->io_flags & ZIO_FLAG_CONFIG_WRITER) {
	zio->io_error = 0;
	} else {
	zio->io_error = SET_ERROR(EIO);
	}
	} else {
	cvd = vdev_draid_spare_get_child(vd, offset);

	if (cvd == NULL) {
	zio->io_error = SET_ERROR(ENXIO);
	} else {
	zio_nowait(zio_vdev_child_io(zio, NULL, cvd,
	offset, zio->io_abd, zio->io_size,
	zio->io_type, zio->io_priority, 0,
	vdev_draid_spare_child_done, zio));
	}
	}
	break;

	case ZIO_TYPE_READ:
	if (VDEV_OFFSET_IS_LABEL(vd, zio->io_offset)) {
	/*
	* Accept probe IOs to simulate the existence of a
	* label. vdev_label_read_config() bypasses the
	* pipeline to read the label configuration and
	* vdev_uberblock_load() skips distributed spares
	* when attempting to locate the best uberblock.
	*/
	if (zio->io_flags & ZIO_FLAG_PROBE) {
	zio->io_error = 0;
	} else {
	zio->io_error = SET_ERROR(EIO);
	}
	} else {
	cvd = vdev_draid_spare_get_child(vd, offset);

	if (cvd == NULL \|\| !vdev_readable(cvd)) {
	zio->io_error = SET_ERROR(ENXIO);
	} else {
	zio_nowait(zio_vdev_child_io(zio, NULL, cvd,
	offset, zio->io_abd, zio->io_size,
	zio->io_type, zio->io_priority, 0,
	vdev_draid_spare_child_done, zio));
	}
	}
	break;

	case ZIO_TYPE_TRIM:
	/* The vdev label ranges are never trimmed */
	ASSERT0(VDEV_OFFSET_IS_LABEL(vd, zio->io_offset));

	cvd = vdev_draid_spare_get_child(vd, offset);

	if (cvd == NULL \|\| !cvd->vdev_has_trim) {
	zio->io_error = SET_ERROR(ENXIO);
	} else {
	zio_nowait(zio_vdev_child_io(zio, NULL, cvd,
	offset, zio->io_abd, zio->io_size,
	zio->io_type, zio->io_priority, 0,
	vdev_draid_spare_child_done, zio));
	}
	break;

	default:
	zio->io_error = SET_ERROR(ENOTSUP);
	break;
	}

	zio_execute(zio);
	}

	/* ARGSUSED */
	static void
	vdev_draid_spare_io_done(zio_t *zio)
	{
	}

	/*
	* Lookup the full spare config in spa->spa_spares.sav_config and
	* return the top_guid and spare_id for the named spare.
	*/
	static int
	vdev_draid_spare_lookup(spa_t spa, nvlist_t nv, uint64_t *top_guidp,
	uint64_t *spare_idp)
	{
	nvlist_t **spares;
	uint_t nspares;
	int error;

	if ((spa->spa_spares.sav_config == NULL) \|\|
	(nvlist_lookup_nvlist_array(spa->spa_spares.sav_config,
	ZPOOL_CONFIG_SPARES, &spares, &nspares) != 0)) {
	return (SET_ERROR(ENOENT));
	}

	char *spare_name;
	error = nvlist_lookup_string(nv, ZPOOL_CONFIG_PATH, &spare_name);
	if (error != 0)
	return (SET_ERROR(EINVAL));

	for (int i = 0; i < nspares; i++) {
	nvlist_t *spare = spares[i];
	uint64_t top_guid, spare_id;
	char type, path;

	/* Skip non-distributed spares */
	error = nvlist_lookup_string(spare, ZPOOL_CONFIG_TYPE, &type);
	if (error != 0 \|\| strcmp(type, VDEV_TYPE_DRAID_SPARE) != 0)
	continue;

	/* Skip spares with the wrong name */
	error = nvlist_lookup_string(spare, ZPOOL_CONFIG_PATH, &path);
	if (error != 0 \|\| strcmp(path, spare_name) != 0)
	continue;

	/* Found the matching spare */
	error = nvlist_lookup_uint64(spare,
	ZPOOL_CONFIG_TOP_GUID, &top_guid);
	if (error == 0) {
	error = nvlist_lookup_uint64(spare,
	ZPOOL_CONFIG_SPARE_ID, &spare_id);
	}

	if (error != 0) {
	return (SET_ERROR(EINVAL));
	} else {
	*top_guidp = top_guid;
	*spare_idp = spare_id;
	return (0);
	}
	}

	return (SET_ERROR(ENOENT));
	}

	/*
	* Initialize private dRAID spare specific fields from the nvlist.
	*/
	static int
	vdev_draid_spare_init(spa_t spa, nvlist_t nv, void **tsd)
	{
	vdev_draid_spare_t *vds;
	uint64_t top_guid = 0;
	uint64_t spare_id;

	/*
	* In the normal case check the list of spares stored in the spa
	* to lookup the top_guid and spare_id for provided spare config.
	* When creating a new pool or adding vdevs the spare list is not
	* yet populated and the values are provided in the passed config.
	*/
	if (vdev_draid_spare_lookup(spa, nv, &top_guid, &spare_id) != 0) {
	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_TOP_GUID,
	&top_guid) != 0)
	return (SET_ERROR(EINVAL));

	if (nvlist_lookup_uint64(nv, ZPOOL_CONFIG_SPARE_ID,
	&spare_id) != 0)
	return (SET_ERROR(EINVAL));
	}

	vds = kmem_alloc(sizeof (vdev_draid_spare_t), KM_SLEEP);
	vds->vds_draid_vdev = NULL;
	vds->vds_top_guid = top_guid;
	vds->vds_spare_id = spare_id;

	*tsd = vds;

	return (0);
	}

	static void
	vdev_draid_spare_fini(vdev_t *vd)
	{
	kmem_free(vd->vdev_tsd, sizeof (vdev_draid_spare_t));
	}

	static void
	vdev_draid_spare_config_generate(vdev_t vd, nvlist_t nv)
	{
	vdev_draid_spare_t *vds = vd->vdev_tsd;

	ASSERT3P(vd->vdev_ops, ==, &vdev_draid_spare_ops);

	fnvlist_add_uint64(nv, ZPOOL_CONFIG_TOP_GUID, vds->vds_top_guid);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_SPARE_ID, vds->vds_spare_id);
	}

	vdev_ops_t vdev_draid_spare_ops = {
	.vdev_op_init = vdev_draid_spare_init,
	.vdev_op_fini = vdev_draid_spare_fini,
	.vdev_op_open = vdev_draid_spare_open,
	.vdev_op_close = vdev_draid_spare_close,
	.vdev_op_asize = vdev_default_asize,
	.vdev_op_min_asize = vdev_default_min_asize,
	.vdev_op_min_alloc = NULL,
	.vdev_op_io_start = vdev_draid_spare_io_start,
	.vdev_op_io_done = vdev_draid_spare_io_done,
	.vdev_op_state_change = NULL,
	.vdev_op_need_resilver = NULL,
	.vdev_op_hold = NULL,
	.vdev_op_rele = NULL,
	.vdev_op_remap = NULL,
	.vdev_op_xlate = vdev_default_xlate,
	.vdev_op_rebuild_asize = NULL,
	.vdev_op_metaslab_init = NULL,
	.vdev_op_config_generate = vdev_draid_spare_config_generate,
	.vdev_op_nparity = NULL,
	.vdev_op_ndisks = NULL,
	.vdev_op_type = VDEV_TYPE_DRAID_SPARE,
	.vdev_op_leaf = B_TRUE,
	};
	diff --git a/sys/contrib/openzfs/module/zfs/vdev_label.c b/sys/contrib/openzfs/module/zfs/vdev_label.c
	index daf53f0a0c8b..a62804ea8f49 100644
	--- a/sys/contrib/openzfs/module/zfs/vdev_label.c
	+++ b/sys/contrib/openzfs/module/zfs/vdev_label.c
	@@ -1,2025 +1,2029 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2012, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2017, Intel Corporation.
	*/

	/*
	* Virtual Device Labels
	* ---------------------
	*
	* The vdev label serves several distinct purposes:
	*
	* 1. Uniquely identify this device as part of a ZFS pool and confirm its
	* identity within the pool.
	*
	* 2. Verify that all the devices given in a configuration are present
	* within the pool.
	*
	* 3. Determine the uberblock for the pool.
	*
	* 4. In case of an import operation, determine the configuration of the
	* toplevel vdev of which it is a part.
	*
	* 5. If an import operation cannot find all the devices in the pool,
	* provide enough information to the administrator to determine which
	* devices are missing.
	*
	* It is important to note that while the kernel is responsible for writing the
	* label, it only consumes the information in the first three cases. The
	* latter information is only consumed in userland when determining the
	* configuration to import a pool.
	*
	*
	* Label Organization
	* ------------------
	*
	* Before describing the contents of the label, it's important to understand how
	* the labels are written and updated with respect to the uberblock.
	*
	* When the pool configuration is altered, either because it was newly created
	* or a device was added, we want to update all the labels such that we can deal
	* with fatal failure at any point. To this end, each disk has two labels which
	* are updated before and after the uberblock is synced. Assuming we have
	* labels and an uberblock with the following transaction groups:
	*
	* L1 UB L2
	* +------+ +------+ +------+
	* \| \| \| \| \| \|
	* \| t10 \| \| t10 \| \| t10 \|
	* \| \| \| \| \| \|
	* +------+ +------+ +------+
	*
	* In this stable state, the labels and the uberblock were all updated within
	* the same transaction group (10). Each label is mirrored and checksummed, so
	* that we can detect when we fail partway through writing the label.
	*
	* In order to identify which labels are valid, the labels are written in the
	* following manner:
	*
	* 1. For each vdev, update 'L1' to the new label
	* 2. Update the uberblock
	* 3. For each vdev, update 'L2' to the new label
	*
	* Given arbitrary failure, we can determine the correct label to use based on
	* the transaction group. If we fail after updating L1 but before updating the
	* UB, we will notice that L1's transaction group is greater than the uberblock,
	* so L2 must be valid. If we fail after writing the uberblock but before
	* writing L2, we will notice that L2's transaction group is less than L1, and
	* therefore L1 is valid.
	*
	* Another added complexity is that not every label is updated when the config
	* is synced. If we add a single device, we do not want to have to re-write
	* every label for every device in the pool. This means that both L1 and L2 may
	* be older than the pool uberblock, because the necessary information is stored
	* on another vdev.
	*
	*
	* On-disk Format
	* --------------
	*
	* The vdev label consists of two distinct parts, and is wrapped within the
	* vdev_label_t structure. The label includes 8k of padding to permit legacy
	* VTOC disk labels, but is otherwise ignored.
	*
	* The first half of the label is a packed nvlist which contains pool wide
	* properties, per-vdev properties, and configuration information. It is
	* described in more detail below.
	*
	* The latter half of the label consists of a redundant array of uberblocks.
	* These uberblocks are updated whenever a transaction group is committed,
	* or when the configuration is updated. When a pool is loaded, we scan each
	* vdev for the 'best' uberblock.
	*
	*
	* Configuration Information
	* -------------------------
	*
	* The nvlist describing the pool and vdev contains the following elements:
	*
	* version ZFS on-disk version
	* name Pool name
	* state Pool state
	* txg Transaction group in which this label was written
	* pool_guid Unique identifier for this pool
	* vdev_tree An nvlist describing vdev tree.
	* features_for_read
	* An nvlist of the features necessary for reading the MOS.
	*
	* Each leaf device label also contains the following:
	*
	* top_guid Unique ID for top-level vdev in which this is contained
	* guid Unique ID for the leaf vdev
	*
	* The 'vs' configuration follows the format described in 'spa_config.c'.
	*/

	#include <sys/zfs_context.h>
	#include <sys/spa.h>
	#include <sys/spa_impl.h>
	#include <sys/dmu.h>
	#include <sys/zap.h>
	#include <sys/vdev.h>
	#include <sys/vdev_impl.h>
	#include <sys/vdev_draid.h>
	#include <sys/uberblock_impl.h>
	#include <sys/metaslab.h>
	#include <sys/metaslab_impl.h>
	#include <sys/zio.h>
	#include <sys/dsl_scan.h>
	#include <sys/abd.h>
	#include <sys/fs/zfs.h>
	#include <sys/byteorder.h>
	#include <sys/zfs_bootenv.h>

	/*
	* Basic routines to read and write from a vdev label.
	* Used throughout the rest of this file.
	*/
	uint64_t
	vdev_label_offset(uint64_t psize, int l, uint64_t offset)
	{
	ASSERT(offset < sizeof (vdev_label_t));
	ASSERT(P2PHASE_TYPED(psize, sizeof (vdev_label_t), uint64_t) == 0);

	return (offset + l * sizeof (vdev_label_t) + (l < VDEV_LABELS / 2 ?
	0 : psize - VDEV_LABELS * sizeof (vdev_label_t)));
	}

	/*
	* Returns back the vdev label associated with the passed in offset.
	*/
	int
	vdev_label_number(uint64_t psize, uint64_t offset)
	{
	int l;

	if (offset >= psize - VDEV_LABEL_END_SIZE) {
	offset -= psize - VDEV_LABEL_END_SIZE;
	offset += (VDEV_LABELS / 2) * sizeof (vdev_label_t);
	}
	l = offset / sizeof (vdev_label_t);
	return (l < VDEV_LABELS ? l : -1);
	}

	static void
	vdev_label_read(zio_t zio, vdev_t vd, int l, abd_t *buf, uint64_t offset,
	uint64_t size, zio_done_func_t done, void private, int flags)
	{
	ASSERT(
	spa_config_held(zio->io_spa, SCL_STATE, RW_READER) == SCL_STATE \|\|
	spa_config_held(zio->io_spa, SCL_STATE, RW_WRITER) == SCL_STATE);
	ASSERT(flags & ZIO_FLAG_CONFIG_WRITER);

	zio_nowait(zio_read_phys(zio, vd,
	vdev_label_offset(vd->vdev_psize, l, offset),
	size, buf, ZIO_CHECKSUM_LABEL, done, private,
	ZIO_PRIORITY_SYNC_READ, flags, B_TRUE));
	}

	void
	vdev_label_write(zio_t zio, vdev_t vd, int l, abd_t *buf, uint64_t offset,
	uint64_t size, zio_done_func_t done, void private, int flags)
	{
	ASSERT(
	spa_config_held(zio->io_spa, SCL_STATE, RW_READER) == SCL_STATE \|\|
	spa_config_held(zio->io_spa, SCL_STATE, RW_WRITER) == SCL_STATE);
	ASSERT(flags & ZIO_FLAG_CONFIG_WRITER);

	zio_nowait(zio_write_phys(zio, vd,
	vdev_label_offset(vd->vdev_psize, l, offset),
	size, buf, ZIO_CHECKSUM_LABEL, done, private,
	ZIO_PRIORITY_SYNC_WRITE, flags, B_TRUE));
	}

	/*
	* Generate the nvlist representing this vdev's stats
	*/
	void
	vdev_config_generate_stats(vdev_t vd, nvlist_t nv)
	{
	nvlist_t *nvx;
	vdev_stat_t *vs;
	vdev_stat_ex_t *vsx;

	vs = kmem_alloc(sizeof (*vs), KM_SLEEP);
	vsx = kmem_alloc(sizeof (*vsx), KM_SLEEP);

	vdev_get_stats_ex(vd, vs, vsx);
	fnvlist_add_uint64_array(nv, ZPOOL_CONFIG_VDEV_STATS,
	(uint64_t )vs, sizeof (vs) / sizeof (uint64_t));

	/*
	* Add extended stats into a special extended stats nvlist. This keeps
	* all the extended stats nicely grouped together. The extended stats
	* nvlist is then added to the main nvlist.
	*/
	nvx = fnvlist_alloc();

	/* ZIOs in flight to disk */
	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_SYNC_R_ACTIVE_QUEUE,
	vsx->vsx_active_queue[ZIO_PRIORITY_SYNC_READ]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_SYNC_W_ACTIVE_QUEUE,
	vsx->vsx_active_queue[ZIO_PRIORITY_SYNC_WRITE]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_ASYNC_R_ACTIVE_QUEUE,
	vsx->vsx_active_queue[ZIO_PRIORITY_ASYNC_READ]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_ASYNC_W_ACTIVE_QUEUE,
	vsx->vsx_active_queue[ZIO_PRIORITY_ASYNC_WRITE]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_SCRUB_ACTIVE_QUEUE,
	vsx->vsx_active_queue[ZIO_PRIORITY_SCRUB]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_TRIM_ACTIVE_QUEUE,
	vsx->vsx_active_queue[ZIO_PRIORITY_TRIM]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_REBUILD_ACTIVE_QUEUE,
	vsx->vsx_active_queue[ZIO_PRIORITY_REBUILD]);

	/* ZIOs pending */
	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_SYNC_R_PEND_QUEUE,
	vsx->vsx_pend_queue[ZIO_PRIORITY_SYNC_READ]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_SYNC_W_PEND_QUEUE,
	vsx->vsx_pend_queue[ZIO_PRIORITY_SYNC_WRITE]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_ASYNC_R_PEND_QUEUE,
	vsx->vsx_pend_queue[ZIO_PRIORITY_ASYNC_READ]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_ASYNC_W_PEND_QUEUE,
	vsx->vsx_pend_queue[ZIO_PRIORITY_ASYNC_WRITE]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_SCRUB_PEND_QUEUE,
	vsx->vsx_pend_queue[ZIO_PRIORITY_SCRUB]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_TRIM_PEND_QUEUE,
	vsx->vsx_pend_queue[ZIO_PRIORITY_TRIM]);

	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_REBUILD_PEND_QUEUE,
	vsx->vsx_pend_queue[ZIO_PRIORITY_REBUILD]);

	/* Histograms */
	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_TOT_R_LAT_HISTO,
	vsx->vsx_total_histo[ZIO_TYPE_READ],
	ARRAY_SIZE(vsx->vsx_total_histo[ZIO_TYPE_READ]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_TOT_W_LAT_HISTO,
	vsx->vsx_total_histo[ZIO_TYPE_WRITE],
	ARRAY_SIZE(vsx->vsx_total_histo[ZIO_TYPE_WRITE]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_DISK_R_LAT_HISTO,
	vsx->vsx_disk_histo[ZIO_TYPE_READ],
	ARRAY_SIZE(vsx->vsx_disk_histo[ZIO_TYPE_READ]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_DISK_W_LAT_HISTO,
	vsx->vsx_disk_histo[ZIO_TYPE_WRITE],
	ARRAY_SIZE(vsx->vsx_disk_histo[ZIO_TYPE_WRITE]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_SYNC_R_LAT_HISTO,
	vsx->vsx_queue_histo[ZIO_PRIORITY_SYNC_READ],
	ARRAY_SIZE(vsx->vsx_queue_histo[ZIO_PRIORITY_SYNC_READ]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_SYNC_W_LAT_HISTO,
	vsx->vsx_queue_histo[ZIO_PRIORITY_SYNC_WRITE],
	ARRAY_SIZE(vsx->vsx_queue_histo[ZIO_PRIORITY_SYNC_WRITE]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_ASYNC_R_LAT_HISTO,
	vsx->vsx_queue_histo[ZIO_PRIORITY_ASYNC_READ],
	ARRAY_SIZE(vsx->vsx_queue_histo[ZIO_PRIORITY_ASYNC_READ]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_ASYNC_W_LAT_HISTO,
	vsx->vsx_queue_histo[ZIO_PRIORITY_ASYNC_WRITE],
	ARRAY_SIZE(vsx->vsx_queue_histo[ZIO_PRIORITY_ASYNC_WRITE]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_SCRUB_LAT_HISTO,
	vsx->vsx_queue_histo[ZIO_PRIORITY_SCRUB],
	ARRAY_SIZE(vsx->vsx_queue_histo[ZIO_PRIORITY_SCRUB]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_TRIM_LAT_HISTO,
	vsx->vsx_queue_histo[ZIO_PRIORITY_TRIM],
	ARRAY_SIZE(vsx->vsx_queue_histo[ZIO_PRIORITY_TRIM]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_REBUILD_LAT_HISTO,
	vsx->vsx_queue_histo[ZIO_PRIORITY_REBUILD],
	ARRAY_SIZE(vsx->vsx_queue_histo[ZIO_PRIORITY_REBUILD]));

	/* Request sizes */
	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_SYNC_IND_R_HISTO,
	vsx->vsx_ind_histo[ZIO_PRIORITY_SYNC_READ],
	ARRAY_SIZE(vsx->vsx_ind_histo[ZIO_PRIORITY_SYNC_READ]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_SYNC_IND_W_HISTO,
	vsx->vsx_ind_histo[ZIO_PRIORITY_SYNC_WRITE],
	ARRAY_SIZE(vsx->vsx_ind_histo[ZIO_PRIORITY_SYNC_WRITE]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_ASYNC_IND_R_HISTO,
	vsx->vsx_ind_histo[ZIO_PRIORITY_ASYNC_READ],
	ARRAY_SIZE(vsx->vsx_ind_histo[ZIO_PRIORITY_ASYNC_READ]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_ASYNC_IND_W_HISTO,
	vsx->vsx_ind_histo[ZIO_PRIORITY_ASYNC_WRITE],
	ARRAY_SIZE(vsx->vsx_ind_histo[ZIO_PRIORITY_ASYNC_WRITE]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_IND_SCRUB_HISTO,
	vsx->vsx_ind_histo[ZIO_PRIORITY_SCRUB],
	ARRAY_SIZE(vsx->vsx_ind_histo[ZIO_PRIORITY_SCRUB]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_IND_TRIM_HISTO,
	vsx->vsx_ind_histo[ZIO_PRIORITY_TRIM],
	ARRAY_SIZE(vsx->vsx_ind_histo[ZIO_PRIORITY_TRIM]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_IND_REBUILD_HISTO,
	vsx->vsx_ind_histo[ZIO_PRIORITY_REBUILD],
	ARRAY_SIZE(vsx->vsx_ind_histo[ZIO_PRIORITY_REBUILD]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_SYNC_AGG_R_HISTO,
	vsx->vsx_agg_histo[ZIO_PRIORITY_SYNC_READ],
	ARRAY_SIZE(vsx->vsx_agg_histo[ZIO_PRIORITY_SYNC_READ]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_SYNC_AGG_W_HISTO,
	vsx->vsx_agg_histo[ZIO_PRIORITY_SYNC_WRITE],
	ARRAY_SIZE(vsx->vsx_agg_histo[ZIO_PRIORITY_SYNC_WRITE]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_ASYNC_AGG_R_HISTO,
	vsx->vsx_agg_histo[ZIO_PRIORITY_ASYNC_READ],
	ARRAY_SIZE(vsx->vsx_agg_histo[ZIO_PRIORITY_ASYNC_READ]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_ASYNC_AGG_W_HISTO,
	vsx->vsx_agg_histo[ZIO_PRIORITY_ASYNC_WRITE],
	ARRAY_SIZE(vsx->vsx_agg_histo[ZIO_PRIORITY_ASYNC_WRITE]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_AGG_SCRUB_HISTO,
	vsx->vsx_agg_histo[ZIO_PRIORITY_SCRUB],
	ARRAY_SIZE(vsx->vsx_agg_histo[ZIO_PRIORITY_SCRUB]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_AGG_TRIM_HISTO,
	vsx->vsx_agg_histo[ZIO_PRIORITY_TRIM],
	ARRAY_SIZE(vsx->vsx_agg_histo[ZIO_PRIORITY_TRIM]));

	fnvlist_add_uint64_array(nvx, ZPOOL_CONFIG_VDEV_AGG_REBUILD_HISTO,
	vsx->vsx_agg_histo[ZIO_PRIORITY_REBUILD],
	ARRAY_SIZE(vsx->vsx_agg_histo[ZIO_PRIORITY_REBUILD]));

	/* IO delays */
	fnvlist_add_uint64(nvx, ZPOOL_CONFIG_VDEV_SLOW_IOS, vs->vs_slow_ios);

	/* Add extended stats nvlist to main nvlist */
	fnvlist_add_nvlist(nv, ZPOOL_CONFIG_VDEV_STATS_EX, nvx);

	fnvlist_free(nvx);
	kmem_free(vs, sizeof (*vs));
	kmem_free(vsx, sizeof (*vsx));
	}

	static void
	root_vdev_actions_getprogress(vdev_t vd, nvlist_t nvl)
	{
	spa_t *spa = vd->vdev_spa;

	if (vd != spa->spa_root_vdev)
	return;

	/* provide either current or previous scan information */
	pool_scan_stat_t ps;
	if (spa_scan_get_stats(spa, &ps) == 0) {
	fnvlist_add_uint64_array(nvl,
	ZPOOL_CONFIG_SCAN_STATS, (uint64_t *)&ps,
	sizeof (pool_scan_stat_t) / sizeof (uint64_t));
	}

	pool_removal_stat_t prs;
	if (spa_removal_get_stats(spa, &prs) == 0) {
	fnvlist_add_uint64_array(nvl,
	ZPOOL_CONFIG_REMOVAL_STATS, (uint64_t *)&prs,
	sizeof (prs) / sizeof (uint64_t));
	}

	pool_checkpoint_stat_t pcs;
	if (spa_checkpoint_get_stats(spa, &pcs) == 0) {
	fnvlist_add_uint64_array(nvl,
	ZPOOL_CONFIG_CHECKPOINT_STATS, (uint64_t *)&pcs,
	sizeof (pcs) / sizeof (uint64_t));
	}
	}

	static void
	top_vdev_actions_getprogress(vdev_t vd, nvlist_t nvl)
	{
	if (vd == vd->vdev_top) {
	vdev_rebuild_stat_t vrs;
	if (vdev_rebuild_get_stats(vd, &vrs) == 0) {
	fnvlist_add_uint64_array(nvl,
	ZPOOL_CONFIG_REBUILD_STATS, (uint64_t *)&vrs,
	sizeof (vrs) / sizeof (uint64_t));
	}
	}
	}

	/*
	* Generate the nvlist representing this vdev's config.
	*/
	nvlist_t *
	vdev_config_generate(spa_t spa, vdev_t vd, boolean_t getstats,
	vdev_config_flag_t flags)
	{
	nvlist_t *nv = NULL;
	vdev_indirect_config_t *vic = &vd->vdev_indirect_config;

	nv = fnvlist_alloc();

	fnvlist_add_string(nv, ZPOOL_CONFIG_TYPE, vd->vdev_ops->vdev_op_type);
	if (!(flags & (VDEV_CONFIG_SPARE \| VDEV_CONFIG_L2CACHE)))
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_ID, vd->vdev_id);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_GUID, vd->vdev_guid);

	if (vd->vdev_path != NULL)
	fnvlist_add_string(nv, ZPOOL_CONFIG_PATH, vd->vdev_path);

	if (vd->vdev_devid != NULL)
	fnvlist_add_string(nv, ZPOOL_CONFIG_DEVID, vd->vdev_devid);

	if (vd->vdev_physpath != NULL)
	fnvlist_add_string(nv, ZPOOL_CONFIG_PHYS_PATH,
	vd->vdev_physpath);

	if (vd->vdev_enc_sysfs_path != NULL)
	fnvlist_add_string(nv, ZPOOL_CONFIG_VDEV_ENC_SYSFS_PATH,
	vd->vdev_enc_sysfs_path);

	if (vd->vdev_fru != NULL)
	fnvlist_add_string(nv, ZPOOL_CONFIG_FRU, vd->vdev_fru);

	if (vd->vdev_ops->vdev_op_config_generate != NULL)
	vd->vdev_ops->vdev_op_config_generate(vd, nv);

	if (vd->vdev_wholedisk != -1ULL) {
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_WHOLE_DISK,
	vd->vdev_wholedisk);
	}

	if (vd->vdev_not_present && !(flags & VDEV_CONFIG_MISSING))
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_NOT_PRESENT, 1);

	if (vd->vdev_isspare)
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_IS_SPARE, 1);

	if (!(flags & (VDEV_CONFIG_SPARE \| VDEV_CONFIG_L2CACHE)) &&
	vd == vd->vdev_top) {
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_METASLAB_ARRAY,
	vd->vdev_ms_array);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_METASLAB_SHIFT,
	vd->vdev_ms_shift);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_ASHIFT, vd->vdev_ashift);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_ASIZE,
	vd->vdev_asize);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_IS_LOG, vd->vdev_islog);
	+ if (vd->vdev_noalloc) {
	+ fnvlist_add_uint64(nv, ZPOOL_CONFIG_NONALLOCATING,
	+ vd->vdev_noalloc);
	+ }
	if (vd->vdev_removing) {
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_REMOVING,
	vd->vdev_removing);
	}

	/* zpool command expects alloc class data */
	if (getstats && vd->vdev_alloc_bias != VDEV_BIAS_NONE) {
	const char *bias = NULL;

	switch (vd->vdev_alloc_bias) {
	case VDEV_BIAS_LOG:
	bias = VDEV_ALLOC_BIAS_LOG;
	break;
	case VDEV_BIAS_SPECIAL:
	bias = VDEV_ALLOC_BIAS_SPECIAL;
	break;
	case VDEV_BIAS_DEDUP:
	bias = VDEV_ALLOC_BIAS_DEDUP;
	break;
	default:
	ASSERT3U(vd->vdev_alloc_bias, ==,
	VDEV_BIAS_NONE);
	}
	fnvlist_add_string(nv, ZPOOL_CONFIG_ALLOCATION_BIAS,
	bias);
	}
	}

	if (vd->vdev_dtl_sm != NULL) {
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_DTL,
	space_map_object(vd->vdev_dtl_sm));
	}

	if (vic->vic_mapping_object != 0) {
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_INDIRECT_OBJECT,
	vic->vic_mapping_object);
	}

	if (vic->vic_births_object != 0) {
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_INDIRECT_BIRTHS,
	vic->vic_births_object);
	}

	if (vic->vic_prev_indirect_vdev != UINT64_MAX) {
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_PREV_INDIRECT_VDEV,
	vic->vic_prev_indirect_vdev);
	}

	if (vd->vdev_crtxg)
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_CREATE_TXG, vd->vdev_crtxg);

	if (vd->vdev_expansion_time)
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_EXPANSION_TIME,
	vd->vdev_expansion_time);

	if (flags & VDEV_CONFIG_MOS) {
	if (vd->vdev_leaf_zap != 0) {
	ASSERT(vd->vdev_ops->vdev_op_leaf);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_VDEV_LEAF_ZAP,
	vd->vdev_leaf_zap);
	}

	if (vd->vdev_top_zap != 0) {
	ASSERT(vd == vd->vdev_top);
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_VDEV_TOP_ZAP,
	vd->vdev_top_zap);
	}

	if (vd->vdev_resilver_deferred) {
	ASSERT(vd->vdev_ops->vdev_op_leaf);
	ASSERT(spa->spa_resilver_deferred);
	fnvlist_add_boolean(nv, ZPOOL_CONFIG_RESILVER_DEFER);
	}
	}

	if (getstats) {
	vdev_config_generate_stats(vd, nv);

	root_vdev_actions_getprogress(vd, nv);
	top_vdev_actions_getprogress(vd, nv);

	/*
	* Note: this can be called from open context
	* (spa_get_stats()), so we need the rwlock to prevent
	* the mapping from being changed by condensing.
	*/
	rw_enter(&vd->vdev_indirect_rwlock, RW_READER);
	if (vd->vdev_indirect_mapping != NULL) {
	ASSERT(vd->vdev_indirect_births != NULL);
	vdev_indirect_mapping_t *vim =
	vd->vdev_indirect_mapping;
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_INDIRECT_SIZE,
	vdev_indirect_mapping_size(vim));
	}
	rw_exit(&vd->vdev_indirect_rwlock);
	if (vd->vdev_mg != NULL &&
	vd->vdev_mg->mg_fragmentation != ZFS_FRAG_INVALID) {
	/*
	* Compute approximately how much memory would be used
	* for the indirect mapping if this device were to
	* be removed.
	*
	* Note: If the frag metric is invalid, then not
	* enough metaslabs have been converted to have
	* histograms.
	*/
	uint64_t seg_count = 0;
	uint64_t to_alloc = vd->vdev_stat.vs_alloc;

	/*
	* There are the same number of allocated segments
	* as free segments, so we will have at least one
	* entry per free segment. However, small free
	* segments (smaller than vdev_removal_max_span)
	* will be combined with adjacent allocated segments
	* as a single mapping.
	*/
	for (int i = 0; i < RANGE_TREE_HISTOGRAM_SIZE; i++) {
	if (i + 1 < highbit64(vdev_removal_max_span)
	- 1) {
	to_alloc +=
	vd->vdev_mg->mg_histogram[i] <<
	(i + 1);
	} else {
	seg_count +=
	vd->vdev_mg->mg_histogram[i];
	}
	}

	/*
	* The maximum length of a mapping is
	* zfs_remove_max_segment, so we need at least one entry
	* per zfs_remove_max_segment of allocated data.
	*/
	seg_count += to_alloc / spa_remove_max_segment(spa);

	fnvlist_add_uint64(nv, ZPOOL_CONFIG_INDIRECT_SIZE,
	seg_count *
	sizeof (vdev_indirect_mapping_entry_phys_t));
	}
	}

	if (!vd->vdev_ops->vdev_op_leaf) {
	nvlist_t **child;
	int c, idx;

	ASSERT(!vd->vdev_ishole);

	child = kmem_alloc(vd->vdev_children * sizeof (nvlist_t *),
	KM_SLEEP);

	for (c = 0, idx = 0; c < vd->vdev_children; c++) {
	vdev_t *cvd = vd->vdev_child[c];

	/*
	* If we're generating an nvlist of removing
	* vdevs then skip over any device which is
	* not being removed.
	*/
	if ((flags & VDEV_CONFIG_REMOVING) &&
	!cvd->vdev_removing)
	continue;

	child[idx++] = vdev_config_generate(spa, cvd,
	getstats, flags);
	}

	if (idx) {
	fnvlist_add_nvlist_array(nv, ZPOOL_CONFIG_CHILDREN,
	- child, idx);
	+ (const nvlist_t * const *)child, idx);
	}

	for (c = 0; c < idx; c++)
	nvlist_free(child[c]);

	kmem_free(child, vd->vdev_children * sizeof (nvlist_t *));

	} else {
	const char *aux = NULL;

	if (vd->vdev_offline && !vd->vdev_tmpoffline)
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_OFFLINE, B_TRUE);
	if (vd->vdev_resilver_txg != 0)
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_RESILVER_TXG,
	vd->vdev_resilver_txg);
	if (vd->vdev_rebuild_txg != 0)
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_REBUILD_TXG,
	vd->vdev_rebuild_txg);
	if (vd->vdev_faulted)
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_FAULTED, B_TRUE);
	if (vd->vdev_degraded)
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_DEGRADED, B_TRUE);
	if (vd->vdev_removed)
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_REMOVED, B_TRUE);
	if (vd->vdev_unspare)
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_UNSPARE, B_TRUE);
	if (vd->vdev_ishole)
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_IS_HOLE, B_TRUE);

	/* Set the reason why we're FAULTED/DEGRADED. */
	switch (vd->vdev_stat.vs_aux) {
	case VDEV_AUX_ERR_EXCEEDED:
	aux = "err_exceeded";
	break;

	case VDEV_AUX_EXTERNAL:
	aux = "external";
	break;
	}

	if (aux != NULL && !vd->vdev_tmpoffline) {
	fnvlist_add_string(nv, ZPOOL_CONFIG_AUX_STATE, aux);
	} else {
	/*
	* We're healthy - clear any previous AUX_STATE values.
	*/
	if (nvlist_exists(nv, ZPOOL_CONFIG_AUX_STATE))
	nvlist_remove_all(nv, ZPOOL_CONFIG_AUX_STATE);
	}

	if (vd->vdev_splitting && vd->vdev_orig_guid != 0LL) {
	fnvlist_add_uint64(nv, ZPOOL_CONFIG_ORIG_GUID,
	vd->vdev_orig_guid);
	}
	}

	return (nv);
	}

	/*
	* Generate a view of the top-level vdevs. If we currently have holes
	* in the namespace, then generate an array which contains a list of holey
	* vdevs. Additionally, add the number of top-level children that currently
	* exist.
	*/
	void
	vdev_top_config_generate(spa_t spa, nvlist_t config)
	{
	vdev_t *rvd = spa->spa_root_vdev;
	uint64_t *array;
	uint_t c, idx;

	array = kmem_alloc(rvd->vdev_children * sizeof (uint64_t), KM_SLEEP);

	for (c = 0, idx = 0; c < rvd->vdev_children; c++) {
	vdev_t *tvd = rvd->vdev_child[c];

	if (tvd->vdev_ishole) {
	array[idx++] = c;
	}
	}

	if (idx) {
	VERIFY(nvlist_add_uint64_array(config, ZPOOL_CONFIG_HOLE_ARRAY,
	array, idx) == 0);
	}

	VERIFY(nvlist_add_uint64(config, ZPOOL_CONFIG_VDEV_CHILDREN,
	rvd->vdev_children) == 0);

	kmem_free(array, rvd->vdev_children * sizeof (uint64_t));
	}

	/*
	* Returns the configuration from the label of the given vdev. For vdevs
	* which don't have a txg value stored on their label (i.e. spares/cache)
	* or have not been completely initialized (txg = 0) just return
	* the configuration from the first valid label we find. Otherwise,
	* find the most up-to-date label that does not exceed the specified
	* 'txg' value.
	*/
	nvlist_t *
	vdev_label_read_config(vdev_t *vd, uint64_t txg)
	{
	spa_t *spa = vd->vdev_spa;
	nvlist_t *config = NULL;
	vdev_phys_t *vp[VDEV_LABELS];
	abd_t *vp_abd[VDEV_LABELS];
	zio_t *zio[VDEV_LABELS];
	uint64_t best_txg = 0;
	uint64_t label_txg = 0;
	int error = 0;
	int flags = ZIO_FLAG_CONFIG_WRITER \| ZIO_FLAG_CANFAIL \|
	ZIO_FLAG_SPECULATIVE;

	ASSERT(vd->vdev_validate_thread == curthread \|\|
	spa_config_held(spa, SCL_STATE_ALL, RW_WRITER) == SCL_STATE_ALL);

	if (!vdev_readable(vd))
	return (NULL);

	/*
	* The label for a dRAID distributed spare is not stored on disk.
	* Instead it is generated when needed which allows us to bypass
	* the pipeline when reading the config from the label.
	*/
	if (vd->vdev_ops == &vdev_draid_spare_ops)
	return (vdev_draid_read_config_spare(vd));

	for (int l = 0; l < VDEV_LABELS; l++) {
	vp_abd[l] = abd_alloc_linear(sizeof (vdev_phys_t), B_TRUE);
	vp[l] = abd_to_buf(vp_abd[l]);
	}

	retry:
	for (int l = 0; l < VDEV_LABELS; l++) {
	zio[l] = zio_root(spa, NULL, NULL, flags);

	vdev_label_read(zio[l], vd, l, vp_abd[l],
	offsetof(vdev_label_t, vl_vdev_phys), sizeof (vdev_phys_t),
	NULL, NULL, flags);
	}
	for (int l = 0; l < VDEV_LABELS; l++) {
	nvlist_t *label = NULL;

	if (zio_wait(zio[l]) == 0 &&
	nvlist_unpack(vp[l]->vp_nvlist, sizeof (vp[l]->vp_nvlist),
	&label, 0) == 0) {
	/*
	* Auxiliary vdevs won't have txg values in their
	* labels and newly added vdevs may not have been
	* completely initialized so just return the
	* configuration from the first valid label we
	* encounter.
	*/
	error = nvlist_lookup_uint64(label,
	ZPOOL_CONFIG_POOL_TXG, &label_txg);
	if ((error \|\| label_txg == 0) && !config) {
	config = label;
	for (l++; l < VDEV_LABELS; l++)
	zio_wait(zio[l]);
	break;
	} else if (label_txg <= txg && label_txg > best_txg) {
	best_txg = label_txg;
	nvlist_free(config);
	config = fnvlist_dup(label);
	}
	}

	if (label != NULL) {
	nvlist_free(label);
	label = NULL;
	}
	}

	if (config == NULL && !(flags & ZIO_FLAG_TRYHARD)) {
	flags \|= ZIO_FLAG_TRYHARD;
	goto retry;
	}

	/*
	* We found a valid label but it didn't pass txg restrictions.
	*/
	if (config == NULL && label_txg != 0) {
	vdev_dbgmsg(vd, "label discarded as txg is too large "
	"(%llu > %llu)", (u_longlong_t)label_txg,
	(u_longlong_t)txg);
	}

	for (int l = 0; l < VDEV_LABELS; l++) {
	abd_free(vp_abd[l]);
	}

	return (config);
	}

	/*
	* Determine if a device is in use. The 'spare_guid' parameter will be filled
	* in with the device guid if this spare is active elsewhere on the system.
	*/
	static boolean_t
	vdev_inuse(vdev_t *vd, uint64_t crtxg, vdev_labeltype_t reason,
	uint64_t spare_guid, uint64_t l2cache_guid)
	{
	spa_t *spa = vd->vdev_spa;
	uint64_t state, pool_guid, device_guid, txg, spare_pool;
	uint64_t vdtxg = 0;
	nvlist_t *label;

	if (spare_guid)
	*spare_guid = 0ULL;
	if (l2cache_guid)
	*l2cache_guid = 0ULL;

	/*
	* Read the label, if any, and perform some basic sanity checks.
	*/
	if ((label = vdev_label_read_config(vd, -1ULL)) == NULL)
	return (B_FALSE);

	(void) nvlist_lookup_uint64(label, ZPOOL_CONFIG_CREATE_TXG,
	&vdtxg);

	if (nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_STATE,
	&state) != 0 \|\|
	nvlist_lookup_uint64(label, ZPOOL_CONFIG_GUID,
	&device_guid) != 0) {
	nvlist_free(label);
	return (B_FALSE);
	}

	if (state != POOL_STATE_SPARE && state != POOL_STATE_L2CACHE &&
	(nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_GUID,
	&pool_guid) != 0 \|\|
	nvlist_lookup_uint64(label, ZPOOL_CONFIG_POOL_TXG,
	&txg) != 0)) {
	nvlist_free(label);
	return (B_FALSE);
	}

	nvlist_free(label);

	/*
	* Check to see if this device indeed belongs to the pool it claims to
	* be a part of. The only way this is allowed is if the device is a hot
	* spare (which we check for later on).
	*/
	if (state != POOL_STATE_SPARE && state != POOL_STATE_L2CACHE &&
	!spa_guid_exists(pool_guid, device_guid) &&
	!spa_spare_exists(device_guid, NULL, NULL) &&
	!spa_l2cache_exists(device_guid, NULL))
	return (B_FALSE);

	/*
	* If the transaction group is zero, then this an initialized (but
	* unused) label. This is only an error if the create transaction
	* on-disk is the same as the one we're using now, in which case the
	* user has attempted to add the same vdev multiple times in the same
	* transaction.
	*/
	if (state != POOL_STATE_SPARE && state != POOL_STATE_L2CACHE &&
	txg == 0 && vdtxg == crtxg)
	return (B_TRUE);

	/*
	* Check to see if this is a spare device. We do an explicit check for
	* spa_has_spare() here because it may be on our pending list of spares
	* to add.
	*/
	if (spa_spare_exists(device_guid, &spare_pool, NULL) \|\|
	spa_has_spare(spa, device_guid)) {
	if (spare_guid)
	*spare_guid = device_guid;

	switch (reason) {
	case VDEV_LABEL_CREATE:
	return (B_TRUE);

	case VDEV_LABEL_REPLACE:
	return (!spa_has_spare(spa, device_guid) \|\|
	spare_pool != 0ULL);

	case VDEV_LABEL_SPARE:
	return (spa_has_spare(spa, device_guid));
	default:
	break;
	}
	}

	/*
	* Check to see if this is an l2cache device.
	*/
	if (spa_l2cache_exists(device_guid, NULL) \|\|
	spa_has_l2cache(spa, device_guid)) {
	if (l2cache_guid)
	*l2cache_guid = device_guid;

	switch (reason) {
	case VDEV_LABEL_CREATE:
	return (B_TRUE);

	case VDEV_LABEL_REPLACE:
	return (!spa_has_l2cache(spa, device_guid));

	case VDEV_LABEL_L2CACHE:
	return (spa_has_l2cache(spa, device_guid));
	default:
	break;
	}
	}

	/*
	* We can't rely on a pool's state if it's been imported
	* read-only. Instead we look to see if the pools is marked
	* read-only in the namespace and set the state to active.
	*/
	if (state != POOL_STATE_SPARE && state != POOL_STATE_L2CACHE &&
	(spa = spa_by_guid(pool_guid, device_guid)) != NULL &&
	spa_mode(spa) == SPA_MODE_READ)
	state = POOL_STATE_ACTIVE;

	/*
	* If the device is marked ACTIVE, then this device is in use by another
	* pool on the system.
	*/
	return (state == POOL_STATE_ACTIVE);
	}

	/*
	* Initialize a vdev label. We check to make sure each leaf device is not in
	* use, and writable. We put down an initial label which we will later
	* overwrite with a complete label. Note that it's important to do this
	* sequentially, not in parallel, so that we catch cases of multiple use of the
	* same leaf vdev in the vdev we're creating -- e.g. mirroring a disk with
	* itself.
	*/
	int
	vdev_label_init(vdev_t *vd, uint64_t crtxg, vdev_labeltype_t reason)
	{
	spa_t *spa = vd->vdev_spa;
	nvlist_t *label;
	vdev_phys_t *vp;
	abd_t *vp_abd;
	abd_t *bootenv;
	uberblock_t *ub;
	abd_t *ub_abd;
	zio_t *zio;
	char *buf;
	size_t buflen;
	int error;
	uint64_t spare_guid = 0, l2cache_guid = 0;
	int flags = ZIO_FLAG_CONFIG_WRITER \| ZIO_FLAG_CANFAIL;

	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	for (int c = 0; c < vd->vdev_children; c++)
	if ((error = vdev_label_init(vd->vdev_child[c],
	crtxg, reason)) != 0)
	return (error);

	/* Track the creation time for this vdev */
	vd->vdev_crtxg = crtxg;

	if (!vd->vdev_ops->vdev_op_leaf \|\| !spa_writeable(spa))
	return (0);

	/*
	* Dead vdevs cannot be initialized.
	*/
	if (vdev_is_dead(vd))
	return (SET_ERROR(EIO));

	/*
	* Determine if the vdev is in use.
	*/
	if (reason != VDEV_LABEL_REMOVE && reason != VDEV_LABEL_SPLIT &&
	vdev_inuse(vd, crtxg, reason, &spare_guid, &l2cache_guid))
	return (SET_ERROR(EBUSY));

	/*
	* If this is a request to add or replace a spare or l2cache device
	* that is in use elsewhere on the system, then we must update the
	* guid (which was initialized to a random value) to reflect the
	* actual GUID (which is shared between multiple pools).
	*/
	if (reason != VDEV_LABEL_REMOVE && reason != VDEV_LABEL_L2CACHE &&
	spare_guid != 0ULL) {
	uint64_t guid_delta = spare_guid - vd->vdev_guid;

	vd->vdev_guid += guid_delta;

	for (vdev_t *pvd = vd; pvd != NULL; pvd = pvd->vdev_parent)
	pvd->vdev_guid_sum += guid_delta;

	/*
	* If this is a replacement, then we want to fallthrough to the
	* rest of the code. If we're adding a spare, then it's already
	* labeled appropriately and we can just return.
	*/
	if (reason == VDEV_LABEL_SPARE)
	return (0);
	ASSERT(reason == VDEV_LABEL_REPLACE \|\|
	reason == VDEV_LABEL_SPLIT);
	}

	if (reason != VDEV_LABEL_REMOVE && reason != VDEV_LABEL_SPARE &&
	l2cache_guid != 0ULL) {
	uint64_t guid_delta = l2cache_guid - vd->vdev_guid;

	vd->vdev_guid += guid_delta;

	for (vdev_t *pvd = vd; pvd != NULL; pvd = pvd->vdev_parent)
	pvd->vdev_guid_sum += guid_delta;

	/*
	* If this is a replacement, then we want to fallthrough to the
	* rest of the code. If we're adding an l2cache, then it's
	* already labeled appropriately and we can just return.
	*/
	if (reason == VDEV_LABEL_L2CACHE)
	return (0);
	ASSERT(reason == VDEV_LABEL_REPLACE);
	}

	/*
	* Initialize its label.
	*/
	vp_abd = abd_alloc_linear(sizeof (vdev_phys_t), B_TRUE);
	abd_zero(vp_abd, sizeof (vdev_phys_t));
	vp = abd_to_buf(vp_abd);

	/*
	* Generate a label describing the pool and our top-level vdev.
	* We mark it as being from txg 0 to indicate that it's not
	* really part of an active pool just yet. The labels will
	* be written again with a meaningful txg by spa_sync().
	*/
	if (reason == VDEV_LABEL_SPARE \|\|
	(reason == VDEV_LABEL_REMOVE && vd->vdev_isspare)) {
	/*
	* For inactive hot spares, we generate a special label that
	* identifies as a mutually shared hot spare. We write the
	* label if we are adding a hot spare, or if we are removing an
	* active hot spare (in which case we want to revert the
	* labels).
	*/
	VERIFY(nvlist_alloc(&label, NV_UNIQUE_NAME, KM_SLEEP) == 0);

	VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_VERSION,
	spa_version(spa)) == 0);
	VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_POOL_STATE,
	POOL_STATE_SPARE) == 0);
	VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_GUID,
	vd->vdev_guid) == 0);
	} else if (reason == VDEV_LABEL_L2CACHE \|\|
	(reason == VDEV_LABEL_REMOVE && vd->vdev_isl2cache)) {
	/*
	* For level 2 ARC devices, add a special label.
	*/
	VERIFY(nvlist_alloc(&label, NV_UNIQUE_NAME, KM_SLEEP) == 0);

	VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_VERSION,
	spa_version(spa)) == 0);
	VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_POOL_STATE,
	POOL_STATE_L2CACHE) == 0);
	VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_GUID,
	vd->vdev_guid) == 0);
	} else {
	uint64_t txg = 0ULL;

	if (reason == VDEV_LABEL_SPLIT)
	txg = spa->spa_uberblock.ub_txg;
	label = spa_config_generate(spa, vd, txg, B_FALSE);

	/*
	* Add our creation time. This allows us to detect multiple
	* vdev uses as described above, and automatically expires if we
	* fail.
	*/
	VERIFY(nvlist_add_uint64(label, ZPOOL_CONFIG_CREATE_TXG,
	crtxg) == 0);
	}

	buf = vp->vp_nvlist;
	buflen = sizeof (vp->vp_nvlist);

	error = nvlist_pack(label, &buf, &buflen, NV_ENCODE_XDR, KM_SLEEP);
	if (error != 0) {
	nvlist_free(label);
	abd_free(vp_abd);
	/* EFAULT means nvlist_pack ran out of room */
	return (SET_ERROR(error == EFAULT ? ENAMETOOLONG : EINVAL));
	}

	/*
	* Initialize uberblock template.
	*/
	ub_abd = abd_alloc_linear(VDEV_UBERBLOCK_RING, B_TRUE);
	abd_zero(ub_abd, VDEV_UBERBLOCK_RING);
	abd_copy_from_buf(ub_abd, &spa->spa_uberblock, sizeof (uberblock_t));
	ub = abd_to_buf(ub_abd);
	ub->ub_txg = 0;

	/* Initialize the 2nd padding area. */
	bootenv = abd_alloc_for_io(VDEV_PAD_SIZE, B_TRUE);
	abd_zero(bootenv, VDEV_PAD_SIZE);

	/*
	* Write everything in parallel.
	*/
	retry:
	zio = zio_root(spa, NULL, NULL, flags);

	for (int l = 0; l < VDEV_LABELS; l++) {

	vdev_label_write(zio, vd, l, vp_abd,
	offsetof(vdev_label_t, vl_vdev_phys),
	sizeof (vdev_phys_t), NULL, NULL, flags);

	/*
	* Skip the 1st padding area.
	* Zero out the 2nd padding area where it might have
	* left over data from previous filesystem format.
	*/
	vdev_label_write(zio, vd, l, bootenv,
	offsetof(vdev_label_t, vl_be),
	VDEV_PAD_SIZE, NULL, NULL, flags);

	vdev_label_write(zio, vd, l, ub_abd,
	offsetof(vdev_label_t, vl_uberblock),
	VDEV_UBERBLOCK_RING, NULL, NULL, flags);
	}

	error = zio_wait(zio);

	if (error != 0 && !(flags & ZIO_FLAG_TRYHARD)) {
	flags \|= ZIO_FLAG_TRYHARD;
	goto retry;
	}

	nvlist_free(label);
	abd_free(bootenv);
	abd_free(ub_abd);
	abd_free(vp_abd);

	/*
	* If this vdev hasn't been previously identified as a spare, then we
	* mark it as such only if a) we are labeling it as a spare, or b) it
	* exists as a spare elsewhere in the system. Do the same for
	* level 2 ARC devices.
	*/
	if (error == 0 && !vd->vdev_isspare &&
	(reason == VDEV_LABEL_SPARE \|\|
	spa_spare_exists(vd->vdev_guid, NULL, NULL)))
	spa_spare_add(vd);

	if (error == 0 && !vd->vdev_isl2cache &&
	(reason == VDEV_LABEL_L2CACHE \|\|
	spa_l2cache_exists(vd->vdev_guid, NULL)))
	spa_l2cache_add(vd);

	return (error);
	}

	/*
	* Done callback for vdev_label_read_bootenv_impl. If this is the first
	* callback to finish, store our abd in the callback pointer. Otherwise, we
	* just free our abd and return.
	*/
	static void
	vdev_label_read_bootenv_done(zio_t *zio)
	{
	zio_t *rio = zio->io_private;
	abd_t **cbp = rio->io_private;

	ASSERT3U(zio->io_size, ==, VDEV_PAD_SIZE);

	if (zio->io_error == 0) {
	mutex_enter(&rio->io_lock);
	if (*cbp == NULL) {
	/* Will free this buffer in vdev_label_read_bootenv. */
	*cbp = zio->io_abd;
	} else {
	abd_free(zio->io_abd);
	}
	mutex_exit(&rio->io_lock);
	} else {
	abd_free(zio->io_abd);
	}
	}

	static void
	vdev_label_read_bootenv_impl(zio_t zio, vdev_t vd, int flags)
	{
	for (int c = 0; c < vd->vdev_children; c++)
	vdev_label_read_bootenv_impl(zio, vd->vdev_child[c], flags);

	/*
	* We just use the first label that has a correct checksum; the
	* bootloader should have rewritten them all to be the same on boot,
	* and any changes we made since boot have been the same across all
	* labels.
	*/
	if (vd->vdev_ops->vdev_op_leaf && vdev_readable(vd)) {
	for (int l = 0; l < VDEV_LABELS; l++) {
	vdev_label_read(zio, vd, l,
	abd_alloc_linear(VDEV_PAD_SIZE, B_FALSE),
	offsetof(vdev_label_t, vl_be), VDEV_PAD_SIZE,
	vdev_label_read_bootenv_done, zio, flags);
	}
	}
	}

	int
	vdev_label_read_bootenv(vdev_t rvd, nvlist_t bootenv)
	{
	nvlist_t *config;
	spa_t *spa = rvd->vdev_spa;
	abd_t *abd = NULL;
	int flags = ZIO_FLAG_CONFIG_WRITER \| ZIO_FLAG_CANFAIL \|
	ZIO_FLAG_SPECULATIVE \| ZIO_FLAG_TRYHARD;

	ASSERT(bootenv);
	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	zio_t *zio = zio_root(spa, NULL, &abd, flags);
	vdev_label_read_bootenv_impl(zio, rvd, flags);
	int err = zio_wait(zio);

	if (abd != NULL) {
	char *buf;
	vdev_boot_envblock_t *vbe = abd_to_buf(abd);

	vbe->vbe_version = ntohll(vbe->vbe_version);
	switch (vbe->vbe_version) {
	case VB_RAW:
	/*
	* if we have textual data in vbe_bootenv, create nvlist
	* with key "envmap".
	*/
	fnvlist_add_uint64(bootenv, BOOTENV_VERSION, VB_RAW);
	vbe->vbe_bootenv[sizeof (vbe->vbe_bootenv) - 1] = '\0';
	fnvlist_add_string(bootenv, GRUB_ENVMAP,
	vbe->vbe_bootenv);
	break;

	case VB_NVLIST:
	err = nvlist_unpack(vbe->vbe_bootenv,
	sizeof (vbe->vbe_bootenv), &config, 0);
	if (err == 0) {
	fnvlist_merge(bootenv, config);
	nvlist_free(config);
	break;
	}
	fallthrough;
	default:
	/* Check for FreeBSD zfs bootonce command string */
	buf = abd_to_buf(abd);
	if (*buf == '\0') {
	fnvlist_add_uint64(bootenv, BOOTENV_VERSION,
	VB_NVLIST);
	break;
	}
	fnvlist_add_string(bootenv, FREEBSD_BOOTONCE, buf);
	}

	/*
	* abd was allocated in vdev_label_read_bootenv_impl()
	*/
	abd_free(abd);
	/*
	* If we managed to read any successfully,
	* return success.
	*/
	return (0);
	}
	return (err);
	}

	int
	vdev_label_write_bootenv(vdev_t vd, nvlist_t env)
	{
	zio_t *zio;
	spa_t *spa = vd->vdev_spa;
	vdev_boot_envblock_t *bootenv;
	int flags = ZIO_FLAG_CONFIG_WRITER \| ZIO_FLAG_CANFAIL;
	int error;
	size_t nvsize;
	char *nvbuf;

	error = nvlist_size(env, &nvsize, NV_ENCODE_XDR);
	if (error != 0)
	return (SET_ERROR(error));

	if (nvsize >= sizeof (bootenv->vbe_bootenv)) {
	return (SET_ERROR(E2BIG));
	}

	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	error = ENXIO;
	for (int c = 0; c < vd->vdev_children; c++) {
	int child_err;

	child_err = vdev_label_write_bootenv(vd->vdev_child[c], env);
	/*
	* As long as any of the disks managed to write all of their
	* labels successfully, return success.
	*/
	if (child_err == 0)
	error = child_err;
	}

	if (!vd->vdev_ops->vdev_op_leaf \|\| vdev_is_dead(vd) \|\|
	!vdev_writeable(vd)) {
	return (error);
	}
	ASSERT3U(sizeof (*bootenv), ==, VDEV_PAD_SIZE);
	abd_t *abd = abd_alloc_for_io(VDEV_PAD_SIZE, B_TRUE);
	abd_zero(abd, VDEV_PAD_SIZE);

	bootenv = abd_borrow_buf_copy(abd, VDEV_PAD_SIZE);
	nvbuf = bootenv->vbe_bootenv;
	nvsize = sizeof (bootenv->vbe_bootenv);

	bootenv->vbe_version = fnvlist_lookup_uint64(env, BOOTENV_VERSION);
	switch (bootenv->vbe_version) {
	case VB_RAW:
	if (nvlist_lookup_string(env, GRUB_ENVMAP, &nvbuf) == 0) {
	(void) strlcpy(bootenv->vbe_bootenv, nvbuf, nvsize);
	}
	error = 0;
	break;

	case VB_NVLIST:
	error = nvlist_pack(env, &nvbuf, &nvsize, NV_ENCODE_XDR,
	KM_SLEEP);
	break;

	default:
	error = EINVAL;
	break;
	}

	if (error == 0) {
	bootenv->vbe_version = htonll(bootenv->vbe_version);
	abd_return_buf_copy(abd, bootenv, VDEV_PAD_SIZE);
	} else {
	abd_free(abd);
	return (SET_ERROR(error));
	}

	retry:
	zio = zio_root(spa, NULL, NULL, flags);
	for (int l = 0; l < VDEV_LABELS; l++) {
	vdev_label_write(zio, vd, l, abd,
	offsetof(vdev_label_t, vl_be),
	VDEV_PAD_SIZE, NULL, NULL, flags);
	}

	error = zio_wait(zio);
	if (error != 0 && !(flags & ZIO_FLAG_TRYHARD)) {
	flags \|= ZIO_FLAG_TRYHARD;
	goto retry;
	}

	abd_free(abd);
	return (error);
	}

	/*
	* ==========================================================================
	* uberblock load/sync
	* ==========================================================================
	*/

	/*
	* Consider the following situation: txg is safely synced to disk. We've
	* written the first uberblock for txg + 1, and then we lose power. When we
	* come back up, we fail to see the uberblock for txg + 1 because, say,
	* it was on a mirrored device and the replica to which we wrote txg + 1
	* is now offline. If we then make some changes and sync txg + 1, and then
	* the missing replica comes back, then for a few seconds we'll have two
	* conflicting uberblocks on disk with the same txg. The solution is simple:
	* among uberblocks with equal txg, choose the one with the latest timestamp.
	*/
	static int
	vdev_uberblock_compare(const uberblock_t ub1, const uberblock_t ub2)
	{
	int cmp = TREE_CMP(ub1->ub_txg, ub2->ub_txg);

	if (likely(cmp))
	return (cmp);

	cmp = TREE_CMP(ub1->ub_timestamp, ub2->ub_timestamp);
	if (likely(cmp))
	return (cmp);

	/*
	* If MMP_VALID(ub) && MMP_SEQ_VALID(ub) then the host has an MMP-aware
	* ZFS, e.g. OpenZFS >= 0.7.
	*
	* If one ub has MMP and the other does not, they were written by
	* different hosts, which matters for MMP. So we treat no MMP/no SEQ as
	* a 0 value.
	*
	* Since timestamp and txg are the same if we get this far, either is
	* acceptable for importing the pool.
	*/
	unsigned int seq1 = 0;
	unsigned int seq2 = 0;

	if (MMP_VALID(ub1) && MMP_SEQ_VALID(ub1))
	seq1 = MMP_SEQ(ub1);

	if (MMP_VALID(ub2) && MMP_SEQ_VALID(ub2))
	seq2 = MMP_SEQ(ub2);

	return (TREE_CMP(seq1, seq2));
	}

	struct ubl_cbdata {
	uberblock_t ubl_ubbest; / Best uberblock */
	vdev_t ubl_vd; / vdev associated with the above */
	};

	static void
	vdev_uberblock_load_done(zio_t *zio)
	{
	vdev_t *vd = zio->io_vd;
	spa_t *spa = zio->io_spa;
	zio_t *rio = zio->io_private;
	uberblock_t *ub = abd_to_buf(zio->io_abd);
	struct ubl_cbdata *cbp = rio->io_private;

	ASSERT3U(zio->io_size, ==, VDEV_UBERBLOCK_SIZE(vd));

	if (zio->io_error == 0 && uberblock_verify(ub) == 0) {
	mutex_enter(&rio->io_lock);
	if (ub->ub_txg <= spa->spa_load_max_txg &&
	vdev_uberblock_compare(ub, cbp->ubl_ubbest) > 0) {
	/*
	* Keep track of the vdev in which this uberblock
	* was found. We will use this information later
	* to obtain the config nvlist associated with
	* this uberblock.
	*/
	cbp->ubl_ubbest = ub;
	cbp->ubl_vd = vd;
	}
	mutex_exit(&rio->io_lock);
	}

	abd_free(zio->io_abd);
	}

	static void
	vdev_uberblock_load_impl(zio_t zio, vdev_t vd, int flags,
	struct ubl_cbdata *cbp)
	{
	for (int c = 0; c < vd->vdev_children; c++)
	vdev_uberblock_load_impl(zio, vd->vdev_child[c], flags, cbp);

	if (vd->vdev_ops->vdev_op_leaf && vdev_readable(vd) &&
	vd->vdev_ops != &vdev_draid_spare_ops) {
	for (int l = 0; l < VDEV_LABELS; l++) {
	for (int n = 0; n < VDEV_UBERBLOCK_COUNT(vd); n++) {
	vdev_label_read(zio, vd, l,
	abd_alloc_linear(VDEV_UBERBLOCK_SIZE(vd),
	B_TRUE), VDEV_UBERBLOCK_OFFSET(vd, n),
	VDEV_UBERBLOCK_SIZE(vd),
	vdev_uberblock_load_done, zio, flags);
	}
	}
	}
	}

	/*
	* Reads the 'best' uberblock from disk along with its associated
	* configuration. First, we read the uberblock array of each label of each
	* vdev, keeping track of the uberblock with the highest txg in each array.
	* Then, we read the configuration from the same vdev as the best uberblock.
	*/
	void
	vdev_uberblock_load(vdev_t rvd, uberblock_t ub, nvlist_t **config)
	{
	zio_t *zio;
	spa_t *spa = rvd->vdev_spa;
	struct ubl_cbdata cb;
	int flags = ZIO_FLAG_CONFIG_WRITER \| ZIO_FLAG_CANFAIL \|
	ZIO_FLAG_SPECULATIVE \| ZIO_FLAG_TRYHARD;

	ASSERT(ub);
	ASSERT(config);

	bzero(ub, sizeof (uberblock_t));
	*config = NULL;

	cb.ubl_ubbest = ub;
	cb.ubl_vd = NULL;

	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
	zio = zio_root(spa, NULL, &cb, flags);
	vdev_uberblock_load_impl(zio, rvd, flags, &cb);
	(void) zio_wait(zio);

	/*
	* It's possible that the best uberblock was discovered on a label
	* that has a configuration which was written in a future txg.
	* Search all labels on this vdev to find the configuration that
	* matches the txg for our uberblock.
	*/
	if (cb.ubl_vd != NULL) {
	vdev_dbgmsg(cb.ubl_vd, "best uberblock found for spa %s. "
	"txg %llu", spa->spa_name, (u_longlong_t)ub->ub_txg);

	*config = vdev_label_read_config(cb.ubl_vd, ub->ub_txg);
	if (*config == NULL && spa->spa_extreme_rewind) {
	vdev_dbgmsg(cb.ubl_vd, "failed to read label config. "
	"Trying again without txg restrictions.");
	*config = vdev_label_read_config(cb.ubl_vd, UINT64_MAX);
	}
	if (*config == NULL) {
	vdev_dbgmsg(cb.ubl_vd, "failed to read label config");
	}
	}
	spa_config_exit(spa, SCL_ALL, FTAG);
	}

	/*
	* For use when a leaf vdev is expanded.
	* The location of labels 2 and 3 changed, and at the new location the
	* uberblock rings are either empty or contain garbage. The sync will write
	* new configs there because the vdev is dirty, but expansion also needs the
	* uberblock rings copied. Read them from label 0 which did not move.
	*
	* Since the point is to populate labels {2,3} with valid uberblocks,
	* we zero uberblocks we fail to read or which are not valid.
	*/

	static void
	vdev_copy_uberblocks(vdev_t *vd)
	{
	abd_t *ub_abd;
	zio_t *write_zio;
	int locks = (SCL_L2ARC \| SCL_ZIO);
	int flags = ZIO_FLAG_CONFIG_WRITER \| ZIO_FLAG_CANFAIL \|
	ZIO_FLAG_SPECULATIVE;

	ASSERT(spa_config_held(vd->vdev_spa, SCL_STATE, RW_READER) ==
	SCL_STATE);
	ASSERT(vd->vdev_ops->vdev_op_leaf);

	/*
	* No uberblocks are stored on distributed spares, they may be
	* safely skipped when expanding a leaf vdev.
	*/
	if (vd->vdev_ops == &vdev_draid_spare_ops)
	return;

	spa_config_enter(vd->vdev_spa, locks, FTAG, RW_READER);

	ub_abd = abd_alloc_linear(VDEV_UBERBLOCK_SIZE(vd), B_TRUE);

	write_zio = zio_root(vd->vdev_spa, NULL, NULL, flags);
	for (int n = 0; n < VDEV_UBERBLOCK_COUNT(vd); n++) {
	const int src_label = 0;
	zio_t *zio;

	zio = zio_root(vd->vdev_spa, NULL, NULL, flags);
	vdev_label_read(zio, vd, src_label, ub_abd,
	VDEV_UBERBLOCK_OFFSET(vd, n), VDEV_UBERBLOCK_SIZE(vd),
	NULL, NULL, flags);

	if (zio_wait(zio) \|\| uberblock_verify(abd_to_buf(ub_abd)))
	abd_zero(ub_abd, VDEV_UBERBLOCK_SIZE(vd));

	for (int l = 2; l < VDEV_LABELS; l++)
	vdev_label_write(write_zio, vd, l, ub_abd,
	VDEV_UBERBLOCK_OFFSET(vd, n),
	VDEV_UBERBLOCK_SIZE(vd), NULL, NULL,
	flags \| ZIO_FLAG_DONT_PROPAGATE);
	}
	(void) zio_wait(write_zio);

	spa_config_exit(vd->vdev_spa, locks, FTAG);

	abd_free(ub_abd);
	}

	/*
	* On success, increment root zio's count of good writes.
	* We only get credit for writes to known-visible vdevs; see spa_vdev_add().
	*/
	static void
	vdev_uberblock_sync_done(zio_t *zio)
	{
	uint64_t *good_writes = zio->io_private;

	if (zio->io_error == 0 && zio->io_vd->vdev_top->vdev_ms_array != 0)
	atomic_inc_64(good_writes);
	}

	/*
	* Write the uberblock to all labels of all leaves of the specified vdev.
	*/
	static void
	vdev_uberblock_sync(zio_t zio, uint64_t good_writes,
	uberblock_t ub, vdev_t vd, int flags)
	{
	for (uint64_t c = 0; c < vd->vdev_children; c++) {
	vdev_uberblock_sync(zio, good_writes,
	ub, vd->vdev_child[c], flags);
	}

	if (!vd->vdev_ops->vdev_op_leaf)
	return;

	if (!vdev_writeable(vd))
	return;

	/*
	* There's no need to write uberblocks to a distributed spare, they
	* are already stored on all the leaves of the parent dRAID. For
	* this same reason vdev_uberblock_load_impl() skips distributed
	* spares when reading uberblocks.
	*/
	if (vd->vdev_ops == &vdev_draid_spare_ops)
	return;

	/* If the vdev was expanded, need to copy uberblock rings. */
	if (vd->vdev_state == VDEV_STATE_HEALTHY &&
	vd->vdev_copy_uberblocks == B_TRUE) {
	vdev_copy_uberblocks(vd);
	vd->vdev_copy_uberblocks = B_FALSE;
	}

	int m = spa_multihost(vd->vdev_spa) ? MMP_BLOCKS_PER_LABEL : 0;
	int n = ub->ub_txg % (VDEV_UBERBLOCK_COUNT(vd) - m);

	/* Copy the uberblock_t into the ABD */
	abd_t *ub_abd = abd_alloc_for_io(VDEV_UBERBLOCK_SIZE(vd), B_TRUE);
	abd_zero(ub_abd, VDEV_UBERBLOCK_SIZE(vd));
	abd_copy_from_buf(ub_abd, ub, sizeof (uberblock_t));

	for (int l = 0; l < VDEV_LABELS; l++)
	vdev_label_write(zio, vd, l, ub_abd,
	VDEV_UBERBLOCK_OFFSET(vd, n), VDEV_UBERBLOCK_SIZE(vd),
	vdev_uberblock_sync_done, good_writes,
	flags \| ZIO_FLAG_DONT_PROPAGATE);

	abd_free(ub_abd);
	}

	/* Sync the uberblocks to all vdevs in svd[] */
	static int
	vdev_uberblock_sync_list(vdev_t *svd, int svdcount, uberblock_t ub, int flags)
	{
	spa_t *spa = svd[0]->vdev_spa;
	zio_t *zio;
	uint64_t good_writes = 0;

	zio = zio_root(spa, NULL, NULL, flags);

	for (int v = 0; v < svdcount; v++)
	vdev_uberblock_sync(zio, &good_writes, ub, svd[v], flags);

	(void) zio_wait(zio);

	/*
	* Flush the uberblocks to disk. This ensures that the odd labels
	* are no longer needed (because the new uberblocks and the even
	* labels are safely on disk), so it is safe to overwrite them.
	*/
	zio = zio_root(spa, NULL, NULL, flags);

	for (int v = 0; v < svdcount; v++) {
	if (vdev_writeable(svd[v])) {
	zio_flush(zio, svd[v]);
	}
	}

	(void) zio_wait(zio);

	return (good_writes >= 1 ? 0 : EIO);
	}

	/*
	* On success, increment the count of good writes for our top-level vdev.
	*/
	static void
	vdev_label_sync_done(zio_t *zio)
	{
	uint64_t *good_writes = zio->io_private;

	if (zio->io_error == 0)
	atomic_inc_64(good_writes);
	}

	/*
	* If there weren't enough good writes, indicate failure to the parent.
	*/
	static void
	vdev_label_sync_top_done(zio_t *zio)
	{
	uint64_t *good_writes = zio->io_private;

	if (*good_writes == 0)
	zio->io_error = SET_ERROR(EIO);

	kmem_free(good_writes, sizeof (uint64_t));
	}

	/*
	* We ignore errors for log and cache devices, simply free the private data.
	*/
	static void
	vdev_label_sync_ignore_done(zio_t *zio)
	{
	kmem_free(zio->io_private, sizeof (uint64_t));
	}

	/*
	* Write all even or odd labels to all leaves of the specified vdev.
	*/
	static void
	vdev_label_sync(zio_t zio, uint64_t good_writes,
	vdev_t *vd, int l, uint64_t txg, int flags)
	{
	nvlist_t *label;
	vdev_phys_t *vp;
	abd_t *vp_abd;
	char *buf;
	size_t buflen;

	for (int c = 0; c < vd->vdev_children; c++) {
	vdev_label_sync(zio, good_writes,
	vd->vdev_child[c], l, txg, flags);
	}

	if (!vd->vdev_ops->vdev_op_leaf)
	return;

	if (!vdev_writeable(vd))
	return;

	/*
	* The top-level config never needs to be written to a distributed
	* spare. When read vdev_dspare_label_read_config() will generate
	* the config for the vdev_label_read_config().
	*/
	if (vd->vdev_ops == &vdev_draid_spare_ops)
	return;

	/*
	* Generate a label describing the top-level config to which we belong.
	*/
	label = spa_config_generate(vd->vdev_spa, vd, txg, B_FALSE);

	vp_abd = abd_alloc_linear(sizeof (vdev_phys_t), B_TRUE);
	abd_zero(vp_abd, sizeof (vdev_phys_t));
	vp = abd_to_buf(vp_abd);

	buf = vp->vp_nvlist;
	buflen = sizeof (vp->vp_nvlist);

	if (!nvlist_pack(label, &buf, &buflen, NV_ENCODE_XDR, KM_SLEEP)) {
	for (; l < VDEV_LABELS; l += 2) {
	vdev_label_write(zio, vd, l, vp_abd,
	offsetof(vdev_label_t, vl_vdev_phys),
	sizeof (vdev_phys_t),
	vdev_label_sync_done, good_writes,
	flags \| ZIO_FLAG_DONT_PROPAGATE);
	}
	}

	abd_free(vp_abd);
	nvlist_free(label);
	}

	static int
	vdev_label_sync_list(spa_t *spa, int l, uint64_t txg, int flags)
	{
	list_t *dl = &spa->spa_config_dirty_list;
	vdev_t *vd;
	zio_t *zio;
	int error;

	/*
	* Write the new labels to disk.
	*/
	zio = zio_root(spa, NULL, NULL, flags);

	for (vd = list_head(dl); vd != NULL; vd = list_next(dl, vd)) {
	uint64_t *good_writes;

	ASSERT(!vd->vdev_ishole);

	good_writes = kmem_zalloc(sizeof (uint64_t), KM_SLEEP);
	zio_t *vio = zio_null(zio, spa, NULL,
	(vd->vdev_islog \|\| vd->vdev_aux != NULL) ?
	vdev_label_sync_ignore_done : vdev_label_sync_top_done,
	good_writes, flags);
	vdev_label_sync(vio, good_writes, vd, l, txg, flags);
	zio_nowait(vio);
	}

	error = zio_wait(zio);

	/*
	* Flush the new labels to disk.
	*/
	zio = zio_root(spa, NULL, NULL, flags);

	for (vd = list_head(dl); vd != NULL; vd = list_next(dl, vd))
	zio_flush(zio, vd);

	(void) zio_wait(zio);

	return (error);
	}

	/*
	* Sync the uberblock and any changes to the vdev configuration.
	*
	* The order of operations is carefully crafted to ensure that
	* if the system panics or loses power at any time, the state on disk
	* is still transactionally consistent. The in-line comments below
	* describe the failure semantics at each stage.
	*
	* Moreover, vdev_config_sync() is designed to be idempotent: if it fails
	* at any time, you can just call it again, and it will resume its work.
	*/
	int
	vdev_config_sync(vdev_t **svd, int svdcount, uint64_t txg)
	{
	spa_t *spa = svd[0]->vdev_spa;
	uberblock_t *ub = &spa->spa_uberblock;
	int error = 0;
	int flags = ZIO_FLAG_CONFIG_WRITER \| ZIO_FLAG_CANFAIL;

	ASSERT(svdcount != 0);
	retry:
	/*
	* Normally, we don't want to try too hard to write every label and
	* uberblock. If there is a flaky disk, we don't want the rest of the
	* sync process to block while we retry. But if we can't write a
	* single label out, we should retry with ZIO_FLAG_TRYHARD before
	* bailing out and declaring the pool faulted.
	*/
	if (error != 0) {
	if ((flags & ZIO_FLAG_TRYHARD) != 0)
	return (error);
	flags \|= ZIO_FLAG_TRYHARD;
	}

	ASSERT(ub->ub_txg <= txg);

	/*
	* If this isn't a resync due to I/O errors,
	* and nothing changed in this transaction group,
	* and the vdev configuration hasn't changed,
	* then there's nothing to do.
	*/
	if (ub->ub_txg < txg) {
	boolean_t changed = uberblock_update(ub, spa->spa_root_vdev,
	txg, spa->spa_mmp.mmp_delay);

	if (!changed && list_is_empty(&spa->spa_config_dirty_list))
	return (0);
	}

	if (txg > spa_freeze_txg(spa))
	return (0);

	ASSERT(txg <= spa->spa_final_txg);

	/*
	* Flush the write cache of every disk that's been written to
	* in this transaction group. This ensures that all blocks
	* written in this txg will be committed to stable storage
	* before any uberblock that references them.
	*/
	zio_t *zio = zio_root(spa, NULL, NULL, flags);

	for (vdev_t *vd =
	txg_list_head(&spa->spa_vdev_txg_list, TXG_CLEAN(txg)); vd != NULL;
	vd = txg_list_next(&spa->spa_vdev_txg_list, vd, TXG_CLEAN(txg)))
	zio_flush(zio, vd);

	(void) zio_wait(zio);

	/*
	* Sync out the even labels (L0, L2) for every dirty vdev. If the
	* system dies in the middle of this process, that's OK: all of the
	* even labels that made it to disk will be newer than any uberblock,
	* and will therefore be considered invalid. The odd labels (L1, L3),
	* which have not yet been touched, will still be valid. We flush
	* the new labels to disk to ensure that all even-label updates
	* are committed to stable storage before the uberblock update.
	*/
	if ((error = vdev_label_sync_list(spa, 0, txg, flags)) != 0) {
	if ((flags & ZIO_FLAG_TRYHARD) != 0) {
	zfs_dbgmsg("vdev_label_sync_list() returned error %d "
	"for pool '%s' when syncing out the even labels "
	"of dirty vdevs", error, spa_name(spa));
	}
	goto retry;
	}

	/*
	* Sync the uberblocks to all vdevs in svd[].
	* If the system dies in the middle of this step, there are two cases
	* to consider, and the on-disk state is consistent either way:
	*
	* (1) If none of the new uberblocks made it to disk, then the
	* previous uberblock will be the newest, and the odd labels
	* (which had not yet been touched) will be valid with respect
	* to that uberblock.
	*
	* (2) If one or more new uberblocks made it to disk, then they
	* will be the newest, and the even labels (which had all
	* been successfully committed) will be valid with respect
	* to the new uberblocks.
	*/
	if ((error = vdev_uberblock_sync_list(svd, svdcount, ub, flags)) != 0) {
	if ((flags & ZIO_FLAG_TRYHARD) != 0) {
	zfs_dbgmsg("vdev_uberblock_sync_list() returned error "
	"%d for pool '%s'", error, spa_name(spa));
	}
	goto retry;
	}

	if (spa_multihost(spa))
	mmp_update_uberblock(spa, ub);

	/*
	* Sync out odd labels for every dirty vdev. If the system dies
	* in the middle of this process, the even labels and the new
	* uberblocks will suffice to open the pool. The next time
	* the pool is opened, the first thing we'll do -- before any
	* user data is modified -- is mark every vdev dirty so that
	* all labels will be brought up to date. We flush the new labels
	* to disk to ensure that all odd-label updates are committed to
	* stable storage before the next transaction group begins.
	*/
	if ((error = vdev_label_sync_list(spa, 1, txg, flags)) != 0) {
	if ((flags & ZIO_FLAG_TRYHARD) != 0) {
	zfs_dbgmsg("vdev_label_sync_list() returned error %d "
	"for pool '%s' when syncing out the odd labels of "
	"dirty vdevs", error, spa_name(spa));
	}
	goto retry;
	}

	return (0);
	}
	diff --git a/sys/contrib/openzfs/module/zfs/vdev_queue.c b/sys/contrib/openzfs/module/zfs/vdev_queue.c
	index cc5b15b8c028..af612ba9c9c6 100644
	--- a/sys/contrib/openzfs/module/zfs/vdev_queue.c
	+++ b/sys/contrib/openzfs/module/zfs/vdev_queue.c
	@@ -1,1121 +1,1121 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright 2009 Sun Microsystems, Inc. All rights reserved.
	* Use is subject to license terms.
	*/

	/*
	* Copyright (c) 2012, 2018 by Delphix. All rights reserved.
	*/

	#include <sys/zfs_context.h>
	#include <sys/vdev_impl.h>
	#include <sys/spa_impl.h>
	#include <sys/zio.h>
	#include <sys/avl.h>
	#include <sys/dsl_pool.h>
	#include <sys/metaslab_impl.h>
	#include <sys/spa.h>
	#include <sys/abd.h>

	/*
	* ZFS I/O Scheduler
	* ---------------
	*
	* ZFS issues I/O operations to leaf vdevs to satisfy and complete zios. The
	* I/O scheduler determines when and in what order those operations are
	* issued. The I/O scheduler divides operations into five I/O classes
	* prioritized in the following order: sync read, sync write, async read,
	* async write, and scrub/resilver. Each queue defines the minimum and
	* maximum number of concurrent operations that may be issued to the device.
	* In addition, the device has an aggregate maximum. Note that the sum of the
	* per-queue minimums must not exceed the aggregate maximum. If the
	* sum of the per-queue maximums exceeds the aggregate maximum, then the
	* number of active i/os may reach zfs_vdev_max_active, in which case no
	* further i/os will be issued regardless of whether all per-queue
	* minimums have been met.
	*
	* For many physical devices, throughput increases with the number of
	* concurrent operations, but latency typically suffers. Further, physical
	* devices typically have a limit at which more concurrent operations have no
	* effect on throughput or can actually cause it to decrease.
	*
	* The scheduler selects the next operation to issue by first looking for an
	* I/O class whose minimum has not been satisfied. Once all are satisfied and
	* the aggregate maximum has not been hit, the scheduler looks for classes
	* whose maximum has not been satisfied. Iteration through the I/O classes is
	* done in the order specified above. No further operations are issued if the
	* aggregate maximum number of concurrent operations has been hit or if there
	* are no operations queued for an I/O class that has not hit its maximum.
	* Every time an i/o is queued or an operation completes, the I/O scheduler
	* looks for new operations to issue.
	*
	* All I/O classes have a fixed maximum number of outstanding operations
	* except for the async write class. Asynchronous writes represent the data
	* that is committed to stable storage during the syncing stage for
	* transaction groups (see txg.c). Transaction groups enter the syncing state
	* periodically so the number of queued async writes will quickly burst up and
	* then bleed down to zero. Rather than servicing them as quickly as possible,
	* the I/O scheduler changes the maximum number of active async write i/os
	* according to the amount of dirty data in the pool (see dsl_pool.c). Since
	* both throughput and latency typically increase with the number of
	* concurrent operations issued to physical devices, reducing the burstiness
	* in the number of concurrent operations also stabilizes the response time of
	* operations from other -- and in particular synchronous -- queues. In broad
	* strokes, the I/O scheduler will issue more concurrent operations from the
	* async write queue as there's more dirty data in the pool.
	*
	* Async Writes
	*
	* The number of concurrent operations issued for the async write I/O class
	* follows a piece-wise linear function defined by a few adjustable points.
	*
	* \| o---------\| <-- zfs_vdev_async_write_max_active
	* ^ \| /^ \|
	* \| \| / \| \|
	* active \| / \| \|
	* I/O \| / \| \|
	* count \| / \| \|
	* \| / \| \|
	* \|------------o \| \| <-- zfs_vdev_async_write_min_active
	* 0\|____________^______\|_________\|
	* 0% \| \| 100% of zfs_dirty_data_max
	* \| \|
	* \| `-- zfs_vdev_async_write_active_max_dirty_percent
	* `--------- zfs_vdev_async_write_active_min_dirty_percent
	*
	* Until the amount of dirty data exceeds a minimum percentage of the dirty
	* data allowed in the pool, the I/O scheduler will limit the number of
	* concurrent operations to the minimum. As that threshold is crossed, the
	* number of concurrent operations issued increases linearly to the maximum at
	* the specified maximum percentage of the dirty data allowed in the pool.
	*
	* Ideally, the amount of dirty data on a busy pool will stay in the sloped
	* part of the function between zfs_vdev_async_write_active_min_dirty_percent
	* and zfs_vdev_async_write_active_max_dirty_percent. If it exceeds the
	* maximum percentage, this indicates that the rate of incoming data is
	* greater than the rate that the backend storage can handle. In this case, we
	* must further throttle incoming writes (see dmu_tx_delay() for details).
	*/

	/*
	* The maximum number of i/os active to each device. Ideally, this will be >=
	* the sum of each queue's max_active.
	*/
	uint32_t zfs_vdev_max_active = 1000;

	/*
	* Per-queue limits on the number of i/os active to each device. If the
	* number of active i/os is < zfs_vdev_max_active, then the min_active comes
	* into play. We will send min_active from each queue round-robin, and then
	* send from queues in the order defined by zio_priority_t up to max_active.
	* Some queues have additional mechanisms to limit number of active I/Os in
	* addition to min_active and max_active, see below.
	*
	* In general, smaller max_active's will lead to lower latency of synchronous
	* operations. Larger max_active's may lead to higher overall throughput,
	* depending on underlying storage.
	*
	* The ratio of the queues' max_actives determines the balance of performance
	* between reads, writes, and scrubs. E.g., increasing
	* zfs_vdev_scrub_max_active will cause the scrub or resilver to complete
	* more quickly, but reads and writes to have higher latency and lower
	* throughput.
	*/
	uint32_t zfs_vdev_sync_read_min_active = 10;
	uint32_t zfs_vdev_sync_read_max_active = 10;
	uint32_t zfs_vdev_sync_write_min_active = 10;
	uint32_t zfs_vdev_sync_write_max_active = 10;
	uint32_t zfs_vdev_async_read_min_active = 1;
	uint32_t zfs_vdev_async_read_max_active = 3;
	uint32_t zfs_vdev_async_write_min_active = 2;
	uint32_t zfs_vdev_async_write_max_active = 10;
	uint32_t zfs_vdev_scrub_min_active = 1;
	uint32_t zfs_vdev_scrub_max_active = 3;
	uint32_t zfs_vdev_removal_min_active = 1;
	uint32_t zfs_vdev_removal_max_active = 2;
	uint32_t zfs_vdev_initializing_min_active = 1;
	uint32_t zfs_vdev_initializing_max_active = 1;
	uint32_t zfs_vdev_trim_min_active = 1;
	uint32_t zfs_vdev_trim_max_active = 2;
	uint32_t zfs_vdev_rebuild_min_active = 1;
	uint32_t zfs_vdev_rebuild_max_active = 3;

	/*
	* When the pool has less than zfs_vdev_async_write_active_min_dirty_percent
	* dirty data, use zfs_vdev_async_write_min_active. When it has more than
	* zfs_vdev_async_write_active_max_dirty_percent, use
	* zfs_vdev_async_write_max_active. The value is linearly interpolated
	* between min and max.
	*/
	int zfs_vdev_async_write_active_min_dirty_percent = 30;
	int zfs_vdev_async_write_active_max_dirty_percent = 60;

	/*
	* For non-interactive I/O (scrub, resilver, removal, initialize and rebuild),
	* the number of concurrently-active I/O's is limited to *_min_active, unless
	* the vdev is "idle". When there are no interactive I/Os active (sync or
	* async), and zfs_vdev_nia_delay I/Os have completed since the last
	* interactive I/O, then the vdev is considered to be "idle", and the number
	* of concurrently-active non-interactive I/O's is increased to *_max_active.
	*/
	uint_t zfs_vdev_nia_delay = 5;

	/*
	* Some HDDs tend to prioritize sequential I/O so high that concurrent
	* random I/O latency reaches several seconds. On some HDDs it happens
	* even if sequential I/Os are submitted one at a time, and so setting
	* *_max_active to 1 does not help. To prevent non-interactive I/Os, like
	* scrub, from monopolizing the device no more than zfs_vdev_nia_credit
	* I/Os can be sent while there are outstanding incomplete interactive
	* I/Os. This enforced wait ensures the HDD services the interactive I/O
	* within a reasonable amount of time.
	*/
	uint_t zfs_vdev_nia_credit = 5;

	/*
	* To reduce IOPs, we aggregate small adjacent I/Os into one large I/O.
	* For read I/Os, we also aggregate across small adjacency gaps; for writes
	* we include spans of optional I/Os to aid aggregation at the disk even when
	* they aren't able to help us aggregate at this level.
	*/
	int zfs_vdev_aggregation_limit = 1 << 20;
	int zfs_vdev_aggregation_limit_non_rotating = SPA_OLD_MAXBLOCKSIZE;
	int zfs_vdev_read_gap_limit = 32 << 10;
	int zfs_vdev_write_gap_limit = 4 << 10;

	/*
	* Define the queue depth percentage for each top-level. This percentage is
	* used in conjunction with zfs_vdev_async_max_active to determine how many
	* allocations a specific top-level vdev should handle. Once the queue depth
	* reaches zfs_vdev_queue_depth_pct * zfs_vdev_async_write_max_active / 100
	* then allocator will stop allocating blocks on that top-level device.
	* The default kernel setting is 1000% which will yield 100 allocations per
	* device. For userland testing, the default setting is 300% which equates
	* to 30 allocations per device.
	*/
	#ifdef _KERNEL
	int zfs_vdev_queue_depth_pct = 1000;
	#else
	int zfs_vdev_queue_depth_pct = 300;
	#endif

	/*
	* When performing allocations for a given metaslab, we want to make sure that
	* there are enough IOs to aggregate together to improve throughput. We want to
	* ensure that there are at least 128k worth of IOs that can be aggregated, and
	* we assume that the average allocation size is 4k, so we need the queue depth
	* to be 32 per allocator to get good aggregation of sequential writes.
	*/
	int zfs_vdev_def_queue_depth = 32;

	/*
	* Allow TRIM I/Os to be aggregated. This should normally not be needed since
	* TRIM I/O for extents up to zfs_trim_extent_bytes_max (128M) can be submitted
	* by the TRIM code in zfs_trim.c.
	*/
	int zfs_vdev_aggregate_trim = 0;

	static int
	vdev_queue_offset_compare(const void x1, const void x2)
	{
	const zio_t z1 = (const zio_t )x1;
	const zio_t z2 = (const zio_t )x2;

	int cmp = TREE_CMP(z1->io_offset, z2->io_offset);

	if (likely(cmp))
	return (cmp);

	return (TREE_PCMP(z1, z2));
	}

	static inline avl_tree_t *
	vdev_queue_class_tree(vdev_queue_t *vq, zio_priority_t p)
	{
	return (&vq->vq_class[p].vqc_queued_tree);
	}

	static inline avl_tree_t *
	vdev_queue_type_tree(vdev_queue_t *vq, zio_type_t t)
	{
	ASSERT(t == ZIO_TYPE_READ \|\| t == ZIO_TYPE_WRITE \|\| t == ZIO_TYPE_TRIM);
	if (t == ZIO_TYPE_READ)
	return (&vq->vq_read_offset_tree);
	else if (t == ZIO_TYPE_WRITE)
	return (&vq->vq_write_offset_tree);
	else
	return (&vq->vq_trim_offset_tree);
	}

	static int
	vdev_queue_timestamp_compare(const void x1, const void x2)
	{
	const zio_t z1 = (const zio_t )x1;
	const zio_t z2 = (const zio_t )x2;

	int cmp = TREE_CMP(z1->io_timestamp, z2->io_timestamp);

	if (likely(cmp))
	return (cmp);

	return (TREE_PCMP(z1, z2));
	}

	static int
	vdev_queue_class_min_active(vdev_queue_t *vq, zio_priority_t p)
	{
	switch (p) {
	case ZIO_PRIORITY_SYNC_READ:
	return (zfs_vdev_sync_read_min_active);
	case ZIO_PRIORITY_SYNC_WRITE:
	return (zfs_vdev_sync_write_min_active);
	case ZIO_PRIORITY_ASYNC_READ:
	return (zfs_vdev_async_read_min_active);
	case ZIO_PRIORITY_ASYNC_WRITE:
	return (zfs_vdev_async_write_min_active);
	case ZIO_PRIORITY_SCRUB:
	return (vq->vq_ia_active == 0 ? zfs_vdev_scrub_min_active :
	MIN(vq->vq_nia_credit, zfs_vdev_scrub_min_active));
	case ZIO_PRIORITY_REMOVAL:
	return (vq->vq_ia_active == 0 ? zfs_vdev_removal_min_active :
	MIN(vq->vq_nia_credit, zfs_vdev_removal_min_active));
	case ZIO_PRIORITY_INITIALIZING:
	return (vq->vq_ia_active == 0 ?zfs_vdev_initializing_min_active:
	MIN(vq->vq_nia_credit, zfs_vdev_initializing_min_active));
	case ZIO_PRIORITY_TRIM:
	return (zfs_vdev_trim_min_active);
	case ZIO_PRIORITY_REBUILD:
	return (vq->vq_ia_active == 0 ? zfs_vdev_rebuild_min_active :
	MIN(vq->vq_nia_credit, zfs_vdev_rebuild_min_active));
	default:
	panic("invalid priority %u", p);
	return (0);
	}
	}

	static int
	vdev_queue_max_async_writes(spa_t *spa)
	{
	int writes;
	uint64_t dirty = 0;
	dsl_pool_t *dp = spa_get_dsl(spa);
	uint64_t min_bytes = zfs_dirty_data_max *
	zfs_vdev_async_write_active_min_dirty_percent / 100;
	uint64_t max_bytes = zfs_dirty_data_max *
	zfs_vdev_async_write_active_max_dirty_percent / 100;

	/*
	* Async writes may occur before the assignment of the spa's
	* dsl_pool_t if a self-healing zio is issued prior to the
	* completion of dmu_objset_open_impl().
	*/
	if (dp == NULL)
	return (zfs_vdev_async_write_max_active);

	/*
	* Sync tasks correspond to interactive user actions. To reduce the
	* execution time of those actions we push data out as fast as possible.
	*/
	dirty = dp->dp_dirty_total;
	if (dirty > max_bytes \|\| spa_has_pending_synctask(spa))
	return (zfs_vdev_async_write_max_active);

	if (dirty < min_bytes)
	return (zfs_vdev_async_write_min_active);

	/*
	* linear interpolation:
	* slope = (max_writes - min_writes) / (max_bytes - min_bytes)
	* move right by min_bytes
	* move up by min_writes
	*/
	writes = (dirty - min_bytes) *
	(zfs_vdev_async_write_max_active -
	zfs_vdev_async_write_min_active) /
	(max_bytes - min_bytes) +
	zfs_vdev_async_write_min_active;
	ASSERT3U(writes, >=, zfs_vdev_async_write_min_active);
	ASSERT3U(writes, <=, zfs_vdev_async_write_max_active);
	return (writes);
	}

	static int
	vdev_queue_class_max_active(spa_t spa, vdev_queue_t vq, zio_priority_t p)
	{
	switch (p) {
	case ZIO_PRIORITY_SYNC_READ:
	return (zfs_vdev_sync_read_max_active);
	case ZIO_PRIORITY_SYNC_WRITE:
	return (zfs_vdev_sync_write_max_active);
	case ZIO_PRIORITY_ASYNC_READ:
	return (zfs_vdev_async_read_max_active);
	case ZIO_PRIORITY_ASYNC_WRITE:
	return (vdev_queue_max_async_writes(spa));
	case ZIO_PRIORITY_SCRUB:
	if (vq->vq_ia_active > 0) {
	return (MIN(vq->vq_nia_credit,
	zfs_vdev_scrub_min_active));
	} else if (vq->vq_nia_credit < zfs_vdev_nia_delay)
	return (MAX(1, zfs_vdev_scrub_min_active));
	return (zfs_vdev_scrub_max_active);
	case ZIO_PRIORITY_REMOVAL:
	if (vq->vq_ia_active > 0) {
	return (MIN(vq->vq_nia_credit,
	zfs_vdev_removal_min_active));
	} else if (vq->vq_nia_credit < zfs_vdev_nia_delay)
	return (MAX(1, zfs_vdev_removal_min_active));
	return (zfs_vdev_removal_max_active);
	case ZIO_PRIORITY_INITIALIZING:
	if (vq->vq_ia_active > 0) {
	return (MIN(vq->vq_nia_credit,
	zfs_vdev_initializing_min_active));
	} else if (vq->vq_nia_credit < zfs_vdev_nia_delay)
	return (MAX(1, zfs_vdev_initializing_min_active));
	return (zfs_vdev_initializing_max_active);
	case ZIO_PRIORITY_TRIM:
	return (zfs_vdev_trim_max_active);
	case ZIO_PRIORITY_REBUILD:
	if (vq->vq_ia_active > 0) {
	return (MIN(vq->vq_nia_credit,
	zfs_vdev_rebuild_min_active));
	} else if (vq->vq_nia_credit < zfs_vdev_nia_delay)
	return (MAX(1, zfs_vdev_rebuild_min_active));
	return (zfs_vdev_rebuild_max_active);
	default:
	panic("invalid priority %u", p);
	return (0);
	}
	}

	/*
	- * Return the i/o class to issue from, or ZIO_PRIORITY_MAX_QUEUEABLE if
	+ * Return the i/o class to issue from, or ZIO_PRIORITY_NUM_QUEUEABLE if
	* there is no eligible class.
	*/
	static zio_priority_t
	vdev_queue_class_to_issue(vdev_queue_t *vq)
	{
	spa_t *spa = vq->vq_vdev->vdev_spa;
	zio_priority_t p, n;

	if (avl_numnodes(&vq->vq_active_tree) >= zfs_vdev_max_active)
	return (ZIO_PRIORITY_NUM_QUEUEABLE);

	/*
	* Find a queue that has not reached its minimum # outstanding i/os.
	* Do round-robin to reduce starvation due to zfs_vdev_max_active
	* and vq_nia_credit limits.
	*/
	for (n = 0; n < ZIO_PRIORITY_NUM_QUEUEABLE; n++) {
	p = (vq->vq_last_prio + n + 1) % ZIO_PRIORITY_NUM_QUEUEABLE;
	if (avl_numnodes(vdev_queue_class_tree(vq, p)) > 0 &&
	vq->vq_class[p].vqc_active <
	vdev_queue_class_min_active(vq, p)) {
	vq->vq_last_prio = p;
	return (p);
	}
	}

	/*
	* If we haven't found a queue, look for one that hasn't reached its
	* maximum # outstanding i/os.
	*/
	for (p = 0; p < ZIO_PRIORITY_NUM_QUEUEABLE; p++) {
	if (avl_numnodes(vdev_queue_class_tree(vq, p)) > 0 &&
	vq->vq_class[p].vqc_active <
	vdev_queue_class_max_active(spa, vq, p)) {
	vq->vq_last_prio = p;
	return (p);
	}
	}

	/* No eligible queued i/os */
	return (ZIO_PRIORITY_NUM_QUEUEABLE);
	}

	void
	vdev_queue_init(vdev_t *vd)
	{
	vdev_queue_t *vq = &vd->vdev_queue;
	zio_priority_t p;

	mutex_init(&vq->vq_lock, NULL, MUTEX_DEFAULT, NULL);
	vq->vq_vdev = vd;
	taskq_init_ent(&vd->vdev_queue.vq_io_search.io_tqent);

	avl_create(&vq->vq_active_tree, vdev_queue_offset_compare,
	sizeof (zio_t), offsetof(struct zio, io_queue_node));
	avl_create(vdev_queue_type_tree(vq, ZIO_TYPE_READ),
	vdev_queue_offset_compare, sizeof (zio_t),
	offsetof(struct zio, io_offset_node));
	avl_create(vdev_queue_type_tree(vq, ZIO_TYPE_WRITE),
	vdev_queue_offset_compare, sizeof (zio_t),
	offsetof(struct zio, io_offset_node));
	avl_create(vdev_queue_type_tree(vq, ZIO_TYPE_TRIM),
	vdev_queue_offset_compare, sizeof (zio_t),
	offsetof(struct zio, io_offset_node));

	for (p = 0; p < ZIO_PRIORITY_NUM_QUEUEABLE; p++) {
	int (compfn) (const void , const void *);

	/*
	* The synchronous/trim i/o queues are dispatched in FIFO rather
	* than LBA order. This provides more consistent latency for
	* these i/os.
	*/
	if (p == ZIO_PRIORITY_SYNC_READ \|\|
	p == ZIO_PRIORITY_SYNC_WRITE \|\|
	p == ZIO_PRIORITY_TRIM) {
	compfn = vdev_queue_timestamp_compare;
	} else {
	compfn = vdev_queue_offset_compare;
	}
	avl_create(vdev_queue_class_tree(vq, p), compfn,
	sizeof (zio_t), offsetof(struct zio, io_queue_node));
	}

	vq->vq_last_offset = 0;
	}

	void
	vdev_queue_fini(vdev_t *vd)
	{
	vdev_queue_t *vq = &vd->vdev_queue;

	for (zio_priority_t p = 0; p < ZIO_PRIORITY_NUM_QUEUEABLE; p++)
	avl_destroy(vdev_queue_class_tree(vq, p));
	avl_destroy(&vq->vq_active_tree);
	avl_destroy(vdev_queue_type_tree(vq, ZIO_TYPE_READ));
	avl_destroy(vdev_queue_type_tree(vq, ZIO_TYPE_WRITE));
	avl_destroy(vdev_queue_type_tree(vq, ZIO_TYPE_TRIM));

	mutex_destroy(&vq->vq_lock);
	}

	static void
	vdev_queue_io_add(vdev_queue_t vq, zio_t zio)
	{
	ASSERT3U(zio->io_priority, <, ZIO_PRIORITY_NUM_QUEUEABLE);
	avl_add(vdev_queue_class_tree(vq, zio->io_priority), zio);
	avl_add(vdev_queue_type_tree(vq, zio->io_type), zio);
	}

	static void
	vdev_queue_io_remove(vdev_queue_t vq, zio_t zio)
	{
	ASSERT3U(zio->io_priority, <, ZIO_PRIORITY_NUM_QUEUEABLE);
	avl_remove(vdev_queue_class_tree(vq, zio->io_priority), zio);
	avl_remove(vdev_queue_type_tree(vq, zio->io_type), zio);
	}

	static boolean_t
	vdev_queue_is_interactive(zio_priority_t p)
	{
	switch (p) {
	case ZIO_PRIORITY_SCRUB:
	case ZIO_PRIORITY_REMOVAL:
	case ZIO_PRIORITY_INITIALIZING:
	case ZIO_PRIORITY_REBUILD:
	return (B_FALSE);
	default:
	return (B_TRUE);
	}
	}

	static void
	vdev_queue_pending_add(vdev_queue_t vq, zio_t zio)
	{
	ASSERT(MUTEX_HELD(&vq->vq_lock));
	ASSERT3U(zio->io_priority, <, ZIO_PRIORITY_NUM_QUEUEABLE);
	vq->vq_class[zio->io_priority].vqc_active++;
	if (vdev_queue_is_interactive(zio->io_priority)) {
	if (++vq->vq_ia_active == 1)
	vq->vq_nia_credit = 1;
	} else if (vq->vq_ia_active > 0) {
	vq->vq_nia_credit--;
	}
	avl_add(&vq->vq_active_tree, zio);
	}

	static void
	vdev_queue_pending_remove(vdev_queue_t vq, zio_t zio)
	{
	ASSERT(MUTEX_HELD(&vq->vq_lock));
	ASSERT3U(zio->io_priority, <, ZIO_PRIORITY_NUM_QUEUEABLE);
	vq->vq_class[zio->io_priority].vqc_active--;
	if (vdev_queue_is_interactive(zio->io_priority)) {
	if (--vq->vq_ia_active == 0)
	vq->vq_nia_credit = 0;
	else
	vq->vq_nia_credit = zfs_vdev_nia_credit;
	} else if (vq->vq_ia_active == 0)
	vq->vq_nia_credit++;
	avl_remove(&vq->vq_active_tree, zio);
	}

	static void
	vdev_queue_agg_io_done(zio_t *aio)
	{
	abd_free(aio->io_abd);
	}

	/*
	* Compute the range spanned by two i/os, which is the endpoint of the last
	* (lio->io_offset + lio->io_size) minus start of the first (fio->io_offset).
	* Conveniently, the gap between fio and lio is given by -IO_SPAN(lio, fio);
	* thus fio and lio are adjacent if and only if IO_SPAN(lio, fio) == 0.
	*/
	#define IO_SPAN(fio, lio) ((lio)->io_offset + (lio)->io_size - (fio)->io_offset)
	#define IO_GAP(fio, lio) (-IO_SPAN(lio, fio))

	/*
	* Sufficiently adjacent io_offset's in ZIOs will be aggregated. We do this
	* by creating a gang ABD from the adjacent ZIOs io_abd's. By using
	* a gang ABD we avoid doing memory copies to and from the parent,
	* child ZIOs. The gang ABD also accounts for gaps between adjacent
	* io_offsets by simply getting the zero ABD for writes or allocating
	* a new ABD for reads and placing them in the gang ABD as well.
	*/
	static zio_t *
	vdev_queue_aggregate(vdev_queue_t vq, zio_t zio)
	{
	zio_t first, last, aio, dio, mandatory, nio;
	uint64_t maxgap = 0;
	uint64_t size;
	uint64_t limit;
	int maxblocksize;
	boolean_t stretch = B_FALSE;
	avl_tree_t *t = vdev_queue_type_tree(vq, zio->io_type);
	enum zio_flag flags = zio->io_flags & ZIO_FLAG_AGG_INHERIT;
	uint64_t next_offset;
	abd_t *abd;

	maxblocksize = spa_maxblocksize(vq->vq_vdev->vdev_spa);
	if (vq->vq_vdev->vdev_nonrot)
	limit = zfs_vdev_aggregation_limit_non_rotating;
	else
	limit = zfs_vdev_aggregation_limit;
	limit = MAX(MIN(limit, maxblocksize), 0);

	if (zio->io_flags & ZIO_FLAG_DONT_AGGREGATE \|\| limit == 0)
	return (NULL);

	/*
	* While TRIM commands could be aggregated based on offset this
	* behavior is disabled until it's determined to be beneficial.
	*/
	if (zio->io_type == ZIO_TYPE_TRIM && !zfs_vdev_aggregate_trim)
	return (NULL);

	/*
	* I/Os to distributed spares are directly dispatched to the dRAID
	* leaf vdevs for aggregation. See the comment at the end of the
	* zio_vdev_io_start() function.
	*/
	ASSERT(vq->vq_vdev->vdev_ops != &vdev_draid_spare_ops);

	first = last = zio;

	if (zio->io_type == ZIO_TYPE_READ)
	maxgap = zfs_vdev_read_gap_limit;

	/*
	* We can aggregate I/Os that are sufficiently adjacent and of
	* the same flavor, as expressed by the AGG_INHERIT flags.
	* The latter requirement is necessary so that certain
	* attributes of the I/O, such as whether it's a normal I/O
	* or a scrub/resilver, can be preserved in the aggregate.
	* We can include optional I/Os, but don't allow them
	* to begin a range as they add no benefit in that situation.
	*/

	/*
	* We keep track of the last non-optional I/O.
	*/
	mandatory = (first->io_flags & ZIO_FLAG_OPTIONAL) ? NULL : first;

	/*
	* Walk backwards through sufficiently contiguous I/Os
	* recording the last non-optional I/O.
	*/
	while ((dio = AVL_PREV(t, first)) != NULL &&
	(dio->io_flags & ZIO_FLAG_AGG_INHERIT) == flags &&
	IO_SPAN(dio, last) <= limit &&
	IO_GAP(dio, first) <= maxgap &&
	dio->io_type == zio->io_type) {
	first = dio;
	if (mandatory == NULL && !(first->io_flags & ZIO_FLAG_OPTIONAL))
	mandatory = first;
	}

	/*
	* Skip any initial optional I/Os.
	*/
	while ((first->io_flags & ZIO_FLAG_OPTIONAL) && first != last) {
	first = AVL_NEXT(t, first);
	ASSERT(first != NULL);
	}


	/*
	* Walk forward through sufficiently contiguous I/Os.
	* The aggregation limit does not apply to optional i/os, so that
	* we can issue contiguous writes even if they are larger than the
	* aggregation limit.
	*/
	while ((dio = AVL_NEXT(t, last)) != NULL &&
	(dio->io_flags & ZIO_FLAG_AGG_INHERIT) == flags &&
	(IO_SPAN(first, dio) <= limit \|\|
	(dio->io_flags & ZIO_FLAG_OPTIONAL)) &&
	IO_SPAN(first, dio) <= maxblocksize &&
	IO_GAP(last, dio) <= maxgap &&
	dio->io_type == zio->io_type) {
	last = dio;
	if (!(last->io_flags & ZIO_FLAG_OPTIONAL))
	mandatory = last;
	}

	/*
	* Now that we've established the range of the I/O aggregation
	* we must decide what to do with trailing optional I/Os.
	* For reads, there's nothing to do. While we are unable to
	* aggregate further, it's possible that a trailing optional
	* I/O would allow the underlying device to aggregate with
	* subsequent I/Os. We must therefore determine if the next
	* non-optional I/O is close enough to make aggregation
	* worthwhile.
	*/
	if (zio->io_type == ZIO_TYPE_WRITE && mandatory != NULL) {
	zio_t *nio = last;
	while ((dio = AVL_NEXT(t, nio)) != NULL &&
	IO_GAP(nio, dio) == 0 &&
	IO_GAP(mandatory, dio) <= zfs_vdev_write_gap_limit) {
	nio = dio;
	if (!(nio->io_flags & ZIO_FLAG_OPTIONAL)) {
	stretch = B_TRUE;
	break;
	}
	}
	}

	if (stretch) {
	/*
	* We are going to include an optional io in our aggregated
	* span, thus closing the write gap. Only mandatory i/os can
	* start aggregated spans, so make sure that the next i/o
	* after our span is mandatory.
	*/
	dio = AVL_NEXT(t, last);
	dio->io_flags &= ~ZIO_FLAG_OPTIONAL;
	} else {
	/* do not include the optional i/o */
	while (last != mandatory && last != first) {
	ASSERT(last->io_flags & ZIO_FLAG_OPTIONAL);
	last = AVL_PREV(t, last);
	ASSERT(last != NULL);
	}
	}

	if (first == last)
	return (NULL);

	size = IO_SPAN(first, last);
	ASSERT3U(size, <=, maxblocksize);

	abd = abd_alloc_gang();
	if (abd == NULL)
	return (NULL);

	aio = zio_vdev_delegated_io(first->io_vd, first->io_offset,
	abd, size, first->io_type, zio->io_priority,
	flags \| ZIO_FLAG_DONT_CACHE \| ZIO_FLAG_DONT_QUEUE,
	vdev_queue_agg_io_done, NULL);
	aio->io_timestamp = first->io_timestamp;

	nio = first;
	next_offset = first->io_offset;
	do {
	dio = nio;
	nio = AVL_NEXT(t, dio);
	zio_add_child(dio, aio);
	vdev_queue_io_remove(vq, dio);

	if (dio->io_offset != next_offset) {
	/* allocate a buffer for a read gap */
	ASSERT3U(dio->io_type, ==, ZIO_TYPE_READ);
	ASSERT3U(dio->io_offset, >, next_offset);
	abd = abd_alloc_for_io(
	dio->io_offset - next_offset, B_TRUE);
	abd_gang_add(aio->io_abd, abd, B_TRUE);
	}
	if (dio->io_abd &&
	(dio->io_size != abd_get_size(dio->io_abd))) {
	/* abd size not the same as IO size */
	ASSERT3U(abd_get_size(dio->io_abd), >, dio->io_size);
	abd = abd_get_offset_size(dio->io_abd, 0, dio->io_size);
	abd_gang_add(aio->io_abd, abd, B_TRUE);
	} else {
	if (dio->io_flags & ZIO_FLAG_NODATA) {
	/* allocate a buffer for a write gap */
	ASSERT3U(dio->io_type, ==, ZIO_TYPE_WRITE);
	ASSERT3P(dio->io_abd, ==, NULL);
	abd_gang_add(aio->io_abd,
	abd_get_zeros(dio->io_size), B_TRUE);
	} else {
	/*
	* We pass B_FALSE to abd_gang_add()
	* because we did not allocate a new
	* ABD, so it is assumed the caller
	* will free this ABD.
	*/
	abd_gang_add(aio->io_abd, dio->io_abd,
	B_FALSE);
	}
	}
	next_offset = dio->io_offset + dio->io_size;
	} while (dio != last);
	ASSERT3U(abd_get_size(aio->io_abd), ==, aio->io_size);

	/*
	* Callers must call zio_vdev_io_bypass() and zio_execute() for
	* aggregated (parent) I/Os so that we could avoid dropping the
	* queue's lock here to avoid a deadlock that we could encounter
	* due to lock order reversal between vq_lock and io_lock in
	* zio_change_priority().
	*/
	return (aio);
	}

	static zio_t *
	vdev_queue_io_to_issue(vdev_queue_t *vq)
	{
	zio_t zio, aio;
	zio_priority_t p;
	avl_index_t idx;
	avl_tree_t *tree;

	again:
	ASSERT(MUTEX_HELD(&vq->vq_lock));

	p = vdev_queue_class_to_issue(vq);

	if (p == ZIO_PRIORITY_NUM_QUEUEABLE) {
	/* No eligible queued i/os */
	return (NULL);
	}

	/*
	* For LBA-ordered queues (async / scrub / initializing), issue the
	* i/o which follows the most recently issued i/o in LBA (offset) order.
	*
	* For FIFO queues (sync/trim), issue the i/o with the lowest timestamp.
	*/
	tree = vdev_queue_class_tree(vq, p);
	vq->vq_io_search.io_timestamp = 0;
	vq->vq_io_search.io_offset = vq->vq_last_offset - 1;
	VERIFY3P(avl_find(tree, &vq->vq_io_search, &idx), ==, NULL);
	zio = avl_nearest(tree, idx, AVL_AFTER);
	if (zio == NULL)
	zio = avl_first(tree);
	ASSERT3U(zio->io_priority, ==, p);

	aio = vdev_queue_aggregate(vq, zio);
	if (aio != NULL) {
	zio = aio;
	} else {
	vdev_queue_io_remove(vq, zio);

	/*
	* If the I/O is or was optional and therefore has no data, we
	* need to simply discard it. We need to drop the vdev queue's
	* lock to avoid a deadlock that we could encounter since this
	* I/O will complete immediately.
	*/
	if (zio->io_flags & ZIO_FLAG_NODATA) {
	mutex_exit(&vq->vq_lock);
	zio_vdev_io_bypass(zio);
	zio_execute(zio);
	mutex_enter(&vq->vq_lock);
	goto again;
	}
	}

	vdev_queue_pending_add(vq, zio);
	vq->vq_last_offset = zio->io_offset + zio->io_size;

	return (zio);
	}

	zio_t *
	vdev_queue_io(zio_t *zio)
	{
	vdev_queue_t *vq = &zio->io_vd->vdev_queue;
	zio_t dio, nio;
	zio_link_t *zl = NULL;

	if (zio->io_flags & ZIO_FLAG_DONT_QUEUE)
	return (zio);

	/*
	* Children i/os inherent their parent's priority, which might
	* not match the child's i/o type. Fix it up here.
	*/
	if (zio->io_type == ZIO_TYPE_READ) {
	ASSERT(zio->io_priority != ZIO_PRIORITY_TRIM);

	if (zio->io_priority != ZIO_PRIORITY_SYNC_READ &&
	zio->io_priority != ZIO_PRIORITY_ASYNC_READ &&
	zio->io_priority != ZIO_PRIORITY_SCRUB &&
	zio->io_priority != ZIO_PRIORITY_REMOVAL &&
	zio->io_priority != ZIO_PRIORITY_INITIALIZING &&
	zio->io_priority != ZIO_PRIORITY_REBUILD) {
	zio->io_priority = ZIO_PRIORITY_ASYNC_READ;
	}
	} else if (zio->io_type == ZIO_TYPE_WRITE) {
	ASSERT(zio->io_priority != ZIO_PRIORITY_TRIM);

	if (zio->io_priority != ZIO_PRIORITY_SYNC_WRITE &&
	zio->io_priority != ZIO_PRIORITY_ASYNC_WRITE &&
	zio->io_priority != ZIO_PRIORITY_REMOVAL &&
	zio->io_priority != ZIO_PRIORITY_INITIALIZING &&
	zio->io_priority != ZIO_PRIORITY_REBUILD) {
	zio->io_priority = ZIO_PRIORITY_ASYNC_WRITE;
	}
	} else {
	ASSERT(zio->io_type == ZIO_TYPE_TRIM);
	ASSERT(zio->io_priority == ZIO_PRIORITY_TRIM);
	}

	zio->io_flags \|= ZIO_FLAG_DONT_CACHE \| ZIO_FLAG_DONT_QUEUE;
	zio->io_timestamp = gethrtime();

	mutex_enter(&vq->vq_lock);
	vdev_queue_io_add(vq, zio);
	nio = vdev_queue_io_to_issue(vq);
	mutex_exit(&vq->vq_lock);

	if (nio == NULL)
	return (NULL);

	if (nio->io_done == vdev_queue_agg_io_done) {
	while ((dio = zio_walk_parents(nio, &zl)) != NULL) {
	ASSERT3U(dio->io_type, ==, nio->io_type);
	zio_vdev_io_bypass(dio);
	zio_execute(dio);
	}
	zio_nowait(nio);
	return (NULL);
	}

	return (nio);
	}

	void
	vdev_queue_io_done(zio_t *zio)
	{
	vdev_queue_t *vq = &zio->io_vd->vdev_queue;
	zio_t dio, nio;
	zio_link_t *zl = NULL;

	hrtime_t now = gethrtime();
	vq->vq_io_complete_ts = now;
	vq->vq_io_delta_ts = zio->io_delta = now - zio->io_timestamp;

	mutex_enter(&vq->vq_lock);
	vdev_queue_pending_remove(vq, zio);

	while ((nio = vdev_queue_io_to_issue(vq)) != NULL) {
	mutex_exit(&vq->vq_lock);
	if (nio->io_done == vdev_queue_agg_io_done) {
	while ((dio = zio_walk_parents(nio, &zl)) != NULL) {
	ASSERT3U(dio->io_type, ==, nio->io_type);
	zio_vdev_io_bypass(dio);
	zio_execute(dio);
	}
	zio_nowait(nio);
	} else {
	zio_vdev_io_reissue(nio);
	zio_execute(nio);
	}
	mutex_enter(&vq->vq_lock);
	}

	mutex_exit(&vq->vq_lock);
	}

	void
	vdev_queue_change_io_priority(zio_t *zio, zio_priority_t priority)
	{
	vdev_queue_t *vq = &zio->io_vd->vdev_queue;
	avl_tree_t *tree;

	/*
	* ZIO_PRIORITY_NOW is used by the vdev cache code and the aggregate zio
	* code to issue IOs without adding them to the vdev queue. In this
	* case, the zio is already going to be issued as quickly as possible
	* and so it doesn't need any reprioritization to help.
	*/
	if (zio->io_priority == ZIO_PRIORITY_NOW)
	return;

	ASSERT3U(zio->io_priority, <, ZIO_PRIORITY_NUM_QUEUEABLE);
	ASSERT3U(priority, <, ZIO_PRIORITY_NUM_QUEUEABLE);

	if (zio->io_type == ZIO_TYPE_READ) {
	if (priority != ZIO_PRIORITY_SYNC_READ &&
	priority != ZIO_PRIORITY_ASYNC_READ &&
	priority != ZIO_PRIORITY_SCRUB)
	priority = ZIO_PRIORITY_ASYNC_READ;
	} else {
	ASSERT(zio->io_type == ZIO_TYPE_WRITE);
	if (priority != ZIO_PRIORITY_SYNC_WRITE &&
	priority != ZIO_PRIORITY_ASYNC_WRITE)
	priority = ZIO_PRIORITY_ASYNC_WRITE;
	}

	mutex_enter(&vq->vq_lock);

	/*
	* If the zio is in none of the queues we can simply change
	* the priority. If the zio is waiting to be submitted we must
	* remove it from the queue and re-insert it with the new priority.
	* Otherwise, the zio is currently active and we cannot change its
	* priority.
	*/
	tree = vdev_queue_class_tree(vq, zio->io_priority);
	if (avl_find(tree, zio, NULL) == zio) {
	avl_remove(vdev_queue_class_tree(vq, zio->io_priority), zio);
	zio->io_priority = priority;
	avl_add(vdev_queue_class_tree(vq, zio->io_priority), zio);
	} else if (avl_find(&vq->vq_active_tree, zio, NULL) != zio) {
	zio->io_priority = priority;
	}

	mutex_exit(&vq->vq_lock);
	}

	/*
	* As these two methods are only used for load calculations we're not
	* concerned if we get an incorrect value on 32bit platforms due to lack of
	* vq_lock mutex use here, instead we prefer to keep it lock free for
	* performance.
	*/
	int
	vdev_queue_length(vdev_t *vd)
	{
	return (avl_numnodes(&vd->vdev_queue.vq_active_tree));
	}

	uint64_t
	vdev_queue_last_offset(vdev_t *vd)
	{
	return (vd->vdev_queue.vq_last_offset);
	}

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, aggregation_limit, INT, ZMOD_RW,
	"Max vdev I/O aggregation size");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, aggregation_limit_non_rotating, INT, ZMOD_RW,
	"Max vdev I/O aggregation size for non-rotating media");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, aggregate_trim, INT, ZMOD_RW,
	"Allow TRIM I/O to be aggregated");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, read_gap_limit, INT, ZMOD_RW,
	"Aggregate read I/O over gap");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, write_gap_limit, INT, ZMOD_RW,
	"Aggregate write I/O over gap");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, max_active, INT, ZMOD_RW,
	"Maximum number of active I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, async_write_active_max_dirty_percent, INT, ZMOD_RW,
	"Async write concurrency max threshold");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, async_write_active_min_dirty_percent, INT, ZMOD_RW,
	"Async write concurrency min threshold");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, async_read_max_active, INT, ZMOD_RW,
	"Max active async read I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, async_read_min_active, INT, ZMOD_RW,
	"Min active async read I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, async_write_max_active, INT, ZMOD_RW,
	"Max active async write I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, async_write_min_active, INT, ZMOD_RW,
	"Min active async write I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, initializing_max_active, INT, ZMOD_RW,
	"Max active initializing I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, initializing_min_active, INT, ZMOD_RW,
	"Min active initializing I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, removal_max_active, INT, ZMOD_RW,
	"Max active removal I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, removal_min_active, INT, ZMOD_RW,
	"Min active removal I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, scrub_max_active, INT, ZMOD_RW,
	"Max active scrub I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, scrub_min_active, INT, ZMOD_RW,
	"Min active scrub I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, sync_read_max_active, INT, ZMOD_RW,
	"Max active sync read I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, sync_read_min_active, INT, ZMOD_RW,
	"Min active sync read I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, sync_write_max_active, INT, ZMOD_RW,
	"Max active sync write I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, sync_write_min_active, INT, ZMOD_RW,
	"Min active sync write I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, trim_max_active, INT, ZMOD_RW,
	"Max active trim/discard I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, trim_min_active, INT, ZMOD_RW,
	"Min active trim/discard I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, rebuild_max_active, INT, ZMOD_RW,
	"Max active rebuild I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, rebuild_min_active, INT, ZMOD_RW,
	"Min active rebuild I/Os per vdev");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, nia_credit, INT, ZMOD_RW,
	"Number of non-interactive I/Os to allow in sequence");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, nia_delay, INT, ZMOD_RW,
	"Number of non-interactive I/Os before _max_active");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_vdev_, queue_depth_pct, INT, ZMOD_RW,
	"Queue depth percentage for each top-level vdev");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/zfs/vdev_removal.c b/sys/contrib/openzfs/module/zfs/vdev_removal.c
	index f762c1df96aa..482b5bb0ba54 100644
	--- a/sys/contrib/openzfs/module/zfs/vdev_removal.c
	+++ b/sys/contrib/openzfs/module/zfs/vdev_removal.c
	@@ -1,2389 +1,2570 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2019, loli10K <ezomori.nozomu@gmail.com>. All rights reserved.
	*/

	#include <sys/zfs_context.h>
	#include <sys/spa_impl.h>
	#include <sys/dmu.h>
	#include <sys/dmu_tx.h>
	#include <sys/zap.h>
	#include <sys/vdev_impl.h>
	#include <sys/metaslab.h>
	#include <sys/metaslab_impl.h>
	#include <sys/uberblock_impl.h>
	#include <sys/txg.h>
	#include <sys/avl.h>
	#include <sys/bpobj.h>
	#include <sys/dsl_pool.h>
	#include <sys/dsl_synctask.h>
	#include <sys/dsl_dir.h>
	#include <sys/arc.h>
	#include <sys/zfeature.h>
	#include <sys/vdev_indirect_births.h>
	#include <sys/vdev_indirect_mapping.h>
	#include <sys/abd.h>
	#include <sys/vdev_initialize.h>
	#include <sys/vdev_trim.h>
	#include <sys/trace_zfs.h>

	/*
	* This file contains the necessary logic to remove vdevs from a
	* storage pool. Currently, the only devices that can be removed
	* are log, cache, and spare devices; and top level vdevs from a pool
	* w/o raidz or mirrors. (Note that members of a mirror can be removed
	* by the detach operation.)
	*
	* Log vdevs are removed by evacuating them and then turning the vdev
	* into a hole vdev while holding spa config locks.
	*
	* Top level vdevs are removed and converted into an indirect vdev via
	* a multi-step process:
	*
	* - Disable allocations from this device (spa_vdev_remove_top).
	*
	* - From a new thread (spa_vdev_remove_thread), copy data from
	* the removing vdev to a different vdev. The copy happens in open
	* context (spa_vdev_copy_impl) and issues a sync task
	* (vdev_mapping_sync) so the sync thread can update the partial
	* indirect mappings in core and on disk.
	*
	* - If a free happens during a removal, it is freed from the
	* removing vdev, and if it has already been copied, from the new
	* location as well (free_from_removing_vdev).
	*
	* - After the removal is completed, the copy thread converts the vdev
	* into an indirect vdev (vdev_remove_complete) before instructing
	* the sync thread to destroy the space maps and finish the removal
	* (spa_finish_removal).
	*/

	typedef struct vdev_copy_arg {
	metaslab_t *vca_msp;
	uint64_t vca_outstanding_bytes;
	uint64_t vca_read_error_bytes;
	uint64_t vca_write_error_bytes;
	kcondvar_t vca_cv;
	kmutex_t vca_lock;
	} vdev_copy_arg_t;

	/*
	* The maximum amount of memory we can use for outstanding i/o while
	* doing a device removal. This determines how much i/o we can have
	* in flight concurrently.
	*/
	int zfs_remove_max_copy_bytes = 64 * 1024 * 1024;

	/*
	* The largest contiguous segment that we will attempt to allocate when
	* removing a device. This can be no larger than SPA_MAXBLOCKSIZE. If
	* there is a performance problem with attempting to allocate large blocks,
	* consider decreasing this.
	*
	* See also the accessor function spa_remove_max_segment().
	*/
	int zfs_remove_max_segment = SPA_MAXBLOCKSIZE;

	/*
	* Ignore hard IO errors during device removal. When set if a device
	* encounters hard IO error during the removal process the removal will
	* not be cancelled. This can result in a normally recoverable block
	* becoming permanently damaged and is not recommended.
	*/
	int zfs_removal_ignore_errors = 0;

	/*
	* Allow a remap segment to span free chunks of at most this size. The main
	* impact of a larger span is that we will read and write larger, more
	* contiguous chunks, with more "unnecessary" data -- trading off bandwidth
	* for iops. The value here was chosen to align with
	* zfs_vdev_read_gap_limit, which is a similar concept when doing regular
	* reads (but there's no reason it has to be the same).
	*
	* Additionally, a higher span will have the following relatively minor
	* effects:
	* - the mapping will be smaller, since one entry can cover more allocated
	* segments
	* - more of the fragmentation in the removing device will be preserved
	* - we'll do larger allocations, which may fail and fall back on smaller
	* allocations
	*/
	int vdev_removal_max_span = 32 * 1024;

	/*
	* This is used by the test suite so that it can ensure that certain
	* actions happen while in the middle of a removal.
	*/
	int zfs_removal_suspend_progress = 0;

	#define VDEV_REMOVAL_ZAP_OBJS "lzap"

	static void spa_vdev_remove_thread(void *arg);
	static int spa_vdev_remove_cancel_impl(spa_t *spa);

	static void
	spa_sync_removing_state(spa_t spa, dmu_tx_t tx)
	{
	VERIFY0(zap_update(spa->spa_dsl_pool->dp_meta_objset,
	DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_REMOVING, sizeof (uint64_t),
	sizeof (spa->spa_removing_phys) / sizeof (uint64_t),
	&spa->spa_removing_phys, tx));
	}

	static nvlist_t *
	spa_nvlist_lookup_by_guid(nvlist_t **nvpp, int count, uint64_t target_guid)
	{
	for (int i = 0; i < count; i++) {
	uint64_t guid =
	fnvlist_lookup_uint64(nvpp[i], ZPOOL_CONFIG_GUID);

	if (guid == target_guid)
	return (nvpp[i]);
	}

	return (NULL);
	}

	+static void
	+vdev_activate(vdev_t *vd)
	+{
	+ metaslab_group_t *mg = vd->vdev_mg;
	+ spa_t *spa = vd->vdev_spa;
	+ uint64_t vdev_space = spa_deflate(spa) ?
	+ vd->vdev_stat.vs_dspace : vd->vdev_stat.vs_space;
	+
	+ ASSERT(!vd->vdev_islog);
	+ ASSERT(vd->vdev_noalloc);
	+
	+ metaslab_group_activate(mg);
	+ metaslab_group_activate(vd->vdev_log_mg);
	+
	+ ASSERT3U(spa->spa_nonallocating_dspace, >=, vdev_space);
	+
	+ spa->spa_nonallocating_dspace -= vdev_space;
	+
	+ vd->vdev_noalloc = B_FALSE;
	+}
	+
	+static int
	+vdev_passivate(vdev_t vd, uint64_t txg)
	+{
	+ spa_t *spa = vd->vdev_spa;
	+ int error;
	+
	+ ASSERT(!vd->vdev_noalloc);
	+
	+ vdev_t *rvd = spa->spa_root_vdev;
	+ metaslab_group_t *mg = vd->vdev_mg;
	+ metaslab_class_t *normal = spa_normal_class(spa);
	+ if (mg->mg_class == normal) {
	+ /*
	+ * We must check that this is not the only allocating device in
	+ * the pool before passivating, otherwise we will not be able
	+ * to make progress because we can't allocate from any vdevs.
	+ */
	+ boolean_t last = B_TRUE;
	+ for (uint64_t id = 0; id < rvd->vdev_children; id++) {
	+ vdev_t *cvd = rvd->vdev_child[id];
	+
	+ if (cvd == vd \|\|
	+ cvd->vdev_ops == &vdev_indirect_ops)
	+ continue;
	+
	+ metaslab_class_t *mc = cvd->vdev_mg->mg_class;
	+ if (mc != normal)
	+ continue;
	+
	+ if (!cvd->vdev_noalloc) {
	+ last = B_FALSE;
	+ break;
	+ }
	+ }
	+ if (last)
	+ return (SET_ERROR(EINVAL));
	+ }
	+
	+ metaslab_group_passivate(mg);
	+ ASSERT(!vd->vdev_islog);
	+ metaslab_group_passivate(vd->vdev_log_mg);
	+
	+ /*
	+ * Wait for the youngest allocations and frees to sync,
	+ * and then wait for the deferral of those frees to finish.
	+ */
	+ spa_vdev_config_exit(spa, NULL,
	+ *txg + TXG_CONCURRENT_STATES + TXG_DEFER_SIZE, 0, FTAG);
	+
	+ /*
	+ * We must ensure that no "stubby" log blocks are allocated
	+ * on the device to be removed. These blocks could be
	+ * written at any time, including while we are in the middle
	+ * of copying them.
	+ */
	+ error = spa_reset_logs(spa);
	+
	+ *txg = spa_vdev_config_enter(spa);
	+
	+ if (error != 0) {
	+ metaslab_group_activate(mg);
	+ ASSERT(!vd->vdev_islog);
	+ if (vd->vdev_log_mg != NULL)
	+ metaslab_group_activate(vd->vdev_log_mg);
	+ return (error);
	+ }
	+
	+ spa->spa_nonallocating_dspace += spa_deflate(spa) ?
	+ vd->vdev_stat.vs_dspace : vd->vdev_stat.vs_space;
	+ vd->vdev_noalloc = B_TRUE;
	+
	+ return (0);
	+}
	+
	+/*
	+ * Turn off allocations for a top-level device from the pool.
	+ *
	+ * Turning off allocations for a top-level device can take a significant
	+ * amount of time. As a result we use the spa_vdev_config_[enter/exit]
	+ * functions which allow us to grab and release the spa_config_lock while
	+ * still holding the namespace lock. During each step the configuration
	+ * is synced out.
	+ */
	+int
	+spa_vdev_noalloc(spa_t *spa, uint64_t guid)
	+{
	+ vdev_t *vd;
	+ uint64_t txg;
	+ int error = 0;
	+
	+ ASSERT(!MUTEX_HELD(&spa_namespace_lock));
	+ ASSERT(spa_writeable(spa));
	+
	+ txg = spa_vdev_enter(spa);
	+
	+ ASSERT(MUTEX_HELD(&spa_namespace_lock));
	+
	+ vd = spa_lookup_by_guid(spa, guid, B_FALSE);
	+
	+ if (vd == NULL)
	+ error = SET_ERROR(ENOENT);
	+ else if (vd->vdev_mg == NULL)
	+ error = SET_ERROR(ZFS_ERR_VDEV_NOTSUP);
	+ else if (!vd->vdev_noalloc)
	+ error = vdev_passivate(vd, &txg);
	+
	+ if (error == 0) {
	+ vdev_dirty_leaves(vd, VDD_DTL, txg);
	+ vdev_config_dirty(vd);
	+ }
	+
	+ error = spa_vdev_exit(spa, NULL, txg, error);
	+
	+ return (error);
	+}
	+
	+int
	+spa_vdev_alloc(spa_t *spa, uint64_t guid)
	+{
	+ vdev_t *vd;
	+ uint64_t txg;
	+ int error = 0;
	+
	+ ASSERT(!MUTEX_HELD(&spa_namespace_lock));
	+ ASSERT(spa_writeable(spa));
	+
	+ txg = spa_vdev_enter(spa);
	+
	+ ASSERT(MUTEX_HELD(&spa_namespace_lock));
	+
	+ vd = spa_lookup_by_guid(spa, guid, B_FALSE);
	+
	+ if (vd == NULL)
	+ error = SET_ERROR(ENOENT);
	+ else if (vd->vdev_mg == NULL)
	+ error = SET_ERROR(ZFS_ERR_VDEV_NOTSUP);
	+ else if (!vd->vdev_removing)
	+ vdev_activate(vd);
	+
	+ if (error == 0) {
	+ vdev_dirty_leaves(vd, VDD_DTL, txg);
	+ vdev_config_dirty(vd);
	+ }
	+
	+ (void) spa_vdev_exit(spa, NULL, txg, error);
	+
	+ return (error);
	+}
	+
	static void
	spa_vdev_remove_aux(nvlist_t config, char name, nvlist_t **dev, int count,
	nvlist_t *dev_to_remove)
	{
	nvlist_t **newdev = NULL;

	if (count > 1)
	newdev = kmem_alloc((count - 1) * sizeof (void *), KM_SLEEP);

	for (int i = 0, j = 0; i < count; i++) {
	if (dev[i] == dev_to_remove)
	continue;
	VERIFY(nvlist_dup(dev[i], &newdev[j++], KM_SLEEP) == 0);
	}

	VERIFY(nvlist_remove(config, name, DATA_TYPE_NVLIST_ARRAY) == 0);
	- VERIFY(nvlist_add_nvlist_array(config, name, newdev, count - 1) == 0);
	+ fnvlist_add_nvlist_array(config, name, (const nvlist_t * const *)newdev,
	+ count - 1);

	for (int i = 0; i < count - 1; i++)
	nvlist_free(newdev[i]);

	if (count > 1)
	kmem_free(newdev, (count - 1) * sizeof (void *));
	}

	static spa_vdev_removal_t *
	spa_vdev_removal_create(vdev_t *vd)
	{
	spa_vdev_removal_t svr = kmem_zalloc(sizeof (svr), KM_SLEEP);
	mutex_init(&svr->svr_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&svr->svr_cv, NULL, CV_DEFAULT, NULL);
	svr->svr_allocd_segs = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0);
	svr->svr_vdev_id = vd->vdev_id;

	for (int i = 0; i < TXG_SIZE; i++) {
	svr->svr_frees[i] = range_tree_create(NULL, RANGE_SEG64, NULL,
	0, 0);
	list_create(&svr->svr_new_segments[i],
	sizeof (vdev_indirect_mapping_entry_t),
	offsetof(vdev_indirect_mapping_entry_t, vime_node));
	}

	return (svr);
	}

	void
	spa_vdev_removal_destroy(spa_vdev_removal_t *svr)
	{
	for (int i = 0; i < TXG_SIZE; i++) {
	ASSERT0(svr->svr_bytes_done[i]);
	ASSERT0(svr->svr_max_offset_to_sync[i]);
	range_tree_destroy(svr->svr_frees[i]);
	list_destroy(&svr->svr_new_segments[i]);
	}

	range_tree_destroy(svr->svr_allocd_segs);
	mutex_destroy(&svr->svr_lock);
	cv_destroy(&svr->svr_cv);
	kmem_free(svr, sizeof (*svr));
	}

	/*
	* This is called as a synctask in the txg in which we will mark this vdev
	* as removing (in the config stored in the MOS).
	*
	* It begins the evacuation of a toplevel vdev by:
	* - initializing the spa_removing_phys which tracks this removal
	* - computing the amount of space to remove for accounting purposes
	* - dirtying all dbufs in the spa_config_object
	* - creating the spa_vdev_removal
	* - starting the spa_vdev_remove_thread
	*/
	static void
	vdev_remove_initiate_sync(void arg, dmu_tx_t tx)
	{
	int vdev_id = (uintptr_t)arg;
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;
	vdev_t *vd = vdev_lookup_top(spa, vdev_id);
	vdev_indirect_config_t *vic = &vd->vdev_indirect_config;
	objset_t *mos = spa->spa_dsl_pool->dp_meta_objset;
	spa_vdev_removal_t *svr = NULL;
	uint64_t txg __maybe_unused = dmu_tx_get_txg(tx);

	ASSERT0(vdev_get_nparity(vd));
	svr = spa_vdev_removal_create(vd);

	ASSERT(vd->vdev_removing);
	ASSERT3P(vd->vdev_indirect_mapping, ==, NULL);

	spa_feature_incr(spa, SPA_FEATURE_DEVICE_REMOVAL, tx);
	if (spa_feature_is_enabled(spa, SPA_FEATURE_OBSOLETE_COUNTS)) {
	/*
	* By activating the OBSOLETE_COUNTS feature, we prevent
	* the pool from being downgraded and ensure that the
	* refcounts are precise.
	*/
	spa_feature_incr(spa, SPA_FEATURE_OBSOLETE_COUNTS, tx);
	uint64_t one = 1;
	VERIFY0(zap_add(spa->spa_meta_objset, vd->vdev_top_zap,
	VDEV_TOP_ZAP_OBSOLETE_COUNTS_ARE_PRECISE, sizeof (one), 1,
	&one, tx));
	boolean_t are_precise __maybe_unused;
	ASSERT0(vdev_obsolete_counts_are_precise(vd, &are_precise));
	ASSERT3B(are_precise, ==, B_TRUE);
	}

	vic->vic_mapping_object = vdev_indirect_mapping_alloc(mos, tx);
	vd->vdev_indirect_mapping =
	vdev_indirect_mapping_open(mos, vic->vic_mapping_object);
	vic->vic_births_object = vdev_indirect_births_alloc(mos, tx);
	vd->vdev_indirect_births =
	vdev_indirect_births_open(mos, vic->vic_births_object);
	spa->spa_removing_phys.sr_removing_vdev = vd->vdev_id;
	spa->spa_removing_phys.sr_start_time = gethrestime_sec();
	spa->spa_removing_phys.sr_end_time = 0;
	spa->spa_removing_phys.sr_state = DSS_SCANNING;
	spa->spa_removing_phys.sr_to_copy = 0;
	spa->spa_removing_phys.sr_copied = 0;

	/*
	* Note: We can't use vdev_stat's vs_alloc for sr_to_copy, because
	* there may be space in the defer tree, which is free, but still
	* counted in vs_alloc.
	*/
	for (uint64_t i = 0; i < vd->vdev_ms_count; i++) {
	metaslab_t *ms = vd->vdev_ms[i];
	if (ms->ms_sm == NULL)
	continue;

	spa->spa_removing_phys.sr_to_copy +=
	metaslab_allocated_space(ms);

	/*
	* Space which we are freeing this txg does not need to
	* be copied.
	*/
	spa->spa_removing_phys.sr_to_copy -=
	range_tree_space(ms->ms_freeing);

	ASSERT0(range_tree_space(ms->ms_freed));
	for (int t = 0; t < TXG_SIZE; t++)
	ASSERT0(range_tree_space(ms->ms_allocating[t]));
	}

	/*
	* Sync tasks are called before metaslab_sync(), so there should
	* be no already-synced metaslabs in the TXG_CLEAN list.
	*/
	ASSERT3P(txg_list_head(&vd->vdev_ms_list, TXG_CLEAN(txg)), ==, NULL);

	spa_sync_removing_state(spa, tx);

	/*
	* All blocks that we need to read the most recent mapping must be
	* stored on concrete vdevs. Therefore, we must dirty anything that
	* is read before spa_remove_init(). Specifically, the
	* spa_config_object. (Note that although we already modified the
	* spa_config_object in spa_sync_removing_state, that may not have
	* modified all blocks of the object.)
	*/
	dmu_object_info_t doi;
	VERIFY0(dmu_object_info(mos, DMU_POOL_DIRECTORY_OBJECT, &doi));
	for (uint64_t offset = 0; offset < doi.doi_max_offset; ) {
	dmu_buf_t *dbuf;
	VERIFY0(dmu_buf_hold(mos, DMU_POOL_DIRECTORY_OBJECT,
	offset, FTAG, &dbuf, 0));
	dmu_buf_will_dirty(dbuf, tx);
	offset += dbuf->db_size;
	dmu_buf_rele(dbuf, FTAG);
	}

	/*
	* Now that we've allocated the im_object, dirty the vdev to ensure
	* that the object gets written to the config on disk.
	*/
	vdev_config_dirty(vd);

	zfs_dbgmsg("starting removal thread for vdev %llu (%px) in txg %llu "
	"im_obj=%llu", (u_longlong_t)vd->vdev_id, vd,
	(u_longlong_t)dmu_tx_get_txg(tx),
	(u_longlong_t)vic->vic_mapping_object);

	spa_history_log_internal(spa, "vdev remove started", tx,
	"%s vdev %llu %s", spa_name(spa), (u_longlong_t)vd->vdev_id,
	(vd->vdev_path != NULL) ? vd->vdev_path : "-");
	/*
	* Setting spa_vdev_removal causes subsequent frees to call
	* free_from_removing_vdev(). Note that we don't need any locking
	* because we are the sync thread, and metaslab_free_impl() is only
	* called from syncing context (potentially from a zio taskq thread,
	* but in any case only when there are outstanding free i/os, which
	* there are not).
	*/
	ASSERT3P(spa->spa_vdev_removal, ==, NULL);
	spa->spa_vdev_removal = svr;
	svr->svr_thread = thread_create(NULL, 0,
	spa_vdev_remove_thread, spa, 0, &p0, TS_RUN, minclsyspri);
	}

	/*
	* When we are opening a pool, we must read the mapping for each
	* indirect vdev in order from most recently removed to least
	* recently removed. We do this because the blocks for the mapping
	* of older indirect vdevs may be stored on more recently removed vdevs.
	* In order to read each indirect mapping object, we must have
	* initialized all more recently removed vdevs.
	*/
	int
	spa_remove_init(spa_t *spa)
	{
	int error;

	error = zap_lookup(spa->spa_dsl_pool->dp_meta_objset,
	DMU_POOL_DIRECTORY_OBJECT,
	DMU_POOL_REMOVING, sizeof (uint64_t),
	sizeof (spa->spa_removing_phys) / sizeof (uint64_t),
	&spa->spa_removing_phys);

	if (error == ENOENT) {
	spa->spa_removing_phys.sr_state = DSS_NONE;
	spa->spa_removing_phys.sr_removing_vdev = -1;
	spa->spa_removing_phys.sr_prev_indirect_vdev = -1;
	spa->spa_indirect_vdevs_loaded = B_TRUE;
	return (0);
	} else if (error != 0) {
	return (error);
	}

	if (spa->spa_removing_phys.sr_state == DSS_SCANNING) {
	/*
	* We are currently removing a vdev. Create and
	* initialize a spa_vdev_removal_t from the bonus
	* buffer of the removing vdevs vdev_im_object, and
	* initialize its partial mapping.
	*/
	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
	vdev_t *vd = vdev_lookup_top(spa,
	spa->spa_removing_phys.sr_removing_vdev);

	if (vd == NULL) {
	spa_config_exit(spa, SCL_STATE, FTAG);
	return (EINVAL);
	}

	vdev_indirect_config_t *vic = &vd->vdev_indirect_config;

	ASSERT(vdev_is_concrete(vd));
	spa_vdev_removal_t *svr = spa_vdev_removal_create(vd);
	ASSERT3U(svr->svr_vdev_id, ==, vd->vdev_id);
	ASSERT(vd->vdev_removing);

	vd->vdev_indirect_mapping = vdev_indirect_mapping_open(
	spa->spa_meta_objset, vic->vic_mapping_object);
	vd->vdev_indirect_births = vdev_indirect_births_open(
	spa->spa_meta_objset, vic->vic_births_object);
	spa_config_exit(spa, SCL_STATE, FTAG);

	spa->spa_vdev_removal = svr;
	}

	spa_config_enter(spa, SCL_STATE, FTAG, RW_READER);
	uint64_t indirect_vdev_id =
	spa->spa_removing_phys.sr_prev_indirect_vdev;
	while (indirect_vdev_id != UINT64_MAX) {
	vdev_t *vd = vdev_lookup_top(spa, indirect_vdev_id);
	vdev_indirect_config_t *vic = &vd->vdev_indirect_config;

	ASSERT3P(vd->vdev_ops, ==, &vdev_indirect_ops);
	vd->vdev_indirect_mapping = vdev_indirect_mapping_open(
	spa->spa_meta_objset, vic->vic_mapping_object);
	vd->vdev_indirect_births = vdev_indirect_births_open(
	spa->spa_meta_objset, vic->vic_births_object);

	indirect_vdev_id = vic->vic_prev_indirect_vdev;
	}
	spa_config_exit(spa, SCL_STATE, FTAG);

	/*
	* Now that we've loaded all the indirect mappings, we can allow
	* reads from other blocks (e.g. via predictive prefetch).
	*/
	spa->spa_indirect_vdevs_loaded = B_TRUE;
	return (0);
	}

	void
	spa_restart_removal(spa_t *spa)
	{
	spa_vdev_removal_t *svr = spa->spa_vdev_removal;

	if (svr == NULL)
	return;

	/*
	* In general when this function is called there is no
	* removal thread running. The only scenario where this
	* is not true is during spa_import() where this function
	* is called twice [once from spa_import_impl() and
	* spa_async_resume()]. Thus, in the scenario where we
	* import a pool that has an ongoing removal we don't
	* want to spawn a second thread.
	*/
	if (svr->svr_thread != NULL)
	return;

	if (!spa_writeable(spa))
	return;

	zfs_dbgmsg("restarting removal of %llu",
	(u_longlong_t)svr->svr_vdev_id);
	svr->svr_thread = thread_create(NULL, 0, spa_vdev_remove_thread, spa,
	0, &p0, TS_RUN, minclsyspri);
	}

	/*
	* Process freeing from a device which is in the middle of being removed.
	* We must handle this carefully so that we attempt to copy freed data,
	* and we correctly free already-copied data.
	*/
	void
	free_from_removing_vdev(vdev_t *vd, uint64_t offset, uint64_t size)
	{
	spa_t *spa = vd->vdev_spa;
	spa_vdev_removal_t *svr = spa->spa_vdev_removal;
	vdev_indirect_mapping_t *vim = vd->vdev_indirect_mapping;
	uint64_t txg = spa_syncing_txg(spa);
	uint64_t max_offset_yet = 0;

	ASSERT(vd->vdev_indirect_config.vic_mapping_object != 0);
	ASSERT3U(vd->vdev_indirect_config.vic_mapping_object, ==,
	vdev_indirect_mapping_object(vim));
	ASSERT3U(vd->vdev_id, ==, svr->svr_vdev_id);

	mutex_enter(&svr->svr_lock);

	/*
	* Remove the segment from the removing vdev's spacemap. This
	* ensures that we will not attempt to copy this space (if the
	* removal thread has not yet visited it), and also ensures
	* that we know what is actually allocated on the new vdevs
	* (needed if we cancel the removal).
	*
	* Note: we must do the metaslab_free_concrete() with the svr_lock
	* held, so that the remove_thread can not load this metaslab and then
	* visit this offset between the time that we metaslab_free_concrete()
	* and when we check to see if it has been visited.
	*
	* Note: The checkpoint flag is set to false as having/taking
	* a checkpoint and removing a device can't happen at the same
	* time.
	*/
	ASSERT(!spa_has_checkpoint(spa));
	metaslab_free_concrete(vd, offset, size, B_FALSE);

	uint64_t synced_size = 0;
	uint64_t synced_offset = 0;
	uint64_t max_offset_synced = vdev_indirect_mapping_max_offset(vim);
	if (offset < max_offset_synced) {
	/*
	* The mapping for this offset is already on disk.
	* Free from the new location.
	*
	* Note that we use svr_max_synced_offset because it is
	* updated atomically with respect to the in-core mapping.
	* By contrast, vim_max_offset is not.
	*
	* This block may be split between a synced entry and an
	* in-flight or unvisited entry. Only process the synced
	* portion of it here.
	*/
	synced_size = MIN(size, max_offset_synced - offset);
	synced_offset = offset;

	ASSERT3U(max_offset_yet, <=, max_offset_synced);
	max_offset_yet = max_offset_synced;

	DTRACE_PROBE3(remove__free__synced,
	spa_t *, spa,
	uint64_t, offset,
	uint64_t, synced_size);

	size -= synced_size;
	offset += synced_size;
	}

	/*
	* Look at all in-flight txgs starting from the currently syncing one
	* and see if a section of this free is being copied. By starting from
	* this txg and iterating forward, we might find that this region
	* was copied in two different txgs and handle it appropriately.
	*/
	for (int i = 0; i < TXG_CONCURRENT_STATES; i++) {
	int txgoff = (txg + i) & TXG_MASK;
	if (size > 0 && offset < svr->svr_max_offset_to_sync[txgoff]) {
	/*
	* The mapping for this offset is in flight, and
	* will be synced in txg+i.
	*/
	uint64_t inflight_size = MIN(size,
	svr->svr_max_offset_to_sync[txgoff] - offset);

	DTRACE_PROBE4(remove__free__inflight,
	spa_t *, spa,
	uint64_t, offset,
	uint64_t, inflight_size,
	uint64_t, txg + i);

	/*
	* We copy data in order of increasing offset.
	* Therefore the max_offset_to_sync[] must increase
	* (or be zero, indicating that nothing is being
	* copied in that txg).
	*/
	if (svr->svr_max_offset_to_sync[txgoff] != 0) {
	ASSERT3U(svr->svr_max_offset_to_sync[txgoff],
	>=, max_offset_yet);
	max_offset_yet =
	svr->svr_max_offset_to_sync[txgoff];
	}

	/*
	* We've already committed to copying this segment:
	* we have allocated space elsewhere in the pool for
	* it and have an IO outstanding to copy the data. We
	* cannot free the space before the copy has
	* completed, or else the copy IO might overwrite any
	* new data. To free that space, we record the
	* segment in the appropriate svr_frees tree and free
	* the mapped space later, in the txg where we have
	* completed the copy and synced the mapping (see
	* vdev_mapping_sync).
	*/
	range_tree_add(svr->svr_frees[txgoff],
	offset, inflight_size);
	size -= inflight_size;
	offset += inflight_size;

	/*
	* This space is already accounted for as being
	* done, because it is being copied in txg+i.
	* However, if i!=0, then it is being copied in
	* a future txg. If we crash after this txg
	* syncs but before txg+i syncs, then the space
	* will be free. Therefore we must account
	* for the space being done in this txg
	* (when it is freed) rather than the future txg
	* (when it will be copied).
	*/
	ASSERT3U(svr->svr_bytes_done[txgoff], >=,
	inflight_size);
	svr->svr_bytes_done[txgoff] -= inflight_size;
	svr->svr_bytes_done[txg & TXG_MASK] += inflight_size;
	}
	}
	ASSERT0(svr->svr_max_offset_to_sync[TXG_CLEAN(txg) & TXG_MASK]);

	if (size > 0) {
	/*
	* The copy thread has not yet visited this offset. Ensure
	* that it doesn't.
	*/

	DTRACE_PROBE3(remove__free__unvisited,
	spa_t *, spa,
	uint64_t, offset,
	uint64_t, size);

	if (svr->svr_allocd_segs != NULL)
	range_tree_clear(svr->svr_allocd_segs, offset, size);

	/*
	* Since we now do not need to copy this data, for
	* accounting purposes we have done our job and can count
	* it as completed.
	*/
	svr->svr_bytes_done[txg & TXG_MASK] += size;
	}
	mutex_exit(&svr->svr_lock);

	/*
	* Now that we have dropped svr_lock, process the synced portion
	* of this free.
	*/
	if (synced_size > 0) {
	vdev_indirect_mark_obsolete(vd, synced_offset, synced_size);

	/*
	* Note: this can only be called from syncing context,
	* and the vdev_indirect_mapping is only changed from the
	* sync thread, so we don't need svr_lock while doing
	* metaslab_free_impl_cb.
	*/
	boolean_t checkpoint = B_FALSE;
	vdev_indirect_ops.vdev_op_remap(vd, synced_offset, synced_size,
	metaslab_free_impl_cb, &checkpoint);
	}
	}

	/*
	* Stop an active removal and update the spa_removing phys.
	*/
	static void
	spa_finish_removal(spa_t spa, dsl_scan_state_t state, dmu_tx_t tx)
	{
	spa_vdev_removal_t *svr = spa->spa_vdev_removal;
	ASSERT3U(dmu_tx_get_txg(tx), ==, spa_syncing_txg(spa));

	/* Ensure the removal thread has completed before we free the svr. */
	spa_vdev_remove_suspend(spa);

	ASSERT(state == DSS_FINISHED \|\| state == DSS_CANCELED);

	if (state == DSS_FINISHED) {
	spa_removing_phys_t *srp = &spa->spa_removing_phys;
	vdev_t *vd = vdev_lookup_top(spa, svr->svr_vdev_id);
	vdev_indirect_config_t *vic = &vd->vdev_indirect_config;

	if (srp->sr_prev_indirect_vdev != -1) {
	vdev_t *pvd;
	pvd = vdev_lookup_top(spa,
	srp->sr_prev_indirect_vdev);
	ASSERT3P(pvd->vdev_ops, ==, &vdev_indirect_ops);
	}

	vic->vic_prev_indirect_vdev = srp->sr_prev_indirect_vdev;
	srp->sr_prev_indirect_vdev = vd->vdev_id;
	}
	spa->spa_removing_phys.sr_state = state;
	spa->spa_removing_phys.sr_end_time = gethrestime_sec();

	spa->spa_vdev_removal = NULL;
	spa_vdev_removal_destroy(svr);

	spa_sync_removing_state(spa, tx);
	spa_notify_waiters(spa);

	vdev_config_dirty(spa->spa_root_vdev);
	}

	static void
	free_mapped_segment_cb(void *arg, uint64_t offset, uint64_t size)
	{
	vdev_t *vd = arg;
	vdev_indirect_mark_obsolete(vd, offset, size);
	boolean_t checkpoint = B_FALSE;
	vdev_indirect_ops.vdev_op_remap(vd, offset, size,
	metaslab_free_impl_cb, &checkpoint);
	}

	/*
	* On behalf of the removal thread, syncs an incremental bit more of
	* the indirect mapping to disk and updates the in-memory mapping.
	* Called as a sync task in every txg that the removal thread makes progress.
	*/
	static void
	vdev_mapping_sync(void arg, dmu_tx_t tx)
	{
	spa_vdev_removal_t *svr = arg;
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;
	vdev_t *vd = vdev_lookup_top(spa, svr->svr_vdev_id);
	vdev_indirect_config_t *vic __maybe_unused = &vd->vdev_indirect_config;
	uint64_t txg = dmu_tx_get_txg(tx);
	vdev_indirect_mapping_t *vim = vd->vdev_indirect_mapping;

	ASSERT(vic->vic_mapping_object != 0);
	ASSERT3U(txg, ==, spa_syncing_txg(spa));

	vdev_indirect_mapping_add_entries(vim,
	&svr->svr_new_segments[txg & TXG_MASK], tx);
	vdev_indirect_births_add_entry(vd->vdev_indirect_births,
	vdev_indirect_mapping_max_offset(vim), dmu_tx_get_txg(tx), tx);

	/*
	* Free the copied data for anything that was freed while the
	* mapping entries were in flight.
	*/
	mutex_enter(&svr->svr_lock);
	range_tree_vacate(svr->svr_frees[txg & TXG_MASK],
	free_mapped_segment_cb, vd);
	ASSERT3U(svr->svr_max_offset_to_sync[txg & TXG_MASK], >=,
	vdev_indirect_mapping_max_offset(vim));
	svr->svr_max_offset_to_sync[txg & TXG_MASK] = 0;
	mutex_exit(&svr->svr_lock);

	spa_sync_removing_state(spa, tx);
	}

	typedef struct vdev_copy_segment_arg {
	spa_t *vcsa_spa;
	dva_t *vcsa_dest_dva;
	uint64_t vcsa_txg;
	range_tree_t *vcsa_obsolete_segs;
	} vdev_copy_segment_arg_t;

	static void
	unalloc_seg(void *arg, uint64_t start, uint64_t size)
	{
	vdev_copy_segment_arg_t *vcsa = arg;
	spa_t *spa = vcsa->vcsa_spa;
	blkptr_t bp = { { { {0} } } };

	BP_SET_BIRTH(&bp, TXG_INITIAL, TXG_INITIAL);
	BP_SET_LSIZE(&bp, size);
	BP_SET_PSIZE(&bp, size);
	BP_SET_COMPRESS(&bp, ZIO_COMPRESS_OFF);
	BP_SET_CHECKSUM(&bp, ZIO_CHECKSUM_OFF);
	BP_SET_TYPE(&bp, DMU_OT_NONE);
	BP_SET_LEVEL(&bp, 0);
	BP_SET_DEDUP(&bp, 0);
	BP_SET_BYTEORDER(&bp, ZFS_HOST_BYTEORDER);

	DVA_SET_VDEV(&bp.blk_dva[0], DVA_GET_VDEV(vcsa->vcsa_dest_dva));
	DVA_SET_OFFSET(&bp.blk_dva[0],
	DVA_GET_OFFSET(vcsa->vcsa_dest_dva) + start);
	DVA_SET_ASIZE(&bp.blk_dva[0], size);

	zio_free(spa, vcsa->vcsa_txg, &bp);
	}

	/*
	* All reads and writes associated with a call to spa_vdev_copy_segment()
	* are done.
	*/
	static void
	spa_vdev_copy_segment_done(zio_t *zio)
	{
	vdev_copy_segment_arg_t *vcsa = zio->io_private;

	range_tree_vacate(vcsa->vcsa_obsolete_segs,
	unalloc_seg, vcsa);
	range_tree_destroy(vcsa->vcsa_obsolete_segs);
	kmem_free(vcsa, sizeof (*vcsa));

	spa_config_exit(zio->io_spa, SCL_STATE, zio->io_spa);
	}

	/*
	* The write of the new location is done.
	*/
	static void
	spa_vdev_copy_segment_write_done(zio_t *zio)
	{
	vdev_copy_arg_t *vca = zio->io_private;

	abd_free(zio->io_abd);

	mutex_enter(&vca->vca_lock);
	vca->vca_outstanding_bytes -= zio->io_size;

	if (zio->io_error != 0)
	vca->vca_write_error_bytes += zio->io_size;

	cv_signal(&vca->vca_cv);
	mutex_exit(&vca->vca_lock);
	}

	/*
	* The read of the old location is done. The parent zio is the write to
	* the new location. Allow it to start.
	*/
	static void
	spa_vdev_copy_segment_read_done(zio_t *zio)
	{
	vdev_copy_arg_t *vca = zio->io_private;

	if (zio->io_error != 0) {
	mutex_enter(&vca->vca_lock);
	vca->vca_read_error_bytes += zio->io_size;
	mutex_exit(&vca->vca_lock);
	}

	zio_nowait(zio_unique_parent(zio));
	}

	/*
	* If the old and new vdevs are mirrors, we will read both sides of the old
	* mirror, and write each copy to the corresponding side of the new mirror.
	* If the old and new vdevs have a different number of children, we will do
	* this as best as possible. Since we aren't verifying checksums, this
	* ensures that as long as there's a good copy of the data, we'll have a
	* good copy after the removal, even if there's silent damage to one side
	* of the mirror. If we're removing a mirror that has some silent damage,
	* we'll have exactly the same damage in the new location (assuming that
	* the new location is also a mirror).
	*
	* We accomplish this by creating a tree of zio_t's, with as many writes as
	* there are "children" of the new vdev (a non-redundant vdev counts as one
	* child, a 2-way mirror has 2 children, etc). Each write has an associated
	* read from a child of the old vdev. Typically there will be the same
	* number of children of the old and new vdevs. However, if there are more
	* children of the new vdev, some child(ren) of the old vdev will be issued
	* multiple reads. If there are more children of the old vdev, some copies
	* will be dropped.
	*
	* For example, the tree of zio_t's for a 2-way mirror is:
	*
	* null
	* / \
	* write(new vdev, child 0) write(new vdev, child 1)
	* \| \|
	* read(old vdev, child 0) read(old vdev, child 1)
	*
	* Child zio's complete before their parents complete. However, zio's
	* created with zio_vdev_child_io() may be issued before their children
	* complete. In this case we need to make sure that the children (reads)
	* complete before the parents (writes) are issued. We do this by not
	* calling zio_nowait() on each write until its corresponding read has
	* completed.
	*
	* The spa_config_lock must be held while zio's created by
	* zio_vdev_child_io() are in progress, to ensure that the vdev tree does
	* not change (e.g. due to a concurrent "zpool attach/detach"). The "null"
	* zio is needed to release the spa_config_lock after all the reads and
	* writes complete. (Note that we can't grab the config lock for each read,
	* because it is not reentrant - we could deadlock with a thread waiting
	* for a write lock.)
	*/
	static void
	spa_vdev_copy_one_child(vdev_copy_arg_t vca, zio_t nzio,
	vdev_t *source_vd, uint64_t source_offset,
	vdev_t *dest_child_vd, uint64_t dest_offset, int dest_id, uint64_t size)
	{
	ASSERT3U(spa_config_held(nzio->io_spa, SCL_ALL, RW_READER), !=, 0);

	/*
	* If the destination child in unwritable then there is no point
	* in issuing the source reads which cannot be written.
	*/
	if (!vdev_writeable(dest_child_vd))
	return;

	mutex_enter(&vca->vca_lock);
	vca->vca_outstanding_bytes += size;
	mutex_exit(&vca->vca_lock);

	abd_t *abd = abd_alloc_for_io(size, B_FALSE);

	vdev_t *source_child_vd = NULL;
	if (source_vd->vdev_ops == &vdev_mirror_ops && dest_id != -1) {
	/*
	* Source and dest are both mirrors. Copy from the same
	* child id as we are copying to (wrapping around if there
	* are more dest children than source children). If the
	* preferred source child is unreadable select another.
	*/
	for (int i = 0; i < source_vd->vdev_children; i++) {
	source_child_vd = source_vd->vdev_child[
	(dest_id + i) % source_vd->vdev_children];
	if (vdev_readable(source_child_vd))
	break;
	}
	} else {
	source_child_vd = source_vd;
	}

	/*
	* There should always be at least one readable source child or
	* the pool would be in a suspended state. Somehow selecting an
	* unreadable child would result in IO errors, the removal process
	* being cancelled, and the pool reverting to its pre-removal state.
	*/
	ASSERT3P(source_child_vd, !=, NULL);

	zio_t *write_zio = zio_vdev_child_io(nzio, NULL,
	dest_child_vd, dest_offset, abd, size,
	ZIO_TYPE_WRITE, ZIO_PRIORITY_REMOVAL,
	ZIO_FLAG_CANFAIL,
	spa_vdev_copy_segment_write_done, vca);

	zio_nowait(zio_vdev_child_io(write_zio, NULL,
	source_child_vd, source_offset, abd, size,
	ZIO_TYPE_READ, ZIO_PRIORITY_REMOVAL,
	ZIO_FLAG_CANFAIL,
	spa_vdev_copy_segment_read_done, vca));
	}

	/*
	* Allocate a new location for this segment, and create the zio_t's to
	* read from the old location and write to the new location.
	*/
	static int
	spa_vdev_copy_segment(vdev_t vd, range_tree_t segs,
	uint64_t maxalloc, uint64_t txg,
	vdev_copy_arg_t vca, zio_alloc_list_t zal)
	{
	metaslab_group_t *mg = vd->vdev_mg;
	spa_t *spa = vd->vdev_spa;
	spa_vdev_removal_t *svr = spa->spa_vdev_removal;
	vdev_indirect_mapping_entry_t *entry;
	dva_t dst = {{ 0 }};
	uint64_t start = range_tree_min(segs);
	ASSERT0(P2PHASE(start, 1 << spa->spa_min_ashift));

	ASSERT3U(maxalloc, <=, SPA_MAXBLOCKSIZE);
	ASSERT0(P2PHASE(maxalloc, 1 << spa->spa_min_ashift));

	uint64_t size = range_tree_span(segs);
	if (range_tree_span(segs) > maxalloc) {
	/*
	* We can't allocate all the segments. Prefer to end
	* the allocation at the end of a segment, thus avoiding
	* additional split blocks.
	*/
	range_seg_max_t search;
	zfs_btree_index_t where;
	rs_set_start(&search, segs, start + maxalloc);
	rs_set_end(&search, segs, start + maxalloc);
	(void) zfs_btree_find(&segs->rt_root, &search, &where);
	range_seg_t *rs = zfs_btree_prev(&segs->rt_root, &where,
	&where);
	if (rs != NULL) {
	size = rs_get_end(rs, segs) - start;
	} else {
	/*
	* There are no segments that end before maxalloc.
	* I.e. the first segment is larger than maxalloc,
	* so we must split it.
	*/
	size = maxalloc;
	}
	}
	ASSERT3U(size, <=, maxalloc);
	ASSERT0(P2PHASE(size, 1 << spa->spa_min_ashift));

	/*
	* An allocation class might not have any remaining vdevs or space
	*/
	metaslab_class_t *mc = mg->mg_class;
	if (mc->mc_groups == 0)
	mc = spa_normal_class(spa);
	int error = metaslab_alloc_dva(spa, mc, size, &dst, 0, NULL, txg, 0,
	zal, 0);
	if (error == ENOSPC && mc != spa_normal_class(spa)) {
	error = metaslab_alloc_dva(spa, spa_normal_class(spa), size,
	&dst, 0, NULL, txg, 0, zal, 0);
	}
	if (error != 0)
	return (error);

	/*
	* Determine the ranges that are not actually needed. Offsets are
	* relative to the start of the range to be copied (i.e. relative to the
	* local variable "start").
	*/
	range_tree_t *obsolete_segs = range_tree_create(NULL, RANGE_SEG64, NULL,
	0, 0);

	zfs_btree_index_t where;
	range_seg_t *rs = zfs_btree_first(&segs->rt_root, &where);
	ASSERT3U(rs_get_start(rs, segs), ==, start);
	uint64_t prev_seg_end = rs_get_end(rs, segs);
	while ((rs = zfs_btree_next(&segs->rt_root, &where, &where)) != NULL) {
	if (rs_get_start(rs, segs) >= start + size) {
	break;
	} else {
	range_tree_add(obsolete_segs,
	prev_seg_end - start,
	rs_get_start(rs, segs) - prev_seg_end);
	}
	prev_seg_end = rs_get_end(rs, segs);
	}
	/* We don't end in the middle of an obsolete range */
	ASSERT3U(start + size, <=, prev_seg_end);

	range_tree_clear(segs, start, size);

	/*
	* We can't have any padding of the allocated size, otherwise we will
	* misunderstand what's allocated, and the size of the mapping. We
	* prevent padding by ensuring that all devices in the pool have the
	* same ashift, and the allocation size is a multiple of the ashift.
	*/
	VERIFY3U(DVA_GET_ASIZE(&dst), ==, size);

	entry = kmem_zalloc(sizeof (vdev_indirect_mapping_entry_t), KM_SLEEP);
	DVA_MAPPING_SET_SRC_OFFSET(&entry->vime_mapping, start);
	entry->vime_mapping.vimep_dst = dst;
	if (spa_feature_is_enabled(spa, SPA_FEATURE_OBSOLETE_COUNTS)) {
	entry->vime_obsolete_count = range_tree_space(obsolete_segs);
	}

	vdev_copy_segment_arg_t vcsa = kmem_zalloc(sizeof (vcsa), KM_SLEEP);
	vcsa->vcsa_dest_dva = &entry->vime_mapping.vimep_dst;
	vcsa->vcsa_obsolete_segs = obsolete_segs;
	vcsa->vcsa_spa = spa;
	vcsa->vcsa_txg = txg;

	/*
	* See comment before spa_vdev_copy_one_child().
	*/
	spa_config_enter(spa, SCL_STATE, spa, RW_READER);
	zio_t *nzio = zio_null(spa->spa_txg_zio[txg & TXG_MASK], spa, NULL,
	spa_vdev_copy_segment_done, vcsa, 0);
	vdev_t *dest_vd = vdev_lookup_top(spa, DVA_GET_VDEV(&dst));
	if (dest_vd->vdev_ops == &vdev_mirror_ops) {
	for (int i = 0; i < dest_vd->vdev_children; i++) {
	vdev_t *child = dest_vd->vdev_child[i];
	spa_vdev_copy_one_child(vca, nzio, vd, start,
	child, DVA_GET_OFFSET(&dst), i, size);
	}
	} else {
	spa_vdev_copy_one_child(vca, nzio, vd, start,
	dest_vd, DVA_GET_OFFSET(&dst), -1, size);
	}
	zio_nowait(nzio);

	list_insert_tail(&svr->svr_new_segments[txg & TXG_MASK], entry);
	ASSERT3U(start + size, <=, vd->vdev_ms_count << vd->vdev_ms_shift);
	vdev_dirty(vd, 0, NULL, txg);

	return (0);
	}

	/*
	* Complete the removal of a toplevel vdev. This is called as a
	* synctask in the same txg that we will sync out the new config (to the
	* MOS object) which indicates that this vdev is indirect.
	*/
	static void
	vdev_remove_complete_sync(void arg, dmu_tx_t tx)
	{
	spa_vdev_removal_t *svr = arg;
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;
	vdev_t *vd = vdev_lookup_top(spa, svr->svr_vdev_id);

	ASSERT3P(vd->vdev_ops, ==, &vdev_indirect_ops);

	for (int i = 0; i < TXG_SIZE; i++) {
	ASSERT0(svr->svr_bytes_done[i]);
	}

	ASSERT3U(spa->spa_removing_phys.sr_copied, ==,
	spa->spa_removing_phys.sr_to_copy);

	vdev_destroy_spacemaps(vd, tx);

	/* destroy leaf zaps, if any */
	ASSERT3P(svr->svr_zaplist, !=, NULL);
	for (nvpair_t *pair = nvlist_next_nvpair(svr->svr_zaplist, NULL);
	pair != NULL;
	pair = nvlist_next_nvpair(svr->svr_zaplist, pair)) {
	vdev_destroy_unlink_zap(vd, fnvpair_value_uint64(pair), tx);
	}
	fnvlist_free(svr->svr_zaplist);

	spa_finish_removal(dmu_tx_pool(tx)->dp_spa, DSS_FINISHED, tx);
	/* vd->vdev_path is not available here */
	spa_history_log_internal(spa, "vdev remove completed", tx,
	"%s vdev %llu", spa_name(spa), (u_longlong_t)vd->vdev_id);
	}

	static void
	vdev_remove_enlist_zaps(vdev_t vd, nvlist_t zlist)
	{
	ASSERT3P(zlist, !=, NULL);
	ASSERT0(vdev_get_nparity(vd));

	if (vd->vdev_leaf_zap != 0) {
	char zkey[32];
	(void) snprintf(zkey, sizeof (zkey), "%s-%llu",
	VDEV_REMOVAL_ZAP_OBJS, (u_longlong_t)vd->vdev_leaf_zap);
	fnvlist_add_uint64(zlist, zkey, vd->vdev_leaf_zap);
	}

	for (uint64_t id = 0; id < vd->vdev_children; id++) {
	vdev_remove_enlist_zaps(vd->vdev_child[id], zlist);
	}
	}

	static void
	vdev_remove_replace_with_indirect(vdev_t *vd, uint64_t txg)
	{
	vdev_t *ivd;
	dmu_tx_t *tx;
	spa_t *spa = vd->vdev_spa;
	spa_vdev_removal_t *svr = spa->spa_vdev_removal;

	/*
	* First, build a list of leaf zaps to be destroyed.
	* This is passed to the sync context thread,
	* which does the actual unlinking.
	*/
	svr->svr_zaplist = fnvlist_alloc();
	vdev_remove_enlist_zaps(vd, svr->svr_zaplist);

	ivd = vdev_add_parent(vd, &vdev_indirect_ops);
	ivd->vdev_removing = 0;

	vd->vdev_leaf_zap = 0;

	vdev_remove_child(ivd, vd);
	vdev_compact_children(ivd);

	ASSERT(!list_link_active(&vd->vdev_state_dirty_node));

	mutex_enter(&svr->svr_lock);
	svr->svr_thread = NULL;
	cv_broadcast(&svr->svr_cv);
	mutex_exit(&svr->svr_lock);

	/* After this, we can not use svr. */
	tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg);
	dsl_sync_task_nowait(spa->spa_dsl_pool,
	vdev_remove_complete_sync, svr, tx);
	dmu_tx_commit(tx);
	}

	/*
	* Complete the removal of a toplevel vdev. This is called in open
	* context by the removal thread after we have copied all vdev's data.
	*/
	static void
	vdev_remove_complete(spa_t *spa)
	{
	uint64_t txg;

	/*
	* Wait for any deferred frees to be synced before we call
	* vdev_metaslab_fini()
	*/
	txg_wait_synced(spa->spa_dsl_pool, 0);
	txg = spa_vdev_enter(spa);
	vdev_t *vd = vdev_lookup_top(spa, spa->spa_vdev_removal->svr_vdev_id);
	ASSERT3P(vd->vdev_initialize_thread, ==, NULL);
	ASSERT3P(vd->vdev_trim_thread, ==, NULL);
	ASSERT3P(vd->vdev_autotrim_thread, ==, NULL);
	+ uint64_t vdev_space = spa_deflate(spa) ?
	+ vd->vdev_stat.vs_dspace : vd->vdev_stat.vs_space;

	sysevent_t *ev = spa_event_create(spa, vd, NULL,
	ESC_ZFS_VDEV_REMOVE_DEV);

	zfs_dbgmsg("finishing device removal for vdev %llu in txg %llu",
	(u_longlong_t)vd->vdev_id, (u_longlong_t)txg);

	+ ASSERT3U(0, !=, vdev_space);
	+ ASSERT3U(spa->spa_nonallocating_dspace, >=, vdev_space);
	+
	+ /* the vdev is no longer part of the dspace */
	+ spa->spa_nonallocating_dspace -= vdev_space;
	+
	/*
	* Discard allocation state.
	*/
	if (vd->vdev_mg != NULL) {
	vdev_metaslab_fini(vd);
	metaslab_group_destroy(vd->vdev_mg);
	vd->vdev_mg = NULL;
	spa_log_sm_set_blocklimit(spa);
	}
	if (vd->vdev_log_mg != NULL) {
	ASSERT0(vd->vdev_ms_count);
	metaslab_group_destroy(vd->vdev_log_mg);
	vd->vdev_log_mg = NULL;
	}
	ASSERT0(vd->vdev_stat.vs_space);
	ASSERT0(vd->vdev_stat.vs_dspace);

	vdev_remove_replace_with_indirect(vd, txg);

	/*
	* We now release the locks, allowing spa_sync to run and finish the
	* removal via vdev_remove_complete_sync in syncing context.
	*
	* Note that we hold on to the vdev_t that has been replaced. Since
	* it isn't part of the vdev tree any longer, it can't be concurrently
	* manipulated, even while we don't have the config lock.
	*/
	(void) spa_vdev_exit(spa, NULL, txg, 0);

	/*
	* Top ZAP should have been transferred to the indirect vdev in
	* vdev_remove_replace_with_indirect.
	*/
	ASSERT0(vd->vdev_top_zap);

	/*
	* Leaf ZAP should have been moved in vdev_remove_replace_with_indirect.
	*/
	ASSERT0(vd->vdev_leaf_zap);

	txg = spa_vdev_enter(spa);
	(void) vdev_label_init(vd, 0, VDEV_LABEL_REMOVE);
	/*
	* Request to update the config and the config cachefile.
	*/
	vdev_config_dirty(spa->spa_root_vdev);
	(void) spa_vdev_exit(spa, vd, txg, 0);

	if (ev != NULL)
	spa_event_post(ev);
	}

	/*
	* Evacuates a segment of size at most max_alloc from the vdev
	* via repeated calls to spa_vdev_copy_segment. If an allocation
	* fails, the pool is probably too fragmented to handle such a
	* large size, so decrease max_alloc so that the caller will not try
	* this size again this txg.
	*/
	static void
	spa_vdev_copy_impl(vdev_t vd, spa_vdev_removal_t svr, vdev_copy_arg_t *vca,
	uint64_t max_alloc, dmu_tx_t tx)
	{
	uint64_t txg = dmu_tx_get_txg(tx);
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;

	mutex_enter(&svr->svr_lock);

	/*
	* Determine how big of a chunk to copy. We can allocate up
	* to max_alloc bytes, and we can span up to vdev_removal_max_span
	* bytes of unallocated space at a time. "segs" will track the
	* allocated segments that we are copying. We may also be copying
	* free segments (of up to vdev_removal_max_span bytes).
	*/
	range_tree_t *segs = range_tree_create(NULL, RANGE_SEG64, NULL, 0, 0);
	for (;;) {
	range_tree_t *rt = svr->svr_allocd_segs;
	range_seg_t *rs = range_tree_first(rt);

	if (rs == NULL)
	break;

	uint64_t seg_length;

	if (range_tree_is_empty(segs)) {
	/* need to truncate the first seg based on max_alloc */
	seg_length = MIN(rs_get_end(rs, rt) - rs_get_start(rs,
	rt), *max_alloc);
	} else {
	if (rs_get_start(rs, rt) - range_tree_max(segs) >
	vdev_removal_max_span) {
	/*
	* Including this segment would cause us to
	* copy a larger unneeded chunk than is allowed.
	*/
	break;
	} else if (rs_get_end(rs, rt) - range_tree_min(segs) >
	*max_alloc) {
	/*
	* This additional segment would extend past
	* max_alloc. Rather than splitting this
	* segment, leave it for the next mapping.
	*/
	break;
	} else {
	seg_length = rs_get_end(rs, rt) -
	rs_get_start(rs, rt);
	}
	}

	range_tree_add(segs, rs_get_start(rs, rt), seg_length);
	range_tree_remove(svr->svr_allocd_segs,
	rs_get_start(rs, rt), seg_length);
	}

	if (range_tree_is_empty(segs)) {
	mutex_exit(&svr->svr_lock);
	range_tree_destroy(segs);
	return;
	}

	if (svr->svr_max_offset_to_sync[txg & TXG_MASK] == 0) {
	dsl_sync_task_nowait(dmu_tx_pool(tx), vdev_mapping_sync,
	svr, tx);
	}

	svr->svr_max_offset_to_sync[txg & TXG_MASK] = range_tree_max(segs);

	/*
	* Note: this is the amount of allocated space
	* that we are taking care of each txg.
	*/
	svr->svr_bytes_done[txg & TXG_MASK] += range_tree_space(segs);

	mutex_exit(&svr->svr_lock);

	zio_alloc_list_t zal;
	metaslab_trace_init(&zal);
	uint64_t thismax = SPA_MAXBLOCKSIZE;
	while (!range_tree_is_empty(segs)) {
	int error = spa_vdev_copy_segment(vd,
	segs, thismax, txg, vca, &zal);

	if (error == ENOSPC) {
	/*
	* Cut our segment in half, and don't try this
	* segment size again this txg. Note that the
	* allocation size must be aligned to the highest
	* ashift in the pool, so that the allocation will
	* not be padded out to a multiple of the ashift,
	* which could cause us to think that this mapping
	* is larger than we intended.
	*/
	ASSERT3U(spa->spa_max_ashift, >=, SPA_MINBLOCKSHIFT);
	ASSERT3U(spa->spa_max_ashift, ==, spa->spa_min_ashift);
	uint64_t attempted =
	MIN(range_tree_span(segs), thismax);
	thismax = P2ROUNDUP(attempted / 2,
	1 << spa->spa_max_ashift);
	/*
	* The minimum-size allocation can not fail.
	*/
	ASSERT3U(attempted, >, 1 << spa->spa_max_ashift);
	*max_alloc = attempted - (1 << spa->spa_max_ashift);
	} else {
	ASSERT0(error);

	/*
	* We've performed an allocation, so reset the
	* alloc trace list.
	*/
	metaslab_trace_fini(&zal);
	metaslab_trace_init(&zal);
	}
	}
	metaslab_trace_fini(&zal);
	range_tree_destroy(segs);
	}

	/*
	* The size of each removal mapping is limited by the tunable
	* zfs_remove_max_segment, but we must adjust this to be a multiple of the
	* pool's ashift, so that we don't try to split individual sectors regardless
	* of the tunable value. (Note that device removal requires that all devices
	* have the same ashift, so there's no difference between spa_min_ashift and
	* spa_max_ashift.) The raw tunable should not be used elsewhere.
	*/
	uint64_t
	spa_remove_max_segment(spa_t *spa)
	{
	return (P2ROUNDUP(zfs_remove_max_segment, 1 << spa->spa_max_ashift));
	}

	/*
	* The removal thread operates in open context. It iterates over all
	* allocated space in the vdev, by loading each metaslab's spacemap.
	* For each contiguous segment of allocated space (capping the segment
	* size at SPA_MAXBLOCKSIZE), we:
	* - Allocate space for it on another vdev.
	* - Create a new mapping from the old location to the new location
	* (as a record in svr_new_segments).
	* - Initiate a physical read zio to get the data off the removing disk.
	* - In the read zio's done callback, initiate a physical write zio to
	* write it to the new vdev.
	* Note that all of this will take effect when a particular TXG syncs.
	* The sync thread ensures that all the phys reads and writes for the syncing
	* TXG have completed (see spa_txg_zio) and writes the new mappings to disk
	* (see vdev_mapping_sync()).
	*/
	static void
	spa_vdev_remove_thread(void *arg)
	{
	spa_t *spa = arg;
	spa_vdev_removal_t *svr = spa->spa_vdev_removal;
	vdev_copy_arg_t vca;
	uint64_t max_alloc = spa_remove_max_segment(spa);
	uint64_t last_txg = 0;

	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	vdev_t *vd = vdev_lookup_top(spa, svr->svr_vdev_id);
	vdev_indirect_mapping_t *vim = vd->vdev_indirect_mapping;
	uint64_t start_offset = vdev_indirect_mapping_max_offset(vim);

	ASSERT3P(vd->vdev_ops, !=, &vdev_indirect_ops);
	ASSERT(vdev_is_concrete(vd));
	ASSERT(vd->vdev_removing);
	ASSERT(vd->vdev_indirect_config.vic_mapping_object != 0);
	ASSERT(vim != NULL);

	mutex_init(&vca.vca_lock, NULL, MUTEX_DEFAULT, NULL);
	cv_init(&vca.vca_cv, NULL, CV_DEFAULT, NULL);
	vca.vca_outstanding_bytes = 0;
	vca.vca_read_error_bytes = 0;
	vca.vca_write_error_bytes = 0;

	mutex_enter(&svr->svr_lock);

	/*
	* Start from vim_max_offset so we pick up where we left off
	* if we are restarting the removal after opening the pool.
	*/
	uint64_t msi;
	for (msi = start_offset >> vd->vdev_ms_shift;
	msi < vd->vdev_ms_count && !svr->svr_thread_exit; msi++) {
	metaslab_t *msp = vd->vdev_ms[msi];
	ASSERT3U(msi, <=, vd->vdev_ms_count);

	ASSERT0(range_tree_space(svr->svr_allocd_segs));

	mutex_enter(&msp->ms_sync_lock);
	mutex_enter(&msp->ms_lock);

	/*
	* Assert nothing in flight -- ms_*tree is empty.
	*/
	for (int i = 0; i < TXG_SIZE; i++) {
	ASSERT0(range_tree_space(msp->ms_allocating[i]));
	}

	/*
	* If the metaslab has ever been allocated from (ms_sm!=NULL),
	* read the allocated segments from the space map object
	* into svr_allocd_segs. Since we do this while holding
	* svr_lock and ms_sync_lock, concurrent frees (which
	* would have modified the space map) will wait for us
	* to finish loading the spacemap, and then take the
	* appropriate action (see free_from_removing_vdev()).
	*/
	if (msp->ms_sm != NULL) {
	VERIFY0(space_map_load(msp->ms_sm,
	svr->svr_allocd_segs, SM_ALLOC));

	range_tree_walk(msp->ms_unflushed_allocs,
	range_tree_add, svr->svr_allocd_segs);
	range_tree_walk(msp->ms_unflushed_frees,
	range_tree_remove, svr->svr_allocd_segs);
	range_tree_walk(msp->ms_freeing,
	range_tree_remove, svr->svr_allocd_segs);

	/*
	* When we are resuming from a paused removal (i.e.
	* when importing a pool with a removal in progress),
	* discard any state that we have already processed.
	*/
	range_tree_clear(svr->svr_allocd_segs, 0, start_offset);
	}
	mutex_exit(&msp->ms_lock);
	mutex_exit(&msp->ms_sync_lock);

	vca.vca_msp = msp;
	zfs_dbgmsg("copying %llu segments for metaslab %llu",
	(u_longlong_t)zfs_btree_numnodes(
	&svr->svr_allocd_segs->rt_root),
	(u_longlong_t)msp->ms_id);

	while (!svr->svr_thread_exit &&
	!range_tree_is_empty(svr->svr_allocd_segs)) {

	mutex_exit(&svr->svr_lock);

	/*
	* We need to periodically drop the config lock so that
	* writers can get in. Additionally, we can't wait
	* for a txg to sync while holding a config lock
	* (since a waiting writer could cause a 3-way deadlock
	* with the sync thread, which also gets a config
	* lock for reader). So we can't hold the config lock
	* while calling dmu_tx_assign().
	*/
	spa_config_exit(spa, SCL_CONFIG, FTAG);

	/*
	* This delay will pause the removal around the point
	* specified by zfs_removal_suspend_progress. We do this
	* solely from the test suite or during debugging.
	*/
	while (zfs_removal_suspend_progress &&
	!svr->svr_thread_exit)
	delay(hz);

	mutex_enter(&vca.vca_lock);
	while (vca.vca_outstanding_bytes >
	zfs_remove_max_copy_bytes) {
	cv_wait(&vca.vca_cv, &vca.vca_lock);
	}
	mutex_exit(&vca.vca_lock);

	dmu_tx_t *tx =
	dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir);

	VERIFY0(dmu_tx_assign(tx, TXG_WAIT));
	uint64_t txg = dmu_tx_get_txg(tx);

	/*
	* Reacquire the vdev_config lock. The vdev_t
	* that we're removing may have changed, e.g. due
	* to a vdev_attach or vdev_detach.
	*/
	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER);
	vd = vdev_lookup_top(spa, svr->svr_vdev_id);

	if (txg != last_txg)
	max_alloc = spa_remove_max_segment(spa);
	last_txg = txg;

	spa_vdev_copy_impl(vd, svr, &vca, &max_alloc, tx);

	dmu_tx_commit(tx);
	mutex_enter(&svr->svr_lock);
	}

	mutex_enter(&vca.vca_lock);
	if (zfs_removal_ignore_errors == 0 &&
	(vca.vca_read_error_bytes > 0 \|\|
	vca.vca_write_error_bytes > 0)) {
	svr->svr_thread_exit = B_TRUE;
	}
	mutex_exit(&vca.vca_lock);
	}

	mutex_exit(&svr->svr_lock);

	spa_config_exit(spa, SCL_CONFIG, FTAG);

	/*
	* Wait for all copies to finish before cleaning up the vca.
	*/
	txg_wait_synced(spa->spa_dsl_pool, 0);
	ASSERT0(vca.vca_outstanding_bytes);

	mutex_destroy(&vca.vca_lock);
	cv_destroy(&vca.vca_cv);

	if (svr->svr_thread_exit) {
	mutex_enter(&svr->svr_lock);
	range_tree_vacate(svr->svr_allocd_segs, NULL, NULL);
	svr->svr_thread = NULL;
	cv_broadcast(&svr->svr_cv);
	mutex_exit(&svr->svr_lock);

	/*
	* During the removal process an unrecoverable read or write
	* error was encountered. The removal process must be
	* cancelled or this damage may become permanent.
	*/
	if (zfs_removal_ignore_errors == 0 &&
	(vca.vca_read_error_bytes > 0 \|\|
	vca.vca_write_error_bytes > 0)) {
	zfs_dbgmsg("canceling removal due to IO errors: "
	"[read_error_bytes=%llu] [write_error_bytes=%llu]",
	(u_longlong_t)vca.vca_read_error_bytes,
	(u_longlong_t)vca.vca_write_error_bytes);
	spa_vdev_remove_cancel_impl(spa);
	}
	} else {
	ASSERT0(range_tree_space(svr->svr_allocd_segs));
	vdev_remove_complete(spa);
	}

	thread_exit();
	}

	void
	spa_vdev_remove_suspend(spa_t *spa)
	{
	spa_vdev_removal_t *svr = spa->spa_vdev_removal;

	if (svr == NULL)
	return;

	mutex_enter(&svr->svr_lock);
	svr->svr_thread_exit = B_TRUE;
	while (svr->svr_thread != NULL)
	cv_wait(&svr->svr_cv, &svr->svr_lock);
	svr->svr_thread_exit = B_FALSE;
	mutex_exit(&svr->svr_lock);
	}

	+/*
	+ * Return true if the "allocating" property has been set to "off"
	+ */
	+static boolean_t
	+vdev_prop_allocating_off(vdev_t *vd)
	+{
	+ uint64_t objid = vd->vdev_top_zap;
	+ uint64_t allocating = 1;
	+
	+ /* no vdev property object => no props */
	+ if (objid != 0) {
	+ spa_t *spa = vd->vdev_spa;
	+ objset_t *mos = spa->spa_meta_objset;
	+
	+ mutex_enter(&spa->spa_props_lock);
	+ (void) zap_lookup(mos, objid, "allocating", sizeof (uint64_t),
	+ 1, &allocating);
	+ mutex_exit(&spa->spa_props_lock);
	+ }
	+ return (allocating == 0);
	+}
	+
	/* ARGSUSED */
	static int
	spa_vdev_remove_cancel_check(void arg, dmu_tx_t tx)
	{
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;

	if (spa->spa_vdev_removal == NULL)
	return (ENOTACTIVE);
	return (0);
	}

	/*
	* Cancel a removal by freeing all entries from the partial mapping
	* and marking the vdev as no longer being removing.
	*/
	/* ARGSUSED */
	static void
	spa_vdev_remove_cancel_sync(void arg, dmu_tx_t tx)
	{
	spa_t *spa = dmu_tx_pool(tx)->dp_spa;
	spa_vdev_removal_t *svr = spa->spa_vdev_removal;
	vdev_t *vd = vdev_lookup_top(spa, svr->svr_vdev_id);
	vdev_indirect_config_t *vic = &vd->vdev_indirect_config;
	vdev_indirect_mapping_t *vim = vd->vdev_indirect_mapping;
	objset_t *mos = spa->spa_meta_objset;

	ASSERT3P(svr->svr_thread, ==, NULL);

	spa_feature_decr(spa, SPA_FEATURE_DEVICE_REMOVAL, tx);

	boolean_t are_precise;
	VERIFY0(vdev_obsolete_counts_are_precise(vd, &are_precise));
	if (are_precise) {
	spa_feature_decr(spa, SPA_FEATURE_OBSOLETE_COUNTS, tx);
	VERIFY0(zap_remove(spa->spa_meta_objset, vd->vdev_top_zap,
	VDEV_TOP_ZAP_OBSOLETE_COUNTS_ARE_PRECISE, tx));
	}

	uint64_t obsolete_sm_object;
	VERIFY0(vdev_obsolete_sm_object(vd, &obsolete_sm_object));
	if (obsolete_sm_object != 0) {
	ASSERT(vd->vdev_obsolete_sm != NULL);
	ASSERT3U(obsolete_sm_object, ==,
	space_map_object(vd->vdev_obsolete_sm));

	space_map_free(vd->vdev_obsolete_sm, tx);
	VERIFY0(zap_remove(spa->spa_meta_objset, vd->vdev_top_zap,
	VDEV_TOP_ZAP_INDIRECT_OBSOLETE_SM, tx));
	space_map_close(vd->vdev_obsolete_sm);
	vd->vdev_obsolete_sm = NULL;
	spa_feature_decr(spa, SPA_FEATURE_OBSOLETE_COUNTS, tx);
	}
	for (int i = 0; i < TXG_SIZE; i++) {
	ASSERT(list_is_empty(&svr->svr_new_segments[i]));
	ASSERT3U(svr->svr_max_offset_to_sync[i], <=,
	vdev_indirect_mapping_max_offset(vim));
	}

	for (uint64_t msi = 0; msi < vd->vdev_ms_count; msi++) {
	metaslab_t *msp = vd->vdev_ms[msi];

	if (msp->ms_start >= vdev_indirect_mapping_max_offset(vim))
	break;

	ASSERT0(range_tree_space(svr->svr_allocd_segs));

	mutex_enter(&msp->ms_lock);

	/*
	* Assert nothing in flight -- ms_*tree is empty.
	*/
	for (int i = 0; i < TXG_SIZE; i++)
	ASSERT0(range_tree_space(msp->ms_allocating[i]));
	for (int i = 0; i < TXG_DEFER_SIZE; i++)
	ASSERT0(range_tree_space(msp->ms_defer[i]));
	ASSERT0(range_tree_space(msp->ms_freed));

	if (msp->ms_sm != NULL) {
	mutex_enter(&svr->svr_lock);
	VERIFY0(space_map_load(msp->ms_sm,
	svr->svr_allocd_segs, SM_ALLOC));

	range_tree_walk(msp->ms_unflushed_allocs,
	range_tree_add, svr->svr_allocd_segs);
	range_tree_walk(msp->ms_unflushed_frees,
	range_tree_remove, svr->svr_allocd_segs);
	range_tree_walk(msp->ms_freeing,
	range_tree_remove, svr->svr_allocd_segs);

	/*
	* Clear everything past what has been synced,
	* because we have not allocated mappings for it yet.
	*/
	uint64_t syncd = vdev_indirect_mapping_max_offset(vim);
	uint64_t sm_end = msp->ms_sm->sm_start +
	msp->ms_sm->sm_size;
	if (sm_end > syncd)
	range_tree_clear(svr->svr_allocd_segs,
	syncd, sm_end - syncd);

	mutex_exit(&svr->svr_lock);
	}
	mutex_exit(&msp->ms_lock);

	mutex_enter(&svr->svr_lock);
	range_tree_vacate(svr->svr_allocd_segs,
	free_mapped_segment_cb, vd);
	mutex_exit(&svr->svr_lock);
	}

	/*
	* Note: this must happen after we invoke free_mapped_segment_cb,
	* because it adds to the obsolete_segments.
	*/
	range_tree_vacate(vd->vdev_obsolete_segments, NULL, NULL);

	ASSERT3U(vic->vic_mapping_object, ==,
	vdev_indirect_mapping_object(vd->vdev_indirect_mapping));
	vdev_indirect_mapping_close(vd->vdev_indirect_mapping);
	vd->vdev_indirect_mapping = NULL;
	vdev_indirect_mapping_free(mos, vic->vic_mapping_object, tx);
	vic->vic_mapping_object = 0;

	ASSERT3U(vic->vic_births_object, ==,
	vdev_indirect_births_object(vd->vdev_indirect_births));
	vdev_indirect_births_close(vd->vdev_indirect_births);
	vd->vdev_indirect_births = NULL;
	vdev_indirect_births_free(mos, vic->vic_births_object, tx);
	vic->vic_births_object = 0;

	/*
	* We may have processed some frees from the removing vdev in this
	* txg, thus increasing svr_bytes_done; discard that here to
	* satisfy the assertions in spa_vdev_removal_destroy().
	* Note that future txg's can not have any bytes_done, because
	* future TXG's are only modified from open context, and we have
	* already shut down the copying thread.
	*/
	svr->svr_bytes_done[dmu_tx_get_txg(tx) & TXG_MASK] = 0;
	spa_finish_removal(spa, DSS_CANCELED, tx);

	vd->vdev_removing = B_FALSE;
	+
	+ if (!vdev_prop_allocating_off(vd)) {
	+ spa_config_enter(spa, SCL_ALLOC \| SCL_VDEV, FTAG, RW_WRITER);
	+ vdev_activate(vd);
	+ spa_config_exit(spa, SCL_ALLOC \| SCL_VDEV, FTAG);
	+ }
	+
	vdev_config_dirty(vd);

	zfs_dbgmsg("canceled device removal for vdev %llu in %llu",
	(u_longlong_t)vd->vdev_id, (u_longlong_t)dmu_tx_get_txg(tx));
	spa_history_log_internal(spa, "vdev remove canceled", tx,
	"%s vdev %llu %s", spa_name(spa),
	(u_longlong_t)vd->vdev_id,
	(vd->vdev_path != NULL) ? vd->vdev_path : "-");
	}

	static int
	spa_vdev_remove_cancel_impl(spa_t *spa)
	{
	- uint64_t vdid = spa->spa_vdev_removal->svr_vdev_id;
	-
	int error = dsl_sync_task(spa->spa_name, spa_vdev_remove_cancel_check,
	spa_vdev_remove_cancel_sync, NULL, 0,
	ZFS_SPACE_CHECK_EXTRA_RESERVED);
	-
	- if (error == 0) {
	- spa_config_enter(spa, SCL_ALLOC \| SCL_VDEV, FTAG, RW_WRITER);
	- vdev_t *vd = vdev_lookup_top(spa, vdid);
	- metaslab_group_activate(vd->vdev_mg);
	- ASSERT(!vd->vdev_islog);
	- metaslab_group_activate(vd->vdev_log_mg);
	- spa_config_exit(spa, SCL_ALLOC \| SCL_VDEV, FTAG);
	- }
	-
	return (error);
	}

	int
	spa_vdev_remove_cancel(spa_t *spa)
	{
	spa_vdev_remove_suspend(spa);

	if (spa->spa_vdev_removal == NULL)
	return (ENOTACTIVE);

	return (spa_vdev_remove_cancel_impl(spa));
	}

	void
	svr_sync(spa_t spa, dmu_tx_t tx)
	{
	spa_vdev_removal_t *svr = spa->spa_vdev_removal;
	int txgoff = dmu_tx_get_txg(tx) & TXG_MASK;

	if (svr == NULL)
	return;

	/*
	* This check is necessary so that we do not dirty the
	* DIRECTORY_OBJECT via spa_sync_removing_state() when there
	* is nothing to do. Dirtying it every time would prevent us
	* from syncing-to-convergence.
	*/
	if (svr->svr_bytes_done[txgoff] == 0)
	return;

	/*
	* Update progress accounting.
	*/
	spa->spa_removing_phys.sr_copied += svr->svr_bytes_done[txgoff];
	svr->svr_bytes_done[txgoff] = 0;

	spa_sync_removing_state(spa, tx);
	}

	static void
	vdev_remove_make_hole_and_free(vdev_t *vd)
	{
	uint64_t id = vd->vdev_id;
	spa_t *spa = vd->vdev_spa;
	vdev_t *rvd = spa->spa_root_vdev;

	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	vdev_free(vd);

	vd = vdev_alloc_common(spa, id, 0, &vdev_hole_ops);
	vdev_add_child(rvd, vd);
	vdev_config_dirty(rvd);

	/*
	* Reassess the health of our root vdev.
	*/
	vdev_reopen(rvd);
	}

	/*
	* Remove a log device. The config lock is held for the specified TXG.
	*/
	static int
	spa_vdev_remove_log(vdev_t vd, uint64_t txg)
	{
	metaslab_group_t *mg = vd->vdev_mg;
	spa_t *spa = vd->vdev_spa;
	int error = 0;

	ASSERT(vd->vdev_islog);
	ASSERT(vd == vd->vdev_top);
	ASSERT3P(vd->vdev_log_mg, ==, NULL);
	ASSERT(MUTEX_HELD(&spa_namespace_lock));

	/*
	* Stop allocating from this vdev.
	*/
	metaslab_group_passivate(mg);

	/*
	* Wait for the youngest allocations and frees to sync,
	* and then wait for the deferral of those frees to finish.
	*/
	spa_vdev_config_exit(spa, NULL,
	*txg + TXG_CONCURRENT_STATES + TXG_DEFER_SIZE, 0, FTAG);

	/*
	* Cancel any initialize or TRIM which was in progress.
	*/
	vdev_initialize_stop_all(vd, VDEV_INITIALIZE_CANCELED);
	vdev_trim_stop_all(vd, VDEV_TRIM_CANCELED);
	vdev_autotrim_stop_wait(vd);

	/*
	* Evacuate the device. We don't hold the config lock as
	* writer since we need to do I/O but we do keep the
	* spa_namespace_lock held. Once this completes the device
	* should no longer have any blocks allocated on it.
	*/
	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	if (vd->vdev_stat.vs_alloc != 0)
	error = spa_reset_logs(spa);

	*txg = spa_vdev_config_enter(spa);

	if (error != 0) {
	metaslab_group_activate(mg);
	ASSERT3P(vd->vdev_log_mg, ==, NULL);
	return (error);
	}
	ASSERT0(vd->vdev_stat.vs_alloc);

	/*
	* The evacuation succeeded. Remove any remaining MOS metadata
	* associated with this vdev, and wait for these changes to sync.
	*/
	vd->vdev_removing = B_TRUE;

	vdev_dirty_leaves(vd, VDD_DTL, *txg);
	vdev_config_dirty(vd);

	/*
	* When the log space map feature is enabled we look at
	* the vdev's top_zap to find the on-disk flush data of
	* the metaslab we just flushed. Thus, while removing a
	* log vdev we make sure to call vdev_metaslab_fini()
	* first, which removes all metaslabs of this vdev from
	* spa_metaslabs_by_flushed before vdev_remove_empty()
	* destroys the top_zap of this log vdev.
	*
	* This avoids the scenario where we flush a metaslab
	* from the log vdev being removed that doesn't have a
	* top_zap and end up failing to lookup its on-disk flush
	* data.
	*
	* We don't call metaslab_group_destroy() right away
	* though (it will be called in vdev_free() later) as
	* during metaslab_sync() of metaslabs from other vdevs
	* we may touch the metaslab group of this vdev through
	* metaslab_class_histogram_verify()
	*/
	vdev_metaslab_fini(vd);
	spa_log_sm_set_blocklimit(spa);

	spa_vdev_config_exit(spa, NULL, *txg, 0, FTAG);
	*txg = spa_vdev_config_enter(spa);

	sysevent_t *ev = spa_event_create(spa, vd, NULL,
	ESC_ZFS_VDEV_REMOVE_DEV);
	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);

	/* The top ZAP should have been destroyed by vdev_remove_empty. */
	ASSERT0(vd->vdev_top_zap);
	/* The leaf ZAP should have been destroyed by vdev_dtl_sync. */
	ASSERT0(vd->vdev_leaf_zap);

	(void) vdev_label_init(vd, 0, VDEV_LABEL_REMOVE);

	if (list_link_active(&vd->vdev_state_dirty_node))
	vdev_state_clean(vd);
	if (list_link_active(&vd->vdev_config_dirty_node))
	vdev_config_clean(vd);

	ASSERT0(vd->vdev_stat.vs_alloc);

	/*
	* Clean up the vdev namespace.
	*/
	vdev_remove_make_hole_and_free(vd);

	if (ev != NULL)
	spa_event_post(ev);

	return (0);
	}

	static int
	spa_vdev_remove_top_check(vdev_t *vd)
	{
	spa_t *spa = vd->vdev_spa;

	if (vd != vd->vdev_top)
	return (SET_ERROR(ENOTSUP));

	if (!vdev_is_concrete(vd))
	return (SET_ERROR(ENOTSUP));

	if (!spa_feature_is_enabled(spa, SPA_FEATURE_DEVICE_REMOVAL))
	return (SET_ERROR(ENOTSUP));

	+ /*
	+ * This device is already being removed
	+ */
	+ if (vd->vdev_removing)
	+ return (SET_ERROR(EALREADY));

	metaslab_class_t *mc = vd->vdev_mg->mg_class;
	metaslab_class_t *normal = spa_normal_class(spa);
	if (mc != normal) {
	/*
	* Space allocated from the special (or dedup) class is
	* included in the DMU's space usage, but it's not included
	* in spa_dspace (or dsl_pool_adjustedsize()). Therefore
	* there is always at least as much free space in the normal
	* class, as is allocated from the special (and dedup) class.
	* As a backup check, we will return ENOSPC if this is
	* violated. See also spa_update_dspace().
	*/
	uint64_t available = metaslab_class_get_space(normal) -
	metaslab_class_get_alloc(normal);
	ASSERT3U(available, >=, vd->vdev_stat.vs_alloc);
	if (available < vd->vdev_stat.vs_alloc)
	return (SET_ERROR(ENOSPC));
	- } else {
	+ } else if (!vd->vdev_noalloc) {
	/* available space in the pool's normal class */
	uint64_t available = dsl_dir_space_available(
	spa->spa_dsl_pool->dp_root_dir, NULL, 0, B_TRUE);
	- if (available <
	- vd->vdev_stat.vs_dspace + spa_get_slop_space(spa)) {
	- /*
	- * This is a normal device. There has to be enough free
	- * space to remove the device and leave double the
	- * "slop" space (i.e. we must leave at least 3% of the
	- * pool free, in addition to the normal slop space).
	- */
	+ if (available < vd->vdev_stat.vs_dspace)
	return (SET_ERROR(ENOSPC));
	- }
	}

	/*
	* There can not be a removal in progress.
	*/
	if (spa->spa_removing_phys.sr_state == DSS_SCANNING)
	return (SET_ERROR(EBUSY));

	/*
	* The device must have all its data.
	*/
	if (!vdev_dtl_empty(vd, DTL_MISSING) \|\|
	!vdev_dtl_empty(vd, DTL_OUTAGE))
	return (SET_ERROR(EBUSY));

	/*
	* The device must be healthy.
	*/
	if (!vdev_readable(vd))
	return (SET_ERROR(EIO));

	/*
	* All vdevs in normal class must have the same ashift.
	*/
	if (spa->spa_max_ashift != spa->spa_min_ashift) {
	return (SET_ERROR(EINVAL));
	}

	/*
	* A removed special/dedup vdev must have same ashift as normal class.
	*/
	ASSERT(!vd->vdev_islog);
	if (vd->vdev_alloc_bias != VDEV_BIAS_NONE &&
	vd->vdev_ashift != spa->spa_max_ashift) {
	return (SET_ERROR(EINVAL));
	}

	/*
	* All vdevs in normal class must have the same ashift
	* and not be raidz or draid.
	*/
	vdev_t *rvd = spa->spa_root_vdev;
	int num_indirect = 0;
	for (uint64_t id = 0; id < rvd->vdev_children; id++) {
	vdev_t *cvd = rvd->vdev_child[id];

	/*
	* A removed special/dedup vdev must have the same ashift
	* across all vdevs in its class.
	*/
	if (vd->vdev_alloc_bias != VDEV_BIAS_NONE &&
	cvd->vdev_alloc_bias == vd->vdev_alloc_bias &&
	cvd->vdev_ashift != vd->vdev_ashift) {
	return (SET_ERROR(EINVAL));
	}
	if (cvd->vdev_ashift != 0 &&
	cvd->vdev_alloc_bias == VDEV_BIAS_NONE)
	ASSERT3U(cvd->vdev_ashift, ==, spa->spa_max_ashift);
	if (cvd->vdev_ops == &vdev_indirect_ops)
	num_indirect++;
	if (!vdev_is_concrete(cvd))
	continue;
	if (vdev_get_nparity(cvd) != 0)
	return (SET_ERROR(EINVAL));
	/*
	* Need the mirror to be mirror of leaf vdevs only
	*/
	if (cvd->vdev_ops == &vdev_mirror_ops) {
	for (uint64_t cid = 0;
	cid < cvd->vdev_children; cid++) {
	if (!cvd->vdev_child[cid]->vdev_ops->
	vdev_op_leaf)
	return (SET_ERROR(EINVAL));
	}
	}
	}

	return (0);
	}

	/*
	* Initiate removal of a top-level vdev, reducing the total space in the pool.
	* The config lock is held for the specified TXG. Once initiated,
	* evacuation of all allocated space (copying it to other vdevs) happens
	* in the background (see spa_vdev_remove_thread()), and can be canceled
	* (see spa_vdev_remove_cancel()). If successful, the vdev will
	* be transformed to an indirect vdev (see spa_vdev_remove_complete()).
	*/
	static int
	spa_vdev_remove_top(vdev_t vd, uint64_t txg)
	{
	spa_t *spa = vd->vdev_spa;
	+ boolean_t set_noalloc = B_FALSE;
	int error;

	/*
	* Check for errors up-front, so that we don't waste time
	* passivating the metaslab group and clearing the ZIL if there
	* are errors.
	*/
	error = spa_vdev_remove_top_check(vd);
	- if (error != 0)
	- return (error);

	/*
	* Stop allocating from this vdev. Note that we must check
	* that this is not the only device in the pool before
	* passivating, otherwise we will not be able to make
	* progress because we can't allocate from any vdevs.
	* The above check for sufficient free space serves this
	* purpose.
	*/
	- metaslab_group_t *mg = vd->vdev_mg;
	- metaslab_group_passivate(mg);
	- ASSERT(!vd->vdev_islog);
	- metaslab_group_passivate(vd->vdev_log_mg);
	-
	- /*
	- * Wait for the youngest allocations and frees to sync,
	- * and then wait for the deferral of those frees to finish.
	- */
	- spa_vdev_config_exit(spa, NULL,
	- *txg + TXG_CONCURRENT_STATES + TXG_DEFER_SIZE, 0, FTAG);
	+ if (error == 0 && !vd->vdev_noalloc) {
	+ set_noalloc = B_TRUE;
	+ error = vdev_passivate(vd, txg);
	+ }

	- /*
	- * We must ensure that no "stubby" log blocks are allocated
	- * on the device to be removed. These blocks could be
	- * written at any time, including while we are in the middle
	- * of copying them.
	- */
	- error = spa_reset_logs(spa);
	+ if (error != 0)
	+ return (error);

	/*
	* We stop any initializing and TRIM that is currently in progress
	* but leave the state as "active". This will allow the process to
	* resume if the removal is canceled sometime later.
	*/
	+
	+ spa_vdev_config_exit(spa, NULL, *txg, 0, FTAG);
	+
	vdev_initialize_stop_all(vd, VDEV_INITIALIZE_ACTIVE);
	vdev_trim_stop_all(vd, VDEV_TRIM_ACTIVE);
	vdev_autotrim_stop_wait(vd);

	*txg = spa_vdev_config_enter(spa);

	/*
	* Things might have changed while the config lock was dropped
	* (e.g. space usage). Check for errors again.
	*/
	- if (error == 0)
	- error = spa_vdev_remove_top_check(vd);
	+ error = spa_vdev_remove_top_check(vd);

	if (error != 0) {
	- metaslab_group_activate(mg);
	- ASSERT(!vd->vdev_islog);
	- metaslab_group_activate(vd->vdev_log_mg);
	+ if (set_noalloc)
	+ vdev_activate(vd);
	spa_async_request(spa, SPA_ASYNC_INITIALIZE_RESTART);
	spa_async_request(spa, SPA_ASYNC_TRIM_RESTART);
	spa_async_request(spa, SPA_ASYNC_AUTOTRIM_RESTART);
	return (error);
	}

	vd->vdev_removing = B_TRUE;

	vdev_dirty_leaves(vd, VDD_DTL, *txg);
	vdev_config_dirty(vd);
	dmu_tx_t tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg);
	dsl_sync_task_nowait(spa->spa_dsl_pool,
	vdev_remove_initiate_sync, (void *)(uintptr_t)vd->vdev_id, tx);
	dmu_tx_commit(tx);

	return (0);
	}

	/*
	* Remove a device from the pool.
	*
	* Removing a device from the vdev namespace requires several steps
	* and can take a significant amount of time. As a result we use
	* the spa_vdev_config_[enter/exit] functions which allow us to
	* grab and release the spa_config_lock while still holding the namespace
	* lock. During each step the configuration is synced out.
	*/
	int
	spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare)
	{
	vdev_t *vd;
	nvlist_t spares, l2cache, *nv;
	uint64_t txg = 0;
	uint_t nspares, nl2cache;
	int error = 0, error_log;
	boolean_t locked = MUTEX_HELD(&spa_namespace_lock);
	sysevent_t *ev = NULL;
	char vd_type = NULL, vd_path = NULL;

	ASSERT(spa_writeable(spa));

	if (!locked)
	txg = spa_vdev_enter(spa);

	ASSERT(MUTEX_HELD(&spa_namespace_lock));
	if (spa_feature_is_active(spa, SPA_FEATURE_POOL_CHECKPOINT)) {
	error = (spa_has_checkpoint(spa)) ?
	ZFS_ERR_CHECKPOINT_EXISTS : ZFS_ERR_DISCARDING_CHECKPOINT;

	if (!locked)
	return (spa_vdev_exit(spa, NULL, txg, error));

	return (error);
	}

	vd = spa_lookup_by_guid(spa, guid, B_FALSE);

	if (spa->spa_spares.sav_vdevs != NULL &&
	nvlist_lookup_nvlist_array(spa->spa_spares.sav_config,
	ZPOOL_CONFIG_SPARES, &spares, &nspares) == 0 &&
	(nv = spa_nvlist_lookup_by_guid(spares, nspares, guid)) != NULL) {
	/*
	* Only remove the hot spare if it's not currently in use
	* in this pool.
	*/
	if (vd == NULL \|\| unspare) {
	char *type;
	boolean_t draid_spare = B_FALSE;

	if (nvlist_lookup_string(nv, ZPOOL_CONFIG_TYPE, &type)
	== 0 && strcmp(type, VDEV_TYPE_DRAID_SPARE) == 0)
	draid_spare = B_TRUE;

	if (vd == NULL && draid_spare) {
	error = SET_ERROR(ENOTSUP);
	} else {
	if (vd == NULL)
	vd = spa_lookup_by_guid(spa,
	guid, B_TRUE);
	ev = spa_event_create(spa, vd, NULL,
	ESC_ZFS_VDEV_REMOVE_AUX);

	vd_type = VDEV_TYPE_SPARE;
	vd_path = spa_strdup(fnvlist_lookup_string(
	nv, ZPOOL_CONFIG_PATH));
	spa_vdev_remove_aux(spa->spa_spares.sav_config,
	ZPOOL_CONFIG_SPARES, spares, nspares, nv);
	spa_load_spares(spa);
	spa->spa_spares.sav_sync = B_TRUE;
	}
	} else {
	error = SET_ERROR(EBUSY);
	}
	} else if (spa->spa_l2cache.sav_vdevs != NULL &&
	nvlist_lookup_nvlist_array(spa->spa_l2cache.sav_config,
	ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache) == 0 &&
	(nv = spa_nvlist_lookup_by_guid(l2cache, nl2cache, guid)) != NULL) {
	vd_type = VDEV_TYPE_L2CACHE;
	vd_path = spa_strdup(fnvlist_lookup_string(
	nv, ZPOOL_CONFIG_PATH));
	/*
	* Cache devices can always be removed.
	*/
	vd = spa_lookup_by_guid(spa, guid, B_TRUE);

	/*
	* Stop trimming the cache device. We need to release the
	* config lock to allow the syncing of TRIM transactions
	* without releasing the spa_namespace_lock. The same
	* strategy is employed in spa_vdev_remove_top().
	*/
	spa_vdev_config_exit(spa, NULL,
	txg + TXG_CONCURRENT_STATES + TXG_DEFER_SIZE, 0, FTAG);
	mutex_enter(&vd->vdev_trim_lock);
	vdev_trim_stop(vd, VDEV_TRIM_CANCELED, NULL);
	mutex_exit(&vd->vdev_trim_lock);
	txg = spa_vdev_config_enter(spa);

	ev = spa_event_create(spa, vd, NULL, ESC_ZFS_VDEV_REMOVE_AUX);
	spa_vdev_remove_aux(spa->spa_l2cache.sav_config,
	ZPOOL_CONFIG_L2CACHE, l2cache, nl2cache, nv);
	spa_load_l2cache(spa);
	spa->spa_l2cache.sav_sync = B_TRUE;
	} else if (vd != NULL && vd->vdev_islog) {
	ASSERT(!locked);
	vd_type = VDEV_TYPE_LOG;
	vd_path = spa_strdup((vd->vdev_path != NULL) ?
	vd->vdev_path : "-");
	error = spa_vdev_remove_log(vd, &txg);
	} else if (vd != NULL) {
	ASSERT(!locked);
	error = spa_vdev_remove_top(vd, &txg);
	} else {
	/*
	* There is no vdev of any kind with the specified guid.
	*/
	error = SET_ERROR(ENOENT);
	}

	error_log = error;

	if (!locked)
	error = spa_vdev_exit(spa, NULL, txg, error);

	/*
	* Logging must be done outside the spa config lock. Otherwise,
	* this code path could end up holding the spa config lock while
	* waiting for a txg_sync so it can write to the internal log.
	* Doing that would prevent the txg sync from actually happening,
	* causing a deadlock.
	*/
	if (error_log == 0 && vd_type != NULL && vd_path != NULL) {
	spa_history_log_internal(spa, "vdev remove", NULL,
	"%s vdev (%s) %s", spa_name(spa), vd_type, vd_path);
	}
	if (vd_path != NULL)
	spa_strfree(vd_path);

	if (ev != NULL)
	spa_event_post(ev);

	return (error);
	}

	int
	spa_removal_get_stats(spa_t spa, pool_removal_stat_t prs)
	{
	prs->prs_state = spa->spa_removing_phys.sr_state;

	if (prs->prs_state == DSS_NONE)
	return (SET_ERROR(ENOENT));

	prs->prs_removing_vdev = spa->spa_removing_phys.sr_removing_vdev;
	prs->prs_start_time = spa->spa_removing_phys.sr_start_time;
	prs->prs_end_time = spa->spa_removing_phys.sr_end_time;
	prs->prs_to_copy = spa->spa_removing_phys.sr_to_copy;
	prs->prs_copied = spa->spa_removing_phys.sr_copied;

	prs->prs_mapping_memory = 0;
	uint64_t indirect_vdev_id =
	spa->spa_removing_phys.sr_prev_indirect_vdev;
	while (indirect_vdev_id != -1) {
	vdev_t *vd = spa->spa_root_vdev->vdev_child[indirect_vdev_id];
	vdev_indirect_config_t *vic = &vd->vdev_indirect_config;
	vdev_indirect_mapping_t *vim = vd->vdev_indirect_mapping;

	ASSERT3P(vd->vdev_ops, ==, &vdev_indirect_ops);
	prs->prs_mapping_memory += vdev_indirect_mapping_size(vim);
	indirect_vdev_id = vic->vic_prev_indirect_vdev;
	}

	return (0);
	}

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM(zfs_vdev, zfs_, removal_ignore_errors, INT, ZMOD_RW,
	"Ignore hard IO errors when removing device");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_, remove_max_segment, INT, ZMOD_RW,
	"Largest contiguous segment to allocate when removing device");

	ZFS_MODULE_PARAM(zfs_vdev, vdev_, removal_max_span, INT, ZMOD_RW,
	"Largest span of free chunks a remap segment can span");

	ZFS_MODULE_PARAM(zfs_vdev, zfs_, removal_suspend_progress, INT, ZMOD_RW,
	"Pause device removal after this many bytes are copied "
	"(debug use only - causes removal to hang)");
	/* END CSTYLED */

	EXPORT_SYMBOL(free_from_removing_vdev);
	EXPORT_SYMBOL(spa_removal_get_stats);
	EXPORT_SYMBOL(spa_remove_init);
	EXPORT_SYMBOL(spa_restart_removal);
	EXPORT_SYMBOL(spa_vdev_removal_destroy);
	EXPORT_SYMBOL(spa_vdev_remove);
	EXPORT_SYMBOL(spa_vdev_remove_cancel);
	EXPORT_SYMBOL(spa_vdev_remove_suspend);
	EXPORT_SYMBOL(svr_sync);
	diff --git a/sys/contrib/openzfs/module/zfs/zfs_fuid.c b/sys/contrib/openzfs/module/zfs/zfs_fuid.c
	index a90bf5feeea1..3aa60034d42a 100644
	--- a/sys/contrib/openzfs/module/zfs/zfs_fuid.c
	+++ b/sys/contrib/openzfs/module/zfs/zfs_fuid.c
	@@ -1,811 +1,811 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
	*/

	#include <sys/zfs_context.h>
	#include <sys/dmu.h>
	#include <sys/avl.h>
	#include <sys/zap.h>
	#include <sys/nvpair.h>
	#ifdef _KERNEL
	#include <sys/sid.h>
	#include <sys/zfs_vfsops.h>
	#include <sys/zfs_znode.h>
	#endif
	#include <sys/zfs_fuid.h>

	/*
	* FUID Domain table(s).
	*
	* The FUID table is stored as a packed nvlist of an array
	* of nvlists which contain an index, domain string and offset
	*
	* During file system initialization the nvlist(s) are read and
	* two AVL trees are created. One tree is keyed by the index number
	* and the other by the domain string. Nodes are never removed from
	* trees, but new entries may be added. If a new entry is added then
	* the zfsvfs->z_fuid_dirty flag is set to true and the caller will then
	* be responsible for calling zfs_fuid_sync() to sync the changes to disk.
	*
	*/

	#define FUID_IDX "fuid_idx"
	#define FUID_DOMAIN "fuid_domain"
	#define FUID_OFFSET "fuid_offset"
	#define FUID_NVP_ARRAY "fuid_nvlist"

	typedef struct fuid_domain {
	avl_node_t f_domnode;
	avl_node_t f_idxnode;
	ksiddomain_t *f_ksid;
	uint64_t f_idx;
	} fuid_domain_t;

	static char *nulldomain = "";

	/*
	* Compare two indexes.
	*/
	static int
	idx_compare(const void arg1, const void arg2)
	{
	const fuid_domain_t node1 = (const fuid_domain_t )arg1;
	const fuid_domain_t node2 = (const fuid_domain_t )arg2;

	return (TREE_CMP(node1->f_idx, node2->f_idx));
	}

	/*
	* Compare two domain strings.
	*/
	static int
	domain_compare(const void arg1, const void arg2)
	{
	const fuid_domain_t node1 = (const fuid_domain_t )arg1;
	const fuid_domain_t node2 = (const fuid_domain_t )arg2;
	int val;

	val = strcmp(node1->f_ksid->kd_name, node2->f_ksid->kd_name);

	return (TREE_ISIGN(val));
	}

	void
	zfs_fuid_avl_tree_create(avl_tree_t idx_tree, avl_tree_t domain_tree)
	{
	avl_create(idx_tree, idx_compare,
	sizeof (fuid_domain_t), offsetof(fuid_domain_t, f_idxnode));
	avl_create(domain_tree, domain_compare,
	sizeof (fuid_domain_t), offsetof(fuid_domain_t, f_domnode));
	}

	/*
	* load initial fuid domain and idx trees. This function is used by
	* both the kernel and zdb.
	*/
	uint64_t
	zfs_fuid_table_load(objset_t os, uint64_t fuid_obj, avl_tree_t idx_tree,
	avl_tree_t *domain_tree)
	{
	dmu_buf_t *db;
	uint64_t fuid_size;

	ASSERT(fuid_obj != 0);
	VERIFY(0 == dmu_bonus_hold(os, fuid_obj,
	FTAG, &db));
	fuid_size = (uint64_t )db->db_data;
	dmu_buf_rele(db, FTAG);

	if (fuid_size) {
	nvlist_t **fuidnvp;
	nvlist_t *nvp = NULL;
	uint_t count;
	char *packed;
	int i;

	packed = kmem_alloc(fuid_size, KM_SLEEP);
	VERIFY(dmu_read(os, fuid_obj, 0,
	fuid_size, packed, DMU_READ_PREFETCH) == 0);
	VERIFY(nvlist_unpack(packed, fuid_size,
	&nvp, 0) == 0);
	VERIFY(nvlist_lookup_nvlist_array(nvp, FUID_NVP_ARRAY,
	&fuidnvp, &count) == 0);

	for (i = 0; i != count; i++) {
	fuid_domain_t *domnode;
	char *domain;
	uint64_t idx;

	VERIFY(nvlist_lookup_string(fuidnvp[i], FUID_DOMAIN,
	&domain) == 0);
	VERIFY(nvlist_lookup_uint64(fuidnvp[i], FUID_IDX,
	&idx) == 0);

	domnode = kmem_alloc(sizeof (fuid_domain_t), KM_SLEEP);

	domnode->f_idx = idx;
	domnode->f_ksid = ksid_lookupdomain(domain);
	avl_add(idx_tree, domnode);
	avl_add(domain_tree, domnode);
	}
	nvlist_free(nvp);
	kmem_free(packed, fuid_size);
	}
	return (fuid_size);
	}

	void
	zfs_fuid_table_destroy(avl_tree_t idx_tree, avl_tree_t domain_tree)
	{
	fuid_domain_t *domnode;
	void *cookie;

	cookie = NULL;
	while ((domnode = avl_destroy_nodes(domain_tree, &cookie)))
	ksiddomain_rele(domnode->f_ksid);

	avl_destroy(domain_tree);
	cookie = NULL;
	while ((domnode = avl_destroy_nodes(idx_tree, &cookie)))
	kmem_free(domnode, sizeof (fuid_domain_t));
	avl_destroy(idx_tree);
	}

	char *
	zfs_fuid_idx_domain(avl_tree_t *idx_tree, uint32_t idx)
	{
	fuid_domain_t searchnode, *findnode;
	avl_index_t loc;

	searchnode.f_idx = idx;

	findnode = avl_find(idx_tree, &searchnode, &loc);

	return (findnode ? findnode->f_ksid->kd_name : nulldomain);
	}

	#ifdef _KERNEL
	/*
	* Load the fuid table(s) into memory.
	*/
	static void
	zfs_fuid_init(zfsvfs_t *zfsvfs)
	{
	rw_enter(&zfsvfs->z_fuid_lock, RW_WRITER);

	if (zfsvfs->z_fuid_loaded) {
	rw_exit(&zfsvfs->z_fuid_lock);
	return;
	}

	zfs_fuid_avl_tree_create(&zfsvfs->z_fuid_idx, &zfsvfs->z_fuid_domain);

	(void) zap_lookup(zfsvfs->z_os, MASTER_NODE_OBJ,
	ZFS_FUID_TABLES, 8, 1, &zfsvfs->z_fuid_obj);
	if (zfsvfs->z_fuid_obj != 0) {
	zfsvfs->z_fuid_size = zfs_fuid_table_load(zfsvfs->z_os,
	zfsvfs->z_fuid_obj, &zfsvfs->z_fuid_idx,
	&zfsvfs->z_fuid_domain);
	}

	zfsvfs->z_fuid_loaded = B_TRUE;
	rw_exit(&zfsvfs->z_fuid_lock);
	}

	/*
	* sync out AVL trees to persistent storage.
	*/
	void
	zfs_fuid_sync(zfsvfs_t zfsvfs, dmu_tx_t tx)
	{
	nvlist_t *nvp;
	nvlist_t **fuids;
	size_t nvsize = 0;
	char *packed;
	dmu_buf_t *db;
	fuid_domain_t *domnode;
	int numnodes;
	int i;

	if (!zfsvfs->z_fuid_dirty) {
	return;
	}

	rw_enter(&zfsvfs->z_fuid_lock, RW_WRITER);

	/*
	* First see if table needs to be created?
	*/
	if (zfsvfs->z_fuid_obj == 0) {
	zfsvfs->z_fuid_obj = dmu_object_alloc(zfsvfs->z_os,
	DMU_OT_FUID, 1 << 14, DMU_OT_FUID_SIZE,
	sizeof (uint64_t), tx);
	VERIFY(zap_add(zfsvfs->z_os, MASTER_NODE_OBJ,
	ZFS_FUID_TABLES, sizeof (uint64_t), 1,
	&zfsvfs->z_fuid_obj, tx) == 0);
	}

	VERIFY(nvlist_alloc(&nvp, NV_UNIQUE_NAME, KM_SLEEP) == 0);

	numnodes = avl_numnodes(&zfsvfs->z_fuid_idx);
	fuids = kmem_alloc(numnodes * sizeof (void *), KM_SLEEP);
	for (i = 0, domnode = avl_first(&zfsvfs->z_fuid_domain); domnode; i++,
	domnode = AVL_NEXT(&zfsvfs->z_fuid_domain, domnode)) {
	VERIFY(nvlist_alloc(&fuids[i], NV_UNIQUE_NAME, KM_SLEEP) == 0);
	VERIFY(nvlist_add_uint64(fuids[i], FUID_IDX,
	domnode->f_idx) == 0);
	VERIFY(nvlist_add_uint64(fuids[i], FUID_OFFSET, 0) == 0);
	VERIFY(nvlist_add_string(fuids[i], FUID_DOMAIN,
	domnode->f_ksid->kd_name) == 0);
	}
	- VERIFY(nvlist_add_nvlist_array(nvp, FUID_NVP_ARRAY,
	- fuids, numnodes) == 0);
	+ fnvlist_add_nvlist_array(nvp, FUID_NVP_ARRAY,
	+ (const nvlist_t * const *)fuids, numnodes);
	for (i = 0; i != numnodes; i++)
	nvlist_free(fuids[i]);
	kmem_free(fuids, numnodes * sizeof (void *));
	VERIFY(nvlist_size(nvp, &nvsize, NV_ENCODE_XDR) == 0);
	packed = kmem_alloc(nvsize, KM_SLEEP);
	VERIFY(nvlist_pack(nvp, &packed, &nvsize,
	NV_ENCODE_XDR, KM_SLEEP) == 0);
	nvlist_free(nvp);
	zfsvfs->z_fuid_size = nvsize;
	dmu_write(zfsvfs->z_os, zfsvfs->z_fuid_obj, 0,
	zfsvfs->z_fuid_size, packed, tx);
	kmem_free(packed, zfsvfs->z_fuid_size);
	VERIFY(0 == dmu_bonus_hold(zfsvfs->z_os, zfsvfs->z_fuid_obj,
	FTAG, &db));
	dmu_buf_will_dirty(db, tx);
	(uint64_t )db->db_data = zfsvfs->z_fuid_size;
	dmu_buf_rele(db, FTAG);

	zfsvfs->z_fuid_dirty = B_FALSE;
	rw_exit(&zfsvfs->z_fuid_lock);
	}

	/*
	* Query domain table for a given domain.
	*
	* If domain isn't found and addok is set, it is added to AVL trees and
	* the zfsvfs->z_fuid_dirty flag will be set to TRUE. It will then be
	* necessary for the caller or another thread to detect the dirty table
	* and sync out the changes.
	*/
	int
	zfs_fuid_find_by_domain(zfsvfs_t zfsvfs, const char domain,
	char **retdomain, boolean_t addok)
	{
	fuid_domain_t searchnode, *findnode;
	avl_index_t loc;
	krw_t rw = RW_READER;

	/*
	* If the dummy "nobody" domain then return an index of 0
	* to cause the created FUID to be a standard POSIX id
	* for the user nobody.
	*/
	if (domain[0] == '\0') {
	if (retdomain)
	*retdomain = nulldomain;
	return (0);
	}

	searchnode.f_ksid = ksid_lookupdomain(domain);
	if (retdomain)
	*retdomain = searchnode.f_ksid->kd_name;
	if (!zfsvfs->z_fuid_loaded)
	zfs_fuid_init(zfsvfs);

	retry:
	rw_enter(&zfsvfs->z_fuid_lock, rw);
	findnode = avl_find(&zfsvfs->z_fuid_domain, &searchnode, &loc);

	if (findnode) {
	rw_exit(&zfsvfs->z_fuid_lock);
	ksiddomain_rele(searchnode.f_ksid);
	return (findnode->f_idx);
	} else if (addok) {
	fuid_domain_t *domnode;
	uint64_t retidx;

	if (rw == RW_READER && !rw_tryupgrade(&zfsvfs->z_fuid_lock)) {
	rw_exit(&zfsvfs->z_fuid_lock);
	rw = RW_WRITER;
	goto retry;
	}

	domnode = kmem_alloc(sizeof (fuid_domain_t), KM_SLEEP);
	domnode->f_ksid = searchnode.f_ksid;

	retidx = domnode->f_idx = avl_numnodes(&zfsvfs->z_fuid_idx) + 1;

	avl_add(&zfsvfs->z_fuid_domain, domnode);
	avl_add(&zfsvfs->z_fuid_idx, domnode);
	zfsvfs->z_fuid_dirty = B_TRUE;
	rw_exit(&zfsvfs->z_fuid_lock);
	return (retidx);
	} else {
	rw_exit(&zfsvfs->z_fuid_lock);
	return (-1);
	}
	}

	/*
	* Query domain table by index, returning domain string
	*
	* Returns a pointer from an avl node of the domain string.
	*
	*/
	const char *
	zfs_fuid_find_by_idx(zfsvfs_t *zfsvfs, uint32_t idx)
	{
	char *domain;

	if (idx == 0 \|\| !zfsvfs->z_use_fuids)
	return (NULL);

	if (!zfsvfs->z_fuid_loaded)
	zfs_fuid_init(zfsvfs);

	rw_enter(&zfsvfs->z_fuid_lock, RW_READER);

	if (zfsvfs->z_fuid_obj \|\| zfsvfs->z_fuid_dirty)
	domain = zfs_fuid_idx_domain(&zfsvfs->z_fuid_idx, idx);
	else
	domain = nulldomain;
	rw_exit(&zfsvfs->z_fuid_lock);

	ASSERT(domain);
	return (domain);
	}

	void
	zfs_fuid_map_ids(znode_t zp, cred_t cr, uid_t uidp, uid_t gidp)
	{
	*uidp = zfs_fuid_map_id(ZTOZSB(zp), KUID_TO_SUID(ZTOUID(zp)),
	cr, ZFS_OWNER);
	*gidp = zfs_fuid_map_id(ZTOZSB(zp), KGID_TO_SGID(ZTOGID(zp)),
	cr, ZFS_GROUP);
	}

	#ifdef __FreeBSD__
	uid_t
	zfs_fuid_map_id(zfsvfs_t *zfsvfs, uint64_t fuid,
	cred_t *cr, zfs_fuid_type_t type)
	{
	uint32_t index = FUID_INDEX(fuid);

	if (index == 0)
	return (fuid);

	return (UID_NOBODY);
	}
	#elif defined(__linux__)
	uid_t
	zfs_fuid_map_id(zfsvfs_t *zfsvfs, uint64_t fuid,
	cred_t *cr, zfs_fuid_type_t type)
	{
	/*
	* The Linux port only supports POSIX IDs, use the passed id.
	*/
	return (fuid);
	}

	#else
	uid_t
	zfs_fuid_map_id(zfsvfs_t *zfsvfs, uint64_t fuid,
	cred_t *cr, zfs_fuid_type_t type)
	{
	uint32_t index = FUID_INDEX(fuid);
	const char *domain;
	uid_t id;

	if (index == 0)
	return (fuid);

	domain = zfs_fuid_find_by_idx(zfsvfs, index);
	ASSERT(domain != NULL);

	if (type == ZFS_OWNER \|\| type == ZFS_ACE_USER) {
	(void) kidmap_getuidbysid(crgetzone(cr), domain,
	FUID_RID(fuid), &id);
	} else {
	(void) kidmap_getgidbysid(crgetzone(cr), domain,
	FUID_RID(fuid), &id);
	}
	return (id);
	}
	#endif

	/*
	* Add a FUID node to the list of fuid's being created for this
	* ACL
	*
	* If ACL has multiple domains, then keep only one copy of each unique
	* domain.
	*/
	void
	zfs_fuid_node_add(zfs_fuid_info_t *fuidpp, const char domain, uint32_t rid,
	uint64_t idx, uint64_t id, zfs_fuid_type_t type)
	{
	zfs_fuid_t *fuid;
	zfs_fuid_domain_t *fuid_domain;
	zfs_fuid_info_t *fuidp;
	uint64_t fuididx;
	boolean_t found = B_FALSE;

	if (*fuidpp == NULL)
	*fuidpp = zfs_fuid_info_alloc();

	fuidp = *fuidpp;
	/*
	* First find fuid domain index in linked list
	*
	* If one isn't found then create an entry.
	*/

	for (fuididx = 1, fuid_domain = list_head(&fuidp->z_domains);
	fuid_domain; fuid_domain = list_next(&fuidp->z_domains,
	fuid_domain), fuididx++) {
	if (idx == fuid_domain->z_domidx) {
	found = B_TRUE;
	break;
	}
	}

	if (!found) {
	fuid_domain = kmem_alloc(sizeof (zfs_fuid_domain_t), KM_SLEEP);
	fuid_domain->z_domain = domain;
	fuid_domain->z_domidx = idx;
	list_insert_tail(&fuidp->z_domains, fuid_domain);
	fuidp->z_domain_str_sz += strlen(domain) + 1;
	fuidp->z_domain_cnt++;
	}

	if (type == ZFS_ACE_USER \|\| type == ZFS_ACE_GROUP) {

	/*
	* Now allocate fuid entry and add it on the end of the list
	*/

	fuid = kmem_alloc(sizeof (zfs_fuid_t), KM_SLEEP);
	fuid->z_id = id;
	fuid->z_domidx = idx;
	fuid->z_logfuid = FUID_ENCODE(fuididx, rid);

	list_insert_tail(&fuidp->z_fuids, fuid);
	fuidp->z_fuid_cnt++;
	} else {
	if (type == ZFS_OWNER)
	fuidp->z_fuid_owner = FUID_ENCODE(fuididx, rid);
	else
	fuidp->z_fuid_group = FUID_ENCODE(fuididx, rid);
	}
	}

	#ifdef HAVE_KSID
	/*
	* Create a file system FUID, based on information in the users cred
	*
	* If cred contains KSID_OWNER then it should be used to determine
	* the uid otherwise cred's uid will be used. By default cred's gid
	* is used unless it's an ephemeral ID in which case KSID_GROUP will
	* be used if it exists.
	*/
	uint64_t
	zfs_fuid_create_cred(zfsvfs_t *zfsvfs, zfs_fuid_type_t type,
	cred_t cr, zfs_fuid_info_t *fuidp)
	{
	uint64_t idx;
	ksid_t *ksid;
	uint32_t rid;
	char *kdomain;
	const char *domain;
	uid_t id;

	VERIFY(type == ZFS_OWNER \|\| type == ZFS_GROUP);

	ksid = crgetsid(cr, (type == ZFS_OWNER) ? KSID_OWNER : KSID_GROUP);

	if (!zfsvfs->z_use_fuids \|\| (ksid == NULL)) {
	id = (type == ZFS_OWNER) ? crgetuid(cr) : crgetgid(cr);

	if (IS_EPHEMERAL(id))
	return ((type == ZFS_OWNER) ? UID_NOBODY : GID_NOBODY);

	return ((uint64_t)id);
	}

	/*
	* ksid is present and FUID is supported
	*/
	id = (type == ZFS_OWNER) ? ksid_getid(ksid) : crgetgid(cr);

	if (!IS_EPHEMERAL(id))
	return ((uint64_t)id);

	if (type == ZFS_GROUP)
	id = ksid_getid(ksid);

	rid = ksid_getrid(ksid);
	domain = ksid_getdomain(ksid);

	idx = zfs_fuid_find_by_domain(zfsvfs, domain, &kdomain, B_TRUE);

	zfs_fuid_node_add(fuidp, kdomain, rid, idx, id, type);

	return (FUID_ENCODE(idx, rid));
	}
	#endif /* HAVE_KSID */

	/*
	* Create a file system FUID for an ACL ace
	* or a chown/chgrp of the file.
	* This is similar to zfs_fuid_create_cred, except that
	* we can't find the domain + rid information in the
	* cred. Instead we have to query Winchester for the
	* domain and rid.
	*
	* During replay operations the domain+rid information is
	* found in the zfs_fuid_info_t that the replay code has
	* attached to the zfsvfs of the file system.
	*/
	uint64_t
	zfs_fuid_create(zfsvfs_t zfsvfs, uint64_t id, cred_t cr,
	zfs_fuid_type_t type, zfs_fuid_info_t **fuidpp)
	{
	#ifdef HAVE_KSID
	const char *domain;
	char *kdomain;
	uint32_t fuid_idx = FUID_INDEX(id);
	uint32_t rid = 0;
	idmap_stat status;
	uint64_t idx = UID_NOBODY;
	zfs_fuid_t *zfuid = NULL;
	zfs_fuid_info_t *fuidp = NULL;

	/*
	* If POSIX ID, or entry is already a FUID then
	* just return the id
	*
	* We may also be handed an already FUID'ized id via
	* chmod.
	*/

	if (!zfsvfs->z_use_fuids \|\| !IS_EPHEMERAL(id) \|\| fuid_idx != 0)
	return (id);

	if (zfsvfs->z_replay) {
	fuidp = zfsvfs->z_fuid_replay;

	/*
	* If we are passed an ephemeral id, but no
	* fuid_info was logged then return NOBODY.
	* This is most likely a result of idmap service
	* not being available.
	*/
	if (fuidp == NULL)
	return (UID_NOBODY);

	VERIFY3U(type, >=, ZFS_OWNER);
	VERIFY3U(type, <=, ZFS_ACE_GROUP);

	switch (type) {
	case ZFS_ACE_USER:
	case ZFS_ACE_GROUP:
	zfuid = list_head(&fuidp->z_fuids);
	rid = FUID_RID(zfuid->z_logfuid);
	idx = FUID_INDEX(zfuid->z_logfuid);
	break;
	case ZFS_OWNER:
	rid = FUID_RID(fuidp->z_fuid_owner);
	idx = FUID_INDEX(fuidp->z_fuid_owner);
	break;
	case ZFS_GROUP:
	rid = FUID_RID(fuidp->z_fuid_group);
	idx = FUID_INDEX(fuidp->z_fuid_group);
	break;
	};
	domain = fuidp->z_domain_table[idx - 1];
	} else {
	if (type == ZFS_OWNER \|\| type == ZFS_ACE_USER)
	status = kidmap_getsidbyuid(crgetzone(cr), id,
	&domain, &rid);
	else
	status = kidmap_getsidbygid(crgetzone(cr), id,
	&domain, &rid);

	if (status != 0) {
	/*
	* When returning nobody we will need to
	* make a dummy fuid table entry for logging
	* purposes.
	*/
	rid = UID_NOBODY;
	domain = nulldomain;
	}
	}

	idx = zfs_fuid_find_by_domain(zfsvfs, domain, &kdomain, B_TRUE);

	if (!zfsvfs->z_replay)
	zfs_fuid_node_add(fuidpp, kdomain,
	rid, idx, id, type);
	else if (zfuid != NULL) {
	list_remove(&fuidp->z_fuids, zfuid);
	kmem_free(zfuid, sizeof (zfs_fuid_t));
	}
	return (FUID_ENCODE(idx, rid));
	#else
	/*
	* The Linux port only supports POSIX IDs, use the passed id.
	*/
	return (id);
	#endif
	}

	void
	zfs_fuid_destroy(zfsvfs_t *zfsvfs)
	{
	rw_enter(&zfsvfs->z_fuid_lock, RW_WRITER);
	if (!zfsvfs->z_fuid_loaded) {
	rw_exit(&zfsvfs->z_fuid_lock);
	return;
	}
	zfs_fuid_table_destroy(&zfsvfs->z_fuid_idx, &zfsvfs->z_fuid_domain);
	rw_exit(&zfsvfs->z_fuid_lock);
	}

	/*
	* Allocate zfs_fuid_info for tracking FUIDs created during
	* zfs_mknode, VOP_SETATTR() or VOP_SETSECATTR()
	*/
	zfs_fuid_info_t *
	zfs_fuid_info_alloc(void)
	{
	zfs_fuid_info_t *fuidp;

	fuidp = kmem_zalloc(sizeof (zfs_fuid_info_t), KM_SLEEP);
	list_create(&fuidp->z_domains, sizeof (zfs_fuid_domain_t),
	offsetof(zfs_fuid_domain_t, z_next));
	list_create(&fuidp->z_fuids, sizeof (zfs_fuid_t),
	offsetof(zfs_fuid_t, z_next));
	return (fuidp);
	}

	/*
	* Release all memory associated with zfs_fuid_info_t
	*/
	void
	zfs_fuid_info_free(zfs_fuid_info_t *fuidp)
	{
	zfs_fuid_t *zfuid;
	zfs_fuid_domain_t *zdomain;

	while ((zfuid = list_head(&fuidp->z_fuids)) != NULL) {
	list_remove(&fuidp->z_fuids, zfuid);
	kmem_free(zfuid, sizeof (zfs_fuid_t));
	}

	if (fuidp->z_domain_table != NULL)
	kmem_free(fuidp->z_domain_table,
	(sizeof (char )) fuidp->z_domain_cnt);

	while ((zdomain = list_head(&fuidp->z_domains)) != NULL) {
	list_remove(&fuidp->z_domains, zdomain);
	kmem_free(zdomain, sizeof (zfs_fuid_domain_t));
	}

	kmem_free(fuidp, sizeof (zfs_fuid_info_t));
	}

	/*
	* Check to see if id is a groupmember. If cred
	* has ksid info then sidlist is checked first
	* and if still not found then POSIX groups are checked
	*
	* Will use a straight FUID compare when possible.
	*/
	boolean_t
	zfs_groupmember(zfsvfs_t zfsvfs, uint64_t id, cred_t cr)
	{
	uid_t gid;

	#ifdef illumos
	ksid_t *ksid = crgetsid(cr, KSID_GROUP);
	ksidlist_t *ksidlist = crgetsidlist(cr);

	if (ksid && ksidlist) {
	int i;
	ksid_t *ksid_groups;
	uint32_t idx = FUID_INDEX(id);
	uint32_t rid = FUID_RID(id);

	ksid_groups = ksidlist->ksl_sids;

	for (i = 0; i != ksidlist->ksl_nsid; i++) {
	if (idx == 0) {
	if (id != IDMAP_WK_CREATOR_GROUP_GID &&
	id == ksid_groups[i].ks_id) {
	return (B_TRUE);
	}
	} else {
	const char *domain;

	domain = zfs_fuid_find_by_idx(zfsvfs, idx);
	ASSERT(domain != NULL);

	if (strcmp(domain,
	IDMAP_WK_CREATOR_SID_AUTHORITY) == 0)
	return (B_FALSE);

	if ((strcmp(domain,
	ksid_groups[i].ks_domain->kd_name) == 0) &&
	rid == ksid_groups[i].ks_rid)
	return (B_TRUE);
	}
	}
	}
	#endif /* illumos */

	/*
	* Not found in ksidlist, check posix groups
	*/
	gid = zfs_fuid_map_id(zfsvfs, id, cr, ZFS_GROUP);
	return (groupmember(gid, cr));
	}

	void
	zfs_fuid_txhold(zfsvfs_t zfsvfs, dmu_tx_t tx)
	{
	if (zfsvfs->z_fuid_obj == 0) {
	dmu_tx_hold_bonus(tx, DMU_NEW_OBJECT);
	dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
	FUID_SIZE_ESTIMATE(zfsvfs));
	dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, FALSE, NULL);
	} else {
	dmu_tx_hold_bonus(tx, zfsvfs->z_fuid_obj);
	dmu_tx_hold_write(tx, zfsvfs->z_fuid_obj, 0,
	FUID_SIZE_ESTIMATE(zfsvfs));
	}
	}

	/*
	* buf must be big enough (eg, 32 bytes)
	*/
	int
	zfs_id_to_fuidstr(zfsvfs_t zfsvfs, const char domain, uid_t rid,
	char *buf, size_t len, boolean_t addok)
	{
	uint64_t fuid;
	int domainid = 0;

	if (domain && domain[0]) {
	domainid = zfs_fuid_find_by_domain(zfsvfs, domain, NULL, addok);
	if (domainid == -1)
	return (SET_ERROR(ENOENT));
	}
	fuid = FUID_ENCODE(domainid, rid);
	(void) snprintf(buf, len, "%llx", (longlong_t)fuid);
	return (0);
	}
	#endif
	diff --git a/sys/contrib/openzfs/module/zfs/zfs_ioctl.c b/sys/contrib/openzfs/module/zfs/zfs_ioctl.c
	index 96a021acbc95..ca2da561220b 100644
	--- a/sys/contrib/openzfs/module/zfs/zfs_ioctl.c
	+++ b/sys/contrib/openzfs/module/zfs/zfs_ioctl.c
	@@ -1,7775 +1,7875 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Portions Copyright 2011 Martin Matuska
	* Copyright 2015, OmniTI Computer Consulting, Inc. All rights reserved.
	* Portions Copyright 2012 Pawel Jakub Dawidek <pawel@dawidek.net>
	* Copyright (c) 2014, 2016 Joyent, Inc. All rights reserved.
	* Copyright 2016 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2014, Joyent, Inc. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
	* Copyright (c) 2013 Steven Hartland. All rights reserved.
	* Copyright (c) 2014 Integros [integros.com]
	* Copyright 2016 Toomas Soome <tsoome@me.com>
	* Copyright (c) 2016 Actifio, Inc. All rights reserved.
	* Copyright (c) 2018, loli10K <ezomori.nozomu@gmail.com>. All rights reserved.
	* Copyright 2017 RackTop Systems.
	* Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
	* Copyright (c) 2019 Datto Inc.
	* Copyright (c) 2019, 2020 by Christian Schwarz. All rights reserved.
	- * Copyright (c) 2019, Klara Inc.
	+ * Copyright (c) 2019, 2021, Klara Inc.
	* Copyright (c) 2019, Allan Jude
	*/

	/*
	* ZFS ioctls.
	*
	* This file handles the ioctls to /dev/zfs, used for configuring ZFS storage
	* pools and filesystems, e.g. with /sbin/zfs and /sbin/zpool.
	*
	* There are two ways that we handle ioctls: the legacy way where almost
	* all of the logic is in the ioctl callback, and the new way where most
	* of the marshalling is handled in the common entry point, zfsdev_ioctl().
	*
	* Non-legacy ioctls should be registered by calling
	* zfs_ioctl_register() from zfs_ioctl_init(). The ioctl is invoked
	* from userland by lzc_ioctl().
	*
	* The registration arguments are as follows:
	*
	* const char *name
	* The name of the ioctl. This is used for history logging. If the
	* ioctl returns successfully (the callback returns 0), and allow_log
	* is true, then a history log entry will be recorded with the input &
	* output nvlists. The log entry can be printed with "zpool history -i".
	*
	* zfs_ioc_t ioc
	* The ioctl request number, which userland will pass to ioctl(2).
	* We want newer versions of libzfs and libzfs_core to run against
	* existing zfs kernel modules (i.e. a deferred reboot after an update).
	* Therefore the ioctl numbers cannot change from release to release.
	*
	* zfs_secpolicy_func_t *secpolicy
	* This function will be called before the zfs_ioc_func_t, to
	* determine if this operation is permitted. It should return EPERM
	* on failure, and 0 on success. Checks include determining if the
	* dataset is visible in this zone, and if the user has either all
	* zfs privileges in the zone (SYS_MOUNT), or has been granted permission
	* to do this operation on this dataset with "zfs allow".
	*
	* zfs_ioc_namecheck_t namecheck
	* This specifies what to expect in the zfs_cmd_t:zc_name -- a pool
	* name, a dataset name, or nothing. If the name is not well-formed,
	* the ioctl will fail and the callback will not be called.
	* Therefore, the callback can assume that the name is well-formed
	* (e.g. is null-terminated, doesn't have more than one '@' character,
	* doesn't have invalid characters).
	*
	* zfs_ioc_poolcheck_t pool_check
	* This specifies requirements on the pool state. If the pool does
	* not meet them (is suspended or is readonly), the ioctl will fail
	* and the callback will not be called. If any checks are specified
	* (i.e. it is not POOL_CHECK_NONE), namecheck must not be NO_NAME.
	* Multiple checks can be or-ed together (e.g. POOL_CHECK_SUSPENDED \|
	* POOL_CHECK_READONLY).
	*
	* zfs_ioc_key_t *nvl_keys
	* The list of expected/allowable innvl input keys. This list is used
	* to validate the nvlist input to the ioctl.
	*
	* boolean_t smush_outnvlist
	* If smush_outnvlist is true, then the output is presumed to be a
	* list of errors, and it will be "smushed" down to fit into the
	* caller's buffer, by removing some entries and replacing them with a
	* single "N_MORE_ERRORS" entry indicating how many were removed. See
	* nvlist_smush() for details. If smush_outnvlist is false, and the
	* outnvlist does not fit into the userland-provided buffer, then the
	* ioctl will fail with ENOMEM.
	*
	* zfs_ioc_func_t *func
	* The callback function that will perform the operation.
	*
	* The callback should return 0 on success, or an error number on
	* failure. If the function fails, the userland ioctl will return -1,
	* and errno will be set to the callback's return value. The callback
	* will be called with the following arguments:
	*
	* const char *name
	* The name of the pool or dataset to operate on, from
	* zfs_cmd_t:zc_name. The 'namecheck' argument specifies the
	* expected type (pool, dataset, or none).
	*
	* nvlist_t *innvl
	* The input nvlist, deserialized from zfs_cmd_t:zc_nvlist_src. Or
	* NULL if no input nvlist was provided. Changes to this nvlist are
	* ignored. If the input nvlist could not be deserialized, the
	* ioctl will fail and the callback will not be called.
	*
	* nvlist_t *outnvl
	* The output nvlist, initially empty. The callback can fill it in,
	* and it will be returned to userland by serializing it into
	* zfs_cmd_t:zc_nvlist_dst. If it is non-empty, and serialization
	* fails (e.g. because the caller didn't supply a large enough
	* buffer), then the overall ioctl will fail. See the
	* 'smush_nvlist' argument above for additional behaviors.
	*
	* There are two typical uses of the output nvlist:
	* - To return state, e.g. property values. In this case,
	* smush_outnvlist should be false. If the buffer was not large
	* enough, the caller will reallocate a larger buffer and try
	* the ioctl again.
	*
	* - To return multiple errors from an ioctl which makes on-disk
	* changes. In this case, smush_outnvlist should be true.
	* Ioctls which make on-disk modifications should generally not
	* use the outnvl if they succeed, because the caller can not
	* distinguish between the operation failing, and
	* deserialization failing.
	*
	* IOCTL Interface Errors
	*
	* The following ioctl input errors can be returned:
	* ZFS_ERR_IOC_CMD_UNAVAIL the ioctl number is not supported by kernel
	* ZFS_ERR_IOC_ARG_UNAVAIL an input argument is not supported by kernel
	* ZFS_ERR_IOC_ARG_REQUIRED a required input argument is missing
	* ZFS_ERR_IOC_ARG_BADTYPE an input argument has an invalid type
	*/

	#include <sys/types.h>
	#include <sys/param.h>
	#include <sys/errno.h>
	#include <sys/uio_impl.h>
	#include <sys/file.h>
	#include <sys/kmem.h>
	#include <sys/cmn_err.h>
	#include <sys/stat.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/zfs_quota.h>
	#include <sys/zfs_vfsops.h>
	#include <sys/zfs_znode.h>
	#include <sys/zap.h>
	#include <sys/spa.h>
	#include <sys/spa_impl.h>
	#include <sys/vdev.h>
	#include <sys/vdev_impl.h>
	#include <sys/dmu.h>
	#include <sys/dsl_dir.h>
	#include <sys/dsl_dataset.h>
	#include <sys/dsl_prop.h>
	#include <sys/dsl_deleg.h>
	#include <sys/dmu_objset.h>
	#include <sys/dmu_impl.h>
	#include <sys/dmu_redact.h>
	#include <sys/dmu_tx.h>
	#include <sys/sunddi.h>
	#include <sys/policy.h>
	#include <sys/zone.h>
	#include <sys/nvpair.h>
	#include <sys/pathname.h>
	#include <sys/fs/zfs.h>
	#include <sys/zfs_ctldir.h>
	#include <sys/zfs_dir.h>
	#include <sys/zfs_onexit.h>
	#include <sys/zvol.h>
	#include <sys/dsl_scan.h>
	#include <sys/fm/util.h>
	#include <sys/dsl_crypt.h>
	#include <sys/rrwlock.h>
	#include <sys/zfs_file.h>

	#include <sys/dmu_recv.h>
	#include <sys/dmu_send.h>
	#include <sys/dmu_recv.h>
	#include <sys/dsl_destroy.h>
	#include <sys/dsl_bookmark.h>
	#include <sys/dsl_userhold.h>
	#include <sys/zfeature.h>
	#include <sys/zcp.h>
	#include <sys/zio_checksum.h>
	#include <sys/vdev_removal.h>
	#include <sys/vdev_impl.h>
	#include <sys/vdev_initialize.h>
	#include <sys/vdev_trim.h>

	#include "zfs_namecheck.h"
	#include "zfs_prop.h"
	#include "zfs_deleg.h"
	#include "zfs_comutil.h"

	#include <sys/lua/lua.h>
	#include <sys/lua/lauxlib.h>
	#include <sys/zfs_ioctl_impl.h>

	kmutex_t zfsdev_state_lock;
	zfsdev_state_t *zfsdev_state_list;

	/*
	* Limit maximum nvlist size. We don't want users passing in insane values
	* for zc->zc_nvlist_src_size, since we will need to allocate that much memory.
	* Defaults to 0=auto which is handled by platform code.
	*/
	unsigned long zfs_max_nvlist_src_size = 0;

	/*
	* When logging the output nvlist of an ioctl in the on-disk history, limit
	* the logged size to this many bytes. This must be less than DMU_MAX_ACCESS.
	* This applies primarily to zfs_ioc_channel_program().
	*/
	unsigned long zfs_history_output_max = 1024 * 1024;

	uint_t zfs_fsyncer_key;
	uint_t zfs_allow_log_key;

	/* DATA_TYPE_ANY is used when zkey_type can vary. */
	#define DATA_TYPE_ANY DATA_TYPE_UNKNOWN

	typedef struct zfs_ioc_vec {
	zfs_ioc_legacy_func_t *zvec_legacy_func;
	zfs_ioc_func_t *zvec_func;
	zfs_secpolicy_func_t *zvec_secpolicy;
	zfs_ioc_namecheck_t zvec_namecheck;
	boolean_t zvec_allow_log;
	zfs_ioc_poolcheck_t zvec_pool_check;
	boolean_t zvec_smush_outnvlist;
	const char *zvec_name;
	const zfs_ioc_key_t *zvec_nvl_keys;
	size_t zvec_nvl_key_count;
	} zfs_ioc_vec_t;

	/* This array is indexed by zfs_userquota_prop_t */
	static const char *userquota_perms[] = {
	ZFS_DELEG_PERM_USERUSED,
	ZFS_DELEG_PERM_USERQUOTA,
	ZFS_DELEG_PERM_GROUPUSED,
	ZFS_DELEG_PERM_GROUPQUOTA,
	ZFS_DELEG_PERM_USEROBJUSED,
	ZFS_DELEG_PERM_USEROBJQUOTA,
	ZFS_DELEG_PERM_GROUPOBJUSED,
	ZFS_DELEG_PERM_GROUPOBJQUOTA,
	ZFS_DELEG_PERM_PROJECTUSED,
	ZFS_DELEG_PERM_PROJECTQUOTA,
	ZFS_DELEG_PERM_PROJECTOBJUSED,
	ZFS_DELEG_PERM_PROJECTOBJQUOTA,
	};

	static int zfs_ioc_userspace_upgrade(zfs_cmd_t *zc);
	static int zfs_ioc_id_quota_upgrade(zfs_cmd_t *zc);
	static int zfs_check_settable(const char name, nvpair_t property,
	cred_t *cr);
	static int zfs_check_clearable(const char dataset, nvlist_t props,
	nvlist_t **errors);
	static int zfs_fill_zplprops_root(uint64_t, nvlist_t , nvlist_t ,
	boolean_t *);
	int zfs_set_prop_nvlist(const char , zprop_source_t, nvlist_t , nvlist_t *);
	static int get_nvlist(uint64_t nvl, uint64_t size, int iflag, nvlist_t **nvp);

	static void
	history_str_free(char *buf)
	{
	kmem_free(buf, HIS_MAX_RECORD_LEN);
	}

	static char *
	history_str_get(zfs_cmd_t *zc)
	{
	char *buf;

	if (zc->zc_history == 0)
	return (NULL);

	buf = kmem_alloc(HIS_MAX_RECORD_LEN, KM_SLEEP);
	if (copyinstr((void *)(uintptr_t)zc->zc_history,
	buf, HIS_MAX_RECORD_LEN, NULL) != 0) {
	history_str_free(buf);
	return (NULL);
	}

	buf[HIS_MAX_RECORD_LEN -1] = '\0';

	return (buf);
	}

	/*
	* Return non-zero if the spa version is less than requested version.
	*/
	static int
	zfs_earlier_version(const char *name, int version)
	{
	spa_t *spa;

	if (spa_open(name, &spa, FTAG) == 0) {
	if (spa_version(spa) < version) {
	spa_close(spa, FTAG);
	return (1);
	}
	spa_close(spa, FTAG);
	}
	return (0);
	}

	/*
	* Return TRUE if the ZPL version is less than requested version.
	*/
	static boolean_t
	zpl_earlier_version(const char *name, int version)
	{
	objset_t *os;
	boolean_t rc = B_TRUE;

	if (dmu_objset_hold(name, FTAG, &os) == 0) {
	uint64_t zplversion;

	if (dmu_objset_type(os) != DMU_OST_ZFS) {
	dmu_objset_rele(os, FTAG);
	return (B_TRUE);
	}
	/* XXX reading from non-owned objset */
	if (zfs_get_zplprop(os, ZFS_PROP_VERSION, &zplversion) == 0)
	rc = zplversion < version;
	dmu_objset_rele(os, FTAG);
	}
	return (rc);
	}

	static void
	zfs_log_history(zfs_cmd_t *zc)
	{
	spa_t *spa;
	char *buf;

	if ((buf = history_str_get(zc)) == NULL)
	return;

	if (spa_open(zc->zc_name, &spa, FTAG) == 0) {
	if (spa_version(spa) >= SPA_VERSION_ZPOOL_HISTORY)
	(void) spa_history_log(spa, buf);
	spa_close(spa, FTAG);
	}
	history_str_free(buf);
	}

	/*
	* Policy for top-level read operations (list pools). Requires no privileges,
	* and can be used in the local zone, as there is no associated dataset.
	*/
	/* ARGSUSED */
	static int
	zfs_secpolicy_none(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (0);
	}

	/*
	* Policy for dataset read operations (list children, get statistics). Requires
	* no privileges, but must be visible in the local zone.
	*/
	/* ARGSUSED */
	static int
	zfs_secpolicy_read(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	if (INGLOBALZONE(curproc) \|\|
	zone_dataset_visible(zc->zc_name, NULL))
	return (0);

	return (SET_ERROR(ENOENT));
	}

	static int
	zfs_dozonecheck_impl(const char dataset, uint64_t zoned, cred_t cr)
	{
	int writable = 1;

	/*
	* The dataset must be visible by this zone -- check this first
	* so they don't see EPERM on something they shouldn't know about.
	*/
	if (!INGLOBALZONE(curproc) &&
	!zone_dataset_visible(dataset, &writable))
	return (SET_ERROR(ENOENT));

	if (INGLOBALZONE(curproc)) {
	/*
	* If the fs is zoned, only root can access it from the
	* global zone.
	*/
	if (secpolicy_zfs(cr) && zoned)
	return (SET_ERROR(EPERM));
	} else {
	/*
	* If we are in a local zone, the 'zoned' property must be set.
	*/
	if (!zoned)
	return (SET_ERROR(EPERM));

	/* must be writable by this zone */
	if (!writable)
	return (SET_ERROR(EPERM));
	}
	return (0);
	}

	static int
	zfs_dozonecheck(const char dataset, cred_t cr)
	{
	uint64_t zoned;

	if (dsl_prop_get_integer(dataset, zfs_prop_to_name(ZFS_PROP_ZONED),
	&zoned, NULL))
	return (SET_ERROR(ENOENT));

	return (zfs_dozonecheck_impl(dataset, zoned, cr));
	}

	static int
	zfs_dozonecheck_ds(const char dataset, dsl_dataset_t ds, cred_t *cr)
	{
	uint64_t zoned;

	if (dsl_prop_get_int_ds(ds, zfs_prop_to_name(ZFS_PROP_ZONED), &zoned))
	return (SET_ERROR(ENOENT));

	return (zfs_dozonecheck_impl(dataset, zoned, cr));
	}

	static int
	zfs_secpolicy_write_perms_ds(const char name, dsl_dataset_t ds,
	const char perm, cred_t cr)
	{
	int error;

	error = zfs_dozonecheck_ds(name, ds, cr);
	if (error == 0) {
	error = secpolicy_zfs(cr);
	if (error != 0)
	error = dsl_deleg_access_impl(ds, perm, cr);
	}
	return (error);
	}

	static int
	zfs_secpolicy_write_perms(const char name, const char perm, cred_t *cr)
	{
	int error;
	dsl_dataset_t *ds;
	dsl_pool_t *dp;

	/*
	* First do a quick check for root in the global zone, which
	* is allowed to do all write_perms. This ensures that zfs_ioc_*
	* will get to handle nonexistent datasets.
	*/
	if (INGLOBALZONE(curproc) && secpolicy_zfs(cr) == 0)
	return (0);

	error = dsl_pool_hold(name, FTAG, &dp);
	if (error != 0)
	return (error);

	error = dsl_dataset_hold(dp, name, FTAG, &ds);
	if (error != 0) {
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	error = zfs_secpolicy_write_perms_ds(name, ds, perm, cr);

	dsl_dataset_rele(ds, FTAG);
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	/*
	* Policy for setting the security label property.
	*
	* Returns 0 for success, non-zero for access and other errors.
	*/
	static int
	zfs_set_slabel_policy(const char name, const char strval, cred_t *cr)
	{
	#ifdef HAVE_MLSLABEL
	char ds_hexsl[MAXNAMELEN];
	bslabel_t ds_sl, new_sl;
	boolean_t new_default = FALSE;
	uint64_t zoned;
	int needed_priv = -1;
	int error;

	/* First get the existing dataset label. */
	error = dsl_prop_get(name, zfs_prop_to_name(ZFS_PROP_MLSLABEL),
	1, sizeof (ds_hexsl), &ds_hexsl, NULL);
	if (error != 0)
	return (SET_ERROR(EPERM));

	if (strcasecmp(strval, ZFS_MLSLABEL_DEFAULT) == 0)
	new_default = TRUE;

	/* The label must be translatable */
	if (!new_default && (hexstr_to_label(strval, &new_sl) != 0))
	return (SET_ERROR(EINVAL));

	/*
	* In a non-global zone, disallow attempts to set a label that
	* doesn't match that of the zone; otherwise no other checks
	* are needed.
	*/
	if (!INGLOBALZONE(curproc)) {
	if (new_default \|\| !blequal(&new_sl, CR_SL(CRED())))
	return (SET_ERROR(EPERM));
	return (0);
	}

	/*
	* For global-zone datasets (i.e., those whose zoned property is
	* "off", verify that the specified new label is valid for the
	* global zone.
	*/
	if (dsl_prop_get_integer(name,
	zfs_prop_to_name(ZFS_PROP_ZONED), &zoned, NULL))
	return (SET_ERROR(EPERM));
	if (!zoned) {
	if (zfs_check_global_label(name, strval) != 0)
	return (SET_ERROR(EPERM));
	}

	/*
	* If the existing dataset label is nondefault, check if the
	* dataset is mounted (label cannot be changed while mounted).
	* Get the zfsvfs_t; if there isn't one, then the dataset isn't
	* mounted (or isn't a dataset, doesn't exist, ...).
	*/
	if (strcasecmp(ds_hexsl, ZFS_MLSLABEL_DEFAULT) != 0) {
	objset_t *os;
	static const char *setsl_tag = "setsl_tag";

	/*
	* Try to own the dataset; abort if there is any error,
	* (e.g., already mounted, in use, or other error).
	*/
	error = dmu_objset_own(name, DMU_OST_ZFS, B_TRUE, B_TRUE,
	setsl_tag, &os);
	if (error != 0)
	return (SET_ERROR(EPERM));

	dmu_objset_disown(os, B_TRUE, setsl_tag);

	if (new_default) {
	needed_priv = PRIV_FILE_DOWNGRADE_SL;
	goto out_check;
	}

	if (hexstr_to_label(strval, &new_sl) != 0)
	return (SET_ERROR(EPERM));

	if (blstrictdom(&ds_sl, &new_sl))
	needed_priv = PRIV_FILE_DOWNGRADE_SL;
	else if (blstrictdom(&new_sl, &ds_sl))
	needed_priv = PRIV_FILE_UPGRADE_SL;
	} else {
	/* dataset currently has a default label */
	if (!new_default)
	needed_priv = PRIV_FILE_UPGRADE_SL;
	}

	out_check:
	if (needed_priv != -1)
	return (PRIV_POLICY(cr, needed_priv, B_FALSE, EPERM, NULL));
	return (0);
	#else
	return (SET_ERROR(ENOTSUP));
	#endif /* HAVE_MLSLABEL */
	}

	static int
	zfs_secpolicy_setprop(const char dsname, zfs_prop_t prop, nvpair_t propval,
	cred_t *cr)
	{
	char *strval;

	/*
	* Check permissions for special properties.
	*/
	switch (prop) {
	default:
	break;
	case ZFS_PROP_ZONED:
	/*
	* Disallow setting of 'zoned' from within a local zone.
	*/
	if (!INGLOBALZONE(curproc))
	return (SET_ERROR(EPERM));
	break;

	case ZFS_PROP_QUOTA:
	case ZFS_PROP_FILESYSTEM_LIMIT:
	case ZFS_PROP_SNAPSHOT_LIMIT:
	if (!INGLOBALZONE(curproc)) {
	uint64_t zoned;
	char setpoint[ZFS_MAX_DATASET_NAME_LEN];
	/*
	* Unprivileged users are allowed to modify the
	* limit on things under (ie. contained by)
	* the thing they own.
	*/
	if (dsl_prop_get_integer(dsname,
	zfs_prop_to_name(ZFS_PROP_ZONED), &zoned, setpoint))
	return (SET_ERROR(EPERM));
	if (!zoned \|\| strlen(dsname) <= strlen(setpoint))
	return (SET_ERROR(EPERM));
	}
	break;

	case ZFS_PROP_MLSLABEL:
	if (!is_system_labeled())
	return (SET_ERROR(EPERM));

	if (nvpair_value_string(propval, &strval) == 0) {
	int err;

	err = zfs_set_slabel_policy(dsname, strval, CRED());
	if (err != 0)
	return (err);
	}
	break;
	}

	return (zfs_secpolicy_write_perms(dsname, zfs_prop_to_name(prop), cr));
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_set_fsacl(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	int error;

	error = zfs_dozonecheck(zc->zc_name, cr);
	if (error != 0)
	return (error);

	/*
	* permission to set permissions will be evaluated later in
	* dsl_deleg_can_allow()
	*/
	return (0);
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_rollback(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (zfs_secpolicy_write_perms(zc->zc_name,
	ZFS_DELEG_PERM_ROLLBACK, cr));
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_send(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	dsl_pool_t *dp;
	dsl_dataset_t *ds;
	const char *cp;
	int error;

	/*
	* Generate the current snapshot name from the given objsetid, then
	* use that name for the secpolicy/zone checks.
	*/
	cp = strchr(zc->zc_name, '@');
	if (cp == NULL)
	return (SET_ERROR(EINVAL));
	error = dsl_pool_hold(zc->zc_name, FTAG, &dp);
	if (error != 0)
	return (error);

	error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &ds);
	if (error != 0) {
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	dsl_dataset_name(ds, zc->zc_name);

	error = zfs_secpolicy_write_perms_ds(zc->zc_name, ds,
	ZFS_DELEG_PERM_SEND, cr);
	dsl_dataset_rele(ds, FTAG);
	dsl_pool_rele(dp, FTAG);

	return (error);
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_send_new(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (zfs_secpolicy_write_perms(zc->zc_name,
	ZFS_DELEG_PERM_SEND, cr));
	}

	static int
	zfs_secpolicy_share(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (SET_ERROR(ENOTSUP));
	}

	static int
	zfs_secpolicy_smb_acl(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (SET_ERROR(ENOTSUP));
	}

	static int
	zfs_get_parent(const char datasetname, char parent, int parentsize)
	{
	char *cp;

	/*
	* Remove the @bla or /bla from the end of the name to get the parent.
	*/
	(void) strncpy(parent, datasetname, parentsize);
	cp = strrchr(parent, '@');
	if (cp != NULL) {
	cp[0] = '\0';
	} else {
	cp = strrchr(parent, '/');
	if (cp == NULL)
	return (SET_ERROR(ENOENT));
	cp[0] = '\0';
	}

	return (0);
	}

	int
	zfs_secpolicy_destroy_perms(const char name, cred_t cr)
	{
	int error;

	if ((error = zfs_secpolicy_write_perms(name,
	ZFS_DELEG_PERM_MOUNT, cr)) != 0)
	return (error);

	return (zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_DESTROY, cr));
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_destroy(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (zfs_secpolicy_destroy_perms(zc->zc_name, cr));
	}

	/*
	* Destroying snapshots with delegated permissions requires
	* descendant mount and destroy permissions.
	*/
	/* ARGSUSED */
	static int
	zfs_secpolicy_destroy_snaps(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	nvlist_t *snaps;
	nvpair_t pair, nextpair;
	int error = 0;

	snaps = fnvlist_lookup_nvlist(innvl, "snaps");

	for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL;
	pair = nextpair) {
	nextpair = nvlist_next_nvpair(snaps, pair);
	error = zfs_secpolicy_destroy_perms(nvpair_name(pair), cr);
	if (error == ENOENT) {
	/*
	* Ignore any snapshots that don't exist (we consider
	* them "already destroyed"). Remove the name from the
	* nvl here in case the snapshot is created between
	* now and when we try to destroy it (in which case
	* we don't want to destroy it since we haven't
	* checked for permission).
	*/
	fnvlist_remove_nvpair(snaps, pair);
	error = 0;
	}
	if (error != 0)
	break;
	}

	return (error);
	}

	int
	zfs_secpolicy_rename_perms(const char from, const char to, cred_t *cr)
	{
	char parentname[ZFS_MAX_DATASET_NAME_LEN];
	int error;

	if ((error = zfs_secpolicy_write_perms(from,
	ZFS_DELEG_PERM_RENAME, cr)) != 0)
	return (error);

	if ((error = zfs_secpolicy_write_perms(from,
	ZFS_DELEG_PERM_MOUNT, cr)) != 0)
	return (error);

	if ((error = zfs_get_parent(to, parentname,
	sizeof (parentname))) != 0)
	return (error);

	if ((error = zfs_secpolicy_write_perms(parentname,
	ZFS_DELEG_PERM_CREATE, cr)) != 0)
	return (error);

	if ((error = zfs_secpolicy_write_perms(parentname,
	ZFS_DELEG_PERM_MOUNT, cr)) != 0)
	return (error);

	return (error);
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_rename(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (zfs_secpolicy_rename_perms(zc->zc_name, zc->zc_value, cr));
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_promote(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	dsl_pool_t *dp;
	dsl_dataset_t *clone;
	int error;

	error = zfs_secpolicy_write_perms(zc->zc_name,
	ZFS_DELEG_PERM_PROMOTE, cr);
	if (error != 0)
	return (error);

	error = dsl_pool_hold(zc->zc_name, FTAG, &dp);
	if (error != 0)
	return (error);

	error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &clone);

	if (error == 0) {
	char parentname[ZFS_MAX_DATASET_NAME_LEN];
	dsl_dataset_t *origin = NULL;
	dsl_dir_t *dd;
	dd = clone->ds_dir;

	error = dsl_dataset_hold_obj(dd->dd_pool,
	dsl_dir_phys(dd)->dd_origin_obj, FTAG, &origin);
	if (error != 0) {
	dsl_dataset_rele(clone, FTAG);
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	error = zfs_secpolicy_write_perms_ds(zc->zc_name, clone,
	ZFS_DELEG_PERM_MOUNT, cr);

	dsl_dataset_name(origin, parentname);
	if (error == 0) {
	error = zfs_secpolicy_write_perms_ds(parentname, origin,
	ZFS_DELEG_PERM_PROMOTE, cr);
	}
	dsl_dataset_rele(clone, FTAG);
	dsl_dataset_rele(origin, FTAG);
	}
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_recv(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	int error;

	if ((error = zfs_secpolicy_write_perms(zc->zc_name,
	ZFS_DELEG_PERM_RECEIVE, cr)) != 0)
	return (error);

	if ((error = zfs_secpolicy_write_perms(zc->zc_name,
	ZFS_DELEG_PERM_MOUNT, cr)) != 0)
	return (error);

	return (zfs_secpolicy_write_perms(zc->zc_name,
	ZFS_DELEG_PERM_CREATE, cr));
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_recv_new(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (zfs_secpolicy_recv(zc, innvl, cr));
	}

	int
	zfs_secpolicy_snapshot_perms(const char name, cred_t cr)
	{
	return (zfs_secpolicy_write_perms(name,
	ZFS_DELEG_PERM_SNAPSHOT, cr));
	}

	/*
	* Check for permission to create each snapshot in the nvlist.
	*/
	/* ARGSUSED */
	static int
	zfs_secpolicy_snapshot(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	nvlist_t *snaps;
	int error = 0;
	nvpair_t *pair;

	snaps = fnvlist_lookup_nvlist(innvl, "snaps");

	for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL;
	pair = nvlist_next_nvpair(snaps, pair)) {
	char *name = nvpair_name(pair);
	char *atp = strchr(name, '@');

	if (atp == NULL) {
	error = SET_ERROR(EINVAL);
	break;
	}
	*atp = '\0';
	error = zfs_secpolicy_snapshot_perms(name, cr);
	*atp = '@';
	if (error != 0)
	break;
	}
	return (error);
	}

	/*
	* Check for permission to create each bookmark in the nvlist.
	*/
	/* ARGSUSED */
	static int
	zfs_secpolicy_bookmark(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	int error = 0;

	for (nvpair_t *pair = nvlist_next_nvpair(innvl, NULL);
	pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) {
	char *name = nvpair_name(pair);
	char *hashp = strchr(name, '#');

	if (hashp == NULL) {
	error = SET_ERROR(EINVAL);
	break;
	}
	*hashp = '\0';
	error = zfs_secpolicy_write_perms(name,
	ZFS_DELEG_PERM_BOOKMARK, cr);
	*hashp = '#';
	if (error != 0)
	break;
	}
	return (error);
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_destroy_bookmarks(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	nvpair_t pair, nextpair;
	int error = 0;

	for (pair = nvlist_next_nvpair(innvl, NULL); pair != NULL;
	pair = nextpair) {
	char *name = nvpair_name(pair);
	char *hashp = strchr(name, '#');
	nextpair = nvlist_next_nvpair(innvl, pair);

	if (hashp == NULL) {
	error = SET_ERROR(EINVAL);
	break;
	}

	*hashp = '\0';
	error = zfs_secpolicy_write_perms(name,
	ZFS_DELEG_PERM_DESTROY, cr);
	*hashp = '#';
	if (error == ENOENT) {
	/*
	* Ignore any filesystems that don't exist (we consider
	* their bookmarks "already destroyed"). Remove
	* the name from the nvl here in case the filesystem
	* is created between now and when we try to destroy
	* the bookmark (in which case we don't want to
	* destroy it since we haven't checked for permission).
	*/
	fnvlist_remove_nvpair(innvl, pair);
	error = 0;
	}
	if (error != 0)
	break;
	}

	return (error);
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_log_history(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	/*
	* Even root must have a proper TSD so that we know what pool
	* to log to.
	*/
	if (tsd_get(zfs_allow_log_key) == NULL)
	return (SET_ERROR(EPERM));
	return (0);
	}

	static int
	zfs_secpolicy_create_clone(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	char parentname[ZFS_MAX_DATASET_NAME_LEN];
	int error;
	char *origin;

	if ((error = zfs_get_parent(zc->zc_name, parentname,
	sizeof (parentname))) != 0)
	return (error);

	if (nvlist_lookup_string(innvl, "origin", &origin) == 0 &&
	(error = zfs_secpolicy_write_perms(origin,
	ZFS_DELEG_PERM_CLONE, cr)) != 0)
	return (error);

	if ((error = zfs_secpolicy_write_perms(parentname,
	ZFS_DELEG_PERM_CREATE, cr)) != 0)
	return (error);

	return (zfs_secpolicy_write_perms(parentname,
	ZFS_DELEG_PERM_MOUNT, cr));
	}

	/*
	* Policy for pool operations - create/destroy pools, add vdevs, etc. Requires
	* SYS_CONFIG privilege, which is not available in a local zone.
	*/
	/* ARGSUSED */
	int
	zfs_secpolicy_config(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	if (secpolicy_sys_config(cr, B_FALSE) != 0)
	return (SET_ERROR(EPERM));

	return (0);
	}

	/*
	* Policy for object to name lookups.
	*/
	/* ARGSUSED */
	static int
	zfs_secpolicy_diff(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	int error;

	if ((error = secpolicy_sys_config(cr, B_FALSE)) == 0)
	return (0);

	error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_DIFF, cr);
	return (error);
	}

	/*
	* Policy for fault injection. Requires all privileges.
	*/
	/* ARGSUSED */
	static int
	zfs_secpolicy_inject(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (secpolicy_zinject(cr));
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_inherit_prop(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	zfs_prop_t prop = zfs_name_to_prop(zc->zc_value);

	if (prop == ZPROP_INVAL) {
	if (!zfs_prop_user(zc->zc_value))
	return (SET_ERROR(EINVAL));
	return (zfs_secpolicy_write_perms(zc->zc_name,
	ZFS_DELEG_PERM_USERPROP, cr));
	} else {
	return (zfs_secpolicy_setprop(zc->zc_name, prop,
	NULL, cr));
	}
	}

	static int
	zfs_secpolicy_userspace_one(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	int err = zfs_secpolicy_read(zc, innvl, cr);
	if (err)
	return (err);

	if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS)
	return (SET_ERROR(EINVAL));

	if (zc->zc_value[0] == 0) {
	/*
	* They are asking about a posix uid/gid. If it's
	* themself, allow it.
	*/
	if (zc->zc_objset_type == ZFS_PROP_USERUSED \|\|
	zc->zc_objset_type == ZFS_PROP_USERQUOTA \|\|
	zc->zc_objset_type == ZFS_PROP_USEROBJUSED \|\|
	zc->zc_objset_type == ZFS_PROP_USEROBJQUOTA) {
	if (zc->zc_guid == crgetuid(cr))
	return (0);
	} else if (zc->zc_objset_type == ZFS_PROP_GROUPUSED \|\|
	zc->zc_objset_type == ZFS_PROP_GROUPQUOTA \|\|
	zc->zc_objset_type == ZFS_PROP_GROUPOBJUSED \|\|
	zc->zc_objset_type == ZFS_PROP_GROUPOBJQUOTA) {
	if (groupmember(zc->zc_guid, cr))
	return (0);
	}
	/* else is for project quota/used */
	}

	return (zfs_secpolicy_write_perms(zc->zc_name,
	userquota_perms[zc->zc_objset_type], cr));
	}

	static int
	zfs_secpolicy_userspace_many(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	int err = zfs_secpolicy_read(zc, innvl, cr);
	if (err)
	return (err);

	if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS)
	return (SET_ERROR(EINVAL));

	return (zfs_secpolicy_write_perms(zc->zc_name,
	userquota_perms[zc->zc_objset_type], cr));
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_userspace_upgrade(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (zfs_secpolicy_setprop(zc->zc_name, ZFS_PROP_VERSION,
	NULL, cr));
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_hold(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	nvpair_t *pair;
	nvlist_t *holds;
	int error;

	holds = fnvlist_lookup_nvlist(innvl, "holds");

	for (pair = nvlist_next_nvpair(holds, NULL); pair != NULL;
	pair = nvlist_next_nvpair(holds, pair)) {
	char fsname[ZFS_MAX_DATASET_NAME_LEN];
	error = dmu_fsname(nvpair_name(pair), fsname);
	if (error != 0)
	return (error);
	error = zfs_secpolicy_write_perms(fsname,
	ZFS_DELEG_PERM_HOLD, cr);
	if (error != 0)
	return (error);
	}
	return (0);
	}

	/* ARGSUSED */
	static int
	zfs_secpolicy_release(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	nvpair_t *pair;
	int error;

	for (pair = nvlist_next_nvpair(innvl, NULL); pair != NULL;
	pair = nvlist_next_nvpair(innvl, pair)) {
	char fsname[ZFS_MAX_DATASET_NAME_LEN];
	error = dmu_fsname(nvpair_name(pair), fsname);
	if (error != 0)
	return (error);
	error = zfs_secpolicy_write_perms(fsname,
	ZFS_DELEG_PERM_RELEASE, cr);
	if (error != 0)
	return (error);
	}
	return (0);
	}

	/*
	* Policy for allowing temporary snapshots to be taken or released
	*/
	static int
	zfs_secpolicy_tmp_snapshot(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	/*
	* A temporary snapshot is the same as a snapshot,
	* hold, destroy and release all rolled into one.
	* Delegated diff alone is sufficient that we allow this.
	*/
	int error;

	if ((error = zfs_secpolicy_write_perms(zc->zc_name,
	ZFS_DELEG_PERM_DIFF, cr)) == 0)
	return (0);

	error = zfs_secpolicy_snapshot_perms(zc->zc_name, cr);

	if (innvl != NULL) {
	if (error == 0)
	error = zfs_secpolicy_hold(zc, innvl, cr);
	if (error == 0)
	error = zfs_secpolicy_release(zc, innvl, cr);
	if (error == 0)
	error = zfs_secpolicy_destroy(zc, innvl, cr);
	}
	return (error);
	}

	static int
	zfs_secpolicy_load_key(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (zfs_secpolicy_write_perms(zc->zc_name,
	ZFS_DELEG_PERM_LOAD_KEY, cr));
	}

	static int
	zfs_secpolicy_change_key(zfs_cmd_t zc, nvlist_t innvl, cred_t *cr)
	{
	return (zfs_secpolicy_write_perms(zc->zc_name,
	ZFS_DELEG_PERM_CHANGE_KEY, cr));
	}

	/*
	* Returns the nvlist as specified by the user in the zfs_cmd_t.
	*/
	static int
	get_nvlist(uint64_t nvl, uint64_t size, int iflag, nvlist_t **nvp)
	{
	char *packed;
	int error;
	nvlist_t *list = NULL;

	/*
	* Read in and unpack the user-supplied nvlist.
	*/
	if (size == 0)
	return (SET_ERROR(EINVAL));

	packed = vmem_alloc(size, KM_SLEEP);

	if ((error = ddi_copyin((void *)(uintptr_t)nvl, packed, size,
	iflag)) != 0) {
	vmem_free(packed, size);
	return (SET_ERROR(EFAULT));
	}

	if ((error = nvlist_unpack(packed, size, &list, 0)) != 0) {
	vmem_free(packed, size);
	return (error);
	}

	vmem_free(packed, size);

	*nvp = list;
	return (0);
	}

	/*
	* Reduce the size of this nvlist until it can be serialized in 'max' bytes.
	* Entries will be removed from the end of the nvlist, and one int32 entry
	* named "N_MORE_ERRORS" will be added indicating how many entries were
	* removed.
	*/
	static int
	nvlist_smush(nvlist_t *errors, size_t max)
	{
	size_t size;

	size = fnvlist_size(errors);

	if (size > max) {
	nvpair_t *more_errors;
	int n = 0;

	if (max < 1024)
	return (SET_ERROR(ENOMEM));

	fnvlist_add_int32(errors, ZPROP_N_MORE_ERRORS, 0);
	more_errors = nvlist_prev_nvpair(errors, NULL);

	do {
	nvpair_t *pair = nvlist_prev_nvpair(errors,
	more_errors);
	fnvlist_remove_nvpair(errors, pair);
	n++;
	size = fnvlist_size(errors);
	} while (size > max);

	fnvlist_remove_nvpair(errors, more_errors);
	fnvlist_add_int32(errors, ZPROP_N_MORE_ERRORS, n);
	ASSERT3U(fnvlist_size(errors), <=, max);
	}

	return (0);
	}

	static int
	put_nvlist(zfs_cmd_t zc, nvlist_t nvl)
	{
	char *packed = NULL;
	int error = 0;
	size_t size;

	size = fnvlist_size(nvl);

	if (size > zc->zc_nvlist_dst_size) {
	error = SET_ERROR(ENOMEM);
	} else {
	packed = fnvlist_pack(nvl, &size);
	if (ddi_copyout(packed, (void *)(uintptr_t)zc->zc_nvlist_dst,
	size, zc->zc_iflags) != 0)
	error = SET_ERROR(EFAULT);
	fnvlist_pack_free(packed, size);
	}

	zc->zc_nvlist_dst_size = size;
	zc->zc_nvlist_dst_filled = B_TRUE;
	return (error);
	}

	int
	getzfsvfs_impl(objset_t os, zfsvfs_t *zfvp)
	{
	int error = 0;
	if (dmu_objset_type(os) != DMU_OST_ZFS) {
	return (SET_ERROR(EINVAL));
	}

	mutex_enter(&os->os_user_ptr_lock);
	*zfvp = dmu_objset_get_user(os);
	/* bump s_active only when non-zero to prevent umount race */
	error = zfs_vfs_ref(zfvp);
	mutex_exit(&os->os_user_ptr_lock);
	return (error);
	}

	int
	getzfsvfs(const char dsname, zfsvfs_t *zfvp)
	{
	objset_t *os;
	int error;

	error = dmu_objset_hold(dsname, FTAG, &os);
	if (error != 0)
	return (error);

	error = getzfsvfs_impl(os, zfvp);
	dmu_objset_rele(os, FTAG);
	return (error);
	}

	/*
	* Find a zfsvfs_t for a mounted filesystem, or create our own, in which
	* case its z_sb will be NULL, and it will be opened as the owner.
	* If 'writer' is set, the z_teardown_lock will be held for RW_WRITER,
	* which prevents all inode ops from running.
	*/
	static int
	zfsvfs_hold(const char name, void tag, zfsvfs_t **zfvp, boolean_t writer)
	{
	int error = 0;

	if (getzfsvfs(name, zfvp) != 0)
	error = zfsvfs_create(name, B_FALSE, zfvp);
	if (error == 0) {
	if (writer)
	ZFS_TEARDOWN_ENTER_WRITE(*zfvp, tag);
	else
	ZFS_TEARDOWN_ENTER_READ(*zfvp, tag);
	if ((*zfvp)->z_unmounted) {
	/*
	* XXX we could probably try again, since the unmounting
	* thread should be just about to disassociate the
	* objset from the zfsvfs.
	*/
	ZFS_TEARDOWN_EXIT(*zfvp, tag);
	return (SET_ERROR(EBUSY));
	}
	}
	return (error);
	}

	static void
	zfsvfs_rele(zfsvfs_t zfsvfs, void tag)
	{
	ZFS_TEARDOWN_EXIT(zfsvfs, tag);

	if (zfs_vfs_held(zfsvfs)) {
	zfs_vfs_rele(zfsvfs);
	} else {
	dmu_objset_disown(zfsvfs->z_os, B_TRUE, zfsvfs);
	zfsvfs_free(zfsvfs);
	}
	}

	static int
	zfs_ioc_pool_create(zfs_cmd_t *zc)
	{
	int error;
	nvlist_t config, props = NULL;
	nvlist_t *rootprops = NULL;
	nvlist_t *zplprops = NULL;
	dsl_crypto_params_t *dcp = NULL;
	const char *spa_name = zc->zc_name;
	boolean_t unload_wkey = B_TRUE;

	if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	zc->zc_iflags, &config)))
	return (error);

	if (zc->zc_nvlist_src_size != 0 && (error =
	get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size,
	zc->zc_iflags, &props))) {
	nvlist_free(config);
	return (error);
	}

	if (props) {
	nvlist_t *nvl = NULL;
	nvlist_t *hidden_args = NULL;
	uint64_t version = SPA_VERSION;
	char *tname;

	(void) nvlist_lookup_uint64(props,
	zpool_prop_to_name(ZPOOL_PROP_VERSION), &version);
	if (!SPA_VERSION_IS_SUPPORTED(version)) {
	error = SET_ERROR(EINVAL);
	goto pool_props_bad;
	}
	(void) nvlist_lookup_nvlist(props, ZPOOL_ROOTFS_PROPS, &nvl);
	if (nvl) {
	error = nvlist_dup(nvl, &rootprops, KM_SLEEP);
	if (error != 0)
	goto pool_props_bad;
	(void) nvlist_remove_all(props, ZPOOL_ROOTFS_PROPS);
	}

	(void) nvlist_lookup_nvlist(props, ZPOOL_HIDDEN_ARGS,
	&hidden_args);
	error = dsl_crypto_params_create_nvlist(DCP_CMD_NONE,
	rootprops, hidden_args, &dcp);
	if (error != 0)
	goto pool_props_bad;
	(void) nvlist_remove_all(props, ZPOOL_HIDDEN_ARGS);

	VERIFY(nvlist_alloc(&zplprops, NV_UNIQUE_NAME, KM_SLEEP) == 0);
	error = zfs_fill_zplprops_root(version, rootprops,
	zplprops, NULL);
	if (error != 0)
	goto pool_props_bad;

	if (nvlist_lookup_string(props,
	zpool_prop_to_name(ZPOOL_PROP_TNAME), &tname) == 0)
	spa_name = tname;
	}

	error = spa_create(zc->zc_name, config, props, zplprops, dcp);

	/*
	* Set the remaining root properties
	*/
	if (!error && (error = zfs_set_prop_nvlist(spa_name,
	ZPROP_SRC_LOCAL, rootprops, NULL)) != 0) {
	(void) spa_destroy(spa_name);
	unload_wkey = B_FALSE; /* spa_destroy() unloads wrapping keys */
	}

	pool_props_bad:
	nvlist_free(rootprops);
	nvlist_free(zplprops);
	nvlist_free(config);
	nvlist_free(props);
	dsl_crypto_params_free(dcp, unload_wkey && !!error);

	return (error);
	}

	static int
	zfs_ioc_pool_destroy(zfs_cmd_t *zc)
	{
	int error;
	zfs_log_history(zc);
	error = spa_destroy(zc->zc_name);

	return (error);
	}

	static int
	zfs_ioc_pool_import(zfs_cmd_t *zc)
	{
	nvlist_t config, props = NULL;
	uint64_t guid;
	int error;

	if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	zc->zc_iflags, &config)) != 0)
	return (error);

	if (zc->zc_nvlist_src_size != 0 && (error =
	get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size,
	zc->zc_iflags, &props))) {
	nvlist_free(config);
	return (error);
	}

	if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &guid) != 0 \|\|
	guid != zc->zc_guid)
	error = SET_ERROR(EINVAL);
	else
	error = spa_import(zc->zc_name, config, props, zc->zc_cookie);

	if (zc->zc_nvlist_dst != 0) {
	int err;

	if ((err = put_nvlist(zc, config)) != 0)
	error = err;
	}

	nvlist_free(config);
	nvlist_free(props);

	return (error);
	}

	static int
	zfs_ioc_pool_export(zfs_cmd_t *zc)
	{
	int error;
	boolean_t force = (boolean_t)zc->zc_cookie;
	boolean_t hardforce = (boolean_t)zc->zc_guid;

	zfs_log_history(zc);
	error = spa_export(zc->zc_name, NULL, force, hardforce);

	return (error);
	}

	static int
	zfs_ioc_pool_configs(zfs_cmd_t *zc)
	{
	nvlist_t *configs;
	int error;

	if ((configs = spa_all_configs(&zc->zc_cookie)) == NULL)
	return (SET_ERROR(EEXIST));

	error = put_nvlist(zc, configs);

	nvlist_free(configs);

	return (error);
	}

	/*
	* inputs:
	* zc_name name of the pool
	*
	* outputs:
	* zc_cookie real errno
	* zc_nvlist_dst config nvlist
	* zc_nvlist_dst_size size of config nvlist
	*/
	static int
	zfs_ioc_pool_stats(zfs_cmd_t *zc)
	{
	nvlist_t *config;
	int error;
	int ret = 0;

	error = spa_get_stats(zc->zc_name, &config, zc->zc_value,
	sizeof (zc->zc_value));

	if (config != NULL) {
	ret = put_nvlist(zc, config);
	nvlist_free(config);

	/*
	* The config may be present even if 'error' is non-zero.
	* In this case we return success, and preserve the real errno
	* in 'zc_cookie'.
	*/
	zc->zc_cookie = error;
	} else {
	ret = error;
	}

	return (ret);
	}

	/*
	* Try to import the given pool, returning pool stats as appropriate so that
	* user land knows which devices are available and overall pool health.
	*/
	static int
	zfs_ioc_pool_tryimport(zfs_cmd_t *zc)
	{
	nvlist_t tryconfig, config = NULL;
	int error;

	if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	zc->zc_iflags, &tryconfig)) != 0)
	return (error);

	config = spa_tryimport(tryconfig);

	nvlist_free(tryconfig);

	if (config == NULL)
	return (SET_ERROR(EINVAL));

	error = put_nvlist(zc, config);
	nvlist_free(config);

	return (error);
	}

	/*
	* inputs:
	* zc_name name of the pool
	* zc_cookie scan func (pool_scan_func_t)
	* zc_flags scrub pause/resume flag (pool_scrub_cmd_t)
	*/
	static int
	zfs_ioc_pool_scan(zfs_cmd_t *zc)
	{
	spa_t *spa;
	int error;

	if (zc->zc_flags >= POOL_SCRUB_FLAGS_END)
	return (SET_ERROR(EINVAL));

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
	return (error);

	if (zc->zc_flags == POOL_SCRUB_PAUSE)
	error = spa_scrub_pause_resume(spa, POOL_SCRUB_PAUSE);
	else if (zc->zc_cookie == POOL_SCAN_NONE)
	error = spa_scan_stop(spa);
	else
	error = spa_scan(spa, zc->zc_cookie);

	spa_close(spa, FTAG);

	return (error);
	}

	static int
	zfs_ioc_pool_freeze(zfs_cmd_t *zc)
	{
	spa_t *spa;
	int error;

	error = spa_open(zc->zc_name, &spa, FTAG);
	if (error == 0) {
	spa_freeze(spa);
	spa_close(spa, FTAG);
	}
	return (error);
	}

	static int
	zfs_ioc_pool_upgrade(zfs_cmd_t *zc)
	{
	spa_t *spa;
	int error;

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
	return (error);

	if (zc->zc_cookie < spa_version(spa) \|\|
	!SPA_VERSION_IS_SUPPORTED(zc->zc_cookie)) {
	spa_close(spa, FTAG);
	return (SET_ERROR(EINVAL));
	}

	spa_upgrade(spa, zc->zc_cookie);
	spa_close(spa, FTAG);

	return (error);
	}

	static int
	zfs_ioc_pool_get_history(zfs_cmd_t *zc)
	{
	spa_t *spa;
	char *hist_buf;
	uint64_t size;
	int error;

	if ((size = zc->zc_history_len) == 0)
	return (SET_ERROR(EINVAL));

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
	return (error);

	if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY) {
	spa_close(spa, FTAG);
	return (SET_ERROR(ENOTSUP));
	}

	hist_buf = vmem_alloc(size, KM_SLEEP);
	if ((error = spa_history_get(spa, &zc->zc_history_offset,
	&zc->zc_history_len, hist_buf)) == 0) {
	error = ddi_copyout(hist_buf,
	(void *)(uintptr_t)zc->zc_history,
	zc->zc_history_len, zc->zc_iflags);
	}

	spa_close(spa, FTAG);
	vmem_free(hist_buf, size);
	return (error);
	}

	static int
	zfs_ioc_pool_reguid(zfs_cmd_t *zc)
	{
	spa_t *spa;
	int error;

	error = spa_open(zc->zc_name, &spa, FTAG);
	if (error == 0) {
	error = spa_change_guid(spa);
	spa_close(spa, FTAG);
	}
	return (error);
	}

	static int
	zfs_ioc_dsobj_to_dsname(zfs_cmd_t *zc)
	{
	return (dsl_dsobj_to_dsname(zc->zc_name, zc->zc_obj, zc->zc_value));
	}

	/*
	* inputs:
	* zc_name name of filesystem
	* zc_obj object to find
	*
	* outputs:
	* zc_value name of object
	*/
	static int
	zfs_ioc_obj_to_path(zfs_cmd_t *zc)
	{
	objset_t *os;
	int error;

	/* XXX reading from objset not owned */
	if ((error = dmu_objset_hold_flags(zc->zc_name, B_TRUE,
	FTAG, &os)) != 0)
	return (error);
	if (dmu_objset_type(os) != DMU_OST_ZFS) {
	dmu_objset_rele_flags(os, B_TRUE, FTAG);
	return (SET_ERROR(EINVAL));
	}
	error = zfs_obj_to_path(os, zc->zc_obj, zc->zc_value,
	sizeof (zc->zc_value));
	dmu_objset_rele_flags(os, B_TRUE, FTAG);

	return (error);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	* zc_obj object to find
	*
	* outputs:
	* zc_stat stats on object
	* zc_value path to object
	*/
	static int
	zfs_ioc_obj_to_stats(zfs_cmd_t *zc)
	{
	objset_t *os;
	int error;

	/* XXX reading from objset not owned */
	if ((error = dmu_objset_hold_flags(zc->zc_name, B_TRUE,
	FTAG, &os)) != 0)
	return (error);
	if (dmu_objset_type(os) != DMU_OST_ZFS) {
	dmu_objset_rele_flags(os, B_TRUE, FTAG);
	return (SET_ERROR(EINVAL));
	}
	error = zfs_obj_to_stats(os, zc->zc_obj, &zc->zc_stat, zc->zc_value,
	sizeof (zc->zc_value));
	dmu_objset_rele_flags(os, B_TRUE, FTAG);

	return (error);
	}

	static int
	zfs_ioc_vdev_add(zfs_cmd_t *zc)
	{
	spa_t *spa;
	int error;
	nvlist_t *config;

	error = spa_open(zc->zc_name, &spa, FTAG);
	if (error != 0)
	return (error);

	error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	zc->zc_iflags, &config);
	if (error == 0) {
	error = spa_vdev_add(spa, config);
	nvlist_free(config);
	}
	spa_close(spa, FTAG);
	return (error);
	}

	/*
	* inputs:
	* zc_name name of the pool
	* zc_guid guid of vdev to remove
	* zc_cookie cancel removal
	*/
	static int
	zfs_ioc_vdev_remove(zfs_cmd_t *zc)
	{
	spa_t *spa;
	int error;

	error = spa_open(zc->zc_name, &spa, FTAG);
	if (error != 0)
	return (error);
	if (zc->zc_cookie != 0) {
	error = spa_vdev_remove_cancel(spa);
	} else {
	error = spa_vdev_remove(spa, zc->zc_guid, B_FALSE);
	}
	spa_close(spa, FTAG);
	return (error);
	}

	static int
	zfs_ioc_vdev_set_state(zfs_cmd_t *zc)
	{
	spa_t *spa;
	int error;
	vdev_state_t newstate = VDEV_STATE_UNKNOWN;

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
	return (error);
	switch (zc->zc_cookie) {
	case VDEV_STATE_ONLINE:
	error = vdev_online(spa, zc->zc_guid, zc->zc_obj, &newstate);
	break;

	case VDEV_STATE_OFFLINE:
	error = vdev_offline(spa, zc->zc_guid, zc->zc_obj);
	break;

	case VDEV_STATE_FAULTED:
	if (zc->zc_obj != VDEV_AUX_ERR_EXCEEDED &&
	zc->zc_obj != VDEV_AUX_EXTERNAL &&
	zc->zc_obj != VDEV_AUX_EXTERNAL_PERSIST)
	zc->zc_obj = VDEV_AUX_ERR_EXCEEDED;

	error = vdev_fault(spa, zc->zc_guid, zc->zc_obj);
	break;

	case VDEV_STATE_DEGRADED:
	if (zc->zc_obj != VDEV_AUX_ERR_EXCEEDED &&
	zc->zc_obj != VDEV_AUX_EXTERNAL)
	zc->zc_obj = VDEV_AUX_ERR_EXCEEDED;

	error = vdev_degrade(spa, zc->zc_guid, zc->zc_obj);
	break;

	default:
	error = SET_ERROR(EINVAL);
	}
	zc->zc_cookie = newstate;
	spa_close(spa, FTAG);
	return (error);
	}

	static int
	zfs_ioc_vdev_attach(zfs_cmd_t *zc)
	{
	spa_t *spa;
	nvlist_t *config;
	int replacing = zc->zc_cookie;
	int rebuild = zc->zc_simple;
	int error;

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
	return (error);

	if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	zc->zc_iflags, &config)) == 0) {
	error = spa_vdev_attach(spa, zc->zc_guid, config, replacing,
	rebuild);
	nvlist_free(config);
	}

	spa_close(spa, FTAG);
	return (error);
	}

	static int
	zfs_ioc_vdev_detach(zfs_cmd_t *zc)
	{
	spa_t *spa;
	int error;

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
	return (error);

	error = spa_vdev_detach(spa, zc->zc_guid, 0, B_FALSE);

	spa_close(spa, FTAG);
	return (error);
	}

	static int
	zfs_ioc_vdev_split(zfs_cmd_t *zc)
	{
	spa_t *spa;
	nvlist_t config, props = NULL;
	int error;
	boolean_t exp = !!(zc->zc_cookie & ZPOOL_EXPORT_AFTER_SPLIT);

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
	return (error);

	if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	zc->zc_iflags, &config))) {
	spa_close(spa, FTAG);
	return (error);
	}

	if (zc->zc_nvlist_src_size != 0 && (error =
	get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size,
	zc->zc_iflags, &props))) {
	spa_close(spa, FTAG);
	nvlist_free(config);
	return (error);
	}

	error = spa_vdev_split_mirror(spa, zc->zc_string, config, props, exp);

	spa_close(spa, FTAG);

	nvlist_free(config);
	nvlist_free(props);

	return (error);
	}

	static int
	zfs_ioc_vdev_setpath(zfs_cmd_t *zc)
	{
	spa_t *spa;
	const char *path = zc->zc_value;
	uint64_t guid = zc->zc_guid;
	int error;

	error = spa_open(zc->zc_name, &spa, FTAG);
	if (error != 0)
	return (error);

	error = spa_vdev_setpath(spa, guid, path);
	spa_close(spa, FTAG);
	return (error);
	}

	static int
	zfs_ioc_vdev_setfru(zfs_cmd_t *zc)
	{
	spa_t *spa;
	const char *fru = zc->zc_value;
	uint64_t guid = zc->zc_guid;
	int error;

	error = spa_open(zc->zc_name, &spa, FTAG);
	if (error != 0)
	return (error);

	error = spa_vdev_setfru(spa, guid, fru);
	spa_close(spa, FTAG);
	return (error);
	}

	static int
	zfs_ioc_objset_stats_impl(zfs_cmd_t zc, objset_t os)
	{
	int error = 0;
	nvlist_t *nv;

	dmu_objset_fast_stat(os, &zc->zc_objset_stats);

	if (zc->zc_nvlist_dst != 0 &&
	(error = dsl_prop_get_all(os, &nv)) == 0) {
	dmu_objset_stats(os, nv);
	/*
	* NB: zvol_get_stats() will read the objset contents,
	* which we aren't supposed to do with a
	* DS_MODE_USER hold, because it could be
	* inconsistent. So this is a bit of a workaround...
	* XXX reading without owning
	*/
	if (!zc->zc_objset_stats.dds_inconsistent &&
	dmu_objset_type(os) == DMU_OST_ZVOL) {
	error = zvol_get_stats(os, nv);
	if (error == EIO) {
	nvlist_free(nv);
	return (error);
	}
	VERIFY0(error);
	}
	if (error == 0)
	error = put_nvlist(zc, nv);
	nvlist_free(nv);
	}

	return (error);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	* zc_nvlist_dst_size size of buffer for property nvlist
	*
	* outputs:
	* zc_objset_stats stats
	* zc_nvlist_dst property nvlist
	* zc_nvlist_dst_size size of property nvlist
	*/
	static int
	zfs_ioc_objset_stats(zfs_cmd_t *zc)
	{
	objset_t *os;
	int error;

	error = dmu_objset_hold(zc->zc_name, FTAG, &os);
	if (error == 0) {
	error = zfs_ioc_objset_stats_impl(zc, os);
	dmu_objset_rele(os, FTAG);
	}

	return (error);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	* zc_nvlist_dst_size size of buffer for property nvlist
	*
	* outputs:
	* zc_nvlist_dst received property nvlist
	* zc_nvlist_dst_size size of received property nvlist
	*
	* Gets received properties (distinct from local properties on or after
	* SPA_VERSION_RECVD_PROPS) for callers who want to differentiate received from
	* local property values.
	*/
	static int
	zfs_ioc_objset_recvd_props(zfs_cmd_t *zc)
	{
	int error = 0;
	nvlist_t *nv;

	/*
	* Without this check, we would return local property values if the
	* caller has not already received properties on or after
	* SPA_VERSION_RECVD_PROPS.
	*/
	if (!dsl_prop_get_hasrecvd(zc->zc_name))
	return (SET_ERROR(ENOTSUP));

	if (zc->zc_nvlist_dst != 0 &&
	(error = dsl_prop_get_received(zc->zc_name, &nv)) == 0) {
	error = put_nvlist(zc, nv);
	nvlist_free(nv);
	}

	return (error);
	}

	static int
	nvl_add_zplprop(objset_t os, nvlist_t props, zfs_prop_t prop)
	{
	uint64_t value;
	int error;

	/*
	* zfs_get_zplprop() will either find a value or give us
	* the default value (if there is one).
	*/
	if ((error = zfs_get_zplprop(os, prop, &value)) != 0)
	return (error);
	VERIFY(nvlist_add_uint64(props, zfs_prop_to_name(prop), value) == 0);
	return (0);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	* zc_nvlist_dst_size size of buffer for zpl property nvlist
	*
	* outputs:
	* zc_nvlist_dst zpl property nvlist
	* zc_nvlist_dst_size size of zpl property nvlist
	*/
	static int
	zfs_ioc_objset_zplprops(zfs_cmd_t *zc)
	{
	objset_t *os;
	int err;

	/* XXX reading without owning */
	if ((err = dmu_objset_hold(zc->zc_name, FTAG, &os)))
	return (err);

	dmu_objset_fast_stat(os, &zc->zc_objset_stats);

	/*
	* NB: nvl_add_zplprop() will read the objset contents,
	* which we aren't supposed to do with a DS_MODE_USER
	* hold, because it could be inconsistent.
	*/
	if (zc->zc_nvlist_dst != 0 &&
	!zc->zc_objset_stats.dds_inconsistent &&
	dmu_objset_type(os) == DMU_OST_ZFS) {
	nvlist_t *nv;

	VERIFY(nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_SLEEP) == 0);
	if ((err = nvl_add_zplprop(os, nv, ZFS_PROP_VERSION)) == 0 &&
	(err = nvl_add_zplprop(os, nv, ZFS_PROP_NORMALIZE)) == 0 &&
	(err = nvl_add_zplprop(os, nv, ZFS_PROP_UTF8ONLY)) == 0 &&
	(err = nvl_add_zplprop(os, nv, ZFS_PROP_CASE)) == 0)
	err = put_nvlist(zc, nv);
	nvlist_free(nv);
	} else {
	err = SET_ERROR(ENOENT);
	}
	dmu_objset_rele(os, FTAG);
	return (err);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	* zc_cookie zap cursor
	* zc_nvlist_dst_size size of buffer for property nvlist
	*
	* outputs:
	* zc_name name of next filesystem
	* zc_cookie zap cursor
	* zc_objset_stats stats
	* zc_nvlist_dst property nvlist
	* zc_nvlist_dst_size size of property nvlist
	*/
	static int
	zfs_ioc_dataset_list_next(zfs_cmd_t *zc)
	{
	objset_t *os;
	int error;
	char *p;
	size_t orig_len = strlen(zc->zc_name);

	top:
	if ((error = dmu_objset_hold(zc->zc_name, FTAG, &os))) {
	if (error == ENOENT)
	error = SET_ERROR(ESRCH);
	return (error);
	}

	p = strrchr(zc->zc_name, '/');
	if (p == NULL \|\| p[1] != '\0')
	(void) strlcat(zc->zc_name, "/", sizeof (zc->zc_name));
	p = zc->zc_name + strlen(zc->zc_name);

	do {
	error = dmu_dir_list_next(os,
	sizeof (zc->zc_name) - (p - zc->zc_name), p,
	NULL, &zc->zc_cookie);
	if (error == ENOENT)
	error = SET_ERROR(ESRCH);
	} while (error == 0 && zfs_dataset_name_hidden(zc->zc_name));
	dmu_objset_rele(os, FTAG);

	/*
	* If it's an internal dataset (ie. with a '$' in its name),
	* don't try to get stats for it, otherwise we'll return ENOENT.
	*/
	if (error == 0 && strchr(zc->zc_name, '$') == NULL) {
	error = zfs_ioc_objset_stats(zc); /* fill in the stats */
	if (error == ENOENT) {
	/* We lost a race with destroy, get the next one. */
	zc->zc_name[orig_len] = '\0';
	goto top;
	}
	}
	return (error);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	* zc_cookie zap cursor
	* zc_nvlist_src iteration range nvlist
	* zc_nvlist_src_size size of iteration range nvlist
	*
	* outputs:
	* zc_name name of next snapshot
	* zc_objset_stats stats
	* zc_nvlist_dst property nvlist
	* zc_nvlist_dst_size size of property nvlist
	*/
	static int
	zfs_ioc_snapshot_list_next(zfs_cmd_t *zc)
	{
	int error;
	objset_t os, ossnap;
	dsl_dataset_t *ds;
	uint64_t min_txg = 0, max_txg = 0;

	if (zc->zc_nvlist_src_size != 0) {
	nvlist_t *props = NULL;
	error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size,
	zc->zc_iflags, &props);
	if (error != 0)
	return (error);
	(void) nvlist_lookup_uint64(props, SNAP_ITER_MIN_TXG,
	&min_txg);
	(void) nvlist_lookup_uint64(props, SNAP_ITER_MAX_TXG,
	&max_txg);
	nvlist_free(props);
	}

	error = dmu_objset_hold(zc->zc_name, FTAG, &os);
	if (error != 0) {
	return (error == ENOENT ? SET_ERROR(ESRCH) : error);
	}

	/*
	* A dataset name of maximum length cannot have any snapshots,
	* so exit immediately.
	*/
	if (strlcat(zc->zc_name, "@", sizeof (zc->zc_name)) >=
	ZFS_MAX_DATASET_NAME_LEN) {
	dmu_objset_rele(os, FTAG);
	return (SET_ERROR(ESRCH));
	}

	while (error == 0) {
	if (issig(JUSTLOOKING) && issig(FORREAL)) {
	error = SET_ERROR(EINTR);
	break;
	}

	error = dmu_snapshot_list_next(os,
	sizeof (zc->zc_name) - strlen(zc->zc_name),
	zc->zc_name + strlen(zc->zc_name), &zc->zc_obj,
	&zc->zc_cookie, NULL);
	if (error == ENOENT) {
	error = SET_ERROR(ESRCH);
	break;
	} else if (error != 0) {
	break;
	}

	error = dsl_dataset_hold_obj(dmu_objset_pool(os), zc->zc_obj,
	FTAG, &ds);
	if (error != 0)
	break;

	if ((min_txg != 0 && dsl_get_creationtxg(ds) < min_txg) \|\|
	(max_txg != 0 && dsl_get_creationtxg(ds) > max_txg)) {
	dsl_dataset_rele(ds, FTAG);
	/* undo snapshot name append */
	*(strchr(zc->zc_name, '@') + 1) = '\0';
	/* skip snapshot */
	continue;
	}

	if (zc->zc_simple) {
	dsl_dataset_rele(ds, FTAG);
	break;
	}

	if ((error = dmu_objset_from_ds(ds, &ossnap)) != 0) {
	dsl_dataset_rele(ds, FTAG);
	break;
	}
	if ((error = zfs_ioc_objset_stats_impl(zc, ossnap)) != 0) {
	dsl_dataset_rele(ds, FTAG);
	break;
	}
	dsl_dataset_rele(ds, FTAG);
	break;
	}

	dmu_objset_rele(os, FTAG);
	/* if we failed, undo the @ that we tacked on to zc_name */
	if (error != 0)
	*strchr(zc->zc_name, '@') = '\0';
	return (error);
	}

	static int
	zfs_prop_set_userquota(const char dsname, nvpair_t pair)
	{
	const char *propname = nvpair_name(pair);
	uint64_t *valary;
	unsigned int vallen;
	const char dash, domain;
	zfs_userquota_prop_t type;
	uint64_t rid;
	uint64_t quota;
	zfsvfs_t *zfsvfs;
	int err;

	if (nvpair_type(pair) == DATA_TYPE_NVLIST) {
	nvlist_t *attrs;
	VERIFY(nvpair_value_nvlist(pair, &attrs) == 0);
	if (nvlist_lookup_nvpair(attrs, ZPROP_VALUE,
	&pair) != 0)
	return (SET_ERROR(EINVAL));
	}

	/*
	* A correctly constructed propname is encoded as
	* userquota@<rid>-<domain>.
	*/
	if ((dash = strchr(propname, '-')) == NULL \|\|
	nvpair_value_uint64_array(pair, &valary, &vallen) != 0 \|\|
	vallen != 3)
	return (SET_ERROR(EINVAL));

	domain = dash + 1;
	type = valary[0];
	rid = valary[1];
	quota = valary[2];

	err = zfsvfs_hold(dsname, FTAG, &zfsvfs, B_FALSE);
	if (err == 0) {
	err = zfs_set_userquota(zfsvfs, type, domain, rid, quota);
	zfsvfs_rele(zfsvfs, FTAG);
	}

	return (err);
	}

	/*
	* If the named property is one that has a special function to set its value,
	* return 0 on success and a positive error code on failure; otherwise if it is
	* not one of the special properties handled by this function, return -1.
	*
	* XXX: It would be better for callers of the property interface if we handled
	* these special cases in dsl_prop.c (in the dsl layer).
	*/
	static int
	zfs_prop_set_special(const char *dsname, zprop_source_t source,
	nvpair_t *pair)
	{
	const char *propname = nvpair_name(pair);
	zfs_prop_t prop = zfs_name_to_prop(propname);
	uint64_t intval = 0;
	const char *strval = NULL;
	int err = -1;

	if (prop == ZPROP_INVAL) {
	if (zfs_prop_userquota(propname))
	return (zfs_prop_set_userquota(dsname, pair));
	return (-1);
	}

	if (nvpair_type(pair) == DATA_TYPE_NVLIST) {
	nvlist_t *attrs;
	VERIFY(nvpair_value_nvlist(pair, &attrs) == 0);
	VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE,
	&pair) == 0);
	}

	/* all special properties are numeric except for keylocation */
	if (zfs_prop_get_type(prop) == PROP_TYPE_STRING) {
	strval = fnvpair_value_string(pair);
	} else {
	intval = fnvpair_value_uint64(pair);
	}

	switch (prop) {
	case ZFS_PROP_QUOTA:
	err = dsl_dir_set_quota(dsname, source, intval);
	break;
	case ZFS_PROP_REFQUOTA:
	err = dsl_dataset_set_refquota(dsname, source, intval);
	break;
	case ZFS_PROP_FILESYSTEM_LIMIT:
	case ZFS_PROP_SNAPSHOT_LIMIT:
	if (intval == UINT64_MAX) {
	/* clearing the limit, just do it */
	err = 0;
	} else {
	err = dsl_dir_activate_fs_ss_limit(dsname);
	}
	/*
	* Set err to -1 to force the zfs_set_prop_nvlist code down the
	* default path to set the value in the nvlist.
	*/
	if (err == 0)
	err = -1;
	break;
	case ZFS_PROP_KEYLOCATION:
	err = dsl_crypto_can_set_keylocation(dsname, strval);

	/*
	* Set err to -1 to force the zfs_set_prop_nvlist code down the
	* default path to set the value in the nvlist.
	*/
	if (err == 0)
	err = -1;
	break;
	case ZFS_PROP_RESERVATION:
	err = dsl_dir_set_reservation(dsname, source, intval);
	break;
	case ZFS_PROP_REFRESERVATION:
	err = dsl_dataset_set_refreservation(dsname, source, intval);
	break;
	case ZFS_PROP_COMPRESSION:
	err = dsl_dataset_set_compression(dsname, source, intval);
	/*
	* Set err to -1 to force the zfs_set_prop_nvlist code down the
	* default path to set the value in the nvlist.
	*/
	if (err == 0)
	err = -1;
	break;
	case ZFS_PROP_VOLSIZE:
	err = zvol_set_volsize(dsname, intval);
	break;
	case ZFS_PROP_SNAPDEV:
	err = zvol_set_snapdev(dsname, source, intval);
	break;
	case ZFS_PROP_VOLMODE:
	err = zvol_set_volmode(dsname, source, intval);
	break;
	case ZFS_PROP_VERSION:
	{
	zfsvfs_t *zfsvfs;

	if ((err = zfsvfs_hold(dsname, FTAG, &zfsvfs, B_TRUE)) != 0)
	break;

	err = zfs_set_version(zfsvfs, intval);
	zfsvfs_rele(zfsvfs, FTAG);

	if (err == 0 && intval >= ZPL_VERSION_USERSPACE) {
	zfs_cmd_t *zc;

	zc = kmem_zalloc(sizeof (zfs_cmd_t), KM_SLEEP);
	(void) strlcpy(zc->zc_name, dsname,
	sizeof (zc->zc_name));
	(void) zfs_ioc_userspace_upgrade(zc);
	(void) zfs_ioc_id_quota_upgrade(zc);
	kmem_free(zc, sizeof (zfs_cmd_t));
	}
	break;
	}
	default:
	err = -1;
	}

	return (err);
	}

	static boolean_t
	zfs_is_namespace_prop(zfs_prop_t prop)
	{
	switch (prop) {

	case ZFS_PROP_ATIME:
	case ZFS_PROP_RELATIME:
	case ZFS_PROP_DEVICES:
	case ZFS_PROP_EXEC:
	case ZFS_PROP_SETUID:
	case ZFS_PROP_READONLY:
	case ZFS_PROP_XATTR:
	case ZFS_PROP_NBMAND:
	return (B_TRUE);

	default:
	return (B_FALSE);
	}
	}

	/*
	* This function is best effort. If it fails to set any of the given properties,
	* it continues to set as many as it can and returns the last error
	* encountered. If the caller provides a non-NULL errlist, it will be filled in
	* with the list of names of all the properties that failed along with the
	* corresponding error numbers.
	*
	* If every property is set successfully, zero is returned and errlist is not
	* modified.
	*/
	int
	zfs_set_prop_nvlist(const char dsname, zprop_source_t source, nvlist_t nvl,
	nvlist_t *errlist)
	{
	nvpair_t *pair;
	nvpair_t *propval;
	int rv = 0;
	uint64_t intval;
	const char *strval;
	boolean_t should_update_mount_cache = B_FALSE;

	nvlist_t *genericnvl = fnvlist_alloc();
	nvlist_t *retrynvl = fnvlist_alloc();
	retry:
	pair = NULL;
	while ((pair = nvlist_next_nvpair(nvl, pair)) != NULL) {
	const char *propname = nvpair_name(pair);
	zfs_prop_t prop = zfs_name_to_prop(propname);
	int err = 0;

	/* decode the property value */
	propval = pair;
	if (nvpair_type(pair) == DATA_TYPE_NVLIST) {
	nvlist_t *attrs;
	attrs = fnvpair_value_nvlist(pair);
	if (nvlist_lookup_nvpair(attrs, ZPROP_VALUE,
	&propval) != 0)
	err = SET_ERROR(EINVAL);
	}

	/* Validate value type */
	if (err == 0 && source == ZPROP_SRC_INHERITED) {
	/* inherited properties are expected to be booleans */
	if (nvpair_type(propval) != DATA_TYPE_BOOLEAN)
	err = SET_ERROR(EINVAL);
	} else if (err == 0 && prop == ZPROP_INVAL) {
	if (zfs_prop_user(propname)) {
	if (nvpair_type(propval) != DATA_TYPE_STRING)
	err = SET_ERROR(EINVAL);
	} else if (zfs_prop_userquota(propname)) {
	if (nvpair_type(propval) !=
	DATA_TYPE_UINT64_ARRAY)
	err = SET_ERROR(EINVAL);
	} else {
	err = SET_ERROR(EINVAL);
	}
	} else if (err == 0) {
	if (nvpair_type(propval) == DATA_TYPE_STRING) {
	if (zfs_prop_get_type(prop) != PROP_TYPE_STRING)
	err = SET_ERROR(EINVAL);
	} else if (nvpair_type(propval) == DATA_TYPE_UINT64) {
	const char *unused;

	intval = fnvpair_value_uint64(propval);

	switch (zfs_prop_get_type(prop)) {
	case PROP_TYPE_NUMBER:
	break;
	case PROP_TYPE_STRING:
	err = SET_ERROR(EINVAL);
	break;
	case PROP_TYPE_INDEX:
	if (zfs_prop_index_to_string(prop,
	intval, &unused) != 0)
	err =
	SET_ERROR(ZFS_ERR_BADPROP);
	break;
	default:
	cmn_err(CE_PANIC,
	"unknown property type");
	}
	} else {
	err = SET_ERROR(EINVAL);
	}
	}

	/* Validate permissions */
	if (err == 0)
	err = zfs_check_settable(dsname, pair, CRED());

	if (err == 0) {
	if (source == ZPROP_SRC_INHERITED)
	err = -1; /* does not need special handling */
	else
	err = zfs_prop_set_special(dsname, source,
	pair);
	if (err == -1) {
	/*
	* For better performance we build up a list of
	* properties to set in a single transaction.
	*/
	err = nvlist_add_nvpair(genericnvl, pair);
	} else if (err != 0 && nvl != retrynvl) {
	/*
	* This may be a spurious error caused by
	* receiving quota and reservation out of order.
	* Try again in a second pass.
	*/
	err = nvlist_add_nvpair(retrynvl, pair);
	}
	}

	if (err != 0) {
	if (errlist != NULL)
	fnvlist_add_int32(errlist, propname, err);
	rv = err;
	}

	if (zfs_is_namespace_prop(prop))
	should_update_mount_cache = B_TRUE;
	}

	if (nvl != retrynvl && !nvlist_empty(retrynvl)) {
	nvl = retrynvl;
	goto retry;
	}

	if (!nvlist_empty(genericnvl) &&
	dsl_props_set(dsname, source, genericnvl) != 0) {
	/*
	* If this fails, we still want to set as many properties as we
	* can, so try setting them individually.
	*/
	pair = NULL;
	while ((pair = nvlist_next_nvpair(genericnvl, pair)) != NULL) {
	const char *propname = nvpair_name(pair);
	int err = 0;

	propval = pair;
	if (nvpair_type(pair) == DATA_TYPE_NVLIST) {
	nvlist_t *attrs;
	attrs = fnvpair_value_nvlist(pair);
	propval = fnvlist_lookup_nvpair(attrs,
	ZPROP_VALUE);
	}

	if (nvpair_type(propval) == DATA_TYPE_STRING) {
	strval = fnvpair_value_string(propval);
	err = dsl_prop_set_string(dsname, propname,
	source, strval);
	} else if (nvpair_type(propval) == DATA_TYPE_BOOLEAN) {
	err = dsl_prop_inherit(dsname, propname,
	source);
	} else {
	intval = fnvpair_value_uint64(propval);
	err = dsl_prop_set_int(dsname, propname, source,
	intval);
	}

	if (err != 0) {
	if (errlist != NULL) {
	fnvlist_add_int32(errlist, propname,
	err);
	}
	rv = err;
	}
	}
	}
	if (should_update_mount_cache)
	zfs_ioctl_update_mount_cache(dsname);

	nvlist_free(genericnvl);
	nvlist_free(retrynvl);

	return (rv);
	}

	/*
	* Check that all the properties are valid user properties.
	*/
	static int
	zfs_check_userprops(nvlist_t *nvl)
	{
	nvpair_t *pair = NULL;

	while ((pair = nvlist_next_nvpair(nvl, pair)) != NULL) {
	const char *propname = nvpair_name(pair);

	if (!zfs_prop_user(propname) \|\|
	nvpair_type(pair) != DATA_TYPE_STRING)
	return (SET_ERROR(EINVAL));

	if (strlen(propname) >= ZAP_MAXNAMELEN)
	return (SET_ERROR(ENAMETOOLONG));

	if (strlen(fnvpair_value_string(pair)) >= ZAP_MAXVALUELEN)
	return (SET_ERROR(E2BIG));
	}
	return (0);
	}

	static void
	props_skip(nvlist_t props, nvlist_t skipped, nvlist_t **newprops)
	{
	nvpair_t *pair;

	VERIFY(nvlist_alloc(newprops, NV_UNIQUE_NAME, KM_SLEEP) == 0);

	pair = NULL;
	while ((pair = nvlist_next_nvpair(props, pair)) != NULL) {
	if (nvlist_exists(skipped, nvpair_name(pair)))
	continue;

	VERIFY(nvlist_add_nvpair(*newprops, pair) == 0);
	}
	}

	static int
	clear_received_props(const char dsname, nvlist_t props,
	nvlist_t *skipped)
	{
	int err = 0;
	nvlist_t *cleared_props = NULL;
	props_skip(props, skipped, &cleared_props);
	if (!nvlist_empty(cleared_props)) {
	/*
	* Acts on local properties until the dataset has received
	* properties at least once on or after SPA_VERSION_RECVD_PROPS.
	*/
	zprop_source_t flags = (ZPROP_SRC_NONE \|
	(dsl_prop_get_hasrecvd(dsname) ? ZPROP_SRC_RECEIVED : 0));
	err = zfs_set_prop_nvlist(dsname, flags, cleared_props, NULL);
	}
	nvlist_free(cleared_props);
	return (err);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	* zc_value name of property to set
	* zc_nvlist_src{_size} nvlist of properties to apply
	* zc_cookie received properties flag
	*
	* outputs:
	* zc_nvlist_dst{_size} error for each unapplied received property
	*/
	static int
	zfs_ioc_set_prop(zfs_cmd_t *zc)
	{
	nvlist_t *nvl;
	boolean_t received = zc->zc_cookie;
	zprop_source_t source = (received ? ZPROP_SRC_RECEIVED :
	ZPROP_SRC_LOCAL);
	nvlist_t *errors;
	int error;

	if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size,
	zc->zc_iflags, &nvl)) != 0)
	return (error);

	if (received) {
	nvlist_t *origprops;

	if (dsl_prop_get_received(zc->zc_name, &origprops) == 0) {
	(void) clear_received_props(zc->zc_name,
	origprops, nvl);
	nvlist_free(origprops);
	}

	error = dsl_prop_set_hasrecvd(zc->zc_name);
	}

	errors = fnvlist_alloc();
	if (error == 0)
	error = zfs_set_prop_nvlist(zc->zc_name, source, nvl, errors);

	if (zc->zc_nvlist_dst != 0 && errors != NULL) {
	(void) put_nvlist(zc, errors);
	}

	nvlist_free(errors);
	nvlist_free(nvl);
	return (error);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	* zc_value name of property to inherit
	* zc_cookie revert to received value if TRUE
	*
	* outputs: none
	*/
	static int
	zfs_ioc_inherit_prop(zfs_cmd_t *zc)
	{
	const char *propname = zc->zc_value;
	zfs_prop_t prop = zfs_name_to_prop(propname);
	boolean_t received = zc->zc_cookie;
	zprop_source_t source = (received
	? ZPROP_SRC_NONE /* revert to received value, if any */
	: ZPROP_SRC_INHERITED); /* explicitly inherit */
	nvlist_t *dummy;
	nvpair_t *pair;
	zprop_type_t type;
	int err;

	if (!received) {
	/*
	* Only check this in the non-received case. We want to allow
	* 'inherit -S' to revert non-inheritable properties like quota
	* and reservation to the received or default values even though
	* they are not considered inheritable.
	*/
	if (prop != ZPROP_INVAL && !zfs_prop_inheritable(prop))
	return (SET_ERROR(EINVAL));
	}

	if (prop == ZPROP_INVAL) {
	if (!zfs_prop_user(propname))
	return (SET_ERROR(EINVAL));

	type = PROP_TYPE_STRING;
	} else if (prop == ZFS_PROP_VOLSIZE \|\| prop == ZFS_PROP_VERSION) {
	return (SET_ERROR(EINVAL));
	} else {
	type = zfs_prop_get_type(prop);
	}

	/*
	* zfs_prop_set_special() expects properties in the form of an
	* nvpair with type info.
	*/
	dummy = fnvlist_alloc();

	switch (type) {
	case PROP_TYPE_STRING:
	VERIFY(0 == nvlist_add_string(dummy, propname, ""));
	break;
	case PROP_TYPE_NUMBER:
	case PROP_TYPE_INDEX:
	VERIFY(0 == nvlist_add_uint64(dummy, propname, 0));
	break;
	default:
	err = SET_ERROR(EINVAL);
	goto errout;
	}

	pair = nvlist_next_nvpair(dummy, NULL);
	if (pair == NULL) {
	err = SET_ERROR(EINVAL);
	} else {
	err = zfs_prop_set_special(zc->zc_name, source, pair);
	if (err == -1) /* property is not "special", needs handling */
	err = dsl_prop_inherit(zc->zc_name, zc->zc_value,
	source);
	}

	errout:
	nvlist_free(dummy);
	return (err);
	}

	static int
	zfs_ioc_pool_set_props(zfs_cmd_t *zc)
	{
	nvlist_t *props;
	spa_t *spa;
	int error;
	nvpair_t *pair;

	if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size,
	zc->zc_iflags, &props)))
	return (error);

	/*
	* If the only property is the configfile, then just do a spa_lookup()
	* to handle the faulted case.
	*/
	pair = nvlist_next_nvpair(props, NULL);
	if (pair != NULL && strcmp(nvpair_name(pair),
	zpool_prop_to_name(ZPOOL_PROP_CACHEFILE)) == 0 &&
	nvlist_next_nvpair(props, pair) == NULL) {
	mutex_enter(&spa_namespace_lock);
	if ((spa = spa_lookup(zc->zc_name)) != NULL) {
	spa_configfile_set(spa, props, B_FALSE);
	spa_write_cachefile(spa, B_FALSE, B_TRUE);
	}
	mutex_exit(&spa_namespace_lock);
	if (spa != NULL) {
	nvlist_free(props);
	return (0);
	}
	}

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) {
	nvlist_free(props);
	return (error);
	}

	error = spa_prop_set(spa, props);

	nvlist_free(props);
	spa_close(spa, FTAG);

	return (error);
	}

	static int
	zfs_ioc_pool_get_props(zfs_cmd_t *zc)
	{
	spa_t *spa;
	int error;
	nvlist_t *nvp = NULL;

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) {
	/*
	* If the pool is faulted, there may be properties we can still
	* get (such as altroot and cachefile), so attempt to get them
	* anyway.
	*/
	mutex_enter(&spa_namespace_lock);
	if ((spa = spa_lookup(zc->zc_name)) != NULL)
	error = spa_prop_get(spa, &nvp);
	mutex_exit(&spa_namespace_lock);
	} else {
	error = spa_prop_get(spa, &nvp);
	spa_close(spa, FTAG);
	}

	if (error == 0 && zc->zc_nvlist_dst != 0)
	error = put_nvlist(zc, nvp);
	else
	error = SET_ERROR(EFAULT);

	nvlist_free(nvp);
	return (error);
	}

	+/*
	+ * innvl: {
	+ * "vdevprops_set_vdev" -> guid
	+ * "vdevprops_set_props" -> { prop -> value }
	+ * }
	+ *
	+ * outnvl: propname -> error code (int32)
	+ */
	+static const zfs_ioc_key_t zfs_keys_vdev_set_props[] = {
	+ {ZPOOL_VDEV_PROPS_SET_VDEV, DATA_TYPE_UINT64, 0},
	+ {ZPOOL_VDEV_PROPS_SET_PROPS, DATA_TYPE_NVLIST, 0}
	+};
	+
	+static int
	+zfs_ioc_vdev_set_props(const char poolname, nvlist_t innvl, nvlist_t *outnvl)
	+{
	+ spa_t *spa;
	+ int error;
	+ vdev_t *vd;
	+ uint64_t vdev_guid;
	+
	+ /* Early validation */
	+ if (nvlist_lookup_uint64(innvl, ZPOOL_VDEV_PROPS_SET_VDEV,
	+ &vdev_guid) != 0)
	+ return (SET_ERROR(EINVAL));
	+
	+ if (outnvl == NULL)
	+ return (SET_ERROR(EINVAL));
	+
	+ if ((error = spa_open(poolname, &spa, FTAG)) != 0)
	+ return (error);
	+
	+ ASSERT(spa_writeable(spa));
	+
	+ if ((vd = spa_lookup_by_guid(spa, vdev_guid, B_TRUE)) == NULL) {
	+ spa_close(spa, FTAG);
	+ return (SET_ERROR(ENOENT));
	+ }
	+
	+ error = vdev_prop_set(vd, innvl, outnvl);
	+
	+ spa_close(spa, FTAG);
	+
	+ return (error);
	+}
	+
	+/*
	+ * innvl: {
	+ * "vdevprops_get_vdev" -> guid
	+ * (optional) "vdevprops_get_props" -> { propname -> propid }
	+ * }
	+ *
	+ * outnvl: propname -> value
	+ */
	+static const zfs_ioc_key_t zfs_keys_vdev_get_props[] = {
	+ {ZPOOL_VDEV_PROPS_GET_VDEV, DATA_TYPE_UINT64, 0},
	+ {ZPOOL_VDEV_PROPS_GET_PROPS, DATA_TYPE_NVLIST, ZK_OPTIONAL}
	+};
	+
	+static int
	+zfs_ioc_vdev_get_props(const char poolname, nvlist_t innvl, nvlist_t *outnvl)
	+{
	+ spa_t *spa;
	+ int error;
	+ vdev_t *vd;
	+ uint64_t vdev_guid;
	+
	+ /* Early validation */
	+ if (nvlist_lookup_uint64(innvl, ZPOOL_VDEV_PROPS_GET_VDEV,
	+ &vdev_guid) != 0)
	+ return (SET_ERROR(EINVAL));
	+
	+ if (outnvl == NULL)
	+ return (SET_ERROR(EINVAL));
	+
	+ if ((error = spa_open(poolname, &spa, FTAG)) != 0)
	+ return (error);
	+
	+ if ((vd = spa_lookup_by_guid(spa, vdev_guid, B_TRUE)) == NULL) {
	+ spa_close(spa, FTAG);
	+ return (SET_ERROR(ENOENT));
	+ }
	+
	+ error = vdev_prop_get(vd, innvl, outnvl);
	+
	+ spa_close(spa, FTAG);
	+
	+ return (error);
	+}
	+
	/*
	* inputs:
	* zc_name name of filesystem
	* zc_nvlist_src{_size} nvlist of delegated permissions
	* zc_perm_action allow/unallow flag
	*
	* outputs: none
	*/
	static int
	zfs_ioc_set_fsacl(zfs_cmd_t *zc)
	{
	int error;
	nvlist_t *fsaclnv = NULL;

	if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size,
	zc->zc_iflags, &fsaclnv)) != 0)
	return (error);

	/*
	* Verify nvlist is constructed correctly
	*/
	if ((error = zfs_deleg_verify_nvlist(fsaclnv)) != 0) {
	nvlist_free(fsaclnv);
	return (SET_ERROR(EINVAL));
	}

	/*
	* If we don't have PRIV_SYS_MOUNT, then validate
	* that user is allowed to hand out each permission in
	* the nvlist(s)
	*/

	error = secpolicy_zfs(CRED());
	if (error != 0) {
	if (zc->zc_perm_action == B_FALSE) {
	error = dsl_deleg_can_allow(zc->zc_name,
	fsaclnv, CRED());
	} else {
	error = dsl_deleg_can_unallow(zc->zc_name,
	fsaclnv, CRED());
	}
	}

	if (error == 0)
	error = dsl_deleg_set(zc->zc_name, fsaclnv, zc->zc_perm_action);

	nvlist_free(fsaclnv);
	return (error);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	*
	* outputs:
	* zc_nvlist_src{_size} nvlist of delegated permissions
	*/
	static int
	zfs_ioc_get_fsacl(zfs_cmd_t *zc)
	{
	nvlist_t *nvp;
	int error;

	if ((error = dsl_deleg_get(zc->zc_name, &nvp)) == 0) {
	error = put_nvlist(zc, nvp);
	nvlist_free(nvp);
	}

	return (error);
	}

	/* ARGSUSED */
	static void
	zfs_create_cb(objset_t os, void arg, cred_t cr, dmu_tx_t tx)
	{
	zfs_creat_t *zct = arg;

	zfs_create_fs(os, cr, zct->zct_zplprops, tx);
	}

	#define ZFS_PROP_UNDEFINED ((uint64_t)-1)

	/*
	* inputs:
	* os parent objset pointer (NULL if root fs)
	* fuids_ok fuids allowed in this version of the spa?
	* sa_ok SAs allowed in this version of the spa?
	* createprops list of properties requested by creator
	*
	* outputs:
	* zplprops values for the zplprops we attach to the master node object
	* is_ci true if requested file system will be purely case-insensitive
	*
	* Determine the settings for utf8only, normalization and
	* casesensitivity. Specific values may have been requested by the
	* creator and/or we can inherit values from the parent dataset. If
	* the file system is of too early a vintage, a creator can not
	* request settings for these properties, even if the requested
	* setting is the default value. We don't actually want to create dsl
	* properties for these, so remove them from the source nvlist after
	* processing.
	*/
	static int
	zfs_fill_zplprops_impl(objset_t *os, uint64_t zplver,
	boolean_t fuids_ok, boolean_t sa_ok, nvlist_t *createprops,
	nvlist_t zplprops, boolean_t is_ci)
	{
	uint64_t sense = ZFS_PROP_UNDEFINED;
	uint64_t norm = ZFS_PROP_UNDEFINED;
	uint64_t u8 = ZFS_PROP_UNDEFINED;
	int error;

	ASSERT(zplprops != NULL);

	/* parent dataset must be a filesystem */
	if (os != NULL && os->os_phys->os_type != DMU_OST_ZFS)
	return (SET_ERROR(ZFS_ERR_WRONG_PARENT));

	/*
	* Pull out creator prop choices, if any.
	*/
	if (createprops) {
	(void) nvlist_lookup_uint64(createprops,
	zfs_prop_to_name(ZFS_PROP_VERSION), &zplver);
	(void) nvlist_lookup_uint64(createprops,
	zfs_prop_to_name(ZFS_PROP_NORMALIZE), &norm);
	(void) nvlist_remove_all(createprops,
	zfs_prop_to_name(ZFS_PROP_NORMALIZE));
	(void) nvlist_lookup_uint64(createprops,
	zfs_prop_to_name(ZFS_PROP_UTF8ONLY), &u8);
	(void) nvlist_remove_all(createprops,
	zfs_prop_to_name(ZFS_PROP_UTF8ONLY));
	(void) nvlist_lookup_uint64(createprops,
	zfs_prop_to_name(ZFS_PROP_CASE), &sense);
	(void) nvlist_remove_all(createprops,
	zfs_prop_to_name(ZFS_PROP_CASE));
	}

	/*
	* If the zpl version requested is whacky or the file system
	* or pool is version is too "young" to support normalization
	* and the creator tried to set a value for one of the props,
	* error out.
	*/
	if ((zplver < ZPL_VERSION_INITIAL \|\| zplver > ZPL_VERSION) \|\|
	(zplver >= ZPL_VERSION_FUID && !fuids_ok) \|\|
	(zplver >= ZPL_VERSION_SA && !sa_ok) \|\|
	(zplver < ZPL_VERSION_NORMALIZATION &&
	(norm != ZFS_PROP_UNDEFINED \|\| u8 != ZFS_PROP_UNDEFINED \|\|
	sense != ZFS_PROP_UNDEFINED)))
	return (SET_ERROR(ENOTSUP));

	/*
	* Put the version in the zplprops
	*/
	VERIFY(nvlist_add_uint64(zplprops,
	zfs_prop_to_name(ZFS_PROP_VERSION), zplver) == 0);

	if (norm == ZFS_PROP_UNDEFINED &&
	(error = zfs_get_zplprop(os, ZFS_PROP_NORMALIZE, &norm)) != 0)
	return (error);
	VERIFY(nvlist_add_uint64(zplprops,
	zfs_prop_to_name(ZFS_PROP_NORMALIZE), norm) == 0);

	/*
	* If we're normalizing, names must always be valid UTF-8 strings.
	*/
	if (norm)
	u8 = 1;
	if (u8 == ZFS_PROP_UNDEFINED &&
	(error = zfs_get_zplprop(os, ZFS_PROP_UTF8ONLY, &u8)) != 0)
	return (error);
	VERIFY(nvlist_add_uint64(zplprops,
	zfs_prop_to_name(ZFS_PROP_UTF8ONLY), u8) == 0);

	if (sense == ZFS_PROP_UNDEFINED &&
	(error = zfs_get_zplprop(os, ZFS_PROP_CASE, &sense)) != 0)
	return (error);
	VERIFY(nvlist_add_uint64(zplprops,
	zfs_prop_to_name(ZFS_PROP_CASE), sense) == 0);

	if (is_ci)
	*is_ci = (sense == ZFS_CASE_INSENSITIVE);

	return (0);
	}

	static int
	zfs_fill_zplprops(const char dataset, nvlist_t createprops,
	nvlist_t zplprops, boolean_t is_ci)
	{
	boolean_t fuids_ok, sa_ok;
	uint64_t zplver = ZPL_VERSION;
	objset_t *os = NULL;
	char parentname[ZFS_MAX_DATASET_NAME_LEN];
	spa_t *spa;
	uint64_t spa_vers;
	int error;

	zfs_get_parent(dataset, parentname, sizeof (parentname));

	if ((error = spa_open(dataset, &spa, FTAG)) != 0)
	return (error);

	spa_vers = spa_version(spa);
	spa_close(spa, FTAG);

	zplver = zfs_zpl_version_map(spa_vers);
	fuids_ok = (zplver >= ZPL_VERSION_FUID);
	sa_ok = (zplver >= ZPL_VERSION_SA);

	/*
	* Open parent object set so we can inherit zplprop values.
	*/
	if ((error = dmu_objset_hold(parentname, FTAG, &os)) != 0)
	return (error);

	error = zfs_fill_zplprops_impl(os, zplver, fuids_ok, sa_ok, createprops,
	zplprops, is_ci);
	dmu_objset_rele(os, FTAG);
	return (error);
	}

	static int
	zfs_fill_zplprops_root(uint64_t spa_vers, nvlist_t *createprops,
	nvlist_t zplprops, boolean_t is_ci)
	{
	boolean_t fuids_ok;
	boolean_t sa_ok;
	uint64_t zplver = ZPL_VERSION;
	int error;

	zplver = zfs_zpl_version_map(spa_vers);
	fuids_ok = (zplver >= ZPL_VERSION_FUID);
	sa_ok = (zplver >= ZPL_VERSION_SA);

	error = zfs_fill_zplprops_impl(NULL, zplver, fuids_ok, sa_ok,
	createprops, zplprops, is_ci);
	return (error);
	}

	/*
	* innvl: {
	* "type" -> dmu_objset_type_t (int32)
	* (optional) "props" -> { prop -> value }
	* (optional) "hidden_args" -> { "wkeydata" -> value }
	* raw uint8_t array of encryption wrapping key data (32 bytes)
	* }
	*
	* outnvl: propname -> error code (int32)
	*/

	static const zfs_ioc_key_t zfs_keys_create[] = {
	{"type", DATA_TYPE_INT32, 0},
	{"props", DATA_TYPE_NVLIST, ZK_OPTIONAL},
	{"hidden_args", DATA_TYPE_NVLIST, ZK_OPTIONAL},
	};

	static int
	zfs_ioc_create(const char fsname, nvlist_t innvl, nvlist_t *outnvl)
	{
	int error = 0;
	zfs_creat_t zct = { 0 };
	nvlist_t *nvprops = NULL;
	nvlist_t *hidden_args = NULL;
	void (cbfunc)(objset_t os, void arg, cred_t cr, dmu_tx_t *tx);
	dmu_objset_type_t type;
	boolean_t is_insensitive = B_FALSE;
	dsl_crypto_params_t *dcp = NULL;

	type = (dmu_objset_type_t)fnvlist_lookup_int32(innvl, "type");
	(void) nvlist_lookup_nvlist(innvl, "props", &nvprops);
	(void) nvlist_lookup_nvlist(innvl, ZPOOL_HIDDEN_ARGS, &hidden_args);

	switch (type) {
	case DMU_OST_ZFS:
	cbfunc = zfs_create_cb;
	break;

	case DMU_OST_ZVOL:
	cbfunc = zvol_create_cb;
	break;

	default:
	cbfunc = NULL;
	break;
	}
	if (strchr(fsname, '@') \|\|
	strchr(fsname, '%'))
	return (SET_ERROR(EINVAL));

	zct.zct_props = nvprops;

	if (cbfunc == NULL)
	return (SET_ERROR(EINVAL));

	if (type == DMU_OST_ZVOL) {
	uint64_t volsize, volblocksize;

	if (nvprops == NULL)
	return (SET_ERROR(EINVAL));
	if (nvlist_lookup_uint64(nvprops,
	zfs_prop_to_name(ZFS_PROP_VOLSIZE), &volsize) != 0)
	return (SET_ERROR(EINVAL));

	if ((error = nvlist_lookup_uint64(nvprops,
	zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE),
	&volblocksize)) != 0 && error != ENOENT)
	return (SET_ERROR(EINVAL));

	if (error != 0)
	volblocksize = zfs_prop_default_numeric(
	ZFS_PROP_VOLBLOCKSIZE);

	if ((error = zvol_check_volblocksize(fsname,
	volblocksize)) != 0 \|\|
	(error = zvol_check_volsize(volsize,
	volblocksize)) != 0)
	return (error);
	} else if (type == DMU_OST_ZFS) {
	int error;

	/*
	* We have to have normalization and
	* case-folding flags correct when we do the
	* file system creation, so go figure them out
	* now.
	*/
	VERIFY(nvlist_alloc(&zct.zct_zplprops,
	NV_UNIQUE_NAME, KM_SLEEP) == 0);
	error = zfs_fill_zplprops(fsname, nvprops,
	zct.zct_zplprops, &is_insensitive);
	if (error != 0) {
	nvlist_free(zct.zct_zplprops);
	return (error);
	}
	}

	error = dsl_crypto_params_create_nvlist(DCP_CMD_NONE, nvprops,
	hidden_args, &dcp);
	if (error != 0) {
	nvlist_free(zct.zct_zplprops);
	return (error);
	}

	error = dmu_objset_create(fsname, type,
	is_insensitive ? DS_FLAG_CI_DATASET : 0, dcp, cbfunc, &zct);

	nvlist_free(zct.zct_zplprops);
	dsl_crypto_params_free(dcp, !!error);

	/*
	* It would be nice to do this atomically.
	*/
	if (error == 0) {
	error = zfs_set_prop_nvlist(fsname, ZPROP_SRC_LOCAL,
	nvprops, outnvl);
	if (error != 0) {
	spa_t *spa;
	int error2;

	/*
	* Volumes will return EBUSY and cannot be destroyed
	* until all asynchronous minor handling (e.g. from
	* setting the volmode property) has completed. Wait for
	* the spa_zvol_taskq to drain then retry.
	*/
	error2 = dsl_destroy_head(fsname);
	while ((error2 == EBUSY) && (type == DMU_OST_ZVOL)) {
	error2 = spa_open(fsname, &spa, FTAG);
	if (error2 == 0) {
	taskq_wait(spa->spa_zvol_taskq);
	spa_close(spa, FTAG);
	}
	error2 = dsl_destroy_head(fsname);
	}
	}
	}
	return (error);
	}

	/*
	* innvl: {
	* "origin" -> name of origin snapshot
	* (optional) "props" -> { prop -> value }
	* (optional) "hidden_args" -> { "wkeydata" -> value }
	* raw uint8_t array of encryption wrapping key data (32 bytes)
	* }
	*
	* outputs:
	* outnvl: propname -> error code (int32)
	*/
	static const zfs_ioc_key_t zfs_keys_clone[] = {
	{"origin", DATA_TYPE_STRING, 0},
	{"props", DATA_TYPE_NVLIST, ZK_OPTIONAL},
	{"hidden_args", DATA_TYPE_NVLIST, ZK_OPTIONAL},
	};

	static int
	zfs_ioc_clone(const char fsname, nvlist_t innvl, nvlist_t *outnvl)
	{
	int error = 0;
	nvlist_t *nvprops = NULL;
	const char *origin_name;

	origin_name = fnvlist_lookup_string(innvl, "origin");
	(void) nvlist_lookup_nvlist(innvl, "props", &nvprops);

	if (strchr(fsname, '@') \|\|
	strchr(fsname, '%'))
	return (SET_ERROR(EINVAL));

	if (dataset_namecheck(origin_name, NULL, NULL) != 0)
	return (SET_ERROR(EINVAL));

	error = dmu_objset_clone(fsname, origin_name);

	/*
	* It would be nice to do this atomically.
	*/
	if (error == 0) {
	error = zfs_set_prop_nvlist(fsname, ZPROP_SRC_LOCAL,
	nvprops, outnvl);
	if (error != 0)
	(void) dsl_destroy_head(fsname);
	}
	return (error);
	}

	static const zfs_ioc_key_t zfs_keys_remap[] = {
	/* no nvl keys */
	};

	/* ARGSUSED */
	static int
	zfs_ioc_remap(const char fsname, nvlist_t innvl, nvlist_t *outnvl)
	{
	/* This IOCTL is no longer supported. */
	return (0);
	}

	/*
	* innvl: {
	* "snaps" -> { snapshot1, snapshot2 }
	* (optional) "props" -> { prop -> value (string) }
	* }
	*
	* outnvl: snapshot -> error code (int32)
	*/
	static const zfs_ioc_key_t zfs_keys_snapshot[] = {
	{"snaps", DATA_TYPE_NVLIST, 0},
	{"props", DATA_TYPE_NVLIST, ZK_OPTIONAL},
	};

	static int
	zfs_ioc_snapshot(const char poolname, nvlist_t innvl, nvlist_t *outnvl)
	{
	nvlist_t *snaps;
	nvlist_t *props = NULL;
	int error, poollen;
	nvpair_t *pair;

	(void) nvlist_lookup_nvlist(innvl, "props", &props);
	if (!nvlist_empty(props) &&
	zfs_earlier_version(poolname, SPA_VERSION_SNAP_PROPS))
	return (SET_ERROR(ENOTSUP));
	if ((error = zfs_check_userprops(props)) != 0)
	return (error);

	snaps = fnvlist_lookup_nvlist(innvl, "snaps");
	poollen = strlen(poolname);
	for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL;
	pair = nvlist_next_nvpair(snaps, pair)) {
	const char *name = nvpair_name(pair);
	char *cp = strchr(name, '@');

	/*
	* The snap name must contain an @, and the part after it must
	* contain only valid characters.
	*/
	if (cp == NULL \|\|
	zfs_component_namecheck(cp + 1, NULL, NULL) != 0)
	return (SET_ERROR(EINVAL));

	/*
	* The snap must be in the specified pool.
	*/
	if (strncmp(name, poolname, poollen) != 0 \|\|
	(name[poollen] != '/' && name[poollen] != '@'))
	return (SET_ERROR(EXDEV));

	/*
	* Check for permission to set the properties on the fs.
	*/
	if (!nvlist_empty(props)) {
	*cp = '\0';
	error = zfs_secpolicy_write_perms(name,
	ZFS_DELEG_PERM_USERPROP, CRED());
	*cp = '@';
	if (error != 0)
	return (error);
	}

	/* This must be the only snap of this fs. */
	for (nvpair_t *pair2 = nvlist_next_nvpair(snaps, pair);
	pair2 != NULL; pair2 = nvlist_next_nvpair(snaps, pair2)) {
	if (strncmp(name, nvpair_name(pair2), cp - name + 1)
	== 0) {
	return (SET_ERROR(EXDEV));
	}
	}
	}

	error = dsl_dataset_snapshot(snaps, props, outnvl);

	return (error);
	}

	/*
	* innvl: "message" -> string
	*/
	static const zfs_ioc_key_t zfs_keys_log_history[] = {
	{"message", DATA_TYPE_STRING, 0},
	};

	/* ARGSUSED */
	static int
	zfs_ioc_log_history(const char unused, nvlist_t innvl, nvlist_t *outnvl)
	{
	const char *message;
	char *poolname;
	spa_t *spa;
	int error;

	/*
	* The poolname in the ioctl is not set, we get it from the TSD,
	* which was set at the end of the last successful ioctl that allows
	* logging. The secpolicy func already checked that it is set.
	* Only one log ioctl is allowed after each successful ioctl, so
	* we clear the TSD here.
	*/
	poolname = tsd_get(zfs_allow_log_key);
	if (poolname == NULL)
	return (SET_ERROR(EINVAL));
	(void) tsd_set(zfs_allow_log_key, NULL);
	error = spa_open(poolname, &spa, FTAG);
	kmem_strfree(poolname);
	if (error != 0)
	return (error);

	message = fnvlist_lookup_string(innvl, "message");

	if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY) {
	spa_close(spa, FTAG);
	return (SET_ERROR(ENOTSUP));
	}

	error = spa_history_log(spa, message);
	spa_close(spa, FTAG);
	return (error);
	}

	/*
	* This ioctl is used to set the bootenv configuration on the current
	* pool. This configuration is stored in the second padding area of the label,
	* and it is used by the bootloader(s) to store the bootloader and/or system
	* specific data.
	* The data is stored as nvlist data stream, and is protected by
	* an embedded checksum.
	* The version can have two possible values:
	* VB_RAW: nvlist should have key GRUB_ENVMAP, value DATA_TYPE_STRING.
	* VB_NVLIST: nvlist with arbitrary <key, value> pairs.
	*/
	static const zfs_ioc_key_t zfs_keys_set_bootenv[] = {
	{"version", DATA_TYPE_UINT64, 0},
	{"<keys>", DATA_TYPE_ANY, ZK_OPTIONAL \| ZK_WILDCARDLIST},
	};

	static int
	zfs_ioc_set_bootenv(const char name, nvlist_t innvl, nvlist_t *outnvl)
	{
	int error;
	spa_t *spa;

	if ((error = spa_open(name, &spa, FTAG)) != 0)
	return (error);
	spa_vdev_state_enter(spa, SCL_ALL);
	error = vdev_label_write_bootenv(spa->spa_root_vdev, innvl);
	(void) spa_vdev_state_exit(spa, NULL, 0);
	spa_close(spa, FTAG);
	return (error);
	}

	static const zfs_ioc_key_t zfs_keys_get_bootenv[] = {
	/* no nvl keys */
	};

	static int
	zfs_ioc_get_bootenv(const char name, nvlist_t innvl, nvlist_t *outnvl)
	{
	spa_t *spa;
	int error;

	if ((error = spa_open(name, &spa, FTAG)) != 0)
	return (error);
	spa_vdev_state_enter(spa, SCL_ALL);
	error = vdev_label_read_bootenv(spa->spa_root_vdev, outnvl);
	(void) spa_vdev_state_exit(spa, NULL, 0);
	spa_close(spa, FTAG);
	return (error);
	}

	/*
	* The dp_config_rwlock must not be held when calling this, because the
	* unmount may need to write out data.
	*
	* This function is best-effort. Callers must deal gracefully if it
	* remains mounted (or is remounted after this call).
	*
	* Returns 0 if the argument is not a snapshot, or it is not currently a
	* filesystem, or we were able to unmount it. Returns error code otherwise.
	*/
	void
	zfs_unmount_snap(const char *snapname)
	{
	if (strchr(snapname, '@') == NULL)
	return;

	(void) zfsctl_snapshot_unmount(snapname, MNT_FORCE);
	}

	/* ARGSUSED */
	static int
	zfs_unmount_snap_cb(const char snapname, void arg)
	{
	zfs_unmount_snap(snapname);
	return (0);
	}

	/*
	* When a clone is destroyed, its origin may also need to be destroyed,
	* in which case it must be unmounted. This routine will do that unmount
	* if necessary.
	*/
	void
	zfs_destroy_unmount_origin(const char *fsname)
	{
	int error;
	objset_t *os;
	dsl_dataset_t *ds;

	error = dmu_objset_hold(fsname, FTAG, &os);
	if (error != 0)
	return;
	ds = dmu_objset_ds(os);
	if (dsl_dir_is_clone(ds->ds_dir) && DS_IS_DEFER_DESTROY(ds->ds_prev)) {
	char originname[ZFS_MAX_DATASET_NAME_LEN];
	dsl_dataset_name(ds->ds_prev, originname);
	dmu_objset_rele(os, FTAG);
	zfs_unmount_snap(originname);
	} else {
	dmu_objset_rele(os, FTAG);
	}
	}

	/*
	* innvl: {
	* "snaps" -> { snapshot1, snapshot2 }
	* (optional boolean) "defer"
	* }
	*
	* outnvl: snapshot -> error code (int32)
	*/
	static const zfs_ioc_key_t zfs_keys_destroy_snaps[] = {
	{"snaps", DATA_TYPE_NVLIST, 0},
	{"defer", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	};

	/* ARGSUSED */
	static int
	zfs_ioc_destroy_snaps(const char poolname, nvlist_t innvl, nvlist_t *outnvl)
	{
	int poollen;
	nvlist_t *snaps;
	nvpair_t *pair;
	boolean_t defer;
	spa_t *spa;

	snaps = fnvlist_lookup_nvlist(innvl, "snaps");
	defer = nvlist_exists(innvl, "defer");

	poollen = strlen(poolname);
	for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL;
	pair = nvlist_next_nvpair(snaps, pair)) {
	const char *name = nvpair_name(pair);

	/*
	* The snap must be in the specified pool to prevent the
	* invalid removal of zvol minors below.
	*/
	if (strncmp(name, poolname, poollen) != 0 \|\|
	(name[poollen] != '/' && name[poollen] != '@'))
	return (SET_ERROR(EXDEV));

	zfs_unmount_snap(nvpair_name(pair));
	if (spa_open(name, &spa, FTAG) == 0) {
	zvol_remove_minors(spa, name, B_TRUE);
	spa_close(spa, FTAG);
	}
	}

	return (dsl_destroy_snapshots_nvl(snaps, defer, outnvl));
	}

	/*
	* Create bookmarks. The bookmark names are of the form <fs>#<bmark>.
	* All bookmarks and snapshots must be in the same pool.
	* dsl_bookmark_create_nvl_validate describes the nvlist schema in more detail.
	*
	* innvl: {
	* new_bookmark1 -> existing_snapshot,
	* new_bookmark2 -> existing_bookmark,
	* }
	*
	* outnvl: bookmark -> error code (int32)
	*
	*/
	static const zfs_ioc_key_t zfs_keys_bookmark[] = {
	{"<bookmark>...", DATA_TYPE_STRING, ZK_WILDCARDLIST},
	};

	/* ARGSUSED */
	static int
	zfs_ioc_bookmark(const char poolname, nvlist_t innvl, nvlist_t *outnvl)
	{
	return (dsl_bookmark_create(innvl, outnvl));
	}

	/*
	* innvl: {
	* property 1, property 2, ...
	* }
	*
	* outnvl: {
	* bookmark name 1 -> { property 1, property 2, ... },
	* bookmark name 2 -> { property 1, property 2, ... }
	* }
	*
	*/
	static const zfs_ioc_key_t zfs_keys_get_bookmarks[] = {
	{"<property>...", DATA_TYPE_BOOLEAN, ZK_WILDCARDLIST \| ZK_OPTIONAL},
	};

	static int
	zfs_ioc_get_bookmarks(const char fsname, nvlist_t innvl, nvlist_t *outnvl)
	{
	return (dsl_get_bookmarks(fsname, innvl, outnvl));
	}

	/*
	* innvl is not used.
	*
	* outnvl: {
	* property 1, property 2, ...
	* }
	*
	*/
	static const zfs_ioc_key_t zfs_keys_get_bookmark_props[] = {
	/* no nvl keys */
	};

	/* ARGSUSED */
	static int
	zfs_ioc_get_bookmark_props(const char bookmark, nvlist_t innvl,
	nvlist_t *outnvl)
	{
	char fsname[ZFS_MAX_DATASET_NAME_LEN];
	char *bmname;

	bmname = strchr(bookmark, '#');
	if (bmname == NULL)
	return (SET_ERROR(EINVAL));
	bmname++;

	(void) strlcpy(fsname, bookmark, sizeof (fsname));
	*(strchr(fsname, '#')) = '\0';

	return (dsl_get_bookmark_props(fsname, bmname, outnvl));
	}

	/*
	* innvl: {
	* bookmark name 1, bookmark name 2
	* }
	*
	* outnvl: bookmark -> error code (int32)
	*
	*/
	static const zfs_ioc_key_t zfs_keys_destroy_bookmarks[] = {
	{"<bookmark>...", DATA_TYPE_BOOLEAN, ZK_WILDCARDLIST},
	};

	static int
	zfs_ioc_destroy_bookmarks(const char poolname, nvlist_t innvl,
	nvlist_t *outnvl)
	{
	int error, poollen;

	poollen = strlen(poolname);
	for (nvpair_t *pair = nvlist_next_nvpair(innvl, NULL);
	pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) {
	const char *name = nvpair_name(pair);
	const char *cp = strchr(name, '#');

	/*
	* The bookmark name must contain an #, and the part after it
	* must contain only valid characters.
	*/
	if (cp == NULL \|\|
	zfs_component_namecheck(cp + 1, NULL, NULL) != 0)
	return (SET_ERROR(EINVAL));

	/*
	* The bookmark must be in the specified pool.
	*/
	if (strncmp(name, poolname, poollen) != 0 \|\|
	(name[poollen] != '/' && name[poollen] != '#'))
	return (SET_ERROR(EXDEV));
	}

	error = dsl_bookmark_destroy(innvl, outnvl);
	return (error);
	}

	static const zfs_ioc_key_t zfs_keys_channel_program[] = {
	{"program", DATA_TYPE_STRING, 0},
	{"arg", DATA_TYPE_ANY, 0},
	{"sync", DATA_TYPE_BOOLEAN_VALUE, ZK_OPTIONAL},
	{"instrlimit", DATA_TYPE_UINT64, ZK_OPTIONAL},
	{"memlimit", DATA_TYPE_UINT64, ZK_OPTIONAL},
	};

	static int
	zfs_ioc_channel_program(const char poolname, nvlist_t innvl,
	nvlist_t *outnvl)
	{
	char *program;
	uint64_t instrlimit, memlimit;
	boolean_t sync_flag;
	nvpair_t *nvarg = NULL;

	program = fnvlist_lookup_string(innvl, ZCP_ARG_PROGRAM);
	if (0 != nvlist_lookup_boolean_value(innvl, ZCP_ARG_SYNC, &sync_flag)) {
	sync_flag = B_TRUE;
	}
	if (0 != nvlist_lookup_uint64(innvl, ZCP_ARG_INSTRLIMIT, &instrlimit)) {
	instrlimit = ZCP_DEFAULT_INSTRLIMIT;
	}
	if (0 != nvlist_lookup_uint64(innvl, ZCP_ARG_MEMLIMIT, &memlimit)) {
	memlimit = ZCP_DEFAULT_MEMLIMIT;
	}
	nvarg = fnvlist_lookup_nvpair(innvl, ZCP_ARG_ARGLIST);

	if (instrlimit == 0 \|\| instrlimit > zfs_lua_max_instrlimit)
	return (SET_ERROR(EINVAL));
	if (memlimit == 0 \|\| memlimit > zfs_lua_max_memlimit)
	return (SET_ERROR(EINVAL));

	return (zcp_eval(poolname, program, sync_flag, instrlimit, memlimit,
	nvarg, outnvl));
	}

	/*
	* innvl: unused
	* outnvl: empty
	*/
	static const zfs_ioc_key_t zfs_keys_pool_checkpoint[] = {
	/* no nvl keys */
	};

	/* ARGSUSED */
	static int
	zfs_ioc_pool_checkpoint(const char poolname, nvlist_t innvl, nvlist_t *outnvl)
	{
	return (spa_checkpoint(poolname));
	}

	/*
	* innvl: unused
	* outnvl: empty
	*/
	static const zfs_ioc_key_t zfs_keys_pool_discard_checkpoint[] = {
	/* no nvl keys */
	};

	/* ARGSUSED */
	static int
	zfs_ioc_pool_discard_checkpoint(const char poolname, nvlist_t innvl,
	nvlist_t *outnvl)
	{
	return (spa_checkpoint_discard(poolname));
	}

	/*
	* inputs:
	* zc_name name of dataset to destroy
	* zc_defer_destroy mark for deferred destroy
	*
	* outputs: none
	*/
	static int
	zfs_ioc_destroy(zfs_cmd_t *zc)
	{
	objset_t *os;
	dmu_objset_type_t ost;
	int err;

	err = dmu_objset_hold(zc->zc_name, FTAG, &os);
	if (err != 0)
	return (err);
	ost = dmu_objset_type(os);
	dmu_objset_rele(os, FTAG);

	if (ost == DMU_OST_ZFS)
	zfs_unmount_snap(zc->zc_name);

	if (strchr(zc->zc_name, '@')) {
	err = dsl_destroy_snapshot(zc->zc_name, zc->zc_defer_destroy);
	} else {
	err = dsl_destroy_head(zc->zc_name);
	if (err == EEXIST) {
	/*
	* It is possible that the given DS may have
	* hidden child (%recv) datasets - "leftovers"
	* resulting from the previously interrupted
	* 'zfs receive'.
	*
	* 6 extra bytes for /%recv
	*/
	char namebuf[ZFS_MAX_DATASET_NAME_LEN + 6];

	if (snprintf(namebuf, sizeof (namebuf), "%s/%s",
	zc->zc_name, recv_clone_name) >=
	sizeof (namebuf))
	return (SET_ERROR(EINVAL));

	/*
	* Try to remove the hidden child (%recv) and after
	* that try to remove the target dataset.
	* If the hidden child (%recv) does not exist
	* the original error (EEXIST) will be returned
	*/
	err = dsl_destroy_head(namebuf);
	if (err == 0)
	err = dsl_destroy_head(zc->zc_name);
	else if (err == ENOENT)
	err = SET_ERROR(EEXIST);
	}
	}

	return (err);
	}

	/*
	* innvl: {
	* "initialize_command" -> POOL_INITIALIZE_{CANCEL\|START\|SUSPEND} (uint64)
	* "initialize_vdevs": { -> guids to initialize (nvlist)
	* "vdev_path_1": vdev_guid_1, (uint64),
	* "vdev_path_2": vdev_guid_2, (uint64),
	* ...
	* },
	* }
	*
	* outnvl: {
	* "initialize_vdevs": { -> initialization errors (nvlist)
	* "vdev_path_1": errno, see function body for possible errnos (uint64)
	* "vdev_path_2": errno, ... (uint64)
	* ...
	* }
	* }
	*
	* EINVAL is returned for an unknown commands or if any of the provided vdev
	* guids have be specified with a type other than uint64.
	*/
	static const zfs_ioc_key_t zfs_keys_pool_initialize[] = {
	{ZPOOL_INITIALIZE_COMMAND, DATA_TYPE_UINT64, 0},
	{ZPOOL_INITIALIZE_VDEVS, DATA_TYPE_NVLIST, 0}
	};

	static int
	zfs_ioc_pool_initialize(const char poolname, nvlist_t innvl, nvlist_t *outnvl)
	{
	uint64_t cmd_type;
	if (nvlist_lookup_uint64(innvl, ZPOOL_INITIALIZE_COMMAND,
	&cmd_type) != 0) {
	return (SET_ERROR(EINVAL));
	}

	if (!(cmd_type == POOL_INITIALIZE_CANCEL \|\|
	cmd_type == POOL_INITIALIZE_START \|\|
	cmd_type == POOL_INITIALIZE_SUSPEND)) {
	return (SET_ERROR(EINVAL));
	}

	nvlist_t *vdev_guids;
	if (nvlist_lookup_nvlist(innvl, ZPOOL_INITIALIZE_VDEVS,
	&vdev_guids) != 0) {
	return (SET_ERROR(EINVAL));
	}

	for (nvpair_t *pair = nvlist_next_nvpair(vdev_guids, NULL);
	pair != NULL; pair = nvlist_next_nvpair(vdev_guids, pair)) {
	uint64_t vdev_guid;
	if (nvpair_value_uint64(pair, &vdev_guid) != 0) {
	return (SET_ERROR(EINVAL));
	}
	}

	spa_t *spa;
	int error = spa_open(poolname, &spa, FTAG);
	if (error != 0)
	return (error);

	nvlist_t *vdev_errlist = fnvlist_alloc();
	int total_errors = spa_vdev_initialize(spa, vdev_guids, cmd_type,
	vdev_errlist);

	if (fnvlist_size(vdev_errlist) > 0) {
	fnvlist_add_nvlist(outnvl, ZPOOL_INITIALIZE_VDEVS,
	vdev_errlist);
	}
	fnvlist_free(vdev_errlist);

	spa_close(spa, FTAG);
	return (total_errors > 0 ? SET_ERROR(EINVAL) : 0);
	}

	/*
	* innvl: {
	* "trim_command" -> POOL_TRIM_{CANCEL\|START\|SUSPEND} (uint64)
	* "trim_vdevs": { -> guids to TRIM (nvlist)
	* "vdev_path_1": vdev_guid_1, (uint64),
	* "vdev_path_2": vdev_guid_2, (uint64),
	* ...
	* },
	* "trim_rate" -> Target TRIM rate in bytes/sec.
	* "trim_secure" -> Set to request a secure TRIM.
	* }
	*
	* outnvl: {
	* "trim_vdevs": { -> TRIM errors (nvlist)
	* "vdev_path_1": errno, see function body for possible errnos (uint64)
	* "vdev_path_2": errno, ... (uint64)
	* ...
	* }
	* }
	*
	* EINVAL is returned for an unknown commands or if any of the provided vdev
	* guids have be specified with a type other than uint64.
	*/
	static const zfs_ioc_key_t zfs_keys_pool_trim[] = {
	{ZPOOL_TRIM_COMMAND, DATA_TYPE_UINT64, 0},
	{ZPOOL_TRIM_VDEVS, DATA_TYPE_NVLIST, 0},
	{ZPOOL_TRIM_RATE, DATA_TYPE_UINT64, ZK_OPTIONAL},
	{ZPOOL_TRIM_SECURE, DATA_TYPE_BOOLEAN_VALUE, ZK_OPTIONAL},
	};

	static int
	zfs_ioc_pool_trim(const char poolname, nvlist_t innvl, nvlist_t *outnvl)
	{
	uint64_t cmd_type;
	if (nvlist_lookup_uint64(innvl, ZPOOL_TRIM_COMMAND, &cmd_type) != 0)
	return (SET_ERROR(EINVAL));

	if (!(cmd_type == POOL_TRIM_CANCEL \|\|
	cmd_type == POOL_TRIM_START \|\|
	cmd_type == POOL_TRIM_SUSPEND)) {
	return (SET_ERROR(EINVAL));
	}

	nvlist_t *vdev_guids;
	if (nvlist_lookup_nvlist(innvl, ZPOOL_TRIM_VDEVS, &vdev_guids) != 0)
	return (SET_ERROR(EINVAL));

	for (nvpair_t *pair = nvlist_next_nvpair(vdev_guids, NULL);
	pair != NULL; pair = nvlist_next_nvpair(vdev_guids, pair)) {
	uint64_t vdev_guid;
	if (nvpair_value_uint64(pair, &vdev_guid) != 0) {
	return (SET_ERROR(EINVAL));
	}
	}

	/* Optional, defaults to maximum rate when not provided */
	uint64_t rate;
	if (nvlist_lookup_uint64(innvl, ZPOOL_TRIM_RATE, &rate) != 0)
	rate = 0;

	/* Optional, defaults to standard TRIM when not provided */
	boolean_t secure;
	if (nvlist_lookup_boolean_value(innvl, ZPOOL_TRIM_SECURE,
	&secure) != 0) {
	secure = B_FALSE;
	}

	spa_t *spa;
	int error = spa_open(poolname, &spa, FTAG);
	if (error != 0)
	return (error);

	nvlist_t *vdev_errlist = fnvlist_alloc();
	int total_errors = spa_vdev_trim(spa, vdev_guids, cmd_type,
	rate, !!zfs_trim_metaslab_skip, secure, vdev_errlist);

	if (fnvlist_size(vdev_errlist) > 0)
	fnvlist_add_nvlist(outnvl, ZPOOL_TRIM_VDEVS, vdev_errlist);

	fnvlist_free(vdev_errlist);

	spa_close(spa, FTAG);
	return (total_errors > 0 ? SET_ERROR(EINVAL) : 0);
	}

	/*
	* This ioctl waits for activity of a particular type to complete. If there is
	* no activity of that type in progress, it returns immediately, and the
	* returned value "waited" is false. If there is activity in progress, and no
	* tag is passed in, the ioctl blocks until all activity of that type is
	* complete, and then returns with "waited" set to true.
	*
	* If a tag is provided, it identifies a particular instance of an activity to
	* wait for. Currently, this is only valid for use with 'initialize', because
	* that is the only activity for which there can be multiple instances running
	* concurrently. In the case of 'initialize', the tag corresponds to the guid of
	* the vdev on which to wait.
	*
	* If a thread waiting in the ioctl receives a signal, the call will return
	* immediately, and the return value will be EINTR.
	*
	* innvl: {
	* "wait_activity" -> int32_t
	* (optional) "wait_tag" -> uint64_t
	* }
	*
	* outnvl: "waited" -> boolean_t
	*/
	static const zfs_ioc_key_t zfs_keys_pool_wait[] = {
	{ZPOOL_WAIT_ACTIVITY, DATA_TYPE_INT32, 0},
	{ZPOOL_WAIT_TAG, DATA_TYPE_UINT64, ZK_OPTIONAL},
	};

	static int
	zfs_ioc_wait(const char name, nvlist_t innvl, nvlist_t *outnvl)
	{
	int32_t activity;
	uint64_t tag;
	boolean_t waited;
	int error;

	if (nvlist_lookup_int32(innvl, ZPOOL_WAIT_ACTIVITY, &activity) != 0)
	return (EINVAL);

	if (nvlist_lookup_uint64(innvl, ZPOOL_WAIT_TAG, &tag) == 0)
	error = spa_wait_tag(name, activity, tag, &waited);
	else
	error = spa_wait(name, activity, &waited);

	if (error == 0)
	fnvlist_add_boolean_value(outnvl, ZPOOL_WAIT_WAITED, waited);

	return (error);
	}

	/*
	* This ioctl waits for activity of a particular type to complete. If there is
	* no activity of that type in progress, it returns immediately, and the
	* returned value "waited" is false. If there is activity in progress, and no
	* tag is passed in, the ioctl blocks until all activity of that type is
	* complete, and then returns with "waited" set to true.
	*
	* If a thread waiting in the ioctl receives a signal, the call will return
	* immediately, and the return value will be EINTR.
	*
	* innvl: {
	* "wait_activity" -> int32_t
	* }
	*
	* outnvl: "waited" -> boolean_t
	*/
	static const zfs_ioc_key_t zfs_keys_fs_wait[] = {
	{ZFS_WAIT_ACTIVITY, DATA_TYPE_INT32, 0},
	};

	static int
	zfs_ioc_wait_fs(const char name, nvlist_t innvl, nvlist_t *outnvl)
	{
	int32_t activity;
	boolean_t waited = B_FALSE;
	int error;
	dsl_pool_t *dp;
	dsl_dir_t *dd;
	dsl_dataset_t *ds;

	if (nvlist_lookup_int32(innvl, ZFS_WAIT_ACTIVITY, &activity) != 0)
	return (SET_ERROR(EINVAL));

	if (activity >= ZFS_WAIT_NUM_ACTIVITIES \|\| activity < 0)
	return (SET_ERROR(EINVAL));

	if ((error = dsl_pool_hold(name, FTAG, &dp)) != 0)
	return (error);

	if ((error = dsl_dataset_hold(dp, name, FTAG, &ds)) != 0) {
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	dd = ds->ds_dir;
	mutex_enter(&dd->dd_activity_lock);
	dd->dd_activity_waiters++;

	/*
	* We get a long-hold here so that the dsl_dataset_t and dsl_dir_t
	* aren't evicted while we're waiting. Normally this is prevented by
	* holding the pool, but we can't do that while we're waiting since
	* that would prevent TXGs from syncing out. Some of the functionality
	* of long-holds (e.g. preventing deletion) is unnecessary for this
	* case, since we would cancel the waiters before proceeding with a
	* deletion. An alternative mechanism for keeping the dataset around
	* could be developed but this is simpler.
	*/
	dsl_dataset_long_hold(ds, FTAG);
	dsl_pool_rele(dp, FTAG);

	error = dsl_dir_wait(dd, ds, activity, &waited);

	dsl_dataset_long_rele(ds, FTAG);
	dd->dd_activity_waiters--;
	if (dd->dd_activity_waiters == 0)
	cv_signal(&dd->dd_activity_cv);
	mutex_exit(&dd->dd_activity_lock);

	dsl_dataset_rele(ds, FTAG);

	if (error == 0)
	fnvlist_add_boolean_value(outnvl, ZFS_WAIT_WAITED, waited);

	return (error);
	}

	/*
	* fsname is name of dataset to rollback (to most recent snapshot)
	*
	* innvl may contain name of expected target snapshot
	*
	* outnvl: "target" -> name of most recent snapshot
	* }
	*/
	static const zfs_ioc_key_t zfs_keys_rollback[] = {
	{"target", DATA_TYPE_STRING, ZK_OPTIONAL},
	};

	/* ARGSUSED */
	static int
	zfs_ioc_rollback(const char fsname, nvlist_t innvl, nvlist_t *outnvl)
	{
	zfsvfs_t *zfsvfs;
	zvol_state_handle_t *zv;
	char *target = NULL;
	int error;

	(void) nvlist_lookup_string(innvl, "target", &target);
	if (target != NULL) {
	const char *cp = strchr(target, '@');

	/*
	* The snap name must contain an @, and the part after it must
	* contain only valid characters.
	*/
	if (cp == NULL \|\|
	zfs_component_namecheck(cp + 1, NULL, NULL) != 0)
	return (SET_ERROR(EINVAL));
	}

	if (getzfsvfs(fsname, &zfsvfs) == 0) {
	dsl_dataset_t *ds;

	ds = dmu_objset_ds(zfsvfs->z_os);
	error = zfs_suspend_fs(zfsvfs);
	if (error == 0) {
	int resume_err;

	error = dsl_dataset_rollback(fsname, target, zfsvfs,
	outnvl);
	resume_err = zfs_resume_fs(zfsvfs, ds);
	error = error ? error : resume_err;
	}
	zfs_vfs_rele(zfsvfs);
	} else if ((zv = zvol_suspend(fsname)) != NULL) {
	error = dsl_dataset_rollback(fsname, target, zvol_tag(zv),
	outnvl);
	zvol_resume(zv);
	} else {
	error = dsl_dataset_rollback(fsname, target, NULL, outnvl);
	}
	return (error);
	}

	static int
	recursive_unmount(const char fsname, void arg)
	{
	const char *snapname = arg;
	char *fullname;

	fullname = kmem_asprintf("%s@%s", fsname, snapname);
	zfs_unmount_snap(fullname);
	kmem_strfree(fullname);

	return (0);
	}

	/*
	*
	* snapname is the snapshot to redact.
	* innvl: {
	* "bookname" -> (string)
	* shortname of the redaction bookmark to generate
	* "snapnv" -> (nvlist, values ignored)
	* snapshots to redact snapname with respect to
	* }
	*
	* outnvl is unused
	*/

	/* ARGSUSED */
	static const zfs_ioc_key_t zfs_keys_redact[] = {
	{"bookname", DATA_TYPE_STRING, 0},
	{"snapnv", DATA_TYPE_NVLIST, 0},
	};
	static int
	zfs_ioc_redact(const char snapname, nvlist_t innvl, nvlist_t *outnvl)
	{
	nvlist_t *redactnvl = NULL;
	char *redactbook = NULL;

	if (nvlist_lookup_nvlist(innvl, "snapnv", &redactnvl) != 0)
	return (SET_ERROR(EINVAL));
	if (fnvlist_num_pairs(redactnvl) == 0)
	return (SET_ERROR(ENXIO));
	if (nvlist_lookup_string(innvl, "bookname", &redactbook) != 0)
	return (SET_ERROR(EINVAL));

	return (dmu_redact_snap(snapname, redactnvl, redactbook));
	}

	/*
	* inputs:
	* zc_name old name of dataset
	* zc_value new name of dataset
	* zc_cookie recursive flag (only valid for snapshots)
	*
	* outputs: none
	*/
	static int
	zfs_ioc_rename(zfs_cmd_t *zc)
	{
	objset_t *os;
	dmu_objset_type_t ost;
	boolean_t recursive = zc->zc_cookie & 1;
	boolean_t nounmount = !!(zc->zc_cookie & 2);
	char *at;
	int err;

	/* "zfs rename" from and to ...%recv datasets should both fail */
	zc->zc_name[sizeof (zc->zc_name) - 1] = '\0';
	zc->zc_value[sizeof (zc->zc_value) - 1] = '\0';
	if (dataset_namecheck(zc->zc_name, NULL, NULL) != 0 \|\|
	dataset_namecheck(zc->zc_value, NULL, NULL) != 0 \|\|
	strchr(zc->zc_name, '%') \|\| strchr(zc->zc_value, '%'))
	return (SET_ERROR(EINVAL));

	err = dmu_objset_hold(zc->zc_name, FTAG, &os);
	if (err != 0)
	return (err);
	ost = dmu_objset_type(os);
	dmu_objset_rele(os, FTAG);

	at = strchr(zc->zc_name, '@');
	if (at != NULL) {
	/* snaps must be in same fs */
	int error;

	if (strncmp(zc->zc_name, zc->zc_value, at - zc->zc_name + 1))
	return (SET_ERROR(EXDEV));
	*at = '\0';
	if (ost == DMU_OST_ZFS && !nounmount) {
	error = dmu_objset_find(zc->zc_name,
	recursive_unmount, at + 1,
	recursive ? DS_FIND_CHILDREN : 0);
	if (error != 0) {
	*at = '@';
	return (error);
	}
	}
	error = dsl_dataset_rename_snapshot(zc->zc_name,
	at + 1, strchr(zc->zc_value, '@') + 1, recursive);
	*at = '@';

	return (error);
	} else {
	return (dsl_dir_rename(zc->zc_name, zc->zc_value));
	}
	}

	static int
	zfs_check_settable(const char dsname, nvpair_t pair, cred_t *cr)
	{
	const char *propname = nvpair_name(pair);
	boolean_t issnap = (strchr(dsname, '@') != NULL);
	zfs_prop_t prop = zfs_name_to_prop(propname);
	uint64_t intval, compval;
	int err;

	if (prop == ZPROP_INVAL) {
	if (zfs_prop_user(propname)) {
	if ((err = zfs_secpolicy_write_perms(dsname,
	ZFS_DELEG_PERM_USERPROP, cr)))
	return (err);
	return (0);
	}

	if (!issnap && zfs_prop_userquota(propname)) {
	const char *perm = NULL;
	const char *uq_prefix =
	zfs_userquota_prop_prefixes[ZFS_PROP_USERQUOTA];
	const char *gq_prefix =
	zfs_userquota_prop_prefixes[ZFS_PROP_GROUPQUOTA];
	const char *uiq_prefix =
	zfs_userquota_prop_prefixes[ZFS_PROP_USEROBJQUOTA];
	const char *giq_prefix =
	zfs_userquota_prop_prefixes[ZFS_PROP_GROUPOBJQUOTA];
	const char *pq_prefix =
	zfs_userquota_prop_prefixes[ZFS_PROP_PROJECTQUOTA];
	const char *piq_prefix = zfs_userquota_prop_prefixes[\
	ZFS_PROP_PROJECTOBJQUOTA];

	if (strncmp(propname, uq_prefix,
	strlen(uq_prefix)) == 0) {
	perm = ZFS_DELEG_PERM_USERQUOTA;
	} else if (strncmp(propname, uiq_prefix,
	strlen(uiq_prefix)) == 0) {
	perm = ZFS_DELEG_PERM_USEROBJQUOTA;
	} else if (strncmp(propname, gq_prefix,
	strlen(gq_prefix)) == 0) {
	perm = ZFS_DELEG_PERM_GROUPQUOTA;
	} else if (strncmp(propname, giq_prefix,
	strlen(giq_prefix)) == 0) {
	perm = ZFS_DELEG_PERM_GROUPOBJQUOTA;
	} else if (strncmp(propname, pq_prefix,
	strlen(pq_prefix)) == 0) {
	perm = ZFS_DELEG_PERM_PROJECTQUOTA;
	} else if (strncmp(propname, piq_prefix,
	strlen(piq_prefix)) == 0) {
	perm = ZFS_DELEG_PERM_PROJECTOBJQUOTA;
	} else {
	/* {USER\|GROUP\|PROJECT}USED are read-only */
	return (SET_ERROR(EINVAL));
	}

	if ((err = zfs_secpolicy_write_perms(dsname, perm, cr)))
	return (err);
	return (0);
	}

	return (SET_ERROR(EINVAL));
	}

	if (issnap)
	return (SET_ERROR(EINVAL));

	if (nvpair_type(pair) == DATA_TYPE_NVLIST) {
	/*
	* dsl_prop_get_all_impl() returns properties in this
	* format.
	*/
	nvlist_t *attrs;
	VERIFY(nvpair_value_nvlist(pair, &attrs) == 0);
	VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE,
	&pair) == 0);
	}

	/*
	* Check that this value is valid for this pool version
	*/
	switch (prop) {
	case ZFS_PROP_COMPRESSION:
	/*
	* If the user specified gzip compression, make sure
	* the SPA supports it. We ignore any errors here since
	* we'll catch them later.
	*/
	if (nvpair_value_uint64(pair, &intval) == 0) {
	compval = ZIO_COMPRESS_ALGO(intval);
	if (compval >= ZIO_COMPRESS_GZIP_1 &&
	compval <= ZIO_COMPRESS_GZIP_9 &&
	zfs_earlier_version(dsname,
	SPA_VERSION_GZIP_COMPRESSION)) {
	return (SET_ERROR(ENOTSUP));
	}

	if (compval == ZIO_COMPRESS_ZLE &&
	zfs_earlier_version(dsname,
	SPA_VERSION_ZLE_COMPRESSION))
	return (SET_ERROR(ENOTSUP));

	if (compval == ZIO_COMPRESS_LZ4) {
	spa_t *spa;

	if ((err = spa_open(dsname, &spa, FTAG)) != 0)
	return (err);

	if (!spa_feature_is_enabled(spa,
	SPA_FEATURE_LZ4_COMPRESS)) {
	spa_close(spa, FTAG);
	return (SET_ERROR(ENOTSUP));
	}
	spa_close(spa, FTAG);
	}

	if (compval == ZIO_COMPRESS_ZSTD) {
	spa_t *spa;

	if ((err = spa_open(dsname, &spa, FTAG)) != 0)
	return (err);

	if (!spa_feature_is_enabled(spa,
	SPA_FEATURE_ZSTD_COMPRESS)) {
	spa_close(spa, FTAG);
	return (SET_ERROR(ENOTSUP));
	}
	spa_close(spa, FTAG);
	}
	}
	break;

	case ZFS_PROP_COPIES:
	if (zfs_earlier_version(dsname, SPA_VERSION_DITTO_BLOCKS))
	return (SET_ERROR(ENOTSUP));
	break;

	case ZFS_PROP_VOLBLOCKSIZE:
	case ZFS_PROP_RECORDSIZE:
	/* Record sizes above 128k need the feature to be enabled */
	if (nvpair_value_uint64(pair, &intval) == 0 &&
	intval > SPA_OLD_MAXBLOCKSIZE) {
	spa_t *spa;

	/*
	* We don't allow setting the property above 1MB,
	* unless the tunable has been changed.
	*/
	if (intval > zfs_max_recordsize \|\|
	intval > SPA_MAXBLOCKSIZE)
	return (SET_ERROR(ERANGE));

	if ((err = spa_open(dsname, &spa, FTAG)) != 0)
	return (err);

	if (!spa_feature_is_enabled(spa,
	SPA_FEATURE_LARGE_BLOCKS)) {
	spa_close(spa, FTAG);
	return (SET_ERROR(ENOTSUP));
	}
	spa_close(spa, FTAG);
	}
	break;

	case ZFS_PROP_DNODESIZE:
	/* Dnode sizes above 512 need the feature to be enabled */
	if (nvpair_value_uint64(pair, &intval) == 0 &&
	intval != ZFS_DNSIZE_LEGACY) {
	spa_t *spa;

	if ((err = spa_open(dsname, &spa, FTAG)) != 0)
	return (err);

	if (!spa_feature_is_enabled(spa,
	SPA_FEATURE_LARGE_DNODE)) {
	spa_close(spa, FTAG);
	return (SET_ERROR(ENOTSUP));
	}
	spa_close(spa, FTAG);
	}
	break;

	case ZFS_PROP_SPECIAL_SMALL_BLOCKS:
	/*
	* This property could require the allocation classes
	* feature to be active for setting, however we allow
	* it so that tests of settable properties succeed.
	* The CLI will issue a warning in this case.
	*/
	break;

	case ZFS_PROP_SHARESMB:
	if (zpl_earlier_version(dsname, ZPL_VERSION_FUID))
	return (SET_ERROR(ENOTSUP));
	break;

	case ZFS_PROP_ACLINHERIT:
	if (nvpair_type(pair) == DATA_TYPE_UINT64 &&
	nvpair_value_uint64(pair, &intval) == 0) {
	if (intval == ZFS_ACL_PASSTHROUGH_X &&
	zfs_earlier_version(dsname,
	SPA_VERSION_PASSTHROUGH_X))
	return (SET_ERROR(ENOTSUP));
	}
	break;
	case ZFS_PROP_CHECKSUM:
	case ZFS_PROP_DEDUP:
	{
	spa_feature_t feature;
	spa_t *spa;
	int err;

	/* dedup feature version checks */
	if (prop == ZFS_PROP_DEDUP &&
	zfs_earlier_version(dsname, SPA_VERSION_DEDUP))
	return (SET_ERROR(ENOTSUP));

	if (nvpair_type(pair) == DATA_TYPE_UINT64 &&
	nvpair_value_uint64(pair, &intval) == 0) {
	/* check prop value is enabled in features */
	feature = zio_checksum_to_feature(
	intval & ZIO_CHECKSUM_MASK);
	if (feature == SPA_FEATURE_NONE)
	break;

	if ((err = spa_open(dsname, &spa, FTAG)) != 0)
	return (err);

	if (!spa_feature_is_enabled(spa, feature)) {
	spa_close(spa, FTAG);
	return (SET_ERROR(ENOTSUP));
	}
	spa_close(spa, FTAG);
	}
	break;
	}

	default:
	break;
	}

	return (zfs_secpolicy_setprop(dsname, prop, pair, CRED()));
	}

	/*
	* Removes properties from the given props list that fail permission checks
	* needed to clear them and to restore them in case of a receive error. For each
	* property, make sure we have both set and inherit permissions.
	*
	* Returns the first error encountered if any permission checks fail. If the
	* caller provides a non-NULL errlist, it also gives the complete list of names
	* of all the properties that failed a permission check along with the
	* corresponding error numbers. The caller is responsible for freeing the
	* returned errlist.
	*
	* If every property checks out successfully, zero is returned and the list
	* pointed at by errlist is NULL.
	*/
	static int
	zfs_check_clearable(const char dataset, nvlist_t props, nvlist_t **errlist)
	{
	zfs_cmd_t *zc;
	nvpair_t pair, next_pair;
	nvlist_t *errors;
	int err, rv = 0;

	if (props == NULL)
	return (0);

	VERIFY(nvlist_alloc(&errors, NV_UNIQUE_NAME, KM_SLEEP) == 0);

	zc = kmem_alloc(sizeof (zfs_cmd_t), KM_SLEEP);
	(void) strlcpy(zc->zc_name, dataset, sizeof (zc->zc_name));
	pair = nvlist_next_nvpair(props, NULL);
	while (pair != NULL) {
	next_pair = nvlist_next_nvpair(props, pair);

	(void) strlcpy(zc->zc_value, nvpair_name(pair),
	sizeof (zc->zc_value));
	if ((err = zfs_check_settable(dataset, pair, CRED())) != 0 \|\|
	(err = zfs_secpolicy_inherit_prop(zc, NULL, CRED())) != 0) {
	VERIFY(nvlist_remove_nvpair(props, pair) == 0);
	VERIFY(nvlist_add_int32(errors,
	zc->zc_value, err) == 0);
	}
	pair = next_pair;
	}
	kmem_free(zc, sizeof (zfs_cmd_t));

	if ((pair = nvlist_next_nvpair(errors, NULL)) == NULL) {
	nvlist_free(errors);
	errors = NULL;
	} else {
	VERIFY(nvpair_value_int32(pair, &rv) == 0);
	}

	if (errlist == NULL)
	nvlist_free(errors);
	else
	*errlist = errors;

	return (rv);
	}

	static boolean_t
	propval_equals(nvpair_t p1, nvpair_t p2)
	{
	if (nvpair_type(p1) == DATA_TYPE_NVLIST) {
	/* dsl_prop_get_all_impl() format */
	nvlist_t *attrs;
	VERIFY(nvpair_value_nvlist(p1, &attrs) == 0);
	VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE,
	&p1) == 0);
	}

	if (nvpair_type(p2) == DATA_TYPE_NVLIST) {
	nvlist_t *attrs;
	VERIFY(nvpair_value_nvlist(p2, &attrs) == 0);
	VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE,
	&p2) == 0);
	}

	if (nvpair_type(p1) != nvpair_type(p2))
	return (B_FALSE);

	if (nvpair_type(p1) == DATA_TYPE_STRING) {
	char valstr1, valstr2;

	VERIFY(nvpair_value_string(p1, (char **)&valstr1) == 0);
	VERIFY(nvpair_value_string(p2, (char **)&valstr2) == 0);
	return (strcmp(valstr1, valstr2) == 0);
	} else {
	uint64_t intval1, intval2;

	VERIFY(nvpair_value_uint64(p1, &intval1) == 0);
	VERIFY(nvpair_value_uint64(p2, &intval2) == 0);
	return (intval1 == intval2);
	}
	}

	/*
	* Remove properties from props if they are not going to change (as determined
	* by comparison with origprops). Remove them from origprops as well, since we
	* do not need to clear or restore properties that won't change.
	*/
	static void
	props_reduce(nvlist_t props, nvlist_t origprops)
	{
	nvpair_t pair, next_pair;

	if (origprops == NULL)
	return; /* all props need to be received */

	pair = nvlist_next_nvpair(props, NULL);
	while (pair != NULL) {
	const char *propname = nvpair_name(pair);
	nvpair_t *match;

	next_pair = nvlist_next_nvpair(props, pair);

	if ((nvlist_lookup_nvpair(origprops, propname,
	&match) != 0) \|\| !propval_equals(pair, match))
	goto next; /* need to set received value */

	/* don't clear the existing received value */
	(void) nvlist_remove_nvpair(origprops, match);
	/* don't bother receiving the property */
	(void) nvlist_remove_nvpair(props, pair);
	next:
	pair = next_pair;
	}
	}

	/*
	* Extract properties that cannot be set PRIOR to the receipt of a dataset.
	* For example, refquota cannot be set until after the receipt of a dataset,
	* because in replication streams, an older/earlier snapshot may exceed the
	* refquota. We want to receive the older/earlier snapshot, but setting
	* refquota pre-receipt will set the dsl's ACTUAL quota, which will prevent
	* the older/earlier snapshot from being received (with EDQUOT).
	*
	* The ZFS test "zfs_receive_011_pos" demonstrates such a scenario.
	*
	* libzfs will need to be judicious handling errors encountered by props
	* extracted by this function.
	*/
	static nvlist_t *
	extract_delay_props(nvlist_t *props)
	{
	nvlist_t *delayprops;
	nvpair_t nvp, tmp;
	static const zfs_prop_t delayable[] = {
	ZFS_PROP_REFQUOTA,
	ZFS_PROP_KEYLOCATION,
	0
	};
	int i;

	VERIFY(nvlist_alloc(&delayprops, NV_UNIQUE_NAME, KM_SLEEP) == 0);

	for (nvp = nvlist_next_nvpair(props, NULL); nvp != NULL;
	nvp = nvlist_next_nvpair(props, nvp)) {
	/*
	* strcmp() is safe because zfs_prop_to_name() always returns
	* a bounded string.
	*/
	for (i = 0; delayable[i] != 0; i++) {
	if (strcmp(zfs_prop_to_name(delayable[i]),
	nvpair_name(nvp)) == 0) {
	break;
	}
	}
	if (delayable[i] != 0) {
	tmp = nvlist_prev_nvpair(props, nvp);
	VERIFY(nvlist_add_nvpair(delayprops, nvp) == 0);
	VERIFY(nvlist_remove_nvpair(props, nvp) == 0);
	nvp = tmp;
	}
	}

	if (nvlist_empty(delayprops)) {
	nvlist_free(delayprops);
	delayprops = NULL;
	}
	return (delayprops);
	}

	static void
	zfs_allow_log_destroy(void *arg)
	{
	char *poolname = arg;

	if (poolname != NULL)
	kmem_strfree(poolname);
	}

	#ifdef ZFS_DEBUG
	static boolean_t zfs_ioc_recv_inject_err;
	#endif

	/*
	* nvlist 'errors' is always allocated. It will contain descriptions of
	* encountered errors, if any. It's the callers responsibility to free.
	*/
	static int
	zfs_ioc_recv_impl(char tofs, char tosnap, char origin, nvlist_t recvprops,
	nvlist_t localprops, nvlist_t hidden_args, boolean_t force,
	boolean_t resumable, int input_fd,
	dmu_replay_record_t begin_record, uint64_t read_bytes,
	uint64_t errflags, nvlist_t *errors)
	{
	dmu_recv_cookie_t drc;
	int error = 0;
	int props_error = 0;
	offset_t off, noff;
	nvlist_t *local_delayprops = NULL;
	nvlist_t *recv_delayprops = NULL;
	nvlist_t origprops = NULL; / existing properties */
	nvlist_t origrecvd = NULL; / existing received properties */
	boolean_t first_recvd_props = B_FALSE;
	boolean_t tofs_was_redacted;
	zfs_file_t *input_fp;

	*read_bytes = 0;
	*errflags = 0;
	*errors = fnvlist_alloc();
	off = 0;

	if ((input_fp = zfs_file_get(input_fd)) == NULL)
	return (SET_ERROR(EBADF));

	noff = off = zfs_file_off(input_fp);
	error = dmu_recv_begin(tofs, tosnap, begin_record, force,
	resumable, localprops, hidden_args, origin, &drc, input_fp,
	&off);
	if (error != 0)
	goto out;
	tofs_was_redacted = dsl_get_redacted(drc.drc_ds);

	/*
	* Set properties before we receive the stream so that they are applied
	* to the new data. Note that we must call dmu_recv_stream() if
	* dmu_recv_begin() succeeds.
	*/
	if (recvprops != NULL && !drc.drc_newfs) {
	if (spa_version(dsl_dataset_get_spa(drc.drc_ds)) >=
	SPA_VERSION_RECVD_PROPS &&
	!dsl_prop_get_hasrecvd(tofs))
	first_recvd_props = B_TRUE;

	/*
	* If new received properties are supplied, they are to
	* completely replace the existing received properties,
	* so stash away the existing ones.
	*/
	if (dsl_prop_get_received(tofs, &origrecvd) == 0) {
	nvlist_t *errlist = NULL;
	/*
	* Don't bother writing a property if its value won't
	* change (and avoid the unnecessary security checks).
	*
	* The first receive after SPA_VERSION_RECVD_PROPS is a
	* special case where we blow away all local properties
	* regardless.
	*/
	if (!first_recvd_props)
	props_reduce(recvprops, origrecvd);
	if (zfs_check_clearable(tofs, origrecvd, &errlist) != 0)
	(void) nvlist_merge(*errors, errlist, 0);
	nvlist_free(errlist);

	if (clear_received_props(tofs, origrecvd,
	first_recvd_props ? NULL : recvprops) != 0)
	*errflags \|= ZPROP_ERR_NOCLEAR;
	} else {
	*errflags \|= ZPROP_ERR_NOCLEAR;
	}
	}

	/*
	* Stash away existing properties so we can restore them on error unless
	* we're doing the first receive after SPA_VERSION_RECVD_PROPS, in which
	* case "origrecvd" will take care of that.
	*/
	if (localprops != NULL && !drc.drc_newfs && !first_recvd_props) {
	objset_t *os;
	if (dmu_objset_hold(tofs, FTAG, &os) == 0) {
	if (dsl_prop_get_all(os, &origprops) != 0) {
	*errflags \|= ZPROP_ERR_NOCLEAR;
	}
	dmu_objset_rele(os, FTAG);
	} else {
	*errflags \|= ZPROP_ERR_NOCLEAR;
	}
	}

	if (recvprops != NULL) {
	props_error = dsl_prop_set_hasrecvd(tofs);

	if (props_error == 0) {
	recv_delayprops = extract_delay_props(recvprops);
	(void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_RECEIVED,
	recvprops, *errors);
	}
	}

	if (localprops != NULL) {
	nvlist_t *oprops = fnvlist_alloc();
	nvlist_t *xprops = fnvlist_alloc();
	nvpair_t *nvp = NULL;

	while ((nvp = nvlist_next_nvpair(localprops, nvp)) != NULL) {
	if (nvpair_type(nvp) == DATA_TYPE_BOOLEAN) {
	/* -x property */
	const char *name = nvpair_name(nvp);
	zfs_prop_t prop = zfs_name_to_prop(name);
	if (prop != ZPROP_INVAL) {
	if (!zfs_prop_inheritable(prop))
	continue;
	} else if (!zfs_prop_user(name))
	continue;
	fnvlist_add_boolean(xprops, name);
	} else {
	/* -o property=value */
	fnvlist_add_nvpair(oprops, nvp);
	}
	}

	local_delayprops = extract_delay_props(oprops);
	(void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_LOCAL,
	oprops, *errors);
	(void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_INHERITED,
	xprops, *errors);

	nvlist_free(oprops);
	nvlist_free(xprops);
	}

	error = dmu_recv_stream(&drc, &off);

	if (error == 0) {
	zfsvfs_t *zfsvfs = NULL;
	zvol_state_handle_t *zv = NULL;

	if (getzfsvfs(tofs, &zfsvfs) == 0) {
	/* online recv */
	dsl_dataset_t *ds;
	int end_err;
	boolean_t stream_is_redacted = DMU_GET_FEATUREFLAGS(
	begin_record->drr_u.drr_begin.
	drr_versioninfo) & DMU_BACKUP_FEATURE_REDACTED;

	ds = dmu_objset_ds(zfsvfs->z_os);
	error = zfs_suspend_fs(zfsvfs);
	/*
	* If the suspend fails, then the recv_end will
	* likely also fail, and clean up after itself.
	*/
	end_err = dmu_recv_end(&drc, zfsvfs);
	/*
	* If the dataset was not redacted, but we received a
	* redacted stream onto it, we need to unmount the
	* dataset. Otherwise, resume the filesystem.
	*/
	if (error == 0 && !drc.drc_newfs &&
	stream_is_redacted && !tofs_was_redacted) {
	error = zfs_end_fs(zfsvfs, ds);
	} else if (error == 0) {
	error = zfs_resume_fs(zfsvfs, ds);
	}
	error = error ? error : end_err;
	zfs_vfs_rele(zfsvfs);
	} else if ((zv = zvol_suspend(tofs)) != NULL) {
	error = dmu_recv_end(&drc, zvol_tag(zv));
	zvol_resume(zv);
	} else {
	error = dmu_recv_end(&drc, NULL);
	}

	/* Set delayed properties now, after we're done receiving. */
	if (recv_delayprops != NULL && error == 0) {
	(void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_RECEIVED,
	recv_delayprops, *errors);
	}
	if (local_delayprops != NULL && error == 0) {
	(void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_LOCAL,
	local_delayprops, *errors);
	}
	}

	/*
	* Merge delayed props back in with initial props, in case
	* we're DEBUG and zfs_ioc_recv_inject_err is set (which means
	* we have to make sure clear_received_props() includes
	* the delayed properties).
	*
	* Since zfs_ioc_recv_inject_err is only in DEBUG kernels,
	* using ASSERT() will be just like a VERIFY.
	*/
	if (recv_delayprops != NULL) {
	ASSERT(nvlist_merge(recvprops, recv_delayprops, 0) == 0);
	nvlist_free(recv_delayprops);
	}
	if (local_delayprops != NULL) {
	ASSERT(nvlist_merge(localprops, local_delayprops, 0) == 0);
	nvlist_free(local_delayprops);
	}
	*read_bytes = off - noff;

	#ifdef ZFS_DEBUG
	if (zfs_ioc_recv_inject_err) {
	zfs_ioc_recv_inject_err = B_FALSE;
	error = 1;
	}
	#endif

	/*
	* On error, restore the original props.
	*/
	if (error != 0 && recvprops != NULL && !drc.drc_newfs) {
	if (clear_received_props(tofs, recvprops, NULL) != 0) {
	/*
	* We failed to clear the received properties.
	* Since we may have left a $recvd value on the
	* system, we can't clear the $hasrecvd flag.
	*/
	*errflags \|= ZPROP_ERR_NORESTORE;
	} else if (first_recvd_props) {
	dsl_prop_unset_hasrecvd(tofs);
	}

	if (origrecvd == NULL && !drc.drc_newfs) {
	/* We failed to stash the original properties. */
	*errflags \|= ZPROP_ERR_NORESTORE;
	}

	/*
	* dsl_props_set() will not convert RECEIVED to LOCAL on or
	* after SPA_VERSION_RECVD_PROPS, so we need to specify LOCAL
	* explicitly if we're restoring local properties cleared in the
	* first new-style receive.
	*/
	if (origrecvd != NULL &&
	zfs_set_prop_nvlist(tofs, (first_recvd_props ?
	ZPROP_SRC_LOCAL : ZPROP_SRC_RECEIVED),
	origrecvd, NULL) != 0) {
	/*
	* We stashed the original properties but failed to
	* restore them.
	*/
	*errflags \|= ZPROP_ERR_NORESTORE;
	}
	}
	if (error != 0 && localprops != NULL && !drc.drc_newfs &&
	!first_recvd_props) {
	nvlist_t *setprops;
	nvlist_t *inheritprops;
	nvpair_t *nvp;

	if (origprops == NULL) {
	/* We failed to stash the original properties. */
	*errflags \|= ZPROP_ERR_NORESTORE;
	goto out;
	}

	/* Restore original props */
	setprops = fnvlist_alloc();
	inheritprops = fnvlist_alloc();
	nvp = NULL;
	while ((nvp = nvlist_next_nvpair(localprops, nvp)) != NULL) {
	const char *name = nvpair_name(nvp);
	const char *source;
	nvlist_t *attrs;

	if (!nvlist_exists(origprops, name)) {
	/*
	* Property was not present or was explicitly
	* inherited before the receive, restore this.
	*/
	fnvlist_add_boolean(inheritprops, name);
	continue;
	}
	attrs = fnvlist_lookup_nvlist(origprops, name);
	source = fnvlist_lookup_string(attrs, ZPROP_SOURCE);

	/* Skip received properties */
	if (strcmp(source, ZPROP_SOURCE_VAL_RECVD) == 0)
	continue;

	if (strcmp(source, tofs) == 0) {
	/* Property was locally set */
	fnvlist_add_nvlist(setprops, name, attrs);
	} else {
	/* Property was implicitly inherited */
	fnvlist_add_boolean(inheritprops, name);
	}
	}

	if (zfs_set_prop_nvlist(tofs, ZPROP_SRC_LOCAL, setprops,
	NULL) != 0)
	*errflags \|= ZPROP_ERR_NORESTORE;
	if (zfs_set_prop_nvlist(tofs, ZPROP_SRC_INHERITED, inheritprops,
	NULL) != 0)
	*errflags \|= ZPROP_ERR_NORESTORE;

	nvlist_free(setprops);
	nvlist_free(inheritprops);
	}
	out:
	zfs_file_put(input_fp);
	nvlist_free(origrecvd);
	nvlist_free(origprops);

	if (error == 0)
	error = props_error;

	return (error);
	}

	/*
	* inputs:
	* zc_name name of containing filesystem (unused)
	* zc_nvlist_src{_size} nvlist of properties to apply
	* zc_nvlist_conf{_size} nvlist of properties to exclude
	* (DATA_TYPE_BOOLEAN) and override (everything else)
	* zc_value name of snapshot to create
	* zc_string name of clone origin (if DRR_FLAG_CLONE)
	* zc_cookie file descriptor to recv from
	* zc_begin_record the BEGIN record of the stream (not byteswapped)
	* zc_guid force flag
	*
	* outputs:
	* zc_cookie number of bytes read
	* zc_obj zprop_errflags_t
	* zc_nvlist_dst{_size} error for each unapplied received property
	*/
	static int
	zfs_ioc_recv(zfs_cmd_t *zc)
	{
	dmu_replay_record_t begin_record;
	nvlist_t *errors = NULL;
	nvlist_t *recvdprops = NULL;
	nvlist_t *localprops = NULL;
	char *origin = NULL;
	char *tosnap;
	char tofs[ZFS_MAX_DATASET_NAME_LEN];
	int error = 0;

	if (dataset_namecheck(zc->zc_value, NULL, NULL) != 0 \|\|
	strchr(zc->zc_value, '@') == NULL \|\|
	strchr(zc->zc_value, '%'))
	return (SET_ERROR(EINVAL));

	(void) strlcpy(tofs, zc->zc_value, sizeof (tofs));
	tosnap = strchr(tofs, '@');
	*tosnap++ = '\0';

	if (zc->zc_nvlist_src != 0 &&
	(error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size,
	zc->zc_iflags, &recvdprops)) != 0)
	return (error);

	if (zc->zc_nvlist_conf != 0 &&
	(error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size,
	zc->zc_iflags, &localprops)) != 0)
	return (error);

	if (zc->zc_string[0])
	origin = zc->zc_string;

	begin_record.drr_type = DRR_BEGIN;
	begin_record.drr_payloadlen = 0;
	begin_record.drr_u.drr_begin = zc->zc_begin_record;

	error = zfs_ioc_recv_impl(tofs, tosnap, origin, recvdprops, localprops,
	NULL, zc->zc_guid, B_FALSE, zc->zc_cookie, &begin_record,
	&zc->zc_cookie, &zc->zc_obj, &errors);
	nvlist_free(recvdprops);
	nvlist_free(localprops);

	/*
	* Now that all props, initial and delayed, are set, report the prop
	* errors to the caller.
	*/
	if (zc->zc_nvlist_dst_size != 0 && errors != NULL &&
	(nvlist_smush(errors, zc->zc_nvlist_dst_size) != 0 \|\|
	put_nvlist(zc, errors) != 0)) {
	/*
	* Caller made zc->zc_nvlist_dst less than the minimum expected
	* size or supplied an invalid address.
	*/
	error = SET_ERROR(EINVAL);
	}

	nvlist_free(errors);

	return (error);
	}

	/*
	* innvl: {
	* "snapname" -> full name of the snapshot to create
	* (optional) "props" -> received properties to set (nvlist)
	* (optional) "localprops" -> override and exclude properties (nvlist)
	* (optional) "origin" -> name of clone origin (DRR_FLAG_CLONE)
	* "begin_record" -> non-byteswapped dmu_replay_record_t
	* "input_fd" -> file descriptor to read stream from (int32)
	* (optional) "force" -> force flag (value ignored)
	* (optional) "resumable" -> resumable flag (value ignored)
	* (optional) "cleanup_fd" -> unused
	* (optional) "action_handle" -> unused
	* (optional) "hidden_args" -> { "wkeydata" -> value }
	* }
	*
	* outnvl: {
	* "read_bytes" -> number of bytes read
	* "error_flags" -> zprop_errflags_t
	* "errors" -> error for each unapplied received property (nvlist)
	* }
	*/
	static const zfs_ioc_key_t zfs_keys_recv_new[] = {
	{"snapname", DATA_TYPE_STRING, 0},
	{"props", DATA_TYPE_NVLIST, ZK_OPTIONAL},
	{"localprops", DATA_TYPE_NVLIST, ZK_OPTIONAL},
	{"origin", DATA_TYPE_STRING, ZK_OPTIONAL},
	{"begin_record", DATA_TYPE_BYTE_ARRAY, 0},
	{"input_fd", DATA_TYPE_INT32, 0},
	{"force", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	{"resumable", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	{"cleanup_fd", DATA_TYPE_INT32, ZK_OPTIONAL},
	{"action_handle", DATA_TYPE_UINT64, ZK_OPTIONAL},
	{"hidden_args", DATA_TYPE_NVLIST, ZK_OPTIONAL},
	};

	static int
	zfs_ioc_recv_new(const char fsname, nvlist_t innvl, nvlist_t *outnvl)
	{
	dmu_replay_record_t *begin_record;
	uint_t begin_record_size;
	nvlist_t *errors = NULL;
	nvlist_t *recvprops = NULL;
	nvlist_t *localprops = NULL;
	nvlist_t *hidden_args = NULL;
	char *snapname;
	char *origin = NULL;
	char *tosnap;
	char tofs[ZFS_MAX_DATASET_NAME_LEN];
	boolean_t force;
	boolean_t resumable;
	uint64_t read_bytes = 0;
	uint64_t errflags = 0;
	int input_fd = -1;
	int error;

	snapname = fnvlist_lookup_string(innvl, "snapname");

	if (dataset_namecheck(snapname, NULL, NULL) != 0 \|\|
	strchr(snapname, '@') == NULL \|\|
	strchr(snapname, '%'))
	return (SET_ERROR(EINVAL));

	(void) strlcpy(tofs, snapname, sizeof (tofs));
	tosnap = strchr(tofs, '@');
	*tosnap++ = '\0';

	error = nvlist_lookup_string(innvl, "origin", &origin);
	if (error && error != ENOENT)
	return (error);

	error = nvlist_lookup_byte_array(innvl, "begin_record",
	(uchar_t **)&begin_record, &begin_record_size);
	if (error != 0 \|\| begin_record_size != sizeof (*begin_record))
	return (SET_ERROR(EINVAL));

	input_fd = fnvlist_lookup_int32(innvl, "input_fd");

	force = nvlist_exists(innvl, "force");
	resumable = nvlist_exists(innvl, "resumable");

	/* we still use "props" here for backwards compatibility */
	error = nvlist_lookup_nvlist(innvl, "props", &recvprops);
	if (error && error != ENOENT)
	return (error);

	error = nvlist_lookup_nvlist(innvl, "localprops", &localprops);
	if (error && error != ENOENT)
	return (error);

	error = nvlist_lookup_nvlist(innvl, ZPOOL_HIDDEN_ARGS, &hidden_args);
	if (error && error != ENOENT)
	return (error);

	error = zfs_ioc_recv_impl(tofs, tosnap, origin, recvprops, localprops,
	hidden_args, force, resumable, input_fd, begin_record,
	&read_bytes, &errflags, &errors);

	fnvlist_add_uint64(outnvl, "read_bytes", read_bytes);
	fnvlist_add_uint64(outnvl, "error_flags", errflags);
	fnvlist_add_nvlist(outnvl, "errors", errors);

	nvlist_free(errors);
	nvlist_free(recvprops);
	nvlist_free(localprops);

	return (error);
	}

	typedef struct dump_bytes_io {
	zfs_file_t *dbi_fp;
	caddr_t dbi_buf;
	int dbi_len;
	int dbi_err;
	} dump_bytes_io_t;

	static void
	dump_bytes_cb(void *arg)
	{
	dump_bytes_io_t dbi = (dump_bytes_io_t )arg;
	zfs_file_t *fp;
	caddr_t buf;

	fp = dbi->dbi_fp;
	buf = dbi->dbi_buf;

	dbi->dbi_err = zfs_file_write(fp, buf, dbi->dbi_len, NULL);
	}

	static int
	dump_bytes(objset_t os, void buf, int len, void *arg)
	{
	dump_bytes_io_t dbi;

	dbi.dbi_fp = arg;
	dbi.dbi_buf = buf;
	dbi.dbi_len = len;

	#if defined(HAVE_LARGE_STACKS)
	dump_bytes_cb(&dbi);
	#else
	/*
	* The vn_rdwr() call is performed in a taskq to ensure that there is
	* always enough stack space to write safely to the target filesystem.
	* The ZIO_TYPE_FREE threads are used because there can be a lot of
	* them and they are used in vdev_file.c for a similar purpose.
	*/
	spa_taskq_dispatch_sync(dmu_objset_spa(os), ZIO_TYPE_FREE,
	ZIO_TASKQ_ISSUE, dump_bytes_cb, &dbi, TQ_SLEEP);
	#endif /* HAVE_LARGE_STACKS */

	return (dbi.dbi_err);
	}

	/*
	* inputs:
	* zc_name name of snapshot to send
	* zc_cookie file descriptor to send stream to
	* zc_obj fromorigin flag (mutually exclusive with zc_fromobj)
	* zc_sendobj objsetid of snapshot to send
	* zc_fromobj objsetid of incremental fromsnap (may be zero)
	* zc_guid if set, estimate size of stream only. zc_cookie is ignored.
	* output size in zc_objset_type.
	* zc_flags lzc_send_flags
	*
	* outputs:
	* zc_objset_type estimated size, if zc_guid is set
	*
	* NOTE: This is no longer the preferred interface, any new functionality
	* should be added to zfs_ioc_send_new() instead.
	*/
	static int
	zfs_ioc_send(zfs_cmd_t *zc)
	{
	int error;
	offset_t off;
	boolean_t estimate = (zc->zc_guid != 0);
	boolean_t embedok = (zc->zc_flags & 0x1);
	boolean_t large_block_ok = (zc->zc_flags & 0x2);
	boolean_t compressok = (zc->zc_flags & 0x4);
	boolean_t rawok = (zc->zc_flags & 0x8);
	boolean_t savedok = (zc->zc_flags & 0x10);

	if (zc->zc_obj != 0) {
	dsl_pool_t *dp;
	dsl_dataset_t *tosnap;

	error = dsl_pool_hold(zc->zc_name, FTAG, &dp);
	if (error != 0)
	return (error);

	error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &tosnap);
	if (error != 0) {
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	if (dsl_dir_is_clone(tosnap->ds_dir))
	zc->zc_fromobj =
	dsl_dir_phys(tosnap->ds_dir)->dd_origin_obj;
	dsl_dataset_rele(tosnap, FTAG);
	dsl_pool_rele(dp, FTAG);
	}

	if (estimate) {
	dsl_pool_t *dp;
	dsl_dataset_t *tosnap;
	dsl_dataset_t *fromsnap = NULL;

	error = dsl_pool_hold(zc->zc_name, FTAG, &dp);
	if (error != 0)
	return (error);

	error = dsl_dataset_hold_obj(dp, zc->zc_sendobj,
	FTAG, &tosnap);
	if (error != 0) {
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	if (zc->zc_fromobj != 0) {
	error = dsl_dataset_hold_obj(dp, zc->zc_fromobj,
	FTAG, &fromsnap);
	if (error != 0) {
	dsl_dataset_rele(tosnap, FTAG);
	dsl_pool_rele(dp, FTAG);
	return (error);
	}
	}

	error = dmu_send_estimate_fast(tosnap, fromsnap, NULL,
	compressok \|\| rawok, savedok, &zc->zc_objset_type);

	if (fromsnap != NULL)
	dsl_dataset_rele(fromsnap, FTAG);
	dsl_dataset_rele(tosnap, FTAG);
	dsl_pool_rele(dp, FTAG);
	} else {
	zfs_file_t *fp;
	dmu_send_outparams_t out = {0};

	if ((fp = zfs_file_get(zc->zc_cookie)) == NULL)
	return (SET_ERROR(EBADF));

	off = zfs_file_off(fp);
	out.dso_outfunc = dump_bytes;
	out.dso_arg = fp;
	out.dso_dryrun = B_FALSE;
	error = dmu_send_obj(zc->zc_name, zc->zc_sendobj,
	zc->zc_fromobj, embedok, large_block_ok, compressok,
	rawok, savedok, zc->zc_cookie, &off, &out);

	zfs_file_put(fp);
	}
	return (error);
	}

	/*
	* inputs:
	* zc_name name of snapshot on which to report progress
	* zc_cookie file descriptor of send stream
	*
	* outputs:
	* zc_cookie number of bytes written in send stream thus far
	* zc_objset_type logical size of data traversed by send thus far
	*/
	static int
	zfs_ioc_send_progress(zfs_cmd_t *zc)
	{
	dsl_pool_t *dp;
	dsl_dataset_t *ds;
	dmu_sendstatus_t *dsp = NULL;
	int error;

	error = dsl_pool_hold(zc->zc_name, FTAG, &dp);
	if (error != 0)
	return (error);

	error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &ds);
	if (error != 0) {
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	mutex_enter(&ds->ds_sendstream_lock);

	/*
	* Iterate over all the send streams currently active on this dataset.
	* If there's one which matches the specified file descriptor _and_ the
	* stream was started by the current process, return the progress of
	* that stream.
	*/

	for (dsp = list_head(&ds->ds_sendstreams); dsp != NULL;
	dsp = list_next(&ds->ds_sendstreams, dsp)) {
	if (dsp->dss_outfd == zc->zc_cookie &&
	zfs_proc_is_caller(dsp->dss_proc))
	break;
	}

	if (dsp != NULL) {
	zc->zc_cookie = atomic_cas_64((volatile uint64_t *)dsp->dss_off,
	0, 0);
	/* This is the closest thing we have to atomic_read_64. */
	zc->zc_objset_type = atomic_cas_64(&dsp->dss_blocks, 0, 0);
	} else {
	error = SET_ERROR(ENOENT);
	}

	mutex_exit(&ds->ds_sendstream_lock);
	dsl_dataset_rele(ds, FTAG);
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	static int
	zfs_ioc_inject_fault(zfs_cmd_t *zc)
	{
	int id, error;

	error = zio_inject_fault(zc->zc_name, (int)zc->zc_guid, &id,
	&zc->zc_inject_record);

	if (error == 0)
	zc->zc_guid = (uint64_t)id;

	return (error);
	}

	static int
	zfs_ioc_clear_fault(zfs_cmd_t *zc)
	{
	return (zio_clear_fault((int)zc->zc_guid));
	}

	static int
	zfs_ioc_inject_list_next(zfs_cmd_t *zc)
	{
	int id = (int)zc->zc_guid;
	int error;

	error = zio_inject_list_next(&id, zc->zc_name, sizeof (zc->zc_name),
	&zc->zc_inject_record);

	zc->zc_guid = id;

	return (error);
	}

	static int
	zfs_ioc_error_log(zfs_cmd_t *zc)
	{
	spa_t *spa;
	int error;
	size_t count = (size_t)zc->zc_nvlist_dst_size;

	if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0)
	return (error);

	error = spa_get_errlog(spa, (void *)(uintptr_t)zc->zc_nvlist_dst,
	&count);
	if (error == 0)
	zc->zc_nvlist_dst_size = count;
	else
	zc->zc_nvlist_dst_size = spa_get_errlog_size(spa);

	spa_close(spa, FTAG);

	return (error);
	}

	static int
	zfs_ioc_clear(zfs_cmd_t *zc)
	{
	spa_t *spa;
	vdev_t *vd;
	int error;

	/*
	* On zpool clear we also fix up missing slogs
	*/
	mutex_enter(&spa_namespace_lock);
	spa = spa_lookup(zc->zc_name);
	if (spa == NULL) {
	mutex_exit(&spa_namespace_lock);
	return (SET_ERROR(EIO));
	}
	if (spa_get_log_state(spa) == SPA_LOG_MISSING) {
	/* we need to let spa_open/spa_load clear the chains */
	spa_set_log_state(spa, SPA_LOG_CLEAR);
	}
	spa->spa_last_open_failed = 0;
	mutex_exit(&spa_namespace_lock);

	if (zc->zc_cookie & ZPOOL_NO_REWIND) {
	error = spa_open(zc->zc_name, &spa, FTAG);
	} else {
	nvlist_t *policy;
	nvlist_t *config = NULL;

	if (zc->zc_nvlist_src == 0)
	return (SET_ERROR(EINVAL));

	if ((error = get_nvlist(zc->zc_nvlist_src,
	zc->zc_nvlist_src_size, zc->zc_iflags, &policy)) == 0) {
	error = spa_open_rewind(zc->zc_name, &spa, FTAG,
	policy, &config);
	if (config != NULL) {
	int err;

	if ((err = put_nvlist(zc, config)) != 0)
	error = err;
	nvlist_free(config);
	}
	nvlist_free(policy);
	}
	}

	if (error != 0)
	return (error);

	/*
	* If multihost is enabled, resuming I/O is unsafe as another
	* host may have imported the pool.
	*/
	if (spa_multihost(spa) && spa_suspended(spa))
	return (SET_ERROR(EINVAL));

	spa_vdev_state_enter(spa, SCL_NONE);

	if (zc->zc_guid == 0) {
	vd = NULL;
	} else {
	vd = spa_lookup_by_guid(spa, zc->zc_guid, B_TRUE);
	if (vd == NULL) {
	error = SET_ERROR(ENODEV);
	(void) spa_vdev_state_exit(spa, NULL, error);
	spa_close(spa, FTAG);
	return (error);
	}
	}

	vdev_clear(spa, vd);

	(void) spa_vdev_state_exit(spa, spa_suspended(spa) ?
	NULL : spa->spa_root_vdev, 0);

	/*
	* Resume any suspended I/Os.
	*/
	if (zio_resume(spa) != 0)
	error = SET_ERROR(EIO);

	spa_close(spa, FTAG);

	return (error);
	}

	/*
	* Reopen all the vdevs associated with the pool.
	*
	* innvl: {
	* "scrub_restart" -> when true and scrub is running, allow to restart
	* scrub as the side effect of the reopen (boolean).
	* }
	*
	* outnvl is unused
	*/
	static const zfs_ioc_key_t zfs_keys_pool_reopen[] = {
	{"scrub_restart", DATA_TYPE_BOOLEAN_VALUE, ZK_OPTIONAL},
	};

	/* ARGSUSED */
	static int
	zfs_ioc_pool_reopen(const char pool, nvlist_t innvl, nvlist_t *outnvl)
	{
	spa_t *spa;
	int error;
	boolean_t rc, scrub_restart = B_TRUE;

	if (innvl) {
	error = nvlist_lookup_boolean_value(innvl,
	"scrub_restart", &rc);
	if (error == 0)
	scrub_restart = rc;
	}

	error = spa_open(pool, &spa, FTAG);
	if (error != 0)
	return (error);

	spa_vdev_state_enter(spa, SCL_NONE);

	/*
	* If the scrub_restart flag is B_FALSE and a scrub is already
	* in progress then set spa_scrub_reopen flag to B_TRUE so that
	* we don't restart the scrub as a side effect of the reopen.
	* Otherwise, let vdev_open() decided if a resilver is required.
	*/

	spa->spa_scrub_reopen = (!scrub_restart &&
	dsl_scan_scrubbing(spa->spa_dsl_pool));
	vdev_reopen(spa->spa_root_vdev);
	spa->spa_scrub_reopen = B_FALSE;

	(void) spa_vdev_state_exit(spa, NULL, 0);
	spa_close(spa, FTAG);
	return (0);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	*
	* outputs:
	* zc_string name of conflicting snapshot, if there is one
	*/
	static int
	zfs_ioc_promote(zfs_cmd_t *zc)
	{
	dsl_pool_t *dp;
	dsl_dataset_t ds, ods;
	char origin[ZFS_MAX_DATASET_NAME_LEN];
	char *cp;
	int error;

	zc->zc_name[sizeof (zc->zc_name) - 1] = '\0';
	if (dataset_namecheck(zc->zc_name, NULL, NULL) != 0 \|\|
	strchr(zc->zc_name, '%'))
	return (SET_ERROR(EINVAL));

	error = dsl_pool_hold(zc->zc_name, FTAG, &dp);
	if (error != 0)
	return (error);

	error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &ds);
	if (error != 0) {
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	if (!dsl_dir_is_clone(ds->ds_dir)) {
	dsl_dataset_rele(ds, FTAG);
	dsl_pool_rele(dp, FTAG);
	return (SET_ERROR(EINVAL));
	}

	error = dsl_dataset_hold_obj(dp,
	dsl_dir_phys(ds->ds_dir)->dd_origin_obj, FTAG, &ods);
	if (error != 0) {
	dsl_dataset_rele(ds, FTAG);
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	dsl_dataset_name(ods, origin);
	dsl_dataset_rele(ods, FTAG);
	dsl_dataset_rele(ds, FTAG);
	dsl_pool_rele(dp, FTAG);

	/*
	* We don't need to unmount all the origin fs's snapshots, but
	* it's easier.
	*/
	cp = strchr(origin, '@');
	if (cp)
	*cp = '\0';
	(void) dmu_objset_find(origin,
	zfs_unmount_snap_cb, NULL, DS_FIND_SNAPSHOTS);
	return (dsl_dataset_promote(zc->zc_name, zc->zc_string));
	}

	/*
	* Retrieve a single {user\|group\|project}{used\|quota}@... property.
	*
	* inputs:
	* zc_name name of filesystem
	* zc_objset_type zfs_userquota_prop_t
	* zc_value domain name (eg. "S-1-234-567-89")
	* zc_guid RID/UID/GID
	*
	* outputs:
	* zc_cookie property value
	*/
	static int
	zfs_ioc_userspace_one(zfs_cmd_t *zc)
	{
	zfsvfs_t *zfsvfs;
	int error;

	if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS)
	return (SET_ERROR(EINVAL));

	error = zfsvfs_hold(zc->zc_name, FTAG, &zfsvfs, B_FALSE);
	if (error != 0)
	return (error);

	error = zfs_userspace_one(zfsvfs,
	zc->zc_objset_type, zc->zc_value, zc->zc_guid, &zc->zc_cookie);
	zfsvfs_rele(zfsvfs, FTAG);

	return (error);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	* zc_cookie zap cursor
	* zc_objset_type zfs_userquota_prop_t
	* zc_nvlist_dst[_size] buffer to fill (not really an nvlist)
	*
	* outputs:
	* zc_nvlist_dst[_size] data buffer (array of zfs_useracct_t)
	* zc_cookie zap cursor
	*/
	static int
	zfs_ioc_userspace_many(zfs_cmd_t *zc)
	{
	zfsvfs_t *zfsvfs;
	int bufsize = zc->zc_nvlist_dst_size;

	if (bufsize <= 0)
	return (SET_ERROR(ENOMEM));

	int error = zfsvfs_hold(zc->zc_name, FTAG, &zfsvfs, B_FALSE);
	if (error != 0)
	return (error);

	void *buf = vmem_alloc(bufsize, KM_SLEEP);

	error = zfs_userspace_many(zfsvfs, zc->zc_objset_type, &zc->zc_cookie,
	buf, &zc->zc_nvlist_dst_size);

	if (error == 0) {
	error = xcopyout(buf,
	(void *)(uintptr_t)zc->zc_nvlist_dst,
	zc->zc_nvlist_dst_size);
	}
	vmem_free(buf, bufsize);
	zfsvfs_rele(zfsvfs, FTAG);

	return (error);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	*
	* outputs:
	* none
	*/
	static int
	zfs_ioc_userspace_upgrade(zfs_cmd_t *zc)
	{
	int error = 0;
	zfsvfs_t *zfsvfs;

	if (getzfsvfs(zc->zc_name, &zfsvfs) == 0) {
	if (!dmu_objset_userused_enabled(zfsvfs->z_os)) {
	/*
	* If userused is not enabled, it may be because the
	* objset needs to be closed & reopened (to grow the
	* objset_phys_t). Suspend/resume the fs will do that.
	*/
	dsl_dataset_t ds, newds;

	ds = dmu_objset_ds(zfsvfs->z_os);
	error = zfs_suspend_fs(zfsvfs);
	if (error == 0) {
	dmu_objset_refresh_ownership(ds, &newds,
	B_TRUE, zfsvfs);
	error = zfs_resume_fs(zfsvfs, newds);
	}
	}
	if (error == 0) {
	mutex_enter(&zfsvfs->z_os->os_upgrade_lock);
	if (zfsvfs->z_os->os_upgrade_id == 0) {
	/* clear potential error code and retry */
	zfsvfs->z_os->os_upgrade_status = 0;
	mutex_exit(&zfsvfs->z_os->os_upgrade_lock);

	dsl_pool_config_enter(
	dmu_objset_pool(zfsvfs->z_os), FTAG);
	dmu_objset_userspace_upgrade(zfsvfs->z_os);
	dsl_pool_config_exit(
	dmu_objset_pool(zfsvfs->z_os), FTAG);
	} else {
	mutex_exit(&zfsvfs->z_os->os_upgrade_lock);
	}

	taskq_wait_id(zfsvfs->z_os->os_spa->spa_upgrade_taskq,
	zfsvfs->z_os->os_upgrade_id);
	error = zfsvfs->z_os->os_upgrade_status;
	}
	zfs_vfs_rele(zfsvfs);
	} else {
	objset_t *os;

	/* XXX kind of reading contents without owning */
	error = dmu_objset_hold_flags(zc->zc_name, B_TRUE, FTAG, &os);
	if (error != 0)
	return (error);

	mutex_enter(&os->os_upgrade_lock);
	if (os->os_upgrade_id == 0) {
	/* clear potential error code and retry */
	os->os_upgrade_status = 0;
	mutex_exit(&os->os_upgrade_lock);

	dmu_objset_userspace_upgrade(os);
	} else {
	mutex_exit(&os->os_upgrade_lock);
	}

	dsl_pool_rele(dmu_objset_pool(os), FTAG);

	taskq_wait_id(os->os_spa->spa_upgrade_taskq, os->os_upgrade_id);
	error = os->os_upgrade_status;

	dsl_dataset_rele_flags(dmu_objset_ds(os), DS_HOLD_FLAG_DECRYPT,
	FTAG);
	}
	return (error);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	*
	* outputs:
	* none
	*/
	static int
	zfs_ioc_id_quota_upgrade(zfs_cmd_t *zc)
	{
	objset_t *os;
	int error;

	error = dmu_objset_hold_flags(zc->zc_name, B_TRUE, FTAG, &os);
	if (error != 0)
	return (error);

	if (dmu_objset_userobjspace_upgradable(os) \|\|
	dmu_objset_projectquota_upgradable(os)) {
	mutex_enter(&os->os_upgrade_lock);
	if (os->os_upgrade_id == 0) {
	/* clear potential error code and retry */
	os->os_upgrade_status = 0;
	mutex_exit(&os->os_upgrade_lock);

	dmu_objset_id_quota_upgrade(os);
	} else {
	mutex_exit(&os->os_upgrade_lock);
	}

	dsl_pool_rele(dmu_objset_pool(os), FTAG);

	taskq_wait_id(os->os_spa->spa_upgrade_taskq, os->os_upgrade_id);
	error = os->os_upgrade_status;
	} else {
	dsl_pool_rele(dmu_objset_pool(os), FTAG);
	}

	dsl_dataset_rele_flags(dmu_objset_ds(os), DS_HOLD_FLAG_DECRYPT, FTAG);

	return (error);
	}

	static int
	zfs_ioc_share(zfs_cmd_t *zc)
	{
	return (SET_ERROR(ENOSYS));
	}

	ace_t full_access[] = {
	{(uid_t)-1, ACE_ALL_PERMS, ACE_EVERYONE, 0}
	};

	/*
	* inputs:
	* zc_name name of containing filesystem
	* zc_obj object # beyond which we want next in-use object #
	*
	* outputs:
	* zc_obj next in-use object #
	*/
	static int
	zfs_ioc_next_obj(zfs_cmd_t *zc)
	{
	objset_t *os = NULL;
	int error;

	error = dmu_objset_hold(zc->zc_name, FTAG, &os);
	if (error != 0)
	return (error);

	error = dmu_object_next(os, &zc->zc_obj, B_FALSE, 0);

	dmu_objset_rele(os, FTAG);
	return (error);
	}

	/*
	* inputs:
	* zc_name name of filesystem
	* zc_value prefix name for snapshot
	* zc_cleanup_fd cleanup-on-exit file descriptor for calling process
	*
	* outputs:
	* zc_value short name of new snapshot
	*/
	static int
	zfs_ioc_tmp_snapshot(zfs_cmd_t *zc)
	{
	char *snap_name;
	char *hold_name;
	minor_t minor;

	zfs_file_t *fp = zfs_onexit_fd_hold(zc->zc_cleanup_fd, &minor);
	if (fp == NULL)
	return (SET_ERROR(EBADF));

	snap_name = kmem_asprintf("%s-%016llx", zc->zc_value,
	(u_longlong_t)ddi_get_lbolt64());
	hold_name = kmem_asprintf("%%%s", zc->zc_value);

	int error = dsl_dataset_snapshot_tmp(zc->zc_name, snap_name, minor,
	hold_name);
	if (error == 0)
	(void) strlcpy(zc->zc_value, snap_name,
	sizeof (zc->zc_value));
	kmem_strfree(snap_name);
	kmem_strfree(hold_name);
	zfs_onexit_fd_rele(fp);
	return (error);
	}

	/*
	* inputs:
	* zc_name name of "to" snapshot
	* zc_value name of "from" snapshot
	* zc_cookie file descriptor to write diff data on
	*
	* outputs:
	* dmu_diff_record_t's to the file descriptor
	*/
	static int
	zfs_ioc_diff(zfs_cmd_t *zc)
	{
	zfs_file_t *fp;
	offset_t off;
	int error;

	if ((fp = zfs_file_get(zc->zc_cookie)) == NULL)
	return (SET_ERROR(EBADF));

	off = zfs_file_off(fp);
	error = dmu_diff(zc->zc_name, zc->zc_value, fp, &off);

	zfs_file_put(fp);

	return (error);
	}

	static int
	zfs_ioc_smb_acl(zfs_cmd_t *zc)
	{
	return (SET_ERROR(ENOTSUP));
	}

	/*
	* innvl: {
	* "holds" -> { snapname -> holdname (string), ... }
	* (optional) "cleanup_fd" -> fd (int32)
	* }
	*
	* outnvl: {
	* snapname -> error value (int32)
	* ...
	* }
	*/
	static const zfs_ioc_key_t zfs_keys_hold[] = {
	{"holds", DATA_TYPE_NVLIST, 0},
	{"cleanup_fd", DATA_TYPE_INT32, ZK_OPTIONAL},
	};

	/* ARGSUSED */
	static int
	zfs_ioc_hold(const char pool, nvlist_t args, nvlist_t *errlist)
	{
	nvpair_t *pair;
	nvlist_t *holds;
	int cleanup_fd = -1;
	int error;
	minor_t minor = 0;
	zfs_file_t *fp = NULL;

	holds = fnvlist_lookup_nvlist(args, "holds");

	/* make sure the user didn't pass us any invalid (empty) tags */
	for (pair = nvlist_next_nvpair(holds, NULL); pair != NULL;
	pair = nvlist_next_nvpair(holds, pair)) {
	char *htag;

	error = nvpair_value_string(pair, &htag);
	if (error != 0)
	return (SET_ERROR(error));

	if (strlen(htag) == 0)
	return (SET_ERROR(EINVAL));
	}

	if (nvlist_lookup_int32(args, "cleanup_fd", &cleanup_fd) == 0) {
	fp = zfs_onexit_fd_hold(cleanup_fd, &minor);
	if (fp == NULL)
	return (SET_ERROR(EBADF));
	}

	error = dsl_dataset_user_hold(holds, minor, errlist);
	if (fp != NULL) {
	ASSERT3U(minor, !=, 0);
	zfs_onexit_fd_rele(fp);
	}
	return (SET_ERROR(error));
	}

	/*
	* innvl is not used.
	*
	* outnvl: {
	* holdname -> time added (uint64 seconds since epoch)
	* ...
	* }
	*/
	static const zfs_ioc_key_t zfs_keys_get_holds[] = {
	/* no nvl keys */
	};

	/* ARGSUSED */
	static int
	zfs_ioc_get_holds(const char snapname, nvlist_t args, nvlist_t *outnvl)
	{
	return (dsl_dataset_get_holds(snapname, outnvl));
	}

	/*
	* innvl: {
	* snapname -> { holdname, ... }
	* ...
	* }
	*
	* outnvl: {
	* snapname -> error value (int32)
	* ...
	* }
	*/
	static const zfs_ioc_key_t zfs_keys_release[] = {
	{"<snapname>...", DATA_TYPE_NVLIST, ZK_WILDCARDLIST},
	};

	/* ARGSUSED */
	static int
	zfs_ioc_release(const char pool, nvlist_t holds, nvlist_t *errlist)
	{
	return (dsl_dataset_user_release(holds, errlist));
	}

	/*
	* inputs:
	* zc_guid flags (ZEVENT_NONBLOCK)
	* zc_cleanup_fd zevent file descriptor
	*
	* outputs:
	* zc_nvlist_dst next nvlist event
	* zc_cookie dropped events since last get
	*/
	static int
	zfs_ioc_events_next(zfs_cmd_t *zc)
	{
	zfs_zevent_t *ze;
	nvlist_t *event = NULL;
	minor_t minor;
	uint64_t dropped = 0;
	int error;

	zfs_file_t *fp = zfs_zevent_fd_hold(zc->zc_cleanup_fd, &minor, &ze);
	if (fp == NULL)
	return (SET_ERROR(EBADF));

	do {
	error = zfs_zevent_next(ze, &event,
	&zc->zc_nvlist_dst_size, &dropped);
	if (event != NULL) {
	zc->zc_cookie = dropped;
	error = put_nvlist(zc, event);
	nvlist_free(event);
	}

	if (zc->zc_guid & ZEVENT_NONBLOCK)
	break;

	if ((error == 0) \|\| (error != ENOENT))
	break;

	error = zfs_zevent_wait(ze);
	if (error != 0)
	break;
	} while (1);

	zfs_zevent_fd_rele(fp);

	return (error);
	}

	/*
	* outputs:
	* zc_cookie cleared events count
	*/
	static int
	zfs_ioc_events_clear(zfs_cmd_t *zc)
	{
	int count;

	zfs_zevent_drain_all(&count);
	zc->zc_cookie = count;

	return (0);
	}

	/*
	* inputs:
	* zc_guid eid \| ZEVENT_SEEK_START \| ZEVENT_SEEK_END
	* zc_cleanup zevent file descriptor
	*/
	static int
	zfs_ioc_events_seek(zfs_cmd_t *zc)
	{
	zfs_zevent_t *ze;
	minor_t minor;
	int error;

	zfs_file_t *fp = zfs_zevent_fd_hold(zc->zc_cleanup_fd, &minor, &ze);
	if (fp == NULL)
	return (SET_ERROR(EBADF));

	error = zfs_zevent_seek(ze, zc->zc_guid);
	zfs_zevent_fd_rele(fp);

	return (error);
	}

	/*
	* inputs:
	* zc_name name of later filesystem or snapshot
	* zc_value full name of old snapshot or bookmark
	*
	* outputs:
	* zc_cookie space in bytes
	* zc_objset_type compressed space in bytes
	* zc_perm_action uncompressed space in bytes
	*/
	static int
	zfs_ioc_space_written(zfs_cmd_t *zc)
	{
	int error;
	dsl_pool_t *dp;
	dsl_dataset_t *new;

	error = dsl_pool_hold(zc->zc_name, FTAG, &dp);
	if (error != 0)
	return (error);
	error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &new);
	if (error != 0) {
	dsl_pool_rele(dp, FTAG);
	return (error);
	}
	if (strchr(zc->zc_value, '#') != NULL) {
	zfs_bookmark_phys_t bmp;
	error = dsl_bookmark_lookup(dp, zc->zc_value,
	new, &bmp);
	if (error == 0) {
	error = dsl_dataset_space_written_bookmark(&bmp, new,
	&zc->zc_cookie,
	&zc->zc_objset_type, &zc->zc_perm_action);
	}
	} else {
	dsl_dataset_t *old;
	error = dsl_dataset_hold(dp, zc->zc_value, FTAG, &old);

	if (error == 0) {
	error = dsl_dataset_space_written(old, new,
	&zc->zc_cookie,
	&zc->zc_objset_type, &zc->zc_perm_action);
	dsl_dataset_rele(old, FTAG);
	}
	}
	dsl_dataset_rele(new, FTAG);
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	/*
	* innvl: {
	* "firstsnap" -> snapshot name
	* }
	*
	* outnvl: {
	* "used" -> space in bytes
	* "compressed" -> compressed space in bytes
	* "uncompressed" -> uncompressed space in bytes
	* }
	*/
	static const zfs_ioc_key_t zfs_keys_space_snaps[] = {
	{"firstsnap", DATA_TYPE_STRING, 0},
	};

	static int
	zfs_ioc_space_snaps(const char lastsnap, nvlist_t innvl, nvlist_t *outnvl)
	{
	int error;
	dsl_pool_t *dp;
	dsl_dataset_t new, old;
	char *firstsnap;
	uint64_t used, comp, uncomp;

	firstsnap = fnvlist_lookup_string(innvl, "firstsnap");

	error = dsl_pool_hold(lastsnap, FTAG, &dp);
	if (error != 0)
	return (error);

	error = dsl_dataset_hold(dp, lastsnap, FTAG, &new);
	if (error == 0 && !new->ds_is_snapshot) {
	dsl_dataset_rele(new, FTAG);
	error = SET_ERROR(EINVAL);
	}
	if (error != 0) {
	dsl_pool_rele(dp, FTAG);
	return (error);
	}
	error = dsl_dataset_hold(dp, firstsnap, FTAG, &old);
	if (error == 0 && !old->ds_is_snapshot) {
	dsl_dataset_rele(old, FTAG);
	error = SET_ERROR(EINVAL);
	}
	if (error != 0) {
	dsl_dataset_rele(new, FTAG);
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	error = dsl_dataset_space_wouldfree(old, new, &used, &comp, &uncomp);
	dsl_dataset_rele(old, FTAG);
	dsl_dataset_rele(new, FTAG);
	dsl_pool_rele(dp, FTAG);
	fnvlist_add_uint64(outnvl, "used", used);
	fnvlist_add_uint64(outnvl, "compressed", comp);
	fnvlist_add_uint64(outnvl, "uncompressed", uncomp);
	return (error);
	}

	/*
	* innvl: {
	* "fd" -> file descriptor to write stream to (int32)
	* (optional) "fromsnap" -> full snap name to send an incremental from
	* (optional) "largeblockok" -> (value ignored)
	* indicates that blocks > 128KB are permitted
	* (optional) "embedok" -> (value ignored)
	* presence indicates DRR_WRITE_EMBEDDED records are permitted
	* (optional) "compressok" -> (value ignored)
	* presence indicates compressed DRR_WRITE records are permitted
	* (optional) "rawok" -> (value ignored)
	* presence indicates raw encrypted records should be used.
	* (optional) "savedok" -> (value ignored)
	* presence indicates we should send a partially received snapshot
	* (optional) "resume_object" and "resume_offset" -> (uint64)
	* if present, resume send stream from specified object and offset.
	* (optional) "redactbook" -> (string)
	* if present, use this bookmark's redaction list to generate a redacted
	* send stream
	* }
	*
	* outnvl is unused
	*/
	static const zfs_ioc_key_t zfs_keys_send_new[] = {
	{"fd", DATA_TYPE_INT32, 0},
	{"fromsnap", DATA_TYPE_STRING, ZK_OPTIONAL},
	{"largeblockok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	{"embedok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	{"compressok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	{"rawok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	{"savedok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	{"resume_object", DATA_TYPE_UINT64, ZK_OPTIONAL},
	{"resume_offset", DATA_TYPE_UINT64, ZK_OPTIONAL},
	{"redactbook", DATA_TYPE_STRING, ZK_OPTIONAL},
	};

	/* ARGSUSED */
	static int
	zfs_ioc_send_new(const char snapname, nvlist_t innvl, nvlist_t *outnvl)
	{
	int error;
	offset_t off;
	char *fromname = NULL;
	int fd;
	zfs_file_t *fp;
	boolean_t largeblockok;
	boolean_t embedok;
	boolean_t compressok;
	boolean_t rawok;
	boolean_t savedok;
	uint64_t resumeobj = 0;
	uint64_t resumeoff = 0;
	char *redactbook = NULL;

	fd = fnvlist_lookup_int32(innvl, "fd");

	(void) nvlist_lookup_string(innvl, "fromsnap", &fromname);

	largeblockok = nvlist_exists(innvl, "largeblockok");
	embedok = nvlist_exists(innvl, "embedok");
	compressok = nvlist_exists(innvl, "compressok");
	rawok = nvlist_exists(innvl, "rawok");
	savedok = nvlist_exists(innvl, "savedok");

	(void) nvlist_lookup_uint64(innvl, "resume_object", &resumeobj);
	(void) nvlist_lookup_uint64(innvl, "resume_offset", &resumeoff);

	(void) nvlist_lookup_string(innvl, "redactbook", &redactbook);

	if ((fp = zfs_file_get(fd)) == NULL)
	return (SET_ERROR(EBADF));

	off = zfs_file_off(fp);

	dmu_send_outparams_t out = {0};
	out.dso_outfunc = dump_bytes;
	out.dso_arg = fp;
	out.dso_dryrun = B_FALSE;
	error = dmu_send(snapname, fromname, embedok, largeblockok,
	compressok, rawok, savedok, resumeobj, resumeoff,
	redactbook, fd, &off, &out);

	zfs_file_put(fp);
	return (error);
	}

	/* ARGSUSED */
	static int
	send_space_sum(objset_t os, void buf, int len, void *arg)
	{
	uint64_t *size = arg;
	*size += len;
	return (0);
	}

	/*
	* Determine approximately how large a zfs send stream will be -- the number
	* of bytes that will be written to the fd supplied to zfs_ioc_send_new().
	*
	* innvl: {
	* (optional) "from" -> full snap or bookmark name to send an incremental
	* from
	* (optional) "largeblockok" -> (value ignored)
	* indicates that blocks > 128KB are permitted
	* (optional) "embedok" -> (value ignored)
	* presence indicates DRR_WRITE_EMBEDDED records are permitted
	* (optional) "compressok" -> (value ignored)
	* presence indicates compressed DRR_WRITE records are permitted
	* (optional) "rawok" -> (value ignored)
	* presence indicates raw encrypted records should be used.
	* (optional) "resume_object" and "resume_offset" -> (uint64)
	* if present, resume send stream from specified object and offset.
	* (optional) "fd" -> file descriptor to use as a cookie for progress
	* tracking (int32)
	* }
	*
	* outnvl: {
	* "space" -> bytes of space (uint64)
	* }
	*/
	static const zfs_ioc_key_t zfs_keys_send_space[] = {
	{"from", DATA_TYPE_STRING, ZK_OPTIONAL},
	{"fromsnap", DATA_TYPE_STRING, ZK_OPTIONAL},
	{"largeblockok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	{"embedok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	{"compressok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	{"rawok", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	{"fd", DATA_TYPE_INT32, ZK_OPTIONAL},
	{"redactbook", DATA_TYPE_STRING, ZK_OPTIONAL},
	{"resume_object", DATA_TYPE_UINT64, ZK_OPTIONAL},
	{"resume_offset", DATA_TYPE_UINT64, ZK_OPTIONAL},
	{"bytes", DATA_TYPE_UINT64, ZK_OPTIONAL},
	};

	static int
	zfs_ioc_send_space(const char snapname, nvlist_t innvl, nvlist_t *outnvl)
	{
	dsl_pool_t *dp;
	dsl_dataset_t *tosnap;
	dsl_dataset_t *fromsnap = NULL;
	int error;
	char *fromname = NULL;
	char *redactlist_book = NULL;
	boolean_t largeblockok;
	boolean_t embedok;
	boolean_t compressok;
	boolean_t rawok;
	boolean_t savedok;
	uint64_t space = 0;
	boolean_t full_estimate = B_FALSE;
	uint64_t resumeobj = 0;
	uint64_t resumeoff = 0;
	uint64_t resume_bytes = 0;
	int32_t fd = -1;
	zfs_bookmark_phys_t zbm = {0};

	error = dsl_pool_hold(snapname, FTAG, &dp);
	if (error != 0)
	return (error);

	error = dsl_dataset_hold(dp, snapname, FTAG, &tosnap);
	if (error != 0) {
	dsl_pool_rele(dp, FTAG);
	return (error);
	}
	(void) nvlist_lookup_int32(innvl, "fd", &fd);

	largeblockok = nvlist_exists(innvl, "largeblockok");
	embedok = nvlist_exists(innvl, "embedok");
	compressok = nvlist_exists(innvl, "compressok");
	rawok = nvlist_exists(innvl, "rawok");
	savedok = nvlist_exists(innvl, "savedok");
	boolean_t from = (nvlist_lookup_string(innvl, "from", &fromname) == 0);
	boolean_t altbook = (nvlist_lookup_string(innvl, "redactbook",
	&redactlist_book) == 0);

	(void) nvlist_lookup_uint64(innvl, "resume_object", &resumeobj);
	(void) nvlist_lookup_uint64(innvl, "resume_offset", &resumeoff);
	(void) nvlist_lookup_uint64(innvl, "bytes", &resume_bytes);

	if (altbook) {
	full_estimate = B_TRUE;
	} else if (from) {
	if (strchr(fromname, '#')) {
	error = dsl_bookmark_lookup(dp, fromname, tosnap, &zbm);

	/*
	* dsl_bookmark_lookup() will fail with EXDEV if
	* the from-bookmark and tosnap are at the same txg.
	* However, it's valid to do a send (and therefore,
	* a send estimate) from and to the same time point,
	* if the bookmark is redacted (the incremental send
	* can change what's redacted on the target). In
	* this case, dsl_bookmark_lookup() fills in zbm
	* but returns EXDEV. Ignore this error.
	*/
	if (error == EXDEV && zbm.zbm_redaction_obj != 0 &&
	zbm.zbm_guid ==
	dsl_dataset_phys(tosnap)->ds_guid)
	error = 0;

	if (error != 0) {
	dsl_dataset_rele(tosnap, FTAG);
	dsl_pool_rele(dp, FTAG);
	return (error);
	}
	if (zbm.zbm_redaction_obj != 0 \|\| !(zbm.zbm_flags &
	ZBM_FLAG_HAS_FBN)) {
	full_estimate = B_TRUE;
	}
	} else if (strchr(fromname, '@')) {
	error = dsl_dataset_hold(dp, fromname, FTAG, &fromsnap);
	if (error != 0) {
	dsl_dataset_rele(tosnap, FTAG);
	dsl_pool_rele(dp, FTAG);
	return (error);
	}

	if (!dsl_dataset_is_before(tosnap, fromsnap, 0)) {
	full_estimate = B_TRUE;
	dsl_dataset_rele(fromsnap, FTAG);
	}
	} else {
	/*
	* from is not properly formatted as a snapshot or
	* bookmark
	*/
	dsl_dataset_rele(tosnap, FTAG);
	dsl_pool_rele(dp, FTAG);
	return (SET_ERROR(EINVAL));
	}
	}

	if (full_estimate) {
	dmu_send_outparams_t out = {0};
	offset_t off = 0;
	out.dso_outfunc = send_space_sum;
	out.dso_arg = &space;
	out.dso_dryrun = B_TRUE;
	/*
	* We have to release these holds so dmu_send can take them. It
	* will do all the error checking we need.
	*/
	dsl_dataset_rele(tosnap, FTAG);
	dsl_pool_rele(dp, FTAG);
	error = dmu_send(snapname, fromname, embedok, largeblockok,
	compressok, rawok, savedok, resumeobj, resumeoff,
	redactlist_book, fd, &off, &out);
	} else {
	error = dmu_send_estimate_fast(tosnap, fromsnap,
	(from && strchr(fromname, '#') != NULL ? &zbm : NULL),
	compressok \|\| rawok, savedok, &space);
	space -= resume_bytes;
	if (fromsnap != NULL)
	dsl_dataset_rele(fromsnap, FTAG);
	dsl_dataset_rele(tosnap, FTAG);
	dsl_pool_rele(dp, FTAG);
	}

	fnvlist_add_uint64(outnvl, "space", space);

	return (error);
	}

	/*
	* Sync the currently open TXG to disk for the specified pool.
	* This is somewhat similar to 'zfs_sync()'.
	* For cases that do not result in error this ioctl will wait for
	* the currently open TXG to commit before returning back to the caller.
	*
	* innvl: {
	* "force" -> when true, force uberblock update even if there is no dirty data.
	* In addition this will cause the vdev configuration to be written
	* out including updating the zpool cache file. (boolean_t)
	* }
	*
	* onvl is unused
	*/
	static const zfs_ioc_key_t zfs_keys_pool_sync[] = {
	{"force", DATA_TYPE_BOOLEAN_VALUE, 0},
	};

	/* ARGSUSED */
	static int
	zfs_ioc_pool_sync(const char pool, nvlist_t innvl, nvlist_t *onvl)
	{
	int err;
	boolean_t rc, force = B_FALSE;
	spa_t *spa;

	if ((err = spa_open(pool, &spa, FTAG)) != 0)
	return (err);

	if (innvl) {
	err = nvlist_lookup_boolean_value(innvl, "force", &rc);
	if (err == 0)
	force = rc;
	}

	if (force) {
	spa_config_enter(spa, SCL_CONFIG, FTAG, RW_WRITER);
	vdev_config_dirty(spa->spa_root_vdev);
	spa_config_exit(spa, SCL_CONFIG, FTAG);
	}
	txg_wait_synced(spa_get_dsl(spa), 0);

	spa_close(spa, FTAG);

	return (0);
	}

	/*
	* Load a user's wrapping key into the kernel.
	* innvl: {
	* "hidden_args" -> { "wkeydata" -> value }
	* raw uint8_t array of encryption wrapping key data (32 bytes)
	* (optional) "noop" -> (value ignored)
	* presence indicated key should only be verified, not loaded
	* }
	*/
	static const zfs_ioc_key_t zfs_keys_load_key[] = {
	{"hidden_args", DATA_TYPE_NVLIST, 0},
	{"noop", DATA_TYPE_BOOLEAN, ZK_OPTIONAL},
	};

	/* ARGSUSED */
	static int
	zfs_ioc_load_key(const char dsname, nvlist_t innvl, nvlist_t *outnvl)
	{
	int ret;
	dsl_crypto_params_t *dcp = NULL;
	nvlist_t *hidden_args;
	boolean_t noop = nvlist_exists(innvl, "noop");

	if (strchr(dsname, '@') != NULL \|\| strchr(dsname, '%') != NULL) {
	ret = SET_ERROR(EINVAL);
	goto error;
	}

	hidden_args = fnvlist_lookup_nvlist(innvl, ZPOOL_HIDDEN_ARGS);

	ret = dsl_crypto_params_create_nvlist(DCP_CMD_NONE, NULL,
	hidden_args, &dcp);
	if (ret != 0)
	goto error;

	ret = spa_keystore_load_wkey(dsname, dcp, noop);
	if (ret != 0)
	goto error;

	dsl_crypto_params_free(dcp, noop);

	return (0);

	error:
	dsl_crypto_params_free(dcp, B_TRUE);
	return (ret);
	}

	/*
	* Unload a user's wrapping key from the kernel.
	* Both innvl and outnvl are unused.
	*/
	static const zfs_ioc_key_t zfs_keys_unload_key[] = {
	/* no nvl keys */
	};

	/* ARGSUSED */
	static int
	zfs_ioc_unload_key(const char dsname, nvlist_t innvl, nvlist_t *outnvl)
	{
	int ret = 0;

	if (strchr(dsname, '@') != NULL \|\| strchr(dsname, '%') != NULL) {
	ret = (SET_ERROR(EINVAL));
	goto out;
	}

	ret = spa_keystore_unload_wkey(dsname);
	if (ret != 0)
	goto out;

	out:
	return (ret);
	}

	/*
	* Changes a user's wrapping key used to decrypt a dataset. The keyformat,
	* keylocation, pbkdf2salt, and pbkdf2iters properties can also be specified
	* here to change how the key is derived in userspace.
	*
	* innvl: {
	* "hidden_args" (optional) -> { "wkeydata" -> value }
	* raw uint8_t array of new encryption wrapping key data (32 bytes)
	* "props" (optional) -> { prop -> value }
	* }
	*
	* outnvl is unused
	*/
	static const zfs_ioc_key_t zfs_keys_change_key[] = {
	{"crypt_cmd", DATA_TYPE_UINT64, ZK_OPTIONAL},
	{"hidden_args", DATA_TYPE_NVLIST, ZK_OPTIONAL},
	{"props", DATA_TYPE_NVLIST, ZK_OPTIONAL},
	};

	/* ARGSUSED */
	static int
	zfs_ioc_change_key(const char dsname, nvlist_t innvl, nvlist_t *outnvl)
	{
	int ret;
	uint64_t cmd = DCP_CMD_NONE;
	dsl_crypto_params_t *dcp = NULL;
	nvlist_t args = NULL, hidden_args = NULL;

	if (strchr(dsname, '@') != NULL \|\| strchr(dsname, '%') != NULL) {
	ret = (SET_ERROR(EINVAL));
	goto error;
	}

	(void) nvlist_lookup_uint64(innvl, "crypt_cmd", &cmd);
	(void) nvlist_lookup_nvlist(innvl, "props", &args);
	(void) nvlist_lookup_nvlist(innvl, ZPOOL_HIDDEN_ARGS, &hidden_args);

	ret = dsl_crypto_params_create_nvlist(cmd, args, hidden_args, &dcp);
	if (ret != 0)
	goto error;

	ret = spa_keystore_change_key(dsname, dcp);
	if (ret != 0)
	goto error;

	dsl_crypto_params_free(dcp, B_FALSE);

	return (0);

	error:
	dsl_crypto_params_free(dcp, B_TRUE);
	return (ret);
	}

	static zfs_ioc_vec_t zfs_ioc_vec[ZFS_IOC_LAST - ZFS_IOC_FIRST];

	static void
	zfs_ioctl_register_legacy(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func,
	zfs_secpolicy_func_t *secpolicy, zfs_ioc_namecheck_t namecheck,
	boolean_t log_history, zfs_ioc_poolcheck_t pool_check)
	{
	zfs_ioc_vec_t *vec = &zfs_ioc_vec[ioc - ZFS_IOC_FIRST];

	ASSERT3U(ioc, >=, ZFS_IOC_FIRST);
	ASSERT3U(ioc, <, ZFS_IOC_LAST);
	ASSERT3P(vec->zvec_legacy_func, ==, NULL);
	ASSERT3P(vec->zvec_func, ==, NULL);

	vec->zvec_legacy_func = func;
	vec->zvec_secpolicy = secpolicy;
	vec->zvec_namecheck = namecheck;
	vec->zvec_allow_log = log_history;
	vec->zvec_pool_check = pool_check;
	}

	/*
	* See the block comment at the beginning of this file for details on
	* each argument to this function.
	*/
	void
	zfs_ioctl_register(const char name, zfs_ioc_t ioc, zfs_ioc_func_t func,
	zfs_secpolicy_func_t *secpolicy, zfs_ioc_namecheck_t namecheck,
	zfs_ioc_poolcheck_t pool_check, boolean_t smush_outnvlist,
	boolean_t allow_log, const zfs_ioc_key_t *nvl_keys, size_t num_keys)
	{
	zfs_ioc_vec_t *vec = &zfs_ioc_vec[ioc - ZFS_IOC_FIRST];

	ASSERT3U(ioc, >=, ZFS_IOC_FIRST);
	ASSERT3U(ioc, <, ZFS_IOC_LAST);
	ASSERT3P(vec->zvec_legacy_func, ==, NULL);
	ASSERT3P(vec->zvec_func, ==, NULL);

	/* if we are logging, the name must be valid */
	ASSERT(!allow_log \|\| namecheck != NO_NAME);

	vec->zvec_name = name;
	vec->zvec_func = func;
	vec->zvec_secpolicy = secpolicy;
	vec->zvec_namecheck = namecheck;
	vec->zvec_pool_check = pool_check;
	vec->zvec_smush_outnvlist = smush_outnvlist;
	vec->zvec_allow_log = allow_log;
	vec->zvec_nvl_keys = nvl_keys;
	vec->zvec_nvl_key_count = num_keys;
	}

	static void
	zfs_ioctl_register_pool(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func,
	zfs_secpolicy_func_t *secpolicy, boolean_t log_history,
	zfs_ioc_poolcheck_t pool_check)
	{
	zfs_ioctl_register_legacy(ioc, func, secpolicy,
	POOL_NAME, log_history, pool_check);
	}

	void
	zfs_ioctl_register_dataset_nolog(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func,
	zfs_secpolicy_func_t *secpolicy, zfs_ioc_poolcheck_t pool_check)
	{
	zfs_ioctl_register_legacy(ioc, func, secpolicy,
	DATASET_NAME, B_FALSE, pool_check);
	}

	static void
	zfs_ioctl_register_pool_modify(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func)
	{
	zfs_ioctl_register_legacy(ioc, func, zfs_secpolicy_config,
	POOL_NAME, B_TRUE, POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY);
	}

	static void
	zfs_ioctl_register_pool_meta(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func,
	zfs_secpolicy_func_t *secpolicy)
	{
	zfs_ioctl_register_legacy(ioc, func, secpolicy,
	NO_NAME, B_FALSE, POOL_CHECK_NONE);
	}

	static void
	zfs_ioctl_register_dataset_read_secpolicy(zfs_ioc_t ioc,
	zfs_ioc_legacy_func_t func, zfs_secpolicy_func_t secpolicy)
	{
	zfs_ioctl_register_legacy(ioc, func, secpolicy,
	DATASET_NAME, B_FALSE, POOL_CHECK_SUSPENDED);
	}

	static void
	zfs_ioctl_register_dataset_read(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func)
	{
	zfs_ioctl_register_dataset_read_secpolicy(ioc, func,
	zfs_secpolicy_read);
	}

	static void
	zfs_ioctl_register_dataset_modify(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func,
	zfs_secpolicy_func_t *secpolicy)
	{
	zfs_ioctl_register_legacy(ioc, func, secpolicy,
	DATASET_NAME, B_TRUE, POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY);
	}

	static void
	zfs_ioctl_init(void)
	{
	zfs_ioctl_register("snapshot", ZFS_IOC_SNAPSHOT,
	zfs_ioc_snapshot, zfs_secpolicy_snapshot, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_snapshot, ARRAY_SIZE(zfs_keys_snapshot));

	zfs_ioctl_register("log_history", ZFS_IOC_LOG_HISTORY,
	zfs_ioc_log_history, zfs_secpolicy_log_history, NO_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_FALSE, B_FALSE,
	zfs_keys_log_history, ARRAY_SIZE(zfs_keys_log_history));

	zfs_ioctl_register("space_snaps", ZFS_IOC_SPACE_SNAPS,
	zfs_ioc_space_snaps, zfs_secpolicy_read, DATASET_NAME,
	POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE,
	zfs_keys_space_snaps, ARRAY_SIZE(zfs_keys_space_snaps));

	zfs_ioctl_register("send", ZFS_IOC_SEND_NEW,
	zfs_ioc_send_new, zfs_secpolicy_send_new, DATASET_NAME,
	POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE,
	zfs_keys_send_new, ARRAY_SIZE(zfs_keys_send_new));

	zfs_ioctl_register("send_space", ZFS_IOC_SEND_SPACE,
	zfs_ioc_send_space, zfs_secpolicy_read, DATASET_NAME,
	POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE,
	zfs_keys_send_space, ARRAY_SIZE(zfs_keys_send_space));

	zfs_ioctl_register("create", ZFS_IOC_CREATE,
	zfs_ioc_create, zfs_secpolicy_create_clone, DATASET_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_create, ARRAY_SIZE(zfs_keys_create));

	zfs_ioctl_register("clone", ZFS_IOC_CLONE,
	zfs_ioc_clone, zfs_secpolicy_create_clone, DATASET_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_clone, ARRAY_SIZE(zfs_keys_clone));

	zfs_ioctl_register("remap", ZFS_IOC_REMAP,
	zfs_ioc_remap, zfs_secpolicy_none, DATASET_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_FALSE, B_TRUE,
	zfs_keys_remap, ARRAY_SIZE(zfs_keys_remap));

	zfs_ioctl_register("destroy_snaps", ZFS_IOC_DESTROY_SNAPS,
	zfs_ioc_destroy_snaps, zfs_secpolicy_destroy_snaps, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_destroy_snaps, ARRAY_SIZE(zfs_keys_destroy_snaps));

	zfs_ioctl_register("hold", ZFS_IOC_HOLD,
	zfs_ioc_hold, zfs_secpolicy_hold, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_hold, ARRAY_SIZE(zfs_keys_hold));
	zfs_ioctl_register("release", ZFS_IOC_RELEASE,
	zfs_ioc_release, zfs_secpolicy_release, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_release, ARRAY_SIZE(zfs_keys_release));

	zfs_ioctl_register("get_holds", ZFS_IOC_GET_HOLDS,
	zfs_ioc_get_holds, zfs_secpolicy_read, DATASET_NAME,
	POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE,
	zfs_keys_get_holds, ARRAY_SIZE(zfs_keys_get_holds));

	zfs_ioctl_register("rollback", ZFS_IOC_ROLLBACK,
	zfs_ioc_rollback, zfs_secpolicy_rollback, DATASET_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_FALSE, B_TRUE,
	zfs_keys_rollback, ARRAY_SIZE(zfs_keys_rollback));

	zfs_ioctl_register("bookmark", ZFS_IOC_BOOKMARK,
	zfs_ioc_bookmark, zfs_secpolicy_bookmark, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_bookmark, ARRAY_SIZE(zfs_keys_bookmark));

	zfs_ioctl_register("get_bookmarks", ZFS_IOC_GET_BOOKMARKS,
	zfs_ioc_get_bookmarks, zfs_secpolicy_read, DATASET_NAME,
	POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE,
	zfs_keys_get_bookmarks, ARRAY_SIZE(zfs_keys_get_bookmarks));

	zfs_ioctl_register("get_bookmark_props", ZFS_IOC_GET_BOOKMARK_PROPS,
	zfs_ioc_get_bookmark_props, zfs_secpolicy_read, ENTITY_NAME,
	POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE, zfs_keys_get_bookmark_props,
	ARRAY_SIZE(zfs_keys_get_bookmark_props));

	zfs_ioctl_register("destroy_bookmarks", ZFS_IOC_DESTROY_BOOKMARKS,
	zfs_ioc_destroy_bookmarks, zfs_secpolicy_destroy_bookmarks,
	POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_destroy_bookmarks,
	ARRAY_SIZE(zfs_keys_destroy_bookmarks));

	zfs_ioctl_register("receive", ZFS_IOC_RECV_NEW,
	zfs_ioc_recv_new, zfs_secpolicy_recv_new, DATASET_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_recv_new, ARRAY_SIZE(zfs_keys_recv_new));
	zfs_ioctl_register("load-key", ZFS_IOC_LOAD_KEY,
	zfs_ioc_load_key, zfs_secpolicy_load_key,
	DATASET_NAME, POOL_CHECK_SUSPENDED, B_TRUE, B_TRUE,
	zfs_keys_load_key, ARRAY_SIZE(zfs_keys_load_key));
	zfs_ioctl_register("unload-key", ZFS_IOC_UNLOAD_KEY,
	zfs_ioc_unload_key, zfs_secpolicy_load_key,
	DATASET_NAME, POOL_CHECK_SUSPENDED, B_TRUE, B_TRUE,
	zfs_keys_unload_key, ARRAY_SIZE(zfs_keys_unload_key));
	zfs_ioctl_register("change-key", ZFS_IOC_CHANGE_KEY,
	zfs_ioc_change_key, zfs_secpolicy_change_key,
	DATASET_NAME, POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY,
	B_TRUE, B_TRUE, zfs_keys_change_key,
	ARRAY_SIZE(zfs_keys_change_key));

	zfs_ioctl_register("sync", ZFS_IOC_POOL_SYNC,
	zfs_ioc_pool_sync, zfs_secpolicy_none, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_FALSE, B_FALSE,
	zfs_keys_pool_sync, ARRAY_SIZE(zfs_keys_pool_sync));
	zfs_ioctl_register("reopen", ZFS_IOC_POOL_REOPEN, zfs_ioc_pool_reopen,
	zfs_secpolicy_config, POOL_NAME, POOL_CHECK_SUSPENDED, B_TRUE,
	B_TRUE, zfs_keys_pool_reopen, ARRAY_SIZE(zfs_keys_pool_reopen));

	zfs_ioctl_register("channel_program", ZFS_IOC_CHANNEL_PROGRAM,
	zfs_ioc_channel_program, zfs_secpolicy_config,
	POOL_NAME, POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE,
	B_TRUE, zfs_keys_channel_program,
	ARRAY_SIZE(zfs_keys_channel_program));

	zfs_ioctl_register("redact", ZFS_IOC_REDACT,
	zfs_ioc_redact, zfs_secpolicy_config, DATASET_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_redact, ARRAY_SIZE(zfs_keys_redact));

	zfs_ioctl_register("zpool_checkpoint", ZFS_IOC_POOL_CHECKPOINT,
	zfs_ioc_pool_checkpoint, zfs_secpolicy_config, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_pool_checkpoint, ARRAY_SIZE(zfs_keys_pool_checkpoint));

	zfs_ioctl_register("zpool_discard_checkpoint",
	ZFS_IOC_POOL_DISCARD_CHECKPOINT, zfs_ioc_pool_discard_checkpoint,
	zfs_secpolicy_config, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_pool_discard_checkpoint,
	ARRAY_SIZE(zfs_keys_pool_discard_checkpoint));

	zfs_ioctl_register("initialize", ZFS_IOC_POOL_INITIALIZE,
	zfs_ioc_pool_initialize, zfs_secpolicy_config, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_pool_initialize, ARRAY_SIZE(zfs_keys_pool_initialize));

	zfs_ioctl_register("trim", ZFS_IOC_POOL_TRIM,
	zfs_ioc_pool_trim, zfs_secpolicy_config, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_TRUE, B_TRUE,
	zfs_keys_pool_trim, ARRAY_SIZE(zfs_keys_pool_trim));

	zfs_ioctl_register("wait", ZFS_IOC_WAIT,
	zfs_ioc_wait, zfs_secpolicy_none, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_FALSE, B_FALSE,
	zfs_keys_pool_wait, ARRAY_SIZE(zfs_keys_pool_wait));

	zfs_ioctl_register("wait_fs", ZFS_IOC_WAIT_FS,
	zfs_ioc_wait_fs, zfs_secpolicy_none, DATASET_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_FALSE, B_FALSE,
	zfs_keys_fs_wait, ARRAY_SIZE(zfs_keys_fs_wait));

	zfs_ioctl_register("set_bootenv", ZFS_IOC_SET_BOOTENV,
	zfs_ioc_set_bootenv, zfs_secpolicy_config, POOL_NAME,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_FALSE, B_TRUE,
	zfs_keys_set_bootenv, ARRAY_SIZE(zfs_keys_set_bootenv));

	zfs_ioctl_register("get_bootenv", ZFS_IOC_GET_BOOTENV,
	zfs_ioc_get_bootenv, zfs_secpolicy_none, POOL_NAME,
	POOL_CHECK_SUSPENDED, B_FALSE, B_TRUE,
	zfs_keys_get_bootenv, ARRAY_SIZE(zfs_keys_get_bootenv));

	+ zfs_ioctl_register("zpool_vdev_get_props", ZFS_IOC_VDEV_GET_PROPS,
	+ zfs_ioc_vdev_get_props, zfs_secpolicy_read, POOL_NAME,
	+ POOL_CHECK_NONE, B_FALSE, B_FALSE, zfs_keys_vdev_get_props,
	+ ARRAY_SIZE(zfs_keys_vdev_get_props));
	+
	+ zfs_ioctl_register("zpool_vdev_set_props", ZFS_IOC_VDEV_SET_PROPS,
	+ zfs_ioc_vdev_set_props, zfs_secpolicy_config, POOL_NAME,
	+ POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY, B_FALSE, B_FALSE,
	+ zfs_keys_vdev_set_props, ARRAY_SIZE(zfs_keys_vdev_set_props));
	+
	/* IOCTLS that use the legacy function signature */

	zfs_ioctl_register_legacy(ZFS_IOC_POOL_FREEZE, zfs_ioc_pool_freeze,
	zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_READONLY);

	zfs_ioctl_register_pool(ZFS_IOC_POOL_CREATE, zfs_ioc_pool_create,
	zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE);
	zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SCAN,
	zfs_ioc_pool_scan);
	zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_UPGRADE,
	zfs_ioc_pool_upgrade);
	zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ADD,
	zfs_ioc_vdev_add);
	zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_REMOVE,
	zfs_ioc_vdev_remove);
	zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SET_STATE,
	zfs_ioc_vdev_set_state);
	zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ATTACH,
	zfs_ioc_vdev_attach);
	zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_DETACH,
	zfs_ioc_vdev_detach);
	zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETPATH,
	zfs_ioc_vdev_setpath);
	zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETFRU,
	zfs_ioc_vdev_setfru);
	zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SET_PROPS,
	zfs_ioc_pool_set_props);
	zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SPLIT,
	zfs_ioc_vdev_split);
	zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_REGUID,
	zfs_ioc_pool_reguid);

	zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_CONFIGS,
	zfs_ioc_pool_configs, zfs_secpolicy_none);
	zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_TRYIMPORT,
	zfs_ioc_pool_tryimport, zfs_secpolicy_config);
	zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_FAULT,
	zfs_ioc_inject_fault, zfs_secpolicy_inject);
	zfs_ioctl_register_pool_meta(ZFS_IOC_CLEAR_FAULT,
	zfs_ioc_clear_fault, zfs_secpolicy_inject);
	zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_LIST_NEXT,
	zfs_ioc_inject_list_next, zfs_secpolicy_inject);

	/*
	* pool destroy, and export don't log the history as part of
	* zfsdev_ioctl, but rather zfs_ioc_pool_export
	* does the logging of those commands.
	*/
	zfs_ioctl_register_pool(ZFS_IOC_POOL_DESTROY, zfs_ioc_pool_destroy,
	zfs_secpolicy_config, B_FALSE, POOL_CHECK_SUSPENDED);
	zfs_ioctl_register_pool(ZFS_IOC_POOL_EXPORT, zfs_ioc_pool_export,
	zfs_secpolicy_config, B_FALSE, POOL_CHECK_SUSPENDED);

	zfs_ioctl_register_pool(ZFS_IOC_POOL_STATS, zfs_ioc_pool_stats,
	zfs_secpolicy_read, B_FALSE, POOL_CHECK_NONE);
	zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_PROPS, zfs_ioc_pool_get_props,
	zfs_secpolicy_read, B_FALSE, POOL_CHECK_NONE);

	zfs_ioctl_register_pool(ZFS_IOC_ERROR_LOG, zfs_ioc_error_log,
	zfs_secpolicy_inject, B_FALSE, POOL_CHECK_SUSPENDED);
	zfs_ioctl_register_pool(ZFS_IOC_DSOBJ_TO_DSNAME,
	zfs_ioc_dsobj_to_dsname,
	zfs_secpolicy_diff, B_FALSE, POOL_CHECK_SUSPENDED);
	zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_HISTORY,
	zfs_ioc_pool_get_history,
	zfs_secpolicy_config, B_FALSE, POOL_CHECK_SUSPENDED);

	zfs_ioctl_register_pool(ZFS_IOC_POOL_IMPORT, zfs_ioc_pool_import,
	zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE);

	zfs_ioctl_register_pool(ZFS_IOC_CLEAR, zfs_ioc_clear,
	zfs_secpolicy_config, B_TRUE, POOL_CHECK_READONLY);

	zfs_ioctl_register_dataset_read(ZFS_IOC_SPACE_WRITTEN,
	zfs_ioc_space_written);
	zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_RECVD_PROPS,
	zfs_ioc_objset_recvd_props);
	zfs_ioctl_register_dataset_read(ZFS_IOC_NEXT_OBJ,
	zfs_ioc_next_obj);
	zfs_ioctl_register_dataset_read(ZFS_IOC_GET_FSACL,
	zfs_ioc_get_fsacl);
	zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_STATS,
	zfs_ioc_objset_stats);
	zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_ZPLPROPS,
	zfs_ioc_objset_zplprops);
	zfs_ioctl_register_dataset_read(ZFS_IOC_DATASET_LIST_NEXT,
	zfs_ioc_dataset_list_next);
	zfs_ioctl_register_dataset_read(ZFS_IOC_SNAPSHOT_LIST_NEXT,
	zfs_ioc_snapshot_list_next);
	zfs_ioctl_register_dataset_read(ZFS_IOC_SEND_PROGRESS,
	zfs_ioc_send_progress);

	zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_DIFF,
	zfs_ioc_diff, zfs_secpolicy_diff);
	zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_STATS,
	zfs_ioc_obj_to_stats, zfs_secpolicy_diff);
	zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_PATH,
	zfs_ioc_obj_to_path, zfs_secpolicy_diff);
	zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_ONE,
	zfs_ioc_userspace_one, zfs_secpolicy_userspace_one);
	zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_MANY,
	zfs_ioc_userspace_many, zfs_secpolicy_userspace_many);
	zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_SEND,
	zfs_ioc_send, zfs_secpolicy_send);

	zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_PROP, zfs_ioc_set_prop,
	zfs_secpolicy_none);
	zfs_ioctl_register_dataset_modify(ZFS_IOC_DESTROY, zfs_ioc_destroy,
	zfs_secpolicy_destroy);
	zfs_ioctl_register_dataset_modify(ZFS_IOC_RENAME, zfs_ioc_rename,
	zfs_secpolicy_rename);
	zfs_ioctl_register_dataset_modify(ZFS_IOC_RECV, zfs_ioc_recv,
	zfs_secpolicy_recv);
	zfs_ioctl_register_dataset_modify(ZFS_IOC_PROMOTE, zfs_ioc_promote,
	zfs_secpolicy_promote);
	zfs_ioctl_register_dataset_modify(ZFS_IOC_INHERIT_PROP,
	zfs_ioc_inherit_prop, zfs_secpolicy_inherit_prop);
	zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_FSACL, zfs_ioc_set_fsacl,
	zfs_secpolicy_set_fsacl);

	zfs_ioctl_register_dataset_nolog(ZFS_IOC_SHARE, zfs_ioc_share,
	zfs_secpolicy_share, POOL_CHECK_NONE);
	zfs_ioctl_register_dataset_nolog(ZFS_IOC_SMB_ACL, zfs_ioc_smb_acl,
	zfs_secpolicy_smb_acl, POOL_CHECK_NONE);
	zfs_ioctl_register_dataset_nolog(ZFS_IOC_USERSPACE_UPGRADE,
	zfs_ioc_userspace_upgrade, zfs_secpolicy_userspace_upgrade,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY);
	zfs_ioctl_register_dataset_nolog(ZFS_IOC_TMP_SNAPSHOT,
	zfs_ioc_tmp_snapshot, zfs_secpolicy_tmp_snapshot,
	POOL_CHECK_SUSPENDED \| POOL_CHECK_READONLY);

	zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_NEXT, zfs_ioc_events_next,
	zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_NONE);
	zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_CLEAR, zfs_ioc_events_clear,
	zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_NONE);
	zfs_ioctl_register_legacy(ZFS_IOC_EVENTS_SEEK, zfs_ioc_events_seek,
	zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_NONE);

	zfs_ioctl_init_os();
	}

	/*
	* Verify that for non-legacy ioctls the input nvlist
	* pairs match against the expected input.
	*
	* Possible errors are:
	* ZFS_ERR_IOC_ARG_UNAVAIL An unrecognized nvpair was encountered
	* ZFS_ERR_IOC_ARG_REQUIRED A required nvpair is missing
	* ZFS_ERR_IOC_ARG_BADTYPE Invalid type for nvpair
	*/
	static int
	zfs_check_input_nvpairs(nvlist_t innvl, const zfs_ioc_vec_t vec)
	{
	const zfs_ioc_key_t *nvl_keys = vec->zvec_nvl_keys;
	boolean_t required_keys_found = B_FALSE;

	/*
	* examine each input pair
	*/
	for (nvpair_t *pair = nvlist_next_nvpair(innvl, NULL);
	pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) {
	char *name = nvpair_name(pair);
	data_type_t type = nvpair_type(pair);
	boolean_t identified = B_FALSE;

	/*
	* check pair against the documented names and type
	*/
	for (int k = 0; k < vec->zvec_nvl_key_count; k++) {
	/* if not a wild card name, check for an exact match */
	if ((nvl_keys[k].zkey_flags & ZK_WILDCARDLIST) == 0 &&
	strcmp(nvl_keys[k].zkey_name, name) != 0)
	continue;

	identified = B_TRUE;

	if (nvl_keys[k].zkey_type != DATA_TYPE_ANY &&
	nvl_keys[k].zkey_type != type) {
	return (SET_ERROR(ZFS_ERR_IOC_ARG_BADTYPE));
	}

	if (nvl_keys[k].zkey_flags & ZK_OPTIONAL)
	continue;

	required_keys_found = B_TRUE;
	break;
	}

	/* allow an 'optional' key, everything else is invalid */
	if (!identified &&
	(strcmp(name, "optional") != 0 \|\|
	type != DATA_TYPE_NVLIST)) {
	return (SET_ERROR(ZFS_ERR_IOC_ARG_UNAVAIL));
	}
	}

	/* verify that all required keys were found */
	for (int k = 0; k < vec->zvec_nvl_key_count; k++) {
	if (nvl_keys[k].zkey_flags & ZK_OPTIONAL)
	continue;

	if (nvl_keys[k].zkey_flags & ZK_WILDCARDLIST) {
	/* at least one non-optional key is expected here */
	if (!required_keys_found)
	return (SET_ERROR(ZFS_ERR_IOC_ARG_REQUIRED));
	continue;
	}

	if (!nvlist_exists(innvl, nvl_keys[k].zkey_name))
	return (SET_ERROR(ZFS_ERR_IOC_ARG_REQUIRED));
	}

	return (0);
	}

	static int
	pool_status_check(const char *name, zfs_ioc_namecheck_t type,
	zfs_ioc_poolcheck_t check)
	{
	spa_t *spa;
	int error;

	ASSERT(type == POOL_NAME \|\| type == DATASET_NAME \|\|
	type == ENTITY_NAME);

	if (check & POOL_CHECK_NONE)
	return (0);

	error = spa_open(name, &spa, FTAG);
	if (error == 0) {
	if ((check & POOL_CHECK_SUSPENDED) && spa_suspended(spa))
	error = SET_ERROR(EAGAIN);
	else if ((check & POOL_CHECK_READONLY) && !spa_writeable(spa))
	error = SET_ERROR(EROFS);
	spa_close(spa, FTAG);
	}
	return (error);
	}

	int
	zfsdev_getminor(zfs_file_t fp, minor_t minorp)
	{
	zfsdev_state_t zs, fpd;

	ASSERT(!MUTEX_HELD(&zfsdev_state_lock));

	fpd = zfs_file_private(fp);
	if (fpd == NULL)
	return (SET_ERROR(EBADF));

	mutex_enter(&zfsdev_state_lock);

	for (zs = zfsdev_state_list; zs != NULL; zs = zs->zs_next) {

	if (zs->zs_minor == -1)
	continue;

	if (fpd == zs) {
	*minorp = fpd->zs_minor;
	mutex_exit(&zfsdev_state_lock);
	return (0);
	}
	}

	mutex_exit(&zfsdev_state_lock);

	return (SET_ERROR(EBADF));
	}

	void *
	zfsdev_get_state(minor_t minor, enum zfsdev_state_type which)
	{
	zfsdev_state_t *zs;

	for (zs = zfsdev_state_list; zs != NULL; zs = zs->zs_next) {
	if (zs->zs_minor == minor) {
	smp_rmb();
	switch (which) {
	case ZST_ONEXIT:
	return (zs->zs_onexit);
	case ZST_ZEVENT:
	return (zs->zs_zevent);
	case ZST_ALL:
	return (zs);
	}
	}
	}

	return (NULL);
	}

	/*
	* Find a free minor number. The zfsdev_state_list is expected to
	* be short since it is only a list of currently open file handles.
	*/
	static minor_t
	zfsdev_minor_alloc(void)
	{
	static minor_t last_minor = 0;
	minor_t m;

	ASSERT(MUTEX_HELD(&zfsdev_state_lock));

	for (m = last_minor + 1; m != last_minor; m++) {
	if (m > ZFSDEV_MAX_MINOR)
	m = 1;
	if (zfsdev_get_state(m, ZST_ALL) == NULL) {
	last_minor = m;
	return (m);
	}
	}

	return (0);
	}

	int
	zfsdev_state_init(void *priv)
	{
	zfsdev_state_t zs, zsprev = NULL;
	minor_t minor;
	boolean_t newzs = B_FALSE;

	ASSERT(MUTEX_HELD(&zfsdev_state_lock));

	minor = zfsdev_minor_alloc();
	if (minor == 0)
	return (SET_ERROR(ENXIO));

	for (zs = zfsdev_state_list; zs != NULL; zs = zs->zs_next) {
	if (zs->zs_minor == -1)
	break;
	zsprev = zs;
	}

	if (!zs) {
	zs = kmem_zalloc(sizeof (zfsdev_state_t), KM_SLEEP);
	newzs = B_TRUE;
	}

	zfsdev_private_set_state(priv, zs);

	zfs_onexit_init((zfs_onexit_t **)&zs->zs_onexit);
	zfs_zevent_init((zfs_zevent_t **)&zs->zs_zevent);

	/*
	* In order to provide for lock-free concurrent read access
	* to the minor list in zfsdev_get_state(), new entries
	* must be completely written before linking them into the
	* list whereas existing entries are already linked; the last
	* operation must be updating zs_minor (from -1 to the new
	* value).
	*/
	if (newzs) {
	zs->zs_minor = minor;
	membar_producer();
	zsprev->zs_next = zs;
	} else {
	membar_producer();
	zs->zs_minor = minor;
	}

	return (0);
	}

	void
	zfsdev_state_destroy(void *priv)
	{
	zfsdev_state_t *zs = zfsdev_private_get_state(priv);

	ASSERT(zs != NULL);
	ASSERT3S(zs->zs_minor, >, 0);

	/*
	* The last reference to this zfsdev file descriptor is being dropped.
	* We don't have to worry about lookup grabbing this state object, and
	* zfsdev_state_init() will not try to reuse this object until it is
	* invalidated by setting zs_minor to -1. Invalidation must be done
	* last, with a memory barrier to ensure ordering. This lets us avoid
	* taking the global zfsdev state lock around destruction.
	*/
	zfs_onexit_destroy(zs->zs_onexit);
	zfs_zevent_destroy(zs->zs_zevent);
	zs->zs_onexit = NULL;
	zs->zs_zevent = NULL;
	membar_producer();
	zs->zs_minor = -1;
	}

	long
	zfsdev_ioctl_common(uint_t vecnum, zfs_cmd_t *zc, int flag)
	{
	int error, cmd;
	const zfs_ioc_vec_t *vec;
	char *saved_poolname = NULL;
	uint64_t max_nvlist_src_size;
	size_t saved_poolname_len = 0;
	nvlist_t *innvl = NULL;
	fstrans_cookie_t cookie;
	hrtime_t start_time = gethrtime();

	cmd = vecnum;
	error = 0;
	if (vecnum >= sizeof (zfs_ioc_vec) / sizeof (zfs_ioc_vec[0]))
	return (SET_ERROR(ZFS_ERR_IOC_CMD_UNAVAIL));

	vec = &zfs_ioc_vec[vecnum];

	/*
	* The registered ioctl list may be sparse, verify that either
	* a normal or legacy handler are registered.
	*/
	if (vec->zvec_func == NULL && vec->zvec_legacy_func == NULL)
	return (SET_ERROR(ZFS_ERR_IOC_CMD_UNAVAIL));

	zc->zc_iflags = flag & FKIOCTL;
	max_nvlist_src_size = zfs_max_nvlist_src_size_os();
	if (zc->zc_nvlist_src_size > max_nvlist_src_size) {
	/*
	* Make sure the user doesn't pass in an insane value for
	* zc_nvlist_src_size. We have to check, since we will end
	* up allocating that much memory inside of get_nvlist(). This
	* prevents a nefarious user from allocating tons of kernel
	* memory.
	*
	* Also, we return EINVAL instead of ENOMEM here. The reason
	* being that returning ENOMEM from an ioctl() has a special
	* connotation; that the user's size value is too small and
	* needs to be expanded to hold the nvlist. See
	* zcmd_expand_dst_nvlist() for details.
	*/
	error = SET_ERROR(EINVAL); /* User's size too big */

	} else if (zc->zc_nvlist_src_size != 0) {
	error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size,
	zc->zc_iflags, &innvl);
	if (error != 0)
	goto out;
	}

	/*
	* Ensure that all pool/dataset names are valid before we pass down to
	* the lower layers.
	*/
	zc->zc_name[sizeof (zc->zc_name) - 1] = '\0';
	switch (vec->zvec_namecheck) {
	case POOL_NAME:
	if (pool_namecheck(zc->zc_name, NULL, NULL) != 0)
	error = SET_ERROR(EINVAL);
	else
	error = pool_status_check(zc->zc_name,
	vec->zvec_namecheck, vec->zvec_pool_check);
	break;

	case DATASET_NAME:
	if (dataset_namecheck(zc->zc_name, NULL, NULL) != 0)
	error = SET_ERROR(EINVAL);
	else
	error = pool_status_check(zc->zc_name,
	vec->zvec_namecheck, vec->zvec_pool_check);
	break;

	case ENTITY_NAME:
	if (entity_namecheck(zc->zc_name, NULL, NULL) != 0) {
	error = SET_ERROR(EINVAL);
	} else {
	error = pool_status_check(zc->zc_name,
	vec->zvec_namecheck, vec->zvec_pool_check);
	}
	break;

	case NO_NAME:
	break;
	}
	/*
	* Ensure that all input pairs are valid before we pass them down
	* to the lower layers.
	*
	* The vectored functions can use fnvlist_lookup_{type} for any
	* required pairs since zfs_check_input_nvpairs() confirmed that
	* they exist and are of the correct type.
	*/
	if (error == 0 && vec->zvec_func != NULL) {
	error = zfs_check_input_nvpairs(innvl, vec);
	if (error != 0)
	goto out;
	}

	if (error == 0) {
	cookie = spl_fstrans_mark();
	error = vec->zvec_secpolicy(zc, innvl, CRED());
	spl_fstrans_unmark(cookie);
	}

	if (error != 0)
	goto out;

	/* legacy ioctls can modify zc_name */
	/*
	* Can't use kmem_strdup() as we might truncate the string and
	* kmem_strfree() would then free with incorrect size.
	*/
	saved_poolname_len = strlen(zc->zc_name) + 1;
	saved_poolname = kmem_alloc(saved_poolname_len, KM_SLEEP);

	strlcpy(saved_poolname, zc->zc_name, saved_poolname_len);
	saved_poolname[strcspn(saved_poolname, "/@#")] = '\0';

	if (vec->zvec_func != NULL) {
	nvlist_t *outnvl;
	int puterror = 0;
	spa_t *spa;
	nvlist_t *lognv = NULL;

	ASSERT(vec->zvec_legacy_func == NULL);

	/*
	* Add the innvl to the lognv before calling the func,
	* in case the func changes the innvl.
	*/
	if (vec->zvec_allow_log) {
	lognv = fnvlist_alloc();
	fnvlist_add_string(lognv, ZPOOL_HIST_IOCTL,
	vec->zvec_name);
	if (!nvlist_empty(innvl)) {
	fnvlist_add_nvlist(lognv, ZPOOL_HIST_INPUT_NVL,
	innvl);
	}
	}

	outnvl = fnvlist_alloc();
	cookie = spl_fstrans_mark();
	error = vec->zvec_func(zc->zc_name, innvl, outnvl);
	spl_fstrans_unmark(cookie);

	/*
	* Some commands can partially execute, modify state, and still
	* return an error. In these cases, attempt to record what
	* was modified.
	*/
	if ((error == 0 \|\|
	(cmd == ZFS_IOC_CHANNEL_PROGRAM && error != EINVAL)) &&
	vec->zvec_allow_log &&
	spa_open(zc->zc_name, &spa, FTAG) == 0) {
	if (!nvlist_empty(outnvl)) {
	size_t out_size = fnvlist_size(outnvl);
	if (out_size > zfs_history_output_max) {
	fnvlist_add_int64(lognv,
	ZPOOL_HIST_OUTPUT_SIZE, out_size);
	} else {
	fnvlist_add_nvlist(lognv,
	ZPOOL_HIST_OUTPUT_NVL, outnvl);
	}
	}
	if (error != 0) {
	fnvlist_add_int64(lognv, ZPOOL_HIST_ERRNO,
	error);
	}
	fnvlist_add_int64(lognv, ZPOOL_HIST_ELAPSED_NS,
	gethrtime() - start_time);
	(void) spa_history_log_nvl(spa, lognv);
	spa_close(spa, FTAG);
	}
	fnvlist_free(lognv);

	if (!nvlist_empty(outnvl) \|\| zc->zc_nvlist_dst_size != 0) {
	int smusherror = 0;
	if (vec->zvec_smush_outnvlist) {
	smusherror = nvlist_smush(outnvl,
	zc->zc_nvlist_dst_size);
	}
	if (smusherror == 0)
	puterror = put_nvlist(zc, outnvl);
	}

	if (puterror != 0)
	error = puterror;

	nvlist_free(outnvl);
	} else {
	cookie = spl_fstrans_mark();
	error = vec->zvec_legacy_func(zc);
	spl_fstrans_unmark(cookie);
	}

	out:
	nvlist_free(innvl);
	if (error == 0 && vec->zvec_allow_log) {
	char *s = tsd_get(zfs_allow_log_key);
	if (s != NULL)
	kmem_strfree(s);
	(void) tsd_set(zfs_allow_log_key, kmem_strdup(saved_poolname));
	}
	if (saved_poolname != NULL)
	kmem_free(saved_poolname, saved_poolname_len);

	return (error);
	}

	int
	zfs_kmod_init(void)
	{
	int error;

	if ((error = zvol_init()) != 0)
	return (error);

	spa_init(SPA_MODE_READ \| SPA_MODE_WRITE);
	zfs_init();

	zfs_ioctl_init();

	mutex_init(&zfsdev_state_lock, NULL, MUTEX_DEFAULT, NULL);
	zfsdev_state_list = kmem_zalloc(sizeof (zfsdev_state_t), KM_SLEEP);
	zfsdev_state_list->zs_minor = -1;

	if ((error = zfsdev_attach()) != 0)
	goto out;

	tsd_create(&zfs_fsyncer_key, NULL);
	tsd_create(&rrw_tsd_key, rrw_tsd_destroy);
	tsd_create(&zfs_allow_log_key, zfs_allow_log_destroy);

	return (0);
	out:
	zfs_fini();
	spa_fini();
	zvol_fini();

	return (error);
	}

	void
	zfs_kmod_fini(void)
	{
	zfsdev_state_t zs, zsnext = NULL;

	zfsdev_detach();

	mutex_destroy(&zfsdev_state_lock);

	for (zs = zfsdev_state_list; zs != NULL; zs = zsnext) {
	zsnext = zs->zs_next;
	if (zs->zs_onexit)
	zfs_onexit_destroy(zs->zs_onexit);
	if (zs->zs_zevent)
	zfs_zevent_destroy(zs->zs_zevent);
	kmem_free(zs, sizeof (zfsdev_state_t));
	}

	zfs_ereport_taskq_fini(); /* run before zfs_fini() on Linux */
	zfs_fini();
	spa_fini();
	zvol_fini();

	tsd_destroy(&zfs_fsyncer_key);
	tsd_destroy(&rrw_tsd_key);
	tsd_destroy(&zfs_allow_log_key);
	}

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM(zfs, zfs_, max_nvlist_src_size, ULONG, ZMOD_RW,
	"Maximum size in bytes allowed for src nvlist passed with ZFS ioctls");

	ZFS_MODULE_PARAM(zfs, zfs_, history_output_max, ULONG, ZMOD_RW,
	"Maximum size in bytes of ZFS ioctl output that will be logged");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/zfs/zio.c b/sys/contrib/openzfs/module/zfs/zio.c
	index c016fa323b41..2079a1e0a9d8 100644
	--- a/sys/contrib/openzfs/module/zfs/zio.c
	+++ b/sys/contrib/openzfs/module/zfs/zio.c
	@@ -1,5054 +1,5054 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
	* Copyright (c) 2011, 2020 by Delphix. All rights reserved.
	* Copyright (c) 2011 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2017, Intel Corporation.
	* Copyright (c) 2019, Klara Inc.
	* Copyright (c) 2019, Allan Jude
	* Copyright (c) 2021, Datto, Inc.
	*/

	#include <sys/sysmacros.h>
	#include <sys/zfs_context.h>
	#include <sys/fm/fs/zfs.h>
	#include <sys/spa.h>
	#include <sys/txg.h>
	#include <sys/spa_impl.h>
	#include <sys/vdev_impl.h>
	#include <sys/vdev_trim.h>
	#include <sys/zio_impl.h>
	#include <sys/zio_compress.h>
	#include <sys/zio_checksum.h>
	#include <sys/dmu_objset.h>
	#include <sys/arc.h>
	#include <sys/ddt.h>
	#include <sys/blkptr.h>
	#include <sys/zfeature.h>
	#include <sys/dsl_scan.h>
	#include <sys/metaslab_impl.h>
	#include <sys/time.h>
	#include <sys/trace_zfs.h>
	#include <sys/abd.h>
	#include <sys/dsl_crypt.h>
	#include <cityhash.h>

	/*
	* ==========================================================================
	* I/O type descriptions
	* ==========================================================================
	*/
	const char *zio_type_name[ZIO_TYPES] = {
	/*
	* Note: Linux kernel thread name length is limited
	* so these names will differ from upstream open zfs.
	*/
	"z_null", "z_rd", "z_wr", "z_fr", "z_cl", "z_ioctl", "z_trim"
	};

	int zio_dva_throttle_enabled = B_TRUE;
	int zio_deadman_log_all = B_FALSE;

	/*
	* ==========================================================================
	* I/O kmem caches
	* ==========================================================================
	*/
	kmem_cache_t *zio_cache;
	kmem_cache_t *zio_link_cache;
	kmem_cache_t *zio_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT];
	kmem_cache_t *zio_data_buf_cache[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT];
	#if defined(ZFS_DEBUG) && !defined(_KERNEL)
	uint64_t zio_buf_cache_allocs[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT];
	uint64_t zio_buf_cache_frees[SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT];
	#endif

	/* Mark IOs as "slow" if they take longer than 30 seconds */
	int zio_slow_io_ms = (30 * MILLISEC);

	#define BP_SPANB(indblkshift, level) \
	(((uint64_t)1) << ((level) * ((indblkshift) - SPA_BLKPTRSHIFT)))
	#define COMPARE_META_LEVEL 0x80000000ul
	/*
	* The following actions directly effect the spa's sync-to-convergence logic.
	* The values below define the sync pass when we start performing the action.
	* Care should be taken when changing these values as they directly impact
	* spa_sync() performance. Tuning these values may introduce subtle performance
	* pathologies and should only be done in the context of performance analysis.
	* These tunables will eventually be removed and replaced with #defines once
	* enough analysis has been done to determine optimal values.
	*
	* The 'zfs_sync_pass_deferred_free' pass must be greater than 1 to ensure that
	* regular blocks are not deferred.
	*
	* Starting in sync pass 8 (zfs_sync_pass_dont_compress), we disable
	* compression (including of metadata). In practice, we don't have this
	* many sync passes, so this has no effect.
	*
	* The original intent was that disabling compression would help the sync
	* passes to converge. However, in practice disabling compression increases
	* the average number of sync passes, because when we turn compression off, a
	* lot of block's size will change and thus we have to re-allocate (not
	* overwrite) them. It also increases the number of 128KB allocations (e.g.
	* for indirect blocks and spacemaps) because these will not be compressed.
	* The 128K allocations are especially detrimental to performance on highly
	* fragmented systems, which may have very few free segments of this size,
	* and may need to load new metaslabs to satisfy 128K allocations.
	*/
	int zfs_sync_pass_deferred_free = 2; /* defer frees starting in this pass */
	int zfs_sync_pass_dont_compress = 8; /* don't compress starting in this pass */
	int zfs_sync_pass_rewrite = 2; /* rewrite new bps starting in this pass */

	/*
	* An allocating zio is one that either currently has the DVA allocate
	* stage set or will have it later in its lifetime.
	*/
	#define IO_IS_ALLOCATING(zio) ((zio)->io_orig_pipeline & ZIO_STAGE_DVA_ALLOCATE)

	/*
	* Enable smaller cores by excluding metadata
	* allocations as well.
	*/
	int zio_exclude_metadata = 0;
	int zio_requeue_io_start_cut_in_line = 1;

	#ifdef ZFS_DEBUG
	int zio_buf_debug_limit = 16384;
	#else
	int zio_buf_debug_limit = 0;
	#endif

	static inline void __zio_execute(zio_t *zio);

	static void zio_taskq_dispatch(zio_t *, zio_taskq_type_t, boolean_t);

	void
	zio_init(void)
	{
	size_t c;

	zio_cache = kmem_cache_create("zio_cache",
	sizeof (zio_t), 0, NULL, NULL, NULL, NULL, NULL, 0);
	zio_link_cache = kmem_cache_create("zio_link_cache",
	sizeof (zio_link_t), 0, NULL, NULL, NULL, NULL, NULL, 0);

	/*
	* For small buffers, we want a cache for each multiple of
	* SPA_MINBLOCKSIZE. For larger buffers, we want a cache
	* for each quarter-power of 2.
	*/
	for (c = 0; c < SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT; c++) {
	size_t size = (c + 1) << SPA_MINBLOCKSHIFT;
	size_t p2 = size;
	size_t align = 0;
	size_t data_cflags, cflags;

	data_cflags = KMC_NODEBUG;
	cflags = (zio_exclude_metadata \|\| size > zio_buf_debug_limit) ?
	KMC_NODEBUG : 0;

	#if defined(_ILP32) && defined(_KERNEL)
	/*
	* Cache size limited to 1M on 32-bit platforms until ARC
	* buffers no longer require virtual address space.
	*/
	if (size > zfs_max_recordsize)
	break;
	#endif

	while (!ISP2(p2))
	p2 &= p2 - 1;

	#ifndef _KERNEL
	/*
	* If we are using watchpoints, put each buffer on its own page,
	* to eliminate the performance overhead of trapping to the
	* kernel when modifying a non-watched buffer that shares the
	* page with a watched buffer.
	*/
	if (arc_watch && !IS_P2ALIGNED(size, PAGESIZE))
	continue;
	/*
	* Here's the problem - on 4K native devices in userland on
	* Linux using O_DIRECT, buffers must be 4K aligned or I/O
	* will fail with EINVAL, causing zdb (and others) to coredump.
	* Since userland probably doesn't need optimized buffer caches,
	* we just force 4K alignment on everything.
	*/
	align = 8 * SPA_MINBLOCKSIZE;
	#else
	if (size < PAGESIZE) {
	align = SPA_MINBLOCKSIZE;
	} else if (IS_P2ALIGNED(size, p2 >> 2)) {
	align = PAGESIZE;
	}
	#endif

	if (align != 0) {
	char name[36];
	if (cflags == data_cflags) {
	/*
	* Resulting kmem caches would be identical.
	* Save memory by creating only one.
	*/
	(void) snprintf(name, sizeof (name),
	"zio_buf_comb_%lu", (ulong_t)size);
	zio_buf_cache[c] = kmem_cache_create(name,
	size, align, NULL, NULL, NULL, NULL, NULL,
	cflags);
	zio_data_buf_cache[c] = zio_buf_cache[c];
	continue;
	}
	(void) snprintf(name, sizeof (name), "zio_buf_%lu",
	(ulong_t)size);
	zio_buf_cache[c] = kmem_cache_create(name, size,
	align, NULL, NULL, NULL, NULL, NULL, cflags);

	(void) snprintf(name, sizeof (name), "zio_data_buf_%lu",
	(ulong_t)size);
	zio_data_buf_cache[c] = kmem_cache_create(name, size,
	align, NULL, NULL, NULL, NULL, NULL, data_cflags);
	}
	}

	while (--c != 0) {
	ASSERT(zio_buf_cache[c] != NULL);
	if (zio_buf_cache[c - 1] == NULL)
	zio_buf_cache[c - 1] = zio_buf_cache[c];

	ASSERT(zio_data_buf_cache[c] != NULL);
	if (zio_data_buf_cache[c - 1] == NULL)
	zio_data_buf_cache[c - 1] = zio_data_buf_cache[c];
	}

	zio_inject_init();

	lz4_init();
	}

	void
	zio_fini(void)
	{
	size_t n = SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT;

	#if defined(ZFS_DEBUG) && !defined(_KERNEL)
	for (size_t i = 0; i < n; i++) {
	if (zio_buf_cache_allocs[i] != zio_buf_cache_frees[i])
	(void) printf("zio_fini: [%d] %llu != %llu\n",
	(int)((i + 1) << SPA_MINBLOCKSHIFT),
	(long long unsigned)zio_buf_cache_allocs[i],
	(long long unsigned)zio_buf_cache_frees[i]);
	}
	#endif

	/*
	* The same kmem cache can show up multiple times in both zio_buf_cache
	* and zio_data_buf_cache. Do a wasteful but trivially correct scan to
	* sort it out.
	*/
	for (size_t i = 0; i < n; i++) {
	kmem_cache_t *cache = zio_buf_cache[i];
	if (cache == NULL)
	continue;
	for (size_t j = i; j < n; j++) {
	if (cache == zio_buf_cache[j])
	zio_buf_cache[j] = NULL;
	if (cache == zio_data_buf_cache[j])
	zio_data_buf_cache[j] = NULL;
	}
	kmem_cache_destroy(cache);
	}

	for (size_t i = 0; i < n; i++) {
	kmem_cache_t *cache = zio_data_buf_cache[i];
	if (cache == NULL)
	continue;
	for (size_t j = i; j < n; j++) {
	if (cache == zio_data_buf_cache[j])
	zio_data_buf_cache[j] = NULL;
	}
	kmem_cache_destroy(cache);
	}

	for (size_t i = 0; i < n; i++) {
	VERIFY3P(zio_buf_cache[i], ==, NULL);
	VERIFY3P(zio_data_buf_cache[i], ==, NULL);
	}

	kmem_cache_destroy(zio_link_cache);
	kmem_cache_destroy(zio_cache);

	zio_inject_fini();

	lz4_fini();
	}

	/*
	* ==========================================================================
	* Allocate and free I/O buffers
	* ==========================================================================
	*/

	/*
	* Use zio_buf_alloc to allocate ZFS metadata. This data will appear in a
	* crashdump if the kernel panics, so use it judiciously. Obviously, it's
	* useful to inspect ZFS metadata, but if possible, we should avoid keeping
	* excess / transient data in-core during a crashdump.
	*/
	void *
	zio_buf_alloc(size_t size)
	{
	size_t c = (size - 1) >> SPA_MINBLOCKSHIFT;

	VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT);
	#if defined(ZFS_DEBUG) && !defined(_KERNEL)
	atomic_add_64(&zio_buf_cache_allocs[c], 1);
	#endif

	return (kmem_cache_alloc(zio_buf_cache[c], KM_PUSHPAGE));
	}

	/*
	* Use zio_data_buf_alloc to allocate data. The data will not appear in a
	* crashdump if the kernel panics. This exists so that we will limit the amount
	* of ZFS data that shows up in a kernel crashdump. (Thus reducing the amount
	* of kernel heap dumped to disk when the kernel panics)
	*/
	void *
	zio_data_buf_alloc(size_t size)
	{
	size_t c = (size - 1) >> SPA_MINBLOCKSHIFT;

	VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT);

	return (kmem_cache_alloc(zio_data_buf_cache[c], KM_PUSHPAGE));
	}

	void
	zio_buf_free(void *buf, size_t size)
	{
	size_t c = (size - 1) >> SPA_MINBLOCKSHIFT;

	VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT);
	#if defined(ZFS_DEBUG) && !defined(_KERNEL)
	atomic_add_64(&zio_buf_cache_frees[c], 1);
	#endif

	kmem_cache_free(zio_buf_cache[c], buf);
	}

	void
	zio_data_buf_free(void *buf, size_t size)
	{
	size_t c = (size - 1) >> SPA_MINBLOCKSHIFT;

	VERIFY3U(c, <, SPA_MAXBLOCKSIZE >> SPA_MINBLOCKSHIFT);

	kmem_cache_free(zio_data_buf_cache[c], buf);
	}

	static void
	zio_abd_free(void *abd, size_t size)
	{
	abd_free((abd_t *)abd);
	}

	/*
	* ==========================================================================
	* Push and pop I/O transform buffers
	* ==========================================================================
	*/
	void
	zio_push_transform(zio_t zio, abd_t data, uint64_t size, uint64_t bufsize,
	zio_transform_func_t *transform)
	{
	zio_transform_t *zt = kmem_alloc(sizeof (zio_transform_t), KM_SLEEP);

	zt->zt_orig_abd = zio->io_abd;
	zt->zt_orig_size = zio->io_size;
	zt->zt_bufsize = bufsize;
	zt->zt_transform = transform;

	zt->zt_next = zio->io_transform_stack;
	zio->io_transform_stack = zt;

	zio->io_abd = data;
	zio->io_size = size;
	}

	void
	zio_pop_transforms(zio_t *zio)
	{
	zio_transform_t *zt;

	while ((zt = zio->io_transform_stack) != NULL) {
	if (zt->zt_transform != NULL)
	zt->zt_transform(zio,
	zt->zt_orig_abd, zt->zt_orig_size);

	if (zt->zt_bufsize != 0)
	abd_free(zio->io_abd);

	zio->io_abd = zt->zt_orig_abd;
	zio->io_size = zt->zt_orig_size;
	zio->io_transform_stack = zt->zt_next;

	kmem_free(zt, sizeof (zio_transform_t));
	}
	}

	/*
	* ==========================================================================
	* I/O transform callbacks for subblocks, decompression, and decryption
	* ==========================================================================
	*/
	static void
	zio_subblock(zio_t zio, abd_t data, uint64_t size)
	{
	ASSERT(zio->io_size > size);

	if (zio->io_type == ZIO_TYPE_READ)
	abd_copy(data, zio->io_abd, size);
	}

	static void
	zio_decompress(zio_t zio, abd_t data, uint64_t size)
	{
	if (zio->io_error == 0) {
	void *tmp = abd_borrow_buf(data, size);
	int ret = zio_decompress_data(BP_GET_COMPRESS(zio->io_bp),
	zio->io_abd, tmp, zio->io_size, size,
	&zio->io_prop.zp_complevel);
	abd_return_buf_copy(data, tmp, size);

	if (zio_injection_enabled && ret == 0)
	ret = zio_handle_fault_injection(zio, EINVAL);

	if (ret != 0)
	zio->io_error = SET_ERROR(EIO);
	}
	}

	static void
	zio_decrypt(zio_t zio, abd_t data, uint64_t size)
	{
	int ret;
	void *tmp;
	blkptr_t *bp = zio->io_bp;
	spa_t *spa = zio->io_spa;
	uint64_t dsobj = zio->io_bookmark.zb_objset;
	uint64_t lsize = BP_GET_LSIZE(bp);
	dmu_object_type_t ot = BP_GET_TYPE(bp);
	uint8_t salt[ZIO_DATA_SALT_LEN];
	uint8_t iv[ZIO_DATA_IV_LEN];
	uint8_t mac[ZIO_DATA_MAC_LEN];
	boolean_t no_crypt = B_FALSE;

	ASSERT(BP_USES_CRYPT(bp));
	ASSERT3U(size, !=, 0);

	if (zio->io_error != 0)
	return;

	/*
	* Verify the cksum of MACs stored in an indirect bp. It will always
	* be possible to verify this since it does not require an encryption
	* key.
	*/
	if (BP_HAS_INDIRECT_MAC_CKSUM(bp)) {
	zio_crypt_decode_mac_bp(bp, mac);

	if (BP_GET_COMPRESS(bp) != ZIO_COMPRESS_OFF) {
	/*
	* We haven't decompressed the data yet, but
	* zio_crypt_do_indirect_mac_checksum() requires
	* decompressed data to be able to parse out the MACs
	* from the indirect block. We decompress it now and
	* throw away the result after we are finished.
	*/
	tmp = zio_buf_alloc(lsize);
	ret = zio_decompress_data(BP_GET_COMPRESS(bp),
	zio->io_abd, tmp, zio->io_size, lsize,
	&zio->io_prop.zp_complevel);
	if (ret != 0) {
	ret = SET_ERROR(EIO);
	goto error;
	}
	ret = zio_crypt_do_indirect_mac_checksum(B_FALSE,
	tmp, lsize, BP_SHOULD_BYTESWAP(bp), mac);
	zio_buf_free(tmp, lsize);
	} else {
	ret = zio_crypt_do_indirect_mac_checksum_abd(B_FALSE,
	zio->io_abd, size, BP_SHOULD_BYTESWAP(bp), mac);
	}
	abd_copy(data, zio->io_abd, size);

	if (zio_injection_enabled && ot != DMU_OT_DNODE && ret == 0) {
	ret = zio_handle_decrypt_injection(spa,
	&zio->io_bookmark, ot, ECKSUM);
	}
	if (ret != 0)
	goto error;

	return;
	}

	/*
	* If this is an authenticated block, just check the MAC. It would be
	* nice to separate this out into its own flag, but for the moment
	* enum zio_flag is out of bits.
	*/
	if (BP_IS_AUTHENTICATED(bp)) {
	if (ot == DMU_OT_OBJSET) {
	ret = spa_do_crypt_objset_mac_abd(B_FALSE, spa,
	dsobj, zio->io_abd, size, BP_SHOULD_BYTESWAP(bp));
	} else {
	zio_crypt_decode_mac_bp(bp, mac);
	ret = spa_do_crypt_mac_abd(B_FALSE, spa, dsobj,
	zio->io_abd, size, mac);
	if (zio_injection_enabled && ret == 0) {
	ret = zio_handle_decrypt_injection(spa,
	&zio->io_bookmark, ot, ECKSUM);
	}
	}
	abd_copy(data, zio->io_abd, size);

	if (ret != 0)
	goto error;

	return;
	}

	zio_crypt_decode_params_bp(bp, salt, iv);

	if (ot == DMU_OT_INTENT_LOG) {
	tmp = abd_borrow_buf_copy(zio->io_abd, sizeof (zil_chain_t));
	zio_crypt_decode_mac_zil(tmp, mac);
	abd_return_buf(zio->io_abd, tmp, sizeof (zil_chain_t));
	} else {
	zio_crypt_decode_mac_bp(bp, mac);
	}

	ret = spa_do_crypt_abd(B_FALSE, spa, &zio->io_bookmark, BP_GET_TYPE(bp),
	BP_GET_DEDUP(bp), BP_SHOULD_BYTESWAP(bp), salt, iv, mac, size, data,
	zio->io_abd, &no_crypt);
	if (no_crypt)
	abd_copy(data, zio->io_abd, size);

	if (ret != 0)
	goto error;

	return;

	error:
	/* assert that the key was found unless this was speculative */
	ASSERT(ret != EACCES \|\| (zio->io_flags & ZIO_FLAG_SPECULATIVE));

	/*
	* If there was a decryption / authentication error return EIO as
	* the io_error. If this was not a speculative zio, create an ereport.
	*/
	if (ret == ECKSUM) {
	zio->io_error = SET_ERROR(EIO);
	if ((zio->io_flags & ZIO_FLAG_SPECULATIVE) == 0) {
	spa_log_error(spa, &zio->io_bookmark);
	(void) zfs_ereport_post(FM_EREPORT_ZFS_AUTHENTICATION,
	spa, NULL, &zio->io_bookmark, zio, 0);
	}
	} else {
	zio->io_error = ret;
	}
	}

	/*
	* ==========================================================================
	* I/O parent/child relationships and pipeline interlocks
	* ==========================================================================
	*/
	zio_t *
	zio_walk_parents(zio_t cio, zio_link_t *zl)
	{
	list_t *pl = &cio->io_parent_list;

	zl = (zl == NULL) ? list_head(pl) : list_next(pl, *zl);
	if (*zl == NULL)
	return (NULL);

	ASSERT((*zl)->zl_child == cio);
	return ((*zl)->zl_parent);
	}

	zio_t *
	zio_walk_children(zio_t pio, zio_link_t *zl)
	{
	list_t *cl = &pio->io_child_list;

	ASSERT(MUTEX_HELD(&pio->io_lock));

	zl = (zl == NULL) ? list_head(cl) : list_next(cl, *zl);
	if (*zl == NULL)
	return (NULL);

	ASSERT((*zl)->zl_parent == pio);
	return ((*zl)->zl_child);
	}

	zio_t *
	zio_unique_parent(zio_t *cio)
	{
	zio_link_t *zl = NULL;
	zio_t *pio = zio_walk_parents(cio, &zl);

	VERIFY3P(zio_walk_parents(cio, &zl), ==, NULL);
	return (pio);
	}

	void
	zio_add_child(zio_t pio, zio_t cio)
	{
	zio_link_t *zl = kmem_cache_alloc(zio_link_cache, KM_SLEEP);

	/*
	* Logical I/Os can have logical, gang, or vdev children.
	* Gang I/Os can have gang or vdev children.
	* Vdev I/Os can only have vdev children.
	* The following ASSERT captures all of these constraints.
	*/
	ASSERT3S(cio->io_child_type, <=, pio->io_child_type);

	zl->zl_parent = pio;
	zl->zl_child = cio;

	mutex_enter(&pio->io_lock);
	mutex_enter(&cio->io_lock);

	ASSERT(pio->io_state[ZIO_WAIT_DONE] == 0);

	for (int w = 0; w < ZIO_WAIT_TYPES; w++)
	pio->io_children[cio->io_child_type][w] += !cio->io_state[w];

	list_insert_head(&pio->io_child_list, zl);
	list_insert_head(&cio->io_parent_list, zl);

	pio->io_child_count++;
	cio->io_parent_count++;

	mutex_exit(&cio->io_lock);
	mutex_exit(&pio->io_lock);
	}

	static void
	zio_remove_child(zio_t pio, zio_t cio, zio_link_t *zl)
	{
	ASSERT(zl->zl_parent == pio);
	ASSERT(zl->zl_child == cio);

	mutex_enter(&pio->io_lock);
	mutex_enter(&cio->io_lock);

	list_remove(&pio->io_child_list, zl);
	list_remove(&cio->io_parent_list, zl);

	pio->io_child_count--;
	cio->io_parent_count--;

	mutex_exit(&cio->io_lock);
	mutex_exit(&pio->io_lock);
	kmem_cache_free(zio_link_cache, zl);
	}

	static boolean_t
	zio_wait_for_children(zio_t *zio, uint8_t childbits, enum zio_wait_type wait)
	{
	boolean_t waiting = B_FALSE;

	mutex_enter(&zio->io_lock);
	ASSERT(zio->io_stall == NULL);
	for (int c = 0; c < ZIO_CHILD_TYPES; c++) {
	if (!(ZIO_CHILD_BIT_IS_SET(childbits, c)))
	continue;

	uint64_t *countp = &zio->io_children[c][wait];
	if (*countp != 0) {
	zio->io_stage >>= 1;
	ASSERT3U(zio->io_stage, !=, ZIO_STAGE_OPEN);
	zio->io_stall = countp;
	waiting = B_TRUE;
	break;
	}
	}
	mutex_exit(&zio->io_lock);
	return (waiting);
	}

	__attribute__((always_inline))
	static inline void
	zio_notify_parent(zio_t pio, zio_t zio, enum zio_wait_type wait,
	zio_t **next_to_executep)
	{
	uint64_t *countp = &pio->io_children[zio->io_child_type][wait];
	int *errorp = &pio->io_child_error[zio->io_child_type];

	mutex_enter(&pio->io_lock);
	if (zio->io_error && !(zio->io_flags & ZIO_FLAG_DONT_PROPAGATE))
	errorp = zio_worst_error(errorp, zio->io_error);
	pio->io_reexecute \|= zio->io_reexecute;
	ASSERT3U(*countp, >, 0);

	(*countp)--;

	if (*countp == 0 && pio->io_stall == countp) {
	zio_taskq_type_t type =
	pio->io_stage < ZIO_STAGE_VDEV_IO_START ? ZIO_TASKQ_ISSUE :
	ZIO_TASKQ_INTERRUPT;
	pio->io_stall = NULL;
	mutex_exit(&pio->io_lock);

	/*
	* If we can tell the caller to execute this parent next, do
	* so. Otherwise dispatch the parent zio as its own task.
	*
	* Having the caller execute the parent when possible reduces
	* locking on the zio taskq's, reduces context switch
	* overhead, and has no recursion penalty. Note that one
	* read from disk typically causes at least 3 zio's: a
	* zio_null(), the logical zio_read(), and then a physical
	* zio. When the physical ZIO completes, we are able to call
	* zio_done() on all 3 of these zio's from one invocation of
	* zio_execute() by returning the parent back to
	* zio_execute(). Since the parent isn't executed until this
	* thread returns back to zio_execute(), the caller should do
	* so promptly.
	*
	* In other cases, dispatching the parent prevents
	* overflowing the stack when we have deeply nested
	* parent-child relationships, as we do with the "mega zio"
	* of writes for spa_sync(), and the chain of ZIL blocks.
	*/
	if (next_to_executep != NULL && *next_to_executep == NULL) {
	*next_to_executep = pio;
	} else {
	zio_taskq_dispatch(pio, type, B_FALSE);
	}
	} else {
	mutex_exit(&pio->io_lock);
	}
	}

	static void
	zio_inherit_child_errors(zio_t *zio, enum zio_child c)
	{
	if (zio->io_child_error[c] != 0 && zio->io_error == 0)
	zio->io_error = zio->io_child_error[c];
	}

	int
	zio_bookmark_compare(const void x1, const void x2)
	{
	const zio_t *z1 = x1;
	const zio_t *z2 = x2;

	if (z1->io_bookmark.zb_objset < z2->io_bookmark.zb_objset)
	return (-1);
	if (z1->io_bookmark.zb_objset > z2->io_bookmark.zb_objset)
	return (1);

	if (z1->io_bookmark.zb_object < z2->io_bookmark.zb_object)
	return (-1);
	if (z1->io_bookmark.zb_object > z2->io_bookmark.zb_object)
	return (1);

	if (z1->io_bookmark.zb_level < z2->io_bookmark.zb_level)
	return (-1);
	if (z1->io_bookmark.zb_level > z2->io_bookmark.zb_level)
	return (1);

	if (z1->io_bookmark.zb_blkid < z2->io_bookmark.zb_blkid)
	return (-1);
	if (z1->io_bookmark.zb_blkid > z2->io_bookmark.zb_blkid)
	return (1);

	if (z1 < z2)
	return (-1);
	if (z1 > z2)
	return (1);

	return (0);
	}

	/*
	* ==========================================================================
	* Create the various types of I/O (read, write, free, etc)
	* ==========================================================================
	*/
	static zio_t *
	zio_create(zio_t pio, spa_t spa, uint64_t txg, const blkptr_t *bp,
	abd_t data, uint64_t lsize, uint64_t psize, zio_done_func_t done,
	void *private, zio_type_t type, zio_priority_t priority,
	enum zio_flag flags, vdev_t *vd, uint64_t offset,
	const zbookmark_phys_t *zb, enum zio_stage stage,
	enum zio_stage pipeline)
	{
	zio_t *zio;

	IMPLY(type != ZIO_TYPE_TRIM, psize <= SPA_MAXBLOCKSIZE);
	ASSERT(P2PHASE(psize, SPA_MINBLOCKSIZE) == 0);
	ASSERT(P2PHASE(offset, SPA_MINBLOCKSIZE) == 0);

	ASSERT(!vd \|\| spa_config_held(spa, SCL_STATE_ALL, RW_READER));
	ASSERT(!bp \|\| !(flags & ZIO_FLAG_CONFIG_WRITER));
	ASSERT(vd \|\| stage == ZIO_STAGE_OPEN);

	IMPLY(lsize != psize, (flags & ZIO_FLAG_RAW_COMPRESS) != 0);

	zio = kmem_cache_alloc(zio_cache, KM_SLEEP);
	bzero(zio, sizeof (zio_t));

	mutex_init(&zio->io_lock, NULL, MUTEX_NOLOCKDEP, NULL);
	cv_init(&zio->io_cv, NULL, CV_DEFAULT, NULL);

	list_create(&zio->io_parent_list, sizeof (zio_link_t),
	offsetof(zio_link_t, zl_parent_node));
	list_create(&zio->io_child_list, sizeof (zio_link_t),
	offsetof(zio_link_t, zl_child_node));
	metaslab_trace_init(&zio->io_alloc_list);

	if (vd != NULL)
	zio->io_child_type = ZIO_CHILD_VDEV;
	else if (flags & ZIO_FLAG_GANG_CHILD)
	zio->io_child_type = ZIO_CHILD_GANG;
	else if (flags & ZIO_FLAG_DDT_CHILD)
	zio->io_child_type = ZIO_CHILD_DDT;
	else
	zio->io_child_type = ZIO_CHILD_LOGICAL;

	if (bp != NULL) {
	zio->io_bp = (blkptr_t *)bp;
	zio->io_bp_copy = *bp;
	zio->io_bp_orig = *bp;
	if (type != ZIO_TYPE_WRITE \|\|
	zio->io_child_type == ZIO_CHILD_DDT)
	zio->io_bp = &zio->io_bp_copy; /* so caller can free */
	if (zio->io_child_type == ZIO_CHILD_LOGICAL)
	zio->io_logical = zio;
	if (zio->io_child_type > ZIO_CHILD_GANG && BP_IS_GANG(bp))
	pipeline \|= ZIO_GANG_STAGES;
	}

	zio->io_spa = spa;
	zio->io_txg = txg;
	zio->io_done = done;
	zio->io_private = private;
	zio->io_type = type;
	zio->io_priority = priority;
	zio->io_vd = vd;
	zio->io_offset = offset;
	zio->io_orig_abd = zio->io_abd = data;
	zio->io_orig_size = zio->io_size = psize;
	zio->io_lsize = lsize;
	zio->io_orig_flags = zio->io_flags = flags;
	zio->io_orig_stage = zio->io_stage = stage;
	zio->io_orig_pipeline = zio->io_pipeline = pipeline;
	zio->io_pipeline_trace = ZIO_STAGE_OPEN;

	zio->io_state[ZIO_WAIT_READY] = (stage >= ZIO_STAGE_READY);
	zio->io_state[ZIO_WAIT_DONE] = (stage >= ZIO_STAGE_DONE);

	if (zb != NULL)
	zio->io_bookmark = *zb;

	if (pio != NULL) {
	zio->io_metaslab_class = pio->io_metaslab_class;
	if (zio->io_logical == NULL)
	zio->io_logical = pio->io_logical;
	if (zio->io_child_type == ZIO_CHILD_GANG)
	zio->io_gang_leader = pio->io_gang_leader;
	zio_add_child(pio, zio);
	}

	taskq_init_ent(&zio->io_tqent);

	return (zio);
	}

	static void
	zio_destroy(zio_t *zio)
	{
	metaslab_trace_fini(&zio->io_alloc_list);
	list_destroy(&zio->io_parent_list);
	list_destroy(&zio->io_child_list);
	mutex_destroy(&zio->io_lock);
	cv_destroy(&zio->io_cv);
	kmem_cache_free(zio_cache, zio);
	}

	zio_t *
	zio_null(zio_t pio, spa_t spa, vdev_t vd, zio_done_func_t done,
	void *private, enum zio_flag flags)
	{
	zio_t *zio;

	zio = zio_create(pio, spa, 0, NULL, NULL, 0, 0, done, private,
	ZIO_TYPE_NULL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL,
	ZIO_STAGE_OPEN, ZIO_INTERLOCK_PIPELINE);

	return (zio);
	}

	zio_t *
	zio_root(spa_t spa, zio_done_func_t done, void *private, enum zio_flag flags)
	{
	return (zio_null(NULL, spa, NULL, done, private, flags));
	}

	static int
	zfs_blkptr_verify_log(spa_t spa, const blkptr_t bp,
	enum blk_verify_flag blk_verify, const char *fmt, ...)
	{
	va_list adx;
	char buf[256];

	va_start(adx, fmt);
	(void) vsnprintf(buf, sizeof (buf), fmt, adx);
	va_end(adx);

	switch (blk_verify) {
	case BLK_VERIFY_HALT:
	dprintf_bp(bp, "blkptr at %p dprintf_bp():", bp);
	zfs_panic_recover("%s: %s", spa_name(spa), buf);
	break;
	case BLK_VERIFY_LOG:
	zfs_dbgmsg("%s: %s", spa_name(spa), buf);
	break;
	case BLK_VERIFY_ONLY:
	break;
	}

	return (1);
	}

	/*
	* Verify the block pointer fields contain reasonable values. This means
	* it only contains known object types, checksum/compression identifiers,
	* block sizes within the maximum allowed limits, valid DVAs, etc.
	*
	* If everything checks out B_TRUE is returned. The zfs_blkptr_verify
	* argument controls the behavior when an invalid field is detected.
	*
	* Modes for zfs_blkptr_verify:
	* 1) BLK_VERIFY_ONLY (evaluate the block)
	* 2) BLK_VERIFY_LOG (evaluate the block and log problems)
	* 3) BLK_VERIFY_HALT (call zfs_panic_recover on error)
	*/
	boolean_t
	zfs_blkptr_verify(spa_t spa, const blkptr_t bp, boolean_t config_held,
	enum blk_verify_flag blk_verify)
	{
	int errors = 0;

	if (!DMU_OT_IS_VALID(BP_GET_TYPE(bp))) {
	errors += zfs_blkptr_verify_log(spa, bp, blk_verify,
	"blkptr at %p has invalid TYPE %llu",
	bp, (longlong_t)BP_GET_TYPE(bp));
	}
	if (BP_GET_CHECKSUM(bp) >= ZIO_CHECKSUM_FUNCTIONS \|\|
	BP_GET_CHECKSUM(bp) <= ZIO_CHECKSUM_ON) {
	errors += zfs_blkptr_verify_log(spa, bp, blk_verify,
	"blkptr at %p has invalid CHECKSUM %llu",
	bp, (longlong_t)BP_GET_CHECKSUM(bp));
	}
	if (BP_GET_COMPRESS(bp) >= ZIO_COMPRESS_FUNCTIONS \|\|
	BP_GET_COMPRESS(bp) <= ZIO_COMPRESS_ON) {
	errors += zfs_blkptr_verify_log(spa, bp, blk_verify,
	"blkptr at %p has invalid COMPRESS %llu",
	bp, (longlong_t)BP_GET_COMPRESS(bp));
	}
	if (BP_GET_LSIZE(bp) > SPA_MAXBLOCKSIZE) {
	errors += zfs_blkptr_verify_log(spa, bp, blk_verify,
	"blkptr at %p has invalid LSIZE %llu",
	bp, (longlong_t)BP_GET_LSIZE(bp));
	}
	if (BP_GET_PSIZE(bp) > SPA_MAXBLOCKSIZE) {
	errors += zfs_blkptr_verify_log(spa, bp, blk_verify,
	"blkptr at %p has invalid PSIZE %llu",
	bp, (longlong_t)BP_GET_PSIZE(bp));
	}

	if (BP_IS_EMBEDDED(bp)) {
	if (BPE_GET_ETYPE(bp) >= NUM_BP_EMBEDDED_TYPES) {
	errors += zfs_blkptr_verify_log(spa, bp, blk_verify,
	"blkptr at %p has invalid ETYPE %llu",
	bp, (longlong_t)BPE_GET_ETYPE(bp));
	}
	}

	/*
	* Do not verify individual DVAs if the config is not trusted. This
	* will be done once the zio is executed in vdev_mirror_map_alloc.
	*/
	if (!spa->spa_trust_config)
	return (errors == 0);

	if (!config_held)
	spa_config_enter(spa, SCL_VDEV, bp, RW_READER);
	else
	ASSERT(spa_config_held(spa, SCL_VDEV, RW_WRITER));
	/*
	* Pool-specific checks.
	*
	* Note: it would be nice to verify that the blk_birth and
	* BP_PHYSICAL_BIRTH() are not too large. However, spa_freeze()
	* allows the birth time of log blocks (and dmu_sync()-ed blocks
	* that are in the log) to be arbitrarily large.
	*/
	for (int i = 0; i < BP_GET_NDVAS(bp); i++) {
	const dva_t *dva = &bp->blk_dva[i];
	uint64_t vdevid = DVA_GET_VDEV(dva);

	if (vdevid >= spa->spa_root_vdev->vdev_children) {
	errors += zfs_blkptr_verify_log(spa, bp, blk_verify,
	"blkptr at %p DVA %u has invalid VDEV %llu",
	bp, i, (longlong_t)vdevid);
	continue;
	}
	vdev_t *vd = spa->spa_root_vdev->vdev_child[vdevid];
	if (vd == NULL) {
	errors += zfs_blkptr_verify_log(spa, bp, blk_verify,
	"blkptr at %p DVA %u has invalid VDEV %llu",
	bp, i, (longlong_t)vdevid);
	continue;
	}
	if (vd->vdev_ops == &vdev_hole_ops) {
	errors += zfs_blkptr_verify_log(spa, bp, blk_verify,
	"blkptr at %p DVA %u has hole VDEV %llu",
	bp, i, (longlong_t)vdevid);
	continue;
	}
	if (vd->vdev_ops == &vdev_missing_ops) {
	/*
	* "missing" vdevs are valid during import, but we
	* don't have their detailed info (e.g. asize), so
	* we can't perform any more checks on them.
	*/
	continue;
	}
	uint64_t offset = DVA_GET_OFFSET(dva);
	uint64_t asize = DVA_GET_ASIZE(dva);
	if (DVA_GET_GANG(dva))
	asize = vdev_gang_header_asize(vd);
	if (offset + asize > vd->vdev_asize) {
	errors += zfs_blkptr_verify_log(spa, bp, blk_verify,
	"blkptr at %p DVA %u has invalid OFFSET %llu",
	bp, i, (longlong_t)offset);
	}
	}
	if (errors > 0)
	dprintf_bp(bp, "blkptr at %p dprintf_bp():", bp);
	if (!config_held)
	spa_config_exit(spa, SCL_VDEV, bp);

	return (errors == 0);
	}

	boolean_t
	zfs_dva_valid(spa_t spa, const dva_t dva, const blkptr_t *bp)
	{
	uint64_t vdevid = DVA_GET_VDEV(dva);

	if (vdevid >= spa->spa_root_vdev->vdev_children)
	return (B_FALSE);

	vdev_t *vd = spa->spa_root_vdev->vdev_child[vdevid];
	if (vd == NULL)
	return (B_FALSE);

	if (vd->vdev_ops == &vdev_hole_ops)
	return (B_FALSE);

	if (vd->vdev_ops == &vdev_missing_ops) {
	return (B_FALSE);
	}

	uint64_t offset = DVA_GET_OFFSET(dva);
	uint64_t asize = DVA_GET_ASIZE(dva);

	if (DVA_GET_GANG(dva))
	asize = vdev_gang_header_asize(vd);
	if (offset + asize > vd->vdev_asize)
	return (B_FALSE);

	return (B_TRUE);
	}

	zio_t *
	zio_read(zio_t pio, spa_t spa, const blkptr_t *bp,
	abd_t data, uint64_t size, zio_done_func_t done, void *private,
	zio_priority_t priority, enum zio_flag flags, const zbookmark_phys_t *zb)
	{
	zio_t *zio;

	zio = zio_create(pio, spa, BP_PHYSICAL_BIRTH(bp), bp,
	data, size, size, done, private,
	ZIO_TYPE_READ, priority, flags, NULL, 0, zb,
	ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ?
	ZIO_DDT_CHILD_READ_PIPELINE : ZIO_READ_PIPELINE);

	return (zio);
	}

	zio_t *
	zio_write(zio_t pio, spa_t spa, uint64_t txg, blkptr_t *bp,
	abd_t data, uint64_t lsize, uint64_t psize, const zio_prop_t zp,
	zio_done_func_t ready, zio_done_func_t children_ready,
	zio_done_func_t physdone, zio_done_func_t done,
	void *private, zio_priority_t priority, enum zio_flag flags,
	const zbookmark_phys_t *zb)
	{
	zio_t *zio;

	ASSERT(zp->zp_checksum >= ZIO_CHECKSUM_OFF &&
	zp->zp_checksum < ZIO_CHECKSUM_FUNCTIONS &&
	zp->zp_compress >= ZIO_COMPRESS_OFF &&
	zp->zp_compress < ZIO_COMPRESS_FUNCTIONS &&
	DMU_OT_IS_VALID(zp->zp_type) &&
	zp->zp_level < 32 &&
	zp->zp_copies > 0 &&
	zp->zp_copies <= spa_max_replication(spa));

	zio = zio_create(pio, spa, txg, bp, data, lsize, psize, done, private,
	ZIO_TYPE_WRITE, priority, flags, NULL, 0, zb,
	ZIO_STAGE_OPEN, (flags & ZIO_FLAG_DDT_CHILD) ?
	ZIO_DDT_CHILD_WRITE_PIPELINE : ZIO_WRITE_PIPELINE);

	zio->io_ready = ready;
	zio->io_children_ready = children_ready;
	zio->io_physdone = physdone;
	zio->io_prop = *zp;

	/*
	* Data can be NULL if we are going to call zio_write_override() to
	* provide the already-allocated BP. But we may need the data to
	* verify a dedup hit (if requested). In this case, don't try to
	* dedup (just take the already-allocated BP verbatim). Encrypted
	* dedup blocks need data as well so we also disable dedup in this
	* case.
	*/
	if (data == NULL &&
	(zio->io_prop.zp_dedup_verify \|\| zio->io_prop.zp_encrypt)) {
	zio->io_prop.zp_dedup = zio->io_prop.zp_dedup_verify = B_FALSE;
	}

	return (zio);
	}

	zio_t *
	zio_rewrite(zio_t pio, spa_t spa, uint64_t txg, blkptr_t bp, abd_t data,
	uint64_t size, zio_done_func_t done, void private,
	zio_priority_t priority, enum zio_flag flags, zbookmark_phys_t *zb)
	{
	zio_t *zio;

	zio = zio_create(pio, spa, txg, bp, data, size, size, done, private,
	ZIO_TYPE_WRITE, priority, flags \| ZIO_FLAG_IO_REWRITE, NULL, 0, zb,
	ZIO_STAGE_OPEN, ZIO_REWRITE_PIPELINE);

	return (zio);
	}

	void
	zio_write_override(zio_t zio, blkptr_t bp, int copies, boolean_t nopwrite)
	{
	ASSERT(zio->io_type == ZIO_TYPE_WRITE);
	ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL);
	ASSERT(zio->io_stage == ZIO_STAGE_OPEN);
	ASSERT(zio->io_txg == spa_syncing_txg(zio->io_spa));

	/*
	* We must reset the io_prop to match the values that existed
	* when the bp was first written by dmu_sync() keeping in mind
	* that nopwrite and dedup are mutually exclusive.
	*/
	zio->io_prop.zp_dedup = nopwrite ? B_FALSE : zio->io_prop.zp_dedup;
	zio->io_prop.zp_nopwrite = nopwrite;
	zio->io_prop.zp_copies = copies;
	zio->io_bp_override = bp;
	}

	void
	zio_free(spa_t spa, uint64_t txg, const blkptr_t bp)
	{

	(void) zfs_blkptr_verify(spa, bp, B_FALSE, BLK_VERIFY_HALT);

	/*
	* The check for EMBEDDED is a performance optimization. We
	* process the free here (by ignoring it) rather than
	* putting it on the list and then processing it in zio_free_sync().
	*/
	if (BP_IS_EMBEDDED(bp))
	return;
	metaslab_check_free(spa, bp);

	/*
	* Frees that are for the currently-syncing txg, are not going to be
	* deferred, and which will not need to do a read (i.e. not GANG or
	* DEDUP), can be processed immediately. Otherwise, put them on the
	* in-memory list for later processing.
	*
	* Note that we only defer frees after zfs_sync_pass_deferred_free
	* when the log space map feature is disabled. [see relevant comment
	* in spa_sync_iterate_to_convergence()]
	*/
	if (BP_IS_GANG(bp) \|\|
	BP_GET_DEDUP(bp) \|\|
	txg != spa->spa_syncing_txg \|\|
	(spa_sync_pass(spa) >= zfs_sync_pass_deferred_free &&
	!spa_feature_is_active(spa, SPA_FEATURE_LOG_SPACEMAP))) {
	bplist_append(&spa->spa_free_bplist[txg & TXG_MASK], bp);
	} else {
	VERIFY3P(zio_free_sync(NULL, spa, txg, bp, 0), ==, NULL);
	}
	}

	/*
	* To improve performance, this function may return NULL if we were able
	* to do the free immediately. This avoids the cost of creating a zio
	* (and linking it to the parent, etc).
	*/
	zio_t *
	zio_free_sync(zio_t pio, spa_t spa, uint64_t txg, const blkptr_t *bp,
	enum zio_flag flags)
	{
	ASSERT(!BP_IS_HOLE(bp));
	ASSERT(spa_syncing_txg(spa) == txg);

	if (BP_IS_EMBEDDED(bp))
	return (NULL);

	metaslab_check_free(spa, bp);
	arc_freed(spa, bp);
	dsl_scan_freed(spa, bp);

	if (BP_IS_GANG(bp) \|\| BP_GET_DEDUP(bp)) {
	/*
	* GANG and DEDUP blocks can induce a read (for the gang block
	* header, or the DDT), so issue them asynchronously so that
	* this thread is not tied up.
	*/
	enum zio_stage stage =
	ZIO_FREE_PIPELINE \| ZIO_STAGE_ISSUE_ASYNC;

	return (zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp),
	BP_GET_PSIZE(bp), NULL, NULL,
	ZIO_TYPE_FREE, ZIO_PRIORITY_NOW,
	flags, NULL, 0, NULL, ZIO_STAGE_OPEN, stage));
	} else {
	metaslab_free(spa, bp, txg, B_FALSE);
	return (NULL);
	}
	}

	zio_t *
	zio_claim(zio_t pio, spa_t spa, uint64_t txg, const blkptr_t *bp,
	zio_done_func_t done, void private, enum zio_flag flags)
	{
	zio_t *zio;

	(void) zfs_blkptr_verify(spa, bp, flags & ZIO_FLAG_CONFIG_WRITER,
	BLK_VERIFY_HALT);

	if (BP_IS_EMBEDDED(bp))
	return (zio_null(pio, spa, NULL, NULL, NULL, 0));

	/*
	* A claim is an allocation of a specific block. Claims are needed
	* to support immediate writes in the intent log. The issue is that
	* immediate writes contain committed data, but in a txg that was
	* not committed. Upon opening the pool after an unclean shutdown,
	* the intent log claims all blocks that contain immediate write data
	* so that the SPA knows they're in use.
	*
	* All claims must be resolved in the first txg -- before the SPA
	* starts allocating blocks -- so that nothing is allocated twice.
	* If txg == 0 we just verify that the block is claimable.
	*/
	ASSERT3U(spa->spa_uberblock.ub_rootbp.blk_birth, <,
	spa_min_claim_txg(spa));
	ASSERT(txg == spa_min_claim_txg(spa) \|\| txg == 0);
	ASSERT(!BP_GET_DEDUP(bp) \|\| !spa_writeable(spa)); /* zdb(8) */

	zio = zio_create(pio, spa, txg, bp, NULL, BP_GET_PSIZE(bp),
	BP_GET_PSIZE(bp), done, private, ZIO_TYPE_CLAIM, ZIO_PRIORITY_NOW,
	flags, NULL, 0, NULL, ZIO_STAGE_OPEN, ZIO_CLAIM_PIPELINE);
	ASSERT0(zio->io_queued_timestamp);

	return (zio);
	}

	zio_t *
	zio_ioctl(zio_t pio, spa_t spa, vdev_t *vd, int cmd,
	zio_done_func_t done, void private, enum zio_flag flags)
	{
	zio_t *zio;
	int c;

	if (vd->vdev_children == 0) {
	zio = zio_create(pio, spa, 0, NULL, NULL, 0, 0, done, private,
	ZIO_TYPE_IOCTL, ZIO_PRIORITY_NOW, flags, vd, 0, NULL,
	ZIO_STAGE_OPEN, ZIO_IOCTL_PIPELINE);

	zio->io_cmd = cmd;
	} else {
	zio = zio_null(pio, spa, NULL, NULL, NULL, flags);

	for (c = 0; c < vd->vdev_children; c++)
	zio_nowait(zio_ioctl(zio, spa, vd->vdev_child[c], cmd,
	done, private, flags));
	}

	return (zio);
	}

	zio_t *
	zio_trim(zio_t pio, vdev_t vd, uint64_t offset, uint64_t size,
	zio_done_func_t done, void private, zio_priority_t priority,
	enum zio_flag flags, enum trim_flag trim_flags)
	{
	zio_t *zio;

	ASSERT0(vd->vdev_children);
	ASSERT0(P2PHASE(offset, 1ULL << vd->vdev_ashift));
	ASSERT0(P2PHASE(size, 1ULL << vd->vdev_ashift));
	ASSERT3U(size, !=, 0);

	zio = zio_create(pio, vd->vdev_spa, 0, NULL, NULL, size, size, done,
	private, ZIO_TYPE_TRIM, priority, flags \| ZIO_FLAG_PHYSICAL,
	vd, offset, NULL, ZIO_STAGE_OPEN, ZIO_TRIM_PIPELINE);
	zio->io_trim_flags = trim_flags;

	return (zio);
	}

	zio_t *
	zio_read_phys(zio_t pio, vdev_t vd, uint64_t offset, uint64_t size,
	abd_t data, int checksum, zio_done_func_t done, void *private,
	zio_priority_t priority, enum zio_flag flags, boolean_t labels)
	{
	zio_t *zio;

	ASSERT(vd->vdev_children == 0);
	ASSERT(!labels \|\| offset + size <= VDEV_LABEL_START_SIZE \|\|
	offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE);
	ASSERT3U(offset + size, <=, vd->vdev_psize);

	zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, size, done,
	private, ZIO_TYPE_READ, priority, flags \| ZIO_FLAG_PHYSICAL, vd,
	offset, NULL, ZIO_STAGE_OPEN, ZIO_READ_PHYS_PIPELINE);

	zio->io_prop.zp_checksum = checksum;

	return (zio);
	}

	zio_t *
	zio_write_phys(zio_t pio, vdev_t vd, uint64_t offset, uint64_t size,
	abd_t data, int checksum, zio_done_func_t done, void *private,
	zio_priority_t priority, enum zio_flag flags, boolean_t labels)
	{
	zio_t *zio;

	ASSERT(vd->vdev_children == 0);
	ASSERT(!labels \|\| offset + size <= VDEV_LABEL_START_SIZE \|\|
	offset >= vd->vdev_psize - VDEV_LABEL_END_SIZE);
	ASSERT3U(offset + size, <=, vd->vdev_psize);

	zio = zio_create(pio, vd->vdev_spa, 0, NULL, data, size, size, done,
	private, ZIO_TYPE_WRITE, priority, flags \| ZIO_FLAG_PHYSICAL, vd,
	offset, NULL, ZIO_STAGE_OPEN, ZIO_WRITE_PHYS_PIPELINE);

	zio->io_prop.zp_checksum = checksum;

	if (zio_checksum_table[checksum].ci_flags & ZCHECKSUM_FLAG_EMBEDDED) {
	/*
	* zec checksums are necessarily destructive -- they modify
	* the end of the write buffer to hold the verifier/checksum.
	* Therefore, we must make a local copy in case the data is
	* being written to multiple places in parallel.
	*/
	abd_t *wbuf = abd_alloc_sametype(data, size);
	abd_copy(wbuf, data, size);

	zio_push_transform(zio, wbuf, size, size, NULL);
	}

	return (zio);
	}

	/*
	* Create a child I/O to do some work for us.
	*/
	zio_t *
	zio_vdev_child_io(zio_t pio, blkptr_t bp, vdev_t *vd, uint64_t offset,
	abd_t *data, uint64_t size, int type, zio_priority_t priority,
	enum zio_flag flags, zio_done_func_t done, void private)
	{
	enum zio_stage pipeline = ZIO_VDEV_CHILD_PIPELINE;
	zio_t *zio;

	/*
	* vdev child I/Os do not propagate their error to the parent.
	* Therefore, for correct operation the caller must check for
	* and handle the error in the child i/o's done callback.
	* The only exceptions are i/os that we don't care about
	* (OPTIONAL or REPAIR).
	*/
	ASSERT((flags & ZIO_FLAG_OPTIONAL) \|\| (flags & ZIO_FLAG_IO_REPAIR) \|\|
	done != NULL);

	if (type == ZIO_TYPE_READ && bp != NULL) {
	/*
	* If we have the bp, then the child should perform the
	* checksum and the parent need not. This pushes error
	* detection as close to the leaves as possible and
	* eliminates redundant checksums in the interior nodes.
	*/
	pipeline \|= ZIO_STAGE_CHECKSUM_VERIFY;
	pio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY;
	}

	if (vd->vdev_ops->vdev_op_leaf) {
	ASSERT0(vd->vdev_children);
	offset += VDEV_LABEL_START_SIZE;
	}

	flags \|= ZIO_VDEV_CHILD_FLAGS(pio);

	/*
	* If we've decided to do a repair, the write is not speculative --
	* even if the original read was.
	*/
	if (flags & ZIO_FLAG_IO_REPAIR)
	flags &= ~ZIO_FLAG_SPECULATIVE;

	/*
	* If we're creating a child I/O that is not associated with a
	* top-level vdev, then the child zio is not an allocating I/O.
	* If this is a retried I/O then we ignore it since we will
	* have already processed the original allocating I/O.
	*/
	if (flags & ZIO_FLAG_IO_ALLOCATING &&
	(vd != vd->vdev_top \|\| (flags & ZIO_FLAG_IO_RETRY))) {
	ASSERT(pio->io_metaslab_class != NULL);
	ASSERT(pio->io_metaslab_class->mc_alloc_throttle_enabled);
	ASSERT(type == ZIO_TYPE_WRITE);
	ASSERT(priority == ZIO_PRIORITY_ASYNC_WRITE);
	ASSERT(!(flags & ZIO_FLAG_IO_REPAIR));
	ASSERT(!(pio->io_flags & ZIO_FLAG_IO_REWRITE) \|\|
	pio->io_child_type == ZIO_CHILD_GANG);

	flags &= ~ZIO_FLAG_IO_ALLOCATING;
	}


	zio = zio_create(pio, pio->io_spa, pio->io_txg, bp, data, size, size,
	done, private, type, priority, flags, vd, offset, &pio->io_bookmark,
	ZIO_STAGE_VDEV_IO_START >> 1, pipeline);
	ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_VDEV);

	zio->io_physdone = pio->io_physdone;
	if (vd->vdev_ops->vdev_op_leaf && zio->io_logical != NULL)
	zio->io_logical->io_phys_children++;

	return (zio);
	}

	zio_t *
	zio_vdev_delegated_io(vdev_t vd, uint64_t offset, abd_t data, uint64_t size,
	zio_type_t type, zio_priority_t priority, enum zio_flag flags,
	zio_done_func_t done, void private)
	{
	zio_t *zio;

	ASSERT(vd->vdev_ops->vdev_op_leaf);

	zio = zio_create(NULL, vd->vdev_spa, 0, NULL,
	data, size, size, done, private, type, priority,
	flags \| ZIO_FLAG_CANFAIL \| ZIO_FLAG_DONT_RETRY \| ZIO_FLAG_DELEGATED,
	vd, offset, NULL,
	ZIO_STAGE_VDEV_IO_START >> 1, ZIO_VDEV_CHILD_PIPELINE);

	return (zio);
	}

	void
	zio_flush(zio_t zio, vdev_t vd)
	{
	zio_nowait(zio_ioctl(zio, zio->io_spa, vd, DKIOCFLUSHWRITECACHE,
	NULL, NULL,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_DONT_PROPAGATE \| ZIO_FLAG_DONT_RETRY));
	}

	void
	zio_shrink(zio_t *zio, uint64_t size)
	{
	ASSERT3P(zio->io_executor, ==, NULL);
	ASSERT3U(zio->io_orig_size, ==, zio->io_size);
	ASSERT3U(size, <=, zio->io_size);

	/*
	* We don't shrink for raidz because of problems with the
	* reconstruction when reading back less than the block size.
	* Note, BP_IS_RAIDZ() assumes no compression.
	*/
	ASSERT(BP_GET_COMPRESS(zio->io_bp) == ZIO_COMPRESS_OFF);
	if (!BP_IS_RAIDZ(zio->io_bp)) {
	/* we are not doing a raw write */
	ASSERT3U(zio->io_size, ==, zio->io_lsize);
	zio->io_orig_size = zio->io_size = zio->io_lsize = size;
	}
	}

	/*
	* ==========================================================================
	* Prepare to read and write logical blocks
	* ==========================================================================
	*/

	static zio_t *
	zio_read_bp_init(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;
	uint64_t psize =
	BP_IS_EMBEDDED(bp) ? BPE_GET_PSIZE(bp) : BP_GET_PSIZE(bp);

	ASSERT3P(zio->io_bp, ==, &zio->io_bp_copy);

	if (BP_GET_COMPRESS(bp) != ZIO_COMPRESS_OFF &&
	zio->io_child_type == ZIO_CHILD_LOGICAL &&
	!(zio->io_flags & ZIO_FLAG_RAW_COMPRESS)) {
	zio_push_transform(zio, abd_alloc_sametype(zio->io_abd, psize),
	psize, psize, zio_decompress);
	}

	if (((BP_IS_PROTECTED(bp) && !(zio->io_flags & ZIO_FLAG_RAW_ENCRYPT)) \|\|
	BP_HAS_INDIRECT_MAC_CKSUM(bp)) &&
	zio->io_child_type == ZIO_CHILD_LOGICAL) {
	zio_push_transform(zio, abd_alloc_sametype(zio->io_abd, psize),
	psize, psize, zio_decrypt);
	}

	if (BP_IS_EMBEDDED(bp) && BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA) {
	int psize = BPE_GET_PSIZE(bp);
	void *data = abd_borrow_buf(zio->io_abd, psize);

	zio->io_pipeline = ZIO_INTERLOCK_PIPELINE;
	decode_embedded_bp_compressed(bp, data);
	abd_return_buf_copy(zio->io_abd, data, psize);
	} else {
	ASSERT(!BP_IS_EMBEDDED(bp));
	ASSERT3P(zio->io_bp, ==, &zio->io_bp_copy);
	}

	if (!DMU_OT_IS_METADATA(BP_GET_TYPE(bp)) && BP_GET_LEVEL(bp) == 0)
	zio->io_flags \|= ZIO_FLAG_DONT_CACHE;

	if (BP_GET_TYPE(bp) == DMU_OT_DDT_ZAP)
	zio->io_flags \|= ZIO_FLAG_DONT_CACHE;

	if (BP_GET_DEDUP(bp) && zio->io_child_type == ZIO_CHILD_LOGICAL)
	zio->io_pipeline = ZIO_DDT_READ_PIPELINE;

	return (zio);
	}

	static zio_t *
	zio_write_bp_init(zio_t *zio)
	{
	if (!IO_IS_ALLOCATING(zio))
	return (zio);

	ASSERT(zio->io_child_type != ZIO_CHILD_DDT);

	if (zio->io_bp_override) {
	blkptr_t *bp = zio->io_bp;
	zio_prop_t *zp = &zio->io_prop;

	ASSERT(bp->blk_birth != zio->io_txg);
	ASSERT(BP_GET_DEDUP(zio->io_bp_override) == 0);

	bp = zio->io_bp_override;
	zio->io_pipeline = ZIO_INTERLOCK_PIPELINE;

	if (BP_IS_EMBEDDED(bp))
	return (zio);

	/*
	* If we've been overridden and nopwrite is set then
	* set the flag accordingly to indicate that a nopwrite
	* has already occurred.
	*/
	if (!BP_IS_HOLE(bp) && zp->zp_nopwrite) {
	ASSERT(!zp->zp_dedup);
	ASSERT3U(BP_GET_CHECKSUM(bp), ==, zp->zp_checksum);
	zio->io_flags \|= ZIO_FLAG_NOPWRITE;
	return (zio);
	}

	ASSERT(!zp->zp_nopwrite);

	if (BP_IS_HOLE(bp) \|\| !zp->zp_dedup)
	return (zio);

	ASSERT((zio_checksum_table[zp->zp_checksum].ci_flags &
	ZCHECKSUM_FLAG_DEDUP) \|\| zp->zp_dedup_verify);

	if (BP_GET_CHECKSUM(bp) == zp->zp_checksum &&
	!zp->zp_encrypt) {
	BP_SET_DEDUP(bp, 1);
	zio->io_pipeline \|= ZIO_STAGE_DDT_WRITE;
	return (zio);
	}

	/*
	* We were unable to handle this as an override bp, treat
	* it as a regular write I/O.
	*/
	zio->io_bp_override = NULL;
	*bp = zio->io_bp_orig;
	zio->io_pipeline = zio->io_orig_pipeline;
	}

	return (zio);
	}

	static zio_t *
	zio_write_compress(zio_t *zio)
	{
	spa_t *spa = zio->io_spa;
	zio_prop_t *zp = &zio->io_prop;
	enum zio_compress compress = zp->zp_compress;
	blkptr_t *bp = zio->io_bp;
	uint64_t lsize = zio->io_lsize;
	uint64_t psize = zio->io_size;
	int pass = 1;

	/*
	* If our children haven't all reached the ready stage,
	* wait for them and then repeat this pipeline stage.
	*/
	if (zio_wait_for_children(zio, ZIO_CHILD_LOGICAL_BIT \|
	ZIO_CHILD_GANG_BIT, ZIO_WAIT_READY)) {
	return (NULL);
	}

	if (!IO_IS_ALLOCATING(zio))
	return (zio);

	if (zio->io_children_ready != NULL) {
	/*
	* Now that all our children are ready, run the callback
	* associated with this zio in case it wants to modify the
	* data to be written.
	*/
	ASSERT3U(zp->zp_level, >, 0);
	zio->io_children_ready(zio);
	}

	ASSERT(zio->io_child_type != ZIO_CHILD_DDT);
	ASSERT(zio->io_bp_override == NULL);

	if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg) {
	/*
	* We're rewriting an existing block, which means we're
	* working on behalf of spa_sync(). For spa_sync() to
	* converge, it must eventually be the case that we don't
	* have to allocate new blocks. But compression changes
	* the blocksize, which forces a reallocate, and makes
	* convergence take longer. Therefore, after the first
	* few passes, stop compressing to ensure convergence.
	*/
	pass = spa_sync_pass(spa);

	ASSERT(zio->io_txg == spa_syncing_txg(spa));
	ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL);
	ASSERT(!BP_GET_DEDUP(bp));

	if (pass >= zfs_sync_pass_dont_compress)
	compress = ZIO_COMPRESS_OFF;

	/* Make sure someone doesn't change their mind on overwrites */
	ASSERT(BP_IS_EMBEDDED(bp) \|\| MIN(zp->zp_copies + BP_IS_GANG(bp),
	spa_max_replication(spa)) == BP_GET_NDVAS(bp));
	}

	/* If it's a compressed write that is not raw, compress the buffer. */
	if (compress != ZIO_COMPRESS_OFF &&
	!(zio->io_flags & ZIO_FLAG_RAW_COMPRESS)) {
	void *cbuf = zio_buf_alloc(lsize);
	psize = zio_compress_data(compress, zio->io_abd, cbuf, lsize,
	zp->zp_complevel);
	if (psize == 0 \|\| psize >= lsize) {
	compress = ZIO_COMPRESS_OFF;
	zio_buf_free(cbuf, lsize);
	} else if (!zp->zp_dedup && !zp->zp_encrypt &&
	psize <= BPE_PAYLOAD_SIZE &&
	zp->zp_level == 0 && !DMU_OT_HAS_FILL(zp->zp_type) &&
	spa_feature_is_enabled(spa, SPA_FEATURE_EMBEDDED_DATA)) {
	encode_embedded_bp_compressed(bp,
	cbuf, compress, lsize, psize);
	BPE_SET_ETYPE(bp, BP_EMBEDDED_TYPE_DATA);
	BP_SET_TYPE(bp, zio->io_prop.zp_type);
	BP_SET_LEVEL(bp, zio->io_prop.zp_level);
	zio_buf_free(cbuf, lsize);
	bp->blk_birth = zio->io_txg;
	zio->io_pipeline = ZIO_INTERLOCK_PIPELINE;
	ASSERT(spa_feature_is_active(spa,
	SPA_FEATURE_EMBEDDED_DATA));
	return (zio);
	} else {
	/*
	* Round compressed size up to the minimum allocation
	* size of the smallest-ashift device, and zero the
	* tail. This ensures that the compressed size of the
	* BP (and thus compressratio property) are correct,
	* in that we charge for the padding used to fill out
	* the last sector.
	*/
	ASSERT3U(spa->spa_min_alloc, >=, SPA_MINBLOCKSHIFT);
	size_t rounded = (size_t)roundup(psize,
	spa->spa_min_alloc);
	if (rounded >= lsize) {
	compress = ZIO_COMPRESS_OFF;
	zio_buf_free(cbuf, lsize);
	psize = lsize;
	} else {
	abd_t *cdata = abd_get_from_buf(cbuf, lsize);
	abd_take_ownership_of_buf(cdata, B_TRUE);
	abd_zero_off(cdata, psize, rounded - psize);
	psize = rounded;
	zio_push_transform(zio, cdata,
	psize, lsize, NULL);
	}
	}

	/*
	* We were unable to handle this as an override bp, treat
	* it as a regular write I/O.
	*/
	zio->io_bp_override = NULL;
	*bp = zio->io_bp_orig;
	zio->io_pipeline = zio->io_orig_pipeline;

	} else if ((zio->io_flags & ZIO_FLAG_RAW_ENCRYPT) != 0 &&
	zp->zp_type == DMU_OT_DNODE) {
	/*
	* The DMU actually relies on the zio layer's compression
	* to free metadnode blocks that have had all contained
	* dnodes freed. As a result, even when doing a raw
	* receive, we must check whether the block can be compressed
	* to a hole.
	*/
	psize = zio_compress_data(ZIO_COMPRESS_EMPTY,
	zio->io_abd, NULL, lsize, zp->zp_complevel);
	if (psize == 0 \|\| psize >= lsize)
	compress = ZIO_COMPRESS_OFF;
	} else if (zio->io_flags & ZIO_FLAG_RAW_COMPRESS) {
	size_t rounded = MIN((size_t)roundup(psize,
	spa->spa_min_alloc), lsize);

	if (rounded != psize) {
	abd_t *cdata = abd_alloc_linear(rounded, B_TRUE);
	abd_zero_off(cdata, psize, rounded - psize);
	abd_copy_off(cdata, zio->io_abd, 0, 0, psize);
	psize = rounded;
	zio_push_transform(zio, cdata,
	psize, rounded, NULL);
	}
	} else {
	ASSERT3U(psize, !=, 0);
	}

	/*
	* The final pass of spa_sync() must be all rewrites, but the first
	* few passes offer a trade-off: allocating blocks defers convergence,
	* but newly allocated blocks are sequential, so they can be written
	* to disk faster. Therefore, we allow the first few passes of
	* spa_sync() to allocate new blocks, but force rewrites after that.
	* There should only be a handful of blocks after pass 1 in any case.
	*/
	if (!BP_IS_HOLE(bp) && bp->blk_birth == zio->io_txg &&
	BP_GET_PSIZE(bp) == psize &&
	pass >= zfs_sync_pass_rewrite) {
	VERIFY3U(psize, !=, 0);
	enum zio_stage gang_stages = zio->io_pipeline & ZIO_GANG_STAGES;

	zio->io_pipeline = ZIO_REWRITE_PIPELINE \| gang_stages;
	zio->io_flags \|= ZIO_FLAG_IO_REWRITE;
	} else {
	BP_ZERO(bp);
	zio->io_pipeline = ZIO_WRITE_PIPELINE;
	}

	if (psize == 0) {
	if (zio->io_bp_orig.blk_birth != 0 &&
	spa_feature_is_active(spa, SPA_FEATURE_HOLE_BIRTH)) {
	BP_SET_LSIZE(bp, lsize);
	BP_SET_TYPE(bp, zp->zp_type);
	BP_SET_LEVEL(bp, zp->zp_level);
	BP_SET_BIRTH(bp, zio->io_txg, 0);
	}
	zio->io_pipeline = ZIO_INTERLOCK_PIPELINE;
	} else {
	ASSERT(zp->zp_checksum != ZIO_CHECKSUM_GANG_HEADER);
	BP_SET_LSIZE(bp, lsize);
	BP_SET_TYPE(bp, zp->zp_type);
	BP_SET_LEVEL(bp, zp->zp_level);
	BP_SET_PSIZE(bp, psize);
	BP_SET_COMPRESS(bp, compress);
	BP_SET_CHECKSUM(bp, zp->zp_checksum);
	BP_SET_DEDUP(bp, zp->zp_dedup);
	BP_SET_BYTEORDER(bp, ZFS_HOST_BYTEORDER);
	if (zp->zp_dedup) {
	ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL);
	ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE));
	ASSERT(!zp->zp_encrypt \|\|
	DMU_OT_IS_ENCRYPTED(zp->zp_type));
	zio->io_pipeline = ZIO_DDT_WRITE_PIPELINE;
	}
	if (zp->zp_nopwrite) {
	ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL);
	ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE));
	zio->io_pipeline \|= ZIO_STAGE_NOP_WRITE;
	}
	}
	return (zio);
	}

	static zio_t *
	zio_free_bp_init(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;

	if (zio->io_child_type == ZIO_CHILD_LOGICAL) {
	if (BP_GET_DEDUP(bp))
	zio->io_pipeline = ZIO_DDT_FREE_PIPELINE;
	}

	ASSERT3P(zio->io_bp, ==, &zio->io_bp_copy);

	return (zio);
	}

	/*
	* ==========================================================================
	* Execute the I/O pipeline
	* ==========================================================================
	*/

	static void
	zio_taskq_dispatch(zio_t *zio, zio_taskq_type_t q, boolean_t cutinline)
	{
	spa_t *spa = zio->io_spa;
	zio_type_t t = zio->io_type;
	int flags = (cutinline ? TQ_FRONT : 0);

	/*
	* If we're a config writer or a probe, the normal issue and
	* interrupt threads may all be blocked waiting for the config lock.
	* In this case, select the otherwise-unused taskq for ZIO_TYPE_NULL.
	*/
	if (zio->io_flags & (ZIO_FLAG_CONFIG_WRITER \| ZIO_FLAG_PROBE))
	t = ZIO_TYPE_NULL;

	/*
	* A similar issue exists for the L2ARC write thread until L2ARC 2.0.
	*/
	if (t == ZIO_TYPE_WRITE && zio->io_vd && zio->io_vd->vdev_aux)
	t = ZIO_TYPE_NULL;

	/*
	* If this is a high priority I/O, then use the high priority taskq if
	* available.
	*/
	if ((zio->io_priority == ZIO_PRIORITY_NOW \|\|
	zio->io_priority == ZIO_PRIORITY_SYNC_WRITE) &&
	spa->spa_zio_taskq[t][q + 1].stqs_count != 0)
	q++;

	ASSERT3U(q, <, ZIO_TASKQ_TYPES);

	/*
	* NB: We are assuming that the zio can only be dispatched
	* to a single taskq at a time. It would be a grievous error
	* to dispatch the zio to another taskq at the same time.
	*/
	ASSERT(taskq_empty_ent(&zio->io_tqent));
	spa_taskq_dispatch_ent(spa, t, q, zio_execute, zio, flags,
	&zio->io_tqent);
	}

	static boolean_t
	zio_taskq_member(zio_t *zio, zio_taskq_type_t q)
	{
	spa_t *spa = zio->io_spa;

	taskq_t *tq = taskq_of_curthread();

	for (zio_type_t t = 0; t < ZIO_TYPES; t++) {
	spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];
	uint_t i;
	for (i = 0; i < tqs->stqs_count; i++) {
	if (tqs->stqs_taskq[i] == tq)
	return (B_TRUE);
	}
	}

	return (B_FALSE);
	}

	static zio_t *
	zio_issue_async(zio_t *zio)
	{
	zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE);

	return (NULL);
	}

	void
	zio_interrupt(void *zio)
	{
	zio_taskq_dispatch(zio, ZIO_TASKQ_INTERRUPT, B_FALSE);
	}

	void
	zio_delay_interrupt(zio_t *zio)
	{
	/*
	* The timeout_generic() function isn't defined in userspace, so
	* rather than trying to implement the function, the zio delay
	* functionality has been disabled for userspace builds.
	*/

	#ifdef _KERNEL
	/*
	* If io_target_timestamp is zero, then no delay has been registered
	* for this IO, thus jump to the end of this function and "skip" the
	* delay; issuing it directly to the zio layer.
	*/
	if (zio->io_target_timestamp != 0) {
	hrtime_t now = gethrtime();

	if (now >= zio->io_target_timestamp) {
	/*
	* This IO has already taken longer than the target
	* delay to complete, so we don't want to delay it
	* any longer; we "miss" the delay and issue it
	* directly to the zio layer. This is likely due to
	* the target latency being set to a value less than
	* the underlying hardware can satisfy (e.g. delay
	* set to 1ms, but the disks take 10ms to complete an
	* IO request).
	*/

	DTRACE_PROBE2(zio__delay__miss, zio_t *, zio,
	hrtime_t, now);

	zio_interrupt(zio);
	} else {
	taskqid_t tid;
	hrtime_t diff = zio->io_target_timestamp - now;
	clock_t expire_at_tick = ddi_get_lbolt() +
	NSEC_TO_TICK(diff);

	DTRACE_PROBE3(zio__delay__hit, zio_t *, zio,
	hrtime_t, now, hrtime_t, diff);

	if (NSEC_TO_TICK(diff) == 0) {
	/* Our delay is less than a jiffy - just spin */
	zfs_sleep_until(zio->io_target_timestamp);
	zio_interrupt(zio);
	} else {
	/*
	* Use taskq_dispatch_delay() in the place of
	* OpenZFS's timeout_generic().
	*/
	tid = taskq_dispatch_delay(system_taskq,
	zio_interrupt, zio, TQ_NOSLEEP,
	expire_at_tick);
	if (tid == TASKQID_INVALID) {
	/*
	* Couldn't allocate a task. Just
	* finish the zio without a delay.
	*/
	zio_interrupt(zio);
	}
	}
	}
	return;
	}
	#endif
	DTRACE_PROBE1(zio__delay__skip, zio_t *, zio);
	zio_interrupt(zio);
	}

	static void
	zio_deadman_impl(zio_t *pio, int ziodepth)
	{
	zio_t cio, cio_next;
	zio_link_t *zl = NULL;
	vdev_t *vd = pio->io_vd;

	if (zio_deadman_log_all \|\| (vd != NULL && vd->vdev_ops->vdev_op_leaf)) {
	vdev_queue_t *vq = vd ? &vd->vdev_queue : NULL;
	zbookmark_phys_t *zb = &pio->io_bookmark;
	uint64_t delta = gethrtime() - pio->io_timestamp;
	uint64_t failmode = spa_get_deadman_failmode(pio->io_spa);

	zfs_dbgmsg("slow zio[%d]: zio=%px timestamp=%llu "
	"delta=%llu queued=%llu io=%llu "
	"path=%s "
	"last=%llu type=%d "
	"priority=%d flags=0x%x stage=0x%x "
	"pipeline=0x%x pipeline-trace=0x%x "
	"objset=%llu object=%llu "
	"level=%llu blkid=%llu "
	"offset=%llu size=%llu "
	"error=%d",
	ziodepth, pio, pio->io_timestamp,
	(u_longlong_t)delta, pio->io_delta, pio->io_delay,
	vd ? vd->vdev_path : "NULL",
	vq ? vq->vq_io_complete_ts : 0, pio->io_type,
	pio->io_priority, pio->io_flags, pio->io_stage,
	pio->io_pipeline, pio->io_pipeline_trace,
	(u_longlong_t)zb->zb_objset, (u_longlong_t)zb->zb_object,
	(u_longlong_t)zb->zb_level, (u_longlong_t)zb->zb_blkid,
	(u_longlong_t)pio->io_offset, (u_longlong_t)pio->io_size,
	pio->io_error);
	(void) zfs_ereport_post(FM_EREPORT_ZFS_DEADMAN,
	pio->io_spa, vd, zb, pio, 0);

	if (failmode == ZIO_FAILURE_MODE_CONTINUE &&
	taskq_empty_ent(&pio->io_tqent)) {
	zio_interrupt(pio);
	}
	}

	mutex_enter(&pio->io_lock);
	for (cio = zio_walk_children(pio, &zl); cio != NULL; cio = cio_next) {
	cio_next = zio_walk_children(pio, &zl);
	zio_deadman_impl(cio, ziodepth + 1);
	}
	mutex_exit(&pio->io_lock);
	}

	/*
	* Log the critical information describing this zio and all of its children
	* using the zfs_dbgmsg() interface then post deadman event for the ZED.
	*/
	void
	zio_deadman(zio_t pio, char tag)
	{
	spa_t *spa = pio->io_spa;
	char *name = spa_name(spa);

	if (!zfs_deadman_enabled \|\| spa_suspended(spa))
	return;

	zio_deadman_impl(pio, 0);

	switch (spa_get_deadman_failmode(spa)) {
	case ZIO_FAILURE_MODE_WAIT:
	zfs_dbgmsg("%s waiting for hung I/O to pool '%s'", tag, name);
	break;

	case ZIO_FAILURE_MODE_CONTINUE:
	zfs_dbgmsg("%s restarting hung I/O for pool '%s'", tag, name);
	break;

	case ZIO_FAILURE_MODE_PANIC:
	fm_panic("%s determined I/O to pool '%s' is hung.", tag, name);
	break;
	}
	}

	/*
	* Execute the I/O pipeline until one of the following occurs:
	* (1) the I/O completes; (2) the pipeline stalls waiting for
	* dependent child I/Os; (3) the I/O issues, so we're waiting
	* for an I/O completion interrupt; (4) the I/O is delegated by
	* vdev-level caching or aggregation; (5) the I/O is deferred
	* due to vdev-level queueing; (6) the I/O is handed off to
	* another thread. In all cases, the pipeline stops whenever
	* there's no CPU work; it never burns a thread in cv_wait_io().
	*
	* There's no locking on io_stage because there's no legitimate way
	* for multiple threads to be attempting to process the same I/O.
	*/
	static zio_pipe_stage_t *zio_pipeline[];

	/*
	* zio_execute() is a wrapper around the static function
	* __zio_execute() so that we can force __zio_execute() to be
	* inlined. This reduces stack overhead which is important
	* because __zio_execute() is called recursively in several zio
	* code paths. zio_execute() itself cannot be inlined because
	* it is externally visible.
	*/
	void
	zio_execute(void *zio)
	{
	fstrans_cookie_t cookie;

	cookie = spl_fstrans_mark();
	__zio_execute(zio);
	spl_fstrans_unmark(cookie);
	}

	/*
	* Used to determine if in the current context the stack is sized large
	* enough to allow zio_execute() to be called recursively. A minimum
	* stack size of 16K is required to avoid needing to re-dispatch the zio.
	*/
	static boolean_t
	zio_execute_stack_check(zio_t *zio)
	{
	#if !defined(HAVE_LARGE_STACKS)
	dsl_pool_t *dp = spa_get_dsl(zio->io_spa);

	/* Executing in txg_sync_thread() context. */
	if (dp && curthread == dp->dp_tx.tx_sync_thread)
	return (B_TRUE);

	/* Pool initialization outside of zio_taskq context. */
	if (dp && spa_is_initializing(dp->dp_spa) &&
	!zio_taskq_member(zio, ZIO_TASKQ_ISSUE) &&
	!zio_taskq_member(zio, ZIO_TASKQ_ISSUE_HIGH))
	return (B_TRUE);
	#endif /* HAVE_LARGE_STACKS */

	return (B_FALSE);
	}

	__attribute__((always_inline))
	static inline void
	__zio_execute(zio_t *zio)
	{
	ASSERT3U(zio->io_queued_timestamp, >, 0);

	while (zio->io_stage < ZIO_STAGE_DONE) {
	enum zio_stage pipeline = zio->io_pipeline;
	enum zio_stage stage = zio->io_stage;

	zio->io_executor = curthread;

	ASSERT(!MUTEX_HELD(&zio->io_lock));
	ASSERT(ISP2(stage));
	ASSERT(zio->io_stall == NULL);

	do {
	stage <<= 1;
	} while ((stage & pipeline) == 0);

	ASSERT(stage <= ZIO_STAGE_DONE);

	/*
	* If we are in interrupt context and this pipeline stage
	* will grab a config lock that is held across I/O,
	* or may wait for an I/O that needs an interrupt thread
	* to complete, issue async to avoid deadlock.
	*
	* For VDEV_IO_START, we cut in line so that the io will
	* be sent to disk promptly.
	*/
	if ((stage & ZIO_BLOCKING_STAGES) && zio->io_vd == NULL &&
	zio_taskq_member(zio, ZIO_TASKQ_INTERRUPT)) {
	boolean_t cut = (stage == ZIO_STAGE_VDEV_IO_START) ?
	zio_requeue_io_start_cut_in_line : B_FALSE;
	zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, cut);
	return;
	}

	/*
	* If the current context doesn't have large enough stacks
	* the zio must be issued asynchronously to prevent overflow.
	*/
	if (zio_execute_stack_check(zio)) {
	boolean_t cut = (stage == ZIO_STAGE_VDEV_IO_START) ?
	zio_requeue_io_start_cut_in_line : B_FALSE;
	zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, cut);
	return;
	}

	zio->io_stage = stage;
	zio->io_pipeline_trace \|= zio->io_stage;

	/*
	* The zio pipeline stage returns the next zio to execute
	* (typically the same as this one), or NULL if we should
	* stop.
	*/
	zio = zio_pipeline[highbit64(stage) - 1](zio);

	if (zio == NULL)
	return;
	}
	}


	/*
	* ==========================================================================
	* Initiate I/O, either sync or async
	* ==========================================================================
	*/
	int
	zio_wait(zio_t *zio)
	{
	/*
	* Some routines, like zio_free_sync(), may return a NULL zio
	* to avoid the performance overhead of creating and then destroying
	* an unneeded zio. For the callers' simplicity, we accept a NULL
	* zio and ignore it.
	*/
	if (zio == NULL)
	return (0);

	long timeout = MSEC_TO_TICK(zfs_deadman_ziotime_ms);
	int error;

	ASSERT3S(zio->io_stage, ==, ZIO_STAGE_OPEN);
	ASSERT3P(zio->io_executor, ==, NULL);

	zio->io_waiter = curthread;
	ASSERT0(zio->io_queued_timestamp);
	zio->io_queued_timestamp = gethrtime();

	__zio_execute(zio);

	mutex_enter(&zio->io_lock);
	while (zio->io_executor != NULL) {
	error = cv_timedwait_io(&zio->io_cv, &zio->io_lock,
	ddi_get_lbolt() + timeout);

	if (zfs_deadman_enabled && error == -1 &&
	gethrtime() - zio->io_queued_timestamp >
	spa_deadman_ziotime(zio->io_spa)) {
	mutex_exit(&zio->io_lock);
	timeout = MSEC_TO_TICK(zfs_deadman_checktime_ms);
	zio_deadman(zio, FTAG);
	mutex_enter(&zio->io_lock);
	}
	}
	mutex_exit(&zio->io_lock);

	error = zio->io_error;
	zio_destroy(zio);

	return (error);
	}

	void
	zio_nowait(zio_t *zio)
	{
	/*
	* See comment in zio_wait().
	*/
	if (zio == NULL)
	return;

	ASSERT3P(zio->io_executor, ==, NULL);

	if (zio->io_child_type == ZIO_CHILD_LOGICAL &&
	zio_unique_parent(zio) == NULL) {
	zio_t *pio;

	/*
	* This is a logical async I/O with no parent to wait for it.
	* We add it to the spa_async_root_zio "Godfather" I/O which
	* will ensure they complete prior to unloading the pool.
	*/
	spa_t *spa = zio->io_spa;
	pio = spa->spa_async_zio_root[CPU_SEQID_UNSTABLE];

	zio_add_child(pio, zio);
	}

	ASSERT0(zio->io_queued_timestamp);
	zio->io_queued_timestamp = gethrtime();
	__zio_execute(zio);
	}

	/*
	* ==========================================================================
	* Reexecute, cancel, or suspend/resume failed I/O
	* ==========================================================================
	*/

	static void
	zio_reexecute(void *arg)
	{
	zio_t *pio = arg;
	zio_t cio, cio_next;

	ASSERT(pio->io_child_type == ZIO_CHILD_LOGICAL);
	ASSERT(pio->io_orig_stage == ZIO_STAGE_OPEN);
	ASSERT(pio->io_gang_leader == NULL);
	ASSERT(pio->io_gang_tree == NULL);

	pio->io_flags = pio->io_orig_flags;
	pio->io_stage = pio->io_orig_stage;
	pio->io_pipeline = pio->io_orig_pipeline;
	pio->io_reexecute = 0;
	pio->io_flags \|= ZIO_FLAG_REEXECUTED;
	pio->io_pipeline_trace = 0;
	pio->io_error = 0;
	for (int w = 0; w < ZIO_WAIT_TYPES; w++)
	pio->io_state[w] = 0;
	for (int c = 0; c < ZIO_CHILD_TYPES; c++)
	pio->io_child_error[c] = 0;

	if (IO_IS_ALLOCATING(pio))
	BP_ZERO(pio->io_bp);

	/*
	* As we reexecute pio's children, new children could be created.
	* New children go to the head of pio's io_child_list, however,
	* so we will (correctly) not reexecute them. The key is that
	* the remainder of pio's io_child_list, from 'cio_next' onward,
	* cannot be affected by any side effects of reexecuting 'cio'.
	*/
	zio_link_t *zl = NULL;
	mutex_enter(&pio->io_lock);
	for (cio = zio_walk_children(pio, &zl); cio != NULL; cio = cio_next) {
	cio_next = zio_walk_children(pio, &zl);
	for (int w = 0; w < ZIO_WAIT_TYPES; w++)
	pio->io_children[cio->io_child_type][w]++;
	mutex_exit(&pio->io_lock);
	zio_reexecute(cio);
	mutex_enter(&pio->io_lock);
	}
	mutex_exit(&pio->io_lock);

	/*
	* Now that all children have been reexecuted, execute the parent.
	* We don't reexecute "The Godfather" I/O here as it's the
	* responsibility of the caller to wait on it.
	*/
	if (!(pio->io_flags & ZIO_FLAG_GODFATHER)) {
	pio->io_queued_timestamp = gethrtime();
	__zio_execute(pio);
	}
	}

	void
	zio_suspend(spa_t spa, zio_t zio, zio_suspend_reason_t reason)
	{
	if (spa_get_failmode(spa) == ZIO_FAILURE_MODE_PANIC)
	fm_panic("Pool '%s' has encountered an uncorrectable I/O "
	"failure and the failure mode property for this pool "
	"is set to panic.", spa_name(spa));

	cmn_err(CE_WARN, "Pool '%s' has encountered an uncorrectable I/O "
	"failure and has been suspended.\n", spa_name(spa));

	(void) zfs_ereport_post(FM_EREPORT_ZFS_IO_FAILURE, spa, NULL,
	NULL, NULL, 0);

	mutex_enter(&spa->spa_suspend_lock);

	if (spa->spa_suspend_zio_root == NULL)
	spa->spa_suspend_zio_root = zio_root(spa, NULL, NULL,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_SPECULATIVE \|
	ZIO_FLAG_GODFATHER);

	spa->spa_suspended = reason;

	if (zio != NULL) {
	ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER));
	ASSERT(zio != spa->spa_suspend_zio_root);
	ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL);
	ASSERT(zio_unique_parent(zio) == NULL);
	ASSERT(zio->io_stage == ZIO_STAGE_DONE);
	zio_add_child(spa->spa_suspend_zio_root, zio);
	}

	mutex_exit(&spa->spa_suspend_lock);
	}

	int
	zio_resume(spa_t *spa)
	{
	zio_t *pio;

	/*
	* Reexecute all previously suspended i/o.
	*/
	mutex_enter(&spa->spa_suspend_lock);
	spa->spa_suspended = ZIO_SUSPEND_NONE;
	cv_broadcast(&spa->spa_suspend_cv);
	pio = spa->spa_suspend_zio_root;
	spa->spa_suspend_zio_root = NULL;
	mutex_exit(&spa->spa_suspend_lock);

	if (pio == NULL)
	return (0);

	zio_reexecute(pio);
	return (zio_wait(pio));
	}

	void
	zio_resume_wait(spa_t *spa)
	{
	mutex_enter(&spa->spa_suspend_lock);
	while (spa_suspended(spa))
	cv_wait(&spa->spa_suspend_cv, &spa->spa_suspend_lock);
	mutex_exit(&spa->spa_suspend_lock);
	}

	/*
	* ==========================================================================
	* Gang blocks.
	*
	* A gang block is a collection of small blocks that looks to the DMU
	* like one large block. When zio_dva_allocate() cannot find a block
	* of the requested size, due to either severe fragmentation or the pool
	* being nearly full, it calls zio_write_gang_block() to construct the
	* block from smaller fragments.
	*
	* A gang block consists of a gang header (zio_gbh_phys_t) and up to
	* three (SPA_GBH_NBLKPTRS) gang members. The gang header is just like
	* an indirect block: it's an array of block pointers. It consumes
	* only one sector and hence is allocatable regardless of fragmentation.
	* The gang header's bps point to its gang members, which hold the data.
	*
	* Gang blocks are self-checksumming, using the bp's <vdev, offset, txg>
	* as the verifier to ensure uniqueness of the SHA256 checksum.
	* Critically, the gang block bp's blk_cksum is the checksum of the data,
	* not the gang header. This ensures that data block signatures (needed for
	* deduplication) are independent of how the block is physically stored.
	*
	* Gang blocks can be nested: a gang member may itself be a gang block.
	* Thus every gang block is a tree in which root and all interior nodes are
	* gang headers, and the leaves are normal blocks that contain user data.
	* The root of the gang tree is called the gang leader.
	*
	* To perform any operation (read, rewrite, free, claim) on a gang block,
	* zio_gang_assemble() first assembles the gang tree (minus data leaves)
	* in the io_gang_tree field of the original logical i/o by recursively
	* reading the gang leader and all gang headers below it. This yields
	* an in-core tree containing the contents of every gang header and the
	* bps for every constituent of the gang block.
	*
	* With the gang tree now assembled, zio_gang_issue() just walks the gang tree
	* and invokes a callback on each bp. To free a gang block, zio_gang_issue()
	* calls zio_free_gang() -- a trivial wrapper around zio_free() -- for each bp.
	* zio_claim_gang() provides a similarly trivial wrapper for zio_claim().
	* zio_read_gang() is a wrapper around zio_read() that omits reading gang
	* headers, since we already have those in io_gang_tree. zio_rewrite_gang()
	* performs a zio_rewrite() of the data or, for gang headers, a zio_rewrite()
	* of the gang header plus zio_checksum_compute() of the data to update the
	* gang header's blk_cksum as described above.
	*
	* The two-phase assemble/issue model solves the problem of partial failure --
	* what if you'd freed part of a gang block but then couldn't read the
	* gang header for another part? Assembling the entire gang tree first
	* ensures that all the necessary gang header I/O has succeeded before
	* starting the actual work of free, claim, or write. Once the gang tree
	* is assembled, free and claim are in-memory operations that cannot fail.
	*
	* In the event that a gang write fails, zio_dva_unallocate() walks the
	* gang tree to immediately free (i.e. insert back into the space map)
	* everything we've allocated. This ensures that we don't get ENOSPC
	* errors during repeated suspend/resume cycles due to a flaky device.
	*
	* Gang rewrites only happen during sync-to-convergence. If we can't assemble
	* the gang tree, we won't modify the block, so we can safely defer the free
	* (knowing that the block is still intact). If we can assemble the gang
	* tree, then even if some of the rewrites fail, zio_dva_unallocate() will free
	* each constituent bp and we can allocate a new block on the next sync pass.
	*
	* In all cases, the gang tree allows complete recovery from partial failure.
	* ==========================================================================
	*/

	static void
	zio_gang_issue_func_done(zio_t *zio)
	{
	abd_free(zio->io_abd);
	}

	static zio_t *
	zio_read_gang(zio_t pio, blkptr_t bp, zio_gang_node_t gn, abd_t data,
	uint64_t offset)
	{
	if (gn != NULL)
	return (pio);

	return (zio_read(pio, pio->io_spa, bp, abd_get_offset(data, offset),
	BP_GET_PSIZE(bp), zio_gang_issue_func_done,
	NULL, pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio),
	&pio->io_bookmark));
	}

	static zio_t *
	zio_rewrite_gang(zio_t pio, blkptr_t bp, zio_gang_node_t gn, abd_t data,
	uint64_t offset)
	{
	zio_t *zio;

	if (gn != NULL) {
	abd_t *gbh_abd =
	abd_get_from_buf(gn->gn_gbh, SPA_GANGBLOCKSIZE);
	zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp,
	gbh_abd, SPA_GANGBLOCKSIZE, zio_gang_issue_func_done, NULL,
	pio->io_priority, ZIO_GANG_CHILD_FLAGS(pio),
	&pio->io_bookmark);
	/*
	* As we rewrite each gang header, the pipeline will compute
	* a new gang block header checksum for it; but no one will
	* compute a new data checksum, so we do that here. The one
	* exception is the gang leader: the pipeline already computed
	* its data checksum because that stage precedes gang assembly.
	* (Presently, nothing actually uses interior data checksums;
	* this is just good hygiene.)
	*/
	if (gn != pio->io_gang_leader->io_gang_tree) {
	abd_t *buf = abd_get_offset(data, offset);

	zio_checksum_compute(zio, BP_GET_CHECKSUM(bp),
	buf, BP_GET_PSIZE(bp));

	abd_free(buf);
	}
	/*
	* If we are here to damage data for testing purposes,
	* leave the GBH alone so that we can detect the damage.
	*/
	if (pio->io_gang_leader->io_flags & ZIO_FLAG_INDUCE_DAMAGE)
	zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES;
	} else {
	zio = zio_rewrite(pio, pio->io_spa, pio->io_txg, bp,
	abd_get_offset(data, offset), BP_GET_PSIZE(bp),
	zio_gang_issue_func_done, NULL, pio->io_priority,
	ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark);
	}

	return (zio);
	}

	/* ARGSUSED */
	static zio_t *
	zio_free_gang(zio_t pio, blkptr_t bp, zio_gang_node_t gn, abd_t data,
	uint64_t offset)
	{
	zio_t *zio = zio_free_sync(pio, pio->io_spa, pio->io_txg, bp,
	ZIO_GANG_CHILD_FLAGS(pio));
	if (zio == NULL) {
	zio = zio_null(pio, pio->io_spa,
	NULL, NULL, NULL, ZIO_GANG_CHILD_FLAGS(pio));
	}
	return (zio);
	}

	/* ARGSUSED */
	static zio_t *
	zio_claim_gang(zio_t pio, blkptr_t bp, zio_gang_node_t gn, abd_t data,
	uint64_t offset)
	{
	return (zio_claim(pio, pio->io_spa, pio->io_txg, bp,
	NULL, NULL, ZIO_GANG_CHILD_FLAGS(pio)));
	}

	static zio_gang_issue_func_t *zio_gang_issue_func[ZIO_TYPES] = {
	NULL,
	zio_read_gang,
	zio_rewrite_gang,
	zio_free_gang,
	zio_claim_gang,
	NULL
	};

	static void zio_gang_tree_assemble_done(zio_t *zio);

	static zio_gang_node_t *
	zio_gang_node_alloc(zio_gang_node_t **gnpp)
	{
	zio_gang_node_t *gn;

	ASSERT(*gnpp == NULL);

	gn = kmem_zalloc(sizeof (*gn), KM_SLEEP);
	gn->gn_gbh = zio_buf_alloc(SPA_GANGBLOCKSIZE);
	*gnpp = gn;

	return (gn);
	}

	static void
	zio_gang_node_free(zio_gang_node_t **gnpp)
	{
	zio_gang_node_t gn = gnpp;

	for (int g = 0; g < SPA_GBH_NBLKPTRS; g++)
	ASSERT(gn->gn_child[g] == NULL);

	zio_buf_free(gn->gn_gbh, SPA_GANGBLOCKSIZE);
	kmem_free(gn, sizeof (*gn));
	*gnpp = NULL;
	}

	static void
	zio_gang_tree_free(zio_gang_node_t **gnpp)
	{
	zio_gang_node_t gn = gnpp;

	if (gn == NULL)
	return;

	for (int g = 0; g < SPA_GBH_NBLKPTRS; g++)
	zio_gang_tree_free(&gn->gn_child[g]);

	zio_gang_node_free(gnpp);
	}

	static void
	zio_gang_tree_assemble(zio_t gio, blkptr_t bp, zio_gang_node_t **gnpp)
	{
	zio_gang_node_t *gn = zio_gang_node_alloc(gnpp);
	abd_t *gbh_abd = abd_get_from_buf(gn->gn_gbh, SPA_GANGBLOCKSIZE);

	ASSERT(gio->io_gang_leader == gio);
	ASSERT(BP_IS_GANG(bp));

	zio_nowait(zio_read(gio, gio->io_spa, bp, gbh_abd, SPA_GANGBLOCKSIZE,
	zio_gang_tree_assemble_done, gn, gio->io_priority,
	ZIO_GANG_CHILD_FLAGS(gio), &gio->io_bookmark));
	}

	static void
	zio_gang_tree_assemble_done(zio_t *zio)
	{
	zio_t *gio = zio->io_gang_leader;
	zio_gang_node_t *gn = zio->io_private;
	blkptr_t *bp = zio->io_bp;

	ASSERT(gio == zio_unique_parent(zio));
	ASSERT(zio->io_child_count == 0);

	if (zio->io_error)
	return;

	/* this ABD was created from a linear buf in zio_gang_tree_assemble */
	if (BP_SHOULD_BYTESWAP(bp))
	byteswap_uint64_array(abd_to_buf(zio->io_abd), zio->io_size);

	ASSERT3P(abd_to_buf(zio->io_abd), ==, gn->gn_gbh);
	ASSERT(zio->io_size == SPA_GANGBLOCKSIZE);
	ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC);

	abd_free(zio->io_abd);

	for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) {
	blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g];
	if (!BP_IS_GANG(gbp))
	continue;
	zio_gang_tree_assemble(gio, gbp, &gn->gn_child[g]);
	}
	}

	static void
	zio_gang_tree_issue(zio_t pio, zio_gang_node_t gn, blkptr_t bp, abd_t data,
	uint64_t offset)
	{
	zio_t *gio = pio->io_gang_leader;
	zio_t *zio;

	ASSERT(BP_IS_GANG(bp) == !!gn);
	ASSERT(BP_GET_CHECKSUM(bp) == BP_GET_CHECKSUM(gio->io_bp));
	ASSERT(BP_GET_LSIZE(bp) == BP_GET_PSIZE(bp) \|\| gn == gio->io_gang_tree);

	/*
	* If you're a gang header, your data is in gn->gn_gbh.
	* If you're a gang member, your data is in 'data' and gn == NULL.
	*/
	zio = zio_gang_issue_func[gio->io_type](pio, bp, gn, data, offset);

	if (gn != NULL) {
	ASSERT(gn->gn_gbh->zg_tail.zec_magic == ZEC_MAGIC);

	for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) {
	blkptr_t *gbp = &gn->gn_gbh->zg_blkptr[g];
	if (BP_IS_HOLE(gbp))
	continue;
	zio_gang_tree_issue(zio, gn->gn_child[g], gbp, data,
	offset);
	offset += BP_GET_PSIZE(gbp);
	}
	}

	if (gn == gio->io_gang_tree)
	ASSERT3U(gio->io_size, ==, offset);

	if (zio != pio)
	zio_nowait(zio);
	}

	static zio_t *
	zio_gang_assemble(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;

	ASSERT(BP_IS_GANG(bp) && zio->io_gang_leader == NULL);
	ASSERT(zio->io_child_type > ZIO_CHILD_GANG);

	zio->io_gang_leader = zio;

	zio_gang_tree_assemble(zio, bp, &zio->io_gang_tree);

	return (zio);
	}

	static zio_t *
	zio_gang_issue(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;

	if (zio_wait_for_children(zio, ZIO_CHILD_GANG_BIT, ZIO_WAIT_DONE)) {
	return (NULL);
	}

	ASSERT(BP_IS_GANG(bp) && zio->io_gang_leader == zio);
	ASSERT(zio->io_child_type > ZIO_CHILD_GANG);

	if (zio->io_child_error[ZIO_CHILD_GANG] == 0)
	zio_gang_tree_issue(zio, zio->io_gang_tree, bp, zio->io_abd,
	0);
	else
	zio_gang_tree_free(&zio->io_gang_tree);

	zio->io_pipeline = ZIO_INTERLOCK_PIPELINE;

	return (zio);
	}

	static void
	zio_write_gang_member_ready(zio_t *zio)
	{
	zio_t *pio = zio_unique_parent(zio);
	dva_t *cdva = zio->io_bp->blk_dva;
	dva_t *pdva = pio->io_bp->blk_dva;
	uint64_t asize;
	zio_t *gio __maybe_unused = zio->io_gang_leader;

	if (BP_IS_HOLE(zio->io_bp))
	return;

	ASSERT(BP_IS_HOLE(&zio->io_bp_orig));

	ASSERT(zio->io_child_type == ZIO_CHILD_GANG);
	ASSERT3U(zio->io_prop.zp_copies, ==, gio->io_prop.zp_copies);
	ASSERT3U(zio->io_prop.zp_copies, <=, BP_GET_NDVAS(zio->io_bp));
	ASSERT3U(pio->io_prop.zp_copies, <=, BP_GET_NDVAS(pio->io_bp));
	ASSERT3U(BP_GET_NDVAS(zio->io_bp), <=, BP_GET_NDVAS(pio->io_bp));

	mutex_enter(&pio->io_lock);
	for (int d = 0; d < BP_GET_NDVAS(zio->io_bp); d++) {
	ASSERT(DVA_GET_GANG(&pdva[d]));
	asize = DVA_GET_ASIZE(&pdva[d]);
	asize += DVA_GET_ASIZE(&cdva[d]);
	DVA_SET_ASIZE(&pdva[d], asize);
	}
	mutex_exit(&pio->io_lock);
	}

	static void
	zio_write_gang_done(zio_t *zio)
	{
	/*
	* The io_abd field will be NULL for a zio with no data. The io_flags
	* will initially have the ZIO_FLAG_NODATA bit flag set, but we can't
	* check for it here as it is cleared in zio_ready.
	*/
	if (zio->io_abd != NULL)
	abd_free(zio->io_abd);
	}

	static zio_t *
	zio_write_gang_block(zio_t pio, metaslab_class_t mc)
	{
	spa_t *spa = pio->io_spa;
	blkptr_t *bp = pio->io_bp;
	zio_t *gio = pio->io_gang_leader;
	zio_t *zio;
	zio_gang_node_t gn, *gnpp;
	zio_gbh_phys_t *gbh;
	abd_t *gbh_abd;
	uint64_t txg = pio->io_txg;
	uint64_t resid = pio->io_size;
	uint64_t lsize;
	int copies = gio->io_prop.zp_copies;
	int gbh_copies;
	zio_prop_t zp;
	int error;
	boolean_t has_data = !(pio->io_flags & ZIO_FLAG_NODATA);

	/*
	* encrypted blocks need DVA[2] free so encrypted gang headers can't
	* have a third copy.
	*/
	gbh_copies = MIN(copies + 1, spa_max_replication(spa));
	if (gio->io_prop.zp_encrypt && gbh_copies >= SPA_DVAS_PER_BP)
	gbh_copies = SPA_DVAS_PER_BP - 1;

	int flags = METASLAB_HINTBP_FAVOR \| METASLAB_GANG_HEADER;
	if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) {
	ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE);
	ASSERT(has_data);

	flags \|= METASLAB_ASYNC_ALLOC;
	VERIFY(zfs_refcount_held(&mc->mc_allocator[pio->io_allocator].
	mca_alloc_slots, pio));

	/*
	* The logical zio has already placed a reservation for
	* 'copies' allocation slots but gang blocks may require
	* additional copies. These additional copies
	* (i.e. gbh_copies - copies) are guaranteed to succeed
	* since metaslab_class_throttle_reserve() always allows
	* additional reservations for gang blocks.
	*/
	VERIFY(metaslab_class_throttle_reserve(mc, gbh_copies - copies,
	pio->io_allocator, pio, flags));
	}

	error = metaslab_alloc(spa, mc, SPA_GANGBLOCKSIZE,
	bp, gbh_copies, txg, pio == gio ? NULL : gio->io_bp, flags,
	&pio->io_alloc_list, pio, pio->io_allocator);
	if (error) {
	if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) {
	ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE);
	ASSERT(has_data);

	/*
	* If we failed to allocate the gang block header then
	* we remove any additional allocation reservations that
	* we placed here. The original reservation will
	* be removed when the logical I/O goes to the ready
	* stage.
	*/
	metaslab_class_throttle_unreserve(mc,
	gbh_copies - copies, pio->io_allocator, pio);
	}

	pio->io_error = error;
	return (pio);
	}

	if (pio == gio) {
	gnpp = &gio->io_gang_tree;
	} else {
	gnpp = pio->io_private;
	ASSERT(pio->io_ready == zio_write_gang_member_ready);
	}

	gn = zio_gang_node_alloc(gnpp);
	gbh = gn->gn_gbh;
	bzero(gbh, SPA_GANGBLOCKSIZE);
	gbh_abd = abd_get_from_buf(gbh, SPA_GANGBLOCKSIZE);

	/*
	* Create the gang header.
	*/
	zio = zio_rewrite(pio, spa, txg, bp, gbh_abd, SPA_GANGBLOCKSIZE,
	zio_write_gang_done, NULL, pio->io_priority,
	ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark);

	/*
	* Create and nowait the gang children.
	*/
	for (int g = 0; resid != 0; resid -= lsize, g++) {
	lsize = P2ROUNDUP(resid / (SPA_GBH_NBLKPTRS - g),
	SPA_MINBLOCKSIZE);
	ASSERT(lsize >= SPA_MINBLOCKSIZE && lsize <= resid);

	zp.zp_checksum = gio->io_prop.zp_checksum;
	zp.zp_compress = ZIO_COMPRESS_OFF;
	zp.zp_complevel = gio->io_prop.zp_complevel;
	zp.zp_type = DMU_OT_NONE;
	zp.zp_level = 0;
	zp.zp_copies = gio->io_prop.zp_copies;
	zp.zp_dedup = B_FALSE;
	zp.zp_dedup_verify = B_FALSE;
	zp.zp_nopwrite = B_FALSE;
	zp.zp_encrypt = gio->io_prop.zp_encrypt;
	zp.zp_byteorder = gio->io_prop.zp_byteorder;
	bzero(zp.zp_salt, ZIO_DATA_SALT_LEN);
	bzero(zp.zp_iv, ZIO_DATA_IV_LEN);
	bzero(zp.zp_mac, ZIO_DATA_MAC_LEN);

	zio_t *cio = zio_write(zio, spa, txg, &gbh->zg_blkptr[g],
	has_data ? abd_get_offset(pio->io_abd, pio->io_size -
	resid) : NULL, lsize, lsize, &zp,
	zio_write_gang_member_ready, NULL, NULL,
	zio_write_gang_done, &gn->gn_child[g], pio->io_priority,
	ZIO_GANG_CHILD_FLAGS(pio), &pio->io_bookmark);

	if (pio->io_flags & ZIO_FLAG_IO_ALLOCATING) {
	ASSERT(pio->io_priority == ZIO_PRIORITY_ASYNC_WRITE);
	ASSERT(has_data);

	/*
	* Gang children won't throttle but we should
	* account for their work, so reserve an allocation
	* slot for them here.
	*/
	VERIFY(metaslab_class_throttle_reserve(mc,
	zp.zp_copies, cio->io_allocator, cio, flags));
	}
	zio_nowait(cio);
	}

	/*
	* Set pio's pipeline to just wait for zio to finish.
	*/
	pio->io_pipeline = ZIO_INTERLOCK_PIPELINE;

	/*
	* We didn't allocate this bp, so make sure it doesn't get unmarked.
	*/
	pio->io_flags &= ~ZIO_FLAG_FASTWRITE;

	zio_nowait(zio);

	return (pio);
	}

	/*
	* The zio_nop_write stage in the pipeline determines if allocating a
	* new bp is necessary. The nopwrite feature can handle writes in
	* either syncing or open context (i.e. zil writes) and as a result is
	* mutually exclusive with dedup.
	*
	* By leveraging a cryptographically secure checksum, such as SHA256, we
	* can compare the checksums of the new data and the old to determine if
	* allocating a new block is required. Note that our requirements for
	* cryptographic strength are fairly weak: there can't be any accidental
	* hash collisions, but we don't need to be secure against intentional
	* (malicious) collisions. To trigger a nopwrite, you have to be able
	* to write the file to begin with, and triggering an incorrect (hash
	* collision) nopwrite is no worse than simply writing to the file.
	* That said, there are no known attacks against the checksum algorithms
	* used for nopwrite, assuming that the salt and the checksums
	* themselves remain secret.
	*/
	static zio_t *
	zio_nop_write(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;
	blkptr_t *bp_orig = &zio->io_bp_orig;
	zio_prop_t *zp = &zio->io_prop;

	ASSERT(BP_GET_LEVEL(bp) == 0);
	ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REWRITE));
	ASSERT(zp->zp_nopwrite);
	ASSERT(!zp->zp_dedup);
	ASSERT(zio->io_bp_override == NULL);
	ASSERT(IO_IS_ALLOCATING(zio));

	/*
	* Check to see if the original bp and the new bp have matching
	* characteristics (i.e. same checksum, compression algorithms, etc).
	* If they don't then just continue with the pipeline which will
	* allocate a new bp.
	*/
	if (BP_IS_HOLE(bp_orig) \|\|
	!(zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_flags &
	ZCHECKSUM_FLAG_NOPWRITE) \|\|
	BP_IS_ENCRYPTED(bp) \|\| BP_IS_ENCRYPTED(bp_orig) \|\|
	BP_GET_CHECKSUM(bp) != BP_GET_CHECKSUM(bp_orig) \|\|
	BP_GET_COMPRESS(bp) != BP_GET_COMPRESS(bp_orig) \|\|
	BP_GET_DEDUP(bp) != BP_GET_DEDUP(bp_orig) \|\|
	zp->zp_copies != BP_GET_NDVAS(bp_orig))
	return (zio);

	/*
	* If the checksums match then reset the pipeline so that we
	* avoid allocating a new bp and issuing any I/O.
	*/
	if (ZIO_CHECKSUM_EQUAL(bp->blk_cksum, bp_orig->blk_cksum)) {
	ASSERT(zio_checksum_table[zp->zp_checksum].ci_flags &
	ZCHECKSUM_FLAG_NOPWRITE);
	ASSERT3U(BP_GET_PSIZE(bp), ==, BP_GET_PSIZE(bp_orig));
	ASSERT3U(BP_GET_LSIZE(bp), ==, BP_GET_LSIZE(bp_orig));
	ASSERT(zp->zp_compress != ZIO_COMPRESS_OFF);
	ASSERT(bcmp(&bp->blk_prop, &bp_orig->blk_prop,
	sizeof (uint64_t)) == 0);

	/*
	* If we're overwriting a block that is currently on an
	* indirect vdev, then ignore the nopwrite request and
	* allow a new block to be allocated on a concrete vdev.
	*/
	spa_config_enter(zio->io_spa, SCL_VDEV, FTAG, RW_READER);
	vdev_t *tvd = vdev_lookup_top(zio->io_spa,
	DVA_GET_VDEV(&bp->blk_dva[0]));
	if (tvd->vdev_ops == &vdev_indirect_ops) {
	spa_config_exit(zio->io_spa, SCL_VDEV, FTAG);
	return (zio);
	}
	spa_config_exit(zio->io_spa, SCL_VDEV, FTAG);

	bp = bp_orig;
	zio->io_pipeline = ZIO_INTERLOCK_PIPELINE;
	zio->io_flags \|= ZIO_FLAG_NOPWRITE;
	}

	return (zio);
	}

	/*
	* ==========================================================================
	* Dedup
	* ==========================================================================
	*/
	static void
	zio_ddt_child_read_done(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;
	ddt_entry_t *dde = zio->io_private;
	ddt_phys_t *ddp;
	zio_t *pio = zio_unique_parent(zio);

	mutex_enter(&pio->io_lock);
	ddp = ddt_phys_select(dde, bp);
	if (zio->io_error == 0)
	ddt_phys_clear(ddp); /* this ddp doesn't need repair */

	if (zio->io_error == 0 && dde->dde_repair_abd == NULL)
	dde->dde_repair_abd = zio->io_abd;
	else
	abd_free(zio->io_abd);
	mutex_exit(&pio->io_lock);
	}

	static zio_t *
	zio_ddt_read_start(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;

	ASSERT(BP_GET_DEDUP(bp));
	ASSERT(BP_GET_PSIZE(bp) == zio->io_size);
	ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL);

	if (zio->io_child_error[ZIO_CHILD_DDT]) {
	ddt_t *ddt = ddt_select(zio->io_spa, bp);
	ddt_entry_t *dde = ddt_repair_start(ddt, bp);
	ddt_phys_t *ddp = dde->dde_phys;
	ddt_phys_t *ddp_self = ddt_phys_select(dde, bp);
	blkptr_t blk;

	ASSERT(zio->io_vsd == NULL);
	zio->io_vsd = dde;

	if (ddp_self == NULL)
	return (zio);

	for (int p = 0; p < DDT_PHYS_TYPES; p++, ddp++) {
	if (ddp->ddp_phys_birth == 0 \|\| ddp == ddp_self)
	continue;
	ddt_bp_create(ddt->ddt_checksum, &dde->dde_key, ddp,
	&blk);
	zio_nowait(zio_read(zio, zio->io_spa, &blk,
	abd_alloc_for_io(zio->io_size, B_TRUE),
	zio->io_size, zio_ddt_child_read_done, dde,
	zio->io_priority, ZIO_DDT_CHILD_FLAGS(zio) \|
	ZIO_FLAG_DONT_PROPAGATE, &zio->io_bookmark));
	}
	return (zio);
	}

	zio_nowait(zio_read(zio, zio->io_spa, bp,
	zio->io_abd, zio->io_size, NULL, NULL, zio->io_priority,
	ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark));

	return (zio);
	}

	static zio_t *
	zio_ddt_read_done(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;

	if (zio_wait_for_children(zio, ZIO_CHILD_DDT_BIT, ZIO_WAIT_DONE)) {
	return (NULL);
	}

	ASSERT(BP_GET_DEDUP(bp));
	ASSERT(BP_GET_PSIZE(bp) == zio->io_size);
	ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL);

	if (zio->io_child_error[ZIO_CHILD_DDT]) {
	ddt_t *ddt = ddt_select(zio->io_spa, bp);
	ddt_entry_t *dde = zio->io_vsd;
	if (ddt == NULL) {
	ASSERT(spa_load_state(zio->io_spa) != SPA_LOAD_NONE);
	return (zio);
	}
	if (dde == NULL) {
	zio->io_stage = ZIO_STAGE_DDT_READ_START >> 1;
	zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_FALSE);
	return (NULL);
	}
	if (dde->dde_repair_abd != NULL) {
	abd_copy(zio->io_abd, dde->dde_repair_abd,
	zio->io_size);
	zio->io_child_error[ZIO_CHILD_DDT] = 0;
	}
	ddt_repair_done(ddt, dde);
	zio->io_vsd = NULL;
	}

	ASSERT(zio->io_vsd == NULL);

	return (zio);
	}

	static boolean_t
	zio_ddt_collision(zio_t zio, ddt_t ddt, ddt_entry_t *dde)
	{
	spa_t *spa = zio->io_spa;
	boolean_t do_raw = !!(zio->io_flags & ZIO_FLAG_RAW);

	ASSERT(!(zio->io_bp_override && do_raw));

	/*
	* Note: we compare the original data, not the transformed data,
	* because when zio->io_bp is an override bp, we will not have
	* pushed the I/O transforms. That's an important optimization
	* because otherwise we'd compress/encrypt all dmu_sync() data twice.
	* However, we should never get a raw, override zio so in these
	* cases we can compare the io_abd directly. This is useful because
	* it allows us to do dedup verification even if we don't have access
	* to the original data (for instance, if the encryption keys aren't
	* loaded).
	*/

	for (int p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) {
	zio_t *lio = dde->dde_lead_zio[p];

	if (lio != NULL && do_raw) {
	return (lio->io_size != zio->io_size \|\|
	abd_cmp(zio->io_abd, lio->io_abd) != 0);
	} else if (lio != NULL) {
	return (lio->io_orig_size != zio->io_orig_size \|\|
	abd_cmp(zio->io_orig_abd, lio->io_orig_abd) != 0);
	}
	}

	for (int p = DDT_PHYS_SINGLE; p <= DDT_PHYS_TRIPLE; p++) {
	ddt_phys_t *ddp = &dde->dde_phys[p];

	if (ddp->ddp_phys_birth != 0 && do_raw) {
	blkptr_t blk = *zio->io_bp;
	uint64_t psize;
	abd_t *tmpabd;
	int error;

	ddt_bp_fill(ddp, &blk, ddp->ddp_phys_birth);
	psize = BP_GET_PSIZE(&blk);

	if (psize != zio->io_size)
	return (B_TRUE);

	ddt_exit(ddt);

	tmpabd = abd_alloc_for_io(psize, B_TRUE);

	error = zio_wait(zio_read(NULL, spa, &blk, tmpabd,
	psize, NULL, NULL, ZIO_PRIORITY_SYNC_READ,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_SPECULATIVE \|
	ZIO_FLAG_RAW, &zio->io_bookmark));

	if (error == 0) {
	if (abd_cmp(tmpabd, zio->io_abd) != 0)
	error = SET_ERROR(ENOENT);
	}

	abd_free(tmpabd);
	ddt_enter(ddt);
	return (error != 0);
	} else if (ddp->ddp_phys_birth != 0) {
	arc_buf_t *abuf = NULL;
	arc_flags_t aflags = ARC_FLAG_WAIT;
	blkptr_t blk = *zio->io_bp;
	int error;

	ddt_bp_fill(ddp, &blk, ddp->ddp_phys_birth);

	if (BP_GET_LSIZE(&blk) != zio->io_orig_size)
	return (B_TRUE);

	ddt_exit(ddt);

	error = arc_read(NULL, spa, &blk,
	arc_getbuf_func, &abuf, ZIO_PRIORITY_SYNC_READ,
	ZIO_FLAG_CANFAIL \| ZIO_FLAG_SPECULATIVE,
	&aflags, &zio->io_bookmark);

	if (error == 0) {
	if (abd_cmp_buf(zio->io_orig_abd, abuf->b_data,
	zio->io_orig_size) != 0)
	error = SET_ERROR(ENOENT);
	arc_buf_destroy(abuf, &abuf);
	}

	ddt_enter(ddt);
	return (error != 0);
	}
	}

	return (B_FALSE);
	}

	static void
	zio_ddt_child_write_ready(zio_t *zio)
	{
	int p = zio->io_prop.zp_copies;
	ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp);
	ddt_entry_t *dde = zio->io_private;
	ddt_phys_t *ddp = &dde->dde_phys[p];
	zio_t *pio;

	if (zio->io_error)
	return;

	ddt_enter(ddt);

	ASSERT(dde->dde_lead_zio[p] == zio);

	ddt_phys_fill(ddp, zio->io_bp);

	zio_link_t *zl = NULL;
	while ((pio = zio_walk_parents(zio, &zl)) != NULL)
	ddt_bp_fill(ddp, pio->io_bp, zio->io_txg);

	ddt_exit(ddt);
	}

	static void
	zio_ddt_child_write_done(zio_t *zio)
	{
	int p = zio->io_prop.zp_copies;
	ddt_t *ddt = ddt_select(zio->io_spa, zio->io_bp);
	ddt_entry_t *dde = zio->io_private;
	ddt_phys_t *ddp = &dde->dde_phys[p];

	ddt_enter(ddt);

	ASSERT(ddp->ddp_refcnt == 0);
	ASSERT(dde->dde_lead_zio[p] == zio);
	dde->dde_lead_zio[p] = NULL;

	if (zio->io_error == 0) {
	zio_link_t *zl = NULL;
	while (zio_walk_parents(zio, &zl) != NULL)
	ddt_phys_addref(ddp);
	} else {
	ddt_phys_clear(ddp);
	}

	ddt_exit(ddt);
	}

	static zio_t *
	zio_ddt_write(zio_t *zio)
	{
	spa_t *spa = zio->io_spa;
	blkptr_t *bp = zio->io_bp;
	uint64_t txg = zio->io_txg;
	zio_prop_t *zp = &zio->io_prop;
	int p = zp->zp_copies;
	zio_t *cio = NULL;
	ddt_t *ddt = ddt_select(spa, bp);
	ddt_entry_t *dde;
	ddt_phys_t *ddp;

	ASSERT(BP_GET_DEDUP(bp));
	ASSERT(BP_GET_CHECKSUM(bp) == zp->zp_checksum);
	ASSERT(BP_IS_HOLE(bp) \|\| zio->io_bp_override);
	ASSERT(!(zio->io_bp_override && (zio->io_flags & ZIO_FLAG_RAW)));

	ddt_enter(ddt);
	dde = ddt_lookup(ddt, bp, B_TRUE);
	ddp = &dde->dde_phys[p];

	if (zp->zp_dedup_verify && zio_ddt_collision(zio, ddt, dde)) {
	/*
	* If we're using a weak checksum, upgrade to a strong checksum
	* and try again. If we're already using a strong checksum,
	* we can't resolve it, so just convert to an ordinary write.
	* (And automatically e-mail a paper to Nature?)
	*/
	if (!(zio_checksum_table[zp->zp_checksum].ci_flags &
	ZCHECKSUM_FLAG_DEDUP)) {
	zp->zp_checksum = spa_dedup_checksum(spa);
	zio_pop_transforms(zio);
	zio->io_stage = ZIO_STAGE_OPEN;
	BP_ZERO(bp);
	} else {
	zp->zp_dedup = B_FALSE;
	BP_SET_DEDUP(bp, B_FALSE);
	}
	ASSERT(!BP_GET_DEDUP(bp));
	zio->io_pipeline = ZIO_WRITE_PIPELINE;
	ddt_exit(ddt);
	return (zio);
	}

	if (ddp->ddp_phys_birth != 0 \|\| dde->dde_lead_zio[p] != NULL) {
	if (ddp->ddp_phys_birth != 0)
	ddt_bp_fill(ddp, bp, txg);
	if (dde->dde_lead_zio[p] != NULL)
	zio_add_child(zio, dde->dde_lead_zio[p]);
	else
	ddt_phys_addref(ddp);
	} else if (zio->io_bp_override) {
	ASSERT(bp->blk_birth == txg);
	ASSERT(BP_EQUAL(bp, zio->io_bp_override));
	ddt_phys_fill(ddp, bp);
	ddt_phys_addref(ddp);
	} else {
	cio = zio_write(zio, spa, txg, bp, zio->io_orig_abd,
	zio->io_orig_size, zio->io_orig_size, zp,
	zio_ddt_child_write_ready, NULL, NULL,
	zio_ddt_child_write_done, dde, zio->io_priority,
	ZIO_DDT_CHILD_FLAGS(zio), &zio->io_bookmark);

	zio_push_transform(cio, zio->io_abd, zio->io_size, 0, NULL);
	dde->dde_lead_zio[p] = cio;
	}

	ddt_exit(ddt);

	zio_nowait(cio);

	return (zio);
	}

	ddt_entry_t freedde; / for debugging */

	static zio_t *
	zio_ddt_free(zio_t *zio)
	{
	spa_t *spa = zio->io_spa;
	blkptr_t *bp = zio->io_bp;
	ddt_t *ddt = ddt_select(spa, bp);
	ddt_entry_t *dde;
	ddt_phys_t *ddp;

	ASSERT(BP_GET_DEDUP(bp));
	ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL);

	ddt_enter(ddt);
	freedde = dde = ddt_lookup(ddt, bp, B_TRUE);
	if (dde) {
	ddp = ddt_phys_select(dde, bp);
	if (ddp)
	ddt_phys_decref(ddp);
	}
	ddt_exit(ddt);

	return (zio);
	}

	/*
	* ==========================================================================
	* Allocate and free blocks
	* ==========================================================================
	*/

	static zio_t *
	zio_io_to_allocate(spa_t *spa, int allocator)
	{
	zio_t *zio;

	ASSERT(MUTEX_HELD(&spa->spa_allocs[allocator].spaa_lock));

	zio = avl_first(&spa->spa_allocs[allocator].spaa_tree);
	if (zio == NULL)
	return (NULL);

	ASSERT(IO_IS_ALLOCATING(zio));

	/*
	* Try to place a reservation for this zio. If we're unable to
	* reserve then we throttle.
	*/
	ASSERT3U(zio->io_allocator, ==, allocator);
	if (!metaslab_class_throttle_reserve(zio->io_metaslab_class,
	zio->io_prop.zp_copies, allocator, zio, 0)) {
	return (NULL);
	}

	avl_remove(&spa->spa_allocs[allocator].spaa_tree, zio);
	ASSERT3U(zio->io_stage, <, ZIO_STAGE_DVA_ALLOCATE);

	return (zio);
	}

	static zio_t *
	zio_dva_throttle(zio_t *zio)
	{
	spa_t *spa = zio->io_spa;
	zio_t *nio;
	metaslab_class_t *mc;

	/* locate an appropriate allocation class */
	mc = spa_preferred_class(spa, zio->io_size, zio->io_prop.zp_type,
	zio->io_prop.zp_level, zio->io_prop.zp_zpl_smallblk);

	if (zio->io_priority == ZIO_PRIORITY_SYNC_WRITE \|\|
	!mc->mc_alloc_throttle_enabled \|\|
	zio->io_child_type == ZIO_CHILD_GANG \|\|
	zio->io_flags & ZIO_FLAG_NODATA) {
	return (zio);
	}

	ASSERT(zio->io_type == ZIO_TYPE_WRITE);
	ASSERT(zio->io_child_type > ZIO_CHILD_GANG);
	ASSERT3U(zio->io_queued_timestamp, >, 0);
	ASSERT(zio->io_stage == ZIO_STAGE_DVA_THROTTLE);

	zbookmark_phys_t *bm = &zio->io_bookmark;
	/*
	* We want to try to use as many allocators as possible to help improve
	* performance, but we also want logically adjacent IOs to be physically
	* adjacent to improve sequential read performance. We chunk each object
	* into 2^20 block regions, and then hash based on the objset, object,
	* level, and region to accomplish both of these goals.
	*/
	int allocator = (uint_t)cityhash4(bm->zb_objset, bm->zb_object,
	bm->zb_level, bm->zb_blkid >> 20) % spa->spa_alloc_count;
	zio->io_allocator = allocator;
	zio->io_metaslab_class = mc;
	mutex_enter(&spa->spa_allocs[allocator].spaa_lock);
	avl_add(&spa->spa_allocs[allocator].spaa_tree, zio);
	nio = zio_io_to_allocate(spa, allocator);
	mutex_exit(&spa->spa_allocs[allocator].spaa_lock);
	return (nio);
	}

	static void
	zio_allocate_dispatch(spa_t *spa, int allocator)
	{
	zio_t *zio;

	mutex_enter(&spa->spa_allocs[allocator].spaa_lock);
	zio = zio_io_to_allocate(spa, allocator);
	mutex_exit(&spa->spa_allocs[allocator].spaa_lock);
	if (zio == NULL)
	return;

	ASSERT3U(zio->io_stage, ==, ZIO_STAGE_DVA_THROTTLE);
	ASSERT0(zio->io_error);
	zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE, B_TRUE);
	}

	static zio_t *
	zio_dva_allocate(zio_t *zio)
	{
	spa_t *spa = zio->io_spa;
	metaslab_class_t *mc;
	blkptr_t *bp = zio->io_bp;
	int error;
	int flags = 0;

	if (zio->io_gang_leader == NULL) {
	ASSERT(zio->io_child_type > ZIO_CHILD_GANG);
	zio->io_gang_leader = zio;
	}

	ASSERT(BP_IS_HOLE(bp));
	ASSERT0(BP_GET_NDVAS(bp));
	ASSERT3U(zio->io_prop.zp_copies, >, 0);
	ASSERT3U(zio->io_prop.zp_copies, <=, spa_max_replication(spa));
	ASSERT3U(zio->io_size, ==, BP_GET_PSIZE(bp));

	flags \|= (zio->io_flags & ZIO_FLAG_FASTWRITE) ? METASLAB_FASTWRITE : 0;
	if (zio->io_flags & ZIO_FLAG_NODATA)
	flags \|= METASLAB_DONT_THROTTLE;
	if (zio->io_flags & ZIO_FLAG_GANG_CHILD)
	flags \|= METASLAB_GANG_CHILD;
	if (zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE)
	flags \|= METASLAB_ASYNC_ALLOC;

	/*
	* if not already chosen, locate an appropriate allocation class
	*/
	mc = zio->io_metaslab_class;
	if (mc == NULL) {
	mc = spa_preferred_class(spa, zio->io_size,
	zio->io_prop.zp_type, zio->io_prop.zp_level,
	zio->io_prop.zp_zpl_smallblk);
	zio->io_metaslab_class = mc;
	}

	/*
	* Try allocating the block in the usual metaslab class.
	* If that's full, allocate it in the normal class.
	* If that's full, allocate as a gang block,
	* and if all are full, the allocation fails (which shouldn't happen).
	*
	* Note that we do not fall back on embedded slog (ZIL) space, to
	* preserve unfragmented slog space, which is critical for decent
	* sync write performance. If a log allocation fails, we will fall
	* back to spa_sync() which is abysmal for performance.
	*/
	error = metaslab_alloc(spa, mc, zio->io_size, bp,
	zio->io_prop.zp_copies, zio->io_txg, NULL, flags,
	&zio->io_alloc_list, zio, zio->io_allocator);

	/*
	* Fallback to normal class when an alloc class is full
	*/
	if (error == ENOSPC && mc != spa_normal_class(spa)) {
	/*
	* If throttling, transfer reservation over to normal class.
	* The io_allocator slot can remain the same even though we
	* are switching classes.
	*/
	if (mc->mc_alloc_throttle_enabled &&
	(zio->io_flags & ZIO_FLAG_IO_ALLOCATING)) {
	metaslab_class_throttle_unreserve(mc,
	zio->io_prop.zp_copies, zio->io_allocator, zio);
	zio->io_flags &= ~ZIO_FLAG_IO_ALLOCATING;

	VERIFY(metaslab_class_throttle_reserve(
	spa_normal_class(spa),
	zio->io_prop.zp_copies, zio->io_allocator, zio,
	flags \| METASLAB_MUST_RESERVE));
	}
	zio->io_metaslab_class = mc = spa_normal_class(spa);
	if (zfs_flags & ZFS_DEBUG_METASLAB_ALLOC) {
	zfs_dbgmsg("%s: metaslab allocation failure, "
	"trying normal class: zio %px, size %llu, error %d",
	spa_name(spa), zio, (u_longlong_t)zio->io_size,
	error);
	}

	error = metaslab_alloc(spa, mc, zio->io_size, bp,
	zio->io_prop.zp_copies, zio->io_txg, NULL, flags,
	&zio->io_alloc_list, zio, zio->io_allocator);
	}

	if (error == ENOSPC && zio->io_size > SPA_MINBLOCKSIZE) {
	if (zfs_flags & ZFS_DEBUG_METASLAB_ALLOC) {
	zfs_dbgmsg("%s: metaslab allocation failure, "
	"trying ganging: zio %px, size %llu, error %d",
	spa_name(spa), zio, (u_longlong_t)zio->io_size,
	error);
	}
	return (zio_write_gang_block(zio, mc));
	}
	if (error != 0) {
	if (error != ENOSPC \|\|
	(zfs_flags & ZFS_DEBUG_METASLAB_ALLOC)) {
	zfs_dbgmsg("%s: metaslab allocation failure: zio %px, "
	"size %llu, error %d",
	spa_name(spa), zio, (u_longlong_t)zio->io_size,
	error);
	}
	zio->io_error = error;
	}

	return (zio);
	}

	static zio_t *
	zio_dva_free(zio_t *zio)
	{
	metaslab_free(zio->io_spa, zio->io_bp, zio->io_txg, B_FALSE);

	return (zio);
	}

	static zio_t *
	zio_dva_claim(zio_t *zio)
	{
	int error;

	error = metaslab_claim(zio->io_spa, zio->io_bp, zio->io_txg);
	if (error)
	zio->io_error = error;

	return (zio);
	}

	/*
	* Undo an allocation. This is used by zio_done() when an I/O fails
	* and we want to give back the block we just allocated.
	* This handles both normal blocks and gang blocks.
	*/
	static void
	zio_dva_unallocate(zio_t zio, zio_gang_node_t gn, blkptr_t *bp)
	{
	ASSERT(bp->blk_birth == zio->io_txg \|\| BP_IS_HOLE(bp));
	ASSERT(zio->io_bp_override == NULL);

	if (!BP_IS_HOLE(bp))
	metaslab_free(zio->io_spa, bp, bp->blk_birth, B_TRUE);

	if (gn != NULL) {
	for (int g = 0; g < SPA_GBH_NBLKPTRS; g++) {
	zio_dva_unallocate(zio, gn->gn_child[g],
	&gn->gn_gbh->zg_blkptr[g]);
	}
	}
	}

	/*
	* Try to allocate an intent log block. Return 0 on success, errno on failure.
	*/
	int
	zio_alloc_zil(spa_t spa, objset_t os, uint64_t txg, blkptr_t *new_bp,
	uint64_t size, boolean_t *slog)
	{
	int error = 1;
	zio_alloc_list_t io_alloc_list;

	ASSERT(txg > spa_syncing_txg(spa));

	metaslab_trace_init(&io_alloc_list);

	/*
	* Block pointer fields are useful to metaslabs for stats and debugging.
	* Fill in the obvious ones before calling into metaslab_alloc().
	*/
	BP_SET_TYPE(new_bp, DMU_OT_INTENT_LOG);
	BP_SET_PSIZE(new_bp, size);
	BP_SET_LEVEL(new_bp, 0);

	/*
	* When allocating a zil block, we don't have information about
	* the final destination of the block except the objset it's part
	* of, so we just hash the objset ID to pick the allocator to get
	* some parallelism.
	*/
	int flags = METASLAB_FASTWRITE \| METASLAB_ZIL;
	int allocator = (uint_t)cityhash4(0, 0, 0,
	os->os_dsl_dataset->ds_object) % spa->spa_alloc_count;
	error = metaslab_alloc(spa, spa_log_class(spa), size, new_bp, 1,
	txg, NULL, flags, &io_alloc_list, NULL, allocator);
	*slog = (error == 0);
	if (error != 0) {
	error = metaslab_alloc(spa, spa_embedded_log_class(spa), size,
	new_bp, 1, txg, NULL, flags,
	&io_alloc_list, NULL, allocator);
	}
	if (error != 0) {
	error = metaslab_alloc(spa, spa_normal_class(spa), size,
	new_bp, 1, txg, NULL, flags,
	&io_alloc_list, NULL, allocator);
	}
	metaslab_trace_fini(&io_alloc_list);

	if (error == 0) {
	BP_SET_LSIZE(new_bp, size);
	BP_SET_PSIZE(new_bp, size);
	BP_SET_COMPRESS(new_bp, ZIO_COMPRESS_OFF);
	BP_SET_CHECKSUM(new_bp,
	spa_version(spa) >= SPA_VERSION_SLIM_ZIL
	? ZIO_CHECKSUM_ZILOG2 : ZIO_CHECKSUM_ZILOG);
	BP_SET_TYPE(new_bp, DMU_OT_INTENT_LOG);
	BP_SET_LEVEL(new_bp, 0);
	BP_SET_DEDUP(new_bp, 0);
	BP_SET_BYTEORDER(new_bp, ZFS_HOST_BYTEORDER);

	/*
	* encrypted blocks will require an IV and salt. We generate
	* these now since we will not be rewriting the bp at
	* rewrite time.
	*/
	if (os->os_encrypted) {
	uint8_t iv[ZIO_DATA_IV_LEN];
	uint8_t salt[ZIO_DATA_SALT_LEN];

	BP_SET_CRYPT(new_bp, B_TRUE);
	VERIFY0(spa_crypt_get_salt(spa,
	dmu_objset_id(os), salt));
	VERIFY0(zio_crypt_generate_iv(iv));

	zio_crypt_encode_params_bp(new_bp, salt, iv);
	}
	} else {
	zfs_dbgmsg("%s: zil block allocation failure: "
	"size %llu, error %d", spa_name(spa), (u_longlong_t)size,
	error);
	}

	return (error);
	}

	/*
	* ==========================================================================
	* Read and write to physical devices
	* ==========================================================================
	*/

	/*
	* Issue an I/O to the underlying vdev. Typically the issue pipeline
	* stops after this stage and will resume upon I/O completion.
	* However, there are instances where the vdev layer may need to
	* continue the pipeline when an I/O was not issued. Since the I/O
	* that was sent to the vdev layer might be different than the one
	* currently active in the pipeline (see vdev_queue_io()), we explicitly
	* force the underlying vdev layers to call either zio_execute() or
	* zio_interrupt() to ensure that the pipeline continues with the correct I/O.
	*/
	static zio_t *
	zio_vdev_io_start(zio_t *zio)
	{
	vdev_t *vd = zio->io_vd;
	uint64_t align;
	spa_t *spa = zio->io_spa;

	zio->io_delay = 0;

	ASSERT(zio->io_error == 0);
	ASSERT(zio->io_child_error[ZIO_CHILD_VDEV] == 0);

	if (vd == NULL) {
	if (!(zio->io_flags & ZIO_FLAG_CONFIG_WRITER))
	spa_config_enter(spa, SCL_ZIO, zio, RW_READER);

	/*
	* The mirror_ops handle multiple DVAs in a single BP.
	*/
	vdev_mirror_ops.vdev_op_io_start(zio);
	return (NULL);
	}

	ASSERT3P(zio->io_logical, !=, zio);
	if (zio->io_type == ZIO_TYPE_WRITE) {
	ASSERT(spa->spa_trust_config);

	/*
	* Note: the code can handle other kinds of writes,
	* but we don't expect them.
	*/
	- if (zio->io_vd->vdev_removing) {
	+ if (zio->io_vd->vdev_noalloc) {
	ASSERT(zio->io_flags &
	(ZIO_FLAG_PHYSICAL \| ZIO_FLAG_SELF_HEAL \|
	ZIO_FLAG_RESILVER \| ZIO_FLAG_INDUCE_DAMAGE));
	}
	}

	align = 1ULL << vd->vdev_top->vdev_ashift;

	if (!(zio->io_flags & ZIO_FLAG_PHYSICAL) &&
	P2PHASE(zio->io_size, align) != 0) {
	/* Transform logical writes to be a full physical block size. */
	uint64_t asize = P2ROUNDUP(zio->io_size, align);
	abd_t *abuf = abd_alloc_sametype(zio->io_abd, asize);
	ASSERT(vd == vd->vdev_top);
	if (zio->io_type == ZIO_TYPE_WRITE) {
	abd_copy(abuf, zio->io_abd, zio->io_size);
	abd_zero_off(abuf, zio->io_size, asize - zio->io_size);
	}
	zio_push_transform(zio, abuf, asize, asize, zio_subblock);
	}

	/*
	* If this is not a physical io, make sure that it is properly aligned
	* before proceeding.
	*/
	if (!(zio->io_flags & ZIO_FLAG_PHYSICAL)) {
	ASSERT0(P2PHASE(zio->io_offset, align));
	ASSERT0(P2PHASE(zio->io_size, align));
	} else {
	/*
	* For physical writes, we allow 512b aligned writes and assume
	* the device will perform a read-modify-write as necessary.
	*/
	ASSERT0(P2PHASE(zio->io_offset, SPA_MINBLOCKSIZE));
	ASSERT0(P2PHASE(zio->io_size, SPA_MINBLOCKSIZE));
	}

	VERIFY(zio->io_type != ZIO_TYPE_WRITE \|\| spa_writeable(spa));

	/*
	* If this is a repair I/O, and there's no self-healing involved --
	* that is, we're just resilvering what we expect to resilver --
	* then don't do the I/O unless zio's txg is actually in vd's DTL.
	* This prevents spurious resilvering.
	*
	* There are a few ways that we can end up creating these spurious
	* resilver i/os:
	*
	* 1. A resilver i/o will be issued if any DVA in the BP has a
	* dirty DTL. The mirror code will issue resilver writes to
	* each DVA, including the one(s) that are not on vdevs with dirty
	* DTLs.
	*
	* 2. With nested replication, which happens when we have a
	* "replacing" or "spare" vdev that's a child of a mirror or raidz.
	* For example, given mirror(replacing(A+B), C), it's likely that
	* only A is out of date (it's the new device). In this case, we'll
	* read from C, then use the data to resilver A+B -- but we don't
	* actually want to resilver B, just A. The top-level mirror has no
	* way to know this, so instead we just discard unnecessary repairs
	* as we work our way down the vdev tree.
	*
	* 3. ZTEST also creates mirrors of mirrors, mirrors of raidz, etc.
	* The same logic applies to any form of nested replication: ditto
	* + mirror, RAID-Z + replacing, etc.
	*
	* However, indirect vdevs point off to other vdevs which may have
	* DTL's, so we never bypass them. The child i/os on concrete vdevs
	* will be properly bypassed instead.
	*
	* Leaf DTL_PARTIAL can be empty when a legitimate write comes from
	* a dRAID spare vdev. For example, when a dRAID spare is first
	* used, its spare blocks need to be written to but the leaf vdev's
	* of such blocks can have empty DTL_PARTIAL.
	*
	* There seemed no clean way to allow such writes while bypassing
	* spurious ones. At this point, just avoid all bypassing for dRAID
	* for correctness.
	*/
	if ((zio->io_flags & ZIO_FLAG_IO_REPAIR) &&
	!(zio->io_flags & ZIO_FLAG_SELF_HEAL) &&
	zio->io_txg != 0 && /* not a delegated i/o */
	vd->vdev_ops != &vdev_indirect_ops &&
	vd->vdev_top->vdev_ops != &vdev_draid_ops &&
	!vdev_dtl_contains(vd, DTL_PARTIAL, zio->io_txg, 1)) {
	ASSERT(zio->io_type == ZIO_TYPE_WRITE);
	zio_vdev_io_bypass(zio);
	return (zio);
	}

	/*
	* Select the next best leaf I/O to process. Distributed spares are
	* excluded since they dispatch the I/O directly to a leaf vdev after
	* applying the dRAID mapping.
	*/
	if (vd->vdev_ops->vdev_op_leaf &&
	vd->vdev_ops != &vdev_draid_spare_ops &&
	(zio->io_type == ZIO_TYPE_READ \|\|
	zio->io_type == ZIO_TYPE_WRITE \|\|
	zio->io_type == ZIO_TYPE_TRIM)) {

	if (zio->io_type == ZIO_TYPE_READ && vdev_cache_read(zio))
	return (zio);

	if ((zio = vdev_queue_io(zio)) == NULL)
	return (NULL);

	if (!vdev_accessible(vd, zio)) {
	zio->io_error = SET_ERROR(ENXIO);
	zio_interrupt(zio);
	return (NULL);
	}
	zio->io_delay = gethrtime();
	}

	vd->vdev_ops->vdev_op_io_start(zio);
	return (NULL);
	}

	static zio_t *
	zio_vdev_io_done(zio_t *zio)
	{
	vdev_t *vd = zio->io_vd;
	vdev_ops_t *ops = vd ? vd->vdev_ops : &vdev_mirror_ops;
	boolean_t unexpected_error = B_FALSE;

	if (zio_wait_for_children(zio, ZIO_CHILD_VDEV_BIT, ZIO_WAIT_DONE)) {
	return (NULL);
	}

	ASSERT(zio->io_type == ZIO_TYPE_READ \|\|
	zio->io_type == ZIO_TYPE_WRITE \|\| zio->io_type == ZIO_TYPE_TRIM);

	if (zio->io_delay)
	zio->io_delay = gethrtime() - zio->io_delay;

	if (vd != NULL && vd->vdev_ops->vdev_op_leaf &&
	vd->vdev_ops != &vdev_draid_spare_ops) {
	vdev_queue_io_done(zio);

	if (zio->io_type == ZIO_TYPE_WRITE)
	vdev_cache_write(zio);

	if (zio_injection_enabled && zio->io_error == 0)
	zio->io_error = zio_handle_device_injections(vd, zio,
	EIO, EILSEQ);

	if (zio_injection_enabled && zio->io_error == 0)
	zio->io_error = zio_handle_label_injection(zio, EIO);

	if (zio->io_error && zio->io_type != ZIO_TYPE_TRIM) {
	if (!vdev_accessible(vd, zio)) {
	zio->io_error = SET_ERROR(ENXIO);
	} else {
	unexpected_error = B_TRUE;
	}
	}
	}

	ops->vdev_op_io_done(zio);

	if (unexpected_error)
	VERIFY(vdev_probe(vd, zio) == NULL);

	return (zio);
	}

	/*
	* This function is used to change the priority of an existing zio that is
	* currently in-flight. This is used by the arc to upgrade priority in the
	* event that a demand read is made for a block that is currently queued
	* as a scrub or async read IO. Otherwise, the high priority read request
	* would end up having to wait for the lower priority IO.
	*/
	void
	zio_change_priority(zio_t *pio, zio_priority_t priority)
	{
	zio_t cio, cio_next;
	zio_link_t *zl = NULL;

	ASSERT3U(priority, <, ZIO_PRIORITY_NUM_QUEUEABLE);

	if (pio->io_vd != NULL && pio->io_vd->vdev_ops->vdev_op_leaf) {
	vdev_queue_change_io_priority(pio, priority);
	} else {
	pio->io_priority = priority;
	}

	mutex_enter(&pio->io_lock);
	for (cio = zio_walk_children(pio, &zl); cio != NULL; cio = cio_next) {
	cio_next = zio_walk_children(pio, &zl);
	zio_change_priority(cio, priority);
	}
	mutex_exit(&pio->io_lock);
	}

	/*
	* For non-raidz ZIOs, we can just copy aside the bad data read from the
	* disk, and use that to finish the checksum ereport later.
	*/
	static void
	zio_vsd_default_cksum_finish(zio_cksum_report_t *zcr,
	const abd_t *good_buf)
	{
	/* no processing needed */
	zfs_ereport_finish_checksum(zcr, good_buf, zcr->zcr_cbdata, B_FALSE);
	}

	/ARGSUSED/
	void
	zio_vsd_default_cksum_report(zio_t zio, zio_cksum_report_t zcr)
	{
	void *abd = abd_alloc_sametype(zio->io_abd, zio->io_size);

	abd_copy(abd, zio->io_abd, zio->io_size);

	zcr->zcr_cbinfo = zio->io_size;
	zcr->zcr_cbdata = abd;
	zcr->zcr_finish = zio_vsd_default_cksum_finish;
	zcr->zcr_free = zio_abd_free;
	}

	static zio_t *
	zio_vdev_io_assess(zio_t *zio)
	{
	vdev_t *vd = zio->io_vd;

	if (zio_wait_for_children(zio, ZIO_CHILD_VDEV_BIT, ZIO_WAIT_DONE)) {
	return (NULL);
	}

	if (vd == NULL && !(zio->io_flags & ZIO_FLAG_CONFIG_WRITER))
	spa_config_exit(zio->io_spa, SCL_ZIO, zio);

	if (zio->io_vsd != NULL) {
	zio->io_vsd_ops->vsd_free(zio);
	zio->io_vsd = NULL;
	}

	if (zio_injection_enabled && zio->io_error == 0)
	zio->io_error = zio_handle_fault_injection(zio, EIO);

	/*
	* If the I/O failed, determine whether we should attempt to retry it.
	*
	* On retry, we cut in line in the issue queue, since we don't want
	* compression/checksumming/etc. work to prevent our (cheap) IO reissue.
	*/
	if (zio->io_error && vd == NULL &&
	!(zio->io_flags & (ZIO_FLAG_DONT_RETRY \| ZIO_FLAG_IO_RETRY))) {
	ASSERT(!(zio->io_flags & ZIO_FLAG_DONT_QUEUE)); /* not a leaf */
	ASSERT(!(zio->io_flags & ZIO_FLAG_IO_BYPASS)); /* not a leaf */
	zio->io_error = 0;
	zio->io_flags \|= ZIO_FLAG_IO_RETRY \|
	ZIO_FLAG_DONT_CACHE \| ZIO_FLAG_DONT_AGGREGATE;
	zio->io_stage = ZIO_STAGE_VDEV_IO_START >> 1;
	zio_taskq_dispatch(zio, ZIO_TASKQ_ISSUE,
	zio_requeue_io_start_cut_in_line);
	return (NULL);
	}

	/*
	* If we got an error on a leaf device, convert it to ENXIO
	* if the device is not accessible at all.
	*/
	if (zio->io_error && vd != NULL && vd->vdev_ops->vdev_op_leaf &&
	!vdev_accessible(vd, zio))
	zio->io_error = SET_ERROR(ENXIO);

	/*
	* If we can't write to an interior vdev (mirror or RAID-Z),
	* set vdev_cant_write so that we stop trying to allocate from it.
	*/
	if (zio->io_error == ENXIO && zio->io_type == ZIO_TYPE_WRITE &&
	vd != NULL && !vd->vdev_ops->vdev_op_leaf) {
	vdev_dbgmsg(vd, "zio_vdev_io_assess(zio=%px) setting "
	"cant_write=TRUE due to write failure with ENXIO",
	zio);
	vd->vdev_cant_write = B_TRUE;
	}

	/*
	* If a cache flush returns ENOTSUP or ENOTTY, we know that no future
	* attempts will ever succeed. In this case we set a persistent
	* boolean flag so that we don't bother with it in the future.
	*/
	if ((zio->io_error == ENOTSUP \|\| zio->io_error == ENOTTY) &&
	zio->io_type == ZIO_TYPE_IOCTL &&
	zio->io_cmd == DKIOCFLUSHWRITECACHE && vd != NULL)
	vd->vdev_nowritecache = B_TRUE;

	if (zio->io_error)
	zio->io_pipeline = ZIO_INTERLOCK_PIPELINE;

	if (vd != NULL && vd->vdev_ops->vdev_op_leaf &&
	zio->io_physdone != NULL) {
	ASSERT(!(zio->io_flags & ZIO_FLAG_DELEGATED));
	ASSERT(zio->io_child_type == ZIO_CHILD_VDEV);
	zio->io_physdone(zio->io_logical);
	}

	return (zio);
	}

	void
	zio_vdev_io_reissue(zio_t *zio)
	{
	ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_START);
	ASSERT(zio->io_error == 0);

	zio->io_stage >>= 1;
	}

	void
	zio_vdev_io_redone(zio_t *zio)
	{
	ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_DONE);

	zio->io_stage >>= 1;
	}

	void
	zio_vdev_io_bypass(zio_t *zio)
	{
	ASSERT(zio->io_stage == ZIO_STAGE_VDEV_IO_START);
	ASSERT(zio->io_error == 0);

	zio->io_flags \|= ZIO_FLAG_IO_BYPASS;
	zio->io_stage = ZIO_STAGE_VDEV_IO_ASSESS >> 1;
	}

	/*
	* ==========================================================================
	* Encrypt and store encryption parameters
	* ==========================================================================
	*/


	/*
	* This function is used for ZIO_STAGE_ENCRYPT. It is responsible for
	* managing the storage of encryption parameters and passing them to the
	* lower-level encryption functions.
	*/
	static zio_t *
	zio_encrypt(zio_t *zio)
	{
	zio_prop_t *zp = &zio->io_prop;
	spa_t *spa = zio->io_spa;
	blkptr_t *bp = zio->io_bp;
	uint64_t psize = BP_GET_PSIZE(bp);
	uint64_t dsobj = zio->io_bookmark.zb_objset;
	dmu_object_type_t ot = BP_GET_TYPE(bp);
	void *enc_buf = NULL;
	abd_t *eabd = NULL;
	uint8_t salt[ZIO_DATA_SALT_LEN];
	uint8_t iv[ZIO_DATA_IV_LEN];
	uint8_t mac[ZIO_DATA_MAC_LEN];
	boolean_t no_crypt = B_FALSE;

	/* the root zio already encrypted the data */
	if (zio->io_child_type == ZIO_CHILD_GANG)
	return (zio);

	/* only ZIL blocks are re-encrypted on rewrite */
	if (!IO_IS_ALLOCATING(zio) && ot != DMU_OT_INTENT_LOG)
	return (zio);

	if (!(zp->zp_encrypt \|\| BP_IS_ENCRYPTED(bp))) {
	BP_SET_CRYPT(bp, B_FALSE);
	return (zio);
	}

	/* if we are doing raw encryption set the provided encryption params */
	if (zio->io_flags & ZIO_FLAG_RAW_ENCRYPT) {
	ASSERT0(BP_GET_LEVEL(bp));
	BP_SET_CRYPT(bp, B_TRUE);
	BP_SET_BYTEORDER(bp, zp->zp_byteorder);
	if (ot != DMU_OT_OBJSET)
	zio_crypt_encode_mac_bp(bp, zp->zp_mac);

	/* dnode blocks must be written out in the provided byteorder */
	if (zp->zp_byteorder != ZFS_HOST_BYTEORDER &&
	ot == DMU_OT_DNODE) {
	void *bswap_buf = zio_buf_alloc(psize);
	abd_t *babd = abd_get_from_buf(bswap_buf, psize);

	ASSERT3U(BP_GET_COMPRESS(bp), ==, ZIO_COMPRESS_OFF);
	abd_copy_to_buf(bswap_buf, zio->io_abd, psize);
	dmu_ot_byteswap[DMU_OT_BYTESWAP(ot)].ob_func(bswap_buf,
	psize);

	abd_take_ownership_of_buf(babd, B_TRUE);
	zio_push_transform(zio, babd, psize, psize, NULL);
	}

	if (DMU_OT_IS_ENCRYPTED(ot))
	zio_crypt_encode_params_bp(bp, zp->zp_salt, zp->zp_iv);
	return (zio);
	}

	/* indirect blocks only maintain a cksum of the lower level MACs */
	if (BP_GET_LEVEL(bp) > 0) {
	BP_SET_CRYPT(bp, B_TRUE);
	VERIFY0(zio_crypt_do_indirect_mac_checksum_abd(B_TRUE,
	zio->io_orig_abd, BP_GET_LSIZE(bp), BP_SHOULD_BYTESWAP(bp),
	mac));
	zio_crypt_encode_mac_bp(bp, mac);
	return (zio);
	}

	/*
	* Objset blocks are a special case since they have 2 256-bit MACs
	* embedded within them.
	*/
	if (ot == DMU_OT_OBJSET) {
	ASSERT0(DMU_OT_IS_ENCRYPTED(ot));
	ASSERT3U(BP_GET_COMPRESS(bp), ==, ZIO_COMPRESS_OFF);
	BP_SET_CRYPT(bp, B_TRUE);
	VERIFY0(spa_do_crypt_objset_mac_abd(B_TRUE, spa, dsobj,
	zio->io_abd, psize, BP_SHOULD_BYTESWAP(bp)));
	return (zio);
	}

	/* unencrypted object types are only authenticated with a MAC */
	if (!DMU_OT_IS_ENCRYPTED(ot)) {
	BP_SET_CRYPT(bp, B_TRUE);
	VERIFY0(spa_do_crypt_mac_abd(B_TRUE, spa, dsobj,
	zio->io_abd, psize, mac));
	zio_crypt_encode_mac_bp(bp, mac);
	return (zio);
	}

	/*
	* Later passes of sync-to-convergence may decide to rewrite data
	* in place to avoid more disk reallocations. This presents a problem
	* for encryption because this constitutes rewriting the new data with
	* the same encryption key and IV. However, this only applies to blocks
	* in the MOS (particularly the spacemaps) and we do not encrypt the
	* MOS. We assert that the zio is allocating or an intent log write
	* to enforce this.
	*/
	ASSERT(IO_IS_ALLOCATING(zio) \|\| ot == DMU_OT_INTENT_LOG);
	ASSERT(BP_GET_LEVEL(bp) == 0 \|\| ot == DMU_OT_INTENT_LOG);
	ASSERT(spa_feature_is_active(spa, SPA_FEATURE_ENCRYPTION));
	ASSERT3U(psize, !=, 0);

	enc_buf = zio_buf_alloc(psize);
	eabd = abd_get_from_buf(enc_buf, psize);
	abd_take_ownership_of_buf(eabd, B_TRUE);

	/*
	* For an explanation of what encryption parameters are stored
	* where, see the block comment in zio_crypt.c.
	*/
	if (ot == DMU_OT_INTENT_LOG) {
	zio_crypt_decode_params_bp(bp, salt, iv);
	} else {
	BP_SET_CRYPT(bp, B_TRUE);
	}

	/* Perform the encryption. This should not fail */
	VERIFY0(spa_do_crypt_abd(B_TRUE, spa, &zio->io_bookmark,
	BP_GET_TYPE(bp), BP_GET_DEDUP(bp), BP_SHOULD_BYTESWAP(bp),
	salt, iv, mac, psize, zio->io_abd, eabd, &no_crypt));

	/* encode encryption metadata into the bp */
	if (ot == DMU_OT_INTENT_LOG) {
	/*
	* ZIL blocks store the MAC in the embedded checksum, so the
	* transform must always be applied.
	*/
	zio_crypt_encode_mac_zil(enc_buf, mac);
	zio_push_transform(zio, eabd, psize, psize, NULL);
	} else {
	BP_SET_CRYPT(bp, B_TRUE);
	zio_crypt_encode_params_bp(bp, salt, iv);
	zio_crypt_encode_mac_bp(bp, mac);

	if (no_crypt) {
	ASSERT3U(ot, ==, DMU_OT_DNODE);
	abd_free(eabd);
	} else {
	zio_push_transform(zio, eabd, psize, psize, NULL);
	}
	}

	return (zio);
	}

	/*
	* ==========================================================================
	* Generate and verify checksums
	* ==========================================================================
	*/
	static zio_t *
	zio_checksum_generate(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;
	enum zio_checksum checksum;

	if (bp == NULL) {
	/*
	* This is zio_write_phys().
	* We're either generating a label checksum, or none at all.
	*/
	checksum = zio->io_prop.zp_checksum;

	if (checksum == ZIO_CHECKSUM_OFF)
	return (zio);

	ASSERT(checksum == ZIO_CHECKSUM_LABEL);
	} else {
	if (BP_IS_GANG(bp) && zio->io_child_type == ZIO_CHILD_GANG) {
	ASSERT(!IO_IS_ALLOCATING(zio));
	checksum = ZIO_CHECKSUM_GANG_HEADER;
	} else {
	checksum = BP_GET_CHECKSUM(bp);
	}
	}

	zio_checksum_compute(zio, checksum, zio->io_abd, zio->io_size);

	return (zio);
	}

	static zio_t *
	zio_checksum_verify(zio_t *zio)
	{
	zio_bad_cksum_t info;
	blkptr_t *bp = zio->io_bp;
	int error;

	ASSERT(zio->io_vd != NULL);

	if (bp == NULL) {
	/*
	* This is zio_read_phys().
	* We're either verifying a label checksum, or nothing at all.
	*/
	if (zio->io_prop.zp_checksum == ZIO_CHECKSUM_OFF)
	return (zio);

	ASSERT3U(zio->io_prop.zp_checksum, ==, ZIO_CHECKSUM_LABEL);
	}

	if ((error = zio_checksum_error(zio, &info)) != 0) {
	zio->io_error = error;
	if (error == ECKSUM &&
	!(zio->io_flags & ZIO_FLAG_SPECULATIVE)) {
	(void) zfs_ereport_start_checksum(zio->io_spa,
	zio->io_vd, &zio->io_bookmark, zio,
	zio->io_offset, zio->io_size, &info);
	mutex_enter(&zio->io_vd->vdev_stat_lock);
	zio->io_vd->vdev_stat.vs_checksum_errors++;
	mutex_exit(&zio->io_vd->vdev_stat_lock);
	}
	}

	return (zio);
	}

	/*
	* Called by RAID-Z to ensure we don't compute the checksum twice.
	*/
	void
	zio_checksum_verified(zio_t *zio)
	{
	zio->io_pipeline &= ~ZIO_STAGE_CHECKSUM_VERIFY;
	}

	/*
	* ==========================================================================
	* Error rank. Error are ranked in the order 0, ENXIO, ECKSUM, EIO, other.
	* An error of 0 indicates success. ENXIO indicates whole-device failure,
	* which may be transient (e.g. unplugged) or permanent. ECKSUM and EIO
	* indicate errors that are specific to one I/O, and most likely permanent.
	* Any other error is presumed to be worse because we weren't expecting it.
	* ==========================================================================
	*/
	int
	zio_worst_error(int e1, int e2)
	{
	static int zio_error_rank[] = { 0, ENXIO, ECKSUM, EIO };
	int r1, r2;

	for (r1 = 0; r1 < sizeof (zio_error_rank) / sizeof (int); r1++)
	if (e1 == zio_error_rank[r1])
	break;

	for (r2 = 0; r2 < sizeof (zio_error_rank) / sizeof (int); r2++)
	if (e2 == zio_error_rank[r2])
	break;

	return (r1 > r2 ? e1 : e2);
	}

	/*
	* ==========================================================================
	* I/O completion
	* ==========================================================================
	*/
	static zio_t *
	zio_ready(zio_t *zio)
	{
	blkptr_t *bp = zio->io_bp;
	zio_t pio, pio_next;
	zio_link_t *zl = NULL;

	if (zio_wait_for_children(zio, ZIO_CHILD_GANG_BIT \| ZIO_CHILD_DDT_BIT,
	ZIO_WAIT_READY)) {
	return (NULL);
	}

	if (zio->io_ready) {
	ASSERT(IO_IS_ALLOCATING(zio));
	ASSERT(bp->blk_birth == zio->io_txg \|\| BP_IS_HOLE(bp) \|\|
	(zio->io_flags & ZIO_FLAG_NOPWRITE));
	ASSERT(zio->io_children[ZIO_CHILD_GANG][ZIO_WAIT_READY] == 0);

	zio->io_ready(zio);
	}

	if (bp != NULL && bp != &zio->io_bp_copy)
	zio->io_bp_copy = *bp;

	if (zio->io_error != 0) {
	zio->io_pipeline = ZIO_INTERLOCK_PIPELINE;

	if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING) {
	ASSERT(IO_IS_ALLOCATING(zio));
	ASSERT(zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE);
	ASSERT(zio->io_metaslab_class != NULL);

	/*
	* We were unable to allocate anything, unreserve and
	* issue the next I/O to allocate.
	*/
	metaslab_class_throttle_unreserve(
	zio->io_metaslab_class, zio->io_prop.zp_copies,
	zio->io_allocator, zio);
	zio_allocate_dispatch(zio->io_spa, zio->io_allocator);
	}
	}

	mutex_enter(&zio->io_lock);
	zio->io_state[ZIO_WAIT_READY] = 1;
	pio = zio_walk_parents(zio, &zl);
	mutex_exit(&zio->io_lock);

	/*
	* As we notify zio's parents, new parents could be added.
	* New parents go to the head of zio's io_parent_list, however,
	* so we will (correctly) not notify them. The remainder of zio's
	* io_parent_list, from 'pio_next' onward, cannot change because
	* all parents must wait for us to be done before they can be done.
	*/
	for (; pio != NULL; pio = pio_next) {
	pio_next = zio_walk_parents(zio, &zl);
	zio_notify_parent(pio, zio, ZIO_WAIT_READY, NULL);
	}

	if (zio->io_flags & ZIO_FLAG_NODATA) {
	if (BP_IS_GANG(bp)) {
	zio->io_flags &= ~ZIO_FLAG_NODATA;
	} else {
	ASSERT((uintptr_t)zio->io_abd < SPA_MAXBLOCKSIZE);
	zio->io_pipeline &= ~ZIO_VDEV_IO_STAGES;
	}
	}

	if (zio_injection_enabled &&
	zio->io_spa->spa_syncing_txg == zio->io_txg)
	zio_handle_ignored_writes(zio);

	return (zio);
	}

	/*
	* Update the allocation throttle accounting.
	*/
	static void
	zio_dva_throttle_done(zio_t *zio)
	{
	zio_t *lio __maybe_unused = zio->io_logical;
	zio_t *pio = zio_unique_parent(zio);
	vdev_t *vd = zio->io_vd;
	int flags = METASLAB_ASYNC_ALLOC;

	ASSERT3P(zio->io_bp, !=, NULL);
	ASSERT3U(zio->io_type, ==, ZIO_TYPE_WRITE);
	ASSERT3U(zio->io_priority, ==, ZIO_PRIORITY_ASYNC_WRITE);
	ASSERT3U(zio->io_child_type, ==, ZIO_CHILD_VDEV);
	ASSERT(vd != NULL);
	ASSERT3P(vd, ==, vd->vdev_top);
	ASSERT(zio_injection_enabled \|\| !(zio->io_flags & ZIO_FLAG_IO_RETRY));
	ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REPAIR));
	ASSERT(zio->io_flags & ZIO_FLAG_IO_ALLOCATING);
	ASSERT(!(lio->io_flags & ZIO_FLAG_IO_REWRITE));
	ASSERT(!(lio->io_orig_flags & ZIO_FLAG_NODATA));

	/*
	* Parents of gang children can have two flavors -- ones that
	* allocated the gang header (will have ZIO_FLAG_IO_REWRITE set)
	* and ones that allocated the constituent blocks. The allocation
	* throttle needs to know the allocating parent zio so we must find
	* it here.
	*/
	if (pio->io_child_type == ZIO_CHILD_GANG) {
	/*
	* If our parent is a rewrite gang child then our grandparent
	* would have been the one that performed the allocation.
	*/
	if (pio->io_flags & ZIO_FLAG_IO_REWRITE)
	pio = zio_unique_parent(pio);
	flags \|= METASLAB_GANG_CHILD;
	}

	ASSERT(IO_IS_ALLOCATING(pio));
	ASSERT3P(zio, !=, zio->io_logical);
	ASSERT(zio->io_logical != NULL);
	ASSERT(!(zio->io_flags & ZIO_FLAG_IO_REPAIR));
	ASSERT0(zio->io_flags & ZIO_FLAG_NOPWRITE);
	ASSERT(zio->io_metaslab_class != NULL);

	mutex_enter(&pio->io_lock);
	metaslab_group_alloc_decrement(zio->io_spa, vd->vdev_id, pio, flags,
	pio->io_allocator, B_TRUE);
	mutex_exit(&pio->io_lock);

	metaslab_class_throttle_unreserve(zio->io_metaslab_class, 1,
	pio->io_allocator, pio);

	/*
	* Call into the pipeline to see if there is more work that
	* needs to be done. If there is work to be done it will be
	* dispatched to another taskq thread.
	*/
	zio_allocate_dispatch(zio->io_spa, pio->io_allocator);
	}

	static zio_t *
	zio_done(zio_t *zio)
	{
	/*
	* Always attempt to keep stack usage minimal here since
	* we can be called recursively up to 19 levels deep.
	*/
	const uint64_t psize = zio->io_size;
	zio_t pio, pio_next;
	zio_link_t *zl = NULL;

	/*
	* If our children haven't all completed,
	* wait for them and then repeat this pipeline stage.
	*/
	if (zio_wait_for_children(zio, ZIO_CHILD_ALL_BITS, ZIO_WAIT_DONE)) {
	return (NULL);
	}

	/*
	* If the allocation throttle is enabled, then update the accounting.
	* We only track child I/Os that are part of an allocating async
	* write. We must do this since the allocation is performed
	* by the logical I/O but the actual write is done by child I/Os.
	*/
	if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING &&
	zio->io_child_type == ZIO_CHILD_VDEV) {
	ASSERT(zio->io_metaslab_class != NULL);
	ASSERT(zio->io_metaslab_class->mc_alloc_throttle_enabled);
	zio_dva_throttle_done(zio);
	}

	/*
	* If the allocation throttle is enabled, verify that
	* we have decremented the refcounts for every I/O that was throttled.
	*/
	if (zio->io_flags & ZIO_FLAG_IO_ALLOCATING) {
	ASSERT(zio->io_type == ZIO_TYPE_WRITE);
	ASSERT(zio->io_priority == ZIO_PRIORITY_ASYNC_WRITE);
	ASSERT(zio->io_bp != NULL);

	metaslab_group_alloc_verify(zio->io_spa, zio->io_bp, zio,
	zio->io_allocator);
	VERIFY(zfs_refcount_not_held(&zio->io_metaslab_class->
	mc_allocator[zio->io_allocator].mca_alloc_slots, zio));
	}


	for (int c = 0; c < ZIO_CHILD_TYPES; c++)
	for (int w = 0; w < ZIO_WAIT_TYPES; w++)
	ASSERT(zio->io_children[c][w] == 0);

	if (zio->io_bp != NULL && !BP_IS_EMBEDDED(zio->io_bp)) {
	ASSERT(zio->io_bp->blk_pad[0] == 0);
	ASSERT(zio->io_bp->blk_pad[1] == 0);
	ASSERT(bcmp(zio->io_bp, &zio->io_bp_copy,
	sizeof (blkptr_t)) == 0 \|\|
	(zio->io_bp == zio_unique_parent(zio)->io_bp));
	if (zio->io_type == ZIO_TYPE_WRITE && !BP_IS_HOLE(zio->io_bp) &&
	zio->io_bp_override == NULL &&
	!(zio->io_flags & ZIO_FLAG_IO_REPAIR)) {
	ASSERT3U(zio->io_prop.zp_copies, <=,
	BP_GET_NDVAS(zio->io_bp));
	ASSERT(BP_COUNT_GANG(zio->io_bp) == 0 \|\|
	(BP_COUNT_GANG(zio->io_bp) ==
	BP_GET_NDVAS(zio->io_bp)));
	}
	if (zio->io_flags & ZIO_FLAG_NOPWRITE)
	VERIFY(BP_EQUAL(zio->io_bp, &zio->io_bp_orig));
	}

	/*
	* If there were child vdev/gang/ddt errors, they apply to us now.
	*/
	zio_inherit_child_errors(zio, ZIO_CHILD_VDEV);
	zio_inherit_child_errors(zio, ZIO_CHILD_GANG);
	zio_inherit_child_errors(zio, ZIO_CHILD_DDT);

	/*
	* If the I/O on the transformed data was successful, generate any
	* checksum reports now while we still have the transformed data.
	*/
	if (zio->io_error == 0) {
	while (zio->io_cksum_report != NULL) {
	zio_cksum_report_t *zcr = zio->io_cksum_report;
	uint64_t align = zcr->zcr_align;
	uint64_t asize = P2ROUNDUP(psize, align);
	abd_t *adata = zio->io_abd;

	if (adata != NULL && asize != psize) {
	adata = abd_alloc(asize, B_TRUE);
	abd_copy(adata, zio->io_abd, psize);
	abd_zero_off(adata, psize, asize - psize);
	}

	zio->io_cksum_report = zcr->zcr_next;
	zcr->zcr_next = NULL;
	zcr->zcr_finish(zcr, adata);
	zfs_ereport_free_checksum(zcr);

	if (adata != NULL && asize != psize)
	abd_free(adata);
	}
	}

	zio_pop_transforms(zio); /* note: may set zio->io_error */

	vdev_stat_update(zio, psize);

	/*
	* If this I/O is attached to a particular vdev is slow, exceeding
	* 30 seconds to complete, post an error described the I/O delay.
	* We ignore these errors if the device is currently unavailable.
	*/
	if (zio->io_delay >= MSEC2NSEC(zio_slow_io_ms)) {
	if (zio->io_vd != NULL && !vdev_is_dead(zio->io_vd)) {
	/*
	* We want to only increment our slow IO counters if
	* the IO is valid (i.e. not if the drive is removed).
	*
	* zfs_ereport_post() will also do these checks, but
	* it can also ratelimit and have other failures, so we
	* need to increment the slow_io counters independent
	* of it.
	*/
	if (zfs_ereport_is_valid(FM_EREPORT_ZFS_DELAY,
	zio->io_spa, zio->io_vd, zio)) {
	mutex_enter(&zio->io_vd->vdev_stat_lock);
	zio->io_vd->vdev_stat.vs_slow_ios++;
	mutex_exit(&zio->io_vd->vdev_stat_lock);

	(void) zfs_ereport_post(FM_EREPORT_ZFS_DELAY,
	zio->io_spa, zio->io_vd, &zio->io_bookmark,
	zio, 0);
	}
	}
	}

	if (zio->io_error) {
	/*
	* If this I/O is attached to a particular vdev,
	* generate an error message describing the I/O failure
	* at the block level. We ignore these errors if the
	* device is currently unavailable.
	*/
	if (zio->io_error != ECKSUM && zio->io_vd != NULL &&
	!vdev_is_dead(zio->io_vd)) {
	int ret = zfs_ereport_post(FM_EREPORT_ZFS_IO,
	zio->io_spa, zio->io_vd, &zio->io_bookmark, zio, 0);
	if (ret != EALREADY) {
	mutex_enter(&zio->io_vd->vdev_stat_lock);
	if (zio->io_type == ZIO_TYPE_READ)
	zio->io_vd->vdev_stat.vs_read_errors++;
	else if (zio->io_type == ZIO_TYPE_WRITE)
	zio->io_vd->vdev_stat.vs_write_errors++;
	mutex_exit(&zio->io_vd->vdev_stat_lock);
	}
	}

	if ((zio->io_error == EIO \|\| !(zio->io_flags &
	(ZIO_FLAG_SPECULATIVE \| ZIO_FLAG_DONT_PROPAGATE))) &&
	zio == zio->io_logical) {
	/*
	* For logical I/O requests, tell the SPA to log the
	* error and generate a logical data ereport.
	*/
	spa_log_error(zio->io_spa, &zio->io_bookmark);
	(void) zfs_ereport_post(FM_EREPORT_ZFS_DATA,
	zio->io_spa, NULL, &zio->io_bookmark, zio, 0);
	}
	}

	if (zio->io_error && zio == zio->io_logical) {
	/*
	* Determine whether zio should be reexecuted. This will
	* propagate all the way to the root via zio_notify_parent().
	*/
	ASSERT(zio->io_vd == NULL && zio->io_bp != NULL);
	ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL);

	if (IO_IS_ALLOCATING(zio) &&
	!(zio->io_flags & ZIO_FLAG_CANFAIL)) {
	if (zio->io_error != ENOSPC)
	zio->io_reexecute \|= ZIO_REEXECUTE_NOW;
	else
	zio->io_reexecute \|= ZIO_REEXECUTE_SUSPEND;
	}

	if ((zio->io_type == ZIO_TYPE_READ \|\|
	zio->io_type == ZIO_TYPE_FREE) &&
	!(zio->io_flags & ZIO_FLAG_SCAN_THREAD) &&
	zio->io_error == ENXIO &&
	spa_load_state(zio->io_spa) == SPA_LOAD_NONE &&
	spa_get_failmode(zio->io_spa) != ZIO_FAILURE_MODE_CONTINUE)
	zio->io_reexecute \|= ZIO_REEXECUTE_SUSPEND;

	if (!(zio->io_flags & ZIO_FLAG_CANFAIL) && !zio->io_reexecute)
	zio->io_reexecute \|= ZIO_REEXECUTE_SUSPEND;

	/*
	* Here is a possibly good place to attempt to do
	* either combinatorial reconstruction or error correction
	* based on checksums. It also might be a good place
	* to send out preliminary ereports before we suspend
	* processing.
	*/
	}

	/*
	* If there were logical child errors, they apply to us now.
	* We defer this until now to avoid conflating logical child
	* errors with errors that happened to the zio itself when
	* updating vdev stats and reporting FMA events above.
	*/
	zio_inherit_child_errors(zio, ZIO_CHILD_LOGICAL);

	if ((zio->io_error \|\| zio->io_reexecute) &&
	IO_IS_ALLOCATING(zio) && zio->io_gang_leader == zio &&
	!(zio->io_flags & (ZIO_FLAG_IO_REWRITE \| ZIO_FLAG_NOPWRITE)))
	zio_dva_unallocate(zio, zio->io_gang_tree, zio->io_bp);

	zio_gang_tree_free(&zio->io_gang_tree);

	/*
	* Godfather I/Os should never suspend.
	*/
	if ((zio->io_flags & ZIO_FLAG_GODFATHER) &&
	(zio->io_reexecute & ZIO_REEXECUTE_SUSPEND))
	zio->io_reexecute &= ~ZIO_REEXECUTE_SUSPEND;

	if (zio->io_reexecute) {
	/*
	* This is a logical I/O that wants to reexecute.
	*
	* Reexecute is top-down. When an i/o fails, if it's not
	* the root, it simply notifies its parent and sticks around.
	* The parent, seeing that it still has children in zio_done(),
	* does the same. This percolates all the way up to the root.
	* The root i/o will reexecute or suspend the entire tree.
	*
	* This approach ensures that zio_reexecute() honors
	* all the original i/o dependency relationships, e.g.
	* parents not executing until children are ready.
	*/
	ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL);

	zio->io_gang_leader = NULL;

	mutex_enter(&zio->io_lock);
	zio->io_state[ZIO_WAIT_DONE] = 1;
	mutex_exit(&zio->io_lock);

	/*
	* "The Godfather" I/O monitors its children but is
	* not a true parent to them. It will track them through
	* the pipeline but severs its ties whenever they get into
	* trouble (e.g. suspended). This allows "The Godfather"
	* I/O to return status without blocking.
	*/
	zl = NULL;
	for (pio = zio_walk_parents(zio, &zl); pio != NULL;
	pio = pio_next) {
	zio_link_t *remove_zl = zl;
	pio_next = zio_walk_parents(zio, &zl);

	if ((pio->io_flags & ZIO_FLAG_GODFATHER) &&
	(zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) {
	zio_remove_child(pio, zio, remove_zl);
	/*
	* This is a rare code path, so we don't
	* bother with "next_to_execute".
	*/
	zio_notify_parent(pio, zio, ZIO_WAIT_DONE,
	NULL);
	}
	}

	if ((pio = zio_unique_parent(zio)) != NULL) {
	/*
	* We're not a root i/o, so there's nothing to do
	* but notify our parent. Don't propagate errors
	* upward since we haven't permanently failed yet.
	*/
	ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER));
	zio->io_flags \|= ZIO_FLAG_DONT_PROPAGATE;
	/*
	* This is a rare code path, so we don't bother with
	* "next_to_execute".
	*/
	zio_notify_parent(pio, zio, ZIO_WAIT_DONE, NULL);
	} else if (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND) {
	/*
	* We'd fail again if we reexecuted now, so suspend
	* until conditions improve (e.g. device comes online).
	*/
	zio_suspend(zio->io_spa, zio, ZIO_SUSPEND_IOERR);
	} else {
	/*
	* Reexecution is potentially a huge amount of work.
	* Hand it off to the otherwise-unused claim taskq.
	*/
	ASSERT(taskq_empty_ent(&zio->io_tqent));
	spa_taskq_dispatch_ent(zio->io_spa,
	ZIO_TYPE_CLAIM, ZIO_TASKQ_ISSUE,
	zio_reexecute, zio, 0, &zio->io_tqent);
	}
	return (NULL);
	}

	ASSERT(zio->io_child_count == 0);
	ASSERT(zio->io_reexecute == 0);
	ASSERT(zio->io_error == 0 \|\| (zio->io_flags & ZIO_FLAG_CANFAIL));

	/*
	* Report any checksum errors, since the I/O is complete.
	*/
	while (zio->io_cksum_report != NULL) {
	zio_cksum_report_t *zcr = zio->io_cksum_report;
	zio->io_cksum_report = zcr->zcr_next;
	zcr->zcr_next = NULL;
	zcr->zcr_finish(zcr, NULL);
	zfs_ereport_free_checksum(zcr);
	}

	if (zio->io_flags & ZIO_FLAG_FASTWRITE && zio->io_bp &&
	!BP_IS_HOLE(zio->io_bp) && !BP_IS_EMBEDDED(zio->io_bp) &&
	!(zio->io_flags & ZIO_FLAG_NOPWRITE)) {
	metaslab_fastwrite_unmark(zio->io_spa, zio->io_bp);
	}

	/*
	* It is the responsibility of the done callback to ensure that this
	* particular zio is no longer discoverable for adoption, and as
	* such, cannot acquire any new parents.
	*/
	if (zio->io_done)
	zio->io_done(zio);

	mutex_enter(&zio->io_lock);
	zio->io_state[ZIO_WAIT_DONE] = 1;
	mutex_exit(&zio->io_lock);

	/*
	* We are done executing this zio. We may want to execute a parent
	* next. See the comment in zio_notify_parent().
	*/
	zio_t *next_to_execute = NULL;
	zl = NULL;
	for (pio = zio_walk_parents(zio, &zl); pio != NULL; pio = pio_next) {
	zio_link_t *remove_zl = zl;
	pio_next = zio_walk_parents(zio, &zl);
	zio_remove_child(pio, zio, remove_zl);
	zio_notify_parent(pio, zio, ZIO_WAIT_DONE, &next_to_execute);
	}

	if (zio->io_waiter != NULL) {
	mutex_enter(&zio->io_lock);
	zio->io_executor = NULL;
	cv_broadcast(&zio->io_cv);
	mutex_exit(&zio->io_lock);
	} else {
	zio_destroy(zio);
	}

	return (next_to_execute);
	}

	/*
	* ==========================================================================
	* I/O pipeline definition
	* ==========================================================================
	*/
	static zio_pipe_stage_t *zio_pipeline[] = {
	NULL,
	zio_read_bp_init,
	zio_write_bp_init,
	zio_free_bp_init,
	zio_issue_async,
	zio_write_compress,
	zio_encrypt,
	zio_checksum_generate,
	zio_nop_write,
	zio_ddt_read_start,
	zio_ddt_read_done,
	zio_ddt_write,
	zio_ddt_free,
	zio_gang_assemble,
	zio_gang_issue,
	zio_dva_throttle,
	zio_dva_allocate,
	zio_dva_free,
	zio_dva_claim,
	zio_ready,
	zio_vdev_io_start,
	zio_vdev_io_done,
	zio_vdev_io_assess,
	zio_checksum_verify,
	zio_done
	};




	/*
	* Compare two zbookmark_phys_t's to see which we would reach first in a
	* pre-order traversal of the object tree.
	*
	* This is simple in every case aside from the meta-dnode object. For all other
	* objects, we traverse them in order (object 1 before object 2, and so on).
	* However, all of these objects are traversed while traversing object 0, since
	* the data it points to is the list of objects. Thus, we need to convert to a
	* canonical representation so we can compare meta-dnode bookmarks to
	* non-meta-dnode bookmarks.
	*
	* We do this by calculating "equivalents" for each field of the zbookmark.
	* zbookmarks outside of the meta-dnode use their own object and level, and
	* calculate the level 0 equivalent (the first L0 blkid that is contained in the
	* blocks this bookmark refers to) by multiplying their blkid by their span
	* (the number of L0 blocks contained within one block at their level).
	* zbookmarks inside the meta-dnode calculate their object equivalent
	* (which is L0equiv * dnodes per data block), use 0 for their L0equiv, and use
	* level + 1<<31 (any value larger than a level could ever be) for their level.
	* This causes them to always compare before a bookmark in their object
	* equivalent, compare appropriately to bookmarks in other objects, and to
	* compare appropriately to other bookmarks in the meta-dnode.
	*/
	int
	zbookmark_compare(uint16_t dbss1, uint8_t ibs1, uint16_t dbss2, uint8_t ibs2,
	const zbookmark_phys_t zb1, const zbookmark_phys_t zb2)
	{
	/*
	* These variables represent the "equivalent" values for the zbookmark,
	* after converting zbookmarks inside the meta dnode to their
	* normal-object equivalents.
	*/
	uint64_t zb1obj, zb2obj;
	uint64_t zb1L0, zb2L0;
	uint64_t zb1level, zb2level;

	if (zb1->zb_object == zb2->zb_object &&
	zb1->zb_level == zb2->zb_level &&
	zb1->zb_blkid == zb2->zb_blkid)
	return (0);

	IMPLY(zb1->zb_level > 0, ibs1 >= SPA_MINBLOCKSHIFT);
	IMPLY(zb2->zb_level > 0, ibs2 >= SPA_MINBLOCKSHIFT);

	/*
	* BP_SPANB calculates the span in blocks.
	*/
	zb1L0 = (zb1->zb_blkid) * BP_SPANB(ibs1, zb1->zb_level);
	zb2L0 = (zb2->zb_blkid) * BP_SPANB(ibs2, zb2->zb_level);

	if (zb1->zb_object == DMU_META_DNODE_OBJECT) {
	zb1obj = zb1L0 * (dbss1 << (SPA_MINBLOCKSHIFT - DNODE_SHIFT));
	zb1L0 = 0;
	zb1level = zb1->zb_level + COMPARE_META_LEVEL;
	} else {
	zb1obj = zb1->zb_object;
	zb1level = zb1->zb_level;
	}

	if (zb2->zb_object == DMU_META_DNODE_OBJECT) {
	zb2obj = zb2L0 * (dbss2 << (SPA_MINBLOCKSHIFT - DNODE_SHIFT));
	zb2L0 = 0;
	zb2level = zb2->zb_level + COMPARE_META_LEVEL;
	} else {
	zb2obj = zb2->zb_object;
	zb2level = zb2->zb_level;
	}

	/* Now that we have a canonical representation, do the comparison. */
	if (zb1obj != zb2obj)
	return (zb1obj < zb2obj ? -1 : 1);
	else if (zb1L0 != zb2L0)
	return (zb1L0 < zb2L0 ? -1 : 1);
	else if (zb1level != zb2level)
	return (zb1level > zb2level ? -1 : 1);
	/*
	* This can (theoretically) happen if the bookmarks have the same object
	* and level, but different blkids, if the block sizes are not the same.
	* There is presently no way to change the indirect block sizes
	*/
	return (0);
	}

	/*
	* This function checks the following: given that last_block is the place that
	* our traversal stopped last time, does that guarantee that we've visited
	* every node under subtree_root? Therefore, we can't just use the raw output
	* of zbookmark_compare. We have to pass in a modified version of
	* subtree_root; by incrementing the block id, and then checking whether
	* last_block is before or equal to that, we can tell whether or not having
	* visited last_block implies that all of subtree_root's children have been
	* visited.
	*/
	boolean_t
	zbookmark_subtree_completed(const dnode_phys_t *dnp,
	const zbookmark_phys_t subtree_root, const zbookmark_phys_t last_block)
	{
	zbookmark_phys_t mod_zb = *subtree_root;
	mod_zb.zb_blkid++;
	ASSERT(last_block->zb_level == 0);

	/* The objset_phys_t isn't before anything. */
	if (dnp == NULL)
	return (B_FALSE);

	/*
	* We pass in 1ULL << (DNODE_BLOCK_SHIFT - SPA_MINBLOCKSHIFT) for the
	* data block size in sectors, because that variable is only used if
	* the bookmark refers to a block in the meta-dnode. Since we don't
	* know without examining it what object it refers to, and there's no
	* harm in passing in this value in other cases, we always pass it in.
	*
	* We pass in 0 for the indirect block size shift because zb2 must be
	* level 0. The indirect block size is only used to calculate the span
	* of the bookmark, but since the bookmark must be level 0, the span is
	* always 1, so the math works out.
	*
	* If you make changes to how the zbookmark_compare code works, be sure
	* to make sure that this code still works afterwards.
	*/
	return (zbookmark_compare(dnp->dn_datablkszsec, dnp->dn_indblkshift,
	1ULL << (DNODE_BLOCK_SHIFT - SPA_MINBLOCKSHIFT), 0, &mod_zb,
	last_block) <= 0);
	}

	EXPORT_SYMBOL(zio_type_name);
	EXPORT_SYMBOL(zio_buf_alloc);
	EXPORT_SYMBOL(zio_data_buf_alloc);
	EXPORT_SYMBOL(zio_buf_free);
	EXPORT_SYMBOL(zio_data_buf_free);

	/* BEGIN CSTYLED */
	ZFS_MODULE_PARAM(zfs_zio, zio_, slow_io_ms, INT, ZMOD_RW,
	"Max I/O completion time (milliseconds) before marking it as slow");

	ZFS_MODULE_PARAM(zfs_zio, zio_, requeue_io_start_cut_in_line, INT, ZMOD_RW,
	"Prioritize requeued I/O");

	ZFS_MODULE_PARAM(zfs, zfs_, sync_pass_deferred_free, INT, ZMOD_RW,
	"Defer frees starting in this pass");

	ZFS_MODULE_PARAM(zfs, zfs_, sync_pass_dont_compress, INT, ZMOD_RW,
	"Don't compress starting in this pass");

	ZFS_MODULE_PARAM(zfs, zfs_, sync_pass_rewrite, INT, ZMOD_RW,
	"Rewrite new bps starting in this pass");

	ZFS_MODULE_PARAM(zfs_zio, zio_, dva_throttle_enabled, INT, ZMOD_RW,
	"Throttle block allocations in the ZIO pipeline");

	ZFS_MODULE_PARAM(zfs_zio, zio_, deadman_log_all, INT, ZMOD_RW,
	"Log all slow ZIOs, not just those with vdevs");
	/* END CSTYLED */
	diff --git a/sys/contrib/openzfs/module/zfs/zvol.c b/sys/contrib/openzfs/module/zfs/zvol.c
	index d50cce7d7357..cb5ee483c1c1 100644
	--- a/sys/contrib/openzfs/module/zfs/zvol.c
	+++ b/sys/contrib/openzfs/module/zfs/zvol.c
	@@ -1,1772 +1,1744 @@
	/*
	* CDDL HEADER START
	*
	* The contents of this file are subject to the terms of the
	* Common Development and Distribution License (the "License").
	* You may not use this file except in compliance with the License.
	*
	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	* or http://www.opensolaris.org/os/licensing.
	* See the License for the specific language governing permissions
	* and limitations under the License.
	*
	* When distributing Covered Code, include this CDDL HEADER in each
	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	* If applicable, add the following below this CDDL HEADER, with the
	* fields enclosed by brackets "[]" replaced with your own identifying
	* information: Portions Copyright [yyyy] [name of copyright owner]
	*
	* CDDL HEADER END
	*/
	/*
	* Copyright (C) 2008-2010 Lawrence Livermore National Security, LLC.
	* Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
	* Rewritten for Linux by Brian Behlendorf <behlendorf1@llnl.gov>.
	* LLNL-CODE-403049.
	*
	* ZFS volume emulation driver.
	*
	* Makes a DMU object look like a volume of arbitrary size, up to 2^64 bytes.
	* Volumes are accessed through the symbolic links named:
	*
	* /dev/<pool_name>/<dataset_name>
	*
	* Volumes are persistent through reboot and module load. No user command
	* needs to be run before opening and using a device.
	*
	* Copyright 2014 Nexenta Systems, Inc. All rights reserved.
	* Copyright (c) 2016 Actifio, Inc. All rights reserved.
	* Copyright (c) 2012, 2019 by Delphix. All rights reserved.
	*/

	/*
	* Note on locking of zvol state structures.
	*
	* These structures are used to maintain internal state used to emulate block
	* devices on top of zvols. In particular, management of device minor number
	* operations - create, remove, rename, and set_snapdev - involves access to
	* these structures. The zvol_state_lock is primarily used to protect the
	* zvol_state_list. The zv->zv_state_lock is used to protect the contents
	* of the zvol_state_t structures, as well as to make sure that when the
	* time comes to remove the structure from the list, it is not in use, and
	* therefore, it can be taken off zvol_state_list and freed.
	*
	* The zv_suspend_lock was introduced to allow for suspending I/O to a zvol,
	* e.g. for the duration of receive and rollback operations. This lock can be
	* held for significant periods of time. Given that it is undesirable to hold
	* mutexes for long periods of time, the following lock ordering applies:
	* - take zvol_state_lock if necessary, to protect zvol_state_list
	* - take zv_suspend_lock if necessary, by the code path in question
	* - take zv_state_lock to protect zvol_state_t
	*
	* The minor operations are issued to spa->spa_zvol_taskq queues, that are
	* single-threaded (to preserve order of minor operations), and are executed
	* through the zvol_task_cb that dispatches the specific operations. Therefore,
	* these operations are serialized per pool. Consequently, we can be certain
	* that for a given zvol, there is only one operation at a time in progress.
	* That is why one can be sure that first, zvol_state_t for a given zvol is
	* allocated and placed on zvol_state_list, and then other minor operations
	* for this zvol are going to proceed in the order of issue.
	*
	*/

	#include <sys/dataset_kstats.h>
	#include <sys/dbuf.h>
	#include <sys/dmu_traverse.h>
	#include <sys/dsl_dataset.h>
	#include <sys/dsl_prop.h>
	#include <sys/dsl_dir.h>
	#include <sys/zap.h>
	#include <sys/zfeature.h>
	#include <sys/zil_impl.h>
	#include <sys/dmu_tx.h>
	#include <sys/zio.h>
	#include <sys/zfs_rlock.h>
	#include <sys/spa_impl.h>
	#include <sys/zvol.h>
	#include <sys/zvol_impl.h>

	unsigned int zvol_inhibit_dev = 0;
	unsigned int zvol_volmode = ZFS_VOLMODE_GEOM;

	struct hlist_head *zvol_htable;
	list_t zvol_state_list;
	krwlock_t zvol_state_lock;
	const zvol_platform_ops_t *ops;

	typedef enum {
	ZVOL_ASYNC_REMOVE_MINORS,
	ZVOL_ASYNC_RENAME_MINORS,
	ZVOL_ASYNC_SET_SNAPDEV,
	ZVOL_ASYNC_SET_VOLMODE,
	ZVOL_ASYNC_MAX
	} zvol_async_op_t;

	typedef struct {
	zvol_async_op_t op;
	char name1[MAXNAMELEN];
	char name2[MAXNAMELEN];
	uint64_t value;
	} zvol_task_t;

	uint64_t
	zvol_name_hash(const char *name)
	{
	int i;
	uint64_t crc = -1ULL;
	const uint8_t p = (const uint8_t )name;
	ASSERT(zfs_crc64_table[128] == ZFS_CRC64_POLY);
	for (i = 0; i < MAXNAMELEN - 1 && *p; i++, p++) {
	crc = (crc >> 8) ^ zfs_crc64_table[(crc ^ (*p)) & 0xFF];
	}
	return (crc);
	}

	/*
	* Find a zvol_state_t given the name and hash generated by zvol_name_hash.
	* If found, return with zv_suspend_lock and zv_state_lock taken, otherwise,
	* return (NULL) without the taking locks. The zv_suspend_lock is always taken
	* before zv_state_lock. The mode argument indicates the mode (including none)
	* for zv_suspend_lock to be taken.
	*/
	zvol_state_t *
	zvol_find_by_name_hash(const char *name, uint64_t hash, int mode)
	{
	zvol_state_t *zv;
	struct hlist_node *p = NULL;

	rw_enter(&zvol_state_lock, RW_READER);
	hlist_for_each(p, ZVOL_HT_HEAD(hash)) {
	zv = hlist_entry(p, zvol_state_t, zv_hlink);
	mutex_enter(&zv->zv_state_lock);
	if (zv->zv_hash == hash &&
	strncmp(zv->zv_name, name, MAXNAMELEN) == 0) {
	/*
	* this is the right zvol, take the locks in the
	* right order
	*/
	if (mode != RW_NONE &&
	!rw_tryenter(&zv->zv_suspend_lock, mode)) {
	mutex_exit(&zv->zv_state_lock);
	rw_enter(&zv->zv_suspend_lock, mode);
	mutex_enter(&zv->zv_state_lock);
	/*
	* zvol cannot be renamed as we continue
	* to hold zvol_state_lock
	*/
	ASSERT(zv->zv_hash == hash &&
	strncmp(zv->zv_name, name, MAXNAMELEN)
	== 0);
	}
	rw_exit(&zvol_state_lock);
	return (zv);
	}
	mutex_exit(&zv->zv_state_lock);
	}
	rw_exit(&zvol_state_lock);

	return (NULL);
	}

	/*
	* Find a zvol_state_t given the name.
	* If found, return with zv_suspend_lock and zv_state_lock taken, otherwise,
	* return (NULL) without the taking locks. The zv_suspend_lock is always taken
	* before zv_state_lock. The mode argument indicates the mode (including none)
	* for zv_suspend_lock to be taken.
	*/
	static zvol_state_t *
	zvol_find_by_name(const char *name, int mode)
	{
	return (zvol_find_by_name_hash(name, zvol_name_hash(name), mode));
	}

	/*
	* ZFS_IOC_CREATE callback handles dmu zvol and zap object creation.
	*/
	void
	zvol_create_cb(objset_t os, void arg, cred_t cr, dmu_tx_t tx)
	{
	zfs_creat_t *zct = arg;
	nvlist_t *nvprops = zct->zct_props;
	int error;
	uint64_t volblocksize, volsize;

	VERIFY(nvlist_lookup_uint64(nvprops,
	zfs_prop_to_name(ZFS_PROP_VOLSIZE), &volsize) == 0);
	if (nvlist_lookup_uint64(nvprops,
	zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), &volblocksize) != 0)
	volblocksize = zfs_prop_default_numeric(ZFS_PROP_VOLBLOCKSIZE);

	/*
	* These properties must be removed from the list so the generic
	* property setting step won't apply to them.
	*/
	VERIFY(nvlist_remove_all(nvprops,
	zfs_prop_to_name(ZFS_PROP_VOLSIZE)) == 0);
	(void) nvlist_remove_all(nvprops,
	zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE));

	error = dmu_object_claim(os, ZVOL_OBJ, DMU_OT_ZVOL, volblocksize,
	DMU_OT_NONE, 0, tx);
	ASSERT(error == 0);

	error = zap_create_claim(os, ZVOL_ZAP_OBJ, DMU_OT_ZVOL_PROP,
	DMU_OT_NONE, 0, tx);
	ASSERT(error == 0);

	error = zap_update(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize, tx);
	ASSERT(error == 0);
	}

	/*
	* ZFS_IOC_OBJSET_STATS entry point.
	*/
	int
	zvol_get_stats(objset_t os, nvlist_t nv)
	{
	int error;
	dmu_object_info_t *doi;
	uint64_t val;

	error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &val);
	if (error)
	return (SET_ERROR(error));

	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_VOLSIZE, val);
	doi = kmem_alloc(sizeof (dmu_object_info_t), KM_SLEEP);
	error = dmu_object_info(os, ZVOL_OBJ, doi);

	if (error == 0) {
	dsl_prop_nvlist_add_uint64(nv, ZFS_PROP_VOLBLOCKSIZE,
	doi->doi_data_block_size);
	}

	kmem_free(doi, sizeof (dmu_object_info_t));

	return (SET_ERROR(error));
	}

	/*
	* Sanity check volume size.
	*/
	int
	zvol_check_volsize(uint64_t volsize, uint64_t blocksize)
	{
	if (volsize == 0)
	return (SET_ERROR(EINVAL));

	if (volsize % blocksize != 0)
	return (SET_ERROR(EINVAL));

	#ifdef _ILP32
	if (volsize - 1 > SPEC_MAXOFFSET_T)
	return (SET_ERROR(EOVERFLOW));
	#endif
	return (0);
	}

	/*
	* Ensure the zap is flushed then inform the VFS of the capacity change.
	*/
	static int
	zvol_update_volsize(uint64_t volsize, objset_t *os)
	{
	dmu_tx_t *tx;
	int error;
	uint64_t txg;

	tx = dmu_tx_create(os);
	dmu_tx_hold_zap(tx, ZVOL_ZAP_OBJ, TRUE, NULL);
	dmu_tx_mark_netfree(tx);
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	dmu_tx_abort(tx);
	return (SET_ERROR(error));
	}
	txg = dmu_tx_get_txg(tx);

	error = zap_update(os, ZVOL_ZAP_OBJ, "size", 8, 1,
	&volsize, tx);
	dmu_tx_commit(tx);

	txg_wait_synced(dmu_objset_pool(os), txg);

	if (error == 0)
	error = dmu_free_long_range(os,
	ZVOL_OBJ, volsize, DMU_OBJECT_END);

	return (error);
	}

	/*
	* Set ZFS_PROP_VOLSIZE set entry point. Note that modifying the volume
	* size will result in a udev "change" event being generated.
	*/
	int
	zvol_set_volsize(const char *name, uint64_t volsize)
	{
	objset_t *os = NULL;
	uint64_t readonly;
	int error;
	boolean_t owned = B_FALSE;

	error = dsl_prop_get_integer(name,
	zfs_prop_to_name(ZFS_PROP_READONLY), &readonly, NULL);
	if (error != 0)
	return (SET_ERROR(error));
	if (readonly)
	return (SET_ERROR(EROFS));

	zvol_state_t *zv = zvol_find_by_name(name, RW_READER);

	ASSERT(zv == NULL \|\| (MUTEX_HELD(&zv->zv_state_lock) &&
	RW_READ_HELD(&zv->zv_suspend_lock)));

	if (zv == NULL \|\| zv->zv_objset == NULL) {
	if (zv != NULL)
	rw_exit(&zv->zv_suspend_lock);
	if ((error = dmu_objset_own(name, DMU_OST_ZVOL, B_FALSE, B_TRUE,
	FTAG, &os)) != 0) {
	if (zv != NULL)
	mutex_exit(&zv->zv_state_lock);
	return (SET_ERROR(error));
	}
	owned = B_TRUE;
	if (zv != NULL)
	zv->zv_objset = os;
	} else {
	os = zv->zv_objset;
	}

	dmu_object_info_t doi = kmem_alloc(sizeof (doi), KM_SLEEP);

	if ((error = dmu_object_info(os, ZVOL_OBJ, doi)) \|\|
	(error = zvol_check_volsize(volsize, doi->doi_data_block_size)))
	goto out;

	error = zvol_update_volsize(volsize, os);
	if (error == 0 && zv != NULL) {
	zv->zv_volsize = volsize;
	zv->zv_changed = 1;
	}
	out:
	kmem_free(doi, sizeof (dmu_object_info_t));

	if (owned) {
	dmu_objset_disown(os, B_TRUE, FTAG);
	if (zv != NULL)
	zv->zv_objset = NULL;
	} else {
	rw_exit(&zv->zv_suspend_lock);
	}

	if (zv != NULL)
	mutex_exit(&zv->zv_state_lock);

	if (error == 0 && zv != NULL)
	ops->zv_update_volsize(zv, volsize);

	return (SET_ERROR(error));
	}

	/*
	* Sanity check volume block size.
	*/
	int
	zvol_check_volblocksize(const char *name, uint64_t volblocksize)
	{
	/* Record sizes above 128k need the feature to be enabled */
	if (volblocksize > SPA_OLD_MAXBLOCKSIZE) {
	spa_t *spa;
	int error;

	if ((error = spa_open(name, &spa, FTAG)) != 0)
	return (error);

	if (!spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_BLOCKS)) {
	spa_close(spa, FTAG);
	return (SET_ERROR(ENOTSUP));
	}

	/*
	* We don't allow setting the property above 1MB,
	* unless the tunable has been changed.
	*/
	if (volblocksize > zfs_max_recordsize)
	return (SET_ERROR(EDOM));

	spa_close(spa, FTAG);
	}

	if (volblocksize < SPA_MINBLOCKSIZE \|\|
	volblocksize > SPA_MAXBLOCKSIZE \|\|
	!ISP2(volblocksize))
	return (SET_ERROR(EDOM));

	return (0);
	}

	/*
	* Replay a TX_TRUNCATE ZIL transaction if asked. TX_TRUNCATE is how we
	* implement DKIOCFREE/free-long-range.
	*/
	static int
	zvol_replay_truncate(void arg1, void arg2, boolean_t byteswap)
	{
	zvol_state_t *zv = arg1;
	lr_truncate_t *lr = arg2;
	uint64_t offset, length;

	if (byteswap)
	byteswap_uint64_array(lr, sizeof (*lr));

	offset = lr->lr_offset;
	length = lr->lr_length;

	dmu_tx_t *tx = dmu_tx_create(zv->zv_objset);
	dmu_tx_mark_netfree(tx);
	int error = dmu_tx_assign(tx, TXG_WAIT);
	if (error != 0) {
	dmu_tx_abort(tx);
	} else {
	zil_replaying(zv->zv_zilog, tx);
	dmu_tx_commit(tx);
	error = dmu_free_long_range(zv->zv_objset, ZVOL_OBJ, offset,
	length);
	}

	return (error);
	}

	/*
	* Replay a TX_WRITE ZIL transaction that didn't get committed
	* after a system failure
	*/
	static int
	zvol_replay_write(void arg1, void arg2, boolean_t byteswap)
	{
	zvol_state_t *zv = arg1;
	lr_write_t *lr = arg2;
	objset_t *os = zv->zv_objset;
	char data = (char )(lr + 1); /* data follows lr_write_t */
	uint64_t offset, length;
	dmu_tx_t *tx;
	int error;

	if (byteswap)
	byteswap_uint64_array(lr, sizeof (*lr));

	offset = lr->lr_offset;
	length = lr->lr_length;

	/* If it's a dmu_sync() block, write the whole block */
	if (lr->lr_common.lrc_reclen == sizeof (lr_write_t)) {
	uint64_t blocksize = BP_GET_LSIZE(&lr->lr_blkptr);
	if (length < blocksize) {
	offset -= offset % blocksize;
	length = blocksize;
	}
	}

	tx = dmu_tx_create(os);
	dmu_tx_hold_write(tx, ZVOL_OBJ, offset, length);
	error = dmu_tx_assign(tx, TXG_WAIT);
	if (error) {
	dmu_tx_abort(tx);
	} else {
	dmu_write(os, ZVOL_OBJ, offset, length, data, tx);
	zil_replaying(zv->zv_zilog, tx);
	dmu_tx_commit(tx);
	}

	return (error);
	}

	static int
	zvol_replay_err(void arg1, void arg2, boolean_t byteswap)
	{
	return (SET_ERROR(ENOTSUP));
	}

	/*
	* Callback vectors for replaying records.
	* Only TX_WRITE and TX_TRUNCATE are needed for zvol.
	*/
	zil_replay_func_t *zvol_replay_vector[TX_MAX_TYPE] = {
	zvol_replay_err, /* no such transaction type */
	zvol_replay_err, /* TX_CREATE */
	zvol_replay_err, /* TX_MKDIR */
	zvol_replay_err, /* TX_MKXATTR */
	zvol_replay_err, /* TX_SYMLINK */
	zvol_replay_err, /* TX_REMOVE */
	zvol_replay_err, /* TX_RMDIR */
	zvol_replay_err, /* TX_LINK */
	zvol_replay_err, /* TX_RENAME */
	zvol_replay_write, /* TX_WRITE */
	zvol_replay_truncate, /* TX_TRUNCATE */
	zvol_replay_err, /* TX_SETATTR */
	zvol_replay_err, /* TX_ACL */
	zvol_replay_err, /* TX_CREATE_ATTR */
	zvol_replay_err, /* TX_CREATE_ACL_ATTR */
	zvol_replay_err, /* TX_MKDIR_ACL */
	zvol_replay_err, /* TX_MKDIR_ATTR */
	zvol_replay_err, /* TX_MKDIR_ACL_ATTR */
	zvol_replay_err, /* TX_WRITE2 */
	};

	/*
	* zvol_log_write() handles synchronous writes using TX_WRITE ZIL transactions.
	*
	* We store data in the log buffers if it's small enough.
	* Otherwise we will later flush the data out via dmu_sync().
	*/
	ssize_t zvol_immediate_write_sz = 32768;

	void
	zvol_log_write(zvol_state_t zv, dmu_tx_t tx, uint64_t offset,
	uint64_t size, int sync)
	{
	uint32_t blocksize = zv->zv_volblocksize;
	zilog_t *zilog = zv->zv_zilog;
	itx_wr_state_t write_state;
	uint64_t sz = size;

	if (zil_replaying(zilog, tx))
	return;

	if (zilog->zl_logbias == ZFS_LOGBIAS_THROUGHPUT)
	write_state = WR_INDIRECT;
	else if (!spa_has_slogs(zilog->zl_spa) &&
	size >= blocksize && blocksize > zvol_immediate_write_sz)
	write_state = WR_INDIRECT;
	else if (sync)
	write_state = WR_COPIED;
	else
	write_state = WR_NEED_COPY;

	while (size) {
	itx_t *itx;
	lr_write_t *lr;
	itx_wr_state_t wr_state = write_state;
	ssize_t len = size;

	if (wr_state == WR_COPIED && size > zil_max_copied_data(zilog))
	wr_state = WR_NEED_COPY;
	else if (wr_state == WR_INDIRECT)
	len = MIN(blocksize - P2PHASE(offset, blocksize), size);

	itx = zil_itx_create(TX_WRITE, sizeof (*lr) +
	(wr_state == WR_COPIED ? len : 0));
	lr = (lr_write_t *)&itx->itx_lr;
	if (wr_state == WR_COPIED && dmu_read_by_dnode(zv->zv_dn,
	offset, len, lr+1, DMU_READ_NO_PREFETCH) != 0) {
	zil_itx_destroy(itx);
	itx = zil_itx_create(TX_WRITE, sizeof (*lr));
	lr = (lr_write_t *)&itx->itx_lr;
	wr_state = WR_NEED_COPY;
	}

	itx->itx_wr_state = wr_state;
	lr->lr_foid = ZVOL_OBJ;
	lr->lr_offset = offset;
	lr->lr_length = len;
	lr->lr_blkoff = 0;
	BP_ZERO(&lr->lr_blkptr);

	itx->itx_private = zv;
	itx->itx_sync = sync;

	(void) zil_itx_assign(zilog, itx, tx);

	offset += len;
	size -= len;
	}

	if (write_state == WR_COPIED \|\| write_state == WR_NEED_COPY) {
	dsl_pool_wrlog_count(zilog->zl_dmu_pool, sz, tx->tx_txg);
	}
	}

	/*
	* Log a DKIOCFREE/free-long-range to the ZIL with TX_TRUNCATE.
	*/
	void
	zvol_log_truncate(zvol_state_t zv, dmu_tx_t tx, uint64_t off, uint64_t len,
	boolean_t sync)
	{
	itx_t *itx;
	lr_truncate_t *lr;
	zilog_t *zilog = zv->zv_zilog;

	if (zil_replaying(zilog, tx))
	return;

	itx = zil_itx_create(TX_TRUNCATE, sizeof (*lr));
	lr = (lr_truncate_t *)&itx->itx_lr;
	lr->lr_foid = ZVOL_OBJ;
	lr->lr_offset = off;
	lr->lr_length = len;

	itx->itx_sync = sync;
	zil_itx_assign(zilog, itx, tx);
	}


	/* ARGSUSED */
	static void
	zvol_get_done(zgd_t *zgd, int error)
	{
	if (zgd->zgd_db)
	dmu_buf_rele(zgd->zgd_db, zgd);

	zfs_rangelock_exit(zgd->zgd_lr);

	kmem_free(zgd, sizeof (zgd_t));
	}

	/*
	* Get data to generate a TX_WRITE intent log record.
	*/
	int
	zvol_get_data(void arg, uint64_t arg2, lr_write_t lr, char *buf,
	struct lwb lwb, zio_t zio)
	{
	zvol_state_t *zv = arg;
	uint64_t offset = lr->lr_offset;
	uint64_t size = lr->lr_length;
	dmu_buf_t *db;
	zgd_t *zgd;
	int error;

	ASSERT3P(lwb, !=, NULL);
	ASSERT3P(zio, !=, NULL);
	ASSERT3U(size, !=, 0);

	zgd = (zgd_t *)kmem_zalloc(sizeof (zgd_t), KM_SLEEP);
	zgd->zgd_lwb = lwb;

	/*
	* Write records come in two flavors: immediate and indirect.
	* For small writes it's cheaper to store the data with the
	* log record (immediate); for large writes it's cheaper to
	* sync the data and get a pointer to it (indirect) so that
	* we don't have to write the data twice.
	*/
	if (buf != NULL) { /* immediate write */
	zgd->zgd_lr = zfs_rangelock_enter(&zv->zv_rangelock, offset,
	size, RL_READER);
	error = dmu_read_by_dnode(zv->zv_dn, offset, size, buf,
	DMU_READ_NO_PREFETCH);
	} else { /* indirect write */
	/*
	* Have to lock the whole block to ensure when it's written out
	* and its checksum is being calculated that no one can change
	* the data. Contrarily to zfs_get_data we need not re-check
	* blocksize after we get the lock because it cannot be changed.
	*/
	size = zv->zv_volblocksize;
	offset = P2ALIGN_TYPED(offset, size, uint64_t);
	zgd->zgd_lr = zfs_rangelock_enter(&zv->zv_rangelock, offset,
	size, RL_READER);
	error = dmu_buf_hold_by_dnode(zv->zv_dn, offset, zgd, &db,
	DMU_READ_NO_PREFETCH);
	if (error == 0) {
	blkptr_t *bp = &lr->lr_blkptr;

	zgd->zgd_db = db;
	zgd->zgd_bp = bp;

	ASSERT(db != NULL);
	ASSERT(db->db_offset == offset);
	ASSERT(db->db_size == size);

	error = dmu_sync(zio, lr->lr_common.lrc_txg,
	zvol_get_done, zgd);

	if (error == 0)
	return (0);
	}
	}

	zvol_get_done(zgd, error);

	return (SET_ERROR(error));
	}

	/*
	* The zvol_state_t's are inserted into zvol_state_list and zvol_htable.
	*/

	void
	zvol_insert(zvol_state_t *zv)
	{
	ASSERT(RW_WRITE_HELD(&zvol_state_lock));
	list_insert_head(&zvol_state_list, zv);
	hlist_add_head(&zv->zv_hlink, ZVOL_HT_HEAD(zv->zv_hash));
	}

	/*
	* Simply remove the zvol from to list of zvols.
	*/
	static void
	zvol_remove(zvol_state_t *zv)
	{
	ASSERT(RW_WRITE_HELD(&zvol_state_lock));
	list_remove(&zvol_state_list, zv);
	hlist_del(&zv->zv_hlink);
	}

	/*
	* Setup zv after we just own the zv->objset
	*/
	static int
	zvol_setup_zv(zvol_state_t *zv)
	{
	uint64_t volsize;
	int error;
	uint64_t ro;
	objset_t *os = zv->zv_objset;

	ASSERT(MUTEX_HELD(&zv->zv_state_lock));
	ASSERT(RW_LOCK_HELD(&zv->zv_suspend_lock));

	zv->zv_zilog = NULL;
	zv->zv_flags &= ~ZVOL_WRITTEN_TO;

	error = dsl_prop_get_integer(zv->zv_name, "readonly", &ro, NULL);
	if (error)
	return (SET_ERROR(error));

	error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize);
	if (error)
	return (SET_ERROR(error));

	error = dnode_hold(os, ZVOL_OBJ, zv, &zv->zv_dn);
	if (error)
	return (SET_ERROR(error));

	ops->zv_set_capacity(zv, volsize >> 9);
	zv->zv_volsize = volsize;

	if (ro \|\| dmu_objset_is_snapshot(os) \|\|
	!spa_writeable(dmu_objset_spa(os))) {
	ops->zv_set_disk_ro(zv, 1);
	zv->zv_flags \|= ZVOL_RDONLY;
	} else {
	ops->zv_set_disk_ro(zv, 0);
	zv->zv_flags &= ~ZVOL_RDONLY;
	}
	return (0);
	}

	/*
	* Shutdown every zv_objset related stuff except zv_objset itself.
	* The is the reverse of zvol_setup_zv.
	*/
	static void
	zvol_shutdown_zv(zvol_state_t *zv)
	{
	ASSERT(MUTEX_HELD(&zv->zv_state_lock) &&
	RW_LOCK_HELD(&zv->zv_suspend_lock));

	if (zv->zv_flags & ZVOL_WRITTEN_TO) {
	ASSERT(zv->zv_zilog != NULL);
	zil_close(zv->zv_zilog);
	}

	zv->zv_zilog = NULL;

	dnode_rele(zv->zv_dn, zv);
	zv->zv_dn = NULL;

	/*
	* Evict cached data. We must write out any dirty data before
	* disowning the dataset.
	*/
	if (zv->zv_flags & ZVOL_WRITTEN_TO)
	txg_wait_synced(dmu_objset_pool(zv->zv_objset), 0);
	(void) dmu_objset_evict_dbufs(zv->zv_objset);
	}

	/*
	* return the proper tag for rollback and recv
	*/
	void *
	zvol_tag(zvol_state_t *zv)
	{
	ASSERT(RW_WRITE_HELD(&zv->zv_suspend_lock));
	return (zv->zv_open_count > 0 ? zv : NULL);
	}

	/*
	* Suspend the zvol for recv and rollback.
	*/
	zvol_state_t *
	zvol_suspend(const char *name)
	{
	zvol_state_t *zv;

	zv = zvol_find_by_name(name, RW_WRITER);

	if (zv == NULL)
	return (NULL);

	/* block all I/O, release in zvol_resume. */
	ASSERT(MUTEX_HELD(&zv->zv_state_lock));
	ASSERT(RW_WRITE_HELD(&zv->zv_suspend_lock));

	atomic_inc(&zv->zv_suspend_ref);

	if (zv->zv_open_count > 0)
	zvol_shutdown_zv(zv);

	/*
	* do not hold zv_state_lock across suspend/resume to
	* avoid locking up zvol lookups
	*/
	mutex_exit(&zv->zv_state_lock);

	/* zv_suspend_lock is released in zvol_resume() */
	return (zv);
	}

	int
	zvol_resume(zvol_state_t *zv)
	{
	int error = 0;

	ASSERT(RW_WRITE_HELD(&zv->zv_suspend_lock));

	mutex_enter(&zv->zv_state_lock);

	if (zv->zv_open_count > 0) {
	VERIFY0(dmu_objset_hold(zv->zv_name, zv, &zv->zv_objset));
	VERIFY3P(zv->zv_objset->os_dsl_dataset->ds_owner, ==, zv);
	VERIFY(dsl_dataset_long_held(zv->zv_objset->os_dsl_dataset));
	dmu_objset_rele(zv->zv_objset, zv);

	error = zvol_setup_zv(zv);
	}

	mutex_exit(&zv->zv_state_lock);

	rw_exit(&zv->zv_suspend_lock);
	/*
	* We need this because we don't hold zvol_state_lock while releasing
	* zv_suspend_lock. zvol_remove_minors_impl thus cannot check
	* zv_suspend_lock to determine it is safe to free because rwlock is
	* not inherent atomic.
	*/
	atomic_dec(&zv->zv_suspend_ref);

	return (SET_ERROR(error));
	}

	int
	zvol_first_open(zvol_state_t *zv, boolean_t readonly)
	{
	objset_t *os;
	- int error, locked = 0;
	- boolean_t ro;
	+ int error;

	ASSERT(RW_READ_HELD(&zv->zv_suspend_lock));
	ASSERT(MUTEX_HELD(&zv->zv_state_lock));
	+ ASSERT(mutex_owned(&spa_namespace_lock));

	- /*
	- * In all other cases the spa_namespace_lock is taken before the
	- * bdev->bd_mutex lock. But in this case the Linux __blkdev_get()
	- * function calls fops->open() with the bdev->bd_mutex lock held.
	- * This deadlock can be easily observed with zvols used as vdevs.
	- *
	- * To avoid a potential lock inversion deadlock we preemptively
	- * try to take the spa_namespace_lock(). Normally it will not
	- * be contended and this is safe because spa_open_common() handles
	- * the case where the caller already holds the spa_namespace_lock.
	- *
	- * When it is contended we risk a lock inversion if we were to
	- * block waiting for the lock. Luckily, the __blkdev_get()
	- * function allows us to return -ERESTARTSYS which will result in
	- * bdev->bd_mutex being dropped, reacquired, and fops->open() being
	- * called again. This process can be repeated safely until both
	- * locks are acquired.
	- */
	- if (!mutex_owned(&spa_namespace_lock)) {
	- locked = mutex_tryenter(&spa_namespace_lock);
	- if (!locked)
	- return (SET_ERROR(EINTR));
	- }
	-
	- ro = (readonly \|\| (strchr(zv->zv_name, '@') != NULL));
	+ boolean_t ro = (readonly \|\| (strchr(zv->zv_name, '@') != NULL));
	error = dmu_objset_own(zv->zv_name, DMU_OST_ZVOL, ro, B_TRUE, zv, &os);
	if (error)
	- goto out_mutex;
	+ return (SET_ERROR(error));

	zv->zv_objset = os;

	error = zvol_setup_zv(zv);
	-
	if (error) {
	dmu_objset_disown(os, 1, zv);
	zv->zv_objset = NULL;
	}

	-out_mutex:
	- if (locked)
	- mutex_exit(&spa_namespace_lock);
	- return (SET_ERROR(error));
	+ return (error);
	}

	void
	zvol_last_close(zvol_state_t *zv)
	{
	ASSERT(RW_READ_HELD(&zv->zv_suspend_lock));
	ASSERT(MUTEX_HELD(&zv->zv_state_lock));

	zvol_shutdown_zv(zv);

	dmu_objset_disown(zv->zv_objset, 1, zv);
	zv->zv_objset = NULL;
	}

	typedef struct minors_job {
	list_t *list;
	list_node_t link;
	/* input */
	char *name;
	/* output */
	int error;
	} minors_job_t;

	/*
	* Prefetch zvol dnodes for the minors_job
	*/
	static void
	zvol_prefetch_minors_impl(void *arg)
	{
	minors_job_t *job = arg;
	char *dsname = job->name;
	objset_t *os = NULL;

	job->error = dmu_objset_own(dsname, DMU_OST_ZVOL, B_TRUE, B_TRUE,
	FTAG, &os);
	if (job->error == 0) {
	dmu_prefetch(os, ZVOL_OBJ, 0, 0, 0, ZIO_PRIORITY_SYNC_READ);
	dmu_objset_disown(os, B_TRUE, FTAG);
	}
	}

	/*
	* Mask errors to continue dmu_objset_find() traversal
	*/
	static int
	zvol_create_snap_minor_cb(const char dsname, void arg)
	{
	minors_job_t *j = arg;
	list_t *minors_list = j->list;
	const char *name = j->name;

	ASSERT0(MUTEX_HELD(&spa_namespace_lock));

	/* skip the designated dataset */
	if (name && strcmp(dsname, name) == 0)
	return (0);

	/* at this point, the dsname should name a snapshot */
	if (strchr(dsname, '@') == 0) {
	dprintf("zvol_create_snap_minor_cb(): "
	"%s is not a snapshot name\n", dsname);
	} else {
	minors_job_t *job;
	char *n = kmem_strdup(dsname);
	if (n == NULL)
	return (0);

	job = kmem_alloc(sizeof (minors_job_t), KM_SLEEP);
	job->name = n;
	job->list = minors_list;
	job->error = 0;
	list_insert_tail(minors_list, job);
	/* don't care if dispatch fails, because job->error is 0 */
	taskq_dispatch(system_taskq, zvol_prefetch_minors_impl, job,
	TQ_SLEEP);
	}

	return (0);
	}

	/*
	* If spa_keystore_load_wkey() is called for an encrypted zvol,
	* we need to look for any clones also using the key. This function
	* is "best effort" - so we just skip over it if there are failures.
	*/
	static void
	zvol_add_clones(const char dsname, list_t minors_list)
	{
	/* Also check if it has clones */
	dsl_dir_t *dd = NULL;
	dsl_pool_t *dp = NULL;

	if (dsl_pool_hold(dsname, FTAG, &dp) != 0)
	return;

	if (!spa_feature_is_enabled(dp->dp_spa,
	SPA_FEATURE_ENCRYPTION))
	goto out;

	if (dsl_dir_hold(dp, dsname, FTAG, &dd, NULL) != 0)
	goto out;

	if (dsl_dir_phys(dd)->dd_clones == 0)
	goto out;

	zap_cursor_t *zc = kmem_alloc(sizeof (zap_cursor_t), KM_SLEEP);
	zap_attribute_t *za = kmem_alloc(sizeof (zap_attribute_t), KM_SLEEP);
	objset_t *mos = dd->dd_pool->dp_meta_objset;

	for (zap_cursor_init(zc, mos, dsl_dir_phys(dd)->dd_clones);
	zap_cursor_retrieve(zc, za) == 0;
	zap_cursor_advance(zc)) {
	dsl_dataset_t *clone;
	minors_job_t *job;

	if (dsl_dataset_hold_obj(dd->dd_pool,
	za->za_first_integer, FTAG, &clone) == 0) {

	char name[ZFS_MAX_DATASET_NAME_LEN];
	dsl_dataset_name(clone, name);

	char *n = kmem_strdup(name);
	job = kmem_alloc(sizeof (minors_job_t), KM_SLEEP);
	job->name = n;
	job->list = minors_list;
	job->error = 0;
	list_insert_tail(minors_list, job);

	dsl_dataset_rele(clone, FTAG);
	}
	}
	zap_cursor_fini(zc);
	kmem_free(za, sizeof (zap_attribute_t));
	kmem_free(zc, sizeof (zap_cursor_t));

	out:
	if (dd != NULL)
	dsl_dir_rele(dd, FTAG);
	if (dp != NULL)
	dsl_pool_rele(dp, FTAG);
	}

	/*
	* Mask errors to continue dmu_objset_find() traversal
	*/
	static int
	zvol_create_minors_cb(const char dsname, void arg)
	{
	uint64_t snapdev;
	int error;
	list_t *minors_list = arg;

	ASSERT0(MUTEX_HELD(&spa_namespace_lock));

	error = dsl_prop_get_integer(dsname, "snapdev", &snapdev, NULL);
	if (error)
	return (0);

	/*
	* Given the name and the 'snapdev' property, create device minor nodes
	* with the linkages to zvols/snapshots as needed.
	* If the name represents a zvol, create a minor node for the zvol, then
	* check if its snapshots are 'visible', and if so, iterate over the
	* snapshots and create device minor nodes for those.
	*/
	if (strchr(dsname, '@') == 0) {
	minors_job_t *job;
	char *n = kmem_strdup(dsname);
	if (n == NULL)
	return (0);

	job = kmem_alloc(sizeof (minors_job_t), KM_SLEEP);
	job->name = n;
	job->list = minors_list;
	job->error = 0;
	list_insert_tail(minors_list, job);
	/* don't care if dispatch fails, because job->error is 0 */
	taskq_dispatch(system_taskq, zvol_prefetch_minors_impl, job,
	TQ_SLEEP);

	zvol_add_clones(dsname, minors_list);

	if (snapdev == ZFS_SNAPDEV_VISIBLE) {
	/*
	* traverse snapshots only, do not traverse children,
	* and skip the 'dsname'
	*/
	error = dmu_objset_find(dsname,
	zvol_create_snap_minor_cb, (void *)job,
	DS_FIND_SNAPSHOTS);
	}
	} else {
	dprintf("zvol_create_minors_cb(): %s is not a zvol name\n",
	dsname);
	}

	return (0);
	}

	/*
	* Create minors for the specified dataset, including children and snapshots.
	* Pay attention to the 'snapdev' property and iterate over the snapshots
	* only if they are 'visible'. This approach allows one to assure that the
	* snapshot metadata is read from disk only if it is needed.
	*
	* The name can represent a dataset to be recursively scanned for zvols and
	* their snapshots, or a single zvol snapshot. If the name represents a
	* dataset, the scan is performed in two nested stages:
	* - scan the dataset for zvols, and
	* - for each zvol, create a minor node, then check if the zvol's snapshots
	* are 'visible', and only then iterate over the snapshots if needed
	*
	* If the name represents a snapshot, a check is performed if the snapshot is
	* 'visible' (which also verifies that the parent is a zvol), and if so,
	* a minor node for that snapshot is created.
	*/
	void
	zvol_create_minors_recursive(const char *name)
	{
	list_t minors_list;
	minors_job_t *job;

	if (zvol_inhibit_dev)
	return;

	/*
	* This is the list for prefetch jobs. Whenever we found a match
	* during dmu_objset_find, we insert a minors_job to the list and do
	* taskq_dispatch to parallel prefetch zvol dnodes. Note we don't need
	* any lock because all list operation is done on the current thread.
	*
	* We will use this list to do zvol_create_minor_impl after prefetch
	* so we don't have to traverse using dmu_objset_find again.
	*/
	list_create(&minors_list, sizeof (minors_job_t),
	offsetof(minors_job_t, link));


	if (strchr(name, '@') != NULL) {
	uint64_t snapdev;

	int error = dsl_prop_get_integer(name, "snapdev",
	&snapdev, NULL);

	if (error == 0 && snapdev == ZFS_SNAPDEV_VISIBLE)
	(void) ops->zv_create_minor(name);
	} else {
	fstrans_cookie_t cookie = spl_fstrans_mark();
	(void) dmu_objset_find(name, zvol_create_minors_cb,
	&minors_list, DS_FIND_CHILDREN);
	spl_fstrans_unmark(cookie);
	}

	taskq_wait_outstanding(system_taskq, 0);

	/*
	* Prefetch is completed, we can do zvol_create_minor_impl
	* sequentially.
	*/
	while ((job = list_head(&minors_list)) != NULL) {
	list_remove(&minors_list, job);
	if (!job->error)
	(void) ops->zv_create_minor(job->name);
	kmem_strfree(job->name);
	kmem_free(job, sizeof (minors_job_t));
	}

	list_destroy(&minors_list);
	}

	void
	zvol_create_minor(const char *name)
	{
	/*
	* Note: the dsl_pool_config_lock must not be held.
	* Minor node creation needs to obtain the zvol_state_lock.
	* zvol_open() obtains the zvol_state_lock and then the dsl pool
	* config lock. Therefore, we can't have the config lock now if
	* we are going to wait for the zvol_state_lock, because it
	* would be a lock order inversion which could lead to deadlock.
	*/

	if (zvol_inhibit_dev)
	return;

	if (strchr(name, '@') != NULL) {
	uint64_t snapdev;

	int error = dsl_prop_get_integer(name,
	"snapdev", &snapdev, NULL);

	if (error == 0 && snapdev == ZFS_SNAPDEV_VISIBLE)
	(void) ops->zv_create_minor(name);
	} else {
	(void) ops->zv_create_minor(name);
	}
	}

	/*
	* Remove minors for specified dataset including children and snapshots.
	*/

	static void
	zvol_free_task(void *arg)
	{
	ops->zv_free(arg);
	}

	void
	zvol_remove_minors_impl(const char *name)
	{
	zvol_state_t zv, zv_next;
	int namelen = ((name) ? strlen(name) : 0);
	taskqid_t t;
	list_t free_list;

	if (zvol_inhibit_dev)
	return;

	list_create(&free_list, sizeof (zvol_state_t),
	offsetof(zvol_state_t, zv_next));

	rw_enter(&zvol_state_lock, RW_WRITER);

	for (zv = list_head(&zvol_state_list); zv != NULL; zv = zv_next) {
	zv_next = list_next(&zvol_state_list, zv);

	mutex_enter(&zv->zv_state_lock);
	if (name == NULL \|\| strcmp(zv->zv_name, name) == 0 \|\|
	(strncmp(zv->zv_name, name, namelen) == 0 &&
	(zv->zv_name[namelen] == '/' \|\|
	zv->zv_name[namelen] == '@'))) {
	/*
	* By holding zv_state_lock here, we guarantee that no
	* one is currently using this zv
	*/

	/* If in use, leave alone */
	if (zv->zv_open_count > 0 \|\|
	atomic_read(&zv->zv_suspend_ref)) {
	mutex_exit(&zv->zv_state_lock);
	continue;
	}

	zvol_remove(zv);

	/*
	* Cleared while holding zvol_state_lock as a writer
	* which will prevent zvol_open() from opening it.
	*/
	ops->zv_clear_private(zv);

	/* Drop zv_state_lock before zvol_free() */
	mutex_exit(&zv->zv_state_lock);

	/* Try parallel zv_free, if failed do it in place */
	t = taskq_dispatch(system_taskq, zvol_free_task, zv,
	TQ_SLEEP);
	if (t == TASKQID_INVALID)
	list_insert_head(&free_list, zv);
	} else {
	mutex_exit(&zv->zv_state_lock);
	}
	}
	rw_exit(&zvol_state_lock);

	/* Drop zvol_state_lock before calling zvol_free() */
	while ((zv = list_head(&free_list)) != NULL) {
	list_remove(&free_list, zv);
	ops->zv_free(zv);
	}
	}

	/* Remove minor for this specific volume only */
	static void
	zvol_remove_minor_impl(const char *name)
	{
	zvol_state_t zv = NULL, zv_next;

	if (zvol_inhibit_dev)
	return;

	rw_enter(&zvol_state_lock, RW_WRITER);

	for (zv = list_head(&zvol_state_list); zv != NULL; zv = zv_next) {
	zv_next = list_next(&zvol_state_list, zv);

	mutex_enter(&zv->zv_state_lock);
	if (strcmp(zv->zv_name, name) == 0) {
	/*
	* By holding zv_state_lock here, we guarantee that no
	* one is currently using this zv
	*/

	/* If in use, leave alone */
	if (zv->zv_open_count > 0 \|\|
	atomic_read(&zv->zv_suspend_ref)) {
	mutex_exit(&zv->zv_state_lock);
	continue;
	}
	zvol_remove(zv);

	ops->zv_clear_private(zv);
	mutex_exit(&zv->zv_state_lock);
	break;
	} else {
	mutex_exit(&zv->zv_state_lock);
	}
	}

	/* Drop zvol_state_lock before calling zvol_free() */
	rw_exit(&zvol_state_lock);

	if (zv != NULL)
	ops->zv_free(zv);
	}

	/*
	* Rename minors for specified dataset including children and snapshots.
	*/
	static void
	zvol_rename_minors_impl(const char oldname, const char newname)
	{
	zvol_state_t zv, zv_next;
	int oldnamelen, newnamelen;

	if (zvol_inhibit_dev)
	return;

	oldnamelen = strlen(oldname);
	newnamelen = strlen(newname);

	rw_enter(&zvol_state_lock, RW_READER);

	for (zv = list_head(&zvol_state_list); zv != NULL; zv = zv_next) {
	zv_next = list_next(&zvol_state_list, zv);

	mutex_enter(&zv->zv_state_lock);

	if (strcmp(zv->zv_name, oldname) == 0) {
	ops->zv_rename_minor(zv, newname);
	} else if (strncmp(zv->zv_name, oldname, oldnamelen) == 0 &&
	(zv->zv_name[oldnamelen] == '/' \|\|
	zv->zv_name[oldnamelen] == '@')) {
	char *name = kmem_asprintf("%s%c%s", newname,
	zv->zv_name[oldnamelen],
	zv->zv_name + oldnamelen + 1);
	ops->zv_rename_minor(zv, name);
	kmem_strfree(name);
	}

	mutex_exit(&zv->zv_state_lock);
	}

	rw_exit(&zvol_state_lock);
	}

	typedef struct zvol_snapdev_cb_arg {
	uint64_t snapdev;
	} zvol_snapdev_cb_arg_t;

	static int
	zvol_set_snapdev_cb(const char dsname, void param)
	{
	zvol_snapdev_cb_arg_t *arg = param;

	if (strchr(dsname, '@') == NULL)
	return (0);

	switch (arg->snapdev) {
	case ZFS_SNAPDEV_VISIBLE:
	(void) ops->zv_create_minor(dsname);
	break;
	case ZFS_SNAPDEV_HIDDEN:
	(void) zvol_remove_minor_impl(dsname);
	break;
	}

	return (0);
	}

	static void
	zvol_set_snapdev_impl(char *name, uint64_t snapdev)
	{
	zvol_snapdev_cb_arg_t arg = {snapdev};
	fstrans_cookie_t cookie = spl_fstrans_mark();
	/*
	* The zvol_set_snapdev_sync() sets snapdev appropriately
	* in the dataset hierarchy. Here, we only scan snapshots.
	*/
	dmu_objset_find(name, zvol_set_snapdev_cb, &arg, DS_FIND_SNAPSHOTS);
	spl_fstrans_unmark(cookie);
	}

	static void
	zvol_set_volmode_impl(char *name, uint64_t volmode)
	{
	fstrans_cookie_t cookie;
	uint64_t old_volmode;
	zvol_state_t *zv;

	if (strchr(name, '@') != NULL)
	return;

	/*
	* It's unfortunate we need to remove minors before we create new ones:
	* this is necessary because our backing gendisk (zvol_state->zv_disk)
	* could be different when we set, for instance, volmode from "geom"
	* to "dev" (or vice versa).
	*/
	zv = zvol_find_by_name(name, RW_NONE);
	if (zv == NULL && volmode == ZFS_VOLMODE_NONE)
	return;
	if (zv != NULL) {
	old_volmode = zv->zv_volmode;
	mutex_exit(&zv->zv_state_lock);
	if (old_volmode == volmode)
	return;
	zvol_wait_close(zv);
	}
	cookie = spl_fstrans_mark();
	switch (volmode) {
	case ZFS_VOLMODE_NONE:
	(void) zvol_remove_minor_impl(name);
	break;
	case ZFS_VOLMODE_GEOM:
	case ZFS_VOLMODE_DEV:
	(void) zvol_remove_minor_impl(name);
	(void) ops->zv_create_minor(name);
	break;
	case ZFS_VOLMODE_DEFAULT:
	(void) zvol_remove_minor_impl(name);
	if (zvol_volmode == ZFS_VOLMODE_NONE)
	break;
	else /* if zvol_volmode is invalid defaults to "geom" */
	(void) ops->zv_create_minor(name);
	break;
	}
	spl_fstrans_unmark(cookie);
	}

	static zvol_task_t *
	zvol_task_alloc(zvol_async_op_t op, const char name1, const char name2,
	uint64_t value)
	{
	zvol_task_t *task;

	/* Never allow tasks on hidden names. */
	if (name1[0] == '$')
	return (NULL);

	task = kmem_zalloc(sizeof (zvol_task_t), KM_SLEEP);
	task->op = op;
	task->value = value;

	strlcpy(task->name1, name1, MAXNAMELEN);
	if (name2 != NULL)
	strlcpy(task->name2, name2, MAXNAMELEN);

	return (task);
	}

	static void
	zvol_task_free(zvol_task_t *task)
	{
	kmem_free(task, sizeof (zvol_task_t));
	}

	/*
	* The worker thread function performed asynchronously.
	*/
	static void
	zvol_task_cb(void *arg)
	{
	zvol_task_t *task = arg;

	switch (task->op) {
	case ZVOL_ASYNC_REMOVE_MINORS:
	zvol_remove_minors_impl(task->name1);
	break;
	case ZVOL_ASYNC_RENAME_MINORS:
	zvol_rename_minors_impl(task->name1, task->name2);
	break;
	case ZVOL_ASYNC_SET_SNAPDEV:
	zvol_set_snapdev_impl(task->name1, task->value);
	break;
	case ZVOL_ASYNC_SET_VOLMODE:
	zvol_set_volmode_impl(task->name1, task->value);
	break;
	default:
	VERIFY(0);
	break;
	}

	zvol_task_free(task);
	}

	typedef struct zvol_set_prop_int_arg {
	const char *zsda_name;
	uint64_t zsda_value;
	zprop_source_t zsda_source;
	dmu_tx_t *zsda_tx;
	} zvol_set_prop_int_arg_t;

	/*
	* Sanity check the dataset for safe use by the sync task. No additional
	* conditions are imposed.
	*/
	static int
	zvol_set_snapdev_check(void arg, dmu_tx_t tx)
	{
	zvol_set_prop_int_arg_t *zsda = arg;
	dsl_pool_t *dp = dmu_tx_pool(tx);
	dsl_dir_t *dd;
	int error;

	error = dsl_dir_hold(dp, zsda->zsda_name, FTAG, &dd, NULL);
	if (error != 0)
	return (error);

	dsl_dir_rele(dd, FTAG);

	return (error);
	}

	/* ARGSUSED */
	static int
	zvol_set_snapdev_sync_cb(dsl_pool_t dp, dsl_dataset_t ds, void *arg)
	{
	char dsname[MAXNAMELEN];
	zvol_task_t *task;
	uint64_t snapdev;

	dsl_dataset_name(ds, dsname);
	if (dsl_prop_get_int_ds(ds, "snapdev", &snapdev) != 0)
	return (0);
	task = zvol_task_alloc(ZVOL_ASYNC_SET_SNAPDEV, dsname, NULL, snapdev);
	if (task == NULL)
	return (0);

	(void) taskq_dispatch(dp->dp_spa->spa_zvol_taskq, zvol_task_cb,
	task, TQ_SLEEP);
	return (0);
	}

	/*
	* Traverse all child datasets and apply snapdev appropriately.
	* We call dsl_prop_set_sync_impl() here to set the value only on the toplevel
	* dataset and read the effective "snapdev" on every child in the callback
	* function: this is because the value is not guaranteed to be the same in the
	* whole dataset hierarchy.
	*/
	static void
	zvol_set_snapdev_sync(void arg, dmu_tx_t tx)
	{
	zvol_set_prop_int_arg_t *zsda = arg;
	dsl_pool_t *dp = dmu_tx_pool(tx);
	dsl_dir_t *dd;
	dsl_dataset_t *ds;
	int error;

	VERIFY0(dsl_dir_hold(dp, zsda->zsda_name, FTAG, &dd, NULL));
	zsda->zsda_tx = tx;

	error = dsl_dataset_hold(dp, zsda->zsda_name, FTAG, &ds);
	if (error == 0) {
	dsl_prop_set_sync_impl(ds, zfs_prop_to_name(ZFS_PROP_SNAPDEV),
	zsda->zsda_source, sizeof (zsda->zsda_value), 1,
	&zsda->zsda_value, zsda->zsda_tx);
	dsl_dataset_rele(ds, FTAG);
	}
	dmu_objset_find_dp(dp, dd->dd_object, zvol_set_snapdev_sync_cb,
	zsda, DS_FIND_CHILDREN);

	dsl_dir_rele(dd, FTAG);
	}

	int
	zvol_set_snapdev(const char *ddname, zprop_source_t source, uint64_t snapdev)
	{
	zvol_set_prop_int_arg_t zsda;

	zsda.zsda_name = ddname;
	zsda.zsda_source = source;
	zsda.zsda_value = snapdev;

	return (dsl_sync_task(ddname, zvol_set_snapdev_check,
	zvol_set_snapdev_sync, &zsda, 0, ZFS_SPACE_CHECK_NONE));
	}

	/*
	* Sanity check the dataset for safe use by the sync task. No additional
	* conditions are imposed.
	*/
	static int
	zvol_set_volmode_check(void arg, dmu_tx_t tx)
	{
	zvol_set_prop_int_arg_t *zsda = arg;
	dsl_pool_t *dp = dmu_tx_pool(tx);
	dsl_dir_t *dd;
	int error;

	error = dsl_dir_hold(dp, zsda->zsda_name, FTAG, &dd, NULL);
	if (error != 0)
	return (error);

	dsl_dir_rele(dd, FTAG);

	return (error);
	}

	/* ARGSUSED */
	static int
	zvol_set_volmode_sync_cb(dsl_pool_t dp, dsl_dataset_t ds, void *arg)
	{
	char dsname[MAXNAMELEN];
	zvol_task_t *task;
	uint64_t volmode;

	dsl_dataset_name(ds, dsname);
	if (dsl_prop_get_int_ds(ds, "volmode", &volmode) != 0)
	return (0);
	task = zvol_task_alloc(ZVOL_ASYNC_SET_VOLMODE, dsname, NULL, volmode);
	if (task == NULL)
	return (0);

	(void) taskq_dispatch(dp->dp_spa->spa_zvol_taskq, zvol_task_cb,
	task, TQ_SLEEP);
	return (0);
	}

	/*
	* Traverse all child datasets and apply volmode appropriately.
	* We call dsl_prop_set_sync_impl() here to set the value only on the toplevel
	* dataset and read the effective "volmode" on every child in the callback
	* function: this is because the value is not guaranteed to be the same in the
	* whole dataset hierarchy.
	*/
	static void
	zvol_set_volmode_sync(void arg, dmu_tx_t tx)
	{
	zvol_set_prop_int_arg_t *zsda = arg;
	dsl_pool_t *dp = dmu_tx_pool(tx);
	dsl_dir_t *dd;
	dsl_dataset_t *ds;
	int error;

	VERIFY0(dsl_dir_hold(dp, zsda->zsda_name, FTAG, &dd, NULL));
	zsda->zsda_tx = tx;

	error = dsl_dataset_hold(dp, zsda->zsda_name, FTAG, &ds);
	if (error == 0) {
	dsl_prop_set_sync_impl(ds, zfs_prop_to_name(ZFS_PROP_VOLMODE),
	zsda->zsda_source, sizeof (zsda->zsda_value), 1,
	&zsda->zsda_value, zsda->zsda_tx);
	dsl_dataset_rele(ds, FTAG);
	}

	dmu_objset_find_dp(dp, dd->dd_object, zvol_set_volmode_sync_cb,
	zsda, DS_FIND_CHILDREN);

	dsl_dir_rele(dd, FTAG);
	}

	int
	zvol_set_volmode(const char *ddname, zprop_source_t source, uint64_t volmode)
	{
	zvol_set_prop_int_arg_t zsda;

	zsda.zsda_name = ddname;
	zsda.zsda_source = source;
	zsda.zsda_value = volmode;

	return (dsl_sync_task(ddname, zvol_set_volmode_check,
	zvol_set_volmode_sync, &zsda, 0, ZFS_SPACE_CHECK_NONE));
	}

	void
	zvol_remove_minors(spa_t spa, const char name, boolean_t async)
	{
	zvol_task_t *task;
	taskqid_t id;

	task = zvol_task_alloc(ZVOL_ASYNC_REMOVE_MINORS, name, NULL, ~0ULL);
	if (task == NULL)
	return;

	id = taskq_dispatch(spa->spa_zvol_taskq, zvol_task_cb, task, TQ_SLEEP);
	if ((async == B_FALSE) && (id != TASKQID_INVALID))
	taskq_wait_id(spa->spa_zvol_taskq, id);
	}

	void
	zvol_rename_minors(spa_t spa, const char name1, const char *name2,
	boolean_t async)
	{
	zvol_task_t *task;
	taskqid_t id;

	task = zvol_task_alloc(ZVOL_ASYNC_RENAME_MINORS, name1, name2, ~0ULL);
	if (task == NULL)
	return;

	id = taskq_dispatch(spa->spa_zvol_taskq, zvol_task_cb, task, TQ_SLEEP);
	if ((async == B_FALSE) && (id != TASKQID_INVALID))
	taskq_wait_id(spa->spa_zvol_taskq, id);
	}

	boolean_t
	zvol_is_zvol(const char *name)
	{

	return (ops->zv_is_zvol(name));
	}

	void
	zvol_register_ops(const zvol_platform_ops_t *zvol_ops)
	{
	ops = zvol_ops;
	}

	int
	zvol_init_impl(void)
	{
	int i;

	list_create(&zvol_state_list, sizeof (zvol_state_t),
	offsetof(zvol_state_t, zv_next));
	rw_init(&zvol_state_lock, NULL, RW_DEFAULT, NULL);

	zvol_htable = kmem_alloc(ZVOL_HT_SIZE * sizeof (struct hlist_head),
	KM_SLEEP);
	for (i = 0; i < ZVOL_HT_SIZE; i++)
	INIT_HLIST_HEAD(&zvol_htable[i]);

	return (0);
	}

	void
	zvol_fini_impl(void)
	{
	zvol_remove_minors_impl(NULL);

	/*
	* The call to "zvol_remove_minors_impl" may dispatch entries to
	* the system_taskq, but it doesn't wait for those entries to
	* complete before it returns. Thus, we must wait for all of the
	* removals to finish, before we can continue.
	*/
	taskq_wait_outstanding(system_taskq, 0);

	kmem_free(zvol_htable, ZVOL_HT_SIZE * sizeof (struct hlist_head));
	list_destroy(&zvol_state_list);
	rw_destroy(&zvol_state_lock);
	}
	diff --git a/sys/contrib/openzfs/module/zstd/include/zstd_compat_wrapper.h b/sys/contrib/openzfs/module/zstd/include/zstd_compat_wrapper.h
	index 71adc78040fb..339713590f96 100644
	--- a/sys/contrib/openzfs/module/zstd/include/zstd_compat_wrapper.h
	+++ b/sys/contrib/openzfs/module/zstd/include/zstd_compat_wrapper.h
	@@ -1,460 +1,461 @@
	/*
	* BSD 3-Clause New License (https://spdx.org/licenses/BSD-3-Clause.html)
	*
	* Redistribution and use in source and binary forms, with or without
	* modification, are permitted provided that the following conditions are met:
	*
	* 1. Redistributions of source code must retain the above copyright notice,
	* this list of conditions and the following disclaimer.
	*
	* 2. Redistributions in binary form must reproduce the above copyright notice,
	* this list of conditions and the following disclaimer in the documentation
	* and/or other materials provided with the distribution.
	*
	* 3. Neither the name of the copyright holder nor the names of its
	* contributors may be used to endorse or promote products derived from this
	* software without specific prior written permission.
	*
	* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
	* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
	* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
	* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
	* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
	* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
	* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
	* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
	* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
	* POSSIBILITY OF SUCH DAMAGE.
	*/

	/*
	* Copyright (c) 2020, Sebastian Gottschall
	*/

	/*
	* This wrapper fixes a problem, in case the ZFS filesystem driver, is compiled
	* statically into the kernel.
	* This will cause a symbol collision with the older in-kernel zstd library.
	* The following macros will simply rename all local zstd symbols and references
	*
	* Note: if the zstd library for zfs is updated to a newer version, this macro
	* list usually needs to be updated.
	* this can be done with some hand crafting of the output of the following
	* script
	* nm zstd.o \| awk '{print "#define "$3 " zfs_" $3}' > macrotable
	*/

	#define BIT_initDStream zfs_BIT_initDStream
	#define BIT_mask zfs_BIT_mask
	#define BIT_reloadDStream zfs_BIT_reloadDStream
	#define ERR_getErrorString zfs_ERR_getErrorString
	#define FSE_NCountWriteBound zfs_FSE_NCountWriteBound
	#define FSE_buildCTable zfs_FSE_buildCTable
	#define FSE_buildCTable_raw zfs_FSE_buildCTable_raw
	#define FSE_buildCTable_rle zfs_FSE_buildCTable_rle
	#define FSE_buildCTable_wksp zfs_FSE_buildCTable_wksp
	#define FSE_buildDTable zfs_FSE_buildDTable
	#define FSE_buildDTable_raw zfs_FSE_buildDTable_raw
	#define FSE_buildDTable_rle zfs_FSE_buildDTable_rle
	#define FSE_compress zfs_FSE_compress
	#define FSE_compress2 zfs_FSE_compress2
	#define FSE_compressBound zfs_FSE_compressBound
	#define FSE_compress_usingCTable zfs_FSE_compress_usingCTable
	#define FSE_compress_usingCTable_generic zfs_FSE_compress_usingCTable_generic
	#define FSE_compress_wksp zfs_FSE_compress_wksp
	#define FSE_createCTable zfs_FSE_createCTable
	#define FSE_createDTable zfs_FSE_createDTable
	#define FSE_decompress zfs_FSE_decompress
	#define FSE_decompress_usingDTable zfs_FSE_decompress_usingDTable
	#define FSE_decompress_wksp zfs_FSE_decompress_wksp
	#define FSE_freeCTable zfs_FSE_freeCTable
	#define FSE_freeDTable zfs_FSE_freeDTable
	#define FSE_getErrorName zfs_FSE_getErrorName
	#define FSE_normalizeCount zfs_FSE_normalizeCount
	#define FSE_optimalTableLog zfs_FSE_optimalTableLog
	#define FSE_optimalTableLog_internal zfs_FSE_optimalTableLog_internal
	#define FSE_readNCount zfs_FSE_readNCount
	#define FSE_versionNumber zfs_FSE_versionNumber
	#define FSE_writeNCount zfs_FSE_writeNCount
	#define HIST_count zfs_HIST_count
	#define HIST_countFast zfs_HIST_countFast
	#define HIST_countFast_wksp zfs_HIST_countFast_wksp
	#define HIST_count_parallel_wksp zfs_HIST_count_parallel_wksp
	#define HIST_count_simple zfs_HIST_count_simple
	#define HIST_count_wksp zfs_HIST_count_wksp
	#define HUF_buildCTable zfs_HUF_buildCTable
	#define HUF_buildCTable_wksp zfs_HUF_buildCTable_wksp
	#define HUF_compress zfs_HUF_compress
	#define HUF_compress1X zfs_HUF_compress1X
	#define HUF_compress1X_repeat zfs_HUF_compress1X_repeat
	#define HUF_compress1X_usingCTable zfs_HUF_compress1X_usingCTable
	#define HUF_compress1X_wksp zfs_HUF_compress1X_wksp
	#define HUF_compress2 zfs_HUF_compress2
	#define HUF_compress4X_repeat zfs_HUF_compress4X_repeat
	#define HUF_compress4X_usingCTable zfs_HUF_compress4X_usingCTable
	#define HUF_compress4X_wksp zfs_HUF_compress4X_wksp
	#define HUF_compressBound zfs_HUF_compressBound
	#define HUF_compressWeights zfs_HUF_compressWeights
	#define HUF_decompress zfs_HUF_decompress
	#define HUF_decompress1X1 zfs_HUF_decompress1X1
	#define HUF_decompress1X1_DCtx zfs_HUF_decompress1X1_DCtx
	#define HUF_decompress1X1_DCtx_wksp zfs_HUF_decompress1X1_DCtx_wksp
	#define HUF_decompress1X1_DCtx_wksp_bmi2 zfs_HUF_decompress1X1_DCtx_wksp_bmi2
	#define HUF_decompress1X1_usingDTable zfs_HUF_decompress1X1_usingDTable
	#define HUF_decompress1X2 zfs_HUF_decompress1X2
	#define HUF_decompress1X2_DCtx zfs_HUF_decompress1X2_DCtx
	#define HUF_decompress1X2_DCtx_wksp zfs_HUF_decompress1X2_DCtx_wksp
	#define HUF_decompress1X2_usingDTable zfs_HUF_decompress1X2_usingDTable
	#define HUF_decompress1X_DCtx zfs_HUF_decompress1X_DCtx
	#define HUF_decompress1X_DCtx_wksp zfs_HUF_decompress1X_DCtx_wksp
	#define HUF_decompress1X_usingDTable zfs_HUF_decompress1X_usingDTable
	#define HUF_decompress1X_usingDTable_bmi2 zfs_HUF_decompress1X_usingDTable_bmi2
	#define HUF_decompress4X1 zfs_HUF_decompress4X1
	#define HUF_decompress4X1_DCtx zfs_HUF_decompress4X1_DCtx
	#define HUF_decompress4X1_DCtx_wksp zfs_HUF_decompress4X1_DCtx_wksp
	#define HUF_decompress4X1_usingDTable zfs_HUF_decompress4X1_usingDTable
	#define HUF_decompress4X2 zfs_HUF_decompress4X2
	#define HUF_decompress4X2_DCtx zfs_HUF_decompress4X2_DCtx
	#define HUF_decompress4X2_DCtx_wksp zfs_HUF_decompress4X2_DCtx_wksp
	#define HUF_decompress4X2_usingDTable zfs_HUF_decompress4X2_usingDTable
	#define HUF_decompress4X_DCtx zfs_HUF_decompress4X_DCtx
	#define HUF_decompress4X_hufOnly zfs_HUF_decompress4X_hufOnly
	#define HUF_decompress4X_hufOnly_wksp zfs_HUF_decompress4X_hufOnly_wksp
	#define HUF_decompress4X_hufOnly_wksp_bmi2 \
	zfs_HUF_decompress4X_hufOnly_wksp_bmi2
	#define HUF_decompress4X_usingDTable zfs_HUF_decompress4X_usingDTable
	#define HUF_decompress4X_usingDTable_bmi2 zfs_HUF_decompress4X_usingDTable_bmi2
	#define HUF_estimateCompressedSize zfs_HUF_estimateCompressedSize
	#define HUF_fillDTableX2Level2 zfs_HUF_fillDTableX2Level2
	#define HUF_getErrorName zfs_HUF_getErrorName
	#define HUF_getNbBits zfs_HUF_getNbBits
	#define HUF_optimalTableLog zfs_HUF_optimalTableLog
	#define HUF_readCTable zfs_HUF_readCTable
	#define HUF_readDTableX1 zfs_HUF_readDTableX1
	#define HUF_readDTableX1_wksp zfs_HUF_readDTableX1_wksp
	#define HUF_readDTableX2 zfs_HUF_readDTableX2
	#define HUF_readDTableX2_wksp zfs_HUF_readDTableX2_wksp
	#define HUF_readStats zfs_HUF_readStats
	#define HUF_selectDecoder zfs_HUF_selectDecoder
	#define HUF_setMaxHeight zfs_HUF_setMaxHeight
	#define HUF_validateCTable zfs_HUF_validateCTable
	#define HUF_writeCTable zfs_HUF_writeCTable
	#define LL_base zfs_LL_base
	#define LL_bits zfs_LL_bits
	#define LL_defaultDTable zfs_LL_defaultDTable
	#define LL_defaultNorm zfs_LL_defaultNorm
	#define ML_base zfs_ML_base
	#define ML_bits zfs_ML_bits
	#define ML_defaultDTable zfs_ML_defaultDTable
	#define ML_defaultNorm zfs_ML_defaultNorm
	#define OF_base zfs_OF_base
	#define OF_bits zfs_OF_bits
	#define OF_defaultDTable zfs_OF_defaultDTable
	#define OF_defaultNorm zfs_OF_defaultNorm
	#define POOL_add zfs_POOL_add
	#define POOL_create zfs_POOL_create
	#define POOL_create_advanced zfs_POOL_create_advanced
	#define POOL_free zfs_POOL_free
	#define POOL_resize zfs_POOL_resize
	#define POOL_sizeof zfs_POOL_sizeof
	#define POOL_tryAdd zfs_POOL_tryAdd
	#define ZSTD_CCtxParams_getParameter zfs_ZSTD_CCtxParams_getParameter
	#define ZSTD_CCtxParams_init zfs_ZSTD_CCtxParams_init
	#define ZSTD_CCtxParams_init_advanced zfs_ZSTD_CCtxParams_init_advanced
	#define ZSTD_CCtxParams_reset zfs_ZSTD_CCtxParams_reset
	#define ZSTD_CCtxParams_setParameter zfs_ZSTD_CCtxParams_setParameter
	#define ZSTD_CCtx_getParameter zfs_ZSTD_CCtx_getParameter
	#define ZSTD_CCtx_loadDictionary zfs_ZSTD_CCtx_loadDictionary
	#define ZSTD_CCtx_loadDictionary_advanced zfs_ZSTD_CCtx_loadDictionary_advanced
	#define ZSTD_CCtx_loadDictionary_byReference \
	zfs_ZSTD_CCtx_loadDictionary_byReference
	#define ZSTD_CCtx_refCDict zfs_ZSTD_CCtx_refCDict
	#define ZSTD_CCtx_refPrefix zfs_ZSTD_CCtx_refPrefix
	#define ZSTD_CCtx_refPrefix_advanced zfs_ZSTD_CCtx_refPrefix_advanced
	#define ZSTD_CCtx_reset zfs_ZSTD_CCtx_reset
	#define ZSTD_CCtx_setParameter zfs_ZSTD_CCtx_setParameter
	#define ZSTD_CCtx_setParametersUsingCCtxParams \
	zfs_ZSTD_CCtx_setParametersUsingCCtxParams
	#define ZSTD_CCtx_setPledgedSrcSize zfs_ZSTD_CCtx_setPledgedSrcSize
	#define ZSTD_CStreamInSize zfs_ZSTD_CStreamInSize
	#define ZSTD_CStreamOutSize zfs_ZSTD_CStreamOutSize
	#define ZSTD_DCtx_loadDictionary zfs_ZSTD_DCtx_loadDictionary
	#define ZSTD_DCtx_loadDictionary_advanced zfs_ZSTD_DCtx_loadDictionary_advanced
	#define ZSTD_DCtx_loadDictionary_byReference \
	zfs_ZSTD_DCtx_loadDictionary_byReference
	#define ZSTD_DCtx_refDDict zfs_ZSTD_DCtx_refDDict
	#define ZSTD_DCtx_refPrefix zfs_ZSTD_DCtx_refPrefix
	#define ZSTD_DCtx_refPrefix_advanced zfs_ZSTD_DCtx_refPrefix_advanced
	#define ZSTD_DCtx_reset zfs_ZSTD_DCtx_reset
	#define ZSTD_DCtx_setFormat zfs_ZSTD_DCtx_setFormat
	#define ZSTD_DCtx_setMaxWindowSize zfs_ZSTD_DCtx_setMaxWindowSize
	#define ZSTD_DCtx_setParameter zfs_ZSTD_DCtx_setParameter
	#define ZSTD_DDict_dictContent zfs_ZSTD_DDict_dictContent
	#define ZSTD_DDict_dictSize zfs_ZSTD_DDict_dictSize
	#define ZSTD_DStreamInSize zfs_ZSTD_DStreamInSize
	#define ZSTD_DStreamOutSize zfs_ZSTD_DStreamOutSize
	#define ZSTD_DUBT_findBestMatch zfs_ZSTD_DUBT_findBestMatch
	#define ZSTD_NCountCost zfs_ZSTD_NCountCost
	#define ZSTD_XXH64_digest zfs_ZSTD_XXH64_digest
	#define ZSTD_adjustCParams zfs_ZSTD_adjustCParams
	#define ZSTD_assignParamsToCCtxParams zfs_ZSTD_assignParamsToCCtxParams
	#define ZSTD_buildCTable zfs_ZSTD_buildCTable
	#define ZSTD_buildFSETable zfs_ZSTD_buildFSETable
	#define ZSTD_buildSeqStore zfs_ZSTD_buildSeqStore
	#define ZSTD_buildSeqTable zfs_ZSTD_buildSeqTable
	#define ZSTD_cParam_getBounds zfs_ZSTD_cParam_getBounds
	#define ZSTD_cParam_withinBounds zfs_ZSTD_cParam_withinBounds
	#define ZSTD_calloc zfs_ZSTD_calloc
	#define ZSTD_checkCParams zfs_ZSTD_checkCParams
	#define ZSTD_checkContinuity zfs_ZSTD_checkContinuity
	#define ZSTD_compress zfs_ZSTD_compress
	#define ZSTD_compress2 zfs_ZSTD_compress2
	#define ZSTD_compressBegin zfs_ZSTD_compressBegin
	#define ZSTD_compressBegin_advanced zfs_ZSTD_compressBegin_advanced
	#define ZSTD_compressBegin_advanced_internal \
	zfs_ZSTD_compressBegin_advanced_internal
	#define ZSTD_compressBegin_usingCDict zfs_ZSTD_compressBegin_usingCDict
	#define ZSTD_compressBegin_usingCDict_advanced \
	zfs_ZSTD_compressBegin_usingCDict_advanced
	#define ZSTD_compressBegin_usingDict zfs_ZSTD_compressBegin_usingDict
	#define ZSTD_compressBlock zfs_ZSTD_compressBlock
	#define ZSTD_compressBlock_btlazy2 zfs_ZSTD_compressBlock_btlazy2
	#define ZSTD_compressBlock_btlazy2_dictMatchState \
	zfs_ZSTD_compressBlock_btlazy2_dictMatchState
	#define ZSTD_compressBlock_btlazy2_extDict \
	zfs_ZSTD_compressBlock_btlazy2_extDict
	#define ZSTD_compressBlock_btopt zfs_ZSTD_compressBlock_btopt
	#define ZSTD_compressBlock_btopt_dictMatchState \
	zfs_ZSTD_compressBlock_btopt_dictMatchState
	#define ZSTD_compressBlock_btopt_extDict zfs_ZSTD_compressBlock_btopt_extDict
	#define ZSTD_compressBlock_btultra zfs_ZSTD_compressBlock_btultra
	#define ZSTD_compressBlock_btultra2 zfs_ZSTD_compressBlock_btultra2
	#define ZSTD_compressBlock_btultra_dictMatchState \
	zfs_ZSTD_compressBlock_btultra_dictMatchState
	#define ZSTD_compressBlock_btultra_extDict \
	zfs_ZSTD_compressBlock_btultra_extDict
	#define ZSTD_compressBlock_doubleFast zfs_ZSTD_compressBlock_doubleFast
	#define ZSTD_compressBlock_doubleFast_dictMatchState \
	zfs_ZSTD_compressBlock_doubleFast_dictMatchState
	#define ZSTD_compressBlock_doubleFast_extDict \
	zfs_ZSTD_compressBlock_doubleFast_extDict
	#define ZSTD_compressBlock_doubleFast_extDict_generic \
	zfs_ZSTD_compressBlock_doubleFast_extDict_generic
	#define ZSTD_compressBlock_fast zfs_ZSTD_compressBlock_fast
	#define ZSTD_compressBlock_fast_dictMatchState \
	zfs_ZSTD_compressBlock_fast_dictMatchState
	#define ZSTD_compressBlock_fast_extDict zfs_ZSTD_compressBlock_fast_extDict
	#define ZSTD_compressBlock_fast_extDict_generic \
	zfs_ZSTD_compressBlock_fast_extDict_generic
	#define ZSTD_compressBlock_greedy zfs_ZSTD_compressBlock_greedy
	#define ZSTD_compressBlock_greedy_dictMatchState \
	zfs_ZSTD_compressBlock_greedy_dictMatchState
	#define ZSTD_compressBlock_greedy_extDict zfs_ZSTD_compressBlock_greedy_extDict
	#define ZSTD_compressBlock_internal zfs_ZSTD_compressBlock_internal
	#define ZSTD_compressBlock_lazy zfs_ZSTD_compressBlock_lazy
	#define ZSTD_compressBlock_lazy2 zfs_ZSTD_compressBlock_lazy2
	#define ZSTD_compressBlock_lazy2_dictMatchState \
	zfs_ZSTD_compressBlock_lazy2_dictMatchState
	#define ZSTD_compressBlock_lazy2_extDict zfs_ZSTD_compressBlock_lazy2_extDict
	#define ZSTD_compressBlock_lazy_dictMatchState \
	zfs_ZSTD_compressBlock_lazy_dictMatchState
	#define ZSTD_compressBlock_lazy_extDict zfs_ZSTD_compressBlock_lazy_extDict
	#define ZSTD_compressBound zfs_ZSTD_compressBound
	#define ZSTD_compressCCtx zfs_ZSTD_compressCCtx
	#define ZSTD_compressContinue zfs_ZSTD_compressContinue
	#define ZSTD_compressContinue_internal zfs_ZSTD_compressContinue_internal
	#define ZSTD_compressEnd zfs_ZSTD_compressEnd
	#define ZSTD_compressLiterals zfs_ZSTD_compressLiterals
	#define ZSTD_compressRleLiteralsBlock zfs_ZSTD_compressRleLiteralsBlock
	#define ZSTD_compressStream zfs_ZSTD_compressStream
	#define ZSTD_compressStream2 zfs_ZSTD_compressStream2
	#define ZSTD_compressStream2_simpleArgs zfs_ZSTD_compressStream2_simpleArgs
	#define ZSTD_compressSuperBlock zfs_ZSTD_compressSuperBlock
	#define ZSTD_compress_advanced zfs_ZSTD_compress_advanced
	#define ZSTD_compress_advanced_internal zfs_ZSTD_compress_advanced_internal
	#define ZSTD_compress_internal zfs_ZSTD_compress_internal
	#define ZSTD_compress_usingCDict zfs_ZSTD_compress_usingCDict
	#define ZSTD_compress_usingCDict_advanced zfs_ZSTD_compress_usingCDict_advanced
	#define ZSTD_compress_usingDict zfs_ZSTD_compress_usingDict
	#define ZSTD_copyCCtx zfs_ZSTD_copyCCtx
	#define ZSTD_copyDCtx zfs_ZSTD_copyDCtx
	#define ZSTD_copyDDictParameters zfs_ZSTD_copyDDictParameters
	#define ZSTD_count zfs_ZSTD_count
	#define ZSTD_count_2segments zfs_ZSTD_count_2segments
	#define ZSTD_createCCtx zfs_ZSTD_createCCtx
	#define ZSTD_createCCtxParams zfs_ZSTD_createCCtxParams
	#define ZSTD_createCCtx_advanced zfs_ZSTD_createCCtx_advanced
	#define ZSTD_createCDict zfs_ZSTD_createCDict
	#define ZSTD_createCDict_advanced zfs_ZSTD_createCDict_advanced
	#define ZSTD_createCDict_byReference zfs_ZSTD_createCDict_byReference
	#define ZSTD_createCStream zfs_ZSTD_createCStream
	#define ZSTD_createCStream_advanced zfs_ZSTD_createCStream_advanced
	#define ZSTD_createDCtx zfs_ZSTD_createDCtx
	#define ZSTD_createDCtx_advanced zfs_ZSTD_createDCtx_advanced
	#define ZSTD_createDDict zfs_ZSTD_createDDict
	#define ZSTD_createDDict_advanced zfs_ZSTD_createDDict_advanced
	#define ZSTD_createDDict_byReference zfs_ZSTD_createDDict_byReference
	#define ZSTD_createDStream zfs_ZSTD_createDStream
	#define ZSTD_createDStream_advanced zfs_ZSTD_createDStream_advanced
	#define ZSTD_crossEntropyCost zfs_ZSTD_crossEntropyCost
	#define ZSTD_cycleLog zfs_ZSTD_cycleLog
	#define ZSTD_dParam_getBounds zfs_ZSTD_dParam_getBounds
	#define ZSTD_decodeLiteralsBlock zfs_ZSTD_decodeLiteralsBlock
	#define ZSTD_decodeSeqHeaders zfs_ZSTD_decodeSeqHeaders
	#define ZSTD_decodingBufferSize_min zfs_ZSTD_decodingBufferSize_min
	#define ZSTD_decompress zfs_ZSTD_decompress
	#define ZSTD_decompressBegin zfs_ZSTD_decompressBegin
	#define ZSTD_decompressBegin_usingDDict zfs_ZSTD_decompressBegin_usingDDict
	#define ZSTD_decompressBegin_usingDict zfs_ZSTD_decompressBegin_usingDict
	#define ZSTD_decompressBlock zfs_ZSTD_decompressBlock
	#define ZSTD_decompressBlock_internal zfs_ZSTD_decompressBlock_internal
	#define ZSTD_decompressBound zfs_ZSTD_decompressBound
	#define ZSTD_decompressContinue zfs_ZSTD_decompressContinue
	#define ZSTD_decompressContinueStream zfs_ZSTD_decompressContinueStream
	#define ZSTD_decompressDCtx zfs_ZSTD_decompressDCtx
	#define ZSTD_decompressMultiFrame zfs_ZSTD_decompressMultiFrame
	#define ZSTD_decompressStream zfs_ZSTD_decompressStream
	#define ZSTD_decompressStream_simpleArgs zfs_ZSTD_decompressStream_simpleArgs
	#define ZSTD_decompress_usingDDict zfs_ZSTD_decompress_usingDDict
	#define ZSTD_decompress_usingDict zfs_ZSTD_decompress_usingDict
	#define ZSTD_defaultCParameters zfs_ZSTD_defaultCParameters
	#define ZSTD_did_fieldSize zfs_ZSTD_did_fieldSize
	#define ZSTD_encodeSequences zfs_ZSTD_encodeSequences
	#define ZSTD_encodeSequences_default zfs_ZSTD_encodeSequences_default
	#define ZSTD_endStream zfs_ZSTD_endStream
	#define ZSTD_estimateCCtxSize zfs_ZSTD_estimateCCtxSize
	#define ZSTD_estimateCCtxSize_usingCCtxParams \
	zfs_ZSTD_estimateCCtxSize_usingCCtxParams
	#define ZSTD_estimateCCtxSize_usingCParams \
	zfs_ZSTD_estimateCCtxSize_usingCParams
	#define ZSTD_estimateCDictSize zfs_ZSTD_estimateCDictSize
	#define ZSTD_estimateCDictSize_advanced zfs_ZSTD_estimateCDictSize_advanced
	#define ZSTD_estimateCStreamSize zfs_ZSTD_estimateCStreamSize
	#define ZSTD_estimateCStreamSize_usingCCtxParams \
	zfs_ZSTD_estimateCStreamSize_usingCCtxParams
	#define ZSTD_estimateCStreamSize_usingCParams \
	zfs_ZSTD_estimateCStreamSize_usingCParams
	#define ZSTD_estimateDCtxSize zfs_ZSTD_estimateDCtxSize
	#define ZSTD_estimateDDictSize zfs_ZSTD_estimateDDictSize
	#define ZSTD_estimateDStreamSize zfs_ZSTD_estimateDStreamSize
	#define ZSTD_estimateDStreamSize_fromFrame \
	zfs_ZSTD_estimateDStreamSize_fromFrame
	#define ZSTD_fcs_fieldSize zfs_ZSTD_fcs_fieldSize
	#define ZSTD_fillDoubleHashTable zfs_ZSTD_fillDoubleHashTable
	#define ZSTD_fillHashTable zfs_ZSTD_fillHashTable
	#define ZSTD_findDecompressedSize zfs_ZSTD_findDecompressedSize
	#define ZSTD_findFrameCompressedSize zfs_ZSTD_findFrameCompressedSize
	#define ZSTD_findFrameSizeInfo zfs_ZSTD_findFrameSizeInfo
	#define ZSTD_flushStream zfs_ZSTD_flushStream
	#define ZSTD_frameHeaderSize zfs_ZSTD_frameHeaderSize
	#define ZSTD_free zfs_ZSTD_free
	#define ZSTD_freeCCtx zfs_ZSTD_freeCCtx
	#define ZSTD_freeCCtxParams zfs_ZSTD_freeCCtxParams
	#define ZSTD_freeCDict zfs_ZSTD_freeCDict
	#define ZSTD_freeCStream zfs_ZSTD_freeCStream
	#define ZSTD_freeDCtx zfs_ZSTD_freeDCtx
	#define ZSTD_freeDDict zfs_ZSTD_freeDDict
	#define ZSTD_freeDStream zfs_ZSTD_freeDStream
	#define ZSTD_fseBitCost zfs_ZSTD_fseBitCost
	#define ZSTD_getBlockSize zfs_ZSTD_getBlockSize
	#define ZSTD_getCParams zfs_ZSTD_getCParams
	#define ZSTD_getCParamsFromCCtxParams zfs_ZSTD_getCParamsFromCCtxParams
	#define ZSTD_getCParamsFromCDict zfs_ZSTD_getCParamsFromCDict
	#define ZSTD_getCParams_internal zfs_ZSTD_getCParams_internal
	#define ZSTD_getDDict zfs_ZSTD_getDDict
	#define ZSTD_getDecompressedSize zfs_ZSTD_getDecompressedSize
	#define ZSTD_getDictID_fromDDict zfs_ZSTD_getDictID_fromDDict
	#define ZSTD_getDictID_fromDict zfs_ZSTD_getDictID_fromDict
	#define ZSTD_getDictID_fromFrame zfs_ZSTD_getDictID_fromFrame
	#define ZSTD_getErrorCode zfs_ZSTD_getErrorCode
	#define ZSTD_getErrorName zfs_ZSTD_getErrorName
	#define ZSTD_getErrorString zfs_ZSTD_getErrorString
	#define ZSTD_getFrameContentSize zfs_ZSTD_getFrameContentSize
	#define ZSTD_getFrameHeader zfs_ZSTD_getFrameHeader
	#define ZSTD_getFrameHeader_advanced zfs_ZSTD_getFrameHeader_advanced
	#define ZSTD_getFrameProgression zfs_ZSTD_getFrameProgression
	#define ZSTD_getParams zfs_ZSTD_getParams
	#define ZSTD_getSeqStore zfs_ZSTD_getSeqStore
	#define ZSTD_getSequences zfs_ZSTD_getSequences
	#define ZSTD_getcBlockSize zfs_ZSTD_getcBlockSize
	#define ZSTD_hashPtr zfs_ZSTD_hashPtr
	#define ZSTD_initCDict_internal zfs_ZSTD_initCDict_internal
	#define ZSTD_initCStream zfs_ZSTD_initCStream
	#define ZSTD_initCStream_advanced zfs_ZSTD_initCStream_advanced
	#define ZSTD_initCStream_internal zfs_ZSTD_initCStream_internal
	#define ZSTD_initCStream_srcSize zfs_ZSTD_initCStream_srcSize
	#define ZSTD_initCStream_usingCDict zfs_ZSTD_initCStream_usingCDict
	#define ZSTD_initCStream_usingCDict_advanced \
	zfs_ZSTD_initCStream_usingCDict_advanced
	#define ZSTD_initCStream_usingDict zfs_ZSTD_initCStream_usingDict
	#define ZSTD_initDDict_internal zfs_ZSTD_initDDict_internal
	#define ZSTD_initDStream zfs_ZSTD_initDStream
	#define ZSTD_initDStream_usingDDict zfs_ZSTD_initDStream_usingDDict
	#define ZSTD_initDStream_usingDict zfs_ZSTD_initDStream_usingDict
	#define ZSTD_initFseState zfs_ZSTD_initFseState
	#define ZSTD_initStaticCCtx zfs_ZSTD_initStaticCCtx
	#define ZSTD_initStaticCDict zfs_ZSTD_initStaticCDict
	#define ZSTD_initStaticCStream zfs_ZSTD_initStaticCStream
	#define ZSTD_initStaticDCtx zfs_ZSTD_initStaticDCtx
	#define ZSTD_initStaticDDict zfs_ZSTD_initStaticDDict
	#define ZSTD_initStaticDStream zfs_ZSTD_initStaticDStream
	#define ZSTD_initStats_ultra zfs_ZSTD_initStats_ultra
	#define ZSTD_insertAndFindFirstIndex zfs_ZSTD_insertAndFindFirstIndex
	#define ZSTD_insertAndFindFirstIndexHash3 zfs_ZSTD_insertAndFindFirstIndexHash3
	#define ZSTD_insertAndFindFirstIndex_internal \
	zfs_ZSTD_insertAndFindFirstIndex_internal
	#define ZSTD_insertBlock zfs_ZSTD_insertBlock
	#define ZSTD_invalidateRepCodes zfs_ZSTD_invalidateRepCodes
	+#define ZSTD_isError zfs_ZSTD_isError
	#define ZSTD_isFrame zfs_ZSTD_isFrame
	#define ZSTD_ldm_adjustParameters zfs_ZSTD_ldm_adjustParameters
	#define ZSTD_ldm_blockCompress zfs_ZSTD_ldm_blockCompress
	#define ZSTD_ldm_fillHashTable zfs_ZSTD_ldm_fillHashTable
	#define ZSTD_ldm_generateSequences zfs_ZSTD_ldm_generateSequences
	#define ZSTD_ldm_getMaxNbSeq zfs_ZSTD_ldm_getMaxNbSeq
	#define ZSTD_ldm_getTableSize zfs_ZSTD_ldm_getTableSize
	#define ZSTD_ldm_skipSequences zfs_ZSTD_ldm_skipSequences
	#define ZSTD_loadCEntropy zfs_ZSTD_loadCEntropy
	#define ZSTD_loadDEntropy zfs_ZSTD_loadDEntropy
	#define ZSTD_loadDictionaryContent zfs_ZSTD_loadDictionaryContent
	#define ZSTD_makeCCtxParamsFromCParams zfs_ZSTD_makeCCtxParamsFromCParams
	#define ZSTD_malloc zfs_ZSTD_malloc
	#define ZSTD_maxCLevel zfs_ZSTD_maxCLevel
	#define ZSTD_minCLevel zfs_ZSTD_minCLevel
	#define ZSTD_nextInputType zfs_ZSTD_nextInputType
	#define ZSTD_nextSrcSizeToDecompress zfs_ZSTD_nextSrcSizeToDecompress
	#define ZSTD_noCompressLiterals zfs_ZSTD_noCompressLiterals
	#define ZSTD_referenceExternalSequences zfs_ZSTD_referenceExternalSequences
	#define ZSTD_rescaleFreqs zfs_ZSTD_rescaleFreqs
	#define ZSTD_resetCCtx_internal zfs_ZSTD_resetCCtx_internal
	#define ZSTD_resetCCtx_usingCDict zfs_ZSTD_resetCCtx_usingCDict
	#define ZSTD_resetCStream zfs_ZSTD_resetCStream
	#define ZSTD_resetDStream zfs_ZSTD_resetDStream
	#define ZSTD_resetSeqStore zfs_ZSTD_resetSeqStore
	#define ZSTD_reset_compressedBlockState zfs_ZSTD_reset_compressedBlockState
	#define ZSTD_safecopy zfs_ZSTD_safecopy
	#define ZSTD_selectBlockCompressor zfs_ZSTD_selectBlockCompressor
	#define ZSTD_selectEncodingType zfs_ZSTD_selectEncodingType
	#define ZSTD_seqToCodes zfs_ZSTD_seqToCodes
	#define ZSTD_sizeof_CCtx zfs_ZSTD_sizeof_CCtx
	#define ZSTD_sizeof_CDict zfs_ZSTD_sizeof_CDict
	#define ZSTD_sizeof_CStream zfs_ZSTD_sizeof_CStream
	#define ZSTD_sizeof_DCtx zfs_ZSTD_sizeof_DCtx
	#define ZSTD_sizeof_DDict zfs_ZSTD_sizeof_DDict
	#define ZSTD_sizeof_DStream zfs_ZSTD_sizeof_DStream
	#define ZSTD_toFlushNow zfs_ZSTD_toFlushNow
	#define ZSTD_updateRep zfs_ZSTD_updateRep
	#define ZSTD_updateStats zfs_ZSTD_updateStats
	#define ZSTD_updateTree zfs_ZSTD_updateTree
	#define ZSTD_versionNumber zfs_ZSTD_versionNumber
	#define ZSTD_versionString zfs_ZSTD_versionString
	#define ZSTD_writeFrameHeader zfs_ZSTD_writeFrameHeader
	#define ZSTD_writeLastEmptyBlock zfs_ZSTD_writeLastEmptyBlock
	#define algoTime zfs_algoTime
	#define attachDictSizeCutoffs zfs_attachDictSizeCutoffs
	#define g_ctx zfs_g_ctx
	#define g_debuglevel zfs_g_debuglevel
	#define kInverseProbabilityLog256 zfs_kInverseProbabilityLog256
	#define repStartValue zfs_repStartValue
	#define FSE_isError zfs_FSE_isError
	#define HUF_isError zfs_HUF_isError
	diff --git a/sys/contrib/openzfs/module/zstd/lib/zstd.c b/sys/contrib/openzfs/module/zstd/lib/zstd.c
	index acdd4d9dac9d..2766e5b74f55 100644
	--- a/sys/contrib/openzfs/module/zstd/lib/zstd.c
	+++ b/sys/contrib/openzfs/module/zstd/lib/zstd.c
	@@ -1,27826 +1,27824 @@
	/*
	* BSD 3-Clause Clear License
	*
	* Redistribution and use in source and binary forms, with or without
	* modification, are permitted provided that the following conditions are met:
	*
	* 1. Redistributions of source code must retain the above copyright notice,
	* this list of conditions and the following disclaimer.
	*
	* 2. Redistributions in binary form must reproduce the above copyright notice,
	* this list of conditions and the following disclaimer in the documentation
	* and/or other materials provided with the distribution.
	*
	* 3. Neither the name of the copyright holder nor the names of its
	* contributors may be used to endorse or promote products derived from this
	* software without specific prior written permission.
	*
	* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
	* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
	* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
	* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
	* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
	* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
	* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
	* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
	* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
	* POSSIBILITY OF SUCH DAMAGE.
	*/

	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc. All rights reserved.
	* Copyright (c) 2019-2020, Michael Niewöhner. All rights reserved.
	*/

	#define MEM_MODULE
	#define XXH_NAMESPACE ZSTD_
	#define XXH_PRIVATE_API
	#define XXH_INLINE_ALL
	#define ZSTD_LEGACY_SUPPORT 0
	#define ZSTD_LIB_DICTBUILDER 0
	#define ZSTD_LIB_DEPRECATED 0
	#define ZSTD_NOBENCH

	/** start inlining common/debug.c **/
	/* ******************************************************************
	* debug
	* Part of FSE library
	* Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */


	/*
	* This module only hosts one global variable
	* which can be used to dynamically influence the verbosity of traces,
	* such as DEBUGLOG and RAWLOG
	*/

	/** start inlining debug.h **/
	/* ******************************************************************
	* debug
	* Part of FSE library
	* Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */


	/*
	* The purpose of this header is to enable debug functions.
	* They regroup assert(), DEBUGLOG() and RAWLOG() for run-time,
	* and DEBUG_STATIC_ASSERT() for compile-time.
	*
	* By default, DEBUGLEVEL==0, which means run-time debug is disabled.
	*
	* Level 1 enables assert() only.
	* Starting level 2, traces can be generated and pushed to stderr.
	* The higher the level, the more verbose the traces.
	*
	* It's possible to dynamically adjust level using variable g_debug_level,
	* which is only declared if DEBUGLEVEL>=2,
	* and is a global variable, not multi-thread protected (use with care)
	*/

	#ifndef DEBUG_H_12987983217
	#define DEBUG_H_12987983217

	#if defined (__cplusplus)
	extern "C" {
	#endif


	/* static assert is triggered at compile time, leaving no runtime artefact.
	* static assert only works with compile-time constants.
	* Also, this variant can only be used inside a function. */
	#define DEBUG_STATIC_ASSERT(c) (void)sizeof(char[(c) ? 1 : -1])


	/* DEBUGLEVEL is expected to be defined externally,
	* typically through compiler command line.
	* Value must be a number. */
	#ifndef DEBUGLEVEL
	# define DEBUGLEVEL 0
	#endif


	/* DEBUGFILE can be defined externally,
	* typically through compiler command line.
	* note : currently useless.
	* Value must be stderr or stdout */
	#ifndef DEBUGFILE
	# define DEBUGFILE stderr
	#endif


	/* recommended values for DEBUGLEVEL :
	* 0 : release mode, no debug, all run-time checks disabled
	* 1 : enables assert() only, no display
	* 2 : reserved, for currently active debug path
	* 3 : events once per object lifetime (CCtx, CDict, etc.)
	* 4 : events once per frame
	* 5 : events once per block
	* 6 : events once per sequence (verbose)
	* 7+: events at every position (very verbose)
	*
	* It's generally inconvenient to output traces > 5.
	* In which case, it's possible to selectively trigger high verbosity levels
	* by modifying g_debug_level.
	*/

	#if (DEBUGLEVEL>=1)
	# include <assert.h>
	#else
	# ifndef assert /* assert may be already defined, due to prior #include <assert.h> */
	# define assert(condition) ((void)0) /* disable assert (default) */
	# endif
	#endif

	#if (DEBUGLEVEL>=2)
	# include <stdio.h>
	extern int g_debuglevel; /* the variable is only declared,
	it actually lives in debug.c,
	and is shared by the whole process.
	It's not thread-safe.
	It's useful when enabling very verbose levels
	on selective conditions (such as position in src) */

	# define RAWLOG(l, ...) { \
	if (l<=g_debuglevel) { \
	fprintf(stderr, __VA_ARGS__); \
	} }
	# define DEBUGLOG(l, ...) { \
	if (l<=g_debuglevel) { \
	fprintf(stderr, __FILE__ ": " __VA_ARGS__); \
	fprintf(stderr, " \n"); \
	} }
	#else
	# define RAWLOG(l, ...) {} /* disabled */
	# define DEBUGLOG(l, ...) {} /* disabled */
	#endif


	#if defined (__cplusplus)
	}
	#endif

	#endif /* DEBUG_H_12987983217 */
	/** ended inlining debug.h **/

	int g_debuglevel = DEBUGLEVEL;
	/** ended inlining common/debug.c **/
	/** start inlining common/entropy_common.c **/
	/* ******************************************************************
	* Common functions of New Generation Entropy library
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - FSE+HUF source repository : https://github.com/Cyan4973/FiniteStateEntropy
	* - Public forum : https://groups.google.com/forum/#!forum/lz4c
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */

	/* *************************************
	* Dependencies
	***************************************/
	/** start inlining mem.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef MEM_H_MODULE
	#define MEM_H_MODULE

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/-***************************************
	* Dependencies
	******************************************/
	#include <stddef.h> /* size_t, ptrdiff_t */
	#include <string.h> /* memcpy */


	/-***************************************
	* Compiler specifics
	******************************************/
	#if defined(_MSC_VER) /* Visual Studio */
	# include <stdlib.h> /* _byteswap_ulong */
	# include <intrin.h> /* _byteswap_* */
	#endif
	#if defined(__GNUC__)
	# define MEM_STATIC static __inline __attribute__((unused))
	#elif defined (__cplusplus) \|\| (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */)
	# define MEM_STATIC static inline
	#elif defined(_MSC_VER)
	# define MEM_STATIC static __inline
	#else
	# define MEM_STATIC static /* this version may generate warnings for unused static functions; disable the relevant warning */
	#endif

	#ifndef __has_builtin
	# define __has_builtin(x) 0 /* compat. with non-clang compilers */
	#endif

	/* code only tested on 32 and 64 bits systems */
	#define MEM_STATIC_ASSERT(c) { enum { MEM_static_assert = 1/(int)(!!(c)) }; }
	MEM_STATIC void MEM_check(void) { MEM_STATIC_ASSERT((sizeof(size_t)==4) \|\| (sizeof(size_t)==8)); }

	/* detects whether we are being compiled under msan */
	#if defined (__has_feature)
	# if __has_feature(memory_sanitizer)
	# define MEMORY_SANITIZER 1
	# endif
	#endif

	#if defined (MEMORY_SANITIZER)
	/* Not all platforms that support msan provide sanitizers/msan_interface.h.
	* We therefore declare the functions we need ourselves, rather than trying to
	* include the header file... */

	#include <stdint.h> /* intptr_t */

	/* Make memory region fully initialized (without changing its contents). */
	void __msan_unpoison(const volatile void *a, size_t size);

	/* Make memory region fully uninitialized (without changing its contents).
	This is a legacy interface that does not update origin information. Use
	__msan_allocated_memory() instead. */
	void __msan_poison(const volatile void *a, size_t size);

	/* Returns the offset of the first (at least partially) poisoned byte in the
	memory range, or -1 if the whole range is good. */
	intptr_t __msan_test_shadow(const volatile void *x, size_t size);
	#endif

	/* detects whether we are being compiled under asan */
	#if defined (__has_feature)
	# if __has_feature(address_sanitizer)
	# define ADDRESS_SANITIZER 1
	# endif
	#elif defined(__SANITIZE_ADDRESS__)
	# define ADDRESS_SANITIZER 1
	#endif

	#if defined (ADDRESS_SANITIZER)
	/* Not all platforms that support asan provide sanitizers/asan_interface.h.
	* We therefore declare the functions we need ourselves, rather than trying to
	* include the header file... */

	/**
	* Marks a memory region (<c>[addr, addr+size)</c>) as unaddressable.
	*
	* This memory must be previously allocated by your program. Instrumented
	* code is forbidden from accessing addresses in this region until it is
	* unpoisoned. This function is not guaranteed to poison the entire region -
	* it could poison only a subregion of <c>[addr, addr+size)</c> due to ASan
	* alignment restrictions.
	*
	* \note This function is not thread-safe because no two threads can poison or
	* unpoison memory in the same memory region simultaneously.
	*
	* \param addr Start of memory region.
	* \param size Size of memory region. */
	void __asan_poison_memory_region(void const volatile *addr, size_t size);

	/**
	* Marks a memory region (<c>[addr, addr+size)</c>) as addressable.
	*
	* This memory must be previously allocated by your program. Accessing
	* addresses in this region is allowed until this region is poisoned again.
	* This function could unpoison a super-region of <c>[addr, addr+size)</c> due
	* to ASan alignment restrictions.
	*
	* \note This function is not thread-safe because no two threads can
	* poison or unpoison memory in the same memory region simultaneously.
	*
	* \param addr Start of memory region.
	* \param size Size of memory region. */
	void __asan_unpoison_memory_region(void const volatile *addr, size_t size);
	#endif


	/-*************************************************************
	* Basic Types
	*****************************************************************/
	#if !defined (__VMS) && (defined (__cplusplus) \|\| (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
	# include <stdint.h>
	typedef uint8_t BYTE;
	typedef uint16_t U16;
	typedef int16_t S16;
	typedef uint32_t U32;
	typedef int32_t S32;
	typedef uint64_t U64;
	typedef int64_t S64;
	#else
	# include <limits.h>
	#if CHAR_BIT != 8
	# error "this implementation requires char to be exactly 8-bit type"
	#endif
	typedef unsigned char BYTE;
	#if USHRT_MAX != 65535
	# error "this implementation requires short to be exactly 16-bit type"
	#endif
	typedef unsigned short U16;
	typedef signed short S16;
	#if UINT_MAX != 4294967295
	# error "this implementation requires int to be exactly 32-bit type"
	#endif
	typedef unsigned int U32;
	typedef signed int S32;
	/* note : there are no limits defined for long long type in C90.
	* limits exist in C99, however, in such case, <stdint.h> is preferred */
	typedef unsigned long long U64;
	typedef signed long long S64;
	#endif


	/-*************************************************************
	* Memory I/O
	*****************************************************************/
	/* MEM_FORCE_MEMORY_ACCESS :
	* By default, access to unaligned memory is controlled by `memcpy()`, which is safe and portable.
	* Unfortunately, on some target/compiler combinations, the generated assembly is sub-optimal.
	* The below switch allow to select different access method for improved performance.
	* Method 0 (default) : use `memcpy()`. Safe and portable.
	* Method 1 : `__packed` statement. It depends on compiler extension (i.e., not portable).
	* This method is safe if your compiler supports it, and generally as fast or faster than `memcpy`.
	* Method 2 : direct access. This method is portable but violate C standard.
	* It can generate buggy code on targets depending on alignment.
	* In some circumstances, it's the only known way to get the most performance (i.e. GCC + ARMv6)
	* See http://fastcompression.blogspot.fr/2015/08/accessing-unaligned-memory.html for details.
	* Prefer these methods in priority order (0 > 1 > 2)
	*/
	#ifndef MEM_FORCE_MEMORY_ACCESS /* can be defined externally, on command line for example */
	# if defined(__GNUC__) && ( defined(__ARM_ARCH_6__) \|\| defined(__ARM_ARCH_6J__) \|\| defined(__ARM_ARCH_6K__) \|\| defined(__ARM_ARCH_6Z__) \|\| defined(__ARM_ARCH_6ZK__) \|\| defined(__ARM_ARCH_6T2__) )
	# define MEM_FORCE_MEMORY_ACCESS 2
	# elif defined(__INTEL_COMPILER) \|\| defined(__GNUC__) \|\| defined(__ICCARM__)
	# define MEM_FORCE_MEMORY_ACCESS 1
	# endif
	#endif

	MEM_STATIC unsigned MEM_32bits(void) { return sizeof(size_t)==4; }
	MEM_STATIC unsigned MEM_64bits(void) { return sizeof(size_t)==8; }

	MEM_STATIC unsigned MEM_isLittleEndian(void)
	{
	const union { U32 u; BYTE c[4]; } one = { 1 }; /* don't use static : performance detrimental */
	return one.c[0];
	}

	#if defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==2)

	/* violates C standard, by lying on structure alignment.
	Only use if no other choice to achieve best performance on target platform */
	MEM_STATIC U16 MEM_read16(const void* memPtr) { return (const U16) memPtr; }
	MEM_STATIC U32 MEM_read32(const void* memPtr) { return (const U32) memPtr; }
	MEM_STATIC U64 MEM_read64(const void* memPtr) { return (const U64) memPtr; }
	MEM_STATIC size_t MEM_readST(const void* memPtr) { return (const size_t) memPtr; }

	MEM_STATIC void MEM_write16(void* memPtr, U16 value) { (U16)memPtr = value; }
	MEM_STATIC void MEM_write32(void* memPtr, U32 value) { (U32)memPtr = value; }
	MEM_STATIC void MEM_write64(void* memPtr, U64 value) { (U64)memPtr = value; }

	#elif defined(MEM_FORCE_MEMORY_ACCESS) && (MEM_FORCE_MEMORY_ACCESS==1)

	/* __pack instructions are safer, but compiler specific, hence potentially problematic for some compilers */
	/* currently only defined for gcc and icc */
	#if defined(_MSC_VER) \|\| (defined(__INTEL_COMPILER) && defined(WIN32))
	__pragma( pack(push, 1) )
	typedef struct { U16 v; } unalign16;
	typedef struct { U32 v; } unalign32;
	typedef struct { U64 v; } unalign64;
	typedef struct { size_t v; } unalignArch;
	__pragma( pack(pop) )
	#else
	typedef struct { U16 v; } __attribute__((packed)) unalign16;
	typedef struct { U32 v; } __attribute__((packed)) unalign32;
	typedef struct { U64 v; } __attribute__((packed)) unalign64;
	typedef struct { size_t v; } __attribute__((packed)) unalignArch;
	#endif

	MEM_STATIC U16 MEM_read16(const void* ptr) { return ((const unalign16*)ptr)->v; }
	MEM_STATIC U32 MEM_read32(const void* ptr) { return ((const unalign32*)ptr)->v; }
	MEM_STATIC U64 MEM_read64(const void* ptr) { return ((const unalign64*)ptr)->v; }
	MEM_STATIC size_t MEM_readST(const void* ptr) { return ((const unalignArch*)ptr)->v; }

	MEM_STATIC void MEM_write16(void* memPtr, U16 value) { ((unalign16*)memPtr)->v = value; }
	MEM_STATIC void MEM_write32(void* memPtr, U32 value) { ((unalign32*)memPtr)->v = value; }
	MEM_STATIC void MEM_write64(void* memPtr, U64 value) { ((unalign64*)memPtr)->v = value; }

	#else

	/* default method, safe and standard.
	can sometimes prove slower */

	MEM_STATIC U16 MEM_read16(const void* memPtr)
	{
	U16 val; memcpy(&val, memPtr, sizeof(val)); return val;
	}

	MEM_STATIC U32 MEM_read32(const void* memPtr)
	{
	U32 val; memcpy(&val, memPtr, sizeof(val)); return val;
	}

	MEM_STATIC U64 MEM_read64(const void* memPtr)
	{
	U64 val; memcpy(&val, memPtr, sizeof(val)); return val;
	}

	MEM_STATIC size_t MEM_readST(const void* memPtr)
	{
	size_t val; memcpy(&val, memPtr, sizeof(val)); return val;
	}

	MEM_STATIC void MEM_write16(void* memPtr, U16 value)
	{
	memcpy(memPtr, &value, sizeof(value));
	}

	MEM_STATIC void MEM_write32(void* memPtr, U32 value)
	{
	memcpy(memPtr, &value, sizeof(value));
	}

	MEM_STATIC void MEM_write64(void* memPtr, U64 value)
	{
	memcpy(memPtr, &value, sizeof(value));
	}

	#endif /* MEM_FORCE_MEMORY_ACCESS */

	MEM_STATIC U32 MEM_swap32(U32 in)
	{
	#if defined(_MSC_VER) /* Visual Studio */
	return _byteswap_ulong(in);
	#elif (defined (__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__ >= 403)) \
	\|\| (defined(__clang__) && __has_builtin(__builtin_bswap32))
	return __builtin_bswap32(in);
	#else
	return ((in << 24) & 0xff000000 ) \|
	((in << 8) & 0x00ff0000 ) \|
	((in >> 8) & 0x0000ff00 ) \|
	((in >> 24) & 0x000000ff );
	#endif
	}

	MEM_STATIC U64 MEM_swap64(U64 in)
	{
	#if defined(_MSC_VER) /* Visual Studio */
	return _byteswap_uint64(in);
	#elif (defined (__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__ >= 403)) \
	\|\| (defined(__clang__) && __has_builtin(__builtin_bswap64))
	return __builtin_bswap64(in);
	#else
	return ((in << 56) & 0xff00000000000000ULL) \|
	((in << 40) & 0x00ff000000000000ULL) \|
	((in << 24) & 0x0000ff0000000000ULL) \|
	((in << 8) & 0x000000ff00000000ULL) \|
	((in >> 8) & 0x00000000ff000000ULL) \|
	((in >> 24) & 0x0000000000ff0000ULL) \|
	((in >> 40) & 0x000000000000ff00ULL) \|
	((in >> 56) & 0x00000000000000ffULL);
	#endif
	}

	MEM_STATIC size_t MEM_swapST(size_t in)
	{
	if (MEM_32bits())
	return (size_t)MEM_swap32((U32)in);
	else
	return (size_t)MEM_swap64((U64)in);
	}

	/=== Little endian r/w ===/

	MEM_STATIC U16 MEM_readLE16(const void* memPtr)
	{
	if (MEM_isLittleEndian())
	return MEM_read16(memPtr);
	else {
	const BYTE* p = (const BYTE*)memPtr;
	return (U16)(p[0] + (p[1]<<8));
	}
	}

	MEM_STATIC void MEM_writeLE16(void* memPtr, U16 val)
	{
	if (MEM_isLittleEndian()) {
	MEM_write16(memPtr, val);
	} else {
	BYTE* p = (BYTE*)memPtr;
	p[0] = (BYTE)val;
	p[1] = (BYTE)(val>>8);
	}
	}

	MEM_STATIC U32 MEM_readLE24(const void* memPtr)
	{
	return MEM_readLE16(memPtr) + (((const BYTE*)memPtr)[2] << 16);
	}

	MEM_STATIC void MEM_writeLE24(void* memPtr, U32 val)
	{
	MEM_writeLE16(memPtr, (U16)val);
	((BYTE*)memPtr)[2] = (BYTE)(val>>16);
	}

	MEM_STATIC U32 MEM_readLE32(const void* memPtr)
	{
	if (MEM_isLittleEndian())
	return MEM_read32(memPtr);
	else
	return MEM_swap32(MEM_read32(memPtr));
	}

	MEM_STATIC void MEM_writeLE32(void* memPtr, U32 val32)
	{
	if (MEM_isLittleEndian())
	MEM_write32(memPtr, val32);
	else
	MEM_write32(memPtr, MEM_swap32(val32));
	}

	MEM_STATIC U64 MEM_readLE64(const void* memPtr)
	{
	if (MEM_isLittleEndian())
	return MEM_read64(memPtr);
	else
	return MEM_swap64(MEM_read64(memPtr));
	}

	MEM_STATIC void MEM_writeLE64(void* memPtr, U64 val64)
	{
	if (MEM_isLittleEndian())
	MEM_write64(memPtr, val64);
	else
	MEM_write64(memPtr, MEM_swap64(val64));
	}

	MEM_STATIC size_t MEM_readLEST(const void* memPtr)
	{
	if (MEM_32bits())
	return (size_t)MEM_readLE32(memPtr);
	else
	return (size_t)MEM_readLE64(memPtr);
	}

	MEM_STATIC void MEM_writeLEST(void* memPtr, size_t val)
	{
	if (MEM_32bits())
	MEM_writeLE32(memPtr, (U32)val);
	else
	MEM_writeLE64(memPtr, (U64)val);
	}

	/=== Big endian r/w ===/

	MEM_STATIC U32 MEM_readBE32(const void* memPtr)
	{
	if (MEM_isLittleEndian())
	return MEM_swap32(MEM_read32(memPtr));
	else
	return MEM_read32(memPtr);
	}

	MEM_STATIC void MEM_writeBE32(void* memPtr, U32 val32)
	{
	if (MEM_isLittleEndian())
	MEM_write32(memPtr, MEM_swap32(val32));
	else
	MEM_write32(memPtr, val32);
	}

	MEM_STATIC U64 MEM_readBE64(const void* memPtr)
	{
	if (MEM_isLittleEndian())
	return MEM_swap64(MEM_read64(memPtr));
	else
	return MEM_read64(memPtr);
	}

	MEM_STATIC void MEM_writeBE64(void* memPtr, U64 val64)
	{
	if (MEM_isLittleEndian())
	MEM_write64(memPtr, MEM_swap64(val64));
	else
	MEM_write64(memPtr, val64);
	}

	MEM_STATIC size_t MEM_readBEST(const void* memPtr)
	{
	if (MEM_32bits())
	return (size_t)MEM_readBE32(memPtr);
	else
	return (size_t)MEM_readBE64(memPtr);
	}

	MEM_STATIC void MEM_writeBEST(void* memPtr, size_t val)
	{
	if (MEM_32bits())
	MEM_writeBE32(memPtr, (U32)val);
	else
	MEM_writeBE64(memPtr, (U64)val);
	}


	#if defined (__cplusplus)
	}
	#endif

	#endif /* MEM_H_MODULE */
	/** ended inlining mem.h **/
	/** start inlining error_private.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/* Note : this module is expected to remain private, do not expose it */

	#ifndef ERROR_H_MODULE
	#define ERROR_H_MODULE

	#if defined (__cplusplus)
	extern "C" {
	#endif


	/* ****************************************
	* Dependencies
	******************************************/
	#include <stddef.h> /* size_t */
	/** start inlining zstd_errors.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_ERRORS_H_398273423
	#define ZSTD_ERRORS_H_398273423

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/===== dependency =====/
	#include <stddef.h> /* size_t */


	/* ===== ZSTDERRORLIB_API : control library symbols visibility ===== */
	#ifndef ZSTDERRORLIB_VISIBILITY
	# if defined(__GNUC__) && (__GNUC__ >= 4)
	# define ZSTDERRORLIB_VISIBILITY __attribute__ ((visibility ("default")))
	# else
	# define ZSTDERRORLIB_VISIBILITY
	# endif
	#endif
	#if defined(ZSTD_DLL_EXPORT) && (ZSTD_DLL_EXPORT==1)
	# define ZSTDERRORLIB_API __declspec(dllexport) ZSTDERRORLIB_VISIBILITY
	#elif defined(ZSTD_DLL_IMPORT) && (ZSTD_DLL_IMPORT==1)
	# define ZSTDERRORLIB_API __declspec(dllimport) ZSTDERRORLIB_VISIBILITY /* It isn't required but allows to generate better code, saving a function pointer load from the IAT and an indirect jump.*/
	#else
	# define ZSTDERRORLIB_API ZSTDERRORLIB_VISIBILITY
	#endif

	/-********************************************
	* Error codes list
	-********************************************
	* Error codes _values_ are pinned down since v1.3.1 only.
	* Therefore, don't rely on values if you may link to any version < v1.3.1.
	*
	* Only values < 100 are considered stable.
	*
	* note 1 : this API shall be used with static linking only.
	* dynamic linking is not yet officially supported.
	* note 2 : Prefer relying on the enum than on its value whenever possible
	* This is the only supported way to use the error list < v1.3.1
	* note 3 : ZSTD_isError() is always correct, whatever the library version.
	**********************************************/
	typedef enum {
	ZSTD_error_no_error = 0,
	ZSTD_error_GENERIC = 1,
	ZSTD_error_prefix_unknown = 10,
	ZSTD_error_version_unsupported = 12,
	ZSTD_error_frameParameter_unsupported = 14,
	ZSTD_error_frameParameter_windowTooLarge = 16,
	ZSTD_error_corruption_detected = 20,
	ZSTD_error_checksum_wrong = 22,
	ZSTD_error_dictionary_corrupted = 30,
	ZSTD_error_dictionary_wrong = 32,
	ZSTD_error_dictionaryCreation_failed = 34,
	ZSTD_error_parameter_unsupported = 40,
	ZSTD_error_parameter_outOfBound = 42,
	ZSTD_error_tableLog_tooLarge = 44,
	ZSTD_error_maxSymbolValue_tooLarge = 46,
	ZSTD_error_maxSymbolValue_tooSmall = 48,
	ZSTD_error_stage_wrong = 60,
	ZSTD_error_init_missing = 62,
	ZSTD_error_memory_allocation = 64,
	ZSTD_error_workSpace_tooSmall= 66,
	ZSTD_error_dstSize_tooSmall = 70,
	ZSTD_error_srcSize_wrong = 72,
	ZSTD_error_dstBuffer_null = 74,
	/* following error codes are __NOT STABLE__, they can be removed or changed in future versions */
	ZSTD_error_frameIndex_tooLarge = 100,
	ZSTD_error_seekableIO = 102,
	ZSTD_error_dstBuffer_wrong = 104,
	ZSTD_error_maxCode = 120 /* never EVER use this value directly, it can change in future versions! Use ZSTD_isError() instead */
	} ZSTD_ErrorCode;

	/*! ZSTD_getErrorCode() :
	convert a `size_t` function result into a `ZSTD_ErrorCode` enum type,
	which can be used to compare with enum list published above */
	ZSTDERRORLIB_API ZSTD_ErrorCode ZSTD_getErrorCode(size_t functionResult);
	ZSTDERRORLIB_API const char* ZSTD_getErrorString(ZSTD_ErrorCode code); /*< Same as ZSTD_getErrorName, but using a `ZSTD_ErrorCode` enum argument /


	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_ERRORS_H_398273423 */
	/** ended inlining zstd_errors.h **/


	/* ****************************************
	* Compiler-specific
	******************************************/
	#if defined(__GNUC__)
	# define ERR_STATIC static __attribute__((unused))
	#elif defined (__cplusplus) \|\| (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */)
	# define ERR_STATIC static inline
	#elif defined(_MSC_VER)
	# define ERR_STATIC static __inline
	#else
	# define ERR_STATIC static /* this version may generate warnings for unused static functions; disable the relevant warning */
	#endif


	/-***************************************
	* Customization (error_public.h)
	******************************************/
	typedef ZSTD_ErrorCode ERR_enum;
	#define PREFIX(name) ZSTD_error_##name


	/-***************************************
	* Error codes handling
	******************************************/
	#undef ERROR /* already defined on Visual Studio */
	#define ERROR(name) ZSTD_ERROR(name)
	#define ZSTD_ERROR(name) ((size_t)-PREFIX(name))

	ERR_STATIC unsigned ERR_isError(size_t code) { return (code > ERROR(maxCode)); }

	ERR_STATIC ERR_enum ERR_getErrorCode(size_t code) { if (!ERR_isError(code)) return (ERR_enum)0; return (ERR_enum) (0-code); }

	/* check and forward error code */
	#define CHECK_V_F(e, f) size_t const e = f; if (ERR_isError(e)) return e
	#define CHECK_F(f) { CHECK_V_F(_var_err__, f); }


	/-***************************************
	* Error Strings
	******************************************/

	const char* ERR_getErrorString(ERR_enum code); /* error_private.c */

	ERR_STATIC const char* ERR_getErrorName(size_t code)
	{
	return ERR_getErrorString(ERR_getErrorCode(code));
	}

	#if defined (__cplusplus)
	}
	#endif

	#endif /* ERROR_H_MODULE */
	/** ended inlining error_private.h **/
	#define FSE_STATIC_LINKING_ONLY /* FSE_MIN_TABLELOG */
	/** start inlining fse.h **/
	/* ******************************************************************
	* FSE : Finite State Entropy codec
	* Public Prototypes declaration
	* Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */

	#if defined (__cplusplus)
	extern "C" {
	#endif

	#ifndef FSE_H
	#define FSE_H


	/-****************************************
	* Dependencies
	******************************************/
	#include <stddef.h> /* size_t, ptrdiff_t */


	/-****************************************
	* FSE_PUBLIC_API : control library symbols visibility
	******************************************/
	#if defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1) && defined(__GNUC__) && (__GNUC__ >= 4)
	# define FSE_PUBLIC_API __attribute__ ((visibility ("default")))
	#elif defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1) /* Visual expected */
	# define FSE_PUBLIC_API __declspec(dllexport)
	#elif defined(FSE_DLL_IMPORT) && (FSE_DLL_IMPORT==1)
	# define FSE_PUBLIC_API __declspec(dllimport) /* It isn't required but allows to generate better code, saving a function pointer load from the IAT and an indirect jump.*/
	#else
	# define FSE_PUBLIC_API
	#endif

	/------ Version ------/
	#define FSE_VERSION_MAJOR 0
	#define FSE_VERSION_MINOR 9
	#define FSE_VERSION_RELEASE 0

	#define FSE_LIB_VERSION FSE_VERSION_MAJOR.FSE_VERSION_MINOR.FSE_VERSION_RELEASE
	#define FSE_QUOTE(str) #str
	#define FSE_EXPAND_AND_QUOTE(str) FSE_QUOTE(str)
	#define FSE_VERSION_STRING FSE_EXPAND_AND_QUOTE(FSE_LIB_VERSION)

	#define FSE_VERSION_NUMBER (FSE_VERSION_MAJOR 100100 + FSE_VERSION_MINOR *100 + FSE_VERSION_RELEASE)
	FSE_PUBLIC_API unsigned FSE_versionNumber(void); /*< library version number; to be used when checking dll version /


	/-***************************************
	* FSE simple functions
	******************************************/
	/*! FSE_compress() :
	Compress content of buffer 'src', of size 'srcSize', into destination buffer 'dst'.
	'dst' buffer must be already allocated. Compression runs faster is dstCapacity >= FSE_compressBound(srcSize).
	@return : size of compressed data (<= dstCapacity).
	Special values : if return == 0, srcData is not compressible => Nothing is stored within dst !!!
	if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression instead.
	if FSE_isError(return), compression failed (more details using FSE_getErrorName())
	*/
	FSE_PUBLIC_API size_t FSE_compress(void* dst, size_t dstCapacity,
	const void* src, size_t srcSize);

	/*! FSE_decompress():
	Decompress FSE data from buffer 'cSrc', of size 'cSrcSize',
	into already allocated destination buffer 'dst', of size 'dstCapacity'.
	@return : size of regenerated data (<= maxDstSize),
	or an error code, which can be tested using FSE_isError() .

	Important : FSE_decompress() does not decompress non-compressible nor RLE data !!!
	Why ? : making this distinction requires a header.
	Header management is intentionally delegated to the user layer, which can better manage special cases.
	*/
	FSE_PUBLIC_API size_t FSE_decompress(void* dst, size_t dstCapacity,
	const void* cSrc, size_t cSrcSize);


	/-****************************************
	* Tool functions
	******************************************/
	FSE_PUBLIC_API size_t FSE_compressBound(size_t size); /* maximum compressed size */

	/* Error Management */
	FSE_PUBLIC_API unsigned FSE_isError(size_t code); /* tells if a return value is an error code */
	FSE_PUBLIC_API const char* FSE_getErrorName(size_t code); /* provides error code string (useful for debugging) */


	/-****************************************
	* FSE advanced functions
	******************************************/
	/*! FSE_compress2() :
	Same as FSE_compress(), but allows the selection of 'maxSymbolValue' and 'tableLog'
	Both parameters can be defined as '0' to mean : use default value
	@return : size of compressed data
	Special values : if return == 0, srcData is not compressible => Nothing is stored within cSrc !!!
	if return == 1, srcData is a single byte symbol * srcSize times. Use RLE compression.
	if FSE_isError(return), it's an error code.
	*/
	FSE_PUBLIC_API size_t FSE_compress2 (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog);


	/-****************************************
	* FSE detailed API
	******************************************/
	/*!
	FSE_compress() does the following:
	1. count symbol occurrence from source[] into table count[] (see hist.h)
	2. normalize counters so that sum(count[]) == Power_of_2 (2^tableLog)
	3. save normalized counters to memory buffer using writeNCount()
	4. build encoding table 'CTable' from normalized counters
	5. encode the data stream using encoding table 'CTable'

	FSE_decompress() does the following:
	1. read normalized counters with readNCount()
	2. build decoding table 'DTable' from normalized counters
	3. decode the data stream using decoding table 'DTable'

	The following API allows targeting specific sub-functions for advanced tasks.
	For example, it's possible to compress several blocks using the same 'CTable',
	or to save and provide normalized distribution using external method.
	*/

	/* * COMPRESSION * */

	/*! FSE_optimalTableLog():
	dynamically downsize 'tableLog' when conditions are met.
	It saves CPU time, by using smaller tables, while preserving or even improving compression ratio.
	@return : recommended tableLog (necessarily <= 'maxTableLog') */
	FSE_PUBLIC_API unsigned FSE_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);

	/*! FSE_normalizeCount():
	normalize counts so that sum(count[]) == Power_of_2 (2^tableLog)
	'normalizedCounter' is a table of short, of minimum size (maxSymbolValue+1).
	@return : tableLog,
	or an errorCode, which can be tested using FSE_isError() */
	FSE_PUBLIC_API size_t FSE_normalizeCount(short* normalizedCounter, unsigned tableLog,
	const unsigned* count, size_t srcSize, unsigned maxSymbolValue);

	/*! FSE_NCountWriteBound():
	Provides the maximum possible size of an FSE normalized table, given 'maxSymbolValue' and 'tableLog'.
	Typically useful for allocation purpose. */
	FSE_PUBLIC_API size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog);

	/*! FSE_writeNCount():
	Compactly save 'normalizedCounter' into 'buffer'.
	@return : size of the compressed table,
	or an errorCode, which can be tested using FSE_isError(). */
	FSE_PUBLIC_API size_t FSE_writeNCount (void* buffer, size_t bufferSize,
	const short* normalizedCounter,
	unsigned maxSymbolValue, unsigned tableLog);

	/*! Constructor and Destructor of FSE_CTable.
	Note that FSE_CTable size depends on 'tableLog' and 'maxSymbolValue' */
	typedef unsigned FSE_CTable; /* don't allocate that. It's only meant to be more restrictive than void* */
	FSE_PUBLIC_API FSE_CTable* FSE_createCTable (unsigned maxSymbolValue, unsigned tableLog);
	FSE_PUBLIC_API void FSE_freeCTable (FSE_CTable* ct);

	/*! FSE_buildCTable():
	Builds `ct`, which must be already allocated, using FSE_createCTable().
	@return : 0, or an errorCode, which can be tested using FSE_isError() */
	FSE_PUBLIC_API size_t FSE_buildCTable(FSE_CTable* ct, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog);

	/*! FSE_compress_usingCTable():
	Compress `src` using `ct` into `dst` which must be already allocated.
	@return : size of compressed data (<= `dstCapacity`),
	or 0 if compressed data could not fit into `dst`,
	or an errorCode, which can be tested using FSE_isError() */
	FSE_PUBLIC_API size_t FSE_compress_usingCTable (void* dst, size_t dstCapacity, const void* src, size_t srcSize, const FSE_CTable* ct);

	/*!
	Tutorial :
	----------
	The first step is to count all symbols. FSE_count() does this job very fast.
	Result will be saved into 'count', a table of unsigned int, which must be already allocated, and have 'maxSymbolValuePtr[0]+1' cells.
	'src' is a table of bytes of size 'srcSize'. All values within 'src' MUST be <= maxSymbolValuePtr[0]
	maxSymbolValuePtr[0] will be updated, with its real value (necessarily <= original value)
	FSE_count() will return the number of occurrence of the most frequent symbol.
	This can be used to know if there is a single symbol within 'src', and to quickly evaluate its compressibility.
	If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).

	The next step is to normalize the frequencies.
	FSE_normalizeCount() will ensure that sum of frequencies is == 2 ^'tableLog'.
	It also guarantees a minimum of 1 to any Symbol with frequency >= 1.
	You can use 'tableLog'==0 to mean "use default tableLog value".
	If you are unsure of which tableLog value to use, you can ask FSE_optimalTableLog(),
	which will provide the optimal valid tableLog given sourceSize, maxSymbolValue, and a user-defined maximum (0 means "default").

	The result of FSE_normalizeCount() will be saved into a table,
	called 'normalizedCounter', which is a table of signed short.
	'normalizedCounter' must be already allocated, and have at least 'maxSymbolValue+1' cells.
	The return value is tableLog if everything proceeded as expected.
	It is 0 if there is a single symbol within distribution.
	If there is an error (ex: invalid tableLog value), the function will return an ErrorCode (which can be tested using FSE_isError()).

	'normalizedCounter' can be saved in a compact manner to a memory area using FSE_writeNCount().
	'buffer' must be already allocated.
	For guaranteed success, buffer size must be at least FSE_headerBound().
	The result of the function is the number of bytes written into 'buffer'.
	If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError(); ex : buffer size too small).

	'normalizedCounter' can then be used to create the compression table 'CTable'.
	The space required by 'CTable' must be already allocated, using FSE_createCTable().
	You can then use FSE_buildCTable() to fill 'CTable'.
	If there is an error, both functions will return an ErrorCode (which can be tested using FSE_isError()).

	'CTable' can then be used to compress 'src', with FSE_compress_usingCTable().
	Similar to FSE_count(), the convention is that 'src' is assumed to be a table of char of size 'srcSize'
	The function returns the size of compressed data (without header), necessarily <= `dstCapacity`.
	If it returns '0', compressed data could not fit into 'dst'.
	If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).
	*/


	/* * DECOMPRESSION * */

	/*! FSE_readNCount():
	Read compactly saved 'normalizedCounter' from 'rBuffer'.
	@return : size read from 'rBuffer',
	or an errorCode, which can be tested using FSE_isError().
	maxSymbolValuePtr[0] and tableLogPtr[0] will also be updated with their respective values */
	FSE_PUBLIC_API size_t FSE_readNCount (short* normalizedCounter,
	unsigned* maxSymbolValuePtr, unsigned* tableLogPtr,
	const void* rBuffer, size_t rBuffSize);

	/*! Constructor and Destructor of FSE_DTable.
	Note that its size depends on 'tableLog' */
	typedef unsigned FSE_DTable; /* don't allocate that. It's just a way to be more restrictive than void* */
	FSE_PUBLIC_API FSE_DTable* FSE_createDTable(unsigned tableLog);
	FSE_PUBLIC_API void FSE_freeDTable(FSE_DTable* dt);

	/*! FSE_buildDTable():
	Builds 'dt', which must be already allocated, using FSE_createDTable().
	return : 0, or an errorCode, which can be tested using FSE_isError() */
	FSE_PUBLIC_API size_t FSE_buildDTable (FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog);

	/*! FSE_decompress_usingDTable():
	Decompress compressed source `cSrc` of size `cSrcSize` using `dt`
	into `dst` which must be already allocated.
	@return : size of regenerated data (necessarily <= `dstCapacity`),
	or an errorCode, which can be tested using FSE_isError() */
	FSE_PUBLIC_API size_t FSE_decompress_usingDTable(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, const FSE_DTable* dt);

	/*!
	Tutorial :
	----------
	(Note : these functions only decompress FSE-compressed blocks.
	If block is uncompressed, use memcpy() instead
	If block is a single repeated byte, use memset() instead )

	The first step is to obtain the normalized frequencies of symbols.
	This can be performed by FSE_readNCount() if it was saved using FSE_writeNCount().
	'normalizedCounter' must be already allocated, and have at least 'maxSymbolValuePtr[0]+1' cells of signed short.
	In practice, that means it's necessary to know 'maxSymbolValue' beforehand,
	or size the table to handle worst case situations (typically 256).
	FSE_readNCount() will provide 'tableLog' and 'maxSymbolValue'.
	The result of FSE_readNCount() is the number of bytes read from 'rBuffer'.
	Note that 'rBufferSize' must be at least 4 bytes, even if useful information is less than that.
	If there is an error, the function will return an error code, which can be tested using FSE_isError().

	The next step is to build the decompression tables 'FSE_DTable' from 'normalizedCounter'.
	This is performed by the function FSE_buildDTable().
	The space required by 'FSE_DTable' must be already allocated using FSE_createDTable().
	If there is an error, the function will return an error code, which can be tested using FSE_isError().

	`FSE_DTable` can then be used to decompress `cSrc`, with FSE_decompress_usingDTable().
	`cSrcSize` must be strictly correct, otherwise decompression will fail.
	FSE_decompress_usingDTable() result will tell how many bytes were regenerated (<=`dstCapacity`).
	If there is an error, the function will return an error code, which can be tested using FSE_isError(). (ex: dst buffer too small)
	*/

	#endif /* FSE_H */

	#if defined(FSE_STATIC_LINKING_ONLY) && !defined(FSE_H_FSE_STATIC_LINKING_ONLY)
	#define FSE_H_FSE_STATIC_LINKING_ONLY

	/* * Dependency * */
	/** start inlining bitstream.h **/
	/* ******************************************************************
	* bitstream
	* Part of FSE library
	* Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */
	#ifndef BITSTREAM_H_MODULE
	#define BITSTREAM_H_MODULE

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/*
	* This API consists of small unitary functions, which must be inlined for best performance.
	* Since link-time-optimization is not available for all compilers,
	* these functions are defined into a .h to be included.
	*/

	/-***************************************
	* Dependencies
	******************************************/
	/** skipping file: mem.h **/
	/** start inlining compiler.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_COMPILER_H
	#define ZSTD_COMPILER_H

	/-******************************************************
	* Compiler specifics
	*********************************************************/
	/* force inlining */

	#if !defined(ZSTD_NO_INLINE)
	#if (defined(__GNUC__) && !defined(__STRICT_ANSI__)) \|\| defined(__cplusplus) \|\| defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L /* C99 */
	# define INLINE_KEYWORD inline
	#else
	# define INLINE_KEYWORD
	#endif

	#if defined(__GNUC__) \|\| defined(__ICCARM__)
	# define FORCE_INLINE_ATTR __attribute__((always_inline))
	#elif defined(_MSC_VER)
	# define FORCE_INLINE_ATTR __forceinline
	#else
	# define FORCE_INLINE_ATTR
	#endif

	#else

	#define INLINE_KEYWORD
	#define FORCE_INLINE_ATTR

	#endif

	/**
	* FORCE_INLINE_TEMPLATE is used to define C "templates", which take constant
	* parameters. They must be inlined for the compiler to eliminate the constant
	* branches.
	*/
	#define FORCE_INLINE_TEMPLATE static INLINE_KEYWORD FORCE_INLINE_ATTR
	/**
	* HINT_INLINE is used to help the compiler generate better code. It is not
	* used for "templates", so it can be tweaked based on the compilers
	* performance.
	*
	* gcc-4.8 and gcc-4.9 have been shown to benefit from leaving off the
	* always_inline attribute.
	*
	* clang up to 5.0.0 (trunk) benefit tremendously from the always_inline
	* attribute.
	*/
	#if !defined(__clang__) && defined(__GNUC__) && __GNUC__ >= 4 && __GNUC_MINOR__ >= 8 && __GNUC__ < 5
	# define HINT_INLINE static INLINE_KEYWORD
	#else
	# define HINT_INLINE static INLINE_KEYWORD FORCE_INLINE_ATTR
	#endif

	/* UNUSED_ATTR tells the compiler it is okay if the function is unused. */
	#if defined(__GNUC__)
	# define UNUSED_ATTR __attribute__((unused))
	#else
	# define UNUSED_ATTR
	#endif

	/* force no inlining */
	#ifdef _MSC_VER
	# define FORCE_NOINLINE static __declspec(noinline)
	#else
	# if defined(__GNUC__) \|\| defined(__ICCARM__)
	# define FORCE_NOINLINE static __attribute__((__noinline__))
	# else
	# define FORCE_NOINLINE static
	# endif
	#endif

	/* target attribute */
	#ifndef __has_attribute
	#define __has_attribute(x) 0 /* Compatibility with non-clang compilers. */
	#endif
	#if defined(__GNUC__) \|\| defined(__ICCARM__)
	# define TARGET_ATTRIBUTE(target) __attribute__((__target__(target)))
	#else
	# define TARGET_ATTRIBUTE(target)
	#endif

	/* Enable runtime BMI2 dispatch based on the CPU.
	* Enabled for clang & gcc >=4.8 on x86 when BMI2 isn't enabled by default.
	*/
	#ifndef DYNAMIC_BMI2
	#if ((defined(__clang__) && __has_attribute(__target__)) \
	\|\| (defined(__GNUC__) \
	&& (__GNUC__ >= 5 \|\| (__GNUC__ == 4 && __GNUC_MINOR__ >= 8)))) \
	&& (defined(__x86_64__) \|\| defined(_M_X86)) \
	&& !defined(__BMI2__)
	# define DYNAMIC_BMI2 1
	#else
	# define DYNAMIC_BMI2 0
	#endif
	#endif

	/* prefetch
	* can be disabled, by declaring NO_PREFETCH build macro */
	#if defined(NO_PREFETCH)
	# define PREFETCH_L1(ptr) (void)(ptr) /* disabled */
	# define PREFETCH_L2(ptr) (void)(ptr) /* disabled */
	#else
	# if defined(_MSC_VER) && (defined(_M_X64) \|\| defined(_M_I86)) /* _mm_prefetch() is not defined outside of x86/x64 */
	# include <mmintrin.h> /* https://msdn.microsoft.com/fr-fr/library/84szxsww(v=vs.90).aspx */
	# define PREFETCH_L1(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T0)
	# define PREFETCH_L2(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T1)
	# elif defined(__aarch64__)
	# define PREFETCH_L1(ptr) __asm__ __volatile__("prfm pldl1keep, %0" ::"Q"(*(ptr)))
	# define PREFETCH_L2(ptr) __asm__ __volatile__("prfm pldl2keep, %0" ::"Q"(*(ptr)))
	# elif defined(__GNUC__) && ( (__GNUC__ >= 4) \|\| ( (__GNUC__ == 3) && (__GNUC_MINOR__ >= 1) ) )
	# define PREFETCH_L1(ptr) __builtin_prefetch((ptr), 0 /* rw==read /, 3 / locality */)
	# define PREFETCH_L2(ptr) __builtin_prefetch((ptr), 0 /* rw==read /, 2 / locality */)
	# else
	# define PREFETCH_L1(ptr) (void)(ptr) /* disabled */
	# define PREFETCH_L2(ptr) (void)(ptr) /* disabled */
	# endif
	#endif /* NO_PREFETCH */

	#define CACHELINE_SIZE 64

	#define PREFETCH_AREA(p, s) { \
	const char* const _ptr = (const char*)(p); \
	size_t const _size = (size_t)(s); \
	size_t _pos; \
	for (_pos=0; _pos<_size; _pos+=CACHELINE_SIZE) { \
	PREFETCH_L2(_ptr + _pos); \
	} \
	}

	/* vectorization
	* older GCC (pre gcc-4.3 picked as the cutoff) uses a different syntax */
	#if !defined(__INTEL_COMPILER) && !defined(__clang__) && defined(__GNUC__)
	# if (__GNUC__ == 4 && __GNUC_MINOR__ > 3) \|\| (__GNUC__ >= 5)
	# define DONT_VECTORIZE __attribute__((optimize("no-tree-vectorize")))
	# else
	# define DONT_VECTORIZE _Pragma("GCC optimize(\"no-tree-vectorize\")")
	# endif
	#else
	# define DONT_VECTORIZE
	#endif

	/* Tell the compiler that a branch is likely or unlikely.
	* Only use these macros if it causes the compiler to generate better code.
	* If you can remove a LIKELY/UNLIKELY annotation without speed changes in gcc
	* and clang, please do.
	*/
	#if defined(__GNUC__)
	#define LIKELY(x) (__builtin_expect((x), 1))
	#define UNLIKELY(x) (__builtin_expect((x), 0))
	#else
	#define LIKELY(x) (x)
	#define UNLIKELY(x) (x)
	#endif

	/* disable warnings */
	#ifdef _MSC_VER /* Visual Studio */
	# include <intrin.h> /* For Visual 2005 */
	# pragma warning(disable : 4100) /* disable: C4100: unreferenced formal parameter */
	# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */
	# pragma warning(disable : 4204) /* disable: C4204: non-constant aggregate initializer */
	# pragma warning(disable : 4214) /* disable: C4214: non-int bitfields */
	# pragma warning(disable : 4324) /* disable: C4324: padded structure */
	#endif

	#endif /* ZSTD_COMPILER_H */
	/** ended inlining compiler.h **/
	/** skipping file: debug.h **/
	/** skipping file: error_private.h **/


	/*=========================================
	* Target specific
	=========================================*/
	#if defined(__BMI__) && defined(__GNUC__)
	# include <immintrin.h> /* support for bextr (experimental) */
	#elif defined(__ICCARM__)
	# include <intrinsics.h>
	#endif

	#define STREAM_ACCUMULATOR_MIN_32 25
	#define STREAM_ACCUMULATOR_MIN_64 57
	#define STREAM_ACCUMULATOR_MIN ((U32)(MEM_32bits() ? STREAM_ACCUMULATOR_MIN_32 : STREAM_ACCUMULATOR_MIN_64))


	/-*****************************************
	* bitStream encoding API (write forward)
	********************************************/
	/* bitStream can mix input from multiple sources.
	* A critical property of these streams is that they encode and decode in reverse direction.
	* So the first bit sequence you add will be the last to be read, like a LIFO stack.
	*/
	typedef struct {
	size_t bitContainer;
	unsigned bitPos;
	char* startPtr;
	char* ptr;
	char* endPtr;
	} BIT_CStream_t;

	MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC, void* dstBuffer, size_t dstCapacity);
	MEM_STATIC void BIT_addBits(BIT_CStream_t* bitC, size_t value, unsigned nbBits);
	MEM_STATIC void BIT_flushBits(BIT_CStream_t* bitC);
	MEM_STATIC size_t BIT_closeCStream(BIT_CStream_t* bitC);

	/* Start with initCStream, providing the size of buffer to write into.
	* bitStream will never write outside of this buffer.
	* `dstCapacity` must be >= sizeof(bitD->bitContainer), otherwise @return will be an error code.
	*
	* bits are first added to a local register.
	* Local register is size_t, hence 64-bits on 64-bits systems, or 32-bits on 32-bits systems.
	* Writing data into memory is an explicit operation, performed by the flushBits function.
	* Hence keep track how many bits are potentially stored into local register to avoid register overflow.
	* After a flushBits, a maximum of 7 bits might still be stored into local register.
	*
	* Avoid storing elements of more than 24 bits if you want compatibility with 32-bits bitstream readers.
	*
	* Last operation is to close the bitStream.
	* The function returns the final size of CStream in bytes.
	* If data couldn't fit into `dstBuffer`, it will return a 0 ( == not storable)
	*/


	/-*******************************************
	* bitStream decoding API (read backward)
	**********************************************/
	typedef struct {
	size_t bitContainer;
	unsigned bitsConsumed;
	const char* ptr;
	const char* start;
	const char* limitPtr;
	} BIT_DStream_t;

	typedef enum { BIT_DStream_unfinished = 0,
	BIT_DStream_endOfBuffer = 1,
	BIT_DStream_completed = 2,
	BIT_DStream_overflow = 3 } BIT_DStream_status; /* result of BIT_reloadDStream() */
	/* 1,2,4,8 would be better for bitmap combinations, but slows down performance a bit ... :( */

	MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, size_t srcSize);
	MEM_STATIC size_t BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits);
	MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD);
	MEM_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t* bitD);


	/* Start by invoking BIT_initDStream().
	* A chunk of the bitStream is then stored into a local register.
	* Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
	* You can then retrieve bitFields stored into the local register, in reverse order.
	* Local register is explicitly reloaded from memory by the BIT_reloadDStream() method.
	* A reload guarantee a minimum of ((8*sizeof(bitD->bitContainer))-7) bits when its result is BIT_DStream_unfinished.
	* Otherwise, it can be less than that, so proceed accordingly.
	* Checking if DStream has reached its end can be performed with BIT_endOfDStream().
	*/


	/-***************************************
	* unsafe API
	******************************************/
	MEM_STATIC void BIT_addBitsFast(BIT_CStream_t* bitC, size_t value, unsigned nbBits);
	/* faster, but works only if value is "clean", meaning all high bits above nbBits are 0 */

	MEM_STATIC void BIT_flushBitsFast(BIT_CStream_t* bitC);
	/* unsafe version; does not check buffer overflow */

	MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits);
	/* faster, but works only if nbBits >= 1 */



	/-*************************************************************
	* Internal functions
	****************************************************************/
	MEM_STATIC unsigned BIT_highbit32 (U32 val)
	{
	assert(val != 0);
	{
	# if defined(_MSC_VER) /* Visual */
	unsigned long r=0;
	return _BitScanReverse ( &r, val ) ? (unsigned)r : 0;
	# elif defined(__GNUC__) && (__GNUC__ >= 3) /* Use GCC Intrinsic */
	return __builtin_clz (val) ^ 31;
	# elif defined(__ICCARM__) /* IAR Intrinsic */
	return 31 - __CLZ(val);
	# else /* Software version */
	static const unsigned DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29,
	11, 14, 16, 18, 22, 25, 3, 30,
	8, 12, 20, 28, 15, 17, 24, 7,
	19, 27, 23, 6, 26, 5, 4, 31 };
	U32 v = val;
	v \|= v >> 1;
	v \|= v >> 2;
	v \|= v >> 4;
	v \|= v >> 8;
	v \|= v >> 16;
	return DeBruijnClz[ (U32) (v * 0x07C4ACDDU) >> 27];
	# endif
	}
	}

	/===== Local Constants =====/
	static const unsigned BIT_mask[] = {
	0, 1, 3, 7, 0xF, 0x1F,
	0x3F, 0x7F, 0xFF, 0x1FF, 0x3FF, 0x7FF,
	0xFFF, 0x1FFF, 0x3FFF, 0x7FFF, 0xFFFF, 0x1FFFF,
	0x3FFFF, 0x7FFFF, 0xFFFFF, 0x1FFFFF, 0x3FFFFF, 0x7FFFFF,
	0xFFFFFF, 0x1FFFFFF, 0x3FFFFFF, 0x7FFFFFF, 0xFFFFFFF, 0x1FFFFFFF,
	0x3FFFFFFF, 0x7FFFFFFF}; /* up to 31 bits */
	#define BIT_MASK_SIZE (sizeof(BIT_mask) / sizeof(BIT_mask[0]))

	/-*************************************************************
	* bitStream encoding
	****************************************************************/
	/*! BIT_initCStream() :
	* `dstCapacity` must be > sizeof(size_t)
	* @return : 0 if success,
	* otherwise an error code (can be tested using ERR_isError()) */
	MEM_STATIC size_t BIT_initCStream(BIT_CStream_t* bitC,
	void* startPtr, size_t dstCapacity)
	{
	bitC->bitContainer = 0;
	bitC->bitPos = 0;
	bitC->startPtr = (char*)startPtr;
	bitC->ptr = bitC->startPtr;
	bitC->endPtr = bitC->startPtr + dstCapacity - sizeof(bitC->bitContainer);
	if (dstCapacity <= sizeof(bitC->bitContainer)) return ERROR(dstSize_tooSmall);
	return 0;
	}

	/*! BIT_addBits() :
	* can add up to 31 bits into `bitC`.
	* Note : does not check for register overflow ! */
	MEM_STATIC void BIT_addBits(BIT_CStream_t* bitC,
	size_t value, unsigned nbBits)
	{
	MEM_STATIC_ASSERT(BIT_MASK_SIZE == 32);
	assert(nbBits < BIT_MASK_SIZE);
	assert(nbBits + bitC->bitPos < sizeof(bitC->bitContainer) * 8);
	bitC->bitContainer \|= (value & BIT_mask[nbBits]) << bitC->bitPos;
	bitC->bitPos += nbBits;
	}

	/*! BIT_addBitsFast() :
	* works only if `value` is _clean_,
	* meaning all high bits above nbBits are 0 */
	MEM_STATIC void BIT_addBitsFast(BIT_CStream_t* bitC,
	size_t value, unsigned nbBits)
	{
	assert((value>>nbBits) == 0);
	assert(nbBits + bitC->bitPos < sizeof(bitC->bitContainer) * 8);
	bitC->bitContainer \|= value << bitC->bitPos;
	bitC->bitPos += nbBits;
	}

	/*! BIT_flushBitsFast() :
	* assumption : bitContainer has not overflowed
	* unsafe version; does not check buffer overflow */
	MEM_STATIC void BIT_flushBitsFast(BIT_CStream_t* bitC)
	{
	size_t const nbBytes = bitC->bitPos >> 3;
	assert(bitC->bitPos < sizeof(bitC->bitContainer) * 8);
	assert(bitC->ptr <= bitC->endPtr);
	MEM_writeLEST(bitC->ptr, bitC->bitContainer);
	bitC->ptr += nbBytes;
	bitC->bitPos &= 7;
	bitC->bitContainer >>= nbBytes*8;
	}

	/*! BIT_flushBits() :
	* assumption : bitContainer has not overflowed
	* safe version; check for buffer overflow, and prevents it.
	* note : does not signal buffer overflow.
	* overflow will be revealed later on using BIT_closeCStream() */
	MEM_STATIC void BIT_flushBits(BIT_CStream_t* bitC)
	{
	size_t const nbBytes = bitC->bitPos >> 3;
	assert(bitC->bitPos < sizeof(bitC->bitContainer) * 8);
	assert(bitC->ptr <= bitC->endPtr);
	MEM_writeLEST(bitC->ptr, bitC->bitContainer);
	bitC->ptr += nbBytes;
	if (bitC->ptr > bitC->endPtr) bitC->ptr = bitC->endPtr;
	bitC->bitPos &= 7;
	bitC->bitContainer >>= nbBytes*8;
	}

	/*! BIT_closeCStream() :
	* @return : size of CStream, in bytes,
	* or 0 if it could not fit into dstBuffer */
	MEM_STATIC size_t BIT_closeCStream(BIT_CStream_t* bitC)
	{
	BIT_addBitsFast(bitC, 1, 1); /* endMark */
	BIT_flushBits(bitC);
	if (bitC->ptr >= bitC->endPtr) return 0; /* overflow detected */
	return (bitC->ptr - bitC->startPtr) + (bitC->bitPos > 0);
	}


	/-*******************************************************
	* bitStream decoding
	**********************************************************/
	/*! BIT_initDStream() :
	* Initialize a BIT_DStream_t.
	* `bitD` : a pointer to an already allocated BIT_DStream_t structure.
	* `srcSize` must be the exact size of the bitStream, in bytes.
	* @return : size of stream (== srcSize), or an errorCode if a problem is detected
	*/
	MEM_STATIC size_t BIT_initDStream(BIT_DStream_t* bitD, const void* srcBuffer, size_t srcSize)
	{
	if (srcSize < 1) { memset(bitD, 0, sizeof(*bitD)); return ERROR(srcSize_wrong); }

	bitD->start = (const char*)srcBuffer;
	bitD->limitPtr = bitD->start + sizeof(bitD->bitContainer);

	if (srcSize >= sizeof(bitD->bitContainer)) { /* normal case */
	bitD->ptr = (const char*)srcBuffer + srcSize - sizeof(bitD->bitContainer);
	bitD->bitContainer = MEM_readLEST(bitD->ptr);
	{ BYTE const lastByte = ((const BYTE*)srcBuffer)[srcSize-1];
	bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0; /* ensures bitsConsumed is always set */
	if (lastByte == 0) return ERROR(GENERIC); /* endMark not present */ }
	} else {
	bitD->ptr = bitD->start;
	bitD->bitContainer = (const BYTE)(bitD->start);
	switch(srcSize)
	{
	case 7: bitD->bitContainer += (size_t)(((const BYTE)(srcBuffer))[6]) << (sizeof(bitD->bitContainer)8 - 16);
	/* fall-through */

	case 6: bitD->bitContainer += (size_t)(((const BYTE)(srcBuffer))[5]) << (sizeof(bitD->bitContainer)8 - 24);
	/* fall-through */

	case 5: bitD->bitContainer += (size_t)(((const BYTE)(srcBuffer))[4]) << (sizeof(bitD->bitContainer)8 - 32);
	/* fall-through */

	case 4: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[3]) << 24;
	/* fall-through */

	case 3: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[2]) << 16;
	/* fall-through */

	case 2: bitD->bitContainer += (size_t)(((const BYTE*)(srcBuffer))[1]) << 8;
	/* fall-through */

	default: break;
	}
	{ BYTE const lastByte = ((const BYTE*)srcBuffer)[srcSize-1];
	bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0;
	if (lastByte == 0) return ERROR(corruption_detected); /* endMark not present */
	}
	bitD->bitsConsumed += (U32)(sizeof(bitD->bitContainer) - srcSize)*8;
	}

	return srcSize;
	}

	MEM_STATIC size_t BIT_getUpperBits(size_t bitContainer, U32 const start)
	{
	return bitContainer >> start;
	}

	MEM_STATIC size_t BIT_getMiddleBits(size_t bitContainer, U32 const start, U32 const nbBits)
	{
	U32 const regMask = sizeof(bitContainer)*8 - 1;
	/* if start > regMask, bitstream is corrupted, and result is undefined */
	assert(nbBits < BIT_MASK_SIZE);
	return (bitContainer >> (start & regMask)) & BIT_mask[nbBits];
	}

	MEM_STATIC size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits)
	{
	assert(nbBits < BIT_MASK_SIZE);
	return bitContainer & BIT_mask[nbBits];
	}

	/*! BIT_lookBits() :
	* Provides next n bits from local register.
	* local register is not modified.
	* On 32-bits, maxNbBits==24.
	* On 64-bits, maxNbBits==56.
	* @return : value extracted */
	MEM_STATIC size_t BIT_lookBits(const BIT_DStream_t* bitD, U32 nbBits)
	{
	/* arbitrate between double-shift and shift+mask */
	#if 1
	/* if bitD->bitsConsumed + nbBits > sizeof(bitD->bitContainer)*8,
	* bitstream is likely corrupted, and result is undefined */
	return BIT_getMiddleBits(bitD->bitContainer, (sizeof(bitD->bitContainer)*8) - bitD->bitsConsumed - nbBits, nbBits);
	#else
	/* this code path is slower on my os-x laptop */
	U32 const regMask = sizeof(bitD->bitContainer)*8 - 1;
	return ((bitD->bitContainer << (bitD->bitsConsumed & regMask)) >> 1) >> ((regMask-nbBits) & regMask);
	#endif
	}

	/*! BIT_lookBitsFast() :
	* unsafe version; only works if nbBits >= 1 */
	MEM_STATIC size_t BIT_lookBitsFast(const BIT_DStream_t* bitD, U32 nbBits)
	{
	U32 const regMask = sizeof(bitD->bitContainer)*8 - 1;
	assert(nbBits >= 1);
	return (bitD->bitContainer << (bitD->bitsConsumed & regMask)) >> (((regMask+1)-nbBits) & regMask);
	}

	MEM_STATIC void BIT_skipBits(BIT_DStream_t* bitD, U32 nbBits)
	{
	bitD->bitsConsumed += nbBits;
	}

	/*! BIT_readBits() :
	* Read (consume) next n bits from local register and update.
	* Pay attention to not read more than nbBits contained into local register.
	* @return : extracted value. */
	MEM_STATIC size_t BIT_readBits(BIT_DStream_t* bitD, unsigned nbBits)
	{
	size_t const value = BIT_lookBits(bitD, nbBits);
	BIT_skipBits(bitD, nbBits);
	return value;
	}

	/*! BIT_readBitsFast() :
	* unsafe version; only works only if nbBits >= 1 */
	MEM_STATIC size_t BIT_readBitsFast(BIT_DStream_t* bitD, unsigned nbBits)
	{
	size_t const value = BIT_lookBitsFast(bitD, nbBits);
	assert(nbBits >= 1);
	BIT_skipBits(bitD, nbBits);
	return value;
	}

	/*! BIT_reloadDStreamFast() :
	* Similar to BIT_reloadDStream(), but with two differences:
	* 1. bitsConsumed <= sizeof(bitD->bitContainer)*8 must hold!
	* 2. Returns BIT_DStream_overflow when bitD->ptr < bitD->limitPtr, at this
	* point you must use BIT_reloadDStream() to reload.
	*/
	MEM_STATIC BIT_DStream_status BIT_reloadDStreamFast(BIT_DStream_t* bitD)
	{
	if (UNLIKELY(bitD->ptr < bitD->limitPtr))
	return BIT_DStream_overflow;
	assert(bitD->bitsConsumed <= sizeof(bitD->bitContainer)*8);
	bitD->ptr -= bitD->bitsConsumed >> 3;
	bitD->bitsConsumed &= 7;
	bitD->bitContainer = MEM_readLEST(bitD->ptr);
	return BIT_DStream_unfinished;
	}

	/*! BIT_reloadDStream() :
	* Refill `bitD` from buffer previously set in BIT_initDStream() .
	* This function is safe, it guarantees it will not read beyond src buffer.
	* @return : status of `BIT_DStream_t` internal register.
	* when status == BIT_DStream_unfinished, internal register is filled with at least 25 or 57 bits */
	MEM_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t* bitD)
	{
	if (bitD->bitsConsumed > (sizeof(bitD->bitContainer)8)) / overflow detected, like end of stream */
	return BIT_DStream_overflow;

	if (bitD->ptr >= bitD->limitPtr) {
	return BIT_reloadDStreamFast(bitD);
	}
	if (bitD->ptr == bitD->start) {
	if (bitD->bitsConsumed < sizeof(bitD->bitContainer)*8) return BIT_DStream_endOfBuffer;
	return BIT_DStream_completed;
	}
	/* start < ptr < limitPtr */
	{ U32 nbBytes = bitD->bitsConsumed >> 3;
	BIT_DStream_status result = BIT_DStream_unfinished;
	if (bitD->ptr - nbBytes < bitD->start) {
	nbBytes = (U32)(bitD->ptr - bitD->start); /* ptr > start */
	result = BIT_DStream_endOfBuffer;
	}
	bitD->ptr -= nbBytes;
	bitD->bitsConsumed -= nbBytes*8;
	bitD->bitContainer = MEM_readLEST(bitD->ptr); /* reminder : srcSize > sizeof(bitD->bitContainer), otherwise bitD->ptr == bitD->start */
	return result;
	}
	}

	/*! BIT_endOfDStream() :
	* @return : 1 if DStream has _exactly_ reached its end (all bits consumed).
	*/
	MEM_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t* DStream)
	{
	return ((DStream->ptr == DStream->start) && (DStream->bitsConsumed == sizeof(DStream->bitContainer)*8));
	}

	#if defined (__cplusplus)
	}
	#endif

	#endif /* BITSTREAM_H_MODULE */
	/** ended inlining bitstream.h **/


	/* *****************************************
	* Static allocation
	*******************************************/
	/* FSE buffer bounds */
	#define FSE_NCOUNTBOUND 512
	#define FSE_BLOCKBOUND(size) (size + (size>>7) + 4 /* fse states / + sizeof(size_t) / bitContainer */)
	#define FSE_COMPRESSBOUND(size) (FSE_NCOUNTBOUND + FSE_BLOCKBOUND(size)) /* Macro version, useful for static allocation */

	/* It is possible to statically allocate FSE CTable/DTable as a table of FSE_CTable/FSE_DTable using below macros */
	#define FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) (1 + (1<<(maxTableLog-1)) + ((maxSymbolValue+1)*2))
	#define FSE_DTABLE_SIZE_U32(maxTableLog) (1 + (1<<maxTableLog))

	/* or use the size to malloc() space directly. Pay attention to alignment restrictions though */
	#define FSE_CTABLE_SIZE(maxTableLog, maxSymbolValue) (FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) * sizeof(FSE_CTable))
	#define FSE_DTABLE_SIZE(maxTableLog) (FSE_DTABLE_SIZE_U32(maxTableLog) * sizeof(FSE_DTable))


	/* *****************************************
	* FSE advanced API
	***************************************** */

	unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus);
	/*< same as FSE_optimalTableLog(), which used `minus==2` /

	/* FSE_compress_wksp() :
	* Same as FSE_compress2(), but using an externally allocated scratch buffer (`workSpace`).
	* FSE_WKSP_SIZE_U32() provides the minimum size required for `workSpace` as a table of FSE_CTable.
	*/
	#define FSE_WKSP_SIZE_U32(maxTableLog, maxSymbolValue) ( FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) + ((maxTableLog > 12) ? (1 << (maxTableLog - 2)) : 1024) )
	size_t FSE_compress_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);

	size_t FSE_buildCTable_raw (FSE_CTable* ct, unsigned nbBits);
	/*< build a fake FSE_CTable, designed for a flat distribution, where each symbol uses nbBits /

	size_t FSE_buildCTable_rle (FSE_CTable* ct, unsigned char symbolValue);
	/*< build a fake FSE_CTable, designed to compress always the same symbolValue /

	/* FSE_buildCTable_wksp() :
	* Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
	* `wkspSize` must be >= `(1<<tableLog)`.
	*/
	size_t FSE_buildCTable_wksp(FSE_CTable* ct, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize);

	size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits);
	/*< build a fake FSE_DTable, designed to read a flat distribution where each symbol uses nbBits /

	size_t FSE_buildDTable_rle (FSE_DTable* dt, unsigned char symbolValue);
	/*< build a fake FSE_DTable, designed to always generate the same symbolValue /

	size_t FSE_decompress_wksp(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, FSE_DTable* workSpace, unsigned maxLog);
	/*< same as FSE_decompress(), using an externally allocated `workSpace` produced with `FSE_DTABLE_SIZE_U32(maxLog)` /

	typedef enum {
	FSE_repeat_none, /*< Cannot use the previous table /
	FSE_repeat_check, /*< Can use the previous table but it must be checked /
	FSE_repeat_valid /*< Can use the previous table and it is assumed to be valid /
	} FSE_repeat;

	/* *****************************************
	* FSE symbol compression API
	*******************************************/
	/*!
	This API consists of small unitary functions, which highly benefit from being inlined.
	Hence their body are included in next section.
	*/
	typedef struct {
	ptrdiff_t value;
	const void* stateTable;
	const void* symbolTT;
	unsigned stateLog;
	} FSE_CState_t;

	static void FSE_initCState(FSE_CState_t* CStatePtr, const FSE_CTable* ct);

	static void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* CStatePtr, unsigned symbol);

	static void FSE_flushCState(BIT_CStream_t* bitC, const FSE_CState_t* CStatePtr);

	/**<
	These functions are inner components of FSE_compress_usingCTable().
	They allow the creation of custom streams, mixing multiple tables and bit sources.

	A key property to keep in mind is that encoding and decoding are done in reverse direction.
	So the first symbol you will encode is the last you will decode, like a LIFO stack.

	You will need a few variables to track your CStream. They are :

	FSE_CTable ct; // Provided by FSE_buildCTable()
	BIT_CStream_t bitStream; // bitStream tracking structure
	FSE_CState_t state; // State tracking structure (can have several)


	The first thing to do is to init bitStream and state.
	size_t errorCode = BIT_initCStream(&bitStream, dstBuffer, maxDstSize);
	FSE_initCState(&state, ct);

	Note that BIT_initCStream() can produce an error code, so its result should be tested, using FSE_isError();
	You can then encode your input data, byte after byte.
	FSE_encodeSymbol() outputs a maximum of 'tableLog' bits at a time.
	Remember decoding will be done in reverse direction.
	FSE_encodeByte(&bitStream, &state, symbol);

	At any time, you can also add any bit sequence.
	Note : maximum allowed nbBits is 25, for compatibility with 32-bits decoders
	BIT_addBits(&bitStream, bitField, nbBits);

	The above methods don't commit data to memory, they just store it into local register, for speed.
	Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
	Writing data to memory is a manual operation, performed by the flushBits function.
	BIT_flushBits(&bitStream);

	Your last FSE encoding operation shall be to flush your last state value(s).
	FSE_flushState(&bitStream, &state);

	Finally, you must close the bitStream.
	The function returns the size of CStream in bytes.
	If data couldn't fit into dstBuffer, it will return a 0 ( == not compressible)
	If there is an error, it returns an errorCode (which can be tested using FSE_isError()).
	size_t size = BIT_closeCStream(&bitStream);
	*/


	/* *****************************************
	* FSE symbol decompression API
	*******************************************/
	typedef struct {
	size_t state;
	const void* table; /* precise table may vary, depending on U16 */
	} FSE_DState_t;


	static void FSE_initDState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD, const FSE_DTable* dt);

	static unsigned char FSE_decodeSymbol(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD);

	static unsigned FSE_endOfDState(const FSE_DState_t* DStatePtr);

	/**<
	Let's now decompose FSE_decompress_usingDTable() into its unitary components.
	You will decode FSE-encoded symbols from the bitStream,
	and also any other bitFields you put in, in reverse order.

	You will need a few variables to track your bitStream. They are :

	BIT_DStream_t DStream; // Stream context
	FSE_DState_t DState; // State context. Multiple ones are possible
	FSE_DTable* DTablePtr; // Decoding table, provided by FSE_buildDTable()

	The first thing to do is to init the bitStream.
	errorCode = BIT_initDStream(&DStream, srcBuffer, srcSize);

	You should then retrieve your initial state(s)
	(in reverse flushing order if you have several ones) :
	errorCode = FSE_initDState(&DState, &DStream, DTablePtr);

	You can then decode your data, symbol after symbol.
	For information the maximum number of bits read by FSE_decodeSymbol() is 'tableLog'.
	Keep in mind that symbols are decoded in reverse order, like a LIFO stack (last in, first out).
	unsigned char symbol = FSE_decodeSymbol(&DState, &DStream);

	You can retrieve any bitfield you eventually stored into the bitStream (in reverse order)
	Note : maximum allowed nbBits is 25, for 32-bits compatibility
	size_t bitField = BIT_readBits(&DStream, nbBits);

	All above operations only read from local register (which size depends on size_t).
	Refueling the register from memory is manually performed by the reload method.
	endSignal = FSE_reloadDStream(&DStream);

	BIT_reloadDStream() result tells if there is still some more data to read from DStream.
	BIT_DStream_unfinished : there is still some data left into the DStream.
	BIT_DStream_endOfBuffer : Dstream reached end of buffer. Its container may no longer be completely filled.
	BIT_DStream_completed : Dstream reached its exact end, corresponding in general to decompression completed.
	BIT_DStream_tooFar : Dstream went too far. Decompression result is corrupted.

	When reaching end of buffer (BIT_DStream_endOfBuffer), progress slowly, notably if you decode multiple symbols per loop,
	to properly detect the exact end of stream.
	After each decoded symbol, check if DStream is fully consumed using this simple test :
	BIT_reloadDStream(&DStream) >= BIT_DStream_completed

	When it's done, verify decompression is fully completed, by checking both DStream and the relevant states.
	Checking if DStream has reached its end is performed by :
	BIT_endOfDStream(&DStream);
	Check also the states. There might be some symbols left there, if some high probability ones (>50%) are possible.
	FSE_endOfDState(&DState);
	*/


	/* *****************************************
	* FSE unsafe API
	*******************************************/
	static unsigned char FSE_decodeSymbolFast(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD);
	/* faster, but works only if nbBits is always >= 1 (otherwise, result will be corrupted) */


	/* *****************************************
	* Implementation of inlined functions
	*******************************************/
	typedef struct {
	int deltaFindState;
	U32 deltaNbBits;
	} FSE_symbolCompressionTransform; /* total 8 bytes */

	MEM_STATIC void FSE_initCState(FSE_CState_t* statePtr, const FSE_CTable* ct)
	{
	const void* ptr = ct;
	const U16* u16ptr = (const U16*) ptr;
	const U32 tableLog = MEM_read16(ptr);
	statePtr->value = (ptrdiff_t)1<<tableLog;
	statePtr->stateTable = u16ptr+2;
	statePtr->symbolTT = ct + 1 + (tableLog ? (1<<(tableLog-1)) : 1);
	statePtr->stateLog = tableLog;
	}


	/*! FSE_initCState2() :
	* Same as FSE_initCState(), but the first symbol to include (which will be the last to be read)
	* uses the smallest state value possible, saving the cost of this symbol */
	MEM_STATIC void FSE_initCState2(FSE_CState_t* statePtr, const FSE_CTable* ct, U32 symbol)
	{
	FSE_initCState(statePtr, ct);
	{ const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol];
	const U16* stateTable = (const U16*)(statePtr->stateTable);
	U32 nbBitsOut = (U32)((symbolTT.deltaNbBits + (1<<15)) >> 16);
	statePtr->value = (nbBitsOut << 16) - symbolTT.deltaNbBits;
	statePtr->value = stateTable[(statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
	}
	}

	MEM_STATIC void FSE_encodeSymbol(BIT_CStream_t* bitC, FSE_CState_t* statePtr, unsigned symbol)
	{
	FSE_symbolCompressionTransform const symbolTT = ((const FSE_symbolCompressionTransform*)(statePtr->symbolTT))[symbol];
	const U16* const stateTable = (const U16*)(statePtr->stateTable);
	U32 const nbBitsOut = (U32)((statePtr->value + symbolTT.deltaNbBits) >> 16);
	BIT_addBits(bitC, statePtr->value, nbBitsOut);
	statePtr->value = stateTable[ (statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
	}

	MEM_STATIC void FSE_flushCState(BIT_CStream_t* bitC, const FSE_CState_t* statePtr)
	{
	BIT_addBits(bitC, statePtr->value, statePtr->stateLog);
	BIT_flushBits(bitC);
	}


	/* FSE_getMaxNbBits() :
	* Approximate maximum cost of a symbol, in bits.
	* Fractional get rounded up (i.e : a symbol with a normalized frequency of 3 gives the same result as a frequency of 2)
	* note 1 : assume symbolValue is valid (<= maxSymbolValue)
	* note 2 : if freq[symbolValue]==0, @return a fake cost of tableLog+1 bits */
	MEM_STATIC U32 FSE_getMaxNbBits(const void* symbolTTPtr, U32 symbolValue)
	{
	const FSE_symbolCompressionTransform* symbolTT = (const FSE_symbolCompressionTransform*) symbolTTPtr;
	return (symbolTT[symbolValue].deltaNbBits + ((1<<16)-1)) >> 16;
	}

	/* FSE_bitCost() :
	* Approximate symbol cost, as fractional value, using fixed-point format (accuracyLog fractional bits)
	* note 1 : assume symbolValue is valid (<= maxSymbolValue)
	* note 2 : if freq[symbolValue]==0, @return a fake cost of tableLog+1 bits */
	MEM_STATIC U32 FSE_bitCost(const void* symbolTTPtr, U32 tableLog, U32 symbolValue, U32 accuracyLog)
	{
	const FSE_symbolCompressionTransform* symbolTT = (const FSE_symbolCompressionTransform*) symbolTTPtr;
	U32 const minNbBits = symbolTT[symbolValue].deltaNbBits >> 16;
	U32 const threshold = (minNbBits+1) << 16;
	assert(tableLog < 16);
	assert(accuracyLog < 31-tableLog); /* ensure enough room for renormalization double shift */
	{ U32 const tableSize = 1 << tableLog;
	U32 const deltaFromThreshold = threshold - (symbolTT[symbolValue].deltaNbBits + tableSize);
	U32 const normalizedDeltaFromThreshold = (deltaFromThreshold << accuracyLog) >> tableLog; /* linear interpolation (very approximate) */
	U32 const bitMultiplier = 1 << accuracyLog;
	assert(symbolTT[symbolValue].deltaNbBits + tableSize <= threshold);
	assert(normalizedDeltaFromThreshold <= bitMultiplier);
	return (minNbBits+1)*bitMultiplier - normalizedDeltaFromThreshold;
	}
	}


	/* ====== Decompression ====== */

	typedef struct {
	U16 tableLog;
	U16 fastMode;
	} FSE_DTableHeader; /* sizeof U32 */

	typedef struct
	{
	unsigned short newState;
	unsigned char symbol;
	unsigned char nbBits;
	} FSE_decode_t; /* size == U32 */

	MEM_STATIC void FSE_initDState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD, const FSE_DTable* dt)
	{
	const void* ptr = dt;
	const FSE_DTableHeader* const DTableH = (const FSE_DTableHeader*)ptr;
	DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog);
	BIT_reloadDStream(bitD);
	DStatePtr->table = dt + 1;
	}

	MEM_STATIC BYTE FSE_peekSymbol(const FSE_DState_t* DStatePtr)
	{
	FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
	return DInfo.symbol;
	}

	MEM_STATIC void FSE_updateState(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD)
	{
	FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
	U32 const nbBits = DInfo.nbBits;
	size_t const lowBits = BIT_readBits(bitD, nbBits);
	DStatePtr->state = DInfo.newState + lowBits;
	}

	MEM_STATIC BYTE FSE_decodeSymbol(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD)
	{
	FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
	U32 const nbBits = DInfo.nbBits;
	BYTE const symbol = DInfo.symbol;
	size_t const lowBits = BIT_readBits(bitD, nbBits);

	DStatePtr->state = DInfo.newState + lowBits;
	return symbol;
	}

	/*! FSE_decodeSymbolFast() :
	unsafe, only works if no symbol has a probability > 50% */
	MEM_STATIC BYTE FSE_decodeSymbolFast(FSE_DState_t* DStatePtr, BIT_DStream_t* bitD)
	{
	FSE_decode_t const DInfo = ((const FSE_decode_t*)(DStatePtr->table))[DStatePtr->state];
	U32 const nbBits = DInfo.nbBits;
	BYTE const symbol = DInfo.symbol;
	size_t const lowBits = BIT_readBitsFast(bitD, nbBits);

	DStatePtr->state = DInfo.newState + lowBits;
	return symbol;
	}

	MEM_STATIC unsigned FSE_endOfDState(const FSE_DState_t* DStatePtr)
	{
	return DStatePtr->state == 0;
	}



	#ifndef FSE_COMMONDEFS_ONLY

	/* **************************************************************
	* Tuning parameters
	****************************************************************/
	/*!MEMORY_USAGE :
	* Memory usage formula : N->2^N Bytes (examples : 10 -> 1KB; 12 -> 4KB ; 16 -> 64KB; 20 -> 1MB; etc.)
	* Increasing memory usage improves compression ratio
	* Reduced memory usage can improve speed, due to cache effect
	* Recommended max value is 14, for 16KB, which nicely fits into Intel x86 L1 cache */
	#ifndef FSE_MAX_MEMORY_USAGE
	# define FSE_MAX_MEMORY_USAGE 14
	#endif
	#ifndef FSE_DEFAULT_MEMORY_USAGE
	# define FSE_DEFAULT_MEMORY_USAGE 13
	#endif

	/*!FSE_MAX_SYMBOL_VALUE :
	* Maximum symbol value authorized.
	* Required for proper stack allocation */
	#ifndef FSE_MAX_SYMBOL_VALUE
	# define FSE_MAX_SYMBOL_VALUE 255
	#endif

	/* **************************************************************
	* template functions type & suffix
	****************************************************************/
	#define FSE_FUNCTION_TYPE BYTE
	#define FSE_FUNCTION_EXTENSION
	#define FSE_DECODE_TYPE FSE_decode_t


	#endif /* !FSE_COMMONDEFS_ONLY */


	/* ***************************************************************
	* Constants
	*****************************************************************/
	#define FSE_MAX_TABLELOG (FSE_MAX_MEMORY_USAGE-2)
	#define FSE_MAX_TABLESIZE (1U<<FSE_MAX_TABLELOG)
	#define FSE_MAXTABLESIZE_MASK (FSE_MAX_TABLESIZE-1)
	#define FSE_DEFAULT_TABLELOG (FSE_DEFAULT_MEMORY_USAGE-2)
	#define FSE_MIN_TABLELOG 5

	#define FSE_TABLELOG_ABSOLUTE_MAX 15
	#if FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX
	# error "FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX is not supported"
	#endif

	#define FSE_TABLESTEP(tableSize) ((tableSize>>1) + (tableSize>>3) + 3)


	#endif /* FSE_STATIC_LINKING_ONLY */


	#if defined (__cplusplus)
	}
	#endif
	/** ended inlining fse.h **/
	#define HUF_STATIC_LINKING_ONLY /* HUF_TABLELOG_ABSOLUTEMAX */
	/** start inlining huf.h **/
	/* ******************************************************************
	* huff0 huffman codec,
	* part of Finite State Entropy library
	* Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */

	#if defined (__cplusplus)
	extern "C" {
	#endif

	#ifndef HUF_H_298734234
	#define HUF_H_298734234

	/* * Dependencies * */
	#include <stddef.h> /* size_t */


	/* * library symbols visibility * */
	/* Note : when linking with -fvisibility=hidden on gcc, or by default on Visual,
	* HUF symbols remain "private" (internal symbols for library only).
	* Set macro FSE_DLL_EXPORT to 1 if you want HUF symbols visible on DLL interface */
	#if defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1) && defined(__GNUC__) && (__GNUC__ >= 4)
	# define HUF_PUBLIC_API __attribute__ ((visibility ("default")))
	#elif defined(FSE_DLL_EXPORT) && (FSE_DLL_EXPORT==1) /* Visual expected */
	# define HUF_PUBLIC_API __declspec(dllexport)
	#elif defined(FSE_DLL_IMPORT) && (FSE_DLL_IMPORT==1)
	# define HUF_PUBLIC_API __declspec(dllimport) /* not required, just to generate faster code (saves a function pointer load from IAT and an indirect jump) */
	#else
	# define HUF_PUBLIC_API
	#endif


	/* ========================== */
	/* * simple functions * */
	/* ========================== */

	/** HUF_compress() :
	* Compress content from buffer 'src', of size 'srcSize', into buffer 'dst'.
	* 'dst' buffer must be already allocated.
	* Compression runs faster if `dstCapacity` >= HUF_compressBound(srcSize).
	* `srcSize` must be <= `HUF_BLOCKSIZE_MAX` == 128 KB.
	* @return : size of compressed data (<= `dstCapacity`).
	* Special values : if return == 0, srcData is not compressible => Nothing is stored within dst !!!
	* if HUF_isError(return), compression failed (more details using HUF_getErrorName())
	*/
	HUF_PUBLIC_API size_t HUF_compress(void* dst, size_t dstCapacity,
	const void* src, size_t srcSize);

	/** HUF_decompress() :
	* Decompress HUF data from buffer 'cSrc', of size 'cSrcSize',
	* into already allocated buffer 'dst', of minimum size 'dstSize'.
	* `originalSize` : must be the *exact* size of original (uncompressed) data.
	* Note : in contrast with FSE, HUF_decompress can regenerate
	* RLE (cSrcSize==1) and uncompressed (cSrcSize==dstSize) data,
	* because it knows size to regenerate (originalSize).
	* @return : size of regenerated data (== originalSize),
	* or an error code, which can be tested using HUF_isError()
	*/
	HUF_PUBLIC_API size_t HUF_decompress(void* dst, size_t originalSize,
	const void* cSrc, size_t cSrcSize);


	/* * Tool functions * */
	#define HUF_BLOCKSIZE_MAX (128 * 1024) /*< maximum input size for a single block compressed with HUF_compress /
	HUF_PUBLIC_API size_t HUF_compressBound(size_t size); /*< maximum compressed size (worst case) /

	/* Error Management */
	HUF_PUBLIC_API unsigned HUF_isError(size_t code); /*< tells if a return value is an error code /
	HUF_PUBLIC_API const char* HUF_getErrorName(size_t code); /*< provides error code string (useful for debugging) /


	/* * Advanced function * */

	/** HUF_compress2() :
	* Same as HUF_compress(), but offers control over `maxSymbolValue` and `tableLog`.
	* `maxSymbolValue` must be <= HUF_SYMBOLVALUE_MAX .
	* `tableLog` must be `<= HUF_TABLELOG_MAX` . */
	HUF_PUBLIC_API size_t HUF_compress2 (void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	unsigned maxSymbolValue, unsigned tableLog);

	/** HUF_compress4X_wksp() :
	* Same as HUF_compress2(), but uses externally allocated `workSpace`.
	* `workspace` must have minimum alignment of 4, and be at least as large as HUF_WORKSPACE_SIZE */
	#define HUF_WORKSPACE_SIZE ((6 << 10) + 256)
	#define HUF_WORKSPACE_SIZE_U32 (HUF_WORKSPACE_SIZE / sizeof(U32))
	HUF_PUBLIC_API size_t HUF_compress4X_wksp (void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	unsigned maxSymbolValue, unsigned tableLog,
	void* workSpace, size_t wkspSize);

	#endif /* HUF_H_298734234 */

	/* ******************************************************************
	* WARNING !!
	* The following section contains advanced and experimental definitions
	* which shall never be used in the context of a dynamic library,
	* because they are not guaranteed to remain stable in the future.
	* Only consider them in association with static linking.
	* *****************************************************************/
	#if defined(HUF_STATIC_LINKING_ONLY) && !defined(HUF_H_HUF_STATIC_LINKING_ONLY)
	#define HUF_H_HUF_STATIC_LINKING_ONLY

	/* * Dependencies * */
	/** skipping file: mem.h **/


	/* * Constants * */
	#define HUF_TABLELOG_MAX 12 /* max runtime value of tableLog (due to static allocation); can be modified up to HUF_ABSOLUTEMAX_TABLELOG */
	#define HUF_TABLELOG_DEFAULT 11 /* default tableLog value when none specified */
	#define HUF_SYMBOLVALUE_MAX 255

	#define HUF_TABLELOG_ABSOLUTEMAX 15 /* absolute limit of HUF_MAX_TABLELOG. Beyond that value, code does not work */
	#if (HUF_TABLELOG_MAX > HUF_TABLELOG_ABSOLUTEMAX)
	# error "HUF_TABLELOG_MAX is too large !"
	#endif


	/* ****************************************
	* Static allocation
	******************************************/
	/* HUF buffer bounds */
	#define HUF_CTABLEBOUND 129
	#define HUF_BLOCKBOUND(size) (size + (size>>8) + 8) /* only true when incompressible is pre-filtered with fast heuristic */
	#define HUF_COMPRESSBOUND(size) (HUF_CTABLEBOUND + HUF_BLOCKBOUND(size)) /* Macro version, useful for static allocation */

	/* static allocation of HUF's Compression Table */
	#define HUF_CTABLE_SIZE_U32(maxSymbolValue) ((maxSymbolValue)+1) /* Use tables of U32, for proper alignment */
	#define HUF_CTABLE_SIZE(maxSymbolValue) (HUF_CTABLE_SIZE_U32(maxSymbolValue) * sizeof(U32))
	#define HUF_CREATE_STATIC_CTABLE(name, maxSymbolValue) \
	U32 name##hb[HUF_CTABLE_SIZE_U32(maxSymbolValue)]; \
	void* name##hv = &(name##hb); \
	HUF_CElt* name = (HUF_CElt)(name##hv) / no final ; */

	/* static allocation of HUF's DTable */
	typedef U32 HUF_DTable;
	#define HUF_DTABLE_SIZE(maxTableLog) (1 + (1<<(maxTableLog)))
	#define HUF_CREATE_STATIC_DTABLEX1(DTable, maxTableLog) \
	HUF_DTable DTable[HUF_DTABLE_SIZE((maxTableLog)-1)] = { ((U32)((maxTableLog)-1) * 0x01000001) }
	#define HUF_CREATE_STATIC_DTABLEX2(DTable, maxTableLog) \
	HUF_DTable DTable[HUF_DTABLE_SIZE(maxTableLog)] = { ((U32)(maxTableLog) * 0x01000001) }


	/* ****************************************
	* Advanced decompression functions
	******************************************/
	size_t HUF_decompress4X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /*< single-symbol decoder /
	#ifndef HUF_FORCE_DECOMPRESS_X1
	size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /*< double-symbols decoder /
	#endif

	size_t HUF_decompress4X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /*< decodes RLE and uncompressed /
	size_t HUF_decompress4X_hufOnly(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /*< considers RLE and uncompressed as errors /
	size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /*< considers RLE and uncompressed as errors /
	size_t HUF_decompress4X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /*< single-symbol decoder /
	size_t HUF_decompress4X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /*< single-symbol decoder /
	#ifndef HUF_FORCE_DECOMPRESS_X1
	size_t HUF_decompress4X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /*< double-symbols decoder /
	size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /*< double-symbols decoder /
	#endif


	/* ****************************************
	* HUF detailed API
	* ****************************************/

	/*! HUF_compress() does the following:
	* 1. count symbol occurrence from source[] into table count[] using FSE_count() (exposed within "fse.h")
	* 2. (optional) refine tableLog using HUF_optimalTableLog()
	* 3. build Huffman table from count using HUF_buildCTable()
	* 4. save Huffman table to memory buffer using HUF_writeCTable()
	* 5. encode the data stream using HUF_compress4X_usingCTable()
	*
	* The following API allows targeting specific sub-functions for advanced tasks.
	* For example, it's possible to compress several blocks using the same 'CTable',
	* or to save and regenerate 'CTable' using external methods.
	*/
	unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
	typedef struct HUF_CElt_s HUF_CElt; /* incomplete type */
	size_t HUF_buildCTable (HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue, unsigned maxNbBits); /* @return : maxNbBits; CTable and count can overlap. In which case, CTable will overwrite count content */
	size_t HUF_writeCTable (void* dst, size_t maxDstSize, const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog);
	size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
	size_t HUF_estimateCompressedSize(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);
	int HUF_validateCTable(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue);

	typedef enum {
	HUF_repeat_none, /*< Cannot use the previous table /
	HUF_repeat_check, /*< Can use the previous table but it must be checked. Note : The previous table must have been constructed by HUF_compress{1, 4}X_repeat /
	HUF_repeat_valid /*< Can use the previous table and it is assumed to be valid /
	} HUF_repeat;
	/** HUF_compress4X_repeat() :
	* Same as HUF_compress4X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
	* If it uses hufTable it does not modify hufTable or repeat.
	* If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
	* If preferRepeat then the old table will always be used if valid. */
	size_t HUF_compress4X_repeat(void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	unsigned maxSymbolValue, unsigned tableLog,
	void* workSpace, size_t wkspSize, /*< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE /
	HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2);

	/** HUF_buildCTable_wksp() :
	* Same as HUF_buildCTable(), but using externally allocated scratch buffer.
	* `workSpace` must be aligned on 4-bytes boundaries, and its size must be >= HUF_CTABLE_WORKSPACE_SIZE.
	*/
	#define HUF_CTABLE_WORKSPACE_SIZE_U32 (2*HUF_SYMBOLVALUE_MAX +1 +1)
	#define HUF_CTABLE_WORKSPACE_SIZE (HUF_CTABLE_WORKSPACE_SIZE_U32 * sizeof(unsigned))
	size_t HUF_buildCTable_wksp (HUF_CElt* tree,
	const unsigned* count, U32 maxSymbolValue, U32 maxNbBits,
	void* workSpace, size_t wkspSize);

	/*! HUF_readStats() :
	* Read compact Huffman tree, saved by HUF_writeCTable().
	* `huffWeight` is destination buffer.
	* @return : size read from `src` , or an error Code .
	* Note : Needed by HUF_readCTable() and HUF_readDTableXn() . */
	size_t HUF_readStats(BYTE* huffWeight, size_t hwSize,
	U32* rankStats, U32* nbSymbolsPtr, U32* tableLogPtr,
	const void* src, size_t srcSize);

	/** HUF_readCTable() :
	* Loading a CTable saved with HUF_writeCTable() */
	size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize, unsigned *hasZeroWeights);

	/** HUF_getNbBits() :
	* Read nbBits from CTable symbolTable, for symbol `symbolValue` presumed <= HUF_SYMBOLVALUE_MAX
	* Note 1 : is not inlined, as HUF_CElt definition is private
	* Note 2 : const void* used, so that it can provide a statically allocated table as argument (which uses type U32) */
	U32 HUF_getNbBits(const void* symbolTable, U32 symbolValue);

	/*
	* HUF_decompress() does the following:
	* 1. select the decompression algorithm (X1, X2) based on pre-computed heuristics
	* 2. build Huffman table from save, using HUF_readDTableX?()
	* 3. decode 1 or 4 segments in parallel using HUF_decompress?X?_usingDTable()
	*/

	/** HUF_selectDecoder() :
	* Tells which decoder is likely to decode faster,
	* based on a set of pre-computed metrics.
	* @return : 0==HUF_decompress4X1, 1==HUF_decompress4X2 .
	* Assumption : 0 < dstSize <= 128 KB */
	U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize);

	/**
	* The minimum workspace size for the `workSpace` used in
	* HUF_readDTableX1_wksp() and HUF_readDTableX2_wksp().
	*
	* The space used depends on HUF_TABLELOG_MAX, ranging from ~1500 bytes when
	* HUF_TABLE_LOG_MAX=12 to ~1850 bytes when HUF_TABLE_LOG_MAX=15.
	* Buffer overflow errors may potentially occur if code modifications result in
	* a required workspace size greater than that specified in the following
	* macro.
	*/
	#define HUF_DECOMPRESS_WORKSPACE_SIZE (2 << 10)
	#define HUF_DECOMPRESS_WORKSPACE_SIZE_U32 (HUF_DECOMPRESS_WORKSPACE_SIZE / sizeof(U32))

	#ifndef HUF_FORCE_DECOMPRESS_X2
	size_t HUF_readDTableX1 (HUF_DTable* DTable, const void* src, size_t srcSize);
	size_t HUF_readDTableX1_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
	#endif
	#ifndef HUF_FORCE_DECOMPRESS_X1
	size_t HUF_readDTableX2 (HUF_DTable* DTable, const void* src, size_t srcSize);
	size_t HUF_readDTableX2_wksp (HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize);
	#endif

	size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
	#ifndef HUF_FORCE_DECOMPRESS_X2
	size_t HUF_decompress4X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
	#endif
	#ifndef HUF_FORCE_DECOMPRESS_X1
	size_t HUF_decompress4X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
	#endif


	/* ====================== */
	/* single stream variants */
	/* ====================== */

	size_t HUF_compress1X (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog);
	size_t HUF_compress1X_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize); /*< `workSpace` must be a table of at least HUF_WORKSPACE_SIZE_U32 unsigned /
	size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable);
	/** HUF_compress1X_repeat() :
	* Same as HUF_compress1X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
	* If it uses hufTable it does not modify hufTable or repeat.
	* If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
	* If preferRepeat then the old table will always be used if valid. */
	size_t HUF_compress1X_repeat(void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	unsigned maxSymbolValue, unsigned tableLog,
	void* workSpace, size_t wkspSize, /*< `workSpace` must be aligned on 4-bytes boundaries, `wkspSize` must be >= HUF_WORKSPACE_SIZE /
	HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2);

	size_t HUF_decompress1X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /* single-symbol decoder */
	#ifndef HUF_FORCE_DECOMPRESS_X1
	size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /* double-symbol decoder */
	#endif

	size_t HUF_decompress1X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize);
	size_t HUF_decompress1X_DCtx_wksp (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize);
	#ifndef HUF_FORCE_DECOMPRESS_X2
	size_t HUF_decompress1X1_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /*< single-symbol decoder /
	size_t HUF_decompress1X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /*< single-symbol decoder /
	#endif
	#ifndef HUF_FORCE_DECOMPRESS_X1
	size_t HUF_decompress1X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize); /*< double-symbols decoder /
	size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize); /*< double-symbols decoder /
	#endif

	size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable); /*< automatic selection of sing or double symbol decoder, based on DTable /
	#ifndef HUF_FORCE_DECOMPRESS_X2
	size_t HUF_decompress1X1_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
	#endif
	#ifndef HUF_FORCE_DECOMPRESS_X1
	size_t HUF_decompress1X2_usingDTable(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable);
	#endif

	/* BMI2 variants.
	* If the CPU has BMI2 support, pass bmi2=1, otherwise pass bmi2=0.
	*/
	size_t HUF_decompress1X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
	#ifndef HUF_FORCE_DECOMPRESS_X2
	size_t HUF_decompress1X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2);
	#endif
	size_t HUF_decompress4X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2);
	size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2);

	#endif /* HUF_STATIC_LINKING_ONLY */

	#if defined (__cplusplus)
	}
	#endif
	/** ended inlining huf.h **/


	/=== Version ===/
	unsigned FSE_versionNumber(void) { return FSE_VERSION_NUMBER; }


	/=== Error Management ===/
	unsigned FSE_isError(size_t code) { return ERR_isError(code); }
	const char* FSE_getErrorName(size_t code) { return ERR_getErrorName(code); }

	unsigned HUF_isError(size_t code) { return ERR_isError(code); }
	const char* HUF_getErrorName(size_t code) { return ERR_getErrorName(code); }


	/-*************************************************************
	* FSE NCount encoding-decoding
	****************************************************************/
	size_t FSE_readNCount (short* normalizedCounter, unsigned* maxSVPtr, unsigned* tableLogPtr,
	const void* headerBuffer, size_t hbSize)
	{
	const BYTE* const istart = (const BYTE*) headerBuffer;
	const BYTE* const iend = istart + hbSize;
	const BYTE* ip = istart;
	int nbBits;
	int remaining;
	int threshold;
	U32 bitStream;
	int bitCount;
	unsigned charnum = 0;
	int previous0 = 0;

	if (hbSize < 4) {
	/* This function only works when hbSize >= 4 */
	char buffer[4];
	memset(buffer, 0, sizeof(buffer));
	memcpy(buffer, headerBuffer, hbSize);
	{ size_t const countSize = FSE_readNCount(normalizedCounter, maxSVPtr, tableLogPtr,
	buffer, sizeof(buffer));
	if (FSE_isError(countSize)) return countSize;
	if (countSize > hbSize) return ERROR(corruption_detected);
	return countSize;
	} }
	assert(hbSize >= 4);

	/* init */
	memset(normalizedCounter, 0, (maxSVPtr+1) sizeof(normalizedCounter[0])); /* all symbols not present in NCount have a frequency of 0 */
	bitStream = MEM_readLE32(ip);
	nbBits = (bitStream & 0xF) + FSE_MIN_TABLELOG; /* extract tableLog */
	if (nbBits > FSE_TABLELOG_ABSOLUTE_MAX) return ERROR(tableLog_tooLarge);
	bitStream >>= 4;
	bitCount = 4;
	*tableLogPtr = nbBits;
	remaining = (1<<nbBits)+1;
	threshold = 1<<nbBits;
	nbBits++;

	while ((remaining>1) & (charnum<=*maxSVPtr)) {
	if (previous0) {
	unsigned n0 = charnum;
	while ((bitStream & 0xFFFF) == 0xFFFF) {
	n0 += 24;
	if (ip < iend-5) {
	ip += 2;
	bitStream = MEM_readLE32(ip) >> bitCount;
	} else {
	bitStream >>= 16;
	bitCount += 16;
	} }
	while ((bitStream & 3) == 3) {
	n0 += 3;
	bitStream >>= 2;
	bitCount += 2;
	}
	n0 += bitStream & 3;
	bitCount += 2;
	if (n0 > *maxSVPtr) return ERROR(maxSymbolValue_tooSmall);
	while (charnum < n0) normalizedCounter[charnum++] = 0;
	if ((ip <= iend-7) \|\| (ip + (bitCount>>3) <= iend-4)) {
	assert((bitCount >> 3) <= 3); /* For first condition to work */
	ip += bitCount>>3;
	bitCount &= 7;
	bitStream = MEM_readLE32(ip) >> bitCount;
	} else {
	bitStream >>= 2;
	} }
	{ int const max = (2*threshold-1) - remaining;
	int count;

	if ((bitStream & (threshold-1)) < (U32)max) {
	count = bitStream & (threshold-1);
	bitCount += nbBits-1;
	} else {
	count = bitStream & (2*threshold-1);
	if (count >= threshold) count -= max;
	bitCount += nbBits;
	}

	count--; /* extra accuracy */
	remaining -= count < 0 ? -count : count; /* -1 means +1 */
	normalizedCounter[charnum++] = (short)count;
	previous0 = !count;
	while (remaining < threshold) {
	nbBits--;
	threshold >>= 1;
	}

	if ((ip <= iend-7) \|\| (ip + (bitCount>>3) <= iend-4)) {
	ip += bitCount>>3;
	bitCount &= 7;
	} else {
	bitCount -= (int)(8 * (iend - 4 - ip));
	ip = iend - 4;
	}
	bitStream = MEM_readLE32(ip) >> (bitCount & 31);
	} } /* while ((remaining>1) & (charnum<=maxSVPtr)) /
	if (remaining != 1) return ERROR(corruption_detected);
	if (bitCount > 32) return ERROR(corruption_detected);
	*maxSVPtr = charnum-1;

	ip += (bitCount+7)>>3;
	return ip-istart;
	}


	/*! HUF_readStats() :
	Read compact Huffman tree, saved by HUF_writeCTable().
	`huffWeight` is destination buffer.
	`rankStats` is assumed to be a table of at least HUF_TABLELOG_MAX U32.
	@return : size read from `src` , or an error Code .
	Note : Needed by HUF_readCTable() and HUF_readDTableX?() .
	*/
	size_t HUF_readStats(BYTE* huffWeight, size_t hwSize, U32* rankStats,
	U32* nbSymbolsPtr, U32* tableLogPtr,
	const void* src, size_t srcSize)
	{
	U32 weightTotal;
	const BYTE* ip = (const BYTE*) src;
	size_t iSize;
	size_t oSize;

	if (!srcSize) return ERROR(srcSize_wrong);
	iSize = ip[0];
	/* memset(huffWeight, 0, hwSize); // is not necessary, even though some analyzer complain ... */

	if (iSize >= 128) { /* special header */
	oSize = iSize - 127;
	iSize = ((oSize+1)/2);
	if (iSize+1 > srcSize) return ERROR(srcSize_wrong);
	if (oSize >= hwSize) return ERROR(corruption_detected);
	ip += 1;
	{ U32 n;
	for (n=0; n<oSize; n+=2) {
	huffWeight[n] = ip[n/2] >> 4;
	huffWeight[n+1] = ip[n/2] & 15;
	} } }
	else { /* header compressed with FSE (normal case) */
	FSE_DTable fseWorkspace[FSE_DTABLE_SIZE_U32(6)]; /* 6 is max possible tableLog for HUF header (maybe even 5, to be tested) */
	if (iSize+1 > srcSize) return ERROR(srcSize_wrong);
	oSize = FSE_decompress_wksp(huffWeight, hwSize-1, ip+1, iSize, fseWorkspace, 6); /* max (hwSize-1) values decoded, as last one is implied */
	if (FSE_isError(oSize)) return oSize;
	}

	/* collect weight stats */
	memset(rankStats, 0, (HUF_TABLELOG_MAX + 1) * sizeof(U32));
	weightTotal = 0;
	{ U32 n; for (n=0; n<oSize; n++) {
	if (huffWeight[n] >= HUF_TABLELOG_MAX) return ERROR(corruption_detected);
	rankStats[huffWeight[n]]++;
	weightTotal += (1 << huffWeight[n]) >> 1;
	} }
	if (weightTotal == 0) return ERROR(corruption_detected);

	/* get last non-null symbol weight (implied, total must be 2^n) */
	{ U32 const tableLog = BIT_highbit32(weightTotal) + 1;
	if (tableLog > HUF_TABLELOG_MAX) return ERROR(corruption_detected);
	*tableLogPtr = tableLog;
	/* determine last weight */
	{ U32 const total = 1 << tableLog;
	U32 const rest = total - weightTotal;
	U32 const verif = 1 << BIT_highbit32(rest);
	U32 const lastWeight = BIT_highbit32(rest) + 1;
	if (verif != rest) return ERROR(corruption_detected); /* last value must be a clean power of 2 */
	huffWeight[oSize] = (BYTE)lastWeight;
	rankStats[lastWeight]++;
	} }

	/* check tree construction validity */
	if ((rankStats[1] < 2) \|\| (rankStats[1] & 1)) return ERROR(corruption_detected); /* by construction : at least 2 elts of rank 1, must be even */

	/* results */
	*nbSymbolsPtr = (U32)(oSize+1);
	return iSize+1;
	}
	/** ended inlining common/entropy_common.c **/
	/** start inlining common/error_private.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/* The purpose of this file is to have a single list of error strings embedded in binary */

	/** skipping file: error_private.h **/

	const char* ERR_getErrorString(ERR_enum code)
	{
	#ifdef ZSTD_STRIP_ERROR_STRINGS
	(void)code;
	return "Error strings stripped";
	#else
	static const char* const notErrorCode = "Unspecified error code";
	switch( code )
	{
	case PREFIX(no_error): return "No error detected";
	case PREFIX(GENERIC): return "Error (generic)";
	case PREFIX(prefix_unknown): return "Unknown frame descriptor";
	case PREFIX(version_unsupported): return "Version not supported";
	case PREFIX(frameParameter_unsupported): return "Unsupported frame parameter";
	case PREFIX(frameParameter_windowTooLarge): return "Frame requires too much memory for decoding";
	case PREFIX(corruption_detected): return "Corrupted block detected";
	case PREFIX(checksum_wrong): return "Restored data doesn't match checksum";
	case PREFIX(parameter_unsupported): return "Unsupported parameter";
	case PREFIX(parameter_outOfBound): return "Parameter is out of bound";
	case PREFIX(init_missing): return "Context should be init first";
	case PREFIX(memory_allocation): return "Allocation error : not enough memory";
	case PREFIX(workSpace_tooSmall): return "workSpace buffer is not large enough";
	case PREFIX(stage_wrong): return "Operation not authorized at current processing stage";
	case PREFIX(tableLog_tooLarge): return "tableLog requires too much memory : unsupported";
	case PREFIX(maxSymbolValue_tooLarge): return "Unsupported max Symbol Value : too large";
	case PREFIX(maxSymbolValue_tooSmall): return "Specified maxSymbolValue is too small";
	case PREFIX(dictionary_corrupted): return "Dictionary is corrupted";
	case PREFIX(dictionary_wrong): return "Dictionary mismatch";
	case PREFIX(dictionaryCreation_failed): return "Cannot create Dictionary from provided samples";
	case PREFIX(dstSize_tooSmall): return "Destination buffer is too small";
	case PREFIX(srcSize_wrong): return "Src size is incorrect";
	case PREFIX(dstBuffer_null): return "Operation on NULL destination buffer";
	/* following error codes are not stable and may be removed or changed in a future version */
	case PREFIX(frameIndex_tooLarge): return "Frame index is too large";
	case PREFIX(seekableIO): return "An I/O error occurred when reading/seeking";
	case PREFIX(dstBuffer_wrong): return "Destination buffer is wrong";
	case PREFIX(maxCode):
	default: return notErrorCode;
	}
	#endif
	}
	/** ended inlining common/error_private.c **/
	/** start inlining common/fse_decompress.c **/
	/* ******************************************************************
	* FSE : Finite State Entropy decoder
	* Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
	* - Public forum : https://groups.google.com/forum/#!forum/lz4c
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */


	/* **************************************************************
	* Includes
	****************************************************************/
	#include <stdlib.h> /* malloc, free, qsort */
	#include <string.h> /* memcpy, memset */
	/** skipping file: bitstream.h **/
	/** skipping file: compiler.h **/
	#define FSE_STATIC_LINKING_ONLY
	/** skipping file: fse.h **/
	/** skipping file: error_private.h **/


	/* **************************************************************
	* Error Management
	****************************************************************/
	#define FSE_isError ERR_isError
	#define FSE_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c) /* use only after variable declarations */


	/* **************************************************************
	* Templates
	****************************************************************/
	/*
	designed to be included
	for type-specific functions (template emulation in C)
	Objective is to write these functions only once, for improved maintenance
	*/

	/* safety checks */
	#ifndef FSE_FUNCTION_EXTENSION
	# error "FSE_FUNCTION_EXTENSION must be defined"
	#endif
	#ifndef FSE_FUNCTION_TYPE
	# error "FSE_FUNCTION_TYPE must be defined"
	#endif

	/* Function names */
	#define FSE_CAT(X,Y) X##Y
	#define FSE_FUNCTION_NAME(X,Y) FSE_CAT(X,Y)
	#define FSE_TYPE_NAME(X,Y) FSE_CAT(X,Y)


	/* Function templates */
	FSE_DTable* FSE_createDTable (unsigned tableLog)
	{
	if (tableLog > FSE_TABLELOG_ABSOLUTE_MAX) tableLog = FSE_TABLELOG_ABSOLUTE_MAX;
	return (FSE_DTable)malloc( FSE_DTABLE_SIZE_U32(tableLog) sizeof (U32) );
	}

	void FSE_freeDTable (FSE_DTable* dt)
	{
	free(dt);
	}

	size_t FSE_buildDTable(FSE_DTable* dt, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog)
	{
	void* const tdPtr = dt+1; /* because dt is unsigned, 32-bits aligned on 32-bits /
	FSE_DECODE_TYPE* const tableDecode = (FSE_DECODE_TYPE*) (tdPtr);
	U16 symbolNext[FSE_MAX_SYMBOL_VALUE+1];

	U32 const maxSV1 = maxSymbolValue + 1;
	U32 const tableSize = 1 << tableLog;
	U32 highThreshold = tableSize-1;

	/* Sanity Checks */
	if (maxSymbolValue > FSE_MAX_SYMBOL_VALUE) return ERROR(maxSymbolValue_tooLarge);
	if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge);

	/* Init, lay down lowprob symbols */
	{ FSE_DTableHeader DTableH;
	DTableH.tableLog = (U16)tableLog;
	DTableH.fastMode = 1;
	{ S16 const largeLimit= (S16)(1 << (tableLog-1));
	U32 s;
	for (s=0; s<maxSV1; s++) {
	if (normalizedCounter[s]==-1) {
	tableDecode[highThreshold--].symbol = (FSE_FUNCTION_TYPE)s;
	symbolNext[s] = 1;
	} else {
	if (normalizedCounter[s] >= largeLimit) DTableH.fastMode=0;
	symbolNext[s] = normalizedCounter[s];
	} } }
	memcpy(dt, &DTableH, sizeof(DTableH));
	}

	/* Spread symbols */
	{ U32 const tableMask = tableSize-1;
	U32 const step = FSE_TABLESTEP(tableSize);
	U32 s, position = 0;
	for (s=0; s<maxSV1; s++) {
	int i;
	for (i=0; i<normalizedCounter[s]; i++) {
	tableDecode[position].symbol = (FSE_FUNCTION_TYPE)s;
	position = (position + step) & tableMask;
	while (position > highThreshold) position = (position + step) & tableMask; /* lowprob area */
	} }
	if (position!=0) return ERROR(GENERIC); /* position must reach all cells once, otherwise normalizedCounter is incorrect */
	}

	/* Build Decoding table */
	{ U32 u;
	for (u=0; u<tableSize; u++) {
	FSE_FUNCTION_TYPE const symbol = (FSE_FUNCTION_TYPE)(tableDecode[u].symbol);
	U32 const nextState = symbolNext[symbol]++;
	tableDecode[u].nbBits = (BYTE) (tableLog - BIT_highbit32(nextState) );
	tableDecode[u].newState = (U16) ( (nextState << tableDecode[u].nbBits) - tableSize);
	} }

	return 0;
	}


	#ifndef FSE_COMMONDEFS_ONLY

	/-******************************************************
	* Decompression (Byte symbols)
	*********************************************************/
	size_t FSE_buildDTable_rle (FSE_DTable* dt, BYTE symbolValue)
	{
	void* ptr = dt;
	FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr;
	void* dPtr = dt + 1;
	FSE_decode_t* const cell = (FSE_decode_t*)dPtr;

	DTableH->tableLog = 0;
	DTableH->fastMode = 0;

	cell->newState = 0;
	cell->symbol = symbolValue;
	cell->nbBits = 0;

	return 0;
	}


	size_t FSE_buildDTable_raw (FSE_DTable* dt, unsigned nbBits)
	{
	void* ptr = dt;
	FSE_DTableHeader* const DTableH = (FSE_DTableHeader*)ptr;
	void* dPtr = dt + 1;
	FSE_decode_t* const dinfo = (FSE_decode_t*)dPtr;
	const unsigned tableSize = 1 << nbBits;
	const unsigned tableMask = tableSize - 1;
	const unsigned maxSV1 = tableMask+1;
	unsigned s;

	/* Sanity checks */
	if (nbBits < 1) return ERROR(GENERIC); /* min size */

	/* Build Decoding Table */
	DTableH->tableLog = (U16)nbBits;
	DTableH->fastMode = 1;
	for (s=0; s<maxSV1; s++) {
	dinfo[s].newState = 0;
	dinfo[s].symbol = (BYTE)s;
	dinfo[s].nbBits = (BYTE)nbBits;
	}

	return 0;
	}

	FORCE_INLINE_TEMPLATE size_t FSE_decompress_usingDTable_generic(
	void* dst, size_t maxDstSize,
	const void* cSrc, size_t cSrcSize,
	const FSE_DTable* dt, const unsigned fast)
	{
	BYTE* const ostart = (BYTE*) dst;
	BYTE* op = ostart;
	BYTE* const omax = op + maxDstSize;
	BYTE* const olimit = omax-3;

	BIT_DStream_t bitD;
	FSE_DState_t state1;
	FSE_DState_t state2;

	/* Init */
	CHECK_F(BIT_initDStream(&bitD, cSrc, cSrcSize));

	FSE_initDState(&state1, &bitD, dt);
	FSE_initDState(&state2, &bitD, dt);

	#define FSE_GETSYMBOL(statePtr) fast ? FSE_decodeSymbolFast(statePtr, &bitD) : FSE_decodeSymbol(statePtr, &bitD)

	/* 4 symbols per loop */
	for ( ; (BIT_reloadDStream(&bitD)==BIT_DStream_unfinished) & (op<olimit) ; op+=4) {
	op[0] = FSE_GETSYMBOL(&state1);

	if (FSE_MAX_TABLELOG2+7 > sizeof(bitD.bitContainer)8) /* This test must be static */
	BIT_reloadDStream(&bitD);

	op[1] = FSE_GETSYMBOL(&state2);

	if (FSE_MAX_TABLELOG4+7 > sizeof(bitD.bitContainer)8) /* This test must be static */
	{ if (BIT_reloadDStream(&bitD) > BIT_DStream_unfinished) { op+=2; break; } }

	op[2] = FSE_GETSYMBOL(&state1);

	if (FSE_MAX_TABLELOG2+7 > sizeof(bitD.bitContainer)8) /* This test must be static */
	BIT_reloadDStream(&bitD);

	op[3] = FSE_GETSYMBOL(&state2);
	}

	/* tail */
	/* note : BIT_reloadDStream(&bitD) >= FSE_DStream_partiallyFilled; Ends at exactly BIT_DStream_completed */
	while (1) {
	if (op>(omax-2)) return ERROR(dstSize_tooSmall);
	*op++ = FSE_GETSYMBOL(&state1);
	if (BIT_reloadDStream(&bitD)==BIT_DStream_overflow) {
	*op++ = FSE_GETSYMBOL(&state2);
	break;
	}

	if (op>(omax-2)) return ERROR(dstSize_tooSmall);
	*op++ = FSE_GETSYMBOL(&state2);
	if (BIT_reloadDStream(&bitD)==BIT_DStream_overflow) {
	*op++ = FSE_GETSYMBOL(&state1);
	break;
	} }

	return op-ostart;
	}


	size_t FSE_decompress_usingDTable(void* dst, size_t originalSize,
	const void* cSrc, size_t cSrcSize,
	const FSE_DTable* dt)
	{
	const void* ptr = dt;
	const FSE_DTableHeader* DTableH = (const FSE_DTableHeader*)ptr;
	const U32 fastMode = DTableH->fastMode;

	/* select fast mode (static) */
	if (fastMode) return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 1);
	return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 0);
	}


	size_t FSE_decompress_wksp(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize, FSE_DTable* workSpace, unsigned maxLog)
	{
	const BYTE* const istart = (const BYTE*)cSrc;
	const BYTE* ip = istart;
	short counting[FSE_MAX_SYMBOL_VALUE+1];
	unsigned tableLog;
	unsigned maxSymbolValue = FSE_MAX_SYMBOL_VALUE;

	/* normal FSE decoding mode */
	size_t const NCountLength = FSE_readNCount (counting, &maxSymbolValue, &tableLog, istart, cSrcSize);
	if (FSE_isError(NCountLength)) return NCountLength;
	/* if (NCountLength >= cSrcSize) return ERROR(srcSize_wrong); / / too small input size; supposed to be already checked in NCountLength, only remaining case : NCountLength==cSrcSize */
	if (tableLog > maxLog) return ERROR(tableLog_tooLarge);
	ip += NCountLength;
	cSrcSize -= NCountLength;

	CHECK_F( FSE_buildDTable (workSpace, counting, maxSymbolValue, tableLog) );

	return FSE_decompress_usingDTable (dst, dstCapacity, ip, cSrcSize, workSpace); /* always return, even if it is an error code */
	}


	typedef FSE_DTable DTable_max_t[FSE_DTABLE_SIZE_U32(FSE_MAX_TABLELOG)];

	size_t FSE_decompress(void* dst, size_t dstCapacity, const void* cSrc, size_t cSrcSize)
	{
	DTable_max_t dt; /* Static analyzer seems unable to understand this table will be properly initialized later */
	return FSE_decompress_wksp(dst, dstCapacity, cSrc, cSrcSize, dt, FSE_MAX_TABLELOG);
	}



	#endif /* FSE_COMMONDEFS_ONLY */
	/** ended inlining common/fse_decompress.c **/
	/** start inlining common/pool.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/


	/* ====== Dependencies ======= */
	#include <stddef.h> /* size_t */
	/** skipping file: debug.h **/
	/** start inlining zstd_internal.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_CCOMMON_H_MODULE
	#define ZSTD_CCOMMON_H_MODULE

	/* this module contains definitions which must be identical
	* across compression, decompression and dictBuilder.
	* It also contains a few functions useful to at least 2 of them
	* and which benefit from being inlined */

	/-************************************
	* Dependencies
	***************************************/
	#if !defined(ZSTD_NO_INTRINSICS) && defined(__ARM_NEON)
	#include <arm_neon.h>
	#endif
	/** skipping file: compiler.h **/
	/** skipping file: mem.h **/
	/** skipping file: debug.h **/
	/** skipping file: error_private.h **/
	#define ZSTD_STATIC_LINKING_ONLY
	/** start inlining ../zstd.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/
	#if defined (__cplusplus)
	extern "C" {
	#endif

	#ifndef ZSTD_H_235446
	#define ZSTD_H_235446

	/* ====== Dependency ======*/
	#include <limits.h> /* INT_MAX */
	#include <stddef.h> /* size_t */


	/* ===== ZSTDLIB_API : control library symbols visibility ===== */
	#ifndef ZSTDLIB_VISIBILITY
	# if defined(__GNUC__) && (__GNUC__ >= 4)
	# define ZSTDLIB_VISIBILITY __attribute__ ((visibility ("default")))
	# else
	# define ZSTDLIB_VISIBILITY
	# endif
	#endif
	#if defined(ZSTD_DLL_EXPORT) && (ZSTD_DLL_EXPORT==1)
	# define ZSTDLIB_API __declspec(dllexport) ZSTDLIB_VISIBILITY
	#elif defined(ZSTD_DLL_IMPORT) && (ZSTD_DLL_IMPORT==1)
	# define ZSTDLIB_API __declspec(dllimport) ZSTDLIB_VISIBILITY /* It isn't required but allows to generate better code, saving a function pointer load from the IAT and an indirect jump.*/
	#else
	# define ZSTDLIB_API ZSTDLIB_VISIBILITY
	#endif


	/*******************************************************************************
	Introduction

	zstd, short for Zstandard, is a fast lossless compression algorithm, targeting
	real-time compression scenarios at zlib-level and better compression ratios.
	The zstd compression library provides in-memory compression and decompression
	functions.

	The library supports regular compression levels from 1 up to ZSTD_maxCLevel(),
	which is currently 22. Levels >= 20, labeled `--ultra`, should be used with
	caution, as they require more memory. The library also offers negative
	compression levels, which extend the range of speed vs. ratio preferences.
	The lower the level, the faster the speed (at the cost of compression).

	Compression can be done in:
	- a single step (described as Simple API)
	- a single step, reusing a context (described as Explicit context)
	- unbounded multiple steps (described as Streaming compression)

	The compression ratio achievable on small data can be highly improved using
	a dictionary. Dictionary compression can be performed in:
	- a single step (described as Simple dictionary API)
	- a single step, reusing a dictionary (described as Bulk-processing
	dictionary API)

	Advanced experimental functions can be accessed using
	`#define ZSTD_STATIC_LINKING_ONLY` before including zstd.h.

	Advanced experimental APIs should never be used with a dynamically-linked
	library. They are not "stable"; their definitions or signatures may change in
	the future. Only static linking is allowed.
	*******************************************************************************/

	/------ Version ------/
	#define ZSTD_VERSION_MAJOR 1
	#define ZSTD_VERSION_MINOR 4
	#define ZSTD_VERSION_RELEASE 5

	#define ZSTD_VERSION_NUMBER (ZSTD_VERSION_MAJOR 100100 + ZSTD_VERSION_MINOR *100 + ZSTD_VERSION_RELEASE)
	ZSTDLIB_API unsigned ZSTD_versionNumber(void); /*< to check runtime library version /

	#define ZSTD_LIB_VERSION ZSTD_VERSION_MAJOR.ZSTD_VERSION_MINOR.ZSTD_VERSION_RELEASE
	#define ZSTD_QUOTE(str) #str
	#define ZSTD_EXPAND_AND_QUOTE(str) ZSTD_QUOTE(str)
	#define ZSTD_VERSION_STRING ZSTD_EXPAND_AND_QUOTE(ZSTD_LIB_VERSION)
	ZSTDLIB_API const char* ZSTD_versionString(void); /* requires v1.3.0+ */

	/* *************************************
	* Default constant
	***************************************/
	#ifndef ZSTD_CLEVEL_DEFAULT
	# define ZSTD_CLEVEL_DEFAULT 3
	#endif

	/* *************************************
	* Constants
	***************************************/

	/* All magic numbers are supposed read/written to/from files/memory using little-endian convention */
	#define ZSTD_MAGICNUMBER 0xFD2FB528 /* valid since v0.8.0 */
	#define ZSTD_MAGIC_DICTIONARY 0xEC30A437 /* valid since v0.7.0 */
	#define ZSTD_MAGIC_SKIPPABLE_START 0x184D2A50 /* all 16 values, from 0x184D2A50 to 0x184D2A5F, signal the beginning of a skippable frame */
	#define ZSTD_MAGIC_SKIPPABLE_MASK 0xFFFFFFF0

	#define ZSTD_BLOCKSIZELOG_MAX 17
	#define ZSTD_BLOCKSIZE_MAX (1<<ZSTD_BLOCKSIZELOG_MAX)



	/***************************************
	* Simple API
	***************************************/
	/*! ZSTD_compress() :
	* Compresses `src` content as a single zstd compressed frame into already allocated `dst`.
	* Hint : compression runs faster if `dstCapacity` >= `ZSTD_compressBound(srcSize)`.
	* @return : compressed size written into `dst` (<= `dstCapacity),
	* or an error code if it fails (which can be tested using ZSTD_isError()). */
	ZSTDLIB_API size_t ZSTD_compress( void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	int compressionLevel);

	/*! ZSTD_decompress() :
	* `compressedSize` : must be the _exact_ size of some number of compressed and/or skippable frames.
	* `dstCapacity` is an upper bound of originalSize to regenerate.
	* If user cannot imply a maximum upper bound, it's better to use streaming mode to decompress data.
	* @return : the number of bytes decompressed into `dst` (<= `dstCapacity`),
	* or an errorCode if it fails (which can be tested using ZSTD_isError()). */
	ZSTDLIB_API size_t ZSTD_decompress( void* dst, size_t dstCapacity,
	const void* src, size_t compressedSize);

	/*! ZSTD_getFrameContentSize() : requires v1.3.0+
	* `src` should point to the start of a ZSTD encoded frame.
	* `srcSize` must be at least as large as the frame header.
	* hint : any size >= `ZSTD_frameHeaderSize_max` is large enough.
	* @return : - decompressed size of `src` frame content, if known
	* - ZSTD_CONTENTSIZE_UNKNOWN if the size cannot be determined
	* - ZSTD_CONTENTSIZE_ERROR if an error occurred (e.g. invalid magic number, srcSize too small)
	* note 1 : a 0 return value means the frame is valid but "empty".
	* note 2 : decompressed size is an optional field, it may not be present, typically in streaming mode.
	* When `return==ZSTD_CONTENTSIZE_UNKNOWN`, data to decompress could be any size.
	* In which case, it's necessary to use streaming mode to decompress data.
	* Optionally, application can rely on some implicit limit,
	* as ZSTD_decompress() only needs an upper bound of decompressed size.
	* (For example, data could be necessarily cut into blocks <= 16 KB).
	* note 3 : decompressed size is always present when compression is completed using single-pass functions,
	* such as ZSTD_compress(), ZSTD_compressCCtx() ZSTD_compress_usingDict() or ZSTD_compress_usingCDict().
	* note 4 : decompressed size can be very large (64-bits value),
	* potentially larger than what local system can handle as a single memory segment.
	* In which case, it's necessary to use streaming mode to decompress data.
	* note 5 : If source is untrusted, decompressed size could be wrong or intentionally modified.
	* Always ensure return value fits within application's authorized limits.
	* Each application can set its own limits.
	* note 6 : This function replaces ZSTD_getDecompressedSize() */
	#define ZSTD_CONTENTSIZE_UNKNOWN (0ULL - 1)
	#define ZSTD_CONTENTSIZE_ERROR (0ULL - 2)
	ZSTDLIB_API unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize);

	/*! ZSTD_getDecompressedSize() :
	* NOTE: This function is now obsolete, in favor of ZSTD_getFrameContentSize().
	* Both functions work the same way, but ZSTD_getDecompressedSize() blends
	* "empty", "unknown" and "error" results to the same return value (0),
	* while ZSTD_getFrameContentSize() gives them separate return values.
	* @return : decompressed size of `src` frame content _if known and not empty_, 0 otherwise. */
	ZSTDLIB_API unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize);

	/*! ZSTD_findFrameCompressedSize() :
	* `src` should point to the start of a ZSTD frame or skippable frame.
	* `srcSize` must be >= first frame size
	* @return : the compressed size of the first frame starting at `src`,
	* suitable to pass as `srcSize` to `ZSTD_decompress` or similar,
	* or an error code if input is invalid */
	ZSTDLIB_API size_t ZSTD_findFrameCompressedSize(const void* src, size_t srcSize);


	/====== Helper functions ======/
	#define ZSTD_COMPRESSBOUND(srcSize) ((srcSize) + ((srcSize)>>8) + (((srcSize) < (128<<10)) ? (((128<<10) - (srcSize)) >> 11) /* margin, from 64 to 0 / : 0)) / this formula ensures that bound(A) + bound(B) <= bound(A+B) as long as A and B >= 128 KB */
	ZSTDLIB_API size_t ZSTD_compressBound(size_t srcSize); /!< maximum compressed size in worst case single-pass scenario /
	ZSTDLIB_API unsigned ZSTD_isError(size_t code); /!< tells if a `size_t` function result is an error code /
	ZSTDLIB_API const char* ZSTD_getErrorName(size_t code); /!< provides readable string from an error code /
	ZSTDLIB_API int ZSTD_minCLevel(void); /!< minimum negative compression level allowed /
	ZSTDLIB_API int ZSTD_maxCLevel(void); /!< maximum compression level available /


	/***************************************
	* Explicit context
	***************************************/
	/*= Compression context
	* When compressing many times,
	* it is recommended to allocate a context just once,
	* and re-use it for each successive compression operation.
	* This will make workload friendlier for system's memory.
	* Note : re-using context is just a speed / resource optimization.
	* It doesn't change the compression ratio, which remains identical.
	* Note 2 : In multi-threaded environments,
	* use one different context per thread for parallel execution.
	*/
	typedef struct ZSTD_CCtx_s ZSTD_CCtx;
	ZSTDLIB_API ZSTD_CCtx* ZSTD_createCCtx(void);
	ZSTDLIB_API size_t ZSTD_freeCCtx(ZSTD_CCtx* cctx);

	/*! ZSTD_compressCCtx() :
	* Same as ZSTD_compress(), using an explicit ZSTD_CCtx.
	* Important : in order to behave similarly to `ZSTD_compress()`,
	* this function compresses at requested compression level,
	* __ignoring any other parameter__ .
	* If any advanced parameter was set using the advanced API,
	* they will all be reset. Only `compressionLevel` remains.
	*/
	ZSTDLIB_API size_t ZSTD_compressCCtx(ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	int compressionLevel);

	/*= Decompression context
	* When decompressing many times,
	* it is recommended to allocate a context only once,
	* and re-use it for each successive compression operation.
	* This will make workload friendlier for system's memory.
	* Use one context per thread for parallel execution. */
	typedef struct ZSTD_DCtx_s ZSTD_DCtx;
	ZSTDLIB_API ZSTD_DCtx* ZSTD_createDCtx(void);
	ZSTDLIB_API size_t ZSTD_freeDCtx(ZSTD_DCtx* dctx);

	/*! ZSTD_decompressDCtx() :
	* Same as ZSTD_decompress(),
	* requires an allocated ZSTD_DCtx.
	* Compatible with sticky parameters.
	*/
	ZSTDLIB_API size_t ZSTD_decompressDCtx(ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize);


	/***************************************
	* Advanced compression API
	***************************************/

	/* API design :
	* Parameters are pushed one by one into an existing context,
	* using ZSTD_CCtx_set*() functions.
	* Pushed parameters are sticky : they are valid for next compressed frame, and any subsequent frame.
	* "sticky" parameters are applicable to `ZSTD_compress2()` and `ZSTD_compressStream*()` !
	* __They do not apply to "simple" one-shot variants such as ZSTD_compressCCtx()__ .
	*
	* It's possible to reset all parameters to "default" using ZSTD_CCtx_reset().
	*
	* This API supercedes all other "advanced" API entry points in the experimental section.
	* In the future, we expect to remove from experimental API entry points which are redundant with this API.
	*/


	/* Compression strategies, listed from fastest to strongest */
	typedef enum { ZSTD_fast=1,
	ZSTD_dfast=2,
	ZSTD_greedy=3,
	ZSTD_lazy=4,
	ZSTD_lazy2=5,
	ZSTD_btlazy2=6,
	ZSTD_btopt=7,
	ZSTD_btultra=8,
	ZSTD_btultra2=9
	/* note : new strategies _might_ be added in the future.
	Only the order (from fast to strong) is guaranteed */
	} ZSTD_strategy;


	typedef enum {

	/* compression parameters
	* Note: When compressing with a ZSTD_CDict these parameters are superseded
	* by the parameters used to construct the ZSTD_CDict.
	* See ZSTD_CCtx_refCDict() for more info (superseded-by-cdict). */
	ZSTD_c_compressionLevel=100, /* Set compression parameters according to pre-defined cLevel table.
	* Note that exact compression parameters are dynamically determined,
	* depending on both compression level and srcSize (when known).
	* Default level is ZSTD_CLEVEL_DEFAULT==3.
	* Special: value 0 means default, which is controlled by ZSTD_CLEVEL_DEFAULT.
	* Note 1 : it's possible to pass a negative compression level.
	* Note 2 : setting a level does not automatically set all other compression parameters
	* to default. Setting this will however eventually dynamically impact the compression
	* parameters which have not been manually set. The manually set
	* ones will 'stick'. */
	/* Advanced compression parameters :
	* It's possible to pin down compression parameters to some specific values.
	* In which case, these values are no longer dynamically selected by the compressor */
	ZSTD_c_windowLog=101, /* Maximum allowed back-reference distance, expressed as power of 2.
	* This will set a memory budget for streaming decompression,
	* with larger values requiring more memory
	* and typically compressing more.
	* Must be clamped between ZSTD_WINDOWLOG_MIN and ZSTD_WINDOWLOG_MAX.
	* Special: value 0 means "use default windowLog".
	* Note: Using a windowLog greater than ZSTD_WINDOWLOG_LIMIT_DEFAULT
	* requires explicitly allowing such size at streaming decompression stage. */
	ZSTD_c_hashLog=102, /* Size of the initial probe table, as a power of 2.
	* Resulting memory usage is (1 << (hashLog+2)).
	* Must be clamped between ZSTD_HASHLOG_MIN and ZSTD_HASHLOG_MAX.
	* Larger tables improve compression ratio of strategies <= dFast,
	* and improve speed of strategies > dFast.
	* Special: value 0 means "use default hashLog". */
	ZSTD_c_chainLog=103, /* Size of the multi-probe search table, as a power of 2.
	* Resulting memory usage is (1 << (chainLog+2)).
	* Must be clamped between ZSTD_CHAINLOG_MIN and ZSTD_CHAINLOG_MAX.
	* Larger tables result in better and slower compression.
	* This parameter is useless for "fast" strategy.
	* It's still useful when using "dfast" strategy,
	* in which case it defines a secondary probe table.
	* Special: value 0 means "use default chainLog". */
	ZSTD_c_searchLog=104, /* Number of search attempts, as a power of 2.
	* More attempts result in better and slower compression.
	* This parameter is useless for "fast" and "dFast" strategies.
	* Special: value 0 means "use default searchLog". */
	ZSTD_c_minMatch=105, /* Minimum size of searched matches.
	* Note that Zstandard can still find matches of smaller size,
	* it just tweaks its search algorithm to look for this size and larger.
	* Larger values increase compression and decompression speed, but decrease ratio.
	* Must be clamped between ZSTD_MINMATCH_MIN and ZSTD_MINMATCH_MAX.
	* Note that currently, for all strategies < btopt, effective minimum is 4.
	* , for all strategies > fast, effective maximum is 6.
	* Special: value 0 means "use default minMatchLength". */
	ZSTD_c_targetLength=106, /* Impact of this field depends on strategy.
	* For strategies btopt, btultra & btultra2:
	* Length of Match considered "good enough" to stop search.
	* Larger values make compression stronger, and slower.
	* For strategy fast:
	* Distance between match sampling.
	* Larger values make compression faster, and weaker.
	* Special: value 0 means "use default targetLength". */
	ZSTD_c_strategy=107, /* See ZSTD_strategy enum definition.
	* The higher the value of selected strategy, the more complex it is,
	* resulting in stronger and slower compression.
	* Special: value 0 means "use default strategy". */

	/* LDM mode parameters */
	ZSTD_c_enableLongDistanceMatching=160, /* Enable long distance matching.
	* This parameter is designed to improve compression ratio
	* for large inputs, by finding large matches at long distance.
	* It increases memory usage and window size.
	* Note: enabling this parameter increases default ZSTD_c_windowLog to 128 MB
	* except when expressly set to a different value. */
	ZSTD_c_ldmHashLog=161, /* Size of the table for long distance matching, as a power of 2.
	* Larger values increase memory usage and compression ratio,
	* but decrease compression speed.
	* Must be clamped between ZSTD_HASHLOG_MIN and ZSTD_HASHLOG_MAX
	* default: windowlog - 7.
	* Special: value 0 means "automatically determine hashlog". */
	ZSTD_c_ldmMinMatch=162, /* Minimum match size for long distance matcher.
	* Larger/too small values usually decrease compression ratio.
	* Must be clamped between ZSTD_LDM_MINMATCH_MIN and ZSTD_LDM_MINMATCH_MAX.
	* Special: value 0 means "use default value" (default: 64). */
	ZSTD_c_ldmBucketSizeLog=163, /* Log size of each bucket in the LDM hash table for collision resolution.
	* Larger values improve collision resolution but decrease compression speed.
	* The maximum value is ZSTD_LDM_BUCKETSIZELOG_MAX.
	* Special: value 0 means "use default value" (default: 3). */
	ZSTD_c_ldmHashRateLog=164, /* Frequency of inserting/looking up entries into the LDM hash table.
	* Must be clamped between 0 and (ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN).
	* Default is MAX(0, (windowLog - ldmHashLog)), optimizing hash table usage.
	* Larger values improve compression speed.
	* Deviating far from default value will likely result in a compression ratio decrease.
	* Special: value 0 means "automatically determine hashRateLog". */

	/* frame parameters */
	ZSTD_c_contentSizeFlag=200, /* Content size will be written into frame header _whenever known_ (default:1)
	* Content size must be known at the beginning of compression.
	* This is automatically the case when using ZSTD_compress2(),
	* For streaming scenarios, content size must be provided with ZSTD_CCtx_setPledgedSrcSize() */
	ZSTD_c_checksumFlag=201, /* A 32-bits checksum of content is written at end of frame (default:0) */
	ZSTD_c_dictIDFlag=202, /* When applicable, dictionary's ID is written into frame header (default:1) */

	/* multi-threading parameters */
	/* These parameters are only useful if multi-threading is enabled (compiled with build macro ZSTD_MULTITHREAD).
	* They return an error otherwise. */
	ZSTD_c_nbWorkers=400, /* Select how many threads will be spawned to compress in parallel.
	* When nbWorkers >= 1, triggers asynchronous mode when used with ZSTD_compressStream*() :
	* ZSTD_compressStream*() consumes input and flush output if possible, but immediately gives back control to caller,
	* while compression work is performed in parallel, within worker threads.
	* (note : a strong exception to this rule is when first invocation of ZSTD_compressStream2() sets ZSTD_e_end :
	* in which case, ZSTD_compressStream2() delegates to ZSTD_compress2(), which is always a blocking call).
	* More workers improve speed, but also increase memory usage.
	* Default value is `0`, aka "single-threaded mode" : no worker is spawned, compression is performed inside Caller's thread, all invocations are blocking */
	ZSTD_c_jobSize=401, /* Size of a compression job. This value is enforced only when nbWorkers >= 1.
	* Each compression job is completed in parallel, so this value can indirectly impact the nb of active threads.
	* 0 means default, which is dynamically determined based on compression parameters.
	* Job size must be a minimum of overlap size, or 1 MB, whichever is largest.
	* The minimum size is automatically and transparently enforced. */
	ZSTD_c_overlapLog=402, /* Control the overlap size, as a fraction of window size.
	* The overlap size is an amount of data reloaded from previous job at the beginning of a new job.
	* It helps preserve compression ratio, while each job is compressed in parallel.
	* This value is enforced only when nbWorkers >= 1.
	* Larger values increase compression ratio, but decrease speed.
	* Possible values range from 0 to 9 :
	* - 0 means "default" : value will be determined by the library, depending on strategy
	* - 1 means "no overlap"
	* - 9 means "full overlap", using a full window size.
	* Each intermediate rank increases/decreases load size by a factor 2 :
	* 9: full window; 8: w/2; 7: w/4; 6: w/8; 5:w/16; 4: w/32; 3:w/64; 2:w/128; 1:no overlap; 0:default
	* default value varies between 6 and 9, depending on strategy */

	/* note : additional experimental parameters are also available
	* within the experimental section of the API.
	* At the time of this writing, they include :
	* ZSTD_c_rsyncable
	* ZSTD_c_format
	* ZSTD_c_forceMaxWindow
	* ZSTD_c_forceAttachDict
	* ZSTD_c_literalCompressionMode
	* ZSTD_c_targetCBlockSize
	* ZSTD_c_srcSizeHint
	* Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
	* note : never ever use experimentalParam? names directly;
	* also, the enums values themselves are unstable and can still change.
	*/
	ZSTD_c_experimentalParam1=500,
	ZSTD_c_experimentalParam2=10,
	ZSTD_c_experimentalParam3=1000,
	ZSTD_c_experimentalParam4=1001,
	ZSTD_c_experimentalParam5=1002,
	ZSTD_c_experimentalParam6=1003,
	ZSTD_c_experimentalParam7=1004
	} ZSTD_cParameter;

	typedef struct {
	size_t error;
	int lowerBound;
	int upperBound;
	} ZSTD_bounds;

	/*! ZSTD_cParam_getBounds() :
	* All parameters must belong to an interval with lower and upper bounds,
	* otherwise they will either trigger an error or be automatically clamped.
	* @return : a structure, ZSTD_bounds, which contains
	* - an error status field, which must be tested using ZSTD_isError()
	* - lower and upper bounds, both inclusive
	*/
	ZSTDLIB_API ZSTD_bounds ZSTD_cParam_getBounds(ZSTD_cParameter cParam);

	/*! ZSTD_CCtx_setParameter() :
	* Set one compression parameter, selected by enum ZSTD_cParameter.
	* All parameters have valid bounds. Bounds can be queried using ZSTD_cParam_getBounds().
	* Providing a value beyond bound will either clamp it, or trigger an error (depending on parameter).
	* Setting a parameter is generally only possible during frame initialization (before starting compression).
	* Exception : when using multi-threading mode (nbWorkers >= 1),
	* the following parameters can be updated _during_ compression (within same frame):
	* => compressionLevel, hashLog, chainLog, searchLog, minMatch, targetLength and strategy.
	* new parameters will be active for next job only (after a flush()).
	* @return : an error code (which can be tested using ZSTD_isError()).
	*/
	ZSTDLIB_API size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int value);

	/*! ZSTD_CCtx_setPledgedSrcSize() :
	* Total input data size to be compressed as a single frame.
	* Value will be written in frame header, unless if explicitly forbidden using ZSTD_c_contentSizeFlag.
	* This value will also be controlled at end of frame, and trigger an error if not respected.
	* @result : 0, or an error code (which can be tested with ZSTD_isError()).
	* Note 1 : pledgedSrcSize==0 actually means zero, aka an empty frame.
	* In order to mean "unknown content size", pass constant ZSTD_CONTENTSIZE_UNKNOWN.
	* ZSTD_CONTENTSIZE_UNKNOWN is default value for any new frame.
	* Note 2 : pledgedSrcSize is only valid once, for the next frame.
	* It's discarded at the end of the frame, and replaced by ZSTD_CONTENTSIZE_UNKNOWN.
	* Note 3 : Whenever all input data is provided and consumed in a single round,
	* for example with ZSTD_compress2(),
	* or invoking immediately ZSTD_compressStream2(,,,ZSTD_e_end),
	* this value is automatically overridden by srcSize instead.
	*/
	ZSTDLIB_API size_t ZSTD_CCtx_setPledgedSrcSize(ZSTD_CCtx* cctx, unsigned long long pledgedSrcSize);

	typedef enum {
	ZSTD_reset_session_only = 1,
	ZSTD_reset_parameters = 2,
	ZSTD_reset_session_and_parameters = 3
	} ZSTD_ResetDirective;

	/*! ZSTD_CCtx_reset() :
	* There are 2 different things that can be reset, independently or jointly :
	* - The session : will stop compressing current frame, and make CCtx ready to start a new one.
	* Useful after an error, or to interrupt any ongoing compression.
	* Any internal data not yet flushed is cancelled.
	* Compression parameters and dictionary remain unchanged.
	* They will be used to compress next frame.
	* Resetting session never fails.
	* - The parameters : changes all parameters back to "default".
	* This removes any reference to any dictionary too.
	* Parameters can only be changed between 2 sessions (i.e. no compression is currently ongoing)
	* otherwise the reset fails, and function returns an error value (which can be tested using ZSTD_isError())
	* - Both : similar to resetting the session, followed by resetting parameters.
	*/
	ZSTDLIB_API size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx, ZSTD_ResetDirective reset);

	/*! ZSTD_compress2() :
	* Behave the same as ZSTD_compressCCtx(), but compression parameters are set using the advanced API.
	* ZSTD_compress2() always starts a new frame.
	* Should cctx hold data from a previously unfinished frame, everything about it is forgotten.
	* - Compression parameters are pushed into CCtx before starting compression, using ZSTD_CCtx_set*()
	* - The function is always blocking, returns when compression is completed.
	* Hint : compression runs faster if `dstCapacity` >= `ZSTD_compressBound(srcSize)`.
	* @return : compressed size written into `dst` (<= `dstCapacity),
	* or an error code if it fails (which can be tested using ZSTD_isError()).
	*/
	ZSTDLIB_API size_t ZSTD_compress2( ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize);


	/***************************************
	* Advanced decompression API
	***************************************/

	/* The advanced API pushes parameters one by one into an existing DCtx context.
	* Parameters are sticky, and remain valid for all following frames
	* using the same DCtx context.
	* It's possible to reset parameters to default values using ZSTD_DCtx_reset().
	* Note : This API is compatible with existing ZSTD_decompressDCtx() and ZSTD_decompressStream().
	* Therefore, no new decompression function is necessary.
	*/

	typedef enum {

	ZSTD_d_windowLogMax=100, /* Select a size limit (in power of 2) beyond which
	* the streaming API will refuse to allocate memory buffer
	* in order to protect the host from unreasonable memory requirements.
	* This parameter is only useful in streaming mode, since no internal buffer is allocated in single-pass mode.
	* By default, a decompression context accepts window sizes <= (1 << ZSTD_WINDOWLOG_LIMIT_DEFAULT).
	* Special: value 0 means "use default maximum windowLog". */

	/* note : additional experimental parameters are also available
	* within the experimental section of the API.
	* At the time of this writing, they include :
	* ZSTD_d_format
	* ZSTD_d_stableOutBuffer
	* Because they are not stable, it's necessary to define ZSTD_STATIC_LINKING_ONLY to access them.
	* note : never ever use experimentalParam? names directly
	*/
	ZSTD_d_experimentalParam1=1000,
	ZSTD_d_experimentalParam2=1001

	} ZSTD_dParameter;

	/*! ZSTD_dParam_getBounds() :
	* All parameters must belong to an interval with lower and upper bounds,
	* otherwise they will either trigger an error or be automatically clamped.
	* @return : a structure, ZSTD_bounds, which contains
	* - an error status field, which must be tested using ZSTD_isError()
	* - both lower and upper bounds, inclusive
	*/
	ZSTDLIB_API ZSTD_bounds ZSTD_dParam_getBounds(ZSTD_dParameter dParam);

	/*! ZSTD_DCtx_setParameter() :
	* Set one compression parameter, selected by enum ZSTD_dParameter.
	* All parameters have valid bounds. Bounds can be queried using ZSTD_dParam_getBounds().
	* Providing a value beyond bound will either clamp it, or trigger an error (depending on parameter).
	* Setting a parameter is only possible during frame initialization (before starting decompression).
	* @return : 0, or an error code (which can be tested using ZSTD_isError()).
	*/
	ZSTDLIB_API size_t ZSTD_DCtx_setParameter(ZSTD_DCtx* dctx, ZSTD_dParameter param, int value);

	/*! ZSTD_DCtx_reset() :
	* Return a DCtx to clean state.
	* Session and parameters can be reset jointly or separately.
	* Parameters can only be reset when no active frame is being decompressed.
	* @return : 0, or an error code, which can be tested with ZSTD_isError()
	*/
	ZSTDLIB_API size_t ZSTD_DCtx_reset(ZSTD_DCtx* dctx, ZSTD_ResetDirective reset);


	/****************************
	* Streaming
	****************************/

	typedef struct ZSTD_inBuffer_s {
	const void* src; /*< start of input buffer /
	size_t size; /*< size of input buffer /
	size_t pos; /*< position where reading stopped. Will be updated. Necessarily 0 <= pos <= size /
	} ZSTD_inBuffer;

	typedef struct ZSTD_outBuffer_s {
	void* dst; /*< start of output buffer /
	size_t size; /*< size of output buffer /
	size_t pos; /*< position where writing stopped. Will be updated. Necessarily 0 <= pos <= size /
	} ZSTD_outBuffer;



	/-**********************************************************************
	* Streaming compression - HowTo
	*
	* A ZSTD_CStream object is required to track streaming operation.
	* Use ZSTD_createCStream() and ZSTD_freeCStream() to create/release resources.
	* ZSTD_CStream objects can be reused multiple times on consecutive compression operations.
	* It is recommended to re-use ZSTD_CStream since it will play nicer with system's memory, by re-using already allocated memory.
	*
	* For parallel execution, use one separate ZSTD_CStream per thread.
	*
	* note : since v1.3.0, ZSTD_CStream and ZSTD_CCtx are the same thing.
	*
	* Parameters are sticky : when starting a new compression on the same context,
	* it will re-use the same sticky parameters as previous compression session.
	* When in doubt, it's recommended to fully initialize the context before usage.
	* Use ZSTD_CCtx_reset() to reset the context and ZSTD_CCtx_setParameter(),
	* ZSTD_CCtx_setPledgedSrcSize(), or ZSTD_CCtx_loadDictionary() and friends to
	* set more specific parameters, the pledged source size, or load a dictionary.
	*
	* Use ZSTD_compressStream2() with ZSTD_e_continue as many times as necessary to
	* consume input stream. The function will automatically update both `pos`
	* fields within `input` and `output`.
	* Note that the function may not consume the entire input, for example, because
	* the output buffer is already full, in which case `input.pos < input.size`.
	* The caller must check if input has been entirely consumed.
	* If not, the caller must make some room to receive more compressed data,
	* and then present again remaining input data.
	* note: ZSTD_e_continue is guaranteed to make some forward progress when called,
	* but doesn't guarantee maximal forward progress. This is especially relevant
	* when compressing with multiple threads. The call won't block if it can
	* consume some input, but if it can't it will wait for some, but not all,
	* output to be flushed.
	* @return : provides a minimum amount of data remaining to be flushed from internal buffers
	* or an error code, which can be tested using ZSTD_isError().
	*
	* At any moment, it's possible to flush whatever data might remain stuck within internal buffer,
	* using ZSTD_compressStream2() with ZSTD_e_flush. `output->pos` will be updated.
	* Note that, if `output->size` is too small, a single invocation with ZSTD_e_flush might not be enough (return code > 0).
	* In which case, make some room to receive more compressed data, and call again ZSTD_compressStream2() with ZSTD_e_flush.
	* You must continue calling ZSTD_compressStream2() with ZSTD_e_flush until it returns 0, at which point you can change the
	* operation.
	* note: ZSTD_e_flush will flush as much output as possible, meaning when compressing with multiple threads, it will
	* block until the flush is complete or the output buffer is full.
	* @return : 0 if internal buffers are entirely flushed,
	* >0 if some data still present within internal buffer (the value is minimal estimation of remaining size),
	* or an error code, which can be tested using ZSTD_isError().
	*
	* Calling ZSTD_compressStream2() with ZSTD_e_end instructs to finish a frame.
	* It will perform a flush and write frame epilogue.
	* The epilogue is required for decoders to consider a frame completed.
	* flush operation is the same, and follows same rules as calling ZSTD_compressStream2() with ZSTD_e_flush.
	* You must continue calling ZSTD_compressStream2() with ZSTD_e_end until it returns 0, at which point you are free to
	* start a new frame.
	* note: ZSTD_e_end will flush as much output as possible, meaning when compressing with multiple threads, it will
	* block until the flush is complete or the output buffer is full.
	* @return : 0 if frame fully completed and fully flushed,
	* >0 if some data still present within internal buffer (the value is minimal estimation of remaining size),
	* or an error code, which can be tested using ZSTD_isError().
	*
	* *******************************************************************/

	typedef ZSTD_CCtx ZSTD_CStream; /*< CCtx and CStream are now effectively same object (>= v1.3.0) /
	/* Continue to distinguish them for compatibility with older versions <= v1.2.0 */
	/===== ZSTD_CStream management functions =====/
	ZSTDLIB_API ZSTD_CStream* ZSTD_createCStream(void);
	ZSTDLIB_API size_t ZSTD_freeCStream(ZSTD_CStream* zcs);

	/===== Streaming compression functions =====/
	typedef enum {
	ZSTD_e_continue=0, /* collect more data, encoder decides when to output compressed result, for optimal compression ratio */
	ZSTD_e_flush=1, /* flush any data provided so far,
	* it creates (at least) one new block, that can be decoded immediately on reception;
	* frame will continue: any future data can still reference previously compressed data, improving compression.
	* note : multithreaded compression will block to flush as much output as possible. */
	ZSTD_e_end=2 /* flush any remaining data _and_ close current frame.
	* note that frame is only closed after compressed data is fully flushed (return value == 0).
	* After that point, any additional data starts a new frame.
	* note : each frame is independent (does not reference any content from previous frame).
	: note : multithreaded compression will block to flush as much output as possible. */
	} ZSTD_EndDirective;

	/*! ZSTD_compressStream2() :
	* Behaves about the same as ZSTD_compressStream, with additional control on end directive.
	* - Compression parameters are pushed into CCtx before starting compression, using ZSTD_CCtx_set*()
	* - Compression parameters cannot be changed once compression is started (save a list of exceptions in multi-threading mode)
	* - output->pos must be <= dstCapacity, input->pos must be <= srcSize
	* - output->pos and input->pos will be updated. They are guaranteed to remain below their respective limit.
	* - When nbWorkers==0 (default), function is blocking : it completes its job before returning to caller.
	* - When nbWorkers>=1, function is non-blocking : it just acquires a copy of input, and distributes jobs to internal worker threads, flush whatever is available,
	* and then immediately returns, just indicating that there is some data remaining to be flushed.
	* The function nonetheless guarantees forward progress : it will return only after it reads or write at least 1+ byte.
	* - Exception : if the first call requests a ZSTD_e_end directive and provides enough dstCapacity, the function delegates to ZSTD_compress2() which is always blocking.
	* - @return provides a minimum amount of data remaining to be flushed from internal buffers
	* or an error code, which can be tested using ZSTD_isError().
	* if @return != 0, flush is not fully completed, there is still some data left within internal buffers.
	* This is useful for ZSTD_e_flush, since in this case more flushes are necessary to empty all buffers.
	* For ZSTD_e_end, @return == 0 when internal buffers are fully flushed and frame is completed.
	* - after a ZSTD_e_end directive, if internal buffer is not fully flushed (@return != 0),
	* only ZSTD_e_end or ZSTD_e_flush operations are allowed.
	* Before starting a new compression job, or changing compression parameters,
	* it is required to fully flush internal buffers.
	*/
	ZSTDLIB_API size_t ZSTD_compressStream2( ZSTD_CCtx* cctx,
	ZSTD_outBuffer* output,
	ZSTD_inBuffer* input,
	ZSTD_EndDirective endOp);


	/* These buffer sizes are softly recommended.
	* They are not required : ZSTD_compressStream*() happily accepts any buffer size, for both input and output.
	* Respecting the recommended size just makes it a bit easier for ZSTD_compressStream*(),
	* reducing the amount of memory shuffling and buffering, resulting in minor performance savings.
	*
	* However, note that these recommendations are from the perspective of a C caller program.
	* If the streaming interface is invoked from some other language,
	* especially managed ones such as Java or Go, through a foreign function interface such as jni or cgo,
	* a major performance rule is to reduce crossing such interface to an absolute minimum.
	* It's not rare that performance ends being spent more into the interface, rather than compression itself.
	* In which cases, prefer using large buffers, as large as practical,
	* for both input and output, to reduce the nb of roundtrips.
	*/
	ZSTDLIB_API size_t ZSTD_CStreamInSize(void); /*< recommended size for input buffer /
	ZSTDLIB_API size_t ZSTD_CStreamOutSize(void); /*< recommended size for output buffer. Guarantee to successfully flush at least one complete compressed block. /


	/* *****************************************************************************
	* This following is a legacy streaming API.
	* It can be replaced by ZSTD_CCtx_reset() and ZSTD_compressStream2().
	* It is redundant, but remains fully supported.
	* Advanced parameters and dictionary compression can only be used through the
	* new API.
	******************************************************************************/

	/*!
	* Equivalent to:
	*
	* ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
	* ZSTD_CCtx_refCDict(zcs, NULL); // clear the dictionary (if any)
	* ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel);
	*/
	ZSTDLIB_API size_t ZSTD_initCStream(ZSTD_CStream* zcs, int compressionLevel);
	/*!
	* Alternative for ZSTD_compressStream2(zcs, output, input, ZSTD_e_continue).
	* NOTE: The return value is different. ZSTD_compressStream() returns a hint for
	* the next read size (if non-zero and not an error). ZSTD_compressStream2()
	* returns the minimum nb of bytes left to flush (if non-zero and not an error).
	*/
	ZSTDLIB_API size_t ZSTD_compressStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output, ZSTD_inBuffer* input);
	/! Equivalent to ZSTD_compressStream2(zcs, output, &emptyInput, ZSTD_e_flush). /
	ZSTDLIB_API size_t ZSTD_flushStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output);
	/! Equivalent to ZSTD_compressStream2(zcs, output, &emptyInput, ZSTD_e_end). /
	ZSTDLIB_API size_t ZSTD_endStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output);


	/-**************************************************************************
	* Streaming decompression - HowTo
	*
	* A ZSTD_DStream object is required to track streaming operations.
	* Use ZSTD_createDStream() and ZSTD_freeDStream() to create/release resources.
	* ZSTD_DStream objects can be re-used multiple times.
	*
	* Use ZSTD_initDStream() to start a new decompression operation.
	* @return : recommended first input size
	* Alternatively, use advanced API to set specific properties.
	*
	* Use ZSTD_decompressStream() repetitively to consume your input.
	* The function will update both `pos` fields.
	* If `input.pos < input.size`, some input has not been consumed.
	* It's up to the caller to present again remaining data.
	* The function tries to flush all data decoded immediately, respecting output buffer size.
	* If `output.pos < output.size`, decoder has flushed everything it could.
	* But if `output.pos == output.size`, there might be some data left within internal buffers.,
	* In which case, call ZSTD_decompressStream() again to flush whatever remains in the buffer.
	* Note : with no additional input provided, amount of data flushed is necessarily <= ZSTD_BLOCKSIZE_MAX.
	* @return : 0 when a frame is completely decoded and fully flushed,
	* or an error code, which can be tested using ZSTD_isError(),
	* or any other value > 0, which means there is still some decoding or flushing to do to complete current frame :
	* the return value is a suggested next input size (just a hint for better latency)
	* that will never request more than the remaining frame size.
	* *******************************************************************************/

	typedef ZSTD_DCtx ZSTD_DStream; /*< DCtx and DStream are now effectively same object (>= v1.3.0) /
	/* For compatibility with versions <= v1.2.0, prefer differentiating them. */
	/===== ZSTD_DStream management functions =====/
	ZSTDLIB_API ZSTD_DStream* ZSTD_createDStream(void);
	ZSTDLIB_API size_t ZSTD_freeDStream(ZSTD_DStream* zds);

	/===== Streaming decompression functions =====/

	/* This function is redundant with the advanced API and equivalent to:
	*
	* ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
	* ZSTD_DCtx_refDDict(zds, NULL);
	*/
	ZSTDLIB_API size_t ZSTD_initDStream(ZSTD_DStream* zds);

	ZSTDLIB_API size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inBuffer* input);

	ZSTDLIB_API size_t ZSTD_DStreamInSize(void); /!< recommended size for input buffer /
	ZSTDLIB_API size_t ZSTD_DStreamOutSize(void); /!< recommended size for output buffer. Guarantee to successfully flush at least one complete block in all circumstances. /


	/**************************
	* Simple dictionary API
	***************************/
	/*! ZSTD_compress_usingDict() :
	* Compression at an explicit compression level using a Dictionary.
	* A dictionary can be any arbitrary data segment (also called a prefix),
	* or a buffer with specified information (see dictBuilder/zdict.h).
	* Note : This function loads the dictionary, resulting in significant startup delay.
	* It's intended for a dictionary used only once.
	* Note 2 : When `dict == NULL \|\| dictSize < 8` no dictionary is used. */
	ZSTDLIB_API size_t ZSTD_compress_usingDict(ZSTD_CCtx* ctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict,size_t dictSize,
	int compressionLevel);

	/*! ZSTD_decompress_usingDict() :
	* Decompression using a known Dictionary.
	* Dictionary must be identical to the one used during compression.
	* Note : This function loads the dictionary, resulting in significant startup delay.
	* It's intended for a dictionary used only once.
	* Note : When `dict == NULL \|\| dictSize < 8` no dictionary is used. */
	ZSTDLIB_API size_t ZSTD_decompress_usingDict(ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict,size_t dictSize);


	/***********************************
	* Bulk processing dictionary API
	**********************************/
	typedef struct ZSTD_CDict_s ZSTD_CDict;

	/*! ZSTD_createCDict() :
	* When compressing multiple messages or blocks using the same dictionary,
	* it's recommended to digest the dictionary only once, since it's a costly operation.
	* ZSTD_createCDict() will create a state from digesting a dictionary.
	* The resulting state can be used for future compression operations with very limited startup cost.
	* ZSTD_CDict can be created once and shared by multiple threads concurrently, since its usage is read-only.
	* @dictBuffer can be released after ZSTD_CDict creation, because its content is copied within CDict.
	* Note 1 : Consider experimental function `ZSTD_createCDict_byReference()` if you prefer to not duplicate @dictBuffer content.
	* Note 2 : A ZSTD_CDict can be created from an empty @dictBuffer,
	* in which case the only thing that it transports is the @compressionLevel.
	* This can be useful in a pipeline featuring ZSTD_compress_usingCDict() exclusively,
	* expecting a ZSTD_CDict parameter with any data, including those without a known dictionary. */
	ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict(const void* dictBuffer, size_t dictSize,
	int compressionLevel);

	/*! ZSTD_freeCDict() :
	* Function frees memory allocated by ZSTD_createCDict(). */
	ZSTDLIB_API size_t ZSTD_freeCDict(ZSTD_CDict* CDict);

	/*! ZSTD_compress_usingCDict() :
	* Compression using a digested Dictionary.
	* Recommended when same dictionary is used multiple times.
	* Note : compression level is _decided at dictionary creation time_,
	* and frame parameters are hardcoded (dictID=yes, contentSize=yes, checksum=no) */
	ZSTDLIB_API size_t ZSTD_compress_usingCDict(ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const ZSTD_CDict* cdict);


	typedef struct ZSTD_DDict_s ZSTD_DDict;

	/*! ZSTD_createDDict() :
	* Create a digested dictionary, ready to start decompression operation without startup delay.
	* dictBuffer can be released after DDict creation, as its content is copied inside DDict. */
	ZSTDLIB_API ZSTD_DDict* ZSTD_createDDict(const void* dictBuffer, size_t dictSize);

	/*! ZSTD_freeDDict() :
	* Function frees memory allocated with ZSTD_createDDict() */
	ZSTDLIB_API size_t ZSTD_freeDDict(ZSTD_DDict* ddict);

	/*! ZSTD_decompress_usingDDict() :
	* Decompression using a digested Dictionary.
	* Recommended when same dictionary is used multiple times. */
	ZSTDLIB_API size_t ZSTD_decompress_usingDDict(ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const ZSTD_DDict* ddict);


	/********************************
	* Dictionary helper functions
	*******************************/

	/*! ZSTD_getDictID_fromDict() :
	* Provides the dictID stored within dictionary.
	* if @return == 0, the dictionary is not conformant with Zstandard specification.
	* It can still be loaded, but as a content-only dictionary. */
	ZSTDLIB_API unsigned ZSTD_getDictID_fromDict(const void* dict, size_t dictSize);

	/*! ZSTD_getDictID_fromDDict() :
	* Provides the dictID of the dictionary loaded into `ddict`.
	* If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
	* Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
	ZSTDLIB_API unsigned ZSTD_getDictID_fromDDict(const ZSTD_DDict* ddict);

	/*! ZSTD_getDictID_fromFrame() :
	* Provides the dictID required to decompressed the frame stored within `src`.
	* If @return == 0, the dictID could not be decoded.
	* This could for one of the following reasons :
	* - The frame does not require a dictionary to be decoded (most common case).
	* - The frame was built with dictID intentionally removed. Whatever dictionary is necessary is a hidden information.
	* Note : this use case also happens when using a non-conformant dictionary.
	* - `srcSize` is too small, and as a result, the frame header could not be decoded (only possible if `srcSize < ZSTD_FRAMEHEADERSIZE_MAX`).
	* - This is not a Zstandard frame.
	* When identifying the exact failure cause, it's possible to use ZSTD_getFrameHeader(), which will provide a more precise error code. */
	ZSTDLIB_API unsigned ZSTD_getDictID_fromFrame(const void* src, size_t srcSize);


	/*******************************************************************************
	* Advanced dictionary and prefix API
	*
	* This API allows dictionaries to be used with ZSTD_compress2(),
	* ZSTD_compressStream2(), and ZSTD_decompress(). Dictionaries are sticky, and
	* only reset with the context is reset with ZSTD_reset_parameters or
	* ZSTD_reset_session_and_parameters. Prefixes are single-use.
	******************************************************************************/


	/*! ZSTD_CCtx_loadDictionary() :
	* Create an internal CDict from `dict` buffer.
	* Decompression will have to use same dictionary.
	* @result : 0, or an error code (which can be tested with ZSTD_isError()).
	* Special: Loading a NULL (or 0-size) dictionary invalidates previous dictionary,
	* meaning "return to no-dictionary mode".
	* Note 1 : Dictionary is sticky, it will be used for all future compressed frames.
	* To return to "no-dictionary" situation, load a NULL dictionary (or reset parameters).
	* Note 2 : Loading a dictionary involves building tables.
	* It's also a CPU consuming operation, with non-negligible impact on latency.
	* Tables are dependent on compression parameters, and for this reason,
	* compression parameters can no longer be changed after loading a dictionary.
	* Note 3 :`dict` content will be copied internally.
	* Use experimental ZSTD_CCtx_loadDictionary_byReference() to reference content instead.
	* In such a case, dictionary buffer must outlive its users.
	* Note 4 : Use ZSTD_CCtx_loadDictionary_advanced()
	* to precisely select how dictionary content must be interpreted. */
	ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, size_t dictSize);

	/*! ZSTD_CCtx_refCDict() :
	* Reference a prepared dictionary, to be used for all next compressed frames.
	* Note that compression parameters are enforced from within CDict,
	* and supersede any compression parameter previously set within CCtx.
	* The parameters ignored are labled as "superseded-by-cdict" in the ZSTD_cParameter enum docs.
	* The ignored parameters will be used again if the CCtx is returned to no-dictionary mode.
	* The dictionary will remain valid for future compressed frames using same CCtx.
	* @result : 0, or an error code (which can be tested with ZSTD_isError()).
	* Special : Referencing a NULL CDict means "return to no-dictionary mode".
	* Note 1 : Currently, only one dictionary can be managed.
	* Referencing a new dictionary effectively "discards" any previous one.
	* Note 2 : CDict is just referenced, its lifetime must outlive its usage within CCtx. */
	ZSTDLIB_API size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict);

	/*! ZSTD_CCtx_refPrefix() :
	* Reference a prefix (single-usage dictionary) for next compressed frame.
	* A prefix is only used once. Tables are discarded at end of frame (ZSTD_e_end).
	* Decompression will need same prefix to properly regenerate data.
	* Compressing with a prefix is similar in outcome as performing a diff and compressing it,
	* but performs much faster, especially during decompression (compression speed is tunable with compression level).
	* @result : 0, or an error code (which can be tested with ZSTD_isError()).
	* Special: Adding any prefix (including NULL) invalidates any previous prefix or dictionary
	* Note 1 : Prefix buffer is referenced. It must outlive compression.
	* Its content must remain unmodified during compression.
	* Note 2 : If the intention is to diff some large src data blob with some prior version of itself,
	* ensure that the window size is large enough to contain the entire source.
	* See ZSTD_c_windowLog.
	* Note 3 : Referencing a prefix involves building tables, which are dependent on compression parameters.
	* It's a CPU consuming operation, with non-negligible impact on latency.
	* If there is a need to use the same prefix multiple times, consider loadDictionary instead.
	* Note 4 : By default, the prefix is interpreted as raw content (ZSTD_dct_rawContent).
	* Use experimental ZSTD_CCtx_refPrefix_advanced() to alter dictionary interpretation. */
	ZSTDLIB_API size_t ZSTD_CCtx_refPrefix(ZSTD_CCtx* cctx,
	const void* prefix, size_t prefixSize);

	/*! ZSTD_DCtx_loadDictionary() :
	* Create an internal DDict from dict buffer,
	* to be used to decompress next frames.
	* The dictionary remains valid for all future frames, until explicitly invalidated.
	* @result : 0, or an error code (which can be tested with ZSTD_isError()).
	* Special : Adding a NULL (or 0-size) dictionary invalidates any previous dictionary,
	* meaning "return to no-dictionary mode".
	* Note 1 : Loading a dictionary involves building tables,
	* which has a non-negligible impact on CPU usage and latency.
	* It's recommended to "load once, use many times", to amortize the cost
	* Note 2 :`dict` content will be copied internally, so `dict` can be released after loading.
	* Use ZSTD_DCtx_loadDictionary_byReference() to reference dictionary content instead.
	* Note 3 : Use ZSTD_DCtx_loadDictionary_advanced() to take control of
	* how dictionary content is loaded and interpreted.
	*/
	ZSTDLIB_API size_t ZSTD_DCtx_loadDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize);

	/*! ZSTD_DCtx_refDDict() :
	* Reference a prepared dictionary, to be used to decompress next frames.
	* The dictionary remains active for decompression of future frames using same DCtx.
	* @result : 0, or an error code (which can be tested with ZSTD_isError()).
	* Note 1 : Currently, only one dictionary can be managed.
	* Referencing a new dictionary effectively "discards" any previous one.
	* Special: referencing a NULL DDict means "return to no-dictionary mode".
	* Note 2 : DDict is just referenced, its lifetime must outlive its usage from DCtx.
	*/
	ZSTDLIB_API size_t ZSTD_DCtx_refDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict);

	/*! ZSTD_DCtx_refPrefix() :
	* Reference a prefix (single-usage dictionary) to decompress next frame.
	* This is the reverse operation of ZSTD_CCtx_refPrefix(),
	* and must use the same prefix as the one used during compression.
	* Prefix is only used once. Reference is discarded at end of frame.
	* End of frame is reached when ZSTD_decompressStream() returns 0.
	* @result : 0, or an error code (which can be tested with ZSTD_isError()).
	* Note 1 : Adding any prefix (including NULL) invalidates any previously set prefix or dictionary
	* Note 2 : Prefix buffer is referenced. It must outlive decompression.
	* Prefix buffer must remain unmodified up to the end of frame,
	* reached when ZSTD_decompressStream() returns 0.
	* Note 3 : By default, the prefix is treated as raw content (ZSTD_dct_rawContent).
	* Use ZSTD_CCtx_refPrefix_advanced() to alter dictMode (Experimental section)
	* Note 4 : Referencing a raw content prefix has almost no cpu nor memory cost.
	* A full dictionary is more costly, as it requires building tables.
	*/
	ZSTDLIB_API size_t ZSTD_DCtx_refPrefix(ZSTD_DCtx* dctx,
	const void* prefix, size_t prefixSize);

	/* === Memory management === */

	/! ZSTD_sizeof_() :
	* These functions give the _current_ memory usage of selected object.
	* Note that object memory usage can evolve (increase or decrease) over time. */
	ZSTDLIB_API size_t ZSTD_sizeof_CCtx(const ZSTD_CCtx* cctx);
	ZSTDLIB_API size_t ZSTD_sizeof_DCtx(const ZSTD_DCtx* dctx);
	ZSTDLIB_API size_t ZSTD_sizeof_CStream(const ZSTD_CStream* zcs);
	ZSTDLIB_API size_t ZSTD_sizeof_DStream(const ZSTD_DStream* zds);
	ZSTDLIB_API size_t ZSTD_sizeof_CDict(const ZSTD_CDict* cdict);
	ZSTDLIB_API size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict);

	#endif /* ZSTD_H_235446 */


	/* **************************************************************************************
	* ADVANCED AND EXPERIMENTAL FUNCTIONS
	****************************************************************************************
	* The definitions in the following section are considered experimental.
	* They are provided for advanced scenarios.
	* They should never be used with a dynamic library, as prototypes may change in the future.
	* Use them only in association with static linking.
	* ***************************************************************************************/

	#if defined(ZSTD_STATIC_LINKING_ONLY) && !defined(ZSTD_H_ZSTD_STATIC_LINKING_ONLY)
	#define ZSTD_H_ZSTD_STATIC_LINKING_ONLY

	/****************************************************************************************
	* experimental API (static linking only)
	****************************************************************************************
	* The following symbols and constants
	* are not planned to join "stable API" status in the near future.
	* They can still change in future versions.
	* Some of them are planned to remain in the static_only section indefinitely.
	* Some of them might be removed in the future (especially when redundant with existing stable functions)
	* ***************************************************************************************/

	#define ZSTD_FRAMEHEADERSIZE_PREFIX(format) ((format) == ZSTD_f_zstd1 ? 5 : 1) /* minimum input size required to query frame header size */
	#define ZSTD_FRAMEHEADERSIZE_MIN(format) ((format) == ZSTD_f_zstd1 ? 6 : 2)
	#define ZSTD_FRAMEHEADERSIZE_MAX 18 /* can be useful for static allocation */
	#define ZSTD_SKIPPABLEHEADERSIZE 8

	/* compression parameter bounds */
	#define ZSTD_WINDOWLOG_MAX_32 30
	#define ZSTD_WINDOWLOG_MAX_64 31
	#define ZSTD_WINDOWLOG_MAX ((int)(sizeof(size_t) == 4 ? ZSTD_WINDOWLOG_MAX_32 : ZSTD_WINDOWLOG_MAX_64))
	#define ZSTD_WINDOWLOG_MIN 10
	#define ZSTD_HASHLOG_MAX ((ZSTD_WINDOWLOG_MAX < 30) ? ZSTD_WINDOWLOG_MAX : 30)
	#define ZSTD_HASHLOG_MIN 6
	#define ZSTD_CHAINLOG_MAX_32 29
	#define ZSTD_CHAINLOG_MAX_64 30
	#define ZSTD_CHAINLOG_MAX ((int)(sizeof(size_t) == 4 ? ZSTD_CHAINLOG_MAX_32 : ZSTD_CHAINLOG_MAX_64))
	#define ZSTD_CHAINLOG_MIN ZSTD_HASHLOG_MIN
	#define ZSTD_SEARCHLOG_MAX (ZSTD_WINDOWLOG_MAX-1)
	#define ZSTD_SEARCHLOG_MIN 1
	#define ZSTD_MINMATCH_MAX 7 /* only for ZSTD_fast, other strategies are limited to 6 */
	#define ZSTD_MINMATCH_MIN 3 /* only for ZSTD_btopt+, faster strategies are limited to 4 */
	#define ZSTD_TARGETLENGTH_MAX ZSTD_BLOCKSIZE_MAX
	#define ZSTD_TARGETLENGTH_MIN 0 /* note : comparing this constant to an unsigned results in a tautological test */
	#define ZSTD_STRATEGY_MIN ZSTD_fast
	#define ZSTD_STRATEGY_MAX ZSTD_btultra2


	#define ZSTD_OVERLAPLOG_MIN 0
	#define ZSTD_OVERLAPLOG_MAX 9

	#define ZSTD_WINDOWLOG_LIMIT_DEFAULT 27 /* by default, the streaming decoder will refuse any frame
	* requiring larger than (1<<ZSTD_WINDOWLOG_LIMIT_DEFAULT) window size,
	* to preserve host's memory from unreasonable requirements.
	* This limit can be overridden using ZSTD_DCtx_setParameter(,ZSTD_d_windowLogMax,).
	* The limit does not apply for one-pass decoders (such as ZSTD_decompress()), since no additional memory is allocated */


	/* LDM parameter bounds */
	#define ZSTD_LDM_HASHLOG_MIN ZSTD_HASHLOG_MIN
	#define ZSTD_LDM_HASHLOG_MAX ZSTD_HASHLOG_MAX
	#define ZSTD_LDM_MINMATCH_MIN 4
	#define ZSTD_LDM_MINMATCH_MAX 4096
	#define ZSTD_LDM_BUCKETSIZELOG_MIN 1
	#define ZSTD_LDM_BUCKETSIZELOG_MAX 8
	#define ZSTD_LDM_HASHRATELOG_MIN 0
	#define ZSTD_LDM_HASHRATELOG_MAX (ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN)

	/* Advanced parameter bounds */
	#define ZSTD_TARGETCBLOCKSIZE_MIN 64
	#define ZSTD_TARGETCBLOCKSIZE_MAX ZSTD_BLOCKSIZE_MAX
	#define ZSTD_SRCSIZEHINT_MIN 0
	#define ZSTD_SRCSIZEHINT_MAX INT_MAX

	/* internal */
	#define ZSTD_HASHLOG3_MAX 17


	/* --- Advanced types --- */

	typedef struct ZSTD_CCtx_params_s ZSTD_CCtx_params;

	typedef struct {
	unsigned int matchPos; /* Match pos in dst */
	/* If seqDef.offset > 3, then this is seqDef.offset - 3
	* If seqDef.offset < 3, then this is the corresponding repeat offset
	* But if seqDef.offset < 3 and litLength == 0, this is the
	* repeat offset before the corresponding repeat offset
	* And if seqDef.offset == 3 and litLength == 0, this is the
	* most recent repeat offset - 1
	*/
	unsigned int offset;
	unsigned int litLength; /* Literal length */
	unsigned int matchLength; /* Match length */
	/* 0 when seq not rep and seqDef.offset otherwise
	* when litLength == 0 this will be <= 4, otherwise <= 3 like normal
	*/
	unsigned int rep;
	} ZSTD_Sequence;

	typedef struct {
	unsigned windowLog; /*< largest match distance : larger == more compression, more memory needed during decompression /
	unsigned chainLog; /*< fully searched segment : larger == more compression, slower, more memory (useless for fast) /
	unsigned hashLog; /*< dispatch table : larger == faster, more memory /
	unsigned searchLog; /*< nb of searches : larger == more compression, slower /
	unsigned minMatch; /*< match length searched : larger == faster decompression, sometimes less compression /
	unsigned targetLength; /*< acceptable match size for optimal parser (only) : larger == more compression, slower /
	ZSTD_strategy strategy; /*< see ZSTD_strategy definition above /
	} ZSTD_compressionParameters;

	typedef struct {
	int contentSizeFlag; /*< 1: content size will be in frame header (when known) /
	int checksumFlag; /*< 1: generate a 32-bits checksum using XXH64 algorithm at end of frame, for error detection /
	int noDictIDFlag; /*< 1: no dictID will be saved into frame header (dictID is only useful for dictionary compression) /
	} ZSTD_frameParameters;

	typedef struct {
	ZSTD_compressionParameters cParams;
	ZSTD_frameParameters fParams;
	} ZSTD_parameters;

	typedef enum {
	ZSTD_dct_auto = 0, /* dictionary is "full" when starting with ZSTD_MAGIC_DICTIONARY, otherwise it is "rawContent" */
	ZSTD_dct_rawContent = 1, /* ensures dictionary is always loaded as rawContent, even if it starts with ZSTD_MAGIC_DICTIONARY */
	ZSTD_dct_fullDict = 2 /* refuses to load a dictionary if it does not respect Zstandard's specification, starting with ZSTD_MAGIC_DICTIONARY */
	} ZSTD_dictContentType_e;

	typedef enum {
	ZSTD_dlm_byCopy = 0, /*< Copy dictionary content internally /
	ZSTD_dlm_byRef = 1 /*< Reference dictionary content -- the dictionary buffer must outlive its users. /
	} ZSTD_dictLoadMethod_e;

	typedef enum {
	ZSTD_f_zstd1 = 0, /* zstd frame format, specified in zstd_compression_format.md (default) */
	ZSTD_f_zstd1_magicless = 1 /* Variant of zstd frame format, without initial 4-bytes magic number.
	* Useful to save 4 bytes per generated frame.
	* Decoder cannot recognise automatically this format, requiring this instruction. */
	} ZSTD_format_e;

	typedef enum {
	/* Note: this enum and the behavior it controls are effectively internal
	* implementation details of the compressor. They are expected to continue
	* to evolve and should be considered only in the context of extremely
	* advanced performance tuning.
	*
	* Zstd currently supports the use of a CDict in three ways:
	*
	* - The contents of the CDict can be copied into the working context. This
	* means that the compression can search both the dictionary and input
	* while operating on a single set of internal tables. This makes
	* the compression faster per-byte of input. However, the initial copy of
	* the CDict's tables incurs a fixed cost at the beginning of the
	* compression. For small compressions (< 8 KB), that copy can dominate
	* the cost of the compression.
	*
	* - The CDict's tables can be used in-place. In this model, compression is
	* slower per input byte, because the compressor has to search two sets of
	* tables. However, this model incurs no start-up cost (as long as the
	* working context's tables can be reused). For small inputs, this can be
	* faster than copying the CDict's tables.
	*
	* - The CDict's tables are not used at all, and instead we use the working
	* context alone to reload the dictionary and use params based on the source
	* size. See ZSTD_compress_insertDictionary() and ZSTD_compress_usingDict().
	* This method is effective when the dictionary sizes are very small relative
	* to the input size, and the input size is fairly large to begin with.
	*
	* Zstd has a simple internal heuristic that selects which strategy to use
	* at the beginning of a compression. However, if experimentation shows that
	* Zstd is making poor choices, it is possible to override that choice with
	* this enum.
	*/
	ZSTD_dictDefaultAttach = 0, /* Use the default heuristic. */
	ZSTD_dictForceAttach = 1, /* Never copy the dictionary. */
	ZSTD_dictForceCopy = 2, /* Always copy the dictionary. */
	ZSTD_dictForceLoad = 3 /* Always reload the dictionary */
	} ZSTD_dictAttachPref_e;

	typedef enum {
	ZSTD_lcm_auto = 0, /**< Automatically determine the compression mode based on the compression level.
	* Negative compression levels will be uncompressed, and positive compression
	* levels will be compressed. */
	ZSTD_lcm_huffman = 1, /**< Always attempt Huffman compression. Uncompressed literals will still be
	* emitted if Huffman compression is not profitable. */
	ZSTD_lcm_uncompressed = 2 /*< Always emit uncompressed literals. /
	} ZSTD_literalCompressionMode_e;


	/***************************************
	* Frame size functions
	***************************************/

	/*! ZSTD_findDecompressedSize() :
	* `src` should point to the start of a series of ZSTD encoded and/or skippable frames
	* `srcSize` must be the _exact_ size of this series
	* (i.e. there should be a frame boundary at `src + srcSize`)
	* @return : - decompressed size of all data in all successive frames
	* - if the decompressed size cannot be determined: ZSTD_CONTENTSIZE_UNKNOWN
	* - if an error occurred: ZSTD_CONTENTSIZE_ERROR
	*
	* note 1 : decompressed size is an optional field, that may not be present, especially in streaming mode.
	* When `return==ZSTD_CONTENTSIZE_UNKNOWN`, data to decompress could be any size.
	* In which case, it's necessary to use streaming mode to decompress data.
	* note 2 : decompressed size is always present when compression is done with ZSTD_compress()
	* note 3 : decompressed size can be very large (64-bits value),
	* potentially larger than what local system can handle as a single memory segment.
	* In which case, it's necessary to use streaming mode to decompress data.
	* note 4 : If source is untrusted, decompressed size could be wrong or intentionally modified.
	* Always ensure result fits within application's authorized limits.
	* Each application can set its own limits.
	* note 5 : ZSTD_findDecompressedSize handles multiple frames, and so it must traverse the input to
	* read each contained frame header. This is fast as most of the data is skipped,
	* however it does mean that all frame data must be present and valid. */
	ZSTDLIB_API unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize);

	/*! ZSTD_decompressBound() :
	* `src` should point to the start of a series of ZSTD encoded and/or skippable frames
	* `srcSize` must be the _exact_ size of this series
	* (i.e. there should be a frame boundary at `src + srcSize`)
	* @return : - upper-bound for the decompressed size of all data in all successive frames
	* - if an error occured: ZSTD_CONTENTSIZE_ERROR
	*
	* note 1 : an error can occur if `src` contains an invalid or incorrectly formatted frame.
	* note 2 : the upper-bound is exact when the decompressed size field is available in every ZSTD encoded frame of `src`.
	* in this case, `ZSTD_findDecompressedSize` and `ZSTD_decompressBound` return the same value.
	* note 3 : when the decompressed size field isn't available, the upper-bound for that frame is calculated by:
	* upper-bound = # blocks * min(128 KB, Window_Size)
	*/
	ZSTDLIB_API unsigned long long ZSTD_decompressBound(const void* src, size_t srcSize);

	/*! ZSTD_frameHeaderSize() :
	* srcSize must be >= ZSTD_FRAMEHEADERSIZE_PREFIX.
	* @return : size of the Frame Header,
	* or an error code (if srcSize is too small) */
	ZSTDLIB_API size_t ZSTD_frameHeaderSize(const void* src, size_t srcSize);

	/*! ZSTD_getSequences() :
	* Extract sequences from the sequence store
	* zc can be used to insert custom compression params.
	* This function invokes ZSTD_compress2
	* @return : number of sequences extracted
	*/
	ZSTDLIB_API size_t ZSTD_getSequences(ZSTD_CCtx* zc, ZSTD_Sequence* outSeqs,
	size_t outSeqsSize, const void* src, size_t srcSize);


	/***************************************
	* Memory management
	***************************************/

	/! ZSTD_estimate() :
	* These functions make it possible to estimate memory usage
	* of a future {D,C}Ctx, before its creation.
	*
	* ZSTD_estimateCCtxSize() will provide a memory budget large enough
	* for any compression level up to selected one.
	* Note : Unlike ZSTD_estimateCStreamSize*(), this estimate
	* does not include space for a window buffer.
	* Therefore, the estimation is only guaranteed for single-shot compressions, not streaming.
	* The estimate will assume the input may be arbitrarily large,
	* which is the worst case.
	*
	* When srcSize can be bound by a known and rather "small" value,
	* this fact can be used to provide a tighter estimation
	* because the CCtx compression context will need less memory.
	* This tighter estimation can be provided by more advanced functions
	* ZSTD_estimateCCtxSize_usingCParams(), which can be used in tandem with ZSTD_getCParams(),
	* and ZSTD_estimateCCtxSize_usingCCtxParams(), which can be used in tandem with ZSTD_CCtxParams_setParameter().
	* Both can be used to estimate memory using custom compression parameters and arbitrary srcSize limits.
	*
	* Note 2 : only single-threaded compression is supported.
	* ZSTD_estimateCCtxSize_usingCCtxParams() will return an error code if ZSTD_c_nbWorkers is >= 1.
	*/
	ZSTDLIB_API size_t ZSTD_estimateCCtxSize(int compressionLevel);
	ZSTDLIB_API size_t ZSTD_estimateCCtxSize_usingCParams(ZSTD_compressionParameters cParams);
	ZSTDLIB_API size_t ZSTD_estimateCCtxSize_usingCCtxParams(const ZSTD_CCtx_params* params);
	ZSTDLIB_API size_t ZSTD_estimateDCtxSize(void);

	/*! ZSTD_estimateCStreamSize() :
	* ZSTD_estimateCStreamSize() will provide a budget large enough for any compression level up to selected one.
	* It will also consider src size to be arbitrarily "large", which is worst case.
	* If srcSize is known to always be small, ZSTD_estimateCStreamSize_usingCParams() can provide a tighter estimation.
	* ZSTD_estimateCStreamSize_usingCParams() can be used in tandem with ZSTD_getCParams() to create cParams from compressionLevel.
	* ZSTD_estimateCStreamSize_usingCCtxParams() can be used in tandem with ZSTD_CCtxParams_setParameter(). Only single-threaded compression is supported. This function will return an error code if ZSTD_c_nbWorkers is >= 1.
	* Note : CStream size estimation is only correct for single-threaded compression.
	* ZSTD_DStream memory budget depends on window Size.
	* This information can be passed manually, using ZSTD_estimateDStreamSize,
	* or deducted from a valid frame Header, using ZSTD_estimateDStreamSize_fromFrame();
	* Note : if streaming is init with function ZSTD_init?Stream_usingDict(),
	* an internal ?Dict will be created, which additional size is not estimated here.
	* In this case, get total size by adding ZSTD_estimate?DictSize */
	ZSTDLIB_API size_t ZSTD_estimateCStreamSize(int compressionLevel);
	ZSTDLIB_API size_t ZSTD_estimateCStreamSize_usingCParams(ZSTD_compressionParameters cParams);
	ZSTDLIB_API size_t ZSTD_estimateCStreamSize_usingCCtxParams(const ZSTD_CCtx_params* params);
	ZSTDLIB_API size_t ZSTD_estimateDStreamSize(size_t windowSize);
	ZSTDLIB_API size_t ZSTD_estimateDStreamSize_fromFrame(const void* src, size_t srcSize);

	/*! ZSTD_estimate?DictSize() :
	* ZSTD_estimateCDictSize() will bet that src size is relatively "small", and content is copied, like ZSTD_createCDict().
	* ZSTD_estimateCDictSize_advanced() makes it possible to control compression parameters precisely, like ZSTD_createCDict_advanced().
	* Note : dictionaries created by reference (`ZSTD_dlm_byRef`) are logically smaller.
	*/
	ZSTDLIB_API size_t ZSTD_estimateCDictSize(size_t dictSize, int compressionLevel);
	ZSTDLIB_API size_t ZSTD_estimateCDictSize_advanced(size_t dictSize, ZSTD_compressionParameters cParams, ZSTD_dictLoadMethod_e dictLoadMethod);
	ZSTDLIB_API size_t ZSTD_estimateDDictSize(size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod);

	/! ZSTD_initStatic() :
	* Initialize an object using a pre-allocated fixed-size buffer.
	* workspace: The memory area to emplace the object into.
	* Provided pointer must be 8-bytes aligned.
	* Buffer must outlive object.
	* workspaceSize: Use ZSTD_estimate*Size() to determine
	* how large workspace must be to support target scenario.
	* @return : pointer to object (same address as workspace, just different type),
	* or NULL if error (size too small, incorrect alignment, etc.)
	* Note : zstd will never resize nor malloc() when using a static buffer.
	* If the object requires more memory than available,
	* zstd will just error out (typically ZSTD_error_memory_allocation).
	* Note 2 : there is no corresponding "free" function.
	* Since workspace is allocated externally, it must be freed externally too.
	* Note 3 : cParams : use ZSTD_getCParams() to convert a compression level
	* into its associated cParams.
	* Limitation 1 : currently not compatible with internal dictionary creation, triggered by
	* ZSTD_CCtx_loadDictionary(), ZSTD_initCStream_usingDict() or ZSTD_initDStream_usingDict().
	* Limitation 2 : static cctx currently not compatible with multi-threading.
	* Limitation 3 : static dctx is incompatible with legacy support.
	*/
	ZSTDLIB_API ZSTD_CCtx* ZSTD_initStaticCCtx(void* workspace, size_t workspaceSize);
	ZSTDLIB_API ZSTD_CStream* ZSTD_initStaticCStream(void* workspace, size_t workspaceSize); /*< same as ZSTD_initStaticCCtx() /

	ZSTDLIB_API ZSTD_DCtx* ZSTD_initStaticDCtx(void* workspace, size_t workspaceSize);
	ZSTDLIB_API ZSTD_DStream* ZSTD_initStaticDStream(void* workspace, size_t workspaceSize); /*< same as ZSTD_initStaticDCtx() /

	ZSTDLIB_API const ZSTD_CDict* ZSTD_initStaticCDict(
	void* workspace, size_t workspaceSize,
	const void* dict, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod,
	ZSTD_dictContentType_e dictContentType,
	ZSTD_compressionParameters cParams);

	ZSTDLIB_API const ZSTD_DDict* ZSTD_initStaticDDict(
	void* workspace, size_t workspaceSize,
	const void* dict, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod,
	ZSTD_dictContentType_e dictContentType);


	/*! Custom memory allocation :
	* These prototypes make it possible to pass your own allocation/free functions.
	* ZSTD_customMem is provided at creation time, using ZSTD_create*_advanced() variants listed below.
	* All allocation/free operations will be completed using these custom variants instead of regular <stdlib.h> ones.
	*/
	typedef void* (ZSTD_allocFunction) (void opaque, size_t size);
	typedef void (ZSTD_freeFunction) (void opaque, void* address);
	typedef struct { ZSTD_allocFunction customAlloc; ZSTD_freeFunction customFree; void* opaque; } ZSTD_customMem;
	static ZSTD_customMem const ZSTD_defaultCMem = { NULL, NULL, NULL }; /*< this constant defers to stdlib's functions /

	ZSTDLIB_API ZSTD_CCtx* ZSTD_createCCtx_advanced(ZSTD_customMem customMem);
	ZSTDLIB_API ZSTD_CStream* ZSTD_createCStream_advanced(ZSTD_customMem customMem);
	ZSTDLIB_API ZSTD_DCtx* ZSTD_createDCtx_advanced(ZSTD_customMem customMem);
	ZSTDLIB_API ZSTD_DStream* ZSTD_createDStream_advanced(ZSTD_customMem customMem);

	ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict_advanced(const void* dict, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod,
	ZSTD_dictContentType_e dictContentType,
	ZSTD_compressionParameters cParams,
	ZSTD_customMem customMem);

	ZSTDLIB_API ZSTD_DDict* ZSTD_createDDict_advanced(const void* dict, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod,
	ZSTD_dictContentType_e dictContentType,
	ZSTD_customMem customMem);



	/***************************************
	* Advanced compression functions
	***************************************/

	/*! ZSTD_createCDict_byReference() :
	* Create a digested dictionary for compression
	* Dictionary content is just referenced, not duplicated.
	* As a consequence, `dictBuffer` must outlive CDict,
	* and its content must remain unmodified throughout the lifetime of CDict.
	* note: equivalent to ZSTD_createCDict_advanced(), with dictLoadMethod==ZSTD_dlm_byRef */
	ZSTDLIB_API ZSTD_CDict* ZSTD_createCDict_byReference(const void* dictBuffer, size_t dictSize, int compressionLevel);

	/*! ZSTD_getCParams() :
	* @return ZSTD_compressionParameters structure for a selected compression level and estimated srcSize.
	* `estimatedSrcSize` value is optional, select 0 if not known */
	ZSTDLIB_API ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel, unsigned long long estimatedSrcSize, size_t dictSize);

	/*! ZSTD_getParams() :
	* same as ZSTD_getCParams(), but @return a full `ZSTD_parameters` object instead of sub-component `ZSTD_compressionParameters`.
	* All fields of `ZSTD_frameParameters` are set to default : contentSize=1, checksum=0, noDictID=0 */
	ZSTDLIB_API ZSTD_parameters ZSTD_getParams(int compressionLevel, unsigned long long estimatedSrcSize, size_t dictSize);

	/*! ZSTD_checkCParams() :
	* Ensure param values remain within authorized range.
	* @return 0 on success, or an error code (can be checked with ZSTD_isError()) */
	ZSTDLIB_API size_t ZSTD_checkCParams(ZSTD_compressionParameters params);

	/*! ZSTD_adjustCParams() :
	* optimize params for a given `srcSize` and `dictSize`.
	* `srcSize` can be unknown, in which case use ZSTD_CONTENTSIZE_UNKNOWN.
	* `dictSize` must be `0` when there is no dictionary.
	* cPar can be invalid : all parameters will be clamped within valid range in the @return struct.
	* This function never fails (wide contract) */
	ZSTDLIB_API ZSTD_compressionParameters ZSTD_adjustCParams(ZSTD_compressionParameters cPar, unsigned long long srcSize, size_t dictSize);

	/*! ZSTD_compress_advanced() :
	* Note : this function is now DEPRECATED.
	* It can be replaced by ZSTD_compress2(), in combination with ZSTD_CCtx_setParameter() and other parameter setters.
	* This prototype will be marked as deprecated and generate compilation warning on reaching v1.5.x */
	ZSTDLIB_API size_t ZSTD_compress_advanced(ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict,size_t dictSize,
	ZSTD_parameters params);

	/*! ZSTD_compress_usingCDict_advanced() :
	* Note : this function is now REDUNDANT.
	* It can be replaced by ZSTD_compress2(), in combination with ZSTD_CCtx_loadDictionary() and other parameter setters.
	* This prototype will be marked as deprecated and generate compilation warning in some future version */
	ZSTDLIB_API size_t ZSTD_compress_usingCDict_advanced(ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const ZSTD_CDict* cdict,
	ZSTD_frameParameters fParams);


	/*! ZSTD_CCtx_loadDictionary_byReference() :
	* Same as ZSTD_CCtx_loadDictionary(), but dictionary content is referenced, instead of being copied into CCtx.
	* It saves some memory, but also requires that `dict` outlives its usage within `cctx` */
	ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary_byReference(ZSTD_CCtx* cctx, const void* dict, size_t dictSize);

	/*! ZSTD_CCtx_loadDictionary_advanced() :
	* Same as ZSTD_CCtx_loadDictionary(), but gives finer control over
	* how to load the dictionary (by copy ? by reference ?)
	* and how to interpret it (automatic ? force raw mode ? full mode only ?) */
	ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary_advanced(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType);

	/*! ZSTD_CCtx_refPrefix_advanced() :
	* Same as ZSTD_CCtx_refPrefix(), but gives finer control over
	* how to interpret prefix content (automatic ? force raw mode (default) ? full mode only ?) */
	ZSTDLIB_API size_t ZSTD_CCtx_refPrefix_advanced(ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType);

	/* === experimental parameters === */
	/* these parameters can be used with ZSTD_setParameter()
	* they are not guaranteed to remain supported in the future */

	/* Enables rsyncable mode,
	* which makes compressed files more rsync friendly
	* by adding periodic synchronization points to the compressed data.
	* The target average block size is ZSTD_c_jobSize / 2.
	* It's possible to modify the job size to increase or decrease
	* the granularity of the synchronization point.
	* Once the jobSize is smaller than the window size,
	* it will result in compression ratio degradation.
	* NOTE 1: rsyncable mode only works when multithreading is enabled.
	* NOTE 2: rsyncable performs poorly in combination with long range mode,
	* since it will decrease the effectiveness of synchronization points,
	* though mileage may vary.
	* NOTE 3: Rsyncable mode limits maximum compression speed to ~400 MB/s.
	* If the selected compression level is already running significantly slower,
	* the overall speed won't be significantly impacted.
	*/
	#define ZSTD_c_rsyncable ZSTD_c_experimentalParam1

	/* Select a compression format.
	* The value must be of type ZSTD_format_e.
	* See ZSTD_format_e enum definition for details */
	#define ZSTD_c_format ZSTD_c_experimentalParam2

	/* Force back-reference distances to remain < windowSize,
	* even when referencing into Dictionary content (default:0) */
	#define ZSTD_c_forceMaxWindow ZSTD_c_experimentalParam3

	/* Controls whether the contents of a CDict
	* are used in place, or copied into the working context.
	* Accepts values from the ZSTD_dictAttachPref_e enum.
	* See the comments on that enum for an explanation of the feature. */
	#define ZSTD_c_forceAttachDict ZSTD_c_experimentalParam4

	/* Controls how the literals are compressed (default is auto).
	* The value must be of type ZSTD_literalCompressionMode_e.
	* See ZSTD_literalCompressionMode_t enum definition for details.
	*/
	#define ZSTD_c_literalCompressionMode ZSTD_c_experimentalParam5

	/* Tries to fit compressed block size to be around targetCBlockSize.
	* No target when targetCBlockSize == 0.
	* There is no guarantee on compressed block size (default:0) */
	#define ZSTD_c_targetCBlockSize ZSTD_c_experimentalParam6

	/* User's best guess of source size.
	* Hint is not valid when srcSizeHint == 0.
	* There is no guarantee that hint is close to actual source size,
	* but compression ratio may regress significantly if guess considerably underestimates */
	#define ZSTD_c_srcSizeHint ZSTD_c_experimentalParam7

	/*! ZSTD_CCtx_getParameter() :
	* Get the requested compression parameter value, selected by enum ZSTD_cParameter,
	* and store it into int* value.
	* @return : 0, or an error code (which can be tested with ZSTD_isError()).
	*/
	ZSTDLIB_API size_t ZSTD_CCtx_getParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int* value);


	/*! ZSTD_CCtx_params :
	* Quick howto :
	* - ZSTD_createCCtxParams() : Create a ZSTD_CCtx_params structure
	* - ZSTD_CCtxParams_setParameter() : Push parameters one by one into
	* an existing ZSTD_CCtx_params structure.
	* This is similar to
	* ZSTD_CCtx_setParameter().
	* - ZSTD_CCtx_setParametersUsingCCtxParams() : Apply parameters to
	* an existing CCtx.
	* These parameters will be applied to
	* all subsequent frames.
	* - ZSTD_compressStream2() : Do compression using the CCtx.
	* - ZSTD_freeCCtxParams() : Free the memory.
	*
	* This can be used with ZSTD_estimateCCtxSize_advanced_usingCCtxParams()
	* for static allocation of CCtx for single-threaded compression.
	*/
	ZSTDLIB_API ZSTD_CCtx_params* ZSTD_createCCtxParams(void);
	ZSTDLIB_API size_t ZSTD_freeCCtxParams(ZSTD_CCtx_params* params);

	/*! ZSTD_CCtxParams_reset() :
	* Reset params to default values.
	*/
	ZSTDLIB_API size_t ZSTD_CCtxParams_reset(ZSTD_CCtx_params* params);

	/*! ZSTD_CCtxParams_init() :
	* Initializes the compression parameters of cctxParams according to
	* compression level. All other parameters are reset to their default values.
	*/
	ZSTDLIB_API size_t ZSTD_CCtxParams_init(ZSTD_CCtx_params* cctxParams, int compressionLevel);

	/*! ZSTD_CCtxParams_init_advanced() :
	* Initializes the compression and frame parameters of cctxParams according to
	* params. All other parameters are reset to their default values.
	*/
	ZSTDLIB_API size_t ZSTD_CCtxParams_init_advanced(ZSTD_CCtx_params* cctxParams, ZSTD_parameters params);

	/*! ZSTD_CCtxParams_setParameter() :
	* Similar to ZSTD_CCtx_setParameter.
	* Set one compression parameter, selected by enum ZSTD_cParameter.
	* Parameters must be applied to a ZSTD_CCtx using ZSTD_CCtx_setParametersUsingCCtxParams().
	* @result : 0, or an error code (which can be tested with ZSTD_isError()).
	*/
	ZSTDLIB_API size_t ZSTD_CCtxParams_setParameter(ZSTD_CCtx_params* params, ZSTD_cParameter param, int value);

	/*! ZSTD_CCtxParams_getParameter() :
	* Similar to ZSTD_CCtx_getParameter.
	* Get the requested value of one compression parameter, selected by enum ZSTD_cParameter.
	* @result : 0, or an error code (which can be tested with ZSTD_isError()).
	*/
	ZSTDLIB_API size_t ZSTD_CCtxParams_getParameter(ZSTD_CCtx_params* params, ZSTD_cParameter param, int* value);

	/*! ZSTD_CCtx_setParametersUsingCCtxParams() :
	* Apply a set of ZSTD_CCtx_params to the compression context.
	* This can be done even after compression is started,
	* if nbWorkers==0, this will have no impact until a new compression is started.
	* if nbWorkers>=1, new parameters will be picked up at next job,
	* with a few restrictions (windowLog, pledgedSrcSize, nbWorkers, jobSize, and overlapLog are not updated).
	*/
	ZSTDLIB_API size_t ZSTD_CCtx_setParametersUsingCCtxParams(
	ZSTD_CCtx* cctx, const ZSTD_CCtx_params* params);

	/*! ZSTD_compressStream2_simpleArgs() :
	* Same as ZSTD_compressStream2(),
	* but using only integral types as arguments.
	* This variant might be helpful for binders from dynamic languages
	* which have troubles handling structures containing memory pointers.
	*/
	ZSTDLIB_API size_t ZSTD_compressStream2_simpleArgs (
	ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity, size_t* dstPos,
	const void* src, size_t srcSize, size_t* srcPos,
	ZSTD_EndDirective endOp);


	/***************************************
	* Advanced decompression functions
	***************************************/

	/*! ZSTD_isFrame() :
	* Tells if the content of `buffer` starts with a valid Frame Identifier.
	* Note : Frame Identifier is 4 bytes. If `size < 4`, @return will always be 0.
	* Note 2 : Legacy Frame Identifiers are considered valid only if Legacy Support is enabled.
	* Note 3 : Skippable Frame Identifiers are considered valid. */
	ZSTDLIB_API unsigned ZSTD_isFrame(const void* buffer, size_t size);

	/*! ZSTD_createDDict_byReference() :
	* Create a digested dictionary, ready to start decompression operation without startup delay.
	* Dictionary content is referenced, and therefore stays in dictBuffer.
	* It is important that dictBuffer outlives DDict,
	* it must remain read accessible throughout the lifetime of DDict */
	ZSTDLIB_API ZSTD_DDict* ZSTD_createDDict_byReference(const void* dictBuffer, size_t dictSize);

	/*! ZSTD_DCtx_loadDictionary_byReference() :
	* Same as ZSTD_DCtx_loadDictionary(),
	* but references `dict` content instead of copying it into `dctx`.
	* This saves memory if `dict` remains around.,
	* However, it's imperative that `dict` remains accessible (and unmodified) while being used, so it must outlive decompression. */
	ZSTDLIB_API size_t ZSTD_DCtx_loadDictionary_byReference(ZSTD_DCtx* dctx, const void* dict, size_t dictSize);

	/*! ZSTD_DCtx_loadDictionary_advanced() :
	* Same as ZSTD_DCtx_loadDictionary(),
	* but gives direct control over
	* how to load the dictionary (by copy ? by reference ?)
	* and how to interpret it (automatic ? force raw mode ? full mode only ?). */
	ZSTDLIB_API size_t ZSTD_DCtx_loadDictionary_advanced(ZSTD_DCtx* dctx, const void* dict, size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType);

	/*! ZSTD_DCtx_refPrefix_advanced() :
	* Same as ZSTD_DCtx_refPrefix(), but gives finer control over
	* how to interpret prefix content (automatic ? force raw mode (default) ? full mode only ?) */
	ZSTDLIB_API size_t ZSTD_DCtx_refPrefix_advanced(ZSTD_DCtx* dctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType);

	/*! ZSTD_DCtx_setMaxWindowSize() :
	* Refuses allocating internal buffers for frames requiring a window size larger than provided limit.
	* This protects a decoder context from reserving too much memory for itself (potential attack scenario).
	* This parameter is only useful in streaming mode, since no internal buffer is allocated in single-pass mode.
	* By default, a decompression context accepts all window sizes <= (1 << ZSTD_WINDOWLOG_LIMIT_DEFAULT)
	* @return : 0, or an error code (which can be tested using ZSTD_isError()).
	*/
	ZSTDLIB_API size_t ZSTD_DCtx_setMaxWindowSize(ZSTD_DCtx* dctx, size_t maxWindowSize);

	/* ZSTD_d_format
	* experimental parameter,
	* allowing selection between ZSTD_format_e input compression formats
	*/
	#define ZSTD_d_format ZSTD_d_experimentalParam1
	/* ZSTD_d_stableOutBuffer
	* Experimental parameter.
	* Default is 0 == disabled. Set to 1 to enable.
	*
	* Tells the decompressor that the ZSTD_outBuffer will ALWAYS be the same
	* between calls, except for the modifications that zstd makes to pos (the
	* caller must not modify pos). This is checked by the decompressor, and
	* decompression will fail if it ever changes. Therefore the ZSTD_outBuffer
	* MUST be large enough to fit the entire decompressed frame. This will be
	* checked when the frame content size is known. The data in the ZSTD_outBuffer
	* in the range [dst, dst + pos) MUST not be modified during decompression
	* or you will get data corruption.
	*
	* When this flags is enabled zstd won't allocate an output buffer, because
	* it can write directly to the ZSTD_outBuffer, but it will still allocate
	* an input buffer large enough to fit any compressed block. This will also
	* avoid the memcpy() from the internal output buffer to the ZSTD_outBuffer.
	* If you need to avoid the input buffer allocation use the buffer-less
	* streaming API.
	*
	* NOTE: So long as the ZSTD_outBuffer always points to valid memory, using
	* this flag is ALWAYS memory safe, and will never access out-of-bounds
	* memory. However, decompression WILL fail if you violate the preconditions.
	*
	* WARNING: The data in the ZSTD_outBuffer in the range [dst, dst + pos) MUST
	* not be modified during decompression or you will get data corruption. This
	* is because zstd needs to reference data in the ZSTD_outBuffer to regenerate
	* matches. Normally zstd maintains its own buffer for this purpose, but passing
	* this flag tells zstd to use the user provided buffer.
	*/
	#define ZSTD_d_stableOutBuffer ZSTD_d_experimentalParam2

	/*! ZSTD_DCtx_setFormat() :
	* Instruct the decoder context about what kind of data to decode next.
	* This instruction is mandatory to decode data without a fully-formed header,
	* such ZSTD_f_zstd1_magicless for example.
	* @return : 0, or an error code (which can be tested using ZSTD_isError()). */
	ZSTDLIB_API size_t ZSTD_DCtx_setFormat(ZSTD_DCtx* dctx, ZSTD_format_e format);

	/*! ZSTD_decompressStream_simpleArgs() :
	* Same as ZSTD_decompressStream(),
	* but using only integral types as arguments.
	* This can be helpful for binders from dynamic languages
	* which have troubles handling structures containing memory pointers.
	*/
	ZSTDLIB_API size_t ZSTD_decompressStream_simpleArgs (
	ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity, size_t* dstPos,
	const void* src, size_t srcSize, size_t* srcPos);


	/********************************************************************
	* Advanced streaming functions
	* Warning : most of these functions are now redundant with the Advanced API.
	* Once Advanced API reaches "stable" status,
	* redundant functions will be deprecated, and then at some point removed.
	********************************************************************/

	/===== Advanced Streaming compression functions =====/
	/**! ZSTD_initCStream_srcSize() :
	* This function is deprecated, and equivalent to:
	* ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
	* ZSTD_CCtx_refCDict(zcs, NULL); // clear the dictionary (if any)
	* ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel);
	* ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
	*
	* pledgedSrcSize must be correct. If it is not known at init time, use
	* ZSTD_CONTENTSIZE_UNKNOWN. Note that, for compatibility with older programs,
	* "0" also disables frame content size field. It may be enabled in the future.
	* Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
	*/
	ZSTDLIB_API size_t
	ZSTD_initCStream_srcSize(ZSTD_CStream* zcs,
	int compressionLevel,
	unsigned long long pledgedSrcSize);

	/**! ZSTD_initCStream_usingDict() :
	* This function is deprecated, and is equivalent to:
	* ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
	* ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel);
	* ZSTD_CCtx_loadDictionary(zcs, dict, dictSize);
	*
	* Creates of an internal CDict (incompatible with static CCtx), except if
	* dict == NULL or dictSize < 8, in which case no dict is used.
	* Note: dict is loaded with ZSTD_dct_auto (treated as a full zstd dictionary if
	* it begins with ZSTD_MAGIC_DICTIONARY, else as raw content) and ZSTD_dlm_byCopy.
	* Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
	*/
	ZSTDLIB_API size_t
	ZSTD_initCStream_usingDict(ZSTD_CStream* zcs,
	const void* dict, size_t dictSize,
	int compressionLevel);

	/**! ZSTD_initCStream_advanced() :
	* This function is deprecated, and is approximately equivalent to:
	* ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
	* // Pseudocode: Set each zstd parameter and leave the rest as-is.
	* for ((param, value) : params) {
	* ZSTD_CCtx_setParameter(zcs, param, value);
	* }
	* ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
	* ZSTD_CCtx_loadDictionary(zcs, dict, dictSize);
	*
	* dict is loaded with ZSTD_dct_auto and ZSTD_dlm_byCopy.
	* pledgedSrcSize must be correct.
	* If srcSize is not known at init time, use value ZSTD_CONTENTSIZE_UNKNOWN.
	* Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
	*/
	ZSTDLIB_API size_t
	ZSTD_initCStream_advanced(ZSTD_CStream* zcs,
	const void* dict, size_t dictSize,
	ZSTD_parameters params,
	unsigned long long pledgedSrcSize);

	/**! ZSTD_initCStream_usingCDict() :
	* This function is deprecated, and equivalent to:
	* ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
	* ZSTD_CCtx_refCDict(zcs, cdict);
	*
	* note : cdict will just be referenced, and must outlive compression session
	* Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
	*/
	ZSTDLIB_API size_t ZSTD_initCStream_usingCDict(ZSTD_CStream* zcs, const ZSTD_CDict* cdict);

	/**! ZSTD_initCStream_usingCDict_advanced() :
	* This function is DEPRECATED, and is approximately equivalent to:
	* ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
	* // Pseudocode: Set each zstd frame parameter and leave the rest as-is.
	* for ((fParam, value) : fParams) {
	* ZSTD_CCtx_setParameter(zcs, fParam, value);
	* }
	* ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
	* ZSTD_CCtx_refCDict(zcs, cdict);
	*
	* same as ZSTD_initCStream_usingCDict(), with control over frame parameters.
	* pledgedSrcSize must be correct. If srcSize is not known at init time, use
	* value ZSTD_CONTENTSIZE_UNKNOWN.
	* Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
	*/
	ZSTDLIB_API size_t
	ZSTD_initCStream_usingCDict_advanced(ZSTD_CStream* zcs,
	const ZSTD_CDict* cdict,
	ZSTD_frameParameters fParams,
	unsigned long long pledgedSrcSize);

	/*! ZSTD_resetCStream() :
	* This function is deprecated, and is equivalent to:
	* ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
	* ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize);
	*
	* start a new frame, using same parameters from previous frame.
	* This is typically useful to skip dictionary loading stage, since it will re-use it in-place.
	* Note that zcs must be init at least once before using ZSTD_resetCStream().
	* If pledgedSrcSize is not known at reset time, use macro ZSTD_CONTENTSIZE_UNKNOWN.
	* If pledgedSrcSize > 0, its value must be correct, as it will be written in header, and controlled at the end.
	* For the time being, pledgedSrcSize==0 is interpreted as "srcSize unknown" for compatibility with older programs,
	* but it will change to mean "empty" in future version, so use macro ZSTD_CONTENTSIZE_UNKNOWN instead.
	* @return : 0, or an error code (which can be tested using ZSTD_isError())
	* Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
	*/
	ZSTDLIB_API size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pledgedSrcSize);


	typedef struct {
	unsigned long long ingested; /* nb input bytes read and buffered */
	unsigned long long consumed; /* nb input bytes actually compressed */
	unsigned long long produced; /* nb of compressed bytes generated and buffered */
	unsigned long long flushed; /* nb of compressed bytes flushed : not provided; can be tracked from caller side */
	unsigned currentJobID; /* MT only : latest started job nb */
	unsigned nbActiveWorkers; /* MT only : nb of workers actively compressing at probe time */
	} ZSTD_frameProgression;

	/* ZSTD_getFrameProgression() :
	* tells how much data has been ingested (read from input)
	* consumed (input actually compressed) and produced (output) for current frame.
	* Note : (ingested - consumed) is amount of input data buffered internally, not yet compressed.
	* Aggregates progression inside active worker threads.
	*/
	ZSTDLIB_API ZSTD_frameProgression ZSTD_getFrameProgression(const ZSTD_CCtx* cctx);

	/*! ZSTD_toFlushNow() :
	* Tell how many bytes are ready to be flushed immediately.
	* Useful for multithreading scenarios (nbWorkers >= 1).
	* Probe the oldest active job, defined as oldest job not yet entirely flushed,
	* and check its output buffer.
	* @return : amount of data stored in oldest job and ready to be flushed immediately.
	* if @return == 0, it means either :
	* + there is no active job (could be checked with ZSTD_frameProgression()), or
	* + oldest job is still actively compressing data,
	* but everything it has produced has also been flushed so far,
	* therefore flush speed is limited by production speed of oldest job
	* irrespective of the speed of concurrent (and newer) jobs.
	*/
	ZSTDLIB_API size_t ZSTD_toFlushNow(ZSTD_CCtx* cctx);


	/===== Advanced Streaming decompression functions =====/
	/**
	* This function is deprecated, and is equivalent to:
	*
	* ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
	* ZSTD_DCtx_loadDictionary(zds, dict, dictSize);
	*
	* note: no dictionary will be used if dict == NULL or dictSize < 8
	* Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
	*/
	ZSTDLIB_API size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t dictSize);

	/**
	* This function is deprecated, and is equivalent to:
	*
	* ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
	* ZSTD_DCtx_refDDict(zds, ddict);
	*
	* note : ddict is referenced, it must outlive decompression session
	* Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
	*/
	ZSTDLIB_API size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* zds, const ZSTD_DDict* ddict);

	/**
	* This function is deprecated, and is equivalent to:
	*
	* ZSTD_DCtx_reset(zds, ZSTD_reset_session_only);
	*
	* re-use decompression parameters from previous init; saves dictionary loading
	* Note : this prototype will be marked as deprecated and generate compilation warnings on reaching v1.5.x
	*/
	ZSTDLIB_API size_t ZSTD_resetDStream(ZSTD_DStream* zds);


	/*********************************************************************
	* Buffer-less and synchronous inner streaming functions
	*
	* This is an advanced API, giving full control over buffer management, for users which need direct control over memory.
	* But it's also a complex one, with several restrictions, documented below.
	* Prefer normal streaming API for an easier experience.
	********************************************************************* */

	/**
	Buffer-less streaming compression (synchronous mode)

	A ZSTD_CCtx object is required to track streaming operations.
	Use ZSTD_createCCtx() / ZSTD_freeCCtx() to manage resource.
	ZSTD_CCtx object can be re-used multiple times within successive compression operations.

	Start by initializing a context.
	Use ZSTD_compressBegin(), or ZSTD_compressBegin_usingDict() for dictionary compression,
	or ZSTD_compressBegin_advanced(), for finer parameter control.
	It's also possible to duplicate a reference context which has already been initialized, using ZSTD_copyCCtx()

	Then, consume your input using ZSTD_compressContinue().
	There are some important considerations to keep in mind when using this advanced function :
	- ZSTD_compressContinue() has no internal buffer. It uses externally provided buffers only.
	- Interface is synchronous : input is consumed entirely and produces 1+ compressed blocks.
	- Caller must ensure there is enough space in `dst` to store compressed data under worst case scenario.
	Worst case evaluation is provided by ZSTD_compressBound().
	ZSTD_compressContinue() doesn't guarantee recover after a failed compression.
	- ZSTD_compressContinue() presumes prior input *is still accessible and unmodified* (up to maximum distance size, see WindowLog).
	It remembers all previous contiguous blocks, plus one separated memory segment (which can itself consists of multiple contiguous blocks)
	- ZSTD_compressContinue() detects that prior input has been overwritten when `src` buffer overlaps.
	In which case, it will "discard" the relevant memory section from its history.

	Finish a frame with ZSTD_compressEnd(), which will write the last block(s) and optional checksum.
	It's possible to use srcSize==0, in which case, it will write a final empty block to end the frame.
	Without last block mark, frames are considered unfinished (hence corrupted) by compliant decoders.

	`ZSTD_CCtx` object can be re-used (ZSTD_compressBegin()) to compress again.
	*/

	/===== Buffer-less streaming compression functions =====/
	ZSTDLIB_API size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel);
	ZSTDLIB_API size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, int compressionLevel);
	ZSTDLIB_API size_t ZSTD_compressBegin_advanced(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, ZSTD_parameters params, unsigned long long pledgedSrcSize); /*< pledgedSrcSize : If srcSize is not known at init time, use ZSTD_CONTENTSIZE_UNKNOWN /
	ZSTDLIB_API size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict); /*< note: fails if cdict==NULL /
	ZSTDLIB_API size_t ZSTD_compressBegin_usingCDict_advanced(ZSTD_CCtx* const cctx, const ZSTD_CDict* const cdict, ZSTD_frameParameters const fParams, unsigned long long const pledgedSrcSize); /* compression parameters are already set within cdict. pledgedSrcSize must be correct. If srcSize is not known, use macro ZSTD_CONTENTSIZE_UNKNOWN */
	ZSTDLIB_API size_t ZSTD_copyCCtx(ZSTD_CCtx* cctx, const ZSTD_CCtx* preparedCCtx, unsigned long long pledgedSrcSize); /*< note: if pledgedSrcSize is not known, use ZSTD_CONTENTSIZE_UNKNOWN /

	ZSTDLIB_API size_t ZSTD_compressContinue(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
	ZSTDLIB_API size_t ZSTD_compressEnd(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);


	/*-
	Buffer-less streaming decompression (synchronous mode)

	A ZSTD_DCtx object is required to track streaming operations.
	Use ZSTD_createDCtx() / ZSTD_freeDCtx() to manage it.
	A ZSTD_DCtx object can be re-used multiple times.

	First typical operation is to retrieve frame parameters, using ZSTD_getFrameHeader().
	Frame header is extracted from the beginning of compressed frame, so providing only the frame's beginning is enough.
	Data fragment must be large enough to ensure successful decoding.
	`ZSTD_frameHeaderSize_max` bytes is guaranteed to always be large enough.
	@result : 0 : successful decoding, the `ZSTD_frameHeader` structure is correctly filled.
	>0 : `srcSize` is too small, please provide at least @result bytes on next attempt.
	errorCode, which can be tested using ZSTD_isError().

	It fills a ZSTD_frameHeader structure with important information to correctly decode the frame,
	such as the dictionary ID, content size, or maximum back-reference distance (`windowSize`).
	Note that these values could be wrong, either because of data corruption, or because a 3rd party deliberately spoofs false information.
	As a consequence, check that values remain within valid application range.
	For example, do not allocate memory blindly, check that `windowSize` is within expectation.
	Each application can set its own limits, depending on local restrictions.
	For extended interoperability, it is recommended to support `windowSize` of at least 8 MB.

	ZSTD_decompressContinue() needs previous data blocks during decompression, up to `windowSize` bytes.
	ZSTD_decompressContinue() is very sensitive to contiguity,
	if 2 blocks don't follow each other, make sure that either the compressor breaks contiguity at the same place,
	or that previous contiguous segment is large enough to properly handle maximum back-reference distance.
	There are multiple ways to guarantee this condition.

	The most memory efficient way is to use a round buffer of sufficient size.
	Sufficient size is determined by invoking ZSTD_decodingBufferSize_min(),
	which can @return an error code if required value is too large for current system (in 32-bits mode).
	In a round buffer methodology, ZSTD_decompressContinue() decompresses each block next to previous one,
	up to the moment there is not enough room left in the buffer to guarantee decoding another full block,
	which maximum size is provided in `ZSTD_frameHeader` structure, field `blockSizeMax`.
	At which point, decoding can resume from the beginning of the buffer.
	Note that already decoded data stored in the buffer should be flushed before being overwritten.

	There are alternatives possible, for example using two or more buffers of size `windowSize` each, though they consume more memory.

	Finally, if you control the compression process, you can also ignore all buffer size rules,
	as long as the encoder and decoder progress in "lock-step",
	aka use exactly the same buffer sizes, break contiguity at the same place, etc.

	Once buffers are setup, start decompression, with ZSTD_decompressBegin().
	If decompression requires a dictionary, use ZSTD_decompressBegin_usingDict() or ZSTD_decompressBegin_usingDDict().

	Then use ZSTD_nextSrcSizeToDecompress() and ZSTD_decompressContinue() alternatively.
	ZSTD_nextSrcSizeToDecompress() tells how many bytes to provide as 'srcSize' to ZSTD_decompressContinue().
	ZSTD_decompressContinue() requires this _exact_ amount of bytes, or it will fail.

	@result of ZSTD_decompressContinue() is the number of bytes regenerated within 'dst' (necessarily <= dstCapacity).
	It can be zero : it just means ZSTD_decompressContinue() has decoded some metadata item.
	It can also be an error code, which can be tested with ZSTD_isError().

	A frame is fully decoded when ZSTD_nextSrcSizeToDecompress() returns zero.
	Context can then be reset to start a new decompression.

	Note : it's possible to know if next input to present is a header or a block, using ZSTD_nextInputType().
	This information is not required to properly decode a frame.

	== Special case : skippable frames ==

	Skippable frames allow integration of user-defined data into a flow of concatenated frames.
	Skippable frames will be ignored (skipped) by decompressor.
	The format of skippable frames is as follows :
	a) Skippable frame ID - 4 Bytes, Little endian format, any value from 0x184D2A50 to 0x184D2A5F
	b) Frame Size - 4 Bytes, Little endian format, unsigned 32-bits
	c) Frame Content - any content (User Data) of length equal to Frame Size
	For skippable frames ZSTD_getFrameHeader() returns zfhPtr->frameType==ZSTD_skippableFrame.
	For skippable frames ZSTD_decompressContinue() always returns 0 : it only skips the content.
	*/

	/===== Buffer-less streaming decompression functions =====/
	typedef enum { ZSTD_frame, ZSTD_skippableFrame } ZSTD_frameType_e;
	typedef struct {
	unsigned long long frameContentSize; /* if == ZSTD_CONTENTSIZE_UNKNOWN, it means this field is not available. 0 means "empty" */
	unsigned long long windowSize; /* can be very large, up to <= frameContentSize */
	unsigned blockSizeMax;
	ZSTD_frameType_e frameType; /* if == ZSTD_skippableFrame, frameContentSize is the size of skippable content */
	unsigned headerSize;
	unsigned dictID;
	unsigned checksumFlag;
	} ZSTD_frameHeader;

	/*! ZSTD_getFrameHeader() :
	* decode Frame Header, or requires larger `srcSize`.
	* @return : 0, `zfhPtr` is correctly filled,
	* >0, `srcSize` is too small, value is wanted `srcSize` amount,
	* or an error code, which can be tested using ZSTD_isError() */
	ZSTDLIB_API size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize); /*< doesn't consume input /
	/*! ZSTD_getFrameHeader_advanced() :
	* same as ZSTD_getFrameHeader(),
	* with added capability to select a format (like ZSTD_f_zstd1_magicless) */
	ZSTDLIB_API size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format);
	ZSTDLIB_API size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize); /*< when frame content size is not known, pass in frameContentSize == ZSTD_CONTENTSIZE_UNKNOWN /

	ZSTDLIB_API size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx);
	ZSTDLIB_API size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx* dctx, const void* dict, size_t dictSize);
	ZSTDLIB_API size_t ZSTD_decompressBegin_usingDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict);

	ZSTDLIB_API size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx* dctx);
	ZSTDLIB_API size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);

	/* misc */
	ZSTDLIB_API void ZSTD_copyDCtx(ZSTD_DCtx* dctx, const ZSTD_DCtx* preparedDCtx);
	typedef enum { ZSTDnit_frameHeader, ZSTDnit_blockHeader, ZSTDnit_block, ZSTDnit_lastBlock, ZSTDnit_checksum, ZSTDnit_skippableFrame } ZSTD_nextInputType_e;
	ZSTDLIB_API ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx* dctx);




	/* ============================ */
	/** Block level API */
	/* ============================ */

	/*!
	Block functions produce and decode raw zstd blocks, without frame metadata.
	Frame metadata cost is typically ~12 bytes, which can be non-negligible for very small blocks (< 100 bytes).
	But users will have to take in charge needed metadata to regenerate data, such as compressed and content sizes.

	A few rules to respect :
	- Compressing and decompressing require a context structure
	+ Use ZSTD_createCCtx() and ZSTD_createDCtx()
	- It is necessary to init context before starting
	+ compression : any ZSTD_compressBegin*() variant, including with dictionary
	+ decompression : any ZSTD_decompressBegin*() variant, including with dictionary
	+ copyCCtx() and copyDCtx() can be used too
	- Block size is limited, it must be <= ZSTD_getBlockSize() <= ZSTD_BLOCKSIZE_MAX == 128 KB
	+ If input is larger than a block size, it's necessary to split input data into multiple blocks
	+ For inputs larger than a single block, consider using regular ZSTD_compress() instead.
	Frame metadata is not that costly, and quickly becomes negligible as source size grows larger than a block.
	- When a block is considered not compressible enough, ZSTD_compressBlock() result will be 0 (zero) !
	===> In which case, nothing is produced into `dst` !
	+ User __must__ test for such outcome and deal directly with uncompressed data
	+ A block cannot be declared incompressible if ZSTD_compressBlock() return value was != 0.
	Doing so would mess up with statistics history, leading to potential data corruption.
	+ ZSTD_decompressBlock() _doesn't accept uncompressed data as input_ !!
	+ In case of multiple successive blocks, should some of them be uncompressed,
	decoder must be informed of their existence in order to follow proper history.
	Use ZSTD_insertBlock() for such a case.
	*/

	/===== Raw zstd block functions =====/
	ZSTDLIB_API size_t ZSTD_getBlockSize (const ZSTD_CCtx* cctx);
	ZSTDLIB_API size_t ZSTD_compressBlock (ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
	ZSTDLIB_API size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);
	ZSTDLIB_API size_t ZSTD_insertBlock (ZSTD_DCtx* dctx, const void* blockStart, size_t blockSize); /*< insert uncompressed block into `dctx` history. Useful for multi-blocks decompression. /


	#endif /* ZSTD_H_ZSTD_STATIC_LINKING_ONLY */

	#if defined (__cplusplus)
	}
	#endif
	/** ended inlining ../zstd.h **/
	#define FSE_STATIC_LINKING_ONLY
	/** skipping file: fse.h **/
	#define HUF_STATIC_LINKING_ONLY
	/** skipping file: huf.h **/
	#ifndef XXH_STATIC_LINKING_ONLY
	# define XXH_STATIC_LINKING_ONLY /* XXH64_state_t */
	#endif
	/** start inlining xxhash.h **/
	/*
	* xxHash - Extremely Fast Hash algorithm
	* Header File
	* Copyright (c) 2012-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - xxHash source repository : https://github.com/Cyan4973/xxHash
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/* Notice extracted from xxHash homepage :

	xxHash is an extremely fast Hash algorithm, running at RAM speed limits.
	It also successfully passes all tests from the SMHasher suite.

	Comparison (single thread, Windows Seven 32 bits, using SMHasher on a Core 2 Duo @3GHz)

	Name Speed Q.Score Author
	xxHash 5.4 GB/s 10
	CrapWow 3.2 GB/s 2 Andrew
	MumurHash 3a 2.7 GB/s 10 Austin Appleby
	SpookyHash 2.0 GB/s 10 Bob Jenkins
	SBox 1.4 GB/s 9 Bret Mulvey
	Lookup3 1.2 GB/s 9 Bob Jenkins
	SuperFastHash 1.2 GB/s 1 Paul Hsieh
	CityHash64 1.05 GB/s 10 Pike & Alakuijala
	FNV 0.55 GB/s 5 Fowler, Noll, Vo
	CRC32 0.43 GB/s 9
	MD5-32 0.33 GB/s 10 Ronald L. Rivest
	SHA1-32 0.28 GB/s 10

	Q.Score is a measure of quality of the hash function.
	It depends on successfully passing SMHasher test set.
	10 is a perfect score.

	A 64-bits version, named XXH64, is available since r35.
	It offers much better speed, but for 64-bits applications only.
	Name Speed on 64 bits Speed on 32 bits
	XXH64 13.8 GB/s 1.9 GB/s
	XXH32 6.8 GB/s 6.0 GB/s
	*/

	#if defined (__cplusplus)
	extern "C" {
	#endif

	#ifndef XXHASH_H_5627135585666179
	#define XXHASH_H_5627135585666179 1


	/* ****************************
	* Definitions
	******************************/
	#include <stddef.h> /* size_t */
	typedef enum { XXH_OK=0, XXH_ERROR } XXH_errorcode;


	/* ****************************
	* API modifier
	******************************/
	/** XXH_PRIVATE_API
	* This is useful if you want to include xxhash functions in `static` mode
	* in order to inline them, and remove their symbol from the public list.
	* Methodology :
	* #define XXH_PRIVATE_API
	* #include "xxhash.h"
	* `xxhash.c` is automatically included.
	* It's not useful to compile and link it as a separate module anymore.
	*/
	#ifdef XXH_PRIVATE_API
	# ifndef XXH_STATIC_LINKING_ONLY
	# define XXH_STATIC_LINKING_ONLY
	# endif
	# if defined(__GNUC__)
	# define XXH_PUBLIC_API static __inline __attribute__((unused))
	# elif defined (__cplusplus) \|\| (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */)
	# define XXH_PUBLIC_API static inline
	# elif defined(_MSC_VER)
	# define XXH_PUBLIC_API static __inline
	# else
	# define XXH_PUBLIC_API static /* this version may generate warnings for unused static functions; disable the relevant warning */
	# endif
	#else
	# define XXH_PUBLIC_API /* do nothing */
	#endif /* XXH_PRIVATE_API */

	/*!XXH_NAMESPACE, aka Namespace Emulation :

	If you want to include _and expose_ xxHash functions from within your own library,
	but also want to avoid symbol collisions with another library which also includes xxHash,

	you can use XXH_NAMESPACE, to automatically prefix any public symbol from xxhash library
	with the value of XXH_NAMESPACE (so avoid to keep it NULL and avoid numeric values).

	Note that no change is required within the calling program as long as it includes `xxhash.h` :
	regular symbol name will be automatically translated by this header.
	*/
	#ifdef XXH_NAMESPACE
	# define XXH_CAT(A,B) A##B
	# define XXH_NAME2(A,B) XXH_CAT(A,B)
	# define XXH32 XXH_NAME2(XXH_NAMESPACE, XXH32)
	# define XXH64 XXH_NAME2(XXH_NAMESPACE, XXH64)
	# define XXH_versionNumber XXH_NAME2(XXH_NAMESPACE, XXH_versionNumber)
	# define XXH32_createState XXH_NAME2(XXH_NAMESPACE, XXH32_createState)
	# define XXH64_createState XXH_NAME2(XXH_NAMESPACE, XXH64_createState)
	# define XXH32_freeState XXH_NAME2(XXH_NAMESPACE, XXH32_freeState)
	# define XXH64_freeState XXH_NAME2(XXH_NAMESPACE, XXH64_freeState)
	# define XXH32_reset XXH_NAME2(XXH_NAMESPACE, XXH32_reset)
	# define XXH64_reset XXH_NAME2(XXH_NAMESPACE, XXH64_reset)
	# define XXH32_update XXH_NAME2(XXH_NAMESPACE, XXH32_update)
	# define XXH64_update XXH_NAME2(XXH_NAMESPACE, XXH64_update)
	# define XXH32_digest XXH_NAME2(XXH_NAMESPACE, XXH32_digest)
	# define XXH64_digest XXH_NAME2(XXH_NAMESPACE, XXH64_digest)
	# define XXH32_copyState XXH_NAME2(XXH_NAMESPACE, XXH32_copyState)
	# define XXH64_copyState XXH_NAME2(XXH_NAMESPACE, XXH64_copyState)
	# define XXH32_canonicalFromHash XXH_NAME2(XXH_NAMESPACE, XXH32_canonicalFromHash)
	# define XXH64_canonicalFromHash XXH_NAME2(XXH_NAMESPACE, XXH64_canonicalFromHash)
	# define XXH32_hashFromCanonical XXH_NAME2(XXH_NAMESPACE, XXH32_hashFromCanonical)
	# define XXH64_hashFromCanonical XXH_NAME2(XXH_NAMESPACE, XXH64_hashFromCanonical)
	#endif


	/* *************************************
	* Version
	***************************************/
	#define XXH_VERSION_MAJOR 0
	#define XXH_VERSION_MINOR 6
	#define XXH_VERSION_RELEASE 2
	#define XXH_VERSION_NUMBER (XXH_VERSION_MAJOR 100100 + XXH_VERSION_MINOR *100 + XXH_VERSION_RELEASE)
	XXH_PUBLIC_API unsigned XXH_versionNumber (void);


	/* ****************************
	* Simple Hash Functions
	******************************/
	typedef unsigned int XXH32_hash_t;
	typedef unsigned long long XXH64_hash_t;

	XXH_PUBLIC_API XXH32_hash_t XXH32 (const void* input, size_t length, unsigned int seed);
	XXH_PUBLIC_API XXH64_hash_t XXH64 (const void* input, size_t length, unsigned long long seed);

	/*!
	XXH32() :
	Calculate the 32-bits hash of sequence "length" bytes stored at memory address "input".
	The memory between input & input+length must be valid (allocated and read-accessible).
	"seed" can be used to alter the result predictably.
	Speed on Core 2 Duo @ 3 GHz (single thread, SMHasher benchmark) : 5.4 GB/s
	XXH64() :
	Calculate the 64-bits hash of sequence of length "len" stored at memory address "input".
	"seed" can be used to alter the result predictably.
	This function runs 2x faster on 64-bits systems, but slower on 32-bits systems (see benchmark).
	*/


	/* ****************************
	* Streaming Hash Functions
	******************************/
	typedef struct XXH32_state_s XXH32_state_t; /* incomplete type */
	typedef struct XXH64_state_s XXH64_state_t; /* incomplete type */

	/! State allocation, compatible with dynamic libraries /

	XXH_PUBLIC_API XXH32_state_t* XXH32_createState(void);
	XXH_PUBLIC_API XXH_errorcode XXH32_freeState(XXH32_state_t* statePtr);

	XXH_PUBLIC_API XXH64_state_t* XXH64_createState(void);
	XXH_PUBLIC_API XXH_errorcode XXH64_freeState(XXH64_state_t* statePtr);


	/* hash streaming */

	XXH_PUBLIC_API XXH_errorcode XXH32_reset (XXH32_state_t* statePtr, unsigned int seed);
	XXH_PUBLIC_API XXH_errorcode XXH32_update (XXH32_state_t* statePtr, const void* input, size_t length);
	XXH_PUBLIC_API XXH32_hash_t XXH32_digest (const XXH32_state_t* statePtr);

	XXH_PUBLIC_API XXH_errorcode XXH64_reset (XXH64_state_t* statePtr, unsigned long long seed);
	XXH_PUBLIC_API XXH_errorcode XXH64_update (XXH64_state_t* statePtr, const void* input, size_t length);
	XXH_PUBLIC_API XXH64_hash_t XXH64_digest (const XXH64_state_t* statePtr);

	/*
	These functions generate the xxHash of an input provided in multiple segments.
	Note that, for small input, they are slower than single-call functions, due to state management.
	For small input, prefer `XXH32()` and `XXH64()` .

	XXH state must first be allocated, using XXH*_createState() .

	Start a new hash by initializing state with a seed, using XXH*_reset().

	Then, feed the hash state by calling XXH*_update() as many times as necessary.
	Obviously, input must be allocated and read accessible.
	The function returns an error code, with 0 meaning OK, and any other value meaning there is an error.

	Finally, a hash value can be produced anytime, by using XXH*_digest().
	This function returns the nn-bits hash as an int or long long.

	It's still possible to continue inserting input into the hash state after a digest,
	and generate some new hashes later on, by calling again XXH*_digest().

	When done, free XXH state space if it was allocated dynamically.
	*/


	/* **************************
	* Utils
	****************************/
	#if !(defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L)) /* ! C99 */
	# define restrict /* disable restrict */
	#endif

	XXH_PUBLIC_API void XXH32_copyState(XXH32_state_t* restrict dst_state, const XXH32_state_t* restrict src_state);
	XXH_PUBLIC_API void XXH64_copyState(XXH64_state_t* restrict dst_state, const XXH64_state_t* restrict src_state);


	/* **************************
	* Canonical representation
	****************************/
	/* Default result type for XXH functions are primitive unsigned 32 and 64 bits.
	* The canonical representation uses human-readable write convention, aka big-endian (large digits first).
	* These functions allow transformation of hash result into and from its canonical format.
	* This way, hash values can be written into a file / memory, and remain comparable on different systems and programs.
	*/
	typedef struct { unsigned char digest[4]; } XXH32_canonical_t;
	typedef struct { unsigned char digest[8]; } XXH64_canonical_t;

	XXH_PUBLIC_API void XXH32_canonicalFromHash(XXH32_canonical_t* dst, XXH32_hash_t hash);
	XXH_PUBLIC_API void XXH64_canonicalFromHash(XXH64_canonical_t* dst, XXH64_hash_t hash);

	XXH_PUBLIC_API XXH32_hash_t XXH32_hashFromCanonical(const XXH32_canonical_t* src);
	XXH_PUBLIC_API XXH64_hash_t XXH64_hashFromCanonical(const XXH64_canonical_t* src);

	#endif /* XXHASH_H_5627135585666179 */



	/* ================================================================================================
	This section contains definitions which are not guaranteed to remain stable.
	They may change in future versions, becoming incompatible with a different version of the library.
	They shall only be used with static linking.
	Never use these definitions in association with dynamic linking !
	=================================================================================================== */
	#if defined(XXH_STATIC_LINKING_ONLY) && !defined(XXH_STATIC_H_3543687687345)
	#define XXH_STATIC_H_3543687687345

	/* These definitions are only meant to allow allocation of XXH state
	statically, on stack, or in a struct for example.
	Do not use members directly. */

	struct XXH32_state_s {
	unsigned total_len_32;
	unsigned large_len;
	unsigned v1;
	unsigned v2;
	unsigned v3;
	unsigned v4;
	unsigned mem32[4]; /* buffer defined as U32 for alignment */
	unsigned memsize;
	unsigned reserved; /* never read nor write, will be removed in a future version */
	}; /* typedef'd to XXH32_state_t */

	struct XXH64_state_s {
	unsigned long long total_len;
	unsigned long long v1;
	unsigned long long v2;
	unsigned long long v3;
	unsigned long long v4;
	unsigned long long mem64[4]; /* buffer defined as U64 for alignment */
	unsigned memsize;
	unsigned reserved[2]; /* never read nor write, will be removed in a future version */
	}; /* typedef'd to XXH64_state_t */


	# ifdef XXH_PRIVATE_API
	/** start inlining xxhash.c **/
	/*
	* xxHash - Fast Hash algorithm
	* Copyright (c) 2012-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - xxHash homepage: http://www.xxhash.com
	* - xxHash source repository : https://github.com/Cyan4973/xxHash
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/


	/* *************************************
	* Tuning parameters
	***************************************/
	/*!XXH_FORCE_MEMORY_ACCESS :
	* By default, access to unaligned memory is controlled by `memcpy()`, which is safe and portable.
	* Unfortunately, on some target/compiler combinations, the generated assembly is sub-optimal.
	* The below switch allow to select different access method for improved performance.
	* Method 0 (default) : use `memcpy()`. Safe and portable.
	* Method 1 : `__packed` statement. It depends on compiler extension (ie, not portable).
	* This method is safe if your compiler supports it, and generally as fast or faster than `memcpy`.
	* Method 2 : direct access. This method doesn't depend on compiler but violate C standard.
	* It can generate buggy code on targets which do not support unaligned memory accesses.
	* But in some circumstances, it's the only known way to get the most performance (ie GCC + ARMv6)
	* See http://stackoverflow.com/a/32095106/646947 for details.
	* Prefer these methods in priority order (0 > 1 > 2)
	*/
	#ifndef XXH_FORCE_MEMORY_ACCESS /* can be defined externally, on command line for example */
	# if defined(__GNUC__) && ( defined(__ARM_ARCH_6__) \|\| defined(__ARM_ARCH_6J__) \|\| defined(__ARM_ARCH_6K__) \|\| defined(__ARM_ARCH_6Z__) \|\| defined(__ARM_ARCH_6ZK__) \|\| defined(__ARM_ARCH_6T2__) )
	# define XXH_FORCE_MEMORY_ACCESS 2
	# elif (defined(__INTEL_COMPILER) && !defined(WIN32)) \|\| \
	(defined(__GNUC__) && ( defined(__ARM_ARCH_7__) \|\| defined(__ARM_ARCH_7A__) \|\| defined(__ARM_ARCH_7R__) \|\| defined(__ARM_ARCH_7M__) \|\| defined(__ARM_ARCH_7S__) )) \|\| \
	defined(__ICCARM__)
	# define XXH_FORCE_MEMORY_ACCESS 1
	# endif
	#endif

	/*!XXH_ACCEPT_NULL_INPUT_POINTER :
	* If the input pointer is a null pointer, xxHash default behavior is to trigger a memory access error, since it is a bad pointer.
	* When this option is enabled, xxHash output for null input pointers will be the same as a null-length input.
	* By default, this option is disabled. To enable it, uncomment below define :
	*/
	/* #define XXH_ACCEPT_NULL_INPUT_POINTER 1 */

	/*!XXH_FORCE_NATIVE_FORMAT :
	* By default, xxHash library provides endian-independent Hash values, based on little-endian convention.
	* Results are therefore identical for little-endian and big-endian CPU.
	* This comes at a performance cost for big-endian CPU, since some swapping is required to emulate little-endian format.
	* Should endian-independence be of no importance for your application, you may set the #define below to 1,
	* to improve speed for Big-endian CPU.
	* This option has no impact on Little_Endian CPU.
	*/
	#ifndef XXH_FORCE_NATIVE_FORMAT /* can be defined externally */
	# define XXH_FORCE_NATIVE_FORMAT 0
	#endif

	/*!XXH_FORCE_ALIGN_CHECK :
	* This is a minor performance trick, only useful with lots of very small keys.
	* It means : check for aligned/unaligned input.
	* The check costs one initial branch per hash; set to 0 when the input data
	* is guaranteed to be aligned.
	*/
	#ifndef XXH_FORCE_ALIGN_CHECK /* can be defined externally */
	# if defined(__i386) \|\| defined(_M_IX86) \|\| defined(__x86_64__) \|\| defined(_M_X64)
	# define XXH_FORCE_ALIGN_CHECK 0
	# else
	# define XXH_FORCE_ALIGN_CHECK 1
	# endif
	#endif


	/* *************************************
	* Includes & Memory related functions
	***************************************/
	/* Modify the local functions below should you wish to use some other memory routines */
	/* for malloc(), free() */
	#include <stdlib.h>
	#include <stddef.h> /* size_t */
	static void* XXH_malloc(size_t s) { return malloc(s); }
	static void XXH_free (void* p) { free(p); }
	/* for memcpy() */
	#include <string.h>
	static void* XXH_memcpy(void* dest, const void* src, size_t size) { return memcpy(dest,src,size); }

	#ifndef XXH_STATIC_LINKING_ONLY
	# define XXH_STATIC_LINKING_ONLY
	#endif
	/** skipping file: xxhash.h **/


	/* *************************************
	* Compiler Specific Options
	***************************************/
	#if (defined(__GNUC__) && !defined(__STRICT_ANSI__)) \|\| defined(__cplusplus) \|\| defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L /* C99 */
	# define INLINE_KEYWORD inline
	#else
	# define INLINE_KEYWORD
	#endif

	#if defined(__GNUC__) \|\| defined(__ICCARM__)
	# define FORCE_INLINE_ATTR __attribute__((always_inline))
	#elif defined(_MSC_VER)
	# define FORCE_INLINE_ATTR __forceinline
	#else
	# define FORCE_INLINE_ATTR
	#endif

	#define FORCE_INLINE_TEMPLATE static INLINE_KEYWORD FORCE_INLINE_ATTR


	#ifdef _MSC_VER
	# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */
	#endif


	/* *************************************
	* Basic Types
	***************************************/
	#ifndef MEM_MODULE
	# define MEM_MODULE
	# if !defined (__VMS) && (defined (__cplusplus) \|\| (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
	# include <stdint.h>
	typedef uint8_t BYTE;
	typedef uint16_t U16;
	typedef uint32_t U32;
	typedef int32_t S32;
	typedef uint64_t U64;
	# else
	typedef unsigned char BYTE;
	typedef unsigned short U16;
	typedef unsigned int U32;
	typedef signed int S32;
	typedef unsigned long long U64; /* if your compiler doesn't support unsigned long long, replace by another 64-bit type here. Note that xxhash.h will also need to be updated. */
	# endif
	#endif


	#if (defined(XXH_FORCE_MEMORY_ACCESS) && (XXH_FORCE_MEMORY_ACCESS==2))

	/* Force direct memory access. Only works on CPU which support unaligned memory access in hardware */
	static U32 XXH_read32(const void* memPtr) { return (const U32) memPtr; }
	static U64 XXH_read64(const void* memPtr) { return (const U64) memPtr; }

	#elif (defined(XXH_FORCE_MEMORY_ACCESS) && (XXH_FORCE_MEMORY_ACCESS==1))

	/* __pack instructions are safer, but compiler specific, hence potentially problematic for some compilers */
	/* currently only defined for gcc and icc */
	typedef union { U32 u32; U64 u64; } __attribute__((packed)) unalign;

	static U32 XXH_read32(const void* ptr) { return ((const unalign*)ptr)->u32; }
	static U64 XXH_read64(const void* ptr) { return ((const unalign*)ptr)->u64; }

	#else

	/* portable and safe solution. Generally efficient.
	* see : http://stackoverflow.com/a/32095106/646947
	*/

	static U32 XXH_read32(const void* memPtr)
	{
	U32 val;
	memcpy(&val, memPtr, sizeof(val));
	return val;
	}

	static U64 XXH_read64(const void* memPtr)
	{
	U64 val;
	memcpy(&val, memPtr, sizeof(val));
	return val;
	}

	#endif /* XXH_FORCE_DIRECT_MEMORY_ACCESS */


	/* ****************************************
	* Compiler-specific Functions and Macros
	******************************************/
	#define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__)

	/* Note : although _rotl exists for minGW (GCC under windows), performance seems poor */
	#if defined(_MSC_VER)
	# define XXH_rotl32(x,r) _rotl(x,r)
	# define XXH_rotl64(x,r) _rotl64(x,r)
	#else
	#if defined(__ICCARM__)
	# include <intrinsics.h>
	# define XXH_rotl32(x,r) __ROR(x,(32 - r))
	#else
	# define XXH_rotl32(x,r) ((x << r) \| (x >> (32 - r)))
	#endif
	# define XXH_rotl64(x,r) ((x << r) \| (x >> (64 - r)))
	#endif

	#if defined(_MSC_VER) /* Visual Studio */
	# define XXH_swap32 _byteswap_ulong
	# define XXH_swap64 _byteswap_uint64
	#elif GCC_VERSION >= 403
	# define XXH_swap32 __builtin_bswap32
	# define XXH_swap64 __builtin_bswap64
	#else
	static U32 XXH_swap32 (U32 x)
	{
	return ((x << 24) & 0xff000000 ) \|
	((x << 8) & 0x00ff0000 ) \|
	((x >> 8) & 0x0000ff00 ) \|
	((x >> 24) & 0x000000ff );
	}
	static U64 XXH_swap64 (U64 x)
	{
	return ((x << 56) & 0xff00000000000000ULL) \|
	((x << 40) & 0x00ff000000000000ULL) \|
	((x << 24) & 0x0000ff0000000000ULL) \|
	((x << 8) & 0x000000ff00000000ULL) \|
	((x >> 8) & 0x00000000ff000000ULL) \|
	((x >> 24) & 0x0000000000ff0000ULL) \|
	((x >> 40) & 0x000000000000ff00ULL) \|
	((x >> 56) & 0x00000000000000ffULL);
	}
	#endif


	/* *************************************
	* Architecture Macros
	***************************************/
	typedef enum { XXH_bigEndian=0, XXH_littleEndian=1 } XXH_endianess;

	/* XXH_CPU_LITTLE_ENDIAN can be defined externally, for example on the compiler command line */
	#ifndef XXH_CPU_LITTLE_ENDIAN
	static const int g_one = 1;
	# define XXH_CPU_LITTLE_ENDIAN ((const char)(&g_one))
	#endif


	/* ***************************
	* Memory reads
	*****************************/
	typedef enum { XXH_aligned, XXH_unaligned } XXH_alignment;

	FORCE_INLINE_TEMPLATE U32 XXH_readLE32_align(const void* ptr, XXH_endianess endian, XXH_alignment align)
	{
	if (align==XXH_unaligned)
	return endian==XXH_littleEndian ? XXH_read32(ptr) : XXH_swap32(XXH_read32(ptr));
	else
	return endian==XXH_littleEndian ? (const U32)ptr : XXH_swap32((const U32)ptr);
	}

	FORCE_INLINE_TEMPLATE U32 XXH_readLE32(const void* ptr, XXH_endianess endian)
	{
	return XXH_readLE32_align(ptr, endian, XXH_unaligned);
	}

	static U32 XXH_readBE32(const void* ptr)
	{
	return XXH_CPU_LITTLE_ENDIAN ? XXH_swap32(XXH_read32(ptr)) : XXH_read32(ptr);
	}

	FORCE_INLINE_TEMPLATE U64 XXH_readLE64_align(const void* ptr, XXH_endianess endian, XXH_alignment align)
	{
	if (align==XXH_unaligned)
	return endian==XXH_littleEndian ? XXH_read64(ptr) : XXH_swap64(XXH_read64(ptr));
	else
	return endian==XXH_littleEndian ? (const U64)ptr : XXH_swap64((const U64)ptr);
	}

	FORCE_INLINE_TEMPLATE U64 XXH_readLE64(const void* ptr, XXH_endianess endian)
	{
	return XXH_readLE64_align(ptr, endian, XXH_unaligned);
	}

	static U64 XXH_readBE64(const void* ptr)
	{
	return XXH_CPU_LITTLE_ENDIAN ? XXH_swap64(XXH_read64(ptr)) : XXH_read64(ptr);
	}


	/* *************************************
	* Macros
	***************************************/
	#define XXH_STATIC_ASSERT(c) { enum { XXH_static_assert = 1/(int)(!!(c)) }; } /* use only after variable declarations */


	/* *************************************
	* Constants
	***************************************/
	static const U32 PRIME32_1 = 2654435761U;
	static const U32 PRIME32_2 = 2246822519U;
	static const U32 PRIME32_3 = 3266489917U;
	static const U32 PRIME32_4 = 668265263U;
	static const U32 PRIME32_5 = 374761393U;

	static const U64 PRIME64_1 = 11400714785074694791ULL;
	static const U64 PRIME64_2 = 14029467366897019727ULL;
	static const U64 PRIME64_3 = 1609587929392839161ULL;
	static const U64 PRIME64_4 = 9650029242287828579ULL;
	static const U64 PRIME64_5 = 2870177450012600261ULL;

	XXH_PUBLIC_API unsigned XXH_versionNumber (void) { return XXH_VERSION_NUMBER; }


	/* **************************
	* Utils
	****************************/
	XXH_PUBLIC_API void XXH32_copyState(XXH32_state_t* restrict dstState, const XXH32_state_t* restrict srcState)
	{
	memcpy(dstState, srcState, sizeof(*dstState));
	}

	XXH_PUBLIC_API void XXH64_copyState(XXH64_state_t* restrict dstState, const XXH64_state_t* restrict srcState)
	{
	memcpy(dstState, srcState, sizeof(*dstState));
	}


	/* ***************************
	* Simple Hash Functions
	*****************************/

	static U32 XXH32_round(U32 seed, U32 input)
	{
	seed += input * PRIME32_2;
	seed = XXH_rotl32(seed, 13);
	seed *= PRIME32_1;
	return seed;
	}

	FORCE_INLINE_TEMPLATE U32 XXH32_endian_align(const void* input, size_t len, U32 seed, XXH_endianess endian, XXH_alignment align)
	{
	const BYTE* p = (const BYTE*)input;
	const BYTE* bEnd = p + len;
	U32 h32;
	#define XXH_get32bits(p) XXH_readLE32_align(p, endian, align)

	#ifdef XXH_ACCEPT_NULL_INPUT_POINTER
	if (p==NULL) {
	len=0;
	bEnd=p=(const BYTE*)(size_t)16;
	}
	#endif

	if (len>=16) {
	const BYTE* const limit = bEnd - 16;
	U32 v1 = seed + PRIME32_1 + PRIME32_2;
	U32 v2 = seed + PRIME32_2;
	U32 v3 = seed + 0;
	U32 v4 = seed - PRIME32_1;

	do {
	v1 = XXH32_round(v1, XXH_get32bits(p)); p+=4;
	v2 = XXH32_round(v2, XXH_get32bits(p)); p+=4;
	v3 = XXH32_round(v3, XXH_get32bits(p)); p+=4;
	v4 = XXH32_round(v4, XXH_get32bits(p)); p+=4;
	} while (p<=limit);

	h32 = XXH_rotl32(v1, 1) + XXH_rotl32(v2, 7) + XXH_rotl32(v3, 12) + XXH_rotl32(v4, 18);
	} else {
	h32 = seed + PRIME32_5;
	}

	h32 += (U32) len;

	while (p+4<=bEnd) {
	h32 += XXH_get32bits(p) * PRIME32_3;
	h32 = XXH_rotl32(h32, 17) * PRIME32_4 ;
	p+=4;
	}

	while (p<bEnd) {
	h32 += (p) PRIME32_5;
	h32 = XXH_rotl32(h32, 11) * PRIME32_1 ;
	p++;
	}

	h32 ^= h32 >> 15;
	h32 *= PRIME32_2;
	h32 ^= h32 >> 13;
	h32 *= PRIME32_3;
	h32 ^= h32 >> 16;

	return h32;
	}


	XXH_PUBLIC_API unsigned int XXH32 (const void* input, size_t len, unsigned int seed)
	{
	#if 0
	/* Simple version, good for code maintenance, but unfortunately slow for small inputs */
	XXH32_CREATESTATE_STATIC(state);
	XXH32_reset(state, seed);
	XXH32_update(state, input, len);
	return XXH32_digest(state);
	#else
	XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN;

	if (XXH_FORCE_ALIGN_CHECK) {
	if ((((size_t)input) & 3) == 0) { /* Input is 4-bytes aligned, leverage the speed benefit */
	if ((endian_detected==XXH_littleEndian) \|\| XXH_FORCE_NATIVE_FORMAT)
	return XXH32_endian_align(input, len, seed, XXH_littleEndian, XXH_aligned);
	else
	return XXH32_endian_align(input, len, seed, XXH_bigEndian, XXH_aligned);
	} }

	if ((endian_detected==XXH_littleEndian) \|\| XXH_FORCE_NATIVE_FORMAT)
	return XXH32_endian_align(input, len, seed, XXH_littleEndian, XXH_unaligned);
	else
	return XXH32_endian_align(input, len, seed, XXH_bigEndian, XXH_unaligned);
	#endif
	}


	static U64 XXH64_round(U64 acc, U64 input)
	{
	acc += input * PRIME64_2;
	acc = XXH_rotl64(acc, 31);
	acc *= PRIME64_1;
	return acc;
	}

	static U64 XXH64_mergeRound(U64 acc, U64 val)
	{
	val = XXH64_round(0, val);
	acc ^= val;
	acc = acc * PRIME64_1 + PRIME64_4;
	return acc;
	}

	FORCE_INLINE_TEMPLATE U64 XXH64_endian_align(const void* input, size_t len, U64 seed, XXH_endianess endian, XXH_alignment align)
	{
	const BYTE* p = (const BYTE*)input;
	const BYTE* const bEnd = p + len;
	U64 h64;
	#define XXH_get64bits(p) XXH_readLE64_align(p, endian, align)

	#ifdef XXH_ACCEPT_NULL_INPUT_POINTER
	if (p==NULL) {
	len=0;
	bEnd=p=(const BYTE*)(size_t)32;
	}
	#endif

	if (len>=32) {
	const BYTE* const limit = bEnd - 32;
	U64 v1 = seed + PRIME64_1 + PRIME64_2;
	U64 v2 = seed + PRIME64_2;
	U64 v3 = seed + 0;
	U64 v4 = seed - PRIME64_1;

	do {
	v1 = XXH64_round(v1, XXH_get64bits(p)); p+=8;
	v2 = XXH64_round(v2, XXH_get64bits(p)); p+=8;
	v3 = XXH64_round(v3, XXH_get64bits(p)); p+=8;
	v4 = XXH64_round(v4, XXH_get64bits(p)); p+=8;
	} while (p<=limit);

	h64 = XXH_rotl64(v1, 1) + XXH_rotl64(v2, 7) + XXH_rotl64(v3, 12) + XXH_rotl64(v4, 18);
	h64 = XXH64_mergeRound(h64, v1);
	h64 = XXH64_mergeRound(h64, v2);
	h64 = XXH64_mergeRound(h64, v3);
	h64 = XXH64_mergeRound(h64, v4);

	} else {
	h64 = seed + PRIME64_5;
	}

	h64 += (U64) len;

	while (p+8<=bEnd) {
	U64 const k1 = XXH64_round(0, XXH_get64bits(p));
	h64 ^= k1;
	h64 = XXH_rotl64(h64,27) * PRIME64_1 + PRIME64_4;
	p+=8;
	}

	if (p+4<=bEnd) {
	h64 ^= (U64)(XXH_get32bits(p)) * PRIME64_1;
	h64 = XXH_rotl64(h64, 23) * PRIME64_2 + PRIME64_3;
	p+=4;
	}

	while (p<bEnd) {
	h64 ^= (p) PRIME64_5;
	h64 = XXH_rotl64(h64, 11) * PRIME64_1;
	p++;
	}

	h64 ^= h64 >> 33;
	h64 *= PRIME64_2;
	h64 ^= h64 >> 29;
	h64 *= PRIME64_3;
	h64 ^= h64 >> 32;

	return h64;
	}


	XXH_PUBLIC_API unsigned long long XXH64 (const void* input, size_t len, unsigned long long seed)
	{
	#if 0
	/* Simple version, good for code maintenance, but unfortunately slow for small inputs */
	XXH64_CREATESTATE_STATIC(state);
	XXH64_reset(state, seed);
	XXH64_update(state, input, len);
	return XXH64_digest(state);
	#else
	XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN;

	if (XXH_FORCE_ALIGN_CHECK) {
	if ((((size_t)input) & 7)==0) { /* Input is aligned, let's leverage the speed advantage */
	if ((endian_detected==XXH_littleEndian) \|\| XXH_FORCE_NATIVE_FORMAT)
	return XXH64_endian_align(input, len, seed, XXH_littleEndian, XXH_aligned);
	else
	return XXH64_endian_align(input, len, seed, XXH_bigEndian, XXH_aligned);
	} }

	if ((endian_detected==XXH_littleEndian) \|\| XXH_FORCE_NATIVE_FORMAT)
	return XXH64_endian_align(input, len, seed, XXH_littleEndian, XXH_unaligned);
	else
	return XXH64_endian_align(input, len, seed, XXH_bigEndian, XXH_unaligned);
	#endif
	}


	/* **************************************************
	* Advanced Hash Functions
	****************************************************/

	XXH_PUBLIC_API XXH32_state_t* XXH32_createState(void)
	{
	return (XXH32_state_t*)XXH_malloc(sizeof(XXH32_state_t));
	}
	XXH_PUBLIC_API XXH_errorcode XXH32_freeState(XXH32_state_t* statePtr)
	{
	XXH_free(statePtr);
	return XXH_OK;
	}

	XXH_PUBLIC_API XXH64_state_t* XXH64_createState(void)
	{
	return (XXH64_state_t*)XXH_malloc(sizeof(XXH64_state_t));
	}
	XXH_PUBLIC_API XXH_errorcode XXH64_freeState(XXH64_state_t* statePtr)
	{
	XXH_free(statePtr);
	return XXH_OK;
	}


	/* Hash feed */

	XXH_PUBLIC_API XXH_errorcode XXH32_reset(XXH32_state_t* statePtr, unsigned int seed)
	{
	XXH32_state_t state; /* using a local state to memcpy() in order to avoid strict-aliasing warnings */
	memset(&state, 0, sizeof(state)-4); /* do not write into reserved, for future removal */
	state.v1 = seed + PRIME32_1 + PRIME32_2;
	state.v2 = seed + PRIME32_2;
	state.v3 = seed + 0;
	state.v4 = seed - PRIME32_1;
	memcpy(statePtr, &state, sizeof(state));
	return XXH_OK;
	}


	XXH_PUBLIC_API XXH_errorcode XXH64_reset(XXH64_state_t* statePtr, unsigned long long seed)
	{
	XXH64_state_t state; /* using a local state to memcpy() in order to avoid strict-aliasing warnings */
	memset(&state, 0, sizeof(state)-8); /* do not write into reserved, for future removal */
	state.v1 = seed + PRIME64_1 + PRIME64_2;
	state.v2 = seed + PRIME64_2;
	state.v3 = seed + 0;
	state.v4 = seed - PRIME64_1;
	memcpy(statePtr, &state, sizeof(state));
	return XXH_OK;
	}


	FORCE_INLINE_TEMPLATE XXH_errorcode XXH32_update_endian (XXH32_state_t* state, const void* input, size_t len, XXH_endianess endian)
	{
	const BYTE* p = (const BYTE*)input;
	const BYTE* const bEnd = p + len;

	#ifdef XXH_ACCEPT_NULL_INPUT_POINTER
	if (input==NULL) return XXH_ERROR;
	#endif

	state->total_len_32 += (unsigned)len;
	state->large_len \|= (len>=16) \| (state->total_len_32>=16);

	if (state->memsize + len < 16) { /* fill in tmp buffer */
	XXH_memcpy((BYTE*)(state->mem32) + state->memsize, input, len);
	state->memsize += (unsigned)len;
	return XXH_OK;
	}

	if (state->memsize) { /* some data left from previous update */
	XXH_memcpy((BYTE*)(state->mem32) + state->memsize, input, 16-state->memsize);
	{ const U32* p32 = state->mem32;
	state->v1 = XXH32_round(state->v1, XXH_readLE32(p32, endian)); p32++;
	state->v2 = XXH32_round(state->v2, XXH_readLE32(p32, endian)); p32++;
	state->v3 = XXH32_round(state->v3, XXH_readLE32(p32, endian)); p32++;
	state->v4 = XXH32_round(state->v4, XXH_readLE32(p32, endian)); p32++;
	}
	p += 16-state->memsize;
	state->memsize = 0;
	}

	if (p <= bEnd-16) {
	const BYTE* const limit = bEnd - 16;
	U32 v1 = state->v1;
	U32 v2 = state->v2;
	U32 v3 = state->v3;
	U32 v4 = state->v4;

	do {
	v1 = XXH32_round(v1, XXH_readLE32(p, endian)); p+=4;
	v2 = XXH32_round(v2, XXH_readLE32(p, endian)); p+=4;
	v3 = XXH32_round(v3, XXH_readLE32(p, endian)); p+=4;
	v4 = XXH32_round(v4, XXH_readLE32(p, endian)); p+=4;
	} while (p<=limit);

	state->v1 = v1;
	state->v2 = v2;
	state->v3 = v3;
	state->v4 = v4;
	}

	if (p < bEnd) {
	XXH_memcpy(state->mem32, p, (size_t)(bEnd-p));
	state->memsize = (unsigned)(bEnd-p);
	}

	return XXH_OK;
	}

	XXH_PUBLIC_API XXH_errorcode XXH32_update (XXH32_state_t* state_in, const void* input, size_t len)
	{
	XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN;

	if ((endian_detected==XXH_littleEndian) \|\| XXH_FORCE_NATIVE_FORMAT)
	return XXH32_update_endian(state_in, input, len, XXH_littleEndian);
	else
	return XXH32_update_endian(state_in, input, len, XXH_bigEndian);
	}



	FORCE_INLINE_TEMPLATE U32 XXH32_digest_endian (const XXH32_state_t* state, XXH_endianess endian)
	{
	const BYTE * p = (const BYTE*)state->mem32;
	const BYTE* const bEnd = (const BYTE*)(state->mem32) + state->memsize;
	U32 h32;

	if (state->large_len) {
	h32 = XXH_rotl32(state->v1, 1) + XXH_rotl32(state->v2, 7) + XXH_rotl32(state->v3, 12) + XXH_rotl32(state->v4, 18);
	} else {
	h32 = state->v3 /* == seed */ + PRIME32_5;
	}

	h32 += state->total_len_32;

	while (p+4<=bEnd) {
	h32 += XXH_readLE32(p, endian) * PRIME32_3;
	h32 = XXH_rotl32(h32, 17) * PRIME32_4;
	p+=4;
	}

	while (p<bEnd) {
	h32 += (p) PRIME32_5;
	h32 = XXH_rotl32(h32, 11) * PRIME32_1;
	p++;
	}

	h32 ^= h32 >> 15;
	h32 *= PRIME32_2;
	h32 ^= h32 >> 13;
	h32 *= PRIME32_3;
	h32 ^= h32 >> 16;

	return h32;
	}


	XXH_PUBLIC_API unsigned int XXH32_digest (const XXH32_state_t* state_in)
	{
	XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN;

	if ((endian_detected==XXH_littleEndian) \|\| XXH_FORCE_NATIVE_FORMAT)
	return XXH32_digest_endian(state_in, XXH_littleEndian);
	else
	return XXH32_digest_endian(state_in, XXH_bigEndian);
	}



	/* ** XXH64 ** */

	FORCE_INLINE_TEMPLATE XXH_errorcode XXH64_update_endian (XXH64_state_t* state, const void* input, size_t len, XXH_endianess endian)
	{
	const BYTE* p = (const BYTE*)input;
	const BYTE* const bEnd = p + len;

	#ifdef XXH_ACCEPT_NULL_INPUT_POINTER
	if (input==NULL) return XXH_ERROR;
	#endif

	state->total_len += len;

	if (state->memsize + len < 32) { /* fill in tmp buffer */
	if (input != NULL) {
	XXH_memcpy(((BYTE*)state->mem64) + state->memsize, input, len);
	}
	state->memsize += (U32)len;
	return XXH_OK;
	}

	if (state->memsize) { /* tmp buffer is full */
	XXH_memcpy(((BYTE*)state->mem64) + state->memsize, input, 32-state->memsize);
	state->v1 = XXH64_round(state->v1, XXH_readLE64(state->mem64+0, endian));
	state->v2 = XXH64_round(state->v2, XXH_readLE64(state->mem64+1, endian));
	state->v3 = XXH64_round(state->v3, XXH_readLE64(state->mem64+2, endian));
	state->v4 = XXH64_round(state->v4, XXH_readLE64(state->mem64+3, endian));
	p += 32-state->memsize;
	state->memsize = 0;
	}

	if (p+32 <= bEnd) {
	const BYTE* const limit = bEnd - 32;
	U64 v1 = state->v1;
	U64 v2 = state->v2;
	U64 v3 = state->v3;
	U64 v4 = state->v4;

	do {
	v1 = XXH64_round(v1, XXH_readLE64(p, endian)); p+=8;
	v2 = XXH64_round(v2, XXH_readLE64(p, endian)); p+=8;
	v3 = XXH64_round(v3, XXH_readLE64(p, endian)); p+=8;
	v4 = XXH64_round(v4, XXH_readLE64(p, endian)); p+=8;
	} while (p<=limit);

	state->v1 = v1;
	state->v2 = v2;
	state->v3 = v3;
	state->v4 = v4;
	}

	if (p < bEnd) {
	XXH_memcpy(state->mem64, p, (size_t)(bEnd-p));
	state->memsize = (unsigned)(bEnd-p);
	}

	return XXH_OK;
	}

	XXH_PUBLIC_API XXH_errorcode XXH64_update (XXH64_state_t* state_in, const void* input, size_t len)
	{
	XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN;

	if ((endian_detected==XXH_littleEndian) \|\| XXH_FORCE_NATIVE_FORMAT)
	return XXH64_update_endian(state_in, input, len, XXH_littleEndian);
	else
	return XXH64_update_endian(state_in, input, len, XXH_bigEndian);
	}



	FORCE_INLINE_TEMPLATE U64 XXH64_digest_endian (const XXH64_state_t* state, XXH_endianess endian)
	{
	const BYTE * p = (const BYTE*)state->mem64;
	const BYTE* const bEnd = (const BYTE*)state->mem64 + state->memsize;
	U64 h64;

	if (state->total_len >= 32) {
	U64 const v1 = state->v1;
	U64 const v2 = state->v2;
	U64 const v3 = state->v3;
	U64 const v4 = state->v4;

	h64 = XXH_rotl64(v1, 1) + XXH_rotl64(v2, 7) + XXH_rotl64(v3, 12) + XXH_rotl64(v4, 18);
	h64 = XXH64_mergeRound(h64, v1);
	h64 = XXH64_mergeRound(h64, v2);
	h64 = XXH64_mergeRound(h64, v3);
	h64 = XXH64_mergeRound(h64, v4);
	} else {
	h64 = state->v3 + PRIME64_5;
	}

	h64 += (U64) state->total_len;

	while (p+8<=bEnd) {
	U64 const k1 = XXH64_round(0, XXH_readLE64(p, endian));
	h64 ^= k1;
	h64 = XXH_rotl64(h64,27) * PRIME64_1 + PRIME64_4;
	p+=8;
	}

	if (p+4<=bEnd) {
	h64 ^= (U64)(XXH_readLE32(p, endian)) * PRIME64_1;
	h64 = XXH_rotl64(h64, 23) * PRIME64_2 + PRIME64_3;
	p+=4;
	}

	while (p<bEnd) {
	h64 ^= (p) PRIME64_5;
	h64 = XXH_rotl64(h64, 11) * PRIME64_1;
	p++;
	}

	h64 ^= h64 >> 33;
	h64 *= PRIME64_2;
	h64 ^= h64 >> 29;
	h64 *= PRIME64_3;
	h64 ^= h64 >> 32;

	return h64;
	}


	XXH_PUBLIC_API unsigned long long XXH64_digest (const XXH64_state_t* state_in)
	{
	XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN;

	if ((endian_detected==XXH_littleEndian) \|\| XXH_FORCE_NATIVE_FORMAT)
	return XXH64_digest_endian(state_in, XXH_littleEndian);
	else
	return XXH64_digest_endian(state_in, XXH_bigEndian);
	}


	/* **************************
	* Canonical representation
	****************************/

	/*! Default XXH result types are basic unsigned 32 and 64 bits.
	* The canonical representation follows human-readable write convention, aka big-endian (large digits first).
	* These functions allow transformation of hash result into and from its canonical format.
	* This way, hash values can be written into a file or buffer, and remain comparable across different systems and programs.
	*/

	XXH_PUBLIC_API void XXH32_canonicalFromHash(XXH32_canonical_t* dst, XXH32_hash_t hash)
	{
	XXH_STATIC_ASSERT(sizeof(XXH32_canonical_t) == sizeof(XXH32_hash_t));
	if (XXH_CPU_LITTLE_ENDIAN) hash = XXH_swap32(hash);
	memcpy(dst, &hash, sizeof(*dst));
	}

	XXH_PUBLIC_API void XXH64_canonicalFromHash(XXH64_canonical_t* dst, XXH64_hash_t hash)
	{
	XXH_STATIC_ASSERT(sizeof(XXH64_canonical_t) == sizeof(XXH64_hash_t));
	if (XXH_CPU_LITTLE_ENDIAN) hash = XXH_swap64(hash);
	memcpy(dst, &hash, sizeof(*dst));
	}

	XXH_PUBLIC_API XXH32_hash_t XXH32_hashFromCanonical(const XXH32_canonical_t* src)
	{
	return XXH_readBE32(src);
	}

	XXH_PUBLIC_API XXH64_hash_t XXH64_hashFromCanonical(const XXH64_canonical_t* src)
	{
	return XXH_readBE64(src);
	}
	/** ended inlining xxhash.c **/
	# endif

	#endif /* XXH_STATIC_LINKING_ONLY && XXH_STATIC_H_3543687687345 */


	#if defined (__cplusplus)
	}
	#endif
	/** ended inlining xxhash.h **/

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/* ---- static assert (debug) --- */
	#define ZSTD_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c)
	-#define ZSTD_isError ERR_isError /* for inlining */
	#define FSE_isError ERR_isError
	#define HUF_isError ERR_isError


	/-************************************
	* shared macros
	***************************************/
	#undef MIN
	#undef MAX
	#define MIN(a,b) ((a)<(b) ? (a) : (b))
	#define MAX(a,b) ((a)>(b) ? (a) : (b))

	/**
	* Ignore: this is an internal helper.
	*
	* This is a helper function to help force C99-correctness during compilation.
	* Under strict compilation modes, variadic macro arguments can't be empty.
	* However, variadic function arguments can be. Using a function therefore lets
	* us statically check that at least one (string) argument was passed,
	* independent of the compilation flags.
	*/
	static INLINE_KEYWORD UNUSED_ATTR
	void _force_has_format_string(const char *format, ...) {
	(void)format;
	}

	/**
	* Ignore: this is an internal helper.
	*
	* We want to force this function invocation to be syntactically correct, but
	* we don't want to force runtime evaluation of its arguments.
	*/
	#define _FORCE_HAS_FORMAT_STRING(...) \
	if (0) { \
	_force_has_format_string(__VA_ARGS__); \
	}

	/**
	* Return the specified error if the condition evaluates to true.
	*
	* In debug modes, prints additional information.
	* In order to do that (particularly, printing the conditional that failed),
	* this can't just wrap RETURN_ERROR().
	*/
	#define RETURN_ERROR_IF(cond, err, ...) \
	if (cond) { \
	RAWLOG(3, "%s:%d: ERROR!: check %s failed, returning %s", \
	__FILE__, __LINE__, ZSTD_QUOTE(cond), ZSTD_QUOTE(ERROR(err))); \
	_FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
	RAWLOG(3, ": " __VA_ARGS__); \
	RAWLOG(3, "\n"); \
	return ERROR(err); \
	}

	/**
	* Unconditionally return the specified error.
	*
	* In debug modes, prints additional information.
	*/
	#define RETURN_ERROR(err, ...) \
	do { \
	RAWLOG(3, "%s:%d: ERROR!: unconditional check failed, returning %s", \
	__FILE__, __LINE__, ZSTD_QUOTE(ERROR(err))); \
	_FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
	RAWLOG(3, ": " __VA_ARGS__); \
	RAWLOG(3, "\n"); \
	return ERROR(err); \
	} while(0);

	/**
	* If the provided expression evaluates to an error code, returns that error code.
	*
	* In debug modes, prints additional information.
	*/
	#define FORWARD_IF_ERROR(err, ...) \
	do { \
	size_t const err_code = (err); \
	if (ERR_isError(err_code)) { \
	RAWLOG(3, "%s:%d: ERROR!: forwarding error in %s: %s", \
	__FILE__, __LINE__, ZSTD_QUOTE(err), ERR_getErrorName(err_code)); \
	_FORCE_HAS_FORMAT_STRING(__VA_ARGS__); \
	RAWLOG(3, ": " __VA_ARGS__); \
	RAWLOG(3, "\n"); \
	return err_code; \
	} \
	} while(0);


	/-************************************
	* Common constants
	***************************************/
	#define ZSTD_OPT_NUM (1<<12)

	#define ZSTD_REP_NUM 3 /* number of repcodes */
	#define ZSTD_REP_MOVE (ZSTD_REP_NUM-1)
	static const U32 repStartValue[ZSTD_REP_NUM] = { 1, 4, 8 };

	#define KB *(1 <<10)
	#define MB *(1 <<20)
	#define GB *(1U<<30)

	#define BIT7 128
	#define BIT6 64
	#define BIT5 32
	#define BIT4 16
	#define BIT1 2
	#define BIT0 1

	#define ZSTD_WINDOWLOG_ABSOLUTEMIN 10
	static const size_t ZSTD_fcs_fieldSize[4] = { 0, 2, 4, 8 };
	static const size_t ZSTD_did_fieldSize[4] = { 0, 1, 2, 4 };

	#define ZSTD_FRAMEIDSIZE 4 /* magic number size */

	#define ZSTD_BLOCKHEADERSIZE 3 /* C standard doesn't allow `static const` variable to be init using another `static const` variable */
	static const size_t ZSTD_blockHeaderSize = ZSTD_BLOCKHEADERSIZE;
	typedef enum { bt_raw, bt_rle, bt_compressed, bt_reserved } blockType_e;

	#define ZSTD_FRAMECHECKSUMSIZE 4

	#define MIN_SEQUENCES_SIZE 1 /* nbSeq==0 */
	#define MIN_CBLOCK_SIZE (1 /litCSize/ + 1 /* RLE or RAW / + MIN_SEQUENCES_SIZE / nbSeq==0 /) / for a non-null block */

	#define HufLog 12
	typedef enum { set_basic, set_rle, set_compressed, set_repeat } symbolEncodingType_e;

	#define LONGNBSEQ 0x7F00

	#define MINMATCH 3

	#define Litbits 8
	#define MaxLit ((1<<Litbits) - 1)
	#define MaxML 52
	#define MaxLL 35
	#define DefaultMaxOff 28
	#define MaxOff 31
	#define MaxSeq MAX(MaxLL, MaxML) /* Assumption : MaxOff < MaxLL,MaxML */
	#define MLFSELog 9
	#define LLFSELog 9
	#define OffFSELog 8
	#define MaxFSELog MAX(MAX(MLFSELog, LLFSELog), OffFSELog)

	static const U32 LL_bits[MaxLL+1] = { 0, 0, 0, 0, 0, 0, 0, 0,
	0, 0, 0, 0, 0, 0, 0, 0,
	1, 1, 1, 1, 2, 2, 3, 3,
	4, 6, 7, 8, 9,10,11,12,
	13,14,15,16 };
	static const S16 LL_defaultNorm[MaxLL+1] = { 4, 3, 2, 2, 2, 2, 2, 2,
	2, 2, 2, 2, 2, 1, 1, 1,
	2, 2, 2, 2, 2, 2, 2, 2,
	2, 3, 2, 1, 1, 1, 1, 1,
	-1,-1,-1,-1 };
	#define LL_DEFAULTNORMLOG 6 /* for static allocation */
	static const U32 LL_defaultNormLog = LL_DEFAULTNORMLOG;

	static const U32 ML_bits[MaxML+1] = { 0, 0, 0, 0, 0, 0, 0, 0,
	0, 0, 0, 0, 0, 0, 0, 0,
	0, 0, 0, 0, 0, 0, 0, 0,
	0, 0, 0, 0, 0, 0, 0, 0,
	1, 1, 1, 1, 2, 2, 3, 3,
	4, 4, 5, 7, 8, 9,10,11,
	12,13,14,15,16 };
	static const S16 ML_defaultNorm[MaxML+1] = { 1, 4, 3, 2, 2, 2, 2, 2,
	2, 1, 1, 1, 1, 1, 1, 1,
	1, 1, 1, 1, 1, 1, 1, 1,
	1, 1, 1, 1, 1, 1, 1, 1,
	1, 1, 1, 1, 1, 1, 1, 1,
	1, 1, 1, 1, 1, 1,-1,-1,
	-1,-1,-1,-1,-1 };
	#define ML_DEFAULTNORMLOG 6 /* for static allocation */
	static const U32 ML_defaultNormLog = ML_DEFAULTNORMLOG;

	static const S16 OF_defaultNorm[DefaultMaxOff+1] = { 1, 1, 1, 1, 1, 1, 2, 2,
	2, 1, 1, 1, 1, 1, 1, 1,
	1, 1, 1, 1, 1, 1, 1, 1,
	-1,-1,-1,-1,-1 };
	#define OF_DEFAULTNORMLOG 5 /* for static allocation */
	static const U32 OF_defaultNormLog = OF_DEFAULTNORMLOG;


	/-******************************************
	* Shared functions to include for inlining
	*********************************************/
	static void ZSTD_copy8(void* dst, const void* src) {
	#if !defined(ZSTD_NO_INTRINSICS) && defined(__ARM_NEON)
	vst1_u8((uint8_t)dst, vld1_u8((const uint8_t)src));
	#else
	memcpy(dst, src, 8);
	#endif
	}

	#define COPY8(d,s) { ZSTD_copy8(d,s); d+=8; s+=8; }
	static void ZSTD_copy16(void* dst, const void* src) {
	#if !defined(ZSTD_NO_INTRINSICS) && defined(__ARM_NEON)
	vst1q_u8((uint8_t)dst, vld1q_u8((const uint8_t)src));
	#else
	memcpy(dst, src, 16);
	#endif
	}
	#define COPY16(d,s) { ZSTD_copy16(d,s); d+=16; s+=16; }

	#define WILDCOPY_OVERLENGTH 32
	#define WILDCOPY_VECLEN 16

	typedef enum {
	ZSTD_no_overlap,
	ZSTD_overlap_src_before_dst
	/* ZSTD_overlap_dst_before_src, */
	} ZSTD_overlap_e;

	/*! ZSTD_wildcopy() :
	* Custom version of memcpy(), can over read/write up to WILDCOPY_OVERLENGTH bytes (if length==0)
	* @param ovtype controls the overlap detection
	* - ZSTD_no_overlap: The source and destination are guaranteed to be at least WILDCOPY_VECLEN bytes apart.
	* - ZSTD_overlap_src_before_dst: The src and dst may overlap, but they MUST be at least 8 bytes apart.
	* The src buffer must be before the dst buffer.
	*/
	MEM_STATIC FORCE_INLINE_ATTR
	void ZSTD_wildcopy(void* dst, const void* src, ptrdiff_t length, ZSTD_overlap_e const ovtype)
	{
	ptrdiff_t diff = (BYTE)dst - (const BYTE)src;
	const BYTE* ip = (const BYTE*)src;
	BYTE* op = (BYTE*)dst;
	BYTE* const oend = op + length;

	assert(diff >= 8 \|\| (ovtype == ZSTD_no_overlap && diff <= -WILDCOPY_VECLEN));

	if (ovtype == ZSTD_overlap_src_before_dst && diff < WILDCOPY_VECLEN) {
	/* Handle short offset copies. */
	do {
	COPY8(op, ip)
	} while (op < oend);
	} else {
	assert(diff >= WILDCOPY_VECLEN \|\| diff <= -WILDCOPY_VECLEN);
	/* Separate out the first COPY16() call because the copy length is
	* almost certain to be short, so the branches have different
	* probabilities. Since it is almost certain to be short, only do
	* one COPY16() in the first call. Then, do two calls per loop since
	* at that point it is more likely to have a high trip count.
	*/
	#ifndef __aarch64__
	do {
	COPY16(op, ip);
	}
	while (op < oend);
	#else
	COPY16(op, ip);
	if (op >= oend) return;
	do {
	COPY16(op, ip);
	COPY16(op, ip);
	}
	while (op < oend);
	#endif
	}
	}

	MEM_STATIC size_t ZSTD_limitCopy(void* dst, size_t dstCapacity, const void* src, size_t srcSize)
	{
	size_t const length = MIN(dstCapacity, srcSize);
	if (length > 0) {
	memcpy(dst, src, length);
	}
	return length;
	}

	/* define "workspace is too large" as this number of times larger than needed */
	#define ZSTD_WORKSPACETOOLARGE_FACTOR 3

	/* when workspace is continuously too large
	* during at least this number of times,
	* context's memory usage is considered wasteful,
	* because it's sized to handle a worst case scenario which rarely happens.
	* In which case, resize it down to free some memory */
	#define ZSTD_WORKSPACETOOLARGE_MAXDURATION 128


	/-******************************************
	* Private declarations
	*********************************************/
	typedef struct seqDef_s {
	U32 offset;
	U16 litLength;
	U16 matchLength;
	} seqDef;

	typedef struct {
	seqDef* sequencesStart;
	seqDef* sequences;
	BYTE* litStart;
	BYTE* lit;
	BYTE* llCode;
	BYTE* mlCode;
	BYTE* ofCode;
	size_t maxNbSeq;
	size_t maxNbLit;
	U32 longLengthID; /* 0 == no longLength; 1 == Lit.longLength; 2 == Match.longLength; */
	U32 longLengthPos;
	} seqStore_t;

	typedef struct {
	U32 litLength;
	U32 matchLength;
	} ZSTD_sequenceLength;

	/**
	* Returns the ZSTD_sequenceLength for the given sequences. It handles the decoding of long sequences
	* indicated by longLengthPos and longLengthID, and adds MINMATCH back to matchLength.
	*/
	MEM_STATIC ZSTD_sequenceLength ZSTD_getSequenceLength(seqStore_t const* seqStore, seqDef const* seq)
	{
	ZSTD_sequenceLength seqLen;
	seqLen.litLength = seq->litLength;
	seqLen.matchLength = seq->matchLength + MINMATCH;
	if (seqStore->longLengthPos == (U32)(seq - seqStore->sequencesStart)) {
	if (seqStore->longLengthID == 1) {
	seqLen.litLength += 0xFFFF;
	}
	if (seqStore->longLengthID == 2) {
	seqLen.matchLength += 0xFFFF;
	}
	}
	return seqLen;
	}

	/**
	* Contains the compressed frame size and an upper-bound for the decompressed frame size.
	* Note: before using `compressedSize`, check for errors using ZSTD_isError().
	* similarly, before using `decompressedBound`, check for errors using:
	* `decompressedBound != ZSTD_CONTENTSIZE_ERROR`
	*/
	typedef struct {
	size_t compressedSize;
	unsigned long long decompressedBound;
	} ZSTD_frameSizeInfo; /* decompress & legacy */

	const seqStore_t* ZSTD_getSeqStore(const ZSTD_CCtx* ctx); /* compress & dictBuilder */
	void ZSTD_seqToCodes(const seqStore_t* seqStorePtr); /* compress, dictBuilder, decodeCorpus (shouldn't get its definition from here) */

	/* custom memory allocation functions */
	void* ZSTD_malloc(size_t size, ZSTD_customMem customMem);
	void* ZSTD_calloc(size_t size, ZSTD_customMem customMem);
	void ZSTD_free(void* ptr, ZSTD_customMem customMem);


	MEM_STATIC U32 ZSTD_highbit32(U32 val) /* compress, dictBuilder, decodeCorpus */
	{
	assert(val != 0);
	{
	# if defined(_MSC_VER) /* Visual */
	unsigned long r=0;
	return _BitScanReverse(&r, val) ? (unsigned)r : 0;
	# elif defined(__GNUC__) && (__GNUC__ >= 3) /* GCC Intrinsic */
	return __builtin_clz (val) ^ 31;
	# elif defined(__ICCARM__) /* IAR Intrinsic */
	return 31 - __CLZ(val);
	# else /* Software version */
	static const U32 DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29, 11, 14, 16, 18, 22, 25, 3, 30, 8, 12, 20, 28, 15, 17, 24, 7, 19, 27, 23, 6, 26, 5, 4, 31 };
	U32 v = val;
	v \|= v >> 1;
	v \|= v >> 2;
	v \|= v >> 4;
	v \|= v >> 8;
	v \|= v >> 16;
	return DeBruijnClz[(v * 0x07C4ACDDU) >> 27];
	# endif
	}
	}


	/* ZSTD_invalidateRepCodes() :
	* ensures next compression will not use repcodes from previous block.
	* Note : only works with regular variant;
	* do not use with extDict variant ! */
	void ZSTD_invalidateRepCodes(ZSTD_CCtx* cctx); /* zstdmt, adaptive_compression (shouldn't get this definition from here) */


	typedef struct {
	blockType_e blockType;
	U32 lastBlock;
	U32 origSize;
	} blockProperties_t; /* declared here for decompress and fullbench */

	/*! ZSTD_getcBlockSize() :
	* Provides the size of compressed block from block header `src` */
	/* Used by: decompress, fullbench (does not get its definition from here) */
	size_t ZSTD_getcBlockSize(const void* src, size_t srcSize,
	blockProperties_t* bpPtr);

	/*! ZSTD_decodeSeqHeaders() :
	* decode sequence header from src */
	/* Used by: decompress, fullbench (does not get its definition from here) */
	size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr,
	const void* src, size_t srcSize);


	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_CCOMMON_H_MODULE */
	/** ended inlining zstd_internal.h **/
	/** start inlining pool.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef POOL_H
	#define POOL_H

	#if defined (__cplusplus)
	extern "C" {
	#endif


	#include <stddef.h> /* size_t */
	#define ZSTD_STATIC_LINKING_ONLY /* ZSTD_customMem */
	/** skipping file: ../zstd.h **/

	typedef struct POOL_ctx_s POOL_ctx;

	/*! POOL_create() :
	* Create a thread pool with at most `numThreads` threads.
	* `numThreads` must be at least 1.
	* The maximum number of queued jobs before blocking is `queueSize`.
	* @return : POOL_ctx pointer on success, else NULL.
	*/
	POOL_ctx* POOL_create(size_t numThreads, size_t queueSize);

	POOL_ctx* POOL_create_advanced(size_t numThreads, size_t queueSize,
	ZSTD_customMem customMem);

	/*! POOL_free() :
	* Free a thread pool returned by POOL_create().
	*/
	void POOL_free(POOL_ctx* ctx);

	/*! POOL_resize() :
	* Expands or shrinks pool's number of threads.
	* This is more efficient than releasing + creating a new context,
	* since it tries to preserve and re-use existing threads.
	* `numThreads` must be at least 1.
	* @return : 0 when resize was successful,
	* !0 (typically 1) if there is an error.
	* note : only numThreads can be resized, queueSize remains unchanged.
	*/
	int POOL_resize(POOL_ctx* ctx, size_t numThreads);

	/*! POOL_sizeof() :
	* @return threadpool memory usage
	* note : compatible with NULL (returns 0 in this case)
	*/
	size_t POOL_sizeof(POOL_ctx* ctx);

	/*! POOL_function :
	* The function type that can be added to a thread pool.
	*/
	typedef void (POOL_function)(void);

	/*! POOL_add() :
	* Add the job `function(opaque)` to the thread pool. `ctx` must be valid.
	* Possibly blocks until there is room in the queue.
	* Note : The function may be executed asynchronously,
	* therefore, `opaque` must live until function has been completed.
	*/
	void POOL_add(POOL_ctx* ctx, POOL_function function, void* opaque);


	/*! POOL_tryAdd() :
	* Add the job `function(opaque)` to thread pool _if_ a worker is available.
	* Returns immediately even if not (does not block).
	* @return : 1 if successful, 0 if not.
	*/
	int POOL_tryAdd(POOL_ctx* ctx, POOL_function function, void* opaque);


	#if defined (__cplusplus)
	}
	#endif

	#endif
	/** ended inlining pool.h **/

	/* ====== Compiler specifics ====== */
	#if defined(_MSC_VER)
	# pragma warning(disable : 4204) /* disable: C4204: non-constant aggregate initializer */
	#endif


	#ifdef ZSTD_MULTITHREAD

	/** start inlining threading.h **/
	/**
	* Copyright (c) 2016 Tino Reichardt
	* All rights reserved.
	*
	* You can contact the author at:
	* - zstdmt source repository: https://github.com/mcmilk/zstdmt
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef THREADING_H_938743
	#define THREADING_H_938743

	/** skipping file: debug.h **/

	#if defined (__cplusplus)
	extern "C" {
	#endif

	#if defined(ZSTD_MULTITHREAD) && defined(_WIN32)

	/**
	* Windows minimalist Pthread Wrapper, based on :
	* http://www.cse.wustl.edu/~schmidt/win32-cv-1.html
	*/
	#ifdef WINVER
	# undef WINVER
	#endif
	#define WINVER 0x0600

	#ifdef _WIN32_WINNT
	# undef _WIN32_WINNT
	#endif
	#define _WIN32_WINNT 0x0600

	#ifndef WIN32_LEAN_AND_MEAN
	# define WIN32_LEAN_AND_MEAN
	#endif

	#undef ERROR /* reported already defined on VS 2015 (Rich Geldreich) */
	#include <windows.h>
	#undef ERROR
	#define ERROR(name) ZSTD_ERROR(name)


	/* mutex */
	#define ZSTD_pthread_mutex_t CRITICAL_SECTION
	#define ZSTD_pthread_mutex_init(a, b) ((void)(b), InitializeCriticalSection((a)), 0)
	#define ZSTD_pthread_mutex_destroy(a) DeleteCriticalSection((a))
	#define ZSTD_pthread_mutex_lock(a) EnterCriticalSection((a))
	#define ZSTD_pthread_mutex_unlock(a) LeaveCriticalSection((a))

	/* condition variable */
	#define ZSTD_pthread_cond_t CONDITION_VARIABLE
	#define ZSTD_pthread_cond_init(a, b) ((void)(b), InitializeConditionVariable((a)), 0)
	#define ZSTD_pthread_cond_destroy(a) ((void)(a))
	#define ZSTD_pthread_cond_wait(a, b) SleepConditionVariableCS((a), (b), INFINITE)
	#define ZSTD_pthread_cond_signal(a) WakeConditionVariable((a))
	#define ZSTD_pthread_cond_broadcast(a) WakeAllConditionVariable((a))

	/* ZSTD_pthread_create() and ZSTD_pthread_join() */
	typedef struct {
	HANDLE handle;
	void* (start_routine)(void);
	void* arg;
	} ZSTD_pthread_t;

	int ZSTD_pthread_create(ZSTD_pthread_t* thread, const void* unused,
	void* (start_routine) (void), void* arg);

	int ZSTD_pthread_join(ZSTD_pthread_t thread, void** value_ptr);

	/**
	* add here more wrappers as required
	*/


	#elif defined(ZSTD_MULTITHREAD) /* posix assumed ; need a better detection method */
	/* === POSIX Systems === */
	# include <pthread.h>

	#if DEBUGLEVEL < 1

	#define ZSTD_pthread_mutex_t pthread_mutex_t
	#define ZSTD_pthread_mutex_init(a, b) pthread_mutex_init((a), (b))
	#define ZSTD_pthread_mutex_destroy(a) pthread_mutex_destroy((a))
	#define ZSTD_pthread_mutex_lock(a) pthread_mutex_lock((a))
	#define ZSTD_pthread_mutex_unlock(a) pthread_mutex_unlock((a))

	#define ZSTD_pthread_cond_t pthread_cond_t
	#define ZSTD_pthread_cond_init(a, b) pthread_cond_init((a), (b))
	#define ZSTD_pthread_cond_destroy(a) pthread_cond_destroy((a))
	#define ZSTD_pthread_cond_wait(a, b) pthread_cond_wait((a), (b))
	#define ZSTD_pthread_cond_signal(a) pthread_cond_signal((a))
	#define ZSTD_pthread_cond_broadcast(a) pthread_cond_broadcast((a))

	#define ZSTD_pthread_t pthread_t
	#define ZSTD_pthread_create(a, b, c, d) pthread_create((a), (b), (c), (d))
	#define ZSTD_pthread_join(a, b) pthread_join((a),(b))

	#else /* DEBUGLEVEL >= 1 */

	/* Debug implementation of threading.
	* In this implementation we use pointers for mutexes and condition variables.
	* This way, if we forget to init/destroy them the program will crash or ASAN
	* will report leaks.
	*/

	#define ZSTD_pthread_mutex_t pthread_mutex_t*
	int ZSTD_pthread_mutex_init(ZSTD_pthread_mutex_t* mutex, pthread_mutexattr_t const* attr);
	int ZSTD_pthread_mutex_destroy(ZSTD_pthread_mutex_t* mutex);
	#define ZSTD_pthread_mutex_lock(a) pthread_mutex_lock(*(a))
	#define ZSTD_pthread_mutex_unlock(a) pthread_mutex_unlock(*(a))

	#define ZSTD_pthread_cond_t pthread_cond_t*
	int ZSTD_pthread_cond_init(ZSTD_pthread_cond_t* cond, pthread_condattr_t const* attr);
	int ZSTD_pthread_cond_destroy(ZSTD_pthread_cond_t* cond);
	#define ZSTD_pthread_cond_wait(a, b) pthread_cond_wait((a), (b))
	#define ZSTD_pthread_cond_signal(a) pthread_cond_signal(*(a))
	#define ZSTD_pthread_cond_broadcast(a) pthread_cond_broadcast(*(a))

	#define ZSTD_pthread_t pthread_t
	#define ZSTD_pthread_create(a, b, c, d) pthread_create((a), (b), (c), (d))
	#define ZSTD_pthread_join(a, b) pthread_join((a),(b))

	#endif

	#else /* ZSTD_MULTITHREAD not defined */
	/* No multithreading support */

	typedef int ZSTD_pthread_mutex_t;
	#define ZSTD_pthread_mutex_init(a, b) ((void)(a), (void)(b), 0)
	#define ZSTD_pthread_mutex_destroy(a) ((void)(a))
	#define ZSTD_pthread_mutex_lock(a) ((void)(a))
	#define ZSTD_pthread_mutex_unlock(a) ((void)(a))

	typedef int ZSTD_pthread_cond_t;
	#define ZSTD_pthread_cond_init(a, b) ((void)(a), (void)(b), 0)
	#define ZSTD_pthread_cond_destroy(a) ((void)(a))
	#define ZSTD_pthread_cond_wait(a, b) ((void)(a), (void)(b))
	#define ZSTD_pthread_cond_signal(a) ((void)(a))
	#define ZSTD_pthread_cond_broadcast(a) ((void)(a))

	/* do not use ZSTD_pthread_t */

	#endif /* ZSTD_MULTITHREAD */

	#if defined (__cplusplus)
	}
	#endif

	#endif /* THREADING_H_938743 */
	/** ended inlining threading.h **/

	/* A job is a function and an opaque argument */
	typedef struct POOL_job_s {
	POOL_function function;
	void *opaque;
	} POOL_job;

	struct POOL_ctx_s {
	ZSTD_customMem customMem;
	/* Keep track of the threads */
	ZSTD_pthread_t* threads;
	size_t threadCapacity;
	size_t threadLimit;

	/* The queue is a circular buffer */
	POOL_job *queue;
	size_t queueHead;
	size_t queueTail;
	size_t queueSize;

	/* The number of threads working on jobs */
	size_t numThreadsBusy;
	/* Indicates if the queue is empty */
	int queueEmpty;

	/* The mutex protects the queue */
	ZSTD_pthread_mutex_t queueMutex;
	/* Condition variable for pushers to wait on when the queue is full */
	ZSTD_pthread_cond_t queuePushCond;
	/* Condition variables for poppers to wait on when the queue is empty */
	ZSTD_pthread_cond_t queuePopCond;
	/* Indicates if the queue is shutting down */
	int shutdown;
	};

	/* POOL_thread() :
	* Work thread for the thread pool.
	* Waits for jobs and executes them.
	* @returns : NULL on failure else non-null.
	*/
	static void* POOL_thread(void* opaque) {
	POOL_ctx* const ctx = (POOL_ctx*)opaque;
	if (!ctx) { return NULL; }
	for (;;) {
	/* Lock the mutex and wait for a non-empty queue or until shutdown */
	ZSTD_pthread_mutex_lock(&ctx->queueMutex);

	while ( ctx->queueEmpty
	\|\| (ctx->numThreadsBusy >= ctx->threadLimit) ) {
	if (ctx->shutdown) {
	/* even if !queueEmpty, (possible if numThreadsBusy >= threadLimit),
	* a few threads will be shutdown while !queueEmpty,
	* but enough threads will remain active to finish the queue */
	ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
	return opaque;
	}
	ZSTD_pthread_cond_wait(&ctx->queuePopCond, &ctx->queueMutex);
	}
	/* Pop a job off the queue */
	{ POOL_job const job = ctx->queue[ctx->queueHead];
	ctx->queueHead = (ctx->queueHead + 1) % ctx->queueSize;
	ctx->numThreadsBusy++;
	ctx->queueEmpty = ctx->queueHead == ctx->queueTail;
	/* Unlock the mutex, signal a pusher, and run the job */
	ZSTD_pthread_cond_signal(&ctx->queuePushCond);
	ZSTD_pthread_mutex_unlock(&ctx->queueMutex);

	job.function(job.opaque);

	/* If the intended queue size was 0, signal after finishing job */
	ZSTD_pthread_mutex_lock(&ctx->queueMutex);
	ctx->numThreadsBusy--;
	if (ctx->queueSize == 1) {
	ZSTD_pthread_cond_signal(&ctx->queuePushCond);
	}
	ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
	}
	} /* for (;;) */
	assert(0); /* Unreachable */
	}

	POOL_ctx* POOL_create(size_t numThreads, size_t queueSize) {
	return POOL_create_advanced(numThreads, queueSize, ZSTD_defaultCMem);
	}

	POOL_ctx* POOL_create_advanced(size_t numThreads, size_t queueSize,
	ZSTD_customMem customMem) {
	POOL_ctx* ctx;
	/* Check parameters */
	if (!numThreads) { return NULL; }
	/* Allocate the context and zero initialize */
	ctx = (POOL_ctx*)ZSTD_calloc(sizeof(POOL_ctx), customMem);
	if (!ctx) { return NULL; }
	/* Initialize the job queue.
	* It needs one extra space since one space is wasted to differentiate
	* empty and full queues.
	*/
	ctx->queueSize = queueSize + 1;
	ctx->queue = (POOL_job)ZSTD_malloc(ctx->queueSize sizeof(POOL_job), customMem);
	ctx->queueHead = 0;
	ctx->queueTail = 0;
	ctx->numThreadsBusy = 0;
	ctx->queueEmpty = 1;
	{
	int error = 0;
	error \|= ZSTD_pthread_mutex_init(&ctx->queueMutex, NULL);
	error \|= ZSTD_pthread_cond_init(&ctx->queuePushCond, NULL);
	error \|= ZSTD_pthread_cond_init(&ctx->queuePopCond, NULL);
	if (error) { POOL_free(ctx); return NULL; }
	}
	ctx->shutdown = 0;
	/* Allocate space for the thread handles */
	ctx->threads = (ZSTD_pthread_t)ZSTD_malloc(numThreads sizeof(ZSTD_pthread_t), customMem);
	ctx->threadCapacity = 0;
	ctx->customMem = customMem;
	/* Check for errors */
	if (!ctx->threads \|\| !ctx->queue) { POOL_free(ctx); return NULL; }
	/* Initialize the threads */
	{ size_t i;
	for (i = 0; i < numThreads; ++i) {
	if (ZSTD_pthread_create(&ctx->threads[i], NULL, &POOL_thread, ctx)) {
	ctx->threadCapacity = i;
	POOL_free(ctx);
	return NULL;
	} }
	ctx->threadCapacity = numThreads;
	ctx->threadLimit = numThreads;
	}
	return ctx;
	}

	/*! POOL_join() :
	Shutdown the queue, wake any sleeping threads, and join all of the threads.
	*/
	static void POOL_join(POOL_ctx* ctx) {
	/* Shut down the queue */
	ZSTD_pthread_mutex_lock(&ctx->queueMutex);
	ctx->shutdown = 1;
	ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
	/* Wake up sleeping threads */
	ZSTD_pthread_cond_broadcast(&ctx->queuePushCond);
	ZSTD_pthread_cond_broadcast(&ctx->queuePopCond);
	/* Join all of the threads */
	{ size_t i;
	for (i = 0; i < ctx->threadCapacity; ++i) {
	ZSTD_pthread_join(ctx->threads[i], NULL); /* note : could fail */
	} }
	}

	void POOL_free(POOL_ctx *ctx) {
	if (!ctx) { return; }
	POOL_join(ctx);
	ZSTD_pthread_mutex_destroy(&ctx->queueMutex);
	ZSTD_pthread_cond_destroy(&ctx->queuePushCond);
	ZSTD_pthread_cond_destroy(&ctx->queuePopCond);
	ZSTD_free(ctx->queue, ctx->customMem);
	ZSTD_free(ctx->threads, ctx->customMem);
	ZSTD_free(ctx, ctx->customMem);
	}



	size_t POOL_sizeof(POOL_ctx *ctx) {
	if (ctx==NULL) return 0; /* supports sizeof NULL */
	return sizeof(*ctx)
	+ ctx->queueSize * sizeof(POOL_job)
	+ ctx->threadCapacity * sizeof(ZSTD_pthread_t);
	}


	/* @return : 0 on success, 1 on error */
	static int POOL_resize_internal(POOL_ctx* ctx, size_t numThreads)
	{
	if (numThreads <= ctx->threadCapacity) {
	if (!numThreads) return 1;
	ctx->threadLimit = numThreads;
	return 0;
	}
	/* numThreads > threadCapacity */
	{ ZSTD_pthread_t* const threadPool = (ZSTD_pthread_t)ZSTD_malloc(numThreads sizeof(ZSTD_pthread_t), ctx->customMem);
	if (!threadPool) return 1;
	/* replace existing thread pool */
	memcpy(threadPool, ctx->threads, ctx->threadCapacity * sizeof(*threadPool));
	ZSTD_free(ctx->threads, ctx->customMem);
	ctx->threads = threadPool;
	/* Initialize additional threads */
	{ size_t threadId;
	for (threadId = ctx->threadCapacity; threadId < numThreads; ++threadId) {
	if (ZSTD_pthread_create(&threadPool[threadId], NULL, &POOL_thread, ctx)) {
	ctx->threadCapacity = threadId;
	return 1;
	} }
	} }
	/* successfully expanded */
	ctx->threadCapacity = numThreads;
	ctx->threadLimit = numThreads;
	return 0;
	}

	/* @return : 0 on success, 1 on error */
	int POOL_resize(POOL_ctx* ctx, size_t numThreads)
	{
	int result;
	if (ctx==NULL) return 1;
	ZSTD_pthread_mutex_lock(&ctx->queueMutex);
	result = POOL_resize_internal(ctx, numThreads);
	ZSTD_pthread_cond_broadcast(&ctx->queuePopCond);
	ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
	return result;
	}

	/**
	* Returns 1 if the queue is full and 0 otherwise.
	*
	* When queueSize is 1 (pool was created with an intended queueSize of 0),
	* then a queue is empty if there is a thread free _and_ no job is waiting.
	*/
	static int isQueueFull(POOL_ctx const* ctx) {
	if (ctx->queueSize > 1) {
	return ctx->queueHead == ((ctx->queueTail + 1) % ctx->queueSize);
	} else {
	return (ctx->numThreadsBusy == ctx->threadLimit) \|\|
	!ctx->queueEmpty;
	}
	}


	static void POOL_add_internal(POOL_ctx* ctx, POOL_function function, void *opaque)
	{
	POOL_job const job = {function, opaque};
	assert(ctx != NULL);
	if (ctx->shutdown) return;

	ctx->queueEmpty = 0;
	ctx->queue[ctx->queueTail] = job;
	ctx->queueTail = (ctx->queueTail + 1) % ctx->queueSize;
	ZSTD_pthread_cond_signal(&ctx->queuePopCond);
	}

	void POOL_add(POOL_ctx* ctx, POOL_function function, void* opaque)
	{
	assert(ctx != NULL);
	ZSTD_pthread_mutex_lock(&ctx->queueMutex);
	/* Wait until there is space in the queue for the new job */
	while (isQueueFull(ctx) && (!ctx->shutdown)) {
	ZSTD_pthread_cond_wait(&ctx->queuePushCond, &ctx->queueMutex);
	}
	POOL_add_internal(ctx, function, opaque);
	ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
	}


	int POOL_tryAdd(POOL_ctx* ctx, POOL_function function, void* opaque)
	{
	assert(ctx != NULL);
	ZSTD_pthread_mutex_lock(&ctx->queueMutex);
	if (isQueueFull(ctx)) {
	ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
	return 0;
	}
	POOL_add_internal(ctx, function, opaque);
	ZSTD_pthread_mutex_unlock(&ctx->queueMutex);
	return 1;
	}


	#else /* ZSTD_MULTITHREAD not defined */

	/* ========================== */
	/* No multi-threading support */
	/* ========================== */


	/* We don't need any data, but if it is empty, malloc() might return NULL. */
	struct POOL_ctx_s {
	int dummy;
	};
	static POOL_ctx g_ctx;

	POOL_ctx* POOL_create(size_t numThreads, size_t queueSize) {
	return POOL_create_advanced(numThreads, queueSize, ZSTD_defaultCMem);
	}

	POOL_ctx* POOL_create_advanced(size_t numThreads, size_t queueSize, ZSTD_customMem customMem) {
	(void)numThreads;
	(void)queueSize;
	(void)customMem;
	return &g_ctx;
	}

	void POOL_free(POOL_ctx* ctx) {
	assert(!ctx \|\| ctx == &g_ctx);
	(void)ctx;
	}

	int POOL_resize(POOL_ctx* ctx, size_t numThreads) {
	(void)ctx; (void)numThreads;
	return 0;
	}

	void POOL_add(POOL_ctx* ctx, POOL_function function, void* opaque) {
	(void)ctx;
	function(opaque);
	}

	int POOL_tryAdd(POOL_ctx* ctx, POOL_function function, void* opaque) {
	(void)ctx;
	function(opaque);
	return 1;
	}

	size_t POOL_sizeof(POOL_ctx* ctx) {
	if (ctx==NULL) return 0; /* supports sizeof NULL */
	assert(ctx == &g_ctx);
	return sizeof(*ctx);
	}

	#endif /* ZSTD_MULTITHREAD */
	/** ended inlining common/pool.c **/
	/** start inlining common/zstd_common.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/



	/-************************************
	* Dependencies
	***************************************/
	#include <stdlib.h> /* malloc, calloc, free */
	#include <string.h> /* memset */
	/** skipping file: error_private.h **/
	/** skipping file: zstd_internal.h **/


	/-***************************************
	* Version
	******************************************/
	unsigned ZSTD_versionNumber(void) { return ZSTD_VERSION_NUMBER; }

	const char* ZSTD_versionString(void) { return ZSTD_VERSION_STRING; }


	/-***************************************
	* ZSTD Error Management
	******************************************/
	-#undef ZSTD_isError /* defined within zstd_internal.h */
	/*! ZSTD_isError() :
	* tells if a return value is an error code
	* symbol is required for external callers */
	unsigned ZSTD_isError(size_t code) { return ERR_isError(code); }

	/*! ZSTD_getErrorName() :
	* provides error code string from function result (useful for debugging) */
	const char* ZSTD_getErrorName(size_t code) { return ERR_getErrorName(code); }

	/*! ZSTD_getError() :
	* convert a `size_t` function result into a proper ZSTD_errorCode enum */
	ZSTD_ErrorCode ZSTD_getErrorCode(size_t code) { return ERR_getErrorCode(code); }

	/*! ZSTD_getErrorString() :
	* provides error code string from enum */
	const char* ZSTD_getErrorString(ZSTD_ErrorCode code) { return ERR_getErrorString(code); }



	/=*************************************************************
	* Custom allocator
	****************************************************************/
	void* ZSTD_malloc(size_t size, ZSTD_customMem customMem)
	{
	if (customMem.customAlloc)
	return customMem.customAlloc(customMem.opaque, size);
	return malloc(size);
	}

	void* ZSTD_calloc(size_t size, ZSTD_customMem customMem)
	{
	if (customMem.customAlloc) {
	/* calloc implemented as malloc+memset;
	* not as efficient as calloc, but next best guess for custom malloc */
	void* const ptr = customMem.customAlloc(customMem.opaque, size);
	memset(ptr, 0, size);
	return ptr;
	}
	return calloc(1, size);
	}

	void ZSTD_free(void* ptr, ZSTD_customMem customMem)
	{
	if (ptr!=NULL) {
	if (customMem.customFree)
	customMem.customFree(customMem.opaque, ptr);
	else
	free(ptr);
	}
	}
	/** ended inlining common/zstd_common.c **/

	/** start inlining compress/fse_compress.c **/
	/* ******************************************************************
	* FSE : Finite State Entropy encoder
	* Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
	* - Public forum : https://groups.google.com/forum/#!forum/lz4c
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */

	/* **************************************************************
	* Includes
	****************************************************************/
	#include <stdlib.h> /* malloc, free, qsort */
	#include <string.h> /* memcpy, memset */
	/** skipping file: ../common/compiler.h **/
	/** skipping file: ../common/mem.h **/
	/** skipping file: ../common/debug.h **/
	/** start inlining hist.h **/
	/* ******************************************************************
	* hist : Histogram functions
	* part of Finite State Entropy project
	* Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
	* - Public forum : https://groups.google.com/forum/#!forum/lz4c
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */

	/* --- dependencies --- */
	#include <stddef.h> /* size_t */


	/* --- simple histogram functions --- */

	/*! HIST_count():
	* Provides the precise count of each byte within a table 'count'.
	* 'count' is a table of unsigned int, of minimum size (*maxSymbolValuePtr+1).
	* Updates *maxSymbolValuePtr with actual largest symbol value detected.
	* @return : count of the most frequent symbol (which isn't identified).
	* or an error code, which can be tested using HIST_isError().
	* note : if return == srcSize, there is only one symbol.
	*/
	size_t HIST_count(unsigned* count, unsigned* maxSymbolValuePtr,
	const void* src, size_t srcSize);

	unsigned HIST_isError(size_t code); /*< tells if a return value is an error code /


	/* --- advanced histogram functions --- */

	#define HIST_WKSP_SIZE_U32 1024
	#define HIST_WKSP_SIZE (HIST_WKSP_SIZE_U32 * sizeof(unsigned))
	/** HIST_count_wksp() :
	* Same as HIST_count(), but using an externally provided scratch buffer.
	* Benefit is this function will use very little stack space.
	* `workSpace` is a writable buffer which must be 4-bytes aligned,
	* `workSpaceSize` must be >= HIST_WKSP_SIZE
	*/
	size_t HIST_count_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
	const void* src, size_t srcSize,
	void* workSpace, size_t workSpaceSize);

	/** HIST_countFast() :
	* same as HIST_count(), but blindly trusts that all byte values within src are <= *maxSymbolValuePtr.
	* This function is unsafe, and will segfault if any value within `src` is `> *maxSymbolValuePtr`
	*/
	size_t HIST_countFast(unsigned* count, unsigned* maxSymbolValuePtr,
	const void* src, size_t srcSize);

	/** HIST_countFast_wksp() :
	* Same as HIST_countFast(), but using an externally provided scratch buffer.
	* `workSpace` is a writable buffer which must be 4-bytes aligned,
	* `workSpaceSize` must be >= HIST_WKSP_SIZE
	*/
	size_t HIST_countFast_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
	const void* src, size_t srcSize,
	void* workSpace, size_t workSpaceSize);

	/*! HIST_count_simple() :
	* Same as HIST_countFast(), this function is unsafe,
	* and will segfault if any value within `src` is `> *maxSymbolValuePtr`.
	* It is also a bit slower for large inputs.
	* However, it does not need any additional memory (not even on stack).
	* @return : count of the most frequent symbol.
	* Note this function doesn't produce any error (i.e. it must succeed).
	*/
	unsigned HIST_count_simple(unsigned* count, unsigned* maxSymbolValuePtr,
	const void* src, size_t srcSize);
	/** ended inlining hist.h **/
	/** skipping file: ../common/bitstream.h **/
	#define FSE_STATIC_LINKING_ONLY
	/** skipping file: ../common/fse.h **/
	/** skipping file: ../common/error_private.h **/


	/* **************************************************************
	* Error Management
	****************************************************************/
	#define FSE_isError ERR_isError


	/* **************************************************************
	* Templates
	****************************************************************/
	/*
	designed to be included
	for type-specific functions (template emulation in C)
	Objective is to write these functions only once, for improved maintenance
	*/

	/* safety checks */
	#ifndef FSE_FUNCTION_EXTENSION
	# error "FSE_FUNCTION_EXTENSION must be defined"
	#endif
	#ifndef FSE_FUNCTION_TYPE
	# error "FSE_FUNCTION_TYPE must be defined"
	#endif

	/* Function names */
	#define FSE_CAT(X,Y) X##Y
	#define FSE_FUNCTION_NAME(X,Y) FSE_CAT(X,Y)
	#define FSE_TYPE_NAME(X,Y) FSE_CAT(X,Y)


	/* Function templates */

	/* FSE_buildCTable_wksp() :
	* Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
	* wkspSize should be sized to handle worst case situation, which is `1<<max_tableLog * sizeof(FSE_FUNCTION_TYPE)`
	* workSpace must also be properly aligned with FSE_FUNCTION_TYPE requirements
	*/
	size_t FSE_buildCTable_wksp(FSE_CTable* ct,
	const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog,
	void* workSpace, size_t wkspSize)
	{
	U32 const tableSize = 1 << tableLog;
	U32 const tableMask = tableSize - 1;
	void* const ptr = ct;
	U16* const tableU16 = ( (U16*) ptr) + 2;
	void* const FSCT = ((U32)ptr) + 1 / header */ + (tableLog ? tableSize>>1 : 1) ;
	FSE_symbolCompressionTransform* const symbolTT = (FSE_symbolCompressionTransform*) (FSCT);
	U32 const step = FSE_TABLESTEP(tableSize);
	U32 cumul[FSE_MAX_SYMBOL_VALUE+2];

	FSE_FUNCTION_TYPE* const tableSymbol = (FSE_FUNCTION_TYPE*)workSpace;
	U32 highThreshold = tableSize-1;

	/* CTable header */
	if (((size_t)1 << tableLog) * sizeof(FSE_FUNCTION_TYPE) > wkspSize) return ERROR(tableLog_tooLarge);
	tableU16[-2] = (U16) tableLog;
	tableU16[-1] = (U16) maxSymbolValue;
	assert(tableLog < 16); /* required for threshold strategy to work */

	/* For explanations on how to distribute symbol values over the table :
	* http://fastcompression.blogspot.fr/2014/02/fse-distributing-symbol-values.html */

	#ifdef __clang_analyzer__
	memset(tableSymbol, 0, sizeof(tableSymbol) tableSize); /* useless initialization, just to keep scan-build happy */
	#endif

	/* symbol start positions */
	{ U32 u;
	cumul[0] = 0;
	for (u=1; u <= maxSymbolValue+1; u++) {
	if (normalizedCounter[u-1]==-1) { /* Low proba symbol */
	cumul[u] = cumul[u-1] + 1;
	tableSymbol[highThreshold--] = (FSE_FUNCTION_TYPE)(u-1);
	} else {
	cumul[u] = cumul[u-1] + normalizedCounter[u-1];
	} }
	cumul[maxSymbolValue+1] = tableSize+1;
	}

	/* Spread symbols */
	{ U32 position = 0;
	U32 symbol;
	for (symbol=0; symbol<=maxSymbolValue; symbol++) {
	int nbOccurrences;
	int const freq = normalizedCounter[symbol];
	for (nbOccurrences=0; nbOccurrences<freq; nbOccurrences++) {
	tableSymbol[position] = (FSE_FUNCTION_TYPE)symbol;
	position = (position + step) & tableMask;
	while (position > highThreshold)
	position = (position + step) & tableMask; /* Low proba area */
	} }

	assert(position==0); /* Must have initialized all positions */
	}

	/* Build table */
	{ U32 u; for (u=0; u<tableSize; u++) {
	FSE_FUNCTION_TYPE s = tableSymbol[u]; /* note : static analyzer may not understand tableSymbol is properly initialized */
	tableU16[cumul[s]++] = (U16) (tableSize+u); /* TableU16 : sorted by symbol order; gives next state value */
	} }

	/* Build Symbol Transformation Table */
	{ unsigned total = 0;
	unsigned s;
	for (s=0; s<=maxSymbolValue; s++) {
	switch (normalizedCounter[s])
	{
	case 0:
	/* filling nonetheless, for compatibility with FSE_getMaxNbBits() */
	symbolTT[s].deltaNbBits = ((tableLog+1) << 16) - (1<<tableLog);
	break;

	case -1:
	case 1:
	symbolTT[s].deltaNbBits = (tableLog << 16) - (1<<tableLog);
	symbolTT[s].deltaFindState = total - 1;
	total ++;
	break;
	default :
	{
	U32 const maxBitsOut = tableLog - BIT_highbit32 (normalizedCounter[s]-1);
	U32 const minStatePlus = normalizedCounter[s] << maxBitsOut;
	symbolTT[s].deltaNbBits = (maxBitsOut << 16) - minStatePlus;
	symbolTT[s].deltaFindState = total - normalizedCounter[s];
	total += normalizedCounter[s];
	} } } }

	#if 0 /* debug : symbol costs */
	DEBUGLOG(5, "\n --- table statistics : ");
	{ U32 symbol;
	for (symbol=0; symbol<=maxSymbolValue; symbol++) {
	DEBUGLOG(5, "%3u: w=%3i, maxBits=%u, fracBits=%.2f",
	symbol, normalizedCounter[symbol],
	FSE_getMaxNbBits(symbolTT, symbol),
	(double)FSE_bitCost(symbolTT, tableLog, symbol, 8) / 256);
	}
	}
	#endif

	return 0;
	}


	size_t FSE_buildCTable(FSE_CTable* ct, const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog)
	{
	FSE_FUNCTION_TYPE tableSymbol[FSE_MAX_TABLESIZE]; /* memset() is not necessary, even if static analyzer complain about it */
	return FSE_buildCTable_wksp(ct, normalizedCounter, maxSymbolValue, tableLog, tableSymbol, sizeof(tableSymbol));
	}



	#ifndef FSE_COMMONDEFS_ONLY


	/-*************************************************************
	* FSE NCount encoding
	****************************************************************/
	size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog)
	{
	size_t const maxHeaderSize = (((maxSymbolValue+1) * tableLog) >> 3) + 3;
	return maxSymbolValue ? maxHeaderSize : FSE_NCOUNTBOUND; /* maxSymbolValue==0 ? use default */
	}

	static size_t
	FSE_writeNCount_generic (void* header, size_t headerBufferSize,
	const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog,
	unsigned writeIsSafe)
	{
	BYTE* const ostart = (BYTE*) header;
	BYTE* out = ostart;
	BYTE* const oend = ostart + headerBufferSize;
	int nbBits;
	const int tableSize = 1 << tableLog;
	int remaining;
	int threshold;
	U32 bitStream = 0;
	int bitCount = 0;
	unsigned symbol = 0;
	unsigned const alphabetSize = maxSymbolValue + 1;
	int previousIs0 = 0;

	/* Table Size */
	bitStream += (tableLog-FSE_MIN_TABLELOG) << bitCount;
	bitCount += 4;

	/* Init */
	remaining = tableSize+1; /* +1 for extra accuracy */
	threshold = tableSize;
	nbBits = tableLog+1;

	while ((symbol < alphabetSize) && (remaining>1)) { /* stops at 1 */
	if (previousIs0) {
	unsigned start = symbol;
	while ((symbol < alphabetSize) && !normalizedCounter[symbol]) symbol++;
	if (symbol == alphabetSize) break; /* incorrect distribution */
	while (symbol >= start+24) {
	start+=24;
	bitStream += 0xFFFFU << bitCount;
	if ((!writeIsSafe) && (out > oend-2))
	return ERROR(dstSize_tooSmall); /* Buffer overflow */
	out[0] = (BYTE) bitStream;
	out[1] = (BYTE)(bitStream>>8);
	out+=2;
	bitStream>>=16;
	}
	while (symbol >= start+3) {
	start+=3;
	bitStream += 3 << bitCount;
	bitCount += 2;
	}
	bitStream += (symbol-start) << bitCount;
	bitCount += 2;
	if (bitCount>16) {
	if ((!writeIsSafe) && (out > oend - 2))
	return ERROR(dstSize_tooSmall); /* Buffer overflow */
	out[0] = (BYTE)bitStream;
	out[1] = (BYTE)(bitStream>>8);
	out += 2;
	bitStream >>= 16;
	bitCount -= 16;
	} }
	{ int count = normalizedCounter[symbol++];
	int const max = (2*threshold-1) - remaining;
	remaining -= count < 0 ? -count : count;
	count++; /* +1 for extra accuracy */
	if (count>=threshold)
	count += max; /* [0..max[ [max..threshold[ (...) [threshold+max 2threshold[ /
	bitStream += count << bitCount;
	bitCount += nbBits;
	bitCount -= (count<max);
	previousIs0 = (count==1);
	if (remaining<1) return ERROR(GENERIC);
	while (remaining<threshold) { nbBits--; threshold>>=1; }
	}
	if (bitCount>16) {
	if ((!writeIsSafe) && (out > oend - 2))
	return ERROR(dstSize_tooSmall); /* Buffer overflow */
	out[0] = (BYTE)bitStream;
	out[1] = (BYTE)(bitStream>>8);
	out += 2;
	bitStream >>= 16;
	bitCount -= 16;
	} }

	if (remaining != 1)
	return ERROR(GENERIC); /* incorrect normalized distribution */
	assert(symbol <= alphabetSize);

	/* flush remaining bitStream */
	if ((!writeIsSafe) && (out > oend - 2))
	return ERROR(dstSize_tooSmall); /* Buffer overflow */
	out[0] = (BYTE)bitStream;
	out[1] = (BYTE)(bitStream>>8);
	out+= (bitCount+7) /8;

	return (out-ostart);
	}


	size_t FSE_writeNCount (void* buffer, size_t bufferSize,
	const short* normalizedCounter, unsigned maxSymbolValue, unsigned tableLog)
	{
	if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge); /* Unsupported */
	if (tableLog < FSE_MIN_TABLELOG) return ERROR(GENERIC); /* Unsupported */

	if (bufferSize < FSE_NCountWriteBound(maxSymbolValue, tableLog))
	return FSE_writeNCount_generic(buffer, bufferSize, normalizedCounter, maxSymbolValue, tableLog, 0);

	return FSE_writeNCount_generic(buffer, bufferSize, normalizedCounter, maxSymbolValue, tableLog, 1 /* write in buffer is safe */);
	}


	/-*************************************************************
	* FSE Compression Code
	****************************************************************/

	FSE_CTable* FSE_createCTable (unsigned maxSymbolValue, unsigned tableLog)
	{
	size_t size;
	if (tableLog > FSE_TABLELOG_ABSOLUTE_MAX) tableLog = FSE_TABLELOG_ABSOLUTE_MAX;
	size = FSE_CTABLE_SIZE_U32 (tableLog, maxSymbolValue) * sizeof(U32);
	return (FSE_CTable*)malloc(size);
	}

	void FSE_freeCTable (FSE_CTable* ct) { free(ct); }

	/* provides the minimum logSize to safely represent a distribution */
	static unsigned FSE_minTableLog(size_t srcSize, unsigned maxSymbolValue)
	{
	U32 minBitsSrc = BIT_highbit32((U32)(srcSize)) + 1;
	U32 minBitsSymbols = BIT_highbit32(maxSymbolValue) + 2;
	U32 minBits = minBitsSrc < minBitsSymbols ? minBitsSrc : minBitsSymbols;
	assert(srcSize > 1); /* Not supported, RLE should be used instead */
	return minBits;
	}

	unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus)
	{
	U32 maxBitsSrc = BIT_highbit32((U32)(srcSize - 1)) - minus;
	U32 tableLog = maxTableLog;
	U32 minBits = FSE_minTableLog(srcSize, maxSymbolValue);
	assert(srcSize > 1); /* Not supported, RLE should be used instead */
	if (tableLog==0) tableLog = FSE_DEFAULT_TABLELOG;
	if (maxBitsSrc < tableLog) tableLog = maxBitsSrc; /* Accuracy can be reduced */
	if (minBits > tableLog) tableLog = minBits; /* Need a minimum to safely represent all symbol values */
	if (tableLog < FSE_MIN_TABLELOG) tableLog = FSE_MIN_TABLELOG;
	if (tableLog > FSE_MAX_TABLELOG) tableLog = FSE_MAX_TABLELOG;
	return tableLog;
	}

	unsigned FSE_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue)
	{
	return FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 2);
	}


	/* Secondary normalization method.
	To be used when primary method fails. */

	static size_t FSE_normalizeM2(short* norm, U32 tableLog, const unsigned* count, size_t total, U32 maxSymbolValue)
	{
	short const NOT_YET_ASSIGNED = -2;
	U32 s;
	U32 distributed = 0;
	U32 ToDistribute;

	/* Init */
	U32 const lowThreshold = (U32)(total >> tableLog);
	U32 lowOne = (U32)((total * 3) >> (tableLog + 1));

	for (s=0; s<=maxSymbolValue; s++) {
	if (count[s] == 0) {
	norm[s]=0;
	continue;
	}
	if (count[s] <= lowThreshold) {
	norm[s] = -1;
	distributed++;
	total -= count[s];
	continue;
	}
	if (count[s] <= lowOne) {
	norm[s] = 1;
	distributed++;
	total -= count[s];
	continue;
	}

	norm[s]=NOT_YET_ASSIGNED;
	}
	ToDistribute = (1 << tableLog) - distributed;

	if (ToDistribute == 0)
	return 0;

	if ((total / ToDistribute) > lowOne) {
	/* risk of rounding to zero */
	lowOne = (U32)((total * 3) / (ToDistribute * 2));
	for (s=0; s<=maxSymbolValue; s++) {
	if ((norm[s] == NOT_YET_ASSIGNED) && (count[s] <= lowOne)) {
	norm[s] = 1;
	distributed++;
	total -= count[s];
	continue;
	} }
	ToDistribute = (1 << tableLog) - distributed;
	}

	if (distributed == maxSymbolValue+1) {
	/* all values are pretty poor;
	probably incompressible data (should have already been detected);
	find max, then give all remaining points to max */
	U32 maxV = 0, maxC = 0;
	for (s=0; s<=maxSymbolValue; s++)
	if (count[s] > maxC) { maxV=s; maxC=count[s]; }
	norm[maxV] += (short)ToDistribute;
	return 0;
	}

	if (total == 0) {
	/* all of the symbols were low enough for the lowOne or lowThreshold */
	for (s=0; ToDistribute > 0; s = (s+1)%(maxSymbolValue+1))
	if (norm[s] > 0) { ToDistribute--; norm[s]++; }
	return 0;
	}

	{ U64 const vStepLog = 62 - tableLog;
	U64 const mid = (1ULL << (vStepLog-1)) - 1;
	U64 const rStep = ((((U64)1<<vStepLog) * ToDistribute) + mid) / total; /* scale on remaining */
	U64 tmpTotal = mid;
	for (s=0; s<=maxSymbolValue; s++) {
	if (norm[s]==NOT_YET_ASSIGNED) {
	U64 const end = tmpTotal + (count[s] * rStep);
	U32 const sStart = (U32)(tmpTotal >> vStepLog);
	U32 const sEnd = (U32)(end >> vStepLog);
	U32 const weight = sEnd - sStart;
	if (weight < 1)
	return ERROR(GENERIC);
	norm[s] = (short)weight;
	tmpTotal = end;
	} } }

	return 0;
	}


	size_t FSE_normalizeCount (short* normalizedCounter, unsigned tableLog,
	const unsigned* count, size_t total,
	unsigned maxSymbolValue)
	{
	/* Sanity checks */
	if (tableLog==0) tableLog = FSE_DEFAULT_TABLELOG;
	if (tableLog < FSE_MIN_TABLELOG) return ERROR(GENERIC); /* Unsupported size */
	if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge); /* Unsupported size */
	if (tableLog < FSE_minTableLog(total, maxSymbolValue)) return ERROR(GENERIC); /* Too small tableLog, compression potentially impossible */

	{ static U32 const rtbTable[] = { 0, 473195, 504333, 520860, 550000, 700000, 750000, 830000 };
	U64 const scale = 62 - tableLog;
	U64 const step = ((U64)1<<62) / total; /* <== here, one division ! */
	U64 const vStep = 1ULL<<(scale-20);
	int stillToDistribute = 1<<tableLog;
	unsigned s;
	unsigned largest=0;
	short largestP=0;
	U32 lowThreshold = (U32)(total >> tableLog);

	for (s=0; s<=maxSymbolValue; s++) {
	if (count[s] == total) return 0; /* rle special case */
	if (count[s] == 0) { normalizedCounter[s]=0; continue; }
	if (count[s] <= lowThreshold) {
	normalizedCounter[s] = -1;
	stillToDistribute--;
	} else {
	short proba = (short)((count[s]*step) >> scale);
	if (proba<8) {
	U64 restToBeat = vStep * rtbTable[proba];
	proba += (count[s]*step) - ((U64)proba<<scale) > restToBeat;
	}
	if (proba > largestP) { largestP=proba; largest=s; }
	normalizedCounter[s] = proba;
	stillToDistribute -= proba;
	} }
	if (-stillToDistribute >= (normalizedCounter[largest] >> 1)) {
	/* corner case, need another normalization method */
	size_t const errorCode = FSE_normalizeM2(normalizedCounter, tableLog, count, total, maxSymbolValue);
	if (FSE_isError(errorCode)) return errorCode;
	}
	else normalizedCounter[largest] += (short)stillToDistribute;
	}

	#if 0
	{ /* Print Table (debug) */
	U32 s;
	U32 nTotal = 0;
	for (s=0; s<=maxSymbolValue; s++)
	RAWLOG(2, "%3i: %4i \n", s, normalizedCounter[s]);
	for (s=0; s<=maxSymbolValue; s++)
	nTotal += abs(normalizedCounter[s]);
	if (nTotal != (1U<<tableLog))
	RAWLOG(2, "Warning !!! Total == %u != %u !!!", nTotal, 1U<<tableLog);
	getchar();
	}
	#endif

	return tableLog;
	}


	/* fake FSE_CTable, for raw (uncompressed) input */
	size_t FSE_buildCTable_raw (FSE_CTable* ct, unsigned nbBits)
	{
	const unsigned tableSize = 1 << nbBits;
	const unsigned tableMask = tableSize - 1;
	const unsigned maxSymbolValue = tableMask;
	void* const ptr = ct;
	U16* const tableU16 = ( (U16*) ptr) + 2;
	void* const FSCT = ((U32)ptr) + 1 / header / + (tableSize>>1); / assumption : tableLog >= 1 */
	FSE_symbolCompressionTransform* const symbolTT = (FSE_symbolCompressionTransform*) (FSCT);
	unsigned s;

	/* Sanity checks */
	if (nbBits < 1) return ERROR(GENERIC); /* min size */

	/* header */
	tableU16[-2] = (U16) nbBits;
	tableU16[-1] = (U16) maxSymbolValue;

	/* Build table */
	for (s=0; s<tableSize; s++)
	tableU16[s] = (U16)(tableSize + s);

	/* Build Symbol Transformation Table */
	{ const U32 deltaNbBits = (nbBits << 16) - (1 << nbBits);
	for (s=0; s<=maxSymbolValue; s++) {
	symbolTT[s].deltaNbBits = deltaNbBits;
	symbolTT[s].deltaFindState = s-1;
	} }

	return 0;
	}

	/* fake FSE_CTable, for rle input (always same symbol) */
	size_t FSE_buildCTable_rle (FSE_CTable* ct, BYTE symbolValue)
	{
	void* ptr = ct;
	U16* tableU16 = ( (U16*) ptr) + 2;
	void* FSCTptr = (U32*)ptr + 2;
	FSE_symbolCompressionTransform* symbolTT = (FSE_symbolCompressionTransform*) FSCTptr;

	/* header */
	tableU16[-2] = (U16) 0;
	tableU16[-1] = (U16) symbolValue;

	/* Build table */
	tableU16[0] = 0;
	tableU16[1] = 0; /* just in case */

	/* Build Symbol Transformation Table */
	symbolTT[symbolValue].deltaNbBits = 0;
	symbolTT[symbolValue].deltaFindState = 0;

	return 0;
	}


	static size_t FSE_compress_usingCTable_generic (void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	const FSE_CTable* ct, const unsigned fast)
	{
	const BYTE* const istart = (const BYTE*) src;
	const BYTE* const iend = istart + srcSize;
	const BYTE* ip=iend;

	BIT_CStream_t bitC;
	FSE_CState_t CState1, CState2;

	/* init */
	if (srcSize <= 2) return 0;
	{ size_t const initError = BIT_initCStream(&bitC, dst, dstSize);
	if (FSE_isError(initError)) return 0; /* not enough space available to write a bitstream */ }

	#define FSE_FLUSHBITS(s) (fast ? BIT_flushBitsFast(s) : BIT_flushBits(s))

	if (srcSize & 1) {
	FSE_initCState2(&CState1, ct, *--ip);
	FSE_initCState2(&CState2, ct, *--ip);
	FSE_encodeSymbol(&bitC, &CState1, *--ip);
	FSE_FLUSHBITS(&bitC);
	} else {
	FSE_initCState2(&CState2, ct, *--ip);
	FSE_initCState2(&CState1, ct, *--ip);
	}

	/* join to mod 4 */
	srcSize -= 2;
	if ((sizeof(bitC.bitContainer)8 > FSE_MAX_TABLELOG4+7 ) && (srcSize & 2)) { /* test bit 2 */
	FSE_encodeSymbol(&bitC, &CState2, *--ip);
	FSE_encodeSymbol(&bitC, &CState1, *--ip);
	FSE_FLUSHBITS(&bitC);
	}

	/* 2 or 4 encoding per loop */
	while ( ip>istart ) {

	FSE_encodeSymbol(&bitC, &CState2, *--ip);

	if (sizeof(bitC.bitContainer)8 < FSE_MAX_TABLELOG2+7 ) /* this test must be static */
	FSE_FLUSHBITS(&bitC);

	FSE_encodeSymbol(&bitC, &CState1, *--ip);

	if (sizeof(bitC.bitContainer)8 > FSE_MAX_TABLELOG4+7 ) { /* this test must be static */
	FSE_encodeSymbol(&bitC, &CState2, *--ip);
	FSE_encodeSymbol(&bitC, &CState1, *--ip);
	}

	FSE_FLUSHBITS(&bitC);
	}

	FSE_flushCState(&bitC, &CState2);
	FSE_flushCState(&bitC, &CState1);
	return BIT_closeCStream(&bitC);
	}

	size_t FSE_compress_usingCTable (void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	const FSE_CTable* ct)
	{
	unsigned const fast = (dstSize >= FSE_BLOCKBOUND(srcSize));

	if (fast)
	return FSE_compress_usingCTable_generic(dst, dstSize, src, srcSize, ct, 1);
	else
	return FSE_compress_usingCTable_generic(dst, dstSize, src, srcSize, ct, 0);
	}


	size_t FSE_compressBound(size_t size) { return FSE_COMPRESSBOUND(size); }

	/* FSE_compress_wksp() :
	* Same as FSE_compress2(), but using an externally allocated scratch buffer (`workSpace`).
	* `wkspSize` size must be `(1<<tableLog)`.
	*/
	size_t FSE_compress_wksp (void* dst, size_t dstSize, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void* workSpace, size_t wkspSize)
	{
	BYTE* const ostart = (BYTE*) dst;
	BYTE* op = ostart;
	BYTE* const oend = ostart + dstSize;

	unsigned count[FSE_MAX_SYMBOL_VALUE+1];
	S16 norm[FSE_MAX_SYMBOL_VALUE+1];
	FSE_CTable* CTable = (FSE_CTable*)workSpace;
	size_t const CTableSize = FSE_CTABLE_SIZE_U32(tableLog, maxSymbolValue);
	void* scratchBuffer = (void*)(CTable + CTableSize);
	size_t const scratchBufferSize = wkspSize - (CTableSize * sizeof(FSE_CTable));

	/* init conditions */
	if (wkspSize < FSE_WKSP_SIZE_U32(tableLog, maxSymbolValue)) return ERROR(tableLog_tooLarge);
	if (srcSize <= 1) return 0; /* Not compressible */
	if (!maxSymbolValue) maxSymbolValue = FSE_MAX_SYMBOL_VALUE;
	if (!tableLog) tableLog = FSE_DEFAULT_TABLELOG;

	/* Scan input and build symbol stats */
	{ CHECK_V_F(maxCount, HIST_count_wksp(count, &maxSymbolValue, src, srcSize, scratchBuffer, scratchBufferSize) );
	if (maxCount == srcSize) return 1; /* only a single symbol in src : rle */
	if (maxCount == 1) return 0; /* each symbol present maximum once => not compressible */
	if (maxCount < (srcSize >> 7)) return 0; /* Heuristic : not compressible enough */
	}

	tableLog = FSE_optimalTableLog(tableLog, srcSize, maxSymbolValue);
	CHECK_F( FSE_normalizeCount(norm, tableLog, count, srcSize, maxSymbolValue) );

	/* Write table description header */
	{ CHECK_V_F(nc_err, FSE_writeNCount(op, oend-op, norm, maxSymbolValue, tableLog) );
	op += nc_err;
	}

	/* Compress */
	CHECK_F( FSE_buildCTable_wksp(CTable, norm, maxSymbolValue, tableLog, scratchBuffer, scratchBufferSize) );
	{ CHECK_V_F(cSize, FSE_compress_usingCTable(op, oend - op, src, srcSize, CTable) );
	if (cSize == 0) return 0; /* not enough space for compressed data */
	op += cSize;
	}

	/* check compressibility */
	if ( (size_t)(op-ostart) >= srcSize-1 ) return 0;

	return op-ostart;
	}

	typedef struct {
	FSE_CTable CTable_max[FSE_CTABLE_SIZE_U32(FSE_MAX_TABLELOG, FSE_MAX_SYMBOL_VALUE)];
	BYTE scratchBuffer[1 << FSE_MAX_TABLELOG];
	} fseWkspMax_t;

	size_t FSE_compress2 (void* dst, size_t dstCapacity, const void* src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog)
	{
	fseWkspMax_t scratchBuffer;
	DEBUG_STATIC_ASSERT(sizeof(scratchBuffer) >= FSE_WKSP_SIZE_U32(FSE_MAX_TABLELOG, FSE_MAX_SYMBOL_VALUE)); /* compilation failures here means scratchBuffer is not large enough */
	if (tableLog > FSE_MAX_TABLELOG) return ERROR(tableLog_tooLarge);
	return FSE_compress_wksp(dst, dstCapacity, src, srcSize, maxSymbolValue, tableLog, &scratchBuffer, sizeof(scratchBuffer));
	}

	size_t FSE_compress (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
	{
	return FSE_compress2(dst, dstCapacity, src, srcSize, FSE_MAX_SYMBOL_VALUE, FSE_DEFAULT_TABLELOG);
	}


	#endif /* FSE_COMMONDEFS_ONLY */
	/** ended inlining compress/fse_compress.c **/
	/** start inlining compress/hist.c **/
	/* ******************************************************************
	* hist : Histogram functions
	* part of Finite State Entropy project
	* Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - FSE source repository : https://github.com/Cyan4973/FiniteStateEntropy
	* - Public forum : https://groups.google.com/forum/#!forum/lz4c
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */

	/* --- dependencies --- */
	/** skipping file: ../common/mem.h **/
	/** skipping file: ../common/debug.h **/
	/** skipping file: ../common/error_private.h **/
	/** skipping file: hist.h **/


	/* --- Error management --- */
	unsigned HIST_isError(size_t code) { return ERR_isError(code); }

	/-*************************************************************
	* Histogram functions
	****************************************************************/
	unsigned HIST_count_simple(unsigned* count, unsigned* maxSymbolValuePtr,
	const void* src, size_t srcSize)
	{
	const BYTE* ip = (const BYTE*)src;
	const BYTE* const end = ip + srcSize;
	unsigned maxSymbolValue = *maxSymbolValuePtr;
	unsigned largestCount=0;

	memset(count, 0, (maxSymbolValue+1) * sizeof(*count));
	if (srcSize==0) { *maxSymbolValuePtr = 0; return 0; }

	while (ip<end) {
	assert(*ip <= maxSymbolValue);
	count[*ip++]++;
	}

	while (!count[maxSymbolValue]) maxSymbolValue--;
	*maxSymbolValuePtr = maxSymbolValue;

	{ U32 s;
	for (s=0; s<=maxSymbolValue; s++)
	if (count[s] > largestCount) largestCount = count[s];
	}

	return largestCount;
	}

	typedef enum { trustInput, checkMaxSymbolValue } HIST_checkInput_e;

	/* HIST_count_parallel_wksp() :
	* store histogram into 4 intermediate tables, recombined at the end.
	* this design makes better use of OoO cpus,
	* and is noticeably faster when some values are heavily repeated.
	* But it needs some additional workspace for intermediate tables.
	* `workSpace` size must be a table of size >= HIST_WKSP_SIZE_U32.
	* @return : largest histogram frequency,
	* or an error code (notably when histogram would be larger than maxSymbolValuePtr). /
	static size_t HIST_count_parallel_wksp(
	unsigned* count, unsigned* maxSymbolValuePtr,
	const void* source, size_t sourceSize,
	HIST_checkInput_e check,
	U32* const workSpace)
	{
	const BYTE* ip = (const BYTE*)source;
	const BYTE* const iend = ip+sourceSize;
	unsigned maxSymbolValue = *maxSymbolValuePtr;
	unsigned max=0;
	U32* const Counting1 = workSpace;
	U32* const Counting2 = Counting1 + 256;
	U32* const Counting3 = Counting2 + 256;
	U32* const Counting4 = Counting3 + 256;

	memset(workSpace, 0, 4256sizeof(unsigned));

	/* safety checks */
	if (!sourceSize) {
	memset(count, 0, maxSymbolValue + 1);
	*maxSymbolValuePtr = 0;
	return 0;
	}
	if (!maxSymbolValue) maxSymbolValue = 255; /* 0 == default */

	/* by stripes of 16 bytes */
	{ U32 cached = MEM_read32(ip); ip += 4;
	while (ip < iend-15) {
	U32 c = cached; cached = MEM_read32(ip); ip += 4;
	Counting1[(BYTE) c ]++;
	Counting2[(BYTE)(c>>8) ]++;
	Counting3[(BYTE)(c>>16)]++;
	Counting4[ c>>24 ]++;
	c = cached; cached = MEM_read32(ip); ip += 4;
	Counting1[(BYTE) c ]++;
	Counting2[(BYTE)(c>>8) ]++;
	Counting3[(BYTE)(c>>16)]++;
	Counting4[ c>>24 ]++;
	c = cached; cached = MEM_read32(ip); ip += 4;
	Counting1[(BYTE) c ]++;
	Counting2[(BYTE)(c>>8) ]++;
	Counting3[(BYTE)(c>>16)]++;
	Counting4[ c>>24 ]++;
	c = cached; cached = MEM_read32(ip); ip += 4;
	Counting1[(BYTE) c ]++;
	Counting2[(BYTE)(c>>8) ]++;
	Counting3[(BYTE)(c>>16)]++;
	Counting4[ c>>24 ]++;
	}
	ip-=4;
	}

	/* finish last symbols */
	while (ip<iend) Counting1[*ip++]++;

	if (check) { /* verify stats will fit into destination table */
	U32 s; for (s=255; s>maxSymbolValue; s--) {
	Counting1[s] += Counting2[s] + Counting3[s] + Counting4[s];
	if (Counting1[s]) return ERROR(maxSymbolValue_tooSmall);
	} }

	{ U32 s;
	if (maxSymbolValue > 255) maxSymbolValue = 255;
	for (s=0; s<=maxSymbolValue; s++) {
	count[s] = Counting1[s] + Counting2[s] + Counting3[s] + Counting4[s];
	if (count[s] > max) max = count[s];
	} }

	while (!count[maxSymbolValue]) maxSymbolValue--;
	*maxSymbolValuePtr = maxSymbolValue;
	return (size_t)max;
	}

	/* HIST_countFast_wksp() :
	* Same as HIST_countFast(), but using an externally provided scratch buffer.
	* `workSpace` is a writable buffer which must be 4-bytes aligned,
	* `workSpaceSize` must be >= HIST_WKSP_SIZE
	*/
	size_t HIST_countFast_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
	const void* source, size_t sourceSize,
	void* workSpace, size_t workSpaceSize)
	{
	if (sourceSize < 1500) /* heuristic threshold */
	return HIST_count_simple(count, maxSymbolValuePtr, source, sourceSize);
	if ((size_t)workSpace & 3) return ERROR(GENERIC); /* must be aligned on 4-bytes boundaries */
	if (workSpaceSize < HIST_WKSP_SIZE) return ERROR(workSpace_tooSmall);
	return HIST_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, trustInput, (U32*)workSpace);
	}

	/* fast variant (unsafe : won't check if src contains values beyond count[] limit) */
	size_t HIST_countFast(unsigned* count, unsigned* maxSymbolValuePtr,
	const void* source, size_t sourceSize)
	{
	unsigned tmpCounters[HIST_WKSP_SIZE_U32];
	return HIST_countFast_wksp(count, maxSymbolValuePtr, source, sourceSize, tmpCounters, sizeof(tmpCounters));
	}

	/* HIST_count_wksp() :
	* Same as HIST_count(), but using an externally provided scratch buffer.
	* `workSpace` size must be table of >= HIST_WKSP_SIZE_U32 unsigned */
	size_t HIST_count_wksp(unsigned* count, unsigned* maxSymbolValuePtr,
	const void* source, size_t sourceSize,
	void* workSpace, size_t workSpaceSize)
	{
	if ((size_t)workSpace & 3) return ERROR(GENERIC); /* must be aligned on 4-bytes boundaries */
	if (workSpaceSize < HIST_WKSP_SIZE) return ERROR(workSpace_tooSmall);
	if (*maxSymbolValuePtr < 255)
	return HIST_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, checkMaxSymbolValue, (U32*)workSpace);
	*maxSymbolValuePtr = 255;
	return HIST_countFast_wksp(count, maxSymbolValuePtr, source, sourceSize, workSpace, workSpaceSize);
	}

	size_t HIST_count(unsigned* count, unsigned* maxSymbolValuePtr,
	const void* src, size_t srcSize)
	{
	unsigned tmpCounters[HIST_WKSP_SIZE_U32];
	return HIST_count_wksp(count, maxSymbolValuePtr, src, srcSize, tmpCounters, sizeof(tmpCounters));
	}
	/** ended inlining compress/hist.c **/
	/** start inlining compress/huf_compress.c **/
	/* ******************************************************************
	* Huffman encoder, part of New Generation Entropy library
	* Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - FSE+HUF source repository : https://github.com/Cyan4973/FiniteStateEntropy
	* - Public forum : https://groups.google.com/forum/#!forum/lz4c
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */

	/* **************************************************************
	* Compiler specifics
	****************************************************************/
	#ifdef _MSC_VER /* Visual Studio */
	# pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */
	#endif


	/* **************************************************************
	* Includes
	****************************************************************/
	#include <string.h> /* memcpy, memset */
	#include <stdio.h> /* printf (debug) */
	/** skipping file: ../common/compiler.h **/
	/** skipping file: ../common/bitstream.h **/
	/** skipping file: hist.h **/
	#define FSE_STATIC_LINKING_ONLY /* FSE_optimalTableLog_internal */
	/** skipping file: ../common/fse.h **/
	#define HUF_STATIC_LINKING_ONLY
	/** skipping file: ../common/huf.h **/
	/** skipping file: ../common/error_private.h **/


	/* **************************************************************
	* Error Management
	****************************************************************/
	#define HUF_isError ERR_isError
	#define HUF_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c) /* use only after variable declarations */


	/* **************************************************************
	* Utils
	****************************************************************/
	unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue)
	{
	return FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 1);
	}


	/* *******************************************************
	* HUF : Huffman block compression
	*********************************************************/
	/* HUF_compressWeights() :
	* Same as FSE_compress(), but dedicated to huff0's weights compression.
	* The use case needs much less stack memory.
	* Note : all elements within weightTable are supposed to be <= HUF_TABLELOG_MAX.
	*/
	#define MAX_FSE_TABLELOG_FOR_HUFF_HEADER 6
	static size_t HUF_compressWeights (void* dst, size_t dstSize, const void* weightTable, size_t wtSize)
	{
	BYTE* const ostart = (BYTE*) dst;
	BYTE* op = ostart;
	BYTE* const oend = ostart + dstSize;

	unsigned maxSymbolValue = HUF_TABLELOG_MAX;
	U32 tableLog = MAX_FSE_TABLELOG_FOR_HUFF_HEADER;

	FSE_CTable CTable[FSE_CTABLE_SIZE_U32(MAX_FSE_TABLELOG_FOR_HUFF_HEADER, HUF_TABLELOG_MAX)];
	BYTE scratchBuffer[1<<MAX_FSE_TABLELOG_FOR_HUFF_HEADER];

	unsigned count[HUF_TABLELOG_MAX+1];
	S16 norm[HUF_TABLELOG_MAX+1];

	/* init conditions */
	if (wtSize <= 1) return 0; /* Not compressible */

	/* Scan input and build symbol stats */
	{ unsigned const maxCount = HIST_count_simple(count, &maxSymbolValue, weightTable, wtSize); /* never fails */
	if (maxCount == wtSize) return 1; /* only a single symbol in src : rle */
	if (maxCount == 1) return 0; /* each symbol present maximum once => not compressible */
	}

	tableLog = FSE_optimalTableLog(tableLog, wtSize, maxSymbolValue);
	CHECK_F( FSE_normalizeCount(norm, tableLog, count, wtSize, maxSymbolValue) );

	/* Write table description header */
	{ CHECK_V_F(hSize, FSE_writeNCount(op, (size_t)(oend-op), norm, maxSymbolValue, tableLog) );
	op += hSize;
	}

	/* Compress */
	CHECK_F( FSE_buildCTable_wksp(CTable, norm, maxSymbolValue, tableLog, scratchBuffer, sizeof(scratchBuffer)) );
	{ CHECK_V_F(cSize, FSE_compress_usingCTable(op, (size_t)(oend - op), weightTable, wtSize, CTable) );
	if (cSize == 0) return 0; /* not enough space for compressed data */
	op += cSize;
	}

	return (size_t)(op-ostart);
	}


	struct HUF_CElt_s {
	U16 val;
	BYTE nbBits;
	}; /* typedef'd to HUF_CElt within "huf.h" */

	/*! HUF_writeCTable() :
	`CTable` : Huffman tree to save, using huf representation.
	@return : size of saved CTable */
	size_t HUF_writeCTable (void* dst, size_t maxDstSize,
	const HUF_CElt* CTable, unsigned maxSymbolValue, unsigned huffLog)
	{
	BYTE bitsToWeight[HUF_TABLELOG_MAX + 1]; /* precomputed conversion table */
	BYTE huffWeight[HUF_SYMBOLVALUE_MAX];
	BYTE* op = (BYTE*)dst;
	U32 n;

	/* check conditions */
	if (maxSymbolValue > HUF_SYMBOLVALUE_MAX) return ERROR(maxSymbolValue_tooLarge);

	/* convert to weight */
	bitsToWeight[0] = 0;
	for (n=1; n<huffLog+1; n++)
	bitsToWeight[n] = (BYTE)(huffLog + 1 - n);
	for (n=0; n<maxSymbolValue; n++)
	huffWeight[n] = bitsToWeight[CTable[n].nbBits];

	/* attempt weights compression by FSE */
	{ CHECK_V_F(hSize, HUF_compressWeights(op+1, maxDstSize-1, huffWeight, maxSymbolValue) );
	if ((hSize>1) & (hSize < maxSymbolValue/2)) { /* FSE compressed */
	op[0] = (BYTE)hSize;
	return hSize+1;
	} }

	/* write raw values as 4-bits (max : 15) */
	if (maxSymbolValue > (256-128)) return ERROR(GENERIC); /* should not happen : likely means source cannot be compressed */
	if (((maxSymbolValue+1)/2) + 1 > maxDstSize) return ERROR(dstSize_tooSmall); /* not enough space within dst buffer */
	op[0] = (BYTE)(128 /special case/ + (maxSymbolValue-1));
	huffWeight[maxSymbolValue] = 0; /* to be sure it doesn't cause msan issue in final combination */
	for (n=0; n<maxSymbolValue; n+=2)
	op[(n/2)+1] = (BYTE)((huffWeight[n] << 4) + huffWeight[n+1]);
	return ((maxSymbolValue+1)/2) + 1;
	}


	size_t HUF_readCTable (HUF_CElt* CTable, unsigned* maxSymbolValuePtr, const void* src, size_t srcSize, unsigned* hasZeroWeights)
	{
	BYTE huffWeight[HUF_SYMBOLVALUE_MAX + 1]; /* init not required, even though some static analyzer may complain */
	U32 rankVal[HUF_TABLELOG_ABSOLUTEMAX + 1]; /* large enough for values from 0 to 16 */
	U32 tableLog = 0;
	U32 nbSymbols = 0;

	/* get symbol weights */
	CHECK_V_F(readSize, HUF_readStats(huffWeight, HUF_SYMBOLVALUE_MAX+1, rankVal, &nbSymbols, &tableLog, src, srcSize));

	/* check result */
	if (tableLog > HUF_TABLELOG_MAX) return ERROR(tableLog_tooLarge);
	if (nbSymbols > *maxSymbolValuePtr+1) return ERROR(maxSymbolValue_tooSmall);

	/* Prepare base value per rank */
	{ U32 n, nextRankStart = 0;
	for (n=1; n<=tableLog; n++) {
	U32 current = nextRankStart;
	nextRankStart += (rankVal[n] << (n-1));
	rankVal[n] = current;
	} }

	/* fill nbBits */
	*hasZeroWeights = 0;
	{ U32 n; for (n=0; n<nbSymbols; n++) {
	const U32 w = huffWeight[n];
	*hasZeroWeights \|= (w == 0);
	CTable[n].nbBits = (BYTE)(tableLog + 1 - w) & -(w != 0);
	} }

	/* fill val */
	{ U16 nbPerRank[HUF_TABLELOG_MAX+2] = {0}; /* support w=0=>n=tableLog+1 */
	U16 valPerRank[HUF_TABLELOG_MAX+2] = {0};
	{ U32 n; for (n=0; n<nbSymbols; n++) nbPerRank[CTable[n].nbBits]++; }
	/* determine stating value per rank */
	valPerRank[tableLog+1] = 0; /* for w==0 */
	{ U16 min = 0;
	U32 n; for (n=tableLog; n>0; n--) { /* start at n=tablelog <-> w=1 */
	valPerRank[n] = min; /* get starting value within each rank */
	min += nbPerRank[n];
	min >>= 1;
	} }
	/* assign value within rank, symbol order */
	{ U32 n; for (n=0; n<nbSymbols; n++) CTable[n].val = valPerRank[CTable[n].nbBits]++; }
	}

	*maxSymbolValuePtr = nbSymbols - 1;
	return readSize;
	}

	U32 HUF_getNbBits(const void* symbolTable, U32 symbolValue)
	{
	const HUF_CElt* table = (const HUF_CElt*)symbolTable;
	assert(symbolValue <= HUF_SYMBOLVALUE_MAX);
	return table[symbolValue].nbBits;
	}


	typedef struct nodeElt_s {
	U32 count;
	U16 parent;
	BYTE byte;
	BYTE nbBits;
	} nodeElt;

	static U32 HUF_setMaxHeight(nodeElt* huffNode, U32 lastNonNull, U32 maxNbBits)
	{
	const U32 largestBits = huffNode[lastNonNull].nbBits;
	if (largestBits <= maxNbBits) return largestBits; /* early exit : no elt > maxNbBits */

	/* there are several too large elements (at least >= 2) */
	{ int totalCost = 0;
	const U32 baseCost = 1 << (largestBits - maxNbBits);
	int n = (int)lastNonNull;

	while (huffNode[n].nbBits > maxNbBits) {
	totalCost += baseCost - (1 << (largestBits - huffNode[n].nbBits));
	huffNode[n].nbBits = (BYTE)maxNbBits;
	n --;
	} /* n stops at huffNode[n].nbBits <= maxNbBits */
	while (huffNode[n].nbBits == maxNbBits) n--; /* n end at index of smallest symbol using < maxNbBits */

	/* renorm totalCost */
	totalCost >>= (largestBits - maxNbBits); /* note : totalCost is necessarily a multiple of baseCost */

	/* repay normalized cost */
	{ U32 const noSymbol = 0xF0F0F0F0;
	U32 rankLast[HUF_TABLELOG_MAX+2];

	/* Get pos of last (smallest) symbol per rank */
	memset(rankLast, 0xF0, sizeof(rankLast));
	{ U32 currentNbBits = maxNbBits;
	int pos;
	for (pos=n ; pos >= 0; pos--) {
	if (huffNode[pos].nbBits >= currentNbBits) continue;
	currentNbBits = huffNode[pos].nbBits; /* < maxNbBits */
	rankLast[maxNbBits-currentNbBits] = (U32)pos;
	} }

	while (totalCost > 0) {
	U32 nBitsToDecrease = BIT_highbit32((U32)totalCost) + 1;
	for ( ; nBitsToDecrease > 1; nBitsToDecrease--) {
	U32 const highPos = rankLast[nBitsToDecrease];
	U32 const lowPos = rankLast[nBitsToDecrease-1];
	if (highPos == noSymbol) continue;
	if (lowPos == noSymbol) break;
	{ U32 const highTotal = huffNode[highPos].count;
	U32 const lowTotal = 2 * huffNode[lowPos].count;
	if (highTotal <= lowTotal) break;
	} }
	/* only triggered when no more rank 1 symbol left => find closest one (note : there is necessarily at least one !) */
	/* HUF_MAX_TABLELOG test just to please gcc 5+; but it should not be necessary */
	while ((nBitsToDecrease<=HUF_TABLELOG_MAX) && (rankLast[nBitsToDecrease] == noSymbol))
	nBitsToDecrease ++;
	totalCost -= 1 << (nBitsToDecrease-1);
	if (rankLast[nBitsToDecrease-1] == noSymbol)
	rankLast[nBitsToDecrease-1] = rankLast[nBitsToDecrease]; /* this rank is no longer empty */
	huffNode[rankLast[nBitsToDecrease]].nbBits ++;
	if (rankLast[nBitsToDecrease] == 0) /* special case, reached largest symbol */
	rankLast[nBitsToDecrease] = noSymbol;
	else {
	rankLast[nBitsToDecrease]--;
	if (huffNode[rankLast[nBitsToDecrease]].nbBits != maxNbBits-nBitsToDecrease)
	rankLast[nBitsToDecrease] = noSymbol; /* this rank is now empty */
	} } /* while (totalCost > 0) */

	while (totalCost < 0) { /* Sometimes, cost correction overshoot */
	if (rankLast[1] == noSymbol) { /* special case : no rank 1 symbol (using maxNbBits-1); let's create one from largest rank 0 (using maxNbBits) */
	while (huffNode[n].nbBits == maxNbBits) n--;
	huffNode[n+1].nbBits--;
	assert(n >= 0);
	rankLast[1] = (U32)(n+1);
	totalCost++;
	continue;
	}
	huffNode[ rankLast[1] + 1 ].nbBits--;
	rankLast[1]++;
	totalCost ++;
	} } } /* there are several too large elements (at least >= 2) */

	return maxNbBits;
	}

	typedef struct {
	U32 base;
	U32 current;
	} rankPos;

	typedef nodeElt huffNodeTable[HUF_CTABLE_WORKSPACE_SIZE_U32];

	#define RANK_POSITION_TABLE_SIZE 32

	typedef struct {
	huffNodeTable huffNodeTbl;
	rankPos rankPosition[RANK_POSITION_TABLE_SIZE];
	} HUF_buildCTable_wksp_tables;

	static void HUF_sort(nodeElt* huffNode, const unsigned* count, U32 maxSymbolValue, rankPos* rankPosition)
	{
	U32 n;

	memset(rankPosition, 0, sizeof(rankPosition) RANK_POSITION_TABLE_SIZE);
	for (n=0; n<=maxSymbolValue; n++) {
	U32 r = BIT_highbit32(count[n] + 1);
	rankPosition[r].base ++;
	}
	for (n=30; n>0; n--) rankPosition[n-1].base += rankPosition[n].base;
	for (n=0; n<32; n++) rankPosition[n].current = rankPosition[n].base;
	for (n=0; n<=maxSymbolValue; n++) {
	U32 const c = count[n];
	U32 const r = BIT_highbit32(c+1) + 1;
	U32 pos = rankPosition[r].current++;
	while ((pos > rankPosition[r].base) && (c > huffNode[pos-1].count)) {
	huffNode[pos] = huffNode[pos-1];
	pos--;
	}
	huffNode[pos].count = c;
	huffNode[pos].byte = (BYTE)n;
	}
	}


	/** HUF_buildCTable_wksp() :
	* Same as HUF_buildCTable(), but using externally allocated scratch buffer.
	* `workSpace` must be aligned on 4-bytes boundaries, and be at least as large as sizeof(HUF_buildCTable_wksp_tables).
	*/
	#define STARTNODE (HUF_SYMBOLVALUE_MAX+1)

	size_t HUF_buildCTable_wksp (HUF_CElt* tree, const unsigned* count, U32 maxSymbolValue, U32 maxNbBits, void* workSpace, size_t wkspSize)
	{
	HUF_buildCTable_wksp_tables* const wksp_tables = (HUF_buildCTable_wksp_tables*)workSpace;
	nodeElt* const huffNode0 = wksp_tables->huffNodeTbl;
	nodeElt* const huffNode = huffNode0+1;
	int nonNullRank;
	int lowS, lowN;
	int nodeNb = STARTNODE;
	int n, nodeRoot;

	/* safety checks */
	if (((size_t)workSpace & 3) != 0) return ERROR(GENERIC); /* must be aligned on 4-bytes boundaries */
	if (wkspSize < sizeof(HUF_buildCTable_wksp_tables))
	return ERROR(workSpace_tooSmall);
	if (maxNbBits == 0) maxNbBits = HUF_TABLELOG_DEFAULT;
	if (maxSymbolValue > HUF_SYMBOLVALUE_MAX)
	return ERROR(maxSymbolValue_tooLarge);
	memset(huffNode0, 0, sizeof(huffNodeTable));

	/* sort, decreasing order */
	HUF_sort(huffNode, count, maxSymbolValue, wksp_tables->rankPosition);

	/* init for parents */
	nonNullRank = (int)maxSymbolValue;
	while(huffNode[nonNullRank].count == 0) nonNullRank--;
	lowS = nonNullRank; nodeRoot = nodeNb + lowS - 1; lowN = nodeNb;
	huffNode[nodeNb].count = huffNode[lowS].count + huffNode[lowS-1].count;
	huffNode[lowS].parent = huffNode[lowS-1].parent = (U16)nodeNb;
	nodeNb++; lowS-=2;
	for (n=nodeNb; n<=nodeRoot; n++) huffNode[n].count = (U32)(1U<<30);
	huffNode0[0].count = (U32)(1U<<31); /* fake entry, strong barrier */

	/* create parents */
	while (nodeNb <= nodeRoot) {
	int const n1 = (huffNode[lowS].count < huffNode[lowN].count) ? lowS-- : lowN++;
	int const n2 = (huffNode[lowS].count < huffNode[lowN].count) ? lowS-- : lowN++;
	huffNode[nodeNb].count = huffNode[n1].count + huffNode[n2].count;
	huffNode[n1].parent = huffNode[n2].parent = (U16)nodeNb;
	nodeNb++;
	}

	/* distribute weights (unlimited tree height) */
	huffNode[nodeRoot].nbBits = 0;
	for (n=nodeRoot-1; n>=STARTNODE; n--)
	huffNode[n].nbBits = huffNode[ huffNode[n].parent ].nbBits + 1;
	for (n=0; n<=nonNullRank; n++)
	huffNode[n].nbBits = huffNode[ huffNode[n].parent ].nbBits + 1;

	/* enforce maxTableLog */
	maxNbBits = HUF_setMaxHeight(huffNode, (U32)nonNullRank, maxNbBits);

	/* fill result into tree (val, nbBits) */
	{ U16 nbPerRank[HUF_TABLELOG_MAX+1] = {0};
	U16 valPerRank[HUF_TABLELOG_MAX+1] = {0};
	int const alphabetSize = (int)(maxSymbolValue + 1);
	if (maxNbBits > HUF_TABLELOG_MAX) return ERROR(GENERIC); /* check fit into table */
	for (n=0; n<=nonNullRank; n++)
	nbPerRank[huffNode[n].nbBits]++;
	/* determine stating value per rank */
	{ U16 min = 0;
	for (n=(int)maxNbBits; n>0; n--) {
	valPerRank[n] = min; /* get starting value within each rank */
	min += nbPerRank[n];
	min >>= 1;
	} }
	for (n=0; n<alphabetSize; n++)
	tree[huffNode[n].byte].nbBits = huffNode[n].nbBits; /* push nbBits per symbol, symbol order */
	for (n=0; n<alphabetSize; n++)
	tree[n].val = valPerRank[tree[n].nbBits]++; /* assign value within rank, symbol order */
	}

	return maxNbBits;
	}

	/** HUF_buildCTable() :
	* @return : maxNbBits
	* Note : count is used before tree is written, so they can safely overlap
	*/
	size_t HUF_buildCTable (HUF_CElt* tree, const unsigned* count, unsigned maxSymbolValue, unsigned maxNbBits)
	{
	HUF_buildCTable_wksp_tables workspace;
	return HUF_buildCTable_wksp(tree, count, maxSymbolValue, maxNbBits, &workspace, sizeof(workspace));
	}

	size_t HUF_estimateCompressedSize(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue)
	{
	size_t nbBits = 0;
	int s;
	for (s = 0; s <= (int)maxSymbolValue; ++s) {
	nbBits += CTable[s].nbBits * count[s];
	}
	return nbBits >> 3;
	}

	int HUF_validateCTable(const HUF_CElt* CTable, const unsigned* count, unsigned maxSymbolValue) {
	int bad = 0;
	int s;
	for (s = 0; s <= (int)maxSymbolValue; ++s) {
	bad \|= (count[s] != 0) & (CTable[s].nbBits == 0);
	}
	return !bad;
	}

	size_t HUF_compressBound(size_t size) { return HUF_COMPRESSBOUND(size); }

	FORCE_INLINE_TEMPLATE void
	HUF_encodeSymbol(BIT_CStream_t* bitCPtr, U32 symbol, const HUF_CElt* CTable)
	{
	BIT_addBitsFast(bitCPtr, CTable[symbol].val, CTable[symbol].nbBits);
	}

	#define HUF_FLUSHBITS(s) BIT_flushBits(s)

	#define HUF_FLUSHBITS_1(stream) \
	if (sizeof((stream)->bitContainer)8 < HUF_TABLELOG_MAX2+7) HUF_FLUSHBITS(stream)

	#define HUF_FLUSHBITS_2(stream) \
	if (sizeof((stream)->bitContainer)8 < HUF_TABLELOG_MAX4+7) HUF_FLUSHBITS(stream)

	FORCE_INLINE_TEMPLATE size_t
	HUF_compress1X_usingCTable_internal_body(void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	const HUF_CElt* CTable)
	{
	const BYTE* ip = (const BYTE*) src;
	BYTE* const ostart = (BYTE*)dst;
	BYTE* const oend = ostart + dstSize;
	BYTE* op = ostart;
	size_t n;
	BIT_CStream_t bitC;

	/* init */
	if (dstSize < 8) return 0; /* not enough space to compress */
	{ size_t const initErr = BIT_initCStream(&bitC, op, (size_t)(oend-op));
	if (HUF_isError(initErr)) return 0; }

	n = srcSize & ~3; /* join to mod 4 */
	switch (srcSize & 3)
	{
	case 3 : HUF_encodeSymbol(&bitC, ip[n+ 2], CTable);
	HUF_FLUSHBITS_2(&bitC);
	/* fall-through */
	case 2 : HUF_encodeSymbol(&bitC, ip[n+ 1], CTable);
	HUF_FLUSHBITS_1(&bitC);
	/* fall-through */
	case 1 : HUF_encodeSymbol(&bitC, ip[n+ 0], CTable);
	HUF_FLUSHBITS(&bitC);
	/* fall-through */
	case 0 : /* fall-through */
	default: break;
	}

	for (; n>0; n-=4) { /* note : n&3==0 at this stage */
	HUF_encodeSymbol(&bitC, ip[n- 1], CTable);
	HUF_FLUSHBITS_1(&bitC);
	HUF_encodeSymbol(&bitC, ip[n- 2], CTable);
	HUF_FLUSHBITS_2(&bitC);
	HUF_encodeSymbol(&bitC, ip[n- 3], CTable);
	HUF_FLUSHBITS_1(&bitC);
	HUF_encodeSymbol(&bitC, ip[n- 4], CTable);
	HUF_FLUSHBITS(&bitC);
	}

	return BIT_closeCStream(&bitC);
	}

	#if DYNAMIC_BMI2

	static TARGET_ATTRIBUTE("bmi2") size_t
	HUF_compress1X_usingCTable_internal_bmi2(void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	const HUF_CElt* CTable)
	{
	return HUF_compress1X_usingCTable_internal_body(dst, dstSize, src, srcSize, CTable);
	}

	static size_t
	HUF_compress1X_usingCTable_internal_default(void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	const HUF_CElt* CTable)
	{
	return HUF_compress1X_usingCTable_internal_body(dst, dstSize, src, srcSize, CTable);
	}

	static size_t
	HUF_compress1X_usingCTable_internal(void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	const HUF_CElt* CTable, const int bmi2)
	{
	if (bmi2) {
	return HUF_compress1X_usingCTable_internal_bmi2(dst, dstSize, src, srcSize, CTable);
	}
	return HUF_compress1X_usingCTable_internal_default(dst, dstSize, src, srcSize, CTable);
	}

	#else

	static size_t
	HUF_compress1X_usingCTable_internal(void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	const HUF_CElt* CTable, const int bmi2)
	{
	(void)bmi2;
	return HUF_compress1X_usingCTable_internal_body(dst, dstSize, src, srcSize, CTable);
	}

	#endif

	size_t HUF_compress1X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable)
	{
	return HUF_compress1X_usingCTable_internal(dst, dstSize, src, srcSize, CTable, /* bmi2 */ 0);
	}


	static size_t
	HUF_compress4X_usingCTable_internal(void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	const HUF_CElt* CTable, int bmi2)
	{
	size_t const segmentSize = (srcSize+3)/4; /* first 3 segments */
	const BYTE* ip = (const BYTE*) src;
	const BYTE* const iend = ip + srcSize;
	BYTE* const ostart = (BYTE*) dst;
	BYTE* const oend = ostart + dstSize;
	BYTE* op = ostart;

	if (dstSize < 6 + 1 + 1 + 1 + 8) return 0; /* minimum space to compress successfully */
	if (srcSize < 12) return 0; /* no saving possible : too small input */
	op += 6; /* jumpTable */

	assert(op <= oend);
	{ CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, bmi2) );
	if (cSize==0) return 0;
	assert(cSize <= 65535);
	MEM_writeLE16(ostart, (U16)cSize);
	op += cSize;
	}

	ip += segmentSize;
	assert(op <= oend);
	{ CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, bmi2) );
	if (cSize==0) return 0;
	assert(cSize <= 65535);
	MEM_writeLE16(ostart+2, (U16)cSize);
	op += cSize;
	}

	ip += segmentSize;
	assert(op <= oend);
	{ CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, segmentSize, CTable, bmi2) );
	if (cSize==0) return 0;
	assert(cSize <= 65535);
	MEM_writeLE16(ostart+4, (U16)cSize);
	op += cSize;
	}

	ip += segmentSize;
	assert(op <= oend);
	assert(ip <= iend);
	{ CHECK_V_F(cSize, HUF_compress1X_usingCTable_internal(op, (size_t)(oend-op), ip, (size_t)(iend-ip), CTable, bmi2) );
	if (cSize==0) return 0;
	op += cSize;
	}

	return (size_t)(op-ostart);
	}

	size_t HUF_compress4X_usingCTable(void* dst, size_t dstSize, const void* src, size_t srcSize, const HUF_CElt* CTable)
	{
	return HUF_compress4X_usingCTable_internal(dst, dstSize, src, srcSize, CTable, /* bmi2 */ 0);
	}

	typedef enum { HUF_singleStream, HUF_fourStreams } HUF_nbStreams_e;

	static size_t HUF_compressCTable_internal(
	BYTE* const ostart, BYTE* op, BYTE* const oend,
	const void* src, size_t srcSize,
	HUF_nbStreams_e nbStreams, const HUF_CElt* CTable, const int bmi2)
	{
	size_t const cSize = (nbStreams==HUF_singleStream) ?
	HUF_compress1X_usingCTable_internal(op, (size_t)(oend - op), src, srcSize, CTable, bmi2) :
	HUF_compress4X_usingCTable_internal(op, (size_t)(oend - op), src, srcSize, CTable, bmi2);
	if (HUF_isError(cSize)) { return cSize; }
	if (cSize==0) { return 0; } /* uncompressible */
	op += cSize;
	/* check compressibility */
	assert(op >= ostart);
	if ((size_t)(op-ostart) >= srcSize-1) { return 0; }
	return (size_t)(op-ostart);
	}

	typedef struct {
	unsigned count[HUF_SYMBOLVALUE_MAX + 1];
	HUF_CElt CTable[HUF_SYMBOLVALUE_MAX + 1];
	HUF_buildCTable_wksp_tables buildCTable_wksp;
	} HUF_compress_tables_t;

	/* HUF_compress_internal() :
	* `workSpace` must a table of at least HUF_WORKSPACE_SIZE_U32 unsigned */
	static size_t
	HUF_compress_internal (void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	unsigned maxSymbolValue, unsigned huffLog,
	HUF_nbStreams_e nbStreams,
	void* workSpace, size_t wkspSize,
	HUF_CElt* oldHufTable, HUF_repeat* repeat, int preferRepeat,
	const int bmi2)
	{
	HUF_compress_tables_t* const table = (HUF_compress_tables_t*)workSpace;
	BYTE* const ostart = (BYTE*)dst;
	BYTE* const oend = ostart + dstSize;
	BYTE* op = ostart;

	HUF_STATIC_ASSERT(sizeof(*table) <= HUF_WORKSPACE_SIZE);

	/* checks & inits */
	if (((size_t)workSpace & 3) != 0) return ERROR(GENERIC); /* must be aligned on 4-bytes boundaries */
	if (wkspSize < HUF_WORKSPACE_SIZE) return ERROR(workSpace_tooSmall);
	if (!srcSize) return 0; /* Uncompressed */
	if (!dstSize) return 0; /* cannot fit anything within dst budget */
	if (srcSize > HUF_BLOCKSIZE_MAX) return ERROR(srcSize_wrong); /* current block size limit */
	if (huffLog > HUF_TABLELOG_MAX) return ERROR(tableLog_tooLarge);
	if (maxSymbolValue > HUF_SYMBOLVALUE_MAX) return ERROR(maxSymbolValue_tooLarge);
	if (!maxSymbolValue) maxSymbolValue = HUF_SYMBOLVALUE_MAX;
	if (!huffLog) huffLog = HUF_TABLELOG_DEFAULT;

	/* Heuristic : If old table is valid, use it for small inputs */
	if (preferRepeat && repeat && *repeat == HUF_repeat_valid) {
	return HUF_compressCTable_internal(ostart, op, oend,
	src, srcSize,
	nbStreams, oldHufTable, bmi2);
	}

	/* Scan input and build symbol stats */
	{ CHECK_V_F(largest, HIST_count_wksp (table->count, &maxSymbolValue, (const BYTE*)src, srcSize, workSpace, wkspSize) );
	if (largest == srcSize) { ostart = ((const BYTE)src)[0]; return 1; } /* single symbol, rle */
	if (largest <= (srcSize >> 7)+4) return 0; /* heuristic : probably not compressible enough */
	}

	/* Check validity of previous table */
	if ( repeat
	&& *repeat == HUF_repeat_check
	&& !HUF_validateCTable(oldHufTable, table->count, maxSymbolValue)) {
	*repeat = HUF_repeat_none;
	}
	/* Heuristic : use existing table for small inputs */
	if (preferRepeat && repeat && *repeat != HUF_repeat_none) {
	return HUF_compressCTable_internal(ostart, op, oend,
	src, srcSize,
	nbStreams, oldHufTable, bmi2);
	}

	/* Build Huffman Tree */
	huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue);
	{ size_t const maxBits = HUF_buildCTable_wksp(table->CTable, table->count,
	maxSymbolValue, huffLog,
	&table->buildCTable_wksp, sizeof(table->buildCTable_wksp));
	CHECK_F(maxBits);
	huffLog = (U32)maxBits;
	/* Zero unused symbols in CTable, so we can check it for validity */
	memset(table->CTable + (maxSymbolValue + 1), 0,
	sizeof(table->CTable) - ((maxSymbolValue + 1) * sizeof(HUF_CElt)));
	}

	/* Write table description header */
	{ CHECK_V_F(hSize, HUF_writeCTable (op, dstSize, table->CTable, maxSymbolValue, huffLog) );
	/* Check if using previous huffman table is beneficial */
	if (repeat && *repeat != HUF_repeat_none) {
	size_t const oldSize = HUF_estimateCompressedSize(oldHufTable, table->count, maxSymbolValue);
	size_t const newSize = HUF_estimateCompressedSize(table->CTable, table->count, maxSymbolValue);
	if (oldSize <= hSize + newSize \|\| hSize + 12 >= srcSize) {
	return HUF_compressCTable_internal(ostart, op, oend,
	src, srcSize,
	nbStreams, oldHufTable, bmi2);
	} }

	/* Use the new huffman table */
	if (hSize + 12ul >= srcSize) { return 0; }
	op += hSize;
	if (repeat) { *repeat = HUF_repeat_none; }
	if (oldHufTable)
	memcpy(oldHufTable, table->CTable, sizeof(table->CTable)); /* Save new table */
	}
	return HUF_compressCTable_internal(ostart, op, oend,
	src, srcSize,
	nbStreams, table->CTable, bmi2);
	}


	size_t HUF_compress1X_wksp (void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	unsigned maxSymbolValue, unsigned huffLog,
	void* workSpace, size_t wkspSize)
	{
	return HUF_compress_internal(dst, dstSize, src, srcSize,
	maxSymbolValue, huffLog, HUF_singleStream,
	workSpace, wkspSize,
	NULL, NULL, 0, 0 /bmi2/);
	}

	size_t HUF_compress1X_repeat (void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	unsigned maxSymbolValue, unsigned huffLog,
	void* workSpace, size_t wkspSize,
	HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2)
	{
	return HUF_compress_internal(dst, dstSize, src, srcSize,
	maxSymbolValue, huffLog, HUF_singleStream,
	workSpace, wkspSize, hufTable,
	repeat, preferRepeat, bmi2);
	}

	size_t HUF_compress1X (void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	unsigned maxSymbolValue, unsigned huffLog)
	{
	unsigned workSpace[HUF_WORKSPACE_SIZE_U32];
	return HUF_compress1X_wksp(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, workSpace, sizeof(workSpace));
	}

	/* HUF_compress4X_repeat():
	* compress input using 4 streams.
	* provide workspace to generate compression tables */
	size_t HUF_compress4X_wksp (void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	unsigned maxSymbolValue, unsigned huffLog,
	void* workSpace, size_t wkspSize)
	{
	return HUF_compress_internal(dst, dstSize, src, srcSize,
	maxSymbolValue, huffLog, HUF_fourStreams,
	workSpace, wkspSize,
	NULL, NULL, 0, 0 /bmi2/);
	}

	/* HUF_compress4X_repeat():
	* compress input using 4 streams.
	* re-use an existing huffman compression table */
	size_t HUF_compress4X_repeat (void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	unsigned maxSymbolValue, unsigned huffLog,
	void* workSpace, size_t wkspSize,
	HUF_CElt* hufTable, HUF_repeat* repeat, int preferRepeat, int bmi2)
	{
	return HUF_compress_internal(dst, dstSize, src, srcSize,
	maxSymbolValue, huffLog, HUF_fourStreams,
	workSpace, wkspSize,
	hufTable, repeat, preferRepeat, bmi2);
	}

	size_t HUF_compress2 (void* dst, size_t dstSize,
	const void* src, size_t srcSize,
	unsigned maxSymbolValue, unsigned huffLog)
	{
	unsigned workSpace[HUF_WORKSPACE_SIZE_U32];
	return HUF_compress4X_wksp(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, workSpace, sizeof(workSpace));
	}

	size_t HUF_compress (void* dst, size_t maxDstSize, const void* src, size_t srcSize)
	{
	return HUF_compress2(dst, maxDstSize, src, srcSize, 255, HUF_TABLELOG_DEFAULT);
	}
	/** ended inlining compress/huf_compress.c **/
	/** start inlining compress/zstd_compress_literals.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/-************************************
	* Dependencies
	***************************************/
	/** start inlining zstd_compress_literals.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_COMPRESS_LITERALS_H
	#define ZSTD_COMPRESS_LITERALS_H

	/** start inlining zstd_compress_internal.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/* This header contains definitions
	* that shall only be used by modules within lib/compress.
	*/

	#ifndef ZSTD_COMPRESS_H
	#define ZSTD_COMPRESS_H

	/-************************************
	* Dependencies
	***************************************/
	/** skipping file: ../common/zstd_internal.h **/
	/** start inlining zstd_cwksp.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_CWKSP_H
	#define ZSTD_CWKSP_H

	/-************************************
	* Dependencies
	***************************************/
	/** skipping file: ../common/zstd_internal.h **/

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/-************************************
	* Constants
	***************************************/

	/* Since the workspace is effectively its own little malloc implementation /
	* arena, when we run under ASAN, we should similarly insert redzones between
	* each internal element of the workspace, so ASAN will catch overruns that
	* reach outside an object but that stay inside the workspace.
	*
	* This defines the size of that redzone.
	*/
	#ifndef ZSTD_CWKSP_ASAN_REDZONE_SIZE
	#define ZSTD_CWKSP_ASAN_REDZONE_SIZE 128
	#endif

	/-************************************
	* Structures
	***************************************/
	typedef enum {
	ZSTD_cwksp_alloc_objects,
	ZSTD_cwksp_alloc_buffers,
	ZSTD_cwksp_alloc_aligned
	} ZSTD_cwksp_alloc_phase_e;

	/**
	* Zstd fits all its internal datastructures into a single continuous buffer,
	* so that it only needs to perform a single OS allocation (or so that a buffer
	* can be provided to it and it can perform no allocations at all). This buffer
	* is called the workspace.
	*
	* Several optimizations complicate that process of allocating memory ranges
	* from this workspace for each internal datastructure:
	*
	* - These different internal datastructures have different setup requirements:
	*
	* - The static objects need to be cleared once and can then be trivially
	* reused for each compression.
	*
	* - Various buffers don't need to be initialized at all--they are always
	* written into before they're read.
	*
	* - The matchstate tables have a unique requirement that they don't need
	* their memory to be totally cleared, but they do need the memory to have
	* some bound, i.e., a guarantee that all values in the memory they've been
	* allocated is less than some maximum value (which is the starting value
	* for the indices that they will then use for compression). When this
	* guarantee is provided to them, they can use the memory without any setup
	* work. When it can't, they have to clear the area.
	*
	* - These buffers also have different alignment requirements.
	*
	* - We would like to reuse the objects in the workspace for multiple
	* compressions without having to perform any expensive reallocation or
	* reinitialization work.
	*
	* - We would like to be able to efficiently reuse the workspace across
	* multiple compressions even when the compression parameters change and
	* we need to resize some of the objects (where possible).
	*
	* To attempt to manage this buffer, given these constraints, the ZSTD_cwksp
	* abstraction was created. It works as follows:
	*
	* Workspace Layout:
	*
	* [ ... workspace ... ]
	* [objects][tables ... ->] free space [<- ... aligned][<- ... buffers]
	*
	* The various objects that live in the workspace are divided into the
	* following categories, and are allocated separately:
	*
	* - Static objects: this is optionally the enclosing ZSTD_CCtx or ZSTD_CDict,
	* so that literally everything fits in a single buffer. Note: if present,
	* this must be the first object in the workspace, since ZSTD_free{CCtx,
	* CDict}() rely on a pointer comparison to see whether one or two frees are
	* required.
	*
	* - Fixed size objects: these are fixed-size, fixed-count objects that are
	* nonetheless "dynamically" allocated in the workspace so that we can
	* control how they're initialized separately from the broader ZSTD_CCtx.
	* Examples:
	* - Entropy Workspace
	* - 2 x ZSTD_compressedBlockState_t
	* - CDict dictionary contents
	*
	* - Tables: these are any of several different datastructures (hash tables,
	* chain tables, binary trees) that all respect a common format: they are
	* uint32_t arrays, all of whose values are between 0 and (nextSrc - base).
	* Their sizes depend on the cparams.
	*
	* - Aligned: these buffers are used for various purposes that require 4 byte
	* alignment, but don't require any initialization before they're used.
	*
	* - Buffers: these buffers are used for various purposes that don't require
	* any alignment or initialization before they're used. This means they can
	* be moved around at no cost for a new compression.
	*
	* Allocating Memory:
	*
	* The various types of objects must be allocated in order, so they can be
	* correctly packed into the workspace buffer. That order is:
	*
	* 1. Objects
	* 2. Buffers
	* 3. Aligned
	* 4. Tables
	*
	* Attempts to reserve objects of different types out of order will fail.
	*/
	typedef struct {
	void* workspace;
	void* workspaceEnd;

	void* objectEnd;
	void* tableEnd;
	void* tableValidEnd;
	void* allocStart;

	int allocFailed;
	int workspaceOversizedDuration;
	ZSTD_cwksp_alloc_phase_e phase;
	} ZSTD_cwksp;

	/-************************************
	* Functions
	***************************************/

	MEM_STATIC size_t ZSTD_cwksp_available_space(ZSTD_cwksp* ws);

	MEM_STATIC void ZSTD_cwksp_assert_internal_consistency(ZSTD_cwksp* ws) {
	(void)ws;
	assert(ws->workspace <= ws->objectEnd);
	assert(ws->objectEnd <= ws->tableEnd);
	assert(ws->objectEnd <= ws->tableValidEnd);
	assert(ws->tableEnd <= ws->allocStart);
	assert(ws->tableValidEnd <= ws->allocStart);
	assert(ws->allocStart <= ws->workspaceEnd);
	}

	/**
	* Align must be a power of 2.
	*/
	MEM_STATIC size_t ZSTD_cwksp_align(size_t size, size_t const align) {
	size_t const mask = align - 1;
	assert((align & mask) == 0);
	return (size + mask) & ~mask;
	}

	/**
	* Use this to determine how much space in the workspace we will consume to
	* allocate this object. (Normally it should be exactly the size of the object,
	* but under special conditions, like ASAN, where we pad each object, it might
	* be larger.)
	*
	* Since tables aren't currently redzoned, you don't need to call through this
	* to figure out how much space you need for the matchState tables. Everything
	* else is though.
	*/
	MEM_STATIC size_t ZSTD_cwksp_alloc_size(size_t size) {
	#if defined (ADDRESS_SANITIZER) && !defined (ZSTD_ASAN_DONT_POISON_WORKSPACE)
	return size + 2 * ZSTD_CWKSP_ASAN_REDZONE_SIZE;
	#else
	return size;
	#endif
	}

	MEM_STATIC void ZSTD_cwksp_internal_advance_phase(
	ZSTD_cwksp* ws, ZSTD_cwksp_alloc_phase_e phase) {
	assert(phase >= ws->phase);
	if (phase > ws->phase) {
	if (ws->phase < ZSTD_cwksp_alloc_buffers &&
	phase >= ZSTD_cwksp_alloc_buffers) {
	ws->tableValidEnd = ws->objectEnd;
	}
	if (ws->phase < ZSTD_cwksp_alloc_aligned &&
	phase >= ZSTD_cwksp_alloc_aligned) {
	/* If unaligned allocations down from a too-large top have left us
	* unaligned, we need to realign our alloc ptr. Technically, this
	* can consume space that is unaccounted for in the neededSpace
	* calculation. However, I believe this can only happen when the
	* workspace is too large, and specifically when it is too large
	* by a larger margin than the space that will be consumed. */
	/* TODO: cleaner, compiler warning friendly way to do this??? */
	ws->allocStart = (BYTE*)ws->allocStart - ((size_t)ws->allocStart & (sizeof(U32)-1));
	if (ws->allocStart < ws->tableValidEnd) {
	ws->tableValidEnd = ws->allocStart;
	}
	}
	ws->phase = phase;
	}
	}

	/**
	* Returns whether this object/buffer/etc was allocated in this workspace.
	*/
	MEM_STATIC int ZSTD_cwksp_owns_buffer(const ZSTD_cwksp* ws, const void* ptr) {
	return (ptr != NULL) && (ws->workspace <= ptr) && (ptr <= ws->workspaceEnd);
	}

	/**
	* Internal function. Do not use directly.
	*/
	MEM_STATIC void* ZSTD_cwksp_reserve_internal(
	ZSTD_cwksp* ws, size_t bytes, ZSTD_cwksp_alloc_phase_e phase) {
	void* alloc;
	void* bottom = ws->tableEnd;
	ZSTD_cwksp_internal_advance_phase(ws, phase);
	alloc = (BYTE *)ws->allocStart - bytes;

	#if defined (ADDRESS_SANITIZER) && !defined (ZSTD_ASAN_DONT_POISON_WORKSPACE)
	/* over-reserve space */
	alloc = (BYTE )alloc - 2 ZSTD_CWKSP_ASAN_REDZONE_SIZE;
	#endif

	DEBUGLOG(5, "cwksp: reserving %p %zd bytes, %zd bytes remaining",
	alloc, bytes, ZSTD_cwksp_available_space(ws) - bytes);
	ZSTD_cwksp_assert_internal_consistency(ws);
	assert(alloc >= bottom);
	if (alloc < bottom) {
	DEBUGLOG(4, "cwksp: alloc failed!");
	ws->allocFailed = 1;
	return NULL;
	}
	if (alloc < ws->tableValidEnd) {
	ws->tableValidEnd = alloc;
	}
	ws->allocStart = alloc;

	#if defined (ADDRESS_SANITIZER) && !defined (ZSTD_ASAN_DONT_POISON_WORKSPACE)
	/* Move alloc so there's ZSTD_CWKSP_ASAN_REDZONE_SIZE unused space on
	* either size. */
	alloc = (BYTE *)alloc + ZSTD_CWKSP_ASAN_REDZONE_SIZE;
	__asan_unpoison_memory_region(alloc, bytes);
	#endif

	return alloc;
	}

	/**
	* Reserves and returns unaligned memory.
	*/
	MEM_STATIC BYTE* ZSTD_cwksp_reserve_buffer(ZSTD_cwksp* ws, size_t bytes) {
	return (BYTE*)ZSTD_cwksp_reserve_internal(ws, bytes, ZSTD_cwksp_alloc_buffers);
	}

	/**
	* Reserves and returns memory sized on and aligned on sizeof(unsigned).
	*/
	MEM_STATIC void* ZSTD_cwksp_reserve_aligned(ZSTD_cwksp* ws, size_t bytes) {
	assert((bytes & (sizeof(U32)-1)) == 0);
	return ZSTD_cwksp_reserve_internal(ws, ZSTD_cwksp_align(bytes, sizeof(U32)), ZSTD_cwksp_alloc_aligned);
	}

	/**
	* Aligned on sizeof(unsigned). These buffers have the special property that
	* their values remain constrained, allowing us to re-use them without
	* memset()-ing them.
	*/
	MEM_STATIC void* ZSTD_cwksp_reserve_table(ZSTD_cwksp* ws, size_t bytes) {
	const ZSTD_cwksp_alloc_phase_e phase = ZSTD_cwksp_alloc_aligned;
	void* alloc = ws->tableEnd;
	void* end = (BYTE *)alloc + bytes;
	void* top = ws->allocStart;

	DEBUGLOG(5, "cwksp: reserving %p table %zd bytes, %zd bytes remaining",
	alloc, bytes, ZSTD_cwksp_available_space(ws) - bytes);
	assert((bytes & (sizeof(U32)-1)) == 0);
	ZSTD_cwksp_internal_advance_phase(ws, phase);
	ZSTD_cwksp_assert_internal_consistency(ws);
	assert(end <= top);
	if (end > top) {
	DEBUGLOG(4, "cwksp: table alloc failed!");
	ws->allocFailed = 1;
	return NULL;
	}
	ws->tableEnd = end;

	#if defined (ADDRESS_SANITIZER) && !defined (ZSTD_ASAN_DONT_POISON_WORKSPACE)
	__asan_unpoison_memory_region(alloc, bytes);
	#endif

	return alloc;
	}

	/**
	* Aligned on sizeof(void*).
	*/
	MEM_STATIC void* ZSTD_cwksp_reserve_object(ZSTD_cwksp* ws, size_t bytes) {
	size_t roundedBytes = ZSTD_cwksp_align(bytes, sizeof(void*));
	void* alloc = ws->objectEnd;
	void* end = (BYTE*)alloc + roundedBytes;

	#if defined (ADDRESS_SANITIZER) && !defined (ZSTD_ASAN_DONT_POISON_WORKSPACE)
	/* over-reserve space */
	end = (BYTE )end + 2 ZSTD_CWKSP_ASAN_REDZONE_SIZE;
	#endif

	DEBUGLOG(5,
	"cwksp: reserving %p object %zd bytes (rounded to %zd), %zd bytes remaining",
	alloc, bytes, roundedBytes, ZSTD_cwksp_available_space(ws) - roundedBytes);
	assert(((size_t)alloc & (sizeof(void*)-1)) == 0);
	assert((bytes & (sizeof(void*)-1)) == 0);
	ZSTD_cwksp_assert_internal_consistency(ws);
	/* we must be in the first phase, no advance is possible */
	if (ws->phase != ZSTD_cwksp_alloc_objects \|\| end > ws->workspaceEnd) {
	DEBUGLOG(4, "cwksp: object alloc failed!");
	ws->allocFailed = 1;
	return NULL;
	}
	ws->objectEnd = end;
	ws->tableEnd = end;
	ws->tableValidEnd = end;

	#if defined (ADDRESS_SANITIZER) && !defined (ZSTD_ASAN_DONT_POISON_WORKSPACE)
	/* Move alloc so there's ZSTD_CWKSP_ASAN_REDZONE_SIZE unused space on
	* either size. */
	alloc = (BYTE *)alloc + ZSTD_CWKSP_ASAN_REDZONE_SIZE;
	__asan_unpoison_memory_region(alloc, bytes);
	#endif

	return alloc;
	}

	MEM_STATIC void ZSTD_cwksp_mark_tables_dirty(ZSTD_cwksp* ws) {
	DEBUGLOG(4, "cwksp: ZSTD_cwksp_mark_tables_dirty");

	#if defined (MEMORY_SANITIZER) && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE)
	/* To validate that the table re-use logic is sound, and that we don't
	* access table space that we haven't cleaned, we re-"poison" the table
	* space every time we mark it dirty. */
	{
	size_t size = (BYTE)ws->tableValidEnd - (BYTE)ws->objectEnd;
	assert(__msan_test_shadow(ws->objectEnd, size) == -1);
	__msan_poison(ws->objectEnd, size);
	}
	#endif

	assert(ws->tableValidEnd >= ws->objectEnd);
	assert(ws->tableValidEnd <= ws->allocStart);
	ws->tableValidEnd = ws->objectEnd;
	ZSTD_cwksp_assert_internal_consistency(ws);
	}

	MEM_STATIC void ZSTD_cwksp_mark_tables_clean(ZSTD_cwksp* ws) {
	DEBUGLOG(4, "cwksp: ZSTD_cwksp_mark_tables_clean");
	assert(ws->tableValidEnd >= ws->objectEnd);
	assert(ws->tableValidEnd <= ws->allocStart);
	if (ws->tableValidEnd < ws->tableEnd) {
	ws->tableValidEnd = ws->tableEnd;
	}
	ZSTD_cwksp_assert_internal_consistency(ws);
	}

	/**
	* Zero the part of the allocated tables not already marked clean.
	*/
	MEM_STATIC void ZSTD_cwksp_clean_tables(ZSTD_cwksp* ws) {
	DEBUGLOG(4, "cwksp: ZSTD_cwksp_clean_tables");
	assert(ws->tableValidEnd >= ws->objectEnd);
	assert(ws->tableValidEnd <= ws->allocStart);
	if (ws->tableValidEnd < ws->tableEnd) {
	memset(ws->tableValidEnd, 0, (BYTE)ws->tableEnd - (BYTE)ws->tableValidEnd);
	}
	ZSTD_cwksp_mark_tables_clean(ws);
	}

	/**
	* Invalidates table allocations.
	* All other allocations remain valid.
	*/
	MEM_STATIC void ZSTD_cwksp_clear_tables(ZSTD_cwksp* ws) {
	DEBUGLOG(4, "cwksp: clearing tables!");

	#if defined (ADDRESS_SANITIZER) && !defined (ZSTD_ASAN_DONT_POISON_WORKSPACE)
	{
	size_t size = (BYTE)ws->tableValidEnd - (BYTE)ws->objectEnd;
	__asan_poison_memory_region(ws->objectEnd, size);
	}
	#endif

	ws->tableEnd = ws->objectEnd;
	ZSTD_cwksp_assert_internal_consistency(ws);
	}

	/**
	* Invalidates all buffer, aligned, and table allocations.
	* Object allocations remain valid.
	*/
	MEM_STATIC void ZSTD_cwksp_clear(ZSTD_cwksp* ws) {
	DEBUGLOG(4, "cwksp: clearing!");

	#if defined (MEMORY_SANITIZER) && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE)
	/* To validate that the context re-use logic is sound, and that we don't
	* access stuff that this compression hasn't initialized, we re-"poison"
	* the workspace (or at least the non-static, non-table parts of it)
	* every time we start a new compression. */
	{
	size_t size = (BYTE)ws->workspaceEnd - (BYTE)ws->tableValidEnd;
	__msan_poison(ws->tableValidEnd, size);
	}
	#endif

	#if defined (ADDRESS_SANITIZER) && !defined (ZSTD_ASAN_DONT_POISON_WORKSPACE)
	{
	size_t size = (BYTE)ws->workspaceEnd - (BYTE)ws->objectEnd;
	__asan_poison_memory_region(ws->objectEnd, size);
	}
	#endif

	ws->tableEnd = ws->objectEnd;
	ws->allocStart = ws->workspaceEnd;
	ws->allocFailed = 0;
	if (ws->phase > ZSTD_cwksp_alloc_buffers) {
	ws->phase = ZSTD_cwksp_alloc_buffers;
	}
	ZSTD_cwksp_assert_internal_consistency(ws);
	}

	/**
	* The provided workspace takes ownership of the buffer [start, start+size).
	* Any existing values in the workspace are ignored (the previously managed
	* buffer, if present, must be separately freed).
	*/
	MEM_STATIC void ZSTD_cwksp_init(ZSTD_cwksp* ws, void* start, size_t size) {
	DEBUGLOG(4, "cwksp: init'ing workspace with %zd bytes", size);
	assert(((size_t)start & (sizeof(void)-1)) == 0); / ensure correct alignment */
	ws->workspace = start;
	ws->workspaceEnd = (BYTE*)start + size;
	ws->objectEnd = ws->workspace;
	ws->tableValidEnd = ws->objectEnd;
	ws->phase = ZSTD_cwksp_alloc_objects;
	ZSTD_cwksp_clear(ws);
	ws->workspaceOversizedDuration = 0;
	ZSTD_cwksp_assert_internal_consistency(ws);
	}

	MEM_STATIC size_t ZSTD_cwksp_create(ZSTD_cwksp* ws, size_t size, ZSTD_customMem customMem) {
	void* workspace = ZSTD_malloc(size, customMem);
	DEBUGLOG(4, "cwksp: creating new workspace with %zd bytes", size);
	RETURN_ERROR_IF(workspace == NULL, memory_allocation, "NULL pointer!");
	ZSTD_cwksp_init(ws, workspace, size);
	return 0;
	}

	MEM_STATIC void ZSTD_cwksp_free(ZSTD_cwksp* ws, ZSTD_customMem customMem) {
	void *ptr = ws->workspace;
	DEBUGLOG(4, "cwksp: freeing workspace");
	memset(ws, 0, sizeof(ZSTD_cwksp));
	ZSTD_free(ptr, customMem);
	}

	/**
	* Moves the management of a workspace from one cwksp to another. The src cwksp
	* is left in an invalid state (src must be re-init()'ed before its used again).
	*/
	MEM_STATIC void ZSTD_cwksp_move(ZSTD_cwksp* dst, ZSTD_cwksp* src) {
	dst = src;
	memset(src, 0, sizeof(ZSTD_cwksp));
	}

	MEM_STATIC size_t ZSTD_cwksp_sizeof(const ZSTD_cwksp* ws) {
	return (size_t)((BYTE)ws->workspaceEnd - (BYTE)ws->workspace);
	}

	MEM_STATIC int ZSTD_cwksp_reserve_failed(const ZSTD_cwksp* ws) {
	return ws->allocFailed;
	}

	/-************************************
	* Functions Checking Free Space
	***************************************/

	MEM_STATIC size_t ZSTD_cwksp_available_space(ZSTD_cwksp* ws) {
	return (size_t)((BYTE)ws->allocStart - (BYTE)ws->tableEnd);
	}

	MEM_STATIC int ZSTD_cwksp_check_available(ZSTD_cwksp* ws, size_t additionalNeededSpace) {
	return ZSTD_cwksp_available_space(ws) >= additionalNeededSpace;
	}

	MEM_STATIC int ZSTD_cwksp_check_too_large(ZSTD_cwksp* ws, size_t additionalNeededSpace) {
	return ZSTD_cwksp_check_available(
	ws, additionalNeededSpace * ZSTD_WORKSPACETOOLARGE_FACTOR);
	}

	MEM_STATIC int ZSTD_cwksp_check_wasteful(ZSTD_cwksp* ws, size_t additionalNeededSpace) {
	return ZSTD_cwksp_check_too_large(ws, additionalNeededSpace)
	&& ws->workspaceOversizedDuration > ZSTD_WORKSPACETOOLARGE_MAXDURATION;
	}

	MEM_STATIC void ZSTD_cwksp_bump_oversized_duration(
	ZSTD_cwksp* ws, size_t additionalNeededSpace) {
	if (ZSTD_cwksp_check_too_large(ws, additionalNeededSpace)) {
	ws->workspaceOversizedDuration++;
	} else {
	ws->workspaceOversizedDuration = 0;
	}
	}

	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_CWKSP_H */
	/** ended inlining zstd_cwksp.h **/
	#ifdef ZSTD_MULTITHREAD
	/** start inlining zstdmt_compress.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTDMT_COMPRESS_H
	#define ZSTDMT_COMPRESS_H

	#if defined (__cplusplus)
	extern "C" {
	#endif


	/* Note : This is an internal API.
	* These APIs used to be exposed with ZSTDLIB_API,
	* because it used to be the only way to invoke MT compression.
	* Now, it's recommended to use ZSTD_compress2 and ZSTD_compressStream2()
	* instead.
	*
	* If you depend on these APIs and can't switch, then define
	* ZSTD_LEGACY_MULTITHREADED_API when making the dynamic library.
	* However, we may completely remove these functions in a future
	* release, so please switch soon.
	*
	* This API requires ZSTD_MULTITHREAD to be defined during compilation,
	* otherwise ZSTDMT_createCCtx*() will fail.
	*/

	#ifdef ZSTD_LEGACY_MULTITHREADED_API
	# define ZSTDMT_API ZSTDLIB_API
	#else
	# define ZSTDMT_API
	#endif

	/* === Dependencies === */
	#include <stddef.h> /* size_t */
	#define ZSTD_STATIC_LINKING_ONLY /* ZSTD_parameters */
	/** skipping file: ../zstd.h **/


	/* === Constants === */
	#ifndef ZSTDMT_NBWORKERS_MAX
	# define ZSTDMT_NBWORKERS_MAX 200
	#endif
	#ifndef ZSTDMT_JOBSIZE_MIN
	# define ZSTDMT_JOBSIZE_MIN (1 MB)
	#endif
	#define ZSTDMT_JOBLOG_MAX (MEM_32bits() ? 29 : 30)
	#define ZSTDMT_JOBSIZE_MAX (MEM_32bits() ? (512 MB) : (1024 MB))


	/* === Memory management === */
	typedef struct ZSTDMT_CCtx_s ZSTDMT_CCtx;
	/* Requires ZSTD_MULTITHREAD to be defined during compilation, otherwise it will return NULL. */
	ZSTDMT_API ZSTDMT_CCtx* ZSTDMT_createCCtx(unsigned nbWorkers);
	/* Requires ZSTD_MULTITHREAD to be defined during compilation, otherwise it will return NULL. */
	ZSTDMT_API ZSTDMT_CCtx* ZSTDMT_createCCtx_advanced(unsigned nbWorkers,
	ZSTD_customMem cMem);
	ZSTDMT_API size_t ZSTDMT_freeCCtx(ZSTDMT_CCtx* mtctx);

	ZSTDMT_API size_t ZSTDMT_sizeof_CCtx(ZSTDMT_CCtx* mtctx);


	/* === Simple one-pass compression function === */

	ZSTDMT_API size_t ZSTDMT_compressCCtx(ZSTDMT_CCtx* mtctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	int compressionLevel);



	/* === Streaming functions === */

	ZSTDMT_API size_t ZSTDMT_initCStream(ZSTDMT_CCtx* mtctx, int compressionLevel);
	ZSTDMT_API size_t ZSTDMT_resetCStream(ZSTDMT_CCtx* mtctx, unsigned long long pledgedSrcSize); /*< if srcSize is not known at reset time, use ZSTD_CONTENTSIZE_UNKNOWN. Note: for compatibility with older programs, 0 means the same as ZSTD_CONTENTSIZE_UNKNOWN, but it will change in the future to mean "empty" /

	ZSTDMT_API size_t ZSTDMT_nextInputSizeHint(const ZSTDMT_CCtx* mtctx);
	ZSTDMT_API size_t ZSTDMT_compressStream(ZSTDMT_CCtx* mtctx, ZSTD_outBuffer* output, ZSTD_inBuffer* input);

	ZSTDMT_API size_t ZSTDMT_flushStream(ZSTDMT_CCtx* mtctx, ZSTD_outBuffer* output); /*< @return : 0 == all flushed; >0 : still some data to be flushed; or an error code (ZSTD_isError()) /
	ZSTDMT_API size_t ZSTDMT_endStream(ZSTDMT_CCtx* mtctx, ZSTD_outBuffer* output); /*< @return : 0 == all flushed; >0 : still some data to be flushed; or an error code (ZSTD_isError()) /


	/* === Advanced functions and parameters === */

	ZSTDMT_API size_t ZSTDMT_compress_advanced(ZSTDMT_CCtx* mtctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const ZSTD_CDict* cdict,
	ZSTD_parameters params,
	int overlapLog);

	ZSTDMT_API size_t ZSTDMT_initCStream_advanced(ZSTDMT_CCtx* mtctx,
	const void* dict, size_t dictSize, /* dict can be released after init, a local copy is preserved within zcs */
	ZSTD_parameters params,
	unsigned long long pledgedSrcSize); /* pledgedSrcSize is optional and can be zero == unknown */

	ZSTDMT_API size_t ZSTDMT_initCStream_usingCDict(ZSTDMT_CCtx* mtctx,
	const ZSTD_CDict* cdict,
	ZSTD_frameParameters fparams,
	unsigned long long pledgedSrcSize); /* note : zero means empty */

	/* ZSTDMT_parameter :
	* List of parameters that can be set using ZSTDMT_setMTCtxParameter() */
	typedef enum {
	ZSTDMT_p_jobSize, /* Each job is compressed in parallel. By default, this value is dynamically determined depending on compression parameters. Can be set explicitly here. */
	ZSTDMT_p_overlapLog, /* Each job may reload a part of previous job to enhance compression ratio; 0 == no overlap, 6(default) == use 1/8th of window, >=9 == use full window. This is a "sticky" parameter : its value will be re-used on next compression job */
	ZSTDMT_p_rsyncable /* Enables rsyncable mode. */
	} ZSTDMT_parameter;

	/* ZSTDMT_setMTCtxParameter() :
	* allow setting individual parameters, one at a time, among a list of enums defined in ZSTDMT_parameter.
	* The function must be called typically after ZSTD_createCCtx() but __before ZSTDMT_init*() !__
	* Parameters not explicitly reset by ZSTDMT_init*() remain the same in consecutive compression sessions.
	* @return : 0, or an error code (which can be tested using ZSTD_isError()) */
	ZSTDMT_API size_t ZSTDMT_setMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter parameter, int value);

	/* ZSTDMT_getMTCtxParameter() :
	* Query the ZSTDMT_CCtx for a parameter value.
	* @return : 0, or an error code (which can be tested using ZSTD_isError()) */
	ZSTDMT_API size_t ZSTDMT_getMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter parameter, int* value);


	/*! ZSTDMT_compressStream_generic() :
	* Combines ZSTDMT_compressStream() with optional ZSTDMT_flushStream() or ZSTDMT_endStream()
	* depending on flush directive.
	* @return : minimum amount of data still to be flushed
	* 0 if fully flushed
	* or an error code
	* note : needs to be init using any ZSTD_initCStream() variant /
	ZSTDMT_API size_t ZSTDMT_compressStream_generic(ZSTDMT_CCtx* mtctx,
	ZSTD_outBuffer* output,
	ZSTD_inBuffer* input,
	ZSTD_EndDirective endOp);


	/* ========================================================
	* === Private interface, for use by ZSTD_compress.c ===
	* === Not exposed in libzstd. Never invoke directly ===
	* ======================================================== */

	/*! ZSTDMT_toFlushNow()
	* Tell how many bytes are ready to be flushed immediately.
	* Probe the oldest active job (not yet entirely flushed) and check its output buffer.
	* If return 0, it means there is no active job,
	* or, it means oldest job is still active, but everything produced has been flushed so far,
	* therefore flushing is limited by speed of oldest job. */
	size_t ZSTDMT_toFlushNow(ZSTDMT_CCtx* mtctx);

	/*! ZSTDMT_CCtxParam_setMTCtxParameter()
	* like ZSTDMT_setMTCtxParameter(), but into a ZSTD_CCtx_Params */
	size_t ZSTDMT_CCtxParam_setMTCtxParameter(ZSTD_CCtx_params* params, ZSTDMT_parameter parameter, int value);

	/*! ZSTDMT_CCtxParam_setNbWorkers()
	* Set nbWorkers, and clamp it.
	* Also reset jobSize and overlapLog */
	size_t ZSTDMT_CCtxParam_setNbWorkers(ZSTD_CCtx_params* params, unsigned nbWorkers);

	/*! ZSTDMT_updateCParams_whileCompressing() :
	* Updates only a selected set of compression parameters, to remain compatible with current frame.
	* New parameters will be applied to next compression job. */
	void ZSTDMT_updateCParams_whileCompressing(ZSTDMT_CCtx* mtctx, const ZSTD_CCtx_params* cctxParams);

	/*! ZSTDMT_getFrameProgression():
	* tells how much data has been consumed (input) and produced (output) for current frame.
	* able to count progression inside worker threads.
	*/
	ZSTD_frameProgression ZSTDMT_getFrameProgression(ZSTDMT_CCtx* mtctx);


	/*! ZSTDMT_initCStream_internal() :
	* Private use only. Init streaming operation.
	* expects params to be valid.
	* must receive dict, or cdict, or none, but not both.
	* @return : 0, or an error code */
	size_t ZSTDMT_initCStream_internal(ZSTDMT_CCtx* zcs,
	const void* dict, size_t dictSize, ZSTD_dictContentType_e dictContentType,
	const ZSTD_CDict* cdict,
	ZSTD_CCtx_params params, unsigned long long pledgedSrcSize);


	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTDMT_COMPRESS_H */
	/** ended inlining zstdmt_compress.h **/
	#endif

	#if defined (__cplusplus)
	extern "C" {
	#endif


	/-************************************
	* Constants
	***************************************/
	#define kSearchStrength 8
	#define HASH_READ_SIZE 8
	#define ZSTD_DUBT_UNSORTED_MARK 1 /* For btlazy2 strategy, index ZSTD_DUBT_UNSORTED_MARK==1 means "unsorted".
	It could be confused for a real successor at index "1", if sorted as larger than its predecessor.
	It's not a big deal though : candidate will just be sorted again.
	Additionally, candidate position 1 will be lost.
	But candidate 1 cannot hide a large tree of candidates, so it's a minimal loss.
	The benefit is that ZSTD_DUBT_UNSORTED_MARK cannot be mishandled after table re-use with a different strategy.
	This constant is required by ZSTD_compressBlock_btlazy2() and ZSTD_reduceTable_internal() */


	/-************************************
	* Context memory management
	***************************************/
	typedef enum { ZSTDcs_created=0, ZSTDcs_init, ZSTDcs_ongoing, ZSTDcs_ending } ZSTD_compressionStage_e;
	typedef enum { zcss_init=0, zcss_load, zcss_flush } ZSTD_cStreamStage;

	typedef struct ZSTD_prefixDict_s {
	const void* dict;
	size_t dictSize;
	ZSTD_dictContentType_e dictContentType;
	} ZSTD_prefixDict;

	typedef struct {
	void* dictBuffer;
	void const* dict;
	size_t dictSize;
	ZSTD_dictContentType_e dictContentType;
	ZSTD_CDict* cdict;
	} ZSTD_localDict;

	typedef struct {
	U32 CTable[HUF_CTABLE_SIZE_U32(255)];
	HUF_repeat repeatMode;
	} ZSTD_hufCTables_t;

	typedef struct {
	FSE_CTable offcodeCTable[FSE_CTABLE_SIZE_U32(OffFSELog, MaxOff)];
	FSE_CTable matchlengthCTable[FSE_CTABLE_SIZE_U32(MLFSELog, MaxML)];
	FSE_CTable litlengthCTable[FSE_CTABLE_SIZE_U32(LLFSELog, MaxLL)];
	FSE_repeat offcode_repeatMode;
	FSE_repeat matchlength_repeatMode;
	FSE_repeat litlength_repeatMode;
	} ZSTD_fseCTables_t;

	typedef struct {
	ZSTD_hufCTables_t huf;
	ZSTD_fseCTables_t fse;
	} ZSTD_entropyCTables_t;

	typedef struct {
	U32 off;
	U32 len;
	} ZSTD_match_t;

	typedef struct {
	int price;
	U32 off;
	U32 mlen;
	U32 litlen;
	U32 rep[ZSTD_REP_NUM];
	} ZSTD_optimal_t;

	typedef enum { zop_dynamic=0, zop_predef } ZSTD_OptPrice_e;

	typedef struct {
	/* All tables are allocated inside cctx->workspace by ZSTD_resetCCtx_internal() */
	unsigned* litFreq; /* table of literals statistics, of size 256 */
	unsigned* litLengthFreq; /* table of litLength statistics, of size (MaxLL+1) */
	unsigned* matchLengthFreq; /* table of matchLength statistics, of size (MaxML+1) */
	unsigned* offCodeFreq; /* table of offCode statistics, of size (MaxOff+1) */
	ZSTD_match_t* matchTable; /* list of found matches, of size ZSTD_OPT_NUM+1 */
	ZSTD_optimal_t* priceTable; /* All positions tracked by optimal parser, of size ZSTD_OPT_NUM+1 */

	U32 litSum; /* nb of literals */
	U32 litLengthSum; /* nb of litLength codes */
	U32 matchLengthSum; /* nb of matchLength codes */
	U32 offCodeSum; /* nb of offset codes */
	U32 litSumBasePrice; /* to compare to log2(litfreq) */
	U32 litLengthSumBasePrice; /* to compare to log2(llfreq) */
	U32 matchLengthSumBasePrice;/* to compare to log2(mlfreq) */
	U32 offCodeSumBasePrice; /* to compare to log2(offreq) */
	ZSTD_OptPrice_e priceType; /* prices can be determined dynamically, or follow a pre-defined cost structure */
	const ZSTD_entropyCTables_t* symbolCosts; /* pre-calculated dictionary statistics */
	ZSTD_literalCompressionMode_e literalCompressionMode;
	} optState_t;

	typedef struct {
	ZSTD_entropyCTables_t entropy;
	U32 rep[ZSTD_REP_NUM];
	} ZSTD_compressedBlockState_t;

	typedef struct {
	BYTE const* nextSrc; /* next block here to continue on current prefix */
	BYTE const* base; /* All regular indexes relative to this position */
	BYTE const* dictBase; /* extDict indexes relative to this position */
	U32 dictLimit; /* below that point, need extDict */
	U32 lowLimit; /* below that point, no more valid data */
	} ZSTD_window_t;

	typedef struct ZSTD_matchState_t ZSTD_matchState_t;
	struct ZSTD_matchState_t {
	ZSTD_window_t window; /* State for window round buffer management */
	U32 loadedDictEnd; /* index of end of dictionary, within context's referential.
	* When loadedDictEnd != 0, a dictionary is in use, and still valid.
	* This relies on a mechanism to set loadedDictEnd=0 when dictionary is no longer within distance.
	* Such mechanism is provided within ZSTD_window_enforceMaxDist() and ZSTD_checkDictValidity().
	* When dict referential is copied into active context (i.e. not attached),
	* loadedDictEnd == dictSize, since referential starts from zero.
	*/
	U32 nextToUpdate; /* index from which to continue table update */
	U32 hashLog3; /* dispatch table for matches of len==3 : larger == faster, more memory */
	U32* hashTable;
	U32* hashTable3;
	U32* chainTable;
	optState_t opt; /* optimal parser state */
	const ZSTD_matchState_t* dictMatchState;
	ZSTD_compressionParameters cParams;
	};

	typedef struct {
	ZSTD_compressedBlockState_t* prevCBlock;
	ZSTD_compressedBlockState_t* nextCBlock;
	ZSTD_matchState_t matchState;
	} ZSTD_blockState_t;

	typedef struct {
	U32 offset;
	U32 checksum;
	} ldmEntry_t;

	typedef struct {
	ZSTD_window_t window; /* State for the window round buffer management */
	ldmEntry_t* hashTable;
	U32 loadedDictEnd;
	BYTE* bucketOffsets; /* Next position in bucket to insert entry */
	U64 hashPower; /* Used to compute the rolling hash.
	* Depends on ldmParams.minMatchLength */
	} ldmState_t;

	typedef struct {
	U32 enableLdm; /* 1 if enable long distance matching */
	U32 hashLog; /* Log size of hashTable */
	U32 bucketSizeLog; /* Log bucket size for collision resolution, at most 8 */
	U32 minMatchLength; /* Minimum match length */
	U32 hashRateLog; /* Log number of entries to skip */
	U32 windowLog; /* Window log for the LDM */
	} ldmParams_t;

	typedef struct {
	U32 offset;
	U32 litLength;
	U32 matchLength;
	} rawSeq;

	typedef struct {
	rawSeq* seq; /* The start of the sequences */
	size_t pos; /* The position where reading stopped. <= size. */
	size_t size; /* The number of sequences. <= capacity. */
	size_t capacity; /* The capacity starting from `seq` pointer */
	} rawSeqStore_t;

	typedef struct {
	int collectSequences;
	ZSTD_Sequence* seqStart;
	size_t seqIndex;
	size_t maxSequences;
	} SeqCollector;

	struct ZSTD_CCtx_params_s {
	ZSTD_format_e format;
	ZSTD_compressionParameters cParams;
	ZSTD_frameParameters fParams;

	int compressionLevel;
	int forceWindow; /* force back-references to respect limit of
	* 1<<wLog, even for dictionary */
	size_t targetCBlockSize; /* Tries to fit compressed block size to be around targetCBlockSize.
	* No target when targetCBlockSize == 0.
	* There is no guarantee on compressed block size */
	int srcSizeHint; /* User's best guess of source size.
	* Hint is not valid when srcSizeHint == 0.
	* There is no guarantee that hint is close to actual source size */

	ZSTD_dictAttachPref_e attachDictPref;
	ZSTD_literalCompressionMode_e literalCompressionMode;

	/* Multithreading: used to pass parameters to mtctx */
	int nbWorkers;
	size_t jobSize;
	int overlapLog;
	int rsyncable;

	/* Long distance matching parameters */
	ldmParams_t ldmParams;

	/* Internal use, for createCCtxParams() and freeCCtxParams() only */
	ZSTD_customMem customMem;
	}; /* typedef'd to ZSTD_CCtx_params within "zstd.h" */

	struct ZSTD_CCtx_s {
	ZSTD_compressionStage_e stage;
	int cParamsChanged; /* == 1 if cParams(except wlog) or compression level are changed in requestedParams. Triggers transmission of new params to ZSTDMT (if available) then reset to 0. */
	int bmi2; /* == 1 if the CPU supports BMI2 and 0 otherwise. CPU support is determined dynamically once per context lifetime. */
	ZSTD_CCtx_params requestedParams;
	ZSTD_CCtx_params appliedParams;
	U32 dictID;

	ZSTD_cwksp workspace; /* manages buffer for dynamic allocations */
	size_t blockSize;
	unsigned long long pledgedSrcSizePlusOne; /* this way, 0 (default) == unknown */
	unsigned long long consumedSrcSize;
	unsigned long long producedCSize;
	XXH64_state_t xxhState;
	ZSTD_customMem customMem;
	size_t staticSize;
	SeqCollector seqCollector;
	int isFirstBlock;
	int initialized;

	seqStore_t seqStore; /* sequences storage ptrs */
	ldmState_t ldmState; /* long distance matching state */
	rawSeq* ldmSequences; /* Storage for the ldm output sequences */
	size_t maxNbLdmSequences;
	rawSeqStore_t externSeqStore; /* Mutable reference to external sequences */
	ZSTD_blockState_t blockState;
	U32* entropyWorkspace; /* entropy workspace of HUF_WORKSPACE_SIZE bytes */

	/* streaming */
	char* inBuff;
	size_t inBuffSize;
	size_t inToCompress;
	size_t inBuffPos;
	size_t inBuffTarget;
	char* outBuff;
	size_t outBuffSize;
	size_t outBuffContentSize;
	size_t outBuffFlushedSize;
	ZSTD_cStreamStage streamStage;
	U32 frameEnded;

	/* Dictionary */
	ZSTD_localDict localDict;
	const ZSTD_CDict* cdict;
	ZSTD_prefixDict prefixDict; /* single-usage dictionary */

	/* Multi-threading */
	#ifdef ZSTD_MULTITHREAD
	ZSTDMT_CCtx* mtctx;
	#endif
	};

	typedef enum { ZSTD_dtlm_fast, ZSTD_dtlm_full } ZSTD_dictTableLoadMethod_e;

	typedef enum { ZSTD_noDict = 0, ZSTD_extDict = 1, ZSTD_dictMatchState = 2 } ZSTD_dictMode_e;


	typedef size_t (*ZSTD_blockCompressor) (
	ZSTD_matchState_t* bs, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	ZSTD_blockCompressor ZSTD_selectBlockCompressor(ZSTD_strategy strat, ZSTD_dictMode_e dictMode);


	MEM_STATIC U32 ZSTD_LLcode(U32 litLength)
	{
	static const BYTE LL_Code[64] = { 0, 1, 2, 3, 4, 5, 6, 7,
	8, 9, 10, 11, 12, 13, 14, 15,
	16, 16, 17, 17, 18, 18, 19, 19,
	20, 20, 20, 20, 21, 21, 21, 21,
	22, 22, 22, 22, 22, 22, 22, 22,
	23, 23, 23, 23, 23, 23, 23, 23,
	24, 24, 24, 24, 24, 24, 24, 24,
	24, 24, 24, 24, 24, 24, 24, 24 };
	static const U32 LL_deltaCode = 19;
	return (litLength > 63) ? ZSTD_highbit32(litLength) + LL_deltaCode : LL_Code[litLength];
	}

	/* ZSTD_MLcode() :
	* note : mlBase = matchLength - MINMATCH;
	* because it's the format it's stored in seqStore->sequences */
	MEM_STATIC U32 ZSTD_MLcode(U32 mlBase)
	{
	static const BYTE ML_Code[128] = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,
	16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31,
	32, 32, 33, 33, 34, 34, 35, 35, 36, 36, 36, 36, 37, 37, 37, 37,
	38, 38, 38, 38, 38, 38, 38, 38, 39, 39, 39, 39, 39, 39, 39, 39,
	40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40,
	41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41,
	42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42,
	42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42 };
	static const U32 ML_deltaCode = 36;
	return (mlBase > 127) ? ZSTD_highbit32(mlBase) + ML_deltaCode : ML_Code[mlBase];
	}

	typedef struct repcodes_s {
	U32 rep[3];
	} repcodes_t;

	MEM_STATIC repcodes_t ZSTD_updateRep(U32 const rep[3], U32 const offset, U32 const ll0)
	{
	repcodes_t newReps;
	if (offset >= ZSTD_REP_NUM) { /* full offset */
	newReps.rep[2] = rep[1];
	newReps.rep[1] = rep[0];
	newReps.rep[0] = offset - ZSTD_REP_MOVE;
	} else { /* repcode */
	U32 const repCode = offset + ll0;
	if (repCode > 0) { /* note : if repCode==0, no change */
	U32 const currentOffset = (repCode==ZSTD_REP_NUM) ? (rep[0] - 1) : rep[repCode];
	newReps.rep[2] = (repCode >= 2) ? rep[1] : rep[2];
	newReps.rep[1] = rep[0];
	newReps.rep[0] = currentOffset;
	} else { /* repCode == 0 */
	memcpy(&newReps, rep, sizeof(newReps));
	}
	}
	return newReps;
	}

	/* ZSTD_cParam_withinBounds:
	* @return 1 if value is within cParam bounds,
	* 0 otherwise */
	MEM_STATIC int ZSTD_cParam_withinBounds(ZSTD_cParameter cParam, int value)
	{
	ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam);
	if (ZSTD_isError(bounds.error)) return 0;
	if (value < bounds.lowerBound) return 0;
	if (value > bounds.upperBound) return 0;
	return 1;
	}

	/* ZSTD_noCompressBlock() :
	* Writes uncompressed block to dst buffer from given src.
	* Returns the size of the block */
	MEM_STATIC size_t ZSTD_noCompressBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize, U32 lastBlock)
	{
	U32 const cBlockHeader24 = lastBlock + (((U32)bt_raw)<<1) + (U32)(srcSize << 3);
	RETURN_ERROR_IF(srcSize + ZSTD_blockHeaderSize > dstCapacity,
	dstSize_tooSmall, "dst buf too small for uncompressed block");
	MEM_writeLE24(dst, cBlockHeader24);
	memcpy((BYTE*)dst + ZSTD_blockHeaderSize, src, srcSize);
	return ZSTD_blockHeaderSize + srcSize;
	}

	MEM_STATIC size_t ZSTD_rleCompressBlock (void* dst, size_t dstCapacity, BYTE src, size_t srcSize, U32 lastBlock)
	{
	BYTE* const op = (BYTE*)dst;
	U32 const cBlockHeader = lastBlock + (((U32)bt_rle)<<1) + (U32)(srcSize << 3);
	RETURN_ERROR_IF(dstCapacity < 4, dstSize_tooSmall, "");
	MEM_writeLE24(op, cBlockHeader);
	op[3] = src;
	return 4;
	}


	/* ZSTD_minGain() :
	* minimum compression required
	* to generate a compress block or a compressed literals section.
	* note : use same formula for both situations */
	MEM_STATIC size_t ZSTD_minGain(size_t srcSize, ZSTD_strategy strat)
	{
	U32 const minlog = (strat>=ZSTD_btultra) ? (U32)(strat) - 1 : 6;
	ZSTD_STATIC_ASSERT(ZSTD_btultra == 8);
	assert(ZSTD_cParam_withinBounds(ZSTD_c_strategy, strat));
	return (srcSize >> minlog) + 2;
	}

	MEM_STATIC int ZSTD_disableLiteralsCompression(const ZSTD_CCtx_params* cctxParams)
	{
	switch (cctxParams->literalCompressionMode) {
	case ZSTD_lcm_huffman:
	return 0;
	case ZSTD_lcm_uncompressed:
	return 1;
	default:
	assert(0 /* impossible: pre-validated */);
	/* fall-through */
	case ZSTD_lcm_auto:
	return (cctxParams->cParams.strategy == ZSTD_fast) && (cctxParams->cParams.targetLength > 0);
	}
	}

	/*! ZSTD_safecopyLiterals() :
	* memcpy() function that won't read beyond more than WILDCOPY_OVERLENGTH bytes past ilimit_w.
	* Only called when the sequence ends past ilimit_w, so it only needs to be optimized for single
	* large copies.
	*/
	static void ZSTD_safecopyLiterals(BYTE* op, BYTE const* ip, BYTE const* const iend, BYTE const* ilimit_w) {
	assert(iend > ilimit_w);
	if (ip <= ilimit_w) {
	ZSTD_wildcopy(op, ip, ilimit_w - ip, ZSTD_no_overlap);
	op += ilimit_w - ip;
	ip = ilimit_w;
	}
	while (ip < iend) op++ = ip++;
	}

	/*! ZSTD_storeSeq() :
	* Store a sequence (litlen, litPtr, offCode and mlBase) into seqStore_t.
	* `offCode` : distance to match + ZSTD_REP_MOVE (values <= ZSTD_REP_MOVE are repCodes).
	* `mlBase` : matchLength - MINMATCH
	* Allowed to overread literals up to litLimit.
	*/
	HINT_INLINE UNUSED_ATTR
	void ZSTD_storeSeq(seqStore_t* seqStorePtr, size_t litLength, const BYTE* literals, const BYTE* litLimit, U32 offCode, size_t mlBase)
	{
	BYTE const* const litLimit_w = litLimit - WILDCOPY_OVERLENGTH;
	BYTE const* const litEnd = literals + litLength;
	#if defined(DEBUGLEVEL) && (DEBUGLEVEL >= 6)
	static const BYTE* g_start = NULL;
	if (g_start==NULL) g_start = (const BYTE)literals; / note : index only works for compression within a single segment */
	{ U32 const pos = (U32)((const BYTE*)literals - g_start);
	DEBUGLOG(6, "Cpos%7u :%3u literals, match%4u bytes at offCode%7u",
	pos, (U32)litLength, (U32)mlBase+MINMATCH, (U32)offCode);
	}
	#endif
	assert((size_t)(seqStorePtr->sequences - seqStorePtr->sequencesStart) < seqStorePtr->maxNbSeq);
	/* copy Literals */
	assert(seqStorePtr->maxNbLit <= 128 KB);
	assert(seqStorePtr->lit + litLength <= seqStorePtr->litStart + seqStorePtr->maxNbLit);
	assert(literals + litLength <= litLimit);
	if (litEnd <= litLimit_w) {
	/* Common case we can use wildcopy.
	* First copy 16 bytes, because literals are likely short.
	*/
	assert(WILDCOPY_OVERLENGTH >= 16);
	ZSTD_copy16(seqStorePtr->lit, literals);
	if (litLength > 16) {
	ZSTD_wildcopy(seqStorePtr->lit+16, literals+16, (ptrdiff_t)litLength-16, ZSTD_no_overlap);
	}
	} else {
	ZSTD_safecopyLiterals(seqStorePtr->lit, literals, litEnd, litLimit_w);
	}
	seqStorePtr->lit += litLength;

	/* literal Length */
	if (litLength>0xFFFF) {
	assert(seqStorePtr->longLengthID == 0); /* there can only be a single long length */
	seqStorePtr->longLengthID = 1;
	seqStorePtr->longLengthPos = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
	}
	seqStorePtr->sequences[0].litLength = (U16)litLength;

	/* match offset */
	seqStorePtr->sequences[0].offset = offCode + 1;

	/* match Length */
	if (mlBase>0xFFFF) {
	assert(seqStorePtr->longLengthID == 0); /* there can only be a single long length */
	seqStorePtr->longLengthID = 2;
	seqStorePtr->longLengthPos = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
	}
	seqStorePtr->sequences[0].matchLength = (U16)mlBase;

	seqStorePtr->sequences++;
	}


	/-************************************
	* Match length counter
	***************************************/
	static unsigned ZSTD_NbCommonBytes (size_t val)
	{
	if (MEM_isLittleEndian()) {
	if (MEM_64bits()) {
	# if defined(_MSC_VER) && defined(_WIN64)
	unsigned long r = 0;
	return _BitScanForward64( &r, (U64)val ) ? (unsigned)(r >> 3) : 0;
	# elif defined(__GNUC__) && (__GNUC__ >= 4)
	return (__builtin_ctzll((U64)val) >> 3);
	# else
	static const int DeBruijnBytePos[64] = { 0, 0, 0, 0, 0, 1, 1, 2,
	0, 3, 1, 3, 1, 4, 2, 7,
	0, 2, 3, 6, 1, 5, 3, 5,
	1, 3, 4, 4, 2, 5, 6, 7,
	7, 0, 1, 2, 3, 3, 4, 6,
	2, 6, 5, 5, 3, 4, 5, 6,
	7, 1, 2, 4, 6, 4, 4, 5,
	7, 2, 6, 5, 7, 6, 7, 7 };
	return DeBruijnBytePos[((U64)((val & -(long long)val) * 0x0218A392CDABBD3FULL)) >> 58];
	# endif
	} else { /* 32 bits */
	# if defined(_MSC_VER)
	unsigned long r=0;
	return _BitScanForward( &r, (U32)val ) ? (unsigned)(r >> 3) : 0;
	# elif defined(__GNUC__) && (__GNUC__ >= 3)
	return (__builtin_ctz((U32)val) >> 3);
	# else
	static const int DeBruijnBytePos[32] = { 0, 0, 3, 0, 3, 1, 3, 0,
	3, 2, 2, 1, 3, 2, 0, 1,
	3, 3, 1, 2, 2, 2, 2, 0,
	3, 1, 2, 0, 1, 0, 1, 1 };
	return DeBruijnBytePos[((U32)((val & -(S32)val) * 0x077CB531U)) >> 27];
	# endif
	}
	} else { /* Big Endian CPU */
	if (MEM_64bits()) {
	# if defined(_MSC_VER) && defined(_WIN64)
	unsigned long r = 0;
	return _BitScanReverse64( &r, val ) ? (unsigned)(r >> 3) : 0;
	# elif defined(__GNUC__) && (__GNUC__ >= 4)
	return (__builtin_clzll(val) >> 3);
	# else
	unsigned r;
	const unsigned n32 = sizeof(size_t)4; / calculate this way due to compiler complaining in 32-bits mode */
	if (!(val>>n32)) { r=4; } else { r=0; val>>=n32; }
	if (!(val>>16)) { r+=2; val>>=8; } else { val>>=24; }
	r += (!val);
	return r;
	# endif
	} else { /* 32 bits */
	# if defined(_MSC_VER)
	unsigned long r = 0;
	return _BitScanReverse( &r, (unsigned long)val ) ? (unsigned)(r >> 3) : 0;
	# elif defined(__GNUC__) && (__GNUC__ >= 3)
	return (__builtin_clz((U32)val) >> 3);
	# else
	unsigned r;
	if (!(val>>16)) { r=2; val>>=8; } else { r=0; val>>=24; }
	r += (!val);
	return r;
	# endif
	} }
	}


	MEM_STATIC size_t ZSTD_count(const BYTE* pIn, const BYTE* pMatch, const BYTE* const pInLimit)
	{
	const BYTE* const pStart = pIn;
	const BYTE* const pInLoopLimit = pInLimit - (sizeof(size_t)-1);

	if (pIn < pInLoopLimit) {
	{ size_t const diff = MEM_readST(pMatch) ^ MEM_readST(pIn);
	if (diff) return ZSTD_NbCommonBytes(diff); }
	pIn+=sizeof(size_t); pMatch+=sizeof(size_t);
	while (pIn < pInLoopLimit) {
	size_t const diff = MEM_readST(pMatch) ^ MEM_readST(pIn);
	if (!diff) { pIn+=sizeof(size_t); pMatch+=sizeof(size_t); continue; }
	pIn += ZSTD_NbCommonBytes(diff);
	return (size_t)(pIn - pStart);
	} }
	if (MEM_64bits() && (pIn<(pInLimit-3)) && (MEM_read32(pMatch) == MEM_read32(pIn))) { pIn+=4; pMatch+=4; }
	if ((pIn<(pInLimit-1)) && (MEM_read16(pMatch) == MEM_read16(pIn))) { pIn+=2; pMatch+=2; }
	if ((pIn<pInLimit) && (pMatch == pIn)) pIn++;
	return (size_t)(pIn - pStart);
	}

	/** ZSTD_count_2segments() :
	* can count match length with `ip` & `match` in 2 different segments.
	* convention : on reaching mEnd, match count continue starting from iStart
	*/
	MEM_STATIC size_t
	ZSTD_count_2segments(const BYTE* ip, const BYTE* match,
	const BYTE* iEnd, const BYTE* mEnd, const BYTE* iStart)
	{
	const BYTE* const vEnd = MIN( ip + (mEnd - match), iEnd);
	size_t const matchLength = ZSTD_count(ip, match, vEnd);
	if (match + matchLength != mEnd) return matchLength;
	DEBUGLOG(7, "ZSTD_count_2segments: found a 2-parts match (current length==%zu)", matchLength);
	DEBUGLOG(7, "distance from match beginning to end dictionary = %zi", mEnd - match);
	DEBUGLOG(7, "distance from current pos to end buffer = %zi", iEnd - ip);
	DEBUGLOG(7, "next byte : ip==%02X, istart==%02X", ip[matchLength], *iStart);
	DEBUGLOG(7, "final match length = %zu", matchLength + ZSTD_count(ip+matchLength, iStart, iEnd));
	return matchLength + ZSTD_count(ip+matchLength, iStart, iEnd);
	}


	/-************************************
	* Hashes
	***************************************/
	static const U32 prime3bytes = 506832829U;
	static U32 ZSTD_hash3(U32 u, U32 h) { return ((u << (32-24)) * prime3bytes) >> (32-h) ; }
	MEM_STATIC size_t ZSTD_hash3Ptr(const void* ptr, U32 h) { return ZSTD_hash3(MEM_readLE32(ptr), h); } /* only in zstd_opt.h */

	static const U32 prime4bytes = 2654435761U;
	static U32 ZSTD_hash4(U32 u, U32 h) { return (u * prime4bytes) >> (32-h) ; }
	static size_t ZSTD_hash4Ptr(const void* ptr, U32 h) { return ZSTD_hash4(MEM_read32(ptr), h); }

	static const U64 prime5bytes = 889523592379ULL;
	static size_t ZSTD_hash5(U64 u, U32 h) { return (size_t)(((u << (64-40)) * prime5bytes) >> (64-h)) ; }
	static size_t ZSTD_hash5Ptr(const void* p, U32 h) { return ZSTD_hash5(MEM_readLE64(p), h); }

	static const U64 prime6bytes = 227718039650203ULL;
	static size_t ZSTD_hash6(U64 u, U32 h) { return (size_t)(((u << (64-48)) * prime6bytes) >> (64-h)) ; }
	static size_t ZSTD_hash6Ptr(const void* p, U32 h) { return ZSTD_hash6(MEM_readLE64(p), h); }

	static const U64 prime7bytes = 58295818150454627ULL;
	static size_t ZSTD_hash7(U64 u, U32 h) { return (size_t)(((u << (64-56)) * prime7bytes) >> (64-h)) ; }
	static size_t ZSTD_hash7Ptr(const void* p, U32 h) { return ZSTD_hash7(MEM_readLE64(p), h); }

	static const U64 prime8bytes = 0xCF1BBCDCB7A56463ULL;
	static size_t ZSTD_hash8(U64 u, U32 h) { return (size_t)(((u) * prime8bytes) >> (64-h)) ; }
	static size_t ZSTD_hash8Ptr(const void* p, U32 h) { return ZSTD_hash8(MEM_readLE64(p), h); }

	MEM_STATIC size_t ZSTD_hashPtr(const void* p, U32 hBits, U32 mls)
	{
	switch(mls)
	{
	default:
	case 4: return ZSTD_hash4Ptr(p, hBits);
	case 5: return ZSTD_hash5Ptr(p, hBits);
	case 6: return ZSTD_hash6Ptr(p, hBits);
	case 7: return ZSTD_hash7Ptr(p, hBits);
	case 8: return ZSTD_hash8Ptr(p, hBits);
	}
	}

	/** ZSTD_ipow() :
	* Return base^exponent.
	*/
	static U64 ZSTD_ipow(U64 base, U64 exponent)
	{
	U64 power = 1;
	while (exponent) {
	if (exponent & 1) power *= base;
	exponent >>= 1;
	base *= base;
	}
	return power;
	}

	#define ZSTD_ROLL_HASH_CHAR_OFFSET 10

	/** ZSTD_rollingHash_append() :
	* Add the buffer to the hash value.
	*/
	static U64 ZSTD_rollingHash_append(U64 hash, void const* buf, size_t size)
	{
	BYTE const* istart = (BYTE const*)buf;
	size_t pos;
	for (pos = 0; pos < size; ++pos) {
	hash *= prime8bytes;
	hash += istart[pos] + ZSTD_ROLL_HASH_CHAR_OFFSET;
	}
	return hash;
	}

	/** ZSTD_rollingHash_compute() :
	* Compute the rolling hash value of the buffer.
	*/
	MEM_STATIC U64 ZSTD_rollingHash_compute(void const* buf, size_t size)
	{
	return ZSTD_rollingHash_append(0, buf, size);
	}

	/** ZSTD_rollingHash_primePower() :
	* Compute the primePower to be passed to ZSTD_rollingHash_rotate() for a hash
	* over a window of length bytes.
	*/
	MEM_STATIC U64 ZSTD_rollingHash_primePower(U32 length)
	{
	return ZSTD_ipow(prime8bytes, length - 1);
	}

	/** ZSTD_rollingHash_rotate() :
	* Rotate the rolling hash by one byte.
	*/
	MEM_STATIC U64 ZSTD_rollingHash_rotate(U64 hash, BYTE toRemove, BYTE toAdd, U64 primePower)
	{
	hash -= (toRemove + ZSTD_ROLL_HASH_CHAR_OFFSET) * primePower;
	hash *= prime8bytes;
	hash += toAdd + ZSTD_ROLL_HASH_CHAR_OFFSET;
	return hash;
	}

	/-************************************
	* Round buffer management
	***************************************/
	#if (ZSTD_WINDOWLOG_MAX_64 > 31)
	# error "ZSTD_WINDOWLOG_MAX is too large : would overflow ZSTD_CURRENT_MAX"
	#endif
	/* Max current allowed */
	#define ZSTD_CURRENT_MAX ((3U << 29) + (1U << ZSTD_WINDOWLOG_MAX))
	/* Maximum chunk size before overflow correction needs to be called again */
	#define ZSTD_CHUNKSIZE_MAX \
	( ((U32)-1) /* Maximum ending current index */ \
	- ZSTD_CURRENT_MAX) /* Maximum beginning lowLimit */

	/**
	* ZSTD_window_clear():
	* Clears the window containing the history by simply setting it to empty.
	*/
	MEM_STATIC void ZSTD_window_clear(ZSTD_window_t* window)
	{
	size_t const endT = (size_t)(window->nextSrc - window->base);
	U32 const end = (U32)endT;

	window->lowLimit = end;
	window->dictLimit = end;
	}

	/**
	* ZSTD_window_hasExtDict():
	* Returns non-zero if the window has a non-empty extDict.
	*/
	MEM_STATIC U32 ZSTD_window_hasExtDict(ZSTD_window_t const window)
	{
	return window.lowLimit < window.dictLimit;
	}

	/**
	* ZSTD_matchState_dictMode():
	* Inspects the provided matchState and figures out what dictMode should be
	* passed to the compressor.
	*/
	MEM_STATIC ZSTD_dictMode_e ZSTD_matchState_dictMode(const ZSTD_matchState_t *ms)
	{
	return ZSTD_window_hasExtDict(ms->window) ?
	ZSTD_extDict :
	ms->dictMatchState != NULL ?
	ZSTD_dictMatchState :
	ZSTD_noDict;
	}

	/**
	* ZSTD_window_needOverflowCorrection():
	* Returns non-zero if the indices are getting too large and need overflow
	* protection.
	*/
	MEM_STATIC U32 ZSTD_window_needOverflowCorrection(ZSTD_window_t const window,
	void const* srcEnd)
	{
	U32 const current = (U32)((BYTE const*)srcEnd - window.base);
	return current > ZSTD_CURRENT_MAX;
	}

	/**
	* ZSTD_window_correctOverflow():
	* Reduces the indices to protect from index overflow.
	* Returns the correction made to the indices, which must be applied to every
	* stored index.
	*
	* The least significant cycleLog bits of the indices must remain the same,
	* which may be 0. Every index up to maxDist in the past must be valid.
	* NOTE: (maxDist & cycleMask) must be zero.
	*/
	MEM_STATIC U32 ZSTD_window_correctOverflow(ZSTD_window_t* window, U32 cycleLog,
	U32 maxDist, void const* src)
	{
	/* preemptive overflow correction:
	* 1. correction is large enough:
	* lowLimit > (3<<29) ==> current > 3<<29 + 1<<windowLog
	* 1<<windowLog <= newCurrent < 1<<chainLog + 1<<windowLog
	*
	* current - newCurrent
	* > (3<<29 + 1<<windowLog) - (1<<windowLog + 1<<chainLog)
	* > (3<<29) - (1<<chainLog)
	* > (3<<29) - (1<<30) (NOTE: chainLog <= 30)
	* > 1<<29
	*
	* 2. (ip+ZSTD_CHUNKSIZE_MAX - cctx->base) doesn't overflow:
	* After correction, current is less than (1<<chainLog + 1<<windowLog).
	* In 64-bit mode we are safe, because we have 64-bit ptrdiff_t.
	* In 32-bit mode we are safe, because (chainLog <= 29), so
	* ip+ZSTD_CHUNKSIZE_MAX - cctx->base < 1<<32.
	* 3. (cctx->lowLimit + 1<<windowLog) < 1<<32:
	* windowLog <= 31 ==> 3<<29 + 1<<windowLog < 7<<29 < 1<<32.
	*/
	U32 const cycleMask = (1U << cycleLog) - 1;
	U32 const current = (U32)((BYTE const*)src - window->base);
	U32 const currentCycle0 = current & cycleMask;
	/* Exclude zero so that newCurrent - maxDist >= 1. */
	U32 const currentCycle1 = currentCycle0 == 0 ? (1U << cycleLog) : currentCycle0;
	U32 const newCurrent = currentCycle1 + maxDist;
	U32 const correction = current - newCurrent;
	assert((maxDist & cycleMask) == 0);
	assert(current > newCurrent);
	/* Loose bound, should be around 1<<29 (see above) */
	assert(correction > 1<<28);

	window->base += correction;
	window->dictBase += correction;
	if (window->lowLimit <= correction) window->lowLimit = 1;
	else window->lowLimit -= correction;
	if (window->dictLimit <= correction) window->dictLimit = 1;
	else window->dictLimit -= correction;

	/* Ensure we can still reference the full window. */
	assert(newCurrent >= maxDist);
	assert(newCurrent - maxDist >= 1);
	/* Ensure that lowLimit and dictLimit didn't underflow. */
	assert(window->lowLimit <= newCurrent);
	assert(window->dictLimit <= newCurrent);

	DEBUGLOG(4, "Correction of 0x%x bytes to lowLimit=0x%x", correction,
	window->lowLimit);
	return correction;
	}

	/**
	* ZSTD_window_enforceMaxDist():
	* Updates lowLimit so that:
	* (srcEnd - base) - lowLimit == maxDist + loadedDictEnd
	*
	* It ensures index is valid as long as index >= lowLimit.
	* This must be called before a block compression call.
	*
	* loadedDictEnd is only defined if a dictionary is in use for current compression.
	* As the name implies, loadedDictEnd represents the index at end of dictionary.
	* The value lies within context's referential, it can be directly compared to blockEndIdx.
	*
	* If loadedDictEndPtr is NULL, no dictionary is in use, and we use loadedDictEnd == 0.
	* If loadedDictEndPtr is not NULL, we set it to zero after updating lowLimit.
	* This is because dictionaries are allowed to be referenced fully
	* as long as the last byte of the dictionary is in the window.
	* Once input has progressed beyond window size, dictionary cannot be referenced anymore.
	*
	* In normal dict mode, the dictionary lies between lowLimit and dictLimit.
	* In dictMatchState mode, lowLimit and dictLimit are the same,
	* and the dictionary is below them.
	* forceWindow and dictMatchState are therefore incompatible.
	*/
	MEM_STATIC void
	ZSTD_window_enforceMaxDist(ZSTD_window_t* window,
	const void* blockEnd,
	U32 maxDist,
	U32* loadedDictEndPtr,
	const ZSTD_matchState_t** dictMatchStatePtr)
	{
	U32 const blockEndIdx = (U32)((BYTE const*)blockEnd - window->base);
	U32 const loadedDictEnd = (loadedDictEndPtr != NULL) ? *loadedDictEndPtr : 0;
	DEBUGLOG(5, "ZSTD_window_enforceMaxDist: blockEndIdx=%u, maxDist=%u, loadedDictEnd=%u",
	(unsigned)blockEndIdx, (unsigned)maxDist, (unsigned)loadedDictEnd);

	/* - When there is no dictionary : loadedDictEnd == 0.
	In which case, the test (blockEndIdx > maxDist) is merely to avoid
	overflowing next operation `newLowLimit = blockEndIdx - maxDist`.
	- When there is a standard dictionary :
	Index referential is copied from the dictionary,
	which means it starts from 0.
	In which case, loadedDictEnd == dictSize,
	and it makes sense to compare `blockEndIdx > maxDist + dictSize`
	since `blockEndIdx` also starts from zero.
	- When there is an attached dictionary :
	loadedDictEnd is expressed within the referential of the context,
	so it can be directly compared against blockEndIdx.
	*/
	if (blockEndIdx > maxDist + loadedDictEnd) {
	U32 const newLowLimit = blockEndIdx - maxDist;
	if (window->lowLimit < newLowLimit) window->lowLimit = newLowLimit;
	if (window->dictLimit < window->lowLimit) {
	DEBUGLOG(5, "Update dictLimit to match lowLimit, from %u to %u",
	(unsigned)window->dictLimit, (unsigned)window->lowLimit);
	window->dictLimit = window->lowLimit;
	}
	/* On reaching window size, dictionaries are invalidated */
	if (loadedDictEndPtr) *loadedDictEndPtr = 0;
	if (dictMatchStatePtr) *dictMatchStatePtr = NULL;
	}
	}

	/* Similar to ZSTD_window_enforceMaxDist(),
	* but only invalidates dictionary
	* when input progresses beyond window size.
	* assumption : loadedDictEndPtr and dictMatchStatePtr are valid (non NULL)
	* loadedDictEnd uses same referential as window->base
	* maxDist is the window size */
	MEM_STATIC void
	ZSTD_checkDictValidity(const ZSTD_window_t* window,
	const void* blockEnd,
	U32 maxDist,
	U32* loadedDictEndPtr,
	const ZSTD_matchState_t** dictMatchStatePtr)
	{
	assert(loadedDictEndPtr != NULL);
	assert(dictMatchStatePtr != NULL);
	{ U32 const blockEndIdx = (U32)((BYTE const*)blockEnd - window->base);
	U32 const loadedDictEnd = *loadedDictEndPtr;
	DEBUGLOG(5, "ZSTD_checkDictValidity: blockEndIdx=%u, maxDist=%u, loadedDictEnd=%u",
	(unsigned)blockEndIdx, (unsigned)maxDist, (unsigned)loadedDictEnd);
	assert(blockEndIdx >= loadedDictEnd);

	if (blockEndIdx > loadedDictEnd + maxDist) {
	/* On reaching window size, dictionaries are invalidated.
	* For simplification, if window size is reached anywhere within next block,
	* the dictionary is invalidated for the full block.
	*/
	DEBUGLOG(6, "invalidating dictionary for current block (distance > windowSize)");
	*loadedDictEndPtr = 0;
	*dictMatchStatePtr = NULL;
	} else {
	if (*loadedDictEndPtr != 0) {
	DEBUGLOG(6, "dictionary considered valid for current block");
	} } }
	}

	MEM_STATIC void ZSTD_window_init(ZSTD_window_t* window) {
	memset(window, 0, sizeof(*window));
	window->base = (BYTE const*)"";
	window->dictBase = (BYTE const*)"";
	window->dictLimit = 1; /* start from 1, so that 1st position is valid */
	window->lowLimit = 1; /* it ensures first and later CCtx usages compress the same */
	window->nextSrc = window->base + 1; /* see issue #1241 */
	}

	/**
	* ZSTD_window_update():
	* Updates the window by appending [src, src + srcSize) to the window.
	* If it is not contiguous, the current prefix becomes the extDict, and we
	* forget about the extDict. Handles overlap of the prefix and extDict.
	* Returns non-zero if the segment is contiguous.
	*/
	MEM_STATIC U32 ZSTD_window_update(ZSTD_window_t* window,
	void const* src, size_t srcSize)
	{
	BYTE const* const ip = (BYTE const*)src;
	U32 contiguous = 1;
	DEBUGLOG(5, "ZSTD_window_update");
	if (srcSize == 0)
	return contiguous;
	assert(window->base != NULL);
	assert(window->dictBase != NULL);
	/* Check if blocks follow each other */
	if (src != window->nextSrc) {
	/* not contiguous */
	size_t const distanceFromBase = (size_t)(window->nextSrc - window->base);
	DEBUGLOG(5, "Non contiguous blocks, new segment starts at %u", window->dictLimit);
	window->lowLimit = window->dictLimit;
	assert(distanceFromBase == (size_t)(U32)distanceFromBase); /* should never overflow */
	window->dictLimit = (U32)distanceFromBase;
	window->dictBase = window->base;
	window->base = ip - distanceFromBase;
	/* ms->nextToUpdate = window->dictLimit; */
	if (window->dictLimit - window->lowLimit < HASH_READ_SIZE) window->lowLimit = window->dictLimit; /* too small extDict */
	contiguous = 0;
	}
	window->nextSrc = ip + srcSize;
	/* if input and dictionary overlap : reduce dictionary (area presumed modified by input) */
	if ( (ip+srcSize > window->dictBase + window->lowLimit)
	& (ip < window->dictBase + window->dictLimit)) {
	ptrdiff_t const highInputIdx = (ip + srcSize) - window->dictBase;
	U32 const lowLimitMax = (highInputIdx > (ptrdiff_t)window->dictLimit) ? window->dictLimit : (U32)highInputIdx;
	window->lowLimit = lowLimitMax;
	DEBUGLOG(5, "Overlapping extDict and input : new lowLimit = %u", window->lowLimit);
	}
	return contiguous;
	}

	/**
	* Returns the lowest allowed match index. It may either be in the ext-dict or the prefix.
	*/
	MEM_STATIC U32 ZSTD_getLowestMatchIndex(const ZSTD_matchState_t* ms, U32 current, unsigned windowLog)
	{
	U32 const maxDistance = 1U << windowLog;
	U32 const lowestValid = ms->window.lowLimit;
	U32 const withinWindow = (current - lowestValid > maxDistance) ? current - maxDistance : lowestValid;
	U32 const isDictionary = (ms->loadedDictEnd != 0);
	U32 const matchLowest = isDictionary ? lowestValid : withinWindow;
	return matchLowest;
	}

	/**
	* Returns the lowest allowed match index in the prefix.
	*/
	MEM_STATIC U32 ZSTD_getLowestPrefixIndex(const ZSTD_matchState_t* ms, U32 current, unsigned windowLog)
	{
	U32 const maxDistance = 1U << windowLog;
	U32 const lowestValid = ms->window.dictLimit;
	U32 const withinWindow = (current - lowestValid > maxDistance) ? current - maxDistance : lowestValid;
	U32 const isDictionary = (ms->loadedDictEnd != 0);
	U32 const matchLowest = isDictionary ? lowestValid : withinWindow;
	return matchLowest;
	}



	/* debug functions */
	#if (DEBUGLEVEL>=2)

	MEM_STATIC double ZSTD_fWeight(U32 rawStat)
	{
	U32 const fp_accuracy = 8;
	U32 const fp_multiplier = (1 << fp_accuracy);
	U32 const newStat = rawStat + 1;
	U32 const hb = ZSTD_highbit32(newStat);
	U32 const BWeight = hb * fp_multiplier;
	U32 const FWeight = (newStat << fp_accuracy) >> hb;
	U32 const weight = BWeight + FWeight;
	assert(hb + fp_accuracy < 31);
	return (double)weight / fp_multiplier;
	}

	/* display a table content,
	* listing each element, its frequency, and its predicted bit cost */
	MEM_STATIC void ZSTD_debugTable(const U32* table, U32 max)
	{
	unsigned u, sum;
	for (u=0, sum=0; u<=max; u++) sum += table[u];
	DEBUGLOG(2, "total nb elts: %u", sum);
	for (u=0; u<=max; u++) {
	DEBUGLOG(2, "%2u: %5u (%.2f)",
	u, table[u], ZSTD_fWeight(sum) - ZSTD_fWeight(table[u]) );
	}
	}

	#endif


	#if defined (__cplusplus)
	}
	#endif

	/* ===============================================================
	* Shared internal declarations
	* These prototypes may be called from sources not in lib/compress
	* =============================================================== */

	/* ZSTD_loadCEntropy() :
	* dict : must point at beginning of a valid zstd dictionary.
	* return : size of dictionary header (size of magic number + dict ID + entropy tables)
	* assumptions : magic number supposed already checked
	* and dictSize >= 8 */
	size_t ZSTD_loadCEntropy(ZSTD_compressedBlockState_t* bs, void* workspace,
	short* offcodeNCount, unsigned* offcodeMaxValue,
	const void* const dict, size_t dictSize);

	void ZSTD_reset_compressedBlockState(ZSTD_compressedBlockState_t* bs);

	/* ==============================================================
	* Private declarations
	* These prototypes shall only be called from within lib/compress
	* ============================================================== */

	/* ZSTD_getCParamsFromCCtxParams() :
	* cParams are built depending on compressionLevel, src size hints,
	* LDM and manually set compression parameters.
	* Note: srcSizeHint == 0 means 0!
	*/
	ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams(
	const ZSTD_CCtx_params* CCtxParams, U64 srcSizeHint, size_t dictSize);

	/*! ZSTD_initCStream_internal() :
	* Private use only. Init streaming operation.
	* expects params to be valid.
	* must receive dict, or cdict, or none, but not both.
	* @return : 0, or an error code */
	size_t ZSTD_initCStream_internal(ZSTD_CStream* zcs,
	const void* dict, size_t dictSize,
	const ZSTD_CDict* cdict,
	const ZSTD_CCtx_params* params, unsigned long long pledgedSrcSize);

	void ZSTD_resetSeqStore(seqStore_t* ssPtr);

	/*! ZSTD_getCParamsFromCDict() :
	* as the name implies */
	ZSTD_compressionParameters ZSTD_getCParamsFromCDict(const ZSTD_CDict* cdict);

	/* ZSTD_compressBegin_advanced_internal() :
	* Private use only. To be called from zstdmt_compress.c. */
	size_t ZSTD_compressBegin_advanced_internal(ZSTD_CCtx* cctx,
	const void* dict, size_t dictSize,
	ZSTD_dictContentType_e dictContentType,
	ZSTD_dictTableLoadMethod_e dtlm,
	const ZSTD_CDict* cdict,
	const ZSTD_CCtx_params* params,
	unsigned long long pledgedSrcSize);

	/* ZSTD_compress_advanced_internal() :
	* Private use only. To be called from zstdmt_compress.c. */
	size_t ZSTD_compress_advanced_internal(ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict,size_t dictSize,
	const ZSTD_CCtx_params* params);


	/* ZSTD_writeLastEmptyBlock() :
	* output an empty Block with end-of-frame mark to complete a frame
	* @return : size of data written into `dst` (== ZSTD_blockHeaderSize (defined in zstd_internal.h))
	* or an error code if `dstCapacity` is too small (<ZSTD_blockHeaderSize)
	*/
	size_t ZSTD_writeLastEmptyBlock(void* dst, size_t dstCapacity);


	/* ZSTD_referenceExternalSequences() :
	* Must be called before starting a compression operation.
	* seqs must parse a prefix of the source.
	* This cannot be used when long range matching is enabled.
	* Zstd will use these sequences, and pass the literals to a secondary block
	* compressor.
	* @return : An error code on failure.
	* NOTE: seqs are not verified! Invalid sequences can cause out-of-bounds memory
	* access and data corruption.
	*/
	size_t ZSTD_referenceExternalSequences(ZSTD_CCtx* cctx, rawSeq* seq, size_t nbSeq);

	/** ZSTD_cycleLog() :
	* condition for correct operation : hashLog > 1 */
	U32 ZSTD_cycleLog(U32 hashLog, ZSTD_strategy strat);

	#endif /* ZSTD_COMPRESS_H */
	/** ended inlining zstd_compress_internal.h **/


	size_t ZSTD_noCompressLiterals (void* dst, size_t dstCapacity, const void* src, size_t srcSize);

	size_t ZSTD_compressRleLiteralsBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize);

	size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
	ZSTD_hufCTables_t* nextHuf,
	ZSTD_strategy strategy, int disableLiteralCompression,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	void* entropyWorkspace, size_t entropyWorkspaceSize,
	const int bmi2);

	#endif /* ZSTD_COMPRESS_LITERALS_H */
	/** ended inlining zstd_compress_literals.h **/

	size_t ZSTD_noCompressLiterals (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
	{
	BYTE* const ostart = (BYTE* const)dst;
	U32 const flSize = 1 + (srcSize>31) + (srcSize>4095);

	RETURN_ERROR_IF(srcSize + flSize > dstCapacity, dstSize_tooSmall, "");

	switch(flSize)
	{
	case 1: /* 2 - 1 - 5 */
	ostart[0] = (BYTE)((U32)set_basic + (srcSize<<3));
	break;
	case 2: /* 2 - 2 - 12 */
	MEM_writeLE16(ostart, (U16)((U32)set_basic + (1<<2) + (srcSize<<4)));
	break;
	case 3: /* 2 - 2 - 20 */
	MEM_writeLE32(ostart, (U32)((U32)set_basic + (3<<2) + (srcSize<<4)));
	break;
	default: /* not necessary : flSize is {1,2,3} */
	assert(0);
	}

	memcpy(ostart + flSize, src, srcSize);
	DEBUGLOG(5, "Raw literals: %u -> %u", (U32)srcSize, (U32)(srcSize + flSize));
	return srcSize + flSize;
	}

	size_t ZSTD_compressRleLiteralsBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
	{
	BYTE* const ostart = (BYTE* const)dst;
	U32 const flSize = 1 + (srcSize>31) + (srcSize>4095);

	(void)dstCapacity; /* dstCapacity already guaranteed to be >=4, hence large enough */

	switch(flSize)
	{
	case 1: /* 2 - 1 - 5 */
	ostart[0] = (BYTE)((U32)set_rle + (srcSize<<3));
	break;
	case 2: /* 2 - 2 - 12 */
	MEM_writeLE16(ostart, (U16)((U32)set_rle + (1<<2) + (srcSize<<4)));
	break;
	case 3: /* 2 - 2 - 20 */
	MEM_writeLE32(ostart, (U32)((U32)set_rle + (3<<2) + (srcSize<<4)));
	break;
	default: /* not necessary : flSize is {1,2,3} */
	assert(0);
	}

	ostart[flSize] = (const BYTE)src;
	DEBUGLOG(5, "RLE literals: %u -> %u", (U32)srcSize, (U32)flSize + 1);
	return flSize+1;
	}

	size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
	ZSTD_hufCTables_t* nextHuf,
	ZSTD_strategy strategy, int disableLiteralCompression,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	void* entropyWorkspace, size_t entropyWorkspaceSize,
	const int bmi2)
	{
	size_t const minGain = ZSTD_minGain(srcSize, strategy);
	size_t const lhSize = 3 + (srcSize >= 1 KB) + (srcSize >= 16 KB);
	BYTE* const ostart = (BYTE*)dst;
	U32 singleStream = srcSize < 256;
	symbolEncodingType_e hType = set_compressed;
	size_t cLitSize;

	DEBUGLOG(5,"ZSTD_compressLiterals (disableLiteralCompression=%i srcSize=%u)",
	disableLiteralCompression, (U32)srcSize);

	/* Prepare nextEntropy assuming reusing the existing table */
	memcpy(nextHuf, prevHuf, sizeof(*prevHuf));

	if (disableLiteralCompression)
	return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);

	/* small ? don't even attempt compression (speed opt) */
	# define COMPRESS_LITERALS_SIZE_MIN 63
	{ size_t const minLitSize = (prevHuf->repeatMode == HUF_repeat_valid) ? 6 : COMPRESS_LITERALS_SIZE_MIN;
	if (srcSize <= minLitSize) return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
	}

	RETURN_ERROR_IF(dstCapacity < lhSize+1, dstSize_tooSmall, "not enough space for compression");
	{ HUF_repeat repeat = prevHuf->repeatMode;
	int const preferRepeat = strategy < ZSTD_lazy ? srcSize <= 1024 : 0;
	if (repeat == HUF_repeat_valid && lhSize == 3) singleStream = 1;
	cLitSize = singleStream ?
	HUF_compress1X_repeat(
	ostart+lhSize, dstCapacity-lhSize, src, srcSize,
	HUF_SYMBOLVALUE_MAX, HUF_TABLELOG_DEFAULT, entropyWorkspace, entropyWorkspaceSize,
	(HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2) :
	HUF_compress4X_repeat(
	ostart+lhSize, dstCapacity-lhSize, src, srcSize,
	HUF_SYMBOLVALUE_MAX, HUF_TABLELOG_DEFAULT, entropyWorkspace, entropyWorkspaceSize,
	(HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2);
	if (repeat != HUF_repeat_none) {
	/* reused the existing table */
	DEBUGLOG(5, "Reusing previous huffman table");
	hType = set_repeat;
	}
	}

	if ((cLitSize==0) \| (cLitSize >= srcSize - minGain) \| ERR_isError(cLitSize)) {
	memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
	return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
	}
	if (cLitSize==1) {
	memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
	return ZSTD_compressRleLiteralsBlock(dst, dstCapacity, src, srcSize);
	}

	if (hType == set_compressed) {
	/* using a newly constructed table */
	nextHuf->repeatMode = HUF_repeat_check;
	}

	/* Build header */
	switch(lhSize)
	{
	case 3: /* 2 - 2 - 10 - 10 */
	{ U32 const lhc = hType + ((!singleStream) << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<14);
	MEM_writeLE24(ostart, lhc);
	break;
	}
	case 4: /* 2 - 2 - 14 - 14 */
	{ U32 const lhc = hType + (2 << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<18);
	MEM_writeLE32(ostart, lhc);
	break;
	}
	case 5: /* 2 - 2 - 18 - 18 */
	{ U32 const lhc = hType + (3 << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<22);
	MEM_writeLE32(ostart, lhc);
	ostart[4] = (BYTE)(cLitSize >> 10);
	break;
	}
	default: /* not possible : lhSize is {3,4,5} */
	assert(0);
	}
	DEBUGLOG(5, "Compressed literals: %u -> %u", (U32)srcSize, (U32)(lhSize+cLitSize));
	return lhSize+cLitSize;
	}
	/** ended inlining compress/zstd_compress_literals.c **/
	/** start inlining compress/zstd_compress_sequences.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/-************************************
	* Dependencies
	***************************************/
	/** start inlining zstd_compress_sequences.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_COMPRESS_SEQUENCES_H
	#define ZSTD_COMPRESS_SEQUENCES_H

	/** skipping file: ../common/fse.h **/
	/** skipping file: ../common/zstd_internal.h **/

	typedef enum {
	ZSTD_defaultDisallowed = 0,
	ZSTD_defaultAllowed = 1
	} ZSTD_defaultPolicy_e;

	symbolEncodingType_e
	ZSTD_selectEncodingType(
	FSE_repeat* repeatMode, unsigned const* count, unsigned const max,
	size_t const mostFrequent, size_t nbSeq, unsigned const FSELog,
	FSE_CTable const* prevCTable,
	short const* defaultNorm, U32 defaultNormLog,
	ZSTD_defaultPolicy_e const isDefaultAllowed,
	ZSTD_strategy const strategy);

	size_t
	ZSTD_buildCTable(void* dst, size_t dstCapacity,
	FSE_CTable* nextCTable, U32 FSELog, symbolEncodingType_e type,
	unsigned* count, U32 max,
	const BYTE* codeTable, size_t nbSeq,
	const S16* defaultNorm, U32 defaultNormLog, U32 defaultMax,
	const FSE_CTable* prevCTable, size_t prevCTableSize,
	void* entropyWorkspace, size_t entropyWorkspaceSize);

	size_t ZSTD_encodeSequences(
	void* dst, size_t dstCapacity,
	FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
	FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
	FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
	seqDef const* sequences, size_t nbSeq, int longOffsets, int bmi2);

	size_t ZSTD_fseBitCost(
	FSE_CTable const* ctable,
	unsigned const* count,
	unsigned const max);

	size_t ZSTD_crossEntropyCost(short const* norm, unsigned accuracyLog,
	unsigned const* count, unsigned const max);
	#endif /* ZSTD_COMPRESS_SEQUENCES_H */
	/** ended inlining zstd_compress_sequences.h **/

	/**
	* -log2(x / 256) lookup table for x in [0, 256).
	* If x == 0: Return 0
	* Else: Return floor(-log2(x / 256) * 256)
	*/
	static unsigned const kInverseProbabilityLog256[256] = {
	0, 2048, 1792, 1642, 1536, 1453, 1386, 1329, 1280, 1236, 1197, 1162,
	1130, 1100, 1073, 1047, 1024, 1001, 980, 960, 941, 923, 906, 889,
	874, 859, 844, 830, 817, 804, 791, 779, 768, 756, 745, 734,
	724, 714, 704, 694, 685, 676, 667, 658, 650, 642, 633, 626,
	618, 610, 603, 595, 588, 581, 574, 567, 561, 554, 548, 542,
	535, 529, 523, 517, 512, 506, 500, 495, 489, 484, 478, 473,
	468, 463, 458, 453, 448, 443, 438, 434, 429, 424, 420, 415,
	411, 407, 402, 398, 394, 390, 386, 382, 377, 373, 370, 366,
	362, 358, 354, 350, 347, 343, 339, 336, 332, 329, 325, 322,
	318, 315, 311, 308, 305, 302, 298, 295, 292, 289, 286, 282,
	279, 276, 273, 270, 267, 264, 261, 258, 256, 253, 250, 247,
	244, 241, 239, 236, 233, 230, 228, 225, 222, 220, 217, 215,
	212, 209, 207, 204, 202, 199, 197, 194, 192, 190, 187, 185,
	182, 180, 178, 175, 173, 171, 168, 166, 164, 162, 159, 157,
	155, 153, 151, 149, 146, 144, 142, 140, 138, 136, 134, 132,
	130, 128, 126, 123, 121, 119, 117, 115, 114, 112, 110, 108,
	106, 104, 102, 100, 98, 96, 94, 93, 91, 89, 87, 85,
	83, 82, 80, 78, 76, 74, 73, 71, 69, 67, 66, 64,
	62, 61, 59, 57, 55, 54, 52, 50, 49, 47, 46, 44,
	42, 41, 39, 37, 36, 34, 33, 31, 30, 28, 26, 25,
	23, 22, 20, 19, 17, 16, 14, 13, 11, 10, 8, 7,
	5, 4, 2, 1,
	};

	static unsigned ZSTD_getFSEMaxSymbolValue(FSE_CTable const* ctable) {
	void const* ptr = ctable;
	U16 const* u16ptr = (U16 const*)ptr;
	U32 const maxSymbolValue = MEM_read16(u16ptr + 1);
	return maxSymbolValue;
	}

	/**
	* Returns the cost in bytes of encoding the normalized count header.
	* Returns an error if any of the helper functions return an error.
	*/
	static size_t ZSTD_NCountCost(unsigned const* count, unsigned const max,
	size_t const nbSeq, unsigned const FSELog)
	{
	BYTE wksp[FSE_NCOUNTBOUND];
	S16 norm[MaxSeq + 1];
	const U32 tableLog = FSE_optimalTableLog(FSELog, nbSeq, max);
	FORWARD_IF_ERROR(FSE_normalizeCount(norm, tableLog, count, nbSeq, max), "");
	return FSE_writeNCount(wksp, sizeof(wksp), norm, max, tableLog);
	}

	/**
	* Returns the cost in bits of encoding the distribution described by count
	* using the entropy bound.
	*/
	static size_t ZSTD_entropyCost(unsigned const* count, unsigned const max, size_t const total)
	{
	unsigned cost = 0;
	unsigned s;
	for (s = 0; s <= max; ++s) {
	unsigned norm = (unsigned)((256 * count[s]) / total);
	if (count[s] != 0 && norm == 0)
	norm = 1;
	assert(count[s] < total);
	cost += count[s] * kInverseProbabilityLog256[norm];
	}
	return cost >> 8;
	}

	/**
	* Returns the cost in bits of encoding the distribution in count using ctable.
	* Returns an error if ctable cannot represent all the symbols in count.
	*/
	size_t ZSTD_fseBitCost(
	FSE_CTable const* ctable,
	unsigned const* count,
	unsigned const max)
	{
	unsigned const kAccuracyLog = 8;
	size_t cost = 0;
	unsigned s;
	FSE_CState_t cstate;
	FSE_initCState(&cstate, ctable);
	if (ZSTD_getFSEMaxSymbolValue(ctable) < max) {
	DEBUGLOG(5, "Repeat FSE_CTable has maxSymbolValue %u < %u",
	ZSTD_getFSEMaxSymbolValue(ctable), max);
	return ERROR(GENERIC);
	}
	for (s = 0; s <= max; ++s) {
	unsigned const tableLog = cstate.stateLog;
	unsigned const badCost = (tableLog + 1) << kAccuracyLog;
	unsigned const bitCost = FSE_bitCost(cstate.symbolTT, tableLog, s, kAccuracyLog);
	if (count[s] == 0)
	continue;
	if (bitCost >= badCost) {
	DEBUGLOG(5, "Repeat FSE_CTable has Prob[%u] == 0", s);
	return ERROR(GENERIC);
	}
	cost += (size_t)count[s] * bitCost;
	}
	return cost >> kAccuracyLog;
	}

	/**
	* Returns the cost in bits of encoding the distribution in count using the
	* table described by norm. The max symbol support by norm is assumed >= max.
	* norm must be valid for every symbol with non-zero probability in count.
	*/
	size_t ZSTD_crossEntropyCost(short const* norm, unsigned accuracyLog,
	unsigned const* count, unsigned const max)
	{
	unsigned const shift = 8 - accuracyLog;
	size_t cost = 0;
	unsigned s;
	assert(accuracyLog <= 8);
	for (s = 0; s <= max; ++s) {
	unsigned const normAcc = (norm[s] != -1) ? (unsigned)norm[s] : 1;
	unsigned const norm256 = normAcc << shift;
	assert(norm256 > 0);
	assert(norm256 < 256);
	cost += count[s] * kInverseProbabilityLog256[norm256];
	}
	return cost >> 8;
	}

	symbolEncodingType_e
	ZSTD_selectEncodingType(
	FSE_repeat* repeatMode, unsigned const* count, unsigned const max,
	size_t const mostFrequent, size_t nbSeq, unsigned const FSELog,
	FSE_CTable const* prevCTable,
	short const* defaultNorm, U32 defaultNormLog,
	ZSTD_defaultPolicy_e const isDefaultAllowed,
	ZSTD_strategy const strategy)
	{
	ZSTD_STATIC_ASSERT(ZSTD_defaultDisallowed == 0 && ZSTD_defaultAllowed != 0);
	if (mostFrequent == nbSeq) {
	*repeatMode = FSE_repeat_none;
	if (isDefaultAllowed && nbSeq <= 2) {
	/* Prefer set_basic over set_rle when there are 2 or less symbols,
	* since RLE uses 1 byte, but set_basic uses 5-6 bits per symbol.
	* If basic encoding isn't possible, always choose RLE.
	*/
	DEBUGLOG(5, "Selected set_basic");
	return set_basic;
	}
	DEBUGLOG(5, "Selected set_rle");
	return set_rle;
	}
	if (strategy < ZSTD_lazy) {
	if (isDefaultAllowed) {
	size_t const staticFse_nbSeq_max = 1000;
	size_t const mult = 10 - strategy;
	size_t const baseLog = 3;
	size_t const dynamicFse_nbSeq_min = (((size_t)1 << defaultNormLog) * mult) >> baseLog; /* 28-36 for offset, 56-72 for lengths */
	assert(defaultNormLog >= 5 && defaultNormLog <= 6); /* xx_DEFAULTNORMLOG */
	assert(mult <= 9 && mult >= 7);
	if ( (*repeatMode == FSE_repeat_valid)
	&& (nbSeq < staticFse_nbSeq_max) ) {
	DEBUGLOG(5, "Selected set_repeat");
	return set_repeat;
	}
	if ( (nbSeq < dynamicFse_nbSeq_min)
	\|\| (mostFrequent < (nbSeq >> (defaultNormLog-1))) ) {
	DEBUGLOG(5, "Selected set_basic");
	/* The format allows default tables to be repeated, but it isn't useful.
	* When using simple heuristics to select encoding type, we don't want
	* to confuse these tables with dictionaries. When running more careful
	* analysis, we don't need to waste time checking both repeating tables
	* and default tables.
	*/
	*repeatMode = FSE_repeat_none;
	return set_basic;
	}
	}
	} else {
	size_t const basicCost = isDefaultAllowed ? ZSTD_crossEntropyCost(defaultNorm, defaultNormLog, count, max) : ERROR(GENERIC);
	size_t const repeatCost = *repeatMode != FSE_repeat_none ? ZSTD_fseBitCost(prevCTable, count, max) : ERROR(GENERIC);
	size_t const NCountCost = ZSTD_NCountCost(count, max, nbSeq, FSELog);
	size_t const compressedCost = (NCountCost << 3) + ZSTD_entropyCost(count, max, nbSeq);

	if (isDefaultAllowed) {
	assert(!ZSTD_isError(basicCost));
	assert(!(*repeatMode == FSE_repeat_valid && ZSTD_isError(repeatCost)));
	}
	assert(!ZSTD_isError(NCountCost));
	assert(compressedCost < ERROR(maxCode));
	DEBUGLOG(5, "Estimated bit costs: basic=%u\trepeat=%u\tcompressed=%u",
	(unsigned)basicCost, (unsigned)repeatCost, (unsigned)compressedCost);
	if (basicCost <= repeatCost && basicCost <= compressedCost) {
	DEBUGLOG(5, "Selected set_basic");
	assert(isDefaultAllowed);
	*repeatMode = FSE_repeat_none;
	return set_basic;
	}
	if (repeatCost <= compressedCost) {
	DEBUGLOG(5, "Selected set_repeat");
	assert(!ZSTD_isError(repeatCost));
	return set_repeat;
	}
	assert(compressedCost < basicCost && compressedCost < repeatCost);
	}
	DEBUGLOG(5, "Selected set_compressed");
	*repeatMode = FSE_repeat_check;
	return set_compressed;
	}

	size_t
	ZSTD_buildCTable(void* dst, size_t dstCapacity,
	FSE_CTable* nextCTable, U32 FSELog, symbolEncodingType_e type,
	unsigned* count, U32 max,
	const BYTE* codeTable, size_t nbSeq,
	const S16* defaultNorm, U32 defaultNormLog, U32 defaultMax,
	const FSE_CTable* prevCTable, size_t prevCTableSize,
	void* entropyWorkspace, size_t entropyWorkspaceSize)
	{
	BYTE* op = (BYTE*)dst;
	const BYTE* const oend = op + dstCapacity;
	DEBUGLOG(6, "ZSTD_buildCTable (dstCapacity=%u)", (unsigned)dstCapacity);

	switch (type) {
	case set_rle:
	FORWARD_IF_ERROR(FSE_buildCTable_rle(nextCTable, (BYTE)max), "");
	RETURN_ERROR_IF(dstCapacity==0, dstSize_tooSmall, "not enough space");
	*op = codeTable[0];
	return 1;
	case set_repeat:
	memcpy(nextCTable, prevCTable, prevCTableSize);
	return 0;
	case set_basic:
	FORWARD_IF_ERROR(FSE_buildCTable_wksp(nextCTable, defaultNorm, defaultMax, defaultNormLog, entropyWorkspace, entropyWorkspaceSize), ""); /* note : could be pre-calculated */
	return 0;
	case set_compressed: {
	S16 norm[MaxSeq + 1];
	size_t nbSeq_1 = nbSeq;
	const U32 tableLog = FSE_optimalTableLog(FSELog, nbSeq, max);
	if (count[codeTable[nbSeq-1]] > 1) {
	count[codeTable[nbSeq-1]]--;
	nbSeq_1--;
	}
	assert(nbSeq_1 > 1);
	FORWARD_IF_ERROR(FSE_normalizeCount(norm, tableLog, count, nbSeq_1, max), "");
	{ size_t const NCountSize = FSE_writeNCount(op, oend - op, norm, max, tableLog); /* overflow protected */
	FORWARD_IF_ERROR(NCountSize, "FSE_writeNCount failed");
	FORWARD_IF_ERROR(FSE_buildCTable_wksp(nextCTable, norm, max, tableLog, entropyWorkspace, entropyWorkspaceSize), "");
	return NCountSize;
	}
	}
	default: assert(0); RETURN_ERROR(GENERIC, "impossible to reach");
	}
	}

	FORCE_INLINE_TEMPLATE size_t
	ZSTD_encodeSequences_body(
	void* dst, size_t dstCapacity,
	FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
	FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
	FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
	seqDef const* sequences, size_t nbSeq, int longOffsets)
	{
	BIT_CStream_t blockStream;
	FSE_CState_t stateMatchLength;
	FSE_CState_t stateOffsetBits;
	FSE_CState_t stateLitLength;

	RETURN_ERROR_IF(
	ERR_isError(BIT_initCStream(&blockStream, dst, dstCapacity)),
	dstSize_tooSmall, "not enough space remaining");
	DEBUGLOG(6, "available space for bitstream : %i (dstCapacity=%u)",
	(int)(blockStream.endPtr - blockStream.startPtr),
	(unsigned)dstCapacity);

	/* first symbols */
	FSE_initCState2(&stateMatchLength, CTable_MatchLength, mlCodeTable[nbSeq-1]);
	FSE_initCState2(&stateOffsetBits, CTable_OffsetBits, ofCodeTable[nbSeq-1]);
	FSE_initCState2(&stateLitLength, CTable_LitLength, llCodeTable[nbSeq-1]);
	BIT_addBits(&blockStream, sequences[nbSeq-1].litLength, LL_bits[llCodeTable[nbSeq-1]]);
	if (MEM_32bits()) BIT_flushBits(&blockStream);
	BIT_addBits(&blockStream, sequences[nbSeq-1].matchLength, ML_bits[mlCodeTable[nbSeq-1]]);
	if (MEM_32bits()) BIT_flushBits(&blockStream);
	if (longOffsets) {
	U32 const ofBits = ofCodeTable[nbSeq-1];
	unsigned const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN-1);
	if (extraBits) {
	BIT_addBits(&blockStream, sequences[nbSeq-1].offset, extraBits);
	BIT_flushBits(&blockStream);
	}
	BIT_addBits(&blockStream, sequences[nbSeq-1].offset >> extraBits,
	ofBits - extraBits);
	} else {
	BIT_addBits(&blockStream, sequences[nbSeq-1].offset, ofCodeTable[nbSeq-1]);
	}
	BIT_flushBits(&blockStream);

	{ size_t n;
	for (n=nbSeq-2 ; n<nbSeq ; n--) { /* intentional underflow */
	BYTE const llCode = llCodeTable[n];
	BYTE const ofCode = ofCodeTable[n];
	BYTE const mlCode = mlCodeTable[n];
	U32 const llBits = LL_bits[llCode];
	U32 const ofBits = ofCode;
	U32 const mlBits = ML_bits[mlCode];
	DEBUGLOG(6, "encoding: litlen:%2u - matchlen:%2u - offCode:%7u",
	(unsigned)sequences[n].litLength,
	(unsigned)sequences[n].matchLength + MINMATCH,
	(unsigned)sequences[n].offset);
	/* 32b/ / 64b*/
	/* (7)/ / (7)*/
	FSE_encodeSymbol(&blockStream, &stateOffsetBits, ofCode); /* 15 / / 15 */
	FSE_encodeSymbol(&blockStream, &stateMatchLength, mlCode); /* 24 / / 24 */
	if (MEM_32bits()) BIT_flushBits(&blockStream); /* (7)*/
	FSE_encodeSymbol(&blockStream, &stateLitLength, llCode); /* 16 / / 33 */
	if (MEM_32bits() \|\| (ofBits+mlBits+llBits >= 64-7-(LLFSELog+MLFSELog+OffFSELog)))
	BIT_flushBits(&blockStream); /* (7)*/
	BIT_addBits(&blockStream, sequences[n].litLength, llBits);
	if (MEM_32bits() && ((llBits+mlBits)>24)) BIT_flushBits(&blockStream);
	BIT_addBits(&blockStream, sequences[n].matchLength, mlBits);
	if (MEM_32bits() \|\| (ofBits+mlBits+llBits > 56)) BIT_flushBits(&blockStream);
	if (longOffsets) {
	unsigned const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN-1);
	if (extraBits) {
	BIT_addBits(&blockStream, sequences[n].offset, extraBits);
	BIT_flushBits(&blockStream); /* (7)*/
	}
	BIT_addBits(&blockStream, sequences[n].offset >> extraBits,
	ofBits - extraBits); /* 31 */
	} else {
	BIT_addBits(&blockStream, sequences[n].offset, ofBits); /* 31 */
	}
	BIT_flushBits(&blockStream); /* (7)*/
	DEBUGLOG(7, "remaining space : %i", (int)(blockStream.endPtr - blockStream.ptr));
	} }

	DEBUGLOG(6, "ZSTD_encodeSequences: flushing ML state with %u bits", stateMatchLength.stateLog);
	FSE_flushCState(&blockStream, &stateMatchLength);
	DEBUGLOG(6, "ZSTD_encodeSequences: flushing Off state with %u bits", stateOffsetBits.stateLog);
	FSE_flushCState(&blockStream, &stateOffsetBits);
	DEBUGLOG(6, "ZSTD_encodeSequences: flushing LL state with %u bits", stateLitLength.stateLog);
	FSE_flushCState(&blockStream, &stateLitLength);

	{ size_t const streamSize = BIT_closeCStream(&blockStream);
	RETURN_ERROR_IF(streamSize==0, dstSize_tooSmall, "not enough space");
	return streamSize;
	}
	}

	static size_t
	ZSTD_encodeSequences_default(
	void* dst, size_t dstCapacity,
	FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
	FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
	FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
	seqDef const* sequences, size_t nbSeq, int longOffsets)
	{
	return ZSTD_encodeSequences_body(dst, dstCapacity,
	CTable_MatchLength, mlCodeTable,
	CTable_OffsetBits, ofCodeTable,
	CTable_LitLength, llCodeTable,
	sequences, nbSeq, longOffsets);
	}


	#if DYNAMIC_BMI2

	static TARGET_ATTRIBUTE("bmi2") size_t
	ZSTD_encodeSequences_bmi2(
	void* dst, size_t dstCapacity,
	FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
	FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
	FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
	seqDef const* sequences, size_t nbSeq, int longOffsets)
	{
	return ZSTD_encodeSequences_body(dst, dstCapacity,
	CTable_MatchLength, mlCodeTable,
	CTable_OffsetBits, ofCodeTable,
	CTable_LitLength, llCodeTable,
	sequences, nbSeq, longOffsets);
	}

	#endif

	size_t ZSTD_encodeSequences(
	void* dst, size_t dstCapacity,
	FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
	FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
	FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
	seqDef const* sequences, size_t nbSeq, int longOffsets, int bmi2)
	{
	DEBUGLOG(5, "ZSTD_encodeSequences: dstCapacity = %u", (unsigned)dstCapacity);
	#if DYNAMIC_BMI2
	if (bmi2) {
	return ZSTD_encodeSequences_bmi2(dst, dstCapacity,
	CTable_MatchLength, mlCodeTable,
	CTable_OffsetBits, ofCodeTable,
	CTable_LitLength, llCodeTable,
	sequences, nbSeq, longOffsets);
	}
	#endif
	(void)bmi2;
	return ZSTD_encodeSequences_default(dst, dstCapacity,
	CTable_MatchLength, mlCodeTable,
	CTable_OffsetBits, ofCodeTable,
	CTable_LitLength, llCodeTable,
	sequences, nbSeq, longOffsets);
	}
	/** ended inlining compress/zstd_compress_sequences.c **/
	/** start inlining compress/zstd_compress_superblock.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/-************************************
	* Dependencies
	***************************************/
	/** start inlining zstd_compress_superblock.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_COMPRESS_ADVANCED_H
	#define ZSTD_COMPRESS_ADVANCED_H

	/-************************************
	* Dependencies
	***************************************/

	/** skipping file: ../zstd.h **/

	/-************************************
	* Target Compressed Block Size
	***************************************/

	/* ZSTD_compressSuperBlock() :
	* Used to compress a super block when targetCBlockSize is being used.
	* The given block will be compressed into multiple sub blocks that are around targetCBlockSize. */
	size_t ZSTD_compressSuperBlock(ZSTD_CCtx* zc,
	void* dst, size_t dstCapacity,
	void const* src, size_t srcSize,
	unsigned lastBlock);

	#endif /* ZSTD_COMPRESS_ADVANCED_H */
	/** ended inlining zstd_compress_superblock.h **/

	/** skipping file: ../common/zstd_internal.h **/
	/** skipping file: hist.h **/
	/** skipping file: zstd_compress_internal.h **/
	/** skipping file: zstd_compress_sequences.h **/
	/** skipping file: zstd_compress_literals.h **/

	/-************************************
	* Superblock entropy buffer structs
	***************************************/
	/** ZSTD_hufCTablesMetadata_t :
	* Stores Literals Block Type for a super-block in hType, and
	* huffman tree description in hufDesBuffer.
	* hufDesSize refers to the size of huffman tree description in bytes.
	* This metadata is populated in ZSTD_buildSuperBlockEntropy_literal() */
	typedef struct {
	symbolEncodingType_e hType;
	BYTE hufDesBuffer[500]; /* TODO give name to this value */
	size_t hufDesSize;
	} ZSTD_hufCTablesMetadata_t;

	/** ZSTD_fseCTablesMetadata_t :
	* Stores symbol compression modes for a super-block in {ll, ol, ml}Type, and
	* fse tables in fseTablesBuffer.
	* fseTablesSize refers to the size of fse tables in bytes.
	* This metadata is populated in ZSTD_buildSuperBlockEntropy_sequences() */
	typedef struct {
	symbolEncodingType_e llType;
	symbolEncodingType_e ofType;
	symbolEncodingType_e mlType;
	BYTE fseTablesBuffer[500]; /* TODO give name to this value */
	size_t fseTablesSize;
	size_t lastCountSize; /* This is to account for bug in 1.3.4. More detail in ZSTD_compressSubBlock_sequences() */
	} ZSTD_fseCTablesMetadata_t;

	typedef struct {
	ZSTD_hufCTablesMetadata_t hufMetadata;
	ZSTD_fseCTablesMetadata_t fseMetadata;
	} ZSTD_entropyCTablesMetadata_t;


	/** ZSTD_buildSuperBlockEntropy_literal() :
	* Builds entropy for the super-block literals.
	* Stores literals block type (raw, rle, compressed, repeat) and
	* huffman description table to hufMetadata.
	* @return : size of huffman description table or error code */
	static size_t ZSTD_buildSuperBlockEntropy_literal(void* const src, size_t srcSize,
	const ZSTD_hufCTables_t* prevHuf,
	ZSTD_hufCTables_t* nextHuf,
	ZSTD_hufCTablesMetadata_t* hufMetadata,
	const int disableLiteralsCompression,
	void* workspace, size_t wkspSize)
	{
	BYTE* const wkspStart = (BYTE*)workspace;
	BYTE* const wkspEnd = wkspStart + wkspSize;
	BYTE* const countWkspStart = wkspStart;
	unsigned* const countWksp = (unsigned*)workspace;
	const size_t countWkspSize = (HUF_SYMBOLVALUE_MAX + 1) * sizeof(unsigned);
	BYTE* const nodeWksp = countWkspStart + countWkspSize;
	const size_t nodeWkspSize = wkspEnd-nodeWksp;
	unsigned maxSymbolValue = 255;
	unsigned huffLog = HUF_TABLELOG_DEFAULT;
	HUF_repeat repeat = prevHuf->repeatMode;

	DEBUGLOG(5, "ZSTD_buildSuperBlockEntropy_literal (srcSize=%zu)", srcSize);

	/* Prepare nextEntropy assuming reusing the existing table */
	memcpy(nextHuf, prevHuf, sizeof(*prevHuf));

	if (disableLiteralsCompression) {
	DEBUGLOG(5, "set_basic - disabled");
	hufMetadata->hType = set_basic;
	return 0;
	}

	/* small ? don't even attempt compression (speed opt) */
	# define COMPRESS_LITERALS_SIZE_MIN 63
	{ size_t const minLitSize = (prevHuf->repeatMode == HUF_repeat_valid) ? 6 : COMPRESS_LITERALS_SIZE_MIN;
	if (srcSize <= minLitSize) {
	DEBUGLOG(5, "set_basic - too small");
	hufMetadata->hType = set_basic;
	return 0;
	}
	}

	/* Scan input and build symbol stats */
	{ size_t const largest = HIST_count_wksp (countWksp, &maxSymbolValue, (const BYTE*)src, srcSize, workspace, wkspSize);
	FORWARD_IF_ERROR(largest, "HIST_count_wksp failed");
	if (largest == srcSize) {
	DEBUGLOG(5, "set_rle");
	hufMetadata->hType = set_rle;
	return 0;
	}
	if (largest <= (srcSize >> 7)+4) {
	DEBUGLOG(5, "set_basic - no gain");
	hufMetadata->hType = set_basic;
	return 0;
	}
	}

	/* Validate the previous Huffman table */
	if (repeat == HUF_repeat_check && !HUF_validateCTable((HUF_CElt const*)prevHuf->CTable, countWksp, maxSymbolValue)) {
	repeat = HUF_repeat_none;
	}

	/* Build Huffman Tree */
	memset(nextHuf->CTable, 0, sizeof(nextHuf->CTable));
	huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue);
	{ size_t const maxBits = HUF_buildCTable_wksp((HUF_CElt*)nextHuf->CTable, countWksp,
	maxSymbolValue, huffLog,
	nodeWksp, nodeWkspSize);
	FORWARD_IF_ERROR(maxBits, "HUF_buildCTable_wksp");
	huffLog = (U32)maxBits;
	{ /* Build and write the CTable */
	size_t const newCSize = HUF_estimateCompressedSize(
	(HUF_CElt*)nextHuf->CTable, countWksp, maxSymbolValue);
	size_t const hSize = HUF_writeCTable(
	hufMetadata->hufDesBuffer, sizeof(hufMetadata->hufDesBuffer),
	(HUF_CElt*)nextHuf->CTable, maxSymbolValue, huffLog);
	/* Check against repeating the previous CTable */
	if (repeat != HUF_repeat_none) {
	size_t const oldCSize = HUF_estimateCompressedSize(
	(HUF_CElt const*)prevHuf->CTable, countWksp, maxSymbolValue);
	if (oldCSize < srcSize && (oldCSize <= hSize + newCSize \|\| hSize + 12 >= srcSize)) {
	DEBUGLOG(5, "set_repeat - smaller");
	memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
	hufMetadata->hType = set_repeat;
	return 0;
	}
	}
	if (newCSize + hSize >= srcSize) {
	DEBUGLOG(5, "set_basic - no gains");
	memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
	hufMetadata->hType = set_basic;
	return 0;
	}
	DEBUGLOG(5, "set_compressed (hSize=%u)", (U32)hSize);
	hufMetadata->hType = set_compressed;
	nextHuf->repeatMode = HUF_repeat_check;
	return hSize;
	}
	}
	}

	/** ZSTD_buildSuperBlockEntropy_sequences() :
	* Builds entropy for the super-block sequences.
	* Stores symbol compression modes and fse table to fseMetadata.
	* @return : size of fse tables or error code */
	static size_t ZSTD_buildSuperBlockEntropy_sequences(seqStore_t* seqStorePtr,
	const ZSTD_fseCTables_t* prevEntropy,
	ZSTD_fseCTables_t* nextEntropy,
	const ZSTD_CCtx_params* cctxParams,
	ZSTD_fseCTablesMetadata_t* fseMetadata,
	void* workspace, size_t wkspSize)
	{
	BYTE* const wkspStart = (BYTE*)workspace;
	BYTE* const wkspEnd = wkspStart + wkspSize;
	BYTE* const countWkspStart = wkspStart;
	unsigned* const countWksp = (unsigned*)workspace;
	const size_t countWkspSize = (MaxSeq + 1) * sizeof(unsigned);
	BYTE* const cTableWksp = countWkspStart + countWkspSize;
	const size_t cTableWkspSize = wkspEnd-cTableWksp;
	ZSTD_strategy const strategy = cctxParams->cParams.strategy;
	FSE_CTable* CTable_LitLength = nextEntropy->litlengthCTable;
	FSE_CTable* CTable_OffsetBits = nextEntropy->offcodeCTable;
	FSE_CTable* CTable_MatchLength = nextEntropy->matchlengthCTable;
	const BYTE* const ofCodeTable = seqStorePtr->ofCode;
	const BYTE* const llCodeTable = seqStorePtr->llCode;
	const BYTE* const mlCodeTable = seqStorePtr->mlCode;
	size_t const nbSeq = seqStorePtr->sequences - seqStorePtr->sequencesStart;
	BYTE* const ostart = fseMetadata->fseTablesBuffer;
	BYTE* const oend = ostart + sizeof(fseMetadata->fseTablesBuffer);
	BYTE* op = ostart;

	assert(cTableWkspSize >= (1 << MaxFSELog) * sizeof(FSE_FUNCTION_TYPE));
	DEBUGLOG(5, "ZSTD_buildSuperBlockEntropy_sequences (nbSeq=%zu)", nbSeq);
	memset(workspace, 0, wkspSize);

	fseMetadata->lastCountSize = 0;
	/* convert length/distances into codes */
	ZSTD_seqToCodes(seqStorePtr);
	/* build CTable for Literal Lengths */
	{ U32 LLtype;
	unsigned max = MaxLL;
	size_t const mostFrequent = HIST_countFast_wksp(countWksp, &max, llCodeTable, nbSeq, workspace, wkspSize); /* can't fail */
	DEBUGLOG(5, "Building LL table");
	nextEntropy->litlength_repeatMode = prevEntropy->litlength_repeatMode;
	LLtype = ZSTD_selectEncodingType(&nextEntropy->litlength_repeatMode,
	countWksp, max, mostFrequent, nbSeq,
	LLFSELog, prevEntropy->litlengthCTable,
	LL_defaultNorm, LL_defaultNormLog,
	ZSTD_defaultAllowed, strategy);
	assert(set_basic < set_compressed && set_rle < set_compressed);
	assert(!(LLtype < set_compressed && nextEntropy->litlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
	{ size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_LitLength, LLFSELog, (symbolEncodingType_e)LLtype,
	countWksp, max, llCodeTable, nbSeq, LL_defaultNorm, LL_defaultNormLog, MaxLL,
	prevEntropy->litlengthCTable, sizeof(prevEntropy->litlengthCTable),
	cTableWksp, cTableWkspSize);
	FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for LitLens failed");
	if (LLtype == set_compressed)
	fseMetadata->lastCountSize = countSize;
	op += countSize;
	fseMetadata->llType = (symbolEncodingType_e) LLtype;
	} }
	/* build CTable for Offsets */
	{ U32 Offtype;
	unsigned max = MaxOff;
	size_t const mostFrequent = HIST_countFast_wksp(countWksp, &max, ofCodeTable, nbSeq, workspace, wkspSize); /* can't fail */
	/* We can only use the basic table if max <= DefaultMaxOff, otherwise the offsets are too large */
	ZSTD_defaultPolicy_e const defaultPolicy = (max <= DefaultMaxOff) ? ZSTD_defaultAllowed : ZSTD_defaultDisallowed;
	DEBUGLOG(5, "Building OF table");
	nextEntropy->offcode_repeatMode = prevEntropy->offcode_repeatMode;
	Offtype = ZSTD_selectEncodingType(&nextEntropy->offcode_repeatMode,
	countWksp, max, mostFrequent, nbSeq,
	OffFSELog, prevEntropy->offcodeCTable,
	OF_defaultNorm, OF_defaultNormLog,
	defaultPolicy, strategy);
	assert(!(Offtype < set_compressed && nextEntropy->offcode_repeatMode != FSE_repeat_none)); /* We don't copy tables */
	{ size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_OffsetBits, OffFSELog, (symbolEncodingType_e)Offtype,
	countWksp, max, ofCodeTable, nbSeq, OF_defaultNorm, OF_defaultNormLog, DefaultMaxOff,
	prevEntropy->offcodeCTable, sizeof(prevEntropy->offcodeCTable),
	cTableWksp, cTableWkspSize);
	FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for Offsets failed");
	if (Offtype == set_compressed)
	fseMetadata->lastCountSize = countSize;
	op += countSize;
	fseMetadata->ofType = (symbolEncodingType_e) Offtype;
	} }
	/* build CTable for MatchLengths */
	{ U32 MLtype;
	unsigned max = MaxML;
	size_t const mostFrequent = HIST_countFast_wksp(countWksp, &max, mlCodeTable, nbSeq, workspace, wkspSize); /* can't fail */
	DEBUGLOG(5, "Building ML table (remaining space : %i)", (int)(oend-op));
	nextEntropy->matchlength_repeatMode = prevEntropy->matchlength_repeatMode;
	MLtype = ZSTD_selectEncodingType(&nextEntropy->matchlength_repeatMode,
	countWksp, max, mostFrequent, nbSeq,
	MLFSELog, prevEntropy->matchlengthCTable,
	ML_defaultNorm, ML_defaultNormLog,
	ZSTD_defaultAllowed, strategy);
	assert(!(MLtype < set_compressed && nextEntropy->matchlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
	{ size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_MatchLength, MLFSELog, (symbolEncodingType_e)MLtype,
	countWksp, max, mlCodeTable, nbSeq, ML_defaultNorm, ML_defaultNormLog, MaxML,
	prevEntropy->matchlengthCTable, sizeof(prevEntropy->matchlengthCTable),
	cTableWksp, cTableWkspSize);
	FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for MatchLengths failed");
	if (MLtype == set_compressed)
	fseMetadata->lastCountSize = countSize;
	op += countSize;
	fseMetadata->mlType = (symbolEncodingType_e) MLtype;
	} }
	assert((size_t) (op-ostart) <= sizeof(fseMetadata->fseTablesBuffer));
	return op-ostart;
	}


	/** ZSTD_buildSuperBlockEntropy() :
	* Builds entropy for the super-block.
	* @return : 0 on success or error code */
	static size_t
	ZSTD_buildSuperBlockEntropy(seqStore_t* seqStorePtr,
	const ZSTD_entropyCTables_t* prevEntropy,
	ZSTD_entropyCTables_t* nextEntropy,
	const ZSTD_CCtx_params* cctxParams,
	ZSTD_entropyCTablesMetadata_t* entropyMetadata,
	void* workspace, size_t wkspSize)
	{
	size_t const litSize = seqStorePtr->lit - seqStorePtr->litStart;
	DEBUGLOG(5, "ZSTD_buildSuperBlockEntropy");
	entropyMetadata->hufMetadata.hufDesSize =
	ZSTD_buildSuperBlockEntropy_literal(seqStorePtr->litStart, litSize,
	&prevEntropy->huf, &nextEntropy->huf,
	&entropyMetadata->hufMetadata,
	ZSTD_disableLiteralsCompression(cctxParams),
	workspace, wkspSize);
	FORWARD_IF_ERROR(entropyMetadata->hufMetadata.hufDesSize, "ZSTD_buildSuperBlockEntropy_literal failed");
	entropyMetadata->fseMetadata.fseTablesSize =
	ZSTD_buildSuperBlockEntropy_sequences(seqStorePtr,
	&prevEntropy->fse, &nextEntropy->fse,
	cctxParams,
	&entropyMetadata->fseMetadata,
	workspace, wkspSize);
	FORWARD_IF_ERROR(entropyMetadata->fseMetadata.fseTablesSize, "ZSTD_buildSuperBlockEntropy_sequences failed");
	return 0;
	}

	/** ZSTD_compressSubBlock_literal() :
	* Compresses literals section for a sub-block.
	* When we have to write the Huffman table we will sometimes choose a header
	* size larger than necessary. This is because we have to pick the header size
	* before we know the table size + compressed size, so we have a bound on the
	* table size. If we guessed incorrectly, we fall back to uncompressed literals.
	*
	* We write the header when writeEntropy=1 and set entropyWrriten=1 when we succeeded
	* in writing the header, otherwise it is set to 0.
	*
	* hufMetadata->hType has literals block type info.
	* If it is set_basic, all sub-blocks literals section will be Raw_Literals_Block.
	* If it is set_rle, all sub-blocks literals section will be RLE_Literals_Block.
	* If it is set_compressed, first sub-block's literals section will be Compressed_Literals_Block
	* If it is set_compressed, first sub-block's literals section will be Treeless_Literals_Block
	* and the following sub-blocks' literals sections will be Treeless_Literals_Block.
	* @return : compressed size of literals section of a sub-block
	* Or 0 if it unable to compress.
	* Or error code */
	static size_t ZSTD_compressSubBlock_literal(const HUF_CElt* hufTable,
	const ZSTD_hufCTablesMetadata_t* hufMetadata,
	const BYTE* literals, size_t litSize,
	void* dst, size_t dstSize,
	const int bmi2, int writeEntropy, int* entropyWritten)
	{
	size_t const header = writeEntropy ? 200 : 0;
	size_t const lhSize = 3 + (litSize >= (1 KB - header)) + (litSize >= (16 KB - header));
	BYTE* const ostart = (BYTE*)dst;
	BYTE* const oend = ostart + dstSize;
	BYTE* op = ostart + lhSize;
	U32 const singleStream = lhSize == 3;
	symbolEncodingType_e hType = writeEntropy ? hufMetadata->hType : set_repeat;
	size_t cLitSize = 0;

	(void)bmi2; /* TODO bmi2... */

	DEBUGLOG(5, "ZSTD_compressSubBlock_literal (litSize=%zu, lhSize=%zu, writeEntropy=%d)", litSize, lhSize, writeEntropy);

	*entropyWritten = 0;
	if (litSize == 0 \|\| hufMetadata->hType == set_basic) {
	DEBUGLOG(5, "ZSTD_compressSubBlock_literal using raw literal");
	return ZSTD_noCompressLiterals(dst, dstSize, literals, litSize);
	} else if (hufMetadata->hType == set_rle) {
	DEBUGLOG(5, "ZSTD_compressSubBlock_literal using rle literal");
	return ZSTD_compressRleLiteralsBlock(dst, dstSize, literals, litSize);
	}

	assert(litSize > 0);
	assert(hufMetadata->hType == set_compressed \|\| hufMetadata->hType == set_repeat);

	if (writeEntropy && hufMetadata->hType == set_compressed) {
	memcpy(op, hufMetadata->hufDesBuffer, hufMetadata->hufDesSize);
	op += hufMetadata->hufDesSize;
	cLitSize += hufMetadata->hufDesSize;
	DEBUGLOG(5, "ZSTD_compressSubBlock_literal (hSize=%zu)", hufMetadata->hufDesSize);
	}

	/* TODO bmi2 */
	{ const size_t cSize = singleStream ? HUF_compress1X_usingCTable(op, oend-op, literals, litSize, hufTable)
	: HUF_compress4X_usingCTable(op, oend-op, literals, litSize, hufTable);
	op += cSize;
	cLitSize += cSize;
	if (cSize == 0 \|\| ERR_isError(cSize)) {
	DEBUGLOG(5, "Failed to write entropy tables %s", ZSTD_getErrorName(cSize));
	return 0;
	}
	/* If we expand and we aren't writing a header then emit uncompressed */
	if (!writeEntropy && cLitSize >= litSize) {
	DEBUGLOG(5, "ZSTD_compressSubBlock_literal using raw literal because uncompressible");
	return ZSTD_noCompressLiterals(dst, dstSize, literals, litSize);
	}
	/* If we are writing headers then allow expansion that doesn't change our header size. */
	if (lhSize < (size_t)(3 + (cLitSize >= 1 KB) + (cLitSize >= 16 KB))) {
	assert(cLitSize > litSize);
	DEBUGLOG(5, "Literals expanded beyond allowed header size");
	return ZSTD_noCompressLiterals(dst, dstSize, literals, litSize);
	}
	DEBUGLOG(5, "ZSTD_compressSubBlock_literal (cSize=%zu)", cSize);
	}

	/* Build header */
	switch(lhSize)
	{
	case 3: /* 2 - 2 - 10 - 10 */
	{ U32 const lhc = hType + ((!singleStream) << 2) + ((U32)litSize<<4) + ((U32)cLitSize<<14);
	MEM_writeLE24(ostart, lhc);
	break;
	}
	case 4: /* 2 - 2 - 14 - 14 */
	{ U32 const lhc = hType + (2 << 2) + ((U32)litSize<<4) + ((U32)cLitSize<<18);
	MEM_writeLE32(ostart, lhc);
	break;
	}
	case 5: /* 2 - 2 - 18 - 18 */
	{ U32 const lhc = hType + (3 << 2) + ((U32)litSize<<4) + ((U32)cLitSize<<22);
	MEM_writeLE32(ostart, lhc);
	ostart[4] = (BYTE)(cLitSize >> 10);
	break;
	}
	default: /* not possible : lhSize is {3,4,5} */
	assert(0);
	}
	*entropyWritten = 1;
	DEBUGLOG(5, "Compressed literals: %u -> %u", (U32)litSize, (U32)(op-ostart));
	return op-ostart;
	}

	static size_t ZSTD_seqDecompressedSize(seqStore_t const* seqStore, const seqDef* sequences, size_t nbSeq, size_t litSize, int lastSequence) {
	const seqDef* const sstart = sequences;
	const seqDef* const send = sequences + nbSeq;
	const seqDef* sp = sstart;
	size_t matchLengthSum = 0;
	size_t litLengthSum = 0;
	while (send-sp > 0) {
	ZSTD_sequenceLength const seqLen = ZSTD_getSequenceLength(seqStore, sp);
	litLengthSum += seqLen.litLength;
	matchLengthSum += seqLen.matchLength;
	sp++;
	}
	assert(litLengthSum <= litSize);
	if (!lastSequence) {
	assert(litLengthSum == litSize);
	}
	return matchLengthSum + litSize;
	}

	/** ZSTD_compressSubBlock_sequences() :
	* Compresses sequences section for a sub-block.
	* fseMetadata->llType, fseMetadata->ofType, and fseMetadata->mlType have
	* symbol compression modes for the super-block.
	* The first successfully compressed block will have these in its header.
	* We set entropyWritten=1 when we succeed in compressing the sequences.
	* The following sub-blocks will always have repeat mode.
	* @return : compressed size of sequences section of a sub-block
	* Or 0 if it is unable to compress
	* Or error code. */
	static size_t ZSTD_compressSubBlock_sequences(const ZSTD_fseCTables_t* fseTables,
	const ZSTD_fseCTablesMetadata_t* fseMetadata,
	const seqDef* sequences, size_t nbSeq,
	const BYTE* llCode, const BYTE* mlCode, const BYTE* ofCode,
	const ZSTD_CCtx_params* cctxParams,
	void* dst, size_t dstCapacity,
	const int bmi2, int writeEntropy, int* entropyWritten)
	{
	const int longOffsets = cctxParams->cParams.windowLog > STREAM_ACCUMULATOR_MIN;
	BYTE* const ostart = (BYTE*)dst;
	BYTE* const oend = ostart + dstCapacity;
	BYTE* op = ostart;
	BYTE* seqHead;

	DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (nbSeq=%zu, writeEntropy=%d, longOffsets=%d)", nbSeq, writeEntropy, longOffsets);

	*entropyWritten = 0;
	/* Sequences Header */
	RETURN_ERROR_IF((oend-op) < 3 /max nbSeq Size/ + 1 /seqHead/,
	dstSize_tooSmall, "");
	if (nbSeq < 0x7F)
	*op++ = (BYTE)nbSeq;
	else if (nbSeq < LONGNBSEQ)
	op[0] = (BYTE)((nbSeq>>8) + 0x80), op[1] = (BYTE)nbSeq, op+=2;
	else
	op[0]=0xFF, MEM_writeLE16(op+1, (U16)(nbSeq - LONGNBSEQ)), op+=3;
	if (nbSeq==0) {
	return op - ostart;
	}

	/* seqHead : flags for FSE encoding type */
	seqHead = op++;

	DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (seqHeadSize=%u)", (unsigned)(op-ostart));

	if (writeEntropy) {
	const U32 LLtype = fseMetadata->llType;
	const U32 Offtype = fseMetadata->ofType;
	const U32 MLtype = fseMetadata->mlType;
	DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (fseTablesSize=%zu)", fseMetadata->fseTablesSize);
	*seqHead = (BYTE)((LLtype<<6) + (Offtype<<4) + (MLtype<<2));
	memcpy(op, fseMetadata->fseTablesBuffer, fseMetadata->fseTablesSize);
	op += fseMetadata->fseTablesSize;
	} else {
	const U32 repeat = set_repeat;
	*seqHead = (BYTE)((repeat<<6) + (repeat<<4) + (repeat<<2));
	}

	{ size_t const bitstreamSize = ZSTD_encodeSequences(
	op, oend - op,
	fseTables->matchlengthCTable, mlCode,
	fseTables->offcodeCTable, ofCode,
	fseTables->litlengthCTable, llCode,
	sequences, nbSeq,
	longOffsets, bmi2);
	FORWARD_IF_ERROR(bitstreamSize, "ZSTD_encodeSequences failed");
	op += bitstreamSize;
	/* zstd versions <= 1.3.4 mistakenly report corruption when
	* FSE_readNCount() receives a buffer < 4 bytes.
	* Fixed by https://github.com/facebook/zstd/pull/1146.
	* This can happen when the last set_compressed table present is 2
	* bytes and the bitstream is only one byte.
	* In this exceedingly rare case, we will simply emit an uncompressed
	* block, since it isn't worth optimizing.
	*/
	#ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
	if (writeEntropy && fseMetadata->lastCountSize && fseMetadata->lastCountSize + bitstreamSize < 4) {
	/* NCountSize >= 2 && bitstreamSize > 0 ==> lastCountSize == 3 */
	assert(fseMetadata->lastCountSize + bitstreamSize == 3);
	DEBUGLOG(5, "Avoiding bug in zstd decoder in versions <= 1.3.4 by "
	"emitting an uncompressed block.");
	return 0;
	}
	#endif
	DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (bitstreamSize=%zu)", bitstreamSize);
	}

	/* zstd versions <= 1.4.0 mistakenly report error when
	* sequences section body size is less than 3 bytes.
	* Fixed by https://github.com/facebook/zstd/pull/1664.
	* This can happen when the previous sequences section block is compressed
	* with rle mode and the current block's sequences section is compressed
	* with repeat mode where sequences section body size can be 1 byte.
	*/
	#ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
	if (op-seqHead < 4) {
	DEBUGLOG(5, "Avoiding bug in zstd decoder in versions <= 1.4.0 by emitting "
	"an uncompressed block when sequences are < 4 bytes");
	return 0;
	}
	#endif

	*entropyWritten = 1;
	return op - ostart;
	}

	/** ZSTD_compressSubBlock() :
	* Compresses a single sub-block.
	* @return : compressed size of the sub-block
	* Or 0 if it failed to compress. */
	static size_t ZSTD_compressSubBlock(const ZSTD_entropyCTables_t* entropy,
	const ZSTD_entropyCTablesMetadata_t* entropyMetadata,
	const seqDef* sequences, size_t nbSeq,
	const BYTE* literals, size_t litSize,
	const BYTE* llCode, const BYTE* mlCode, const BYTE* ofCode,
	const ZSTD_CCtx_params* cctxParams,
	void* dst, size_t dstCapacity,
	const int bmi2,
	int writeLitEntropy, int writeSeqEntropy,
	int* litEntropyWritten, int* seqEntropyWritten,
	U32 lastBlock)
	{
	BYTE* const ostart = (BYTE*)dst;
	BYTE* const oend = ostart + dstCapacity;
	BYTE* op = ostart + ZSTD_blockHeaderSize;
	DEBUGLOG(5, "ZSTD_compressSubBlock (litSize=%zu, nbSeq=%zu, writeLitEntropy=%d, writeSeqEntropy=%d, lastBlock=%d)",
	litSize, nbSeq, writeLitEntropy, writeSeqEntropy, lastBlock);
	{ size_t cLitSize = ZSTD_compressSubBlock_literal((const HUF_CElt*)entropy->huf.CTable,
	&entropyMetadata->hufMetadata, literals, litSize,
	op, oend-op, bmi2, writeLitEntropy, litEntropyWritten);
	FORWARD_IF_ERROR(cLitSize, "ZSTD_compressSubBlock_literal failed");
	if (cLitSize == 0) return 0;
	op += cLitSize;
	}
	{ size_t cSeqSize = ZSTD_compressSubBlock_sequences(&entropy->fse,
	&entropyMetadata->fseMetadata,
	sequences, nbSeq,
	llCode, mlCode, ofCode,
	cctxParams,
	op, oend-op,
	bmi2, writeSeqEntropy, seqEntropyWritten);
	FORWARD_IF_ERROR(cSeqSize, "ZSTD_compressSubBlock_sequences failed");
	if (cSeqSize == 0) return 0;
	op += cSeqSize;
	}
	/* Write block header */
	{ size_t cSize = (op-ostart)-ZSTD_blockHeaderSize;
	U32 const cBlockHeader24 = lastBlock + (((U32)bt_compressed)<<1) + (U32)(cSize << 3);
	MEM_writeLE24(ostart, cBlockHeader24);
	}
	return op-ostart;
	}

	static size_t ZSTD_estimateSubBlockSize_literal(const BYTE* literals, size_t litSize,
	const ZSTD_hufCTables_t* huf,
	const ZSTD_hufCTablesMetadata_t* hufMetadata,
	void* workspace, size_t wkspSize,
	int writeEntropy)
	{
	unsigned* const countWksp = (unsigned*)workspace;
	unsigned maxSymbolValue = 255;
	size_t literalSectionHeaderSize = 3; /* Use hard coded size of 3 bytes */

	if (hufMetadata->hType == set_basic) return litSize;
	else if (hufMetadata->hType == set_rle) return 1;
	else if (hufMetadata->hType == set_compressed \|\| hufMetadata->hType == set_repeat) {
	size_t const largest = HIST_count_wksp (countWksp, &maxSymbolValue, (const BYTE*)literals, litSize, workspace, wkspSize);
	if (ZSTD_isError(largest)) return litSize;
	{ size_t cLitSizeEstimate = HUF_estimateCompressedSize((const HUF_CElt*)huf->CTable, countWksp, maxSymbolValue);
	if (writeEntropy) cLitSizeEstimate += hufMetadata->hufDesSize;
	return cLitSizeEstimate + literalSectionHeaderSize;
	} }
	assert(0); /* impossible */
	return 0;
	}

	static size_t ZSTD_estimateSubBlockSize_symbolType(symbolEncodingType_e type,
	const BYTE* codeTable, unsigned maxCode,
	size_t nbSeq, const FSE_CTable* fseCTable,
	const U32* additionalBits,
	short const* defaultNorm, U32 defaultNormLog,
	void* workspace, size_t wkspSize)
	{
	unsigned* const countWksp = (unsigned*)workspace;
	const BYTE* ctp = codeTable;
	const BYTE* const ctStart = ctp;
	const BYTE* const ctEnd = ctStart + nbSeq;
	size_t cSymbolTypeSizeEstimateInBits = 0;
	unsigned max = maxCode;

	HIST_countFast_wksp(countWksp, &max, codeTable, nbSeq, workspace, wkspSize); /* can't fail */
	if (type == set_basic) {
	cSymbolTypeSizeEstimateInBits = ZSTD_crossEntropyCost(defaultNorm, defaultNormLog, countWksp, max);
	} else if (type == set_rle) {
	cSymbolTypeSizeEstimateInBits = 0;
	} else if (type == set_compressed \|\| type == set_repeat) {
	cSymbolTypeSizeEstimateInBits = ZSTD_fseBitCost(fseCTable, countWksp, max);
	}
	if (ZSTD_isError(cSymbolTypeSizeEstimateInBits)) return nbSeq * 10;
	while (ctp < ctEnd) {
	if (additionalBits) cSymbolTypeSizeEstimateInBits += additionalBits[*ctp];
	else cSymbolTypeSizeEstimateInBits += ctp; / for offset, offset code is also the number of additional bits */
	ctp++;
	}
	return cSymbolTypeSizeEstimateInBits / 8;
	}

	static size_t ZSTD_estimateSubBlockSize_sequences(const BYTE* ofCodeTable,
	const BYTE* llCodeTable,
	const BYTE* mlCodeTable,
	size_t nbSeq,
	const ZSTD_fseCTables_t* fseTables,
	const ZSTD_fseCTablesMetadata_t* fseMetadata,
	void* workspace, size_t wkspSize,
	int writeEntropy)
	{
	size_t sequencesSectionHeaderSize = 3; /* Use hard coded size of 3 bytes */
	size_t cSeqSizeEstimate = 0;
	cSeqSizeEstimate += ZSTD_estimateSubBlockSize_symbolType(fseMetadata->ofType, ofCodeTable, MaxOff,
	nbSeq, fseTables->offcodeCTable, NULL,
	OF_defaultNorm, OF_defaultNormLog,
	workspace, wkspSize);
	cSeqSizeEstimate += ZSTD_estimateSubBlockSize_symbolType(fseMetadata->llType, llCodeTable, MaxLL,
	nbSeq, fseTables->litlengthCTable, LL_bits,
	LL_defaultNorm, LL_defaultNormLog,
	workspace, wkspSize);
	cSeqSizeEstimate += ZSTD_estimateSubBlockSize_symbolType(fseMetadata->mlType, mlCodeTable, MaxML,
	nbSeq, fseTables->matchlengthCTable, ML_bits,
	ML_defaultNorm, ML_defaultNormLog,
	workspace, wkspSize);
	if (writeEntropy) cSeqSizeEstimate += fseMetadata->fseTablesSize;
	return cSeqSizeEstimate + sequencesSectionHeaderSize;
	}

	static size_t ZSTD_estimateSubBlockSize(const BYTE* literals, size_t litSize,
	const BYTE* ofCodeTable,
	const BYTE* llCodeTable,
	const BYTE* mlCodeTable,
	size_t nbSeq,
	const ZSTD_entropyCTables_t* entropy,
	const ZSTD_entropyCTablesMetadata_t* entropyMetadata,
	void* workspace, size_t wkspSize,
	int writeLitEntropy, int writeSeqEntropy) {
	size_t cSizeEstimate = 0;
	cSizeEstimate += ZSTD_estimateSubBlockSize_literal(literals, litSize,
	&entropy->huf, &entropyMetadata->hufMetadata,
	workspace, wkspSize, writeLitEntropy);
	cSizeEstimate += ZSTD_estimateSubBlockSize_sequences(ofCodeTable, llCodeTable, mlCodeTable,
	nbSeq, &entropy->fse, &entropyMetadata->fseMetadata,
	workspace, wkspSize, writeSeqEntropy);
	return cSizeEstimate + ZSTD_blockHeaderSize;
	}

	static int ZSTD_needSequenceEntropyTables(ZSTD_fseCTablesMetadata_t const* fseMetadata)
	{
	if (fseMetadata->llType == set_compressed \|\| fseMetadata->llType == set_rle)
	return 1;
	if (fseMetadata->mlType == set_compressed \|\| fseMetadata->mlType == set_rle)
	return 1;
	if (fseMetadata->ofType == set_compressed \|\| fseMetadata->ofType == set_rle)
	return 1;
	return 0;
	}

	/** ZSTD_compressSubBlock_multi() :
	* Breaks super-block into multiple sub-blocks and compresses them.
	* Entropy will be written to the first block.
	* The following blocks will use repeat mode to compress.
	* All sub-blocks are compressed blocks (no raw or rle blocks).
	* @return : compressed size of the super block (which is multiple ZSTD blocks)
	* Or 0 if it failed to compress. */
	static size_t ZSTD_compressSubBlock_multi(const seqStore_t* seqStorePtr,
	const ZSTD_compressedBlockState_t* prevCBlock,
	ZSTD_compressedBlockState_t* nextCBlock,
	const ZSTD_entropyCTablesMetadata_t* entropyMetadata,
	const ZSTD_CCtx_params* cctxParams,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const int bmi2, U32 lastBlock,
	void* workspace, size_t wkspSize)
	{
	const seqDef* const sstart = seqStorePtr->sequencesStart;
	const seqDef* const send = seqStorePtr->sequences;
	const seqDef* sp = sstart;
	const BYTE* const lstart = seqStorePtr->litStart;
	const BYTE* const lend = seqStorePtr->lit;
	const BYTE* lp = lstart;
	BYTE const* ip = (BYTE const*)src;
	BYTE const* const iend = ip + srcSize;
	BYTE* const ostart = (BYTE*)dst;
	BYTE* const oend = ostart + dstCapacity;
	BYTE* op = ostart;
	const BYTE* llCodePtr = seqStorePtr->llCode;
	const BYTE* mlCodePtr = seqStorePtr->mlCode;
	const BYTE* ofCodePtr = seqStorePtr->ofCode;
	size_t targetCBlockSize = cctxParams->targetCBlockSize;
	size_t litSize, seqCount;
	int writeLitEntropy = entropyMetadata->hufMetadata.hType == set_compressed;
	int writeSeqEntropy = 1;
	int lastSequence = 0;

	DEBUGLOG(5, "ZSTD_compressSubBlock_multi (litSize=%u, nbSeq=%u)",
	(unsigned)(lend-lp), (unsigned)(send-sstart));

	litSize = 0;
	seqCount = 0;
	do {
	size_t cBlockSizeEstimate = 0;
	if (sstart == send) {
	lastSequence = 1;
	} else {
	const seqDef* const sequence = sp + seqCount;
	lastSequence = sequence == send - 1;
	litSize += ZSTD_getSequenceLength(seqStorePtr, sequence).litLength;
	seqCount++;
	}
	if (lastSequence) {
	assert(lp <= lend);
	assert(litSize <= (size_t)(lend - lp));
	litSize = (size_t)(lend - lp);
	}
	/* I think there is an optimization opportunity here.
	* Calling ZSTD_estimateSubBlockSize for every sequence can be wasteful
	* since it recalculates estimate from scratch.
	* For example, it would recount literal distribution and symbol codes everytime.
	*/
	cBlockSizeEstimate = ZSTD_estimateSubBlockSize(lp, litSize, ofCodePtr, llCodePtr, mlCodePtr, seqCount,
	&nextCBlock->entropy, entropyMetadata,
	workspace, wkspSize, writeLitEntropy, writeSeqEntropy);
	if (cBlockSizeEstimate > targetCBlockSize \|\| lastSequence) {
	int litEntropyWritten = 0;
	int seqEntropyWritten = 0;
	const size_t decompressedSize = ZSTD_seqDecompressedSize(seqStorePtr, sp, seqCount, litSize, lastSequence);
	const size_t cSize = ZSTD_compressSubBlock(&nextCBlock->entropy, entropyMetadata,
	sp, seqCount,
	lp, litSize,
	llCodePtr, mlCodePtr, ofCodePtr,
	cctxParams,
	op, oend-op,
	bmi2, writeLitEntropy, writeSeqEntropy,
	&litEntropyWritten, &seqEntropyWritten,
	lastBlock && lastSequence);
	FORWARD_IF_ERROR(cSize, "ZSTD_compressSubBlock failed");
	if (cSize > 0 && cSize < decompressedSize) {
	DEBUGLOG(5, "Committed the sub-block");
	assert(ip + decompressedSize <= iend);
	ip += decompressedSize;
	sp += seqCount;
	lp += litSize;
	op += cSize;
	llCodePtr += seqCount;
	mlCodePtr += seqCount;
	ofCodePtr += seqCount;
	litSize = 0;
	seqCount = 0;
	/* Entropy only needs to be written once */
	if (litEntropyWritten) {
	writeLitEntropy = 0;
	}
	if (seqEntropyWritten) {
	writeSeqEntropy = 0;
	}
	}
	}
	} while (!lastSequence);
	if (writeLitEntropy) {
	DEBUGLOG(5, "ZSTD_compressSubBlock_multi has literal entropy tables unwritten");
	memcpy(&nextCBlock->entropy.huf, &prevCBlock->entropy.huf, sizeof(prevCBlock->entropy.huf));
	}
	if (writeSeqEntropy && ZSTD_needSequenceEntropyTables(&entropyMetadata->fseMetadata)) {
	/* If we haven't written our entropy tables, then we've violated our contract and
	* must emit an uncompressed block.
	*/
	DEBUGLOG(5, "ZSTD_compressSubBlock_multi has sequence entropy tables unwritten");
	return 0;
	}
	if (ip < iend) {
	size_t const cSize = ZSTD_noCompressBlock(op, oend - op, ip, iend - ip, lastBlock);
	DEBUGLOG(5, "ZSTD_compressSubBlock_multi last sub-block uncompressed, %zu bytes", (size_t)(iend - ip));
	FORWARD_IF_ERROR(cSize, "ZSTD_noCompressBlock failed");
	assert(cSize != 0);
	op += cSize;
	/* We have to regenerate the repcodes because we've skipped some sequences */
	if (sp < send) {
	seqDef const* seq;
	repcodes_t rep;
	memcpy(&rep, prevCBlock->rep, sizeof(rep));
	for (seq = sstart; seq < sp; ++seq) {
	rep = ZSTD_updateRep(rep.rep, seq->offset - 1, ZSTD_getSequenceLength(seqStorePtr, seq).litLength == 0);
	}
	memcpy(nextCBlock->rep, &rep, sizeof(rep));
	}
	}
	DEBUGLOG(5, "ZSTD_compressSubBlock_multi compressed");
	return op-ostart;
	}

	size_t ZSTD_compressSuperBlock(ZSTD_CCtx* zc,
	void* dst, size_t dstCapacity,
	void const* src, size_t srcSize,
	unsigned lastBlock) {
	ZSTD_entropyCTablesMetadata_t entropyMetadata;

	FORWARD_IF_ERROR(ZSTD_buildSuperBlockEntropy(&zc->seqStore,
	&zc->blockState.prevCBlock->entropy,
	&zc->blockState.nextCBlock->entropy,
	&zc->appliedParams,
	&entropyMetadata,
	zc->entropyWorkspace, HUF_WORKSPACE_SIZE /* statically allocated in resetCCtx */), "");

	return ZSTD_compressSubBlock_multi(&zc->seqStore,
	zc->blockState.prevCBlock,
	zc->blockState.nextCBlock,
	&entropyMetadata,
	&zc->appliedParams,
	dst, dstCapacity,
	src, srcSize,
	zc->bmi2, lastBlock,
	zc->entropyWorkspace, HUF_WORKSPACE_SIZE /* statically allocated in resetCCtx */);
	}
	/** ended inlining compress/zstd_compress_superblock.c **/
	/** start inlining compress/zstd_compress.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/-************************************
	* Dependencies
	***************************************/
	#include <limits.h> /* INT_MAX */
	#include <string.h> /* memset */
	/** start inlining ../common/cpu.h **/
	/*
	* Copyright (c) 2018-2020, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_COMMON_CPU_H
	#define ZSTD_COMMON_CPU_H

	/**
	* Implementation taken from folly/CpuId.h
	* https://github.com/facebook/folly/blob/master/folly/CpuId.h
	*/

	#include <string.h>

	/** skipping file: mem.h **/

	#ifdef _MSC_VER
	#include <intrin.h>
	#endif

	typedef struct {
	U32 f1c;
	U32 f1d;
	U32 f7b;
	U32 f7c;
	} ZSTD_cpuid_t;

	MEM_STATIC ZSTD_cpuid_t ZSTD_cpuid(void) {
	U32 f1c = 0;
	U32 f1d = 0;
	U32 f7b = 0;
	U32 f7c = 0;
	#if defined(_MSC_VER) && (defined(_M_X64) \|\| defined(_M_IX86))
	int reg[4];
	__cpuid((int*)reg, 0);
	{
	int const n = reg[0];
	if (n >= 1) {
	__cpuid((int*)reg, 1);
	f1c = (U32)reg[2];
	f1d = (U32)reg[3];
	}
	if (n >= 7) {
	__cpuidex((int*)reg, 7, 0);
	f7b = (U32)reg[1];
	f7c = (U32)reg[2];
	}
	}
	#elif defined(__i386__) && defined(__PIC__) && !defined(__clang__) && defined(__GNUC__)
	/* The following block like the normal cpuid branch below, but gcc
	* reserves ebx for use of its pic register so we must specially
	* handle the save and restore to avoid clobbering the register
	*/
	U32 n;
	__asm__(
	"pushl %%ebx\n\t"
	"cpuid\n\t"
	"popl %%ebx\n\t"
	: "=a"(n)
	: "a"(0)
	: "ecx", "edx");
	if (n >= 1) {
	U32 f1a;
	__asm__(
	"pushl %%ebx\n\t"
	"cpuid\n\t"
	"popl %%ebx\n\t"
	: "=a"(f1a), "=c"(f1c), "=d"(f1d)
	: "a"(1));
	}
	if (n >= 7) {
	__asm__(
	"pushl %%ebx\n\t"
	"cpuid\n\t"
	"movl %%ebx, %%eax\n\t"
	"popl %%ebx"
	: "=a"(f7b), "=c"(f7c)
	: "a"(7), "c"(0)
	: "edx");
	}
	#elif defined(__x86_64__) \|\| defined(_M_X64) \|\| defined(__i386__)
	U32 n;
	__asm__("cpuid" : "=a"(n) : "a"(0) : "ebx", "ecx", "edx");
	if (n >= 1) {
	U32 f1a;
	__asm__("cpuid" : "=a"(f1a), "=c"(f1c), "=d"(f1d) : "a"(1) : "ebx");
	}
	if (n >= 7) {
	U32 f7a;
	__asm__("cpuid"
	: "=a"(f7a), "=b"(f7b), "=c"(f7c)
	: "a"(7), "c"(0)
	: "edx");
	}
	#endif
	{
	ZSTD_cpuid_t cpuid;
	cpuid.f1c = f1c;
	cpuid.f1d = f1d;
	cpuid.f7b = f7b;
	cpuid.f7c = f7c;
	return cpuid;
	}
	}

	#define X(name, r, bit) \
	MEM_STATIC int ZSTD_cpuid_##name(ZSTD_cpuid_t const cpuid) { \
	return ((cpuid.r) & (1U << bit)) != 0; \
	}

	/* cpuid(1): Processor Info and Feature Bits. */
	#define C(name, bit) X(name, f1c, bit)
	C(sse3, 0)
	C(pclmuldq, 1)
	C(dtes64, 2)
	C(monitor, 3)
	C(dscpl, 4)
	C(vmx, 5)
	C(smx, 6)
	C(eist, 7)
	C(tm2, 8)
	C(ssse3, 9)
	C(cnxtid, 10)
	C(fma, 12)
	C(cx16, 13)
	C(xtpr, 14)
	C(pdcm, 15)
	C(pcid, 17)
	C(dca, 18)
	C(sse41, 19)
	C(sse42, 20)
	C(x2apic, 21)
	C(movbe, 22)
	C(popcnt, 23)
	C(tscdeadline, 24)
	C(aes, 25)
	C(xsave, 26)
	C(osxsave, 27)
	C(avx, 28)
	C(f16c, 29)
	C(rdrand, 30)
	#undef C
	#define D(name, bit) X(name, f1d, bit)
	D(fpu, 0)
	D(vme, 1)
	D(de, 2)
	D(pse, 3)
	D(tsc, 4)
	D(msr, 5)
	D(pae, 6)
	D(mce, 7)
	D(cx8, 8)
	D(apic, 9)
	D(sep, 11)
	D(mtrr, 12)
	D(pge, 13)
	D(mca, 14)
	D(cmov, 15)
	D(pat, 16)
	D(pse36, 17)
	D(psn, 18)
	D(clfsh, 19)
	D(ds, 21)
	D(acpi, 22)
	D(mmx, 23)
	D(fxsr, 24)
	D(sse, 25)
	D(sse2, 26)
	D(ss, 27)
	D(htt, 28)
	D(tm, 29)
	D(pbe, 31)
	#undef D

	/* cpuid(7): Extended Features. */
	#define B(name, bit) X(name, f7b, bit)
	B(bmi1, 3)
	B(hle, 4)
	B(avx2, 5)
	B(smep, 7)
	B(bmi2, 8)
	B(erms, 9)
	B(invpcid, 10)
	B(rtm, 11)
	B(mpx, 14)
	B(avx512f, 16)
	B(avx512dq, 17)
	B(rdseed, 18)
	B(adx, 19)
	B(smap, 20)
	B(avx512ifma, 21)
	B(pcommit, 22)
	B(clflushopt, 23)
	B(clwb, 24)
	B(avx512pf, 26)
	B(avx512er, 27)
	B(avx512cd, 28)
	B(sha, 29)
	B(avx512bw, 30)
	B(avx512vl, 31)
	#undef B
	#define C(name, bit) X(name, f7c, bit)
	C(prefetchwt1, 0)
	C(avx512vbmi, 1)
	#undef C

	#undef X

	#endif /* ZSTD_COMMON_CPU_H */
	/** ended inlining ../common/cpu.h **/
	/** skipping file: ../common/mem.h **/
	/** skipping file: hist.h **/
	#define FSE_STATIC_LINKING_ONLY /* FSE_encodeSymbol */
	/** skipping file: ../common/fse.h **/
	#define HUF_STATIC_LINKING_ONLY
	/** skipping file: ../common/huf.h **/
	/** skipping file: zstd_compress_internal.h **/
	/** skipping file: zstd_compress_sequences.h **/
	/** skipping file: zstd_compress_literals.h **/
	/** start inlining zstd_fast.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_FAST_H
	#define ZSTD_FAST_H

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/** skipping file: ../common/mem.h **/
	/** skipping file: zstd_compress_internal.h **/

	void ZSTD_fillHashTable(ZSTD_matchState_t* ms,
	void const* end, ZSTD_dictTableLoadMethod_e dtlm);
	size_t ZSTD_compressBlock_fast(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_fast_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_fast_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);

	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_FAST_H */
	/** ended inlining zstd_fast.h **/
	/** start inlining zstd_double_fast.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_DOUBLE_FAST_H
	#define ZSTD_DOUBLE_FAST_H

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/** skipping file: ../common/mem.h **/
	/** skipping file: zstd_compress_internal.h **/

	void ZSTD_fillDoubleHashTable(ZSTD_matchState_t* ms,
	void const* end, ZSTD_dictTableLoadMethod_e dtlm);
	size_t ZSTD_compressBlock_doubleFast(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_doubleFast_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_doubleFast_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);


	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_DOUBLE_FAST_H */
	/** ended inlining zstd_double_fast.h **/
	/** start inlining zstd_lazy.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_LAZY_H
	#define ZSTD_LAZY_H

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/** skipping file: zstd_compress_internal.h **/

	U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip);

	void ZSTD_preserveUnsortedMark (U32* const table, U32 const size, U32 const reducerValue); /! used in ZSTD_reduceIndex(). preemptively increase value of ZSTD_DUBT_UNSORTED_MARK /

	size_t ZSTD_compressBlock_btlazy2(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_lazy2(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_lazy(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_greedy(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);

	size_t ZSTD_compressBlock_btlazy2_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_lazy2_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_lazy_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_greedy_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);

	size_t ZSTD_compressBlock_greedy_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_lazy_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_lazy2_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_btlazy2_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);

	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_LAZY_H */
	/** ended inlining zstd_lazy.h **/
	/** start inlining zstd_opt.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_OPT_H
	#define ZSTD_OPT_H

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/** skipping file: zstd_compress_internal.h **/

	/* used in ZSTD_loadDictionaryContent() */
	void ZSTD_updateTree(ZSTD_matchState_t* ms, const BYTE* ip, const BYTE* iend);

	size_t ZSTD_compressBlock_btopt(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_btultra(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_btultra2(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);


	size_t ZSTD_compressBlock_btopt_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_btultra_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);

	size_t ZSTD_compressBlock_btopt_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);
	size_t ZSTD_compressBlock_btultra_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);

	/* note : no btultra2 variant for extDict nor dictMatchState,
	* because btultra2 is not meant to work with dictionaries
	* and is only specific for the first block (no prefix) */

	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_OPT_H */
	/** ended inlining zstd_opt.h **/
	/** start inlining zstd_ldm.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_LDM_H
	#define ZSTD_LDM_H

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/** skipping file: zstd_compress_internal.h **/
	/** skipping file: ../zstd.h **/

	/-************************************
	* Long distance matching
	***************************************/

	#define ZSTD_LDM_DEFAULT_WINDOW_LOG ZSTD_WINDOWLOG_LIMIT_DEFAULT

	void ZSTD_ldm_fillHashTable(
	ldmState_t* state, const BYTE* ip,
	const BYTE* iend, ldmParams_t const* params);

	/**
	* ZSTD_ldm_generateSequences():
	*
	* Generates the sequences using the long distance match finder.
	* Generates long range matching sequences in `sequences`, which parse a prefix
	* of the source. `sequences` must be large enough to store every sequence,
	* which can be checked with `ZSTD_ldm_getMaxNbSeq()`.
	* @returns 0 or an error code.
	*
	* NOTE: The user must have called ZSTD_window_update() for all of the input
	* they have, even if they pass it to ZSTD_ldm_generateSequences() in chunks.
	* NOTE: This function returns an error if it runs out of space to store
	* sequences.
	*/
	size_t ZSTD_ldm_generateSequences(
	ldmState_t* ldms, rawSeqStore_t* sequences,
	ldmParams_t const* params, void const* src, size_t srcSize);

	/**
	* ZSTD_ldm_blockCompress():
	*
	* Compresses a block using the predefined sequences, along with a secondary
	* block compressor. The literals section of every sequence is passed to the
	* secondary block compressor, and those sequences are interspersed with the
	* predefined sequences. Returns the length of the last literals.
	* Updates `rawSeqStore.pos` to indicate how many sequences have been consumed.
	* `rawSeqStore.seq` may also be updated to split the last sequence between two
	* blocks.
	* @return The length of the last literals.
	*
	* NOTE: The source must be at most the maximum block size, but the predefined
	* sequences can be any size, and may be longer than the block. In the case that
	* they are longer than the block, the last sequences may need to be split into
	* two. We handle that case correctly, and update `rawSeqStore` appropriately.
	* NOTE: This function does not return any errors.
	*/
	size_t ZSTD_ldm_blockCompress(rawSeqStore_t* rawSeqStore,
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize);

	/**
	* ZSTD_ldm_skipSequences():
	*
	* Skip past `srcSize` bytes worth of sequences in `rawSeqStore`.
	* Avoids emitting matches less than `minMatch` bytes.
	* Must be called for data with is not passed to ZSTD_ldm_blockCompress().
	*/
	void ZSTD_ldm_skipSequences(rawSeqStore_t* rawSeqStore, size_t srcSize,
	U32 const minMatch);


	/** ZSTD_ldm_getTableSize() :
	* Estimate the space needed for long distance matching tables or 0 if LDM is
	* disabled.
	*/
	size_t ZSTD_ldm_getTableSize(ldmParams_t params);

	/** ZSTD_ldm_getSeqSpace() :
	* Return an upper bound on the number of sequences that can be produced by
	* the long distance matcher, or 0 if LDM is disabled.
	*/
	size_t ZSTD_ldm_getMaxNbSeq(ldmParams_t params, size_t maxChunkSize);

	/** ZSTD_ldm_adjustParameters() :
	* If the params->hashRateLog is not set, set it to its default value based on
	* windowLog and params->hashLog.
	*
	* Ensures that params->bucketSizeLog is <= params->hashLog (setting it to
	* params->hashLog if it is not).
	*
	* Ensures that the minMatchLength >= targetLength during optimal parsing.
	*/
	void ZSTD_ldm_adjustParameters(ldmParams_t* params,
	ZSTD_compressionParameters const* cParams);

	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_FAST_H */
	/** ended inlining zstd_ldm.h **/
	/** skipping file: zstd_compress_superblock.h **/


	/-************************************
	* Helper functions
	***************************************/
	/* ZSTD_compressBound()
	* Note that the result from this function is only compatible with the "normal"
	* full-block strategy.
	* When there are a lot of small blocks due to frequent flush in streaming mode
	* the overhead of headers can make the compressed data to be larger than the
	* return value of ZSTD_compressBound().
	*/
	size_t ZSTD_compressBound(size_t srcSize) {
	return ZSTD_COMPRESSBOUND(srcSize);
	}


	/-************************************
	* Context memory management
	***************************************/
	struct ZSTD_CDict_s {
	const void* dictContent;
	size_t dictContentSize;
	U32* entropyWorkspace; /* entropy workspace of HUF_WORKSPACE_SIZE bytes */
	ZSTD_cwksp workspace;
	ZSTD_matchState_t matchState;
	ZSTD_compressedBlockState_t cBlockState;
	ZSTD_customMem customMem;
	U32 dictID;
	int compressionLevel; /* 0 indicates that advanced API was used to select CDict params */
	}; /* typedef'd to ZSTD_CDict within "zstd.h" */

	ZSTD_CCtx* ZSTD_createCCtx(void)
	{
	return ZSTD_createCCtx_advanced(ZSTD_defaultCMem);
	}

	static void ZSTD_initCCtx(ZSTD_CCtx* cctx, ZSTD_customMem memManager)
	{
	assert(cctx != NULL);
	memset(cctx, 0, sizeof(*cctx));
	cctx->customMem = memManager;
	cctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid());
	{ size_t const err = ZSTD_CCtx_reset(cctx, ZSTD_reset_parameters);
	assert(!ZSTD_isError(err));
	(void)err;
	}
	}

	ZSTD_CCtx* ZSTD_createCCtx_advanced(ZSTD_customMem customMem)
	{
	ZSTD_STATIC_ASSERT(zcss_init==0);
	ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN==(0ULL - 1));
	if (!customMem.customAlloc ^ !customMem.customFree) return NULL;
	{ ZSTD_CCtx* const cctx = (ZSTD_CCtx*)ZSTD_malloc(sizeof(ZSTD_CCtx), customMem);
	if (!cctx) return NULL;
	ZSTD_initCCtx(cctx, customMem);
	return cctx;
	}
	}

	ZSTD_CCtx* ZSTD_initStaticCCtx(void* workspace, size_t workspaceSize)
	{
	ZSTD_cwksp ws;
	ZSTD_CCtx* cctx;
	if (workspaceSize <= sizeof(ZSTD_CCtx)) return NULL; /* minimum size */
	if ((size_t)workspace & 7) return NULL; /* must be 8-aligned */
	ZSTD_cwksp_init(&ws, workspace, workspaceSize);

	cctx = (ZSTD_CCtx*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CCtx));
	if (cctx == NULL) return NULL;

	memset(cctx, 0, sizeof(ZSTD_CCtx));
	ZSTD_cwksp_move(&cctx->workspace, &ws);
	cctx->staticSize = workspaceSize;

	/* statically sized space. entropyWorkspace never moves (but prev/next block swap places) */
	if (!ZSTD_cwksp_check_available(&cctx->workspace, HUF_WORKSPACE_SIZE + 2 * sizeof(ZSTD_compressedBlockState_t))) return NULL;
	cctx->blockState.prevCBlock = (ZSTD_compressedBlockState_t*)ZSTD_cwksp_reserve_object(&cctx->workspace, sizeof(ZSTD_compressedBlockState_t));
	cctx->blockState.nextCBlock = (ZSTD_compressedBlockState_t*)ZSTD_cwksp_reserve_object(&cctx->workspace, sizeof(ZSTD_compressedBlockState_t));
	cctx->entropyWorkspace = (U32*)ZSTD_cwksp_reserve_object(&cctx->workspace, HUF_WORKSPACE_SIZE);
	cctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid());
	return cctx;
	}

	/**
	* Clears and frees all of the dictionaries in the CCtx.
	*/
	static void ZSTD_clearAllDicts(ZSTD_CCtx* cctx)
	{
	ZSTD_free(cctx->localDict.dictBuffer, cctx->customMem);
	ZSTD_freeCDict(cctx->localDict.cdict);
	memset(&cctx->localDict, 0, sizeof(cctx->localDict));
	memset(&cctx->prefixDict, 0, sizeof(cctx->prefixDict));
	cctx->cdict = NULL;
	}

	static size_t ZSTD_sizeof_localDict(ZSTD_localDict dict)
	{
	size_t const bufferSize = dict.dictBuffer != NULL ? dict.dictSize : 0;
	size_t const cdictSize = ZSTD_sizeof_CDict(dict.cdict);
	return bufferSize + cdictSize;
	}

	static void ZSTD_freeCCtxContent(ZSTD_CCtx* cctx)
	{
	assert(cctx != NULL);
	assert(cctx->staticSize == 0);
	ZSTD_clearAllDicts(cctx);
	#ifdef ZSTD_MULTITHREAD
	ZSTDMT_freeCCtx(cctx->mtctx); cctx->mtctx = NULL;
	#endif
	ZSTD_cwksp_free(&cctx->workspace, cctx->customMem);
	}

	size_t ZSTD_freeCCtx(ZSTD_CCtx* cctx)
	{
	if (cctx==NULL) return 0; /* support free on NULL */
	RETURN_ERROR_IF(cctx->staticSize, memory_allocation,
	"not compatible with static CCtx");
	{
	int cctxInWorkspace = ZSTD_cwksp_owns_buffer(&cctx->workspace, cctx);
	ZSTD_freeCCtxContent(cctx);
	if (!cctxInWorkspace) {
	ZSTD_free(cctx, cctx->customMem);
	}
	}
	return 0;
	}


	static size_t ZSTD_sizeof_mtctx(const ZSTD_CCtx* cctx)
	{
	#ifdef ZSTD_MULTITHREAD
	return ZSTDMT_sizeof_CCtx(cctx->mtctx);
	#else
	(void)cctx;
	return 0;
	#endif
	}


	size_t ZSTD_sizeof_CCtx(const ZSTD_CCtx* cctx)
	{
	if (cctx==NULL) return 0; /* support sizeof on NULL */
	/* cctx may be in the workspace */
	return (cctx->workspace.workspace == cctx ? 0 : sizeof(*cctx))
	+ ZSTD_cwksp_sizeof(&cctx->workspace)
	+ ZSTD_sizeof_localDict(cctx->localDict)
	+ ZSTD_sizeof_mtctx(cctx);
	}

	size_t ZSTD_sizeof_CStream(const ZSTD_CStream* zcs)
	{
	return ZSTD_sizeof_CCtx(zcs); /* same object */
	}

	/* private API call, for dictBuilder only */
	const seqStore_t* ZSTD_getSeqStore(const ZSTD_CCtx* ctx) { return &(ctx->seqStore); }

	static ZSTD_CCtx_params ZSTD_makeCCtxParamsFromCParams(
	ZSTD_compressionParameters cParams)
	{
	ZSTD_CCtx_params cctxParams;
	memset(&cctxParams, 0, sizeof(cctxParams));
	cctxParams.cParams = cParams;
	cctxParams.compressionLevel = ZSTD_CLEVEL_DEFAULT; /* should not matter, as all cParams are presumed properly defined */
	assert(!ZSTD_checkCParams(cParams));
	cctxParams.fParams.contentSizeFlag = 1;
	return cctxParams;
	}

	static ZSTD_CCtx_params* ZSTD_createCCtxParams_advanced(
	ZSTD_customMem customMem)
	{
	ZSTD_CCtx_params* params;
	if (!customMem.customAlloc ^ !customMem.customFree) return NULL;
	params = (ZSTD_CCtx_params*)ZSTD_calloc(
	sizeof(ZSTD_CCtx_params), customMem);
	if (!params) { return NULL; }
	params->customMem = customMem;
	params->compressionLevel = ZSTD_CLEVEL_DEFAULT;
	params->fParams.contentSizeFlag = 1;
	return params;
	}

	ZSTD_CCtx_params* ZSTD_createCCtxParams(void)
	{
	return ZSTD_createCCtxParams_advanced(ZSTD_defaultCMem);
	}

	size_t ZSTD_freeCCtxParams(ZSTD_CCtx_params* params)
	{
	if (params == NULL) { return 0; }
	ZSTD_free(params, params->customMem);
	return 0;
	}

	size_t ZSTD_CCtxParams_reset(ZSTD_CCtx_params* params)
	{
	return ZSTD_CCtxParams_init(params, ZSTD_CLEVEL_DEFAULT);
	}

	size_t ZSTD_CCtxParams_init(ZSTD_CCtx_params* cctxParams, int compressionLevel) {
	RETURN_ERROR_IF(!cctxParams, GENERIC, "NULL pointer!");
	memset(cctxParams, 0, sizeof(*cctxParams));
	cctxParams->compressionLevel = compressionLevel;
	cctxParams->fParams.contentSizeFlag = 1;
	return 0;
	}

	size_t ZSTD_CCtxParams_init_advanced(ZSTD_CCtx_params* cctxParams, ZSTD_parameters params)
	{
	RETURN_ERROR_IF(!cctxParams, GENERIC, "NULL pointer!");
	FORWARD_IF_ERROR( ZSTD_checkCParams(params.cParams) , "");
	memset(cctxParams, 0, sizeof(*cctxParams));
	assert(!ZSTD_checkCParams(params.cParams));
	cctxParams->cParams = params.cParams;
	cctxParams->fParams = params.fParams;
	cctxParams->compressionLevel = ZSTD_CLEVEL_DEFAULT; /* should not matter, as all cParams are presumed properly defined */
	return 0;
	}

	/* ZSTD_assignParamsToCCtxParams() :
	* params is presumed valid at this stage */
	static ZSTD_CCtx_params ZSTD_assignParamsToCCtxParams(
	const ZSTD_CCtx_params* cctxParams, const ZSTD_parameters* params)
	{
	ZSTD_CCtx_params ret = *cctxParams;
	assert(!ZSTD_checkCParams(params->cParams));
	ret.cParams = params->cParams;
	ret.fParams = params->fParams;
	ret.compressionLevel = ZSTD_CLEVEL_DEFAULT; /* should not matter, as all cParams are presumed properly defined */
	return ret;
	}

	ZSTD_bounds ZSTD_cParam_getBounds(ZSTD_cParameter param)
	{
	ZSTD_bounds bounds = { 0, 0, 0 };

	switch(param)
	{
	case ZSTD_c_compressionLevel:
	bounds.lowerBound = ZSTD_minCLevel();
	bounds.upperBound = ZSTD_maxCLevel();
	return bounds;

	case ZSTD_c_windowLog:
	bounds.lowerBound = ZSTD_WINDOWLOG_MIN;
	bounds.upperBound = ZSTD_WINDOWLOG_MAX;
	return bounds;

	case ZSTD_c_hashLog:
	bounds.lowerBound = ZSTD_HASHLOG_MIN;
	bounds.upperBound = ZSTD_HASHLOG_MAX;
	return bounds;

	case ZSTD_c_chainLog:
	bounds.lowerBound = ZSTD_CHAINLOG_MIN;
	bounds.upperBound = ZSTD_CHAINLOG_MAX;
	return bounds;

	case ZSTD_c_searchLog:
	bounds.lowerBound = ZSTD_SEARCHLOG_MIN;
	bounds.upperBound = ZSTD_SEARCHLOG_MAX;
	return bounds;

	case ZSTD_c_minMatch:
	bounds.lowerBound = ZSTD_MINMATCH_MIN;
	bounds.upperBound = ZSTD_MINMATCH_MAX;
	return bounds;

	case ZSTD_c_targetLength:
	bounds.lowerBound = ZSTD_TARGETLENGTH_MIN;
	bounds.upperBound = ZSTD_TARGETLENGTH_MAX;
	return bounds;

	case ZSTD_c_strategy:
	bounds.lowerBound = ZSTD_STRATEGY_MIN;
	bounds.upperBound = ZSTD_STRATEGY_MAX;
	return bounds;

	case ZSTD_c_contentSizeFlag:
	bounds.lowerBound = 0;
	bounds.upperBound = 1;
	return bounds;

	case ZSTD_c_checksumFlag:
	bounds.lowerBound = 0;
	bounds.upperBound = 1;
	return bounds;

	case ZSTD_c_dictIDFlag:
	bounds.lowerBound = 0;
	bounds.upperBound = 1;
	return bounds;

	case ZSTD_c_nbWorkers:
	bounds.lowerBound = 0;
	#ifdef ZSTD_MULTITHREAD
	bounds.upperBound = ZSTDMT_NBWORKERS_MAX;
	#else
	bounds.upperBound = 0;
	#endif
	return bounds;

	case ZSTD_c_jobSize:
	bounds.lowerBound = 0;
	#ifdef ZSTD_MULTITHREAD
	bounds.upperBound = ZSTDMT_JOBSIZE_MAX;
	#else
	bounds.upperBound = 0;
	#endif
	return bounds;

	case ZSTD_c_overlapLog:
	#ifdef ZSTD_MULTITHREAD
	bounds.lowerBound = ZSTD_OVERLAPLOG_MIN;
	bounds.upperBound = ZSTD_OVERLAPLOG_MAX;
	#else
	bounds.lowerBound = 0;
	bounds.upperBound = 0;
	#endif
	return bounds;

	case ZSTD_c_enableLongDistanceMatching:
	bounds.lowerBound = 0;
	bounds.upperBound = 1;
	return bounds;

	case ZSTD_c_ldmHashLog:
	bounds.lowerBound = ZSTD_LDM_HASHLOG_MIN;
	bounds.upperBound = ZSTD_LDM_HASHLOG_MAX;
	return bounds;

	case ZSTD_c_ldmMinMatch:
	bounds.lowerBound = ZSTD_LDM_MINMATCH_MIN;
	bounds.upperBound = ZSTD_LDM_MINMATCH_MAX;
	return bounds;

	case ZSTD_c_ldmBucketSizeLog:
	bounds.lowerBound = ZSTD_LDM_BUCKETSIZELOG_MIN;
	bounds.upperBound = ZSTD_LDM_BUCKETSIZELOG_MAX;
	return bounds;

	case ZSTD_c_ldmHashRateLog:
	bounds.lowerBound = ZSTD_LDM_HASHRATELOG_MIN;
	bounds.upperBound = ZSTD_LDM_HASHRATELOG_MAX;
	return bounds;

	/* experimental parameters */
	case ZSTD_c_rsyncable:
	bounds.lowerBound = 0;
	bounds.upperBound = 1;
	return bounds;

	case ZSTD_c_forceMaxWindow :
	bounds.lowerBound = 0;
	bounds.upperBound = 1;
	return bounds;

	case ZSTD_c_format:
	ZSTD_STATIC_ASSERT(ZSTD_f_zstd1 < ZSTD_f_zstd1_magicless);
	bounds.lowerBound = ZSTD_f_zstd1;
	bounds.upperBound = ZSTD_f_zstd1_magicless; /* note : how to ensure at compile time that this is the highest value enum ? */
	return bounds;

	case ZSTD_c_forceAttachDict:
	ZSTD_STATIC_ASSERT(ZSTD_dictDefaultAttach < ZSTD_dictForceCopy);
	bounds.lowerBound = ZSTD_dictDefaultAttach;
	bounds.upperBound = ZSTD_dictForceLoad; /* note : how to ensure at compile time that this is the highest value enum ? */
	return bounds;

	case ZSTD_c_literalCompressionMode:
	ZSTD_STATIC_ASSERT(ZSTD_lcm_auto < ZSTD_lcm_huffman && ZSTD_lcm_huffman < ZSTD_lcm_uncompressed);
	bounds.lowerBound = ZSTD_lcm_auto;
	bounds.upperBound = ZSTD_lcm_uncompressed;
	return bounds;

	case ZSTD_c_targetCBlockSize:
	bounds.lowerBound = ZSTD_TARGETCBLOCKSIZE_MIN;
	bounds.upperBound = ZSTD_TARGETCBLOCKSIZE_MAX;
	return bounds;

	case ZSTD_c_srcSizeHint:
	bounds.lowerBound = ZSTD_SRCSIZEHINT_MIN;
	bounds.upperBound = ZSTD_SRCSIZEHINT_MAX;
	return bounds;

	default:
	bounds.error = ERROR(parameter_unsupported);
	return bounds;
	}
	}

	/* ZSTD_cParam_clampBounds:
	* Clamps the value into the bounded range.
	*/
	static size_t ZSTD_cParam_clampBounds(ZSTD_cParameter cParam, int* value)
	{
	ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam);
	if (ZSTD_isError(bounds.error)) return bounds.error;
	if (value < bounds.lowerBound) value = bounds.lowerBound;
	if (value > bounds.upperBound) value = bounds.upperBound;
	return 0;
	}

	#define BOUNDCHECK(cParam, val) { \
	RETURN_ERROR_IF(!ZSTD_cParam_withinBounds(cParam,val), \
	parameter_outOfBound, "Param out of bounds"); \
	}


	static int ZSTD_isUpdateAuthorized(ZSTD_cParameter param)
	{
	switch(param)
	{
	case ZSTD_c_compressionLevel:
	case ZSTD_c_hashLog:
	case ZSTD_c_chainLog:
	case ZSTD_c_searchLog:
	case ZSTD_c_minMatch:
	case ZSTD_c_targetLength:
	case ZSTD_c_strategy:
	return 1;

	case ZSTD_c_format:
	case ZSTD_c_windowLog:
	case ZSTD_c_contentSizeFlag:
	case ZSTD_c_checksumFlag:
	case ZSTD_c_dictIDFlag:
	case ZSTD_c_forceMaxWindow :
	case ZSTD_c_nbWorkers:
	case ZSTD_c_jobSize:
	case ZSTD_c_overlapLog:
	case ZSTD_c_rsyncable:
	case ZSTD_c_enableLongDistanceMatching:
	case ZSTD_c_ldmHashLog:
	case ZSTD_c_ldmMinMatch:
	case ZSTD_c_ldmBucketSizeLog:
	case ZSTD_c_ldmHashRateLog:
	case ZSTD_c_forceAttachDict:
	case ZSTD_c_literalCompressionMode:
	case ZSTD_c_targetCBlockSize:
	case ZSTD_c_srcSizeHint:
	default:
	return 0;
	}
	}

	size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int value)
	{
	DEBUGLOG(4, "ZSTD_CCtx_setParameter (%i, %i)", (int)param, value);
	if (cctx->streamStage != zcss_init) {
	if (ZSTD_isUpdateAuthorized(param)) {
	cctx->cParamsChanged = 1;
	} else {
	RETURN_ERROR(stage_wrong, "can only set params in ctx init stage");
	} }

	switch(param)
	{
	case ZSTD_c_nbWorkers:
	RETURN_ERROR_IF((value!=0) && cctx->staticSize, parameter_unsupported,
	"MT not compatible with static alloc");
	break;

	case ZSTD_c_compressionLevel:
	case ZSTD_c_windowLog:
	case ZSTD_c_hashLog:
	case ZSTD_c_chainLog:
	case ZSTD_c_searchLog:
	case ZSTD_c_minMatch:
	case ZSTD_c_targetLength:
	case ZSTD_c_strategy:
	case ZSTD_c_ldmHashRateLog:
	case ZSTD_c_format:
	case ZSTD_c_contentSizeFlag:
	case ZSTD_c_checksumFlag:
	case ZSTD_c_dictIDFlag:
	case ZSTD_c_forceMaxWindow:
	case ZSTD_c_forceAttachDict:
	case ZSTD_c_literalCompressionMode:
	case ZSTD_c_jobSize:
	case ZSTD_c_overlapLog:
	case ZSTD_c_rsyncable:
	case ZSTD_c_enableLongDistanceMatching:
	case ZSTD_c_ldmHashLog:
	case ZSTD_c_ldmMinMatch:
	case ZSTD_c_ldmBucketSizeLog:
	case ZSTD_c_targetCBlockSize:
	case ZSTD_c_srcSizeHint:
	break;

	default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
	}
	return ZSTD_CCtxParams_setParameter(&cctx->requestedParams, param, value);
	}

	size_t ZSTD_CCtxParams_setParameter(ZSTD_CCtx_params* CCtxParams,
	ZSTD_cParameter param, int value)
	{
	DEBUGLOG(4, "ZSTD_CCtxParams_setParameter (%i, %i)", (int)param, value);
	switch(param)
	{
	case ZSTD_c_format :
	BOUNDCHECK(ZSTD_c_format, value);
	CCtxParams->format = (ZSTD_format_e)value;
	return (size_t)CCtxParams->format;

	case ZSTD_c_compressionLevel : {
	FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), "");
	if (value) { /* 0 : does not change current level */
	CCtxParams->compressionLevel = value;
	}
	if (CCtxParams->compressionLevel >= 0) return (size_t)CCtxParams->compressionLevel;
	return 0; /* return type (size_t) cannot represent negative values */
	}

	case ZSTD_c_windowLog :
	if (value!=0) /* 0 => use default */
	BOUNDCHECK(ZSTD_c_windowLog, value);
	CCtxParams->cParams.windowLog = (U32)value;
	return CCtxParams->cParams.windowLog;

	case ZSTD_c_hashLog :
	if (value!=0) /* 0 => use default */
	BOUNDCHECK(ZSTD_c_hashLog, value);
	CCtxParams->cParams.hashLog = (U32)value;
	return CCtxParams->cParams.hashLog;

	case ZSTD_c_chainLog :
	if (value!=0) /* 0 => use default */
	BOUNDCHECK(ZSTD_c_chainLog, value);
	CCtxParams->cParams.chainLog = (U32)value;
	return CCtxParams->cParams.chainLog;

	case ZSTD_c_searchLog :
	if (value!=0) /* 0 => use default */
	BOUNDCHECK(ZSTD_c_searchLog, value);
	CCtxParams->cParams.searchLog = (U32)value;
	return (size_t)value;

	case ZSTD_c_minMatch :
	if (value!=0) /* 0 => use default */
	BOUNDCHECK(ZSTD_c_minMatch, value);
	CCtxParams->cParams.minMatch = value;
	return CCtxParams->cParams.minMatch;

	case ZSTD_c_targetLength :
	BOUNDCHECK(ZSTD_c_targetLength, value);
	CCtxParams->cParams.targetLength = value;
	return CCtxParams->cParams.targetLength;

	case ZSTD_c_strategy :
	if (value!=0) /* 0 => use default */
	BOUNDCHECK(ZSTD_c_strategy, value);
	CCtxParams->cParams.strategy = (ZSTD_strategy)value;
	return (size_t)CCtxParams->cParams.strategy;

	case ZSTD_c_contentSizeFlag :
	/* Content size written in frame header _when known_ (default:1) */
	DEBUGLOG(4, "set content size flag = %u", (value!=0));
	CCtxParams->fParams.contentSizeFlag = value != 0;
	return CCtxParams->fParams.contentSizeFlag;

	case ZSTD_c_checksumFlag :
	/* A 32-bits content checksum will be calculated and written at end of frame (default:0) */
	CCtxParams->fParams.checksumFlag = value != 0;
	return CCtxParams->fParams.checksumFlag;

	case ZSTD_c_dictIDFlag : /* When applicable, dictionary's dictID is provided in frame header (default:1) */
	DEBUGLOG(4, "set dictIDFlag = %u", (value!=0));
	CCtxParams->fParams.noDictIDFlag = !value;
	return !CCtxParams->fParams.noDictIDFlag;

	case ZSTD_c_forceMaxWindow :
	CCtxParams->forceWindow = (value != 0);
	return CCtxParams->forceWindow;

	case ZSTD_c_forceAttachDict : {
	const ZSTD_dictAttachPref_e pref = (ZSTD_dictAttachPref_e)value;
	BOUNDCHECK(ZSTD_c_forceAttachDict, pref);
	CCtxParams->attachDictPref = pref;
	return CCtxParams->attachDictPref;
	}

	case ZSTD_c_literalCompressionMode : {
	const ZSTD_literalCompressionMode_e lcm = (ZSTD_literalCompressionMode_e)value;
	BOUNDCHECK(ZSTD_c_literalCompressionMode, lcm);
	CCtxParams->literalCompressionMode = lcm;
	return CCtxParams->literalCompressionMode;
	}

	case ZSTD_c_nbWorkers :
	#ifndef ZSTD_MULTITHREAD
	RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
	return 0;
	#else
	FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), "");
	CCtxParams->nbWorkers = value;
	return CCtxParams->nbWorkers;
	#endif

	case ZSTD_c_jobSize :
	#ifndef ZSTD_MULTITHREAD
	RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
	return 0;
	#else
	/* Adjust to the minimum non-default value. */
	if (value != 0 && value < ZSTDMT_JOBSIZE_MIN)
	value = ZSTDMT_JOBSIZE_MIN;
	FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), "");
	assert(value >= 0);
	CCtxParams->jobSize = value;
	return CCtxParams->jobSize;
	#endif

	case ZSTD_c_overlapLog :
	#ifndef ZSTD_MULTITHREAD
	RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
	return 0;
	#else
	FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(ZSTD_c_overlapLog, &value), "");
	CCtxParams->overlapLog = value;
	return CCtxParams->overlapLog;
	#endif

	case ZSTD_c_rsyncable :
	#ifndef ZSTD_MULTITHREAD
	RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
	return 0;
	#else
	FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(ZSTD_c_overlapLog, &value), "");
	CCtxParams->rsyncable = value;
	return CCtxParams->rsyncable;
	#endif

	case ZSTD_c_enableLongDistanceMatching :
	CCtxParams->ldmParams.enableLdm = (value!=0);
	return CCtxParams->ldmParams.enableLdm;

	case ZSTD_c_ldmHashLog :
	if (value!=0) /* 0 ==> auto */
	BOUNDCHECK(ZSTD_c_ldmHashLog, value);
	CCtxParams->ldmParams.hashLog = value;
	return CCtxParams->ldmParams.hashLog;

	case ZSTD_c_ldmMinMatch :
	if (value!=0) /* 0 ==> default */
	BOUNDCHECK(ZSTD_c_ldmMinMatch, value);
	CCtxParams->ldmParams.minMatchLength = value;
	return CCtxParams->ldmParams.minMatchLength;

	case ZSTD_c_ldmBucketSizeLog :
	if (value!=0) /* 0 ==> default */
	BOUNDCHECK(ZSTD_c_ldmBucketSizeLog, value);
	CCtxParams->ldmParams.bucketSizeLog = value;
	return CCtxParams->ldmParams.bucketSizeLog;

	case ZSTD_c_ldmHashRateLog :
	RETURN_ERROR_IF(value > ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN,
	parameter_outOfBound, "Param out of bounds!");
	CCtxParams->ldmParams.hashRateLog = value;
	return CCtxParams->ldmParams.hashRateLog;

	case ZSTD_c_targetCBlockSize :
	if (value!=0) /* 0 ==> default */
	BOUNDCHECK(ZSTD_c_targetCBlockSize, value);
	CCtxParams->targetCBlockSize = value;
	return CCtxParams->targetCBlockSize;

	case ZSTD_c_srcSizeHint :
	if (value!=0) /* 0 ==> default */
	BOUNDCHECK(ZSTD_c_srcSizeHint, value);
	CCtxParams->srcSizeHint = value;
	return CCtxParams->srcSizeHint;

	default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
	}
	}

	size_t ZSTD_CCtx_getParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int* value)
	{
	return ZSTD_CCtxParams_getParameter(&cctx->requestedParams, param, value);
	}

	size_t ZSTD_CCtxParams_getParameter(
	ZSTD_CCtx_params* CCtxParams, ZSTD_cParameter param, int* value)
	{
	switch(param)
	{
	case ZSTD_c_format :
	*value = CCtxParams->format;
	break;
	case ZSTD_c_compressionLevel :
	*value = CCtxParams->compressionLevel;
	break;
	case ZSTD_c_windowLog :
	*value = (int)CCtxParams->cParams.windowLog;
	break;
	case ZSTD_c_hashLog :
	*value = (int)CCtxParams->cParams.hashLog;
	break;
	case ZSTD_c_chainLog :
	*value = (int)CCtxParams->cParams.chainLog;
	break;
	case ZSTD_c_searchLog :
	*value = CCtxParams->cParams.searchLog;
	break;
	case ZSTD_c_minMatch :
	*value = CCtxParams->cParams.minMatch;
	break;
	case ZSTD_c_targetLength :
	*value = CCtxParams->cParams.targetLength;
	break;
	case ZSTD_c_strategy :
	*value = (unsigned)CCtxParams->cParams.strategy;
	break;
	case ZSTD_c_contentSizeFlag :
	*value = CCtxParams->fParams.contentSizeFlag;
	break;
	case ZSTD_c_checksumFlag :
	*value = CCtxParams->fParams.checksumFlag;
	break;
	case ZSTD_c_dictIDFlag :
	*value = !CCtxParams->fParams.noDictIDFlag;
	break;
	case ZSTD_c_forceMaxWindow :
	*value = CCtxParams->forceWindow;
	break;
	case ZSTD_c_forceAttachDict :
	*value = CCtxParams->attachDictPref;
	break;
	case ZSTD_c_literalCompressionMode :
	*value = CCtxParams->literalCompressionMode;
	break;
	case ZSTD_c_nbWorkers :
	#ifndef ZSTD_MULTITHREAD
	assert(CCtxParams->nbWorkers == 0);
	#endif
	*value = CCtxParams->nbWorkers;
	break;
	case ZSTD_c_jobSize :
	#ifndef ZSTD_MULTITHREAD
	RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
	#else
	assert(CCtxParams->jobSize <= INT_MAX);
	*value = (int)CCtxParams->jobSize;
	break;
	#endif
	case ZSTD_c_overlapLog :
	#ifndef ZSTD_MULTITHREAD
	RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
	#else
	*value = CCtxParams->overlapLog;
	break;
	#endif
	case ZSTD_c_rsyncable :
	#ifndef ZSTD_MULTITHREAD
	RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
	#else
	*value = CCtxParams->rsyncable;
	break;
	#endif
	case ZSTD_c_enableLongDistanceMatching :
	*value = CCtxParams->ldmParams.enableLdm;
	break;
	case ZSTD_c_ldmHashLog :
	*value = CCtxParams->ldmParams.hashLog;
	break;
	case ZSTD_c_ldmMinMatch :
	*value = CCtxParams->ldmParams.minMatchLength;
	break;
	case ZSTD_c_ldmBucketSizeLog :
	*value = CCtxParams->ldmParams.bucketSizeLog;
	break;
	case ZSTD_c_ldmHashRateLog :
	*value = CCtxParams->ldmParams.hashRateLog;
	break;
	case ZSTD_c_targetCBlockSize :
	*value = (int)CCtxParams->targetCBlockSize;
	break;
	case ZSTD_c_srcSizeHint :
	*value = (int)CCtxParams->srcSizeHint;
	break;
	default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
	}
	return 0;
	}

	/** ZSTD_CCtx_setParametersUsingCCtxParams() :
	* just applies `params` into `cctx`
	* no action is performed, parameters are merely stored.
	* If ZSTDMT is enabled, parameters are pushed to cctx->mtctx.
	* This is possible even if a compression is ongoing.
	* In which case, new parameters will be applied on the fly, starting with next compression job.
	*/
	size_t ZSTD_CCtx_setParametersUsingCCtxParams(
	ZSTD_CCtx* cctx, const ZSTD_CCtx_params* params)
	{
	DEBUGLOG(4, "ZSTD_CCtx_setParametersUsingCCtxParams");
	RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
	"The context is in the wrong stage!");
	RETURN_ERROR_IF(cctx->cdict, stage_wrong,
	"Can't override parameters with cdict attached (some must "
	"be inherited from the cdict).");

	cctx->requestedParams = *params;
	return 0;
	}

	ZSTDLIB_API size_t ZSTD_CCtx_setPledgedSrcSize(ZSTD_CCtx* cctx, unsigned long long pledgedSrcSize)
	{
	DEBUGLOG(4, "ZSTD_CCtx_setPledgedSrcSize to %u bytes", (U32)pledgedSrcSize);
	RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
	"Can't set pledgedSrcSize when not in init stage.");
	cctx->pledgedSrcSizePlusOne = pledgedSrcSize+1;
	return 0;
	}

	/**
	* Initializes the local dict using the requested parameters.
	* NOTE: This does not use the pledged src size, because it may be used for more
	* than one compression.
	*/
	static size_t ZSTD_initLocalDict(ZSTD_CCtx* cctx)
	{
	ZSTD_localDict* const dl = &cctx->localDict;
	ZSTD_compressionParameters const cParams = ZSTD_getCParamsFromCCtxParams(
	&cctx->requestedParams, ZSTD_CONTENTSIZE_UNKNOWN, dl->dictSize);
	if (dl->dict == NULL) {
	/* No local dictionary. */
	assert(dl->dictBuffer == NULL);
	assert(dl->cdict == NULL);
	assert(dl->dictSize == 0);
	return 0;
	}
	if (dl->cdict != NULL) {
	assert(cctx->cdict == dl->cdict);
	/* Local dictionary already initialized. */
	return 0;
	}
	assert(dl->dictSize > 0);
	assert(cctx->cdict == NULL);
	assert(cctx->prefixDict.dict == NULL);

	dl->cdict = ZSTD_createCDict_advanced(
	dl->dict,
	dl->dictSize,
	ZSTD_dlm_byRef,
	dl->dictContentType,
	cParams,
	cctx->customMem);
	RETURN_ERROR_IF(!dl->cdict, memory_allocation, "ZSTD_createCDict_advanced failed");
	cctx->cdict = dl->cdict;
	return 0;
	}

	size_t ZSTD_CCtx_loadDictionary_advanced(
	ZSTD_CCtx* cctx, const void* dict, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType)
	{
	RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
	"Can't load a dictionary when ctx is not in init stage.");
	RETURN_ERROR_IF(cctx->staticSize, memory_allocation,
	"no malloc for static CCtx");
	DEBUGLOG(4, "ZSTD_CCtx_loadDictionary_advanced (size: %u)", (U32)dictSize);
	ZSTD_clearAllDicts(cctx); /* in case one already exists */
	if (dict == NULL \|\| dictSize == 0) /* no dictionary mode */
	return 0;
	if (dictLoadMethod == ZSTD_dlm_byRef) {
	cctx->localDict.dict = dict;
	} else {
	void* dictBuffer = ZSTD_malloc(dictSize, cctx->customMem);
	RETURN_ERROR_IF(!dictBuffer, memory_allocation, "NULL pointer!");
	memcpy(dictBuffer, dict, dictSize);
	cctx->localDict.dictBuffer = dictBuffer;
	cctx->localDict.dict = dictBuffer;
	}
	cctx->localDict.dictSize = dictSize;
	cctx->localDict.dictContentType = dictContentType;
	return 0;
	}

	ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary_byReference(
	ZSTD_CCtx* cctx, const void* dict, size_t dictSize)
	{
	return ZSTD_CCtx_loadDictionary_advanced(
	cctx, dict, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto);
	}

	ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, size_t dictSize)
	{
	return ZSTD_CCtx_loadDictionary_advanced(
	cctx, dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto);
	}


	size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict)
	{
	RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
	"Can't ref a dict when ctx not in init stage.");
	/* Free the existing local cdict (if any) to save memory. */
	ZSTD_clearAllDicts(cctx);
	cctx->cdict = cdict;
	return 0;
	}

	size_t ZSTD_CCtx_refPrefix(ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize)
	{
	return ZSTD_CCtx_refPrefix_advanced(cctx, prefix, prefixSize, ZSTD_dct_rawContent);
	}

	size_t ZSTD_CCtx_refPrefix_advanced(
	ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType)
	{
	RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
	"Can't ref a prefix when ctx not in init stage.");
	ZSTD_clearAllDicts(cctx);
	if (prefix != NULL && prefixSize > 0) {
	cctx->prefixDict.dict = prefix;
	cctx->prefixDict.dictSize = prefixSize;
	cctx->prefixDict.dictContentType = dictContentType;
	}
	return 0;
	}

	/*! ZSTD_CCtx_reset() :
	* Also dumps dictionary */
	size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx, ZSTD_ResetDirective reset)
	{
	if ( (reset == ZSTD_reset_session_only)
	\|\| (reset == ZSTD_reset_session_and_parameters) ) {
	cctx->streamStage = zcss_init;
	cctx->pledgedSrcSizePlusOne = 0;
	}
	if ( (reset == ZSTD_reset_parameters)
	\|\| (reset == ZSTD_reset_session_and_parameters) ) {
	RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
	"Can't reset parameters only when not in init stage.");
	ZSTD_clearAllDicts(cctx);
	return ZSTD_CCtxParams_reset(&cctx->requestedParams);
	}
	return 0;
	}


	/** ZSTD_checkCParams() :
	control CParam values remain within authorized range.
	@return : 0, or an error code if one value is beyond authorized range */
	size_t ZSTD_checkCParams(ZSTD_compressionParameters cParams)
	{
	BOUNDCHECK(ZSTD_c_windowLog, (int)cParams.windowLog);
	BOUNDCHECK(ZSTD_c_chainLog, (int)cParams.chainLog);
	BOUNDCHECK(ZSTD_c_hashLog, (int)cParams.hashLog);
	BOUNDCHECK(ZSTD_c_searchLog, (int)cParams.searchLog);
	BOUNDCHECK(ZSTD_c_minMatch, (int)cParams.minMatch);
	BOUNDCHECK(ZSTD_c_targetLength,(int)cParams.targetLength);
	BOUNDCHECK(ZSTD_c_strategy, cParams.strategy);
	return 0;
	}

	/** ZSTD_clampCParams() :
	* make CParam values within valid range.
	* @return : valid CParams */
	static ZSTD_compressionParameters
	ZSTD_clampCParams(ZSTD_compressionParameters cParams)
	{
	# define CLAMP_TYPE(cParam, val, type) { \
	ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam); \
	if ((int)val<bounds.lowerBound) val=(type)bounds.lowerBound; \
	else if ((int)val>bounds.upperBound) val=(type)bounds.upperBound; \
	}
	# define CLAMP(cParam, val) CLAMP_TYPE(cParam, val, unsigned)
	CLAMP(ZSTD_c_windowLog, cParams.windowLog);
	CLAMP(ZSTD_c_chainLog, cParams.chainLog);
	CLAMP(ZSTD_c_hashLog, cParams.hashLog);
	CLAMP(ZSTD_c_searchLog, cParams.searchLog);
	CLAMP(ZSTD_c_minMatch, cParams.minMatch);
	CLAMP(ZSTD_c_targetLength,cParams.targetLength);
	CLAMP_TYPE(ZSTD_c_strategy,cParams.strategy, ZSTD_strategy);
	return cParams;
	}

	/** ZSTD_cycleLog() :
	* condition for correct operation : hashLog > 1 */
	U32 ZSTD_cycleLog(U32 hashLog, ZSTD_strategy strat)
	{
	U32 const btScale = ((U32)strat >= (U32)ZSTD_btlazy2);
	return hashLog - btScale;
	}

	/** ZSTD_adjustCParams_internal() :
	* optimize `cPar` for a specified input (`srcSize` and `dictSize`).
	* mostly downsize to reduce memory consumption and initialization latency.
	* `srcSize` can be ZSTD_CONTENTSIZE_UNKNOWN when not known.
	* note : `srcSize==0` means 0!
	* condition : cPar is presumed validated (can be checked using ZSTD_checkCParams()). */
	static ZSTD_compressionParameters
	ZSTD_adjustCParams_internal(ZSTD_compressionParameters cPar,
	unsigned long long srcSize,
	size_t dictSize)
	{
	static const U64 minSrcSize = 513; /* (1<<9) + 1 */
	static const U64 maxWindowResize = 1ULL << (ZSTD_WINDOWLOG_MAX-1);
	assert(ZSTD_checkCParams(cPar)==0);

	if (dictSize && srcSize == ZSTD_CONTENTSIZE_UNKNOWN)
	srcSize = minSrcSize;

	/* resize windowLog if input is small enough, to use less memory */
	if ( (srcSize < maxWindowResize)
	&& (dictSize < maxWindowResize) ) {
	U32 const tSize = (U32)(srcSize + dictSize);
	static U32 const hashSizeMin = 1 << ZSTD_HASHLOG_MIN;
	U32 const srcLog = (tSize < hashSizeMin) ? ZSTD_HASHLOG_MIN :
	ZSTD_highbit32(tSize-1) + 1;
	if (cPar.windowLog > srcLog) cPar.windowLog = srcLog;
	}
	if (cPar.hashLog > cPar.windowLog+1) cPar.hashLog = cPar.windowLog+1;
	{ U32 const cycleLog = ZSTD_cycleLog(cPar.chainLog, cPar.strategy);
	if (cycleLog > cPar.windowLog)
	cPar.chainLog -= (cycleLog - cPar.windowLog);
	}

	if (cPar.windowLog < ZSTD_WINDOWLOG_ABSOLUTEMIN)
	cPar.windowLog = ZSTD_WINDOWLOG_ABSOLUTEMIN; /* minimum wlog required for valid frame header */

	return cPar;
	}

	ZSTD_compressionParameters
	ZSTD_adjustCParams(ZSTD_compressionParameters cPar,
	unsigned long long srcSize,
	size_t dictSize)
	{
	cPar = ZSTD_clampCParams(cPar); /* resulting cPar is necessarily valid (all parameters within range) */
	if (srcSize == 0) srcSize = ZSTD_CONTENTSIZE_UNKNOWN;
	return ZSTD_adjustCParams_internal(cPar, srcSize, dictSize);
	}

	static ZSTD_compressionParameters ZSTD_getCParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize);
	static ZSTD_parameters ZSTD_getParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize);

	ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams(
	const ZSTD_CCtx_params* CCtxParams, U64 srcSizeHint, size_t dictSize)
	{
	ZSTD_compressionParameters cParams;
	if (srcSizeHint == ZSTD_CONTENTSIZE_UNKNOWN && CCtxParams->srcSizeHint > 0) {
	srcSizeHint = CCtxParams->srcSizeHint;
	}
	cParams = ZSTD_getCParams_internal(CCtxParams->compressionLevel, srcSizeHint, dictSize);
	if (CCtxParams->ldmParams.enableLdm) cParams.windowLog = ZSTD_LDM_DEFAULT_WINDOW_LOG;
	if (CCtxParams->cParams.windowLog) cParams.windowLog = CCtxParams->cParams.windowLog;
	if (CCtxParams->cParams.hashLog) cParams.hashLog = CCtxParams->cParams.hashLog;
	if (CCtxParams->cParams.chainLog) cParams.chainLog = CCtxParams->cParams.chainLog;
	if (CCtxParams->cParams.searchLog) cParams.searchLog = CCtxParams->cParams.searchLog;
	if (CCtxParams->cParams.minMatch) cParams.minMatch = CCtxParams->cParams.minMatch;
	if (CCtxParams->cParams.targetLength) cParams.targetLength = CCtxParams->cParams.targetLength;
	if (CCtxParams->cParams.strategy) cParams.strategy = CCtxParams->cParams.strategy;
	assert(!ZSTD_checkCParams(cParams));
	/* srcSizeHint == 0 means 0 */
	return ZSTD_adjustCParams_internal(cParams, srcSizeHint, dictSize);
	}

	static size_t
	ZSTD_sizeof_matchState(const ZSTD_compressionParameters* const cParams,
	const U32 forCCtx)
	{
	size_t const chainSize = (cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cParams->chainLog);
	size_t const hSize = ((size_t)1) << cParams->hashLog;
	U32 const hashLog3 = (forCCtx && cParams->minMatch==3) ? MIN(ZSTD_HASHLOG3_MAX, cParams->windowLog) : 0;
	size_t const h3Size = hashLog3 ? ((size_t)1) << hashLog3 : 0;
	/* We don't use ZSTD_cwksp_alloc_size() here because the tables aren't
	* surrounded by redzones in ASAN. */
	size_t const tableSpace = chainSize * sizeof(U32)
	+ hSize * sizeof(U32)
	+ h3Size * sizeof(U32);
	size_t const optPotentialSpace =
	ZSTD_cwksp_alloc_size((MaxML+1) * sizeof(U32))
	+ ZSTD_cwksp_alloc_size((MaxLL+1) * sizeof(U32))
	+ ZSTD_cwksp_alloc_size((MaxOff+1) * sizeof(U32))
	+ ZSTD_cwksp_alloc_size((1<<Litbits) * sizeof(U32))
	+ ZSTD_cwksp_alloc_size((ZSTD_OPT_NUM+1) * sizeof(ZSTD_match_t))
	+ ZSTD_cwksp_alloc_size((ZSTD_OPT_NUM+1) * sizeof(ZSTD_optimal_t));
	size_t const optSpace = (forCCtx && (cParams->strategy >= ZSTD_btopt))
	? optPotentialSpace
	: 0;
	DEBUGLOG(4, "chainSize: %u - hSize: %u - h3Size: %u",
	(U32)chainSize, (U32)hSize, (U32)h3Size);
	return tableSpace + optSpace;
	}

	size_t ZSTD_estimateCCtxSize_usingCCtxParams(const ZSTD_CCtx_params* params)
	{
	RETURN_ERROR_IF(params->nbWorkers > 0, GENERIC, "Estimate CCtx size is supported for single-threaded compression only.");
	{ ZSTD_compressionParameters const cParams =
	ZSTD_getCParamsFromCCtxParams(params, ZSTD_CONTENTSIZE_UNKNOWN, 0);
	size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, (size_t)1 << cParams.windowLog);
	U32 const divider = (cParams.minMatch==3) ? 3 : 4;
	size_t const maxNbSeq = blockSize / divider;
	size_t const tokenSpace = ZSTD_cwksp_alloc_size(WILDCOPY_OVERLENGTH + blockSize)
	+ ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(seqDef))
	+ 3 * ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(BYTE));
	size_t const entropySpace = ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE);
	size_t const blockStateSpace = 2 * ZSTD_cwksp_alloc_size(sizeof(ZSTD_compressedBlockState_t));
	size_t const matchStateSize = ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 1);

	size_t const ldmSpace = ZSTD_ldm_getTableSize(params->ldmParams);
	size_t const ldmSeqSpace = ZSTD_cwksp_alloc_size(ZSTD_ldm_getMaxNbSeq(params->ldmParams, blockSize) * sizeof(rawSeq));

	/* estimateCCtxSize is for one-shot compression. So no buffers should
	* be needed. However, we still allocate two 0-sized buffers, which can
	* take space under ASAN. */
	size_t const bufferSpace = ZSTD_cwksp_alloc_size(0)
	+ ZSTD_cwksp_alloc_size(0);

	size_t const cctxSpace = ZSTD_cwksp_alloc_size(sizeof(ZSTD_CCtx));

	size_t const neededSpace =
	cctxSpace +
	entropySpace +
	blockStateSpace +
	ldmSpace +
	ldmSeqSpace +
	matchStateSize +
	tokenSpace +
	bufferSpace;

	DEBUGLOG(5, "estimate workspace : %u", (U32)neededSpace);
	return neededSpace;
	}
	}

	size_t ZSTD_estimateCCtxSize_usingCParams(ZSTD_compressionParameters cParams)
	{
	ZSTD_CCtx_params const params = ZSTD_makeCCtxParamsFromCParams(cParams);
	return ZSTD_estimateCCtxSize_usingCCtxParams(&params);
	}

	static size_t ZSTD_estimateCCtxSize_internal(int compressionLevel)
	{
	ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, 0);
	return ZSTD_estimateCCtxSize_usingCParams(cParams);
	}

	size_t ZSTD_estimateCCtxSize(int compressionLevel)
	{
	int level;
	size_t memBudget = 0;
	for (level=MIN(compressionLevel, 1); level<=compressionLevel; level++) {
	size_t const newMB = ZSTD_estimateCCtxSize_internal(level);
	if (newMB > memBudget) memBudget = newMB;
	}
	return memBudget;
	}

	size_t ZSTD_estimateCStreamSize_usingCCtxParams(const ZSTD_CCtx_params* params)
	{
	RETURN_ERROR_IF(params->nbWorkers > 0, GENERIC, "Estimate CCtx size is supported for single-threaded compression only.");
	{ ZSTD_compressionParameters const cParams =
	ZSTD_getCParamsFromCCtxParams(params, ZSTD_CONTENTSIZE_UNKNOWN, 0);
	size_t const CCtxSize = ZSTD_estimateCCtxSize_usingCCtxParams(params);
	size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, (size_t)1 << cParams.windowLog);
	size_t const inBuffSize = ((size_t)1 << cParams.windowLog) + blockSize;
	size_t const outBuffSize = ZSTD_compressBound(blockSize) + 1;
	size_t const streamingSize = ZSTD_cwksp_alloc_size(inBuffSize)
	+ ZSTD_cwksp_alloc_size(outBuffSize);

	return CCtxSize + streamingSize;
	}
	}

	size_t ZSTD_estimateCStreamSize_usingCParams(ZSTD_compressionParameters cParams)
	{
	ZSTD_CCtx_params const params = ZSTD_makeCCtxParamsFromCParams(cParams);
	return ZSTD_estimateCStreamSize_usingCCtxParams(&params);
	}

	static size_t ZSTD_estimateCStreamSize_internal(int compressionLevel)
	{
	ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, 0);
	return ZSTD_estimateCStreamSize_usingCParams(cParams);
	}

	size_t ZSTD_estimateCStreamSize(int compressionLevel)
	{
	int level;
	size_t memBudget = 0;
	for (level=MIN(compressionLevel, 1); level<=compressionLevel; level++) {
	size_t const newMB = ZSTD_estimateCStreamSize_internal(level);
	if (newMB > memBudget) memBudget = newMB;
	}
	return memBudget;
	}

	/* ZSTD_getFrameProgression():
	* tells how much data has been consumed (input) and produced (output) for current frame.
	* able to count progression inside worker threads (non-blocking mode).
	*/
	ZSTD_frameProgression ZSTD_getFrameProgression(const ZSTD_CCtx* cctx)
	{
	#ifdef ZSTD_MULTITHREAD
	if (cctx->appliedParams.nbWorkers > 0) {
	return ZSTDMT_getFrameProgression(cctx->mtctx);
	}
	#endif
	{ ZSTD_frameProgression fp;
	size_t const buffered = (cctx->inBuff == NULL) ? 0 :
	cctx->inBuffPos - cctx->inToCompress;
	if (buffered) assert(cctx->inBuffPos >= cctx->inToCompress);
	assert(buffered <= ZSTD_BLOCKSIZE_MAX);
	fp.ingested = cctx->consumedSrcSize + buffered;
	fp.consumed = cctx->consumedSrcSize;
	fp.produced = cctx->producedCSize;
	fp.flushed = cctx->producedCSize; /* simplified; some data might still be left within streaming output buffer */
	fp.currentJobID = 0;
	fp.nbActiveWorkers = 0;
	return fp;
	} }

	/*! ZSTD_toFlushNow()
	* Only useful for multithreading scenarios currently (nbWorkers >= 1).
	*/
	size_t ZSTD_toFlushNow(ZSTD_CCtx* cctx)
	{
	#ifdef ZSTD_MULTITHREAD
	if (cctx->appliedParams.nbWorkers > 0) {
	return ZSTDMT_toFlushNow(cctx->mtctx);
	}
	#endif
	(void)cctx;
	return 0; /* over-simplification; could also check if context is currently running in streaming mode, and in which case, report how many bytes are left to be flushed within output buffer */
	}

	static void ZSTD_assertEqualCParams(ZSTD_compressionParameters cParams1,
	ZSTD_compressionParameters cParams2)
	{
	(void)cParams1;
	(void)cParams2;
	assert(cParams1.windowLog == cParams2.windowLog);
	assert(cParams1.chainLog == cParams2.chainLog);
	assert(cParams1.hashLog == cParams2.hashLog);
	assert(cParams1.searchLog == cParams2.searchLog);
	assert(cParams1.minMatch == cParams2.minMatch);
	assert(cParams1.targetLength == cParams2.targetLength);
	assert(cParams1.strategy == cParams2.strategy);
	}

	void ZSTD_reset_compressedBlockState(ZSTD_compressedBlockState_t* bs)
	{
	int i;
	for (i = 0; i < ZSTD_REP_NUM; ++i)
	bs->rep[i] = repStartValue[i];
	bs->entropy.huf.repeatMode = HUF_repeat_none;
	bs->entropy.fse.offcode_repeatMode = FSE_repeat_none;
	bs->entropy.fse.matchlength_repeatMode = FSE_repeat_none;
	bs->entropy.fse.litlength_repeatMode = FSE_repeat_none;
	}

	/*! ZSTD_invalidateMatchState()
	* Invalidate all the matches in the match finder tables.
	* Requires nextSrc and base to be set (can be NULL).
	*/
	static void ZSTD_invalidateMatchState(ZSTD_matchState_t* ms)
	{
	ZSTD_window_clear(&ms->window);

	ms->nextToUpdate = ms->window.dictLimit;
	ms->loadedDictEnd = 0;
	ms->opt.litLengthSum = 0; /* force reset of btopt stats */
	ms->dictMatchState = NULL;
	}

	/**
	* Indicates whether this compression proceeds directly from user-provided
	* source buffer to user-provided destination buffer (ZSTDb_not_buffered), or
	* whether the context needs to buffer the input/output (ZSTDb_buffered).
	*/
	typedef enum {
	ZSTDb_not_buffered,
	ZSTDb_buffered
	} ZSTD_buffered_policy_e;

	/**
	* Controls, for this matchState reset, whether the tables need to be cleared /
	* prepared for the coming compression (ZSTDcrp_makeClean), or whether the
	* tables can be left unclean (ZSTDcrp_leaveDirty), because we know that a
	* subsequent operation will overwrite the table space anyways (e.g., copying
	* the matchState contents in from a CDict).
	*/
	typedef enum {
	ZSTDcrp_makeClean,
	ZSTDcrp_leaveDirty
	} ZSTD_compResetPolicy_e;

	/**
	* Controls, for this matchState reset, whether indexing can continue where it
	* left off (ZSTDirp_continue), or whether it needs to be restarted from zero
	* (ZSTDirp_reset).
	*/
	typedef enum {
	ZSTDirp_continue,
	ZSTDirp_reset
	} ZSTD_indexResetPolicy_e;

	typedef enum {
	ZSTD_resetTarget_CDict,
	ZSTD_resetTarget_CCtx
	} ZSTD_resetTarget_e;

	static size_t
	ZSTD_reset_matchState(ZSTD_matchState_t* ms,
	ZSTD_cwksp* ws,
	const ZSTD_compressionParameters* cParams,
	const ZSTD_compResetPolicy_e crp,
	const ZSTD_indexResetPolicy_e forceResetIndex,
	const ZSTD_resetTarget_e forWho)
	{
	size_t const chainSize = (cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cParams->chainLog);
	size_t const hSize = ((size_t)1) << cParams->hashLog;
	U32 const hashLog3 = ((forWho == ZSTD_resetTarget_CCtx) && cParams->minMatch==3) ? MIN(ZSTD_HASHLOG3_MAX, cParams->windowLog) : 0;
	size_t const h3Size = hashLog3 ? ((size_t)1) << hashLog3 : 0;

	DEBUGLOG(4, "reset indices : %u", forceResetIndex == ZSTDirp_reset);
	if (forceResetIndex == ZSTDirp_reset) {
	ZSTD_window_init(&ms->window);
	ZSTD_cwksp_mark_tables_dirty(ws);
	}

	ms->hashLog3 = hashLog3;

	ZSTD_invalidateMatchState(ms);

	assert(!ZSTD_cwksp_reserve_failed(ws)); /* check that allocation hasn't already failed */

	ZSTD_cwksp_clear_tables(ws);

	DEBUGLOG(5, "reserving table space");
	/* table Space */
	ms->hashTable = (U32)ZSTD_cwksp_reserve_table(ws, hSize sizeof(U32));
	ms->chainTable = (U32)ZSTD_cwksp_reserve_table(ws, chainSize sizeof(U32));
	ms->hashTable3 = (U32)ZSTD_cwksp_reserve_table(ws, h3Size sizeof(U32));
	RETURN_ERROR_IF(ZSTD_cwksp_reserve_failed(ws), memory_allocation,
	"failed a workspace allocation in ZSTD_reset_matchState");

	DEBUGLOG(4, "reset table : %u", crp!=ZSTDcrp_leaveDirty);
	if (crp!=ZSTDcrp_leaveDirty) {
	/* reset tables only */
	ZSTD_cwksp_clean_tables(ws);
	}

	/* opt parser space */
	if ((forWho == ZSTD_resetTarget_CCtx) && (cParams->strategy >= ZSTD_btopt)) {
	DEBUGLOG(4, "reserving optimal parser space");
	ms->opt.litFreq = (unsigned)ZSTD_cwksp_reserve_aligned(ws, (1<<Litbits) sizeof(unsigned));
	ms->opt.litLengthFreq = (unsigned)ZSTD_cwksp_reserve_aligned(ws, (MaxLL+1) sizeof(unsigned));
	ms->opt.matchLengthFreq = (unsigned)ZSTD_cwksp_reserve_aligned(ws, (MaxML+1) sizeof(unsigned));
	ms->opt.offCodeFreq = (unsigned)ZSTD_cwksp_reserve_aligned(ws, (MaxOff+1) sizeof(unsigned));
	ms->opt.matchTable = (ZSTD_match_t)ZSTD_cwksp_reserve_aligned(ws, (ZSTD_OPT_NUM+1) sizeof(ZSTD_match_t));
	ms->opt.priceTable = (ZSTD_optimal_t)ZSTD_cwksp_reserve_aligned(ws, (ZSTD_OPT_NUM+1) sizeof(ZSTD_optimal_t));
	}

	ms->cParams = *cParams;

	RETURN_ERROR_IF(ZSTD_cwksp_reserve_failed(ws), memory_allocation,
	"failed a workspace allocation in ZSTD_reset_matchState");

	return 0;
	}

	/* ZSTD_indexTooCloseToMax() :
	* minor optimization : prefer memset() rather than reduceIndex()
	* which is measurably slow in some circumstances (reported for Visual Studio).
	* Works when re-using a context for a lot of smallish inputs :
	* if all inputs are smaller than ZSTD_INDEXOVERFLOW_MARGIN,
	* memset() will be triggered before reduceIndex().
	*/
	#define ZSTD_INDEXOVERFLOW_MARGIN (16 MB)
	static int ZSTD_indexTooCloseToMax(ZSTD_window_t w)
	{
	return (size_t)(w.nextSrc - w.base) > (ZSTD_CURRENT_MAX - ZSTD_INDEXOVERFLOW_MARGIN);
	}

	/*! ZSTD_resetCCtx_internal() :
	note : `params` are assumed fully validated at this stage */
	static size_t ZSTD_resetCCtx_internal(ZSTD_CCtx* zc,
	ZSTD_CCtx_params params,
	U64 const pledgedSrcSize,
	ZSTD_compResetPolicy_e const crp,
	ZSTD_buffered_policy_e const zbuff)
	{
	ZSTD_cwksp* const ws = &zc->workspace;
	DEBUGLOG(4, "ZSTD_resetCCtx_internal: pledgedSrcSize=%u, wlog=%u",
	(U32)pledgedSrcSize, params.cParams.windowLog);
	assert(!ZSTD_isError(ZSTD_checkCParams(params.cParams)));

	zc->isFirstBlock = 1;

	if (params.ldmParams.enableLdm) {
	/* Adjust long distance matching parameters */
	ZSTD_ldm_adjustParameters(&params.ldmParams, &params.cParams);
	assert(params.ldmParams.hashLog >= params.ldmParams.bucketSizeLog);
	assert(params.ldmParams.hashRateLog < 32);
	zc->ldmState.hashPower = ZSTD_rollingHash_primePower(params.ldmParams.minMatchLength);
	}

	{ size_t const windowSize = MAX(1, (size_t)MIN(((U64)1 << params.cParams.windowLog), pledgedSrcSize));
	size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, windowSize);
	U32 const divider = (params.cParams.minMatch==3) ? 3 : 4;
	size_t const maxNbSeq = blockSize / divider;
	size_t const tokenSpace = ZSTD_cwksp_alloc_size(WILDCOPY_OVERLENGTH + blockSize)
	+ ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(seqDef))
	+ 3 * ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(BYTE));
	size_t const buffOutSize = (zbuff==ZSTDb_buffered) ? ZSTD_compressBound(blockSize)+1 : 0;
	size_t const buffInSize = (zbuff==ZSTDb_buffered) ? windowSize + blockSize : 0;
	size_t const matchStateSize = ZSTD_sizeof_matchState(&params.cParams, /* forCCtx */ 1);
	size_t const maxNbLdmSeq = ZSTD_ldm_getMaxNbSeq(params.ldmParams, blockSize);

	ZSTD_indexResetPolicy_e needsIndexReset = zc->initialized ? ZSTDirp_continue : ZSTDirp_reset;

	if (ZSTD_indexTooCloseToMax(zc->blockState.matchState.window)) {
	needsIndexReset = ZSTDirp_reset;
	}

	if (!zc->staticSize) ZSTD_cwksp_bump_oversized_duration(ws, 0);

	/* Check if workspace is large enough, alloc a new one if needed */
	{ size_t const cctxSpace = zc->staticSize ? ZSTD_cwksp_alloc_size(sizeof(ZSTD_CCtx)) : 0;
	size_t const entropySpace = ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE);
	size_t const blockStateSpace = 2 * ZSTD_cwksp_alloc_size(sizeof(ZSTD_compressedBlockState_t));
	size_t const bufferSpace = ZSTD_cwksp_alloc_size(buffInSize) + ZSTD_cwksp_alloc_size(buffOutSize);
	size_t const ldmSpace = ZSTD_ldm_getTableSize(params.ldmParams);
	size_t const ldmSeqSpace = ZSTD_cwksp_alloc_size(maxNbLdmSeq * sizeof(rawSeq));

	size_t const neededSpace =
	cctxSpace +
	entropySpace +
	blockStateSpace +
	ldmSpace +
	ldmSeqSpace +
	matchStateSize +
	tokenSpace +
	bufferSpace;

	int const workspaceTooSmall = ZSTD_cwksp_sizeof(ws) < neededSpace;
	int const workspaceWasteful = ZSTD_cwksp_check_wasteful(ws, neededSpace);

	DEBUGLOG(4, "Need %zuKB workspace, including %zuKB for match state, and %zuKB for buffers",
	neededSpace>>10, matchStateSize>>10, bufferSpace>>10);
	DEBUGLOG(4, "windowSize: %zu - blockSize: %zu", windowSize, blockSize);

	if (workspaceTooSmall \|\| workspaceWasteful) {
	DEBUGLOG(4, "Resize workspaceSize from %zuKB to %zuKB",
	ZSTD_cwksp_sizeof(ws) >> 10,
	neededSpace >> 10);

	RETURN_ERROR_IF(zc->staticSize, memory_allocation, "static cctx : no resize");

	needsIndexReset = ZSTDirp_reset;

	ZSTD_cwksp_free(ws, zc->customMem);
	FORWARD_IF_ERROR(ZSTD_cwksp_create(ws, neededSpace, zc->customMem), "");

	DEBUGLOG(5, "reserving object space");
	/* Statically sized space.
	* entropyWorkspace never moves,
	* though prev/next block swap places */
	assert(ZSTD_cwksp_check_available(ws, 2 * sizeof(ZSTD_compressedBlockState_t)));
	zc->blockState.prevCBlock = (ZSTD_compressedBlockState_t*) ZSTD_cwksp_reserve_object(ws, sizeof(ZSTD_compressedBlockState_t));
	RETURN_ERROR_IF(zc->blockState.prevCBlock == NULL, memory_allocation, "couldn't allocate prevCBlock");
	zc->blockState.nextCBlock = (ZSTD_compressedBlockState_t*) ZSTD_cwksp_reserve_object(ws, sizeof(ZSTD_compressedBlockState_t));
	RETURN_ERROR_IF(zc->blockState.nextCBlock == NULL, memory_allocation, "couldn't allocate nextCBlock");
	zc->entropyWorkspace = (U32*) ZSTD_cwksp_reserve_object(ws, HUF_WORKSPACE_SIZE);
	RETURN_ERROR_IF(zc->blockState.nextCBlock == NULL, memory_allocation, "couldn't allocate entropyWorkspace");
	} }

	ZSTD_cwksp_clear(ws);

	/* init params */
	zc->appliedParams = params;
	zc->blockState.matchState.cParams = params.cParams;
	zc->pledgedSrcSizePlusOne = pledgedSrcSize+1;
	zc->consumedSrcSize = 0;
	zc->producedCSize = 0;
	if (pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN)
	zc->appliedParams.fParams.contentSizeFlag = 0;
	DEBUGLOG(4, "pledged content size : %u ; flag : %u",
	(unsigned)pledgedSrcSize, zc->appliedParams.fParams.contentSizeFlag);
	zc->blockSize = blockSize;

	XXH64_reset(&zc->xxhState, 0);
	zc->stage = ZSTDcs_init;
	zc->dictID = 0;

	ZSTD_reset_compressedBlockState(zc->blockState.prevCBlock);

	/* ZSTD_wildcopy() is used to copy into the literals buffer,
	* so we have to oversize the buffer by WILDCOPY_OVERLENGTH bytes.
	*/
	zc->seqStore.litStart = ZSTD_cwksp_reserve_buffer(ws, blockSize + WILDCOPY_OVERLENGTH);
	zc->seqStore.maxNbLit = blockSize;

	/* buffers */
	zc->inBuffSize = buffInSize;
	zc->inBuff = (char*)ZSTD_cwksp_reserve_buffer(ws, buffInSize);
	zc->outBuffSize = buffOutSize;
	zc->outBuff = (char*)ZSTD_cwksp_reserve_buffer(ws, buffOutSize);

	/* ldm bucketOffsets table */
	if (params.ldmParams.enableLdm) {
	/* TODO: avoid memset? */
	size_t const ldmBucketSize =
	((size_t)1) << (params.ldmParams.hashLog -
	params.ldmParams.bucketSizeLog);
	zc->ldmState.bucketOffsets = ZSTD_cwksp_reserve_buffer(ws, ldmBucketSize);
	memset(zc->ldmState.bucketOffsets, 0, ldmBucketSize);
	}

	/* sequences storage */
	ZSTD_referenceExternalSequences(zc, NULL, 0);
	zc->seqStore.maxNbSeq = maxNbSeq;
	zc->seqStore.llCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE));
	zc->seqStore.mlCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE));
	zc->seqStore.ofCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE));
	zc->seqStore.sequencesStart = (seqDef)ZSTD_cwksp_reserve_aligned(ws, maxNbSeq sizeof(seqDef));

	FORWARD_IF_ERROR(ZSTD_reset_matchState(
	&zc->blockState.matchState,
	ws,
	&params.cParams,
	crp,
	needsIndexReset,
	ZSTD_resetTarget_CCtx), "");

	/* ldm hash table */
	if (params.ldmParams.enableLdm) {
	/* TODO: avoid memset? */
	size_t const ldmHSize = ((size_t)1) << params.ldmParams.hashLog;
	zc->ldmState.hashTable = (ldmEntry_t)ZSTD_cwksp_reserve_aligned(ws, ldmHSize sizeof(ldmEntry_t));
	memset(zc->ldmState.hashTable, 0, ldmHSize * sizeof(ldmEntry_t));
	zc->ldmSequences = (rawSeq)ZSTD_cwksp_reserve_aligned(ws, maxNbLdmSeq sizeof(rawSeq));
	zc->maxNbLdmSequences = maxNbLdmSeq;

	ZSTD_window_init(&zc->ldmState.window);
	ZSTD_window_clear(&zc->ldmState.window);
	zc->ldmState.loadedDictEnd = 0;
	}

	DEBUGLOG(3, "wksp: finished allocating, %zd bytes remain available", ZSTD_cwksp_available_space(ws));
	zc->initialized = 1;

	return 0;
	}
	}

	/* ZSTD_invalidateRepCodes() :
	* ensures next compression will not use repcodes from previous block.
	* Note : only works with regular variant;
	* do not use with extDict variant ! */
	void ZSTD_invalidateRepCodes(ZSTD_CCtx* cctx) {
	int i;
	for (i=0; i<ZSTD_REP_NUM; i++) cctx->blockState.prevCBlock->rep[i] = 0;
	assert(!ZSTD_window_hasExtDict(cctx->blockState.matchState.window));
	}

	/* These are the approximate sizes for each strategy past which copying the
	* dictionary tables into the working context is faster than using them
	* in-place.
	*/
	static const size_t attachDictSizeCutoffs[ZSTD_STRATEGY_MAX+1] = {
	8 KB, /* unused */
	8 KB, /* ZSTD_fast */
	16 KB, /* ZSTD_dfast */
	32 KB, /* ZSTD_greedy */
	32 KB, /* ZSTD_lazy */
	32 KB, /* ZSTD_lazy2 */
	32 KB, /* ZSTD_btlazy2 */
	32 KB, /* ZSTD_btopt */
	8 KB, /* ZSTD_btultra */
	8 KB /* ZSTD_btultra2 */
	};

	static int ZSTD_shouldAttachDict(const ZSTD_CDict* cdict,
	const ZSTD_CCtx_params* params,
	U64 pledgedSrcSize)
	{
	size_t cutoff = attachDictSizeCutoffs[cdict->matchState.cParams.strategy];
	return ( pledgedSrcSize <= cutoff
	\|\| pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN
	\|\| params->attachDictPref == ZSTD_dictForceAttach )
	&& params->attachDictPref != ZSTD_dictForceCopy
	&& !params->forceWindow; /* dictMatchState isn't correctly
	* handled in _enforceMaxDist */
	}

	static size_t
	ZSTD_resetCCtx_byAttachingCDict(ZSTD_CCtx* cctx,
	const ZSTD_CDict* cdict,
	ZSTD_CCtx_params params,
	U64 pledgedSrcSize,
	ZSTD_buffered_policy_e zbuff)
	{
	{ const ZSTD_compressionParameters* const cdict_cParams = &cdict->matchState.cParams;
	unsigned const windowLog = params.cParams.windowLog;
	assert(windowLog != 0);
	/* Resize working context table params for input only, since the dict
	* has its own tables. */
	/* pledgeSrcSize == 0 means 0! */
	params.cParams = ZSTD_adjustCParams_internal(*cdict_cParams, pledgedSrcSize, 0);
	params.cParams.windowLog = windowLog;
	FORWARD_IF_ERROR(ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize,
	ZSTDcrp_makeClean, zbuff), "");
	assert(cctx->appliedParams.cParams.strategy == cdict_cParams->strategy);
	}

	{ const U32 cdictEnd = (U32)( cdict->matchState.window.nextSrc
	- cdict->matchState.window.base);
	const U32 cdictLen = cdictEnd - cdict->matchState.window.dictLimit;
	if (cdictLen == 0) {
	/* don't even attach dictionaries with no contents */
	DEBUGLOG(4, "skipping attaching empty dictionary");
	} else {
	DEBUGLOG(4, "attaching dictionary into context");
	cctx->blockState.matchState.dictMatchState = &cdict->matchState;

	/* prep working match state so dict matches never have negative indices
	* when they are translated to the working context's index space. */
	if (cctx->blockState.matchState.window.dictLimit < cdictEnd) {
	cctx->blockState.matchState.window.nextSrc =
	cctx->blockState.matchState.window.base + cdictEnd;
	ZSTD_window_clear(&cctx->blockState.matchState.window);
	}
	/* loadedDictEnd is expressed within the referential of the active context */
	cctx->blockState.matchState.loadedDictEnd = cctx->blockState.matchState.window.dictLimit;
	} }

	cctx->dictID = cdict->dictID;

	/* copy block state */
	memcpy(cctx->blockState.prevCBlock, &cdict->cBlockState, sizeof(cdict->cBlockState));

	return 0;
	}

	static size_t ZSTD_resetCCtx_byCopyingCDict(ZSTD_CCtx* cctx,
	const ZSTD_CDict* cdict,
	ZSTD_CCtx_params params,
	U64 pledgedSrcSize,
	ZSTD_buffered_policy_e zbuff)
	{
	const ZSTD_compressionParameters *cdict_cParams = &cdict->matchState.cParams;

	DEBUGLOG(4, "copying dictionary into context");

	{ unsigned const windowLog = params.cParams.windowLog;
	assert(windowLog != 0);
	/* Copy only compression parameters related to tables. */
	params.cParams = *cdict_cParams;
	params.cParams.windowLog = windowLog;
	FORWARD_IF_ERROR(ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize,
	ZSTDcrp_leaveDirty, zbuff), "");
	assert(cctx->appliedParams.cParams.strategy == cdict_cParams->strategy);
	assert(cctx->appliedParams.cParams.hashLog == cdict_cParams->hashLog);
	assert(cctx->appliedParams.cParams.chainLog == cdict_cParams->chainLog);
	}

	ZSTD_cwksp_mark_tables_dirty(&cctx->workspace);

	/* copy tables */
	{ size_t const chainSize = (cdict_cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cdict_cParams->chainLog);
	size_t const hSize = (size_t)1 << cdict_cParams->hashLog;

	memcpy(cctx->blockState.matchState.hashTable,
	cdict->matchState.hashTable,
	hSize * sizeof(U32));
	memcpy(cctx->blockState.matchState.chainTable,
	cdict->matchState.chainTable,
	chainSize * sizeof(U32));
	}

	/* Zero the hashTable3, since the cdict never fills it */
	{ int const h3log = cctx->blockState.matchState.hashLog3;
	size_t const h3Size = h3log ? ((size_t)1 << h3log) : 0;
	assert(cdict->matchState.hashLog3 == 0);
	memset(cctx->blockState.matchState.hashTable3, 0, h3Size * sizeof(U32));
	}

	ZSTD_cwksp_mark_tables_clean(&cctx->workspace);

	/* copy dictionary offsets */
	{ ZSTD_matchState_t const* srcMatchState = &cdict->matchState;
	ZSTD_matchState_t* dstMatchState = &cctx->blockState.matchState;
	dstMatchState->window = srcMatchState->window;
	dstMatchState->nextToUpdate = srcMatchState->nextToUpdate;
	dstMatchState->loadedDictEnd= srcMatchState->loadedDictEnd;
	}

	cctx->dictID = cdict->dictID;

	/* copy block state */
	memcpy(cctx->blockState.prevCBlock, &cdict->cBlockState, sizeof(cdict->cBlockState));

	return 0;
	}

	/* We have a choice between copying the dictionary context into the working
	* context, or referencing the dictionary context from the working context
	* in-place. We decide here which strategy to use. */
	static size_t ZSTD_resetCCtx_usingCDict(ZSTD_CCtx* cctx,
	const ZSTD_CDict* cdict,
	const ZSTD_CCtx_params* params,
	U64 pledgedSrcSize,
	ZSTD_buffered_policy_e zbuff)
	{

	DEBUGLOG(4, "ZSTD_resetCCtx_usingCDict (pledgedSrcSize=%u)",
	(unsigned)pledgedSrcSize);

	if (ZSTD_shouldAttachDict(cdict, params, pledgedSrcSize)) {
	return ZSTD_resetCCtx_byAttachingCDict(
	cctx, cdict, *params, pledgedSrcSize, zbuff);
	} else {
	return ZSTD_resetCCtx_byCopyingCDict(
	cctx, cdict, *params, pledgedSrcSize, zbuff);
	}
	}

	/*! ZSTD_copyCCtx_internal() :
	* Duplicate an existing context `srcCCtx` into another one `dstCCtx`.
	* Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()).
	* The "context", in this case, refers to the hash and chain tables,
	* entropy tables, and dictionary references.
	* `windowLog` value is enforced if != 0, otherwise value is copied from srcCCtx.
	* @return : 0, or an error code */
	static size_t ZSTD_copyCCtx_internal(ZSTD_CCtx* dstCCtx,
	const ZSTD_CCtx* srcCCtx,
	ZSTD_frameParameters fParams,
	U64 pledgedSrcSize,
	ZSTD_buffered_policy_e zbuff)
	{
	DEBUGLOG(5, "ZSTD_copyCCtx_internal");
	RETURN_ERROR_IF(srcCCtx->stage!=ZSTDcs_init, stage_wrong,
	"Can't copy a ctx that's not in init stage.");

	memcpy(&dstCCtx->customMem, &srcCCtx->customMem, sizeof(ZSTD_customMem));
	{ ZSTD_CCtx_params params = dstCCtx->requestedParams;
	/* Copy only compression parameters related to tables. */
	params.cParams = srcCCtx->appliedParams.cParams;
	params.fParams = fParams;
	ZSTD_resetCCtx_internal(dstCCtx, params, pledgedSrcSize,
	ZSTDcrp_leaveDirty, zbuff);
	assert(dstCCtx->appliedParams.cParams.windowLog == srcCCtx->appliedParams.cParams.windowLog);
	assert(dstCCtx->appliedParams.cParams.strategy == srcCCtx->appliedParams.cParams.strategy);
	assert(dstCCtx->appliedParams.cParams.hashLog == srcCCtx->appliedParams.cParams.hashLog);
	assert(dstCCtx->appliedParams.cParams.chainLog == srcCCtx->appliedParams.cParams.chainLog);
	assert(dstCCtx->blockState.matchState.hashLog3 == srcCCtx->blockState.matchState.hashLog3);
	}

	ZSTD_cwksp_mark_tables_dirty(&dstCCtx->workspace);

	/* copy tables */
	{ size_t const chainSize = (srcCCtx->appliedParams.cParams.strategy == ZSTD_fast) ? 0 : ((size_t)1 << srcCCtx->appliedParams.cParams.chainLog);
	size_t const hSize = (size_t)1 << srcCCtx->appliedParams.cParams.hashLog;
	int const h3log = srcCCtx->blockState.matchState.hashLog3;
	size_t const h3Size = h3log ? ((size_t)1 << h3log) : 0;

	memcpy(dstCCtx->blockState.matchState.hashTable,
	srcCCtx->blockState.matchState.hashTable,
	hSize * sizeof(U32));
	memcpy(dstCCtx->blockState.matchState.chainTable,
	srcCCtx->blockState.matchState.chainTable,
	chainSize * sizeof(U32));
	memcpy(dstCCtx->blockState.matchState.hashTable3,
	srcCCtx->blockState.matchState.hashTable3,
	h3Size * sizeof(U32));
	}

	ZSTD_cwksp_mark_tables_clean(&dstCCtx->workspace);

	/* copy dictionary offsets */
	{
	const ZSTD_matchState_t* srcMatchState = &srcCCtx->blockState.matchState;
	ZSTD_matchState_t* dstMatchState = &dstCCtx->blockState.matchState;
	dstMatchState->window = srcMatchState->window;
	dstMatchState->nextToUpdate = srcMatchState->nextToUpdate;
	dstMatchState->loadedDictEnd= srcMatchState->loadedDictEnd;
	}
	dstCCtx->dictID = srcCCtx->dictID;

	/* copy block state */
	memcpy(dstCCtx->blockState.prevCBlock, srcCCtx->blockState.prevCBlock, sizeof(*srcCCtx->blockState.prevCBlock));

	return 0;
	}

	/*! ZSTD_copyCCtx() :
	* Duplicate an existing context `srcCCtx` into another one `dstCCtx`.
	* Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()).
	* pledgedSrcSize==0 means "unknown".
	* @return : 0, or an error code */
	size_t ZSTD_copyCCtx(ZSTD_CCtx* dstCCtx, const ZSTD_CCtx* srcCCtx, unsigned long long pledgedSrcSize)
	{
	ZSTD_frameParameters fParams = { 1 /content/, 0 /checksum/, 0 /noDictID/ };
	ZSTD_buffered_policy_e const zbuff = (ZSTD_buffered_policy_e)(srcCCtx->inBuffSize>0);
	ZSTD_STATIC_ASSERT((U32)ZSTDb_buffered==1);
	if (pledgedSrcSize==0) pledgedSrcSize = ZSTD_CONTENTSIZE_UNKNOWN;
	fParams.contentSizeFlag = (pledgedSrcSize != ZSTD_CONTENTSIZE_UNKNOWN);

	return ZSTD_copyCCtx_internal(dstCCtx, srcCCtx,
	fParams, pledgedSrcSize,
	zbuff);
	}


	#define ZSTD_ROWSIZE 16
	/*! ZSTD_reduceTable() :
	* reduce table indexes by `reducerValue`, or squash to zero.
	* PreserveMark preserves "unsorted mark" for btlazy2 strategy.
	* It must be set to a clear 0/1 value, to remove branch during inlining.
	* Presume table size is a multiple of ZSTD_ROWSIZE
	* to help auto-vectorization */
	FORCE_INLINE_TEMPLATE void
	ZSTD_reduceTable_internal (U32* const table, U32 const size, U32 const reducerValue, int const preserveMark)
	{
	int const nbRows = (int)size / ZSTD_ROWSIZE;
	int cellNb = 0;
	int rowNb;
	assert((size & (ZSTD_ROWSIZE-1)) == 0); /* multiple of ZSTD_ROWSIZE */
	assert(size < (1U<<31)); /* can be casted to int */

	#if defined (MEMORY_SANITIZER) && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE)
	/* To validate that the table re-use logic is sound, and that we don't
	* access table space that we haven't cleaned, we re-"poison" the table
	* space every time we mark it dirty.
	*
	* This function however is intended to operate on those dirty tables and
	* re-clean them. So when this function is used correctly, we can unpoison
	* the memory it operated on. This introduces a blind spot though, since
	* if we now try to operate on __actually__ poisoned memory, we will not
	* detect that. */
	__msan_unpoison(table, size * sizeof(U32));
	#endif

	for (rowNb=0 ; rowNb < nbRows ; rowNb++) {
	int column;
	for (column=0; column<ZSTD_ROWSIZE; column++) {
	if (preserveMark) {
	U32 const adder = (table[cellNb] == ZSTD_DUBT_UNSORTED_MARK) ? reducerValue : 0;
	table[cellNb] += adder;
	}
	if (table[cellNb] < reducerValue) table[cellNb] = 0;
	else table[cellNb] -= reducerValue;
	cellNb++;
	} }
	}

	static void ZSTD_reduceTable(U32* const table, U32 const size, U32 const reducerValue)
	{
	ZSTD_reduceTable_internal(table, size, reducerValue, 0);
	}

	static void ZSTD_reduceTable_btlazy2(U32* const table, U32 const size, U32 const reducerValue)
	{
	ZSTD_reduceTable_internal(table, size, reducerValue, 1);
	}

	/*! ZSTD_reduceIndex() :
	* rescale all indexes to avoid future overflow (indexes are U32) */
	static void ZSTD_reduceIndex (ZSTD_matchState_t* ms, ZSTD_CCtx_params const* params, const U32 reducerValue)
	{
	{ U32 const hSize = (U32)1 << params->cParams.hashLog;
	ZSTD_reduceTable(ms->hashTable, hSize, reducerValue);
	}

	if (params->cParams.strategy != ZSTD_fast) {
	U32 const chainSize = (U32)1 << params->cParams.chainLog;
	if (params->cParams.strategy == ZSTD_btlazy2)
	ZSTD_reduceTable_btlazy2(ms->chainTable, chainSize, reducerValue);
	else
	ZSTD_reduceTable(ms->chainTable, chainSize, reducerValue);
	}

	if (ms->hashLog3) {
	U32 const h3Size = (U32)1 << ms->hashLog3;
	ZSTD_reduceTable(ms->hashTable3, h3Size, reducerValue);
	}
	}


	/-******************************************************
	* Block entropic compression
	*********************************************************/

	/* See doc/zstd_compression_format.md for detailed format description */

	void ZSTD_seqToCodes(const seqStore_t* seqStorePtr)
	{
	const seqDef* const sequences = seqStorePtr->sequencesStart;
	BYTE* const llCodeTable = seqStorePtr->llCode;
	BYTE* const ofCodeTable = seqStorePtr->ofCode;
	BYTE* const mlCodeTable = seqStorePtr->mlCode;
	U32 const nbSeq = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
	U32 u;
	assert(nbSeq <= seqStorePtr->maxNbSeq);
	for (u=0; u<nbSeq; u++) {
	U32 const llv = sequences[u].litLength;
	U32 const mlv = sequences[u].matchLength;
	llCodeTable[u] = (BYTE)ZSTD_LLcode(llv);
	ofCodeTable[u] = (BYTE)ZSTD_highbit32(sequences[u].offset);
	mlCodeTable[u] = (BYTE)ZSTD_MLcode(mlv);
	}
	if (seqStorePtr->longLengthID==1)
	llCodeTable[seqStorePtr->longLengthPos] = MaxLL;
	if (seqStorePtr->longLengthID==2)
	mlCodeTable[seqStorePtr->longLengthPos] = MaxML;
	}

	/* ZSTD_useTargetCBlockSize():
	* Returns if target compressed block size param is being used.
	* If used, compression will do best effort to make a compressed block size to be around targetCBlockSize.
	* Returns 1 if true, 0 otherwise. */
	static int ZSTD_useTargetCBlockSize(const ZSTD_CCtx_params* cctxParams)
	{
	DEBUGLOG(5, "ZSTD_useTargetCBlockSize (targetCBlockSize=%zu)", cctxParams->targetCBlockSize);
	return (cctxParams->targetCBlockSize != 0);
	}

	/* ZSTD_compressSequences_internal():
	* actually compresses both literals and sequences */
	MEM_STATIC size_t
	ZSTD_compressSequences_internal(seqStore_t* seqStorePtr,
	const ZSTD_entropyCTables_t* prevEntropy,
	ZSTD_entropyCTables_t* nextEntropy,
	const ZSTD_CCtx_params* cctxParams,
	void* dst, size_t dstCapacity,
	void* entropyWorkspace, size_t entropyWkspSize,
	const int bmi2)
	{
	const int longOffsets = cctxParams->cParams.windowLog > STREAM_ACCUMULATOR_MIN;
	ZSTD_strategy const strategy = cctxParams->cParams.strategy;
	unsigned count[MaxSeq+1];
	FSE_CTable* CTable_LitLength = nextEntropy->fse.litlengthCTable;
	FSE_CTable* CTable_OffsetBits = nextEntropy->fse.offcodeCTable;
	FSE_CTable* CTable_MatchLength = nextEntropy->fse.matchlengthCTable;
	U32 LLtype, Offtype, MLtype; /* compressed, raw or rle */
	const seqDef* const sequences = seqStorePtr->sequencesStart;
	const BYTE* const ofCodeTable = seqStorePtr->ofCode;
	const BYTE* const llCodeTable = seqStorePtr->llCode;
	const BYTE* const mlCodeTable = seqStorePtr->mlCode;
	BYTE* const ostart = (BYTE*)dst;
	BYTE* const oend = ostart + dstCapacity;
	BYTE* op = ostart;
	size_t const nbSeq = (size_t)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
	BYTE* seqHead;
	BYTE* lastNCount = NULL;

	DEBUGLOG(5, "ZSTD_compressSequences_internal (nbSeq=%zu)", nbSeq);
	ZSTD_STATIC_ASSERT(HUF_WORKSPACE_SIZE >= (1<<MAX(MLFSELog,LLFSELog)));

	/* Compress literals */
	{ const BYTE* const literals = seqStorePtr->litStart;
	size_t const litSize = (size_t)(seqStorePtr->lit - literals);
	size_t const cSize = ZSTD_compressLiterals(
	&prevEntropy->huf, &nextEntropy->huf,
	cctxParams->cParams.strategy,
	ZSTD_disableLiteralsCompression(cctxParams),
	op, dstCapacity,
	literals, litSize,
	entropyWorkspace, entropyWkspSize,
	bmi2);
	FORWARD_IF_ERROR(cSize, "ZSTD_compressLiterals failed");
	assert(cSize <= dstCapacity);
	op += cSize;
	}

	/* Sequences Header */
	RETURN_ERROR_IF((oend-op) < 3 /max nbSeq Size/ + 1 /seqHead/,
	dstSize_tooSmall, "Can't fit seq hdr in output buf!");
	if (nbSeq < 128) {
	*op++ = (BYTE)nbSeq;
	} else if (nbSeq < LONGNBSEQ) {
	op[0] = (BYTE)((nbSeq>>8) + 0x80);
	op[1] = (BYTE)nbSeq;
	op+=2;
	} else {
	op[0]=0xFF;
	MEM_writeLE16(op+1, (U16)(nbSeq - LONGNBSEQ));
	op+=3;
	}
	assert(op <= oend);
	if (nbSeq==0) {
	/* Copy the old tables over as if we repeated them */
	memcpy(&nextEntropy->fse, &prevEntropy->fse, sizeof(prevEntropy->fse));
	return (size_t)(op - ostart);
	}

	/* seqHead : flags for FSE encoding type */
	seqHead = op++;
	assert(op <= oend);

	/* convert length/distances into codes */
	ZSTD_seqToCodes(seqStorePtr);
	/* build CTable for Literal Lengths */
	{ unsigned max = MaxLL;
	size_t const mostFrequent = HIST_countFast_wksp(count, &max, llCodeTable, nbSeq, entropyWorkspace, entropyWkspSize); /* can't fail */
	DEBUGLOG(5, "Building LL table");
	nextEntropy->fse.litlength_repeatMode = prevEntropy->fse.litlength_repeatMode;
	LLtype = ZSTD_selectEncodingType(&nextEntropy->fse.litlength_repeatMode,
	count, max, mostFrequent, nbSeq,
	LLFSELog, prevEntropy->fse.litlengthCTable,
	LL_defaultNorm, LL_defaultNormLog,
	ZSTD_defaultAllowed, strategy);
	assert(set_basic < set_compressed && set_rle < set_compressed);
	assert(!(LLtype < set_compressed && nextEntropy->fse.litlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
	{ size_t const countSize = ZSTD_buildCTable(
	op, (size_t)(oend - op),
	CTable_LitLength, LLFSELog, (symbolEncodingType_e)LLtype,
	count, max, llCodeTable, nbSeq,
	LL_defaultNorm, LL_defaultNormLog, MaxLL,
	prevEntropy->fse.litlengthCTable,
	sizeof(prevEntropy->fse.litlengthCTable),
	entropyWorkspace, entropyWkspSize);
	FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for LitLens failed");
	if (LLtype == set_compressed)
	lastNCount = op;
	op += countSize;
	assert(op <= oend);
	} }
	/* build CTable for Offsets */
	{ unsigned max = MaxOff;
	size_t const mostFrequent = HIST_countFast_wksp(
	count, &max, ofCodeTable, nbSeq, entropyWorkspace, entropyWkspSize); /* can't fail */
	/* We can only use the basic table if max <= DefaultMaxOff, otherwise the offsets are too large */
	ZSTD_defaultPolicy_e const defaultPolicy = (max <= DefaultMaxOff) ? ZSTD_defaultAllowed : ZSTD_defaultDisallowed;
	DEBUGLOG(5, "Building OF table");
	nextEntropy->fse.offcode_repeatMode = prevEntropy->fse.offcode_repeatMode;
	Offtype = ZSTD_selectEncodingType(&nextEntropy->fse.offcode_repeatMode,
	count, max, mostFrequent, nbSeq,
	OffFSELog, prevEntropy->fse.offcodeCTable,
	OF_defaultNorm, OF_defaultNormLog,
	defaultPolicy, strategy);
	assert(!(Offtype < set_compressed && nextEntropy->fse.offcode_repeatMode != FSE_repeat_none)); /* We don't copy tables */
	{ size_t const countSize = ZSTD_buildCTable(
	op, (size_t)(oend - op),
	CTable_OffsetBits, OffFSELog, (symbolEncodingType_e)Offtype,
	count, max, ofCodeTable, nbSeq,
	OF_defaultNorm, OF_defaultNormLog, DefaultMaxOff,
	prevEntropy->fse.offcodeCTable,
	sizeof(prevEntropy->fse.offcodeCTable),
	entropyWorkspace, entropyWkspSize);
	FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for Offsets failed");
	if (Offtype == set_compressed)
	lastNCount = op;
	op += countSize;
	assert(op <= oend);
	} }
	/* build CTable for MatchLengths */
	{ unsigned max = MaxML;
	size_t const mostFrequent = HIST_countFast_wksp(
	count, &max, mlCodeTable, nbSeq, entropyWorkspace, entropyWkspSize); /* can't fail */
	DEBUGLOG(5, "Building ML table (remaining space : %i)", (int)(oend-op));
	nextEntropy->fse.matchlength_repeatMode = prevEntropy->fse.matchlength_repeatMode;
	MLtype = ZSTD_selectEncodingType(&nextEntropy->fse.matchlength_repeatMode,
	count, max, mostFrequent, nbSeq,
	MLFSELog, prevEntropy->fse.matchlengthCTable,
	ML_defaultNorm, ML_defaultNormLog,
	ZSTD_defaultAllowed, strategy);
	assert(!(MLtype < set_compressed && nextEntropy->fse.matchlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
	{ size_t const countSize = ZSTD_buildCTable(
	op, (size_t)(oend - op),
	CTable_MatchLength, MLFSELog, (symbolEncodingType_e)MLtype,
	count, max, mlCodeTable, nbSeq,
	ML_defaultNorm, ML_defaultNormLog, MaxML,
	prevEntropy->fse.matchlengthCTable,
	sizeof(prevEntropy->fse.matchlengthCTable),
	entropyWorkspace, entropyWkspSize);
	FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for MatchLengths failed");
	if (MLtype == set_compressed)
	lastNCount = op;
	op += countSize;
	assert(op <= oend);
	} }

	*seqHead = (BYTE)((LLtype<<6) + (Offtype<<4) + (MLtype<<2));

	{ size_t const bitstreamSize = ZSTD_encodeSequences(
	op, (size_t)(oend - op),
	CTable_MatchLength, mlCodeTable,
	CTable_OffsetBits, ofCodeTable,
	CTable_LitLength, llCodeTable,
	sequences, nbSeq,
	longOffsets, bmi2);
	FORWARD_IF_ERROR(bitstreamSize, "ZSTD_encodeSequences failed");
	op += bitstreamSize;
	assert(op <= oend);
	/* zstd versions <= 1.3.4 mistakenly report corruption when
	* FSE_readNCount() receives a buffer < 4 bytes.
	* Fixed by https://github.com/facebook/zstd/pull/1146.
	* This can happen when the last set_compressed table present is 2
	* bytes and the bitstream is only one byte.
	* In this exceedingly rare case, we will simply emit an uncompressed
	* block, since it isn't worth optimizing.
	*/
	if (lastNCount && (op - lastNCount) < 4) {
	/* NCountSize >= 2 && bitstreamSize > 0 ==> lastCountSize == 3 */
	assert(op - lastNCount == 3);
	DEBUGLOG(5, "Avoiding bug in zstd decoder in versions <= 1.3.4 by "
	"emitting an uncompressed block.");
	return 0;
	}
	}

	DEBUGLOG(5, "compressed block size : %u", (unsigned)(op - ostart));
	return (size_t)(op - ostart);
	}

	MEM_STATIC size_t
	ZSTD_compressSequences(seqStore_t* seqStorePtr,
	const ZSTD_entropyCTables_t* prevEntropy,
	ZSTD_entropyCTables_t* nextEntropy,
	const ZSTD_CCtx_params* cctxParams,
	void* dst, size_t dstCapacity,
	size_t srcSize,
	void* entropyWorkspace, size_t entropyWkspSize,
	int bmi2)
	{
	size_t const cSize = ZSTD_compressSequences_internal(
	seqStorePtr, prevEntropy, nextEntropy, cctxParams,
	dst, dstCapacity,
	entropyWorkspace, entropyWkspSize, bmi2);
	if (cSize == 0) return 0;
	/* When srcSize <= dstCapacity, there is enough space to write a raw uncompressed block.
	* Since we ran out of space, block must be not compressible, so fall back to raw uncompressed block.
	*/
	if ((cSize == ERROR(dstSize_tooSmall)) & (srcSize <= dstCapacity))
	return 0; /* block not compressed */
	FORWARD_IF_ERROR(cSize, "ZSTD_compressSequences_internal failed");

	/* Check compressibility */
	{ size_t const maxCSize = srcSize - ZSTD_minGain(srcSize, cctxParams->cParams.strategy);
	if (cSize >= maxCSize) return 0; /* block not compressed */
	}

	return cSize;
	}

	/* ZSTD_selectBlockCompressor() :
	* Not static, but internal use only (used by long distance matcher)
	* assumption : strat is a valid strategy */
	ZSTD_blockCompressor ZSTD_selectBlockCompressor(ZSTD_strategy strat, ZSTD_dictMode_e dictMode)
	{
	static const ZSTD_blockCompressor blockCompressor[3][ZSTD_STRATEGY_MAX+1] = {
	{ ZSTD_compressBlock_fast /* default for 0 */,
	ZSTD_compressBlock_fast,
	ZSTD_compressBlock_doubleFast,
	ZSTD_compressBlock_greedy,
	ZSTD_compressBlock_lazy,
	ZSTD_compressBlock_lazy2,
	ZSTD_compressBlock_btlazy2,
	ZSTD_compressBlock_btopt,
	ZSTD_compressBlock_btultra,
	ZSTD_compressBlock_btultra2 },
	{ ZSTD_compressBlock_fast_extDict /* default for 0 */,
	ZSTD_compressBlock_fast_extDict,
	ZSTD_compressBlock_doubleFast_extDict,
	ZSTD_compressBlock_greedy_extDict,
	ZSTD_compressBlock_lazy_extDict,
	ZSTD_compressBlock_lazy2_extDict,
	ZSTD_compressBlock_btlazy2_extDict,
	ZSTD_compressBlock_btopt_extDict,
	ZSTD_compressBlock_btultra_extDict,
	ZSTD_compressBlock_btultra_extDict },
	{ ZSTD_compressBlock_fast_dictMatchState /* default for 0 */,
	ZSTD_compressBlock_fast_dictMatchState,
	ZSTD_compressBlock_doubleFast_dictMatchState,
	ZSTD_compressBlock_greedy_dictMatchState,
	ZSTD_compressBlock_lazy_dictMatchState,
	ZSTD_compressBlock_lazy2_dictMatchState,
	ZSTD_compressBlock_btlazy2_dictMatchState,
	ZSTD_compressBlock_btopt_dictMatchState,
	ZSTD_compressBlock_btultra_dictMatchState,
	ZSTD_compressBlock_btultra_dictMatchState }
	};
	ZSTD_blockCompressor selectedCompressor;
	ZSTD_STATIC_ASSERT((unsigned)ZSTD_fast == 1);

	assert(ZSTD_cParam_withinBounds(ZSTD_c_strategy, strat));
	selectedCompressor = blockCompressor[(int)dictMode][(int)strat];
	assert(selectedCompressor != NULL);
	return selectedCompressor;
	}

	static void ZSTD_storeLastLiterals(seqStore_t* seqStorePtr,
	const BYTE* anchor, size_t lastLLSize)
	{
	memcpy(seqStorePtr->lit, anchor, lastLLSize);
	seqStorePtr->lit += lastLLSize;
	}

	void ZSTD_resetSeqStore(seqStore_t* ssPtr)
	{
	ssPtr->lit = ssPtr->litStart;
	ssPtr->sequences = ssPtr->sequencesStart;
	ssPtr->longLengthID = 0;
	}

	typedef enum { ZSTDbss_compress, ZSTDbss_noCompress } ZSTD_buildSeqStore_e;

	static size_t ZSTD_buildSeqStore(ZSTD_CCtx* zc, const void* src, size_t srcSize)
	{
	ZSTD_matchState_t* const ms = &zc->blockState.matchState;
	DEBUGLOG(5, "ZSTD_buildSeqStore (srcSize=%zu)", srcSize);
	assert(srcSize <= ZSTD_BLOCKSIZE_MAX);
	/* Assert that we have correctly flushed the ctx params into the ms's copy */
	ZSTD_assertEqualCParams(zc->appliedParams.cParams, ms->cParams);
	if (srcSize < MIN_CBLOCK_SIZE+ZSTD_blockHeaderSize+1) {
	ZSTD_ldm_skipSequences(&zc->externSeqStore, srcSize, zc->appliedParams.cParams.minMatch);
	return ZSTDbss_noCompress; /* don't even attempt compression below a certain srcSize */
	}
	ZSTD_resetSeqStore(&(zc->seqStore));
	/* required for optimal parser to read stats from dictionary */
	ms->opt.symbolCosts = &zc->blockState.prevCBlock->entropy;
	/* tell the optimal parser how we expect to compress literals */
	ms->opt.literalCompressionMode = zc->appliedParams.literalCompressionMode;
	/* a gap between an attached dict and the current window is not safe,
	* they must remain adjacent,
	* and when that stops being the case, the dict must be unset */
	assert(ms->dictMatchState == NULL \|\| ms->loadedDictEnd == ms->window.dictLimit);

	/* limited update after a very long match */
	{ const BYTE* const base = ms->window.base;
	const BYTE* const istart = (const BYTE*)src;
	const U32 current = (U32)(istart-base);
	if (sizeof(ptrdiff_t)==8) assert(istart - base < (ptrdiff_t)(U32)(-1)); /* ensure no overflow */
	if (current > ms->nextToUpdate + 384)
	ms->nextToUpdate = current - MIN(192, (U32)(current - ms->nextToUpdate - 384));
	}

	/* select and store sequences */
	{ ZSTD_dictMode_e const dictMode = ZSTD_matchState_dictMode(ms);
	size_t lastLLSize;
	{ int i;
	for (i = 0; i < ZSTD_REP_NUM; ++i)
	zc->blockState.nextCBlock->rep[i] = zc->blockState.prevCBlock->rep[i];
	}
	if (zc->externSeqStore.pos < zc->externSeqStore.size) {
	assert(!zc->appliedParams.ldmParams.enableLdm);
	/* Updates ldmSeqStore.pos */
	lastLLSize =
	ZSTD_ldm_blockCompress(&zc->externSeqStore,
	ms, &zc->seqStore,
	zc->blockState.nextCBlock->rep,
	src, srcSize);
	assert(zc->externSeqStore.pos <= zc->externSeqStore.size);
	} else if (zc->appliedParams.ldmParams.enableLdm) {
	rawSeqStore_t ldmSeqStore = {NULL, 0, 0, 0};

	ldmSeqStore.seq = zc->ldmSequences;
	ldmSeqStore.capacity = zc->maxNbLdmSequences;
	/* Updates ldmSeqStore.size */
	FORWARD_IF_ERROR(ZSTD_ldm_generateSequences(&zc->ldmState, &ldmSeqStore,
	&zc->appliedParams.ldmParams,
	src, srcSize), "");
	/* Updates ldmSeqStore.pos */
	lastLLSize =
	ZSTD_ldm_blockCompress(&ldmSeqStore,
	ms, &zc->seqStore,
	zc->blockState.nextCBlock->rep,
	src, srcSize);
	assert(ldmSeqStore.pos == ldmSeqStore.size);
	} else { /* not long range mode */
	ZSTD_blockCompressor const blockCompressor = ZSTD_selectBlockCompressor(zc->appliedParams.cParams.strategy, dictMode);
	lastLLSize = blockCompressor(ms, &zc->seqStore, zc->blockState.nextCBlock->rep, src, srcSize);
	}
	{ const BYTE* const lastLiterals = (const BYTE*)src + srcSize - lastLLSize;
	ZSTD_storeLastLiterals(&zc->seqStore, lastLiterals, lastLLSize);
	} }
	return ZSTDbss_compress;
	}

	static void ZSTD_copyBlockSequences(ZSTD_CCtx* zc)
	{
	const seqStore_t* seqStore = ZSTD_getSeqStore(zc);
	const seqDef* seqs = seqStore->sequencesStart;
	size_t seqsSize = seqStore->sequences - seqs;

	ZSTD_Sequence* outSeqs = &zc->seqCollector.seqStart[zc->seqCollector.seqIndex];
	size_t i; size_t position; int repIdx;

	assert(zc->seqCollector.seqIndex + 1 < zc->seqCollector.maxSequences);
	for (i = 0, position = 0; i < seqsSize; ++i) {
	outSeqs[i].offset = seqs[i].offset;
	outSeqs[i].litLength = seqs[i].litLength;
	outSeqs[i].matchLength = seqs[i].matchLength + MINMATCH;

	if (i == seqStore->longLengthPos) {
	if (seqStore->longLengthID == 1) {
	outSeqs[i].litLength += 0x10000;
	} else if (seqStore->longLengthID == 2) {
	outSeqs[i].matchLength += 0x10000;
	}
	}

	if (outSeqs[i].offset <= ZSTD_REP_NUM) {
	outSeqs[i].rep = outSeqs[i].offset;
	repIdx = (unsigned int)i - outSeqs[i].offset;

	if (outSeqs[i].litLength == 0) {
	if (outSeqs[i].offset < 3) {
	--repIdx;
	} else {
	repIdx = (unsigned int)i - 1;
	}
	++outSeqs[i].rep;
	}
	assert(repIdx >= -3);
	outSeqs[i].offset = repIdx >= 0 ? outSeqs[repIdx].offset : repStartValue[-repIdx - 1];
	if (outSeqs[i].rep == 4) {
	--outSeqs[i].offset;
	}
	} else {
	outSeqs[i].offset -= ZSTD_REP_NUM;
	}

	position += outSeqs[i].litLength;
	outSeqs[i].matchPos = (unsigned int)position;
	position += outSeqs[i].matchLength;
	}
	zc->seqCollector.seqIndex += seqsSize;
	}

	size_t ZSTD_getSequences(ZSTD_CCtx* zc, ZSTD_Sequence* outSeqs,
	size_t outSeqsSize, const void* src, size_t srcSize)
	{
	const size_t dstCapacity = ZSTD_compressBound(srcSize);
	void* dst = ZSTD_malloc(dstCapacity, ZSTD_defaultCMem);
	SeqCollector seqCollector;

	RETURN_ERROR_IF(dst == NULL, memory_allocation, "NULL pointer!");

	seqCollector.collectSequences = 1;
	seqCollector.seqStart = outSeqs;
	seqCollector.seqIndex = 0;
	seqCollector.maxSequences = outSeqsSize;
	zc->seqCollector = seqCollector;

	ZSTD_compress2(zc, dst, dstCapacity, src, srcSize);
	ZSTD_free(dst, ZSTD_defaultCMem);
	return zc->seqCollector.seqIndex;
	}

	/* Returns true if the given block is a RLE block */
	static int ZSTD_isRLE(const BYTE *ip, size_t length) {
	size_t i;
	if (length < 2) return 1;
	for (i = 1; i < length; ++i) {
	if (ip[0] != ip[i]) return 0;
	}
	return 1;
	}

	/* Returns true if the given block may be RLE.
	* This is just a heuristic based on the compressibility.
	* It may return both false positives and false negatives.
	*/
	static int ZSTD_maybeRLE(seqStore_t const* seqStore)
	{
	size_t const nbSeqs = (size_t)(seqStore->sequences - seqStore->sequencesStart);
	size_t const nbLits = (size_t)(seqStore->lit - seqStore->litStart);

	return nbSeqs < 4 && nbLits < 10;
	}

	static void ZSTD_confirmRepcodesAndEntropyTables(ZSTD_CCtx* zc)
	{
	ZSTD_compressedBlockState_t* const tmp = zc->blockState.prevCBlock;
	zc->blockState.prevCBlock = zc->blockState.nextCBlock;
	zc->blockState.nextCBlock = tmp;
	}

	static size_t ZSTD_compressBlock_internal(ZSTD_CCtx* zc,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize, U32 frame)
	{
	/* This the upper bound for the length of an rle block.
	* This isn't the actual upper bound. Finding the real threshold
	* needs further investigation.
	*/
	const U32 rleMaxLength = 25;
	size_t cSize;
	const BYTE* ip = (const BYTE*)src;
	BYTE* op = (BYTE*)dst;
	DEBUGLOG(5, "ZSTD_compressBlock_internal (dstCapacity=%u, dictLimit=%u, nextToUpdate=%u)",
	(unsigned)dstCapacity, (unsigned)zc->blockState.matchState.window.dictLimit,
	(unsigned)zc->blockState.matchState.nextToUpdate);

	{ const size_t bss = ZSTD_buildSeqStore(zc, src, srcSize);
	FORWARD_IF_ERROR(bss, "ZSTD_buildSeqStore failed");
	if (bss == ZSTDbss_noCompress) { cSize = 0; goto out; }
	}

	if (zc->seqCollector.collectSequences) {
	ZSTD_copyBlockSequences(zc);
	return 0;
	}

	/* encode sequences and literals */
	cSize = ZSTD_compressSequences(&zc->seqStore,
	&zc->blockState.prevCBlock->entropy, &zc->blockState.nextCBlock->entropy,
	&zc->appliedParams,
	dst, dstCapacity,
	srcSize,
	zc->entropyWorkspace, HUF_WORKSPACE_SIZE /* statically allocated in resetCCtx */,
	zc->bmi2);

	if (frame &&
	/* We don't want to emit our first block as a RLE even if it qualifies because
	* doing so will cause the decoder (cli only) to throw a "should consume all input error."
	* This is only an issue for zstd <= v1.4.3
	*/
	!zc->isFirstBlock &&
	cSize < rleMaxLength &&
	ZSTD_isRLE(ip, srcSize))
	{
	cSize = 1;
	op[0] = ip[0];
	}

	out:
	if (!ZSTD_isError(cSize) && cSize > 1) {
	ZSTD_confirmRepcodesAndEntropyTables(zc);
	}
	/* We check that dictionaries have offset codes available for the first
	* block. After the first block, the offcode table might not have large
	* enough codes to represent the offsets in the data.
	*/
	if (zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode == FSE_repeat_valid)
	zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode = FSE_repeat_check;

	return cSize;
	}

	static size_t ZSTD_compressBlock_targetCBlockSize_body(ZSTD_CCtx* zc,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const size_t bss, U32 lastBlock)
	{
	DEBUGLOG(6, "Attempting ZSTD_compressSuperBlock()");
	if (bss == ZSTDbss_compress) {
	if (/* We don't want to emit our first block as a RLE even if it qualifies because
	* doing so will cause the decoder (cli only) to throw a "should consume all input error."
	* This is only an issue for zstd <= v1.4.3
	*/
	!zc->isFirstBlock &&
	ZSTD_maybeRLE(&zc->seqStore) &&
	ZSTD_isRLE((BYTE const*)src, srcSize))
	{
	return ZSTD_rleCompressBlock(dst, dstCapacity, (BYTE const)src, srcSize, lastBlock);
	}
	/* Attempt superblock compression.
	*
	* Note that compressed size of ZSTD_compressSuperBlock() is not bound by the
	* standard ZSTD_compressBound(). This is a problem, because even if we have
	* space now, taking an extra byte now could cause us to run out of space later
	* and violate ZSTD_compressBound().
	*
	* Define blockBound(blockSize) = blockSize + ZSTD_blockHeaderSize.
	*
	* In order to respect ZSTD_compressBound() we must attempt to emit a raw
	* uncompressed block in these cases:
	* * cSize == 0: Return code for an uncompressed block.
	* * cSize == dstSize_tooSmall: We may have expanded beyond blockBound(srcSize).
	* ZSTD_noCompressBlock() will return dstSize_tooSmall if we are really out of
	* output space.
	* * cSize >= blockBound(srcSize): We have expanded the block too much so
	* emit an uncompressed block.
	*/
	{
	size_t const cSize = ZSTD_compressSuperBlock(zc, dst, dstCapacity, src, srcSize, lastBlock);
	if (cSize != ERROR(dstSize_tooSmall)) {
	size_t const maxCSize = srcSize - ZSTD_minGain(srcSize, zc->appliedParams.cParams.strategy);
	FORWARD_IF_ERROR(cSize, "ZSTD_compressSuperBlock failed");
	if (cSize != 0 && cSize < maxCSize + ZSTD_blockHeaderSize) {
	ZSTD_confirmRepcodesAndEntropyTables(zc);
	return cSize;
	}
	}
	}
	}

	DEBUGLOG(6, "Resorting to ZSTD_noCompressBlock()");
	/* Superblock compression failed, attempt to emit a single no compress block.
	* The decoder will be able to stream this block since it is uncompressed.
	*/
	return ZSTD_noCompressBlock(dst, dstCapacity, src, srcSize, lastBlock);
	}

	static size_t ZSTD_compressBlock_targetCBlockSize(ZSTD_CCtx* zc,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	U32 lastBlock)
	{
	size_t cSize = 0;
	const size_t bss = ZSTD_buildSeqStore(zc, src, srcSize);
	DEBUGLOG(5, "ZSTD_compressBlock_targetCBlockSize (dstCapacity=%u, dictLimit=%u, nextToUpdate=%u, srcSize=%zu)",
	(unsigned)dstCapacity, (unsigned)zc->blockState.matchState.window.dictLimit, (unsigned)zc->blockState.matchState.nextToUpdate, srcSize);
	FORWARD_IF_ERROR(bss, "ZSTD_buildSeqStore failed");

	cSize = ZSTD_compressBlock_targetCBlockSize_body(zc, dst, dstCapacity, src, srcSize, bss, lastBlock);
	FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_targetCBlockSize_body failed");

	if (zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode == FSE_repeat_valid)
	zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode = FSE_repeat_check;

	return cSize;
	}

	static void ZSTD_overflowCorrectIfNeeded(ZSTD_matchState_t* ms,
	ZSTD_cwksp* ws,
	ZSTD_CCtx_params const* params,
	void const* ip,
	void const* iend)
	{
	if (ZSTD_window_needOverflowCorrection(ms->window, iend)) {
	U32 const maxDist = (U32)1 << params->cParams.windowLog;
	U32 const cycleLog = ZSTD_cycleLog(params->cParams.chainLog, params->cParams.strategy);
	U32 const correction = ZSTD_window_correctOverflow(&ms->window, cycleLog, maxDist, ip);
	ZSTD_STATIC_ASSERT(ZSTD_CHAINLOG_MAX <= 30);
	ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX_32 <= 30);
	ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX <= 31);
	ZSTD_cwksp_mark_tables_dirty(ws);
	ZSTD_reduceIndex(ms, params, correction);
	ZSTD_cwksp_mark_tables_clean(ws);
	if (ms->nextToUpdate < correction) ms->nextToUpdate = 0;
	else ms->nextToUpdate -= correction;
	/* invalidate dictionaries on overflow correction */
	ms->loadedDictEnd = 0;
	ms->dictMatchState = NULL;
	}
	}

	/*! ZSTD_compress_frameChunk() :
	* Compress a chunk of data into one or multiple blocks.
	* All blocks will be terminated, all input will be consumed.
	* Function will issue an error if there is not enough `dstCapacity` to hold the compressed content.
	* Frame is supposed already started (header already produced)
	* @return : compressed size, or an error code
	*/
	static size_t ZSTD_compress_frameChunk (ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	U32 lastFrameChunk)
	{
	size_t blockSize = cctx->blockSize;
	size_t remaining = srcSize;
	const BYTE* ip = (const BYTE*)src;
	BYTE* const ostart = (BYTE*)dst;
	BYTE* op = ostart;
	U32 const maxDist = (U32)1 << cctx->appliedParams.cParams.windowLog;

	assert(cctx->appliedParams.cParams.windowLog <= ZSTD_WINDOWLOG_MAX);

	DEBUGLOG(5, "ZSTD_compress_frameChunk (blockSize=%u)", (unsigned)blockSize);
	if (cctx->appliedParams.fParams.checksumFlag && srcSize)
	XXH64_update(&cctx->xxhState, src, srcSize);

	while (remaining) {
	ZSTD_matchState_t* const ms = &cctx->blockState.matchState;
	U32 const lastBlock = lastFrameChunk & (blockSize >= remaining);

	RETURN_ERROR_IF(dstCapacity < ZSTD_blockHeaderSize + MIN_CBLOCK_SIZE,
	dstSize_tooSmall,
	"not enough space to store compressed block");
	if (remaining < blockSize) blockSize = remaining;

	ZSTD_overflowCorrectIfNeeded(
	ms, &cctx->workspace, &cctx->appliedParams, ip, ip + blockSize);
	ZSTD_checkDictValidity(&ms->window, ip + blockSize, maxDist, &ms->loadedDictEnd, &ms->dictMatchState);

	/* Ensure hash/chain table insertion resumes no sooner than lowlimit */
	if (ms->nextToUpdate < ms->window.lowLimit) ms->nextToUpdate = ms->window.lowLimit;

	{ size_t cSize;
	if (ZSTD_useTargetCBlockSize(&cctx->appliedParams)) {
	cSize = ZSTD_compressBlock_targetCBlockSize(cctx, op, dstCapacity, ip, blockSize, lastBlock);
	FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_targetCBlockSize failed");
	assert(cSize > 0);
	assert(cSize <= blockSize + ZSTD_blockHeaderSize);
	} else {
	cSize = ZSTD_compressBlock_internal(cctx,
	op+ZSTD_blockHeaderSize, dstCapacity-ZSTD_blockHeaderSize,
	ip, blockSize, 1 /* frame */);
	FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_internal failed");

	if (cSize == 0) { /* block is not compressible */
	cSize = ZSTD_noCompressBlock(op, dstCapacity, ip, blockSize, lastBlock);
	FORWARD_IF_ERROR(cSize, "ZSTD_noCompressBlock failed");
	} else {
	U32 const cBlockHeader = cSize == 1 ?
	lastBlock + (((U32)bt_rle)<<1) + (U32)(blockSize << 3) :
	lastBlock + (((U32)bt_compressed)<<1) + (U32)(cSize << 3);
	MEM_writeLE24(op, cBlockHeader);
	cSize += ZSTD_blockHeaderSize;
	}
	}


	ip += blockSize;
	assert(remaining >= blockSize);
	remaining -= blockSize;
	op += cSize;
	assert(dstCapacity >= cSize);
	dstCapacity -= cSize;
	cctx->isFirstBlock = 0;
	DEBUGLOG(5, "ZSTD_compress_frameChunk: adding a block of size %u",
	(unsigned)cSize);
	} }

	if (lastFrameChunk && (op>ostart)) cctx->stage = ZSTDcs_ending;
	return (size_t)(op-ostart);
	}


	static size_t ZSTD_writeFrameHeader(void* dst, size_t dstCapacity,
	const ZSTD_CCtx_params* params, U64 pledgedSrcSize, U32 dictID)
	{ BYTE* const op = (BYTE*)dst;
	U32 const dictIDSizeCodeLength = (dictID>0) + (dictID>=256) + (dictID>=65536); /* 0-3 */
	U32 const dictIDSizeCode = params->fParams.noDictIDFlag ? 0 : dictIDSizeCodeLength; /* 0-3 */
	U32 const checksumFlag = params->fParams.checksumFlag>0;
	U32 const windowSize = (U32)1 << params->cParams.windowLog;
	U32 const singleSegment = params->fParams.contentSizeFlag && (windowSize >= pledgedSrcSize);
	BYTE const windowLogByte = (BYTE)((params->cParams.windowLog - ZSTD_WINDOWLOG_ABSOLUTEMIN) << 3);
	U32 const fcsCode = params->fParams.contentSizeFlag ?
	(pledgedSrcSize>=256) + (pledgedSrcSize>=65536+256) + (pledgedSrcSize>=0xFFFFFFFFU) : 0; /* 0-3 */
	BYTE const frameHeaderDescriptionByte = (BYTE)(dictIDSizeCode + (checksumFlag<<2) + (singleSegment<<5) + (fcsCode<<6) );
	size_t pos=0;

	assert(!(params->fParams.contentSizeFlag && pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN));
	RETURN_ERROR_IF(dstCapacity < ZSTD_FRAMEHEADERSIZE_MAX, dstSize_tooSmall,
	"dst buf is too small to fit worst-case frame header size.");
	DEBUGLOG(4, "ZSTD_writeFrameHeader : dictIDFlag : %u ; dictID : %u ; dictIDSizeCode : %u",
	!params->fParams.noDictIDFlag, (unsigned)dictID, (unsigned)dictIDSizeCode);

	if (params->format == ZSTD_f_zstd1) {
	MEM_writeLE32(dst, ZSTD_MAGICNUMBER);
	pos = 4;
	}
	op[pos++] = frameHeaderDescriptionByte;
	if (!singleSegment) op[pos++] = windowLogByte;
	switch(dictIDSizeCode)
	{
	default: assert(0); /* impossible */
	case 0 : break;
	case 1 : op[pos] = (BYTE)(dictID); pos++; break;
	case 2 : MEM_writeLE16(op+pos, (U16)dictID); pos+=2; break;
	case 3 : MEM_writeLE32(op+pos, dictID); pos+=4; break;
	}
	switch(fcsCode)
	{
	default: assert(0); /* impossible */
	case 0 : if (singleSegment) op[pos++] = (BYTE)(pledgedSrcSize); break;
	case 1 : MEM_writeLE16(op+pos, (U16)(pledgedSrcSize-256)); pos+=2; break;
	case 2 : MEM_writeLE32(op+pos, (U32)(pledgedSrcSize)); pos+=4; break;
	case 3 : MEM_writeLE64(op+pos, (U64)(pledgedSrcSize)); pos+=8; break;
	}
	return pos;
	}

	/* ZSTD_writeLastEmptyBlock() :
	* output an empty Block with end-of-frame mark to complete a frame
	* @return : size of data written into `dst` (== ZSTD_blockHeaderSize (defined in zstd_internal.h))
	* or an error code if `dstCapacity` is too small (<ZSTD_blockHeaderSize)
	*/
	size_t ZSTD_writeLastEmptyBlock(void* dst, size_t dstCapacity)
	{
	RETURN_ERROR_IF(dstCapacity < ZSTD_blockHeaderSize, dstSize_tooSmall,
	"dst buf is too small to write frame trailer empty block.");
	{ U32 const cBlockHeader24 = 1 /lastBlock/ + (((U32)bt_raw)<<1); /* 0 size */
	MEM_writeLE24(dst, cBlockHeader24);
	return ZSTD_blockHeaderSize;
	}
	}

	size_t ZSTD_referenceExternalSequences(ZSTD_CCtx* cctx, rawSeq* seq, size_t nbSeq)
	{
	RETURN_ERROR_IF(cctx->stage != ZSTDcs_init, stage_wrong,
	"wrong cctx stage");
	RETURN_ERROR_IF(cctx->appliedParams.ldmParams.enableLdm,
	parameter_unsupported,
	"incompatible with ldm");
	cctx->externSeqStore.seq = seq;
	cctx->externSeqStore.size = nbSeq;
	cctx->externSeqStore.capacity = nbSeq;
	cctx->externSeqStore.pos = 0;
	return 0;
	}


	static size_t ZSTD_compressContinue_internal (ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	U32 frame, U32 lastFrameChunk)
	{
	ZSTD_matchState_t* const ms = &cctx->blockState.matchState;
	size_t fhSize = 0;

	DEBUGLOG(5, "ZSTD_compressContinue_internal, stage: %u, srcSize: %u",
	cctx->stage, (unsigned)srcSize);
	RETURN_ERROR_IF(cctx->stage==ZSTDcs_created, stage_wrong,
	"missing init (ZSTD_compressBegin)");

	if (frame && (cctx->stage==ZSTDcs_init)) {
	fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, &cctx->appliedParams,
	cctx->pledgedSrcSizePlusOne-1, cctx->dictID);
	FORWARD_IF_ERROR(fhSize, "ZSTD_writeFrameHeader failed");
	assert(fhSize <= dstCapacity);
	dstCapacity -= fhSize;
	dst = (char*)dst + fhSize;
	cctx->stage = ZSTDcs_ongoing;
	}

	if (!srcSize) return fhSize; /* do not generate an empty block if no input */

	if (!ZSTD_window_update(&ms->window, src, srcSize)) {
	ms->nextToUpdate = ms->window.dictLimit;
	}
	if (cctx->appliedParams.ldmParams.enableLdm) {
	ZSTD_window_update(&cctx->ldmState.window, src, srcSize);
	}

	if (!frame) {
	/* overflow check and correction for block mode */
	ZSTD_overflowCorrectIfNeeded(
	ms, &cctx->workspace, &cctx->appliedParams,
	src, (BYTE const*)src + srcSize);
	}

	DEBUGLOG(5, "ZSTD_compressContinue_internal (blockSize=%u)", (unsigned)cctx->blockSize);
	{ size_t const cSize = frame ?
	ZSTD_compress_frameChunk (cctx, dst, dstCapacity, src, srcSize, lastFrameChunk) :
	ZSTD_compressBlock_internal (cctx, dst, dstCapacity, src, srcSize, 0 /* frame */);
	FORWARD_IF_ERROR(cSize, "%s", frame ? "ZSTD_compress_frameChunk failed" : "ZSTD_compressBlock_internal failed");
	cctx->consumedSrcSize += srcSize;
	cctx->producedCSize += (cSize + fhSize);
	assert(!(cctx->appliedParams.fParams.contentSizeFlag && cctx->pledgedSrcSizePlusOne == 0));
	if (cctx->pledgedSrcSizePlusOne != 0) { /* control src size */
	ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN == (unsigned long long)-1);
	RETURN_ERROR_IF(
	cctx->consumedSrcSize+1 > cctx->pledgedSrcSizePlusOne,
	srcSize_wrong,
	"error : pledgedSrcSize = %u, while realSrcSize >= %u",
	(unsigned)cctx->pledgedSrcSizePlusOne-1,
	(unsigned)cctx->consumedSrcSize);
	}
	return cSize + fhSize;
	}
	}

	size_t ZSTD_compressContinue (ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize)
	{
	DEBUGLOG(5, "ZSTD_compressContinue (srcSize=%u)", (unsigned)srcSize);
	return ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 1 /* frame mode /, 0 / last chunk */);
	}


	size_t ZSTD_getBlockSize(const ZSTD_CCtx* cctx)
	{
	ZSTD_compressionParameters const cParams = cctx->appliedParams.cParams;
	assert(!ZSTD_checkCParams(cParams));
	return MIN (ZSTD_BLOCKSIZE_MAX, (U32)1 << cParams.windowLog);
	}

	size_t ZSTD_compressBlock(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize)
	{
	DEBUGLOG(5, "ZSTD_compressBlock: srcSize = %u", (unsigned)srcSize);
	{ size_t const blockSizeMax = ZSTD_getBlockSize(cctx);
	RETURN_ERROR_IF(srcSize > blockSizeMax, srcSize_wrong, "input is larger than a block"); }

	return ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 0 /* frame mode /, 0 / last chunk */);
	}

	/*! ZSTD_loadDictionaryContent() :
	* @return : 0, or an error code
	*/
	static size_t ZSTD_loadDictionaryContent(ZSTD_matchState_t* ms,
	ldmState_t* ls,
	ZSTD_cwksp* ws,
	ZSTD_CCtx_params const* params,
	const void* src, size_t srcSize,
	ZSTD_dictTableLoadMethod_e dtlm)
	{
	const BYTE* ip = (const BYTE*) src;
	const BYTE* const iend = ip + srcSize;

	ZSTD_window_update(&ms->window, src, srcSize);
	ms->loadedDictEnd = params->forceWindow ? 0 : (U32)(iend - ms->window.base);

	if (params->ldmParams.enableLdm && ls != NULL) {
	ZSTD_window_update(&ls->window, src, srcSize);
	ls->loadedDictEnd = params->forceWindow ? 0 : (U32)(iend - ls->window.base);
	}

	/* Assert that we the ms params match the params we're being given */
	ZSTD_assertEqualCParams(params->cParams, ms->cParams);

	if (srcSize <= HASH_READ_SIZE) return 0;

	while (iend - ip > HASH_READ_SIZE) {
	size_t const remaining = (size_t)(iend - ip);
	size_t const chunk = MIN(remaining, ZSTD_CHUNKSIZE_MAX);
	const BYTE* const ichunk = ip + chunk;

	ZSTD_overflowCorrectIfNeeded(ms, ws, params, ip, ichunk);

	if (params->ldmParams.enableLdm && ls != NULL)
	ZSTD_ldm_fillHashTable(ls, (const BYTE)src, (const BYTE)src + srcSize, &params->ldmParams);

	switch(params->cParams.strategy)
	{
	case ZSTD_fast:
	ZSTD_fillHashTable(ms, ichunk, dtlm);
	break;
	case ZSTD_dfast:
	ZSTD_fillDoubleHashTable(ms, ichunk, dtlm);
	break;

	case ZSTD_greedy:
	case ZSTD_lazy:
	case ZSTD_lazy2:
	if (chunk >= HASH_READ_SIZE)
	ZSTD_insertAndFindFirstIndex(ms, ichunk-HASH_READ_SIZE);
	break;

	case ZSTD_btlazy2: /* we want the dictionary table fully sorted */
	case ZSTD_btopt:
	case ZSTD_btultra:
	case ZSTD_btultra2:
	if (chunk >= HASH_READ_SIZE)
	ZSTD_updateTree(ms, ichunk-HASH_READ_SIZE, ichunk);
	break;

	default:
	assert(0); /* not possible : not a valid strategy id */
	}

	ip = ichunk;
	}

	ms->nextToUpdate = (U32)(iend - ms->window.base);
	return 0;
	}


	/* Dictionaries that assign zero probability to symbols that show up causes problems
	when FSE encoding. Refuse dictionaries that assign zero probability to symbols
	that we may encounter during compression.
	NOTE: This behavior is not standard and could be improved in the future. */
	static size_t ZSTD_checkDictNCount(short* normalizedCounter, unsigned dictMaxSymbolValue, unsigned maxSymbolValue) {
	U32 s;
	RETURN_ERROR_IF(dictMaxSymbolValue < maxSymbolValue, dictionary_corrupted, "dict fse tables don't have all symbols");
	for (s = 0; s <= maxSymbolValue; ++s) {
	RETURN_ERROR_IF(normalizedCounter[s] == 0, dictionary_corrupted, "dict fse tables don't have all symbols");
	}
	return 0;
	}

	size_t ZSTD_loadCEntropy(ZSTD_compressedBlockState_t* bs, void* workspace,
	short* offcodeNCount, unsigned* offcodeMaxValue,
	const void* const dict, size_t dictSize)
	{
	const BYTE* dictPtr = (const BYTE)dict; / skip magic num and dict ID */
	const BYTE* const dictEnd = dictPtr + dictSize;
	dictPtr += 8;
	bs->entropy.huf.repeatMode = HUF_repeat_check;

	{ unsigned maxSymbolValue = 255;
	unsigned hasZeroWeights = 1;
	size_t const hufHeaderSize = HUF_readCTable((HUF_CElt*)bs->entropy.huf.CTable, &maxSymbolValue, dictPtr,
	dictEnd-dictPtr, &hasZeroWeights);

	/* We only set the loaded table as valid if it contains all non-zero
	* weights. Otherwise, we set it to check */
	if (!hasZeroWeights)
	bs->entropy.huf.repeatMode = HUF_repeat_valid;

	RETURN_ERROR_IF(HUF_isError(hufHeaderSize), dictionary_corrupted, "");
	RETURN_ERROR_IF(maxSymbolValue < 255, dictionary_corrupted, "");
	dictPtr += hufHeaderSize;
	}

	{ unsigned offcodeLog;
	size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, offcodeMaxValue, &offcodeLog, dictPtr, dictEnd-dictPtr);
	RETURN_ERROR_IF(FSE_isError(offcodeHeaderSize), dictionary_corrupted, "");
	RETURN_ERROR_IF(offcodeLog > OffFSELog, dictionary_corrupted, "");
	/* Defer checking offcodeMaxValue because we need to know the size of the dictionary content */
	/* fill all offset symbols to avoid garbage at end of table */
	RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp(
	bs->entropy.fse.offcodeCTable,
	offcodeNCount, MaxOff, offcodeLog,
	workspace, HUF_WORKSPACE_SIZE)),
	dictionary_corrupted, "");
	dictPtr += offcodeHeaderSize;
	}

	{ short matchlengthNCount[MaxML+1];
	unsigned matchlengthMaxValue = MaxML, matchlengthLog;
	size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, dictEnd-dictPtr);
	RETURN_ERROR_IF(FSE_isError(matchlengthHeaderSize), dictionary_corrupted, "");
	RETURN_ERROR_IF(matchlengthLog > MLFSELog, dictionary_corrupted, "");
	/* Every match length code must have non-zero probability */
	FORWARD_IF_ERROR( ZSTD_checkDictNCount(matchlengthNCount, matchlengthMaxValue, MaxML), "");
	RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp(
	bs->entropy.fse.matchlengthCTable,
	matchlengthNCount, matchlengthMaxValue, matchlengthLog,
	workspace, HUF_WORKSPACE_SIZE)),
	dictionary_corrupted, "");
	dictPtr += matchlengthHeaderSize;
	}

	{ short litlengthNCount[MaxLL+1];
	unsigned litlengthMaxValue = MaxLL, litlengthLog;
	size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, dictEnd-dictPtr);
	RETURN_ERROR_IF(FSE_isError(litlengthHeaderSize), dictionary_corrupted, "");
	RETURN_ERROR_IF(litlengthLog > LLFSELog, dictionary_corrupted, "");
	/* Every literal length code must have non-zero probability */
	FORWARD_IF_ERROR( ZSTD_checkDictNCount(litlengthNCount, litlengthMaxValue, MaxLL), "");
	RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp(
	bs->entropy.fse.litlengthCTable,
	litlengthNCount, litlengthMaxValue, litlengthLog,
	workspace, HUF_WORKSPACE_SIZE)),
	dictionary_corrupted, "");
	dictPtr += litlengthHeaderSize;
	}

	RETURN_ERROR_IF(dictPtr+12 > dictEnd, dictionary_corrupted, "");
	bs->rep[0] = MEM_readLE32(dictPtr+0);
	bs->rep[1] = MEM_readLE32(dictPtr+4);
	bs->rep[2] = MEM_readLE32(dictPtr+8);
	dictPtr += 12;

	return dictPtr - (const BYTE*)dict;
	}

	/* Dictionary format :
	* See :
	* https://github.com/facebook/zstd/blob/master/doc/zstd_compression_format.md#dictionary-format
	*/
	/*! ZSTD_loadZstdDictionary() :
	* @return : dictID, or an error code
	* assumptions : magic number supposed already checked
	* dictSize supposed >= 8
	*/
	static size_t ZSTD_loadZstdDictionary(ZSTD_compressedBlockState_t* bs,
	ZSTD_matchState_t* ms,
	ZSTD_cwksp* ws,
	ZSTD_CCtx_params const* params,
	const void* dict, size_t dictSize,
	ZSTD_dictTableLoadMethod_e dtlm,
	void* workspace)
	{
	const BYTE* dictPtr = (const BYTE*)dict;
	const BYTE* const dictEnd = dictPtr + dictSize;
	short offcodeNCount[MaxOff+1];
	unsigned offcodeMaxValue = MaxOff;
	size_t dictID;
	size_t eSize;

	ZSTD_STATIC_ASSERT(HUF_WORKSPACE_SIZE >= (1<<MAX(MLFSELog,LLFSELog)));
	assert(dictSize >= 8);
	assert(MEM_readLE32(dictPtr) == ZSTD_MAGIC_DICTIONARY);

	dictID = params->fParams.noDictIDFlag ? 0 : MEM_readLE32(dictPtr + 4 /* skip magic number */ );
	eSize = ZSTD_loadCEntropy(bs, workspace, offcodeNCount, &offcodeMaxValue, dict, dictSize);
	FORWARD_IF_ERROR(eSize, "ZSTD_loadCEntropy failed");
	dictPtr += eSize;

	{ size_t const dictContentSize = (size_t)(dictEnd - dictPtr);
	U32 offcodeMax = MaxOff;
	if (dictContentSize <= ((U32)-1) - 128 KB) {
	U32 const maxOffset = (U32)dictContentSize + 128 KB; /* The maximum offset that must be supported */
	offcodeMax = ZSTD_highbit32(maxOffset); /* Calculate minimum offset code required to represent maxOffset */
	}
	/* All offset values <= dictContentSize + 128 KB must be representable */
	FORWARD_IF_ERROR(ZSTD_checkDictNCount(offcodeNCount, offcodeMaxValue, MIN(offcodeMax, MaxOff)), "");
	/* All repCodes must be <= dictContentSize and != 0*/
	{ U32 u;
	for (u=0; u<3; u++) {
	RETURN_ERROR_IF(bs->rep[u] == 0, dictionary_corrupted, "");
	RETURN_ERROR_IF(bs->rep[u] > dictContentSize, dictionary_corrupted, "");
	} }

	bs->entropy.fse.offcode_repeatMode = FSE_repeat_valid;
	bs->entropy.fse.matchlength_repeatMode = FSE_repeat_valid;
	bs->entropy.fse.litlength_repeatMode = FSE_repeat_valid;
	FORWARD_IF_ERROR(ZSTD_loadDictionaryContent(
	ms, NULL, ws, params, dictPtr, dictContentSize, dtlm), "");
	return dictID;
	}
	}

	/** ZSTD_compress_insertDictionary() :
	* @return : dictID, or an error code */
	static size_t
	ZSTD_compress_insertDictionary(ZSTD_compressedBlockState_t* bs,
	ZSTD_matchState_t* ms,
	ldmState_t* ls,
	ZSTD_cwksp* ws,
	const ZSTD_CCtx_params* params,
	const void* dict, size_t dictSize,
	ZSTD_dictContentType_e dictContentType,
	ZSTD_dictTableLoadMethod_e dtlm,
	void* workspace)
	{
	DEBUGLOG(4, "ZSTD_compress_insertDictionary (dictSize=%u)", (U32)dictSize);
	if ((dict==NULL) \|\| (dictSize<8)) {
	RETURN_ERROR_IF(dictContentType == ZSTD_dct_fullDict, dictionary_wrong, "");
	return 0;
	}

	ZSTD_reset_compressedBlockState(bs);

	/* dict restricted modes */
	if (dictContentType == ZSTD_dct_rawContent)
	return ZSTD_loadDictionaryContent(ms, ls, ws, params, dict, dictSize, dtlm);

	if (MEM_readLE32(dict) != ZSTD_MAGIC_DICTIONARY) {
	if (dictContentType == ZSTD_dct_auto) {
	DEBUGLOG(4, "raw content dictionary detected");
	return ZSTD_loadDictionaryContent(
	ms, ls, ws, params, dict, dictSize, dtlm);
	}
	RETURN_ERROR_IF(dictContentType == ZSTD_dct_fullDict, dictionary_wrong, "");
	assert(0); /* impossible */
	}

	/* dict as full zstd dictionary */
	return ZSTD_loadZstdDictionary(
	bs, ms, ws, params, dict, dictSize, dtlm, workspace);
	}

	#define ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF (128 KB)
	#define ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER (6)

	/*! ZSTD_compressBegin_internal() :
	* @return : 0, or an error code */
	static size_t ZSTD_compressBegin_internal(ZSTD_CCtx* cctx,
	const void* dict, size_t dictSize,
	ZSTD_dictContentType_e dictContentType,
	ZSTD_dictTableLoadMethod_e dtlm,
	const ZSTD_CDict* cdict,
	const ZSTD_CCtx_params* params, U64 pledgedSrcSize,
	ZSTD_buffered_policy_e zbuff)
	{
	DEBUGLOG(4, "ZSTD_compressBegin_internal: wlog=%u", params->cParams.windowLog);
	/* params are supposed to be fully validated at this point */
	assert(!ZSTD_isError(ZSTD_checkCParams(params->cParams)));
	assert(!((dict) && (cdict))); /* either dict or cdict, not both */
	if ( (cdict)
	&& (cdict->dictContentSize > 0)
	&& ( pledgedSrcSize < ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF
	\|\| pledgedSrcSize < cdict->dictContentSize * ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER
	\|\| pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN
	\|\| cdict->compressionLevel == 0)
	&& (params->attachDictPref != ZSTD_dictForceLoad) ) {
	return ZSTD_resetCCtx_usingCDict(cctx, cdict, params, pledgedSrcSize, zbuff);
	}

	FORWARD_IF_ERROR( ZSTD_resetCCtx_internal(cctx, *params, pledgedSrcSize,
	ZSTDcrp_makeClean, zbuff) , "");
	{ size_t const dictID = cdict ?
	ZSTD_compress_insertDictionary(
	cctx->blockState.prevCBlock, &cctx->blockState.matchState,
	&cctx->ldmState, &cctx->workspace, &cctx->appliedParams, cdict->dictContent,
	cdict->dictContentSize, dictContentType, dtlm,
	cctx->entropyWorkspace)
	: ZSTD_compress_insertDictionary(
	cctx->blockState.prevCBlock, &cctx->blockState.matchState,
	&cctx->ldmState, &cctx->workspace, &cctx->appliedParams, dict, dictSize,
	dictContentType, dtlm, cctx->entropyWorkspace);
	FORWARD_IF_ERROR(dictID, "ZSTD_compress_insertDictionary failed");
	assert(dictID <= UINT_MAX);
	cctx->dictID = (U32)dictID;
	}
	return 0;
	}

	size_t ZSTD_compressBegin_advanced_internal(ZSTD_CCtx* cctx,
	const void* dict, size_t dictSize,
	ZSTD_dictContentType_e dictContentType,
	ZSTD_dictTableLoadMethod_e dtlm,
	const ZSTD_CDict* cdict,
	const ZSTD_CCtx_params* params,
	unsigned long long pledgedSrcSize)
	{
	DEBUGLOG(4, "ZSTD_compressBegin_advanced_internal: wlog=%u", params->cParams.windowLog);
	/* compression parameters verification and optimization */
	FORWARD_IF_ERROR( ZSTD_checkCParams(params->cParams) , "");
	return ZSTD_compressBegin_internal(cctx,
	dict, dictSize, dictContentType, dtlm,
	cdict,
	params, pledgedSrcSize,
	ZSTDb_not_buffered);
	}

	/*! ZSTD_compressBegin_advanced() :
	* @return : 0, or an error code */
	size_t ZSTD_compressBegin_advanced(ZSTD_CCtx* cctx,
	const void* dict, size_t dictSize,
	ZSTD_parameters params, unsigned long long pledgedSrcSize)
	{
	ZSTD_CCtx_params const cctxParams =
	ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, &params);
	return ZSTD_compressBegin_advanced_internal(cctx,
	dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast,
	NULL /cdict/,
	&cctxParams, pledgedSrcSize);
	}

	size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, int compressionLevel)
	{
	ZSTD_parameters const params = ZSTD_getParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize);
	ZSTD_CCtx_params const cctxParams =
	ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, &params);
	DEBUGLOG(4, "ZSTD_compressBegin_usingDict (dictSize=%u)", (unsigned)dictSize);
	return ZSTD_compressBegin_internal(cctx, dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast, NULL,
	&cctxParams, ZSTD_CONTENTSIZE_UNKNOWN, ZSTDb_not_buffered);
	}

	size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel)
	{
	return ZSTD_compressBegin_usingDict(cctx, NULL, 0, compressionLevel);
	}


	/*! ZSTD_writeEpilogue() :
	* Ends a frame.
	* @return : nb of bytes written into dst (or an error code) */
	static size_t ZSTD_writeEpilogue(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity)
	{
	BYTE* const ostart = (BYTE*)dst;
	BYTE* op = ostart;
	size_t fhSize = 0;

	DEBUGLOG(4, "ZSTD_writeEpilogue");
	RETURN_ERROR_IF(cctx->stage == ZSTDcs_created, stage_wrong, "init missing");

	/* special case : empty frame */
	if (cctx->stage == ZSTDcs_init) {
	fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, &cctx->appliedParams, 0, 0);
	FORWARD_IF_ERROR(fhSize, "ZSTD_writeFrameHeader failed");
	dstCapacity -= fhSize;
	op += fhSize;
	cctx->stage = ZSTDcs_ongoing;
	}

	if (cctx->stage != ZSTDcs_ending) {
	/* write one last empty block, make it the "last" block */
	U32 const cBlockHeader24 = 1 /* last block */ + (((U32)bt_raw)<<1) + 0;
	RETURN_ERROR_IF(dstCapacity<4, dstSize_tooSmall, "no room for epilogue");
	MEM_writeLE32(op, cBlockHeader24);
	op += ZSTD_blockHeaderSize;
	dstCapacity -= ZSTD_blockHeaderSize;
	}

	if (cctx->appliedParams.fParams.checksumFlag) {
	U32 const checksum = (U32) XXH64_digest(&cctx->xxhState);
	RETURN_ERROR_IF(dstCapacity<4, dstSize_tooSmall, "no room for checksum");
	DEBUGLOG(4, "ZSTD_writeEpilogue: write checksum : %08X", (unsigned)checksum);
	MEM_writeLE32(op, checksum);
	op += 4;
	}

	cctx->stage = ZSTDcs_created; /* return to "created but no init" status */
	return op-ostart;
	}

	size_t ZSTD_compressEnd (ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize)
	{
	size_t endResult;
	size_t const cSize = ZSTD_compressContinue_internal(cctx,
	dst, dstCapacity, src, srcSize,
	1 /* frame mode /, 1 / last chunk */);
	FORWARD_IF_ERROR(cSize, "ZSTD_compressContinue_internal failed");
	endResult = ZSTD_writeEpilogue(cctx, (char*)dst + cSize, dstCapacity-cSize);
	FORWARD_IF_ERROR(endResult, "ZSTD_writeEpilogue failed");
	assert(!(cctx->appliedParams.fParams.contentSizeFlag && cctx->pledgedSrcSizePlusOne == 0));
	if (cctx->pledgedSrcSizePlusOne != 0) { /* control src size */
	ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN == (unsigned long long)-1);
	DEBUGLOG(4, "end of frame : controlling src size");
	RETURN_ERROR_IF(
	cctx->pledgedSrcSizePlusOne != cctx->consumedSrcSize+1,
	srcSize_wrong,
	"error : pledgedSrcSize = %u, while realSrcSize = %u",
	(unsigned)cctx->pledgedSrcSizePlusOne-1,
	(unsigned)cctx->consumedSrcSize);
	}
	return cSize + endResult;
	}


	static size_t ZSTD_compress_internal (ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict,size_t dictSize,
	const ZSTD_parameters* params)
	{
	ZSTD_CCtx_params const cctxParams =
	ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, params);
	DEBUGLOG(4, "ZSTD_compress_internal");
	return ZSTD_compress_advanced_internal(cctx,
	dst, dstCapacity,
	src, srcSize,
	dict, dictSize,
	&cctxParams);
	}

	size_t ZSTD_compress_advanced (ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict,size_t dictSize,
	ZSTD_parameters params)
	{
	DEBUGLOG(4, "ZSTD_compress_advanced");
	FORWARD_IF_ERROR(ZSTD_checkCParams(params.cParams), "");
	return ZSTD_compress_internal(cctx,
	dst, dstCapacity,
	src, srcSize,
	dict, dictSize,
	&params);
	}

	/* Internal */
	size_t ZSTD_compress_advanced_internal(
	ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict,size_t dictSize,
	const ZSTD_CCtx_params* params)
	{
	DEBUGLOG(4, "ZSTD_compress_advanced_internal (srcSize:%u)", (unsigned)srcSize);
	FORWARD_IF_ERROR( ZSTD_compressBegin_internal(cctx,
	dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast, NULL,
	params, srcSize, ZSTDb_not_buffered) , "");
	return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize);
	}

	size_t ZSTD_compress_usingDict(ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict, size_t dictSize,
	int compressionLevel)
	{
	ZSTD_parameters const params = ZSTD_getParams_internal(compressionLevel, srcSize, dict ? dictSize : 0);
	ZSTD_CCtx_params cctxParams = ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, &params);
	DEBUGLOG(4, "ZSTD_compress_usingDict (srcSize=%u)", (unsigned)srcSize);
	assert(params.fParams.contentSizeFlag == 1);
	return ZSTD_compress_advanced_internal(cctx, dst, dstCapacity, src, srcSize, dict, dictSize, &cctxParams);
	}

	size_t ZSTD_compressCCtx(ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	int compressionLevel)
	{
	DEBUGLOG(4, "ZSTD_compressCCtx (srcSize=%u)", (unsigned)srcSize);
	assert(cctx != NULL);
	return ZSTD_compress_usingDict(cctx, dst, dstCapacity, src, srcSize, NULL, 0, compressionLevel);
	}

	size_t ZSTD_compress(void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	int compressionLevel)
	{
	size_t result;
	ZSTD_CCtx ctxBody;
	ZSTD_initCCtx(&ctxBody, ZSTD_defaultCMem);
	result = ZSTD_compressCCtx(&ctxBody, dst, dstCapacity, src, srcSize, compressionLevel);
	ZSTD_freeCCtxContent(&ctxBody); /* can't free ctxBody itself, as it's on stack; free only heap content */
	return result;
	}


	/* ===== Dictionary API ===== */

	/*! ZSTD_estimateCDictSize_advanced() :
	* Estimate amount of memory that will be needed to create a dictionary with following arguments */
	size_t ZSTD_estimateCDictSize_advanced(
	size_t dictSize, ZSTD_compressionParameters cParams,
	ZSTD_dictLoadMethod_e dictLoadMethod)
	{
	DEBUGLOG(5, "sizeof(ZSTD_CDict) : %u", (unsigned)sizeof(ZSTD_CDict));
	return ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict))
	+ ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE)
	+ ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0)
	+ (dictLoadMethod == ZSTD_dlm_byRef ? 0
	: ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void *))));
	}

	size_t ZSTD_estimateCDictSize(size_t dictSize, int compressionLevel)
	{
	ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize);
	return ZSTD_estimateCDictSize_advanced(dictSize, cParams, ZSTD_dlm_byCopy);
	}

	size_t ZSTD_sizeof_CDict(const ZSTD_CDict* cdict)
	{
	if (cdict==NULL) return 0; /* support sizeof on NULL */
	DEBUGLOG(5, "sizeof(cdict) : %u", (unsigned)sizeof(cdict));
	/* cdict may be in the workspace */
	return (cdict->workspace.workspace == cdict ? 0 : sizeof(*cdict))
	+ ZSTD_cwksp_sizeof(&cdict->workspace);
	}

	static size_t ZSTD_initCDict_internal(
	ZSTD_CDict* cdict,
	const void* dictBuffer, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod,
	ZSTD_dictContentType_e dictContentType,
	ZSTD_compressionParameters cParams)
	{
	DEBUGLOG(3, "ZSTD_initCDict_internal (dictContentType:%u)", (unsigned)dictContentType);
	assert(!ZSTD_checkCParams(cParams));
	cdict->matchState.cParams = cParams;
	if ((dictLoadMethod == ZSTD_dlm_byRef) \|\| (!dictBuffer) \|\| (!dictSize)) {
	cdict->dictContent = dictBuffer;
	} else {
	void internalBuffer = ZSTD_cwksp_reserve_object(&cdict->workspace, ZSTD_cwksp_align(dictSize, sizeof(void)));
	RETURN_ERROR_IF(!internalBuffer, memory_allocation, "NULL pointer!");
	cdict->dictContent = internalBuffer;
	memcpy(internalBuffer, dictBuffer, dictSize);
	}
	cdict->dictContentSize = dictSize;

	cdict->entropyWorkspace = (U32*)ZSTD_cwksp_reserve_object(&cdict->workspace, HUF_WORKSPACE_SIZE);


	/* Reset the state to no dictionary */
	ZSTD_reset_compressedBlockState(&cdict->cBlockState);
	FORWARD_IF_ERROR(ZSTD_reset_matchState(
	&cdict->matchState,
	&cdict->workspace,
	&cParams,
	ZSTDcrp_makeClean,
	ZSTDirp_reset,
	ZSTD_resetTarget_CDict), "");
	/* (Maybe) load the dictionary
	* Skips loading the dictionary if it is < 8 bytes.
	*/
	{ ZSTD_CCtx_params params;
	memset(&params, 0, sizeof(params));
	params.compressionLevel = ZSTD_CLEVEL_DEFAULT;
	params.fParams.contentSizeFlag = 1;
	params.cParams = cParams;
	{ size_t const dictID = ZSTD_compress_insertDictionary(
	&cdict->cBlockState, &cdict->matchState, NULL, &cdict->workspace,
	&params, cdict->dictContent, cdict->dictContentSize,
	dictContentType, ZSTD_dtlm_full, cdict->entropyWorkspace);
	FORWARD_IF_ERROR(dictID, "ZSTD_compress_insertDictionary failed");
	assert(dictID <= (size_t)(U32)-1);
	cdict->dictID = (U32)dictID;
	}
	}

	return 0;
	}

	ZSTD_CDict* ZSTD_createCDict_advanced(const void* dictBuffer, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod,
	ZSTD_dictContentType_e dictContentType,
	ZSTD_compressionParameters cParams, ZSTD_customMem customMem)
	{
	DEBUGLOG(3, "ZSTD_createCDict_advanced, mode %u", (unsigned)dictContentType);
	if (!customMem.customAlloc ^ !customMem.customFree) return NULL;

	{ size_t const workspaceSize =
	ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict)) +
	ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE) +
	ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0) +
	(dictLoadMethod == ZSTD_dlm_byRef ? 0
	: ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void*))));
	void* const workspace = ZSTD_malloc(workspaceSize, customMem);
	ZSTD_cwksp ws;
	ZSTD_CDict* cdict;

	if (!workspace) {
	ZSTD_free(workspace, customMem);
	return NULL;
	}

	ZSTD_cwksp_init(&ws, workspace, workspaceSize);

	cdict = (ZSTD_CDict*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CDict));
	assert(cdict != NULL);
	ZSTD_cwksp_move(&cdict->workspace, &ws);
	cdict->customMem = customMem;
	cdict->compressionLevel = 0; /* signals advanced API usage */

	if (ZSTD_isError( ZSTD_initCDict_internal(cdict,
	dictBuffer, dictSize,
	dictLoadMethod, dictContentType,
	cParams) )) {
	ZSTD_freeCDict(cdict);
	return NULL;
	}

	return cdict;
	}
	}

	ZSTD_CDict* ZSTD_createCDict(const void* dict, size_t dictSize, int compressionLevel)
	{
	ZSTD_compressionParameters cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize);
	ZSTD_CDict* cdict = ZSTD_createCDict_advanced(dict, dictSize,
	ZSTD_dlm_byCopy, ZSTD_dct_auto,
	cParams, ZSTD_defaultCMem);
	if (cdict)
	cdict->compressionLevel = compressionLevel == 0 ? ZSTD_CLEVEL_DEFAULT : compressionLevel;
	return cdict;
	}

	ZSTD_CDict* ZSTD_createCDict_byReference(const void* dict, size_t dictSize, int compressionLevel)
	{
	ZSTD_compressionParameters cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize);
	return ZSTD_createCDict_advanced(dict, dictSize,
	ZSTD_dlm_byRef, ZSTD_dct_auto,
	cParams, ZSTD_defaultCMem);
	}

	size_t ZSTD_freeCDict(ZSTD_CDict* cdict)
	{
	if (cdict==NULL) return 0; /* support free on NULL */
	{ ZSTD_customMem const cMem = cdict->customMem;
	int cdictInWorkspace = ZSTD_cwksp_owns_buffer(&cdict->workspace, cdict);
	ZSTD_cwksp_free(&cdict->workspace, cMem);
	if (!cdictInWorkspace) {
	ZSTD_free(cdict, cMem);
	}
	return 0;
	}
	}

	/*! ZSTD_initStaticCDict_advanced() :
	* Generate a digested dictionary in provided memory area.
	* workspace: The memory area to emplace the dictionary into.
	* Provided pointer must 8-bytes aligned.
	* It must outlive dictionary usage.
	* workspaceSize: Use ZSTD_estimateCDictSize()
	* to determine how large workspace must be.
	* cParams : use ZSTD_getCParams() to transform a compression level
	* into its relevants cParams.
	* @return : pointer to ZSTD_CDict*, or NULL if error (size too small)
	* Note : there is no corresponding "free" function.
	* Since workspace was allocated externally, it must be freed externally.
	*/
	const ZSTD_CDict* ZSTD_initStaticCDict(
	void* workspace, size_t workspaceSize,
	const void* dict, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod,
	ZSTD_dictContentType_e dictContentType,
	ZSTD_compressionParameters cParams)
	{
	size_t const matchStateSize = ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0);
	size_t const neededSize = ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict))
	+ (dictLoadMethod == ZSTD_dlm_byRef ? 0
	: ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void*))))
	+ ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE)
	+ matchStateSize;
	ZSTD_CDict* cdict;

	if ((size_t)workspace & 7) return NULL; /* 8-aligned */

	{
	ZSTD_cwksp ws;
	ZSTD_cwksp_init(&ws, workspace, workspaceSize);
	cdict = (ZSTD_CDict*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CDict));
	if (cdict == NULL) return NULL;
	ZSTD_cwksp_move(&cdict->workspace, &ws);
	}

	DEBUGLOG(4, "(workspaceSize < neededSize) : (%u < %u) => %u",
	(unsigned)workspaceSize, (unsigned)neededSize, (unsigned)(workspaceSize < neededSize));
	if (workspaceSize < neededSize) return NULL;

	if (ZSTD_isError( ZSTD_initCDict_internal(cdict,
	dict, dictSize,
	dictLoadMethod, dictContentType,
	cParams) ))
	return NULL;

	return cdict;
	}

	ZSTD_compressionParameters ZSTD_getCParamsFromCDict(const ZSTD_CDict* cdict)
	{
	assert(cdict != NULL);
	return cdict->matchState.cParams;
	}

	/* ZSTD_compressBegin_usingCDict_advanced() :
	* cdict must be != NULL */
	size_t ZSTD_compressBegin_usingCDict_advanced(
	ZSTD_CCtx* const cctx, const ZSTD_CDict* const cdict,
	ZSTD_frameParameters const fParams, unsigned long long const pledgedSrcSize)
	{
	DEBUGLOG(4, "ZSTD_compressBegin_usingCDict_advanced");
	RETURN_ERROR_IF(cdict==NULL, dictionary_wrong, "NULL pointer!");
	{ ZSTD_CCtx_params params = cctx->requestedParams;
	params.cParams = ( pledgedSrcSize < ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF
	\|\| pledgedSrcSize < cdict->dictContentSize * ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER
	\|\| pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN
	\|\| cdict->compressionLevel == 0 )
	&& (params.attachDictPref != ZSTD_dictForceLoad) ?
	ZSTD_getCParamsFromCDict(cdict)
	: ZSTD_getCParams(cdict->compressionLevel,
	pledgedSrcSize,
	cdict->dictContentSize);
	/* Increase window log to fit the entire dictionary and source if the
	* source size is known. Limit the increase to 19, which is the
	* window log for compression level 1 with the largest source size.
	*/
	if (pledgedSrcSize != ZSTD_CONTENTSIZE_UNKNOWN) {
	U32 const limitedSrcSize = (U32)MIN(pledgedSrcSize, 1U << 19);
	U32 const limitedSrcLog = limitedSrcSize > 1 ? ZSTD_highbit32(limitedSrcSize - 1) + 1 : 1;
	params.cParams.windowLog = MAX(params.cParams.windowLog, limitedSrcLog);
	}
	params.fParams = fParams;
	return ZSTD_compressBegin_internal(cctx,
	NULL, 0, ZSTD_dct_auto, ZSTD_dtlm_fast,
	cdict,
	&params, pledgedSrcSize,
	ZSTDb_not_buffered);
	}
	}

	/* ZSTD_compressBegin_usingCDict() :
	* pledgedSrcSize=0 means "unknown"
	* if pledgedSrcSize>0, it will enable contentSizeFlag */
	size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict)
	{
	ZSTD_frameParameters const fParams = { 0 /content/, 0 /checksum/, 0 /noDictID/ };
	DEBUGLOG(4, "ZSTD_compressBegin_usingCDict : dictIDFlag == %u", !fParams.noDictIDFlag);
	return ZSTD_compressBegin_usingCDict_advanced(cctx, cdict, fParams, ZSTD_CONTENTSIZE_UNKNOWN);
	}

	size_t ZSTD_compress_usingCDict_advanced(ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const ZSTD_CDict* cdict, ZSTD_frameParameters fParams)
	{
	FORWARD_IF_ERROR(ZSTD_compressBegin_usingCDict_advanced(cctx, cdict, fParams, srcSize), ""); /* will check if cdict != NULL */
	return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize);
	}

	/*! ZSTD_compress_usingCDict() :
	* Compression using a digested Dictionary.
	* Faster startup than ZSTD_compress_usingDict(), recommended when same dictionary is used multiple times.
	* Note that compression parameters are decided at CDict creation time
	* while frame parameters are hardcoded */
	size_t ZSTD_compress_usingCDict(ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const ZSTD_CDict* cdict)
	{
	ZSTD_frameParameters const fParams = { 1 /content/, 0 /checksum/, 0 /noDictID/ };
	return ZSTD_compress_usingCDict_advanced(cctx, dst, dstCapacity, src, srcSize, cdict, fParams);
	}



	/* ******************************************************************
	* Streaming
	********************************************************************/

	ZSTD_CStream* ZSTD_createCStream(void)
	{
	DEBUGLOG(3, "ZSTD_createCStream");
	return ZSTD_createCStream_advanced(ZSTD_defaultCMem);
	}

	ZSTD_CStream* ZSTD_initStaticCStream(void *workspace, size_t workspaceSize)
	{
	return ZSTD_initStaticCCtx(workspace, workspaceSize);
	}

	ZSTD_CStream* ZSTD_createCStream_advanced(ZSTD_customMem customMem)
	{ /* CStream and CCtx are now same object */
	return ZSTD_createCCtx_advanced(customMem);
	}

	size_t ZSTD_freeCStream(ZSTD_CStream* zcs)
	{
	return ZSTD_freeCCtx(zcs); /* same object */
	}



	/====== Initialization ======/

	size_t ZSTD_CStreamInSize(void) { return ZSTD_BLOCKSIZE_MAX; }

	size_t ZSTD_CStreamOutSize(void)
	{
	return ZSTD_compressBound(ZSTD_BLOCKSIZE_MAX) + ZSTD_blockHeaderSize + 4 /* 32-bits hash */ ;
	}

	static size_t ZSTD_resetCStream_internal(ZSTD_CStream* cctx,
	const void* const dict, size_t const dictSize, ZSTD_dictContentType_e const dictContentType,
	const ZSTD_CDict* const cdict,
	ZSTD_CCtx_params params, unsigned long long const pledgedSrcSize)
	{
	DEBUGLOG(4, "ZSTD_resetCStream_internal");
	/* Finalize the compression parameters */
	params.cParams = ZSTD_getCParamsFromCCtxParams(&params, pledgedSrcSize, dictSize);
	/* params are supposed to be fully validated at this point */
	assert(!ZSTD_isError(ZSTD_checkCParams(params.cParams)));
	assert(!((dict) && (cdict))); /* either dict or cdict, not both */

	FORWARD_IF_ERROR( ZSTD_compressBegin_internal(cctx,
	dict, dictSize, dictContentType, ZSTD_dtlm_fast,
	cdict,
	&params, pledgedSrcSize,
	ZSTDb_buffered) , "");

	cctx->inToCompress = 0;
	cctx->inBuffPos = 0;
	cctx->inBuffTarget = cctx->blockSize
	+ (cctx->blockSize == pledgedSrcSize); /* for small input: avoid automatic flush on reaching end of block, since it would require to add a 3-bytes null block to end frame */
	cctx->outBuffContentSize = cctx->outBuffFlushedSize = 0;
	cctx->streamStage = zcss_load;
	cctx->frameEnded = 0;
	return 0; /* ready to go */
	}

	/* ZSTD_resetCStream():
	* pledgedSrcSize == 0 means "unknown" */
	size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pss)
	{
	/* temporary : 0 interpreted as "unknown" during transition period.
	* Users willing to specify "unknown" must use ZSTD_CONTENTSIZE_UNKNOWN.
	* 0 will be interpreted as "empty" in the future.
	*/
	U64 const pledgedSrcSize = (pss==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss;
	DEBUGLOG(4, "ZSTD_resetCStream: pledgedSrcSize = %u", (unsigned)pledgedSrcSize);
	FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
	return 0;
	}

	/*! ZSTD_initCStream_internal() :
	* Note : for lib/compress only. Used by zstdmt_compress.c.
	* Assumption 1 : params are valid
	* Assumption 2 : either dict, or cdict, is defined, not both */
	size_t ZSTD_initCStream_internal(ZSTD_CStream* zcs,
	const void* dict, size_t dictSize, const ZSTD_CDict* cdict,
	const ZSTD_CCtx_params* params,
	unsigned long long pledgedSrcSize)
	{
	DEBUGLOG(4, "ZSTD_initCStream_internal");
	FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
	assert(!ZSTD_isError(ZSTD_checkCParams(params->cParams)));
	zcs->requestedParams = *params;
	assert(!((dict) && (cdict))); /* either dict or cdict, not both */
	if (dict) {
	FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , "");
	} else {
	/* Dictionary is cleared if !cdict */
	FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , "");
	}
	return 0;
	}

	/* ZSTD_initCStream_usingCDict_advanced() :
	* same as ZSTD_initCStream_usingCDict(), with control over frame parameters */
	size_t ZSTD_initCStream_usingCDict_advanced(ZSTD_CStream* zcs,
	const ZSTD_CDict* cdict,
	ZSTD_frameParameters fParams,
	unsigned long long pledgedSrcSize)
	{
	DEBUGLOG(4, "ZSTD_initCStream_usingCDict_advanced");
	FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
	zcs->requestedParams.fParams = fParams;
	FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , "");
	return 0;
	}

	/* note : cdict must outlive compression session */
	size_t ZSTD_initCStream_usingCDict(ZSTD_CStream* zcs, const ZSTD_CDict* cdict)
	{
	DEBUGLOG(4, "ZSTD_initCStream_usingCDict");
	FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , "");
	return 0;
	}


	/* ZSTD_initCStream_advanced() :
	* pledgedSrcSize must be exact.
	* if srcSize is not known at init time, use value ZSTD_CONTENTSIZE_UNKNOWN.
	* dict is loaded with default parameters ZSTD_dct_auto and ZSTD_dlm_byCopy. */
	size_t ZSTD_initCStream_advanced(ZSTD_CStream* zcs,
	const void* dict, size_t dictSize,
	ZSTD_parameters params, unsigned long long pss)
	{
	/* for compatibility with older programs relying on this behavior.
	* Users should now specify ZSTD_CONTENTSIZE_UNKNOWN.
	* This line will be removed in the future.
	*/
	U64 const pledgedSrcSize = (pss==0 && params.fParams.contentSizeFlag==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss;
	DEBUGLOG(4, "ZSTD_initCStream_advanced");
	FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
	FORWARD_IF_ERROR( ZSTD_checkCParams(params.cParams) , "");
	zcs->requestedParams = ZSTD_assignParamsToCCtxParams(&zcs->requestedParams, &params);
	FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , "");
	return 0;
	}

	size_t ZSTD_initCStream_usingDict(ZSTD_CStream* zcs, const void* dict, size_t dictSize, int compressionLevel)
	{
	DEBUGLOG(4, "ZSTD_initCStream_usingDict");
	FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , "");
	return 0;
	}

	size_t ZSTD_initCStream_srcSize(ZSTD_CStream* zcs, int compressionLevel, unsigned long long pss)
	{
	/* temporary : 0 interpreted as "unknown" during transition period.
	* Users willing to specify "unknown" must use ZSTD_CONTENTSIZE_UNKNOWN.
	* 0 will be interpreted as "empty" in the future.
	*/
	U64 const pledgedSrcSize = (pss==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss;
	DEBUGLOG(4, "ZSTD_initCStream_srcSize");
	FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, NULL) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
	return 0;
	}

	size_t ZSTD_initCStream(ZSTD_CStream* zcs, int compressionLevel)
	{
	DEBUGLOG(4, "ZSTD_initCStream");
	FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, NULL) , "");
	FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , "");
	return 0;
	}

	/====== Compression ======/

	static size_t ZSTD_nextInputSizeHint(const ZSTD_CCtx* cctx)
	{
	size_t hintInSize = cctx->inBuffTarget - cctx->inBuffPos;
	if (hintInSize==0) hintInSize = cctx->blockSize;
	return hintInSize;
	}

	/** ZSTD_compressStream_generic():
	* internal function for all compressStream() variants
	* non-static, because can be called from zstdmt_compress.c
	* @return : hint size for next input */
	static size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs,
	ZSTD_outBuffer* output,
	ZSTD_inBuffer* input,
	ZSTD_EndDirective const flushMode)
	{
	const char* const istart = (const char*)input->src;
	const char* const iend = input->size != 0 ? istart + input->size : istart;
	const char* ip = input->pos != 0 ? istart + input->pos : istart;
	char* const ostart = (char*)output->dst;
	char* const oend = output->size != 0 ? ostart + output->size : ostart;
	char* op = output->pos != 0 ? ostart + output->pos : ostart;
	U32 someMoreWork = 1;

	/* check expectations */
	DEBUGLOG(5, "ZSTD_compressStream_generic, flush=%u", (unsigned)flushMode);
	assert(zcs->inBuff != NULL);
	assert(zcs->inBuffSize > 0);
	assert(zcs->outBuff != NULL);
	assert(zcs->outBuffSize > 0);
	assert(output->pos <= output->size);
	assert(input->pos <= input->size);

	while (someMoreWork) {
	switch(zcs->streamStage)
	{
	case zcss_init:
	RETURN_ERROR(init_missing, "call ZSTD_initCStream() first!");

	case zcss_load:
	if ( (flushMode == ZSTD_e_end)
	&& ((size_t)(oend-op) >= ZSTD_compressBound(iend-ip)) /* enough dstCapacity */
	&& (zcs->inBuffPos == 0) ) {
	/* shortcut to compression pass directly into output buffer */
	size_t const cSize = ZSTD_compressEnd(zcs,
	op, oend-op, ip, iend-ip);
	DEBUGLOG(4, "ZSTD_compressEnd : cSize=%u", (unsigned)cSize);
	FORWARD_IF_ERROR(cSize, "ZSTD_compressEnd failed");
	ip = iend;
	op += cSize;
	zcs->frameEnded = 1;
	ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
	someMoreWork = 0; break;
	}
	/* complete loading into inBuffer */
	{ size_t const toLoad = zcs->inBuffTarget - zcs->inBuffPos;
	size_t const loaded = ZSTD_limitCopy(
	zcs->inBuff + zcs->inBuffPos, toLoad,
	ip, iend-ip);
	zcs->inBuffPos += loaded;
	if (loaded != 0)
	ip += loaded;
	if ( (flushMode == ZSTD_e_continue)
	&& (zcs->inBuffPos < zcs->inBuffTarget) ) {
	/* not enough input to fill full block : stop here */
	someMoreWork = 0; break;
	}
	if ( (flushMode == ZSTD_e_flush)
	&& (zcs->inBuffPos == zcs->inToCompress) ) {
	/* empty */
	someMoreWork = 0; break;
	}
	}
	/* compress current block (note : this stage cannot be stopped in the middle) */
	DEBUGLOG(5, "stream compression stage (flushMode==%u)", flushMode);
	{ void* cDst;
	size_t cSize;
	size_t const iSize = zcs->inBuffPos - zcs->inToCompress;
	size_t oSize = oend-op;
	unsigned const lastBlock = (flushMode == ZSTD_e_end) && (ip==iend);
	if (oSize >= ZSTD_compressBound(iSize))
	cDst = op; /* compress into output buffer, to skip flush stage */
	else
	cDst = zcs->outBuff, oSize = zcs->outBuffSize;
	cSize = lastBlock ?
	ZSTD_compressEnd(zcs, cDst, oSize,
	zcs->inBuff + zcs->inToCompress, iSize) :
	ZSTD_compressContinue(zcs, cDst, oSize,
	zcs->inBuff + zcs->inToCompress, iSize);
	FORWARD_IF_ERROR(cSize, "%s", lastBlock ? "ZSTD_compressEnd failed" : "ZSTD_compressContinue failed");
	zcs->frameEnded = lastBlock;
	/* prepare next block */
	zcs->inBuffTarget = zcs->inBuffPos + zcs->blockSize;
	if (zcs->inBuffTarget > zcs->inBuffSize)
	zcs->inBuffPos = 0, zcs->inBuffTarget = zcs->blockSize;
	DEBUGLOG(5, "inBuffTarget:%u / inBuffSize:%u",
	(unsigned)zcs->inBuffTarget, (unsigned)zcs->inBuffSize);
	if (!lastBlock)
	assert(zcs->inBuffTarget <= zcs->inBuffSize);
	zcs->inToCompress = zcs->inBuffPos;
	if (cDst == op) { /* no need to flush */
	op += cSize;
	if (zcs->frameEnded) {
	DEBUGLOG(5, "Frame completed directly in outBuffer");
	someMoreWork = 0;
	ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
	}
	break;
	}
	zcs->outBuffContentSize = cSize;
	zcs->outBuffFlushedSize = 0;
	zcs->streamStage = zcss_flush; /* pass-through to flush stage */
	}
	/* fall-through */
	case zcss_flush:
	DEBUGLOG(5, "flush stage");
	{ size_t const toFlush = zcs->outBuffContentSize - zcs->outBuffFlushedSize;
	size_t const flushed = ZSTD_limitCopy(op, (size_t)(oend-op),
	zcs->outBuff + zcs->outBuffFlushedSize, toFlush);
	DEBUGLOG(5, "toFlush: %u into %u ==> flushed: %u",
	(unsigned)toFlush, (unsigned)(oend-op), (unsigned)flushed);
	if (flushed)
	op += flushed;
	zcs->outBuffFlushedSize += flushed;
	if (toFlush!=flushed) {
	/* flush not fully completed, presumably because dst is too small */
	assert(op==oend);
	someMoreWork = 0;
	break;
	}
	zcs->outBuffContentSize = zcs->outBuffFlushedSize = 0;
	if (zcs->frameEnded) {
	DEBUGLOG(5, "Frame completed on flush");
	someMoreWork = 0;
	ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only);
	break;
	}
	zcs->streamStage = zcss_load;
	break;
	}

	default: /* impossible */
	assert(0);
	}
	}

	input->pos = ip - istart;
	output->pos = op - ostart;
	if (zcs->frameEnded) return 0;
	return ZSTD_nextInputSizeHint(zcs);
	}

	static size_t ZSTD_nextInputSizeHint_MTorST(const ZSTD_CCtx* cctx)
	{
	#ifdef ZSTD_MULTITHREAD
	if (cctx->appliedParams.nbWorkers >= 1) {
	assert(cctx->mtctx != NULL);
	return ZSTDMT_nextInputSizeHint(cctx->mtctx);
	}
	#endif
	return ZSTD_nextInputSizeHint(cctx);

	}

	size_t ZSTD_compressStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output, ZSTD_inBuffer* input)
	{
	FORWARD_IF_ERROR( ZSTD_compressStream2(zcs, output, input, ZSTD_e_continue) , "");
	return ZSTD_nextInputSizeHint_MTorST(zcs);
	}


	size_t ZSTD_compressStream2( ZSTD_CCtx* cctx,
	ZSTD_outBuffer* output,
	ZSTD_inBuffer* input,
	ZSTD_EndDirective endOp)
	{
	DEBUGLOG(5, "ZSTD_compressStream2, endOp=%u ", (unsigned)endOp);
	/* check conditions */
	RETURN_ERROR_IF(output->pos > output->size, GENERIC, "invalid buffer");
	RETURN_ERROR_IF(input->pos > input->size, GENERIC, "invalid buffer");
	assert(cctx!=NULL);

	/* transparent initialization stage */
	if (cctx->streamStage == zcss_init) {
	ZSTD_CCtx_params params = cctx->requestedParams;
	ZSTD_prefixDict const prefixDict = cctx->prefixDict;
	FORWARD_IF_ERROR( ZSTD_initLocalDict(cctx) , ""); /* Init the local dict if present. */
	memset(&cctx->prefixDict, 0, sizeof(cctx->prefixDict)); /* single usage */
	assert(prefixDict.dict==NULL \|\| cctx->cdict==NULL); /* only one can be set */
	DEBUGLOG(4, "ZSTD_compressStream2 : transparent init stage");
	if (endOp == ZSTD_e_end) cctx->pledgedSrcSizePlusOne = input->size + 1; /* auto-fix pledgedSrcSize */
	params.cParams = ZSTD_getCParamsFromCCtxParams(
	&cctx->requestedParams, cctx->pledgedSrcSizePlusOne-1, 0 /dictSize/);


	#ifdef ZSTD_MULTITHREAD
	if ((cctx->pledgedSrcSizePlusOne-1) <= ZSTDMT_JOBSIZE_MIN) {
	params.nbWorkers = 0; /* do not invoke multi-threading when src size is too small */
	}
	if (params.nbWorkers > 0) {
	/* mt context creation */
	if (cctx->mtctx == NULL) {
	DEBUGLOG(4, "ZSTD_compressStream2: creating new mtctx for nbWorkers=%u",
	params.nbWorkers);
	cctx->mtctx = ZSTDMT_createCCtx_advanced((U32)params.nbWorkers, cctx->customMem);
	RETURN_ERROR_IF(cctx->mtctx == NULL, memory_allocation, "NULL pointer!");
	}
	/* mt compression */
	DEBUGLOG(4, "call ZSTDMT_initCStream_internal as nbWorkers=%u", params.nbWorkers);
	FORWARD_IF_ERROR( ZSTDMT_initCStream_internal(
	cctx->mtctx,
	prefixDict.dict, prefixDict.dictSize, prefixDict.dictContentType,
	cctx->cdict, params, cctx->pledgedSrcSizePlusOne-1) , "");
	cctx->streamStage = zcss_load;
	cctx->appliedParams.nbWorkers = params.nbWorkers;
	} else
	#endif
	{ FORWARD_IF_ERROR( ZSTD_resetCStream_internal(cctx,
	prefixDict.dict, prefixDict.dictSize, prefixDict.dictContentType,
	cctx->cdict,
	params, cctx->pledgedSrcSizePlusOne-1) , "");
	assert(cctx->streamStage == zcss_load);
	assert(cctx->appliedParams.nbWorkers == 0);
	} }
	/* end of transparent initialization stage */

	/* compression stage */
	#ifdef ZSTD_MULTITHREAD
	if (cctx->appliedParams.nbWorkers > 0) {
	int const forceMaxProgress = (endOp == ZSTD_e_flush \|\| endOp == ZSTD_e_end);
	size_t flushMin;
	assert(forceMaxProgress \|\| endOp == ZSTD_e_continue /* Protection for a new flush type */);
	if (cctx->cParamsChanged) {
	ZSTDMT_updateCParams_whileCompressing(cctx->mtctx, &cctx->requestedParams);
	cctx->cParamsChanged = 0;
	}
	do {
	flushMin = ZSTDMT_compressStream_generic(cctx->mtctx, output, input, endOp);
	if ( ZSTD_isError(flushMin)
	\|\| (endOp == ZSTD_e_end && flushMin == 0) ) { /* compression completed */
	ZSTD_CCtx_reset(cctx, ZSTD_reset_session_only);
	}
	FORWARD_IF_ERROR(flushMin, "ZSTDMT_compressStream_generic failed");
	} while (forceMaxProgress && flushMin != 0 && output->pos < output->size);
	DEBUGLOG(5, "completed ZSTD_compressStream2 delegating to ZSTDMT_compressStream_generic");
	/* Either we don't require maximum forward progress, we've finished the
	* flush, or we are out of output space.
	*/
	assert(!forceMaxProgress \|\| flushMin == 0 \|\| output->pos == output->size);
	return flushMin;
	}
	#endif
	FORWARD_IF_ERROR( ZSTD_compressStream_generic(cctx, output, input, endOp) , "");
	DEBUGLOG(5, "completed ZSTD_compressStream2");
	return cctx->outBuffContentSize - cctx->outBuffFlushedSize; /* remaining to flush */
	}

	size_t ZSTD_compressStream2_simpleArgs (
	ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity, size_t* dstPos,
	const void* src, size_t srcSize, size_t* srcPos,
	ZSTD_EndDirective endOp)
	{
	ZSTD_outBuffer output = { dst, dstCapacity, *dstPos };
	ZSTD_inBuffer input = { src, srcSize, *srcPos };
	/* ZSTD_compressStream2() will check validity of dstPos and srcPos */
	size_t const cErr = ZSTD_compressStream2(cctx, &output, &input, endOp);
	*dstPos = output.pos;
	*srcPos = input.pos;
	return cErr;
	}

	size_t ZSTD_compress2(ZSTD_CCtx* cctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize)
	{
	DEBUGLOG(4, "ZSTD_compress2 (srcSize=%u)", (unsigned)srcSize);
	ZSTD_CCtx_reset(cctx, ZSTD_reset_session_only);
	{ size_t oPos = 0;
	size_t iPos = 0;
	size_t const result = ZSTD_compressStream2_simpleArgs(cctx,
	dst, dstCapacity, &oPos,
	src, srcSize, &iPos,
	ZSTD_e_end);
	FORWARD_IF_ERROR(result, "ZSTD_compressStream2_simpleArgs failed");
	if (result != 0) { /* compression not completed, due to lack of output space */
	assert(oPos == dstCapacity);
	RETURN_ERROR(dstSize_tooSmall, "");
	}
	assert(iPos == srcSize); /* all input is expected consumed */
	return oPos;
	}
	}

	/====== Finalize ======/

	/*! ZSTD_flushStream() :
	* @return : amount of data remaining to flush */
	size_t ZSTD_flushStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output)
	{
	ZSTD_inBuffer input = { NULL, 0, 0 };
	return ZSTD_compressStream2(zcs, output, &input, ZSTD_e_flush);
	}


	size_t ZSTD_endStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output)
	{
	ZSTD_inBuffer input = { NULL, 0, 0 };
	size_t const remainingToFlush = ZSTD_compressStream2(zcs, output, &input, ZSTD_e_end);
	FORWARD_IF_ERROR( remainingToFlush , "ZSTD_compressStream2 failed");
	if (zcs->appliedParams.nbWorkers > 0) return remainingToFlush; /* minimal estimation */
	/* single thread mode : attempt to calculate remaining to flush more precisely */
	{ size_t const lastBlockSize = zcs->frameEnded ? 0 : ZSTD_BLOCKHEADERSIZE;
	size_t const checksumSize = (size_t)(zcs->frameEnded ? 0 : zcs->appliedParams.fParams.checksumFlag * 4);
	size_t const toFlush = remainingToFlush + lastBlockSize + checksumSize;
	DEBUGLOG(4, "ZSTD_endStream : remaining to flush : %u", (unsigned)toFlush);
	return toFlush;
	}
	}


	/-===== Pre-defined compression levels =====-/

	#define ZSTD_MAX_CLEVEL 22
	int ZSTD_maxCLevel(void) { return ZSTD_MAX_CLEVEL; }
	int ZSTD_minCLevel(void) { return (int)-ZSTD_TARGETLENGTH_MAX; }

	static const ZSTD_compressionParameters ZSTD_defaultCParameters[4][ZSTD_MAX_CLEVEL+1] = {
	{ /* "default" - for any srcSize > 256 KB */
	/* W, C, H, S, L, TL, strat */
	{ 19, 12, 13, 1, 6, 1, ZSTD_fast }, /* base for negative levels */
	{ 19, 13, 14, 1, 7, 0, ZSTD_fast }, /* level 1 */
	{ 20, 15, 16, 1, 6, 0, ZSTD_fast }, /* level 2 */
	{ 21, 16, 17, 1, 5, 0, ZSTD_dfast }, /* level 3 */
	{ 21, 18, 18, 1, 5, 0, ZSTD_dfast }, /* level 4 */
	{ 21, 18, 19, 2, 5, 2, ZSTD_greedy }, /* level 5 */
	{ 21, 19, 19, 3, 5, 4, ZSTD_greedy }, /* level 6 */
	{ 21, 19, 19, 3, 5, 8, ZSTD_lazy }, /* level 7 */
	{ 21, 19, 19, 3, 5, 16, ZSTD_lazy2 }, /* level 8 */
	{ 21, 19, 20, 4, 5, 16, ZSTD_lazy2 }, /* level 9 */
	{ 22, 20, 21, 4, 5, 16, ZSTD_lazy2 }, /* level 10 */
	{ 22, 21, 22, 4, 5, 16, ZSTD_lazy2 }, /* level 11 */
	{ 22, 21, 22, 5, 5, 16, ZSTD_lazy2 }, /* level 12 */
	{ 22, 21, 22, 5, 5, 32, ZSTD_btlazy2 }, /* level 13 */
	{ 22, 22, 23, 5, 5, 32, ZSTD_btlazy2 }, /* level 14 */
	{ 22, 23, 23, 6, 5, 32, ZSTD_btlazy2 }, /* level 15 */
	{ 22, 22, 22, 5, 5, 48, ZSTD_btopt }, /* level 16 */
	{ 23, 23, 22, 5, 4, 64, ZSTD_btopt }, /* level 17 */
	{ 23, 23, 22, 6, 3, 64, ZSTD_btultra }, /* level 18 */
	{ 23, 24, 22, 7, 3,256, ZSTD_btultra2}, /* level 19 */
	{ 25, 25, 23, 7, 3,256, ZSTD_btultra2}, /* level 20 */
	{ 26, 26, 24, 7, 3,512, ZSTD_btultra2}, /* level 21 */
	{ 27, 27, 25, 9, 3,999, ZSTD_btultra2}, /* level 22 */
	},
	{ /* for srcSize <= 256 KB */
	/* W, C, H, S, L, T, strat */
	{ 18, 12, 13, 1, 5, 1, ZSTD_fast }, /* base for negative levels */
	{ 18, 13, 14, 1, 6, 0, ZSTD_fast }, /* level 1 */
	{ 18, 14, 14, 1, 5, 0, ZSTD_dfast }, /* level 2 */
	{ 18, 16, 16, 1, 4, 0, ZSTD_dfast }, /* level 3 */
	{ 18, 16, 17, 2, 5, 2, ZSTD_greedy }, /* level 4.*/
	{ 18, 18, 18, 3, 5, 2, ZSTD_greedy }, /* level 5.*/
	{ 18, 18, 19, 3, 5, 4, ZSTD_lazy }, /* level 6.*/
	{ 18, 18, 19, 4, 4, 4, ZSTD_lazy }, /* level 7 */
	{ 18, 18, 19, 4, 4, 8, ZSTD_lazy2 }, /* level 8 */
	{ 18, 18, 19, 5, 4, 8, ZSTD_lazy2 }, /* level 9 */
	{ 18, 18, 19, 6, 4, 8, ZSTD_lazy2 }, /* level 10 */
	{ 18, 18, 19, 5, 4, 12, ZSTD_btlazy2 }, /* level 11.*/
	{ 18, 19, 19, 7, 4, 12, ZSTD_btlazy2 }, /* level 12.*/
	{ 18, 18, 19, 4, 4, 16, ZSTD_btopt }, /* level 13 */
	{ 18, 18, 19, 4, 3, 32, ZSTD_btopt }, /* level 14.*/
	{ 18, 18, 19, 6, 3,128, ZSTD_btopt }, /* level 15.*/
	{ 18, 19, 19, 6, 3,128, ZSTD_btultra }, /* level 16.*/
	{ 18, 19, 19, 8, 3,256, ZSTD_btultra }, /* level 17.*/
	{ 18, 19, 19, 6, 3,128, ZSTD_btultra2}, /* level 18.*/
	{ 18, 19, 19, 8, 3,256, ZSTD_btultra2}, /* level 19.*/
	{ 18, 19, 19, 10, 3,512, ZSTD_btultra2}, /* level 20.*/
	{ 18, 19, 19, 12, 3,512, ZSTD_btultra2}, /* level 21.*/
	{ 18, 19, 19, 13, 3,999, ZSTD_btultra2}, /* level 22.*/
	},
	{ /* for srcSize <= 128 KB */
	/* W, C, H, S, L, T, strat */
	{ 17, 12, 12, 1, 5, 1, ZSTD_fast }, /* base for negative levels */
	{ 17, 12, 13, 1, 6, 0, ZSTD_fast }, /* level 1 */
	{ 17, 13, 15, 1, 5, 0, ZSTD_fast }, /* level 2 */
	{ 17, 15, 16, 2, 5, 0, ZSTD_dfast }, /* level 3 */
	{ 17, 17, 17, 2, 4, 0, ZSTD_dfast }, /* level 4 */
	{ 17, 16, 17, 3, 4, 2, ZSTD_greedy }, /* level 5 */
	{ 17, 17, 17, 3, 4, 4, ZSTD_lazy }, /* level 6 */
	{ 17, 17, 17, 3, 4, 8, ZSTD_lazy2 }, /* level 7 */
	{ 17, 17, 17, 4, 4, 8, ZSTD_lazy2 }, /* level 8 */
	{ 17, 17, 17, 5, 4, 8, ZSTD_lazy2 }, /* level 9 */
	{ 17, 17, 17, 6, 4, 8, ZSTD_lazy2 }, /* level 10 */
	{ 17, 17, 17, 5, 4, 8, ZSTD_btlazy2 }, /* level 11 */
	{ 17, 18, 17, 7, 4, 12, ZSTD_btlazy2 }, /* level 12 */
	{ 17, 18, 17, 3, 4, 12, ZSTD_btopt }, /* level 13.*/
	{ 17, 18, 17, 4, 3, 32, ZSTD_btopt }, /* level 14.*/
	{ 17, 18, 17, 6, 3,256, ZSTD_btopt }, /* level 15.*/
	{ 17, 18, 17, 6, 3,128, ZSTD_btultra }, /* level 16.*/
	{ 17, 18, 17, 8, 3,256, ZSTD_btultra }, /* level 17.*/
	{ 17, 18, 17, 10, 3,512, ZSTD_btultra }, /* level 18.*/
	{ 17, 18, 17, 5, 3,256, ZSTD_btultra2}, /* level 19.*/
	{ 17, 18, 17, 7, 3,512, ZSTD_btultra2}, /* level 20.*/
	{ 17, 18, 17, 9, 3,512, ZSTD_btultra2}, /* level 21.*/
	{ 17, 18, 17, 11, 3,999, ZSTD_btultra2}, /* level 22.*/
	},
	{ /* for srcSize <= 16 KB */
	/* W, C, H, S, L, T, strat */
	{ 14, 12, 13, 1, 5, 1, ZSTD_fast }, /* base for negative levels */
	{ 14, 14, 15, 1, 5, 0, ZSTD_fast }, /* level 1 */
	{ 14, 14, 15, 1, 4, 0, ZSTD_fast }, /* level 2 */
	{ 14, 14, 15, 2, 4, 0, ZSTD_dfast }, /* level 3 */
	{ 14, 14, 14, 4, 4, 2, ZSTD_greedy }, /* level 4 */
	{ 14, 14, 14, 3, 4, 4, ZSTD_lazy }, /* level 5.*/
	{ 14, 14, 14, 4, 4, 8, ZSTD_lazy2 }, /* level 6 */
	{ 14, 14, 14, 6, 4, 8, ZSTD_lazy2 }, /* level 7 */
	{ 14, 14, 14, 8, 4, 8, ZSTD_lazy2 }, /* level 8.*/
	{ 14, 15, 14, 5, 4, 8, ZSTD_btlazy2 }, /* level 9.*/
	{ 14, 15, 14, 9, 4, 8, ZSTD_btlazy2 }, /* level 10.*/
	{ 14, 15, 14, 3, 4, 12, ZSTD_btopt }, /* level 11.*/
	{ 14, 15, 14, 4, 3, 24, ZSTD_btopt }, /* level 12.*/
	{ 14, 15, 14, 5, 3, 32, ZSTD_btultra }, /* level 13.*/
	{ 14, 15, 15, 6, 3, 64, ZSTD_btultra }, /* level 14.*/
	{ 14, 15, 15, 7, 3,256, ZSTD_btultra }, /* level 15.*/
	{ 14, 15, 15, 5, 3, 48, ZSTD_btultra2}, /* level 16.*/
	{ 14, 15, 15, 6, 3,128, ZSTD_btultra2}, /* level 17.*/
	{ 14, 15, 15, 7, 3,256, ZSTD_btultra2}, /* level 18.*/
	{ 14, 15, 15, 8, 3,256, ZSTD_btultra2}, /* level 19.*/
	{ 14, 15, 15, 8, 3,512, ZSTD_btultra2}, /* level 20.*/
	{ 14, 15, 15, 9, 3,512, ZSTD_btultra2}, /* level 21.*/
	{ 14, 15, 15, 10, 3,999, ZSTD_btultra2}, /* level 22.*/
	},
	};

	/*! ZSTD_getCParams_internal() :
	* @return ZSTD_compressionParameters structure for a selected compression level, srcSize and dictSize.
	* Note: srcSizeHint 0 means 0, use ZSTD_CONTENTSIZE_UNKNOWN for unknown.
	* Use dictSize == 0 for unknown or unused. */
	static ZSTD_compressionParameters ZSTD_getCParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize)
	{
	int const unknown = srcSizeHint == ZSTD_CONTENTSIZE_UNKNOWN;
	size_t const addedSize = unknown && dictSize > 0 ? 500 : 0;
	U64 const rSize = unknown && dictSize == 0 ? ZSTD_CONTENTSIZE_UNKNOWN : srcSizeHint+dictSize+addedSize;
	U32 const tableID = (rSize <= 256 KB) + (rSize <= 128 KB) + (rSize <= 16 KB);
	int row = compressionLevel;
	DEBUGLOG(5, "ZSTD_getCParams_internal (cLevel=%i)", compressionLevel);
	if (compressionLevel == 0) row = ZSTD_CLEVEL_DEFAULT; /* 0 == default */
	if (compressionLevel < 0) row = 0; /* entry 0 is baseline for fast mode */
	if (compressionLevel > ZSTD_MAX_CLEVEL) row = ZSTD_MAX_CLEVEL;
	{ ZSTD_compressionParameters cp = ZSTD_defaultCParameters[tableID][row];
	if (compressionLevel < 0) cp.targetLength = (unsigned)(-compressionLevel); /* acceleration factor */
	/* refine parameters based on srcSize & dictSize */
	return ZSTD_adjustCParams_internal(cp, srcSizeHint, dictSize);
	}
	}

	/*! ZSTD_getCParams() :
	* @return ZSTD_compressionParameters structure for a selected compression level, srcSize and dictSize.
	* Size values are optional, provide 0 if not known or unused */
	ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize)
	{
	if (srcSizeHint == 0) srcSizeHint = ZSTD_CONTENTSIZE_UNKNOWN;
	return ZSTD_getCParams_internal(compressionLevel, srcSizeHint, dictSize);
	}

	/*! ZSTD_getParams() :
	* same idea as ZSTD_getCParams()
	* @return a `ZSTD_parameters` structure (instead of `ZSTD_compressionParameters`).
	* Fields of `ZSTD_frameParameters` are set to default values */
	static ZSTD_parameters ZSTD_getParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize) {
	ZSTD_parameters params;
	ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, srcSizeHint, dictSize);
	DEBUGLOG(5, "ZSTD_getParams (cLevel=%i)", compressionLevel);
	memset(&params, 0, sizeof(params));
	params.cParams = cParams;
	params.fParams.contentSizeFlag = 1;
	return params;
	}

	/*! ZSTD_getParams() :
	* same idea as ZSTD_getCParams()
	* @return a `ZSTD_parameters` structure (instead of `ZSTD_compressionParameters`).
	* Fields of `ZSTD_frameParameters` are set to default values */
	ZSTD_parameters ZSTD_getParams(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize) {
	if (srcSizeHint == 0) srcSizeHint = ZSTD_CONTENTSIZE_UNKNOWN;
	return ZSTD_getParams_internal(compressionLevel, srcSizeHint, dictSize);
	}
	/** ended inlining compress/zstd_compress.c **/
	/** start inlining compress/zstd_double_fast.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/** skipping file: zstd_compress_internal.h **/
	/** skipping file: zstd_double_fast.h **/


	void ZSTD_fillDoubleHashTable(ZSTD_matchState_t* ms,
	void const* end, ZSTD_dictTableLoadMethod_e dtlm)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32* const hashLarge = ms->hashTable;
	U32 const hBitsL = cParams->hashLog;
	U32 const mls = cParams->minMatch;
	U32* const hashSmall = ms->chainTable;
	U32 const hBitsS = cParams->chainLog;
	const BYTE* const base = ms->window.base;
	const BYTE* ip = base + ms->nextToUpdate;
	const BYTE* const iend = ((const BYTE*)end) - HASH_READ_SIZE;
	const U32 fastHashFillStep = 3;

	/* Always insert every fastHashFillStep position into the hash tables.
	* Insert the other positions into the large hash table if their entry
	* is empty.
	*/
	for (; ip + fastHashFillStep - 1 <= iend; ip += fastHashFillStep) {
	U32 const current = (U32)(ip - base);
	U32 i;
	for (i = 0; i < fastHashFillStep; ++i) {
	size_t const smHash = ZSTD_hashPtr(ip + i, hBitsS, mls);
	size_t const lgHash = ZSTD_hashPtr(ip + i, hBitsL, 8);
	if (i == 0)
	hashSmall[smHash] = current + i;
	if (i == 0 \|\| hashLarge[lgHash] == 0)
	hashLarge[lgHash] = current + i;
	/* Only load extra positions for ZSTD_dtlm_full */
	if (dtlm == ZSTD_dtlm_fast)
	break;
	} }
	}


	FORCE_INLINE_TEMPLATE
	size_t ZSTD_compressBlock_doubleFast_generic(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize,
	U32 const mls /* template */, ZSTD_dictMode_e const dictMode)
	{
	ZSTD_compressionParameters const* cParams = &ms->cParams;
	U32* const hashLong = ms->hashTable;
	const U32 hBitsL = cParams->hashLog;
	U32* const hashSmall = ms->chainTable;
	const U32 hBitsS = cParams->chainLog;
	const BYTE* const base = ms->window.base;
	const BYTE* const istart = (const BYTE*)src;
	const BYTE* ip = istart;
	const BYTE* anchor = istart;
	const U32 endIndex = (U32)((size_t)(istart - base) + srcSize);
	/* presumes that, if there is a dictionary, it must be using Attach mode */
	const U32 prefixLowestIndex = ZSTD_getLowestPrefixIndex(ms, endIndex, cParams->windowLog);
	const BYTE* const prefixLowest = base + prefixLowestIndex;
	const BYTE* const iend = istart + srcSize;
	const BYTE* const ilimit = iend - HASH_READ_SIZE;
	U32 offset_1=rep[0], offset_2=rep[1];
	U32 offsetSaved = 0;

	const ZSTD_matchState_t* const dms = ms->dictMatchState;
	const ZSTD_compressionParameters* const dictCParams =
	dictMode == ZSTD_dictMatchState ?
	&dms->cParams : NULL;
	const U32* const dictHashLong = dictMode == ZSTD_dictMatchState ?
	dms->hashTable : NULL;
	const U32* const dictHashSmall = dictMode == ZSTD_dictMatchState ?
	dms->chainTable : NULL;
	const U32 dictStartIndex = dictMode == ZSTD_dictMatchState ?
	dms->window.dictLimit : 0;
	const BYTE* const dictBase = dictMode == ZSTD_dictMatchState ?
	dms->window.base : NULL;
	const BYTE* const dictStart = dictMode == ZSTD_dictMatchState ?
	dictBase + dictStartIndex : NULL;
	const BYTE* const dictEnd = dictMode == ZSTD_dictMatchState ?
	dms->window.nextSrc : NULL;
	const U32 dictIndexDelta = dictMode == ZSTD_dictMatchState ?
	prefixLowestIndex - (U32)(dictEnd - dictBase) :
	0;
	const U32 dictHBitsL = dictMode == ZSTD_dictMatchState ?
	dictCParams->hashLog : hBitsL;
	const U32 dictHBitsS = dictMode == ZSTD_dictMatchState ?
	dictCParams->chainLog : hBitsS;
	const U32 dictAndPrefixLength = (U32)((ip - prefixLowest) + (dictEnd - dictStart));

	DEBUGLOG(5, "ZSTD_compressBlock_doubleFast_generic");

	assert(dictMode == ZSTD_noDict \|\| dictMode == ZSTD_dictMatchState);

	/* if a dictionary is attached, it must be within window range */
	if (dictMode == ZSTD_dictMatchState) {
	assert(ms->window.dictLimit + (1U << cParams->windowLog) >= endIndex);
	}

	/* init */
	ip += (dictAndPrefixLength == 0);
	if (dictMode == ZSTD_noDict) {
	U32 const current = (U32)(ip - base);
	U32 const windowLow = ZSTD_getLowestPrefixIndex(ms, current, cParams->windowLog);
	U32 const maxRep = current - windowLow;
	if (offset_2 > maxRep) offsetSaved = offset_2, offset_2 = 0;
	if (offset_1 > maxRep) offsetSaved = offset_1, offset_1 = 0;
	}
	if (dictMode == ZSTD_dictMatchState) {
	/* dictMatchState repCode checks don't currently handle repCode == 0
	* disabling. */
	assert(offset_1 <= dictAndPrefixLength);
	assert(offset_2 <= dictAndPrefixLength);
	}

	/* Main Search Loop */
	while (ip < ilimit) { /* < instead of <=, because repcode check at (ip+1) */
	size_t mLength;
	U32 offset;
	size_t const h2 = ZSTD_hashPtr(ip, hBitsL, 8);
	size_t const h = ZSTD_hashPtr(ip, hBitsS, mls);
	size_t const dictHL = ZSTD_hashPtr(ip, dictHBitsL, 8);
	size_t const dictHS = ZSTD_hashPtr(ip, dictHBitsS, mls);
	U32 const current = (U32)(ip-base);
	U32 const matchIndexL = hashLong[h2];
	U32 matchIndexS = hashSmall[h];
	const BYTE* matchLong = base + matchIndexL;
	const BYTE* match = base + matchIndexS;
	const U32 repIndex = current + 1 - offset_1;
	const BYTE* repMatch = (dictMode == ZSTD_dictMatchState
	&& repIndex < prefixLowestIndex) ?
	dictBase + (repIndex - dictIndexDelta) :
	base + repIndex;
	hashLong[h2] = hashSmall[h] = current; /* update hash tables */

	/* check dictMatchState repcode */
	if (dictMode == ZSTD_dictMatchState
	&& ((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
	&& (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
	const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
	mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
	ip++;
	ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
	goto _match_stored;
	}

	/* check noDict repcode */
	if ( dictMode == ZSTD_noDict
	&& ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1)))) {
	mLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4;
	ip++;
	ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
	goto _match_stored;
	}

	if (matchIndexL > prefixLowestIndex) {
	/* check prefix long match */
	if (MEM_read64(matchLong) == MEM_read64(ip)) {
	mLength = ZSTD_count(ip+8, matchLong+8, iend) + 8;
	offset = (U32)(ip-matchLong);
	while (((ip>anchor) & (matchLong>prefixLowest)) && (ip[-1] == matchLong[-1])) { ip--; matchLong--; mLength++; } /* catch up */
	goto _match_found;
	}
	} else if (dictMode == ZSTD_dictMatchState) {
	/* check dictMatchState long match */
	U32 const dictMatchIndexL = dictHashLong[dictHL];
	const BYTE* dictMatchL = dictBase + dictMatchIndexL;
	assert(dictMatchL < dictEnd);

	if (dictMatchL > dictStart && MEM_read64(dictMatchL) == MEM_read64(ip)) {
	mLength = ZSTD_count_2segments(ip+8, dictMatchL+8, iend, dictEnd, prefixLowest) + 8;
	offset = (U32)(current - dictMatchIndexL - dictIndexDelta);
	while (((ip>anchor) & (dictMatchL>dictStart)) && (ip[-1] == dictMatchL[-1])) { ip--; dictMatchL--; mLength++; } /* catch up */
	goto _match_found;
	} }

	if (matchIndexS > prefixLowestIndex) {
	/* check prefix short match */
	if (MEM_read32(match) == MEM_read32(ip)) {
	goto _search_next_long;
	}
	} else if (dictMode == ZSTD_dictMatchState) {
	/* check dictMatchState short match */
	U32 const dictMatchIndexS = dictHashSmall[dictHS];
	match = dictBase + dictMatchIndexS;
	matchIndexS = dictMatchIndexS + dictIndexDelta;

	if (match > dictStart && MEM_read32(match) == MEM_read32(ip)) {
	goto _search_next_long;
	} }

	ip += ((ip-anchor) >> kSearchStrength) + 1;
	#if defined(__aarch64__)
	PREFETCH_L1(ip+256);
	#endif
	continue;

	_search_next_long:

	{ size_t const hl3 = ZSTD_hashPtr(ip+1, hBitsL, 8);
	size_t const dictHLNext = ZSTD_hashPtr(ip+1, dictHBitsL, 8);
	U32 const matchIndexL3 = hashLong[hl3];
	const BYTE* matchL3 = base + matchIndexL3;
	hashLong[hl3] = current + 1;

	/* check prefix long +1 match */
	if (matchIndexL3 > prefixLowestIndex) {
	if (MEM_read64(matchL3) == MEM_read64(ip+1)) {
	mLength = ZSTD_count(ip+9, matchL3+8, iend) + 8;
	ip++;
	offset = (U32)(ip-matchL3);
	while (((ip>anchor) & (matchL3>prefixLowest)) && (ip[-1] == matchL3[-1])) { ip--; matchL3--; mLength++; } /* catch up */
	goto _match_found;
	}
	} else if (dictMode == ZSTD_dictMatchState) {
	/* check dict long +1 match */
	U32 const dictMatchIndexL3 = dictHashLong[dictHLNext];
	const BYTE* dictMatchL3 = dictBase + dictMatchIndexL3;
	assert(dictMatchL3 < dictEnd);
	if (dictMatchL3 > dictStart && MEM_read64(dictMatchL3) == MEM_read64(ip+1)) {
	mLength = ZSTD_count_2segments(ip+1+8, dictMatchL3+8, iend, dictEnd, prefixLowest) + 8;
	ip++;
	offset = (U32)(current + 1 - dictMatchIndexL3 - dictIndexDelta);
	while (((ip>anchor) & (dictMatchL3>dictStart)) && (ip[-1] == dictMatchL3[-1])) { ip--; dictMatchL3--; mLength++; } /* catch up */
	goto _match_found;
	} } }

	/* if no long +1 match, explore the short match we found */
	if (dictMode == ZSTD_dictMatchState && matchIndexS < prefixLowestIndex) {
	mLength = ZSTD_count_2segments(ip+4, match+4, iend, dictEnd, prefixLowest) + 4;
	offset = (U32)(current - matchIndexS);
	while (((ip>anchor) & (match>dictStart)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
	} else {
	mLength = ZSTD_count(ip+4, match+4, iend) + 4;
	offset = (U32)(ip - match);
	while (((ip>anchor) & (match>prefixLowest)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
	}

	/* fall-through */

	_match_found:
	offset_2 = offset_1;
	offset_1 = offset;

	ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);

	_match_stored:
	/* match found */
	ip += mLength;
	anchor = ip;

	if (ip <= ilimit) {
	/* Complementary insertion */
	/* done after iLimit test, as candidates could be > iend-8 */
	{ U32 const indexToInsert = current+2;
	hashLong[ZSTD_hashPtr(base+indexToInsert, hBitsL, 8)] = indexToInsert;
	hashLong[ZSTD_hashPtr(ip-2, hBitsL, 8)] = (U32)(ip-2-base);
	hashSmall[ZSTD_hashPtr(base+indexToInsert, hBitsS, mls)] = indexToInsert;
	hashSmall[ZSTD_hashPtr(ip-1, hBitsS, mls)] = (U32)(ip-1-base);
	}

	/* check immediate repcode */
	if (dictMode == ZSTD_dictMatchState) {
	while (ip <= ilimit) {
	U32 const current2 = (U32)(ip-base);
	U32 const repIndex2 = current2 - offset_2;
	const BYTE* repMatch2 = dictMode == ZSTD_dictMatchState
	&& repIndex2 < prefixLowestIndex ?
	dictBase + repIndex2 - dictIndexDelta :
	base + repIndex2;
	if ( ((U32)((prefixLowestIndex-1) - (U32)repIndex2) >= 3 /* intentional overflow */)
	&& (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
	const BYTE* const repEnd2 = repIndex2 < prefixLowestIndex ? dictEnd : iend;
	size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixLowest) + 4;
	U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
	ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH);
	hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = current2;
	hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = current2;
	ip += repLength2;
	anchor = ip;
	continue;
	}
	break;
	} }

	if (dictMode == ZSTD_noDict) {
	while ( (ip <= ilimit)
	&& ( (offset_2>0)
	& (MEM_read32(ip) == MEM_read32(ip - offset_2)) )) {
	/* store sequence */
	size_t const rLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4;
	U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff; /* swap offset_2 <=> offset_1 */
	hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = (U32)(ip-base);
	hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = (U32)(ip-base);
	ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, rLength-MINMATCH);
	ip += rLength;
	anchor = ip;
	continue; /* faster when present ... (?) */
	} } }
	} /* while (ip < ilimit) */

	/* save reps for next block */
	rep[0] = offset_1 ? offset_1 : offsetSaved;
	rep[1] = offset_2 ? offset_2 : offsetSaved;

	/* Return the last literals size */
	return (size_t)(iend - anchor);
	}


	size_t ZSTD_compressBlock_doubleFast(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	const U32 mls = ms->cParams.minMatch;
	switch(mls)
	{
	default: /* includes case 3 */
	case 4 :
	return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 4, ZSTD_noDict);
	case 5 :
	return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 5, ZSTD_noDict);
	case 6 :
	return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 6, ZSTD_noDict);
	case 7 :
	return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 7, ZSTD_noDict);
	}
	}


	size_t ZSTD_compressBlock_doubleFast_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	const U32 mls = ms->cParams.minMatch;
	switch(mls)
	{
	default: /* includes case 3 */
	case 4 :
	return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 4, ZSTD_dictMatchState);
	case 5 :
	return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 5, ZSTD_dictMatchState);
	case 6 :
	return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 6, ZSTD_dictMatchState);
	case 7 :
	return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, 7, ZSTD_dictMatchState);
	}
	}


	static size_t ZSTD_compressBlock_doubleFast_extDict_generic(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize,
	U32 const mls /* template */)
	{
	ZSTD_compressionParameters const* cParams = &ms->cParams;
	U32* const hashLong = ms->hashTable;
	U32 const hBitsL = cParams->hashLog;
	U32* const hashSmall = ms->chainTable;
	U32 const hBitsS = cParams->chainLog;
	const BYTE* const istart = (const BYTE*)src;
	const BYTE* ip = istart;
	const BYTE* anchor = istart;
	const BYTE* const iend = istart + srcSize;
	const BYTE* const ilimit = iend - 8;
	const BYTE* const base = ms->window.base;
	const U32 endIndex = (U32)((size_t)(istart - base) + srcSize);
	const U32 lowLimit = ZSTD_getLowestMatchIndex(ms, endIndex, cParams->windowLog);
	const U32 dictStartIndex = lowLimit;
	const U32 dictLimit = ms->window.dictLimit;
	const U32 prefixStartIndex = (dictLimit > lowLimit) ? dictLimit : lowLimit;
	const BYTE* const prefixStart = base + prefixStartIndex;
	const BYTE* const dictBase = ms->window.dictBase;
	const BYTE* const dictStart = dictBase + dictStartIndex;
	const BYTE* const dictEnd = dictBase + prefixStartIndex;
	U32 offset_1=rep[0], offset_2=rep[1];

	DEBUGLOG(5, "ZSTD_compressBlock_doubleFast_extDict_generic (srcSize=%zu)", srcSize);

	/* if extDict is invalidated due to maxDistance, switch to "regular" variant */
	if (prefixStartIndex == dictStartIndex)
	return ZSTD_compressBlock_doubleFast_generic(ms, seqStore, rep, src, srcSize, mls, ZSTD_noDict);

	/* Search Loop */
	while (ip < ilimit) { /* < instead of <=, because (ip+1) */
	const size_t hSmall = ZSTD_hashPtr(ip, hBitsS, mls);
	const U32 matchIndex = hashSmall[hSmall];
	const BYTE* const matchBase = matchIndex < prefixStartIndex ? dictBase : base;
	const BYTE* match = matchBase + matchIndex;

	const size_t hLong = ZSTD_hashPtr(ip, hBitsL, 8);
	const U32 matchLongIndex = hashLong[hLong];
	const BYTE* const matchLongBase = matchLongIndex < prefixStartIndex ? dictBase : base;
	const BYTE* matchLong = matchLongBase + matchLongIndex;

	const U32 current = (U32)(ip-base);
	const U32 repIndex = current + 1 - offset_1; /* offset_1 expected <= current +1 */
	const BYTE* const repBase = repIndex < prefixStartIndex ? dictBase : base;
	const BYTE* const repMatch = repBase + repIndex;
	size_t mLength;
	hashSmall[hSmall] = hashLong[hLong] = current; /* update hash table */

	if ((((U32)((prefixStartIndex-1) - repIndex) >= 3) /* intentional underflow : ensure repIndex doesn't overlap dict + prefix */
	& (repIndex > dictStartIndex))
	&& (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
	const BYTE* repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
	mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixStart) + 4;
	ip++;
	ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
	} else {
	if ((matchLongIndex > dictStartIndex) && (MEM_read64(matchLong) == MEM_read64(ip))) {
	const BYTE* const matchEnd = matchLongIndex < prefixStartIndex ? dictEnd : iend;
	const BYTE* const lowMatchPtr = matchLongIndex < prefixStartIndex ? dictStart : prefixStart;
	U32 offset;
	mLength = ZSTD_count_2segments(ip+8, matchLong+8, iend, matchEnd, prefixStart) + 8;
	offset = current - matchLongIndex;
	while (((ip>anchor) & (matchLong>lowMatchPtr)) && (ip[-1] == matchLong[-1])) { ip--; matchLong--; mLength++; } /* catch up */
	offset_2 = offset_1;
	offset_1 = offset;
	ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);

	} else if ((matchIndex > dictStartIndex) && (MEM_read32(match) == MEM_read32(ip))) {
	size_t const h3 = ZSTD_hashPtr(ip+1, hBitsL, 8);
	U32 const matchIndex3 = hashLong[h3];
	const BYTE* const match3Base = matchIndex3 < prefixStartIndex ? dictBase : base;
	const BYTE* match3 = match3Base + matchIndex3;
	U32 offset;
	hashLong[h3] = current + 1;
	if ( (matchIndex3 > dictStartIndex) && (MEM_read64(match3) == MEM_read64(ip+1)) ) {
	const BYTE* const matchEnd = matchIndex3 < prefixStartIndex ? dictEnd : iend;
	const BYTE* const lowMatchPtr = matchIndex3 < prefixStartIndex ? dictStart : prefixStart;
	mLength = ZSTD_count_2segments(ip+9, match3+8, iend, matchEnd, prefixStart) + 8;
	ip++;
	offset = current+1 - matchIndex3;
	while (((ip>anchor) & (match3>lowMatchPtr)) && (ip[-1] == match3[-1])) { ip--; match3--; mLength++; } /* catch up */
	} else {
	const BYTE* const matchEnd = matchIndex < prefixStartIndex ? dictEnd : iend;
	const BYTE* const lowMatchPtr = matchIndex < prefixStartIndex ? dictStart : prefixStart;
	mLength = ZSTD_count_2segments(ip+4, match+4, iend, matchEnd, prefixStart) + 4;
	offset = current - matchIndex;
	while (((ip>anchor) & (match>lowMatchPtr)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
	}
	offset_2 = offset_1;
	offset_1 = offset;
	ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);

	} else {
	ip += ((ip-anchor) >> kSearchStrength) + 1;
	continue;
	} }

	/* move to next sequence start */
	ip += mLength;
	anchor = ip;

	if (ip <= ilimit) {
	/* Complementary insertion */
	/* done after iLimit test, as candidates could be > iend-8 */
	{ U32 const indexToInsert = current+2;
	hashLong[ZSTD_hashPtr(base+indexToInsert, hBitsL, 8)] = indexToInsert;
	hashLong[ZSTD_hashPtr(ip-2, hBitsL, 8)] = (U32)(ip-2-base);
	hashSmall[ZSTD_hashPtr(base+indexToInsert, hBitsS, mls)] = indexToInsert;
	hashSmall[ZSTD_hashPtr(ip-1, hBitsS, mls)] = (U32)(ip-1-base);
	}

	/* check immediate repcode */
	while (ip <= ilimit) {
	U32 const current2 = (U32)(ip-base);
	U32 const repIndex2 = current2 - offset_2;
	const BYTE* repMatch2 = repIndex2 < prefixStartIndex ? dictBase + repIndex2 : base + repIndex2;
	if ( (((U32)((prefixStartIndex-1) - repIndex2) >= 3) /* intentional overflow : ensure repIndex2 doesn't overlap dict + prefix */
	& (repIndex2 > dictStartIndex))
	&& (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
	const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
	size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
	U32 const tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
	ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH);
	hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = current2;
	hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = current2;
	ip += repLength2;
	anchor = ip;
	continue;
	}
	break;
	} } }

	/* save reps for next block */
	rep[0] = offset_1;
	rep[1] = offset_2;

	/* Return the last literals size */
	return (size_t)(iend - anchor);
	}


	size_t ZSTD_compressBlock_doubleFast_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	U32 const mls = ms->cParams.minMatch;
	switch(mls)
	{
	default: /* includes case 3 */
	case 4 :
	return ZSTD_compressBlock_doubleFast_extDict_generic(ms, seqStore, rep, src, srcSize, 4);
	case 5 :
	return ZSTD_compressBlock_doubleFast_extDict_generic(ms, seqStore, rep, src, srcSize, 5);
	case 6 :
	return ZSTD_compressBlock_doubleFast_extDict_generic(ms, seqStore, rep, src, srcSize, 6);
	case 7 :
	return ZSTD_compressBlock_doubleFast_extDict_generic(ms, seqStore, rep, src, srcSize, 7);
	}
	}
	/** ended inlining compress/zstd_double_fast.c **/
	/** start inlining compress/zstd_fast.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/** skipping file: zstd_compress_internal.h **/
	/** skipping file: zstd_fast.h **/


	void ZSTD_fillHashTable(ZSTD_matchState_t* ms,
	const void* const end,
	ZSTD_dictTableLoadMethod_e dtlm)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32* const hashTable = ms->hashTable;
	U32 const hBits = cParams->hashLog;
	U32 const mls = cParams->minMatch;
	const BYTE* const base = ms->window.base;
	const BYTE* ip = base + ms->nextToUpdate;
	const BYTE* const iend = ((const BYTE*)end) - HASH_READ_SIZE;
	const U32 fastHashFillStep = 3;

	/* Always insert every fastHashFillStep position into the hash table.
	* Insert the other positions if their hash entry is empty.
	*/
	for ( ; ip + fastHashFillStep < iend + 2; ip += fastHashFillStep) {
	U32 const current = (U32)(ip - base);
	size_t const hash0 = ZSTD_hashPtr(ip, hBits, mls);
	hashTable[hash0] = current;
	if (dtlm == ZSTD_dtlm_fast) continue;
	/* Only load extra positions for ZSTD_dtlm_full */
	{ U32 p;
	for (p = 1; p < fastHashFillStep; ++p) {
	size_t const hash = ZSTD_hashPtr(ip + p, hBits, mls);
	if (hashTable[hash] == 0) { /* not yet filled */
	hashTable[hash] = current + p;
	} } } }
	}


	FORCE_INLINE_TEMPLATE size_t
	ZSTD_compressBlock_fast_generic(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize,
	U32 const mls)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32* const hashTable = ms->hashTable;
	U32 const hlog = cParams->hashLog;
	/* support stepSize of 0 */
	size_t const stepSize = cParams->targetLength + !(cParams->targetLength) + 1;
	const BYTE* const base = ms->window.base;
	const BYTE* const istart = (const BYTE*)src;
	/* We check ip0 (ip + 0) and ip1 (ip + 1) each loop */
	const BYTE* ip0 = istart;
	const BYTE* ip1;
	const BYTE* anchor = istart;
	const U32 endIndex = (U32)((size_t)(istart - base) + srcSize);
	const U32 prefixStartIndex = ZSTD_getLowestPrefixIndex(ms, endIndex, cParams->windowLog);
	const BYTE* const prefixStart = base + prefixStartIndex;
	const BYTE* const iend = istart + srcSize;
	const BYTE* const ilimit = iend - HASH_READ_SIZE;
	U32 offset_1=rep[0], offset_2=rep[1];
	U32 offsetSaved = 0;

	/* init */
	DEBUGLOG(5, "ZSTD_compressBlock_fast_generic");
	ip0 += (ip0 == prefixStart);
	ip1 = ip0 + 1;
	{ U32 const current = (U32)(ip0 - base);
	U32 const windowLow = ZSTD_getLowestPrefixIndex(ms, current, cParams->windowLog);
	U32 const maxRep = current - windowLow;
	if (offset_2 > maxRep) offsetSaved = offset_2, offset_2 = 0;
	if (offset_1 > maxRep) offsetSaved = offset_1, offset_1 = 0;
	}

	/* Main Search Loop */
	#ifdef __INTEL_COMPILER
	/* From intel 'The vector pragma indicates that the loop should be
	* vectorized if it is legal to do so'. Can be used together with
	* #pragma ivdep (but have opted to exclude that because intel
	* warns against using it).*/
	#pragma vector always
	#endif
	while (ip1 < ilimit) { /* < instead of <=, because check at ip0+2 */
	size_t mLength;
	BYTE const* ip2 = ip0 + 2;
	size_t const h0 = ZSTD_hashPtr(ip0, hlog, mls);
	U32 const val0 = MEM_read32(ip0);
	size_t const h1 = ZSTD_hashPtr(ip1, hlog, mls);
	U32 const val1 = MEM_read32(ip1);
	U32 const current0 = (U32)(ip0-base);
	U32 const current1 = (U32)(ip1-base);
	U32 const matchIndex0 = hashTable[h0];
	U32 const matchIndex1 = hashTable[h1];
	BYTE const* repMatch = ip2 - offset_1;
	const BYTE* match0 = base + matchIndex0;
	const BYTE* match1 = base + matchIndex1;
	U32 offcode;

	#if defined(__aarch64__)
	PREFETCH_L1(ip0+256);
	#endif

	hashTable[h0] = current0; /* update hash table */
	hashTable[h1] = current1; /* update hash table */

	assert(ip0 + 1 == ip1);

	if ((offset_1 > 0) & (MEM_read32(repMatch) == MEM_read32(ip2))) {
	mLength = (ip2[-1] == repMatch[-1]) ? 1 : 0;
	ip0 = ip2 - mLength;
	match0 = repMatch - mLength;
	mLength += 4;
	offcode = 0;
	goto _match;
	}
	if ((matchIndex0 > prefixStartIndex) && MEM_read32(match0) == val0) {
	/* found a regular match */
	goto _offset;
	}
	if ((matchIndex1 > prefixStartIndex) && MEM_read32(match1) == val1) {
	/* found a regular match after one literal */
	ip0 = ip1;
	match0 = match1;
	goto _offset;
	}
	{ size_t const step = ((size_t)(ip0-anchor) >> (kSearchStrength - 1)) + stepSize;
	assert(step >= 2);
	ip0 += step;
	ip1 += step;
	continue;
	}
	_offset: /* Requires: ip0, match0 */
	/* Compute the offset code */
	offset_2 = offset_1;
	offset_1 = (U32)(ip0-match0);
	offcode = offset_1 + ZSTD_REP_MOVE;
	mLength = 4;
	/* Count the backwards match length */
	while (((ip0>anchor) & (match0>prefixStart))
	&& (ip0[-1] == match0[-1])) { ip0--; match0--; mLength++; } /* catch up */

	_match: /* Requires: ip0, match0, offcode */
	/* Count the forward length */
	mLength += ZSTD_count(ip0+mLength, match0+mLength, iend);
	ZSTD_storeSeq(seqStore, (size_t)(ip0-anchor), anchor, iend, offcode, mLength-MINMATCH);
	/* match found */
	ip0 += mLength;
	anchor = ip0;

	if (ip0 <= ilimit) {
	/* Fill Table */
	assert(base+current0+2 > istart); /* check base overflow */
	hashTable[ZSTD_hashPtr(base+current0+2, hlog, mls)] = current0+2; /* here because current+2 could be > iend-8 */
	hashTable[ZSTD_hashPtr(ip0-2, hlog, mls)] = (U32)(ip0-2-base);

	if (offset_2 > 0) { /* offset_2==0 means offset_2 is invalidated */
	while ( (ip0 <= ilimit) && (MEM_read32(ip0) == MEM_read32(ip0 - offset_2)) ) {
	/* store sequence */
	size_t const rLength = ZSTD_count(ip0+4, ip0+4-offset_2, iend) + 4;
	{ U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff; } /* swap offset_2 <=> offset_1 */
	hashTable[ZSTD_hashPtr(ip0, hlog, mls)] = (U32)(ip0-base);
	ip0 += rLength;
	ZSTD_storeSeq(seqStore, 0 /litLen/, anchor, iend, 0 /offCode/, rLength-MINMATCH);
	anchor = ip0;
	continue; /* faster when present (confirmed on gcc-8) ... (?) */
	} } }
	ip1 = ip0 + 1;
	}

	/* save reps for next block */
	rep[0] = offset_1 ? offset_1 : offsetSaved;
	rep[1] = offset_2 ? offset_2 : offsetSaved;

	/* Return the last literals size */
	return (size_t)(iend - anchor);
	}


	size_t ZSTD_compressBlock_fast(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	U32 const mls = ms->cParams.minMatch;
	assert(ms->dictMatchState == NULL);
	switch(mls)
	{
	default: /* includes case 3 */
	case 4 :
	return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 4);
	case 5 :
	return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 5);
	case 6 :
	return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 6);
	case 7 :
	return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, 7);
	}
	}

	FORCE_INLINE_TEMPLATE
	size_t ZSTD_compressBlock_fast_dictMatchState_generic(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize, U32 const mls)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32* const hashTable = ms->hashTable;
	U32 const hlog = cParams->hashLog;
	/* support stepSize of 0 */
	U32 const stepSize = cParams->targetLength + !(cParams->targetLength);
	const BYTE* const base = ms->window.base;
	const BYTE* const istart = (const BYTE*)src;
	const BYTE* ip = istart;
	const BYTE* anchor = istart;
	const U32 prefixStartIndex = ms->window.dictLimit;
	const BYTE* const prefixStart = base + prefixStartIndex;
	const BYTE* const iend = istart + srcSize;
	const BYTE* const ilimit = iend - HASH_READ_SIZE;
	U32 offset_1=rep[0], offset_2=rep[1];
	U32 offsetSaved = 0;

	const ZSTD_matchState_t* const dms = ms->dictMatchState;
	const ZSTD_compressionParameters* const dictCParams = &dms->cParams ;
	const U32* const dictHashTable = dms->hashTable;
	const U32 dictStartIndex = dms->window.dictLimit;
	const BYTE* const dictBase = dms->window.base;
	const BYTE* const dictStart = dictBase + dictStartIndex;
	const BYTE* const dictEnd = dms->window.nextSrc;
	const U32 dictIndexDelta = prefixStartIndex - (U32)(dictEnd - dictBase);
	const U32 dictAndPrefixLength = (U32)(ip - prefixStart + dictEnd - dictStart);
	const U32 dictHLog = dictCParams->hashLog;

	/* if a dictionary is still attached, it necessarily means that
	* it is within window size. So we just check it. */
	const U32 maxDistance = 1U << cParams->windowLog;
	const U32 endIndex = (U32)((size_t)(ip - base) + srcSize);
	assert(endIndex - prefixStartIndex <= maxDistance);
	(void)maxDistance; (void)endIndex; /* these variables are not used when assert() is disabled */

	/* ensure there will be no no underflow
	* when translating a dict index into a local index */
	assert(prefixStartIndex >= (U32)(dictEnd - dictBase));

	/* init */
	DEBUGLOG(5, "ZSTD_compressBlock_fast_dictMatchState_generic");
	ip += (dictAndPrefixLength == 0);
	/* dictMatchState repCode checks don't currently handle repCode == 0
	* disabling. */
	assert(offset_1 <= dictAndPrefixLength);
	assert(offset_2 <= dictAndPrefixLength);

	/* Main Search Loop */
	while (ip < ilimit) { /* < instead of <=, because repcode check at (ip+1) */
	size_t mLength;
	size_t const h = ZSTD_hashPtr(ip, hlog, mls);
	U32 const current = (U32)(ip-base);
	U32 const matchIndex = hashTable[h];
	const BYTE* match = base + matchIndex;
	const U32 repIndex = current + 1 - offset_1;
	const BYTE* repMatch = (repIndex < prefixStartIndex) ?
	dictBase + (repIndex - dictIndexDelta) :
	base + repIndex;
	hashTable[h] = current; /* update hash table */

	if ( ((U32)((prefixStartIndex-1) - repIndex) >= 3) /* intentional underflow : ensure repIndex isn't overlapping dict + prefix */
	&& (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
	const BYTE* const repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
	mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixStart) + 4;
	ip++;
	ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
	} else if ( (matchIndex <= prefixStartIndex) ) {
	size_t const dictHash = ZSTD_hashPtr(ip, dictHLog, mls);
	U32 const dictMatchIndex = dictHashTable[dictHash];
	const BYTE* dictMatch = dictBase + dictMatchIndex;
	if (dictMatchIndex <= dictStartIndex \|\|
	MEM_read32(dictMatch) != MEM_read32(ip)) {
	assert(stepSize >= 1);
	ip += ((ip-anchor) >> kSearchStrength) + stepSize;
	continue;
	} else {
	/* found a dict match */
	U32 const offset = (U32)(current-dictMatchIndex-dictIndexDelta);
	mLength = ZSTD_count_2segments(ip+4, dictMatch+4, iend, dictEnd, prefixStart) + 4;
	while (((ip>anchor) & (dictMatch>dictStart))
	&& (ip[-1] == dictMatch[-1])) {
	ip--; dictMatch--; mLength++;
	} /* catch up */
	offset_2 = offset_1;
	offset_1 = offset;
	ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
	}
	} else if (MEM_read32(match) != MEM_read32(ip)) {
	/* it's not a match, and we're not going to check the dictionary */
	assert(stepSize >= 1);
	ip += ((ip-anchor) >> kSearchStrength) + stepSize;
	continue;
	} else {
	/* found a regular match */
	U32 const offset = (U32)(ip-match);
	mLength = ZSTD_count(ip+4, match+4, iend) + 4;
	while (((ip>anchor) & (match>prefixStart))
	&& (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
	offset_2 = offset_1;
	offset_1 = offset;
	ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
	}

	/* match found */
	ip += mLength;
	anchor = ip;

	if (ip <= ilimit) {
	/* Fill Table */
	assert(base+current+2 > istart); /* check base overflow */
	hashTable[ZSTD_hashPtr(base+current+2, hlog, mls)] = current+2; /* here because current+2 could be > iend-8 */
	hashTable[ZSTD_hashPtr(ip-2, hlog, mls)] = (U32)(ip-2-base);

	/* check immediate repcode */
	while (ip <= ilimit) {
	U32 const current2 = (U32)(ip-base);
	U32 const repIndex2 = current2 - offset_2;
	const BYTE* repMatch2 = repIndex2 < prefixStartIndex ?
	dictBase - dictIndexDelta + repIndex2 :
	base + repIndex2;
	if ( ((U32)((prefixStartIndex-1) - (U32)repIndex2) >= 3 /* intentional overflow */)
	&& (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
	const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
	size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
	U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
	ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH);
	hashTable[ZSTD_hashPtr(ip, hlog, mls)] = current2;
	ip += repLength2;
	anchor = ip;
	continue;
	}
	break;
	}
	}
	}

	/* save reps for next block */
	rep[0] = offset_1 ? offset_1 : offsetSaved;
	rep[1] = offset_2 ? offset_2 : offsetSaved;

	/* Return the last literals size */
	return (size_t)(iend - anchor);
	}

	size_t ZSTD_compressBlock_fast_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	U32 const mls = ms->cParams.minMatch;
	assert(ms->dictMatchState != NULL);
	switch(mls)
	{
	default: /* includes case 3 */
	case 4 :
	return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 4);
	case 5 :
	return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 5);
	case 6 :
	return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 6);
	case 7 :
	return ZSTD_compressBlock_fast_dictMatchState_generic(ms, seqStore, rep, src, srcSize, 7);
	}
	}


	static size_t ZSTD_compressBlock_fast_extDict_generic(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize, U32 const mls)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32* const hashTable = ms->hashTable;
	U32 const hlog = cParams->hashLog;
	/* support stepSize of 0 */
	U32 const stepSize = cParams->targetLength + !(cParams->targetLength);
	const BYTE* const base = ms->window.base;
	const BYTE* const dictBase = ms->window.dictBase;
	const BYTE* const istart = (const BYTE*)src;
	const BYTE* ip = istart;
	const BYTE* anchor = istart;
	const U32 endIndex = (U32)((size_t)(istart - base) + srcSize);
	const U32 lowLimit = ZSTD_getLowestMatchIndex(ms, endIndex, cParams->windowLog);
	const U32 dictStartIndex = lowLimit;
	const BYTE* const dictStart = dictBase + dictStartIndex;
	const U32 dictLimit = ms->window.dictLimit;
	const U32 prefixStartIndex = dictLimit < lowLimit ? lowLimit : dictLimit;
	const BYTE* const prefixStart = base + prefixStartIndex;
	const BYTE* const dictEnd = dictBase + prefixStartIndex;
	const BYTE* const iend = istart + srcSize;
	const BYTE* const ilimit = iend - 8;
	U32 offset_1=rep[0], offset_2=rep[1];

	DEBUGLOG(5, "ZSTD_compressBlock_fast_extDict_generic (offset_1=%u)", offset_1);

	/* switch to "regular" variant if extDict is invalidated due to maxDistance */
	if (prefixStartIndex == dictStartIndex)
	return ZSTD_compressBlock_fast_generic(ms, seqStore, rep, src, srcSize, mls);

	/* Search Loop */
	while (ip < ilimit) { /* < instead of <=, because (ip+1) */
	const size_t h = ZSTD_hashPtr(ip, hlog, mls);
	const U32 matchIndex = hashTable[h];
	const BYTE* const matchBase = matchIndex < prefixStartIndex ? dictBase : base;
	const BYTE* match = matchBase + matchIndex;
	const U32 current = (U32)(ip-base);
	const U32 repIndex = current + 1 - offset_1;
	const BYTE* const repBase = repIndex < prefixStartIndex ? dictBase : base;
	const BYTE* const repMatch = repBase + repIndex;
	hashTable[h] = current; /* update hash table */
	DEBUGLOG(7, "offset_1 = %u , current = %u", offset_1, current);
	assert(offset_1 <= current +1); /* check repIndex */

	if ( (((U32)((prefixStartIndex-1) - repIndex) >= 3) /* intentional underflow */ & (repIndex > dictStartIndex))
	&& (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
	const BYTE* const repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
	size_t const rLength = ZSTD_count_2segments(ip+1 +4, repMatch +4, iend, repMatchEnd, prefixStart) + 4;
	ip++;
	ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, rLength-MINMATCH);
	ip += rLength;
	anchor = ip;
	} else {
	if ( (matchIndex < dictStartIndex) \|\|
	(MEM_read32(match) != MEM_read32(ip)) ) {
	assert(stepSize >= 1);
	ip += ((ip-anchor) >> kSearchStrength) + stepSize;
	continue;
	}
	{ const BYTE* const matchEnd = matchIndex < prefixStartIndex ? dictEnd : iend;
	const BYTE* const lowMatchPtr = matchIndex < prefixStartIndex ? dictStart : prefixStart;
	U32 const offset = current - matchIndex;
	size_t mLength = ZSTD_count_2segments(ip+4, match+4, iend, matchEnd, prefixStart) + 4;
	while (((ip>anchor) & (match>lowMatchPtr)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
	offset_2 = offset_1; offset_1 = offset; /* update offset history */
	ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
	ip += mLength;
	anchor = ip;
	} }

	if (ip <= ilimit) {
	/* Fill Table */
	hashTable[ZSTD_hashPtr(base+current+2, hlog, mls)] = current+2;
	hashTable[ZSTD_hashPtr(ip-2, hlog, mls)] = (U32)(ip-2-base);
	/* check immediate repcode */
	while (ip <= ilimit) {
	U32 const current2 = (U32)(ip-base);
	U32 const repIndex2 = current2 - offset_2;
	const BYTE* const repMatch2 = repIndex2 < prefixStartIndex ? dictBase + repIndex2 : base + repIndex2;
	if ( (((U32)((prefixStartIndex-1) - repIndex2) >= 3) & (repIndex2 > dictStartIndex)) /* intentional overflow */
	&& (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
	const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
	size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
	{ U32 const tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; } /* swap offset_2 <=> offset_1 */
	ZSTD_storeSeq(seqStore, 0 /litlen/, anchor, iend, 0 /offcode/, repLength2-MINMATCH);
	hashTable[ZSTD_hashPtr(ip, hlog, mls)] = current2;
	ip += repLength2;
	anchor = ip;
	continue;
	}
	break;
	} } }

	/* save reps for next block */
	rep[0] = offset_1;
	rep[1] = offset_2;

	/* Return the last literals size */
	return (size_t)(iend - anchor);
	}


	size_t ZSTD_compressBlock_fast_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	U32 const mls = ms->cParams.minMatch;
	switch(mls)
	{
	default: /* includes case 3 */
	case 4 :
	return ZSTD_compressBlock_fast_extDict_generic(ms, seqStore, rep, src, srcSize, 4);
	case 5 :
	return ZSTD_compressBlock_fast_extDict_generic(ms, seqStore, rep, src, srcSize, 5);
	case 6 :
	return ZSTD_compressBlock_fast_extDict_generic(ms, seqStore, rep, src, srcSize, 6);
	case 7 :
	return ZSTD_compressBlock_fast_extDict_generic(ms, seqStore, rep, src, srcSize, 7);
	}
	}
	/** ended inlining compress/zstd_fast.c **/
	/** start inlining compress/zstd_lazy.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/** skipping file: zstd_compress_internal.h **/
	/** skipping file: zstd_lazy.h **/


	/-************************************
	* Binary Tree search
	***************************************/

	static void
	ZSTD_updateDUBT(ZSTD_matchState_t* ms,
	const BYTE* ip, const BYTE* iend,
	U32 mls)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32* const hashTable = ms->hashTable;
	U32 const hashLog = cParams->hashLog;

	U32* const bt = ms->chainTable;
	U32 const btLog = cParams->chainLog - 1;
	U32 const btMask = (1 << btLog) - 1;

	const BYTE* const base = ms->window.base;
	U32 const target = (U32)(ip - base);
	U32 idx = ms->nextToUpdate;

	if (idx != target)
	DEBUGLOG(7, "ZSTD_updateDUBT, from %u to %u (dictLimit:%u)",
	idx, target, ms->window.dictLimit);
	assert(ip + 8 <= iend); /* condition for ZSTD_hashPtr */
	(void)iend;

	assert(idx >= ms->window.dictLimit); /* condition for valid base+idx */
	for ( ; idx < target ; idx++) {
	size_t const h = ZSTD_hashPtr(base + idx, hashLog, mls); /* assumption : ip + 8 <= iend */
	U32 const matchIndex = hashTable[h];

	U32* const nextCandidatePtr = bt + 2*(idx&btMask);
	U32* const sortMarkPtr = nextCandidatePtr + 1;

	DEBUGLOG(8, "ZSTD_updateDUBT: insert %u", idx);
	hashTable[h] = idx; /* Update Hash Table */
	nextCandidatePtr = matchIndex; / update BT like a chain */
	*sortMarkPtr = ZSTD_DUBT_UNSORTED_MARK;
	}
	ms->nextToUpdate = target;
	}


	/** ZSTD_insertDUBT1() :
	* sort one already inserted but unsorted position
	* assumption : current >= btlow == (current - btmask)
	* doesn't fail */
	static void
	ZSTD_insertDUBT1(ZSTD_matchState_t* ms,
	U32 current, const BYTE* inputEnd,
	U32 nbCompares, U32 btLow,
	const ZSTD_dictMode_e dictMode)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32* const bt = ms->chainTable;
	U32 const btLog = cParams->chainLog - 1;
	U32 const btMask = (1 << btLog) - 1;
	size_t commonLengthSmaller=0, commonLengthLarger=0;
	const BYTE* const base = ms->window.base;
	const BYTE* const dictBase = ms->window.dictBase;
	const U32 dictLimit = ms->window.dictLimit;
	const BYTE* const ip = (current>=dictLimit) ? base + current : dictBase + current;
	const BYTE* const iend = (current>=dictLimit) ? inputEnd : dictBase + dictLimit;
	const BYTE* const dictEnd = dictBase + dictLimit;
	const BYTE* const prefixStart = base + dictLimit;
	const BYTE* match;
	U32* smallerPtr = bt + 2*(current&btMask);
	U32* largerPtr = smallerPtr + 1;
	U32 matchIndex = smallerPtr; / this candidate is unsorted : next sorted candidate is reached through smallerPtr, while largerPtr contains previous unsorted candidate (which is already saved and can be overwritten) */
	U32 dummy32; /* to be nullified at the end */
	U32 const windowValid = ms->window.lowLimit;
	U32 const maxDistance = 1U << cParams->windowLog;
	U32 const windowLow = (current - windowValid > maxDistance) ? current - maxDistance : windowValid;


	DEBUGLOG(8, "ZSTD_insertDUBT1(%u) (dictLimit=%u, lowLimit=%u)",
	current, dictLimit, windowLow);
	assert(current >= btLow);
	assert(ip < iend); /* condition for ZSTD_count */

	while (nbCompares-- && (matchIndex > windowLow)) {
	U32* const nextPtr = bt + 2*(matchIndex & btMask);
	size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
	assert(matchIndex < current);
	/* note : all candidates are now supposed sorted,
	* but it's still possible to have nextPtr[1] == ZSTD_DUBT_UNSORTED_MARK
	* when a real index has the same value as ZSTD_DUBT_UNSORTED_MARK */

	if ( (dictMode != ZSTD_extDict)
	\|\| (matchIndex+matchLength >= dictLimit) /* both in current segment*/
	\|\| (current < dictLimit) /* both in extDict */) {
	const BYTE* const mBase = ( (dictMode != ZSTD_extDict)
	\|\| (matchIndex+matchLength >= dictLimit)) ?
	base : dictBase;
	assert( (matchIndex+matchLength >= dictLimit) /* might be wrong if extDict is incorrectly set to 0 */
	\|\| (current < dictLimit) );
	match = mBase + matchIndex;
	matchLength += ZSTD_count(ip+matchLength, match+matchLength, iend);
	} else {
	match = dictBase + matchIndex;
	matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart);
	if (matchIndex+matchLength >= dictLimit)
	match = base + matchIndex; /* preparation for next read of match[matchLength] */
	}

	DEBUGLOG(8, "ZSTD_insertDUBT1: comparing %u with %u : found %u common bytes ",
	current, matchIndex, (U32)matchLength);

	if (ip+matchLength == iend) { /* equal : no way to know if inf or sup */
	break; /* drop , to guarantee consistency ; miss a bit of compression, but other solutions can corrupt tree */
	}

	if (match[matchLength] < ip[matchLength]) { /* necessarily within buffer */
	/* match is smaller than current */
	smallerPtr = matchIndex; / update smaller idx */
	commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
	if (matchIndex <= btLow) { smallerPtr=&dummy32; break; } /* beyond tree size, stop searching */
	DEBUGLOG(8, "ZSTD_insertDUBT1: %u (>btLow=%u) is smaller : next => %u",
	matchIndex, btLow, nextPtr[1]);
	smallerPtr = nextPtr+1; /* new "candidate" => larger than match, which was smaller than target */
	matchIndex = nextPtr[1]; /* new matchIndex, larger than previous and closer to current */
	} else {
	/* match is larger than current */
	*largerPtr = matchIndex;
	commonLengthLarger = matchLength;
	if (matchIndex <= btLow) { largerPtr=&dummy32; break; } /* beyond tree size, stop searching */
	DEBUGLOG(8, "ZSTD_insertDUBT1: %u (>btLow=%u) is larger => %u",
	matchIndex, btLow, nextPtr[0]);
	largerPtr = nextPtr;
	matchIndex = nextPtr[0];
	} }

	smallerPtr = largerPtr = 0;
	}


	static size_t
	ZSTD_DUBT_findBetterDictMatch (
	ZSTD_matchState_t* ms,
	const BYTE* const ip, const BYTE* const iend,
	size_t* offsetPtr,
	size_t bestLength,
	U32 nbCompares,
	U32 const mls,
	const ZSTD_dictMode_e dictMode)
	{
	const ZSTD_matchState_t * const dms = ms->dictMatchState;
	const ZSTD_compressionParameters* const dmsCParams = &dms->cParams;
	const U32 * const dictHashTable = dms->hashTable;
	U32 const hashLog = dmsCParams->hashLog;
	size_t const h = ZSTD_hashPtr(ip, hashLog, mls);
	U32 dictMatchIndex = dictHashTable[h];

	const BYTE* const base = ms->window.base;
	const BYTE* const prefixStart = base + ms->window.dictLimit;
	U32 const current = (U32)(ip-base);
	const BYTE* const dictBase = dms->window.base;
	const BYTE* const dictEnd = dms->window.nextSrc;
	U32 const dictHighLimit = (U32)(dms->window.nextSrc - dms->window.base);
	U32 const dictLowLimit = dms->window.lowLimit;
	U32 const dictIndexDelta = ms->window.lowLimit - dictHighLimit;

	U32* const dictBt = dms->chainTable;
	U32 const btLog = dmsCParams->chainLog - 1;
	U32 const btMask = (1 << btLog) - 1;
	U32 const btLow = (btMask >= dictHighLimit - dictLowLimit) ? dictLowLimit : dictHighLimit - btMask;

	size_t commonLengthSmaller=0, commonLengthLarger=0;

	(void)dictMode;
	assert(dictMode == ZSTD_dictMatchState);

	while (nbCompares-- && (dictMatchIndex > dictLowLimit)) {
	U32* const nextPtr = dictBt + 2*(dictMatchIndex & btMask);
	size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
	const BYTE* match = dictBase + dictMatchIndex;
	matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart);
	if (dictMatchIndex+matchLength >= dictHighLimit)
	match = base + dictMatchIndex + dictIndexDelta; /* to prepare for next usage of match[matchLength] */

	if (matchLength > bestLength) {
	U32 matchIndex = dictMatchIndex + dictIndexDelta;
	if ( (4*(int)(matchLength-bestLength)) > (int)(ZSTD_highbit32(current-matchIndex+1) - ZSTD_highbit32((U32)offsetPtr[0]+1)) ) {
	DEBUGLOG(9, "ZSTD_DUBT_findBetterDictMatch(%u) : found better match length %u -> %u and offsetCode %u -> %u (dictMatchIndex %u, matchIndex %u)",
	current, (U32)bestLength, (U32)matchLength, (U32)*offsetPtr, ZSTD_REP_MOVE + current - matchIndex, dictMatchIndex, matchIndex);
	bestLength = matchLength, *offsetPtr = ZSTD_REP_MOVE + current - matchIndex;
	}
	if (ip+matchLength == iend) { /* reached end of input : ip[matchLength] is not valid, no way to know if it's larger or smaller than match */
	break; /* drop, to guarantee consistency (miss a little bit of compression) */
	}
	}

	if (match[matchLength] < ip[matchLength]) {
	if (dictMatchIndex <= btLow) { break; } /* beyond tree size, stop the search */
	commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
	dictMatchIndex = nextPtr[1]; /* new matchIndex larger than previous (closer to current) */
	} else {
	/* match is larger than current */
	if (dictMatchIndex <= btLow) { break; } /* beyond tree size, stop the search */
	commonLengthLarger = matchLength;
	dictMatchIndex = nextPtr[0];
	}
	}

	if (bestLength >= MINMATCH) {
	U32 const mIndex = current - ((U32)*offsetPtr - ZSTD_REP_MOVE); (void)mIndex;
	DEBUGLOG(8, "ZSTD_DUBT_findBetterDictMatch(%u) : found match of length %u and offsetCode %u (pos %u)",
	current, (U32)bestLength, (U32)*offsetPtr, mIndex);
	}
	return bestLength;

	}


	static size_t
	ZSTD_DUBT_findBestMatch(ZSTD_matchState_t* ms,
	const BYTE* const ip, const BYTE* const iend,
	size_t* offsetPtr,
	U32 const mls,
	const ZSTD_dictMode_e dictMode)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32* const hashTable = ms->hashTable;
	U32 const hashLog = cParams->hashLog;
	size_t const h = ZSTD_hashPtr(ip, hashLog, mls);
	U32 matchIndex = hashTable[h];

	const BYTE* const base = ms->window.base;
	U32 const current = (U32)(ip-base);
	U32 const windowLow = ZSTD_getLowestMatchIndex(ms, current, cParams->windowLog);

	U32* const bt = ms->chainTable;
	U32 const btLog = cParams->chainLog - 1;
	U32 const btMask = (1 << btLog) - 1;
	U32 const btLow = (btMask >= current) ? 0 : current - btMask;
	U32 const unsortLimit = MAX(btLow, windowLow);

	U32* nextCandidate = bt + 2*(matchIndex&btMask);
	U32* unsortedMark = bt + 2*(matchIndex&btMask) + 1;
	U32 nbCompares = 1U << cParams->searchLog;
	U32 nbCandidates = nbCompares;
	U32 previousCandidate = 0;

	DEBUGLOG(7, "ZSTD_DUBT_findBestMatch (%u) ", current);
	assert(ip <= iend-8); /* required for h calculation */

	/* reach end of unsorted candidates list */
	while ( (matchIndex > unsortLimit)
	&& (*unsortedMark == ZSTD_DUBT_UNSORTED_MARK)
	&& (nbCandidates > 1) ) {
	DEBUGLOG(8, "ZSTD_DUBT_findBestMatch: candidate %u is unsorted",
	matchIndex);
	unsortedMark = previousCandidate; / the unsortedMark becomes a reversed chain, to move up back to original position */
	previousCandidate = matchIndex;
	matchIndex = *nextCandidate;
	nextCandidate = bt + 2*(matchIndex&btMask);
	unsortedMark = bt + 2*(matchIndex&btMask) + 1;
	nbCandidates --;
	}

	/* nullify last candidate if it's still unsorted
	* simplification, detrimental to compression ratio, beneficial for speed */
	if ( (matchIndex > unsortLimit)
	&& (*unsortedMark==ZSTD_DUBT_UNSORTED_MARK) ) {
	DEBUGLOG(7, "ZSTD_DUBT_findBestMatch: nullify last unsorted candidate %u",
	matchIndex);
	nextCandidate = unsortedMark = 0;
	}

	/* batch sort stacked candidates */
	matchIndex = previousCandidate;
	while (matchIndex) { /* will end on matchIndex == 0 */
	U32* const nextCandidateIdxPtr = bt + 2*(matchIndex&btMask) + 1;
	U32 const nextCandidateIdx = *nextCandidateIdxPtr;
	ZSTD_insertDUBT1(ms, matchIndex, iend,
	nbCandidates, unsortLimit, dictMode);
	matchIndex = nextCandidateIdx;
	nbCandidates++;
	}

	/* find longest match */
	{ size_t commonLengthSmaller = 0, commonLengthLarger = 0;
	const BYTE* const dictBase = ms->window.dictBase;
	const U32 dictLimit = ms->window.dictLimit;
	const BYTE* const dictEnd = dictBase + dictLimit;
	const BYTE* const prefixStart = base + dictLimit;
	U32* smallerPtr = bt + 2*(current&btMask);
	U32* largerPtr = bt + 2*(current&btMask) + 1;
	U32 matchEndIdx = current + 8 + 1;
	U32 dummy32; /* to be nullified at the end */
	size_t bestLength = 0;

	matchIndex = hashTable[h];
	hashTable[h] = current; /* Update Hash Table */

	while (nbCompares-- && (matchIndex > windowLow)) {
	U32* const nextPtr = bt + 2*(matchIndex & btMask);
	size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
	const BYTE* match;

	if ((dictMode != ZSTD_extDict) \|\| (matchIndex+matchLength >= dictLimit)) {
	match = base + matchIndex;
	matchLength += ZSTD_count(ip+matchLength, match+matchLength, iend);
	} else {
	match = dictBase + matchIndex;
	matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart);
	if (matchIndex+matchLength >= dictLimit)
	match = base + matchIndex; /* to prepare for next usage of match[matchLength] */
	}

	if (matchLength > bestLength) {
	if (matchLength > matchEndIdx - matchIndex)
	matchEndIdx = matchIndex + (U32)matchLength;
	if ( (4*(int)(matchLength-bestLength)) > (int)(ZSTD_highbit32(current-matchIndex+1) - ZSTD_highbit32((U32)offsetPtr[0]+1)) )
	bestLength = matchLength, *offsetPtr = ZSTD_REP_MOVE + current - matchIndex;
	if (ip+matchLength == iend) { /* equal : no way to know if inf or sup */
	if (dictMode == ZSTD_dictMatchState) {
	nbCompares = 0; /* in addition to avoiding checking any
	* further in this loop, make sure we
	* skip checking in the dictionary. */
	}
	break; /* drop, to guarantee consistency (miss a little bit of compression) */
	}
	}

	if (match[matchLength] < ip[matchLength]) {
	/* match is smaller than current */
	smallerPtr = matchIndex; / update smaller idx */
	commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
	if (matchIndex <= btLow) { smallerPtr=&dummy32; break; } /* beyond tree size, stop the search */
	smallerPtr = nextPtr+1; /* new "smaller" => larger of match */
	matchIndex = nextPtr[1]; /* new matchIndex larger than previous (closer to current) */
	} else {
	/* match is larger than current */
	*largerPtr = matchIndex;
	commonLengthLarger = matchLength;
	if (matchIndex <= btLow) { largerPtr=&dummy32; break; } /* beyond tree size, stop the search */
	largerPtr = nextPtr;
	matchIndex = nextPtr[0];
	} }

	smallerPtr = largerPtr = 0;

	if (dictMode == ZSTD_dictMatchState && nbCompares) {
	bestLength = ZSTD_DUBT_findBetterDictMatch(
	ms, ip, iend,
	offsetPtr, bestLength, nbCompares,
	mls, dictMode);
	}

	assert(matchEndIdx > current+8); /* ensure nextToUpdate is increased */
	ms->nextToUpdate = matchEndIdx - 8; /* skip repetitive patterns */
	if (bestLength >= MINMATCH) {
	U32 const mIndex = current - ((U32)*offsetPtr - ZSTD_REP_MOVE); (void)mIndex;
	DEBUGLOG(8, "ZSTD_DUBT_findBestMatch(%u) : found match of length %u and offsetCode %u (pos %u)",
	current, (U32)bestLength, (U32)*offsetPtr, mIndex);
	}
	return bestLength;
	}
	}


	/** ZSTD_BtFindBestMatch() : Tree updater, providing best match */
	FORCE_INLINE_TEMPLATE size_t
	ZSTD_BtFindBestMatch( ZSTD_matchState_t* ms,
	const BYTE* const ip, const BYTE* const iLimit,
	size_t* offsetPtr,
	const U32 mls /* template */,
	const ZSTD_dictMode_e dictMode)
	{
	DEBUGLOG(7, "ZSTD_BtFindBestMatch");
	if (ip < ms->window.base + ms->nextToUpdate) return 0; /* skipped area */
	ZSTD_updateDUBT(ms, ip, iLimit, mls);
	return ZSTD_DUBT_findBestMatch(ms, ip, iLimit, offsetPtr, mls, dictMode);
	}


	static size_t
	ZSTD_BtFindBestMatch_selectMLS ( ZSTD_matchState_t* ms,
	const BYTE* ip, const BYTE* const iLimit,
	size_t* offsetPtr)
	{
	switch(ms->cParams.minMatch)
	{
	default : /* includes case 3 */
	case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_noDict);
	case 5 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 5, ZSTD_noDict);
	case 7 :
	case 6 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 6, ZSTD_noDict);
	}
	}


	static size_t ZSTD_BtFindBestMatch_dictMatchState_selectMLS (
	ZSTD_matchState_t* ms,
	const BYTE* ip, const BYTE* const iLimit,
	size_t* offsetPtr)
	{
	switch(ms->cParams.minMatch)
	{
	default : /* includes case 3 */
	case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_dictMatchState);
	case 5 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 5, ZSTD_dictMatchState);
	case 7 :
	case 6 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 6, ZSTD_dictMatchState);
	}
	}


	static size_t ZSTD_BtFindBestMatch_extDict_selectMLS (
	ZSTD_matchState_t* ms,
	const BYTE* ip, const BYTE* const iLimit,
	size_t* offsetPtr)
	{
	switch(ms->cParams.minMatch)
	{
	default : /* includes case 3 */
	case 4 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 4, ZSTD_extDict);
	case 5 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 5, ZSTD_extDict);
	case 7 :
	case 6 : return ZSTD_BtFindBestMatch(ms, ip, iLimit, offsetPtr, 6, ZSTD_extDict);
	}
	}



	/* *********************************
	* Hash Chain
	***********************************/
	#define NEXT_IN_CHAIN(d, mask) chainTable[(d) & (mask)]

	/* Update chains up to ip (excluded)
	Assumption : always within prefix (i.e. not within extDict) */
	static U32 ZSTD_insertAndFindFirstIndex_internal(
	ZSTD_matchState_t* ms,
	const ZSTD_compressionParameters* const cParams,
	const BYTE* ip, U32 const mls)
	{
	U32* const hashTable = ms->hashTable;
	const U32 hashLog = cParams->hashLog;
	U32* const chainTable = ms->chainTable;
	const U32 chainMask = (1 << cParams->chainLog) - 1;
	const BYTE* const base = ms->window.base;
	const U32 target = (U32)(ip - base);
	U32 idx = ms->nextToUpdate;

	while(idx < target) { /* catch up */
	size_t const h = ZSTD_hashPtr(base+idx, hashLog, mls);
	NEXT_IN_CHAIN(idx, chainMask) = hashTable[h];
	hashTable[h] = idx;
	idx++;
	}

	ms->nextToUpdate = target;
	return hashTable[ZSTD_hashPtr(ip, hashLog, mls)];
	}

	U32 ZSTD_insertAndFindFirstIndex(ZSTD_matchState_t* ms, const BYTE* ip) {
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	return ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, ms->cParams.minMatch);
	}


	/* inlining is important to hardwire a hot branch (template emulation) */
	FORCE_INLINE_TEMPLATE
	size_t ZSTD_HcFindBestMatch_generic (
	ZSTD_matchState_t* ms,
	const BYTE* const ip, const BYTE* const iLimit,
	size_t* offsetPtr,
	const U32 mls, const ZSTD_dictMode_e dictMode)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32* const chainTable = ms->chainTable;
	const U32 chainSize = (1 << cParams->chainLog);
	const U32 chainMask = chainSize-1;
	const BYTE* const base = ms->window.base;
	const BYTE* const dictBase = ms->window.dictBase;
	const U32 dictLimit = ms->window.dictLimit;
	const BYTE* const prefixStart = base + dictLimit;
	const BYTE* const dictEnd = dictBase + dictLimit;
	const U32 current = (U32)(ip-base);
	const U32 maxDistance = 1U << cParams->windowLog;
	const U32 lowestValid = ms->window.lowLimit;
	const U32 withinMaxDistance = (current - lowestValid > maxDistance) ? current - maxDistance : lowestValid;
	const U32 isDictionary = (ms->loadedDictEnd != 0);
	const U32 lowLimit = isDictionary ? lowestValid : withinMaxDistance;
	const U32 minChain = current > chainSize ? current - chainSize : 0;
	U32 nbAttempts = 1U << cParams->searchLog;
	size_t ml=4-1;

	/* HC4 match finder */
	U32 matchIndex = ZSTD_insertAndFindFirstIndex_internal(ms, cParams, ip, mls);

	for ( ; (matchIndex>lowLimit) & (nbAttempts>0) ; nbAttempts--) {
	size_t currentMl=0;
	if ((dictMode != ZSTD_extDict) \|\| matchIndex >= dictLimit) {
	const BYTE* const match = base + matchIndex;
	assert(matchIndex >= dictLimit); /* ensures this is true if dictMode != ZSTD_extDict */
	if (match[ml] == ip[ml]) /* potentially better */
	currentMl = ZSTD_count(ip, match, iLimit);
	} else {
	const BYTE* const match = dictBase + matchIndex;
	assert(match+4 <= dictEnd);
	if (MEM_read32(match) == MEM_read32(ip)) /* assumption : matchIndex <= dictLimit-4 (by table construction) */
	currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, dictEnd, prefixStart) + 4;
	}

	/* save best solution */
	if (currentMl > ml) {
	ml = currentMl;
	*offsetPtr = current - matchIndex + ZSTD_REP_MOVE;
	if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */
	}

	if (matchIndex <= minChain) break;
	matchIndex = NEXT_IN_CHAIN(matchIndex, chainMask);
	}

	if (dictMode == ZSTD_dictMatchState) {
	const ZSTD_matchState_t* const dms = ms->dictMatchState;
	const U32* const dmsChainTable = dms->chainTable;
	const U32 dmsChainSize = (1 << dms->cParams.chainLog);
	const U32 dmsChainMask = dmsChainSize - 1;
	const U32 dmsLowestIndex = dms->window.dictLimit;
	const BYTE* const dmsBase = dms->window.base;
	const BYTE* const dmsEnd = dms->window.nextSrc;
	const U32 dmsSize = (U32)(dmsEnd - dmsBase);
	const U32 dmsIndexDelta = dictLimit - dmsSize;
	const U32 dmsMinChain = dmsSize > dmsChainSize ? dmsSize - dmsChainSize : 0;

	matchIndex = dms->hashTable[ZSTD_hashPtr(ip, dms->cParams.hashLog, mls)];

	for ( ; (matchIndex>dmsLowestIndex) & (nbAttempts>0) ; nbAttempts--) {
	size_t currentMl=0;
	const BYTE* const match = dmsBase + matchIndex;
	assert(match+4 <= dmsEnd);
	if (MEM_read32(match) == MEM_read32(ip)) /* assumption : matchIndex <= dictLimit-4 (by table construction) */
	currentMl = ZSTD_count_2segments(ip+4, match+4, iLimit, dmsEnd, prefixStart) + 4;

	/* save best solution */
	if (currentMl > ml) {
	ml = currentMl;
	*offsetPtr = current - (matchIndex + dmsIndexDelta) + ZSTD_REP_MOVE;
	if (ip+currentMl == iLimit) break; /* best possible, avoids read overflow on next attempt */
	}

	if (matchIndex <= dmsMinChain) break;
	matchIndex = dmsChainTable[matchIndex & dmsChainMask];
	}
	}

	return ml;
	}


	FORCE_INLINE_TEMPLATE size_t ZSTD_HcFindBestMatch_selectMLS (
	ZSTD_matchState_t* ms,
	const BYTE* ip, const BYTE* const iLimit,
	size_t* offsetPtr)
	{
	switch(ms->cParams.minMatch)
	{
	default : /* includes case 3 */
	case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_noDict);
	case 5 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 5, ZSTD_noDict);
	case 7 :
	case 6 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 6, ZSTD_noDict);
	}
	}


	static size_t ZSTD_HcFindBestMatch_dictMatchState_selectMLS (
	ZSTD_matchState_t* ms,
	const BYTE* ip, const BYTE* const iLimit,
	size_t* offsetPtr)
	{
	switch(ms->cParams.minMatch)
	{
	default : /* includes case 3 */
	case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_dictMatchState);
	case 5 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 5, ZSTD_dictMatchState);
	case 7 :
	case 6 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 6, ZSTD_dictMatchState);
	}
	}


	FORCE_INLINE_TEMPLATE size_t ZSTD_HcFindBestMatch_extDict_selectMLS (
	ZSTD_matchState_t* ms,
	const BYTE* ip, const BYTE* const iLimit,
	size_t* offsetPtr)
	{
	switch(ms->cParams.minMatch)
	{
	default : /* includes case 3 */
	case 4 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 4, ZSTD_extDict);
	case 5 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 5, ZSTD_extDict);
	case 7 :
	case 6 : return ZSTD_HcFindBestMatch_generic(ms, ip, iLimit, offsetPtr, 6, ZSTD_extDict);
	}
	}


	/* *******************************
	* Common parser - lazy strategy
	*********************************/
	typedef enum { search_hashChain, search_binaryTree } searchMethod_e;

	FORCE_INLINE_TEMPLATE size_t
	ZSTD_compressBlock_lazy_generic(
	ZSTD_matchState_t* ms, seqStore_t* seqStore,
	U32 rep[ZSTD_REP_NUM],
	const void* src, size_t srcSize,
	const searchMethod_e searchMethod, const U32 depth,
	ZSTD_dictMode_e const dictMode)
	{
	const BYTE* const istart = (const BYTE*)src;
	const BYTE* ip = istart;
	const BYTE* anchor = istart;
	const BYTE* const iend = istart + srcSize;
	const BYTE* const ilimit = iend - 8;
	const BYTE* const base = ms->window.base;
	const U32 prefixLowestIndex = ms->window.dictLimit;
	const BYTE* const prefixLowest = base + prefixLowestIndex;

	typedef size_t (*searchMax_f)(
	ZSTD_matchState_t* ms,
	const BYTE* ip, const BYTE* iLimit, size_t* offsetPtr);
	searchMax_f const searchMax = dictMode == ZSTD_dictMatchState ?
	(searchMethod==search_binaryTree ? ZSTD_BtFindBestMatch_dictMatchState_selectMLS
	: ZSTD_HcFindBestMatch_dictMatchState_selectMLS) :
	(searchMethod==search_binaryTree ? ZSTD_BtFindBestMatch_selectMLS
	: ZSTD_HcFindBestMatch_selectMLS);
	U32 offset_1 = rep[0], offset_2 = rep[1], savedOffset=0;

	const ZSTD_matchState_t* const dms = ms->dictMatchState;
	const U32 dictLowestIndex = dictMode == ZSTD_dictMatchState ?
	dms->window.dictLimit : 0;
	const BYTE* const dictBase = dictMode == ZSTD_dictMatchState ?
	dms->window.base : NULL;
	const BYTE* const dictLowest = dictMode == ZSTD_dictMatchState ?
	dictBase + dictLowestIndex : NULL;
	const BYTE* const dictEnd = dictMode == ZSTD_dictMatchState ?
	dms->window.nextSrc : NULL;
	const U32 dictIndexDelta = dictMode == ZSTD_dictMatchState ?
	prefixLowestIndex - (U32)(dictEnd - dictBase) :
	0;
	const U32 dictAndPrefixLength = (U32)((ip - prefixLowest) + (dictEnd - dictLowest));

	DEBUGLOG(5, "ZSTD_compressBlock_lazy_generic (dictMode=%u)", (U32)dictMode);

	/* init */
	ip += (dictAndPrefixLength == 0);
	if (dictMode == ZSTD_noDict) {
	U32 const current = (U32)(ip - base);
	U32 const windowLow = ZSTD_getLowestPrefixIndex(ms, current, ms->cParams.windowLog);
	U32 const maxRep = current - windowLow;
	if (offset_2 > maxRep) savedOffset = offset_2, offset_2 = 0;
	if (offset_1 > maxRep) savedOffset = offset_1, offset_1 = 0;
	}
	if (dictMode == ZSTD_dictMatchState) {
	/* dictMatchState repCode checks don't currently handle repCode == 0
	* disabling. */
	assert(offset_1 <= dictAndPrefixLength);
	assert(offset_2 <= dictAndPrefixLength);
	}

	/* Match Loop */
	#if defined(__GNUC__) && defined(__x86_64__)
	/* I've measured random a 5% speed loss on levels 5 & 6 (greedy) when the
	* code alignment is perturbed. To fix the instability align the loop on 32-bytes.
	*/
	__asm__(".p2align 5");
	#endif
	while (ip < ilimit) {
	size_t matchLength=0;
	size_t offset=0;
	const BYTE* start=ip+1;

	/* check repCode */
	if (dictMode == ZSTD_dictMatchState) {
	const U32 repIndex = (U32)(ip - base) + 1 - offset_1;
	const BYTE* repMatch = (dictMode == ZSTD_dictMatchState
	&& repIndex < prefixLowestIndex) ?
	dictBase + (repIndex - dictIndexDelta) :
	base + repIndex;
	if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
	&& (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
	const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
	matchLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
	if (depth==0) goto _storeSequence;
	}
	}
	if ( dictMode == ZSTD_noDict
	&& ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1)))) {
	matchLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4;
	if (depth==0) goto _storeSequence;
	}

	/* first search (depth 0) */
	{ size_t offsetFound = 999999999;
	size_t const ml2 = searchMax(ms, ip, iend, &offsetFound);
	if (ml2 > matchLength)
	matchLength = ml2, start = ip, offset=offsetFound;
	}

	if (matchLength < 4) {
	ip += ((ip-anchor) >> kSearchStrength) + 1; /* jump faster over incompressible sections */
	continue;
	}

	/* let's try to find a better solution */
	if (depth>=1)
	while (ip<ilimit) {
	ip ++;
	if ( (dictMode == ZSTD_noDict)
	&& (offset) && ((offset_1>0) & (MEM_read32(ip) == MEM_read32(ip - offset_1)))) {
	size_t const mlRep = ZSTD_count(ip+4, ip+4-offset_1, iend) + 4;
	int const gain2 = (int)(mlRep * 3);
	int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)offset+1) + 1);
	if ((mlRep >= 4) && (gain2 > gain1))
	matchLength = mlRep, offset = 0, start = ip;
	}
	if (dictMode == ZSTD_dictMatchState) {
	const U32 repIndex = (U32)(ip - base) - offset_1;
	const BYTE* repMatch = repIndex < prefixLowestIndex ?
	dictBase + (repIndex - dictIndexDelta) :
	base + repIndex;
	if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
	&& (MEM_read32(repMatch) == MEM_read32(ip)) ) {
	const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
	size_t const mlRep = ZSTD_count_2segments(ip+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
	int const gain2 = (int)(mlRep * 3);
	int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)offset+1) + 1);
	if ((mlRep >= 4) && (gain2 > gain1))
	matchLength = mlRep, offset = 0, start = ip;
	}
	}
	{ size_t offset2=999999999;
	size_t const ml2 = searchMax(ms, ip, iend, &offset2);
	int const gain2 = (int)(ml24 - ZSTD_highbit32((U32)offset2+1)); / raw approx */
	int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 4);
	if ((ml2 >= 4) && (gain2 > gain1)) {
	matchLength = ml2, offset = offset2, start = ip;
	continue; /* search a better one */
	} }

	/* let's find an even better one */
	if ((depth==2) && (ip<ilimit)) {
	ip ++;
	if ( (dictMode == ZSTD_noDict)
	&& (offset) && ((offset_1>0) & (MEM_read32(ip) == MEM_read32(ip - offset_1)))) {
	size_t const mlRep = ZSTD_count(ip+4, ip+4-offset_1, iend) + 4;
	int const gain2 = (int)(mlRep * 4);
	int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 1);
	if ((mlRep >= 4) && (gain2 > gain1))
	matchLength = mlRep, offset = 0, start = ip;
	}
	if (dictMode == ZSTD_dictMatchState) {
	const U32 repIndex = (U32)(ip - base) - offset_1;
	const BYTE* repMatch = repIndex < prefixLowestIndex ?
	dictBase + (repIndex - dictIndexDelta) :
	base + repIndex;
	if (((U32)((prefixLowestIndex-1) - repIndex) >= 3 /* intentional underflow */)
	&& (MEM_read32(repMatch) == MEM_read32(ip)) ) {
	const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
	size_t const mlRep = ZSTD_count_2segments(ip+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
	int const gain2 = (int)(mlRep * 4);
	int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 1);
	if ((mlRep >= 4) && (gain2 > gain1))
	matchLength = mlRep, offset = 0, start = ip;
	}
	}
	{ size_t offset2=999999999;
	size_t const ml2 = searchMax(ms, ip, iend, &offset2);
	int const gain2 = (int)(ml24 - ZSTD_highbit32((U32)offset2+1)); / raw approx */
	int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 7);
	if ((ml2 >= 4) && (gain2 > gain1)) {
	matchLength = ml2, offset = offset2, start = ip;
	continue;
	} } }
	break; /* nothing found : store previous solution */
	}

	/* NOTE:
	* start[-offset+ZSTD_REP_MOVE-1] is undefined behavior.
	* (-offset+ZSTD_REP_MOVE-1) is unsigned, and is added to start, which
	* overflows the pointer, which is undefined behavior.
	*/
	/* catch up */
	if (offset) {
	if (dictMode == ZSTD_noDict) {
	while ( ((start > anchor) & (start - (offset-ZSTD_REP_MOVE) > prefixLowest))
	&& (start[-1] == (start-(offset-ZSTD_REP_MOVE))[-1]) ) /* only search for offset within prefix */
	{ start--; matchLength++; }
	}
	if (dictMode == ZSTD_dictMatchState) {
	U32 const matchIndex = (U32)((start-base) - (offset - ZSTD_REP_MOVE));
	const BYTE* match = (matchIndex < prefixLowestIndex) ? dictBase + matchIndex - dictIndexDelta : base + matchIndex;
	const BYTE* const mStart = (matchIndex < prefixLowestIndex) ? dictLowest : prefixLowest;
	while ((start>anchor) && (match>mStart) && (start[-1] == match[-1])) { start--; match--; matchLength++; } /* catch up */
	}
	offset_2 = offset_1; offset_1 = (U32)(offset - ZSTD_REP_MOVE);
	}
	/* store sequence */
	_storeSequence:
	{ size_t const litLength = start - anchor;
	ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offset, matchLength-MINMATCH);
	anchor = ip = start + matchLength;
	}

	/* check immediate repcode */
	if (dictMode == ZSTD_dictMatchState) {
	while (ip <= ilimit) {
	U32 const current2 = (U32)(ip-base);
	U32 const repIndex = current2 - offset_2;
	const BYTE* repMatch = dictMode == ZSTD_dictMatchState
	&& repIndex < prefixLowestIndex ?
	dictBase - dictIndexDelta + repIndex :
	base + repIndex;
	if ( ((U32)((prefixLowestIndex-1) - (U32)repIndex) >= 3 /* intentional overflow */)
	&& (MEM_read32(repMatch) == MEM_read32(ip)) ) {
	const BYTE* const repEnd2 = repIndex < prefixLowestIndex ? dictEnd : iend;
	matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd2, prefixLowest) + 4;
	offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap offset_2 <=> offset_1 */
	ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH);
	ip += matchLength;
	anchor = ip;
	continue;
	}
	break;
	}
	}

	if (dictMode == ZSTD_noDict) {
	while ( ((ip <= ilimit) & (offset_2>0))
	&& (MEM_read32(ip) == MEM_read32(ip - offset_2)) ) {
	/* store sequence */
	matchLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4;
	offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap repcodes */
	ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH);
	ip += matchLength;
	anchor = ip;
	continue; /* faster when present ... (?) */
	} } }

	/* Save reps for next block */
	rep[0] = offset_1 ? offset_1 : savedOffset;
	rep[1] = offset_2 ? offset_2 : savedOffset;

	/* Return the last literals size */
	return (size_t)(iend - anchor);
	}


	size_t ZSTD_compressBlock_btlazy2(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_noDict);
	}

	size_t ZSTD_compressBlock_lazy2(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_noDict);
	}

	size_t ZSTD_compressBlock_lazy(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_noDict);
	}

	size_t ZSTD_compressBlock_greedy(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_noDict);
	}

	size_t ZSTD_compressBlock_btlazy2_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2, ZSTD_dictMatchState);
	}

	size_t ZSTD_compressBlock_lazy2_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2, ZSTD_dictMatchState);
	}

	size_t ZSTD_compressBlock_lazy_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1, ZSTD_dictMatchState);
	}

	size_t ZSTD_compressBlock_greedy_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	return ZSTD_compressBlock_lazy_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0, ZSTD_dictMatchState);
	}


	FORCE_INLINE_TEMPLATE
	size_t ZSTD_compressBlock_lazy_extDict_generic(
	ZSTD_matchState_t* ms, seqStore_t* seqStore,
	U32 rep[ZSTD_REP_NUM],
	const void* src, size_t srcSize,
	const searchMethod_e searchMethod, const U32 depth)
	{
	const BYTE* const istart = (const BYTE*)src;
	const BYTE* ip = istart;
	const BYTE* anchor = istart;
	const BYTE* const iend = istart + srcSize;
	const BYTE* const ilimit = iend - 8;
	const BYTE* const base = ms->window.base;
	const U32 dictLimit = ms->window.dictLimit;
	const BYTE* const prefixStart = base + dictLimit;
	const BYTE* const dictBase = ms->window.dictBase;
	const BYTE* const dictEnd = dictBase + dictLimit;
	const BYTE* const dictStart = dictBase + ms->window.lowLimit;
	const U32 windowLog = ms->cParams.windowLog;

	typedef size_t (*searchMax_f)(
	ZSTD_matchState_t* ms,
	const BYTE* ip, const BYTE* iLimit, size_t* offsetPtr);
	searchMax_f searchMax = searchMethod==search_binaryTree ? ZSTD_BtFindBestMatch_extDict_selectMLS : ZSTD_HcFindBestMatch_extDict_selectMLS;

	U32 offset_1 = rep[0], offset_2 = rep[1];

	DEBUGLOG(5, "ZSTD_compressBlock_lazy_extDict_generic");

	/* init */
	ip += (ip == prefixStart);

	/* Match Loop */
	#if defined(__GNUC__) && defined(__x86_64__)
	/* I've measured random a 5% speed loss on levels 5 & 6 (greedy) when the
	* code alignment is perturbed. To fix the instability align the loop on 32-bytes.
	*/
	__asm__(".p2align 5");
	#endif
	while (ip < ilimit) {
	size_t matchLength=0;
	size_t offset=0;
	const BYTE* start=ip+1;
	U32 current = (U32)(ip-base);

	/* check repCode */
	{ const U32 windowLow = ZSTD_getLowestMatchIndex(ms, current+1, windowLog);
	const U32 repIndex = (U32)(current+1 - offset_1);
	const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
	const BYTE* const repMatch = repBase + repIndex;
	if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow)) /* intentional overflow */
	if (MEM_read32(ip+1) == MEM_read32(repMatch)) {
	/* repcode detected we should take it */
	const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
	matchLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repEnd, prefixStart) + 4;
	if (depth==0) goto _storeSequence;
	} }

	/* first search (depth 0) */
	{ size_t offsetFound = 999999999;
	size_t const ml2 = searchMax(ms, ip, iend, &offsetFound);
	if (ml2 > matchLength)
	matchLength = ml2, start = ip, offset=offsetFound;
	}

	if (matchLength < 4) {
	ip += ((ip-anchor) >> kSearchStrength) + 1; /* jump faster over incompressible sections */
	continue;
	}

	/* let's try to find a better solution */
	if (depth>=1)
	while (ip<ilimit) {
	ip ++;
	current++;
	/* check repCode */
	if (offset) {
	const U32 windowLow = ZSTD_getLowestMatchIndex(ms, current, windowLog);
	const U32 repIndex = (U32)(current - offset_1);
	const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
	const BYTE* const repMatch = repBase + repIndex;
	if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow)) /* intentional overflow */
	if (MEM_read32(ip) == MEM_read32(repMatch)) {
	/* repcode detected */
	const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
	size_t const repLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
	int const gain2 = (int)(repLength * 3);
	int const gain1 = (int)(matchLength*3 - ZSTD_highbit32((U32)offset+1) + 1);
	if ((repLength >= 4) && (gain2 > gain1))
	matchLength = repLength, offset = 0, start = ip;
	} }

	/* search match, depth 1 */
	{ size_t offset2=999999999;
	size_t const ml2 = searchMax(ms, ip, iend, &offset2);
	int const gain2 = (int)(ml24 - ZSTD_highbit32((U32)offset2+1)); / raw approx */
	int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 4);
	if ((ml2 >= 4) && (gain2 > gain1)) {
	matchLength = ml2, offset = offset2, start = ip;
	continue; /* search a better one */
	} }

	/* let's find an even better one */
	if ((depth==2) && (ip<ilimit)) {
	ip ++;
	current++;
	/* check repCode */
	if (offset) {
	const U32 windowLow = ZSTD_getLowestMatchIndex(ms, current, windowLog);
	const U32 repIndex = (U32)(current - offset_1);
	const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
	const BYTE* const repMatch = repBase + repIndex;
	if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow)) /* intentional overflow */
	if (MEM_read32(ip) == MEM_read32(repMatch)) {
	/* repcode detected */
	const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
	size_t const repLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
	int const gain2 = (int)(repLength * 4);
	int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 1);
	if ((repLength >= 4) && (gain2 > gain1))
	matchLength = repLength, offset = 0, start = ip;
	} }

	/* search match, depth 2 */
	{ size_t offset2=999999999;
	size_t const ml2 = searchMax(ms, ip, iend, &offset2);
	int const gain2 = (int)(ml24 - ZSTD_highbit32((U32)offset2+1)); / raw approx */
	int const gain1 = (int)(matchLength*4 - ZSTD_highbit32((U32)offset+1) + 7);
	if ((ml2 >= 4) && (gain2 > gain1)) {
	matchLength = ml2, offset = offset2, start = ip;
	continue;
	} } }
	break; /* nothing found : store previous solution */
	}

	/* catch up */
	if (offset) {
	U32 const matchIndex = (U32)((start-base) - (offset - ZSTD_REP_MOVE));
	const BYTE* match = (matchIndex < dictLimit) ? dictBase + matchIndex : base + matchIndex;
	const BYTE* const mStart = (matchIndex < dictLimit) ? dictStart : prefixStart;
	while ((start>anchor) && (match>mStart) && (start[-1] == match[-1])) { start--; match--; matchLength++; } /* catch up */
	offset_2 = offset_1; offset_1 = (U32)(offset - ZSTD_REP_MOVE);
	}

	/* store sequence */
	_storeSequence:
	{ size_t const litLength = start - anchor;
	ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offset, matchLength-MINMATCH);
	anchor = ip = start + matchLength;
	}

	/* check immediate repcode */
	while (ip <= ilimit) {
	const U32 repCurrent = (U32)(ip-base);
	const U32 windowLow = ZSTD_getLowestMatchIndex(ms, repCurrent, windowLog);
	const U32 repIndex = repCurrent - offset_2;
	const BYTE* const repBase = repIndex < dictLimit ? dictBase : base;
	const BYTE* const repMatch = repBase + repIndex;
	if (((U32)((dictLimit-1) - repIndex) >= 3) & (repIndex > windowLow)) /* intentional overflow */
	if (MEM_read32(ip) == MEM_read32(repMatch)) {
	/* repcode detected we should take it */
	const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
	matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
	offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap offset history */
	ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH);
	ip += matchLength;
	anchor = ip;
	continue; /* faster when present ... (?) */
	}
	break;
	} }

	/* Save reps for next block */
	rep[0] = offset_1;
	rep[1] = offset_2;

	/* Return the last literals size */
	return (size_t)(iend - anchor);
	}


	size_t ZSTD_compressBlock_greedy_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 0);
	}

	size_t ZSTD_compressBlock_lazy_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)

	{
	return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 1);
	}

	size_t ZSTD_compressBlock_lazy2_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)

	{
	return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_hashChain, 2);
	}

	size_t ZSTD_compressBlock_btlazy2_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)

	{
	return ZSTD_compressBlock_lazy_extDict_generic(ms, seqStore, rep, src, srcSize, search_binaryTree, 2);
	}
	/** ended inlining compress/zstd_lazy.c **/
	/** start inlining compress/zstd_ldm.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/** skipping file: zstd_ldm.h **/

	/** skipping file: ../common/debug.h **/
	/** skipping file: zstd_fast.h **/
	/** skipping file: zstd_double_fast.h **/

	#define LDM_BUCKET_SIZE_LOG 3
	#define LDM_MIN_MATCH_LENGTH 64
	#define LDM_HASH_RLOG 7
	#define LDM_HASH_CHAR_OFFSET 10

	void ZSTD_ldm_adjustParameters(ldmParams_t* params,
	ZSTD_compressionParameters const* cParams)
	{
	params->windowLog = cParams->windowLog;
	ZSTD_STATIC_ASSERT(LDM_BUCKET_SIZE_LOG <= ZSTD_LDM_BUCKETSIZELOG_MAX);
	DEBUGLOG(4, "ZSTD_ldm_adjustParameters");
	if (!params->bucketSizeLog) params->bucketSizeLog = LDM_BUCKET_SIZE_LOG;
	if (!params->minMatchLength) params->minMatchLength = LDM_MIN_MATCH_LENGTH;
	if (cParams->strategy >= ZSTD_btopt) {
	/* Get out of the way of the optimal parser */
	U32 const minMatch = MAX(cParams->targetLength, params->minMatchLength);
	assert(minMatch >= ZSTD_LDM_MINMATCH_MIN);
	assert(minMatch <= ZSTD_LDM_MINMATCH_MAX);
	params->minMatchLength = minMatch;
	}
	if (params->hashLog == 0) {
	params->hashLog = MAX(ZSTD_HASHLOG_MIN, params->windowLog - LDM_HASH_RLOG);
	assert(params->hashLog <= ZSTD_HASHLOG_MAX);
	}
	if (params->hashRateLog == 0) {
	params->hashRateLog = params->windowLog < params->hashLog
	? 0
	: params->windowLog - params->hashLog;
	}
	params->bucketSizeLog = MIN(params->bucketSizeLog, params->hashLog);
	}

	size_t ZSTD_ldm_getTableSize(ldmParams_t params)
	{
	size_t const ldmHSize = ((size_t)1) << params.hashLog;
	size_t const ldmBucketSizeLog = MIN(params.bucketSizeLog, params.hashLog);
	size_t const ldmBucketSize = ((size_t)1) << (params.hashLog - ldmBucketSizeLog);
	size_t const totalSize = ZSTD_cwksp_alloc_size(ldmBucketSize)
	+ ZSTD_cwksp_alloc_size(ldmHSize * sizeof(ldmEntry_t));
	return params.enableLdm ? totalSize : 0;
	}

	size_t ZSTD_ldm_getMaxNbSeq(ldmParams_t params, size_t maxChunkSize)
	{
	return params.enableLdm ? (maxChunkSize / params.minMatchLength) : 0;
	}

	/** ZSTD_ldm_getSmallHash() :
	* numBits should be <= 32
	* If numBits==0, returns 0.
	* @return : the most significant numBits of value. */
	static U32 ZSTD_ldm_getSmallHash(U64 value, U32 numBits)
	{
	assert(numBits <= 32);
	return numBits == 0 ? 0 : (U32)(value >> (64 - numBits));
	}

	/** ZSTD_ldm_getChecksum() :
	* numBitsToDiscard should be <= 32
	* @return : the next most significant 32 bits after numBitsToDiscard */
	static U32 ZSTD_ldm_getChecksum(U64 hash, U32 numBitsToDiscard)
	{
	assert(numBitsToDiscard <= 32);
	return (hash >> (64 - 32 - numBitsToDiscard)) & 0xFFFFFFFF;
	}

	/** ZSTD_ldm_getTag() ;
	* Given the hash, returns the most significant numTagBits bits
	* after (32 + hbits) bits.
	*
	* If there are not enough bits remaining, return the last
	* numTagBits bits. */
	static U32 ZSTD_ldm_getTag(U64 hash, U32 hbits, U32 numTagBits)
	{
	assert(numTagBits < 32 && hbits <= 32);
	if (32 - hbits < numTagBits) {
	return hash & (((U32)1 << numTagBits) - 1);
	} else {
	return (hash >> (32 - hbits - numTagBits)) & (((U32)1 << numTagBits) - 1);
	}
	}

	/** ZSTD_ldm_getBucket() :
	* Returns a pointer to the start of the bucket associated with hash. */
	static ldmEntry_t* ZSTD_ldm_getBucket(
	ldmState_t* ldmState, size_t hash, ldmParams_t const ldmParams)
	{
	return ldmState->hashTable + (hash << ldmParams.bucketSizeLog);
	}

	/** ZSTD_ldm_insertEntry() :
	* Insert the entry with corresponding hash into the hash table */
	static void ZSTD_ldm_insertEntry(ldmState_t* ldmState,
	size_t const hash, const ldmEntry_t entry,
	ldmParams_t const ldmParams)
	{
	BYTE* const bucketOffsets = ldmState->bucketOffsets;
	*(ZSTD_ldm_getBucket(ldmState, hash, ldmParams) + bucketOffsets[hash]) = entry;
	bucketOffsets[hash]++;
	bucketOffsets[hash] &= ((U32)1 << ldmParams.bucketSizeLog) - 1;
	}

	/** ZSTD_ldm_makeEntryAndInsertByTag() :
	*
	* Gets the small hash, checksum, and tag from the rollingHash.
	*
	* If the tag matches (1 << ldmParams.hashRateLog)-1, then
	* creates an ldmEntry from the offset, and inserts it into the hash table.
	*
	* hBits is the length of the small hash, which is the most significant hBits
	* of rollingHash. The checksum is the next 32 most significant bits, followed
	* by ldmParams.hashRateLog bits that make up the tag. */
	static void ZSTD_ldm_makeEntryAndInsertByTag(ldmState_t* ldmState,
	U64 const rollingHash,
	U32 const hBits,
	U32 const offset,
	ldmParams_t const ldmParams)
	{
	U32 const tag = ZSTD_ldm_getTag(rollingHash, hBits, ldmParams.hashRateLog);
	U32 const tagMask = ((U32)1 << ldmParams.hashRateLog) - 1;
	if (tag == tagMask) {
	U32 const hash = ZSTD_ldm_getSmallHash(rollingHash, hBits);
	U32 const checksum = ZSTD_ldm_getChecksum(rollingHash, hBits);
	ldmEntry_t entry;
	entry.offset = offset;
	entry.checksum = checksum;
	ZSTD_ldm_insertEntry(ldmState, hash, entry, ldmParams);
	}
	}

	/** ZSTD_ldm_countBackwardsMatch() :
	* Returns the number of bytes that match backwards before pIn and pMatch.
	*
	* We count only bytes where pMatch >= pBase and pIn >= pAnchor. */
	static size_t ZSTD_ldm_countBackwardsMatch(
	const BYTE* pIn, const BYTE* pAnchor,
	const BYTE* pMatch, const BYTE* pBase)
	{
	size_t matchLength = 0;
	while (pIn > pAnchor && pMatch > pBase && pIn[-1] == pMatch[-1]) {
	pIn--;
	pMatch--;
	matchLength++;
	}
	return matchLength;
	}

	/** ZSTD_ldm_fillFastTables() :
	*
	* Fills the relevant tables for the ZSTD_fast and ZSTD_dfast strategies.
	* This is similar to ZSTD_loadDictionaryContent.
	*
	* The tables for the other strategies are filled within their
	* block compressors. */
	static size_t ZSTD_ldm_fillFastTables(ZSTD_matchState_t* ms,
	void const* end)
	{
	const BYTE* const iend = (const BYTE*)end;

	switch(ms->cParams.strategy)
	{
	case ZSTD_fast:
	ZSTD_fillHashTable(ms, iend, ZSTD_dtlm_fast);
	break;

	case ZSTD_dfast:
	ZSTD_fillDoubleHashTable(ms, iend, ZSTD_dtlm_fast);
	break;

	case ZSTD_greedy:
	case ZSTD_lazy:
	case ZSTD_lazy2:
	case ZSTD_btlazy2:
	case ZSTD_btopt:
	case ZSTD_btultra:
	case ZSTD_btultra2:
	break;
	default:
	assert(0); /* not possible : not a valid strategy id */
	}

	return 0;
	}

	/** ZSTD_ldm_fillLdmHashTable() :
	*
	* Fills hashTable from (lastHashed + 1) to iend (non-inclusive).
	* lastHash is the rolling hash that corresponds to lastHashed.
	*
	* Returns the rolling hash corresponding to position iend-1. */
	static U64 ZSTD_ldm_fillLdmHashTable(ldmState_t* state,
	U64 lastHash, const BYTE* lastHashed,
	const BYTE* iend, const BYTE* base,
	U32 hBits, ldmParams_t const ldmParams)
	{
	U64 rollingHash = lastHash;
	const BYTE* cur = lastHashed + 1;

	while (cur < iend) {
	rollingHash = ZSTD_rollingHash_rotate(rollingHash, cur[-1],
	cur[ldmParams.minMatchLength-1],
	state->hashPower);
	ZSTD_ldm_makeEntryAndInsertByTag(state,
	rollingHash, hBits,
	(U32)(cur - base), ldmParams);
	++cur;
	}
	return rollingHash;
	}

	void ZSTD_ldm_fillHashTable(
	ldmState_t* state, const BYTE* ip,
	const BYTE* iend, ldmParams_t const* params)
	{
	DEBUGLOG(5, "ZSTD_ldm_fillHashTable");
	if ((size_t)(iend - ip) >= params->minMatchLength) {
	U64 startingHash = ZSTD_rollingHash_compute(ip, params->minMatchLength);
	ZSTD_ldm_fillLdmHashTable(
	state, startingHash, ip, iend - params->minMatchLength, state->window.base,
	params->hashLog - params->bucketSizeLog,
	*params);
	}
	}


	/** ZSTD_ldm_limitTableUpdate() :
	*
	* Sets cctx->nextToUpdate to a position corresponding closer to anchor
	* if it is far way
	* (after a long match, only update tables a limited amount). */
	static void ZSTD_ldm_limitTableUpdate(ZSTD_matchState_t* ms, const BYTE* anchor)
	{
	U32 const current = (U32)(anchor - ms->window.base);
	if (current > ms->nextToUpdate + 1024) {
	ms->nextToUpdate =
	current - MIN(512, current - ms->nextToUpdate - 1024);
	}
	}

	static size_t ZSTD_ldm_generateSequences_internal(
	ldmState_t* ldmState, rawSeqStore_t* rawSeqStore,
	ldmParams_t const* params, void const* src, size_t srcSize)
	{
	/* LDM parameters */
	int const extDict = ZSTD_window_hasExtDict(ldmState->window);
	U32 const minMatchLength = params->minMatchLength;
	U64 const hashPower = ldmState->hashPower;
	U32 const hBits = params->hashLog - params->bucketSizeLog;
	U32 const ldmBucketSize = 1U << params->bucketSizeLog;
	U32 const hashRateLog = params->hashRateLog;
	U32 const ldmTagMask = (1U << params->hashRateLog) - 1;
	/* Prefix and extDict parameters */
	U32 const dictLimit = ldmState->window.dictLimit;
	U32 const lowestIndex = extDict ? ldmState->window.lowLimit : dictLimit;
	BYTE const* const base = ldmState->window.base;
	BYTE const* const dictBase = extDict ? ldmState->window.dictBase : NULL;
	BYTE const* const dictStart = extDict ? dictBase + lowestIndex : NULL;
	BYTE const* const dictEnd = extDict ? dictBase + dictLimit : NULL;
	BYTE const* const lowPrefixPtr = base + dictLimit;
	/* Input bounds */
	BYTE const* const istart = (BYTE const*)src;
	BYTE const* const iend = istart + srcSize;
	BYTE const* const ilimit = iend - MAX(minMatchLength, HASH_READ_SIZE);
	/* Input positions */
	BYTE const* anchor = istart;
	BYTE const* ip = istart;
	/* Rolling hash */
	BYTE const* lastHashed = NULL;
	U64 rollingHash = 0;

	while (ip <= ilimit) {
	size_t mLength;
	U32 const current = (U32)(ip - base);
	size_t forwardMatchLength = 0, backwardMatchLength = 0;
	ldmEntry_t* bestEntry = NULL;
	if (ip != istart) {
	rollingHash = ZSTD_rollingHash_rotate(rollingHash, lastHashed[0],
	lastHashed[minMatchLength],
	hashPower);
	} else {
	rollingHash = ZSTD_rollingHash_compute(ip, minMatchLength);
	}
	lastHashed = ip;

	/* Do not insert and do not look for a match */
	if (ZSTD_ldm_getTag(rollingHash, hBits, hashRateLog) != ldmTagMask) {
	ip++;
	continue;
	}

	/* Get the best entry and compute the match lengths */
	{
	ldmEntry_t* const bucket =
	ZSTD_ldm_getBucket(ldmState,
	ZSTD_ldm_getSmallHash(rollingHash, hBits),
	*params);
	ldmEntry_t* cur;
	size_t bestMatchLength = 0;
	U32 const checksum = ZSTD_ldm_getChecksum(rollingHash, hBits);

	for (cur = bucket; cur < bucket + ldmBucketSize; ++cur) {
	size_t curForwardMatchLength, curBackwardMatchLength,
	curTotalMatchLength;
	if (cur->checksum != checksum \|\| cur->offset <= lowestIndex) {
	continue;
	}
	if (extDict) {
	BYTE const* const curMatchBase =
	cur->offset < dictLimit ? dictBase : base;
	BYTE const* const pMatch = curMatchBase + cur->offset;
	BYTE const* const matchEnd =
	cur->offset < dictLimit ? dictEnd : iend;
	BYTE const* const lowMatchPtr =
	cur->offset < dictLimit ? dictStart : lowPrefixPtr;

	curForwardMatchLength = ZSTD_count_2segments(
	ip, pMatch, iend,
	matchEnd, lowPrefixPtr);
	if (curForwardMatchLength < minMatchLength) {
	continue;
	}
	curBackwardMatchLength =
	ZSTD_ldm_countBackwardsMatch(ip, anchor, pMatch,
	lowMatchPtr);
	curTotalMatchLength = curForwardMatchLength +
	curBackwardMatchLength;
	} else { /* !extDict */
	BYTE const* const pMatch = base + cur->offset;
	curForwardMatchLength = ZSTD_count(ip, pMatch, iend);
	if (curForwardMatchLength < minMatchLength) {
	continue;
	}
	curBackwardMatchLength =
	ZSTD_ldm_countBackwardsMatch(ip, anchor, pMatch,
	lowPrefixPtr);
	curTotalMatchLength = curForwardMatchLength +
	curBackwardMatchLength;
	}

	if (curTotalMatchLength > bestMatchLength) {
	bestMatchLength = curTotalMatchLength;
	forwardMatchLength = curForwardMatchLength;
	backwardMatchLength = curBackwardMatchLength;
	bestEntry = cur;
	}
	}
	}

	/* No match found -- continue searching */
	if (bestEntry == NULL) {
	ZSTD_ldm_makeEntryAndInsertByTag(ldmState, rollingHash,
	hBits, current,
	*params);
	ip++;
	continue;
	}

	/* Match found */
	mLength = forwardMatchLength + backwardMatchLength;
	ip -= backwardMatchLength;

	{
	/* Store the sequence:
	* ip = current - backwardMatchLength
	* The match is at (bestEntry->offset - backwardMatchLength)
	*/
	U32 const matchIndex = bestEntry->offset;
	U32 const offset = current - matchIndex;
	rawSeq* const seq = rawSeqStore->seq + rawSeqStore->size;

	/* Out of sequence storage */
	if (rawSeqStore->size == rawSeqStore->capacity)
	return ERROR(dstSize_tooSmall);
	seq->litLength = (U32)(ip - anchor);
	seq->matchLength = (U32)mLength;
	seq->offset = offset;
	rawSeqStore->size++;
	}

	/* Insert the current entry into the hash table */
	ZSTD_ldm_makeEntryAndInsertByTag(ldmState, rollingHash, hBits,
	(U32)(lastHashed - base),
	*params);

	assert(ip + backwardMatchLength == lastHashed);

	/* Fill the hash table from lastHashed+1 to ip+mLength*/
	/* Heuristic: don't need to fill the entire table at end of block */
	if (ip + mLength <= ilimit) {
	rollingHash = ZSTD_ldm_fillLdmHashTable(
	ldmState, rollingHash, lastHashed,
	ip + mLength, base, hBits, *params);
	lastHashed = ip + mLength - 1;
	}
	ip += mLength;
	anchor = ip;
	}
	return iend - anchor;
	}

	/*! ZSTD_ldm_reduceTable() :
	* reduce table indexes by `reducerValue` */
	static void ZSTD_ldm_reduceTable(ldmEntry_t* const table, U32 const size,
	U32 const reducerValue)
	{
	U32 u;
	for (u = 0; u < size; u++) {
	if (table[u].offset < reducerValue) table[u].offset = 0;
	else table[u].offset -= reducerValue;
	}
	}

	size_t ZSTD_ldm_generateSequences(
	ldmState_t* ldmState, rawSeqStore_t* sequences,
	ldmParams_t const* params, void const* src, size_t srcSize)
	{
	U32 const maxDist = 1U << params->windowLog;
	BYTE const* const istart = (BYTE const*)src;
	BYTE const* const iend = istart + srcSize;
	size_t const kMaxChunkSize = 1 << 20;
	size_t const nbChunks = (srcSize / kMaxChunkSize) + ((srcSize % kMaxChunkSize) != 0);
	size_t chunk;
	size_t leftoverSize = 0;

	assert(ZSTD_CHUNKSIZE_MAX >= kMaxChunkSize);
	/* Check that ZSTD_window_update() has been called for this chunk prior
	* to passing it to this function.
	*/
	assert(ldmState->window.nextSrc >= (BYTE const*)src + srcSize);
	/* The input could be very large (in zstdmt), so it must be broken up into
	* chunks to enforce the maximum distance and handle overflow correction.
	*/
	assert(sequences->pos <= sequences->size);
	assert(sequences->size <= sequences->capacity);
	for (chunk = 0; chunk < nbChunks && sequences->size < sequences->capacity; ++chunk) {
	BYTE const* const chunkStart = istart + chunk * kMaxChunkSize;
	size_t const remaining = (size_t)(iend - chunkStart);
	BYTE const *const chunkEnd =
	(remaining < kMaxChunkSize) ? iend : chunkStart + kMaxChunkSize;
	size_t const chunkSize = chunkEnd - chunkStart;
	size_t newLeftoverSize;
	size_t const prevSize = sequences->size;

	assert(chunkStart < iend);
	/* 1. Perform overflow correction if necessary. */
	if (ZSTD_window_needOverflowCorrection(ldmState->window, chunkEnd)) {
	U32 const ldmHSize = 1U << params->hashLog;
	U32 const correction = ZSTD_window_correctOverflow(
	&ldmState->window, /* cycleLog */ 0, maxDist, chunkStart);
	ZSTD_ldm_reduceTable(ldmState->hashTable, ldmHSize, correction);
	/* invalidate dictionaries on overflow correction */
	ldmState->loadedDictEnd = 0;
	}
	/* 2. We enforce the maximum offset allowed.
	*
	* kMaxChunkSize should be small enough that we don't lose too much of
	* the window through early invalidation.
	* TODO: * Test the chunk size.
	* * Try invalidation after the sequence generation and test the
	* the offset against maxDist directly.
	*
	* NOTE: Because of dictionaries + sequence splitting we MUST make sure
	* that any offset used is valid at the END of the sequence, since it may
	* be split into two sequences. This condition holds when using
	* ZSTD_window_enforceMaxDist(), but if we move to checking offsets
	* against maxDist directly, we'll have to carefully handle that case.
	*/
	ZSTD_window_enforceMaxDist(&ldmState->window, chunkEnd, maxDist, &ldmState->loadedDictEnd, NULL);
	/* 3. Generate the sequences for the chunk, and get newLeftoverSize. */
	newLeftoverSize = ZSTD_ldm_generateSequences_internal(
	ldmState, sequences, params, chunkStart, chunkSize);
	if (ZSTD_isError(newLeftoverSize))
	return newLeftoverSize;
	/* 4. We add the leftover literals from previous iterations to the first
	* newly generated sequence, or add the `newLeftoverSize` if none are
	* generated.
	*/
	/* Prepend the leftover literals from the last call */
	if (prevSize < sequences->size) {
	sequences->seq[prevSize].litLength += (U32)leftoverSize;
	leftoverSize = newLeftoverSize;
	} else {
	assert(newLeftoverSize == chunkSize);
	leftoverSize += chunkSize;
	}
	}
	return 0;
	}

	void ZSTD_ldm_skipSequences(rawSeqStore_t* rawSeqStore, size_t srcSize, U32 const minMatch) {
	while (srcSize > 0 && rawSeqStore->pos < rawSeqStore->size) {
	rawSeq* seq = rawSeqStore->seq + rawSeqStore->pos;
	if (srcSize <= seq->litLength) {
	/* Skip past srcSize literals */
	seq->litLength -= (U32)srcSize;
	return;
	}
	srcSize -= seq->litLength;
	seq->litLength = 0;
	if (srcSize < seq->matchLength) {
	/* Skip past the first srcSize of the match */
	seq->matchLength -= (U32)srcSize;
	if (seq->matchLength < minMatch) {
	/* The match is too short, omit it */
	if (rawSeqStore->pos + 1 < rawSeqStore->size) {
	seq[1].litLength += seq[0].matchLength;
	}
	rawSeqStore->pos++;
	}
	return;
	}
	srcSize -= seq->matchLength;
	seq->matchLength = 0;
	rawSeqStore->pos++;
	}
	}

	/**
	* If the sequence length is longer than remaining then the sequence is split
	* between this block and the next.
	*
	* Returns the current sequence to handle, or if the rest of the block should
	* be literals, it returns a sequence with offset == 0.
	*/
	static rawSeq maybeSplitSequence(rawSeqStore_t* rawSeqStore,
	U32 const remaining, U32 const minMatch)
	{
	rawSeq sequence = rawSeqStore->seq[rawSeqStore->pos];
	assert(sequence.offset > 0);
	/* Likely: No partial sequence */
	if (remaining >= sequence.litLength + sequence.matchLength) {
	rawSeqStore->pos++;
	return sequence;
	}
	/* Cut the sequence short (offset == 0 ==> rest is literals). */
	if (remaining <= sequence.litLength) {
	sequence.offset = 0;
	} else if (remaining < sequence.litLength + sequence.matchLength) {
	sequence.matchLength = remaining - sequence.litLength;
	if (sequence.matchLength < minMatch) {
	sequence.offset = 0;
	}
	}
	/* Skip past `remaining` bytes for the future sequences. */
	ZSTD_ldm_skipSequences(rawSeqStore, remaining, minMatch);
	return sequence;
	}

	size_t ZSTD_ldm_blockCompress(rawSeqStore_t* rawSeqStore,
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	void const* src, size_t srcSize)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	unsigned const minMatch = cParams->minMatch;
	ZSTD_blockCompressor const blockCompressor =
	ZSTD_selectBlockCompressor(cParams->strategy, ZSTD_matchState_dictMode(ms));
	/* Input bounds */
	BYTE const* const istart = (BYTE const*)src;
	BYTE const* const iend = istart + srcSize;
	/* Input positions */
	BYTE const* ip = istart;

	DEBUGLOG(5, "ZSTD_ldm_blockCompress: srcSize=%zu", srcSize);
	assert(rawSeqStore->pos <= rawSeqStore->size);
	assert(rawSeqStore->size <= rawSeqStore->capacity);
	/* Loop through each sequence and apply the block compressor to the lits */
	while (rawSeqStore->pos < rawSeqStore->size && ip < iend) {
	/* maybeSplitSequence updates rawSeqStore->pos */
	rawSeq const sequence = maybeSplitSequence(rawSeqStore,
	(U32)(iend - ip), minMatch);
	int i;
	/* End signal */
	if (sequence.offset == 0)
	break;

	assert(ip + sequence.litLength + sequence.matchLength <= iend);

	/* Fill tables for block compressor */
	ZSTD_ldm_limitTableUpdate(ms, ip);
	ZSTD_ldm_fillFastTables(ms, ip);
	/* Run the block compressor */
	DEBUGLOG(5, "pos %u : calling block compressor on segment of size %u", (unsigned)(ip-istart), sequence.litLength);
	{
	size_t const newLitLength =
	blockCompressor(ms, seqStore, rep, ip, sequence.litLength);
	ip += sequence.litLength;
	/* Update the repcodes */
	for (i = ZSTD_REP_NUM - 1; i > 0; i--)
	rep[i] = rep[i-1];
	rep[0] = sequence.offset;
	/* Store the sequence */
	ZSTD_storeSeq(seqStore, newLitLength, ip - newLitLength, iend,
	sequence.offset + ZSTD_REP_MOVE,
	sequence.matchLength - MINMATCH);
	ip += sequence.matchLength;
	}
	}
	/* Fill the tables for the block compressor */
	ZSTD_ldm_limitTableUpdate(ms, ip);
	ZSTD_ldm_fillFastTables(ms, ip);
	/* Compress the last literals */
	return blockCompressor(ms, seqStore, rep, ip, iend - ip);
	}
	/** ended inlining compress/zstd_ldm.c **/
	/** start inlining compress/zstd_opt.c **/
	/*
	* Copyright (c) 2016-2020, Przemyslaw Skibinski, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/** skipping file: zstd_compress_internal.h **/
	/** skipping file: hist.h **/
	/** skipping file: zstd_opt.h **/


	#define ZSTD_LITFREQ_ADD 2 /* scaling factor for litFreq, so that frequencies adapt faster to new stats */
	#define ZSTD_FREQ_DIV 4 /* log factor when using previous stats to init next stats */
	#define ZSTD_MAX_PRICE (1<<30)

	#define ZSTD_PREDEF_THRESHOLD 1024 /* if srcSize < ZSTD_PREDEF_THRESHOLD, symbols' cost is assumed static, directly determined by pre-defined distributions */


	/-************************************
	* Price functions for optimal parser
	***************************************/

	#if 0 /* approximation at bit level */
	# define BITCOST_ACCURACY 0
	# define BITCOST_MULTIPLIER (1 << BITCOST_ACCURACY)
	# define WEIGHT(stat) ((void)opt, ZSTD_bitWeight(stat))
	#elif 0 /* fractional bit accuracy */
	# define BITCOST_ACCURACY 8
	# define BITCOST_MULTIPLIER (1 << BITCOST_ACCURACY)
	# define WEIGHT(stat,opt) ((void)opt, ZSTD_fracWeight(stat))
	#else /* opt==approx, ultra==accurate */
	# define BITCOST_ACCURACY 8
	# define BITCOST_MULTIPLIER (1 << BITCOST_ACCURACY)
	# define WEIGHT(stat,opt) (opt ? ZSTD_fracWeight(stat) : ZSTD_bitWeight(stat))
	#endif

	MEM_STATIC U32 ZSTD_bitWeight(U32 stat)
	{
	return (ZSTD_highbit32(stat+1) * BITCOST_MULTIPLIER);
	}

	MEM_STATIC U32 ZSTD_fracWeight(U32 rawStat)
	{
	U32 const stat = rawStat + 1;
	U32 const hb = ZSTD_highbit32(stat);
	U32 const BWeight = hb * BITCOST_MULTIPLIER;
	U32 const FWeight = (stat << BITCOST_ACCURACY) >> hb;
	U32 const weight = BWeight + FWeight;
	assert(hb + BITCOST_ACCURACY < 31);
	return weight;
	}

	#if (DEBUGLEVEL>=2)
	/* debugging function,
	* @return price in bytes as fractional value
	* for debug messages only */
	MEM_STATIC double ZSTD_fCost(U32 price)
	{
	return (double)price / (BITCOST_MULTIPLIER*8);
	}
	#endif

	static int ZSTD_compressedLiterals(optState_t const* const optPtr)
	{
	return optPtr->literalCompressionMode != ZSTD_lcm_uncompressed;
	}

	static void ZSTD_setBasePrices(optState_t* optPtr, int optLevel)
	{
	if (ZSTD_compressedLiterals(optPtr))
	optPtr->litSumBasePrice = WEIGHT(optPtr->litSum, optLevel);
	optPtr->litLengthSumBasePrice = WEIGHT(optPtr->litLengthSum, optLevel);
	optPtr->matchLengthSumBasePrice = WEIGHT(optPtr->matchLengthSum, optLevel);
	optPtr->offCodeSumBasePrice = WEIGHT(optPtr->offCodeSum, optLevel);
	}


	/* ZSTD_downscaleStat() :
	* reduce all elements in table by a factor 2^(ZSTD_FREQ_DIV+malus)
	* return the resulting sum of elements */
	static U32 ZSTD_downscaleStat(unsigned* table, U32 lastEltIndex, int malus)
	{
	U32 s, sum=0;
	DEBUGLOG(5, "ZSTD_downscaleStat (nbElts=%u)", (unsigned)lastEltIndex+1);
	assert(ZSTD_FREQ_DIV+malus > 0 && ZSTD_FREQ_DIV+malus < 31);
	for (s=0; s<lastEltIndex+1; s++) {
	table[s] = 1 + (table[s] >> (ZSTD_FREQ_DIV+malus));
	sum += table[s];
	}
	return sum;
	}

	/* ZSTD_rescaleFreqs() :
	* if first block (detected by optPtr->litLengthSum == 0) : init statistics
	* take hints from dictionary if there is one
	* or init from zero, using src for literals stats, or flat 1 for match symbols
	* otherwise downscale existing stats, to be used as seed for next block.
	*/
	static void
	ZSTD_rescaleFreqs(optState_t* const optPtr,
	const BYTE* const src, size_t const srcSize,
	int const optLevel)
	{
	int const compressedLiterals = ZSTD_compressedLiterals(optPtr);
	DEBUGLOG(5, "ZSTD_rescaleFreqs (srcSize=%u)", (unsigned)srcSize);
	optPtr->priceType = zop_dynamic;

	if (optPtr->litLengthSum == 0) { /* first block : init */
	if (srcSize <= ZSTD_PREDEF_THRESHOLD) { /* heuristic */
	DEBUGLOG(5, "(srcSize <= ZSTD_PREDEF_THRESHOLD) => zop_predef");
	optPtr->priceType = zop_predef;
	}

	assert(optPtr->symbolCosts != NULL);
	if (optPtr->symbolCosts->huf.repeatMode == HUF_repeat_valid) {
	/* huffman table presumed generated by dictionary */
	optPtr->priceType = zop_dynamic;

	if (compressedLiterals) {
	unsigned lit;
	assert(optPtr->litFreq != NULL);
	optPtr->litSum = 0;
	for (lit=0; lit<=MaxLit; lit++) {
	U32 const scaleLog = 11; /* scale to 2K */
	U32 const bitCost = HUF_getNbBits(optPtr->symbolCosts->huf.CTable, lit);
	assert(bitCost <= scaleLog);
	optPtr->litFreq[lit] = bitCost ? 1 << (scaleLog-bitCost) : 1 /minimum to calculate cost/;
	optPtr->litSum += optPtr->litFreq[lit];
	} }

	{ unsigned ll;
	FSE_CState_t llstate;
	FSE_initCState(&llstate, optPtr->symbolCosts->fse.litlengthCTable);
	optPtr->litLengthSum = 0;
	for (ll=0; ll<=MaxLL; ll++) {
	U32 const scaleLog = 10; /* scale to 1K */
	U32 const bitCost = FSE_getMaxNbBits(llstate.symbolTT, ll);
	assert(bitCost < scaleLog);
	optPtr->litLengthFreq[ll] = bitCost ? 1 << (scaleLog-bitCost) : 1 /minimum to calculate cost/;
	optPtr->litLengthSum += optPtr->litLengthFreq[ll];
	} }

	{ unsigned ml;
	FSE_CState_t mlstate;
	FSE_initCState(&mlstate, optPtr->symbolCosts->fse.matchlengthCTable);
	optPtr->matchLengthSum = 0;
	for (ml=0; ml<=MaxML; ml++) {
	U32 const scaleLog = 10;
	U32 const bitCost = FSE_getMaxNbBits(mlstate.symbolTT, ml);
	assert(bitCost < scaleLog);
	optPtr->matchLengthFreq[ml] = bitCost ? 1 << (scaleLog-bitCost) : 1 /minimum to calculate cost/;
	optPtr->matchLengthSum += optPtr->matchLengthFreq[ml];
	} }

	{ unsigned of;
	FSE_CState_t ofstate;
	FSE_initCState(&ofstate, optPtr->symbolCosts->fse.offcodeCTable);
	optPtr->offCodeSum = 0;
	for (of=0; of<=MaxOff; of++) {
	U32 const scaleLog = 10;
	U32 const bitCost = FSE_getMaxNbBits(ofstate.symbolTT, of);
	assert(bitCost < scaleLog);
	optPtr->offCodeFreq[of] = bitCost ? 1 << (scaleLog-bitCost) : 1 /minimum to calculate cost/;
	optPtr->offCodeSum += optPtr->offCodeFreq[of];
	} }

	} else { /* not a dictionary */

	assert(optPtr->litFreq != NULL);
	if (compressedLiterals) {
	unsigned lit = MaxLit;
	HIST_count_simple(optPtr->litFreq, &lit, src, srcSize); /* use raw first block to init statistics */
	optPtr->litSum = ZSTD_downscaleStat(optPtr->litFreq, MaxLit, 1);
	}

	{ unsigned ll;
	for (ll=0; ll<=MaxLL; ll++)
	optPtr->litLengthFreq[ll] = 1;
	}
	optPtr->litLengthSum = MaxLL+1;

	{ unsigned ml;
	for (ml=0; ml<=MaxML; ml++)
	optPtr->matchLengthFreq[ml] = 1;
	}
	optPtr->matchLengthSum = MaxML+1;

	{ unsigned of;
	for (of=0; of<=MaxOff; of++)
	optPtr->offCodeFreq[of] = 1;
	}
	optPtr->offCodeSum = MaxOff+1;

	}

	} else { /* new block : re-use previous statistics, scaled down */

	if (compressedLiterals)
	optPtr->litSum = ZSTD_downscaleStat(optPtr->litFreq, MaxLit, 1);
	optPtr->litLengthSum = ZSTD_downscaleStat(optPtr->litLengthFreq, MaxLL, 0);
	optPtr->matchLengthSum = ZSTD_downscaleStat(optPtr->matchLengthFreq, MaxML, 0);
	optPtr->offCodeSum = ZSTD_downscaleStat(optPtr->offCodeFreq, MaxOff, 0);
	}

	ZSTD_setBasePrices(optPtr, optLevel);
	}

	/* ZSTD_rawLiteralsCost() :
	* price of literals (only) in specified segment (which length can be 0).
	* does not include price of literalLength symbol */
	static U32 ZSTD_rawLiteralsCost(const BYTE* const literals, U32 const litLength,
	const optState_t* const optPtr,
	int optLevel)
	{
	if (litLength == 0) return 0;

	if (!ZSTD_compressedLiterals(optPtr))
	return (litLength << 3) * BITCOST_MULTIPLIER; /* Uncompressed - 8 bytes per literal. */

	if (optPtr->priceType == zop_predef)
	return (litLength6) BITCOST_MULTIPLIER; /* 6 bit per literal - no statistic used */

	/* dynamic statistics */
	{ U32 price = litLength * optPtr->litSumBasePrice;
	U32 u;
	for (u=0; u < litLength; u++) {
	assert(WEIGHT(optPtr->litFreq[literals[u]], optLevel) <= optPtr->litSumBasePrice); /* literal cost should never be negative */
	price -= WEIGHT(optPtr->litFreq[literals[u]], optLevel);
	}
	return price;
	}
	}

	/* ZSTD_litLengthPrice() :
	* cost of literalLength symbol */
	static U32 ZSTD_litLengthPrice(U32 const litLength, const optState_t* const optPtr, int optLevel)
	{
	if (optPtr->priceType == zop_predef) return WEIGHT(litLength, optLevel);

	/* dynamic statistics */
	{ U32 const llCode = ZSTD_LLcode(litLength);
	return (LL_bits[llCode] * BITCOST_MULTIPLIER)
	+ optPtr->litLengthSumBasePrice
	- WEIGHT(optPtr->litLengthFreq[llCode], optLevel);
	}
	}

	/* ZSTD_getMatchPrice() :
	* Provides the cost of the match part (offset + matchLength) of a sequence
	* Must be combined with ZSTD_fullLiteralsCost() to get the full cost of a sequence.
	* optLevel: when <2, favors small offset for decompression speed (improved cache efficiency) */
	FORCE_INLINE_TEMPLATE U32
	ZSTD_getMatchPrice(U32 const offset,
	U32 const matchLength,
	const optState_t* const optPtr,
	int const optLevel)
	{
	U32 price;
	U32 const offCode = ZSTD_highbit32(offset+1);
	U32 const mlBase = matchLength - MINMATCH;
	assert(matchLength >= MINMATCH);

	if (optPtr->priceType == zop_predef) /* fixed scheme, do not use statistics */
	return WEIGHT(mlBase, optLevel) + ((16 + offCode) * BITCOST_MULTIPLIER);

	/* dynamic statistics */
	price = (offCode * BITCOST_MULTIPLIER) + (optPtr->offCodeSumBasePrice - WEIGHT(optPtr->offCodeFreq[offCode], optLevel));
	if ((optLevel<2) /static/ && offCode >= 20)
	price += (offCode-19)2 BITCOST_MULTIPLIER; /* handicap for long distance offsets, favor decompression speed */

	/* match Length */
	{ U32 const mlCode = ZSTD_MLcode(mlBase);
	price += (ML_bits[mlCode] * BITCOST_MULTIPLIER) + (optPtr->matchLengthSumBasePrice - WEIGHT(optPtr->matchLengthFreq[mlCode], optLevel));
	}

	price += BITCOST_MULTIPLIER / 5; /* heuristic : make matches a bit more costly to favor less sequences -> faster decompression speed */

	DEBUGLOG(8, "ZSTD_getMatchPrice(ml:%u) = %u", matchLength, price);
	return price;
	}

	/* ZSTD_updateStats() :
	* assumption : literals + litLengtn <= iend */
	static void ZSTD_updateStats(optState_t* const optPtr,
	U32 litLength, const BYTE* literals,
	U32 offsetCode, U32 matchLength)
	{
	/* literals */
	if (ZSTD_compressedLiterals(optPtr)) {
	U32 u;
	for (u=0; u < litLength; u++)
	optPtr->litFreq[literals[u]] += ZSTD_LITFREQ_ADD;
	optPtr->litSum += litLength*ZSTD_LITFREQ_ADD;
	}

	/* literal Length */
	{ U32 const llCode = ZSTD_LLcode(litLength);
	optPtr->litLengthFreq[llCode]++;
	optPtr->litLengthSum++;
	}

	/* match offset code (0-2=>repCode; 3+=>offset+2) */
	{ U32 const offCode = ZSTD_highbit32(offsetCode+1);
	assert(offCode <= MaxOff);
	optPtr->offCodeFreq[offCode]++;
	optPtr->offCodeSum++;
	}

	/* match Length */
	{ U32 const mlBase = matchLength - MINMATCH;
	U32 const mlCode = ZSTD_MLcode(mlBase);
	optPtr->matchLengthFreq[mlCode]++;
	optPtr->matchLengthSum++;
	}
	}


	/* ZSTD_readMINMATCH() :
	* function safe only for comparisons
	* assumption : memPtr must be at least 4 bytes before end of buffer */
	MEM_STATIC U32 ZSTD_readMINMATCH(const void* memPtr, U32 length)
	{
	switch (length)
	{
	default :
	case 4 : return MEM_read32(memPtr);
	case 3 : if (MEM_isLittleEndian())
	return MEM_read32(memPtr)<<8;
	else
	return MEM_read32(memPtr)>>8;
	}
	}


	/* Update hashTable3 up to ip (excluded)
	Assumption : always within prefix (i.e. not within extDict) */
	static U32 ZSTD_insertAndFindFirstIndexHash3 (ZSTD_matchState_t* ms,
	U32* nextToUpdate3,
	const BYTE* const ip)
	{
	U32* const hashTable3 = ms->hashTable3;
	U32 const hashLog3 = ms->hashLog3;
	const BYTE* const base = ms->window.base;
	U32 idx = *nextToUpdate3;
	U32 const target = (U32)(ip - base);
	size_t const hash3 = ZSTD_hash3Ptr(ip, hashLog3);
	assert(hashLog3 > 0);

	while(idx < target) {
	hashTable3[ZSTD_hash3Ptr(base+idx, hashLog3)] = idx;
	idx++;
	}

	*nextToUpdate3 = target;
	return hashTable3[hash3];
	}


	/-************************************
	* Binary Tree search
	***************************************/
	/** ZSTD_insertBt1() : add one or multiple positions to tree.
	* ip : assumed <= iend-8 .
	* @return : nb of positions added */
	static U32 ZSTD_insertBt1(
	ZSTD_matchState_t* ms,
	const BYTE* const ip, const BYTE* const iend,
	U32 const mls, const int extDict)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32* const hashTable = ms->hashTable;
	U32 const hashLog = cParams->hashLog;
	size_t const h = ZSTD_hashPtr(ip, hashLog, mls);
	U32* const bt = ms->chainTable;
	U32 const btLog = cParams->chainLog - 1;
	U32 const btMask = (1 << btLog) - 1;
	U32 matchIndex = hashTable[h];
	size_t commonLengthSmaller=0, commonLengthLarger=0;
	const BYTE* const base = ms->window.base;
	const BYTE* const dictBase = ms->window.dictBase;
	const U32 dictLimit = ms->window.dictLimit;
	const BYTE* const dictEnd = dictBase + dictLimit;
	const BYTE* const prefixStart = base + dictLimit;
	const BYTE* match;
	const U32 current = (U32)(ip-base);
	const U32 btLow = btMask >= current ? 0 : current - btMask;
	U32* smallerPtr = bt + 2*(current&btMask);
	U32* largerPtr = smallerPtr + 1;
	U32 dummy32; /* to be nullified at the end */
	U32 const windowLow = ms->window.lowLimit;
	U32 matchEndIdx = current+8+1;
	size_t bestLength = 8;
	U32 nbCompares = 1U << cParams->searchLog;
	#ifdef ZSTD_C_PREDICT
	U32 predictedSmall = (bt + 2((current-1)&btMask) + 0);
	U32 predictedLarge = (bt + 2((current-1)&btMask) + 1);
	predictedSmall += (predictedSmall>0);
	predictedLarge += (predictedLarge>0);
	#endif /* ZSTD_C_PREDICT */

	DEBUGLOG(8, "ZSTD_insertBt1 (%u)", current);

	assert(ip <= iend-8); /* required for h calculation */
	hashTable[h] = current; /* Update Hash Table */

	assert(windowLow > 0);
	while (nbCompares-- && (matchIndex >= windowLow)) {
	U32* const nextPtr = bt + 2*(matchIndex & btMask);
	size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
	assert(matchIndex < current);

	#ifdef ZSTD_C_PREDICT /* note : can create issues when hlog small <= 11 */
	const U32* predictPtr = bt + 2((matchIndex-1) & btMask); / written this way, as bt is a roll buffer */
	if (matchIndex == predictedSmall) {
	/* no need to check length, result known */
	*smallerPtr = matchIndex;
	if (matchIndex <= btLow) { smallerPtr=&dummy32; break; } /* beyond tree size, stop the search */
	smallerPtr = nextPtr+1; /* new "smaller" => larger of match */
	matchIndex = nextPtr[1]; /* new matchIndex larger than previous (closer to current) */
	predictedSmall = predictPtr[1] + (predictPtr[1]>0);
	continue;
	}
	if (matchIndex == predictedLarge) {
	*largerPtr = matchIndex;
	if (matchIndex <= btLow) { largerPtr=&dummy32; break; } /* beyond tree size, stop the search */
	largerPtr = nextPtr;
	matchIndex = nextPtr[0];
	predictedLarge = predictPtr[0] + (predictPtr[0]>0);
	continue;
	}
	#endif

	if (!extDict \|\| (matchIndex+matchLength >= dictLimit)) {
	assert(matchIndex+matchLength >= dictLimit); /* might be wrong if actually extDict */
	match = base + matchIndex;
	matchLength += ZSTD_count(ip+matchLength, match+matchLength, iend);
	} else {
	match = dictBase + matchIndex;
	matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iend, dictEnd, prefixStart);
	if (matchIndex+matchLength >= dictLimit)
	match = base + matchIndex; /* to prepare for next usage of match[matchLength] */
	}

	if (matchLength > bestLength) {
	bestLength = matchLength;
	if (matchLength > matchEndIdx - matchIndex)
	matchEndIdx = matchIndex + (U32)matchLength;
	}

	if (ip+matchLength == iend) { /* equal : no way to know if inf or sup */
	break; /* drop , to guarantee consistency ; miss a bit of compression, but other solutions can corrupt tree */
	}

	if (match[matchLength] < ip[matchLength]) { /* necessarily within buffer */
	/* match is smaller than current */
	smallerPtr = matchIndex; / update smaller idx */
	commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
	if (matchIndex <= btLow) { smallerPtr=&dummy32; break; } /* beyond tree size, stop searching */
	smallerPtr = nextPtr+1; /* new "candidate" => larger than match, which was smaller than target */
	matchIndex = nextPtr[1]; /* new matchIndex, larger than previous and closer to current */
	} else {
	/* match is larger than current */
	*largerPtr = matchIndex;
	commonLengthLarger = matchLength;
	if (matchIndex <= btLow) { largerPtr=&dummy32; break; } /* beyond tree size, stop searching */
	largerPtr = nextPtr;
	matchIndex = nextPtr[0];
	} }

	smallerPtr = largerPtr = 0;
	{ U32 positions = 0;
	if (bestLength > 384) positions = MIN(192, (U32)(bestLength - 384)); /* speed optimization */
	assert(matchEndIdx > current + 8);
	return MAX(positions, matchEndIdx - (current + 8));
	}
	}

	FORCE_INLINE_TEMPLATE
	void ZSTD_updateTree_internal(
	ZSTD_matchState_t* ms,
	const BYTE* const ip, const BYTE* const iend,
	const U32 mls, const ZSTD_dictMode_e dictMode)
	{
	const BYTE* const base = ms->window.base;
	U32 const target = (U32)(ip - base);
	U32 idx = ms->nextToUpdate;
	DEBUGLOG(6, "ZSTD_updateTree_internal, from %u to %u (dictMode:%u)",
	idx, target, dictMode);

	while(idx < target) {
	U32 const forward = ZSTD_insertBt1(ms, base+idx, iend, mls, dictMode == ZSTD_extDict);
	assert(idx < (U32)(idx + forward));
	idx += forward;
	}
	assert((size_t)(ip - base) <= (size_t)(U32)(-1));
	assert((size_t)(iend - base) <= (size_t)(U32)(-1));
	ms->nextToUpdate = target;
	}

	void ZSTD_updateTree(ZSTD_matchState_t* ms, const BYTE* ip, const BYTE* iend) {
	ZSTD_updateTree_internal(ms, ip, iend, ms->cParams.minMatch, ZSTD_noDict);
	}

	FORCE_INLINE_TEMPLATE
	U32 ZSTD_insertBtAndGetAllMatches (
	ZSTD_match_t* matches, /* store result (found matches) in this table (presumed large enough) */
	ZSTD_matchState_t* ms,
	U32* nextToUpdate3,
	const BYTE* const ip, const BYTE* const iLimit, const ZSTD_dictMode_e dictMode,
	const U32 rep[ZSTD_REP_NUM],
	U32 const ll0, /* tells if associated literal length is 0 or not. This value must be 0 or 1 */
	const U32 lengthToBeat,
	U32 const mls /* template */)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32 const sufficient_len = MIN(cParams->targetLength, ZSTD_OPT_NUM -1);
	const BYTE* const base = ms->window.base;
	U32 const current = (U32)(ip-base);
	U32 const hashLog = cParams->hashLog;
	U32 const minMatch = (mls==3) ? 3 : 4;
	U32* const hashTable = ms->hashTable;
	size_t const h = ZSTD_hashPtr(ip, hashLog, mls);
	U32 matchIndex = hashTable[h];
	U32* const bt = ms->chainTable;
	U32 const btLog = cParams->chainLog - 1;
	U32 const btMask= (1U << btLog) - 1;
	size_t commonLengthSmaller=0, commonLengthLarger=0;
	const BYTE* const dictBase = ms->window.dictBase;
	U32 const dictLimit = ms->window.dictLimit;
	const BYTE* const dictEnd = dictBase + dictLimit;
	const BYTE* const prefixStart = base + dictLimit;
	U32 const btLow = (btMask >= current) ? 0 : current - btMask;
	U32 const windowLow = ZSTD_getLowestMatchIndex(ms, current, cParams->windowLog);
	U32 const matchLow = windowLow ? windowLow : 1;
	U32* smallerPtr = bt + 2*(current&btMask);
	U32* largerPtr = bt + 2*(current&btMask) + 1;
	U32 matchEndIdx = current+8+1; /* farthest referenced position of any match => detects repetitive patterns */
	U32 dummy32; /* to be nullified at the end */
	U32 mnum = 0;
	U32 nbCompares = 1U << cParams->searchLog;

	const ZSTD_matchState_t* dms = dictMode == ZSTD_dictMatchState ? ms->dictMatchState : NULL;
	const ZSTD_compressionParameters* const dmsCParams =
	dictMode == ZSTD_dictMatchState ? &dms->cParams : NULL;
	const BYTE* const dmsBase = dictMode == ZSTD_dictMatchState ? dms->window.base : NULL;
	const BYTE* const dmsEnd = dictMode == ZSTD_dictMatchState ? dms->window.nextSrc : NULL;
	U32 const dmsHighLimit = dictMode == ZSTD_dictMatchState ? (U32)(dmsEnd - dmsBase) : 0;
	U32 const dmsLowLimit = dictMode == ZSTD_dictMatchState ? dms->window.lowLimit : 0;
	U32 const dmsIndexDelta = dictMode == ZSTD_dictMatchState ? windowLow - dmsHighLimit : 0;
	U32 const dmsHashLog = dictMode == ZSTD_dictMatchState ? dmsCParams->hashLog : hashLog;
	U32 const dmsBtLog = dictMode == ZSTD_dictMatchState ? dmsCParams->chainLog - 1 : btLog;
	U32 const dmsBtMask = dictMode == ZSTD_dictMatchState ? (1U << dmsBtLog) - 1 : 0;
	U32 const dmsBtLow = dictMode == ZSTD_dictMatchState && dmsBtMask < dmsHighLimit - dmsLowLimit ? dmsHighLimit - dmsBtMask : dmsLowLimit;

	size_t bestLength = lengthToBeat-1;
	DEBUGLOG(8, "ZSTD_insertBtAndGetAllMatches: current=%u", current);

	/* check repCode */
	assert(ll0 <= 1); /* necessarily 1 or 0 */
	{ U32 const lastR = ZSTD_REP_NUM + ll0;
	U32 repCode;
	for (repCode = ll0; repCode < lastR; repCode++) {
	U32 const repOffset = (repCode==ZSTD_REP_NUM) ? (rep[0] - 1) : rep[repCode];
	U32 const repIndex = current - repOffset;
	U32 repLen = 0;
	assert(current >= dictLimit);
	if (repOffset-1 /* intentional overflow, discards 0 and -1 / < current-dictLimit) { / equivalent to `current > repIndex >= dictLimit` */
	/* We must validate the repcode offset because when we're using a dictionary the
	* valid offset range shrinks when the dictionary goes out of bounds.
	*/
	if ((repIndex >= windowLow) & (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(ip - repOffset, minMatch))) {
	repLen = (U32)ZSTD_count(ip+minMatch, ip+minMatch-repOffset, iLimit) + minMatch;
	}
	} else { /* repIndex < dictLimit \|\| repIndex >= current */
	const BYTE* const repMatch = dictMode == ZSTD_dictMatchState ?
	dmsBase + repIndex - dmsIndexDelta :
	dictBase + repIndex;
	assert(current >= windowLow);
	if ( dictMode == ZSTD_extDict
	&& ( ((repOffset-1) /intentional overflow/ < current - windowLow) /* equivalent to `current > repIndex >= windowLow` */
	& (((U32)((dictLimit-1) - repIndex) >= 3) ) /* intentional overflow : do not test positions overlapping 2 memory segments */)
	&& (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(repMatch, minMatch)) ) {
	repLen = (U32)ZSTD_count_2segments(ip+minMatch, repMatch+minMatch, iLimit, dictEnd, prefixStart) + minMatch;
	}
	if (dictMode == ZSTD_dictMatchState
	&& ( ((repOffset-1) /intentional overflow/ < current - (dmsLowLimit + dmsIndexDelta)) /* equivalent to `current > repIndex >= dmsLowLimit` */
	& ((U32)((dictLimit-1) - repIndex) >= 3) ) /* intentional overflow : do not test positions overlapping 2 memory segments */
	&& (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(repMatch, minMatch)) ) {
	repLen = (U32)ZSTD_count_2segments(ip+minMatch, repMatch+minMatch, iLimit, dmsEnd, prefixStart) + minMatch;
	} }
	/* save longer solution */
	if (repLen > bestLength) {
	DEBUGLOG(8, "found repCode %u (ll0:%u, offset:%u) of length %u",
	repCode, ll0, repOffset, repLen);
	bestLength = repLen;
	matches[mnum].off = repCode - ll0;
	matches[mnum].len = (U32)repLen;
	mnum++;
	if ( (repLen > sufficient_len)
	\| (ip+repLen == iLimit) ) { /* best possible */
	return mnum;
	} } } }

	/* HC3 match finder */
	if ((mls == 3) /static/ && (bestLength < mls)) {
	U32 const matchIndex3 = ZSTD_insertAndFindFirstIndexHash3(ms, nextToUpdate3, ip);
	if ((matchIndex3 >= matchLow)
	& (current - matchIndex3 < (1<<18)) /heuristic : longer distance likely too expensive/ ) {
	size_t mlen;
	if ((dictMode == ZSTD_noDict) /static/ \|\| (dictMode == ZSTD_dictMatchState) /static/ \|\| (matchIndex3 >= dictLimit)) {
	const BYTE* const match = base + matchIndex3;
	mlen = ZSTD_count(ip, match, iLimit);
	} else {
	const BYTE* const match = dictBase + matchIndex3;
	mlen = ZSTD_count_2segments(ip, match, iLimit, dictEnd, prefixStart);
	}

	/* save best solution */
	if (mlen >= mls /* == 3 > bestLength */) {
	DEBUGLOG(8, "found small match with hlog3, of length %u",
	(U32)mlen);
	bestLength = mlen;
	assert(current > matchIndex3);
	assert(mnum==0); /* no prior solution */
	matches[0].off = (current - matchIndex3) + ZSTD_REP_MOVE;
	matches[0].len = (U32)mlen;
	mnum = 1;
	if ( (mlen > sufficient_len) \|
	(ip+mlen == iLimit) ) { /* best possible length */
	ms->nextToUpdate = current+1; /* skip insertion */
	return 1;
	} } }
	/* no dictMatchState lookup: dicts don't have a populated HC3 table */
	}

	hashTable[h] = current; /* Update Hash Table */

	while (nbCompares-- && (matchIndex >= matchLow)) {
	U32* const nextPtr = bt + 2*(matchIndex & btMask);
	const BYTE* match;
	size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
	assert(current > matchIndex);

	if ((dictMode == ZSTD_noDict) \|\| (dictMode == ZSTD_dictMatchState) \|\| (matchIndex+matchLength >= dictLimit)) {
	assert(matchIndex+matchLength >= dictLimit); /* ensure the condition is correct when !extDict */
	match = base + matchIndex;
	if (matchIndex >= dictLimit) assert(memcmp(match, ip, matchLength) == 0); /* ensure early section of match is equal as expected */
	matchLength += ZSTD_count(ip+matchLength, match+matchLength, iLimit);
	} else {
	match = dictBase + matchIndex;
	assert(memcmp(match, ip, matchLength) == 0); /* ensure early section of match is equal as expected */
	matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iLimit, dictEnd, prefixStart);
	if (matchIndex+matchLength >= dictLimit)
	match = base + matchIndex; /* prepare for match[matchLength] read */
	}

	if (matchLength > bestLength) {
	DEBUGLOG(8, "found match of length %u at distance %u (offCode=%u)",
	(U32)matchLength, current - matchIndex, current - matchIndex + ZSTD_REP_MOVE);
	assert(matchEndIdx > matchIndex);
	if (matchLength > matchEndIdx - matchIndex)
	matchEndIdx = matchIndex + (U32)matchLength;
	bestLength = matchLength;
	matches[mnum].off = (current - matchIndex) + ZSTD_REP_MOVE;
	matches[mnum].len = (U32)matchLength;
	mnum++;
	if ( (matchLength > ZSTD_OPT_NUM)
	\| (ip+matchLength == iLimit) /* equal : no way to know if inf or sup */) {
	if (dictMode == ZSTD_dictMatchState) nbCompares = 0; /* break should also skip searching dms */
	break; /* drop, to preserve bt consistency (miss a little bit of compression) */
	}
	}

	if (match[matchLength] < ip[matchLength]) {
	/* match smaller than current */
	smallerPtr = matchIndex; / update smaller idx */
	commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
	if (matchIndex <= btLow) { smallerPtr=&dummy32; break; } /* beyond tree size, stop the search */
	smallerPtr = nextPtr+1; /* new candidate => larger than match, which was smaller than current */
	matchIndex = nextPtr[1]; /* new matchIndex, larger than previous, closer to current */
	} else {
	*largerPtr = matchIndex;
	commonLengthLarger = matchLength;
	if (matchIndex <= btLow) { largerPtr=&dummy32; break; } /* beyond tree size, stop the search */
	largerPtr = nextPtr;
	matchIndex = nextPtr[0];
	} }

	smallerPtr = largerPtr = 0;

	if (dictMode == ZSTD_dictMatchState && nbCompares) {
	size_t const dmsH = ZSTD_hashPtr(ip, dmsHashLog, mls);
	U32 dictMatchIndex = dms->hashTable[dmsH];
	const U32* const dmsBt = dms->chainTable;
	commonLengthSmaller = commonLengthLarger = 0;
	while (nbCompares-- && (dictMatchIndex > dmsLowLimit)) {
	const U32* const nextPtr = dmsBt + 2*(dictMatchIndex & dmsBtMask);
	size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
	const BYTE* match = dmsBase + dictMatchIndex;
	matchLength += ZSTD_count_2segments(ip+matchLength, match+matchLength, iLimit, dmsEnd, prefixStart);
	if (dictMatchIndex+matchLength >= dmsHighLimit)
	match = base + dictMatchIndex + dmsIndexDelta; /* to prepare for next usage of match[matchLength] */

	if (matchLength > bestLength) {
	matchIndex = dictMatchIndex + dmsIndexDelta;
	DEBUGLOG(8, "found dms match of length %u at distance %u (offCode=%u)",
	(U32)matchLength, current - matchIndex, current - matchIndex + ZSTD_REP_MOVE);
	if (matchLength > matchEndIdx - matchIndex)
	matchEndIdx = matchIndex + (U32)matchLength;
	bestLength = matchLength;
	matches[mnum].off = (current - matchIndex) + ZSTD_REP_MOVE;
	matches[mnum].len = (U32)matchLength;
	mnum++;
	if ( (matchLength > ZSTD_OPT_NUM)
	\| (ip+matchLength == iLimit) /* equal : no way to know if inf or sup */) {
	break; /* drop, to guarantee consistency (miss a little bit of compression) */
	}
	}

	if (dictMatchIndex <= dmsBtLow) { break; } /* beyond tree size, stop the search */
	if (match[matchLength] < ip[matchLength]) {
	commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
	dictMatchIndex = nextPtr[1]; /* new matchIndex larger than previous (closer to current) */
	} else {
	/* match is larger than current */
	commonLengthLarger = matchLength;
	dictMatchIndex = nextPtr[0];
	}
	}
	}

	assert(matchEndIdx > current+8);
	ms->nextToUpdate = matchEndIdx - 8; /* skip repetitive patterns */
	return mnum;
	}


	FORCE_INLINE_TEMPLATE U32 ZSTD_BtGetAllMatches (
	ZSTD_match_t* matches, /* store result (match found, increasing size) in this table */
	ZSTD_matchState_t* ms,
	U32* nextToUpdate3,
	const BYTE* ip, const BYTE* const iHighLimit, const ZSTD_dictMode_e dictMode,
	const U32 rep[ZSTD_REP_NUM],
	U32 const ll0,
	U32 const lengthToBeat)
	{
	const ZSTD_compressionParameters* const cParams = &ms->cParams;
	U32 const matchLengthSearch = cParams->minMatch;
	DEBUGLOG(8, "ZSTD_BtGetAllMatches");
	if (ip < ms->window.base + ms->nextToUpdate) return 0; /* skipped area */
	ZSTD_updateTree_internal(ms, ip, iHighLimit, matchLengthSearch, dictMode);
	switch(matchLengthSearch)
	{
	case 3 : return ZSTD_insertBtAndGetAllMatches(matches, ms, nextToUpdate3, ip, iHighLimit, dictMode, rep, ll0, lengthToBeat, 3);
	default :
	case 4 : return ZSTD_insertBtAndGetAllMatches(matches, ms, nextToUpdate3, ip, iHighLimit, dictMode, rep, ll0, lengthToBeat, 4);
	case 5 : return ZSTD_insertBtAndGetAllMatches(matches, ms, nextToUpdate3, ip, iHighLimit, dictMode, rep, ll0, lengthToBeat, 5);
	case 7 :
	case 6 : return ZSTD_insertBtAndGetAllMatches(matches, ms, nextToUpdate3, ip, iHighLimit, dictMode, rep, ll0, lengthToBeat, 6);
	}
	}


	/-******************************
	* Optimal parser
	*********************************/


	static U32 ZSTD_totalLen(ZSTD_optimal_t sol)
	{
	return sol.litlen + sol.mlen;
	}

	#if 0 /* debug */

	static void
	listStats(const U32* table, int lastEltID)
	{
	int const nbElts = lastEltID + 1;
	int enb;
	for (enb=0; enb < nbElts; enb++) {
	(void)table;
	/* RAWLOG(2, "%3i:%3i, ", enb, table[enb]); */
	RAWLOG(2, "%4i,", table[enb]);
	}
	RAWLOG(2, " \n");
	}

	#endif

	FORCE_INLINE_TEMPLATE size_t
	ZSTD_compressBlock_opt_generic(ZSTD_matchState_t* ms,
	seqStore_t* seqStore,
	U32 rep[ZSTD_REP_NUM],
	const void* src, size_t srcSize,
	const int optLevel,
	const ZSTD_dictMode_e dictMode)
	{
	optState_t* const optStatePtr = &ms->opt;
	const BYTE* const istart = (const BYTE*)src;
	const BYTE* ip = istart;
	const BYTE* anchor = istart;
	const BYTE* const iend = istart + srcSize;
	const BYTE* const ilimit = iend - 8;
	const BYTE* const base = ms->window.base;
	const BYTE* const prefixStart = base + ms->window.dictLimit;
	const ZSTD_compressionParameters* const cParams = &ms->cParams;

	U32 const sufficient_len = MIN(cParams->targetLength, ZSTD_OPT_NUM -1);
	U32 const minMatch = (cParams->minMatch == 3) ? 3 : 4;
	U32 nextToUpdate3 = ms->nextToUpdate;

	ZSTD_optimal_t* const opt = optStatePtr->priceTable;
	ZSTD_match_t* const matches = optStatePtr->matchTable;
	ZSTD_optimal_t lastSequence;

	/* init */
	DEBUGLOG(5, "ZSTD_compressBlock_opt_generic: current=%u, prefix=%u, nextToUpdate=%u",
	(U32)(ip - base), ms->window.dictLimit, ms->nextToUpdate);
	assert(optLevel <= 2);
	ZSTD_rescaleFreqs(optStatePtr, (const BYTE*)src, srcSize, optLevel);
	ip += (ip==prefixStart);

	/* Match Loop */
	while (ip < ilimit) {
	U32 cur, last_pos = 0;

	/* find first match */
	{ U32 const litlen = (U32)(ip - anchor);
	U32 const ll0 = !litlen;
	U32 const nbMatches = ZSTD_BtGetAllMatches(matches, ms, &nextToUpdate3, ip, iend, dictMode, rep, ll0, minMatch);
	if (!nbMatches) { ip++; continue; }

	/* initialize opt[0] */
	{ U32 i ; for (i=0; i<ZSTD_REP_NUM; i++) opt[0].rep[i] = rep[i]; }
	opt[0].mlen = 0; /* means is_a_literal */
	opt[0].litlen = litlen;
	/* We don't need to include the actual price of the literals because
	* it is static for the duration of the forward pass, and is included
	* in every price. We include the literal length to avoid negative
	* prices when we subtract the previous literal length.
	*/
	opt[0].price = ZSTD_litLengthPrice(litlen, optStatePtr, optLevel);

	/* large match -> immediate encoding */
	{ U32 const maxML = matches[nbMatches-1].len;
	U32 const maxOffset = matches[nbMatches-1].off;
	DEBUGLOG(6, "found %u matches of maxLength=%u and maxOffCode=%u at cPos=%u => start new series",
	nbMatches, maxML, maxOffset, (U32)(ip-prefixStart));

	if (maxML > sufficient_len) {
	lastSequence.litlen = litlen;
	lastSequence.mlen = maxML;
	lastSequence.off = maxOffset;
	DEBUGLOG(6, "large match (%u>%u), immediate encoding",
	maxML, sufficient_len);
	cur = 0;
	last_pos = ZSTD_totalLen(lastSequence);
	goto _shortestPath;
	} }

	/* set prices for first matches starting position == 0 */
	{ U32 const literalsPrice = opt[0].price + ZSTD_litLengthPrice(0, optStatePtr, optLevel);
	U32 pos;
	U32 matchNb;
	for (pos = 1; pos < minMatch; pos++) {
	opt[pos].price = ZSTD_MAX_PRICE; /* mlen, litlen and price will be fixed during forward scanning */
	}
	for (matchNb = 0; matchNb < nbMatches; matchNb++) {
	U32 const offset = matches[matchNb].off;
	U32 const end = matches[matchNb].len;
	for ( ; pos <= end ; pos++ ) {
	U32 const matchPrice = ZSTD_getMatchPrice(offset, pos, optStatePtr, optLevel);
	U32 const sequencePrice = literalsPrice + matchPrice;
	DEBUGLOG(7, "rPos:%u => set initial price : %.2f",
	pos, ZSTD_fCost(sequencePrice));
	opt[pos].mlen = pos;
	opt[pos].off = offset;
	opt[pos].litlen = litlen;
	opt[pos].price = sequencePrice;
	} }
	last_pos = pos-1;
	}
	}

	/* check further positions */
	for (cur = 1; cur <= last_pos; cur++) {
	const BYTE* const inr = ip + cur;
	assert(cur < ZSTD_OPT_NUM);
	DEBUGLOG(7, "cPos:%zi==rPos:%u", inr-istart, cur)

	/* Fix current position with one literal if cheaper */
	{ U32 const litlen = (opt[cur-1].mlen == 0) ? opt[cur-1].litlen + 1 : 1;
	int const price = opt[cur-1].price
	+ ZSTD_rawLiteralsCost(ip+cur-1, 1, optStatePtr, optLevel)
	+ ZSTD_litLengthPrice(litlen, optStatePtr, optLevel)
	- ZSTD_litLengthPrice(litlen-1, optStatePtr, optLevel);
	assert(price < 1000000000); /* overflow check */
	if (price <= opt[cur].price) {
	DEBUGLOG(7, "cPos:%zi==rPos:%u : better price (%.2f<=%.2f) using literal (ll==%u) (hist:%u,%u,%u)",
	inr-istart, cur, ZSTD_fCost(price), ZSTD_fCost(opt[cur].price), litlen,
	opt[cur-1].rep[0], opt[cur-1].rep[1], opt[cur-1].rep[2]);
	opt[cur].mlen = 0;
	opt[cur].off = 0;
	opt[cur].litlen = litlen;
	opt[cur].price = price;
	} else {
	DEBUGLOG(7, "cPos:%zi==rPos:%u : literal would cost more (%.2f>%.2f) (hist:%u,%u,%u)",
	inr-istart, cur, ZSTD_fCost(price), ZSTD_fCost(opt[cur].price),
	opt[cur].rep[0], opt[cur].rep[1], opt[cur].rep[2]);
	}
	}

	/* Set the repcodes of the current position. We must do it here
	* because we rely on the repcodes of the 2nd to last sequence being
	* correct to set the next chunks repcodes during the backward
	* traversal.
	*/
	ZSTD_STATIC_ASSERT(sizeof(opt[cur].rep) == sizeof(repcodes_t));
	assert(cur >= opt[cur].mlen);
	if (opt[cur].mlen != 0) {
	U32 const prev = cur - opt[cur].mlen;
	repcodes_t newReps = ZSTD_updateRep(opt[prev].rep, opt[cur].off, opt[cur].litlen==0);
	memcpy(opt[cur].rep, &newReps, sizeof(repcodes_t));
	} else {
	memcpy(opt[cur].rep, opt[cur - 1].rep, sizeof(repcodes_t));
	}

	/* last match must start at a minimum distance of 8 from oend */
	if (inr > ilimit) continue;

	if (cur == last_pos) break;

	if ( (optLevel==0) /static_test/
	&& (opt[cur+1].price <= opt[cur].price + (BITCOST_MULTIPLIER/2)) ) {
	DEBUGLOG(7, "move to next rPos:%u : price is <=", cur+1);
	continue; /* skip unpromising positions; about ~+6% speed, -0.01 ratio */
	}

	{ U32 const ll0 = (opt[cur].mlen != 0);
	U32 const litlen = (opt[cur].mlen == 0) ? opt[cur].litlen : 0;
	U32 const previousPrice = opt[cur].price;
	U32 const basePrice = previousPrice + ZSTD_litLengthPrice(0, optStatePtr, optLevel);
	U32 const nbMatches = ZSTD_BtGetAllMatches(matches, ms, &nextToUpdate3, inr, iend, dictMode, opt[cur].rep, ll0, minMatch);
	U32 matchNb;
	if (!nbMatches) {
	DEBUGLOG(7, "rPos:%u : no match found", cur);
	continue;
	}

	{ U32 const maxML = matches[nbMatches-1].len;
	DEBUGLOG(7, "cPos:%zi==rPos:%u, found %u matches, of maxLength=%u",
	inr-istart, cur, nbMatches, maxML);

	if ( (maxML > sufficient_len)
	\|\| (cur + maxML >= ZSTD_OPT_NUM) ) {
	lastSequence.mlen = maxML;
	lastSequence.off = matches[nbMatches-1].off;
	lastSequence.litlen = litlen;
	cur -= (opt[cur].mlen==0) ? opt[cur].litlen : 0; /* last sequence is actually only literals, fix cur to last match - note : may underflow, in which case, it's first sequence, and it's okay */
	last_pos = cur + ZSTD_totalLen(lastSequence);
	if (cur > ZSTD_OPT_NUM) cur = 0; /* underflow => first match */
	goto _shortestPath;
	} }

	/* set prices using matches found at position == cur */
	for (matchNb = 0; matchNb < nbMatches; matchNb++) {
	U32 const offset = matches[matchNb].off;
	U32 const lastML = matches[matchNb].len;
	U32 const startML = (matchNb>0) ? matches[matchNb-1].len+1 : minMatch;
	U32 mlen;

	DEBUGLOG(7, "testing match %u => offCode=%4u, mlen=%2u, llen=%2u",
	matchNb, matches[matchNb].off, lastML, litlen);

	for (mlen = lastML; mlen >= startML; mlen--) { /* scan downward */
	U32 const pos = cur + mlen;
	int const price = basePrice + ZSTD_getMatchPrice(offset, mlen, optStatePtr, optLevel);

	if ((pos > last_pos) \|\| (price < opt[pos].price)) {
	DEBUGLOG(7, "rPos:%u (ml=%2u) => new better price (%.2f<%.2f)",
	pos, mlen, ZSTD_fCost(price), ZSTD_fCost(opt[pos].price));
	while (last_pos < pos) { opt[last_pos+1].price = ZSTD_MAX_PRICE; last_pos++; } /* fill empty positions */
	opt[pos].mlen = mlen;
	opt[pos].off = offset;
	opt[pos].litlen = litlen;
	opt[pos].price = price;
	} else {
	DEBUGLOG(7, "rPos:%u (ml=%2u) => new price is worse (%.2f>=%.2f)",
	pos, mlen, ZSTD_fCost(price), ZSTD_fCost(opt[pos].price));
	if (optLevel==0) break; /* early update abort; gets ~+10% speed for about -0.01 ratio loss */
	}
	} } }
	} /* for (cur = 1; cur <= last_pos; cur++) */

	lastSequence = opt[last_pos];
	cur = last_pos > ZSTD_totalLen(lastSequence) ? last_pos - ZSTD_totalLen(lastSequence) : 0; /* single sequence, and it starts before `ip` */
	assert(cur < ZSTD_OPT_NUM); /* control overflow*/

	_shortestPath: /* cur, last_pos, best_mlen, best_off have to be set */
	assert(opt[0].mlen == 0);

	/* Set the next chunk's repcodes based on the repcodes of the beginning
	* of the last match, and the last sequence. This avoids us having to
	* update them while traversing the sequences.
	*/
	if (lastSequence.mlen != 0) {
	repcodes_t reps = ZSTD_updateRep(opt[cur].rep, lastSequence.off, lastSequence.litlen==0);
	memcpy(rep, &reps, sizeof(reps));
	} else {
	memcpy(rep, opt[cur].rep, sizeof(repcodes_t));
	}

	{ U32 const storeEnd = cur + 1;
	U32 storeStart = storeEnd;
	U32 seqPos = cur;

	DEBUGLOG(6, "start reverse traversal (last_pos:%u, cur:%u)",
	last_pos, cur); (void)last_pos;
	assert(storeEnd < ZSTD_OPT_NUM);
	DEBUGLOG(6, "last sequence copied into pos=%u (llen=%u,mlen=%u,ofc=%u)",
	storeEnd, lastSequence.litlen, lastSequence.mlen, lastSequence.off);
	opt[storeEnd] = lastSequence;
	while (seqPos > 0) {
	U32 const backDist = ZSTD_totalLen(opt[seqPos]);
	storeStart--;
	DEBUGLOG(6, "sequence from rPos=%u copied into pos=%u (llen=%u,mlen=%u,ofc=%u)",
	seqPos, storeStart, opt[seqPos].litlen, opt[seqPos].mlen, opt[seqPos].off);
	opt[storeStart] = opt[seqPos];
	seqPos = (seqPos > backDist) ? seqPos - backDist : 0;
	}

	/* save sequences */
	DEBUGLOG(6, "sending selected sequences into seqStore")
	{ U32 storePos;
	for (storePos=storeStart; storePos <= storeEnd; storePos++) {
	U32 const llen = opt[storePos].litlen;
	U32 const mlen = opt[storePos].mlen;
	U32 const offCode = opt[storePos].off;
	U32 const advance = llen + mlen;
	DEBUGLOG(6, "considering seq starting at %zi, llen=%u, mlen=%u",
	anchor - istart, (unsigned)llen, (unsigned)mlen);

	if (mlen==0) { /* only literals => must be last "sequence", actually starting a new stream of sequences */
	assert(storePos == storeEnd); /* must be last sequence */
	ip = anchor + llen; /* last "sequence" is a bunch of literals => don't progress anchor */
	continue; /* will finish */
	}

	assert(anchor + llen <= iend);
	ZSTD_updateStats(optStatePtr, llen, anchor, offCode, mlen);
	ZSTD_storeSeq(seqStore, llen, anchor, iend, offCode, mlen-MINMATCH);
	anchor += advance;
	ip = anchor;
	} }
	ZSTD_setBasePrices(optStatePtr, optLevel);
	}
	} /* while (ip < ilimit) */

	/* Return the last literals size */
	return (size_t)(iend - anchor);
	}


	size_t ZSTD_compressBlock_btopt(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	const void* src, size_t srcSize)
	{
	DEBUGLOG(5, "ZSTD_compressBlock_btopt");
	return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 0 /optLevel/, ZSTD_noDict);
	}


	/* used in 2-pass strategy */
	static U32 ZSTD_upscaleStat(unsigned* table, U32 lastEltIndex, int bonus)
	{
	U32 s, sum=0;
	assert(ZSTD_FREQ_DIV+bonus >= 0);
	for (s=0; s<lastEltIndex+1; s++) {
	table[s] <<= ZSTD_FREQ_DIV+bonus;
	table[s]--;
	sum += table[s];
	}
	return sum;
	}

	/* used in 2-pass strategy */
	MEM_STATIC void ZSTD_upscaleStats(optState_t* optPtr)
	{
	if (ZSTD_compressedLiterals(optPtr))
	optPtr->litSum = ZSTD_upscaleStat(optPtr->litFreq, MaxLit, 0);
	optPtr->litLengthSum = ZSTD_upscaleStat(optPtr->litLengthFreq, MaxLL, 0);
	optPtr->matchLengthSum = ZSTD_upscaleStat(optPtr->matchLengthFreq, MaxML, 0);
	optPtr->offCodeSum = ZSTD_upscaleStat(optPtr->offCodeFreq, MaxOff, 0);
	}

	/* ZSTD_initStats_ultra():
	* make a first compression pass, just to seed stats with more accurate starting values.
	* only works on first block, with no dictionary and no ldm.
	* this function cannot error, hence its contract must be respected.
	*/
	static void
	ZSTD_initStats_ultra(ZSTD_matchState_t* ms,
	seqStore_t* seqStore,
	U32 rep[ZSTD_REP_NUM],
	const void* src, size_t srcSize)
	{
	U32 tmpRep[ZSTD_REP_NUM]; /* updated rep codes will sink here */
	memcpy(tmpRep, rep, sizeof(tmpRep));

	DEBUGLOG(4, "ZSTD_initStats_ultra (srcSize=%zu)", srcSize);
	assert(ms->opt.litLengthSum == 0); /* first block */
	assert(seqStore->sequences == seqStore->sequencesStart); /* no ldm */
	assert(ms->window.dictLimit == ms->window.lowLimit); /* no dictionary */
	assert(ms->window.dictLimit - ms->nextToUpdate <= 1); /* no prefix (note: intentional overflow, defined as 2-complement) */

	ZSTD_compressBlock_opt_generic(ms, seqStore, tmpRep, src, srcSize, 2 /optLevel/, ZSTD_noDict); /* generate stats into ms->opt*/

	/* invalidate first scan from history */
	ZSTD_resetSeqStore(seqStore);
	ms->window.base -= srcSize;
	ms->window.dictLimit += (U32)srcSize;
	ms->window.lowLimit = ms->window.dictLimit;
	ms->nextToUpdate = ms->window.dictLimit;

	/* re-inforce weight of collected statistics */
	ZSTD_upscaleStats(&ms->opt);
	}

	size_t ZSTD_compressBlock_btultra(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	const void* src, size_t srcSize)
	{
	DEBUGLOG(5, "ZSTD_compressBlock_btultra (srcSize=%zu)", srcSize);
	return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /optLevel/, ZSTD_noDict);
	}

	size_t ZSTD_compressBlock_btultra2(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	const void* src, size_t srcSize)
	{
	U32 const current = (U32)((const BYTE*)src - ms->window.base);
	DEBUGLOG(5, "ZSTD_compressBlock_btultra2 (srcSize=%zu)", srcSize);

	/* 2-pass strategy:
	* this strategy makes a first pass over first block to collect statistics
	* and seed next round's statistics with it.
	* After 1st pass, function forgets everything, and starts a new block.
	* Consequently, this can only work if no data has been previously loaded in tables,
	* aka, no dictionary, no prefix, no ldm preprocessing.
	* The compression ratio gain is generally small (~0.5% on first block),
	* the cost is 2x cpu time on first block. */
	assert(srcSize <= ZSTD_BLOCKSIZE_MAX);
	if ( (ms->opt.litLengthSum==0) /* first block */
	&& (seqStore->sequences == seqStore->sequencesStart) /* no ldm */
	&& (ms->window.dictLimit == ms->window.lowLimit) /* no dictionary */
	&& (current == ms->window.dictLimit) /* start of frame, nothing already loaded nor skipped */
	&& (srcSize > ZSTD_PREDEF_THRESHOLD)
	) {
	ZSTD_initStats_ultra(ms, seqStore, rep, src, srcSize);
	}

	return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /optLevel/, ZSTD_noDict);
	}

	size_t ZSTD_compressBlock_btopt_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	const void* src, size_t srcSize)
	{
	return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 0 /optLevel/, ZSTD_dictMatchState);
	}

	size_t ZSTD_compressBlock_btultra_dictMatchState(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	const void* src, size_t srcSize)
	{
	return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /optLevel/, ZSTD_dictMatchState);
	}

	size_t ZSTD_compressBlock_btopt_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	const void* src, size_t srcSize)
	{
	return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 0 /optLevel/, ZSTD_extDict);
	}

	size_t ZSTD_compressBlock_btultra_extDict(
	ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
	const void* src, size_t srcSize)
	{
	return ZSTD_compressBlock_opt_generic(ms, seqStore, rep, src, srcSize, 2 /optLevel/, ZSTD_extDict);
	}

	/* note : no btultra2 variant for extDict nor dictMatchState,
	* because btultra2 is not meant to work with dictionaries
	* and is only specific for the first block (no prefix) */
	/** ended inlining compress/zstd_opt.c **/

	/** start inlining decompress/huf_decompress.c **/
	/* ******************************************************************
	* huff0 huffman decoder,
	* part of Finite State Entropy library
	* Copyright (c) 2013-2020, Yann Collet, Facebook, Inc.
	*
	* You can contact the author at :
	* - FSE+HUF source repository : https://github.com/Cyan4973/FiniteStateEntropy
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	****************************************************************** */

	/* **************************************************************
	* Dependencies
	****************************************************************/
	#include <string.h> /* memcpy, memset */
	/** skipping file: ../common/compiler.h **/
	/** skipping file: ../common/bitstream.h **/
	/** skipping file: ../common/fse.h **/
	#define HUF_STATIC_LINKING_ONLY
	/** skipping file: ../common/huf.h **/
	/** skipping file: ../common/error_private.h **/

	/* **************************************************************
	* Macros
	****************************************************************/

	/* These two optional macros force the use one way or another of the two
	* Huffman decompression implementations. You can't force in both directions
	* at the same time.
	*/
	#if defined(HUF_FORCE_DECOMPRESS_X1) && \
	defined(HUF_FORCE_DECOMPRESS_X2)
	#error "Cannot force the use of the X1 and X2 decoders at the same time!"
	#endif


	/* **************************************************************
	* Error Management
	****************************************************************/
	#define HUF_isError ERR_isError


	/* **************************************************************
	* Byte alignment for workSpace management
	****************************************************************/
	#define HUF_ALIGN(x, a) HUF_ALIGN_MASK((x), (a) - 1)
	#define HUF_ALIGN_MASK(x, mask) (((x) + (mask)) & ~(mask))


	/* **************************************************************
	* BMI2 Variant Wrappers
	****************************************************************/
	#if DYNAMIC_BMI2

	#define HUF_DGEN(fn) \
	\
	static size_t fn##_default( \
	void* dst, size_t dstSize, \
	const void* cSrc, size_t cSrcSize, \
	const HUF_DTable* DTable) \
	{ \
	return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable); \
	} \
	\
	static TARGET_ATTRIBUTE("bmi2") size_t fn##_bmi2( \
	void* dst, size_t dstSize, \
	const void* cSrc, size_t cSrcSize, \
	const HUF_DTable* DTable) \
	{ \
	return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable); \
	} \
	\
	static size_t fn(void* dst, size_t dstSize, void const* cSrc, \
	size_t cSrcSize, HUF_DTable const* DTable, int bmi2) \
	{ \
	if (bmi2) { \
	return fn##_bmi2(dst, dstSize, cSrc, cSrcSize, DTable); \
	} \
	return fn##_default(dst, dstSize, cSrc, cSrcSize, DTable); \
	}

	#else

	#define HUF_DGEN(fn) \
	static size_t fn(void* dst, size_t dstSize, void const* cSrc, \
	size_t cSrcSize, HUF_DTable const* DTable, int bmi2) \
	{ \
	(void)bmi2; \
	return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable); \
	}

	#endif


	/-**************************/
	/* generic DTableDesc */
	/-**************************/
	typedef struct { BYTE maxTableLog; BYTE tableType; BYTE tableLog; BYTE reserved; } DTableDesc;

	static DTableDesc HUF_getDTableDesc(const HUF_DTable* table)
	{
	DTableDesc dtd;
	memcpy(&dtd, table, sizeof(dtd));
	return dtd;
	}


	#ifndef HUF_FORCE_DECOMPRESS_X2

	/-**************************/
	/* single-symbol decoding */
	/-**************************/
	typedef struct { BYTE byte; BYTE nbBits; } HUF_DEltX1; /* single-symbol decoding */

	size_t HUF_readDTableX1_wksp(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize)
	{
	U32 tableLog = 0;
	U32 nbSymbols = 0;
	size_t iSize;
	void* const dtPtr = DTable + 1;
	HUF_DEltX1* const dt = (HUF_DEltX1*)dtPtr;

	U32* rankVal;
	BYTE* huffWeight;
	size_t spaceUsed32 = 0;

	rankVal = (U32 *)workSpace + spaceUsed32;
	spaceUsed32 += HUF_TABLELOG_ABSOLUTEMAX + 1;
	huffWeight = (BYTE )((U32 )workSpace + spaceUsed32);
	spaceUsed32 += HUF_ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;

	if ((spaceUsed32 << 2) > wkspSize) return ERROR(tableLog_tooLarge);

	DEBUG_STATIC_ASSERT(sizeof(DTableDesc) == sizeof(HUF_DTable));
	/* memset(huffWeight, 0, sizeof(huffWeight)); / / is not necessary, even though some analyzer complain ... */

	iSize = HUF_readStats(huffWeight, HUF_SYMBOLVALUE_MAX + 1, rankVal, &nbSymbols, &tableLog, src, srcSize);
	if (HUF_isError(iSize)) return iSize;

	/* Table header */
	{ DTableDesc dtd = HUF_getDTableDesc(DTable);
	if (tableLog > (U32)(dtd.maxTableLog+1)) return ERROR(tableLog_tooLarge); /* DTable too small, Huffman tree cannot fit in */
	dtd.tableType = 0;
	dtd.tableLog = (BYTE)tableLog;
	memcpy(DTable, &dtd, sizeof(dtd));
	}

	/* Calculate starting value for each rank */
	{ U32 n, nextRankStart = 0;
	for (n=1; n<tableLog+1; n++) {
	U32 const current = nextRankStart;
	nextRankStart += (rankVal[n] << (n-1));
	rankVal[n] = current;
	} }

	/* fill DTable */
	{ U32 n;
	size_t const nEnd = nbSymbols;
	for (n=0; n<nEnd; n++) {
	size_t const w = huffWeight[n];
	size_t const length = (1 << w) >> 1;
	size_t const uStart = rankVal[w];
	size_t const uEnd = uStart + length;
	size_t u;
	HUF_DEltX1 D;
	D.byte = (BYTE)n;
	D.nbBits = (BYTE)(tableLog + 1 - w);
	rankVal[w] = (U32)uEnd;
	if (length < 4) {
	/* Use length in the loop bound so the compiler knows it is short. */
	for (u = 0; u < length; ++u)
	dt[uStart + u] = D;
	} else {
	/* Unroll the loop 4 times, we know it is a power of 2. */
	for (u = uStart; u < uEnd; u += 4) {
	dt[u + 0] = D;
	dt[u + 1] = D;
	dt[u + 2] = D;
	dt[u + 3] = D;
	} } } }
	return iSize;
	}

	size_t HUF_readDTableX1(HUF_DTable* DTable, const void* src, size_t srcSize)
	{
	U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
	return HUF_readDTableX1_wksp(DTable, src, srcSize,
	workSpace, sizeof(workSpace));
	}

	FORCE_INLINE_TEMPLATE BYTE
	HUF_decodeSymbolX1(BIT_DStream_t* Dstream, const HUF_DEltX1* dt, const U32 dtLog)
	{
	size_t const val = BIT_lookBitsFast(Dstream, dtLog); /* note : dtLog >= 1 */
	BYTE const c = dt[val].byte;
	BIT_skipBits(Dstream, dt[val].nbBits);
	return c;
	}

	#define HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr) \
	*ptr++ = HUF_decodeSymbolX1(DStreamPtr, dt, dtLog)

	#define HUF_DECODE_SYMBOLX1_1(ptr, DStreamPtr) \
	if (MEM_64bits() \|\| (HUF_TABLELOG_MAX<=12)) \
	HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr)

	#define HUF_DECODE_SYMBOLX1_2(ptr, DStreamPtr) \
	if (MEM_64bits()) \
	HUF_DECODE_SYMBOLX1_0(ptr, DStreamPtr)

	HINT_INLINE size_t
	HUF_decodeStreamX1(BYTE* p, BIT_DStream_t* const bitDPtr, BYTE* const pEnd, const HUF_DEltX1* const dt, const U32 dtLog)
	{
	BYTE* const pStart = p;

	/* up to 4 symbols at a time */
	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-3)) {
	HUF_DECODE_SYMBOLX1_2(p, bitDPtr);
	HUF_DECODE_SYMBOLX1_1(p, bitDPtr);
	HUF_DECODE_SYMBOLX1_2(p, bitDPtr);
	HUF_DECODE_SYMBOLX1_0(p, bitDPtr);
	}

	/* [0-3] symbols remaining */
	if (MEM_32bits())
	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd))
	HUF_DECODE_SYMBOLX1_0(p, bitDPtr);

	/* no more data to retrieve from bitstream, no need to reload */
	while (p < pEnd)
	HUF_DECODE_SYMBOLX1_0(p, bitDPtr);

	return pEnd-pStart;
	}

	FORCE_INLINE_TEMPLATE size_t
	HUF_decompress1X1_usingDTable_internal_body(
	void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	const HUF_DTable* DTable)
	{
	BYTE* op = (BYTE*)dst;
	BYTE* const oend = op + dstSize;
	const void* dtPtr = DTable + 1;
	const HUF_DEltX1* const dt = (const HUF_DEltX1*)dtPtr;
	BIT_DStream_t bitD;
	DTableDesc const dtd = HUF_getDTableDesc(DTable);
	U32 const dtLog = dtd.tableLog;

	CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) );

	HUF_decodeStreamX1(op, &bitD, oend, dt, dtLog);

	if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);

	return dstSize;
	}

	FORCE_INLINE_TEMPLATE size_t
	HUF_decompress4X1_usingDTable_internal_body(
	void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	const HUF_DTable* DTable)
	{
	/* Check */
	if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */

	{ const BYTE* const istart = (const BYTE*) cSrc;
	BYTE* const ostart = (BYTE*) dst;
	BYTE* const oend = ostart + dstSize;
	BYTE* const olimit = oend - 3;
	const void* const dtPtr = DTable + 1;
	const HUF_DEltX1* const dt = (const HUF_DEltX1*)dtPtr;

	/* Init */
	BIT_DStream_t bitD1;
	BIT_DStream_t bitD2;
	BIT_DStream_t bitD3;
	BIT_DStream_t bitD4;
	size_t const length1 = MEM_readLE16(istart);
	size_t const length2 = MEM_readLE16(istart+2);
	size_t const length3 = MEM_readLE16(istart+4);
	size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
	const BYTE* const istart1 = istart + 6; /* jumpTable */
	const BYTE* const istart2 = istart1 + length1;
	const BYTE* const istart3 = istart2 + length2;
	const BYTE* const istart4 = istart3 + length3;
	const size_t segmentSize = (dstSize+3) / 4;
	BYTE* const opStart2 = ostart + segmentSize;
	BYTE* const opStart3 = opStart2 + segmentSize;
	BYTE* const opStart4 = opStart3 + segmentSize;
	BYTE* op1 = ostart;
	BYTE* op2 = opStart2;
	BYTE* op3 = opStart3;
	BYTE* op4 = opStart4;
	DTableDesc const dtd = HUF_getDTableDesc(DTable);
	U32 const dtLog = dtd.tableLog;
	U32 endSignal = 1;

	if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */
	CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
	CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
	CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
	CHECK_F( BIT_initDStream(&bitD4, istart4, length4) );

	/* up to 16 symbols per loop (4 symbols per stream) in 64-bit mode */
	for ( ; (endSignal) & (op4 < olimit) ; ) {
	HUF_DECODE_SYMBOLX1_2(op1, &bitD1);
	HUF_DECODE_SYMBOLX1_2(op2, &bitD2);
	HUF_DECODE_SYMBOLX1_2(op3, &bitD3);
	HUF_DECODE_SYMBOLX1_2(op4, &bitD4);
	HUF_DECODE_SYMBOLX1_1(op1, &bitD1);
	HUF_DECODE_SYMBOLX1_1(op2, &bitD2);
	HUF_DECODE_SYMBOLX1_1(op3, &bitD3);
	HUF_DECODE_SYMBOLX1_1(op4, &bitD4);
	HUF_DECODE_SYMBOLX1_2(op1, &bitD1);
	HUF_DECODE_SYMBOLX1_2(op2, &bitD2);
	HUF_DECODE_SYMBOLX1_2(op3, &bitD3);
	HUF_DECODE_SYMBOLX1_2(op4, &bitD4);
	HUF_DECODE_SYMBOLX1_0(op1, &bitD1);
	HUF_DECODE_SYMBOLX1_0(op2, &bitD2);
	HUF_DECODE_SYMBOLX1_0(op3, &bitD3);
	HUF_DECODE_SYMBOLX1_0(op4, &bitD4);
	endSignal &= BIT_reloadDStreamFast(&bitD1) == BIT_DStream_unfinished;
	endSignal &= BIT_reloadDStreamFast(&bitD2) == BIT_DStream_unfinished;
	endSignal &= BIT_reloadDStreamFast(&bitD3) == BIT_DStream_unfinished;
	endSignal &= BIT_reloadDStreamFast(&bitD4) == BIT_DStream_unfinished;
	}

	/* check corruption */
	/* note : should not be necessary : op# advance in lock step, and we control op4.
	* but curiously, binary generated by gcc 7.2 & 7.3 with -mbmi2 runs faster when >=1 test is present */
	if (op1 > opStart2) return ERROR(corruption_detected);
	if (op2 > opStart3) return ERROR(corruption_detected);
	if (op3 > opStart4) return ERROR(corruption_detected);
	/* note : op4 supposed already verified within main loop */

	/* finish bitStreams one by one */
	HUF_decodeStreamX1(op1, &bitD1, opStart2, dt, dtLog);
	HUF_decodeStreamX1(op2, &bitD2, opStart3, dt, dtLog);
	HUF_decodeStreamX1(op3, &bitD3, opStart4, dt, dtLog);
	HUF_decodeStreamX1(op4, &bitD4, oend, dt, dtLog);

	/* check */
	{ U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
	if (!endCheck) return ERROR(corruption_detected); }

	/* decoded size */
	return dstSize;
	}
	}


	typedef size_t (HUF_decompress_usingDTable_t)(void dst, size_t dstSize,
	const void *cSrc,
	size_t cSrcSize,
	const HUF_DTable *DTable);

	HUF_DGEN(HUF_decompress1X1_usingDTable_internal)
	HUF_DGEN(HUF_decompress4X1_usingDTable_internal)



	size_t HUF_decompress1X1_usingDTable(
	void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	const HUF_DTable* DTable)
	{
	DTableDesc dtd = HUF_getDTableDesc(DTable);
	if (dtd.tableType != 0) return ERROR(GENERIC);
	return HUF_decompress1X1_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
	}

	size_t HUF_decompress1X1_DCtx_wksp(HUF_DTable* DCtx, void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	void* workSpace, size_t wkspSize)
	{
	const BYTE* ip = (const BYTE*) cSrc;

	size_t const hSize = HUF_readDTableX1_wksp(DCtx, cSrc, cSrcSize, workSpace, wkspSize);
	if (HUF_isError(hSize)) return hSize;
	if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
	ip += hSize; cSrcSize -= hSize;

	return HUF_decompress1X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0);
	}


	size_t HUF_decompress1X1_DCtx(HUF_DTable* DCtx, void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize)
	{
	U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
	return HUF_decompress1X1_DCtx_wksp(DCtx, dst, dstSize, cSrc, cSrcSize,
	workSpace, sizeof(workSpace));
	}

	size_t HUF_decompress1X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
	{
	HUF_CREATE_STATIC_DTABLEX1(DTable, HUF_TABLELOG_MAX);
	return HUF_decompress1X1_DCtx (DTable, dst, dstSize, cSrc, cSrcSize);
	}

	size_t HUF_decompress4X1_usingDTable(
	void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	const HUF_DTable* DTable)
	{
	DTableDesc dtd = HUF_getDTableDesc(DTable);
	if (dtd.tableType != 0) return ERROR(GENERIC);
	return HUF_decompress4X1_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
	}

	static size_t HUF_decompress4X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	void* workSpace, size_t wkspSize, int bmi2)
	{
	const BYTE* ip = (const BYTE*) cSrc;

	size_t const hSize = HUF_readDTableX1_wksp (dctx, cSrc, cSrcSize,
	workSpace, wkspSize);
	if (HUF_isError(hSize)) return hSize;
	if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
	ip += hSize; cSrcSize -= hSize;

	return HUF_decompress4X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
	}

	size_t HUF_decompress4X1_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	void* workSpace, size_t wkspSize)
	{
	return HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, 0);
	}


	size_t HUF_decompress4X1_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
	{
	U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
	return HUF_decompress4X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize,
	workSpace, sizeof(workSpace));
	}
	size_t HUF_decompress4X1 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
	{
	HUF_CREATE_STATIC_DTABLEX1(DTable, HUF_TABLELOG_MAX);
	return HUF_decompress4X1_DCtx(DTable, dst, dstSize, cSrc, cSrcSize);
	}

	#endif /* HUF_FORCE_DECOMPRESS_X2 */


	#ifndef HUF_FORCE_DECOMPRESS_X1

	/* *************************/
	/* double-symbols decoding */
	/* *************************/

	typedef struct { U16 sequence; BYTE nbBits; BYTE length; } HUF_DEltX2; /* double-symbols decoding */
	typedef struct { BYTE symbol; BYTE weight; } sortedSymbol_t;
	typedef U32 rankValCol_t[HUF_TABLELOG_MAX + 1];
	typedef rankValCol_t rankVal_t[HUF_TABLELOG_MAX];


	/* HUF_fillDTableX2Level2() :
	* `rankValOrigin` must be a table of at least (HUF_TABLELOG_MAX + 1) U32 */
	static void HUF_fillDTableX2Level2(HUF_DEltX2* DTable, U32 sizeLog, const U32 consumed,
	const U32* rankValOrigin, const int minWeight,
	const sortedSymbol_t* sortedSymbols, const U32 sortedListSize,
	U32 nbBitsBaseline, U16 baseSeq)
	{
	HUF_DEltX2 DElt;
	U32 rankVal[HUF_TABLELOG_MAX + 1];

	/* get pre-calculated rankVal */
	memcpy(rankVal, rankValOrigin, sizeof(rankVal));

	/* fill skipped values */
	if (minWeight>1) {
	U32 i, skipSize = rankVal[minWeight];
	MEM_writeLE16(&(DElt.sequence), baseSeq);
	DElt.nbBits = (BYTE)(consumed);
	DElt.length = 1;
	for (i = 0; i < skipSize; i++)
	DTable[i] = DElt;
	}

	/* fill DTable */
	{ U32 s; for (s=0; s<sortedListSize; s++) { /* note : sortedSymbols already skipped */
	const U32 symbol = sortedSymbols[s].symbol;
	const U32 weight = sortedSymbols[s].weight;
	const U32 nbBits = nbBitsBaseline - weight;
	const U32 length = 1 << (sizeLog-nbBits);
	const U32 start = rankVal[weight];
	U32 i = start;
	const U32 end = start + length;

	MEM_writeLE16(&(DElt.sequence), (U16)(baseSeq + (symbol << 8)));
	DElt.nbBits = (BYTE)(nbBits + consumed);
	DElt.length = 2;
	do { DTable[i++] = DElt; } while (i<end); /* since length >= 1 */

	rankVal[weight] += length;
	} }
	}


	static void HUF_fillDTableX2(HUF_DEltX2* DTable, const U32 targetLog,
	const sortedSymbol_t* sortedList, const U32 sortedListSize,
	const U32* rankStart, rankVal_t rankValOrigin, const U32 maxWeight,
	const U32 nbBitsBaseline)
	{
	U32 rankVal[HUF_TABLELOG_MAX + 1];
	const int scaleLog = nbBitsBaseline - targetLog; /* note : targetLog >= srcLog, hence scaleLog <= 1 */
	const U32 minBits = nbBitsBaseline - maxWeight;
	U32 s;

	memcpy(rankVal, rankValOrigin, sizeof(rankVal));

	/* fill DTable */
	for (s=0; s<sortedListSize; s++) {
	const U16 symbol = sortedList[s].symbol;
	const U32 weight = sortedList[s].weight;
	const U32 nbBits = nbBitsBaseline - weight;
	const U32 start = rankVal[weight];
	const U32 length = 1 << (targetLog-nbBits);

	if (targetLog-nbBits >= minBits) { /* enough room for a second symbol */
	U32 sortedRank;
	int minWeight = nbBits + scaleLog;
	if (minWeight < 1) minWeight = 1;
	sortedRank = rankStart[minWeight];
	HUF_fillDTableX2Level2(DTable+start, targetLog-nbBits, nbBits,
	rankValOrigin[nbBits], minWeight,
	sortedList+sortedRank, sortedListSize-sortedRank,
	nbBitsBaseline, symbol);
	} else {
	HUF_DEltX2 DElt;
	MEM_writeLE16(&(DElt.sequence), symbol);
	DElt.nbBits = (BYTE)(nbBits);
	DElt.length = 1;
	{ U32 const end = start + length;
	U32 u;
	for (u = start; u < end; u++) DTable[u] = DElt;
	} }
	rankVal[weight] += length;
	}
	}

	size_t HUF_readDTableX2_wksp(HUF_DTable* DTable,
	const void* src, size_t srcSize,
	void* workSpace, size_t wkspSize)
	{
	U32 tableLog, maxW, sizeOfSort, nbSymbols;
	DTableDesc dtd = HUF_getDTableDesc(DTable);
	U32 const maxTableLog = dtd.maxTableLog;
	size_t iSize;
	void* dtPtr = DTable+1; /* force compiler to avoid strict-aliasing */
	HUF_DEltX2* const dt = (HUF_DEltX2*)dtPtr;
	U32 *rankStart;

	rankValCol_t* rankVal;
	U32* rankStats;
	U32* rankStart0;
	sortedSymbol_t* sortedSymbol;
	BYTE* weightList;
	size_t spaceUsed32 = 0;

	rankVal = (rankValCol_t )((U32 )workSpace + spaceUsed32);
	spaceUsed32 += (sizeof(rankValCol_t) * HUF_TABLELOG_MAX) >> 2;
	rankStats = (U32 *)workSpace + spaceUsed32;
	spaceUsed32 += HUF_TABLELOG_MAX + 1;
	rankStart0 = (U32 *)workSpace + spaceUsed32;
	spaceUsed32 += HUF_TABLELOG_MAX + 2;
	sortedSymbol = (sortedSymbol_t )workSpace + (spaceUsed32 sizeof(U32)) / sizeof(sortedSymbol_t);
	spaceUsed32 += HUF_ALIGN(sizeof(sortedSymbol_t) * (HUF_SYMBOLVALUE_MAX + 1), sizeof(U32)) >> 2;
	weightList = (BYTE )((U32 )workSpace + spaceUsed32);
	spaceUsed32 += HUF_ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;

	if ((spaceUsed32 << 2) > wkspSize) return ERROR(tableLog_tooLarge);

	rankStart = rankStart0 + 1;
	memset(rankStats, 0, sizeof(U32) * (2 * HUF_TABLELOG_MAX + 2 + 1));

	DEBUG_STATIC_ASSERT(sizeof(HUF_DEltX2) == sizeof(HUF_DTable)); /* if compiler fails here, assertion is wrong */
	if (maxTableLog > HUF_TABLELOG_MAX) return ERROR(tableLog_tooLarge);
	/* memset(weightList, 0, sizeof(weightList)); / / is not necessary, even though some analyzer complain ... */

	iSize = HUF_readStats(weightList, HUF_SYMBOLVALUE_MAX + 1, rankStats, &nbSymbols, &tableLog, src, srcSize);
	if (HUF_isError(iSize)) return iSize;

	/* check result */
	if (tableLog > maxTableLog) return ERROR(tableLog_tooLarge); /* DTable can't fit code depth */

	/* find maxWeight */
	for (maxW = tableLog; rankStats[maxW]==0; maxW--) {} /* necessarily finds a solution before 0 */

	/* Get start index of each weight */
	{ U32 w, nextRankStart = 0;
	for (w=1; w<maxW+1; w++) {
	U32 current = nextRankStart;
	nextRankStart += rankStats[w];
	rankStart[w] = current;
	}
	rankStart[0] = nextRankStart; /* put all 0w symbols at the end of sorted list*/
	sizeOfSort = nextRankStart;
	}

	/* sort symbols by weight */
	{ U32 s;
	for (s=0; s<nbSymbols; s++) {
	U32 const w = weightList[s];
	U32 const r = rankStart[w]++;
	sortedSymbol[r].symbol = (BYTE)s;
	sortedSymbol[r].weight = (BYTE)w;
	}
	rankStart[0] = 0; /* forget 0w symbols; this is beginning of weight(1) */
	}

	/* Build rankVal */
	{ U32* const rankVal0 = rankVal[0];
	{ int const rescale = (maxTableLog-tableLog) - 1; /* tableLog <= maxTableLog */
	U32 nextRankVal = 0;
	U32 w;
	for (w=1; w<maxW+1; w++) {
	U32 current = nextRankVal;
	nextRankVal += rankStats[w] << (w+rescale);
	rankVal0[w] = current;
	} }
	{ U32 const minBits = tableLog+1 - maxW;
	U32 consumed;
	for (consumed = minBits; consumed < maxTableLog - minBits + 1; consumed++) {
	U32* const rankValPtr = rankVal[consumed];
	U32 w;
	for (w = 1; w < maxW+1; w++) {
	rankValPtr[w] = rankVal0[w] >> consumed;
	} } } }

	HUF_fillDTableX2(dt, maxTableLog,
	sortedSymbol, sizeOfSort,
	rankStart0, rankVal, maxW,
	tableLog+1);

	dtd.tableLog = (BYTE)maxTableLog;
	dtd.tableType = 1;
	memcpy(DTable, &dtd, sizeof(dtd));
	return iSize;
	}

	size_t HUF_readDTableX2(HUF_DTable* DTable, const void* src, size_t srcSize)
	{
	U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
	return HUF_readDTableX2_wksp(DTable, src, srcSize,
	workSpace, sizeof(workSpace));
	}


	FORCE_INLINE_TEMPLATE U32
	HUF_decodeSymbolX2(void* op, BIT_DStream_t* DStream, const HUF_DEltX2* dt, const U32 dtLog)
	{
	size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
	memcpy(op, dt+val, 2);
	BIT_skipBits(DStream, dt[val].nbBits);
	return dt[val].length;
	}

	FORCE_INLINE_TEMPLATE U32
	HUF_decodeLastSymbolX2(void* op, BIT_DStream_t* DStream, const HUF_DEltX2* dt, const U32 dtLog)
	{
	size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
	memcpy(op, dt+val, 1);
	if (dt[val].length==1) BIT_skipBits(DStream, dt[val].nbBits);
	else {
	if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) {
	BIT_skipBits(DStream, dt[val].nbBits);
	if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8))
	/* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */
	DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8);
	} }
	return 1;
	}

	#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) \
	ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)

	#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr) \
	if (MEM_64bits() \|\| (HUF_TABLELOG_MAX<=12)) \
	ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)

	#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \
	if (MEM_64bits()) \
	ptr += HUF_decodeSymbolX2(ptr, DStreamPtr, dt, dtLog)

	HINT_INLINE size_t
	HUF_decodeStreamX2(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* const pEnd,
	const HUF_DEltX2* const dt, const U32 dtLog)
	{
	BYTE* const pStart = p;

	/* up to 8 symbols at a time */
	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-(sizeof(bitDPtr->bitContainer)-1))) {
	HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
	HUF_DECODE_SYMBOLX2_1(p, bitDPtr);
	HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
	HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
	}

	/* closer to end : up to 2 symbols at a time */
	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p <= pEnd-2))
	HUF_DECODE_SYMBOLX2_0(p, bitDPtr);

	while (p <= pEnd-2)
	HUF_DECODE_SYMBOLX2_0(p, bitDPtr); /* no need to reload : reached the end of DStream */

	if (p < pEnd)
	p += HUF_decodeLastSymbolX2(p, bitDPtr, dt, dtLog);

	return p-pStart;
	}

	FORCE_INLINE_TEMPLATE size_t
	HUF_decompress1X2_usingDTable_internal_body(
	void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	const HUF_DTable* DTable)
	{
	BIT_DStream_t bitD;

	/* Init */
	CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) );

	/* decode */
	{ BYTE* const ostart = (BYTE*) dst;
	BYTE* const oend = ostart + dstSize;
	const void* const dtPtr = DTable+1; /* force compiler to not use strict-aliasing */
	const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
	DTableDesc const dtd = HUF_getDTableDesc(DTable);
	HUF_decodeStreamX2(ostart, &bitD, oend, dt, dtd.tableLog);
	}

	/* check */
	if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);

	/* decoded size */
	return dstSize;
	}

	FORCE_INLINE_TEMPLATE size_t
	HUF_decompress4X2_usingDTable_internal_body(
	void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	const HUF_DTable* DTable)
	{
	if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */

	{ const BYTE* const istart = (const BYTE*) cSrc;
	BYTE* const ostart = (BYTE*) dst;
	BYTE* const oend = ostart + dstSize;
	BYTE* const olimit = oend - (sizeof(size_t)-1);
	const void* const dtPtr = DTable+1;
	const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;

	/* Init */
	BIT_DStream_t bitD1;
	BIT_DStream_t bitD2;
	BIT_DStream_t bitD3;
	BIT_DStream_t bitD4;
	size_t const length1 = MEM_readLE16(istart);
	size_t const length2 = MEM_readLE16(istart+2);
	size_t const length3 = MEM_readLE16(istart+4);
	size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
	const BYTE* const istart1 = istart + 6; /* jumpTable */
	const BYTE* const istart2 = istart1 + length1;
	const BYTE* const istart3 = istart2 + length2;
	const BYTE* const istart4 = istart3 + length3;
	size_t const segmentSize = (dstSize+3) / 4;
	BYTE* const opStart2 = ostart + segmentSize;
	BYTE* const opStart3 = opStart2 + segmentSize;
	BYTE* const opStart4 = opStart3 + segmentSize;
	BYTE* op1 = ostart;
	BYTE* op2 = opStart2;
	BYTE* op3 = opStart3;
	BYTE* op4 = opStart4;
	U32 endSignal = 1;
	DTableDesc const dtd = HUF_getDTableDesc(DTable);
	U32 const dtLog = dtd.tableLog;

	if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */
	CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
	CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
	CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
	CHECK_F( BIT_initDStream(&bitD4, istart4, length4) );

	/* 16-32 symbols per loop (4-8 symbols per stream) */
	for ( ; (endSignal) & (op4 < olimit); ) {
	#if defined(__clang__) && (defined(__x86_64__) \|\| defined(__i386__))
	HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
	HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
	HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
	HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
	HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
	HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
	HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
	HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
	endSignal &= BIT_reloadDStreamFast(&bitD1) == BIT_DStream_unfinished;
	endSignal &= BIT_reloadDStreamFast(&bitD2) == BIT_DStream_unfinished;
	HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
	HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
	HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
	HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
	HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
	HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
	HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
	HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
	endSignal &= BIT_reloadDStreamFast(&bitD3) == BIT_DStream_unfinished;
	endSignal &= BIT_reloadDStreamFast(&bitD4) == BIT_DStream_unfinished;
	#else
	HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
	HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
	HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
	HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
	HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
	HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
	HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
	HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
	HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
	HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
	HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
	HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
	HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
	HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
	HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
	HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
	endSignal = (U32)LIKELY(
	(BIT_reloadDStreamFast(&bitD1) == BIT_DStream_unfinished)
	& (BIT_reloadDStreamFast(&bitD2) == BIT_DStream_unfinished)
	& (BIT_reloadDStreamFast(&bitD3) == BIT_DStream_unfinished)
	& (BIT_reloadDStreamFast(&bitD4) == BIT_DStream_unfinished));
	#endif
	}

	/* check corruption */
	if (op1 > opStart2) return ERROR(corruption_detected);
	if (op2 > opStart3) return ERROR(corruption_detected);
	if (op3 > opStart4) return ERROR(corruption_detected);
	/* note : op4 already verified within main loop */

	/* finish bitStreams one by one */
	HUF_decodeStreamX2(op1, &bitD1, opStart2, dt, dtLog);
	HUF_decodeStreamX2(op2, &bitD2, opStart3, dt, dtLog);
	HUF_decodeStreamX2(op3, &bitD3, opStart4, dt, dtLog);
	HUF_decodeStreamX2(op4, &bitD4, oend, dt, dtLog);

	/* check */
	{ U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
	if (!endCheck) return ERROR(corruption_detected); }

	/* decoded size */
	return dstSize;
	}
	}

	HUF_DGEN(HUF_decompress1X2_usingDTable_internal)
	HUF_DGEN(HUF_decompress4X2_usingDTable_internal)

	size_t HUF_decompress1X2_usingDTable(
	void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	const HUF_DTable* DTable)
	{
	DTableDesc dtd = HUF_getDTableDesc(DTable);
	if (dtd.tableType != 1) return ERROR(GENERIC);
	return HUF_decompress1X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
	}

	size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable* DCtx, void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	void* workSpace, size_t wkspSize)
	{
	const BYTE* ip = (const BYTE*) cSrc;

	size_t const hSize = HUF_readDTableX2_wksp(DCtx, cSrc, cSrcSize,
	workSpace, wkspSize);
	if (HUF_isError(hSize)) return hSize;
	if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
	ip += hSize; cSrcSize -= hSize;

	return HUF_decompress1X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx, /* bmi2 */ 0);
	}


	size_t HUF_decompress1X2_DCtx(HUF_DTable* DCtx, void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize)
	{
	U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
	return HUF_decompress1X2_DCtx_wksp(DCtx, dst, dstSize, cSrc, cSrcSize,
	workSpace, sizeof(workSpace));
	}

	size_t HUF_decompress1X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
	{
	HUF_CREATE_STATIC_DTABLEX2(DTable, HUF_TABLELOG_MAX);
	return HUF_decompress1X2_DCtx(DTable, dst, dstSize, cSrc, cSrcSize);
	}

	size_t HUF_decompress4X2_usingDTable(
	void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	const HUF_DTable* DTable)
	{
	DTableDesc dtd = HUF_getDTableDesc(DTable);
	if (dtd.tableType != 1) return ERROR(GENERIC);
	return HUF_decompress4X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
	}

	static size_t HUF_decompress4X2_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	void* workSpace, size_t wkspSize, int bmi2)
	{
	const BYTE* ip = (const BYTE*) cSrc;

	size_t hSize = HUF_readDTableX2_wksp(dctx, cSrc, cSrcSize,
	workSpace, wkspSize);
	if (HUF_isError(hSize)) return hSize;
	if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
	ip += hSize; cSrcSize -= hSize;

	return HUF_decompress4X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
	}

	size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	void* workSpace, size_t wkspSize)
	{
	return HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, /* bmi2 */ 0);
	}


	size_t HUF_decompress4X2_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize)
	{
	U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
	return HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize,
	workSpace, sizeof(workSpace));
	}

	size_t HUF_decompress4X2 (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
	{
	HUF_CREATE_STATIC_DTABLEX2(DTable, HUF_TABLELOG_MAX);
	return HUF_decompress4X2_DCtx(DTable, dst, dstSize, cSrc, cSrcSize);
	}

	#endif /* HUF_FORCE_DECOMPRESS_X1 */


	/* ***********************************/
	/* Universal decompression selectors */
	/* ***********************************/

	size_t HUF_decompress1X_usingDTable(void* dst, size_t maxDstSize,
	const void* cSrc, size_t cSrcSize,
	const HUF_DTable* DTable)
	{
	DTableDesc const dtd = HUF_getDTableDesc(DTable);
	#if defined(HUF_FORCE_DECOMPRESS_X1)
	(void)dtd;
	assert(dtd.tableType == 0);
	return HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
	#elif defined(HUF_FORCE_DECOMPRESS_X2)
	(void)dtd;
	assert(dtd.tableType == 1);
	return HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
	#else
	return dtd.tableType ? HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
	HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
	#endif
	}

	size_t HUF_decompress4X_usingDTable(void* dst, size_t maxDstSize,
	const void* cSrc, size_t cSrcSize,
	const HUF_DTable* DTable)
	{
	DTableDesc const dtd = HUF_getDTableDesc(DTable);
	#if defined(HUF_FORCE_DECOMPRESS_X1)
	(void)dtd;
	assert(dtd.tableType == 0);
	return HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
	#elif defined(HUF_FORCE_DECOMPRESS_X2)
	(void)dtd;
	assert(dtd.tableType == 1);
	return HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
	#else
	return dtd.tableType ? HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0) :
	HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, /* bmi2 */ 0);
	#endif
	}


	#if !defined(HUF_FORCE_DECOMPRESS_X1) && !defined(HUF_FORCE_DECOMPRESS_X2)
	typedef struct { U32 tableTime; U32 decode256Time; } algo_time_t;
	static const algo_time_t algoTime[16 /* Quantization /][3 / single, double, quad */] =
	{
	/* single, double, quad */
	{{0,0}, {1,1}, {2,2}}, /* Q==0 : impossible */
	{{0,0}, {1,1}, {2,2}}, /* Q==1 : impossible */
	{{ 38,130}, {1313, 74}, {2151, 38}}, /* Q == 2 : 12-18% */
	{{ 448,128}, {1353, 74}, {2238, 41}}, /* Q == 3 : 18-25% */
	{{ 556,128}, {1353, 74}, {2238, 47}}, /* Q == 4 : 25-32% */
	{{ 714,128}, {1418, 74}, {2436, 53}}, /* Q == 5 : 32-38% */
	{{ 883,128}, {1437, 74}, {2464, 61}}, /* Q == 6 : 38-44% */
	{{ 897,128}, {1515, 75}, {2622, 68}}, /* Q == 7 : 44-50% */
	{{ 926,128}, {1613, 75}, {2730, 75}}, /* Q == 8 : 50-56% */
	{{ 947,128}, {1729, 77}, {3359, 77}}, /* Q == 9 : 56-62% */
	{{1107,128}, {2083, 81}, {4006, 84}}, /* Q ==10 : 62-69% */
	{{1177,128}, {2379, 87}, {4785, 88}}, /* Q ==11 : 69-75% */
	{{1242,128}, {2415, 93}, {5155, 84}}, /* Q ==12 : 75-81% */
	{{1349,128}, {2644,106}, {5260,106}}, /* Q ==13 : 81-87% */
	{{1455,128}, {2422,124}, {4174,124}}, /* Q ==14 : 87-93% */
	{{ 722,128}, {1891,145}, {1936,146}}, /* Q ==15 : 93-99% */
	};
	#endif

	/** HUF_selectDecoder() :
	* Tells which decoder is likely to decode faster,
	* based on a set of pre-computed metrics.
	* @return : 0==HUF_decompress4X1, 1==HUF_decompress4X2 .
	* Assumption : 0 < dstSize <= 128 KB */
	U32 HUF_selectDecoder (size_t dstSize, size_t cSrcSize)
	{
	assert(dstSize > 0);
	assert(dstSize <= 128*1024);
	#if defined(HUF_FORCE_DECOMPRESS_X1)
	(void)dstSize;
	(void)cSrcSize;
	return 0;
	#elif defined(HUF_FORCE_DECOMPRESS_X2)
	(void)dstSize;
	(void)cSrcSize;
	return 1;
	#else
	/* decoder timing evaluation */
	{ U32 const Q = (cSrcSize >= dstSize) ? 15 : (U32)(cSrcSize * 16 / dstSize); /* Q < 16 */
	U32 const D256 = (U32)(dstSize >> 8);
	U32 const DTime0 = algoTime[Q][0].tableTime + (algoTime[Q][0].decode256Time * D256);
	U32 DTime1 = algoTime[Q][1].tableTime + (algoTime[Q][1].decode256Time * D256);
	DTime1 += DTime1 >> 3; /* advantage to algorithm using less memory, to reduce cache eviction */
	return DTime1 < DTime0;
	}
	#endif
	}


	typedef size_t (decompressionAlgo)(void dst, size_t dstSize, const void* cSrc, size_t cSrcSize);

	size_t HUF_decompress (void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
	{
	#if !defined(HUF_FORCE_DECOMPRESS_X1) && !defined(HUF_FORCE_DECOMPRESS_X2)
	static const decompressionAlgo decompress[2] = { HUF_decompress4X1, HUF_decompress4X2 };
	#endif

	/* validation checks */
	if (dstSize == 0) return ERROR(dstSize_tooSmall);
	if (cSrcSize > dstSize) return ERROR(corruption_detected); /* invalid */
	if (cSrcSize == dstSize) { memcpy(dst, cSrc, dstSize); return dstSize; } /* not compressed */
	if (cSrcSize == 1) { memset(dst, (const BYTE)cSrc, dstSize); return dstSize; } /* RLE */

	{ U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
	#if defined(HUF_FORCE_DECOMPRESS_X1)
	(void)algoNb;
	assert(algoNb == 0);
	return HUF_decompress4X1(dst, dstSize, cSrc, cSrcSize);
	#elif defined(HUF_FORCE_DECOMPRESS_X2)
	(void)algoNb;
	assert(algoNb == 1);
	return HUF_decompress4X2(dst, dstSize, cSrc, cSrcSize);
	#else
	return decompress[algoNb](dst, dstSize, cSrc, cSrcSize);
	#endif
	}
	}

	size_t HUF_decompress4X_DCtx (HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
	{
	/* validation checks */
	if (dstSize == 0) return ERROR(dstSize_tooSmall);
	if (cSrcSize > dstSize) return ERROR(corruption_detected); /* invalid */
	if (cSrcSize == dstSize) { memcpy(dst, cSrc, dstSize); return dstSize; } /* not compressed */
	if (cSrcSize == 1) { memset(dst, (const BYTE)cSrc, dstSize); return dstSize; } /* RLE */

	{ U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
	#if defined(HUF_FORCE_DECOMPRESS_X1)
	(void)algoNb;
	assert(algoNb == 0);
	return HUF_decompress4X1_DCtx(dctx, dst, dstSize, cSrc, cSrcSize);
	#elif defined(HUF_FORCE_DECOMPRESS_X2)
	(void)algoNb;
	assert(algoNb == 1);
	return HUF_decompress4X2_DCtx(dctx, dst, dstSize, cSrc, cSrcSize);
	#else
	return algoNb ? HUF_decompress4X2_DCtx(dctx, dst, dstSize, cSrc, cSrcSize) :
	HUF_decompress4X1_DCtx(dctx, dst, dstSize, cSrc, cSrcSize) ;
	#endif
	}
	}

	size_t HUF_decompress4X_hufOnly(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize)
	{
	U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
	return HUF_decompress4X_hufOnly_wksp(dctx, dst, dstSize, cSrc, cSrcSize,
	workSpace, sizeof(workSpace));
	}


	size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable* dctx, void* dst,
	size_t dstSize, const void* cSrc,
	size_t cSrcSize, void* workSpace,
	size_t wkspSize)
	{
	/* validation checks */
	if (dstSize == 0) return ERROR(dstSize_tooSmall);
	if (cSrcSize == 0) return ERROR(corruption_detected);

	{ U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
	#if defined(HUF_FORCE_DECOMPRESS_X1)
	(void)algoNb;
	assert(algoNb == 0);
	return HUF_decompress4X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
	#elif defined(HUF_FORCE_DECOMPRESS_X2)
	(void)algoNb;
	assert(algoNb == 1);
	return HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
	#else
	return algoNb ? HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
	cSrcSize, workSpace, wkspSize):
	HUF_decompress4X1_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize);
	#endif
	}
	}

	size_t HUF_decompress1X_DCtx_wksp(HUF_DTable* dctx, void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize,
	void* workSpace, size_t wkspSize)
	{
	/* validation checks */
	if (dstSize == 0) return ERROR(dstSize_tooSmall);
	if (cSrcSize > dstSize) return ERROR(corruption_detected); /* invalid */
	if (cSrcSize == dstSize) { memcpy(dst, cSrc, dstSize); return dstSize; } /* not compressed */
	if (cSrcSize == 1) { memset(dst, (const BYTE)cSrc, dstSize); return dstSize; } /* RLE */

	{ U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
	#if defined(HUF_FORCE_DECOMPRESS_X1)
	(void)algoNb;
	assert(algoNb == 0);
	return HUF_decompress1X1_DCtx_wksp(dctx, dst, dstSize, cSrc,
	cSrcSize, workSpace, wkspSize);
	#elif defined(HUF_FORCE_DECOMPRESS_X2)
	(void)algoNb;
	assert(algoNb == 1);
	return HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
	cSrcSize, workSpace, wkspSize);
	#else
	return algoNb ? HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc,
	cSrcSize, workSpace, wkspSize):
	HUF_decompress1X1_DCtx_wksp(dctx, dst, dstSize, cSrc,
	cSrcSize, workSpace, wkspSize);
	#endif
	}
	}

	size_t HUF_decompress1X_DCtx(HUF_DTable* dctx, void* dst, size_t dstSize,
	const void* cSrc, size_t cSrcSize)
	{
	U32 workSpace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];
	return HUF_decompress1X_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize,
	workSpace, sizeof(workSpace));
	}


	size_t HUF_decompress1X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2)
	{
	DTableDesc const dtd = HUF_getDTableDesc(DTable);
	#if defined(HUF_FORCE_DECOMPRESS_X1)
	(void)dtd;
	assert(dtd.tableType == 0);
	return HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
	#elif defined(HUF_FORCE_DECOMPRESS_X2)
	(void)dtd;
	assert(dtd.tableType == 1);
	return HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
	#else
	return dtd.tableType ? HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
	HUF_decompress1X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
	#endif
	}

	#ifndef HUF_FORCE_DECOMPRESS_X2
	size_t HUF_decompress1X1_DCtx_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2)
	{
	const BYTE* ip = (const BYTE*) cSrc;

	size_t const hSize = HUF_readDTableX1_wksp(dctx, cSrc, cSrcSize, workSpace, wkspSize);
	if (HUF_isError(hSize)) return hSize;
	if (hSize >= cSrcSize) return ERROR(srcSize_wrong);
	ip += hSize; cSrcSize -= hSize;

	return HUF_decompress1X1_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx, bmi2);
	}
	#endif

	size_t HUF_decompress4X_usingDTable_bmi2(void* dst, size_t maxDstSize, const void* cSrc, size_t cSrcSize, const HUF_DTable* DTable, int bmi2)
	{
	DTableDesc const dtd = HUF_getDTableDesc(DTable);
	#if defined(HUF_FORCE_DECOMPRESS_X1)
	(void)dtd;
	assert(dtd.tableType == 0);
	return HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
	#elif defined(HUF_FORCE_DECOMPRESS_X2)
	(void)dtd;
	assert(dtd.tableType == 1);
	return HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
	#else
	return dtd.tableType ? HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2) :
	HUF_decompress4X1_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable, bmi2);
	#endif
	}

	size_t HUF_decompress4X_hufOnly_wksp_bmi2(HUF_DTable* dctx, void* dst, size_t dstSize, const void* cSrc, size_t cSrcSize, void* workSpace, size_t wkspSize, int bmi2)
	{
	/* validation checks */
	if (dstSize == 0) return ERROR(dstSize_tooSmall);
	if (cSrcSize == 0) return ERROR(corruption_detected);

	{ U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
	#if defined(HUF_FORCE_DECOMPRESS_X1)
	(void)algoNb;
	assert(algoNb == 0);
	return HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
	#elif defined(HUF_FORCE_DECOMPRESS_X2)
	(void)algoNb;
	assert(algoNb == 1);
	return HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
	#else
	return algoNb ? HUF_decompress4X2_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2) :
	HUF_decompress4X1_DCtx_wksp_bmi2(dctx, dst, dstSize, cSrc, cSrcSize, workSpace, wkspSize, bmi2);
	#endif
	}
	}
	/** ended inlining decompress/huf_decompress.c **/
	/** start inlining decompress/zstd_ddict.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/* zstd_ddict.c :
	* concentrates all logic that needs to know the internals of ZSTD_DDict object */

	/-******************************************************
	* Dependencies
	*********************************************************/
	#include <string.h> /* memcpy, memmove, memset */
	/** skipping file: ../common/cpu.h **/
	/** skipping file: ../common/mem.h **/
	#define FSE_STATIC_LINKING_ONLY
	/** skipping file: ../common/fse.h **/
	#define HUF_STATIC_LINKING_ONLY
	/** skipping file: ../common/huf.h **/
	/** start inlining zstd_decompress_internal.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/


	/* zstd_decompress_internal:
	* objects and definitions shared within lib/decompress modules */

	#ifndef ZSTD_DECOMPRESS_INTERNAL_H
	#define ZSTD_DECOMPRESS_INTERNAL_H


	/-******************************************************
	* Dependencies
	*********************************************************/
	/** skipping file: ../common/mem.h **/
	/** skipping file: ../common/zstd_internal.h **/



	/-******************************************************
	* Constants
	*********************************************************/
	static const U32 LL_base[MaxLL+1] = {
	0, 1, 2, 3, 4, 5, 6, 7,
	8, 9, 10, 11, 12, 13, 14, 15,
	16, 18, 20, 22, 24, 28, 32, 40,
	48, 64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000,
	0x2000, 0x4000, 0x8000, 0x10000 };

	static const U32 OF_base[MaxOff+1] = {
	0, 1, 1, 5, 0xD, 0x1D, 0x3D, 0x7D,
	0xFD, 0x1FD, 0x3FD, 0x7FD, 0xFFD, 0x1FFD, 0x3FFD, 0x7FFD,
	0xFFFD, 0x1FFFD, 0x3FFFD, 0x7FFFD, 0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD,
	0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD, 0x1FFFFFFD, 0x3FFFFFFD, 0x7FFFFFFD };

	static const U32 OF_bits[MaxOff+1] = {
	0, 1, 2, 3, 4, 5, 6, 7,
	8, 9, 10, 11, 12, 13, 14, 15,
	16, 17, 18, 19, 20, 21, 22, 23,
	24, 25, 26, 27, 28, 29, 30, 31 };

	static const U32 ML_base[MaxML+1] = {
	3, 4, 5, 6, 7, 8, 9, 10,
	11, 12, 13, 14, 15, 16, 17, 18,
	19, 20, 21, 22, 23, 24, 25, 26,
	27, 28, 29, 30, 31, 32, 33, 34,
	35, 37, 39, 41, 43, 47, 51, 59,
	67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803,
	0x1003, 0x2003, 0x4003, 0x8003, 0x10003 };


	/-******************************************************
	* Decompression types
	*********************************************************/
	typedef struct {
	U32 fastMode;
	U32 tableLog;
	} ZSTD_seqSymbol_header;

	typedef struct {
	U16 nextState;
	BYTE nbAdditionalBits;
	BYTE nbBits;
	U32 baseValue;
	} ZSTD_seqSymbol;

	#define SEQSYMBOL_TABLE_SIZE(log) (1 + (1 << (log)))

	typedef struct {
	ZSTD_seqSymbol LLTable[SEQSYMBOL_TABLE_SIZE(LLFSELog)]; /* Note : Space reserved for FSE Tables */
	ZSTD_seqSymbol OFTable[SEQSYMBOL_TABLE_SIZE(OffFSELog)]; /* is also used as temporary workspace while building hufTable during DDict creation */
	ZSTD_seqSymbol MLTable[SEQSYMBOL_TABLE_SIZE(MLFSELog)]; /* and therefore must be at least HUF_DECOMPRESS_WORKSPACE_SIZE large */
	HUF_DTable hufTable[HUF_DTABLE_SIZE(HufLog)]; /* can accommodate HUF_decompress4X */
	U32 rep[ZSTD_REP_NUM];
	} ZSTD_entropyDTables_t;

	typedef enum { ZSTDds_getFrameHeaderSize, ZSTDds_decodeFrameHeader,
	ZSTDds_decodeBlockHeader, ZSTDds_decompressBlock,
	ZSTDds_decompressLastBlock, ZSTDds_checkChecksum,
	ZSTDds_decodeSkippableHeader, ZSTDds_skipFrame } ZSTD_dStage;

	typedef enum { zdss_init=0, zdss_loadHeader,
	zdss_read, zdss_load, zdss_flush } ZSTD_dStreamStage;

	typedef enum {
	ZSTD_use_indefinitely = -1, /* Use the dictionary indefinitely */
	ZSTD_dont_use = 0, /* Do not use the dictionary (if one exists free it) */
	ZSTD_use_once = 1 /* Use the dictionary once and set to ZSTD_dont_use */
	} ZSTD_dictUses_e;

	typedef enum {
	ZSTD_obm_buffered = 0, /* Buffer the output */
	ZSTD_obm_stable = 1 /* ZSTD_outBuffer is stable */
	} ZSTD_outBufferMode_e;

	struct ZSTD_DCtx_s
	{
	const ZSTD_seqSymbol* LLTptr;
	const ZSTD_seqSymbol* MLTptr;
	const ZSTD_seqSymbol* OFTptr;
	const HUF_DTable* HUFptr;
	ZSTD_entropyDTables_t entropy;
	U32 workspace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32]; /* space needed when building huffman tables */
	const void* previousDstEnd; /* detect continuity */
	const void* prefixStart; /* start of current segment */
	const void* virtualStart; /* virtual start of previous segment if it was just before current one */
	const void* dictEnd; /* end of previous segment */
	size_t expected;
	ZSTD_frameHeader fParams;
	U64 decodedSize;
	blockType_e bType; /* used in ZSTD_decompressContinue(), store blockType between block header decoding and block decompression stages */
	ZSTD_dStage stage;
	U32 litEntropy;
	U32 fseEntropy;
	XXH64_state_t xxhState;
	size_t headerSize;
	ZSTD_format_e format;
	const BYTE* litPtr;
	ZSTD_customMem customMem;
	size_t litSize;
	size_t rleSize;
	size_t staticSize;
	int bmi2; /* == 1 if the CPU supports BMI2 and 0 otherwise. CPU support is determined dynamically once per context lifetime. */

	/* dictionary */
	ZSTD_DDict* ddictLocal;
	const ZSTD_DDict* ddict; /* set by ZSTD_initDStream_usingDDict(), or ZSTD_DCtx_refDDict() */
	U32 dictID;
	int ddictIsCold; /* if == 1 : dictionary is "new" for working context, and presumed "cold" (not in cpu cache) */
	ZSTD_dictUses_e dictUses;

	/* streaming */
	ZSTD_dStreamStage streamStage;
	char* inBuff;
	size_t inBuffSize;
	size_t inPos;
	size_t maxWindowSize;
	char* outBuff;
	size_t outBuffSize;
	size_t outStart;
	size_t outEnd;
	size_t lhSize;
	void* legacyContext;
	U32 previousLegacyVersion;
	U32 legacyVersion;
	U32 hostageByte;
	int noForwardProgress;
	ZSTD_outBufferMode_e outBufferMode;
	ZSTD_outBuffer expectedOutBuffer;

	/* workspace */
	BYTE litBuffer[ZSTD_BLOCKSIZE_MAX + WILDCOPY_OVERLENGTH];
	BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX];

	size_t oversizedDuration;

	#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
	void const* dictContentBeginForFuzzing;
	void const* dictContentEndForFuzzing;
	#endif
	}; /* typedef'd to ZSTD_DCtx within "zstd.h" */


	/-******************************************************
	* Shared internal functions
	*********************************************************/

	/*! ZSTD_loadDEntropy() :
	* dict : must point at beginning of a valid zstd dictionary.
	* @return : size of dictionary header (size of magic number + dict ID + entropy tables) */
	size_t ZSTD_loadDEntropy(ZSTD_entropyDTables_t* entropy,
	const void* const dict, size_t const dictSize);

	/*! ZSTD_checkContinuity() :
	* check if next `dst` follows previous position, where decompression ended.
	* If yes, do nothing (continue on current segment).
	* If not, classify previous segment as "external dictionary", and start a new segment.
	* This function cannot fail. */
	void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst);


	#endif /* ZSTD_DECOMPRESS_INTERNAL_H */
	/** ended inlining zstd_decompress_internal.h **/
	/** start inlining zstd_ddict.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/


	#ifndef ZSTD_DDICT_H
	#define ZSTD_DDICT_H

	/-******************************************************
	* Dependencies
	*********************************************************/
	#include <stddef.h> /* size_t */
	/** skipping file: ../zstd.h **/


	/-******************************************************
	* Interface
	*********************************************************/

	/* note: several prototypes are already published in `zstd.h` :
	* ZSTD_createDDict()
	* ZSTD_createDDict_byReference()
	* ZSTD_createDDict_advanced()
	* ZSTD_freeDDict()
	* ZSTD_initStaticDDict()
	* ZSTD_sizeof_DDict()
	* ZSTD_estimateDDictSize()
	* ZSTD_getDictID_fromDict()
	*/

	const void* ZSTD_DDict_dictContent(const ZSTD_DDict* ddict);
	size_t ZSTD_DDict_dictSize(const ZSTD_DDict* ddict);

	void ZSTD_copyDDictParameters(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict);



	#endif /* ZSTD_DDICT_H */
	/** ended inlining zstd_ddict.h **/

	#if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>=1)
	/** start inlining ../legacy/zstd_legacy.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_LEGACY_H
	#define ZSTD_LEGACY_H

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/* *************************************
	* Includes
	***************************************/
	/** skipping file: ../common/mem.h **/
	/** skipping file: ../common/error_private.h **/
	/** skipping file: ../common/zstd_internal.h **/

	#if !defined (ZSTD_LEGACY_SUPPORT) \|\| (ZSTD_LEGACY_SUPPORT == 0)
	# undef ZSTD_LEGACY_SUPPORT
	# define ZSTD_LEGACY_SUPPORT 8
	#endif

	#if (ZSTD_LEGACY_SUPPORT <= 1)
	/** start inlining zstd_v01.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_V01_H_28739879432
	#define ZSTD_V01_H_28739879432

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/* *************************************
	* Includes
	***************************************/
	#include <stddef.h> /* size_t */


	/* *************************************
	* Simple one-step function
	***************************************/
	/**
	ZSTDv01_decompress() : decompress ZSTD frames compliant with v0.1.x format
	compressedSize : is the exact source size
	maxOriginalSize : is the size of the 'dst' buffer, which must be already allocated.
	It must be equal or larger than originalSize, otherwise decompression will fail.
	return : the number of bytes decompressed into destination buffer (originalSize)
	or an errorCode if it fails (which can be tested using ZSTDv01_isError())
	*/
	size_t ZSTDv01_decompress( void* dst, size_t maxOriginalSize,
	const void* src, size_t compressedSize);

	/**
	ZSTDv01_findFrameSizeInfoLegacy() : get the source length and decompressed bound of a ZSTD frame compliant with v0.1.x format
	srcSize : The size of the 'src' buffer, at least as large as the frame pointed to by 'src'
	cSize (output parameter) : the number of bytes that would be read to decompress this frame
	or an error code if it fails (which can be tested using ZSTDv01_isError())
	dBound (output parameter) : an upper-bound for the decompressed size of the data in the frame
	or ZSTD_CONTENTSIZE_ERROR if an error occurs

	note : assumes `cSize` and `dBound` are _not_ NULL.
	*/
	void ZSTDv01_findFrameSizeInfoLegacy(const void *src, size_t srcSize,
	size_t* cSize, unsigned long long* dBound);

	/**
	ZSTDv01_isError() : tells if the result of ZSTDv01_decompress() is an error
	*/
	unsigned ZSTDv01_isError(size_t code);


	/* *************************************
	* Advanced functions
	***************************************/
	typedef struct ZSTDv01_Dctx_s ZSTDv01_Dctx;
	ZSTDv01_Dctx* ZSTDv01_createDCtx(void);
	size_t ZSTDv01_freeDCtx(ZSTDv01_Dctx* dctx);

	size_t ZSTDv01_decompressDCtx(void* ctx,
	void* dst, size_t maxOriginalSize,
	const void* src, size_t compressedSize);

	/* *************************************
	* Streaming functions
	***************************************/
	size_t ZSTDv01_resetDCtx(ZSTDv01_Dctx* dctx);

	size_t ZSTDv01_nextSrcSizeToDecompress(ZSTDv01_Dctx* dctx);
	size_t ZSTDv01_decompressContinue(ZSTDv01_Dctx* dctx, void* dst, size_t maxDstSize, const void* src, size_t srcSize);
	/**
	Use above functions alternatively.
	ZSTD_nextSrcSizeToDecompress() tells how much bytes to provide as 'srcSize' to ZSTD_decompressContinue().
	ZSTD_decompressContinue() will use previous data blocks to improve compression if they are located prior to current block.
	Result is the number of bytes regenerated within 'dst'.
	It can be zero, which is not an error; it just means ZSTD_decompressContinue() has decoded some header.
	*/

	/* *************************************
	* Prefix - version detection
	***************************************/
	#define ZSTDv01_magicNumber 0xFD2FB51E /* Big Endian version */
	#define ZSTDv01_magicNumberLE 0x1EB52FFD /* Little Endian version */


	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_V01_H_28739879432 */
	/** ended inlining zstd_v01.h **/
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 2)
	/** start inlining zstd_v02.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_V02_H_4174539423
	#define ZSTD_V02_H_4174539423

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/* *************************************
	* Includes
	***************************************/
	#include <stddef.h> /* size_t */


	/* *************************************
	* Simple one-step function
	***************************************/
	/**
	ZSTDv02_decompress() : decompress ZSTD frames compliant with v0.2.x format
	compressedSize : is the exact source size
	maxOriginalSize : is the size of the 'dst' buffer, which must be already allocated.
	It must be equal or larger than originalSize, otherwise decompression will fail.
	return : the number of bytes decompressed into destination buffer (originalSize)
	or an errorCode if it fails (which can be tested using ZSTDv01_isError())
	*/
	size_t ZSTDv02_decompress( void* dst, size_t maxOriginalSize,
	const void* src, size_t compressedSize);

	/**
	ZSTDv02_findFrameSizeInfoLegacy() : get the source length and decompressed bound of a ZSTD frame compliant with v0.2.x format
	srcSize : The size of the 'src' buffer, at least as large as the frame pointed to by 'src'
	cSize (output parameter) : the number of bytes that would be read to decompress this frame
	or an error code if it fails (which can be tested using ZSTDv01_isError())
	dBound (output parameter) : an upper-bound for the decompressed size of the data in the frame
	or ZSTD_CONTENTSIZE_ERROR if an error occurs

	note : assumes `cSize` and `dBound` are _not_ NULL.
	*/
	void ZSTDv02_findFrameSizeInfoLegacy(const void *src, size_t srcSize,
	size_t* cSize, unsigned long long* dBound);

	/**
	ZSTDv02_isError() : tells if the result of ZSTDv02_decompress() is an error
	*/
	unsigned ZSTDv02_isError(size_t code);


	/* *************************************
	* Advanced functions
	***************************************/
	typedef struct ZSTDv02_Dctx_s ZSTDv02_Dctx;
	ZSTDv02_Dctx* ZSTDv02_createDCtx(void);
	size_t ZSTDv02_freeDCtx(ZSTDv02_Dctx* dctx);

	size_t ZSTDv02_decompressDCtx(void* ctx,
	void* dst, size_t maxOriginalSize,
	const void* src, size_t compressedSize);

	/* *************************************
	* Streaming functions
	***************************************/
	size_t ZSTDv02_resetDCtx(ZSTDv02_Dctx* dctx);

	size_t ZSTDv02_nextSrcSizeToDecompress(ZSTDv02_Dctx* dctx);
	size_t ZSTDv02_decompressContinue(ZSTDv02_Dctx* dctx, void* dst, size_t maxDstSize, const void* src, size_t srcSize);
	/**
	Use above functions alternatively.
	ZSTD_nextSrcSizeToDecompress() tells how much bytes to provide as 'srcSize' to ZSTD_decompressContinue().
	ZSTD_decompressContinue() will use previous data blocks to improve compression if they are located prior to current block.
	Result is the number of bytes regenerated within 'dst'.
	It can be zero, which is not an error; it just means ZSTD_decompressContinue() has decoded some header.
	*/

	/* *************************************
	* Prefix - version detection
	***************************************/
	#define ZSTDv02_magicNumber 0xFD2FB522 /* v0.2 */


	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_V02_H_4174539423 */
	/** ended inlining zstd_v02.h **/
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 3)
	/** start inlining zstd_v03.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_V03_H_298734209782
	#define ZSTD_V03_H_298734209782

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/* *************************************
	* Includes
	***************************************/
	#include <stddef.h> /* size_t */


	/* *************************************
	* Simple one-step function
	***************************************/
	/**
	ZSTDv03_decompress() : decompress ZSTD frames compliant with v0.3.x format
	compressedSize : is the exact source size
	maxOriginalSize : is the size of the 'dst' buffer, which must be already allocated.
	It must be equal or larger than originalSize, otherwise decompression will fail.
	return : the number of bytes decompressed into destination buffer (originalSize)
	or an errorCode if it fails (which can be tested using ZSTDv01_isError())
	*/
	size_t ZSTDv03_decompress( void* dst, size_t maxOriginalSize,
	const void* src, size_t compressedSize);

	/**
	ZSTDv03_findFrameSizeInfoLegacy() : get the source length and decompressed bound of a ZSTD frame compliant with v0.3.x format
	srcSize : The size of the 'src' buffer, at least as large as the frame pointed to by 'src'
	cSize (output parameter) : the number of bytes that would be read to decompress this frame
	or an error code if it fails (which can be tested using ZSTDv01_isError())
	dBound (output parameter) : an upper-bound for the decompressed size of the data in the frame
	or ZSTD_CONTENTSIZE_ERROR if an error occurs

	note : assumes `cSize` and `dBound` are _not_ NULL.
	*/
	void ZSTDv03_findFrameSizeInfoLegacy(const void *src, size_t srcSize,
	size_t* cSize, unsigned long long* dBound);

	/**
	ZSTDv03_isError() : tells if the result of ZSTDv03_decompress() is an error
	*/
	unsigned ZSTDv03_isError(size_t code);


	/* *************************************
	* Advanced functions
	***************************************/
	typedef struct ZSTDv03_Dctx_s ZSTDv03_Dctx;
	ZSTDv03_Dctx* ZSTDv03_createDCtx(void);
	size_t ZSTDv03_freeDCtx(ZSTDv03_Dctx* dctx);

	size_t ZSTDv03_decompressDCtx(void* ctx,
	void* dst, size_t maxOriginalSize,
	const void* src, size_t compressedSize);

	/* *************************************
	* Streaming functions
	***************************************/
	size_t ZSTDv03_resetDCtx(ZSTDv03_Dctx* dctx);

	size_t ZSTDv03_nextSrcSizeToDecompress(ZSTDv03_Dctx* dctx);
	size_t ZSTDv03_decompressContinue(ZSTDv03_Dctx* dctx, void* dst, size_t maxDstSize, const void* src, size_t srcSize);
	/**
	Use above functions alternatively.
	ZSTD_nextSrcSizeToDecompress() tells how much bytes to provide as 'srcSize' to ZSTD_decompressContinue().
	ZSTD_decompressContinue() will use previous data blocks to improve compression if they are located prior to current block.
	Result is the number of bytes regenerated within 'dst'.
	It can be zero, which is not an error; it just means ZSTD_decompressContinue() has decoded some header.
	*/

	/* *************************************
	* Prefix - version detection
	***************************************/
	#define ZSTDv03_magicNumber 0xFD2FB523 /* v0.3 */


	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_V03_H_298734209782 */
	/** ended inlining zstd_v03.h **/
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 4)
	/** start inlining zstd_v04.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTD_V04_H_91868324769238
	#define ZSTD_V04_H_91868324769238

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/* *************************************
	* Includes
	***************************************/
	#include <stddef.h> /* size_t */


	/* *************************************
	* Simple one-step function
	***************************************/
	/**
	ZSTDv04_decompress() : decompress ZSTD frames compliant with v0.4.x format
	compressedSize : is the exact source size
	maxOriginalSize : is the size of the 'dst' buffer, which must be already allocated.
	It must be equal or larger than originalSize, otherwise decompression will fail.
	return : the number of bytes decompressed into destination buffer (originalSize)
	or an errorCode if it fails (which can be tested using ZSTDv01_isError())
	*/
	size_t ZSTDv04_decompress( void* dst, size_t maxOriginalSize,
	const void* src, size_t compressedSize);

	/**
	ZSTDv04_findFrameSizeInfoLegacy() : get the source length and decompressed bound of a ZSTD frame compliant with v0.4.x format
	srcSize : The size of the 'src' buffer, at least as large as the frame pointed to by 'src'
	cSize (output parameter) : the number of bytes that would be read to decompress this frame
	or an error code if it fails (which can be tested using ZSTDv01_isError())
	dBound (output parameter) : an upper-bound for the decompressed size of the data in the frame
	or ZSTD_CONTENTSIZE_ERROR if an error occurs

	note : assumes `cSize` and `dBound` are _not_ NULL.
	*/
	void ZSTDv04_findFrameSizeInfoLegacy(const void *src, size_t srcSize,
	size_t* cSize, unsigned long long* dBound);

	/**
	ZSTDv04_isError() : tells if the result of ZSTDv04_decompress() is an error
	*/
	unsigned ZSTDv04_isError(size_t code);


	/* *************************************
	* Advanced functions
	***************************************/
	typedef struct ZSTDv04_Dctx_s ZSTDv04_Dctx;
	ZSTDv04_Dctx* ZSTDv04_createDCtx(void);
	size_t ZSTDv04_freeDCtx(ZSTDv04_Dctx* dctx);

	size_t ZSTDv04_decompressDCtx(ZSTDv04_Dctx* dctx,
	void* dst, size_t maxOriginalSize,
	const void* src, size_t compressedSize);


	/* *************************************
	* Direct Streaming
	***************************************/
	size_t ZSTDv04_resetDCtx(ZSTDv04_Dctx* dctx);

	size_t ZSTDv04_nextSrcSizeToDecompress(ZSTDv04_Dctx* dctx);
	size_t ZSTDv04_decompressContinue(ZSTDv04_Dctx* dctx, void* dst, size_t maxDstSize, const void* src, size_t srcSize);
	/**
	Use above functions alternatively.
	ZSTD_nextSrcSizeToDecompress() tells how much bytes to provide as 'srcSize' to ZSTD_decompressContinue().
	ZSTD_decompressContinue() will use previous data blocks to improve compression if they are located prior to current block.
	Result is the number of bytes regenerated within 'dst'.
	It can be zero, which is not an error; it just means ZSTD_decompressContinue() has decoded some header.
	*/


	/* *************************************
	* Buffered Streaming
	***************************************/
	typedef struct ZBUFFv04_DCtx_s ZBUFFv04_DCtx;
	ZBUFFv04_DCtx* ZBUFFv04_createDCtx(void);
	size_t ZBUFFv04_freeDCtx(ZBUFFv04_DCtx* dctx);

	size_t ZBUFFv04_decompressInit(ZBUFFv04_DCtx* dctx);
	size_t ZBUFFv04_decompressWithDictionary(ZBUFFv04_DCtx* dctx, const void* dict, size_t dictSize);

	size_t ZBUFFv04_decompressContinue(ZBUFFv04_DCtx* dctx, void* dst, size_t* maxDstSizePtr, const void* src, size_t* srcSizePtr);

	/ **********************************************
	* Streaming decompression
	*
	* A ZBUFF_DCtx object is required to track streaming operation.
	* Use ZBUFF_createDCtx() and ZBUFF_freeDCtx() to create/release resources.
	* Use ZBUFF_decompressInit() to start a new decompression operation.
	* ZBUFF_DCtx objects can be reused multiple times.
	*
	* Optionally, a reference to a static dictionary can be set, using ZBUFF_decompressWithDictionary()
	* It must be the same content as the one set during compression phase.
	* Dictionary content must remain accessible during the decompression process.
	*
	* Use ZBUFF_decompressContinue() repetitively to consume your input.
	* srcSizePtr and maxDstSizePtr can be any size.
	* The function will report how many bytes were read or written by modifying srcSizePtr and maxDstSizePtr.
	* Note that it may not consume the entire input, in which case it's up to the caller to present remaining input again.
	* The content of dst will be overwritten (up to *maxDstSizePtr) at each function call, so save its content if it matters or change dst.
	* @return : a hint to preferred nb of bytes to use as input for next function call (it's only a hint, to improve latency)
	* or 0 when a frame is completely decoded
	* or an error code, which can be tested using ZBUFF_isError().
	*
	* Hint : recommended buffer sizes (not compulsory) : ZBUFF_recommendedDInSize / ZBUFF_recommendedDOutSize
	* output : ZBUFF_recommendedDOutSize==128 KB block size is the internal unit, it ensures it's always possible to write a full block when it's decoded.
	* input : ZBUFF_recommendedDInSize==128Kb+3; just follow indications from ZBUFF_decompressContinue() to minimize latency. It should always be <= 128 KB + 3 .
	* **************************************************/
	unsigned ZBUFFv04_isError(size_t errorCode);
	const char* ZBUFFv04_getErrorName(size_t errorCode);


	/** The below functions provide recommended buffer sizes for Compression or Decompression operations.
	* These sizes are not compulsory, they just tend to offer better latency */
	size_t ZBUFFv04_recommendedDInSize(void);
	size_t ZBUFFv04_recommendedDOutSize(void);


	/* *************************************
	* Prefix - version detection
	***************************************/
	#define ZSTDv04_magicNumber 0xFD2FB524 /* v0.4 */


	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_V04_H_91868324769238 */
	/** ended inlining zstd_v04.h **/
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 5)
	/** start inlining zstd_v05.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTDv05_H
	#define ZSTDv05_H

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/-************************************
	* Dependencies
	***************************************/
	#include <stddef.h> /* size_t */
	/** skipping file: ../common/mem.h **/


	/* *************************************
	* Simple functions
	***************************************/
	/*! ZSTDv05_decompress() :
	`compressedSize` : is the _exact_ size of the compressed blob, otherwise decompression will fail.
	`dstCapacity` must be large enough, equal or larger than originalSize.
	@return : the number of bytes decompressed into `dst` (<= `dstCapacity`),
	or an errorCode if it fails (which can be tested using ZSTDv05_isError()) */
	size_t ZSTDv05_decompress( void* dst, size_t dstCapacity,
	const void* src, size_t compressedSize);

	/**
	ZSTDv05_findFrameSizeInfoLegacy() : get the source length and decompressed bound of a ZSTD frame compliant with v0.5.x format
	srcSize : The size of the 'src' buffer, at least as large as the frame pointed to by 'src'
	cSize (output parameter) : the number of bytes that would be read to decompress this frame
	or an error code if it fails (which can be tested using ZSTDv01_isError())
	dBound (output parameter) : an upper-bound for the decompressed size of the data in the frame
	or ZSTD_CONTENTSIZE_ERROR if an error occurs

	note : assumes `cSize` and `dBound` are _not_ NULL.
	*/
	void ZSTDv05_findFrameSizeInfoLegacy(const void *src, size_t srcSize,
	size_t* cSize, unsigned long long* dBound);

	/* *************************************
	* Helper functions
	***************************************/
	/* Error Management */
	unsigned ZSTDv05_isError(size_t code); /!< tells if a `size_t` function result is an error code /
	const char* ZSTDv05_getErrorName(size_t code); /!< provides readable string for an error code /


	/* *************************************
	* Explicit memory management
	***************************************/
	/** Decompression context */
	typedef struct ZSTDv05_DCtx_s ZSTDv05_DCtx;
	ZSTDv05_DCtx* ZSTDv05_createDCtx(void);
	size_t ZSTDv05_freeDCtx(ZSTDv05_DCtx* dctx); /!< @return : errorCode /

	/** ZSTDv05_decompressDCtx() :
	* Same as ZSTDv05_decompress(), but requires an already allocated ZSTDv05_DCtx (see ZSTDv05_createDCtx()) */
	size_t ZSTDv05_decompressDCtx(ZSTDv05_DCtx* ctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);


	/-**********************
	* Simple Dictionary API
	*************************/
	/*! ZSTDv05_decompress_usingDict() :
	* Decompression using a pre-defined Dictionary content (see dictBuilder).
	* Dictionary must be identical to the one used during compression, otherwise regenerated data will be corrupted.
	* Note : dict can be NULL, in which case, it's equivalent to ZSTDv05_decompressDCtx() */
	size_t ZSTDv05_decompress_usingDict(ZSTDv05_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict,size_t dictSize);

	/-***********************
	* Advanced Streaming API
	***************************/
	typedef enum { ZSTDv05_fast, ZSTDv05_greedy, ZSTDv05_lazy, ZSTDv05_lazy2, ZSTDv05_btlazy2, ZSTDv05_opt, ZSTDv05_btopt } ZSTDv05_strategy;
	typedef struct {
	U64 srcSize;
	U32 windowLog; /* the only useful information to retrieve */
	U32 contentLog; U32 hashLog; U32 searchLog; U32 searchLength; U32 targetLength; ZSTDv05_strategy strategy;
	} ZSTDv05_parameters;
	size_t ZSTDv05_getFrameParams(ZSTDv05_parameters* params, const void* src, size_t srcSize);

	size_t ZSTDv05_decompressBegin_usingDict(ZSTDv05_DCtx* dctx, const void* dict, size_t dictSize);
	void ZSTDv05_copyDCtx(ZSTDv05_DCtx* dstDCtx, const ZSTDv05_DCtx* srcDCtx);
	size_t ZSTDv05_nextSrcSizeToDecompress(ZSTDv05_DCtx* dctx);
	size_t ZSTDv05_decompressContinue(ZSTDv05_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);


	/-**********************
	* ZBUFF API
	*************************/
	typedef struct ZBUFFv05_DCtx_s ZBUFFv05_DCtx;
	ZBUFFv05_DCtx* ZBUFFv05_createDCtx(void);
	size_t ZBUFFv05_freeDCtx(ZBUFFv05_DCtx* dctx);

	size_t ZBUFFv05_decompressInit(ZBUFFv05_DCtx* dctx);
	size_t ZBUFFv05_decompressInitDictionary(ZBUFFv05_DCtx* dctx, const void* dict, size_t dictSize);

	size_t ZBUFFv05_decompressContinue(ZBUFFv05_DCtx* dctx,
	void* dst, size_t* dstCapacityPtr,
	const void* src, size_t* srcSizePtr);

	/-**************************************************************************
	* Streaming decompression
	*
	* A ZBUFFv05_DCtx object is required to track streaming operations.
	* Use ZBUFFv05_createDCtx() and ZBUFFv05_freeDCtx() to create/release resources.
	* Use ZBUFFv05_decompressInit() to start a new decompression operation,
	* or ZBUFFv05_decompressInitDictionary() if decompression requires a dictionary.
	* Note that ZBUFFv05_DCtx objects can be reused multiple times.
	*
	* Use ZBUFFv05_decompressContinue() repetitively to consume your input.
	* srcSizePtr and dstCapacityPtr can be any size.
	* The function will report how many bytes were read or written by modifying srcSizePtr and dstCapacityPtr.
	* Note that it may not consume the entire input, in which case it's up to the caller to present remaining input again.
	* The content of @dst will be overwritten (up to *dstCapacityPtr) at each function call, so save its content if it matters or change @dst.
	* @return : a hint to preferred nb of bytes to use as input for next function call (it's only a hint, to help latency)
	* or 0 when a frame is completely decoded
	* or an error code, which can be tested using ZBUFFv05_isError().
	*
	* Hint : recommended buffer sizes (not compulsory) : ZBUFFv05_recommendedDInSize() / ZBUFFv05_recommendedDOutSize()
	* output : ZBUFFv05_recommendedDOutSize==128 KB block size is the internal unit, it ensures it's always possible to write a full block when decoded.
	* input : ZBUFFv05_recommendedDInSize==128Kb+3; just follow indications from ZBUFFv05_decompressContinue() to minimize latency. It should always be <= 128 KB + 3 .
	* *******************************************************************************/


	/* *************************************
	* Tool functions
	***************************************/
	unsigned ZBUFFv05_isError(size_t errorCode);
	const char* ZBUFFv05_getErrorName(size_t errorCode);

	/** Functions below provide recommended buffer sizes for Compression or Decompression operations.
	* These sizes are just hints, and tend to offer better latency */
	size_t ZBUFFv05_recommendedDInSize(void);
	size_t ZBUFFv05_recommendedDOutSize(void);



	/-************************************
	* Constants
	***************************************/
	#define ZSTDv05_MAGICNUMBER 0xFD2FB525 /* v0.5 */




	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTDv0505_H */
	/** ended inlining zstd_v05.h **/
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 6)
	/** start inlining zstd_v06.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTDv06_H
	#define ZSTDv06_H

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/====== Dependency ======/
	#include <stddef.h> /* size_t */


	/====== Export for Windows ======/
	/*!
	* ZSTDv06_DLL_EXPORT :
	* Enable exporting of functions when building a Windows DLL
	*/
	#if defined(_WIN32) && defined(ZSTDv06_DLL_EXPORT) && (ZSTDv06_DLL_EXPORT==1)
	# define ZSTDLIBv06_API __declspec(dllexport)
	#else
	# define ZSTDLIBv06_API
	#endif


	/* *************************************
	* Simple functions
	***************************************/
	/*! ZSTDv06_decompress() :
	`compressedSize` : is the _exact_ size of the compressed blob, otherwise decompression will fail.
	`dstCapacity` must be large enough, equal or larger than originalSize.
	@return : the number of bytes decompressed into `dst` (<= `dstCapacity`),
	or an errorCode if it fails (which can be tested using ZSTDv06_isError()) */
	ZSTDLIBv06_API size_t ZSTDv06_decompress( void* dst, size_t dstCapacity,
	const void* src, size_t compressedSize);

	/**
	ZSTDv06_findFrameSizeInfoLegacy() : get the source length and decompressed bound of a ZSTD frame compliant with v0.6.x format
	srcSize : The size of the 'src' buffer, at least as large as the frame pointed to by 'src'
	cSize (output parameter) : the number of bytes that would be read to decompress this frame
	or an error code if it fails (which can be tested using ZSTDv01_isError())
	dBound (output parameter) : an upper-bound for the decompressed size of the data in the frame
	or ZSTD_CONTENTSIZE_ERROR if an error occurs

	note : assumes `cSize` and `dBound` are _not_ NULL.
	*/
	void ZSTDv06_findFrameSizeInfoLegacy(const void *src, size_t srcSize,
	size_t* cSize, unsigned long long* dBound);

	/* *************************************
	* Helper functions
	***************************************/
	ZSTDLIBv06_API size_t ZSTDv06_compressBound(size_t srcSize); /!< maximum compressed size (worst case scenario) /

	/* Error Management */
	ZSTDLIBv06_API unsigned ZSTDv06_isError(size_t code); /!< tells if a `size_t` function result is an error code /
	ZSTDLIBv06_API const char* ZSTDv06_getErrorName(size_t code); /!< provides readable string for an error code /


	/* *************************************
	* Explicit memory management
	***************************************/
	/** Decompression context */
	typedef struct ZSTDv06_DCtx_s ZSTDv06_DCtx;
	ZSTDLIBv06_API ZSTDv06_DCtx* ZSTDv06_createDCtx(void);
	ZSTDLIBv06_API size_t ZSTDv06_freeDCtx(ZSTDv06_DCtx* dctx); /!< @return : errorCode /

	/** ZSTDv06_decompressDCtx() :
	* Same as ZSTDv06_decompress(), but requires an already allocated ZSTDv06_DCtx (see ZSTDv06_createDCtx()) */
	ZSTDLIBv06_API size_t ZSTDv06_decompressDCtx(ZSTDv06_DCtx* ctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);


	/-**********************
	* Dictionary API
	*************************/
	/*! ZSTDv06_decompress_usingDict() :
	* Decompression using a pre-defined Dictionary content (see dictBuilder).
	* Dictionary must be identical to the one used during compression, otherwise regenerated data will be corrupted.
	* Note : dict can be NULL, in which case, it's equivalent to ZSTDv06_decompressDCtx() */
	ZSTDLIBv06_API size_t ZSTDv06_decompress_usingDict(ZSTDv06_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict,size_t dictSize);


	/-***********************
	* Advanced Streaming API
	***************************/
	struct ZSTDv06_frameParams_s { unsigned long long frameContentSize; unsigned windowLog; };
	typedef struct ZSTDv06_frameParams_s ZSTDv06_frameParams;

	ZSTDLIBv06_API size_t ZSTDv06_getFrameParams(ZSTDv06_frameParams* fparamsPtr, const void* src, size_t srcSize); /*< doesn't consume input /
	ZSTDLIBv06_API size_t ZSTDv06_decompressBegin_usingDict(ZSTDv06_DCtx* dctx, const void* dict, size_t dictSize);
	ZSTDLIBv06_API void ZSTDv06_copyDCtx(ZSTDv06_DCtx* dctx, const ZSTDv06_DCtx* preparedDCtx);

	ZSTDLIBv06_API size_t ZSTDv06_nextSrcSizeToDecompress(ZSTDv06_DCtx* dctx);
	ZSTDLIBv06_API size_t ZSTDv06_decompressContinue(ZSTDv06_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);



	/* *************************************
	* ZBUFF API
	***************************************/

	typedef struct ZBUFFv06_DCtx_s ZBUFFv06_DCtx;
	ZSTDLIBv06_API ZBUFFv06_DCtx* ZBUFFv06_createDCtx(void);
	ZSTDLIBv06_API size_t ZBUFFv06_freeDCtx(ZBUFFv06_DCtx* dctx);

	ZSTDLIBv06_API size_t ZBUFFv06_decompressInit(ZBUFFv06_DCtx* dctx);
	ZSTDLIBv06_API size_t ZBUFFv06_decompressInitDictionary(ZBUFFv06_DCtx* dctx, const void* dict, size_t dictSize);

	ZSTDLIBv06_API size_t ZBUFFv06_decompressContinue(ZBUFFv06_DCtx* dctx,
	void* dst, size_t* dstCapacityPtr,
	const void* src, size_t* srcSizePtr);

	/-**************************************************************************
	* Streaming decompression howto
	*
	* A ZBUFFv06_DCtx object is required to track streaming operations.
	* Use ZBUFFv06_createDCtx() and ZBUFFv06_freeDCtx() to create/release resources.
	* Use ZBUFFv06_decompressInit() to start a new decompression operation,
	* or ZBUFFv06_decompressInitDictionary() if decompression requires a dictionary.
	* Note that ZBUFFv06_DCtx objects can be re-init multiple times.
	*
	* Use ZBUFFv06_decompressContinue() repetitively to consume your input.
	* srcSizePtr and dstCapacityPtr can be any size.
	* The function will report how many bytes were read or written by modifying srcSizePtr and dstCapacityPtr.
	* Note that it may not consume the entire input, in which case it's up to the caller to present remaining input again.
	* The content of `dst` will be overwritten (up to *dstCapacityPtr) at each function call, so save its content if it matters, or change `dst`.
	* @return : a hint to preferred nb of bytes to use as input for next function call (it's only a hint, to help latency),
	* or 0 when a frame is completely decoded,
	* or an error code, which can be tested using ZBUFFv06_isError().
	*
	* Hint : recommended buffer sizes (not compulsory) : ZBUFFv06_recommendedDInSize() and ZBUFFv06_recommendedDOutSize()
	* output : ZBUFFv06_recommendedDOutSize== 128 KB block size is the internal unit, it ensures it's always possible to write a full block when decoded.
	* input : ZBUFFv06_recommendedDInSize == 128KB + 3;
	* just follow indications from ZBUFFv06_decompressContinue() to minimize latency. It should always be <= 128 KB + 3 .
	* *******************************************************************************/


	/* *************************************
	* Tool functions
	***************************************/
	ZSTDLIBv06_API unsigned ZBUFFv06_isError(size_t errorCode);
	ZSTDLIBv06_API const char* ZBUFFv06_getErrorName(size_t errorCode);

	/** Functions below provide recommended buffer sizes for Compression or Decompression operations.
	* These sizes are just hints, they tend to offer better latency */
	ZSTDLIBv06_API size_t ZBUFFv06_recommendedDInSize(void);
	ZSTDLIBv06_API size_t ZBUFFv06_recommendedDOutSize(void);


	/-************************************
	* Constants
	***************************************/
	#define ZSTDv06_MAGICNUMBER 0xFD2FB526 /* v0.6 */



	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTDv06_BUFFERED_H */
	/** ended inlining zstd_v06.h **/
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 7)
	/** start inlining zstd_v07.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	#ifndef ZSTDv07_H_235446
	#define ZSTDv07_H_235446

	#if defined (__cplusplus)
	extern "C" {
	#endif

	/====== Dependency ======/
	#include <stddef.h> /* size_t */


	/====== Export for Windows ======/
	/*!
	* ZSTDv07_DLL_EXPORT :
	* Enable exporting of functions when building a Windows DLL
	*/
	#if defined(_WIN32) && defined(ZSTDv07_DLL_EXPORT) && (ZSTDv07_DLL_EXPORT==1)
	# define ZSTDLIBv07_API __declspec(dllexport)
	#else
	# define ZSTDLIBv07_API
	#endif


	/* *************************************
	* Simple API
	***************************************/
	/*! ZSTDv07_getDecompressedSize() :
	* @return : decompressed size if known, 0 otherwise.
	note 1 : if `0`, follow up with ZSTDv07_getFrameParams() to know precise failure cause.
	note 2 : decompressed size could be wrong or intentionally modified !
	always ensure results fit within application's authorized limits */
	unsigned long long ZSTDv07_getDecompressedSize(const void* src, size_t srcSize);

	/*! ZSTDv07_decompress() :
	`compressedSize` : must be _exact_ size of compressed input, otherwise decompression will fail.
	`dstCapacity` must be equal or larger than originalSize.
	@return : the number of bytes decompressed into `dst` (<= `dstCapacity`),
	or an errorCode if it fails (which can be tested using ZSTDv07_isError()) */
	ZSTDLIBv07_API size_t ZSTDv07_decompress( void* dst, size_t dstCapacity,
	const void* src, size_t compressedSize);

	/**
	ZSTDv07_findFrameSizeInfoLegacy() : get the source length and decompressed bound of a ZSTD frame compliant with v0.7.x format
	srcSize : The size of the 'src' buffer, at least as large as the frame pointed to by 'src'
	cSize (output parameter) : the number of bytes that would be read to decompress this frame
	or an error code if it fails (which can be tested using ZSTDv01_isError())
	dBound (output parameter) : an upper-bound for the decompressed size of the data in the frame
	or ZSTD_CONTENTSIZE_ERROR if an error occurs

	note : assumes `cSize` and `dBound` are _not_ NULL.
	*/
	void ZSTDv07_findFrameSizeInfoLegacy(const void *src, size_t srcSize,
	size_t* cSize, unsigned long long* dBound);

	/====== Helper functions ======/
	ZSTDLIBv07_API unsigned ZSTDv07_isError(size_t code); /!< tells if a `size_t` function result is an error code /
	ZSTDLIBv07_API const char* ZSTDv07_getErrorName(size_t code); /!< provides readable string from an error code /


	/-************************************
	* Explicit memory management
	***************************************/
	/** Decompression context */
	typedef struct ZSTDv07_DCtx_s ZSTDv07_DCtx;
	ZSTDLIBv07_API ZSTDv07_DCtx* ZSTDv07_createDCtx(void);
	ZSTDLIBv07_API size_t ZSTDv07_freeDCtx(ZSTDv07_DCtx* dctx); /!< @return : errorCode /

	/** ZSTDv07_decompressDCtx() :
	* Same as ZSTDv07_decompress(), requires an allocated ZSTDv07_DCtx (see ZSTDv07_createDCtx()) */
	ZSTDLIBv07_API size_t ZSTDv07_decompressDCtx(ZSTDv07_DCtx* ctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize);


	/-***********************
	* Simple dictionary API
	***************************/
	/*! ZSTDv07_decompress_usingDict() :
	* Decompression using a pre-defined Dictionary content (see dictBuilder).
	* Dictionary must be identical to the one used during compression.
	* Note : This function load the dictionary, resulting in a significant startup time */
	ZSTDLIBv07_API size_t ZSTDv07_decompress_usingDict(ZSTDv07_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict,size_t dictSize);


	/-*************************
	* Advanced Dictionary API
	****************************/
	/*! ZSTDv07_createDDict() :
	* Create a digested dictionary, ready to start decompression operation without startup delay.
	* `dict` can be released after creation */
	typedef struct ZSTDv07_DDict_s ZSTDv07_DDict;
	ZSTDLIBv07_API ZSTDv07_DDict* ZSTDv07_createDDict(const void* dict, size_t dictSize);
	ZSTDLIBv07_API size_t ZSTDv07_freeDDict(ZSTDv07_DDict* ddict);

	/*! ZSTDv07_decompress_usingDDict() :
	* Decompression using a pre-digested Dictionary
	* Faster startup than ZSTDv07_decompress_usingDict(), recommended when same dictionary is used multiple times. */
	ZSTDLIBv07_API size_t ZSTDv07_decompress_usingDDict(ZSTDv07_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const ZSTDv07_DDict* ddict);

	typedef struct {
	unsigned long long frameContentSize;
	unsigned windowSize;
	unsigned dictID;
	unsigned checksumFlag;
	} ZSTDv07_frameParams;

	ZSTDLIBv07_API size_t ZSTDv07_getFrameParams(ZSTDv07_frameParams* fparamsPtr, const void* src, size_t srcSize); /*< doesn't consume input /




	/* *************************************
	* Streaming functions
	***************************************/
	typedef struct ZBUFFv07_DCtx_s ZBUFFv07_DCtx;
	ZSTDLIBv07_API ZBUFFv07_DCtx* ZBUFFv07_createDCtx(void);
	ZSTDLIBv07_API size_t ZBUFFv07_freeDCtx(ZBUFFv07_DCtx* dctx);

	ZSTDLIBv07_API size_t ZBUFFv07_decompressInit(ZBUFFv07_DCtx* dctx);
	ZSTDLIBv07_API size_t ZBUFFv07_decompressInitDictionary(ZBUFFv07_DCtx* dctx, const void* dict, size_t dictSize);

	ZSTDLIBv07_API size_t ZBUFFv07_decompressContinue(ZBUFFv07_DCtx* dctx,
	void* dst, size_t* dstCapacityPtr,
	const void* src, size_t* srcSizePtr);

	/-**************************************************************************
	* Streaming decompression howto
	*
	* A ZBUFFv07_DCtx object is required to track streaming operations.
	* Use ZBUFFv07_createDCtx() and ZBUFFv07_freeDCtx() to create/release resources.
	* Use ZBUFFv07_decompressInit() to start a new decompression operation,
	* or ZBUFFv07_decompressInitDictionary() if decompression requires a dictionary.
	* Note that ZBUFFv07_DCtx objects can be re-init multiple times.
	*
	* Use ZBUFFv07_decompressContinue() repetitively to consume your input.
	* srcSizePtr and dstCapacityPtr can be any size.
	* The function will report how many bytes were read or written by modifying srcSizePtr and dstCapacityPtr.
	* Note that it may not consume the entire input, in which case it's up to the caller to present remaining input again.
	* The content of `dst` will be overwritten (up to *dstCapacityPtr) at each function call, so save its content if it matters, or change `dst`.
	* @return : a hint to preferred nb of bytes to use as input for next function call (it's only a hint, to help latency),
	* or 0 when a frame is completely decoded,
	* or an error code, which can be tested using ZBUFFv07_isError().
	*
	* Hint : recommended buffer sizes (not compulsory) : ZBUFFv07_recommendedDInSize() and ZBUFFv07_recommendedDOutSize()
	* output : ZBUFFv07_recommendedDOutSize== 128 KB block size is the internal unit, it ensures it's always possible to write a full block when decoded.
	* input : ZBUFFv07_recommendedDInSize == 128KB + 3;
	* just follow indications from ZBUFFv07_decompressContinue() to minimize latency. It should always be <= 128 KB + 3 .
	* *******************************************************************************/


	/* *************************************
	* Tool functions
	***************************************/
	ZSTDLIBv07_API unsigned ZBUFFv07_isError(size_t errorCode);
	ZSTDLIBv07_API const char* ZBUFFv07_getErrorName(size_t errorCode);

	/** Functions below provide recommended buffer sizes for Compression or Decompression operations.
	* These sizes are just hints, they tend to offer better latency */
	ZSTDLIBv07_API size_t ZBUFFv07_recommendedDInSize(void);
	ZSTDLIBv07_API size_t ZBUFFv07_recommendedDOutSize(void);


	/-************************************
	* Constants
	***************************************/
	#define ZSTDv07_MAGICNUMBER 0xFD2FB527 /* v0.7 */


	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTDv07_H_235446 */
	/** ended inlining zstd_v07.h **/
	#endif

	/** ZSTD_isLegacy() :
	@return : > 0 if supported by legacy decoder. 0 otherwise.
	return value is the version.
	*/
	MEM_STATIC unsigned ZSTD_isLegacy(const void* src, size_t srcSize)
	{
	U32 magicNumberLE;
	if (srcSize<4) return 0;
	magicNumberLE = MEM_readLE32(src);
	switch(magicNumberLE)
	{
	#if (ZSTD_LEGACY_SUPPORT <= 1)
	case ZSTDv01_magicNumberLE:return 1;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 2)
	case ZSTDv02_magicNumber : return 2;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 3)
	case ZSTDv03_magicNumber : return 3;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 4)
	case ZSTDv04_magicNumber : return 4;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 5)
	case ZSTDv05_MAGICNUMBER : return 5;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 6)
	case ZSTDv06_MAGICNUMBER : return 6;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 7)
	case ZSTDv07_MAGICNUMBER : return 7;
	#endif
	default : return 0;
	}
	}


	MEM_STATIC unsigned long long ZSTD_getDecompressedSize_legacy(const void* src, size_t srcSize)
	{
	U32 const version = ZSTD_isLegacy(src, srcSize);
	if (version < 5) return 0; /* no decompressed size in frame header, or not a legacy format */
	#if (ZSTD_LEGACY_SUPPORT <= 5)
	if (version==5) {
	ZSTDv05_parameters fParams;
	size_t const frResult = ZSTDv05_getFrameParams(&fParams, src, srcSize);
	if (frResult != 0) return 0;
	return fParams.srcSize;
	}
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 6)
	if (version==6) {
	ZSTDv06_frameParams fParams;
	size_t const frResult = ZSTDv06_getFrameParams(&fParams, src, srcSize);
	if (frResult != 0) return 0;
	return fParams.frameContentSize;
	}
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 7)
	if (version==7) {
	ZSTDv07_frameParams fParams;
	size_t const frResult = ZSTDv07_getFrameParams(&fParams, src, srcSize);
	if (frResult != 0) return 0;
	return fParams.frameContentSize;
	}
	#endif
	return 0; /* should not be possible */
	}


	MEM_STATIC size_t ZSTD_decompressLegacy(
	void* dst, size_t dstCapacity,
	const void* src, size_t compressedSize,
	const void* dict,size_t dictSize)
	{
	U32 const version = ZSTD_isLegacy(src, compressedSize);
	(void)dst; (void)dstCapacity; (void)dict; (void)dictSize; /* unused when ZSTD_LEGACY_SUPPORT >= 8 */
	switch(version)
	{
	#if (ZSTD_LEGACY_SUPPORT <= 1)
	case 1 :
	return ZSTDv01_decompress(dst, dstCapacity, src, compressedSize);
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 2)
	case 2 :
	return ZSTDv02_decompress(dst, dstCapacity, src, compressedSize);
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 3)
	case 3 :
	return ZSTDv03_decompress(dst, dstCapacity, src, compressedSize);
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 4)
	case 4 :
	return ZSTDv04_decompress(dst, dstCapacity, src, compressedSize);
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 5)
	case 5 :
	{ size_t result;
	ZSTDv05_DCtx* const zd = ZSTDv05_createDCtx();
	if (zd==NULL) return ERROR(memory_allocation);
	result = ZSTDv05_decompress_usingDict(zd, dst, dstCapacity, src, compressedSize, dict, dictSize);
	ZSTDv05_freeDCtx(zd);
	return result;
	}
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 6)
	case 6 :
	{ size_t result;
	ZSTDv06_DCtx* const zd = ZSTDv06_createDCtx();
	if (zd==NULL) return ERROR(memory_allocation);
	result = ZSTDv06_decompress_usingDict(zd, dst, dstCapacity, src, compressedSize, dict, dictSize);
	ZSTDv06_freeDCtx(zd);
	return result;
	}
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 7)
	case 7 :
	{ size_t result;
	ZSTDv07_DCtx* const zd = ZSTDv07_createDCtx();
	if (zd==NULL) return ERROR(memory_allocation);
	result = ZSTDv07_decompress_usingDict(zd, dst, dstCapacity, src, compressedSize, dict, dictSize);
	ZSTDv07_freeDCtx(zd);
	return result;
	}
	#endif
	default :
	return ERROR(prefix_unknown);
	}
	}

	MEM_STATIC ZSTD_frameSizeInfo ZSTD_findFrameSizeInfoLegacy(const void *src, size_t srcSize)
	{
	ZSTD_frameSizeInfo frameSizeInfo;
	U32 const version = ZSTD_isLegacy(src, srcSize);
	switch(version)
	{
	#if (ZSTD_LEGACY_SUPPORT <= 1)
	case 1 :
	ZSTDv01_findFrameSizeInfoLegacy(src, srcSize,
	&frameSizeInfo.compressedSize,
	&frameSizeInfo.decompressedBound);
	break;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 2)
	case 2 :
	ZSTDv02_findFrameSizeInfoLegacy(src, srcSize,
	&frameSizeInfo.compressedSize,
	&frameSizeInfo.decompressedBound);
	break;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 3)
	case 3 :
	ZSTDv03_findFrameSizeInfoLegacy(src, srcSize,
	&frameSizeInfo.compressedSize,
	&frameSizeInfo.decompressedBound);
	break;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 4)
	case 4 :
	ZSTDv04_findFrameSizeInfoLegacy(src, srcSize,
	&frameSizeInfo.compressedSize,
	&frameSizeInfo.decompressedBound);
	break;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 5)
	case 5 :
	ZSTDv05_findFrameSizeInfoLegacy(src, srcSize,
	&frameSizeInfo.compressedSize,
	&frameSizeInfo.decompressedBound);
	break;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 6)
	case 6 :
	ZSTDv06_findFrameSizeInfoLegacy(src, srcSize,
	&frameSizeInfo.compressedSize,
	&frameSizeInfo.decompressedBound);
	break;
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 7)
	case 7 :
	ZSTDv07_findFrameSizeInfoLegacy(src, srcSize,
	&frameSizeInfo.compressedSize,
	&frameSizeInfo.decompressedBound);
	break;
	#endif
	default :
	frameSizeInfo.compressedSize = ERROR(prefix_unknown);
	frameSizeInfo.decompressedBound = ZSTD_CONTENTSIZE_ERROR;
	break;
	}
	if (!ZSTD_isError(frameSizeInfo.compressedSize) && frameSizeInfo.compressedSize > srcSize) {
	frameSizeInfo.compressedSize = ERROR(srcSize_wrong);
	frameSizeInfo.decompressedBound = ZSTD_CONTENTSIZE_ERROR;
	}
	return frameSizeInfo;
	}

	MEM_STATIC size_t ZSTD_findFrameCompressedSizeLegacy(const void *src, size_t srcSize)
	{
	ZSTD_frameSizeInfo frameSizeInfo = ZSTD_findFrameSizeInfoLegacy(src, srcSize);
	return frameSizeInfo.compressedSize;
	}

	MEM_STATIC size_t ZSTD_freeLegacyStreamContext(void* legacyContext, U32 version)
	{
	switch(version)
	{
	default :
	case 1 :
	case 2 :
	case 3 :
	(void)legacyContext;
	return ERROR(version_unsupported);
	#if (ZSTD_LEGACY_SUPPORT <= 4)
	case 4 : return ZBUFFv04_freeDCtx((ZBUFFv04_DCtx*)legacyContext);
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 5)
	case 5 : return ZBUFFv05_freeDCtx((ZBUFFv05_DCtx*)legacyContext);
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 6)
	case 6 : return ZBUFFv06_freeDCtx((ZBUFFv06_DCtx*)legacyContext);
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 7)
	case 7 : return ZBUFFv07_freeDCtx((ZBUFFv07_DCtx*)legacyContext);
	#endif
	}
	}


	MEM_STATIC size_t ZSTD_initLegacyStream(void** legacyContext, U32 prevVersion, U32 newVersion,
	const void* dict, size_t dictSize)
	{
	DEBUGLOG(5, "ZSTD_initLegacyStream for v0.%u", newVersion);
	if (prevVersion != newVersion) ZSTD_freeLegacyStreamContext(*legacyContext, prevVersion);
	switch(newVersion)
	{
	default :
	case 1 :
	case 2 :
	case 3 :
	(void)dict; (void)dictSize;
	return 0;
	#if (ZSTD_LEGACY_SUPPORT <= 4)
	case 4 :
	{
	ZBUFFv04_DCtx* dctx = (prevVersion != newVersion) ? ZBUFFv04_createDCtx() : (ZBUFFv04_DCtx)legacyContext;
	if (dctx==NULL) return ERROR(memory_allocation);
	ZBUFFv04_decompressInit(dctx);
	ZBUFFv04_decompressWithDictionary(dctx, dict, dictSize);
	*legacyContext = dctx;
	return 0;
	}
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 5)
	case 5 :
	{
	ZBUFFv05_DCtx* dctx = (prevVersion != newVersion) ? ZBUFFv05_createDCtx() : (ZBUFFv05_DCtx)legacyContext;
	if (dctx==NULL) return ERROR(memory_allocation);
	ZBUFFv05_decompressInitDictionary(dctx, dict, dictSize);
	*legacyContext = dctx;
	return 0;
	}
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 6)
	case 6 :
	{
	ZBUFFv06_DCtx* dctx = (prevVersion != newVersion) ? ZBUFFv06_createDCtx() : (ZBUFFv06_DCtx)legacyContext;
	if (dctx==NULL) return ERROR(memory_allocation);
	ZBUFFv06_decompressInitDictionary(dctx, dict, dictSize);
	*legacyContext = dctx;
	return 0;
	}
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 7)
	case 7 :
	{
	ZBUFFv07_DCtx* dctx = (prevVersion != newVersion) ? ZBUFFv07_createDCtx() : (ZBUFFv07_DCtx)legacyContext;
	if (dctx==NULL) return ERROR(memory_allocation);
	ZBUFFv07_decompressInitDictionary(dctx, dict, dictSize);
	*legacyContext = dctx;
	return 0;
	}
	#endif
	}
	}



	MEM_STATIC size_t ZSTD_decompressLegacyStream(void* legacyContext, U32 version,
	ZSTD_outBuffer* output, ZSTD_inBuffer* input)
	{
	DEBUGLOG(5, "ZSTD_decompressLegacyStream for v0.%u", version);
	switch(version)
	{
	default :
	case 1 :
	case 2 :
	case 3 :
	(void)legacyContext; (void)output; (void)input;
	return ERROR(version_unsupported);
	#if (ZSTD_LEGACY_SUPPORT <= 4)
	case 4 :
	{
	ZBUFFv04_DCtx* dctx = (ZBUFFv04_DCtx*) legacyContext;
	const void* src = (const char*)input->src + input->pos;
	size_t readSize = input->size - input->pos;
	void* dst = (char*)output->dst + output->pos;
	size_t decodedSize = output->size - output->pos;
	size_t const hintSize = ZBUFFv04_decompressContinue(dctx, dst, &decodedSize, src, &readSize);
	output->pos += decodedSize;
	input->pos += readSize;
	return hintSize;
	}
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 5)
	case 5 :
	{
	ZBUFFv05_DCtx* dctx = (ZBUFFv05_DCtx*) legacyContext;
	const void* src = (const char*)input->src + input->pos;
	size_t readSize = input->size - input->pos;
	void* dst = (char*)output->dst + output->pos;
	size_t decodedSize = output->size - output->pos;
	size_t const hintSize = ZBUFFv05_decompressContinue(dctx, dst, &decodedSize, src, &readSize);
	output->pos += decodedSize;
	input->pos += readSize;
	return hintSize;
	}
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 6)
	case 6 :
	{
	ZBUFFv06_DCtx* dctx = (ZBUFFv06_DCtx*) legacyContext;
	const void* src = (const char*)input->src + input->pos;
	size_t readSize = input->size - input->pos;
	void* dst = (char*)output->dst + output->pos;
	size_t decodedSize = output->size - output->pos;
	size_t const hintSize = ZBUFFv06_decompressContinue(dctx, dst, &decodedSize, src, &readSize);
	output->pos += decodedSize;
	input->pos += readSize;
	return hintSize;
	}
	#endif
	#if (ZSTD_LEGACY_SUPPORT <= 7)
	case 7 :
	{
	ZBUFFv07_DCtx* dctx = (ZBUFFv07_DCtx*) legacyContext;
	const void* src = (const char*)input->src + input->pos;
	size_t readSize = input->size - input->pos;
	void* dst = (char*)output->dst + output->pos;
	size_t decodedSize = output->size - output->pos;
	size_t const hintSize = ZBUFFv07_decompressContinue(dctx, dst, &decodedSize, src, &readSize);
	output->pos += decodedSize;
	input->pos += readSize;
	return hintSize;
	}
	#endif
	}
	}


	#if defined (__cplusplus)
	}
	#endif

	#endif /* ZSTD_LEGACY_H */
	/** ended inlining ../legacy/zstd_legacy.h **/
	#endif



	/-******************************************************
	* Types
	*********************************************************/
	struct ZSTD_DDict_s {
	void* dictBuffer;
	const void* dictContent;
	size_t dictSize;
	ZSTD_entropyDTables_t entropy;
	U32 dictID;
	U32 entropyPresent;
	ZSTD_customMem cMem;
	}; /* typedef'd to ZSTD_DDict within "zstd.h" */

	const void* ZSTD_DDict_dictContent(const ZSTD_DDict* ddict)
	{
	assert(ddict != NULL);
	return ddict->dictContent;
	}

	size_t ZSTD_DDict_dictSize(const ZSTD_DDict* ddict)
	{
	assert(ddict != NULL);
	return ddict->dictSize;
	}

	void ZSTD_copyDDictParameters(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
	{
	DEBUGLOG(4, "ZSTD_copyDDictParameters");
	assert(dctx != NULL);
	assert(ddict != NULL);
	dctx->dictID = ddict->dictID;
	dctx->prefixStart = ddict->dictContent;
	dctx->virtualStart = ddict->dictContent;
	dctx->dictEnd = (const BYTE*)ddict->dictContent + ddict->dictSize;
	dctx->previousDstEnd = dctx->dictEnd;
	#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
	dctx->dictContentBeginForFuzzing = dctx->prefixStart;
	dctx->dictContentEndForFuzzing = dctx->previousDstEnd;
	#endif
	if (ddict->entropyPresent) {
	dctx->litEntropy = 1;
	dctx->fseEntropy = 1;
	dctx->LLTptr = ddict->entropy.LLTable;
	dctx->MLTptr = ddict->entropy.MLTable;
	dctx->OFTptr = ddict->entropy.OFTable;
	dctx->HUFptr = ddict->entropy.hufTable;
	dctx->entropy.rep[0] = ddict->entropy.rep[0];
	dctx->entropy.rep[1] = ddict->entropy.rep[1];
	dctx->entropy.rep[2] = ddict->entropy.rep[2];
	} else {
	dctx->litEntropy = 0;
	dctx->fseEntropy = 0;
	}
	}


	static size_t
	ZSTD_loadEntropy_intoDDict(ZSTD_DDict* ddict,
	ZSTD_dictContentType_e dictContentType)
	{
	ddict->dictID = 0;
	ddict->entropyPresent = 0;
	if (dictContentType == ZSTD_dct_rawContent) return 0;

	if (ddict->dictSize < 8) {
	if (dictContentType == ZSTD_dct_fullDict)
	return ERROR(dictionary_corrupted); /* only accept specified dictionaries */
	return 0; /* pure content mode */
	}
	{ U32 const magic = MEM_readLE32(ddict->dictContent);
	if (magic != ZSTD_MAGIC_DICTIONARY) {
	if (dictContentType == ZSTD_dct_fullDict)
	return ERROR(dictionary_corrupted); /* only accept specified dictionaries */
	return 0; /* pure content mode */
	}
	}
	ddict->dictID = MEM_readLE32((const char*)ddict->dictContent + ZSTD_FRAMEIDSIZE);

	/* load entropy tables */
	RETURN_ERROR_IF(ZSTD_isError(ZSTD_loadDEntropy(
	&ddict->entropy, ddict->dictContent, ddict->dictSize)),
	dictionary_corrupted, "");
	ddict->entropyPresent = 1;
	return 0;
	}


	static size_t ZSTD_initDDict_internal(ZSTD_DDict* ddict,
	const void* dict, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod,
	ZSTD_dictContentType_e dictContentType)
	{
	if ((dictLoadMethod == ZSTD_dlm_byRef) \|\| (!dict) \|\| (!dictSize)) {
	ddict->dictBuffer = NULL;
	ddict->dictContent = dict;
	if (!dict) dictSize = 0;
	} else {
	void* const internalBuffer = ZSTD_malloc(dictSize, ddict->cMem);
	ddict->dictBuffer = internalBuffer;
	ddict->dictContent = internalBuffer;
	if (!internalBuffer) return ERROR(memory_allocation);
	memcpy(internalBuffer, dict, dictSize);
	}
	ddict->dictSize = dictSize;
	ddict->entropy.hufTable[0] = (HUF_DTable)((HufLog)0x1000001); / cover both little and big endian */

	/* parse dictionary content */
	FORWARD_IF_ERROR( ZSTD_loadEntropy_intoDDict(ddict, dictContentType) , "");

	return 0;
	}

	ZSTD_DDict* ZSTD_createDDict_advanced(const void* dict, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod,
	ZSTD_dictContentType_e dictContentType,
	ZSTD_customMem customMem)
	{
	if (!customMem.customAlloc ^ !customMem.customFree) return NULL;

	{ ZSTD_DDict* const ddict = (ZSTD_DDict*) ZSTD_malloc(sizeof(ZSTD_DDict), customMem);
	if (ddict == NULL) return NULL;
	ddict->cMem = customMem;
	{ size_t const initResult = ZSTD_initDDict_internal(ddict,
	dict, dictSize,
	dictLoadMethod, dictContentType);
	if (ZSTD_isError(initResult)) {
	ZSTD_freeDDict(ddict);
	return NULL;
	} }
	return ddict;
	}
	}

	/*! ZSTD_createDDict() :
	* Create a digested dictionary, to start decompression without startup delay.
	* `dict` content is copied inside DDict.
	* Consequently, `dict` can be released after `ZSTD_DDict` creation */
	ZSTD_DDict* ZSTD_createDDict(const void* dict, size_t dictSize)
	{
	ZSTD_customMem const allocator = { NULL, NULL, NULL };
	return ZSTD_createDDict_advanced(dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto, allocator);
	}

	/*! ZSTD_createDDict_byReference() :
	* Create a digested dictionary, to start decompression without startup delay.
	* Dictionary content is simply referenced, it will be accessed during decompression.
	* Warning : dictBuffer must outlive DDict (DDict must be freed before dictBuffer) */
	ZSTD_DDict* ZSTD_createDDict_byReference(const void* dictBuffer, size_t dictSize)
	{
	ZSTD_customMem const allocator = { NULL, NULL, NULL };
	return ZSTD_createDDict_advanced(dictBuffer, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto, allocator);
	}


	const ZSTD_DDict* ZSTD_initStaticDDict(
	void* sBuffer, size_t sBufferSize,
	const void* dict, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod,
	ZSTD_dictContentType_e dictContentType)
	{
	size_t const neededSpace = sizeof(ZSTD_DDict)
	+ (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize);
	ZSTD_DDict* const ddict = (ZSTD_DDict*)sBuffer;
	assert(sBuffer != NULL);
	assert(dict != NULL);
	if ((size_t)sBuffer & 7) return NULL; /* 8-aligned */
	if (sBufferSize < neededSpace) return NULL;
	if (dictLoadMethod == ZSTD_dlm_byCopy) {
	memcpy(ddict+1, dict, dictSize); /* local copy */
	dict = ddict+1;
	}
	if (ZSTD_isError( ZSTD_initDDict_internal(ddict,
	dict, dictSize,
	ZSTD_dlm_byRef, dictContentType) ))
	return NULL;
	return ddict;
	}


	size_t ZSTD_freeDDict(ZSTD_DDict* ddict)
	{
	if (ddict==NULL) return 0; /* support free on NULL */
	{ ZSTD_customMem const cMem = ddict->cMem;
	ZSTD_free(ddict->dictBuffer, cMem);
	ZSTD_free(ddict, cMem);
	return 0;
	}
	}

	/*! ZSTD_estimateDDictSize() :
	* Estimate amount of memory that will be needed to create a dictionary for decompression.
	* Note : dictionary created by reference using ZSTD_dlm_byRef are smaller */
	size_t ZSTD_estimateDDictSize(size_t dictSize, ZSTD_dictLoadMethod_e dictLoadMethod)
	{
	return sizeof(ZSTD_DDict) + (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize);
	}

	size_t ZSTD_sizeof_DDict(const ZSTD_DDict* ddict)
	{
	if (ddict==NULL) return 0; /* support sizeof on NULL */
	return sizeof(*ddict) + (ddict->dictBuffer ? ddict->dictSize : 0) ;
	}

	/*! ZSTD_getDictID_fromDDict() :
	* Provides the dictID of the dictionary loaded into `ddict`.
	* If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
	* Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
	unsigned ZSTD_getDictID_fromDDict(const ZSTD_DDict* ddict)
	{
	if (ddict==NULL) return 0;
	return ZSTD_getDictID_fromDict(ddict->dictContent, ddict->dictSize);
	}
	/** ended inlining decompress/zstd_ddict.c **/
	/** start inlining decompress/zstd_decompress.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/


	/* ***************************************************************
	* Tuning parameters
	*****************************************************************/
	/*!
	* HEAPMODE :
	* Select how default decompression function ZSTD_decompress() allocates its context,
	* on stack (0), or into heap (1, default; requires malloc()).
	* Note that functions with explicit context such as ZSTD_decompressDCtx() are unaffected.
	*/
	#ifndef ZSTD_HEAPMODE
	# define ZSTD_HEAPMODE 1
	#endif

	/*!
	* LEGACY_SUPPORT :
	* if set to 1+, ZSTD_decompress() can decode older formats (v0.1+)
	*/
	#ifndef ZSTD_LEGACY_SUPPORT
	# define ZSTD_LEGACY_SUPPORT 0
	#endif

	/*!
	* MAXWINDOWSIZE_DEFAULT :
	* maximum window size accepted by DStream __by default__.
	* Frames requiring more memory will be rejected.
	* It's possible to set a different limit using ZSTD_DCtx_setMaxWindowSize().
	*/
	#ifndef ZSTD_MAXWINDOWSIZE_DEFAULT
	# define ZSTD_MAXWINDOWSIZE_DEFAULT (((U32)1 << ZSTD_WINDOWLOG_LIMIT_DEFAULT) + 1)
	#endif

	/*!
	* NO_FORWARD_PROGRESS_MAX :
	* maximum allowed nb of calls to ZSTD_decompressStream()
	* without any forward progress
	* (defined as: no byte read from input, and no byte flushed to output)
	* before triggering an error.
	*/
	#ifndef ZSTD_NO_FORWARD_PROGRESS_MAX
	# define ZSTD_NO_FORWARD_PROGRESS_MAX 16
	#endif


	/-******************************************************
	* Dependencies
	*********************************************************/
	#include <string.h> /* memcpy, memmove, memset */
	/** skipping file: ../common/cpu.h **/
	/** skipping file: ../common/mem.h **/
	#define FSE_STATIC_LINKING_ONLY
	/** skipping file: ../common/fse.h **/
	#define HUF_STATIC_LINKING_ONLY
	/** skipping file: ../common/huf.h **/
	/** skipping file: ../common/zstd_internal.h **/
	/** skipping file: zstd_decompress_internal.h **/
	/** skipping file: zstd_ddict.h **/
	/** start inlining zstd_decompress_block.h **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/


	#ifndef ZSTD_DEC_BLOCK_H
	#define ZSTD_DEC_BLOCK_H

	/-******************************************************
	* Dependencies
	*********************************************************/
	#include <stddef.h> /* size_t */
	/** skipping file: ../zstd.h **/
	/** skipping file: ../common/zstd_internal.h **/
	/** skipping file: zstd_decompress_internal.h **/


	/* === Prototypes === */

	/* note: prototypes already published within `zstd.h` :
	* ZSTD_decompressBlock()
	*/

	/* note: prototypes already published within `zstd_internal.h` :
	* ZSTD_getcBlockSize()
	* ZSTD_decodeSeqHeaders()
	*/


	/* ZSTD_decompressBlock_internal() :
	* decompress block, starting at `src`,
	* into destination buffer `dst`.
	* @return : decompressed block size,
	* or an error code (which can be tested using ZSTD_isError())
	*/
	size_t ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize, const int frame);

	/* ZSTD_buildFSETable() :
	* generate FSE decoding table for one symbol (ll, ml or off)
	* this function must be called with valid parameters only
	* (dt is large enough, normalizedCounter distribution total is a power of 2, max is within range, etc.)
	* in which case it cannot fail.
	* Internal use only.
	*/
	void ZSTD_buildFSETable(ZSTD_seqSymbol* dt,
	const short* normalizedCounter, unsigned maxSymbolValue,
	const U32* baseValue, const U32* nbAdditionalBits,
	unsigned tableLog);


	#endif /* ZSTD_DEC_BLOCK_H */
	/** ended inlining zstd_decompress_block.h **/

	#if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>=1)
	/** skipping file: ../legacy/zstd_legacy.h **/
	#endif


	/-************************************************************
	* Context management
	***************************************************************/
	size_t ZSTD_sizeof_DCtx (const ZSTD_DCtx* dctx)
	{
	if (dctx==NULL) return 0; /* support sizeof NULL */
	return sizeof(*dctx)
	+ ZSTD_sizeof_DDict(dctx->ddictLocal)
	+ dctx->inBuffSize + dctx->outBuffSize;
	}

	size_t ZSTD_estimateDCtxSize(void) { return sizeof(ZSTD_DCtx); }


	static size_t ZSTD_startingInputLength(ZSTD_format_e format)
	{
	size_t const startingInputLength = ZSTD_FRAMEHEADERSIZE_PREFIX(format);
	/* only supports formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless */
	assert( (format == ZSTD_f_zstd1) \|\| (format == ZSTD_f_zstd1_magicless) );
	return startingInputLength;
	}

	static void ZSTD_initDCtx_internal(ZSTD_DCtx* dctx)
	{
	dctx->format = ZSTD_f_zstd1; /* ZSTD_decompressBegin() invokes ZSTD_startingInputLength() with argument dctx->format */
	dctx->staticSize = 0;
	dctx->maxWindowSize = ZSTD_MAXWINDOWSIZE_DEFAULT;
	dctx->ddict = NULL;
	dctx->ddictLocal = NULL;
	dctx->dictEnd = NULL;
	dctx->ddictIsCold = 0;
	dctx->dictUses = ZSTD_dont_use;
	dctx->inBuff = NULL;
	dctx->inBuffSize = 0;
	dctx->outBuffSize = 0;
	dctx->streamStage = zdss_init;
	dctx->legacyContext = NULL;
	dctx->previousLegacyVersion = 0;
	dctx->noForwardProgress = 0;
	dctx->oversizedDuration = 0;
	dctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid());
	dctx->outBufferMode = ZSTD_obm_buffered;
	#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
	dctx->dictContentEndForFuzzing = NULL;
	#endif
	}

	ZSTD_DCtx* ZSTD_initStaticDCtx(void *workspace, size_t workspaceSize)
	{
	ZSTD_DCtx* const dctx = (ZSTD_DCtx*) workspace;

	if ((size_t)workspace & 7) return NULL; /* 8-aligned */
	if (workspaceSize < sizeof(ZSTD_DCtx)) return NULL; /* minimum size */

	ZSTD_initDCtx_internal(dctx);
	dctx->staticSize = workspaceSize;
	dctx->inBuff = (char*)(dctx+1);
	return dctx;
	}

	ZSTD_DCtx* ZSTD_createDCtx_advanced(ZSTD_customMem customMem)
	{
	if (!customMem.customAlloc ^ !customMem.customFree) return NULL;

	{ ZSTD_DCtx* const dctx = (ZSTD_DCtx)ZSTD_malloc(sizeof(dctx), customMem);
	if (!dctx) return NULL;
	dctx->customMem = customMem;
	ZSTD_initDCtx_internal(dctx);
	return dctx;
	}
	}

	ZSTD_DCtx* ZSTD_createDCtx(void)
	{
	DEBUGLOG(3, "ZSTD_createDCtx");
	return ZSTD_createDCtx_advanced(ZSTD_defaultCMem);
	}

	static void ZSTD_clearDict(ZSTD_DCtx* dctx)
	{
	ZSTD_freeDDict(dctx->ddictLocal);
	dctx->ddictLocal = NULL;
	dctx->ddict = NULL;
	dctx->dictUses = ZSTD_dont_use;
	}

	size_t ZSTD_freeDCtx(ZSTD_DCtx* dctx)
	{
	if (dctx==NULL) return 0; /* support free on NULL */
	RETURN_ERROR_IF(dctx->staticSize, memory_allocation, "not compatible with static DCtx");
	{ ZSTD_customMem const cMem = dctx->customMem;
	ZSTD_clearDict(dctx);
	ZSTD_free(dctx->inBuff, cMem);
	dctx->inBuff = NULL;
	#if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
	if (dctx->legacyContext)
	ZSTD_freeLegacyStreamContext(dctx->legacyContext, dctx->previousLegacyVersion);
	#endif
	ZSTD_free(dctx, cMem);
	return 0;
	}
	}

	/* no longer useful */
	void ZSTD_copyDCtx(ZSTD_DCtx* dstDCtx, const ZSTD_DCtx* srcDCtx)
	{
	size_t const toCopy = (size_t)((char)(&dstDCtx->inBuff) - (char)dstDCtx);
	memcpy(dstDCtx, srcDCtx, toCopy); /* no need to copy workspace */
	}


	/-************************************************************
	* Frame header decoding
	***************************************************************/

	/*! ZSTD_isFrame() :
	* Tells if the content of `buffer` starts with a valid Frame Identifier.
	* Note : Frame Identifier is 4 bytes. If `size < 4`, @return will always be 0.
	* Note 2 : Legacy Frame Identifiers are considered valid only if Legacy Support is enabled.
	* Note 3 : Skippable Frame Identifiers are considered valid. */
	unsigned ZSTD_isFrame(const void* buffer, size_t size)
	{
	if (size < ZSTD_FRAMEIDSIZE) return 0;
	{ U32 const magic = MEM_readLE32(buffer);
	if (magic == ZSTD_MAGICNUMBER) return 1;
	if ((magic & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) return 1;
	}
	#if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
	if (ZSTD_isLegacy(buffer, size)) return 1;
	#endif
	return 0;
	}

	/** ZSTD_frameHeaderSize_internal() :
	* srcSize must be large enough to reach header size fields.
	* note : only works for formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless.
	* @return : size of the Frame Header
	* or an error code, which can be tested with ZSTD_isError() */
	static size_t ZSTD_frameHeaderSize_internal(const void* src, size_t srcSize, ZSTD_format_e format)
	{
	size_t const minInputSize = ZSTD_startingInputLength(format);
	RETURN_ERROR_IF(srcSize < minInputSize, srcSize_wrong, "");

	{ BYTE const fhd = ((const BYTE*)src)[minInputSize-1];
	U32 const dictID= fhd & 3;
	U32 const singleSegment = (fhd >> 5) & 1;
	U32 const fcsId = fhd >> 6;
	return minInputSize + !singleSegment
	+ ZSTD_did_fieldSize[dictID] + ZSTD_fcs_fieldSize[fcsId]
	+ (singleSegment && !fcsId);
	}
	}

	/** ZSTD_frameHeaderSize() :
	* srcSize must be >= ZSTD_frameHeaderSize_prefix.
	* @return : size of the Frame Header,
	* or an error code (if srcSize is too small) */
	size_t ZSTD_frameHeaderSize(const void* src, size_t srcSize)
	{
	return ZSTD_frameHeaderSize_internal(src, srcSize, ZSTD_f_zstd1);
	}


	/** ZSTD_getFrameHeader_advanced() :
	* decode Frame Header, or require larger `srcSize`.
	* note : only works for formats ZSTD_f_zstd1 and ZSTD_f_zstd1_magicless
	* @return : 0, `zfhPtr` is correctly filled,
	* >0, `srcSize` is too small, value is wanted `srcSize` amount,
	* or an error code, which can be tested using ZSTD_isError() */
	size_t ZSTD_getFrameHeader_advanced(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize, ZSTD_format_e format)
	{
	const BYTE* ip = (const BYTE*)src;
	size_t const minInputSize = ZSTD_startingInputLength(format);

	memset(zfhPtr, 0, sizeof(zfhPtr)); / not strictly necessary, but static analyzer do not understand that zfhPtr is only going to be read only if return value is zero, since they are 2 different signals */
	if (srcSize < minInputSize) return minInputSize;
	RETURN_ERROR_IF(src==NULL, GENERIC, "invalid parameter");

	if ( (format != ZSTD_f_zstd1_magicless)
	&& (MEM_readLE32(src) != ZSTD_MAGICNUMBER) ) {
	if ((MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
	/* skippable frame */
	if (srcSize < ZSTD_SKIPPABLEHEADERSIZE)
	return ZSTD_SKIPPABLEHEADERSIZE; /* magic number + frame length */
	memset(zfhPtr, 0, sizeof(*zfhPtr));
	zfhPtr->frameContentSize = MEM_readLE32((const char *)src + ZSTD_FRAMEIDSIZE);
	zfhPtr->frameType = ZSTD_skippableFrame;
	return 0;
	}
	RETURN_ERROR(prefix_unknown, "");
	}

	/* ensure there is enough `srcSize` to fully read/decode frame header */
	{ size_t const fhsize = ZSTD_frameHeaderSize_internal(src, srcSize, format);
	if (srcSize < fhsize) return fhsize;
	zfhPtr->headerSize = (U32)fhsize;
	}

	{ BYTE const fhdByte = ip[minInputSize-1];
	size_t pos = minInputSize;
	U32 const dictIDSizeCode = fhdByte&3;
	U32 const checksumFlag = (fhdByte>>2)&1;
	U32 const singleSegment = (fhdByte>>5)&1;
	U32 const fcsID = fhdByte>>6;
	U64 windowSize = 0;
	U32 dictID = 0;
	U64 frameContentSize = ZSTD_CONTENTSIZE_UNKNOWN;
	RETURN_ERROR_IF((fhdByte & 0x08) != 0, frameParameter_unsupported,
	"reserved bits, must be zero");

	if (!singleSegment) {
	BYTE const wlByte = ip[pos++];
	U32 const windowLog = (wlByte >> 3) + ZSTD_WINDOWLOG_ABSOLUTEMIN;
	RETURN_ERROR_IF(windowLog > ZSTD_WINDOWLOG_MAX, frameParameter_windowTooLarge, "");
	windowSize = (1ULL << windowLog);
	windowSize += (windowSize >> 3) * (wlByte&7);
	}
	switch(dictIDSizeCode)
	{
	default: assert(0); /* impossible */
	case 0 : break;
	case 1 : dictID = ip[pos]; pos++; break;
	case 2 : dictID = MEM_readLE16(ip+pos); pos+=2; break;
	case 3 : dictID = MEM_readLE32(ip+pos); pos+=4; break;
	}
	switch(fcsID)
	{
	default: assert(0); /* impossible */
	case 0 : if (singleSegment) frameContentSize = ip[pos]; break;
	case 1 : frameContentSize = MEM_readLE16(ip+pos)+256; break;
	case 2 : frameContentSize = MEM_readLE32(ip+pos); break;
	case 3 : frameContentSize = MEM_readLE64(ip+pos); break;
	}
	if (singleSegment) windowSize = frameContentSize;

	zfhPtr->frameType = ZSTD_frame;
	zfhPtr->frameContentSize = frameContentSize;
	zfhPtr->windowSize = windowSize;
	zfhPtr->blockSizeMax = (unsigned) MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
	zfhPtr->dictID = dictID;
	zfhPtr->checksumFlag = checksumFlag;
	}
	return 0;
	}

	/** ZSTD_getFrameHeader() :
	* decode Frame Header, or require larger `srcSize`.
	* note : this function does not consume input, it only reads it.
	* @return : 0, `zfhPtr` is correctly filled,
	* >0, `srcSize` is too small, value is wanted `srcSize` amount,
	* or an error code, which can be tested using ZSTD_isError() */
	size_t ZSTD_getFrameHeader(ZSTD_frameHeader* zfhPtr, const void* src, size_t srcSize)
	{
	return ZSTD_getFrameHeader_advanced(zfhPtr, src, srcSize, ZSTD_f_zstd1);
	}


	/** ZSTD_getFrameContentSize() :
	* compatible with legacy mode
	* @return : decompressed size of the single frame pointed to be `src` if known, otherwise
	* - ZSTD_CONTENTSIZE_UNKNOWN if the size cannot be determined
	* - ZSTD_CONTENTSIZE_ERROR if an error occurred (e.g. invalid magic number, srcSize too small) */
	unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize)
	{
	#if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
	if (ZSTD_isLegacy(src, srcSize)) {
	unsigned long long const ret = ZSTD_getDecompressedSize_legacy(src, srcSize);
	return ret == 0 ? ZSTD_CONTENTSIZE_UNKNOWN : ret;
	}
	#endif
	{ ZSTD_frameHeader zfh;
	if (ZSTD_getFrameHeader(&zfh, src, srcSize) != 0)
	return ZSTD_CONTENTSIZE_ERROR;
	if (zfh.frameType == ZSTD_skippableFrame) {
	return 0;
	} else {
	return zfh.frameContentSize;
	} }
	}

	static size_t readSkippableFrameSize(void const* src, size_t srcSize)
	{
	size_t const skippableHeaderSize = ZSTD_SKIPPABLEHEADERSIZE;
	U32 sizeU32;

	RETURN_ERROR_IF(srcSize < ZSTD_SKIPPABLEHEADERSIZE, srcSize_wrong, "");

	sizeU32 = MEM_readLE32((BYTE const*)src + ZSTD_FRAMEIDSIZE);
	RETURN_ERROR_IF((U32)(sizeU32 + ZSTD_SKIPPABLEHEADERSIZE) < sizeU32,
	frameParameter_unsupported, "");
	{
	size_t const skippableSize = skippableHeaderSize + sizeU32;
	RETURN_ERROR_IF(skippableSize > srcSize, srcSize_wrong, "");
	return skippableSize;
	}
	}

	/** ZSTD_findDecompressedSize() :
	* compatible with legacy mode
	* `srcSize` must be the exact length of some number of ZSTD compressed and/or
	* skippable frames
	* @return : decompressed size of the frames contained */
	unsigned long long ZSTD_findDecompressedSize(const void* src, size_t srcSize)
	{
	unsigned long long totalDstSize = 0;

	while (srcSize >= ZSTD_startingInputLength(ZSTD_f_zstd1)) {
	U32 const magicNumber = MEM_readLE32(src);

	if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
	size_t const skippableSize = readSkippableFrameSize(src, srcSize);
	if (ZSTD_isError(skippableSize)) {
	return ZSTD_CONTENTSIZE_ERROR;
	}
	assert(skippableSize <= srcSize);

	src = (const BYTE *)src + skippableSize;
	srcSize -= skippableSize;
	continue;
	}

	{ unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
	if (ret >= ZSTD_CONTENTSIZE_ERROR) return ret;

	/* check for overflow */
	if (totalDstSize + ret < totalDstSize) return ZSTD_CONTENTSIZE_ERROR;
	totalDstSize += ret;
	}
	{ size_t const frameSrcSize = ZSTD_findFrameCompressedSize(src, srcSize);
	if (ZSTD_isError(frameSrcSize)) {
	return ZSTD_CONTENTSIZE_ERROR;
	}

	src = (const BYTE *)src + frameSrcSize;
	srcSize -= frameSrcSize;
	}
	} /* while (srcSize >= ZSTD_frameHeaderSize_prefix) */

	if (srcSize) return ZSTD_CONTENTSIZE_ERROR;

	return totalDstSize;
	}

	/** ZSTD_getDecompressedSize() :
	* compatible with legacy mode
	* @return : decompressed size if known, 0 otherwise
	note : 0 can mean any of the following :
	- frame content is empty
	- decompressed size field is not present in frame header
	- frame header unknown / not supported
	- frame header not complete (`srcSize` too small) */
	unsigned long long ZSTD_getDecompressedSize(const void* src, size_t srcSize)
	{
	unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
	ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_ERROR < ZSTD_CONTENTSIZE_UNKNOWN);
	return (ret >= ZSTD_CONTENTSIZE_ERROR) ? 0 : ret;
	}


	/** ZSTD_decodeFrameHeader() :
	* `headerSize` must be the size provided by ZSTD_frameHeaderSize().
	* @return : 0 if success, or an error code, which can be tested using ZSTD_isError() */
	static size_t ZSTD_decodeFrameHeader(ZSTD_DCtx* dctx, const void* src, size_t headerSize)
	{
	size_t const result = ZSTD_getFrameHeader_advanced(&(dctx->fParams), src, headerSize, dctx->format);
	if (ZSTD_isError(result)) return result; /* invalid header */
	RETURN_ERROR_IF(result>0, srcSize_wrong, "headerSize too small");
	#ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
	/* Skip the dictID check in fuzzing mode, because it makes the search
	* harder.
	*/
	RETURN_ERROR_IF(dctx->fParams.dictID && (dctx->dictID != dctx->fParams.dictID),
	dictionary_wrong, "");
	#endif
	if (dctx->fParams.checksumFlag) XXH64_reset(&dctx->xxhState, 0);
	return 0;
	}

	static ZSTD_frameSizeInfo ZSTD_errorFrameSizeInfo(size_t ret)
	{
	ZSTD_frameSizeInfo frameSizeInfo;
	frameSizeInfo.compressedSize = ret;
	frameSizeInfo.decompressedBound = ZSTD_CONTENTSIZE_ERROR;
	return frameSizeInfo;
	}

	static ZSTD_frameSizeInfo ZSTD_findFrameSizeInfo(const void* src, size_t srcSize)
	{
	ZSTD_frameSizeInfo frameSizeInfo;
	memset(&frameSizeInfo, 0, sizeof(ZSTD_frameSizeInfo));

	#if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
	if (ZSTD_isLegacy(src, srcSize))
	return ZSTD_findFrameSizeInfoLegacy(src, srcSize);
	#endif

	if ((srcSize >= ZSTD_SKIPPABLEHEADERSIZE)
	&& (MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
	frameSizeInfo.compressedSize = readSkippableFrameSize(src, srcSize);
	assert(ZSTD_isError(frameSizeInfo.compressedSize) \|\|
	frameSizeInfo.compressedSize <= srcSize);
	return frameSizeInfo;
	} else {
	const BYTE* ip = (const BYTE*)src;
	const BYTE* const ipstart = ip;
	size_t remainingSize = srcSize;
	size_t nbBlocks = 0;
	ZSTD_frameHeader zfh;

	/* Extract Frame Header */
	{ size_t const ret = ZSTD_getFrameHeader(&zfh, src, srcSize);
	if (ZSTD_isError(ret))
	return ZSTD_errorFrameSizeInfo(ret);
	if (ret > 0)
	return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));
	}

	ip += zfh.headerSize;
	remainingSize -= zfh.headerSize;

	/* Iterate over each block */
	while (1) {
	blockProperties_t blockProperties;
	size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSize, &blockProperties);
	if (ZSTD_isError(cBlockSize))
	return ZSTD_errorFrameSizeInfo(cBlockSize);

	if (ZSTD_blockHeaderSize + cBlockSize > remainingSize)
	return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));

	ip += ZSTD_blockHeaderSize + cBlockSize;
	remainingSize -= ZSTD_blockHeaderSize + cBlockSize;
	nbBlocks++;

	if (blockProperties.lastBlock) break;
	}

	/* Final frame content checksum */
	if (zfh.checksumFlag) {
	if (remainingSize < 4)
	return ZSTD_errorFrameSizeInfo(ERROR(srcSize_wrong));
	ip += 4;
	}

	frameSizeInfo.compressedSize = ip - ipstart;
	frameSizeInfo.decompressedBound = (zfh.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN)
	? zfh.frameContentSize
	: nbBlocks * zfh.blockSizeMax;
	return frameSizeInfo;
	}
	}

	/** ZSTD_findFrameCompressedSize() :
	* compatible with legacy mode
	* `src` must point to the start of a ZSTD frame, ZSTD legacy frame, or skippable frame
	* `srcSize` must be at least as large as the frame contained
	* @return : the compressed size of the frame starting at `src` */
	size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize)
	{
	ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
	return frameSizeInfo.compressedSize;
	}

	/** ZSTD_decompressBound() :
	* compatible with legacy mode
	* `src` must point to the start of a ZSTD frame or a skippeable frame
	* `srcSize` must be at least as large as the frame contained
	* @return : the maximum decompressed size of the compressed source
	*/
	unsigned long long ZSTD_decompressBound(const void* src, size_t srcSize)
	{
	unsigned long long bound = 0;
	/* Iterate over each frame */
	while (srcSize > 0) {
	ZSTD_frameSizeInfo const frameSizeInfo = ZSTD_findFrameSizeInfo(src, srcSize);
	size_t const compressedSize = frameSizeInfo.compressedSize;
	unsigned long long const decompressedBound = frameSizeInfo.decompressedBound;
	if (ZSTD_isError(compressedSize) \|\| decompressedBound == ZSTD_CONTENTSIZE_ERROR)
	return ZSTD_CONTENTSIZE_ERROR;
	assert(srcSize >= compressedSize);
	src = (const BYTE*)src + compressedSize;
	srcSize -= compressedSize;
	bound += decompressedBound;
	}
	return bound;
	}


	/-************************************************************
	* Frame decoding
	***************************************************************/

	/** ZSTD_insertBlock() :
	* insert `src` block into `dctx` history. Useful to track uncompressed blocks. */
	size_t ZSTD_insertBlock(ZSTD_DCtx* dctx, const void* blockStart, size_t blockSize)
	{
	DEBUGLOG(5, "ZSTD_insertBlock: %u bytes", (unsigned)blockSize);
	ZSTD_checkContinuity(dctx, blockStart);
	dctx->previousDstEnd = (const char*)blockStart + blockSize;
	return blockSize;
	}


	static size_t ZSTD_copyRawBlock(void* dst, size_t dstCapacity,
	const void* src, size_t srcSize)
	{
	DEBUGLOG(5, "ZSTD_copyRawBlock");
	if (dst == NULL) {
	if (srcSize == 0) return 0;
	RETURN_ERROR(dstBuffer_null, "");
	}
	RETURN_ERROR_IF(srcSize > dstCapacity, dstSize_tooSmall, "");
	memcpy(dst, src, srcSize);
	return srcSize;
	}

	static size_t ZSTD_setRleBlock(void* dst, size_t dstCapacity,
	BYTE b,
	size_t regenSize)
	{
	if (dst == NULL) {
	if (regenSize == 0) return 0;
	RETURN_ERROR(dstBuffer_null, "");
	}
	RETURN_ERROR_IF(regenSize > dstCapacity, dstSize_tooSmall, "");
	memset(dst, b, regenSize);
	return regenSize;
	}


	/*! ZSTD_decompressFrame() :
	* @dctx must be properly initialized
	* will update srcPtr and srcSizePtr,
	* to make srcPtr progress by one frame. /
	static size_t ZSTD_decompressFrame(ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void** srcPtr, size_t *srcSizePtr)
	{
	const BYTE* ip = (const BYTE)(srcPtr);
	BYTE* const ostart = (BYTE* const)dst;
	BYTE* const oend = dstCapacity != 0 ? ostart + dstCapacity : ostart;
	BYTE* op = ostart;
	size_t remainingSrcSize = *srcSizePtr;

	DEBUGLOG(4, "ZSTD_decompressFrame (srcSize:%i)", (int)*srcSizePtr);

	/* check */
	RETURN_ERROR_IF(
	remainingSrcSize < ZSTD_FRAMEHEADERSIZE_MIN(dctx->format)+ZSTD_blockHeaderSize,
	srcSize_wrong, "");

	/* Frame Header */
	{ size_t const frameHeaderSize = ZSTD_frameHeaderSize_internal(
	ip, ZSTD_FRAMEHEADERSIZE_PREFIX(dctx->format), dctx->format);
	if (ZSTD_isError(frameHeaderSize)) return frameHeaderSize;
	RETURN_ERROR_IF(remainingSrcSize < frameHeaderSize+ZSTD_blockHeaderSize,
	srcSize_wrong, "");
	FORWARD_IF_ERROR( ZSTD_decodeFrameHeader(dctx, ip, frameHeaderSize) , "");
	ip += frameHeaderSize; remainingSrcSize -= frameHeaderSize;
	}

	/* Loop on each block */
	while (1) {
	size_t decodedSize;
	blockProperties_t blockProperties;
	size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSrcSize, &blockProperties);
	if (ZSTD_isError(cBlockSize)) return cBlockSize;

	ip += ZSTD_blockHeaderSize;
	remainingSrcSize -= ZSTD_blockHeaderSize;
	RETURN_ERROR_IF(cBlockSize > remainingSrcSize, srcSize_wrong, "");

	switch(blockProperties.blockType)
	{
	case bt_compressed:
	decodedSize = ZSTD_decompressBlock_internal(dctx, op, oend-op, ip, cBlockSize, /* frame */ 1);
	break;
	case bt_raw :
	decodedSize = ZSTD_copyRawBlock(op, oend-op, ip, cBlockSize);
	break;
	case bt_rle :
	decodedSize = ZSTD_setRleBlock(op, oend-op, *ip, blockProperties.origSize);
	break;
	case bt_reserved :
	default:
	RETURN_ERROR(corruption_detected, "invalid block type");
	}

	if (ZSTD_isError(decodedSize)) return decodedSize;
	if (dctx->fParams.checksumFlag)
	XXH64_update(&dctx->xxhState, op, decodedSize);
	if (decodedSize != 0)
	op += decodedSize;
	assert(ip != NULL);
	ip += cBlockSize;
	remainingSrcSize -= cBlockSize;
	if (blockProperties.lastBlock) break;
	}

	if (dctx->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN) {
	RETURN_ERROR_IF((U64)(op-ostart) != dctx->fParams.frameContentSize,
	corruption_detected, "");
	}
	if (dctx->fParams.checksumFlag) { /* Frame content checksum verification */
	U32 const checkCalc = (U32)XXH64_digest(&dctx->xxhState);
	U32 checkRead;
	RETURN_ERROR_IF(remainingSrcSize<4, checksum_wrong, "");
	checkRead = MEM_readLE32(ip);
	RETURN_ERROR_IF(checkRead != checkCalc, checksum_wrong, "");
	ip += 4;
	remainingSrcSize -= 4;
	}

	/* Allow caller to get size read */
	*srcPtr = ip;
	*srcSizePtr = remainingSrcSize;
	return op-ostart;
	}

	static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict, size_t dictSize,
	const ZSTD_DDict* ddict)
	{
	void* const dststart = dst;
	int moreThan1Frame = 0;

	DEBUGLOG(5, "ZSTD_decompressMultiFrame");
	assert(dict==NULL \|\| ddict==NULL); /* either dict or ddict set, not both */

	if (ddict) {
	dict = ZSTD_DDict_dictContent(ddict);
	dictSize = ZSTD_DDict_dictSize(ddict);
	}

	while (srcSize >= ZSTD_startingInputLength(dctx->format)) {

	#if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT >= 1)
	if (ZSTD_isLegacy(src, srcSize)) {
	size_t decodedSize;
	size_t const frameSize = ZSTD_findFrameCompressedSizeLegacy(src, srcSize);
	if (ZSTD_isError(frameSize)) return frameSize;
	RETURN_ERROR_IF(dctx->staticSize, memory_allocation,
	"legacy support is not compatible with static dctx");

	decodedSize = ZSTD_decompressLegacy(dst, dstCapacity, src, frameSize, dict, dictSize);
	if (ZSTD_isError(decodedSize)) return decodedSize;

	assert(decodedSize <=- dstCapacity);
	dst = (BYTE*)dst + decodedSize;
	dstCapacity -= decodedSize;

	src = (const BYTE*)src + frameSize;
	srcSize -= frameSize;

	continue;
	}
	#endif

	{ U32 const magicNumber = MEM_readLE32(src);
	DEBUGLOG(4, "reading magic number %08X (expecting %08X)",
	(unsigned)magicNumber, ZSTD_MAGICNUMBER);
	if ((magicNumber & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) {
	size_t const skippableSize = readSkippableFrameSize(src, srcSize);
	FORWARD_IF_ERROR(skippableSize, "readSkippableFrameSize failed");
	assert(skippableSize <= srcSize);

	src = (const BYTE *)src + skippableSize;
	srcSize -= skippableSize;
	continue;
	} }

	if (ddict) {
	/* we were called from ZSTD_decompress_usingDDict */
	FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDDict(dctx, ddict), "");
	} else {
	/* this will initialize correctly with no dict if dict == NULL, so
	* use this in all cases but ddict */
	FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDict(dctx, dict, dictSize), "");
	}
	ZSTD_checkContinuity(dctx, dst);

	{ const size_t res = ZSTD_decompressFrame(dctx, dst, dstCapacity,
	&src, &srcSize);
	RETURN_ERROR_IF(
	(ZSTD_getErrorCode(res) == ZSTD_error_prefix_unknown)
	&& (moreThan1Frame==1),
	srcSize_wrong,
	"at least one frame successfully completed, but following "
	"bytes are garbage: it's more likely to be a srcSize error, "
	"specifying more bytes than compressed size of frame(s). This "
	"error message replaces ERROR(prefix_unknown), which would be "
	"confusing, as the first header is actually correct. Note that "
	"one could be unlucky, it might be a corruption error instead, "
	"happening right at the place where we expect zstd magic "
	"bytes. But this is _much_ less likely than a srcSize field "
	"error.");
	if (ZSTD_isError(res)) return res;
	assert(res <= dstCapacity);
	if (res != 0)
	dst = (BYTE*)dst + res;
	dstCapacity -= res;
	}
	moreThan1Frame = 1;
	} /* while (srcSize >= ZSTD_frameHeaderSize_prefix) */

	RETURN_ERROR_IF(srcSize, srcSize_wrong, "input not entirely consumed");

	return (BYTE)dst - (BYTE)dststart;
	}

	size_t ZSTD_decompress_usingDict(ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const void* dict, size_t dictSize)
	{
	return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize, dict, dictSize, NULL);
	}


	static ZSTD_DDict const* ZSTD_getDDict(ZSTD_DCtx* dctx)
	{
	switch (dctx->dictUses) {
	default:
	assert(0 /* Impossible */);
	/* fall-through */
	case ZSTD_dont_use:
	ZSTD_clearDict(dctx);
	return NULL;
	case ZSTD_use_indefinitely:
	return dctx->ddict;
	case ZSTD_use_once:
	dctx->dictUses = ZSTD_dont_use;
	return dctx->ddict;
	}
	}

	size_t ZSTD_decompressDCtx(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize)
	{
	return ZSTD_decompress_usingDDict(dctx, dst, dstCapacity, src, srcSize, ZSTD_getDDict(dctx));
	}


	size_t ZSTD_decompress(void* dst, size_t dstCapacity, const void* src, size_t srcSize)
	{
	#if defined(ZSTD_HEAPMODE) && (ZSTD_HEAPMODE>=1)
	size_t regenSize;
	ZSTD_DCtx* const dctx = ZSTD_createDCtx();
	RETURN_ERROR_IF(dctx==NULL, memory_allocation, "NULL pointer!");
	regenSize = ZSTD_decompressDCtx(dctx, dst, dstCapacity, src, srcSize);
	ZSTD_freeDCtx(dctx);
	return regenSize;
	#else /* stack mode */
	ZSTD_DCtx dctx;
	ZSTD_initDCtx_internal(&dctx);
	return ZSTD_decompressDCtx(&dctx, dst, dstCapacity, src, srcSize);
	#endif
	}


	/-*************************************
	* Advanced Streaming Decompression API
	* Bufferless and synchronous
	****************************************/
	size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx* dctx) { return dctx->expected; }

	/**
	* Similar to ZSTD_nextSrcSizeToDecompress(), but when when a block input can be streamed,
	* we allow taking a partial block as the input. Currently only raw uncompressed blocks can
	* be streamed.
	*
	* For blocks that can be streamed, this allows us to reduce the latency until we produce
	* output, and avoid copying the input.
	*
	* @param inputSize - The total amount of input that the caller currently has.
	*/
	static size_t ZSTD_nextSrcSizeToDecompressWithInputSize(ZSTD_DCtx* dctx, size_t inputSize) {
	if (!(dctx->stage == ZSTDds_decompressBlock \|\| dctx->stage == ZSTDds_decompressLastBlock))
	return dctx->expected;
	if (dctx->bType != bt_raw)
	return dctx->expected;
	return MIN(MAX(inputSize, 1), dctx->expected);
	}

	ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx* dctx) {
	switch(dctx->stage)
	{
	default: /* should not happen */
	assert(0);
	case ZSTDds_getFrameHeaderSize:
	case ZSTDds_decodeFrameHeader:
	return ZSTDnit_frameHeader;
	case ZSTDds_decodeBlockHeader:
	return ZSTDnit_blockHeader;
	case ZSTDds_decompressBlock:
	return ZSTDnit_block;
	case ZSTDds_decompressLastBlock:
	return ZSTDnit_lastBlock;
	case ZSTDds_checkChecksum:
	return ZSTDnit_checksum;
	case ZSTDds_decodeSkippableHeader:
	case ZSTDds_skipFrame:
	return ZSTDnit_skippableFrame;
	}
	}

	static int ZSTD_isSkipFrame(ZSTD_DCtx* dctx) { return dctx->stage == ZSTDds_skipFrame; }

	/** ZSTD_decompressContinue() :
	* srcSize : must be the exact nb of bytes expected (see ZSTD_nextSrcSizeToDecompress())
	* @return : nb of bytes generated into `dst` (necessarily <= `dstCapacity)
	* or an error code, which can be tested using ZSTD_isError() */
	size_t ZSTD_decompressContinue(ZSTD_DCtx* dctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize)
	{
	DEBUGLOG(5, "ZSTD_decompressContinue (srcSize:%u)", (unsigned)srcSize);
	/* Sanity check */
	RETURN_ERROR_IF(srcSize != ZSTD_nextSrcSizeToDecompressWithInputSize(dctx, srcSize), srcSize_wrong, "not allowed");
	if (dstCapacity) ZSTD_checkContinuity(dctx, dst);

	switch (dctx->stage)
	{
	case ZSTDds_getFrameHeaderSize :
	assert(src != NULL);
	if (dctx->format == ZSTD_f_zstd1) { /* allows header */
	assert(srcSize >= ZSTD_FRAMEIDSIZE); /* to read skippable magic number */
	if ((MEM_readLE32(src) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) { /* skippable frame */
	memcpy(dctx->headerBuffer, src, srcSize);
	dctx->expected = ZSTD_SKIPPABLEHEADERSIZE - srcSize; /* remaining to load to get full skippable frame header */
	dctx->stage = ZSTDds_decodeSkippableHeader;
	return 0;
	} }
	dctx->headerSize = ZSTD_frameHeaderSize_internal(src, srcSize, dctx->format);
	if (ZSTD_isError(dctx->headerSize)) return dctx->headerSize;
	memcpy(dctx->headerBuffer, src, srcSize);
	dctx->expected = dctx->headerSize - srcSize;
	dctx->stage = ZSTDds_decodeFrameHeader;
	return 0;

	case ZSTDds_decodeFrameHeader:
	assert(src != NULL);
	memcpy(dctx->headerBuffer + (dctx->headerSize - srcSize), src, srcSize);
	FORWARD_IF_ERROR(ZSTD_decodeFrameHeader(dctx, dctx->headerBuffer, dctx->headerSize), "");
	dctx->expected = ZSTD_blockHeaderSize;
	dctx->stage = ZSTDds_decodeBlockHeader;
	return 0;

	case ZSTDds_decodeBlockHeader:
	{ blockProperties_t bp;
	size_t const cBlockSize = ZSTD_getcBlockSize(src, ZSTD_blockHeaderSize, &bp);
	if (ZSTD_isError(cBlockSize)) return cBlockSize;
	RETURN_ERROR_IF(cBlockSize > dctx->fParams.blockSizeMax, corruption_detected, "Block Size Exceeds Maximum");
	dctx->expected = cBlockSize;
	dctx->bType = bp.blockType;
	dctx->rleSize = bp.origSize;
	if (cBlockSize) {
	dctx->stage = bp.lastBlock ? ZSTDds_decompressLastBlock : ZSTDds_decompressBlock;
	return 0;
	}
	/* empty block */
	if (bp.lastBlock) {
	if (dctx->fParams.checksumFlag) {
	dctx->expected = 4;
	dctx->stage = ZSTDds_checkChecksum;
	} else {
	dctx->expected = 0; /* end of frame */
	dctx->stage = ZSTDds_getFrameHeaderSize;
	}
	} else {
	dctx->expected = ZSTD_blockHeaderSize; /* jump to next header */
	dctx->stage = ZSTDds_decodeBlockHeader;
	}
	return 0;
	}

	case ZSTDds_decompressLastBlock:
	case ZSTDds_decompressBlock:
	DEBUGLOG(5, "ZSTD_decompressContinue: case ZSTDds_decompressBlock");
	{ size_t rSize;
	switch(dctx->bType)
	{
	case bt_compressed:
	DEBUGLOG(5, "ZSTD_decompressContinue: case bt_compressed");
	rSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize, /* frame */ 1);
	dctx->expected = 0; /* Streaming not supported */
	break;
	case bt_raw :
	assert(srcSize <= dctx->expected);
	rSize = ZSTD_copyRawBlock(dst, dstCapacity, src, srcSize);
	FORWARD_IF_ERROR(rSize, "ZSTD_copyRawBlock failed");
	assert(rSize == srcSize);
	dctx->expected -= rSize;
	break;
	case bt_rle :
	rSize = ZSTD_setRleBlock(dst, dstCapacity, (const BYTE)src, dctx->rleSize);
	dctx->expected = 0; /* Streaming not supported */
	break;
	case bt_reserved : /* should never happen */
	default:
	RETURN_ERROR(corruption_detected, "invalid block type");
	}
	FORWARD_IF_ERROR(rSize, "");
	RETURN_ERROR_IF(rSize > dctx->fParams.blockSizeMax, corruption_detected, "Decompressed Block Size Exceeds Maximum");
	DEBUGLOG(5, "ZSTD_decompressContinue: decoded size from block : %u", (unsigned)rSize);
	dctx->decodedSize += rSize;
	if (dctx->fParams.checksumFlag) XXH64_update(&dctx->xxhState, dst, rSize);
	dctx->previousDstEnd = (char*)dst + rSize;

	/* Stay on the same stage until we are finished streaming the block. */
	if (dctx->expected > 0) {
	return rSize;
	}

	if (dctx->stage == ZSTDds_decompressLastBlock) { /* end of frame */
	DEBUGLOG(4, "ZSTD_decompressContinue: decoded size from frame : %u", (unsigned)dctx->decodedSize);
	RETURN_ERROR_IF(
	dctx->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
	&& dctx->decodedSize != dctx->fParams.frameContentSize,
	corruption_detected, "");
	if (dctx->fParams.checksumFlag) { /* another round for frame checksum */
	dctx->expected = 4;
	dctx->stage = ZSTDds_checkChecksum;
	} else {
	dctx->expected = 0; /* ends here */
	dctx->stage = ZSTDds_getFrameHeaderSize;
	}
	} else {
	dctx->stage = ZSTDds_decodeBlockHeader;
	dctx->expected = ZSTD_blockHeaderSize;
	}
	return rSize;
	}

	case ZSTDds_checkChecksum:
	assert(srcSize == 4); /* guaranteed by dctx->expected */
	{ U32 const h32 = (U32)XXH64_digest(&dctx->xxhState);
	U32 const check32 = MEM_readLE32(src);
	DEBUGLOG(4, "ZSTD_decompressContinue: checksum : calculated %08X :: %08X read", (unsigned)h32, (unsigned)check32);
	RETURN_ERROR_IF(check32 != h32, checksum_wrong, "");
	dctx->expected = 0;
	dctx->stage = ZSTDds_getFrameHeaderSize;
	return 0;
	}

	case ZSTDds_decodeSkippableHeader:
	assert(src != NULL);
	assert(srcSize <= ZSTD_SKIPPABLEHEADERSIZE);
	memcpy(dctx->headerBuffer + (ZSTD_SKIPPABLEHEADERSIZE - srcSize), src, srcSize); /* complete skippable header */
	dctx->expected = MEM_readLE32(dctx->headerBuffer + ZSTD_FRAMEIDSIZE); /* note : dctx->expected can grow seriously large, beyond local buffer size */
	dctx->stage = ZSTDds_skipFrame;
	return 0;

	case ZSTDds_skipFrame:
	dctx->expected = 0;
	dctx->stage = ZSTDds_getFrameHeaderSize;
	return 0;

	default:
	assert(0); /* impossible */
	RETURN_ERROR(GENERIC, "impossible to reach"); /* some compiler require default to do something */
	}
	}


	static size_t ZSTD_refDictContent(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
	{
	dctx->dictEnd = dctx->previousDstEnd;
	dctx->virtualStart = (const char)dict - ((const char)(dctx->previousDstEnd) - (const char*)(dctx->prefixStart));
	dctx->prefixStart = dict;
	dctx->previousDstEnd = (const char*)dict + dictSize;
	#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
	dctx->dictContentBeginForFuzzing = dctx->prefixStart;
	dctx->dictContentEndForFuzzing = dctx->previousDstEnd;
	#endif
	return 0;
	}

	/*! ZSTD_loadDEntropy() :
	* dict : must point at beginning of a valid zstd dictionary.
	* @return : size of entropy tables read */
	size_t
	ZSTD_loadDEntropy(ZSTD_entropyDTables_t* entropy,
	const void* const dict, size_t const dictSize)
	{
	const BYTE* dictPtr = (const BYTE*)dict;
	const BYTE* const dictEnd = dictPtr + dictSize;

	RETURN_ERROR_IF(dictSize <= 8, dictionary_corrupted, "dict is too small");
	assert(MEM_readLE32(dict) == ZSTD_MAGIC_DICTIONARY); /* dict must be valid */
	dictPtr += 8; /* skip header = magic + dictID */

	ZSTD_STATIC_ASSERT(offsetof(ZSTD_entropyDTables_t, OFTable) == offsetof(ZSTD_entropyDTables_t, LLTable) + sizeof(entropy->LLTable));
	ZSTD_STATIC_ASSERT(offsetof(ZSTD_entropyDTables_t, MLTable) == offsetof(ZSTD_entropyDTables_t, OFTable) + sizeof(entropy->OFTable));
	ZSTD_STATIC_ASSERT(sizeof(entropy->LLTable) + sizeof(entropy->OFTable) + sizeof(entropy->MLTable) >= HUF_DECOMPRESS_WORKSPACE_SIZE);
	{ void* const workspace = &entropy->LLTable; /* use fse tables as temporary workspace; implies fse tables are grouped together */
	size_t const workspaceSize = sizeof(entropy->LLTable) + sizeof(entropy->OFTable) + sizeof(entropy->MLTable);
	#ifdef HUF_FORCE_DECOMPRESS_X1
	/* in minimal huffman, we always use X1 variants */
	size_t const hSize = HUF_readDTableX1_wksp(entropy->hufTable,
	dictPtr, dictEnd - dictPtr,
	workspace, workspaceSize);
	#else
	size_t const hSize = HUF_readDTableX2_wksp(entropy->hufTable,
	dictPtr, dictEnd - dictPtr,
	workspace, workspaceSize);
	#endif
	RETURN_ERROR_IF(HUF_isError(hSize), dictionary_corrupted, "");
	dictPtr += hSize;
	}

	{ short offcodeNCount[MaxOff+1];
	unsigned offcodeMaxValue = MaxOff, offcodeLog;
	size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dictPtr, dictEnd-dictPtr);
	RETURN_ERROR_IF(FSE_isError(offcodeHeaderSize), dictionary_corrupted, "");
	RETURN_ERROR_IF(offcodeMaxValue > MaxOff, dictionary_corrupted, "");
	RETURN_ERROR_IF(offcodeLog > OffFSELog, dictionary_corrupted, "");
	ZSTD_buildFSETable( entropy->OFTable,
	offcodeNCount, offcodeMaxValue,
	OF_base, OF_bits,
	offcodeLog);
	dictPtr += offcodeHeaderSize;
	}

	{ short matchlengthNCount[MaxML+1];
	unsigned matchlengthMaxValue = MaxML, matchlengthLog;
	size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, dictEnd-dictPtr);
	RETURN_ERROR_IF(FSE_isError(matchlengthHeaderSize), dictionary_corrupted, "");
	RETURN_ERROR_IF(matchlengthMaxValue > MaxML, dictionary_corrupted, "");
	RETURN_ERROR_IF(matchlengthLog > MLFSELog, dictionary_corrupted, "");
	ZSTD_buildFSETable( entropy->MLTable,
	matchlengthNCount, matchlengthMaxValue,
	ML_base, ML_bits,
	matchlengthLog);
	dictPtr += matchlengthHeaderSize;
	}

	{ short litlengthNCount[MaxLL+1];
	unsigned litlengthMaxValue = MaxLL, litlengthLog;
	size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, dictEnd-dictPtr);
	RETURN_ERROR_IF(FSE_isError(litlengthHeaderSize), dictionary_corrupted, "");
	RETURN_ERROR_IF(litlengthMaxValue > MaxLL, dictionary_corrupted, "");
	RETURN_ERROR_IF(litlengthLog > LLFSELog, dictionary_corrupted, "");
	ZSTD_buildFSETable( entropy->LLTable,
	litlengthNCount, litlengthMaxValue,
	LL_base, LL_bits,
	litlengthLog);
	dictPtr += litlengthHeaderSize;
	}

	RETURN_ERROR_IF(dictPtr+12 > dictEnd, dictionary_corrupted, "");
	{ int i;
	size_t const dictContentSize = (size_t)(dictEnd - (dictPtr+12));
	for (i=0; i<3; i++) {
	U32 const rep = MEM_readLE32(dictPtr); dictPtr += 4;
	RETURN_ERROR_IF(rep==0 \|\| rep > dictContentSize,
	dictionary_corrupted, "");
	entropy->rep[i] = rep;
	} }

	return dictPtr - (const BYTE*)dict;
	}

	static size_t ZSTD_decompress_insertDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
	{
	if (dictSize < 8) return ZSTD_refDictContent(dctx, dict, dictSize);
	{ U32 const magic = MEM_readLE32(dict);
	if (magic != ZSTD_MAGIC_DICTIONARY) {
	return ZSTD_refDictContent(dctx, dict, dictSize); /* pure content mode */
	} }
	dctx->dictID = MEM_readLE32((const char*)dict + ZSTD_FRAMEIDSIZE);

	/* load entropy tables */
	{ size_t const eSize = ZSTD_loadDEntropy(&dctx->entropy, dict, dictSize);
	RETURN_ERROR_IF(ZSTD_isError(eSize), dictionary_corrupted, "");
	dict = (const char*)dict + eSize;
	dictSize -= eSize;
	}
	dctx->litEntropy = dctx->fseEntropy = 1;

	/* reference dictionary content */
	return ZSTD_refDictContent(dctx, dict, dictSize);
	}

	size_t ZSTD_decompressBegin(ZSTD_DCtx* dctx)
	{
	assert(dctx != NULL);
	dctx->expected = ZSTD_startingInputLength(dctx->format); /* dctx->format must be properly set */
	dctx->stage = ZSTDds_getFrameHeaderSize;
	dctx->decodedSize = 0;
	dctx->previousDstEnd = NULL;
	dctx->prefixStart = NULL;
	dctx->virtualStart = NULL;
	dctx->dictEnd = NULL;
	dctx->entropy.hufTable[0] = (HUF_DTable)((HufLog)0x1000001); / cover both little and big endian */
	dctx->litEntropy = dctx->fseEntropy = 0;
	dctx->dictID = 0;
	dctx->bType = bt_reserved;
	ZSTD_STATIC_ASSERT(sizeof(dctx->entropy.rep) == sizeof(repStartValue));
	memcpy(dctx->entropy.rep, repStartValue, sizeof(repStartValue)); /* initial repcodes */
	dctx->LLTptr = dctx->entropy.LLTable;
	dctx->MLTptr = dctx->entropy.MLTable;
	dctx->OFTptr = dctx->entropy.OFTable;
	dctx->HUFptr = dctx->entropy.hufTable;
	return 0;
	}

	size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
	{
	FORWARD_IF_ERROR( ZSTD_decompressBegin(dctx) , "");
	if (dict && dictSize)
	RETURN_ERROR_IF(
	ZSTD_isError(ZSTD_decompress_insertDictionary(dctx, dict, dictSize)),
	dictionary_corrupted, "");
	return 0;
	}


	/* ====== ZSTD_DDict ====== */

	size_t ZSTD_decompressBegin_usingDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
	{
	DEBUGLOG(4, "ZSTD_decompressBegin_usingDDict");
	assert(dctx != NULL);
	if (ddict) {
	const char* const dictStart = (const char*)ZSTD_DDict_dictContent(ddict);
	size_t const dictSize = ZSTD_DDict_dictSize(ddict);
	const void* const dictEnd = dictStart + dictSize;
	dctx->ddictIsCold = (dctx->dictEnd != dictEnd);
	DEBUGLOG(4, "DDict is %s",
	dctx->ddictIsCold ? "~cold~" : "hot!");
	}
	FORWARD_IF_ERROR( ZSTD_decompressBegin(dctx) , "");
	if (ddict) { /* NULL ddict is equivalent to no dictionary */
	ZSTD_copyDDictParameters(dctx, ddict);
	}
	return 0;
	}

	/*! ZSTD_getDictID_fromDict() :
	* Provides the dictID stored within dictionary.
	* if @return == 0, the dictionary is not conformant with Zstandard specification.
	* It can still be loaded, but as a content-only dictionary. */
	unsigned ZSTD_getDictID_fromDict(const void* dict, size_t dictSize)
	{
	if (dictSize < 8) return 0;
	if (MEM_readLE32(dict) != ZSTD_MAGIC_DICTIONARY) return 0;
	return MEM_readLE32((const char*)dict + ZSTD_FRAMEIDSIZE);
	}

	/*! ZSTD_getDictID_fromFrame() :
	* Provides the dictID required to decompress frame stored within `src`.
	* If @return == 0, the dictID could not be decoded.
	* This could for one of the following reasons :
	* - The frame does not require a dictionary (most common case).
	* - The frame was built with dictID intentionally removed.
	* Needed dictionary is a hidden information.
	* Note : this use case also happens when using a non-conformant dictionary.
	* - `srcSize` is too small, and as a result, frame header could not be decoded.
	* Note : possible if `srcSize < ZSTD_FRAMEHEADERSIZE_MAX`.
	* - This is not a Zstandard frame.
	* When identifying the exact failure cause, it's possible to use
	* ZSTD_getFrameHeader(), which will provide a more precise error code. */
	unsigned ZSTD_getDictID_fromFrame(const void* src, size_t srcSize)
	{
	ZSTD_frameHeader zfp = { 0, 0, 0, ZSTD_frame, 0, 0, 0 };
	size_t const hError = ZSTD_getFrameHeader(&zfp, src, srcSize);
	if (ZSTD_isError(hError)) return 0;
	return zfp.dictID;
	}


	/*! ZSTD_decompress_usingDDict() :
	* Decompression using a pre-digested Dictionary
	* Use dictionary without significant overhead. */
	size_t ZSTD_decompress_usingDDict(ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize,
	const ZSTD_DDict* ddict)
	{
	/* pass content and size in case legacy frames are encountered */
	return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize,
	NULL, 0,
	ddict);
	}


	/*=====================================
	* Streaming decompression
	====================================/

	ZSTD_DStream* ZSTD_createDStream(void)
	{
	DEBUGLOG(3, "ZSTD_createDStream");
	return ZSTD_createDStream_advanced(ZSTD_defaultCMem);
	}

	ZSTD_DStream* ZSTD_initStaticDStream(void *workspace, size_t workspaceSize)
	{
	return ZSTD_initStaticDCtx(workspace, workspaceSize);
	}

	ZSTD_DStream* ZSTD_createDStream_advanced(ZSTD_customMem customMem)
	{
	return ZSTD_createDCtx_advanced(customMem);
	}

	size_t ZSTD_freeDStream(ZSTD_DStream* zds)
	{
	return ZSTD_freeDCtx(zds);
	}


	/* * Initialization * */

	size_t ZSTD_DStreamInSize(void) { return ZSTD_BLOCKSIZE_MAX + ZSTD_blockHeaderSize; }
	size_t ZSTD_DStreamOutSize(void) { return ZSTD_BLOCKSIZE_MAX; }

	size_t ZSTD_DCtx_loadDictionary_advanced(ZSTD_DCtx* dctx,
	const void* dict, size_t dictSize,
	ZSTD_dictLoadMethod_e dictLoadMethod,
	ZSTD_dictContentType_e dictContentType)
	{
	RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
	ZSTD_clearDict(dctx);
	if (dict && dictSize != 0) {
	dctx->ddictLocal = ZSTD_createDDict_advanced(dict, dictSize, dictLoadMethod, dictContentType, dctx->customMem);
	RETURN_ERROR_IF(dctx->ddictLocal == NULL, memory_allocation, "NULL pointer!");
	dctx->ddict = dctx->ddictLocal;
	dctx->dictUses = ZSTD_use_indefinitely;
	}
	return 0;
	}

	size_t ZSTD_DCtx_loadDictionary_byReference(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
	{
	return ZSTD_DCtx_loadDictionary_advanced(dctx, dict, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto);
	}

	size_t ZSTD_DCtx_loadDictionary(ZSTD_DCtx* dctx, const void* dict, size_t dictSize)
	{
	return ZSTD_DCtx_loadDictionary_advanced(dctx, dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto);
	}

	size_t ZSTD_DCtx_refPrefix_advanced(ZSTD_DCtx* dctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType)
	{
	FORWARD_IF_ERROR(ZSTD_DCtx_loadDictionary_advanced(dctx, prefix, prefixSize, ZSTD_dlm_byRef, dictContentType), "");
	dctx->dictUses = ZSTD_use_once;
	return 0;
	}

	size_t ZSTD_DCtx_refPrefix(ZSTD_DCtx* dctx, const void* prefix, size_t prefixSize)
	{
	return ZSTD_DCtx_refPrefix_advanced(dctx, prefix, prefixSize, ZSTD_dct_rawContent);
	}


	/* ZSTD_initDStream_usingDict() :
	* return : expected size, aka ZSTD_startingInputLength().
	* this function cannot fail */
	size_t ZSTD_initDStream_usingDict(ZSTD_DStream* zds, const void* dict, size_t dictSize)
	{
	DEBUGLOG(4, "ZSTD_initDStream_usingDict");
	FORWARD_IF_ERROR( ZSTD_DCtx_reset(zds, ZSTD_reset_session_only) , "");
	FORWARD_IF_ERROR( ZSTD_DCtx_loadDictionary(zds, dict, dictSize) , "");
	return ZSTD_startingInputLength(zds->format);
	}

	/* note : this variant can't fail */
	size_t ZSTD_initDStream(ZSTD_DStream* zds)
	{
	DEBUGLOG(4, "ZSTD_initDStream");
	return ZSTD_initDStream_usingDDict(zds, NULL);
	}

	/* ZSTD_initDStream_usingDDict() :
	* ddict will just be referenced, and must outlive decompression session
	* this function cannot fail */
	size_t ZSTD_initDStream_usingDDict(ZSTD_DStream* dctx, const ZSTD_DDict* ddict)
	{
	FORWARD_IF_ERROR( ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only) , "");
	FORWARD_IF_ERROR( ZSTD_DCtx_refDDict(dctx, ddict) , "");
	return ZSTD_startingInputLength(dctx->format);
	}

	/* ZSTD_resetDStream() :
	* return : expected size, aka ZSTD_startingInputLength().
	* this function cannot fail */
	size_t ZSTD_resetDStream(ZSTD_DStream* dctx)
	{
	FORWARD_IF_ERROR(ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only), "");
	return ZSTD_startingInputLength(dctx->format);
	}


	size_t ZSTD_DCtx_refDDict(ZSTD_DCtx* dctx, const ZSTD_DDict* ddict)
	{
	RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
	ZSTD_clearDict(dctx);
	if (ddict) {
	dctx->ddict = ddict;
	dctx->dictUses = ZSTD_use_indefinitely;
	}
	return 0;
	}

	/* ZSTD_DCtx_setMaxWindowSize() :
	* note : no direct equivalence in ZSTD_DCtx_setParameter,
	* since this version sets windowSize, and the other sets windowLog */
	size_t ZSTD_DCtx_setMaxWindowSize(ZSTD_DCtx* dctx, size_t maxWindowSize)
	{
	ZSTD_bounds const bounds = ZSTD_dParam_getBounds(ZSTD_d_windowLogMax);
	size_t const min = (size_t)1 << bounds.lowerBound;
	size_t const max = (size_t)1 << bounds.upperBound;
	RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
	RETURN_ERROR_IF(maxWindowSize < min, parameter_outOfBound, "");
	RETURN_ERROR_IF(maxWindowSize > max, parameter_outOfBound, "");
	dctx->maxWindowSize = maxWindowSize;
	return 0;
	}

	size_t ZSTD_DCtx_setFormat(ZSTD_DCtx* dctx, ZSTD_format_e format)
	{
	return ZSTD_DCtx_setParameter(dctx, ZSTD_d_format, format);
	}

	ZSTD_bounds ZSTD_dParam_getBounds(ZSTD_dParameter dParam)
	{
	ZSTD_bounds bounds = { 0, 0, 0 };
	switch(dParam) {
	case ZSTD_d_windowLogMax:
	bounds.lowerBound = ZSTD_WINDOWLOG_ABSOLUTEMIN;
	bounds.upperBound = ZSTD_WINDOWLOG_MAX;
	return bounds;
	case ZSTD_d_format:
	bounds.lowerBound = (int)ZSTD_f_zstd1;
	bounds.upperBound = (int)ZSTD_f_zstd1_magicless;
	ZSTD_STATIC_ASSERT(ZSTD_f_zstd1 < ZSTD_f_zstd1_magicless);
	return bounds;
	case ZSTD_d_stableOutBuffer:
	bounds.lowerBound = (int)ZSTD_obm_buffered;
	bounds.upperBound = (int)ZSTD_obm_stable;
	return bounds;
	default:;
	}
	bounds.error = ERROR(parameter_unsupported);
	return bounds;
	}

	/* ZSTD_dParam_withinBounds:
	* @return 1 if value is within dParam bounds,
	* 0 otherwise */
	static int ZSTD_dParam_withinBounds(ZSTD_dParameter dParam, int value)
	{
	ZSTD_bounds const bounds = ZSTD_dParam_getBounds(dParam);
	if (ZSTD_isError(bounds.error)) return 0;
	if (value < bounds.lowerBound) return 0;
	if (value > bounds.upperBound) return 0;
	return 1;
	}

	#define CHECK_DBOUNDS(p,v) { \
	RETURN_ERROR_IF(!ZSTD_dParam_withinBounds(p, v), parameter_outOfBound, ""); \
	}

	size_t ZSTD_DCtx_setParameter(ZSTD_DCtx* dctx, ZSTD_dParameter dParam, int value)
	{
	RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
	switch(dParam) {
	case ZSTD_d_windowLogMax:
	if (value == 0) value = ZSTD_WINDOWLOG_LIMIT_DEFAULT;
	CHECK_DBOUNDS(ZSTD_d_windowLogMax, value);
	dctx->maxWindowSize = ((size_t)1) << value;
	return 0;
	case ZSTD_d_format:
	CHECK_DBOUNDS(ZSTD_d_format, value);
	dctx->format = (ZSTD_format_e)value;
	return 0;
	case ZSTD_d_stableOutBuffer:
	CHECK_DBOUNDS(ZSTD_d_stableOutBuffer, value);
	dctx->outBufferMode = (ZSTD_outBufferMode_e)value;
	return 0;
	default:;
	}
	RETURN_ERROR(parameter_unsupported, "");
	}

	size_t ZSTD_DCtx_reset(ZSTD_DCtx* dctx, ZSTD_ResetDirective reset)
	{
	if ( (reset == ZSTD_reset_session_only)
	\|\| (reset == ZSTD_reset_session_and_parameters) ) {
	dctx->streamStage = zdss_init;
	dctx->noForwardProgress = 0;
	}
	if ( (reset == ZSTD_reset_parameters)
	\|\| (reset == ZSTD_reset_session_and_parameters) ) {
	RETURN_ERROR_IF(dctx->streamStage != zdss_init, stage_wrong, "");
	ZSTD_clearDict(dctx);
	dctx->format = ZSTD_f_zstd1;
	dctx->maxWindowSize = ZSTD_MAXWINDOWSIZE_DEFAULT;
	}
	return 0;
	}


	size_t ZSTD_sizeof_DStream(const ZSTD_DStream* dctx)
	{
	return ZSTD_sizeof_DCtx(dctx);
	}

	size_t ZSTD_decodingBufferSize_min(unsigned long long windowSize, unsigned long long frameContentSize)
	{
	size_t const blockSize = (size_t) MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
	unsigned long long const neededRBSize = windowSize + blockSize + (WILDCOPY_OVERLENGTH * 2);
	unsigned long long const neededSize = MIN(frameContentSize, neededRBSize);
	size_t const minRBSize = (size_t) neededSize;
	RETURN_ERROR_IF((unsigned long long)minRBSize != neededSize,
	frameParameter_windowTooLarge, "");
	return minRBSize;
	}

	size_t ZSTD_estimateDStreamSize(size_t windowSize)
	{
	size_t const blockSize = MIN(windowSize, ZSTD_BLOCKSIZE_MAX);
	size_t const inBuffSize = blockSize; /* no block can be larger */
	size_t const outBuffSize = ZSTD_decodingBufferSize_min(windowSize, ZSTD_CONTENTSIZE_UNKNOWN);
	return ZSTD_estimateDCtxSize() + inBuffSize + outBuffSize;
	}

	size_t ZSTD_estimateDStreamSize_fromFrame(const void* src, size_t srcSize)
	{
	U32 const windowSizeMax = 1U << ZSTD_WINDOWLOG_MAX; /* note : should be user-selectable, but requires an additional parameter (or a dctx) */
	ZSTD_frameHeader zfh;
	size_t const err = ZSTD_getFrameHeader(&zfh, src, srcSize);
	if (ZSTD_isError(err)) return err;
	RETURN_ERROR_IF(err>0, srcSize_wrong, "");
	RETURN_ERROR_IF(zfh.windowSize > windowSizeMax,
	frameParameter_windowTooLarge, "");
	return ZSTD_estimateDStreamSize((size_t)zfh.windowSize);
	}


	/* *** Decompression *** */

	static int ZSTD_DCtx_isOverflow(ZSTD_DStream* zds, size_t const neededInBuffSize, size_t const neededOutBuffSize)
	{
	return (zds->inBuffSize + zds->outBuffSize) >= (neededInBuffSize + neededOutBuffSize) * ZSTD_WORKSPACETOOLARGE_FACTOR;
	}

	static void ZSTD_DCtx_updateOversizedDuration(ZSTD_DStream* zds, size_t const neededInBuffSize, size_t const neededOutBuffSize)
	{
	if (ZSTD_DCtx_isOverflow(zds, neededInBuffSize, neededOutBuffSize))
	zds->oversizedDuration++;
	else
	zds->oversizedDuration = 0;
	}

	static int ZSTD_DCtx_isOversizedTooLong(ZSTD_DStream* zds)
	{
	return zds->oversizedDuration >= ZSTD_WORKSPACETOOLARGE_MAXDURATION;
	}

	/* Checks that the output buffer hasn't changed if ZSTD_obm_stable is used. */
	static size_t ZSTD_checkOutBuffer(ZSTD_DStream const* zds, ZSTD_outBuffer const* output)
	{
	ZSTD_outBuffer const expect = zds->expectedOutBuffer;
	/* No requirement when ZSTD_obm_stable is not enabled. */
	if (zds->outBufferMode != ZSTD_obm_stable)
	return 0;
	/* Any buffer is allowed in zdss_init, this must be the same for every other call until
	* the context is reset.
	*/
	if (zds->streamStage == zdss_init)
	return 0;
	/* The buffer must match our expectation exactly. */
	if (expect.dst == output->dst && expect.pos == output->pos && expect.size == output->size)
	return 0;
	RETURN_ERROR(dstBuffer_wrong, "ZSTD_obm_stable enabled but output differs!");
	}

	/* Calls ZSTD_decompressContinue() with the right parameters for ZSTD_decompressStream()
	* and updates the stage and the output buffer state. This call is extracted so it can be
	* used both when reading directly from the ZSTD_inBuffer, and in buffered input mode.
	* NOTE: You must break after calling this function since the streamStage is modified.
	*/
	static size_t ZSTD_decompressContinueStream(
	ZSTD_DStream* zds, char** op, char* oend,
	void const* src, size_t srcSize) {
	int const isSkipFrame = ZSTD_isSkipFrame(zds);
	if (zds->outBufferMode == ZSTD_obm_buffered) {
	size_t const dstSize = isSkipFrame ? 0 : zds->outBuffSize - zds->outStart;
	size_t const decodedSize = ZSTD_decompressContinue(zds,
	zds->outBuff + zds->outStart, dstSize, src, srcSize);
	FORWARD_IF_ERROR(decodedSize, "");
	if (!decodedSize && !isSkipFrame) {
	zds->streamStage = zdss_read;
	} else {
	zds->outEnd = zds->outStart + decodedSize;
	zds->streamStage = zdss_flush;
	}
	} else {
	/* Write directly into the output buffer */
	size_t const dstSize = isSkipFrame ? 0 : oend - *op;
	size_t const decodedSize = ZSTD_decompressContinue(zds, *op, dstSize, src, srcSize);
	FORWARD_IF_ERROR(decodedSize, "");
	*op += decodedSize;
	/* Flushing is not needed. */
	zds->streamStage = zdss_read;
	assert(*op <= oend);
	assert(zds->outBufferMode == ZSTD_obm_stable);
	}
	return 0;
	}

	size_t ZSTD_decompressStream(ZSTD_DStream* zds, ZSTD_outBuffer* output, ZSTD_inBuffer* input)
	{
	const char* const src = (const char*)input->src;
	const char* const istart = input->pos != 0 ? src + input->pos : src;
	const char* const iend = input->size != 0 ? src + input->size : src;
	const char* ip = istart;
	char* const dst = (char*)output->dst;
	char* const ostart = output->pos != 0 ? dst + output->pos : dst;
	char* const oend = output->size != 0 ? dst + output->size : dst;
	char* op = ostart;
	U32 someMoreWork = 1;

	DEBUGLOG(5, "ZSTD_decompressStream");
	RETURN_ERROR_IF(
	input->pos > input->size,
	srcSize_wrong,
	"forbidden. in: pos: %u vs size: %u",
	(U32)input->pos, (U32)input->size);
	RETURN_ERROR_IF(
	output->pos > output->size,
	dstSize_tooSmall,
	"forbidden. out: pos: %u vs size: %u",
	(U32)output->pos, (U32)output->size);
	DEBUGLOG(5, "input size : %u", (U32)(input->size - input->pos));
	FORWARD_IF_ERROR(ZSTD_checkOutBuffer(zds, output), "");

	while (someMoreWork) {
	switch(zds->streamStage)
	{
	case zdss_init :
	DEBUGLOG(5, "stage zdss_init => transparent reset ");
	zds->streamStage = zdss_loadHeader;
	zds->lhSize = zds->inPos = zds->outStart = zds->outEnd = 0;
	zds->legacyVersion = 0;
	zds->hostageByte = 0;
	zds->expectedOutBuffer = *output;
	/* fall-through */

	case zdss_loadHeader :
	DEBUGLOG(5, "stage zdss_loadHeader (srcSize : %u)", (U32)(iend - ip));
	#if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>=1)
	if (zds->legacyVersion) {
	RETURN_ERROR_IF(zds->staticSize, memory_allocation,
	"legacy support is incompatible with static dctx");
	{ size_t const hint = ZSTD_decompressLegacyStream(zds->legacyContext, zds->legacyVersion, output, input);
	if (hint==0) zds->streamStage = zdss_init;
	return hint;
	} }
	#endif
	{ size_t const hSize = ZSTD_getFrameHeader_advanced(&zds->fParams, zds->headerBuffer, zds->lhSize, zds->format);
	DEBUGLOG(5, "header size : %u", (U32)hSize);
	if (ZSTD_isError(hSize)) {
	#if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>=1)
	U32 const legacyVersion = ZSTD_isLegacy(istart, iend-istart);
	if (legacyVersion) {
	ZSTD_DDict const* const ddict = ZSTD_getDDict(zds);
	const void* const dict = ddict ? ZSTD_DDict_dictContent(ddict) : NULL;
	size_t const dictSize = ddict ? ZSTD_DDict_dictSize(ddict) : 0;
	DEBUGLOG(5, "ZSTD_decompressStream: detected legacy version v0.%u", legacyVersion);
	RETURN_ERROR_IF(zds->staticSize, memory_allocation,
	"legacy support is incompatible with static dctx");
	FORWARD_IF_ERROR(ZSTD_initLegacyStream(&zds->legacyContext,
	zds->previousLegacyVersion, legacyVersion,
	dict, dictSize), "");
	zds->legacyVersion = zds->previousLegacyVersion = legacyVersion;
	{ size_t const hint = ZSTD_decompressLegacyStream(zds->legacyContext, legacyVersion, output, input);
	if (hint==0) zds->streamStage = zdss_init; /* or stay in stage zdss_loadHeader */
	return hint;
	} }
	#endif
	return hSize; /* error */
	}
	if (hSize != 0) { /* need more input */
	size_t const toLoad = hSize - zds->lhSize; /* if hSize!=0, hSize > zds->lhSize */
	size_t const remainingInput = (size_t)(iend-ip);
	assert(iend >= ip);
	if (toLoad > remainingInput) { /* not enough input to load full header */
	if (remainingInput > 0) {
	memcpy(zds->headerBuffer + zds->lhSize, ip, remainingInput);
	zds->lhSize += remainingInput;
	}
	input->pos = input->size;
	return (MAX((size_t)ZSTD_FRAMEHEADERSIZE_MIN(zds->format), hSize) - zds->lhSize) + ZSTD_blockHeaderSize; /* remaining header bytes + next block header */
	}
	assert(ip != NULL);
	memcpy(zds->headerBuffer + zds->lhSize, ip, toLoad); zds->lhSize = hSize; ip += toLoad;
	break;
	} }

	/* check for single-pass mode opportunity */
	if (zds->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
	&& zds->fParams.frameType != ZSTD_skippableFrame
	&& (U64)(size_t)(oend-op) >= zds->fParams.frameContentSize) {
	size_t const cSize = ZSTD_findFrameCompressedSize(istart, iend-istart);
	if (cSize <= (size_t)(iend-istart)) {
	/* shortcut : using single-pass mode */
	size_t const decompressedSize = ZSTD_decompress_usingDDict(zds, op, oend-op, istart, cSize, ZSTD_getDDict(zds));
	if (ZSTD_isError(decompressedSize)) return decompressedSize;
	DEBUGLOG(4, "shortcut to single-pass ZSTD_decompress_usingDDict()")
	ip = istart + cSize;
	op += decompressedSize;
	zds->expected = 0;
	zds->streamStage = zdss_init;
	someMoreWork = 0;
	break;
	} }

	/* Check output buffer is large enough for ZSTD_odm_stable. */
	if (zds->outBufferMode == ZSTD_obm_stable
	&& zds->fParams.frameType != ZSTD_skippableFrame
	&& zds->fParams.frameContentSize != ZSTD_CONTENTSIZE_UNKNOWN
	&& (U64)(size_t)(oend-op) < zds->fParams.frameContentSize) {
	RETURN_ERROR(dstSize_tooSmall, "ZSTD_obm_stable passed but ZSTD_outBuffer is too small");
	}

	/* Consume header (see ZSTDds_decodeFrameHeader) */
	DEBUGLOG(4, "Consume header");
	FORWARD_IF_ERROR(ZSTD_decompressBegin_usingDDict(zds, ZSTD_getDDict(zds)), "");

	if ((MEM_readLE32(zds->headerBuffer) & ZSTD_MAGIC_SKIPPABLE_MASK) == ZSTD_MAGIC_SKIPPABLE_START) { /* skippable frame */
	zds->expected = MEM_readLE32(zds->headerBuffer + ZSTD_FRAMEIDSIZE);
	zds->stage = ZSTDds_skipFrame;
	} else {
	FORWARD_IF_ERROR(ZSTD_decodeFrameHeader(zds, zds->headerBuffer, zds->lhSize), "");
	zds->expected = ZSTD_blockHeaderSize;
	zds->stage = ZSTDds_decodeBlockHeader;
	}

	/* control buffer memory usage */
	DEBUGLOG(4, "Control max memory usage (%u KB <= max %u KB)",
	(U32)(zds->fParams.windowSize >>10),
	(U32)(zds->maxWindowSize >> 10) );
	zds->fParams.windowSize = MAX(zds->fParams.windowSize, 1U << ZSTD_WINDOWLOG_ABSOLUTEMIN);
	RETURN_ERROR_IF(zds->fParams.windowSize > zds->maxWindowSize,
	frameParameter_windowTooLarge, "");

	/* Adapt buffer sizes to frame header instructions */
	{ size_t const neededInBuffSize = MAX(zds->fParams.blockSizeMax, 4 /* frame checksum */);
	size_t const neededOutBuffSize = zds->outBufferMode == ZSTD_obm_buffered
	? ZSTD_decodingBufferSize_min(zds->fParams.windowSize, zds->fParams.frameContentSize)
	: 0;

	ZSTD_DCtx_updateOversizedDuration(zds, neededInBuffSize, neededOutBuffSize);

	{ int const tooSmall = (zds->inBuffSize < neededInBuffSize) \|\| (zds->outBuffSize < neededOutBuffSize);
	int const tooLarge = ZSTD_DCtx_isOversizedTooLong(zds);

	if (tooSmall \|\| tooLarge) {
	size_t const bufferSize = neededInBuffSize + neededOutBuffSize;
	DEBUGLOG(4, "inBuff : from %u to %u",
	(U32)zds->inBuffSize, (U32)neededInBuffSize);
	DEBUGLOG(4, "outBuff : from %u to %u",
	(U32)zds->outBuffSize, (U32)neededOutBuffSize);
	if (zds->staticSize) { /* static DCtx */
	DEBUGLOG(4, "staticSize : %u", (U32)zds->staticSize);
	assert(zds->staticSize >= sizeof(ZSTD_DCtx)); /* controlled at init */
	RETURN_ERROR_IF(
	bufferSize > zds->staticSize - sizeof(ZSTD_DCtx),
	memory_allocation, "");
	} else {
	ZSTD_free(zds->inBuff, zds->customMem);
	zds->inBuffSize = 0;
	zds->outBuffSize = 0;
	zds->inBuff = (char*)ZSTD_malloc(bufferSize, zds->customMem);
	RETURN_ERROR_IF(zds->inBuff == NULL, memory_allocation, "");
	}
	zds->inBuffSize = neededInBuffSize;
	zds->outBuff = zds->inBuff + zds->inBuffSize;
	zds->outBuffSize = neededOutBuffSize;
	} } }
	zds->streamStage = zdss_read;
	/* fall-through */

	case zdss_read:
	DEBUGLOG(5, "stage zdss_read");
	{ size_t const neededInSize = ZSTD_nextSrcSizeToDecompressWithInputSize(zds, iend - ip);
	DEBUGLOG(5, "neededInSize = %u", (U32)neededInSize);
	if (neededInSize==0) { /* end of frame */
	zds->streamStage = zdss_init;
	someMoreWork = 0;
	break;
	}
	if ((size_t)(iend-ip) >= neededInSize) { /* decode directly from src */
	FORWARD_IF_ERROR(ZSTD_decompressContinueStream(zds, &op, oend, ip, neededInSize), "");
	ip += neededInSize;
	/* Function modifies the stage so we must break */
	break;
	} }
	if (ip==iend) { someMoreWork = 0; break; } /* no more input */
	zds->streamStage = zdss_load;
	/* fall-through */

	case zdss_load:
	{ size_t const neededInSize = ZSTD_nextSrcSizeToDecompress(zds);
	size_t const toLoad = neededInSize - zds->inPos;
	int const isSkipFrame = ZSTD_isSkipFrame(zds);
	size_t loadedSize;
	/* At this point we shouldn't be decompressing a block that we can stream. */
	assert(neededInSize == ZSTD_nextSrcSizeToDecompressWithInputSize(zds, iend - ip));
	if (isSkipFrame) {
	loadedSize = MIN(toLoad, (size_t)(iend-ip));
	} else {
	RETURN_ERROR_IF(toLoad > zds->inBuffSize - zds->inPos,
	corruption_detected,
	"should never happen");
	loadedSize = ZSTD_limitCopy(zds->inBuff + zds->inPos, toLoad, ip, iend-ip);
	}
	ip += loadedSize;
	zds->inPos += loadedSize;
	if (loadedSize < toLoad) { someMoreWork = 0; break; } /* not enough input, wait for more */

	/* decode loaded input */
	zds->inPos = 0; /* input is consumed */
	FORWARD_IF_ERROR(ZSTD_decompressContinueStream(zds, &op, oend, zds->inBuff, neededInSize), "");
	/* Function modifies the stage so we must break */
	break;
	}
	case zdss_flush:
	{ size_t const toFlushSize = zds->outEnd - zds->outStart;
	size_t const flushedSize = ZSTD_limitCopy(op, oend-op, zds->outBuff + zds->outStart, toFlushSize);
	op += flushedSize;
	zds->outStart += flushedSize;
	if (flushedSize == toFlushSize) { /* flush completed */
	zds->streamStage = zdss_read;
	if ( (zds->outBuffSize < zds->fParams.frameContentSize)
	&& (zds->outStart + zds->fParams.blockSizeMax > zds->outBuffSize) ) {
	DEBUGLOG(5, "restart filling outBuff from beginning (left:%i, needed:%u)",
	(int)(zds->outBuffSize - zds->outStart),
	(U32)zds->fParams.blockSizeMax);
	zds->outStart = zds->outEnd = 0;
	}
	break;
	} }
	/* cannot complete flush */
	someMoreWork = 0;
	break;

	default:
	assert(0); /* impossible */
	RETURN_ERROR(GENERIC, "impossible to reach"); /* some compiler require default to do something */
	} }

	/* result */
	input->pos = (size_t)(ip - (const char*)(input->src));
	output->pos = (size_t)(op - (char*)(output->dst));

	/* Update the expected output buffer for ZSTD_obm_stable. */
	zds->expectedOutBuffer = *output;

	if ((ip==istart) && (op==ostart)) { /* no forward progress */
	zds->noForwardProgress ++;
	if (zds->noForwardProgress >= ZSTD_NO_FORWARD_PROGRESS_MAX) {
	RETURN_ERROR_IF(op==oend, dstSize_tooSmall, "");
	RETURN_ERROR_IF(ip==iend, srcSize_wrong, "");
	assert(0);
	}
	} else {
	zds->noForwardProgress = 0;
	}
	{ size_t nextSrcSizeHint = ZSTD_nextSrcSizeToDecompress(zds);
	if (!nextSrcSizeHint) { /* frame fully decoded */
	if (zds->outEnd == zds->outStart) { /* output fully flushed */
	if (zds->hostageByte) {
	if (input->pos >= input->size) {
	/* can't release hostage (not present) */
	zds->streamStage = zdss_read;
	return 1;
	}
	input->pos++; /* release hostage */
	} /* zds->hostageByte */
	return 0;
	} /* zds->outEnd == zds->outStart */
	if (!zds->hostageByte) { /* output not fully flushed; keep last byte as hostage; will be released when all output is flushed */
	input->pos--; /* note : pos > 0, otherwise, impossible to finish reading last block */
	zds->hostageByte=1;
	}
	return 1;
	} /* nextSrcSizeHint==0 */
	nextSrcSizeHint += ZSTD_blockHeaderSize * (ZSTD_nextInputType(zds) == ZSTDnit_block); /* preload header of next block */
	assert(zds->inPos <= nextSrcSizeHint);
	nextSrcSizeHint -= zds->inPos; /* part already loaded*/
	return nextSrcSizeHint;
	}
	}

	size_t ZSTD_decompressStream_simpleArgs (
	ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity, size_t* dstPos,
	const void* src, size_t srcSize, size_t* srcPos)
	{
	ZSTD_outBuffer output = { dst, dstCapacity, *dstPos };
	ZSTD_inBuffer input = { src, srcSize, *srcPos };
	/* ZSTD_compress_generic() will check validity of dstPos and srcPos */
	size_t const cErr = ZSTD_decompressStream(dctx, &output, &input);
	*dstPos = output.pos;
	*srcPos = input.pos;
	return cErr;
	}
	/** ended inlining decompress/zstd_decompress.c **/
	/** start inlining decompress/zstd_decompress_block.c **/
	/*
	* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
	* All rights reserved.
	*
	* This source code is licensed under both the BSD-style license (found in the
	* LICENSE file in the root directory of this source tree) and the GPLv2 (found
	* in the COPYING file in the root directory of this source tree).
	* You may select, at your option, one of the above-listed licenses.
	*/

	/* zstd_decompress_block :
	* this module takes care of decompressing _compressed_ block */

	/-******************************************************
	* Dependencies
	*********************************************************/
	#include <string.h> /* memcpy, memmove, memset */
	/** skipping file: ../common/compiler.h **/
	/** skipping file: ../common/cpu.h **/
	/** skipping file: ../common/mem.h **/
	#define FSE_STATIC_LINKING_ONLY
	/** skipping file: ../common/fse.h **/
	#define HUF_STATIC_LINKING_ONLY
	/** skipping file: ../common/huf.h **/
	/** skipping file: ../common/zstd_internal.h **/
	/** skipping file: zstd_decompress_internal.h **/
	/** skipping file: zstd_ddict.h **/
	/** skipping file: zstd_decompress_block.h **/

	/_******************************************************
	* Macros
	**********************************************************/

	/* These two optional macros force the use one way or another of the two
	* ZSTD_decompressSequences implementations. You can't force in both directions
	* at the same time.
	*/
	#if defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
	defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
	#error "Cannot force the use of the short and the long ZSTD_decompressSequences variants!"
	#endif


	/_******************************************************
	* Memory operations
	**********************************************************/
	static void ZSTD_copy4(void* dst, const void* src) { memcpy(dst, src, 4); }


	/-************************************************************
	* Block decoding
	***************************************************************/

	/*! ZSTD_getcBlockSize() :
	* Provides the size of compressed block from block header `src` */
	size_t ZSTD_getcBlockSize(const void* src, size_t srcSize,
	blockProperties_t* bpPtr)
	{
	RETURN_ERROR_IF(srcSize < ZSTD_blockHeaderSize, srcSize_wrong, "");

	{ U32 const cBlockHeader = MEM_readLE24(src);
	U32 const cSize = cBlockHeader >> 3;
	bpPtr->lastBlock = cBlockHeader & 1;
	bpPtr->blockType = (blockType_e)((cBlockHeader >> 1) & 3);
	bpPtr->origSize = cSize; /* only useful for RLE */
	if (bpPtr->blockType == bt_rle) return 1;
	RETURN_ERROR_IF(bpPtr->blockType == bt_reserved, corruption_detected, "");
	return cSize;
	}
	}


	/* Hidden declaration for fullbench */
	size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx,
	const void* src, size_t srcSize);
	/*! ZSTD_decodeLiteralsBlock() :
	* @return : nb of bytes read from src (< srcSize )
	* note : symbol not declared but exposed for fullbench */
	size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx* dctx,
	const void* src, size_t srcSize) /* note : srcSize < BLOCKSIZE */
	{
	DEBUGLOG(5, "ZSTD_decodeLiteralsBlock");
	RETURN_ERROR_IF(srcSize < MIN_CBLOCK_SIZE, corruption_detected, "");

	{ const BYTE* const istart = (const BYTE*) src;
	symbolEncodingType_e const litEncType = (symbolEncodingType_e)(istart[0] & 3);

	switch(litEncType)
	{
	case set_repeat:
	DEBUGLOG(5, "set_repeat flag : re-using stats from previous compressed literals block");
	RETURN_ERROR_IF(dctx->litEntropy==0, dictionary_corrupted, "");
	/* fall-through */

	case set_compressed:
	RETURN_ERROR_IF(srcSize < 5, corruption_detected, "srcSize >= MIN_CBLOCK_SIZE == 3; here we need up to 5 for case 3");
	{ size_t lhSize, litSize, litCSize;
	U32 singleStream=0;
	U32 const lhlCode = (istart[0] >> 2) & 3;
	U32 const lhc = MEM_readLE32(istart);
	size_t hufSuccess;
	switch(lhlCode)
	{
	case 0: case 1: default: /* note : default is impossible, since lhlCode into [0..3] */
	/* 2 - 2 - 10 - 10 */
	singleStream = !lhlCode;
	lhSize = 3;
	litSize = (lhc >> 4) & 0x3FF;
	litCSize = (lhc >> 14) & 0x3FF;
	break;
	case 2:
	/* 2 - 2 - 14 - 14 */
	lhSize = 4;
	litSize = (lhc >> 4) & 0x3FFF;
	litCSize = lhc >> 18;
	break;
	case 3:
	/* 2 - 2 - 18 - 18 */
	lhSize = 5;
	litSize = (lhc >> 4) & 0x3FFFF;
	litCSize = (lhc >> 22) + ((size_t)istart[4] << 10);
	break;
	}
	RETURN_ERROR_IF(litSize > ZSTD_BLOCKSIZE_MAX, corruption_detected, "");
	RETURN_ERROR_IF(litCSize + lhSize > srcSize, corruption_detected, "");

	/* prefetch huffman table if cold */
	if (dctx->ddictIsCold && (litSize > 768 /* heuristic */)) {
	PREFETCH_AREA(dctx->HUFptr, sizeof(dctx->entropy.hufTable));
	}

	if (litEncType==set_repeat) {
	if (singleStream) {
	hufSuccess = HUF_decompress1X_usingDTable_bmi2(
	dctx->litBuffer, litSize, istart+lhSize, litCSize,
	dctx->HUFptr, dctx->bmi2);
	} else {
	hufSuccess = HUF_decompress4X_usingDTable_bmi2(
	dctx->litBuffer, litSize, istart+lhSize, litCSize,
	dctx->HUFptr, dctx->bmi2);
	}
	} else {
	if (singleStream) {
	#if defined(HUF_FORCE_DECOMPRESS_X2)
	hufSuccess = HUF_decompress1X_DCtx_wksp(
	dctx->entropy.hufTable, dctx->litBuffer, litSize,
	istart+lhSize, litCSize, dctx->workspace,
	sizeof(dctx->workspace));
	#else
	hufSuccess = HUF_decompress1X1_DCtx_wksp_bmi2(
	dctx->entropy.hufTable, dctx->litBuffer, litSize,
	istart+lhSize, litCSize, dctx->workspace,
	sizeof(dctx->workspace), dctx->bmi2);
	#endif
	} else {
	hufSuccess = HUF_decompress4X_hufOnly_wksp_bmi2(
	dctx->entropy.hufTable, dctx->litBuffer, litSize,
	istart+lhSize, litCSize, dctx->workspace,
	sizeof(dctx->workspace), dctx->bmi2);
	}
	}

	RETURN_ERROR_IF(HUF_isError(hufSuccess), corruption_detected, "");

	dctx->litPtr = dctx->litBuffer;
	dctx->litSize = litSize;
	dctx->litEntropy = 1;
	if (litEncType==set_compressed) dctx->HUFptr = dctx->entropy.hufTable;
	memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
	return litCSize + lhSize;
	}

	case set_basic:
	{ size_t litSize, lhSize;
	U32 const lhlCode = ((istart[0]) >> 2) & 3;
	switch(lhlCode)
	{
	case 0: case 2: default: /* note : default is impossible, since lhlCode into [0..3] */
	lhSize = 1;
	litSize = istart[0] >> 3;
	break;
	case 1:
	lhSize = 2;
	litSize = MEM_readLE16(istart) >> 4;
	break;
	case 3:
	lhSize = 3;
	litSize = MEM_readLE24(istart) >> 4;
	break;
	}

	if (lhSize+litSize+WILDCOPY_OVERLENGTH > srcSize) { /* risk reading beyond src buffer with wildcopy */
	RETURN_ERROR_IF(litSize+lhSize > srcSize, corruption_detected, "");
	memcpy(dctx->litBuffer, istart+lhSize, litSize);
	dctx->litPtr = dctx->litBuffer;
	dctx->litSize = litSize;
	memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
	return lhSize+litSize;
	}
	/* direct reference into compressed stream */
	dctx->litPtr = istart+lhSize;
	dctx->litSize = litSize;
	return lhSize+litSize;
	}

	case set_rle:
	{ U32 const lhlCode = ((istart[0]) >> 2) & 3;
	size_t litSize, lhSize;
	switch(lhlCode)
	{
	case 0: case 2: default: /* note : default is impossible, since lhlCode into [0..3] */
	lhSize = 1;
	litSize = istart[0] >> 3;
	break;
	case 1:
	lhSize = 2;
	litSize = MEM_readLE16(istart) >> 4;
	break;
	case 3:
	lhSize = 3;
	litSize = MEM_readLE24(istart) >> 4;
	RETURN_ERROR_IF(srcSize<4, corruption_detected, "srcSize >= MIN_CBLOCK_SIZE == 3; here we need lhSize+1 = 4");
	break;
	}
	RETURN_ERROR_IF(litSize > ZSTD_BLOCKSIZE_MAX, corruption_detected, "");
	memset(dctx->litBuffer, istart[lhSize], litSize + WILDCOPY_OVERLENGTH);
	dctx->litPtr = dctx->litBuffer;
	dctx->litSize = litSize;
	return lhSize+1;
	}
	default:
	RETURN_ERROR(corruption_detected, "impossible");
	}
	}
	}

	/* Default FSE distribution tables.
	* These are pre-calculated FSE decoding tables using default distributions as defined in specification :
	* https://github.com/facebook/zstd/blob/master/doc/zstd_compression_format.md#default-distributions
	* They were generated programmatically with following method :
	* - start from default distributions, present in /lib/common/zstd_internal.h
	* - generate tables normally, using ZSTD_buildFSETable()
	* - printout the content of tables
	* - pretify output, report below, test with fuzzer to ensure it's correct */

	/* Default FSE distribution table for Literal Lengths */
	static const ZSTD_seqSymbol LL_defaultDTable[(1<<LL_DEFAULTNORMLOG)+1] = {
	{ 1, 1, 1, LL_DEFAULTNORMLOG}, /* header : fastMode, tableLog */
	/* nextState, nbAddBits, nbBits, baseVal */
	{ 0, 0, 4, 0}, { 16, 0, 4, 0},
	{ 32, 0, 5, 1}, { 0, 0, 5, 3},
	{ 0, 0, 5, 4}, { 0, 0, 5, 6},
	{ 0, 0, 5, 7}, { 0, 0, 5, 9},
	{ 0, 0, 5, 10}, { 0, 0, 5, 12},
	{ 0, 0, 6, 14}, { 0, 1, 5, 16},
	{ 0, 1, 5, 20}, { 0, 1, 5, 22},
	{ 0, 2, 5, 28}, { 0, 3, 5, 32},
	{ 0, 4, 5, 48}, { 32, 6, 5, 64},
	{ 0, 7, 5, 128}, { 0, 8, 6, 256},
	{ 0, 10, 6, 1024}, { 0, 12, 6, 4096},
	{ 32, 0, 4, 0}, { 0, 0, 4, 1},
	{ 0, 0, 5, 2}, { 32, 0, 5, 4},
	{ 0, 0, 5, 5}, { 32, 0, 5, 7},
	{ 0, 0, 5, 8}, { 32, 0, 5, 10},
	{ 0, 0, 5, 11}, { 0, 0, 6, 13},
	{ 32, 1, 5, 16}, { 0, 1, 5, 18},
	{ 32, 1, 5, 22}, { 0, 2, 5, 24},
	{ 32, 3, 5, 32}, { 0, 3, 5, 40},
	{ 0, 6, 4, 64}, { 16, 6, 4, 64},
	{ 32, 7, 5, 128}, { 0, 9, 6, 512},
	{ 0, 11, 6, 2048}, { 48, 0, 4, 0},
	{ 16, 0, 4, 1}, { 32, 0, 5, 2},
	{ 32, 0, 5, 3}, { 32, 0, 5, 5},
	{ 32, 0, 5, 6}, { 32, 0, 5, 8},
	{ 32, 0, 5, 9}, { 32, 0, 5, 11},
	{ 32, 0, 5, 12}, { 0, 0, 6, 15},
	{ 32, 1, 5, 18}, { 32, 1, 5, 20},
	{ 32, 2, 5, 24}, { 32, 2, 5, 28},
	{ 32, 3, 5, 40}, { 32, 4, 5, 48},
	{ 0, 16, 6,65536}, { 0, 15, 6,32768},
	{ 0, 14, 6,16384}, { 0, 13, 6, 8192},
	}; /* LL_defaultDTable */

	/* Default FSE distribution table for Offset Codes */
	static const ZSTD_seqSymbol OF_defaultDTable[(1<<OF_DEFAULTNORMLOG)+1] = {
	{ 1, 1, 1, OF_DEFAULTNORMLOG}, /* header : fastMode, tableLog */
	/* nextState, nbAddBits, nbBits, baseVal */
	{ 0, 0, 5, 0}, { 0, 6, 4, 61},
	{ 0, 9, 5, 509}, { 0, 15, 5,32765},
	{ 0, 21, 5,2097149}, { 0, 3, 5, 5},
	{ 0, 7, 4, 125}, { 0, 12, 5, 4093},
	{ 0, 18, 5,262141}, { 0, 23, 5,8388605},
	{ 0, 5, 5, 29}, { 0, 8, 4, 253},
	{ 0, 14, 5,16381}, { 0, 20, 5,1048573},
	{ 0, 2, 5, 1}, { 16, 7, 4, 125},
	{ 0, 11, 5, 2045}, { 0, 17, 5,131069},
	{ 0, 22, 5,4194301}, { 0, 4, 5, 13},
	{ 16, 8, 4, 253}, { 0, 13, 5, 8189},
	{ 0, 19, 5,524285}, { 0, 1, 5, 1},
	{ 16, 6, 4, 61}, { 0, 10, 5, 1021},
	{ 0, 16, 5,65533}, { 0, 28, 5,268435453},
	{ 0, 27, 5,134217725}, { 0, 26, 5,67108861},
	{ 0, 25, 5,33554429}, { 0, 24, 5,16777213},
	}; /* OF_defaultDTable */


	/* Default FSE distribution table for Match Lengths */
	static const ZSTD_seqSymbol ML_defaultDTable[(1<<ML_DEFAULTNORMLOG)+1] = {
	{ 1, 1, 1, ML_DEFAULTNORMLOG}, /* header : fastMode, tableLog */
	/* nextState, nbAddBits, nbBits, baseVal */
	{ 0, 0, 6, 3}, { 0, 0, 4, 4},
	{ 32, 0, 5, 5}, { 0, 0, 5, 6},
	{ 0, 0, 5, 8}, { 0, 0, 5, 9},
	{ 0, 0, 5, 11}, { 0, 0, 6, 13},
	{ 0, 0, 6, 16}, { 0, 0, 6, 19},
	{ 0, 0, 6, 22}, { 0, 0, 6, 25},
	{ 0, 0, 6, 28}, { 0, 0, 6, 31},
	{ 0, 0, 6, 34}, { 0, 1, 6, 37},
	{ 0, 1, 6, 41}, { 0, 2, 6, 47},
	{ 0, 3, 6, 59}, { 0, 4, 6, 83},
	{ 0, 7, 6, 131}, { 0, 9, 6, 515},
	{ 16, 0, 4, 4}, { 0, 0, 4, 5},
	{ 32, 0, 5, 6}, { 0, 0, 5, 7},
	{ 32, 0, 5, 9}, { 0, 0, 5, 10},
	{ 0, 0, 6, 12}, { 0, 0, 6, 15},
	{ 0, 0, 6, 18}, { 0, 0, 6, 21},
	{ 0, 0, 6, 24}, { 0, 0, 6, 27},
	{ 0, 0, 6, 30}, { 0, 0, 6, 33},
	{ 0, 1, 6, 35}, { 0, 1, 6, 39},
	{ 0, 2, 6, 43}, { 0, 3, 6, 51},
	{ 0, 4, 6, 67}, { 0, 5, 6, 99},
	{ 0, 8, 6, 259}, { 32, 0, 4, 4},
	{ 48, 0, 4, 4}, { 16, 0, 4, 5},
	{ 32, 0, 5, 7}, { 32, 0, 5, 8},
	{ 32, 0, 5, 10}, { 32, 0, 5, 11},
	{ 0, 0, 6, 14}, { 0, 0, 6, 17},
	{ 0, 0, 6, 20}, { 0, 0, 6, 23},
	{ 0, 0, 6, 26}, { 0, 0, 6, 29},
	{ 0, 0, 6, 32}, { 0, 16, 6,65539},
	{ 0, 15, 6,32771}, { 0, 14, 6,16387},
	{ 0, 13, 6, 8195}, { 0, 12, 6, 4099},
	{ 0, 11, 6, 2051}, { 0, 10, 6, 1027},
	}; /* ML_defaultDTable */


	static void ZSTD_buildSeqTable_rle(ZSTD_seqSymbol* dt, U32 baseValue, U32 nbAddBits)
	{
	void* ptr = dt;
	ZSTD_seqSymbol_header* const DTableH = (ZSTD_seqSymbol_header*)ptr;
	ZSTD_seqSymbol* const cell = dt + 1;

	DTableH->tableLog = 0;
	DTableH->fastMode = 0;

	cell->nbBits = 0;
	cell->nextState = 0;
	assert(nbAddBits < 255);
	cell->nbAdditionalBits = (BYTE)nbAddBits;
	cell->baseValue = baseValue;
	}


	/* ZSTD_buildFSETable() :
	* generate FSE decoding table for one symbol (ll, ml or off)
	* cannot fail if input is valid =>
	* all inputs are presumed validated at this stage */
	void
	ZSTD_buildFSETable(ZSTD_seqSymbol* dt,
	const short* normalizedCounter, unsigned maxSymbolValue,
	const U32* baseValue, const U32* nbAdditionalBits,
	unsigned tableLog)
	{
	ZSTD_seqSymbol* const tableDecode = dt+1;
	U16 symbolNext[MaxSeq+1];

	U32 const maxSV1 = maxSymbolValue + 1;
	U32 const tableSize = 1 << tableLog;
	U32 highThreshold = tableSize-1;

	/* Sanity Checks */
	assert(maxSymbolValue <= MaxSeq);
	assert(tableLog <= MaxFSELog);

	/* Init, lay down lowprob symbols */
	{ ZSTD_seqSymbol_header DTableH;
	DTableH.tableLog = tableLog;
	DTableH.fastMode = 1;
	{ S16 const largeLimit= (S16)(1 << (tableLog-1));
	U32 s;
	for (s=0; s<maxSV1; s++) {
	if (normalizedCounter[s]==-1) {
	tableDecode[highThreshold--].baseValue = s;
	symbolNext[s] = 1;
	} else {
	if (normalizedCounter[s] >= largeLimit) DTableH.fastMode=0;
	assert(normalizedCounter[s]>=0);
	symbolNext[s] = (U16)normalizedCounter[s];
	} } }
	memcpy(dt, &DTableH, sizeof(DTableH));
	}

	/* Spread symbols */
	{ U32 const tableMask = tableSize-1;
	U32 const step = FSE_TABLESTEP(tableSize);
	U32 s, position = 0;
	for (s=0; s<maxSV1; s++) {
	int i;
	for (i=0; i<normalizedCounter[s]; i++) {
	tableDecode[position].baseValue = s;
	position = (position + step) & tableMask;
	while (position > highThreshold) position = (position + step) & tableMask; /* lowprob area */
	} }
	assert(position == 0); /* position must reach all cells once, otherwise normalizedCounter is incorrect */
	}

	/* Build Decoding table */
	{ U32 u;
	for (u=0; u<tableSize; u++) {
	U32 const symbol = tableDecode[u].baseValue;
	U32 const nextState = symbolNext[symbol]++;
	tableDecode[u].nbBits = (BYTE) (tableLog - BIT_highbit32(nextState) );
	tableDecode[u].nextState = (U16) ( (nextState << tableDecode[u].nbBits) - tableSize);
	assert(nbAdditionalBits[symbol] < 255);
	tableDecode[u].nbAdditionalBits = (BYTE)nbAdditionalBits[symbol];
	tableDecode[u].baseValue = baseValue[symbol];
	} }
	}


	/*! ZSTD_buildSeqTable() :
	* @return : nb bytes read from src,
	* or an error code if it fails */
	static size_t ZSTD_buildSeqTable(ZSTD_seqSymbol* DTableSpace, const ZSTD_seqSymbol** DTablePtr,
	symbolEncodingType_e type, unsigned max, U32 maxLog,
	const void* src, size_t srcSize,
	const U32* baseValue, const U32* nbAdditionalBits,
	const ZSTD_seqSymbol* defaultTable, U32 flagRepeatTable,
	int ddictIsCold, int nbSeq)
	{
	switch(type)
	{
	case set_rle :
	RETURN_ERROR_IF(!srcSize, srcSize_wrong, "");
	RETURN_ERROR_IF(((const BYTE)src) > max, corruption_detected, "");
	{ U32 const symbol = (const BYTE)src;
	U32 const baseline = baseValue[symbol];
	U32 const nbBits = nbAdditionalBits[symbol];
	ZSTD_buildSeqTable_rle(DTableSpace, baseline, nbBits);
	}
	*DTablePtr = DTableSpace;
	return 1;
	case set_basic :
	*DTablePtr = defaultTable;
	return 0;
	case set_repeat:
	RETURN_ERROR_IF(!flagRepeatTable, corruption_detected, "");
	/* prefetch FSE table if used */
	if (ddictIsCold && (nbSeq > 24 /* heuristic */)) {
	const void* const pStart = *DTablePtr;
	size_t const pSize = sizeof(ZSTD_seqSymbol) * (SEQSYMBOL_TABLE_SIZE(maxLog));
	PREFETCH_AREA(pStart, pSize);
	}
	return 0;
	case set_compressed :
	{ unsigned tableLog;
	S16 norm[MaxSeq+1];
	size_t const headerSize = FSE_readNCount(norm, &max, &tableLog, src, srcSize);
	RETURN_ERROR_IF(FSE_isError(headerSize), corruption_detected, "");
	RETURN_ERROR_IF(tableLog > maxLog, corruption_detected, "");
	ZSTD_buildFSETable(DTableSpace, norm, max, baseValue, nbAdditionalBits, tableLog);
	*DTablePtr = DTableSpace;
	return headerSize;
	}
	default :
	assert(0);
	RETURN_ERROR(GENERIC, "impossible");
	}
	}

	size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr,
	const void* src, size_t srcSize)
	{
	const BYTE* const istart = (const BYTE* const)src;
	const BYTE* const iend = istart + srcSize;
	const BYTE* ip = istart;
	int nbSeq;
	DEBUGLOG(5, "ZSTD_decodeSeqHeaders");

	/* check */
	RETURN_ERROR_IF(srcSize < MIN_SEQUENCES_SIZE, srcSize_wrong, "");

	/* SeqHead */
	nbSeq = *ip++;
	if (!nbSeq) {
	*nbSeqPtr=0;
	RETURN_ERROR_IF(srcSize != 1, srcSize_wrong, "");
	return 1;
	}
	if (nbSeq > 0x7F) {
	if (nbSeq == 0xFF) {
	RETURN_ERROR_IF(ip+2 > iend, srcSize_wrong, "");
	nbSeq = MEM_readLE16(ip) + LONGNBSEQ, ip+=2;
	} else {
	RETURN_ERROR_IF(ip >= iend, srcSize_wrong, "");
	nbSeq = ((nbSeq-0x80)<<8) + *ip++;
	}
	}
	*nbSeqPtr = nbSeq;

	/* FSE table descriptors */
	RETURN_ERROR_IF(ip+1 > iend, srcSize_wrong, ""); /* minimum possible size: 1 byte for symbol encoding types */
	{ symbolEncodingType_e const LLtype = (symbolEncodingType_e)(*ip >> 6);
	symbolEncodingType_e const OFtype = (symbolEncodingType_e)((*ip >> 4) & 3);
	symbolEncodingType_e const MLtype = (symbolEncodingType_e)((*ip >> 2) & 3);
	ip++;

	/* Build DTables */
	{ size_t const llhSize = ZSTD_buildSeqTable(dctx->entropy.LLTable, &dctx->LLTptr,
	LLtype, MaxLL, LLFSELog,
	ip, iend-ip,
	LL_base, LL_bits,
	LL_defaultDTable, dctx->fseEntropy,
	dctx->ddictIsCold, nbSeq);
	RETURN_ERROR_IF(ZSTD_isError(llhSize), corruption_detected, "ZSTD_buildSeqTable failed");
	ip += llhSize;
	}

	{ size_t const ofhSize = ZSTD_buildSeqTable(dctx->entropy.OFTable, &dctx->OFTptr,
	OFtype, MaxOff, OffFSELog,
	ip, iend-ip,
	OF_base, OF_bits,
	OF_defaultDTable, dctx->fseEntropy,
	dctx->ddictIsCold, nbSeq);
	RETURN_ERROR_IF(ZSTD_isError(ofhSize), corruption_detected, "ZSTD_buildSeqTable failed");
	ip += ofhSize;
	}

	{ size_t const mlhSize = ZSTD_buildSeqTable(dctx->entropy.MLTable, &dctx->MLTptr,
	MLtype, MaxML, MLFSELog,
	ip, iend-ip,
	ML_base, ML_bits,
	ML_defaultDTable, dctx->fseEntropy,
	dctx->ddictIsCold, nbSeq);
	RETURN_ERROR_IF(ZSTD_isError(mlhSize), corruption_detected, "ZSTD_buildSeqTable failed");
	ip += mlhSize;
	}
	}

	return ip-istart;
	}


	typedef struct {
	size_t litLength;
	size_t matchLength;
	size_t offset;
	const BYTE* match;
	} seq_t;

	typedef struct {
	size_t state;
	const ZSTD_seqSymbol* table;
	} ZSTD_fseState;

	typedef struct {
	BIT_DStream_t DStream;
	ZSTD_fseState stateLL;
	ZSTD_fseState stateOffb;
	ZSTD_fseState stateML;
	size_t prevOffset[ZSTD_REP_NUM];
	const BYTE* prefixStart;
	const BYTE* dictEnd;
	size_t pos;
	} seqState_t;

	/*! ZSTD_overlapCopy8() :
	* Copies 8 bytes from ip to op and updates op and ip where ip <= op.
	* If the offset is < 8 then the offset is spread to at least 8 bytes.
	*
	* Precondition: ip <= op
	* Postcondition: op - op >= 8
	*/
	HINT_INLINE void ZSTD_overlapCopy8(BYTE op, BYTE const ip, size_t offset) {
	assert(ip <= op);
	if (offset < 8) {
	/* close range match, overlap */
	static const U32 dec32table[] = { 0, 1, 2, 1, 4, 4, 4, 4 }; /* added */
	static const int dec64table[] = { 8, 8, 8, 7, 8, 9,10,11 }; /* subtracted */
	int const sub2 = dec64table[offset];
	(op)[0] = (ip)[0];
	(op)[1] = (ip)[1];
	(op)[2] = (ip)[2];
	(op)[3] = (ip)[3];
	*ip += dec32table[offset];
	ZSTD_copy4(op+4, ip);
	*ip -= sub2;
	} else {
	ZSTD_copy8(op, ip);
	}
	*ip += 8;
	*op += 8;
	assert(op - ip >= 8);
	}

	/*! ZSTD_safecopy() :
	* Specialized version of memcpy() that is allowed to READ up to WILDCOPY_OVERLENGTH past the input buffer
	* and write up to 16 bytes past oend_w (op >= oend_w is allowed).
	* This function is only called in the uncommon case where the sequence is near the end of the block. It
	* should be fast for a single long sequence, but can be slow for several short sequences.
	*
	* @param ovtype controls the overlap detection
	* - ZSTD_no_overlap: The source and destination are guaranteed to be at least WILDCOPY_VECLEN bytes apart.
	* - ZSTD_overlap_src_before_dst: The src and dst may overlap and may be any distance apart.
	* The src buffer must be before the dst buffer.
	*/
	static void ZSTD_safecopy(BYTE* op, BYTE* const oend_w, BYTE const* ip, ptrdiff_t length, ZSTD_overlap_e ovtype) {
	ptrdiff_t const diff = op - ip;
	BYTE* const oend = op + length;

	assert((ovtype == ZSTD_no_overlap && (diff <= -8 \|\| diff >= 8 \|\| op >= oend_w)) \|\|
	(ovtype == ZSTD_overlap_src_before_dst && diff >= 0));

	if (length < 8) {
	/* Handle short lengths. */
	while (op < oend) op++ = ip++;
	return;
	}
	if (ovtype == ZSTD_overlap_src_before_dst) {
	/* Copy 8 bytes and ensure the offset >= 8 when there can be overlap. */
	assert(length >= 8);
	ZSTD_overlapCopy8(&op, &ip, diff);
	assert(op - ip >= 8);
	assert(op <= oend);
	}

	if (oend <= oend_w) {
	/* No risk of overwrite. */
	ZSTD_wildcopy(op, ip, length, ovtype);
	return;
	}
	if (op <= oend_w) {
	/* Wildcopy until we get close to the end. */
	assert(oend > oend_w);
	ZSTD_wildcopy(op, ip, oend_w - op, ovtype);
	ip += oend_w - op;
	op = oend_w;
	}
	/* Handle the leftovers. */
	while (op < oend) op++ = ip++;
	}

	/* ZSTD_execSequenceEnd():
	* This version handles cases that are near the end of the output buffer. It requires
	* more careful checks to make sure there is no overflow. By separating out these hard
	* and unlikely cases, we can speed up the common cases.
	*
	* NOTE: This function needs to be fast for a single long sequence, but doesn't need
	* to be optimized for many small sequences, since those fall into ZSTD_execSequence().
	*/
	FORCE_NOINLINE
	size_t ZSTD_execSequenceEnd(BYTE* op,
	BYTE* const oend, seq_t sequence,
	const BYTE** litPtr, const BYTE* const litLimit,
	const BYTE* const prefixStart, const BYTE* const virtualStart, const BYTE* const dictEnd)
	{
	BYTE* const oLitEnd = op + sequence.litLength;
	size_t const sequenceLength = sequence.litLength + sequence.matchLength;
	const BYTE* const iLitEnd = *litPtr + sequence.litLength;
	const BYTE* match = oLitEnd - sequence.offset;
	BYTE* const oend_w = oend - WILDCOPY_OVERLENGTH;

	/* bounds checks : careful of address space overflow in 32-bit mode */
	RETURN_ERROR_IF(sequenceLength > (size_t)(oend - op), dstSize_tooSmall, "last match must fit within dstBuffer");
	RETURN_ERROR_IF(sequence.litLength > (size_t)(litLimit - *litPtr), corruption_detected, "try to read beyond literal buffer");
	assert(op < op + sequenceLength);
	assert(oLitEnd < op + sequenceLength);

	/* copy literals */
	ZSTD_safecopy(op, oend_w, *litPtr, sequence.litLength, ZSTD_no_overlap);
	op = oLitEnd;
	*litPtr = iLitEnd;

	/* copy Match */
	if (sequence.offset > (size_t)(oLitEnd - prefixStart)) {
	/* offset beyond prefix */
	RETURN_ERROR_IF(sequence.offset > (size_t)(oLitEnd - virtualStart), corruption_detected, "");
	match = dictEnd - (prefixStart-match);
	if (match + sequence.matchLength <= dictEnd) {
	memmove(oLitEnd, match, sequence.matchLength);
	return sequenceLength;
	}
	/* span extDict & currentPrefixSegment */
	{ size_t const length1 = dictEnd - match;
	memmove(oLitEnd, match, length1);
	op = oLitEnd + length1;
	sequence.matchLength -= length1;
	match = prefixStart;
	} }
	ZSTD_safecopy(op, oend_w, match, sequence.matchLength, ZSTD_overlap_src_before_dst);
	return sequenceLength;
	}

	HINT_INLINE
	size_t ZSTD_execSequence(BYTE* op,
	BYTE* const oend, seq_t sequence,
	const BYTE** litPtr, const BYTE* const litLimit,
	const BYTE* const prefixStart, const BYTE* const virtualStart, const BYTE* const dictEnd)
	{
	BYTE* const oLitEnd = op + sequence.litLength;
	size_t const sequenceLength = sequence.litLength + sequence.matchLength;
	BYTE* const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */
	BYTE* const oend_w = oend - WILDCOPY_OVERLENGTH; /* risk : address space underflow on oend=NULL */
	const BYTE* const iLitEnd = *litPtr + sequence.litLength;
	const BYTE* match = oLitEnd - sequence.offset;

	assert(op != NULL /* Precondition */);
	assert(oend_w < oend /* No underflow */);
	/* Handle edge cases in a slow path:
	* - Read beyond end of literals
	* - Match end is within WILDCOPY_OVERLIMIT of oend
	* - 32-bit mode and the match length overflows
	*/
	if (UNLIKELY(
	iLitEnd > litLimit \|\|
	oMatchEnd > oend_w \|\|
	(MEM_32bits() && (size_t)(oend - op) < sequenceLength + WILDCOPY_OVERLENGTH)))
	return ZSTD_execSequenceEnd(op, oend, sequence, litPtr, litLimit, prefixStart, virtualStart, dictEnd);

	/* Assumptions (everything else goes into ZSTD_execSequenceEnd()) */
	assert(op <= oLitEnd /* No overflow */);
	assert(oLitEnd < oMatchEnd /* Non-zero match & no overflow */);
	assert(oMatchEnd <= oend /* No underflow */);
	assert(iLitEnd <= litLimit /* Literal length is in bounds */);
	assert(oLitEnd <= oend_w /* Can wildcopy literals */);
	assert(oMatchEnd <= oend_w /* Can wildcopy matches */);

	/* Copy Literals:
	* Split out litLength <= 16 since it is nearly always true. +1.6% on gcc-9.
	* We likely don't need the full 32-byte wildcopy.
	*/
	assert(WILDCOPY_OVERLENGTH >= 16);
	ZSTD_copy16(op, (*litPtr));
	if (UNLIKELY(sequence.litLength > 16)) {
	ZSTD_wildcopy(op+16, (*litPtr)+16, sequence.litLength-16, ZSTD_no_overlap);
	}
	op = oLitEnd;
	litPtr = iLitEnd; / update for next sequence */

	/* Copy Match */
	if (sequence.offset > (size_t)(oLitEnd - prefixStart)) {
	/* offset beyond prefix -> go into extDict */
	RETURN_ERROR_IF(UNLIKELY(sequence.offset > (size_t)(oLitEnd - virtualStart)), corruption_detected, "");
	match = dictEnd + (match - prefixStart);
	if (match + sequence.matchLength <= dictEnd) {
	memmove(oLitEnd, match, sequence.matchLength);
	return sequenceLength;
	}
	/* span extDict & currentPrefixSegment */
	{ size_t const length1 = dictEnd - match;
	memmove(oLitEnd, match, length1);
	op = oLitEnd + length1;
	sequence.matchLength -= length1;
	match = prefixStart;
	} }
	/* Match within prefix of 1 or more bytes */
	assert(op <= oMatchEnd);
	assert(oMatchEnd <= oend_w);
	assert(match >= prefixStart);
	assert(sequence.matchLength >= 1);

	/* Nearly all offsets are >= WILDCOPY_VECLEN bytes, which means we can use wildcopy
	* without overlap checking.
	*/
	if (LIKELY(sequence.offset >= WILDCOPY_VECLEN)) {
	/* We bet on a full wildcopy for matches, since we expect matches to be
	* longer than literals (in general). In silesia, ~10% of matches are longer
	* than 16 bytes.
	*/
	ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength, ZSTD_no_overlap);
	return sequenceLength;
	}
	assert(sequence.offset < WILDCOPY_VECLEN);

	/* Copy 8 bytes and spread the offset to be >= 8. */
	ZSTD_overlapCopy8(&op, &match, sequence.offset);

	/* If the match length is > 8 bytes, then continue with the wildcopy. */
	if (sequence.matchLength > 8) {
	assert(op < oMatchEnd);
	ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength-8, ZSTD_overlap_src_before_dst);
	}
	return sequenceLength;
	}

	static void
	ZSTD_initFseState(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD, const ZSTD_seqSymbol* dt)
	{
	const void* ptr = dt;
	const ZSTD_seqSymbol_header* const DTableH = (const ZSTD_seqSymbol_header*)ptr;
	DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog);
	DEBUGLOG(6, "ZSTD_initFseState : val=%u using %u bits",
	(U32)DStatePtr->state, DTableH->tableLog);
	BIT_reloadDStream(bitD);
	DStatePtr->table = dt + 1;
	}

	FORCE_INLINE_TEMPLATE void
	ZSTD_updateFseState(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD)
	{
	ZSTD_seqSymbol const DInfo = DStatePtr->table[DStatePtr->state];
	U32 const nbBits = DInfo.nbBits;
	size_t const lowBits = BIT_readBits(bitD, nbBits);
	DStatePtr->state = DInfo.nextState + lowBits;
	}

	FORCE_INLINE_TEMPLATE void
	ZSTD_updateFseStateWithDInfo(ZSTD_fseState* DStatePtr, BIT_DStream_t* bitD, ZSTD_seqSymbol const DInfo)
	{
	U32 const nbBits = DInfo.nbBits;
	size_t const lowBits = BIT_readBits(bitD, nbBits);
	DStatePtr->state = DInfo.nextState + lowBits;
	}

	/* We need to add at most (ZSTD_WINDOWLOG_MAX_32 - 1) bits to read the maximum
	* offset bits. But we can only read at most (STREAM_ACCUMULATOR_MIN_32 - 1)
	* bits before reloading. This value is the maximum number of bytes we read
	* after reloading when we are decoding long offsets.
	*/
	#define LONG_OFFSETS_MAX_EXTRA_BITS_32 \
	(ZSTD_WINDOWLOG_MAX_32 > STREAM_ACCUMULATOR_MIN_32 \
	? ZSTD_WINDOWLOG_MAX_32 - STREAM_ACCUMULATOR_MIN_32 \
	: 0)

	typedef enum { ZSTD_lo_isRegularOffset, ZSTD_lo_isLongOffset=1 } ZSTD_longOffset_e;
	typedef enum { ZSTD_p_noPrefetch=0, ZSTD_p_prefetch=1 } ZSTD_prefetch_e;

	FORCE_INLINE_TEMPLATE seq_t
	ZSTD_decodeSequence(seqState_t* seqState, const ZSTD_longOffset_e longOffsets, const ZSTD_prefetch_e prefetch)
	{
	seq_t seq;
	ZSTD_seqSymbol const llDInfo = seqState->stateLL.table[seqState->stateLL.state];
	ZSTD_seqSymbol const mlDInfo = seqState->stateML.table[seqState->stateML.state];
	ZSTD_seqSymbol const ofDInfo = seqState->stateOffb.table[seqState->stateOffb.state];
	U32 const llBase = llDInfo.baseValue;
	U32 const mlBase = mlDInfo.baseValue;
	U32 const ofBase = ofDInfo.baseValue;
	BYTE const llBits = llDInfo.nbAdditionalBits;
	BYTE const mlBits = mlDInfo.nbAdditionalBits;
	BYTE const ofBits = ofDInfo.nbAdditionalBits;
	BYTE const totalBits = llBits+mlBits+ofBits;

	/* sequence */
	{ size_t offset;
	if (ofBits > 1) {
	ZSTD_STATIC_ASSERT(ZSTD_lo_isLongOffset == 1);
	ZSTD_STATIC_ASSERT(LONG_OFFSETS_MAX_EXTRA_BITS_32 == 5);
	assert(ofBits <= MaxOff);
	if (MEM_32bits() && longOffsets && (ofBits >= STREAM_ACCUMULATOR_MIN_32)) {
	U32 const extraBits = ofBits - MIN(ofBits, 32 - seqState->DStream.bitsConsumed);
	offset = ofBase + (BIT_readBitsFast(&seqState->DStream, ofBits - extraBits) << extraBits);
	BIT_reloadDStream(&seqState->DStream);
	if (extraBits) offset += BIT_readBitsFast(&seqState->DStream, extraBits);
	assert(extraBits <= LONG_OFFSETS_MAX_EXTRA_BITS_32); /* to avoid another reload */
	} else {
	offset = ofBase + BIT_readBitsFast(&seqState->DStream, ofBits/>0/); /* <= (ZSTD_WINDOWLOG_MAX-1) bits */
	if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream);
	}
	seqState->prevOffset[2] = seqState->prevOffset[1];
	seqState->prevOffset[1] = seqState->prevOffset[0];
	seqState->prevOffset[0] = offset;
	} else {
	U32 const ll0 = (llBase == 0);
	if (LIKELY((ofBits == 0))) {
	if (LIKELY(!ll0))
	offset = seqState->prevOffset[0];
	else {
	offset = seqState->prevOffset[1];
	seqState->prevOffset[1] = seqState->prevOffset[0];
	seqState->prevOffset[0] = offset;
	}
	} else {
	offset = ofBase + ll0 + BIT_readBitsFast(&seqState->DStream, 1);
	{ size_t temp = (offset==3) ? seqState->prevOffset[0] - 1 : seqState->prevOffset[offset];
	temp += !temp; /* 0 is not valid; input is corrupted; force offset to 1 */
	if (offset != 1) seqState->prevOffset[2] = seqState->prevOffset[1];
	seqState->prevOffset[1] = seqState->prevOffset[0];
	seqState->prevOffset[0] = offset = temp;
	} } }
	seq.offset = offset;
	}

	seq.matchLength = mlBase;
	if (mlBits > 0)
	seq.matchLength += BIT_readBitsFast(&seqState->DStream, mlBits/>0/);

	if (MEM_32bits() && (mlBits+llBits >= STREAM_ACCUMULATOR_MIN_32-LONG_OFFSETS_MAX_EXTRA_BITS_32))
	BIT_reloadDStream(&seqState->DStream);
	if (MEM_64bits() && UNLIKELY(totalBits >= STREAM_ACCUMULATOR_MIN_64-(LLFSELog+MLFSELog+OffFSELog)))
	BIT_reloadDStream(&seqState->DStream);
	/* Ensure there are enough bits to read the rest of data in 64-bit mode. */
	ZSTD_STATIC_ASSERT(16+LLFSELog+MLFSELog+OffFSELog < STREAM_ACCUMULATOR_MIN_64);

	seq.litLength = llBase;
	if (llBits > 0)
	seq.litLength += BIT_readBitsFast(&seqState->DStream, llBits/>0/);

	if (MEM_32bits())
	BIT_reloadDStream(&seqState->DStream);

	DEBUGLOG(6, "seq: litL=%u, matchL=%u, offset=%u",
	(U32)seq.litLength, (U32)seq.matchLength, (U32)seq.offset);

	if (prefetch == ZSTD_p_prefetch) {
	size_t const pos = seqState->pos + seq.litLength;
	const BYTE* const matchBase = (seq.offset > pos) ? seqState->dictEnd : seqState->prefixStart;
	seq.match = matchBase + pos - seq.offset; /* note : this operation can overflow when seq.offset is really too large, which can only happen when input is corrupted.
	* No consequence though : no memory access will occur, offset is only used for prefetching */
	seqState->pos = pos + seq.matchLength;
	}

	/* ANS state update
	* gcc-9.0.0 does 2.5% worse with ZSTD_updateFseStateWithDInfo().
	* clang-9.2.0 does 7% worse with ZSTD_updateFseState().
	* Naturally it seems like ZSTD_updateFseStateWithDInfo() should be the
	* better option, so it is the default for other compilers. But, if you
	* measure that it is worse, please put up a pull request.
	*/
	{
	#if defined(__GNUC__) && !defined(__clang__)
	const int kUseUpdateFseState = 1;
	#else
	const int kUseUpdateFseState = 0;
	#endif
	if (kUseUpdateFseState) {
	ZSTD_updateFseState(&seqState->stateLL, &seqState->DStream); /* <= 9 bits */
	ZSTD_updateFseState(&seqState->stateML, &seqState->DStream); /* <= 9 bits */
	if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream); /* <= 18 bits */
	ZSTD_updateFseState(&seqState->stateOffb, &seqState->DStream); /* <= 8 bits */
	} else {
	ZSTD_updateFseStateWithDInfo(&seqState->stateLL, &seqState->DStream, llDInfo); /* <= 9 bits */
	ZSTD_updateFseStateWithDInfo(&seqState->stateML, &seqState->DStream, mlDInfo); /* <= 9 bits */
	if (MEM_32bits()) BIT_reloadDStream(&seqState->DStream); /* <= 18 bits */
	ZSTD_updateFseStateWithDInfo(&seqState->stateOffb, &seqState->DStream, ofDInfo); /* <= 8 bits */
	}
	}

	return seq;
	}

	#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
	static int ZSTD_dictionaryIsActive(ZSTD_DCtx const* dctx, BYTE const* prefixStart, BYTE const* oLitEnd)
	{
	size_t const windowSize = dctx->fParams.windowSize;
	/* No dictionary used. */
	if (dctx->dictContentEndForFuzzing == NULL) return 0;
	/* Dictionary is our prefix. */
	if (prefixStart == dctx->dictContentBeginForFuzzing) return 1;
	/* Dictionary is not our ext-dict. */
	if (dctx->dictEnd != dctx->dictContentEndForFuzzing) return 0;
	/* Dictionary is not within our window size. */
	if ((size_t)(oLitEnd - prefixStart) >= windowSize) return 0;
	/* Dictionary is active. */
	return 1;
	}

	MEM_STATIC void ZSTD_assertValidSequence(
	ZSTD_DCtx const* dctx,
	BYTE const* op, BYTE const* oend,
	seq_t const seq,
	BYTE const* prefixStart, BYTE const* virtualStart)
	{
	size_t const windowSize = dctx->fParams.windowSize;
	size_t const sequenceSize = seq.litLength + seq.matchLength;
	BYTE const* const oLitEnd = op + seq.litLength;
	DEBUGLOG(6, "Checking sequence: litL=%u matchL=%u offset=%u",
	(U32)seq.litLength, (U32)seq.matchLength, (U32)seq.offset);
	assert(op <= oend);
	assert((size_t)(oend - op) >= sequenceSize);
	assert(sequenceSize <= ZSTD_BLOCKSIZE_MAX);
	if (ZSTD_dictionaryIsActive(dctx, prefixStart, oLitEnd)) {
	size_t const dictSize = (size_t)((char const)dctx->dictContentEndForFuzzing - (char const)dctx->dictContentBeginForFuzzing);
	/* Offset must be within the dictionary. */
	assert(seq.offset <= (size_t)(oLitEnd - virtualStart));
	assert(seq.offset <= windowSize + dictSize);
	} else {
	/* Offset must be within our window. */
	assert(seq.offset <= windowSize);
	}
	}
	#endif

	#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
	FORCE_INLINE_TEMPLATE size_t
	DONT_VECTORIZE
	ZSTD_decompressSequences_body( ZSTD_DCtx* dctx,
	void* dst, size_t maxDstSize,
	const void* seqStart, size_t seqSize, int nbSeq,
	const ZSTD_longOffset_e isLongOffset,
	const int frame)
	{
	const BYTE* ip = (const BYTE*)seqStart;
	const BYTE* const iend = ip + seqSize;
	BYTE* const ostart = (BYTE* const)dst;
	BYTE* const oend = ostart + maxDstSize;
	BYTE* op = ostart;
	const BYTE* litPtr = dctx->litPtr;
	const BYTE* const litEnd = litPtr + dctx->litSize;
	const BYTE* const prefixStart = (const BYTE*) (dctx->prefixStart);
	const BYTE* const vBase = (const BYTE*) (dctx->virtualStart);
	const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd);
	DEBUGLOG(5, "ZSTD_decompressSequences_body");
	(void)frame;

	/* Regen sequences */
	if (nbSeq) {
	seqState_t seqState;
	size_t error = 0;
	dctx->fseEntropy = 1;
	{ U32 i; for (i=0; i<ZSTD_REP_NUM; i++) seqState.prevOffset[i] = dctx->entropy.rep[i]; }
	RETURN_ERROR_IF(
	ERR_isError(BIT_initDStream(&seqState.DStream, ip, iend-ip)),
	corruption_detected, "");
	ZSTD_initFseState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
	ZSTD_initFseState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
	ZSTD_initFseState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
	assert(dst != NULL);

	ZSTD_STATIC_ASSERT(
	BIT_DStream_unfinished < BIT_DStream_completed &&
	BIT_DStream_endOfBuffer < BIT_DStream_completed &&
	BIT_DStream_completed < BIT_DStream_overflow);

	#if defined(__GNUC__) && defined(__x86_64__)
	/* Align the decompression loop to 32 + 16 bytes.
	*
	* zstd compiled with gcc-9 on an Intel i9-9900k shows 10% decompression
	* speed swings based on the alignment of the decompression loop. This
	* performance swing is caused by parts of the decompression loop falling
	* out of the DSB. The entire decompression loop should fit in the DSB,
	* when it can't we get much worse performance. You can measure if you've
	* hit the good case or the bad case with this perf command for some
	* compressed file test.zst:
	*
	* perf stat -e cycles -e instructions -e idq.all_dsb_cycles_any_uops \
	* -e idq.all_mite_cycles_any_uops -- ./zstd -tq test.zst
	*
	* If you see most cycles served out of the MITE you've hit the bad case.
	* If you see most cycles served out of the DSB you've hit the good case.
	* If it is pretty even then you may be in an okay case.
	*
	* I've been able to reproduce this issue on the following CPUs:
	* - Kabylake: Macbook Pro (15-inch, 2019) 2.4 GHz Intel Core i9
	* Use Instruments->Counters to get DSB/MITE cycles.
	* I never got performance swings, but I was able to
	* go from the good case of mostly DSB to half of the
	* cycles served from MITE.
	* - Coffeelake: Intel i9-9900k
	*
	* I haven't been able to reproduce the instability or DSB misses on any
	* of the following CPUS:
	* - Haswell
	* - Broadwell: Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GH
	* - Skylake
	*
	* If you are seeing performance stability this script can help test.
	* It tests on 4 commits in zstd where I saw performance change.
	*
	* https://gist.github.com/terrelln/9889fc06a423fd5ca6e99351564473f4
	*/
	__asm__(".p2align 5");
	__asm__("nop");
	__asm__(".p2align 4");
	#endif
	for ( ; ; ) {
	seq_t const sequence = ZSTD_decodeSequence(&seqState, isLongOffset, ZSTD_p_noPrefetch);
	size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequence, &litPtr, litEnd, prefixStart, vBase, dictEnd);
	#if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
	assert(!ZSTD_isError(oneSeqSize));
	if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequence, prefixStart, vBase);
	#endif
	DEBUGLOG(6, "regenerated sequence size : %u", (U32)oneSeqSize);
	BIT_reloadDStream(&(seqState.DStream));
	/* gcc and clang both don't like early returns in this loop.
	* gcc doesn't like early breaks either.
	* Instead save an error and report it at the end.
	* When there is an error, don't increment op, so we don't
	* overwrite.
	*/
	if (UNLIKELY(ZSTD_isError(oneSeqSize))) error = oneSeqSize;
	else op += oneSeqSize;
	if (UNLIKELY(!--nbSeq)) break;
	}

	/* check if reached exact end */
	DEBUGLOG(5, "ZSTD_decompressSequences_body: after decode loop, remaining nbSeq : %i", nbSeq);
	if (ZSTD_isError(error)) return error;
	RETURN_ERROR_IF(nbSeq, corruption_detected, "");
	RETURN_ERROR_IF(BIT_reloadDStream(&seqState.DStream) < BIT_DStream_completed, corruption_detected, "");
	/* save reps for next block */
	{ U32 i; for (i=0; i<ZSTD_REP_NUM; i++) dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]); }
	}

	/* last literal segment */
	{ size_t const lastLLSize = litEnd - litPtr;
	RETURN_ERROR_IF(lastLLSize > (size_t)(oend-op), dstSize_tooSmall, "");
	if (op != NULL) {
	memcpy(op, litPtr, lastLLSize);
	op += lastLLSize;
	}
	}

	return op-ostart;
	}

	static size_t
	ZSTD_decompressSequences_default(ZSTD_DCtx* dctx,
	void* dst, size_t maxDstSize,
	const void* seqStart, size_t seqSize, int nbSeq,
	const ZSTD_longOffset_e isLongOffset,
	const int frame)
	{
	return ZSTD_decompressSequences_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
	}
	#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */

	#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
	FORCE_INLINE_TEMPLATE size_t
	ZSTD_decompressSequencesLong_body(
	ZSTD_DCtx* dctx,
	void* dst, size_t maxDstSize,
	const void* seqStart, size_t seqSize, int nbSeq,
	const ZSTD_longOffset_e isLongOffset,
	const int frame)
	{
	const BYTE* ip = (const BYTE*)seqStart;
	const BYTE* const iend = ip + seqSize;
	BYTE* const ostart = (BYTE* const)dst;
	BYTE* const oend = ostart + maxDstSize;
	BYTE* op = ostart;
	const BYTE* litPtr = dctx->litPtr;
	const BYTE* const litEnd = litPtr + dctx->litSize;
	const BYTE* const prefixStart = (const BYTE*) (dctx->prefixStart);
	const BYTE* const dictStart = (const BYTE*) (dctx->virtualStart);
	const BYTE* const dictEnd = (const BYTE*) (dctx->dictEnd);
	(void)frame;

	/* Regen sequences */
	if (nbSeq) {
	#define STORED_SEQS 4
	#define STORED_SEQS_MASK (STORED_SEQS-1)
	#define ADVANCED_SEQS 4
	seq_t sequences[STORED_SEQS];
	int const seqAdvance = MIN(nbSeq, ADVANCED_SEQS);
	seqState_t seqState;
	int seqNb;
	dctx->fseEntropy = 1;
	{ int i; for (i=0; i<ZSTD_REP_NUM; i++) seqState.prevOffset[i] = dctx->entropy.rep[i]; }
	seqState.prefixStart = prefixStart;
	seqState.pos = (size_t)(op-prefixStart);
	seqState.dictEnd = dictEnd;
	assert(dst != NULL);
	assert(iend >= ip);
	RETURN_ERROR_IF(
	ERR_isError(BIT_initDStream(&seqState.DStream, ip, iend-ip)),
	corruption_detected, "");
	ZSTD_initFseState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
	ZSTD_initFseState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
	ZSTD_initFseState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);

	/* prepare in advance */
	for (seqNb=0; (BIT_reloadDStream(&seqState.DStream) <= BIT_DStream_completed) && (seqNb<seqAdvance); seqNb++) {
	sequences[seqNb] = ZSTD_decodeSequence(&seqState, isLongOffset, ZSTD_p_prefetch);
	PREFETCH_L1(sequences[seqNb].match); PREFETCH_L1(sequences[seqNb].match + sequences[seqNb].matchLength - 1); /* note : it's safe to invoke PREFETCH() on any memory address, including invalid ones */
	}
	RETURN_ERROR_IF(seqNb<seqAdvance, corruption_detected, "");

	/* decode and decompress */
	for ( ; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && (seqNb<nbSeq) ; seqNb++) {
	seq_t const sequence = ZSTD_decodeSequence(&seqState, isLongOffset, ZSTD_p_prefetch);
	size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequences[(seqNb-ADVANCED_SEQS) & STORED_SEQS_MASK], &litPtr, litEnd, prefixStart, dictStart, dictEnd);
	#if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
	assert(!ZSTD_isError(oneSeqSize));
	if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequences[(seqNb-ADVANCED_SEQS) & STORED_SEQS_MASK], prefixStart, dictStart);
	#endif
	if (ZSTD_isError(oneSeqSize)) return oneSeqSize;
	PREFETCH_L1(sequence.match); PREFETCH_L1(sequence.match + sequence.matchLength - 1); /* note : it's safe to invoke PREFETCH() on any memory address, including invalid ones */
	sequences[seqNb & STORED_SEQS_MASK] = sequence;
	op += oneSeqSize;
	}
	RETURN_ERROR_IF(seqNb<nbSeq, corruption_detected, "");

	/* finish queue */
	seqNb -= seqAdvance;
	for ( ; seqNb<nbSeq ; seqNb++) {
	size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequences[seqNb&STORED_SEQS_MASK], &litPtr, litEnd, prefixStart, dictStart, dictEnd);
	#if defined(FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION) && defined(FUZZING_ASSERT_VALID_SEQUENCE)
	assert(!ZSTD_isError(oneSeqSize));
	if (frame) ZSTD_assertValidSequence(dctx, op, oend, sequences[seqNb&STORED_SEQS_MASK], prefixStart, dictStart);
	#endif
	if (ZSTD_isError(oneSeqSize)) return oneSeqSize;
	op += oneSeqSize;
	}

	/* save reps for next block */
	{ U32 i; for (i=0; i<ZSTD_REP_NUM; i++) dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]); }
	}

	/* last literal segment */
	{ size_t const lastLLSize = litEnd - litPtr;
	RETURN_ERROR_IF(lastLLSize > (size_t)(oend-op), dstSize_tooSmall, "");
	if (op != NULL) {
	memcpy(op, litPtr, lastLLSize);
	op += lastLLSize;
	}
	}

	return op-ostart;
	}

	static size_t
	ZSTD_decompressSequencesLong_default(ZSTD_DCtx* dctx,
	void* dst, size_t maxDstSize,
	const void* seqStart, size_t seqSize, int nbSeq,
	const ZSTD_longOffset_e isLongOffset,
	const int frame)
	{
	return ZSTD_decompressSequencesLong_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
	}
	#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */



	#if DYNAMIC_BMI2

	#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
	static TARGET_ATTRIBUTE("bmi2") size_t
	DONT_VECTORIZE
	ZSTD_decompressSequences_bmi2(ZSTD_DCtx* dctx,
	void* dst, size_t maxDstSize,
	const void* seqStart, size_t seqSize, int nbSeq,
	const ZSTD_longOffset_e isLongOffset,
	const int frame)
	{
	return ZSTD_decompressSequences_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
	}
	#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */

	#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
	static TARGET_ATTRIBUTE("bmi2") size_t
	ZSTD_decompressSequencesLong_bmi2(ZSTD_DCtx* dctx,
	void* dst, size_t maxDstSize,
	const void* seqStart, size_t seqSize, int nbSeq,
	const ZSTD_longOffset_e isLongOffset,
	const int frame)
	{
	return ZSTD_decompressSequencesLong_body(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
	}
	#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */

	#endif /* DYNAMIC_BMI2 */

	typedef size_t (*ZSTD_decompressSequences_t)(
	ZSTD_DCtx* dctx,
	void* dst, size_t maxDstSize,
	const void* seqStart, size_t seqSize, int nbSeq,
	const ZSTD_longOffset_e isLongOffset,
	const int frame);

	#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
	static size_t
	ZSTD_decompressSequences(ZSTD_DCtx* dctx, void* dst, size_t maxDstSize,
	const void* seqStart, size_t seqSize, int nbSeq,
	const ZSTD_longOffset_e isLongOffset,
	const int frame)
	{
	DEBUGLOG(5, "ZSTD_decompressSequences");
	#if DYNAMIC_BMI2
	if (dctx->bmi2) {
	return ZSTD_decompressSequences_bmi2(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
	}
	#endif
	return ZSTD_decompressSequences_default(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
	}
	#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG */


	#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
	/* ZSTD_decompressSequencesLong() :
	* decompression function triggered when a minimum share of offsets is considered "long",
	* aka out of cache.
	* note : "long" definition seems overloaded here, sometimes meaning "wider than bitstream register", and sometimes meaning "farther than memory cache distance".
	* This function will try to mitigate main memory latency through the use of prefetching */
	static size_t
	ZSTD_decompressSequencesLong(ZSTD_DCtx* dctx,
	void* dst, size_t maxDstSize,
	const void* seqStart, size_t seqSize, int nbSeq,
	const ZSTD_longOffset_e isLongOffset,
	const int frame)
	{
	DEBUGLOG(5, "ZSTD_decompressSequencesLong");
	#if DYNAMIC_BMI2
	if (dctx->bmi2) {
	return ZSTD_decompressSequencesLong_bmi2(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
	}
	#endif
	return ZSTD_decompressSequencesLong_default(dctx, dst, maxDstSize, seqStart, seqSize, nbSeq, isLongOffset, frame);
	}
	#endif /* ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT */



	#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
	!defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
	/* ZSTD_getLongOffsetsShare() :
	* condition : offTable must be valid
	* @return : "share" of long offsets (arbitrarily defined as > (1<<23))
	* compared to maximum possible of (1<<OffFSELog) */
	static unsigned
	ZSTD_getLongOffsetsShare(const ZSTD_seqSymbol* offTable)
	{
	const void* ptr = offTable;
	U32 const tableLog = ((const ZSTD_seqSymbol_header*)ptr)[0].tableLog;
	const ZSTD_seqSymbol* table = offTable + 1;
	U32 const max = 1 << tableLog;
	U32 u, total = 0;
	DEBUGLOG(5, "ZSTD_getLongOffsetsShare: (tableLog=%u)", tableLog);

	assert(max <= (1 << OffFSELog)); /* max not too large */
	for (u=0; u<max; u++) {
	if (table[u].nbAdditionalBits > 22) total += 1;
	}

	assert(tableLog <= OffFSELog);
	total <<= (OffFSELog - tableLog); /* scale to OffFSELog */

	return total;
	}
	#endif

	size_t
	ZSTD_decompressBlock_internal(ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize, const int frame)
	{ /* blockType == blockCompressed */
	const BYTE* ip = (const BYTE*)src;
	/* isLongOffset must be true if there are long offsets.
	* Offsets are long if they are larger than 2^STREAM_ACCUMULATOR_MIN.
	* We don't expect that to be the case in 64-bit mode.
	* In block mode, window size is not known, so we have to be conservative.
	* (note: but it could be evaluated from current-lowLimit)
	*/
	ZSTD_longOffset_e const isLongOffset = (ZSTD_longOffset_e)(MEM_32bits() && (!frame \|\| (dctx->fParams.windowSize > (1ULL << STREAM_ACCUMULATOR_MIN))));
	DEBUGLOG(5, "ZSTD_decompressBlock_internal (size : %u)", (U32)srcSize);

	RETURN_ERROR_IF(srcSize >= ZSTD_BLOCKSIZE_MAX, srcSize_wrong, "");

	/* Decode literals section */
	{ size_t const litCSize = ZSTD_decodeLiteralsBlock(dctx, src, srcSize);
	DEBUGLOG(5, "ZSTD_decodeLiteralsBlock : %u", (U32)litCSize);
	if (ZSTD_isError(litCSize)) return litCSize;
	ip += litCSize;
	srcSize -= litCSize;
	}

	/* Build Decoding Tables */
	{
	/* These macros control at build-time which decompressor implementation
	* we use. If neither is defined, we do some inspection and dispatch at
	* runtime.
	*/
	#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
	!defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
	int usePrefetchDecoder = dctx->ddictIsCold;
	#endif
	int nbSeq;
	size_t const seqHSize = ZSTD_decodeSeqHeaders(dctx, &nbSeq, ip, srcSize);
	if (ZSTD_isError(seqHSize)) return seqHSize;
	ip += seqHSize;
	srcSize -= seqHSize;

	RETURN_ERROR_IF(dst == NULL && nbSeq > 0, dstSize_tooSmall, "NULL not handled");

	#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
	!defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
	if ( !usePrefetchDecoder
	&& (!frame \|\| (dctx->fParams.windowSize > (1<<24)))
	&& (nbSeq>ADVANCED_SEQS) ) { /* could probably use a larger nbSeq limit */
	U32 const shareLongOffsets = ZSTD_getLongOffsetsShare(dctx->OFTptr);
	U32 const minShare = MEM_64bits() ? 7 : 20; /* heuristic values, correspond to 2.73% and 7.81% */
	usePrefetchDecoder = (shareLongOffsets >= minShare);
	}
	#endif

	dctx->ddictIsCold = 0;

	#if !defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT) && \
	!defined(ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG)
	if (usePrefetchDecoder)
	#endif
	#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_SHORT
	return ZSTD_decompressSequencesLong(dctx, dst, dstCapacity, ip, srcSize, nbSeq, isLongOffset, frame);
	#endif

	#ifndef ZSTD_FORCE_DECOMPRESS_SEQUENCES_LONG
	/* else */
	return ZSTD_decompressSequences(dctx, dst, dstCapacity, ip, srcSize, nbSeq, isLongOffset, frame);
	#endif
	}
	}


	void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst)
	{
	if (dst != dctx->previousDstEnd) { /* not contiguous */
	dctx->dictEnd = dctx->previousDstEnd;
	dctx->virtualStart = (const char)dst - ((const char)(dctx->previousDstEnd) - (const char*)(dctx->prefixStart));
	dctx->prefixStart = dst;
	dctx->previousDstEnd = dst;
	}
	}


	size_t ZSTD_decompressBlock(ZSTD_DCtx* dctx,
	void* dst, size_t dstCapacity,
	const void* src, size_t srcSize)
	{
	size_t dSize;
	ZSTD_checkContinuity(dctx, dst);
	dSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize, /* frame */ 0);
	dctx->previousDstEnd = (char*)dst + dSize;
	return dSize;
	}
	/** ended inlining decompress/zstd_decompress_block.c **/
	diff --git a/sys/contrib/openzfs/rpm/generic/zfs-dkms.spec.in b/sys/contrib/openzfs/rpm/generic/zfs-dkms.spec.in
	index e0e0efe2a1de..02be716aa964 100644
	--- a/sys/contrib/openzfs/rpm/generic/zfs-dkms.spec.in
	+++ b/sys/contrib/openzfs/rpm/generic/zfs-dkms.spec.in
	@@ -1,109 +1,113 @@
	%{?!packager: %define packager Brian Behlendorf <behlendorf1@llnl.gov>}

	%if ! 0%{?rhel}%{?fedora}%{?mageia}%{?suse_version}
	%define not_rpm 1
	%endif

	# Exclude input files from mangling
	%global __brp_mangle_shebangs_exclude_from ^/usr/src/.*$

	%define module @PACKAGE@
	%define mkconf scripts/dkms.mkconf

	Name: %{module}-dkms

	Version: @VERSION@
	Release: @RELEASE@%{?dist}
	Summary: Kernel module(s) (dkms)

	Group: System Environment/Kernel
	License: @ZFS_META_LICENSE@
	URL: https://github.com/openzfs/zfs
	Source0: %{module}-%{version}.tar.gz
	BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
	BuildArch: noarch

	Requires: dkms >= 2.2.0.3
	+Requires(post): dkms >= 2.2.0.3
	+Requires(preun): dkms >= 2.2.0.3
	Requires: gcc, make, perl, diffutils
	+Requires(post): gcc, make, perl, diffutils
	%if 0%{?rhel}%{?fedora}%{?mageia}%{?suse_version}
	Requires: kernel-devel >= @ZFS_META_KVER_MIN@, kernel-devel <= @ZFS_META_KVER_MAX@.999
	+Requires(post): kernel-devel >= @ZFS_META_KVER_MIN@, kernel-devel <= @ZFS_META_KVER_MAX@.999
	Obsoletes: spl-dkms
	%endif
	Provides: %{module}-kmod = %{version}
	AutoReqProv: no

	%if 0%{?rhel}%{?fedora}%{?suse_version}
	# We don't directly use it, but if this isn't installed, rpmbuild as root can
	# crash+corrupt rpmdb
	# See issue #12071
	BuildRequires: ncompress
	%endif

	%description
	This package contains the dkms ZFS kernel modules.

	%prep
	%setup -q -n %{module}-%{version}

	%build
	%{mkconf} -n %{module} -v %{version} -f dkms.conf

	%install
	if [ "$RPM_BUILD_ROOT" != "/" ]; then
	rm -rf $RPM_BUILD_ROOT
	fi
	mkdir -p $RPM_BUILD_ROOT/usr/src/
	cp -rf ${RPM_BUILD_DIR}/%{module}-%{version} $RPM_BUILD_ROOT/usr/src/

	%clean
	if [ "$RPM_BUILD_ROOT" != "/" ]; then
	rm -rf $RPM_BUILD_ROOT
	fi

	%files
	%defattr(-,root,root)
	/usr/src/%{module}-%{version}

	%post
	for POSTINST in /usr/lib/dkms/common.postinst; do
	if [ -f $POSTINST ]; then
	$POSTINST %{module} %{version}
	exit $?
	fi
	echo "WARNING: $POSTINST does not exist."
	done
	echo -e "ERROR: DKMS version is too old and %{module} was not"
	echo -e "built with legacy DKMS support."
	echo -e "You must either rebuild %{module} with legacy postinst"
	echo -e "support or upgrade DKMS to a more current version."
	exit 1

	%preun
	# Are we doing an upgrade?
	if [ "$1" = "1" -o "$1" = "upgrade" ] ; then
	# Yes we are. Are we upgrading to a new ZFS version?
	NEWEST_VER=$(dkms status zfs \| tr -d , \| sort -r -V \| awk '/installed/{print $2; exit}')
	if [ "$NEWEST_VER" != "%{version}" ] ; then
	# Yes, it's a new ZFS version. We'll uninstall the old module
	# later on in this script.
	true
	else
	# No, it's probably an upgrade of the same ZFS version
	# to a new distro (zfs-dkms-0.7.12.fc28->zfs-dkms-0.7.12.fc29).
	# Don't remove our modules, since the rebuild for the new
	# distro will automatically delete the old modules.
	exit 0
	fi
	fi

	# If we're here then we're doing an uninstall (not upgrade).
	CONFIG_H="/var/lib/dkms/%{module}/%{version}///%{module}_config.h"
	SPEC_META_ALIAS="@PACKAGE@-@VERSION@-@RELEASE@"
	DKMS_META_ALIAS=`cat $CONFIG_H 2>/dev/null \|
	awk -F'"' '/META_ALIAS\s+"/ { print $2; exit 0 }'`
	if [ "$SPEC_META_ALIAS" = "$DKMS_META_ALIAS" ]; then
	echo -e
	echo -e "Uninstall of %{module} module ($SPEC_META_ALIAS) beginning:"
	dkms remove -m %{module} -v %{version} --all %{!?not_rpm:--rpm_safe_upgrade}
	fi
	exit 0
	diff --git a/sys/contrib/openzfs/rpm/generic/zfs.spec.in b/sys/contrib/openzfs/rpm/generic/zfs.spec.in
	index 4a37ae8ce1d5..25beadce7c1f 100644
	--- a/sys/contrib/openzfs/rpm/generic/zfs.spec.in
	+++ b/sys/contrib/openzfs/rpm/generic/zfs.spec.in
	@@ -1,567 +1,569 @@
	%global _sbindir /sbin
	%global _libdir /%{_lib}

	# Set the default udev directory based on distribution.
	%if %{undefined _udevdir}
	%if 0%{?fedora} >= 17 \|\| 0%{?rhel} >= 7 \|\| 0%{?centos} >= 7
	%global _udevdir %{_prefix}/lib/udev
	%else
	%global _udevdir /lib/udev
	%endif
	%endif

	# Set the default udevrule directory based on distribution.
	%if %{undefined _udevruledir}
	%if 0%{?fedora} >= 17 \|\| 0%{?rhel} >= 7 \|\| 0%{?centos} >= 7
	%global _udevruledir %{_prefix}/lib/udev/rules.d
	%else
	%global _udevruledir /lib/udev/rules.d
	%endif
	%endif

	# Set the default dracut directory based on distribution.
	%if %{undefined _dracutdir}
	%if 0%{?fedora} >= 17 \|\| 0%{?rhel} >= 7 \|\| 0%{?centos} >= 7
	%global _dracutdir %{_prefix}/lib/dracut
	%else
	%global _dracutdir %{_prefix}/share/dracut
	%endif
	%endif

	%if %{undefined _initconfdir}
	%global _initconfdir /etc/sysconfig
	%endif

	%if %{undefined _unitdir}
	%global _unitdir %{_prefix}/lib/systemd/system
	%endif

	%if %{undefined _presetdir}
	%global _presetdir %{_prefix}/lib/systemd/system-preset
	%endif

	%if %{undefined _modulesloaddir}
	%global _modulesloaddir %{_prefix}/lib/modules-load.d
	%endif

	%if %{undefined _systemdgeneratordir}
	%global _systemdgeneratordir %{_prefix}/lib/systemd/system-generators
	%endif

	%if %{undefined _pkgconfigdir}
	%global _pkgconfigdir %{_prefix}/%{_lib}/pkgconfig
	%endif

	%bcond_with debug
	%bcond_with debuginfo
	%bcond_with asan
	%bcond_with systemd
	%bcond_with pam

	# Generic enable switch for systemd
	%if %{with systemd}
	%define _systemd 1
	%endif

	# RHEL >= 7 comes with systemd
	%if 0%{?rhel} >= 7
	%define _systemd 1
	%endif

	# Fedora >= 15 comes with systemd, but only >= 18 has
	# the proper macros
	%if 0%{?fedora} >= 18
	%define _systemd 1
	%endif

	# opensuse >= 12.1 comes with systemd, but only >= 13.1
	# has the proper macros
	%if 0%{?suse_version} >= 1310
	%define _systemd 1
	%endif

	# When not specified default to distribution provided version. This
	# is normally Python 3, but for RHEL <= 7 only Python 2 is provided.
	%if %{undefined __use_python}
	%if 0%{?rhel} && 0%{?rhel} <= 7
	%define __python /usr/bin/python2
	%define __python_pkg_version 2
	%define __python_cffi_pkg python-cffi
	%define __python_setuptools_pkg python-setuptools
	%else
	%define __python /usr/bin/python3
	%define __python_pkg_version 3
	%define __python_cffi_pkg python3-cffi
	%define __python_setuptools_pkg python3-setuptools
	%endif
	%else
	%define __python %{__use_python}
	%define __python_pkg_version %{__use_python_pkg_version}
	%define __python_cffi_pkg python%{__python_pkg_version}-cffi
	%define __python_setuptools_pkg python%{__python_pkg_version}-setuptools
	%endif
	%define __python_sitelib %(%{__python} -Esc "from distutils.sysconfig import get_python_lib; print(get_python_lib())")

	# By default python-pyzfs is enabled, with the exception of
	# RHEL 6 which by default uses Python 2.6 which is too old.
	%if 0%{?rhel} == 6
	%bcond_with pyzfs
	%else
	%bcond_without pyzfs
	%endif

	Name: @PACKAGE@
	Version: @VERSION@
	Release: @RELEASE@%{?dist}
	Summary: Commands to control the kernel modules and libraries

	Group: System Environment/Kernel
	License: @ZFS_META_LICENSE@
	URL: https://github.com/openzfs/zfs
	Source0: %{name}-%{version}.tar.gz
	BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
	Requires: libzpool5 = %{version}
	Requires: libnvpair3 = %{version}
	Requires: libuutil3 = %{version}
	Requires: libzfs5 = %{version}
	Requires: %{name}-kmod = %{version}
	Provides: %{name}-kmod-common = %{version}
	Obsoletes: spl

	# zfs-fuse provides the same commands and man pages that OpenZFS does.
	# Renaming those on either side would conflict with all available documentation.
	Conflicts: zfs-fuse

	%if 0%{?rhel}%{?fedora}%{?suse_version}
	BuildRequires: gcc, make
	BuildRequires: zlib-devel
	BuildRequires: libuuid-devel
	BuildRequires: libblkid-devel
	BuildRequires: libudev-devel
	BuildRequires: libattr-devel
	BuildRequires: openssl-devel
	# We don't directly use it, but if this isn't installed, rpmbuild as root can
	# crash+corrupt rpmdb
	# See issue #12071
	BuildRequires: ncompress
	%if 0%{?fedora} >= 28 \|\| 0%{?rhel} >= 8 \|\| 0%{?centos} >= 8
	BuildRequires: libtirpc-devel
	%endif

	Requires: openssl
	%if 0%{?_systemd}
	BuildRequires: systemd
	%endif
	%endif

	%if 0%{?_systemd}
	Requires(post): systemd
	Requires(preun): systemd
	Requires(postun): systemd
	%endif

	# The zpool iostat/status -c scripts call some utilities like lsblk and iostat
	Requires: util-linux
	Requires: sysstat

	%description
	This package contains the core ZFS command line utilities.

	%package -n libzpool5
	Summary: Native ZFS pool library for Linux
	Group: System Environment/Kernel
	Obsoletes: libzpool2
	Obsoletes: libzpool4

	%description -n libzpool5
	This package contains the zpool library, which provides support
	for managing zpools

	%if %{defined ldconfig_scriptlets}
	%ldconfig_scriptlets -n libzpool5
	%else
	%post -n libzpool5 -p /sbin/ldconfig
	%postun -n libzpool5 -p /sbin/ldconfig
	%endif

	%package -n libnvpair3
	Summary: Solaris name-value library for Linux
	Group: System Environment/Kernel
	Obsoletes: libnvpair1

	%description -n libnvpair3
	This package contains routines for packing and unpacking name-value
	pairs. This functionality is used to portably transport data across
	process boundaries, between kernel and user space, and can be used
	to write self describing data structures on disk.

	%if %{defined ldconfig_scriptlets}
	%ldconfig_scriptlets -n libnvpair3
	%else
	%post -n libnvpair3 -p /sbin/ldconfig
	%postun -n libnvpair3 -p /sbin/ldconfig
	%endif

	%package -n libuutil3
	Summary: Solaris userland utility library for Linux
	Group: System Environment/Kernel
	Obsoletes: libuutil1

	%description -n libuutil3
	This library provides a variety of compatibility functions for OpenZFS:
	* libspl: The Solaris Porting Layer userland library, which provides APIs
	that make it possible to run Solaris user code in a Linux environment
	with relatively minimal modification.
	* libavl: The Adelson-Velskii Landis balanced binary tree manipulation
	library.
	* libefi: The Extensible Firmware Interface library for GUID disk
	partitioning.
	* libshare: NFS, SMB, and iSCSI service integration for ZFS.

	%if %{defined ldconfig_scriptlets}
	%ldconfig_scriptlets -n libuutil3
	%else
	%post -n libuutil3 -p /sbin/ldconfig
	%postun -n libuutil3 -p /sbin/ldconfig
	%endif

	# The library version is encoded in the package name. When updating the
	# version information it is important to add an obsoletes line below for
	# the previous version of the package.
	%package -n libzfs5
	Summary: Native ZFS filesystem library for Linux
	Group: System Environment/Kernel
	Obsoletes: libzfs2
	Obsoletes: libzfs4

	%description -n libzfs5
	This package provides support for managing ZFS filesystems

	%if %{defined ldconfig_scriptlets}
	%ldconfig_scriptlets -n libzfs5
	%else
	%post -n libzfs5 -p /sbin/ldconfig
	%postun -n libzfs5 -p /sbin/ldconfig
	%endif

	%package -n libzfs5-devel
	Summary: Development headers
	Group: System Environment/Kernel
	Requires: libzfs5 = %{version}
	Requires: libzpool5 = %{version}
	Requires: libnvpair3 = %{version}
	Requires: libuutil3 = %{version}
	Provides: libzpool5-devel
	Provides: libnvpair3-devel
	Provides: libuutil3-devel
	Obsoletes: zfs-devel
	Obsoletes: libzfs2-devel
	Obsoletes: libzfs4-devel

	%description -n libzfs5-devel
	This package contains the header files needed for building additional
	applications against the ZFS libraries.

	%package test
	Summary: Test infrastructure
	Group: System Environment/Kernel
	Requires: %{name}%{?_isa} = %{version}-%{release}
	Requires: parted
	Requires: lsscsi
	Requires: mdadm
	Requires: bc
	Requires: ksh
	Requires: fio
	Requires: acl
	Requires: sudo
	Requires: sysstat
	Requires: libaio
	Requires: python%{__python_pkg_version}
	%if 0%{?rhel}%{?fedora}%{?suse_version}
	BuildRequires: libaio-devel
	%endif
	AutoReqProv: no

	%description test
	This package contains test infrastructure and support scripts for
	validating the file system.

	%package dracut
	Summary: Dracut module
	Group: System Environment/Kernel
	BuildArch: noarch
	Requires: %{name} >= %{version}
	Requires: dracut
	Requires: /usr/bin/awk
	Requires: grep

	%description dracut
	This package contains a dracut module used to construct an initramfs
	image which is ZFS aware.

	%if %{with pyzfs}
	%package -n python%{__python_pkg_version}-pyzfs
	Summary: Python %{python_version} wrapper for libzfs_core
	Group: Development/Languages/Python
	License: Apache-2.0
	BuildArch: noarch
	Requires: libzfs5 = %{version}
	Requires: libnvpair3 = %{version}
	Requires: libffi
	Requires: python%{__python_pkg_version}
	Requires: %{__python_cffi_pkg}
	%if 0%{?rhel}%{?fedora}%{?suse_version}
	%if 0%{?rhel} >= 8 \|\| 0%{?centos} >= 8 \|\| 0%{?fedora} >= 28
	BuildRequires: python3-packaging
	%else
	BuildRequires: python-packaging
	%endif
	BuildRequires: python%{__python_pkg_version}-devel
	BuildRequires: %{__python_cffi_pkg}
	BuildRequires: %{__python_setuptools_pkg}
	BuildRequires: libffi-devel
	%endif

	%description -n python%{__python_pkg_version}-pyzfs
	This package provides a python wrapper for the libzfs_core C library.
	%endif

	%if 0%{?_initramfs}
	%package initramfs
	Summary: Initramfs module
	Group: System Environment/Kernel
	Requires: %{name}%{?_isa} = %{version}-%{release}
	Requires: %{name} = %{version}-%{release}
	Requires: initramfs-tools

	%description initramfs
	This package contains a initramfs module used to construct an initramfs
	image which is ZFS aware.
	%endif

	%prep
	%if %{with debug}
	%define debug --enable-debug
	%else
	%define debug --disable-debug
	%endif

	%if %{with debuginfo}
	%define debuginfo --enable-debuginfo
	%else
	%define debuginfo --disable-debuginfo
	%endif

	%if %{with asan}
	%define asan --enable-asan
	%else
	%define asan --disable-asan
	%endif

	%if 0%{?_systemd}
	%define systemd --enable-systemd --with-systemdunitdir=%{_unitdir} --with-systemdpresetdir=%{_presetdir} --with-systemdmodulesloaddir=%{_modulesloaddir} --with-systemdgeneratordir=%{_systemdgeneratordir} --disable-sysvinit
	%define systemd_svcs zfs-import-cache.service zfs-import-scan.service zfs-mount.service zfs-share.service zfs-zed.service zfs.target zfs-import.target zfs-volume-wait.service zfs-volumes.target
	%else
	%define systemd --enable-sysvinit --disable-systemd
	%endif

	%if %{with pyzfs}
	%define pyzfs --enable-pyzfs
	%else
	%define pyzfs --disable-pyzfs
	%endif

	%if %{with pam}
	%define pam --enable-pam
	%else
	%define pam --disable-pam
	%endif

	%setup -q

	%build
	%configure \
	--with-config=user \
	--with-udevdir=%{_udevdir} \
	--with-udevruledir=%{_udevruledir} \
	--with-dracutdir=%{_dracutdir} \
	--with-pamconfigsdir=%{_datadir}/pam-configs \
	--with-pammoduledir=%{_libdir}/security \
	--with-python=%{__python} \
	--with-pkgconfigdir=%{_pkgconfigdir} \
	--disable-static \
	%{debug} \
	%{debuginfo} \
	%{asan} \
	%{systemd} \
	%{pam} \
	%{pyzfs}
	make %{?_smp_mflags}

	%install
	%{__rm} -rf $RPM_BUILD_ROOT
	make install DESTDIR=%{?buildroot}
	find %{?buildroot}%{_libdir} -name '*.la' -exec rm -f {} \;
	%if 0%{!?__brp_mangle_shebangs:1}
	find %{?buildroot}%{_bindir} \
	$ -name arc_summary -or -name arcstat -or -name dbufstat $ \
	-exec %{__sed} -i 's\|^#!.*\|#!%{__python}\|' {} \;
	find %{?buildroot}%{_datadir} \
	$ -name test-runner.py -or -name zts-report.py $ \
	-exec %{__sed} -i 's\|^#!.*\|#!%{__python}\|' {} \;
	%endif

	%post
	%if 0%{?_systemd}
	%if 0%{?systemd_post:1}
	%systemd_post %{systemd_svcs}
	%else
	if [ "$1" = "1" -o "$1" = "install" ] ; then
	# Initial installation
	systemctl preset %{systemd_svcs} >/dev/null \|\| true
	fi
	%endif
	%else
	if [ -x /sbin/chkconfig ]; then
	/sbin/chkconfig --add zfs-import
	+ /sbin/chkconfig --add zfs-load-key
	/sbin/chkconfig --add zfs-mount
	/sbin/chkconfig --add zfs-share
	/sbin/chkconfig --add zfs-zed
	fi
	%endif
	exit 0

	# On RHEL/CentOS 7 the static nodes aren't refreshed by default after
	# installing a package. This is the default behavior for Fedora.
	%posttrans
	%if 0%{?rhel} == 7 \|\| 0%{?centos} == 7
	systemctl restart kmod-static-nodes
	systemctl restart systemd-tmpfiles-setup-dev
	udevadm trigger
	%endif

	%preun
	%if 0%{?_systemd}
	%if 0%{?systemd_preun:1}
	%systemd_preun %{systemd_svcs}
	%else
	if [ "$1" = "0" -o "$1" = "remove" ] ; then
	# Package removal, not upgrade
	systemctl --no-reload disable %{systemd_svcs} >/dev/null \|\| true
	systemctl stop %{systemd_svcs} >/dev/null \|\| true
	fi
	%endif
	%else
	if [ "$1" = "0" -o "$1" = "remove" ] && [ -x /sbin/chkconfig ]; then
	/sbin/chkconfig --del zfs-import
	+ /sbin/chkconfig --del zfs-load-key
	/sbin/chkconfig --del zfs-mount
	/sbin/chkconfig --del zfs-share
	/sbin/chkconfig --del zfs-zed
	fi
	%endif
	exit 0

	%postun
	%if 0%{?_systemd}
	%if 0%{?systemd_postun:1}
	%systemd_postun %{systemd_svcs}
	%else
	systemctl --system daemon-reload >/dev/null \|\| true
	%endif
	%endif

	%files
	# Core utilities
	%{_sbindir}/*
	%{_bindir}/raidz_test
	%{_sbindir}/zgenhostid
	%{_bindir}/zvol_wait
	# Optional Python 2/3 scripts
	%{_bindir}/arc_summary
	%{_bindir}/arcstat
	%{_bindir}/dbufstat
	# Man pages
	%{_mandir}/man1/*
	%{_mandir}/man4/*
	%{_mandir}/man5/*
	%{_mandir}/man7/*
	%{_mandir}/man8/*
	# Configuration files and scripts
	%{_libexecdir}/%{name}
	%{_udevdir}/vdev_id
	%{_udevdir}/zvol_id
	%{_udevdir}/rules.d/*
	%{_datadir}/%{name}/compatibility.d
	%if ! 0%{?_systemd} \|\| 0%{?_initramfs}
	# Files needed for sysvinit and initramfs-tools
	%{_sysconfdir}/%{name}/zfs-functions
	%config(noreplace) %{_initconfdir}/zfs
	%else
	%exclude %{_sysconfdir}/%{name}/zfs-functions
	%exclude %{_initconfdir}/zfs
	%endif
	%if 0%{?_systemd}
	%{_unitdir}/*
	%{_presetdir}/*
	%{_modulesloaddir}/*
	%{_systemdgeneratordir}/*
	%else
	%config(noreplace) %{_sysconfdir}/init.d/*
	%endif
	%config(noreplace) %{_sysconfdir}/%{name}/zed.d/*
	%config(noreplace) %{_sysconfdir}/%{name}/zpool.d/*
	%config(noreplace) %{_sysconfdir}/%{name}/vdev_id.conf.*.example
	%attr(440, root, root) %config(noreplace) %{_sysconfdir}/sudoers.d/*
	%if %{with pam}
	%{_libdir}/security/*
	%{_datadir}/pam-configs/*
	%endif

	%files -n libzpool5
	%{_libdir}/libzpool.so.*

	%files -n libnvpair3
	%{_libdir}/libnvpair.so.*

	%files -n libuutil3
	%{_libdir}/libuutil.so.*

	%files -n libzfs5
	%{_libdir}/libzfs.so.

	%files -n libzfs5-devel
	%{_pkgconfigdir}/libzfs.pc
	%{_pkgconfigdir}/libzfsbootenv.pc
	%{_pkgconfigdir}/libzfs_core.pc
	%{_libdir}/*.so
	%{_includedir}/*
	%doc AUTHORS COPYRIGHT LICENSE NOTICE README.md

	%files test
	%{_datadir}/%{name}/zfs-tests
	%{_datadir}/%{name}/test-runner
	%{_datadir}/%{name}/runfiles
	%{_datadir}/%{name}/*.sh

	%files dracut
	%doc contrib/dracut/README.dracut.markdown
	%{_dracutdir}/modules.d/*

	%if %{with pyzfs}
	%files -n python%{__python_pkg_version}-pyzfs
	%doc contrib/pyzfs/README
	%doc contrib/pyzfs/LICENSE
	%defattr(-,root,root,-)
	%{__python_sitelib}/libzfs_core/*
	%{__python_sitelib}/pyzfs*
	%endif

	%if 0%{?_initramfs}
	%files initramfs
	%doc contrib/initramfs/README.initramfs.markdown
	/usr/share/initramfs-tools/*
	%else
	# Since we're not building the initramfs package,
	# ignore those files.
	%exclude /usr/share/initramfs-tools
	%endif
	diff --git a/sys/contrib/openzfs/scripts/Makefile.am b/sys/contrib/openzfs/scripts/Makefile.am
	index 6c59fd7d4faf..77b1269a9de4 100644
	--- a/sys/contrib/openzfs/scripts/Makefile.am
	+++ b/sys/contrib/openzfs/scripts/Makefile.am
	@@ -1,90 +1,89 @@
	include $(top_srcdir)/config/Shellcheck.am

	pkgdatadir = $(datadir)/@PACKAGE@

	dist_pkgdata_SCRIPTS = \
	zimport.sh \
	zfs.sh \
	zfs-tests.sh \
	zloop.sh \
	zfs-helpers.sh

	EXTRA_SCRIPTS = \
	commitcheck.sh \
	common.sh.in \
	dkms.mkconf \
	dkms.postbuild \
	kmodtool \
	make_gitrev.sh \
	man-dates.sh \
	paxcheck.sh \
	mancheck.sh

	EXTRA_DIST = \
	cstyle.pl \
	enum-extract.pl \
	zfs2zol-patch.sed \
	zol2zfs-patch.sed \
	$(EXTRA_SCRIPTS)

	-SHELLCHECK_IGNORE = ,SC1117
	SHELLCHECKSCRIPTS = $(EXTRA_SCRIPTS)

	define EXTRA_ENVIRONMENT

	# Only required for in-tree use
	export INTREE="yes"
	export GDB="libtool --mode=execute gdb"
	export LDMOD=/sbin/insmod

	export CMD_DIR=@abs_top_builddir@/cmd
	export UDEV_RULE_DIR=@abs_top_builddir@/udev/rules.d
	export ZEDLET_ETC_DIR=$$CMD_DIR/zed/zed.d
	export ZEDLET_LIBEXEC_DIR=$$CMD_DIR/zed/zed.d
	export ZPOOL_SCRIPT_DIR=$$CMD_DIR/zpool/zpool.d
	export ZPOOL_SCRIPTS_PATH=$$CMD_DIR/zpool/zpool.d
	export ZPOOL_COMPAT_DIR=$$CMD_DIR/zpool/compatibility.d
	export CONTRIB_DIR=@abs_top_builddir@/contrib
	export LIB_DIR=@abs_top_builddir@/lib
	export SYSCONF_DIR=@abs_top_builddir@/etc

	export INSTALL_UDEV_DIR=@udevdir@
	export INSTALL_UDEV_RULE_DIR=@udevruledir@
	export INSTALL_MOUNT_HELPER_DIR=@mounthelperdir@
	export INSTALL_SYSCONF_DIR=@sysconfdir@
	export INSTALL_PYTHON_DIR=@pythonsitedir@

	export KMOD_SPL=@abs_top_builddir@/module/spl/spl.ko
	export KMOD_ZAVL=@abs_top_builddir@/module/avl/zavl.ko
	export KMOD_ZNVPAIR=@abs_top_builddir@/module/nvpair/znvpair.ko
	export KMOD_ZUNICODE=@abs_top_builddir@/module/unicode/zunicode.ko
	export KMOD_ZCOMMON=@abs_top_builddir@/module/zcommon/zcommon.ko
	export KMOD_ZLUA=@abs_top_builddir@/module/lua/zlua.ko
	export KMOD_ICP=@abs_top_builddir@/module/icp/icp.ko
	export KMOD_ZFS=@abs_top_builddir@/module/zfs/zfs.ko
	export KMOD_FREEBSD=@abs_top_builddir@/module/openzfs.ko
	export KMOD_ZZSTD=@abs_top_builddir@/module/zstd/zzstd.ko
	endef

	export EXTRA_ENVIRONMENT

	all-local:
	-$(SED) -e '\\|^export BIN_DIR=\|s\|$$\|@abs_top_builddir@/bin\|' \
	-e '\\|^export SBIN_DIR=\|s\|$$\|@abs_top_builddir@/bin\|' \
	-e '\\|^export LIBEXEC_DIR=\|s\|$$\|@abs_top_builddir@/bin\|' \
	-e '\\|^export ZTS_DIR=\|s\|$$\|@abs_top_srcdir@/tests\|' \
	-e '\\|^export SCRIPT_DIR=\|s\|$$\|@abs_top_srcdir@/scripts\|' \
	$(abs_top_srcdir)/scripts/common.sh.in >common.sh
	-echo "$$EXTRA_ENVIRONMENT" >>common.sh

	clean-local:
	-$(RM) common.sh

	install-data-hook:
	-$(SED) -e '\\|^export BIN_DIR=\|s\|$$\|@bindir@\|' \
	-e '\\|^export SBIN_DIR=\|s\|$$\|@sbindir@\|' \
	-e '\\|^export LIBEXEC_DIR=\|s\|$$\|@zfsexecdir@\|' \
	-e '\\|^export ZTS_DIR=\|s\|$$\|@datadir@/@PACKAGE@\|' \
	-e '\\|^export SCRIPT_DIR=\|s\|$$\|@datadir@/@PACKAGE@\|' \
	$(abs_top_srcdir)/scripts/common.sh.in \
	>$(DESTDIR)$(datadir)/@PACKAGE@/common.sh
	diff --git a/sys/contrib/openzfs/scripts/dkms.mkconf b/sys/contrib/openzfs/scripts/dkms.mkconf
	index 9d12a8c3b3fd..ebe38e31916a 100755
	--- a/sys/contrib/openzfs/scripts/dkms.mkconf
	+++ b/sys/contrib/openzfs/scripts/dkms.mkconf
	@@ -1,123 +1,122 @@
	#!/bin/sh

	PROG=$0

	pkgcfg=/etc/sysconfig/zfs

	while getopts "n:v:c:f:" opt; do
	case $opt in
	n) pkgname=$OPTARG ;;
	v) pkgver=$OPTARG ;;
	c) pkgcfg=$OPTARG ;;
	f) filename=$OPTARG ;;
	*) err=1 ;;
	esac
	done

	if [ -z "${pkgname}" ] \|\| [ -z "${pkgver}" ] \|\| [ -z "${filename}" ] \|\|
	[ -n "${err}" ]; then
	echo "Usage: $PROG -n <pkgname> -v <pkgver> -c <pkgcfg> -f <filename>"
	exit 1
	fi

	exec cat >"${filename}" <<EOF
	PACKAGE_NAME="${pkgname}"
	PACKAGE_VERSION="${pkgver}"
	PACKAGE_CONFIG="${pkgcfg}"
	NO_WEAK_MODULES="yes"
	PRE_BUILD="configure
	--prefix=/usr
	--with-config=kernel
	--with-linux=\$(
	case \`lsb_release -is\` in
	(Debian\|Devuan)
	if [[ -e \${kernel_source_dir/%build/source} ]]
	then
	echo \${kernel_source_dir/%build/source}
	else
	# A kpkg exception for Proxmox 2.0
	echo \${kernel_source_dir}
	fi
	;;
	(*)
	echo \${kernel_source_dir}
	;;
	esac
	)
	--with-linux-obj=\${kernel_source_dir}
	\$(
	[[ -n \"\${ICP_ROOT}\" ]] && \\
	{
	echo --with-qat=\"\${ICP_ROOT}\"
	}
	)
	\$(
	[[ -r \${PACKAGE_CONFIG} ]] \\
	&& source \${PACKAGE_CONFIG} \\
	&& shopt -q -s extglob \\
	&& \\
	{
	if [[ \${ZFS_DKMS_ENABLE_DEBUG,,} == @(y\|yes) ]]
	then
	echo --enable-debug
	fi
	if [[ \${ZFS_DKMS_ENABLE_DEBUGINFO,,} == @(y\|yes) ]]
	then
	echo --enable-debuginfo
	fi
	}
	)
	"
	POST_BUILD="scripts/dkms.postbuild
	-n \${PACKAGE_NAME}
	-v \${PACKAGE_VERSION}
	-a \${arch}
	-k \${kernelver}
	-t \${dkms_tree}
	"
	AUTOINSTALL="yes"
	-REMAKE_INITRD="no"
	MAKE[0]="make"
	STRIP[0]="\$(
	[[ -r \${PACKAGE_CONFIG} ]] \\
	&& source \${PACKAGE_CONFIG} \\
	&& shopt -q -s extglob \\
	&& [[ \${ZFS_DKMS_DISABLE_STRIP,,} == @(y\|yes) ]] \\
	&& echo -n no
	)"
	STRIP[1]="\${STRIP[0]}"
	STRIP[2]="\${STRIP[0]}"
	STRIP[3]="\${STRIP[0]}"
	STRIP[4]="\${STRIP[0]}"
	STRIP[5]="\${STRIP[0]}"
	STRIP[6]="\${STRIP[0]}"
	STRIP[7]="\${STRIP[0]}"
	STRIP[8]="\${STRIP[0]}"
	BUILT_MODULE_NAME[0]="zavl"
	BUILT_MODULE_LOCATION[0]="module/avl/"
	DEST_MODULE_LOCATION[0]="/extra/avl/avl"
	BUILT_MODULE_NAME[1]="znvpair"
	BUILT_MODULE_LOCATION[1]="module/nvpair/"
	DEST_MODULE_LOCATION[1]="/extra/nvpair/znvpair"
	BUILT_MODULE_NAME[2]="zunicode"
	BUILT_MODULE_LOCATION[2]="module/unicode/"
	DEST_MODULE_LOCATION[2]="/extra/unicode/zunicode"
	BUILT_MODULE_NAME[3]="zcommon"
	BUILT_MODULE_LOCATION[3]="module/zcommon/"
	DEST_MODULE_LOCATION[3]="/extra/zcommon/zcommon"
	BUILT_MODULE_NAME[4]="zfs"
	BUILT_MODULE_LOCATION[4]="module/zfs/"
	DEST_MODULE_LOCATION[4]="/extra/zfs/zfs"
	BUILT_MODULE_NAME[5]="icp"
	BUILT_MODULE_LOCATION[5]="module/icp/"
	DEST_MODULE_LOCATION[5]="/extra/icp/icp"
	BUILT_MODULE_NAME[6]="zlua"
	BUILT_MODULE_LOCATION[6]="module/lua/"
	DEST_MODULE_LOCATION[6]="/extra/lua/zlua"
	BUILT_MODULE_NAME[7]="spl"
	BUILT_MODULE_LOCATION[7]="module/spl/"
	DEST_MODULE_LOCATION[7]="/extra/spl/spl"
	BUILT_MODULE_NAME[8]="zzstd"
	BUILT_MODULE_LOCATION[8]="module/zstd/"
	DEST_MODULE_LOCATION[8]="/extra/zstd/zzstd"
	EOF
	diff --git a/sys/contrib/openzfs/scripts/mancheck.sh b/sys/contrib/openzfs/scripts/mancheck.sh
	index 0793cc48fa97..347b2e086470 100755
	--- a/sys/contrib/openzfs/scripts/mancheck.sh
	+++ b/sys/contrib/openzfs/scripts/mancheck.sh
	@@ -1,53 +1,53 @@
	#!/bin/sh
	#
	# Permission to use, copy, modify, and/or distribute this software for
	# any purpose with or without fee is hereby granted.
	#
	# THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
	# WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
	# MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
	# ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
	# WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN
	# AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT
	# OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
	#
	-# shellcheck disable=SC2086,SC2250
	+# shellcheck disable=SC2086

	trap 'rm -f "$stdout_file" "$stderr_file" "$result_file"' EXIT

	if [ "$#" -eq 0 ]; then
	echo "Usage: $0 manpage-directory..."
	exit 1
	fi

	if ! command -v mandoc > /dev/null; then
	echo "skipping mancheck because mandoc is not installed"
	exit 0
	fi

	IFS="
	"
	files="$(find "$@" -type f -name '[1-9]')" \|\| exit 1

	add_excl="$(awk '
	/^.\\" lint-ok:/ {
	print "-e"
	$1 = "mandoc:"
	$2 = FILENAME ":[[:digit:]]+:[[:digit:]]+:"
	print
	}' $files)"

	# Redirect to file instead of 2>&1ing because mandoc flushes inconsistently(?) which tears lines
	# https://github.com/openzfs/zfs/pull/12129/checks?check_run_id=2701608671#step:5:3
	stdout_file="$(mktemp)"
	stderr_file="$(mktemp)"
	mandoc -Tlint $files 1>"$stdout_file" 2>"$stderr_file"
	result_file="$(mktemp)"
	grep -vhE -e 'mandoc: outdated mandoc.db' -e 'STYLE: referenced manual not found' $add_excl "$stdout_file" "$stderr_file" > "$result_file"

	if [ -s "$result_file" ]; then
	cat "$result_file"
	exit 1
	else
	echo "no errors found"
	fi
	diff --git a/sys/contrib/openzfs/scripts/zfs-tests.sh b/sys/contrib/openzfs/scripts/zfs-tests.sh
	index 797d08706338..f871a51d34c2 100755
	--- a/sys/contrib/openzfs/scripts/zfs-tests.sh
	+++ b/sys/contrib/openzfs/scripts/zfs-tests.sh
	@@ -1,716 +1,748 @@
	#!/bin/sh
	#
	# CDDL HEADER START
	#
	# The contents of this file are subject to the terms of the
	# Common Development and Distribution License, Version 1.0 only
	# (the "License"). You may not use this file except in compliance
	# with the License.
	#
	# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	# or http://www.opensolaris.org/os/licensing.
	# See the License for the specific language governing permissions
	# and limitations under the License.
	#
	# When distributing Covered Code, include this CDDL HEADER in each
	# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	# If applicable, add the following below this CDDL HEADER, with the
	# fields enclosed by brackets "[]" replaced with your own identifying
	# information: Portions Copyright [yyyy] [name of copyright owner]
	#
	# CDDL HEADER END
	#

	+#
	+# Copyright 2020 OmniOS Community Edition (OmniOSce) Association.
	+#
	+
	BASE_DIR=$(dirname "$0")
	SCRIPT_COMMON=common.sh
	if [ -f "${BASE_DIR}/${SCRIPT_COMMON}" ]; then
	. "${BASE_DIR}/${SCRIPT_COMMON}"
	else
	echo "Missing helper script ${SCRIPT_COMMON}" && exit 1
	fi

	PROG=zfs-tests.sh
	VERBOSE="no"
	QUIET=""
	CLEANUP="yes"
	CLEANUPALL="no"
	LOOPBACK="yes"
	STACK_TRACER="no"
	FILESIZE="4G"
	DEFAULT_RUNFILES="common.run,$(uname \| tr '[:upper:]' '[:lower:]').run"
	RUNFILES=${RUNFILES:-$DEFAULT_RUNFILES}
	FILEDIR=${FILEDIR:-/var/tmp}
	DISKS=${DISKS:-""}
	SINGLETEST=""
	SINGLETESTUSER="root"
	TAGS=""
	ITERATIONS=1
	ZFS_DBGMSG="$STF_SUITE/callbacks/zfs_dbgmsg.ksh"
	ZFS_DMESG="$STF_SUITE/callbacks/zfs_dmesg.ksh"
	UNAME=$(uname -s)
	+RERUN=""

	# Override some defaults if on FreeBSD
	if [ "$UNAME" = "FreeBSD" ] ; then
	TESTFAIL_CALLBACKS=${TESTFAIL_CALLBACKS:-"$ZFS_DMESG"}
	LOSETUP=/sbin/mdconfig
	DMSETUP=/sbin/gpart
	else
	ZFS_MMP="$STF_SUITE/callbacks/zfs_mmp.ksh"
	TESTFAIL_CALLBACKS=${TESTFAIL_CALLBACKS:-"$ZFS_DBGMSG:$ZFS_DMESG:$ZFS_MMP"}
	LOSETUP=${LOSETUP:-/sbin/losetup}
	DMSETUP=${DMSETUP:-/sbin/dmsetup}
	fi

	#
	# Log an informational message when additional verbosity is enabled.
	#
	msg() {
	if [ "$VERBOSE" = "yes" ]; then
	echo "$@"
	fi
	}

	#
	# Log a failure message, cleanup, and return an error.
	#
	fail() {
	echo "$PROG: $1" >&2
	cleanup
	exit 1
	}

	cleanup_freebsd_loopback() {
	for TEST_LOOPBACK in ${LOOPBACKS}; do
	if [ -c "/dev/${TEST_LOOPBACK}" ]; then
	sudo "${LOSETUP}" -d -u "${TEST_LOOPBACK}" \|\|
	echo "Failed to destroy: ${TEST_LOOPBACK}"
	fi
	done
	}

	cleanup_linux_loopback() {
	for TEST_LOOPBACK in ${LOOPBACKS}; do
	LOOP_DEV="${TEST_LOOPBACK##*/}"
	DM_DEV=$(sudo "${DMSETUP}" ls 2>/dev/null \| \
	grep "${LOOP_DEV}" \| cut -f1)

	if [ -n "$DM_DEV" ]; then
	sudo "${DMSETUP}" remove "${DM_DEV}" \|\|
	echo "Failed to remove: ${DM_DEV}"
	fi

	if [ -n "${TEST_LOOPBACK}" ]; then
	sudo "${LOSETUP}" -d "${TEST_LOOPBACK}" \|\|
	echo "Failed to remove: ${TEST_LOOPBACK}"
	fi
	done
	}

	#
	# Attempt to remove loopback devices and files which where created earlier
	# by this script to run the test framework. The '-k' option may be passed
	# to the script to suppress cleanup for debugging purposes.
	#
	cleanup() {
	if [ "$CLEANUP" = "no" ]; then
	return 0
	fi


	if [ "$LOOPBACK" = "yes" ]; then
	if [ "$UNAME" = "FreeBSD" ] ; then
	cleanup_freebsd_loopback
	else
	cleanup_linux_loopback
	fi
	fi

	for TEST_FILE in ${FILES}; do
	rm -f "${TEST_FILE}" >/dev/null 2>&1
	done

	if [ "$STF_PATH_REMOVE" = "yes" ] && [ -d "$STF_PATH" ]; then
	rm -Rf "$STF_PATH"
	fi
	}
	trap cleanup EXIT

	#
	# Attempt to remove all testpools (testpool.XXX), unopened dm devices,
	# loopback devices, and files. This is a useful way to cleanup a previous
	# test run failure which has left the system in an unknown state. This can
	# be dangerous and should only be used in a dedicated test environment.
	#
	cleanup_all() {
	TEST_POOLS=$(sudo "$ZPOOL" list -H -o name \| grep testpool)
	if [ "$UNAME" = "FreeBSD" ] ; then
	TEST_LOOPBACKS=$(sudo "${LOSETUP}" -l)
	else
	TEST_LOOPBACKS=$(sudo "${LOSETUP}" -a\|grep file-vdev\|cut -f1 -d:)
	fi
	TEST_FILES=$(ls /var/tmp/file-vdev* 2>/dev/null)

	msg
	msg "--- Cleanup ---"
	msg "Removing pool(s): $(echo "${TEST_POOLS}" \| tr '\n' ' ')"
	for TEST_POOL in $TEST_POOLS; do
	sudo "$ZPOOL" destroy "${TEST_POOL}"
	done

	if [ "$UNAME" != "FreeBSD" ] ; then
	msg "Removing dm(s): $(sudo "${DMSETUP}" ls \|
	grep loop \| tr '\n' ' ')"
	sudo "${DMSETUP}" remove_all
	fi

	msg "Removing loopback(s): $(echo "${TEST_LOOPBACKS}" \| tr '\n' ' ')"
	for TEST_LOOPBACK in $TEST_LOOPBACKS; do
	if [ "$UNAME" = "FreeBSD" ] ; then
	sudo "${LOSETUP}" -d -u "${TEST_LOOPBACK}"
	else
	sudo "${LOSETUP}" -d "${TEST_LOOPBACK}"
	fi
	done

	msg "Removing files(s): $(echo "${TEST_FILES}" \| tr '\n' ' ')"
	for TEST_FILE in $TEST_FILES; do
	sudo rm -f "${TEST_FILE}"
	done
	}

	#
	# Takes a name as the only arguments and looks for the following variations
	# on that name. If one is found it is returned.
	#
	# $RUNFILE_DIR/<name>
	# $RUNFILE_DIR/<name>.run
	# <name>
	# <name>.run
	#
	find_runfile() {
	NAME=$1
	RESULT=""

	if [ -f "$RUNFILE_DIR/$NAME" ]; then
	RESULT="$RUNFILE_DIR/$NAME"
	elif [ -f "$RUNFILE_DIR/$NAME.run" ]; then
	RESULT="$RUNFILE_DIR/$NAME.run"
	elif [ -f "$NAME" ]; then
	RESULT="$NAME"
	elif [ -f "$NAME.run" ]; then
	RESULT="$NAME.run"
	fi

	echo "$RESULT"
	}

	#
	# Symlink file if it appears under any of the given paths.
	#
	create_links() {
	dir_list="$1"
	file_list="$2"

	[ -n "$STF_PATH" ] \|\| fail "STF_PATH wasn't correctly set"

	for i in $file_list; do
	for j in $dir_list; do
	[ ! -e "$STF_PATH/$i" ] \|\| continue

	if [ ! -d "$j/$i" ] && [ -e "$j/$i" ]; then
	ln -sf "$j/$i" "$STF_PATH/$i" \|\| \
	fail "Couldn't link $i"
	break
	fi
	done

	[ ! -e "$STF_PATH/$i" ] && \
	STF_MISSING_BIN="$STF_MISSING_BIN $i"
	done
	STF_MISSING_BIN=${STF_MISSING_BIN# }
	}

	#
	# Constrain the path to limit the available binaries to a known set.
	# When running in-tree a top level ./bin/ directory is created for
	# convenience, otherwise a temporary directory is used.
	#
	constrain_path() {
	. "$STF_SUITE/include/commands.cfg"

	# On FreeBSD, base system zfs utils are in /sbin and OpenZFS utils
	# install to /usr/local/sbin. To avoid testing the wrong utils we
	# need /usr/local to come before / in the path search order.
	SYSTEM_DIRS="/usr/local/bin /usr/local/sbin"
	SYSTEM_DIRS="$SYSTEM_DIRS /usr/bin /usr/sbin /bin /sbin $LIBEXEC_DIR"

	if [ "$INTREE" = "yes" ]; then
	# Constrained path set to ./zfs/bin/
	STF_PATH="$BIN_DIR"
	STF_PATH_REMOVE="no"
	STF_MISSING_BIN=""
	if [ ! -d "$STF_PATH" ]; then
	mkdir "$STF_PATH"
	chmod 755 "$STF_PATH" \|\| fail "Couldn't chmod $STF_PATH"
	fi

	# Special case links for standard zfs utilities
	DIRS="$(find "$CMD_DIR" -type d \( ! -name .deps -a \
	! -name .libs \) -print \| tr '\n' ' ')"
	create_links "$DIRS" "$ZFS_FILES"

	# Special case links for zfs test suite utilities
	DIRS="$(find "$STF_SUITE" -type d \( ! -name .deps -a \
	! -name .libs \) -print \| tr '\n' ' ')"
	create_links "$DIRS" "$ZFSTEST_FILES"
	else
	# Constrained path set to /var/tmp/constrained_path.*
	SYSTEMDIR=${SYSTEMDIR:-/var/tmp/constrained_path.XXXXXX}
	STF_PATH=$(mktemp -d "$SYSTEMDIR")
	STF_PATH_REMOVE="yes"
	STF_MISSING_BIN=""

	chmod 755 "$STF_PATH" \|\| fail "Couldn't chmod $STF_PATH"

	# Special case links for standard zfs utilities
	create_links "$SYSTEM_DIRS" "$ZFS_FILES"

	# Special case links for zfs test suite utilities
	create_links "$STF_SUITE/bin" "$ZFSTEST_FILES"
	fi

	# Standard system utilities
	SYSTEM_FILES="$SYSTEM_FILES_COMMON"
	if [ "$UNAME" = "FreeBSD" ] ; then
	SYSTEM_FILES="$SYSTEM_FILES $SYSTEM_FILES_FREEBSD"
	else
	SYSTEM_FILES="$SYSTEM_FILES $SYSTEM_FILES_LINUX"
	fi
	create_links "$SYSTEM_DIRS" "$SYSTEM_FILES"

	# Exceptions
	ln -fs "$STF_PATH/awk" "$STF_PATH/nawk"
	if [ "$UNAME" = "Linux" ] ; then
	ln -fs /sbin/fsck.ext4 "$STF_PATH/fsck"
	ln -fs /sbin/mkfs.ext4 "$STF_PATH/newfs"
	ln -fs "$STF_PATH/gzip" "$STF_PATH/compress"
	ln -fs "$STF_PATH/gunzip" "$STF_PATH/uncompress"
	ln -fs "$STF_PATH/exportfs" "$STF_PATH/share"
	ln -fs "$STF_PATH/exportfs" "$STF_PATH/unshare"
	elif [ "$UNAME" = "FreeBSD" ] ; then
	ln -fs /usr/local/bin/ksh93 "$STF_PATH/ksh"
	fi
	}

	#
	# Output a useful usage message.
	#
	usage() {
	cat << EOF
	USAGE:
	$0 [-hvqxkfS] [-s SIZE] [-r RUNFILES] [-t PATH] [-u USER]

	DESCRIPTION:
	ZFS Test Suite launch script

	OPTIONS:
	-h Show this message
	-v Verbose zfs-tests.sh output
	-q Quiet test-runner output
	-x Remove all testpools, dm, lo, and files (unsafe)
	-k Disable cleanup after test failure
	-f Use files only, disables block device tests
	-S Enable stack tracer (negative performance impact)
	-c Only create and populate constrained path
	+ -R Automatically rerun failing tests
	-n NFSFILE Use the nfsfile to determine the NFS configuration
	-I NUM Number of iterations
	-d DIR Use DIR for files and loopback devices
	-s SIZE Use vdevs of SIZE (default: 4G)
	-r RUNFILES Run tests in RUNFILES (default: ${DEFAULT_RUNFILES})
	-t PATH Run single test at PATH relative to test suite
	-T TAGS Comma separated list of tags (default: 'functional')
	-u USER Run single test as USER (default: root)

	EXAMPLES:
	# Run the default (linux) suite of tests and output the configuration used.
	$0 -v

	# Run a smaller suite of tests designed to run more quickly.
	$0 -r linux-fast

	# Run a single test
	$0 -t tests/functional/cli_root/zfs_bookmark/zfs_bookmark_cliargs.ksh

	# Cleanup a previous run of the test suite prior to testing, run the
	# default (linux) suite of tests and perform no cleanup on exit.
	$0 -x

	EOF
	}

	-while getopts 'hvqxkfScn:d:s:r:?t:T:u:I:' OPTION; do
	+while getopts 'hvqxkfScRn:d:s:r:?t:T:u:I:' OPTION; do
	case $OPTION in
	h)
	usage
	exit 1
	;;
	v)
	VERBOSE="yes"
	;;
	q)
	QUIET="yes"
	;;
	x)
	CLEANUPALL="yes"
	;;
	k)
	CLEANUP="no"
	;;
	f)
	LOOPBACK="no"
	;;
	S)
	STACK_TRACER="yes"
	;;
	c)
	constrain_path
	exit
	;;
	+ R)
	+ RERUN="yes"
	+ ;;
	n)
	nfsfile=$OPTARG
	[ -f "$nfsfile" ] \|\| fail "Cannot read file: $nfsfile"
	export NFS=1
	. "$nfsfile"
	;;
	d)
	FILEDIR="$OPTARG"
	;;
	I)
	ITERATIONS="$OPTARG"
	if [ "$ITERATIONS" -le 0 ]; then
	fail "Iterations must be greater than 0."
	fi
	;;
	s)
	FILESIZE="$OPTARG"
	;;
	r)
	RUNFILES="$OPTARG"
	;;
	t)
	if [ -n "$SINGLETEST" ]; then
	fail "-t can only be provided once."
	fi
	SINGLETEST="$OPTARG"
	;;
	T)
	TAGS="$OPTARG"
	;;
	u)
	SINGLETESTUSER="$OPTARG"
	;;
	?)
	usage
	exit
	;;
	esac
	done

	shift $((OPTIND-1))

	FILES=${FILES:-"$FILEDIR/file-vdev0 $FILEDIR/file-vdev1 $FILEDIR/file-vdev2"}
	LOOPBACKS=${LOOPBACKS:-""}

	if [ -n "$SINGLETEST" ]; then
	if [ -n "$TAGS" ]; then
	fail "-t and -T are mutually exclusive."
	fi
	RUNFILE_DIR="/var/tmp"
	RUNFILES="zfs-tests.$$.run"
	SINGLEQUIET="False"

	if [ -n "$QUIET" ]; then
	SINGLEQUIET="True"
	fi

	cat >$RUNFILE_DIR/$RUNFILES << EOF
	[DEFAULT]
	pre =
	quiet = $SINGLEQUIET
	pre_user = root
	user = $SINGLETESTUSER
	timeout = 600
	post_user = root
	post =
	outputdir = /var/tmp/test_results
	EOF
	SINGLETESTDIR=$(dirname "$SINGLETEST")
	SINGLETESTFILE=$(basename "$SINGLETEST")
	SETUPSCRIPT=
	CLEANUPSCRIPT=

	if [ -f "$STF_SUITE/$SINGLETESTDIR/setup.ksh" ]; then
	SETUPSCRIPT="setup"
	fi

	if [ -f "$STF_SUITE/$SINGLETESTDIR/cleanup.ksh" ]; then
	CLEANUPSCRIPT="cleanup"
	fi

	cat >>$RUNFILE_DIR/$RUNFILES << EOF

	[$SINGLETESTDIR]
	tests = ['$SINGLETESTFILE']
	pre = $SETUPSCRIPT
	post = $CLEANUPSCRIPT
	tags = ['functional']
	EOF
	fi

	#
	# Use default tag if none was specified
	#
	TAGS=${TAGS:='functional'}

	#
	# Attempt to locate the runfiles describing the test workload.
	#
	R=""
	IFS=,
	for RUNFILE in $RUNFILES; do
	if [ -n "$RUNFILE" ]; then
	SAVED_RUNFILE="$RUNFILE"
	RUNFILE=$(find_runfile "$RUNFILE")
	[ -z "$RUNFILE" ] && fail "Cannot find runfile: $SAVED_RUNFILE"
	R="$R,$RUNFILE"
	fi

	if [ ! -r "$RUNFILE" ]; then
	fail "Cannot read runfile: $RUNFILE"
	fi
	done
	unset IFS
	RUNFILES=${R#,}

	#
	# This script should not be run as root. Instead the test user, which may
	# be a normal user account, needs to be configured such that it can
	# run commands via sudo passwordlessly.
	#
	if [ "$(id -u)" = "0" ]; then
	fail "This script must not be run as root."
	fi

	if [ "$(sudo whoami)" != "root" ]; then
	fail "Passwordless sudo access required."
	fi

	#
	# Constrain the available binaries to a known set.
	#
	constrain_path

	#
	# Check if ksh exists
	#
	if [ "$UNAME" = "FreeBSD" ]; then
	sudo ln -fs /usr/local/bin/ksh93 /bin/ksh
	fi
	[ -e "$STF_PATH/ksh" ] \|\| fail "This test suite requires ksh."
	[ -e "$STF_SUITE/include/default.cfg" ] \|\| fail \
	"Missing $STF_SUITE/include/default.cfg file."

	#
	# Verify the ZFS module stack is loaded.
	#
	if [ "$STACK_TRACER" = "yes" ]; then
	sudo "${ZFS_SH}" -S >/dev/null 2>&1
	else
	sudo "${ZFS_SH}" >/dev/null 2>&1
	fi

	#
	# Attempt to cleanup all previous state for a new test run.
	#
	if [ "$CLEANUPALL" = "yes" ]; then
	cleanup_all
	fi

	#
	# By default preserve any existing pools
	# NOTE: Since 'zpool list' outputs a newline-delimited list convert $KEEP from
	# space-delimited to newline-delimited.
	#
	if [ -z "${KEEP}" ]; then
	KEEP="$(sudo "$ZPOOL" list -H -o name)"
	if [ -z "${KEEP}" ]; then
	KEEP="rpool"
	fi
	else
	KEEP="$(echo "$KEEP" \| tr '[:blank:]' '\n')"
	fi

	#
	# NOTE: The following environment variables are undocumented
	# and should be used for testing purposes only:
	#
	# __ZFS_POOL_EXCLUDE - don't iterate over the pools it lists
	# __ZFS_POOL_RESTRICT - iterate only over the pools it lists
	#
	# See libzfs/libzfs_config.c for more information.
	#
	if [ "$UNAME" = "FreeBSD" ] ; then
	__ZFS_POOL_EXCLUDE="$(echo "$KEEP" \| tr -s '\n' ' ')"
	else
	__ZFS_POOL_EXCLUDE="$(echo "$KEEP" \| sed ':a;N;s/\n/ /g;ba')"
	fi

	. "$STF_SUITE/include/default.cfg"

	#
	# No DISKS have been provided so a basic file or loopback based devices
	# must be created for the test suite to use.
	#
	if [ -z "${DISKS}" ]; then
	#
	# If this is a performance run, prevent accidental use of
	# loopback devices.
	#
	[ "$TAGS" = "perf" ] && fail "Running perf tests without disks."

	#
	# Create sparse files for the test suite. These may be used
	# directory or have loopback devices layered on them.
	#
	for TEST_FILE in ${FILES}; do
	[ -f "$TEST_FILE" ] && fail "Failed file exists: ${TEST_FILE}"
	truncate -s "${FILESIZE}" "${TEST_FILE}" \|\|
	fail "Failed creating: ${TEST_FILE} ($?)"
	done

	#
	# If requested setup loopback devices backed by the sparse files.
	#
	if [ "$LOOPBACK" = "yes" ]; then
	test -x "$LOSETUP" \|\| fail "$LOSETUP utility must be installed"

	for TEST_FILE in ${FILES}; do
	if [ "$UNAME" = "FreeBSD" ] ; then
	MDDEVICE=$(sudo "${LOSETUP}" -a -t vnode -f "${TEST_FILE}")
	if [ -z "$MDDEVICE" ] ; then
	fail "Failed: ${TEST_FILE} -> loopback"
	fi
	DISKS="$DISKS $MDDEVICE"
	LOOPBACKS="$LOOPBACKS $MDDEVICE"
	else
	TEST_LOOPBACK=$(sudo "${LOSETUP}" -f)
	sudo "${LOSETUP}" "${TEST_LOOPBACK}" "${TEST_FILE}" \|\|
	fail "Failed: ${TEST_FILE} -> ${TEST_LOOPBACK}"
	BASELOOPBACK="${TEST_LOOPBACK##*/}"
	DISKS="$DISKS $BASELOOPBACK"
	LOOPBACKS="$LOOPBACKS $TEST_LOOPBACK"
	fi
	done
	DISKS=${DISKS# }
	LOOPBACKS=${LOOPBACKS# }
	else
	DISKS="$FILES"
	fi
	fi

	#
	# It may be desirable to test with fewer disks than the default when running
	# the performance tests, but the functional tests require at least three.
	#
	NUM_DISKS=$(echo "${DISKS}" \| awk '{print NF}')
	if [ "$TAGS" != "perf" ]; then
	[ "$NUM_DISKS" -lt 3 ] && fail "Not enough disks ($NUM_DISKS/3 minimum)"
	fi

	#
	# Disable SELinux until the ZFS Test Suite has been updated accordingly.
	#
	if [ -x "$STF_PATH/setenforce" ]; then
	sudo setenforce permissive >/dev/null 2>&1
	fi

	#
	# Enable internal ZFS debug log and clear it.
	#
	if [ -e /sys/module/zfs/parameters/zfs_dbgmsg_enable ]; then
	sudo /bin/sh -c "echo 1 >/sys/module/zfs/parameters/zfs_dbgmsg_enable"
	sudo /bin/sh -c "echo 0 >/proc/spl/kstat/zfs/dbgmsg"
	fi

	msg
	msg "--- Configuration ---"
	msg "Runfiles: $RUNFILES"
	msg "STF_TOOLS: $STF_TOOLS"
	msg "STF_SUITE: $STF_SUITE"
	msg "STF_PATH: $STF_PATH"
	msg "FILEDIR: $FILEDIR"
	msg "FILES: $FILES"
	msg "LOOPBACKS: $LOOPBACKS"
	msg "DISKS: $DISKS"
	msg "NUM_DISKS: $NUM_DISKS"
	msg "FILESIZE: $FILESIZE"
	msg "ITERATIONS: $ITERATIONS"
	msg "TAGS: $TAGS"
	msg "STACK_TRACER: $STACK_TRACER"
	msg "Keep pool(s): $KEEP"
	msg "Missing util(s): $STF_MISSING_BIN"
	msg ""

	export STF_TOOLS
	export STF_SUITE
	export STF_PATH
	export DISKS
	export FILEDIR
	export KEEP
	export __ZFS_POOL_EXCLUDE
	export TESTFAIL_CALLBACKS
	export PATH=$STF_PATH

	if [ "$UNAME" = "FreeBSD" ] ; then
	mkdir -p "$FILEDIR" \|\| true
	RESULTS_FILE=$(mktemp -u "${FILEDIR}/zts-results.XXXXXX")
	REPORT_FILE=$(mktemp -u "${FILEDIR}/zts-report.XXXXXX")
	else
	RESULTS_FILE=$(mktemp -u -t zts-results.XXXXXX -p "$FILEDIR")
	REPORT_FILE=$(mktemp -u -t zts-report.XXXXXX -p "$FILEDIR")
	fi

	#
	# Run all the tests as specified.
	#
	msg "${TEST_RUNNER} ${QUIET:+-q}" \
	"-c \"${RUNFILES}\"" \
	"-T \"${TAGS}\"" \
	"-i \"${STF_SUITE}\"" \
	"-I \"${ITERATIONS}\""
	${TEST_RUNNER} ${QUIET:+-q} \
	-c "${RUNFILES}" \
	-T "${TAGS}" \
	-i "${STF_SUITE}" \
	-I "${ITERATIONS}" \
	2>&1 \| tee "$RESULTS_FILE"
	-
	#
	# Analyze the results.
	#
	-${ZTS_REPORT} "$RESULTS_FILE" >"$REPORT_FILE"
	+${ZTS_REPORT} ${RERUN:+--no-maybes} "$RESULTS_FILE" >"$REPORT_FILE"
	RESULT=$?
	+
	+if [ "$RESULT" -eq "2" ] && [ -n "$RERUN" ]; then
	+ MAYBES="$($ZTS_REPORT --list-maybes)"
	+ TEMP_RESULTS_FILE=$(mktemp -u -t zts-results-tmp.XXXXX -p "$FILEDIR")
	+ TEST_LIST=$(mktemp -u -t test-list.XXXXX -p "$FILEDIR")
	+ grep "^Test:.*\[FAIL\]" "$RESULTS_FILE" >"$TEMP_RESULTS_FILE"
	+ for test_name in $MAYBES; do
	+ grep "$test_name " "$TEMP_RESULTS_FILE" >>"$TEST_LIST"
	+ done
	+ ${TEST_RUNNER} ${QUIET:+-q} \
	+ -c "${RUNFILES}" \
	+ -T "${TAGS}" \
	+ -i "${STF_SUITE}" \
	+ -I "${ITERATIONS}" \
	+ -l "${TEST_LIST}" \
	+ 2>&1 \| tee "$RESULTS_FILE"
	+ #
	+ # Analyze the results.
	+ #
	+ ${ZTS_REPORT} --no-maybes "$RESULTS_FILE" >"$REPORT_FILE"
	+ RESULT=$?
	+fi
	+
	+
	cat "$REPORT_FILE"

	RESULTS_DIR=$(awk '/^Log directory/ { print $3 }' "$RESULTS_FILE")
	if [ -d "$RESULTS_DIR" ]; then
	cat "$RESULTS_FILE" "$REPORT_FILE" >"$RESULTS_DIR/results"
	fi

	rm -f "$RESULTS_FILE" "$REPORT_FILE"

	if [ -n "$SINGLETEST" ]; then
	rm -f "$RUNFILES" >/dev/null 2>&1
	fi

	exit ${RESULT}
	diff --git a/sys/contrib/openzfs/tests/README.md b/sys/contrib/openzfs/tests/README.md
	index 72b994fa9fa9..a01ffbe12cf7 100644
	--- a/sys/contrib/openzfs/tests/README.md
	+++ b/sys/contrib/openzfs/tests/README.md
	@@ -1,152 +1,257 @@
	# ZFS Test Suite README

	-1) Building and installing the ZFS Test Suite
	+### 1) Building and installing the ZFS Test Suite

	The ZFS Test Suite runs under the test-runner framework. This framework
	is built along side the standard ZFS utilities and is included as part of
	zfs-test package. The zfs-test package can be built from source as follows:

	$ ./configure
	$ make pkg-utils

	The resulting packages can be installed using the rpm or dpkg command as
	appropriate for your distributions. Alternately, if you have installed
	ZFS from a distributions repository (not from source) the zfs-test package
	may be provided for your distribution.

	- Installed from source
	$ rpm -ivh ./zfs-test*.rpm, or
	$ dpkg -i ./zfs-test*.deb,

	- Installed from package repository
	$ yum install zfs-test
	$ apt-get install zfs-test

	-2) Running the ZFS Test Suite
	+### 2) Running the ZFS Test Suite

	The pre-requisites for running the ZFS Test Suite are:

	* Three scratch disks
	* Specify the disks you wish to use in the $DISKS variable, as a
	space delimited list like this: DISKS='vdb vdc vdd'. By default
	the zfs-tests.sh script will construct three loopback devices to
	be used for testing: DISKS='loop0 loop1 loop2'.
	* A non-root user with a full set of basic privileges and the ability
	to sudo(8) to root without a password to run the test.
	* Specify any pools you wish to preserve as a space delimited list in
	the $KEEP variable. All pools detected at the start of testing are
	added automatically.
	* The ZFS Test Suite will add users and groups to test machine to
	verify functionality. Therefore it is strongly advised that a
	dedicated test machine, which can be a VM, be used for testing.

	Once the pre-requisites are satisfied simply run the zfs-tests.sh script:

	$ /usr/share/zfs/zfs-tests.sh

	Alternately, the zfs-tests.sh script can be run from the source tree to allow
	developers to rapidly validate their work. In this mode the ZFS utilities and
	modules from the source tree will be used (rather than those installed on the
	system). In order to avoid certain types of failures you will need to ensure
	the ZFS udev rules are installed. This can be done manually or by ensuring
	some version of ZFS is installed on the system.

	$ ./scripts/zfs-tests.sh

	The following zfs-tests.sh options are supported:

	-v Verbose zfs-tests.sh output When specified additional
	information describing the test environment will be logged
	prior to invoking test-runner. This includes the runfile
	being used, the DISKS targeted, pools to keep, etc.

	-q Quiet test-runner output. When specified it is passed to
	test-runner(1) which causes output to be written to the
	console only for tests that do not pass and the results
	summary.

	-x Remove all testpools, dm, lo, and files (unsafe). When
	specified the script will attempt to remove any leftover
	configuration from a previous test run. This includes
	destroying any pools named testpool, unused DM devices,
	and loopback devices backed by file-vdevs. This operation
	can be DANGEROUS because it is possible that the script
	will mistakenly remove a resource not related to the testing.

	-k Disable cleanup after test failure. When specified the
	zfs-tests.sh script will not perform any additional cleanup
	when test-runner exists. This is useful when the results of
	a specific test need to be preserved for further analysis.

	-f Use sparse files directly instead of loopback devices for
	the testing. When running in this mode certain tests will
	be skipped which depend on real block devices.

	-c Only create and populate constrained path

	-I NUM Number of iterations

	-d DIR Create sparse files for vdevs in the DIR directory. By
	default these files are created under /var/tmp/.

	-s SIZE Use vdevs of SIZE (default: 4G)

	-r RUNFILES Run tests in RUNFILES (default: common.run,linux.run)

	-t PATH Run single test at PATH relative to test suite

	-T TAGS Comma separated list of tags (default: 'functional')

	-u USER Run single test as USER (default: root)


	The ZFS Test Suite allows the user to specify a subset of the tests via a
	runfile or list of tags.

	The format of the runfile is explained in test-runner(1), and
	the files that zfs-tests.sh uses are available for reference under
	/usr/share/zfs/runfiles. To specify a custom runfile, use the -r option:

	$ /usr/share/zfs/zfs-tests.sh -r my_tests.run

	Otherwise user can set needed tags to run only specific tests.

	-3) Test results
	+### 3) Test results

	While the ZFS Test Suite is running, one informational line is printed at the
	end of each test, and a results summary is printed at the end of the run. The
	results summary includes the location of the complete logs, which is logged in
	the form `/var/tmp/test_results/[ISO 8601 date]`. A normal test run launched
	with the `zfs-tests.sh` wrapper script will look something like this:

	$ /usr/share/zfs/zfs-tests.sh -v -d /tmp/test

	--- Configuration ---
	Runfile: /usr/share/zfs/runfiles/linux.run
	STF_TOOLS: /usr/share/zfs/test-runner
	STF_SUITE: /usr/share/zfs/zfs-tests
	STF_PATH: /var/tmp/constrained_path.G0Sf
	FILEDIR: /tmp/test
	FILES: /tmp/test/file-vdev0 /tmp/test/file-vdev1 /tmp/test/file-vdev2
	LOOPBACKS: /dev/loop0 /dev/loop1 /dev/loop2
	DISKS: loop0 loop1 loop2
	NUM_DISKS: 3
	FILESIZE: 4G
	ITERATIONS: 1
	TAGS: functional
	Keep pool(s): rpool


	/usr/share/zfs/test-runner/bin/test-runner.py -c /usr/share/zfs/runfiles/linux.run \
	-T functional -i /usr/share/zfs/zfs-tests -I 1
	Test: /usr/share/zfs/zfs-tests/tests/functional/arc/setup (run as root) [00:00] [PASS]
	...more than 1100 additional tests...
	Test: /usr/share/zfs/zfs-tests/tests/functional/zvol/zvol_swap/cleanup (run as root) [00:00] [PASS]

	Results Summary
	SKIP 52
	PASS 1129
	-
	+
	Running Time: 02:35:33
	Percent passed: 95.6%
	Log directory: /var/tmp/test_results/20180515T054509
	+
	+### 4) Example of adding and running test-case (zpool_example)
	+
	+ This broadly boils down to 5 steps
	+ 1. Create/Set password-less sudo for user running test case.
	+ 2. Edit configure.ac, Makefile.am appropriately
	+ 3. Create/Modify .run files
	+ 4. Create actual test-scripts
	+ 5. Run Test case
	+
	+ Will look at each of them in depth.
	+
	+ * Set password-less sudo for 'Test' user as test script cannot be run as root
	+ * Edit file configure.ac and include line under AC_CONFIG_FILES section
	+ ~~~~
	+ tests/zfs-tests/tests/functional/cli_root/zpool_example/Makefile
	+ ~~~~
	+ * Edit file tests/runfiles/Makefile.am and add line zpool_example.
	+ ~~~~
	+ pkgdatadir = $(datadir)/@PACKAGE@/runfiles
	+ dist_pkgdata_DATA = \
	+ zpool_example.run \
	+ common.run \
	+ freebsd.run \
	+ linux.run \
	+ longevity.run \
	+ perf-regression.run \
	+ sanity.run \
	+ sunos.run
	+ ~~~~
	+ * Create file tests/runfiles/zpool_example.run. This defines the most
	+ common properties when run with test-runner.py or zfs-tests.sh.
	+ ~~~~
	+ [DEFAULT]
	+ timeout = 600
	+ outputdir = /var/tmp/test_results
	+ tags = ['functional']
	+
	+ tests = ['zpool_example_001_pos']
	+ ~~~~
	+ If adding test-case to an already existing suite the runfile would
	+ already be present and it needs to be only updated. For example, adding
	+ zpool_example_002_pos to the above runfile only update the "tests ="
	+ section of the runfile as shown below
	+ ~~~~
	+ [DEFAULT]
	+ timeout = 600
	+ outputdir = /var/tmp/test_results
	+ tags = ['functional']
	+
	+ tests = ['zpool_example_001_pos', 'zpool_example_002_pos']
	+ ~~~~
	+
	+ * Edit tests/zfs-tests/tests/functional/cli_root/Makefile.am and add line
	+ under SUBDIRS.
	+ ~~~~
	+ zpool_example \ (Make sure to escape the line end as there will be other folders names following)
	+ ~~~~
	+ * Create new file tests/zfs-tests/tests/functional/cli_root/zpool_example/Makefile.am
	+ the contents of the file could be as below. What it says it that now we have
	+ a test case zpool_example_001_pos.ksh
	+ ~~~~
	+ pkgdatadir = $(datadir)/@PACKAGE@/zfs-tests/tests/functional/cli_root/zpool_example
	+ dist_pkgdata_SCRIPTS = \
	+ zpool_example_001_pos.ksh
	+ ~~~~
	+ * We can now create our test-case zpool_example_001_pos.ksh under
	+ tests/zfs-tests/tests/functional/cli_root/zpool_example/.
	+ ~~~~
	+ # DESCRIPTION:
	+ # zpool_example Test
	+ #
	+ # STRATEGY:
	+ # 1. Demo a very basic test case
	+ #
	+
	+ DISKS_DEV1="/dev/loop0"
	+ DISKS_DEV2="/dev/loop1"
	+ TESTPOOL=EXAMPLE_POOL
	+
	+ function cleanup
	+ {
	+ # Cleanup
	+ destroy_pool $TESTPOOL
	+ log_must rm -f $DISKS_DEV1
	+ log_must rm -f $DISKS_DEV2
	+ }
	+
	+ log_assert "zpool_example"
	+ # Run function "cleanup" on exit
	+ log_onexit cleanup
	+
	+ # Prep backend device
	+ log_must dd if=/dev/zero of=$DISKS_DEV1 bs=512 count=140000
	+ log_must dd if=/dev/zero of=$DISKS_DEV2 bs=512 count=140000
	+
	+ # Create pool
	+ log_must zpool create $TESTPOOL $type $DISKS_DEV1 $DISKS_DEV2
	+
	+ log_pass "zpool_example"
	+ ~~~~
	+ * Run Test case, which can be done in two ways. Described in detail above in
	+ section 2.
	+ * test-runner.py (This takes run file as input. See zpool_example.run)
	+ * zfs-tests.sh. Can execute the run file or individual tests
	diff --git a/sys/contrib/openzfs/tests/runfiles/common.run b/sys/contrib/openzfs/tests/runfiles/common.run
	index 89cd58c2aa32..d5052172d269 100644
	--- a/sys/contrib/openzfs/tests/runfiles/common.run
	+++ b/sys/contrib/openzfs/tests/runfiles/common.run
	@@ -1,959 +1,959 @@
	#
	# This file and its contents are supplied under the terms of the
	# Common Development and Distribution License ("CDDL"), version 1.0.
	# You may only use this file in accordance with the terms of version
	# 1.0 of the CDDL.
	#
	# A full copy of the text of the CDDL should have accompanied this
	# source. A copy of the CDDL is also available via the Internet at
	# http://www.illumos.org/license/CDDL.
	#
	# This run file contains all of the common functional tests. When
	# adding a new test consider also adding it to the sanity.run file
	# if the new test runs to completion in only a few seconds.
	#
	# Approximate run time: 4-5 hours
	#

	[DEFAULT]
	pre = setup
	quiet = False
	pre_user = root
	user = root
	timeout = 600
	post_user = root
	post = cleanup
	failsafe_user = root
	failsafe = callbacks/zfs_failsafe
	outputdir = /var/tmp/test_results
	tags = ['functional']

	[tests/functional/acl/off]
	tests = ['posixmode']
	tags = ['functional', 'acl']

	[tests/functional/alloc_class]
	tests = ['alloc_class_001_pos', 'alloc_class_002_neg', 'alloc_class_003_pos',
	'alloc_class_004_pos', 'alloc_class_005_pos', 'alloc_class_006_pos',
	'alloc_class_007_pos', 'alloc_class_008_pos', 'alloc_class_009_pos',
	'alloc_class_010_pos', 'alloc_class_011_neg', 'alloc_class_012_pos',
	'alloc_class_013_pos']
	tags = ['functional', 'alloc_class']

	[tests/functional/arc]
	tests = ['dbufstats_001_pos', 'dbufstats_002_pos', 'dbufstats_003_pos',
	'arcstats_runtime_tuning']
	tags = ['functional', 'arc']

	[tests/functional/atime]
	tests = ['atime_001_pos', 'atime_002_neg', 'root_atime_off', 'root_atime_on']
	tags = ['functional', 'atime']

	[tests/functional/bootfs]
	tests = ['bootfs_001_pos', 'bootfs_002_neg', 'bootfs_003_pos',
	'bootfs_004_neg', 'bootfs_005_neg', 'bootfs_006_pos', 'bootfs_007_pos',
	'bootfs_008_pos']
	tags = ['functional', 'bootfs']

	[tests/functional/btree]
	tests = ['btree_positive', 'btree_negative']
	tags = ['functional', 'btree']
	pre =
	post =

	[tests/functional/cache]
	tests = ['cache_001_pos', 'cache_002_pos', 'cache_003_pos', 'cache_004_neg',
	'cache_005_neg', 'cache_006_pos', 'cache_007_neg', 'cache_008_neg',
	'cache_009_pos', 'cache_010_pos', 'cache_011_pos', 'cache_012_pos']
	tags = ['functional', 'cache']

	[tests/functional/cachefile]
	tests = ['cachefile_001_pos', 'cachefile_002_pos', 'cachefile_003_pos',
	'cachefile_004_pos']
	tags = ['functional', 'cachefile']

	[tests/functional/casenorm]
	tests = ['case_all_values', 'norm_all_values', 'mixed_create_failure',
	'sensitive_none_lookup', 'sensitive_none_delete',
	'sensitive_formd_lookup', 'sensitive_formd_delete',
	'insensitive_none_lookup', 'insensitive_none_delete',
	'insensitive_formd_lookup', 'insensitive_formd_delete',
	'mixed_none_lookup', 'mixed_none_lookup_ci', 'mixed_none_delete',
	'mixed_formd_lookup', 'mixed_formd_lookup_ci', 'mixed_formd_delete']
	tags = ['functional', 'casenorm']

	[tests/functional/channel_program/lua_core]
	tests = ['tst.args_to_lua', 'tst.divide_by_zero', 'tst.exists',
	'tst.integer_illegal', 'tst.integer_overflow', 'tst.language_functions_neg',
	'tst.language_functions_pos', 'tst.large_prog', 'tst.libraries',
	'tst.memory_limit', 'tst.nested_neg', 'tst.nested_pos', 'tst.nvlist_to_lua',
	'tst.recursive_neg', 'tst.recursive_pos', 'tst.return_large',
	'tst.return_nvlist_neg', 'tst.return_nvlist_pos',
	'tst.return_recursive_table', 'tst.stack_gsub', 'tst.timeout']
	tags = ['functional', 'channel_program', 'lua_core']

	[tests/functional/channel_program/synctask_core]
	tests = ['tst.destroy_fs', 'tst.destroy_snap', 'tst.get_count_and_limit',
	'tst.get_index_props', 'tst.get_mountpoint', 'tst.get_neg',
	'tst.get_number_props', 'tst.get_string_props', 'tst.get_type',
	'tst.get_userquota', 'tst.get_written', 'tst.inherit', 'tst.list_bookmarks',
	'tst.list_children', 'tst.list_clones', 'tst.list_holds',
	'tst.list_snapshots', 'tst.list_system_props',
	'tst.list_user_props', 'tst.parse_args_neg','tst.promote_conflict',
	'tst.promote_multiple', 'tst.promote_simple', 'tst.rollback_mult',
	'tst.rollback_one', 'tst.set_props', 'tst.snapshot_destroy', 'tst.snapshot_neg',
	'tst.snapshot_recursive', 'tst.snapshot_simple',
	'tst.bookmark.create', 'tst.bookmark.copy',
	'tst.terminate_by_signal'
	]
	tags = ['functional', 'channel_program', 'synctask_core']

	[tests/functional/checksum]
	tests = ['run_edonr_test', 'run_sha2_test', 'run_skein_test', 'filetest_001_pos',
	'filetest_002_pos']
	tags = ['functional', 'checksum']

	[tests/functional/clean_mirror]
	tests = [ 'clean_mirror_001_pos', 'clean_mirror_002_pos',
	'clean_mirror_003_pos', 'clean_mirror_004_pos']
	tags = ['functional', 'clean_mirror']

	[tests/functional/cli_root/zdb]
	tests = ['zdb_002_pos', 'zdb_003_pos', 'zdb_004_pos', 'zdb_005_pos',
	'zdb_006_pos', 'zdb_args_neg', 'zdb_args_pos',
	'zdb_block_size_histogram', 'zdb_checksum', 'zdb_decompress',
	'zdb_display_block', 'zdb_label_checksum', 'zdb_object_range_neg',
	'zdb_object_range_pos', 'zdb_objset_id', 'zdb_decompress_zstd',
	'zdb_recover', 'zdb_recover_2']
	pre =
	post =
	tags = ['functional', 'cli_root', 'zdb']

	[tests/functional/cli_root/zfs]
	tests = ['zfs_001_neg', 'zfs_002_pos']
	tags = ['functional', 'cli_root', 'zfs']

	[tests/functional/cli_root/zfs_bookmark]
	tests = ['zfs_bookmark_cliargs']
	tags = ['functional', 'cli_root', 'zfs_bookmark']

	[tests/functional/cli_root/zfs_change-key]
	tests = ['zfs_change-key', 'zfs_change-key_child', 'zfs_change-key_format',
	'zfs_change-key_inherit', 'zfs_change-key_load', 'zfs_change-key_location',
	'zfs_change-key_pbkdf2iters', 'zfs_change-key_clones']
	tags = ['functional', 'cli_root', 'zfs_change-key']

	[tests/functional/cli_root/zfs_clone]
	tests = ['zfs_clone_001_neg', 'zfs_clone_002_pos', 'zfs_clone_003_pos',
	'zfs_clone_004_pos', 'zfs_clone_005_pos', 'zfs_clone_006_pos',
	'zfs_clone_007_pos', 'zfs_clone_008_neg', 'zfs_clone_009_neg',
	'zfs_clone_010_pos', 'zfs_clone_encrypted', 'zfs_clone_deeply_nested']
	tags = ['functional', 'cli_root', 'zfs_clone']

	[tests/functional/cli_root/zfs_copies]
	tests = ['zfs_copies_001_pos', 'zfs_copies_002_pos', 'zfs_copies_003_pos',
	'zfs_copies_004_neg', 'zfs_copies_005_neg', 'zfs_copies_006_pos']
	tags = ['functional', 'cli_root', 'zfs_copies']

	[tests/functional/cli_root/zfs_create]
	tests = ['zfs_create_001_pos', 'zfs_create_002_pos', 'zfs_create_003_pos',
	'zfs_create_004_pos', 'zfs_create_005_pos', 'zfs_create_006_pos',
	'zfs_create_007_pos', 'zfs_create_008_neg', 'zfs_create_009_neg',
	'zfs_create_010_neg', 'zfs_create_011_pos', 'zfs_create_012_pos',
	'zfs_create_013_pos', 'zfs_create_014_pos', 'zfs_create_encrypted',
	'zfs_create_crypt_combos', 'zfs_create_dryrun', 'zfs_create_nomount',
	'zfs_create_verbose']
	tags = ['functional', 'cli_root', 'zfs_create']

	[tests/functional/cli_root/zfs_destroy]
	tests = ['zfs_clone_livelist_condense_and_disable',
	'zfs_clone_livelist_condense_races', 'zfs_clone_livelist_dedup',
	'zfs_destroy_001_pos', 'zfs_destroy_002_pos', 'zfs_destroy_003_pos',
	'zfs_destroy_004_pos', 'zfs_destroy_005_neg', 'zfs_destroy_006_neg',
	'zfs_destroy_007_neg', 'zfs_destroy_008_pos', 'zfs_destroy_009_pos',
	'zfs_destroy_010_pos', 'zfs_destroy_011_pos', 'zfs_destroy_012_pos',
	'zfs_destroy_013_neg', 'zfs_destroy_014_pos', 'zfs_destroy_015_pos',
	'zfs_destroy_016_pos', 'zfs_destroy_clone_livelist',
	'zfs_destroy_dev_removal', 'zfs_destroy_dev_removal_condense']
	tags = ['functional', 'cli_root', 'zfs_destroy']

	[tests/functional/cli_root/zfs_diff]
	tests = ['zfs_diff_changes', 'zfs_diff_cliargs', 'zfs_diff_timestamp',
	- 'zfs_diff_types', 'zfs_diff_encrypted']
	+ 'zfs_diff_types', 'zfs_diff_encrypted', 'zfs_diff_mangle']
	tags = ['functional', 'cli_root', 'zfs_diff']

	[tests/functional/cli_root/zfs_get]
	tests = ['zfs_get_001_pos', 'zfs_get_002_pos', 'zfs_get_003_pos',
	'zfs_get_004_pos', 'zfs_get_005_neg', 'zfs_get_006_neg', 'zfs_get_007_neg',
	'zfs_get_008_pos', 'zfs_get_009_pos', 'zfs_get_010_neg']
	tags = ['functional', 'cli_root', 'zfs_get']

	[tests/functional/cli_root/zfs_ids_to_path]
	tests = ['zfs_ids_to_path_001_pos']
	tags = ['functional', 'cli_root', 'zfs_ids_to_path']

	[tests/functional/cli_root/zfs_inherit]
	tests = ['zfs_inherit_001_neg', 'zfs_inherit_002_neg', 'zfs_inherit_003_pos',
	'zfs_inherit_mountpoint']
	tags = ['functional', 'cli_root', 'zfs_inherit']

	[tests/functional/cli_root/zfs_load-key]
	tests = ['zfs_load-key', 'zfs_load-key_all', 'zfs_load-key_file',
	'zfs_load-key_https', 'zfs_load-key_location', 'zfs_load-key_noop',
	'zfs_load-key_recursive']
	tags = ['functional', 'cli_root', 'zfs_load-key']

	[tests/functional/cli_root/zfs_mount]
	tests = ['zfs_mount_001_pos', 'zfs_mount_002_pos', 'zfs_mount_003_pos',
	'zfs_mount_004_pos', 'zfs_mount_005_pos', 'zfs_mount_007_pos',
	'zfs_mount_009_neg', 'zfs_mount_010_neg', 'zfs_mount_011_neg',
	'zfs_mount_012_pos', 'zfs_mount_all_001_pos', 'zfs_mount_encrypted',
	'zfs_mount_remount', 'zfs_mount_all_fail', 'zfs_mount_all_mountpoints',
	'zfs_mount_test_race']
	tags = ['functional', 'cli_root', 'zfs_mount']

	[tests/functional/cli_root/zfs_program]
	tests = ['zfs_program_json']
	tags = ['functional', 'cli_root', 'zfs_program']

	[tests/functional/cli_root/zfs_promote]
	tests = ['zfs_promote_001_pos', 'zfs_promote_002_pos', 'zfs_promote_003_pos',
	'zfs_promote_004_pos', 'zfs_promote_005_pos', 'zfs_promote_006_neg',
	'zfs_promote_007_neg', 'zfs_promote_008_pos', 'zfs_promote_encryptionroot']
	tags = ['functional', 'cli_root', 'zfs_promote']

	[tests/functional/cli_root/zfs_property]
	tests = ['zfs_written_property_001_pos']
	tags = ['functional', 'cli_root', 'zfs_property']

	[tests/functional/cli_root/zfs_receive]
	tests = ['zfs_receive_001_pos', 'zfs_receive_002_pos', 'zfs_receive_003_pos',
	'zfs_receive_004_neg', 'zfs_receive_005_neg', 'zfs_receive_006_pos',
	'zfs_receive_007_neg', 'zfs_receive_008_pos', 'zfs_receive_009_neg',
	'zfs_receive_010_pos', 'zfs_receive_011_pos', 'zfs_receive_012_pos',
	'zfs_receive_013_pos', 'zfs_receive_014_pos', 'zfs_receive_015_pos',
	'zfs_receive_016_pos', 'receive-o-x_props_override',
	'receive-o-x_props_aliases',
	'zfs_receive_from_encrypted', 'zfs_receive_to_encrypted',
	'zfs_receive_raw', 'zfs_receive_raw_incremental', 'zfs_receive_-e',
	'zfs_receive_raw_-d', 'zfs_receive_from_zstd', 'zfs_receive_new_props']
	tags = ['functional', 'cli_root', 'zfs_receive']

	[tests/functional/cli_root/zfs_rename]
	tests = ['zfs_rename_001_pos', 'zfs_rename_002_pos', 'zfs_rename_003_pos',
	'zfs_rename_004_neg', 'zfs_rename_005_neg', 'zfs_rename_006_pos',
	'zfs_rename_007_pos', 'zfs_rename_008_pos', 'zfs_rename_009_neg',
	'zfs_rename_010_neg', 'zfs_rename_011_pos', 'zfs_rename_012_neg',
	'zfs_rename_013_pos', 'zfs_rename_014_neg', 'zfs_rename_encrypted_child',
	'zfs_rename_to_encrypted', 'zfs_rename_mountpoint', 'zfs_rename_nounmount']
	tags = ['functional', 'cli_root', 'zfs_rename']

	[tests/functional/cli_root/zfs_reservation]
	tests = ['zfs_reservation_001_pos', 'zfs_reservation_002_pos']
	tags = ['functional', 'cli_root', 'zfs_reservation']

	[tests/functional/cli_root/zfs_rollback]
	tests = ['zfs_rollback_001_pos', 'zfs_rollback_002_pos',
	'zfs_rollback_003_neg', 'zfs_rollback_004_neg']
	tags = ['functional', 'cli_root', 'zfs_rollback']

	[tests/functional/cli_root/zfs_send]
	tests = ['zfs_send_001_pos', 'zfs_send_002_pos', 'zfs_send_003_pos',
	'zfs_send_004_neg', 'zfs_send_005_pos', 'zfs_send_006_pos',
	'zfs_send_007_pos', 'zfs_send_encrypted', 'zfs_send_raw',
	'zfs_send_sparse', 'zfs_send-b', 'zfs_send_skip_missing']
	tags = ['functional', 'cli_root', 'zfs_send']

	[tests/functional/cli_root/zfs_set]
	tests = ['cache_001_pos', 'cache_002_neg', 'canmount_001_pos',
	'canmount_002_pos', 'canmount_003_pos', 'canmount_004_pos',
	'checksum_001_pos', 'compression_001_pos', 'mountpoint_001_pos',
	'mountpoint_002_pos', 'reservation_001_neg', 'user_property_002_pos',
	'share_mount_001_neg', 'snapdir_001_pos', 'onoffs_001_pos',
	'user_property_001_pos', 'user_property_003_neg', 'readonly_001_pos',
	'user_property_004_pos', 'version_001_neg', 'zfs_set_001_neg',
	'zfs_set_002_neg', 'zfs_set_003_neg', 'property_alias_001_pos',
	'mountpoint_003_pos', 'ro_props_001_pos', 'zfs_set_keylocation',
	'zfs_set_feature_activation']
	tags = ['functional', 'cli_root', 'zfs_set']

	[tests/functional/cli_root/zfs_share]
	tests = ['zfs_share_001_pos', 'zfs_share_002_pos', 'zfs_share_003_pos',
	'zfs_share_004_pos', 'zfs_share_006_pos', 'zfs_share_008_neg',
	'zfs_share_010_neg', 'zfs_share_011_pos', 'zfs_share_concurrent_shares']
	tags = ['functional', 'cli_root', 'zfs_share']

	[tests/functional/cli_root/zfs_snapshot]
	tests = ['zfs_snapshot_001_neg', 'zfs_snapshot_002_neg',
	'zfs_snapshot_003_neg', 'zfs_snapshot_004_neg', 'zfs_snapshot_005_neg',
	'zfs_snapshot_006_pos', 'zfs_snapshot_007_neg', 'zfs_snapshot_008_neg',
	'zfs_snapshot_009_pos']
	tags = ['functional', 'cli_root', 'zfs_snapshot']

	[tests/functional/cli_root/zfs_unload-key]
	tests = ['zfs_unload-key', 'zfs_unload-key_all', 'zfs_unload-key_recursive']
	tags = ['functional', 'cli_root', 'zfs_unload-key']

	[tests/functional/cli_root/zfs_unmount]
	tests = ['zfs_unmount_001_pos', 'zfs_unmount_002_pos', 'zfs_unmount_003_pos',
	'zfs_unmount_004_pos', 'zfs_unmount_005_pos', 'zfs_unmount_006_pos',
	'zfs_unmount_007_neg', 'zfs_unmount_008_neg', 'zfs_unmount_009_pos',
	'zfs_unmount_all_001_pos', 'zfs_unmount_nested', 'zfs_unmount_unload_keys']
	tags = ['functional', 'cli_root', 'zfs_unmount']

	[tests/functional/cli_root/zfs_unshare]
	tests = ['zfs_unshare_001_pos', 'zfs_unshare_002_pos', 'zfs_unshare_003_pos',
	'zfs_unshare_004_neg', 'zfs_unshare_005_neg', 'zfs_unshare_006_pos',
	'zfs_unshare_007_pos']
	tags = ['functional', 'cli_root', 'zfs_unshare']

	[tests/functional/cli_root/zfs_upgrade]
	tests = ['zfs_upgrade_001_pos', 'zfs_upgrade_002_pos', 'zfs_upgrade_003_pos',
	'zfs_upgrade_004_pos', 'zfs_upgrade_005_pos', 'zfs_upgrade_006_neg',
	'zfs_upgrade_007_neg']
	tags = ['functional', 'cli_root', 'zfs_upgrade']

	[tests/functional/cli_root/zfs_wait]
	tests = ['zfs_wait_deleteq']
	tags = ['functional', 'cli_root', 'zfs_wait']

	[tests/functional/cli_root/zhack]
	tests = ['zhack_label_checksum']
	pre =
	post =
	tags = ['functional', 'cli_root', 'zhack']

	[tests/functional/cli_root/zpool]
	tests = ['zpool_001_neg', 'zpool_002_pos', 'zpool_003_pos', 'zpool_colors']
	tags = ['functional', 'cli_root', 'zpool']

	[tests/functional/cli_root/zpool_add]
	tests = ['zpool_add_001_pos', 'zpool_add_002_pos', 'zpool_add_003_pos',
	'zpool_add_004_pos', 'zpool_add_006_pos', 'zpool_add_007_neg',
	'zpool_add_008_neg', 'zpool_add_009_neg', 'zpool_add_010_pos',
	'add-o_ashift', 'add_prop_ashift', 'zpool_add_dryrun_output']
	tags = ['functional', 'cli_root', 'zpool_add']

	[tests/functional/cli_root/zpool_attach]
	tests = ['zpool_attach_001_neg', 'attach-o_ashift']
	tags = ['functional', 'cli_root', 'zpool_attach']

	[tests/functional/cli_root/zpool_clear]
	tests = ['zpool_clear_001_pos', 'zpool_clear_002_neg', 'zpool_clear_003_neg',
	'zpool_clear_readonly']
	tags = ['functional', 'cli_root', 'zpool_clear']

	[tests/functional/cli_root/zpool_create]
	tests = ['zpool_create_001_pos', 'zpool_create_002_pos',
	'zpool_create_003_pos', 'zpool_create_004_pos', 'zpool_create_005_pos',
	'zpool_create_006_pos', 'zpool_create_007_neg', 'zpool_create_008_pos',
	'zpool_create_009_neg', 'zpool_create_010_neg', 'zpool_create_011_neg',
	'zpool_create_012_neg', 'zpool_create_014_neg', 'zpool_create_015_neg',
	'zpool_create_017_neg', 'zpool_create_018_pos', 'zpool_create_019_pos',
	'zpool_create_020_pos', 'zpool_create_021_pos', 'zpool_create_022_pos',
	'zpool_create_023_neg', 'zpool_create_024_pos',
	'zpool_create_encrypted', 'zpool_create_crypt_combos',
	'zpool_create_draid_001_pos', 'zpool_create_draid_002_pos',
	'zpool_create_draid_003_pos', 'zpool_create_draid_004_pos',
	'zpool_create_features_001_pos', 'zpool_create_features_002_pos',
	'zpool_create_features_003_pos', 'zpool_create_features_004_neg',
	'zpool_create_features_005_pos', 'zpool_create_features_006_pos',
	'zpool_create_features_007_pos', 'zpool_create_features_008_pos',
	'zpool_create_features_009_pos', 'create-o_ashift',
	'zpool_create_tempname', 'zpool_create_dryrun_output']
	tags = ['functional', 'cli_root', 'zpool_create']

	[tests/functional/cli_root/zpool_destroy]
	tests = ['zpool_destroy_001_pos', 'zpool_destroy_002_pos',
	'zpool_destroy_003_neg']
	pre =
	post =
	tags = ['functional', 'cli_root', 'zpool_destroy']

	[tests/functional/cli_root/zpool_detach]
	tests = ['zpool_detach_001_neg']
	tags = ['functional', 'cli_root', 'zpool_detach']

	[tests/functional/cli_root/zpool_events]
	tests = ['zpool_events_clear', 'zpool_events_cliargs', 'zpool_events_follow',
	'zpool_events_poolname', 'zpool_events_errors', 'zpool_events_duplicates',
	'zpool_events_clear_retained']
	tags = ['functional', 'cli_root', 'zpool_events']

	[tests/functional/cli_root/zpool_export]
	tests = ['zpool_export_001_pos', 'zpool_export_002_pos',
	'zpool_export_003_neg', 'zpool_export_004_pos']
	tags = ['functional', 'cli_root', 'zpool_export']

	[tests/functional/cli_root/zpool_get]
	tests = ['zpool_get_001_pos', 'zpool_get_002_pos', 'zpool_get_003_pos',
	'zpool_get_004_neg', 'zpool_get_005_pos']
	tags = ['functional', 'cli_root', 'zpool_get']

	[tests/functional/cli_root/zpool_history]
	tests = ['zpool_history_001_neg', 'zpool_history_002_pos']
	tags = ['functional', 'cli_root', 'zpool_history']

	[tests/functional/cli_root/zpool_import]
	tests = ['zpool_import_001_pos', 'zpool_import_002_pos',
	'zpool_import_003_pos', 'zpool_import_004_pos', 'zpool_import_005_pos',
	'zpool_import_006_pos', 'zpool_import_007_pos', 'zpool_import_008_pos',
	'zpool_import_009_neg', 'zpool_import_010_pos', 'zpool_import_011_neg',
	'zpool_import_012_pos', 'zpool_import_013_neg', 'zpool_import_014_pos',
	'zpool_import_015_pos', 'zpool_import_016_pos', 'zpool_import_017_pos',
	'zpool_import_features_001_pos', 'zpool_import_features_002_neg',
	'zpool_import_features_003_pos', 'zpool_import_missing_001_pos',
	'zpool_import_missing_002_pos', 'zpool_import_missing_003_pos',
	'zpool_import_rename_001_pos', 'zpool_import_all_001_pos',
	'zpool_import_encrypted', 'zpool_import_encrypted_load',
	'zpool_import_errata3', 'zpool_import_errata4',
	'import_cachefile_device_added',
	'import_cachefile_device_removed',
	'import_cachefile_device_replaced',
	'import_cachefile_mirror_attached',
	'import_cachefile_mirror_detached',
	'import_cachefile_paths_changed',
	'import_cachefile_shared_device',
	'import_devices_missing',
	'import_paths_changed',
	'import_rewind_config_changed',
	'import_rewind_device_replaced']
	tags = ['functional', 'cli_root', 'zpool_import']
	timeout = 1200

	[tests/functional/cli_root/zpool_labelclear]
	tests = ['zpool_labelclear_active', 'zpool_labelclear_exported',
	'zpool_labelclear_removed', 'zpool_labelclear_valid']
	pre =
	post =
	tags = ['functional', 'cli_root', 'zpool_labelclear']

	[tests/functional/cli_root/zpool_initialize]
	tests = ['zpool_initialize_attach_detach_add_remove',
	'zpool_initialize_fault_export_import_online',
	'zpool_initialize_import_export',
	'zpool_initialize_offline_export_import_online',
	'zpool_initialize_online_offline',
	'zpool_initialize_split',
	'zpool_initialize_start_and_cancel_neg',
	'zpool_initialize_start_and_cancel_pos',
	'zpool_initialize_suspend_resume',
	'zpool_initialize_unsupported_vdevs',
	'zpool_initialize_verify_checksums',
	'zpool_initialize_verify_initialized']
	pre =
	tags = ['functional', 'cli_root', 'zpool_initialize']

	[tests/functional/cli_root/zpool_offline]
	tests = ['zpool_offline_001_pos', 'zpool_offline_002_neg',
	'zpool_offline_003_pos']
	tags = ['functional', 'cli_root', 'zpool_offline']

	[tests/functional/cli_root/zpool_online]
	tests = ['zpool_online_001_pos', 'zpool_online_002_neg']
	tags = ['functional', 'cli_root', 'zpool_online']

	[tests/functional/cli_root/zpool_remove]
	tests = ['zpool_remove_001_neg', 'zpool_remove_002_pos',
	'zpool_remove_003_pos']
	tags = ['functional', 'cli_root', 'zpool_remove']

	[tests/functional/cli_root/zpool_replace]
	tests = ['zpool_replace_001_neg', 'replace-o_ashift', 'replace_prop_ashift']
	tags = ['functional', 'cli_root', 'zpool_replace']

	[tests/functional/cli_root/zpool_resilver]
	tests = ['zpool_resilver_bad_args', 'zpool_resilver_restart']
	tags = ['functional', 'cli_root', 'zpool_resilver']

	[tests/functional/cli_root/zpool_scrub]
	tests = ['zpool_scrub_001_neg', 'zpool_scrub_002_pos', 'zpool_scrub_003_pos',
	'zpool_scrub_004_pos', 'zpool_scrub_005_pos',
	'zpool_scrub_encrypted_unloaded', 'zpool_scrub_print_repairing',
	'zpool_scrub_offline_device', 'zpool_scrub_multiple_copies']
	tags = ['functional', 'cli_root', 'zpool_scrub']

	[tests/functional/cli_root/zpool_set]
	tests = ['zpool_set_001_pos', 'zpool_set_002_neg', 'zpool_set_003_neg',
	'zpool_set_ashift', 'zpool_set_features']
	tags = ['functional', 'cli_root', 'zpool_set']

	[tests/functional/cli_root/zpool_split]
	tests = ['zpool_split_cliargs', 'zpool_split_devices',
	'zpool_split_encryption', 'zpool_split_props', 'zpool_split_vdevs',
	'zpool_split_resilver', 'zpool_split_indirect',
	'zpool_split_dryrun_output']
	tags = ['functional', 'cli_root', 'zpool_split']

	[tests/functional/cli_root/zpool_status]
	tests = ['zpool_status_001_pos', 'zpool_status_002_pos',
	'zpool_status_features_001_pos']
	tags = ['functional', 'cli_root', 'zpool_status']

	[tests/functional/cli_root/zpool_sync]
	tests = ['zpool_sync_001_pos', 'zpool_sync_002_neg']
	tags = ['functional', 'cli_root', 'zpool_sync']

	[tests/functional/cli_root/zpool_trim]
	tests = ['zpool_trim_attach_detach_add_remove',
	'zpool_trim_fault_export_import_online',
	'zpool_trim_import_export', 'zpool_trim_multiple', 'zpool_trim_neg',
	'zpool_trim_offline_export_import_online', 'zpool_trim_online_offline',
	'zpool_trim_partial', 'zpool_trim_rate', 'zpool_trim_rate_neg',
	'zpool_trim_secure', 'zpool_trim_split', 'zpool_trim_start_and_cancel_neg',
	'zpool_trim_start_and_cancel_pos', 'zpool_trim_suspend_resume',
	'zpool_trim_unsupported_vdevs', 'zpool_trim_verify_checksums',
	'zpool_trim_verify_trimmed']
	tags = ['functional', 'zpool_trim']

	[tests/functional/cli_root/zpool_upgrade]
	tests = ['zpool_upgrade_001_pos', 'zpool_upgrade_002_pos',
	'zpool_upgrade_003_pos', 'zpool_upgrade_004_pos',
	'zpool_upgrade_005_neg', 'zpool_upgrade_006_neg',
	'zpool_upgrade_007_pos', 'zpool_upgrade_008_pos',
	'zpool_upgrade_009_neg', 'zpool_upgrade_features_001_pos']
	tags = ['functional', 'cli_root', 'zpool_upgrade']

	[tests/functional/cli_root/zpool_wait]
	tests = ['zpool_wait_discard', 'zpool_wait_freeing',
	'zpool_wait_initialize_basic', 'zpool_wait_initialize_cancel',
	'zpool_wait_initialize_flag', 'zpool_wait_multiple',
	'zpool_wait_no_activity', 'zpool_wait_remove', 'zpool_wait_remove_cancel',
	'zpool_wait_trim_basic', 'zpool_wait_trim_cancel', 'zpool_wait_trim_flag',
	'zpool_wait_usage']
	tags = ['functional', 'cli_root', 'zpool_wait']

	[tests/functional/cli_root/zpool_wait/scan]
	tests = ['zpool_wait_replace_cancel', 'zpool_wait_rebuild',
	'zpool_wait_resilver', 'zpool_wait_scrub_cancel',
	'zpool_wait_replace', 'zpool_wait_scrub_basic', 'zpool_wait_scrub_flag']
	tags = ['functional', 'cli_root', 'zpool_wait']

	[tests/functional/cli_user/misc]
	tests = ['zdb_001_neg', 'zfs_001_neg', 'zfs_allow_001_neg',
	'zfs_clone_001_neg', 'zfs_create_001_neg', 'zfs_destroy_001_neg',
	'zfs_get_001_neg', 'zfs_inherit_001_neg', 'zfs_mount_001_neg',
	'zfs_promote_001_neg', 'zfs_receive_001_neg', 'zfs_rename_001_neg',
	'zfs_rollback_001_neg', 'zfs_send_001_neg', 'zfs_set_001_neg',
	'zfs_share_001_neg', 'zfs_snapshot_001_neg', 'zfs_unallow_001_neg',
	'zfs_unmount_001_neg', 'zfs_unshare_001_neg', 'zfs_upgrade_001_neg',
	'zpool_001_neg', 'zpool_add_001_neg', 'zpool_attach_001_neg',
	'zpool_clear_001_neg', 'zpool_create_001_neg', 'zpool_destroy_001_neg',
	'zpool_detach_001_neg', 'zpool_export_001_neg', 'zpool_get_001_neg',
	'zpool_history_001_neg', 'zpool_import_001_neg', 'zpool_import_002_neg',
	'zpool_offline_001_neg', 'zpool_online_001_neg', 'zpool_remove_001_neg',
	'zpool_replace_001_neg', 'zpool_scrub_001_neg', 'zpool_set_001_neg',
	'zpool_status_001_neg', 'zpool_upgrade_001_neg', 'arcstat_001_pos',
	'arc_summary_001_pos', 'arc_summary_002_neg', 'zpool_wait_privilege']
	user =
	tags = ['functional', 'cli_user', 'misc']

	[tests/functional/cli_user/zfs_list]
	tests = ['zfs_list_001_pos', 'zfs_list_002_pos', 'zfs_list_003_pos',
	'zfs_list_004_neg', 'zfs_list_007_pos', 'zfs_list_008_neg']
	user =
	tags = ['functional', 'cli_user', 'zfs_list']

	[tests/functional/cli_user/zpool_iostat]
	tests = ['zpool_iostat_001_neg', 'zpool_iostat_002_pos',
	'zpool_iostat_003_neg', 'zpool_iostat_004_pos',
	'zpool_iostat_005_pos', 'zpool_iostat_-c_disable',
	'zpool_iostat_-c_homedir', 'zpool_iostat_-c_searchpath']
	user =
	tags = ['functional', 'cli_user', 'zpool_iostat']

	[tests/functional/cli_user/zpool_list]
	tests = ['zpool_list_001_pos', 'zpool_list_002_neg']
	user =
	tags = ['functional', 'cli_user', 'zpool_list']

	[tests/functional/cli_user/zpool_status]
	tests = ['zpool_status_003_pos', 'zpool_status_-c_disable',
	'zpool_status_-c_homedir', 'zpool_status_-c_searchpath']
	user =
	tags = ['functional', 'cli_user', 'zpool_status']

	[tests/functional/compression]
	tests = ['compress_001_pos', 'compress_002_pos', 'compress_003_pos',
	'l2arc_compressed_arc', 'l2arc_compressed_arc_disabled',
	'l2arc_encrypted', 'l2arc_encrypted_no_compressed_arc']
	tags = ['functional', 'compression']

	[tests/functional/cp_files]
	tests = ['cp_files_001_pos']
	tags = ['functional', 'cp_files']

	[tests/functional/crtime]
	tests = ['crtime_001_pos' ]
	tags = ['functional', 'crtime']

	[tests/functional/ctime]
	tests = ['ctime_001_pos' ]
	tags = ['functional', 'ctime']

	[tests/functional/deadman]
	tests = ['deadman_ratelimit', 'deadman_sync', 'deadman_zio']
	pre =
	post =
	tags = ['functional', 'deadman']

	[tests/functional/delegate]
	tests = ['zfs_allow_001_pos', 'zfs_allow_002_pos', 'zfs_allow_003_pos',
	'zfs_allow_004_pos', 'zfs_allow_005_pos', 'zfs_allow_006_pos',
	'zfs_allow_007_pos', 'zfs_allow_008_pos', 'zfs_allow_009_neg',
	'zfs_allow_010_pos', 'zfs_allow_011_neg', 'zfs_allow_012_neg',
	'zfs_unallow_001_pos', 'zfs_unallow_002_pos', 'zfs_unallow_003_pos',
	'zfs_unallow_004_pos', 'zfs_unallow_005_pos', 'zfs_unallow_006_pos',
	'zfs_unallow_007_neg', 'zfs_unallow_008_neg']
	tags = ['functional', 'delegate']

	[tests/functional/exec]
	tests = ['exec_001_pos', 'exec_002_neg']
	tags = ['functional', 'exec']

	[tests/functional/fallocate]
	tests = ['fallocate_punch-hole']
	tags = ['functional', 'fallocate']

	[tests/functional/features/async_destroy]
	tests = ['async_destroy_001_pos']
	tags = ['functional', 'features', 'async_destroy']

	[tests/functional/features/large_dnode]
	tests = ['large_dnode_001_pos', 'large_dnode_003_pos', 'large_dnode_004_neg',
	'large_dnode_005_pos', 'large_dnode_007_neg', 'large_dnode_009_pos']
	tags = ['functional', 'features', 'large_dnode']

	[tests/functional/grow]
	pre =
	post =
	tests = ['grow_pool_001_pos', 'grow_replicas_001_pos']
	tags = ['functional', 'grow']

	[tests/functional/history]
	tests = ['history_001_pos', 'history_002_pos', 'history_003_pos',
	'history_004_pos', 'history_005_neg', 'history_006_neg',
	'history_007_pos', 'history_008_pos', 'history_009_pos',
	'history_010_pos']
	tags = ['functional', 'history']

	[tests/functional/hkdf]
	tests = ['run_hkdf_test']
	tags = ['functional', 'hkdf']

	[tests/functional/inheritance]
	tests = ['inherit_001_pos']
	pre =
	tags = ['functional', 'inheritance']

	[tests/functional/io]
	tests = ['sync', 'psync', 'posixaio', 'mmap']
	tags = ['functional', 'io']

	[tests/functional/inuse]
	tests = ['inuse_004_pos', 'inuse_005_pos', 'inuse_008_pos', 'inuse_009_pos']
	post =
	tags = ['functional', 'inuse']

	[tests/functional/large_files]
	tests = ['large_files_001_pos', 'large_files_002_pos']
	tags = ['functional', 'large_files']

	[tests/functional/largest_pool]
	tests = ['largest_pool_001_pos']
	pre =
	post =
	tags = ['functional', 'largest_pool']

	[tests/functional/limits]
	tests = ['filesystem_count', 'filesystem_limit', 'snapshot_count',
	'snapshot_limit']
	tags = ['functional', 'limits']

	[tests/functional/link_count]
	tests = ['link_count_001', 'link_count_root_inode']
	tags = ['functional', 'link_count']

	[tests/functional/migration]
	tests = ['migration_001_pos', 'migration_002_pos', 'migration_003_pos',
	'migration_004_pos', 'migration_005_pos', 'migration_006_pos',
	'migration_007_pos', 'migration_008_pos', 'migration_009_pos',
	'migration_010_pos', 'migration_011_pos', 'migration_012_pos']
	tags = ['functional', 'migration']

	[tests/functional/mmap]
	tests = ['mmap_write_001_pos', 'mmap_read_001_pos', 'mmap_seek_001_pos']
	tags = ['functional', 'mmap']

	[tests/functional/mount]
	tests = ['umount_001', 'umountall_001']
	tags = ['functional', 'mount']

	[tests/functional/mv_files]
	tests = ['mv_files_001_pos', 'mv_files_002_pos', 'random_creation']
	tags = ['functional', 'mv_files']

	[tests/functional/nestedfs]
	tests = ['nestedfs_001_pos']
	tags = ['functional', 'nestedfs']

	[tests/functional/no_space]
	tests = ['enospc_001_pos', 'enospc_002_pos', 'enospc_003_pos',
	'enospc_df']
	tags = ['functional', 'no_space']

	[tests/functional/nopwrite]
	tests = ['nopwrite_copies', 'nopwrite_mtime', 'nopwrite_negative',
	'nopwrite_promoted_clone', 'nopwrite_recsize', 'nopwrite_sync',
	'nopwrite_varying_compression', 'nopwrite_volume']
	tags = ['functional', 'nopwrite']

	[tests/functional/online_offline]
	tests = ['online_offline_001_pos', 'online_offline_002_neg',
	'online_offline_003_neg']
	tags = ['functional', 'online_offline']

	[tests/functional/pool_checkpoint]
	tests = ['checkpoint_after_rewind', 'checkpoint_big_rewind',
	'checkpoint_capacity', 'checkpoint_conf_change', 'checkpoint_discard',
	'checkpoint_discard_busy', 'checkpoint_discard_many',
	'checkpoint_indirect', 'checkpoint_invalid', 'checkpoint_lun_expsz',
	'checkpoint_open', 'checkpoint_removal', 'checkpoint_rewind',
	'checkpoint_ro_rewind', 'checkpoint_sm_scale', 'checkpoint_twice',
	'checkpoint_vdev_add', 'checkpoint_zdb', 'checkpoint_zhack_feat']
	tags = ['functional', 'pool_checkpoint']
	timeout = 1800

	[tests/functional/pool_names]
	tests = ['pool_names_001_pos', 'pool_names_002_neg']
	pre =
	post =
	tags = ['functional', 'pool_names']

	[tests/functional/poolversion]
	tests = ['poolversion_001_pos', 'poolversion_002_pos']
	tags = ['functional', 'poolversion']

	[tests/functional/pyzfs]
	tests = ['pyzfs_unittest']
	pre =
	post =
	tags = ['functional', 'pyzfs']

	[tests/functional/quota]
	tests = ['quota_001_pos', 'quota_002_pos', 'quota_003_pos',
	'quota_004_pos', 'quota_005_pos', 'quota_006_neg']
	tags = ['functional', 'quota']

	[tests/functional/redacted_send]
	tests = ['redacted_compressed', 'redacted_contents', 'redacted_deleted',
	'redacted_disabled_feature', 'redacted_embedded', 'redacted_holes',
	'redacted_incrementals', 'redacted_largeblocks', 'redacted_many_clones',
	'redacted_mixed_recsize', 'redacted_mounts', 'redacted_negative',
	'redacted_origin', 'redacted_panic', 'redacted_props', 'redacted_resume',
	'redacted_size', 'redacted_volume']
	tags = ['functional', 'redacted_send']

	[tests/functional/raidz]
	tests = ['raidz_001_neg', 'raidz_002_pos', 'raidz_003_pos', 'raidz_004_pos']
	tags = ['functional', 'raidz']

	[tests/functional/redundancy]
	tests = ['redundancy_draid', 'redundancy_draid1', 'redundancy_draid2',
	'redundancy_draid3', 'redundancy_draid_damaged', 'redundancy_draid_spare1',
	'redundancy_draid_spare2', 'redundancy_draid_spare3', 'redundancy_mirror',
	'redundancy_raidz', 'redundancy_raidz1', 'redundancy_raidz2',
	'redundancy_raidz3', 'redundancy_stripe']
	tags = ['functional', 'redundancy']
	timeout = 1200

	[tests/functional/refquota]
	tests = ['refquota_001_pos', 'refquota_002_pos', 'refquota_003_pos',
	'refquota_004_pos', 'refquota_005_pos', 'refquota_006_neg',
	'refquota_007_neg', 'refquota_008_neg']
	tags = ['functional', 'refquota']

	[tests/functional/refreserv]
	tests = ['refreserv_001_pos', 'refreserv_002_pos', 'refreserv_003_pos',
	'refreserv_004_pos', 'refreserv_005_pos', 'refreserv_multi_raidz',
	'refreserv_raidz']
	tags = ['functional', 'refreserv']

	[tests/functional/removal]
	pre =
	tests = ['removal_all_vdev', 'removal_cancel', 'removal_check_space',
	'removal_condense_export', 'removal_multiple_indirection',
	'removal_nopwrite', 'removal_remap_deadlists',
	'removal_resume_export', 'removal_sanity', 'removal_with_add',
	'removal_with_create_fs', 'removal_with_dedup',
	'removal_with_errors', 'removal_with_export',
	'removal_with_ganging', 'removal_with_faulted',
	'removal_with_remove', 'removal_with_scrub', 'removal_with_send',
	'removal_with_send_recv', 'removal_with_snapshot',
	'removal_with_write', 'removal_with_zdb', 'remove_expanded',
	'remove_mirror', 'remove_mirror_sanity', 'remove_raidz',
	'remove_indirect', 'remove_attach_mirror']
	tags = ['functional', 'removal']

	[tests/functional/rename_dirs]
	tests = ['rename_dirs_001_pos']
	tags = ['functional', 'rename_dirs']

	[tests/functional/replacement]
	tests = ['attach_import', 'attach_multiple', 'attach_rebuild',
	'attach_resilver', 'detach', 'rebuild_disabled_feature',
	'rebuild_multiple', 'rebuild_raidz', 'replace_import', 'replace_rebuild',
	'replace_resilver', 'resilver_restart_001', 'resilver_restart_002',
	'scrub_cancel']
	tags = ['functional', 'replacement']

	[tests/functional/reservation]
	tests = ['reservation_001_pos', 'reservation_002_pos', 'reservation_003_pos',
	'reservation_004_pos', 'reservation_005_pos', 'reservation_006_pos',
	'reservation_007_pos', 'reservation_008_pos', 'reservation_009_pos',
	'reservation_010_pos', 'reservation_011_pos', 'reservation_012_pos',
	'reservation_013_pos', 'reservation_014_pos', 'reservation_015_pos',
	'reservation_016_pos', 'reservation_017_pos', 'reservation_018_pos',
	'reservation_019_pos', 'reservation_020_pos', 'reservation_021_neg',
	'reservation_022_pos']
	tags = ['functional', 'reservation']

	[tests/functional/rootpool]
	tests = ['rootpool_002_neg', 'rootpool_003_neg', 'rootpool_007_pos']
	tags = ['functional', 'rootpool']

	[tests/functional/rsend]
	tests = ['recv_dedup', 'recv_dedup_encrypted_zvol', 'rsend_001_pos',
	'rsend_002_pos', 'rsend_003_pos', 'rsend_004_pos', 'rsend_005_pos',
	'rsend_006_pos', 'rsend_007_pos', 'rsend_008_pos', 'rsend_009_pos',
	'rsend_010_pos', 'rsend_011_pos', 'rsend_012_pos', 'rsend_013_pos',
	'rsend_014_pos', 'rsend_016_neg', 'rsend_019_pos', 'rsend_020_pos',
	'rsend_021_pos', 'rsend_022_pos', 'rsend_024_pos',
	'send-c_verify_ratio', 'send-c_verify_contents', 'send-c_props',
	'send-c_incremental', 'send-c_volume', 'send-c_zstreamdump',
	'send-c_lz4_disabled', 'send-c_recv_lz4_disabled',
	'send-c_mixed_compression', 'send-c_stream_size_estimate',
	'send-c_embedded_blocks', 'send-c_resume', 'send-cpL_varied_recsize',
	'send-c_recv_dedup', 'send-L_toggle', 'send_encrypted_hierarchy',
	'send_encrypted_props', 'send_encrypted_truncated_files',
	'send_freeobjects', 'send_realloc_files',
	'send_realloc_encrypted_files', 'send_spill_block', 'send_holds',
	'send_hole_birth', 'send_mixed_raw', 'send-wR_encrypted_zvol',
	'send_partial_dataset', 'send_invalid', 'send_doall']
	tags = ['functional', 'rsend']

	[tests/functional/scrub_mirror]
	tests = ['scrub_mirror_001_pos', 'scrub_mirror_002_pos',
	'scrub_mirror_003_pos', 'scrub_mirror_004_pos']
	tags = ['functional', 'scrub_mirror']

	[tests/functional/slog]
	tests = ['slog_001_pos', 'slog_002_pos', 'slog_003_pos', 'slog_004_pos',
	'slog_005_pos', 'slog_006_pos', 'slog_007_pos', 'slog_008_neg',
	'slog_009_neg', 'slog_010_neg', 'slog_011_neg', 'slog_012_neg',
	'slog_013_pos', 'slog_014_pos', 'slog_015_neg', 'slog_replay_fs_001',
	'slog_replay_fs_002', 'slog_replay_volume']
	tags = ['functional', 'slog']

	[tests/functional/snapshot]
	tests = ['clone_001_pos', 'rollback_001_pos', 'rollback_002_pos',
	'rollback_003_pos', 'snapshot_001_pos', 'snapshot_002_pos',
	'snapshot_003_pos', 'snapshot_004_pos', 'snapshot_005_pos',
	'snapshot_006_pos', 'snapshot_007_pos', 'snapshot_008_pos',
	'snapshot_009_pos', 'snapshot_010_pos', 'snapshot_011_pos',
	'snapshot_012_pos', 'snapshot_013_pos', 'snapshot_014_pos',
	'snapshot_017_pos']
	tags = ['functional', 'snapshot']

	[tests/functional/snapused]
	tests = ['snapused_001_pos', 'snapused_002_pos', 'snapused_003_pos',
	'snapused_004_pos', 'snapused_005_pos']
	tags = ['functional', 'snapused']

	[tests/functional/sparse]
	tests = ['sparse_001_pos']
	tags = ['functional', 'sparse']

	[tests/functional/suid]
	tests = ['suid_write_to_suid', 'suid_write_to_sgid', 'suid_write_to_suid_sgid',
	'suid_write_to_none']
	tags = ['functional', 'suid']

	[tests/functional/threadsappend]
	tests = ['threadsappend_001_pos']
	tags = ['functional', 'threadsappend']

	[tests/functional/trim]
	tests = ['autotrim_integrity', 'autotrim_config', 'autotrim_trim_integrity',
	'trim_integrity', 'trim_config', 'trim_l2arc']
	tags = ['functional', 'trim']

	[tests/functional/truncate]
	tests = ['truncate_001_pos', 'truncate_002_pos', 'truncate_timestamps']
	tags = ['functional', 'truncate']

	[tests/functional/upgrade]
	tests = ['upgrade_userobj_001_pos', 'upgrade_readonly_pool']
	tags = ['functional', 'upgrade']

	[tests/functional/userquota]
	tests = [
	'userquota_001_pos', 'userquota_002_pos', 'userquota_003_pos',
	'userquota_004_pos', 'userquota_005_neg', 'userquota_006_pos',
	'userquota_007_pos', 'userquota_008_pos', 'userquota_009_pos',
	'userquota_010_pos', 'userquota_011_pos', 'userquota_012_neg',
	'userspace_001_pos', 'userspace_002_pos', 'userspace_encrypted']
	tags = ['functional', 'userquota']

	[tests/functional/vdev_zaps]
	tests = ['vdev_zaps_001_pos', 'vdev_zaps_002_pos', 'vdev_zaps_003_pos',
	'vdev_zaps_004_pos', 'vdev_zaps_005_pos', 'vdev_zaps_006_pos',
	'vdev_zaps_007_pos']
	tags = ['functional', 'vdev_zaps']

	[tests/functional/write_dirs]
	tests = ['write_dirs_001_pos', 'write_dirs_002_pos']
	tags = ['functional', 'write_dirs']

	[tests/functional/xattr]
	tests = ['xattr_001_pos', 'xattr_002_neg', 'xattr_003_neg', 'xattr_004_pos',
	'xattr_005_pos', 'xattr_006_pos', 'xattr_007_neg',
	'xattr_011_pos', 'xattr_012_pos', 'xattr_013_pos']
	tags = ['functional', 'xattr']

	[tests/functional/zvol/zvol_ENOSPC]
	tests = ['zvol_ENOSPC_001_pos']
	tags = ['functional', 'zvol', 'zvol_ENOSPC']

	[tests/functional/zvol/zvol_cli]
	tests = ['zvol_cli_001_pos', 'zvol_cli_002_pos', 'zvol_cli_003_neg']
	tags = ['functional', 'zvol', 'zvol_cli']

	[tests/functional/zvol/zvol_misc]
	tests = ['zvol_misc_002_pos', 'zvol_misc_hierarchy', 'zvol_misc_rename_inuse',
	'zvol_misc_snapdev', 'zvol_misc_volmode', 'zvol_misc_zil']
	tags = ['functional', 'zvol', 'zvol_misc']

	[tests/functional/zvol/zvol_swap]
	tests = ['zvol_swap_001_pos', 'zvol_swap_002_pos', 'zvol_swap_004_pos']
	tags = ['functional', 'zvol', 'zvol_swap']

	[tests/functional/libzfs]
	tests = ['many_fds', 'libzfs_input']
	tags = ['functional', 'libzfs']

	[tests/functional/log_spacemap]
	tests = ['log_spacemap_import_logs']
	pre =
	post =
	tags = ['functional', 'log_spacemap']

	[tests/functional/l2arc]
	tests = ['l2arc_arcstats_pos', 'l2arc_mfuonly_pos', 'l2arc_l2miss_pos',
	'persist_l2arc_001_pos', 'persist_l2arc_002_pos',
	'persist_l2arc_003_neg', 'persist_l2arc_004_pos', 'persist_l2arc_005_pos']
	tags = ['functional', 'l2arc']

	[tests/functional/zpool_influxdb]
	tests = ['zpool_influxdb']
	tags = ['functional', 'zpool_influxdb']
	diff --git a/sys/contrib/openzfs/tests/runfiles/linux.run b/sys/contrib/openzfs/tests/runfiles/linux.run
	index eab9c5dc7743..58eefec319ea 100644
	--- a/sys/contrib/openzfs/tests/runfiles/linux.run
	+++ b/sys/contrib/openzfs/tests/runfiles/linux.run
	@@ -1,170 +1,170 @@
	#
	# This file and its contents are supplied under the terms of the
	# Common Development and Distribution License ("CDDL"), version 1.0.
	# You may only use this file in accordance with the terms of version
	# 1.0 of the CDDL.
	#
	# A full copy of the text of the CDDL should have accompanied this
	# source. A copy of the CDDL is also available via the Internet at
	# http://www.illumos.org/license/CDDL.
	#

	[DEFAULT]
	pre = setup
	quiet = False
	pre_user = root
	user = root
	timeout = 600
	post_user = root
	post = cleanup
	failsafe_user = root
	failsafe = callbacks/zfs_failsafe
	outputdir = /var/tmp/test_results
	tags = ['functional']

	[tests/functional/acl/posix:Linux]
	tests = ['posix_001_pos', 'posix_002_pos', 'posix_003_pos', 'posix_004_pos']
	tags = ['functional', 'acl', 'posix']

	[tests/functional/acl/posix-sa:Linux]
	tests = ['posix_001_pos', 'posix_002_pos', 'posix_003_pos', 'posix_004_pos']
	tags = ['functional', 'acl', 'posix-sa']

	[tests/functional/atime:Linux]
	tests = ['atime_003_pos', 'root_relatime_on']
	tags = ['functional', 'atime']

	[tests/functional/chattr:Linux]
	tests = ['chattr_001_pos', 'chattr_002_neg']
	tags = ['functional', 'chattr']

	[tests/functional/cli_root/zfs:Linux]
	tests = ['zfs_003_neg']
	tags = ['functional', 'cli_root', 'zfs']

	[tests/functional/cli_root/zfs_mount:Linux]
	tests = ['zfs_mount_006_pos', 'zfs_mount_008_pos', 'zfs_mount_013_pos',
	'zfs_mount_014_neg', 'zfs_multi_mount']
	tags = ['functional', 'cli_root', 'zfs_mount']

	[tests/functional/cli_root/zfs_share:Linux]
	tests = ['zfs_share_005_pos', 'zfs_share_007_neg', 'zfs_share_009_neg',
	'zfs_share_012_pos', 'zfs_share_013_pos']
	tags = ['functional', 'cli_root', 'zfs_share']

	[tests/functional/cli_root/zfs_sysfs:Linux]
	tests = ['zfeature_set_unsupported', 'zfs_get_unsupported',
	'zfs_set_unsupported', 'zfs_sysfs_live', 'zpool_get_unsupported',
	'zpool_set_unsupported']
	tags = ['functional', 'cli_root', 'zfs_sysfs']

	[tests/functional/cli_root/zpool_add:Linux]
	tests = ['add_nested_replacing_spare']
	tags = ['functional', 'cli_root', 'zpool_add']

	[tests/functional/cli_root/zpool_expand:Linux]
	tests = ['zpool_expand_001_pos', 'zpool_expand_002_pos',
	'zpool_expand_003_neg', 'zpool_expand_004_pos', 'zpool_expand_005_pos']
	tags = ['functional', 'cli_root', 'zpool_expand']

	[tests/functional/cli_root/zpool_reopen:Linux]
	tests = ['zpool_reopen_001_pos', 'zpool_reopen_002_pos',
	'zpool_reopen_003_pos', 'zpool_reopen_004_pos', 'zpool_reopen_005_pos',
	'zpool_reopen_006_neg', 'zpool_reopen_007_pos']
	tags = ['functional', 'cli_root', 'zpool_reopen']

	[tests/functional/cli_root/zpool_split:Linux]
	tests = ['zpool_split_wholedisk']
	tags = ['functional', 'cli_root', 'zpool_split']

	[tests/functional/compression:Linux]
	tests = ['compress_004_pos']
	tags = ['functional', 'compression']

	[tests/functional/devices:Linux]
	tests = ['devices_001_pos', 'devices_002_neg', 'devices_003_pos']
	tags = ['functional', 'devices']

	[tests/functional/events:Linux]
	tests = ['events_001_pos', 'events_002_pos', 'zed_rc_filter', 'zed_fd_spill']
	tags = ['functional', 'events']

	[tests/functional/fallocate:Linux]
	tests = ['fallocate_prealloc']
	tags = ['functional', 'fallocate']

	[tests/functional/fault:Linux]
	tests = ['auto_offline_001_pos', 'auto_online_001_pos', 'auto_online_002_pos',
	'auto_replace_001_pos', 'auto_spare_001_pos', 'auto_spare_002_pos',
	'auto_spare_multiple', 'auto_spare_ashift', 'auto_spare_shared',
	'decrypt_fault', 'decompress_fault', 'scrub_after_resilver',
	'zpool_status_-s']
	tags = ['functional', 'fault']

	[tests/functional/features/large_dnode:Linux]
	tests = ['large_dnode_002_pos', 'large_dnode_006_pos', 'large_dnode_008_pos']
	tags = ['functional', 'features', 'large_dnode']

	[tests/functional/io:Linux]
	tests = ['libaio', 'io_uring']
	tags = ['functional', 'io']

	[tests/functional/mmap:Linux]
	tests = ['mmap_libaio_001_pos']
	tags = ['functional', 'mmap']

	[tests/functional/mmp:Linux]
	tests = ['mmp_on_thread', 'mmp_on_uberblocks', 'mmp_on_off', 'mmp_interval',
	'mmp_active_import', 'mmp_inactive_import', 'mmp_exported_import',
	'mmp_write_uberblocks', 'mmp_reset_interval', 'multihost_history',
	'mmp_on_zdb', 'mmp_write_distribution', 'mmp_hostid']
	tags = ['functional', 'mmp']

	[tests/functional/mount:Linux]
	tests = ['umount_unlinked_drain']
	tags = ['functional', 'mount']

	[tests/functional/pam:Linux]
	-tests = ['pam_basic', 'pam_nounmount']
	+tests = ['pam_basic', 'pam_nounmount', 'pam_short_password']
	tags = ['functional', 'pam']

	[tests/functional/procfs:Linux]
	tests = ['procfs_list_basic', 'procfs_list_concurrent_readers',
	'procfs_list_stale_read', 'pool_state']
	tags = ['functional', 'procfs']

	[tests/functional/projectquota:Linux]
	tests = ['projectid_001_pos', 'projectid_002_pos', 'projectid_003_pos',
	'projectquota_001_pos', 'projectquota_002_pos', 'projectquota_003_pos',
	'projectquota_004_neg', 'projectquota_005_pos', 'projectquota_006_pos',
	'projectquota_007_pos', 'projectquota_008_pos', 'projectquota_009_pos',
	'projectspace_001_pos', 'projectspace_002_pos', 'projectspace_003_pos',
	'projectspace_004_pos',
	'projecttree_001_pos', 'projecttree_002_pos', 'projecttree_003_neg']
	tags = ['functional', 'projectquota']

	[tests/functional/rsend:Linux]
	tests = ['send_realloc_dnode_size', 'send_encrypted_files']
	tags = ['functional', 'rsend']

	[tests/functional/snapshot:Linux]
	tests = ['snapshot_015_pos', 'snapshot_016_pos']
	tags = ['functional', 'snapshot']

	[tests/functional/tmpfile:Linux]
	tests = ['tmpfile_001_pos', 'tmpfile_002_pos', 'tmpfile_003_pos',
	'tmpfile_stat_mode']
	tags = ['functional', 'tmpfile']

	[tests/functional/upgrade:Linux]
	tests = ['upgrade_projectquota_001_pos']
	tags = ['functional', 'upgrade']

	[tests/functional/user_namespace:Linux]
	tests = ['user_namespace_001']
	tags = ['functional', 'user_namespace']

	[tests/functional/userquota:Linux]
	tests = ['groupspace_001_pos', 'groupspace_002_pos', 'groupspace_003_pos',
	'userquota_013_pos', 'userspace_003_pos']
	tags = ['functional', 'userquota']
	diff --git a/sys/contrib/openzfs/tests/runfiles/sanity.run b/sys/contrib/openzfs/tests/runfiles/sanity.run
	index 2a9b4adf5fb9..8057d8148c93 100644
	--- a/sys/contrib/openzfs/tests/runfiles/sanity.run
	+++ b/sys/contrib/openzfs/tests/runfiles/sanity.run
	@@ -1,623 +1,622 @@
	#
	# This file and its contents are supplied under the terms of the
	# Common Development and Distribution License ("CDDL"), version 1.0.
	# You may only use this file in accordance with the terms of version
	# 1.0 of the CDDL.
	#
	# A full copy of the text of the CDDL should have accompanied this
	# source. A copy of the CDDL is also available via the Internet at
	# http://www.illumos.org/license/CDDL.
	#
	# This run file contains a subset of functional tests which exercise
	# as much functionality as possible while still executing relatively
	# quickly. The included tests should take no more than a few seconds
	# each to run at most. This provides a convenient way to sanity test a
	# change before committing to a full test run which takes several hours.
	#
	# Approximate run time: 15 minutes
	#

	[DEFAULT]
	pre = setup
	quiet = False
	pre_user = root
	user = root
	timeout = 180
	post_user = root
	post = cleanup
	failsafe_user = root
	failsafe = callbacks/zfs_failsafe
	outputdir = /var/tmp/test_results
	tags = ['functional']

	[tests/functional/acl/off]
	tests = ['posixmode']
	tags = ['functional', 'acl']

	[tests/functional/alloc_class]
	tests = ['alloc_class_003_pos', 'alloc_class_004_pos', 'alloc_class_005_pos',
	'alloc_class_006_pos', 'alloc_class_008_pos', 'alloc_class_010_pos',
	'alloc_class_011_neg']
	tags = ['functional', 'alloc_class']

	[tests/functional/arc]
	tests = ['dbufstats_001_pos', 'dbufstats_002_pos', 'arcstats_runtime_tuning']
	tags = ['functional', 'arc']

	[tests/functional/bootfs]
	tests = ['bootfs_004_neg', 'bootfs_007_pos']
	tags = ['functional', 'bootfs']

	[tests/functional/cache]
	tests = ['cache_004_neg', 'cache_005_neg', 'cache_007_neg', 'cache_010_pos']
	tags = ['functional', 'cache']

	[tests/functional/cachefile]
	tests = ['cachefile_001_pos', 'cachefile_002_pos', 'cachefile_003_pos',
	'cachefile_004_pos']
	tags = ['functional', 'cachefile']

	[tests/functional/casenorm]
	tests = ['case_all_values', 'norm_all_values', 'sensitive_none_lookup',
	'sensitive_none_delete', 'insensitive_none_lookup',
	'insensitive_none_delete', 'mixed_none_lookup', 'mixed_none_delete']
	tags = ['functional', 'casenorm']

	[tests/functional/channel_program/lua_core]
	tests = ['tst.args_to_lua', 'tst.divide_by_zero', 'tst.exists',
	'tst.integer_illegal', 'tst.integer_overflow', 'tst.language_functions_neg',
	'tst.language_functions_pos', 'tst.large_prog', 'tst.libraries',
	'tst.memory_limit', 'tst.nested_neg', 'tst.nested_pos', 'tst.nvlist_to_lua',
	'tst.recursive_neg', 'tst.recursive_pos', 'tst.return_large',
	'tst.return_nvlist_neg', 'tst.return_nvlist_pos',
	'tst.return_recursive_table', 'tst.stack_gsub', 'tst.timeout']
	tags = ['functional', 'channel_program', 'lua_core']

	[tests/functional/channel_program/synctask_core]
	tests = ['tst.destroy_fs', 'tst.destroy_snap', 'tst.get_count_and_limit',
	'tst.get_index_props', 'tst.get_mountpoint', 'tst.get_neg',
	'tst.get_number_props', 'tst.get_string_props', 'tst.get_type',
	'tst.get_userquota', 'tst.get_written', 'tst.inherit', 'tst.list_bookmarks',
	'tst.list_children', 'tst.list_clones', 'tst.list_holds',
	'tst.list_snapshots', 'tst.list_system_props',
	'tst.list_user_props', 'tst.parse_args_neg','tst.promote_conflict',
	'tst.promote_multiple', 'tst.promote_simple', 'tst.rollback_mult',
	'tst.rollback_one', 'tst.set_props', 'tst.snapshot_destroy',
	'tst.snapshot_neg', 'tst.snapshot_recursive', 'tst.snapshot_simple',
	'tst.bookmark.create', 'tst.bookmark.copy']
	tags = ['functional', 'channel_program', 'synctask_core']

	[tests/functional/cli_root/zdb]
	tests = ['zdb_003_pos', 'zdb_004_pos', 'zdb_005_pos']
	pre =
	post =
	tags = ['functional', 'cli_root', 'zdb']

	[tests/functional/cli_root/zfs]
	tests = ['zfs_001_neg', 'zfs_002_pos']
	tags = ['functional', 'cli_root', 'zfs']

	[tests/functional/cli_root/zfs_bookmark]
	tests = ['zfs_bookmark_cliargs']
	tags = ['functional', 'cli_root', 'zfs_bookmark']

	[tests/functional/cli_root/zfs_change-key]
	tests = ['zfs_change-key', 'zfs_change-key_child', 'zfs_change-key_format',
	'zfs_change-key_inherit', 'zfs_change-key_load', 'zfs_change-key_location',
	'zfs_change-key_pbkdf2iters', 'zfs_change-key_clones']
	tags = ['functional', 'cli_root', 'zfs_change-key']

	[tests/functional/cli_root/zfs_clone]
	tests = ['zfs_clone_001_neg', 'zfs_clone_002_pos', 'zfs_clone_003_pos',
	'zfs_clone_004_pos', 'zfs_clone_005_pos', 'zfs_clone_006_pos',
	'zfs_clone_007_pos', 'zfs_clone_008_neg', 'zfs_clone_009_neg',
	'zfs_clone_encrypted']
	tags = ['functional', 'cli_root', 'zfs_clone']

	[tests/functional/cli_root/zfs_create]
	tests = ['zfs_create_001_pos', 'zfs_create_002_pos', 'zfs_create_003_pos',
	'zfs_create_004_pos', 'zfs_create_005_pos', 'zfs_create_006_pos',
	'zfs_create_007_pos', 'zfs_create_011_pos', 'zfs_create_012_pos',
	'zfs_create_013_pos', 'zfs_create_014_pos', 'zfs_create_encrypted',
	'zfs_create_dryrun', 'zfs_create_verbose']
	tags = ['functional', 'cli_root', 'zfs_create']

	[tests/functional/cli_root/zfs_destroy]
	tests = ['zfs_destroy_002_pos', 'zfs_destroy_003_pos',
	'zfs_destroy_004_pos', 'zfs_destroy_006_neg', 'zfs_destroy_007_neg',
	'zfs_destroy_008_pos', 'zfs_destroy_009_pos', 'zfs_destroy_010_pos',
	'zfs_destroy_011_pos', 'zfs_destroy_012_pos', 'zfs_destroy_013_neg',
	'zfs_destroy_014_pos', 'zfs_destroy_dev_removal',
	'zfs_destroy_dev_removal_condense']
	tags = ['functional', 'cli_root', 'zfs_destroy']

	[tests/functional/cli_root/zfs_diff]
	tests = ['zfs_diff_cliargs', 'zfs_diff_encrypted']
	tags = ['functional', 'cli_root', 'zfs_diff']

	[tests/functional/cli_root/zfs_get]
	tests = ['zfs_get_003_pos', 'zfs_get_006_neg', 'zfs_get_007_neg',
	'zfs_get_010_neg']
	tags = ['functional', 'cli_root', 'zfs_get']

	[tests/functional/cli_root/zfs_inherit]
	tests = ['zfs_inherit_001_neg', 'zfs_inherit_003_pos', 'zfs_inherit_mountpoint']
	tags = ['functional', 'cli_root', 'zfs_inherit']

	[tests/functional/cli_root/zfs_load-key]
	tests = ['zfs_load-key', 'zfs_load-key_all', 'zfs_load-key_file',
	'zfs_load-key_https', 'zfs_load-key_location', 'zfs_load-key_noop',
	'zfs_load-key_recursive']
	tags = ['functional', 'cli_root', 'zfs_load-key']

	[tests/functional/cli_root/zfs_mount]
	tests = ['zfs_mount_001_pos', 'zfs_mount_002_pos', 'zfs_mount_003_pos',
	'zfs_mount_004_pos', 'zfs_mount_005_pos', 'zfs_mount_007_pos',
	'zfs_mount_009_neg', 'zfs_mount_010_neg', 'zfs_mount_011_neg',
	'zfs_mount_012_pos', 'zfs_mount_encrypted', 'zfs_mount_remount',
	'zfs_mount_all_fail', 'zfs_mount_all_mountpoints', 'zfs_mount_test_race']
	tags = ['functional', 'cli_root', 'zfs_mount']

	[tests/functional/cli_root/zfs_program]
	tests = ['zfs_program_json']
	tags = ['functional', 'cli_root', 'zfs_program']

	[tests/functional/cli_root/zfs_promote]
	tests = ['zfs_promote_001_pos', 'zfs_promote_002_pos', 'zfs_promote_003_pos',
	'zfs_promote_004_pos', 'zfs_promote_005_pos', 'zfs_promote_006_neg',
	'zfs_promote_007_neg', 'zfs_promote_008_pos', 'zfs_promote_encryptionroot']
	tags = ['functional', 'cli_root', 'zfs_promote']

	[tests/functional/cli_root/zfs_receive]
	tests = ['zfs_receive_001_pos', 'zfs_receive_002_pos', 'zfs_receive_003_pos',
	'zfs_receive_004_neg', 'zfs_receive_005_neg', 'zfs_receive_006_pos',
	'zfs_receive_007_neg', 'zfs_receive_008_pos', 'zfs_receive_009_neg',
	'zfs_receive_010_pos', 'zfs_receive_011_pos', 'zfs_receive_012_pos',
	'zfs_receive_013_pos', 'zfs_receive_014_pos', 'zfs_receive_015_pos',
	'zfs_receive_016_pos', 'zfs_receive_from_encrypted',
	'zfs_receive_to_encrypted', 'zfs_receive_raw',
	'zfs_receive_raw_incremental', 'zfs_receive_-e',
	'zfs_receive_raw_-d', 'zfs_receive_from_zstd', 'zfs_receive_new_props']
	tags = ['functional', 'cli_root', 'zfs_receive']

	[tests/functional/cli_root/zfs_rename]
	tests = ['zfs_rename_003_pos', 'zfs_rename_004_neg',
	'zfs_rename_005_neg', 'zfs_rename_006_pos', 'zfs_rename_007_pos',
	'zfs_rename_008_pos', 'zfs_rename_009_neg', 'zfs_rename_010_neg',
	'zfs_rename_011_pos', 'zfs_rename_012_neg', 'zfs_rename_013_pos',
	'zfs_rename_encrypted_child', 'zfs_rename_to_encrypted',
	'zfs_rename_mountpoint', 'zfs_rename_nounmount']
	tags = ['functional', 'cli_root', 'zfs_rename']

	[tests/functional/cli_root/zfs_reservation]
	tests = ['zfs_reservation_001_pos', 'zfs_reservation_002_pos']
	tags = ['functional', 'cli_root', 'zfs_reservation']

	[tests/functional/cli_root/zfs_rollback]
	tests = ['zfs_rollback_003_neg', 'zfs_rollback_004_neg']
	tags = ['functional', 'cli_root', 'zfs_rollback']

	[tests/functional/cli_root/zfs_send]
	tests = ['zfs_send_001_pos', 'zfs_send_002_pos', 'zfs_send_003_pos',
	'zfs_send_004_neg', 'zfs_send_005_pos', 'zfs_send_encrypted',
	'zfs_send_raw']
	tags = ['functional', 'cli_root', 'zfs_send']

	[tests/functional/cli_root/zfs_set]
	tests = ['cache_001_pos', 'cache_002_neg', 'canmount_001_pos',
	'canmount_002_pos', 'canmount_003_pos', 'canmount_004_pos',
	'checksum_001_pos', 'compression_001_pos', 'mountpoint_001_pos',
	'mountpoint_002_pos', 'user_property_002_pos',
	'share_mount_001_neg', 'snapdir_001_pos', 'onoffs_001_pos',
	'user_property_001_pos', 'user_property_003_neg', 'readonly_001_pos',
	'user_property_004_pos', 'version_001_neg',
	'zfs_set_003_neg', 'property_alias_001_pos',
	'zfs_set_keylocation', 'zfs_set_feature_activation']
	tags = ['functional', 'cli_root', 'zfs_set']

	[tests/functional/cli_root/zfs_snapshot]
	tests = ['zfs_snapshot_001_neg', 'zfs_snapshot_002_neg',
	'zfs_snapshot_003_neg', 'zfs_snapshot_006_pos', 'zfs_snapshot_007_neg']
	tags = ['functional', 'cli_root', 'zfs_snapshot']

	[tests/functional/cli_root/zfs_unload-key]
	tests = ['zfs_unload-key', 'zfs_unload-key_all', 'zfs_unload-key_recursive']
	tags = ['functional', 'cli_root', 'zfs_unload-key']

	[tests/functional/cli_root/zfs_unmount]
	tests = ['zfs_unmount_001_pos', 'zfs_unmount_002_pos', 'zfs_unmount_003_pos',
	'zfs_unmount_004_pos', 'zfs_unmount_007_neg', 'zfs_unmount_008_neg',
	'zfs_unmount_009_pos', 'zfs_unmount_unload_keys']
	tags = ['functional', 'cli_root', 'zfs_unmount']

	[tests/functional/cli_root/zfs_upgrade]
	tests = ['zfs_upgrade_001_pos', 'zfs_upgrade_002_pos', 'zfs_upgrade_006_neg',
	'zfs_upgrade_007_neg']
	tags = ['functional', 'cli_root', 'zfs_upgrade']

	[tests/functional/cli_root/zfs_wait]
	tests = ['zfs_wait_deleteq']
	tags = ['functional', 'cli_root', 'zfs_wait']

	[tests/functional/cli_root/zpool]
	tests = ['zpool_001_neg', 'zpool_003_pos', 'zpool_colors']
	tags = ['functional', 'cli_root', 'zpool']

	[tests/functional/cli_root/zpool_add]
	tests = ['zpool_add_002_pos', 'zpool_add_003_pos',
	'zpool_add_004_pos', 'zpool_add_006_pos', 'zpool_add_007_neg',
	'zpool_add_008_neg', 'zpool_add_009_neg']
	tags = ['functional', 'cli_root', 'zpool_add']

	[tests/functional/cli_root/zpool_attach]
	tests = ['zpool_attach_001_neg']
	tags = ['functional', 'cli_root', 'zpool_attach']

	[tests/functional/cli_root/zpool_clear]
	tests = ['zpool_clear_002_neg']
	tags = ['functional', 'cli_root', 'zpool_clear']

	[tests/functional/cli_root/zpool_create]
	tests = ['zpool_create_001_pos', 'zpool_create_002_pos',
	'zpool_create_003_pos', 'zpool_create_004_pos', 'zpool_create_007_neg',
	'zpool_create_008_pos', 'zpool_create_010_neg', 'zpool_create_011_neg',
	'zpool_create_012_neg', 'zpool_create_014_neg', 'zpool_create_015_neg',
	'zpool_create_017_neg', 'zpool_create_018_pos', 'zpool_create_019_pos',
	'zpool_create_020_pos', 'zpool_create_021_pos', 'zpool_create_022_pos',
	'zpool_create_encrypted',
	'zpool_create_features_001_pos', 'zpool_create_features_002_pos',
	'zpool_create_features_003_pos', 'zpool_create_features_004_neg',
	'zpool_create_features_005_pos']
	tags = ['functional', 'cli_root', 'zpool_create']

	[tests/functional/cli_root/zpool_destroy]
	tests = ['zpool_destroy_001_pos', 'zpool_destroy_002_pos',
	'zpool_destroy_003_neg']
	pre =
	post =
	tags = ['functional', 'cli_root', 'zpool_destroy']

	[tests/functional/cli_root/zpool_detach]
	tests = ['zpool_detach_001_neg']
	tags = ['functional', 'cli_root', 'zpool_detach']

	[tests/functional/cli_root/zpool_events]
	tests = ['zpool_events_clear', 'zpool_events_follow', 'zpool_events_poolname']
	tags = ['functional', 'cli_root', 'zpool_events']

	[tests/functional/cli_root/zpool_export]
	tests = ['zpool_export_001_pos', 'zpool_export_002_pos', 'zpool_export_003_neg']
	tags = ['functional', 'cli_root', 'zpool_export']

	[tests/functional/cli_root/zpool_get]
	tests = ['zpool_get_001_pos', 'zpool_get_002_pos', 'zpool_get_003_pos',
	'zpool_get_004_neg', 'zpool_get_005_pos']
	tags = ['functional', 'cli_root', 'zpool_get']

	[tests/functional/cli_root/zpool_history]
	tests = ['zpool_history_001_neg', 'zpool_history_002_pos']
	tags = ['functional', 'cli_root', 'zpool_history']

	[tests/functional/cli_root/zpool_import]
	tests = ['zpool_import_003_pos', 'zpool_import_010_pos', 'zpool_import_011_neg',
	'zpool_import_014_pos', 'zpool_import_features_001_pos',
	'zpool_import_all_001_pos', 'zpool_import_encrypted']
	tags = ['functional', 'cli_root', 'zpool_import']

	[tests/functional/cli_root/zpool_labelclear]
	tests = ['zpool_labelclear_active', 'zpool_labelclear_exported',
	'zpool_labelclear_removed', 'zpool_labelclear_valid']
	pre =
	post =
	tags = ['functional', 'cli_root', 'zpool_labelclear']

	[tests/functional/cli_root/zpool_initialize]
	tests = ['zpool_initialize_online_offline']
	pre =
	tags = ['functional', 'cli_root', 'zpool_initialize']

	[tests/functional/cli_root/zpool_offline]
	tests = ['zpool_offline_001_pos', 'zpool_offline_002_neg']
	tags = ['functional', 'cli_root', 'zpool_offline']

	[tests/functional/cli_root/zpool_online]
	tests = ['zpool_online_001_pos', 'zpool_online_002_neg']
	tags = ['functional', 'cli_root', 'zpool_online']

	[tests/functional/cli_root/zpool_remove]
	tests = ['zpool_remove_001_neg', 'zpool_remove_002_pos',
	'zpool_remove_003_pos']
	tags = ['functional', 'cli_root', 'zpool_remove']

	[tests/functional/cli_root/zpool_replace]
	tests = ['zpool_replace_001_neg']
	tags = ['functional', 'cli_root', 'zpool_replace']

	[tests/functional/cli_root/zpool_resilver]
	tests = ['zpool_resilver_bad_args']
	tags = ['functional', 'cli_root', 'zpool_resilver']

	[tests/functional/cli_root/zpool_scrub]
	tests = ['zpool_scrub_001_neg', 'zpool_scrub_003_pos',
	'zpool_scrub_encrypted_unloaded', 'zpool_scrub_print_repairing',
	'zpool_scrub_offline_device', 'zpool_scrub_multiple_copies']
	tags = ['functional', 'cli_root', 'zpool_scrub']

	[tests/functional/cli_root/zpool_set]
	tests = ['zpool_set_001_pos', 'zpool_set_002_neg', 'zpool_set_003_neg',
	'zpool_set_ashift', 'zpool_set_features']
	tags = ['functional', 'cli_root', 'zpool_set']

	[tests/functional/cli_root/zpool_split]
	tests = ['zpool_split_cliargs', 'zpool_split_devices',
	'zpool_split_props', 'zpool_split_vdevs', 'zpool_split_indirect']
	tags = ['functional', 'cli_root', 'zpool_split']

	[tests/functional/cli_root/zpool_status]
	tests = ['zpool_status_001_pos', 'zpool_status_002_pos']
	tags = ['functional', 'cli_root', 'zpool_status']

	[tests/functional/cli_root/zpool_sync]
	tests = ['zpool_sync_002_neg']
	tags = ['functional', 'cli_root', 'zpool_sync']

	[tests/functional/cli_root/zpool_trim]
	tests = ['zpool_trim_attach_detach_add_remove', 'zpool_trim_neg',
	'zpool_trim_offline_export_import_online', 'zpool_trim_online_offline',
	'zpool_trim_rate_neg', 'zpool_trim_secure', 'zpool_trim_split',
	'zpool_trim_start_and_cancel_neg', 'zpool_trim_start_and_cancel_pos']
	tags = ['functional', 'zpool_trim']

	[tests/functional/cli_root/zpool_upgrade]
	tests = ['zpool_upgrade_001_pos', 'zpool_upgrade_003_pos',
	'zpool_upgrade_005_neg', 'zpool_upgrade_006_neg',
	'zpool_upgrade_009_neg']
	tags = ['functional', 'cli_root', 'zpool_upgrade']

	[tests/functional/cli_root/zpool_wait]
	tests = ['zpool_wait_no_activity', 'zpool_wait_usage']
	tags = ['functional', 'cli_root', 'zpool_wait']

	[tests/functional/cli_root/zpool_wait/scan]
	tests = ['zpool_wait_scrub_flag']
	tags = ['functional', 'cli_root', 'zpool_wait']

	[tests/functional/cli_user/misc]
	tests = ['zdb_001_neg', 'zfs_001_neg', 'zfs_allow_001_neg',
	'zfs_clone_001_neg', 'zfs_create_001_neg', 'zfs_destroy_001_neg',
	'zfs_get_001_neg', 'zfs_inherit_001_neg', 'zfs_mount_001_neg',
	'zfs_promote_001_neg', 'zfs_receive_001_neg', 'zfs_rename_001_neg',
	'zfs_rollback_001_neg', 'zfs_send_001_neg', 'zfs_set_001_neg',
	'zfs_snapshot_001_neg', 'zfs_unallow_001_neg',
	'zfs_unmount_001_neg', 'zfs_upgrade_001_neg',
	'zpool_001_neg', 'zpool_add_001_neg', 'zpool_attach_001_neg',
	'zpool_clear_001_neg', 'zpool_create_001_neg', 'zpool_destroy_001_neg',
	'zpool_detach_001_neg', 'zpool_export_001_neg', 'zpool_get_001_neg',
	'zpool_history_001_neg', 'zpool_offline_001_neg', 'zpool_online_001_neg',
	'zpool_remove_001_neg', 'zpool_scrub_001_neg', 'zpool_set_001_neg',
	'zpool_status_001_neg', 'zpool_upgrade_001_neg', 'arcstat_001_pos',
	'arc_summary_001_pos', 'arc_summary_002_neg', 'zpool_wait_privilege']
	user =
	tags = ['functional', 'cli_user', 'misc']

	[tests/functional/cli_user/zpool_iostat]
	tests = ['zpool_iostat_001_neg', 'zpool_iostat_002_pos',
	'zpool_iostat_003_neg', 'zpool_iostat_004_pos',
	'zpool_iostat_-c_disable',
	'zpool_iostat_-c_homedir', 'zpool_iostat_-c_searchpath']
	user =
	tags = ['functional', 'cli_user', 'zpool_iostat']

	[tests/functional/cli_user/zpool_list]
	tests = ['zpool_list_001_pos', 'zpool_list_002_neg']
	user =
	tags = ['functional', 'cli_user', 'zpool_list']

	[tests/functional/compression]
	tests = ['compress_003_pos','compress_zstd_bswap']
	tags = ['functional', 'compression']

	[tests/functional/exec]
	tests = ['exec_001_pos', 'exec_002_neg']
	tags = ['functional', 'exec']

	[tests/functional/features/large_dnode]
	tests = ['large_dnode_003_pos', 'large_dnode_004_neg',
	'large_dnode_005_pos', 'large_dnode_007_neg']
	tags = ['functional', 'features', 'large_dnode']

	[tests/functional/grow]
	pre =
	post =
	tests = ['grow_pool_001_pos', 'grow_replicas_001_pos']
	tags = ['functional', 'grow']

	[tests/functional/history]
	-tests = ['history_004_pos', 'history_005_neg', 'history_006_neg',
	- 'history_007_pos', 'history_008_pos', 'history_009_pos']
	+tests = ['history_004_pos', 'history_005_neg', 'history_007_pos',
	+ 'history_009_pos']
	tags = ['functional', 'history']

	[tests/functional/hkdf]
	tests = ['run_hkdf_test']
	tags = ['functional', 'hkdf']

	[tests/functional/inuse]
	tests = ['inuse_004_pos', 'inuse_005_pos']
	post =
	tags = ['functional', 'inuse']

	[tests/functional/large_files]
	tests = ['large_files_001_pos', 'large_files_002_pos']
	tags = ['functional', 'large_files']

	[tests/functional/libzfs]
	tests = ['many_fds', 'libzfs_input']
	tags = ['functional', 'libzfs']

	[tests/functional/limits]
	tests = ['filesystem_count', 'snapshot_count']
	tags = ['functional', 'limits']

	[tests/functional/link_count]
	tests = ['link_count_root_inode']
	tags = ['functional', 'link_count']

	[tests/functional/log_spacemap]
	tests = ['log_spacemap_import_logs']
	pre =
	post =
	tags = ['functional', 'log_spacemap']

	[tests/functional/migration]
	tests = ['migration_001_pos', 'migration_002_pos', 'migration_003_pos',
	'migration_004_pos', 'migration_005_pos', 'migration_006_pos',
	'migration_007_pos', 'migration_008_pos', 'migration_009_pos',
	'migration_010_pos', 'migration_011_pos', 'migration_012_pos']
	tags = ['functional', 'migration']

	[tests/functional/mmap]
	tests = ['mmap_read_001_pos']
	tags = ['functional', 'mmap']

	[tests/functional/nestedfs]
	tests = ['nestedfs_001_pos']
	tags = ['functional', 'nestedfs']

	[tests/functional/nopwrite]
	tests = ['nopwrite_sync', 'nopwrite_volume']
	tags = ['functional', 'nopwrite']

	[tests/functional/pool_checkpoint]
	tests = ['checkpoint_conf_change', 'checkpoint_discard_many',
	'checkpoint_removal', 'checkpoint_sm_scale', 'checkpoint_twice']
	tags = ['functional', 'pool_checkpoint']
	timeout = 1800

	[tests/functional/poolversion]
	tests = ['poolversion_001_pos', 'poolversion_002_pos']
	tags = ['functional', 'poolversion']

	[tests/functional/redacted_send]
	tests = ['redacted_compressed', 'redacted_contents', 'redacted_deleted',
	'redacted_disabled_feature', 'redacted_incrementals',
	'redacted_largeblocks', 'redacted_mixed_recsize', 'redacted_negative',
	'redacted_origin', 'redacted_props', 'redacted_resume', 'redacted_size']
	tags = ['functional', 'redacted_send']

	[tests/functional/raidz]
	tests = ['raidz_001_neg']
	tags = ['functional', 'raidz']

	[tests/functional/refquota]
	tests = ['refquota_001_pos', 'refquota_002_pos', 'refquota_003_pos',
	'refquota_004_pos', 'refquota_005_pos', 'refquota_006_neg',
	'refquota_007_neg']
	tags = ['functional', 'refquota']

	[tests/functional/refreserv]
	tests = ['refreserv_001_pos', 'refreserv_002_pos', 'refreserv_003_pos',
	'refreserv_005_pos', 'refreserv_multi_raidz']
	tags = ['functional', 'refreserv']

	[tests/functional/removal]
	pre =
	tests = ['removal_all_vdev', 'removal_sanity', 'removal_with_dedup',
	'removal_with_ganging', 'removal_with_faulted']
	tags = ['functional', 'removal']

	[tests/functional/replacement]
	tests = ['rebuild_raidz']
	tags = ['functional', 'replacement']

	[tests/functional/reservation]
	tests = ['reservation_001_pos', 'reservation_002_pos', 'reservation_003_pos',
	'reservation_004_pos', 'reservation_005_pos', 'reservation_006_pos',
	'reservation_007_pos', 'reservation_008_pos', 'reservation_009_pos',
	'reservation_010_pos', 'reservation_011_pos', 'reservation_012_pos',
	'reservation_014_pos', 'reservation_015_pos',
	'reservation_016_pos', 'reservation_017_pos', 'reservation_018_pos',
	'reservation_019_pos', 'reservation_020_pos', 'reservation_021_neg',
	'reservation_022_pos']
	tags = ['functional', 'reservation']

	[tests/functional/rsend]
	tests = ['recv_dedup', 'recv_dedup_encrypted_zvol', 'rsend_001_pos',
	'rsend_002_pos', 'rsend_003_pos', 'rsend_004_pos', 'rsend_005_pos',
	'rsend_006_pos', 'rsend_009_pos', 'rsend_010_pos', 'rsend_011_pos',
	'rsend_014_pos', 'rsend_016_neg', 'send-c_verify_contents',
	'send-c_volume', 'send-c_zstreamdump', 'send-c_recv_dedup',
	'send-L_toggle', 'send_encrypted_hierarchy', 'send_encrypted_props',
	'send_encrypted_truncated_files', 'send_freeobjects', 'send_holds',
	'send_mixed_raw', 'send-wR_encrypted_zvol', 'send_partial_dataset',
	'send_invalid']
	tags = ['functional', 'rsend']

	[tests/functional/scrub_mirror]
	tests = ['scrub_mirror_001_pos', 'scrub_mirror_002_pos']
	tags = ['functional', 'scrub_mirror']

	[tests/functional/slog]
	tests = ['slog_008_neg', 'slog_009_neg', 'slog_010_neg']
	tags = ['functional', 'slog']

	[tests/functional/snapshot]
	tests = ['clone_001_pos', 'rollback_001_pos', 'rollback_002_pos',
	- 'rollback_003_pos', 'snapshot_001_pos', 'snapshot_002_pos',
	- 'snapshot_003_pos', 'snapshot_004_pos', 'snapshot_005_pos',
	- 'snapshot_006_pos', 'snapshot_007_pos', 'snapshot_008_pos',
	- 'snapshot_009_pos', 'snapshot_010_pos', 'snapshot_011_pos',
	- 'snapshot_012_pos', 'snapshot_013_pos', 'snapshot_014_pos',
	- 'snapshot_017_pos']
	+ 'snapshot_001_pos', 'snapshot_002_pos', 'snapshot_003_pos',
	+ 'snapshot_004_pos', 'snapshot_005_pos', 'snapshot_006_pos',
	+ 'snapshot_007_pos', 'snapshot_008_pos', 'snapshot_009_pos',
	+ 'snapshot_010_pos', 'snapshot_011_pos', 'snapshot_012_pos',
	+ 'snapshot_013_pos', 'snapshot_014_pos', 'snapshot_017_pos']
	tags = ['functional', 'snapshot']

	[tests/functional/snapused]
	tests = ['snapused_002_pos', 'snapused_004_pos', 'snapused_005_pos']
	tags = ['functional', 'snapused']

	[tests/functional/sparse]
	tests = ['sparse_001_pos']
	tags = ['functional', 'sparse']

	[tests/functional/suid]
	tests = ['suid_write_to_suid', 'suid_write_to_sgid', 'suid_write_to_suid_sgid',
	'suid_write_to_none']
	tags = ['functional', 'suid']

	[tests/functional/threadsappend]
	tests = ['threadsappend_001_pos']
	tags = ['functional', 'threadsappend']

	[tests/functional/truncate]
	tests = ['truncate_001_pos', 'truncate_002_pos']
	tags = ['functional', 'truncate']

	[tests/functional/upgrade]
	tests = ['upgrade_userobj_001_pos', 'upgrade_readonly_pool']
	tags = ['functional', 'upgrade']

	[tests/functional/vdev_zaps]
	tests = ['vdev_zaps_001_pos', 'vdev_zaps_003_pos', 'vdev_zaps_004_pos',
	'vdev_zaps_005_pos', 'vdev_zaps_006_pos']
	tags = ['functional', 'vdev_zaps']

	[tests/functional/xattr]
	tests = ['xattr_001_pos', 'xattr_002_neg', 'xattr_003_neg', 'xattr_004_pos',
	'xattr_005_pos', 'xattr_006_pos', 'xattr_007_neg',
	'xattr_011_pos', 'xattr_013_pos']
	tags = ['functional', 'xattr']

	[tests/functional/zvol/zvol_ENOSPC]
	tests = ['zvol_ENOSPC_001_pos']
	tags = ['functional', 'zvol', 'zvol_ENOSPC']

	[tests/functional/zvol/zvol_cli]
	tests = ['zvol_cli_001_pos', 'zvol_cli_002_pos', 'zvol_cli_003_neg']
	tags = ['functional', 'zvol', 'zvol_cli']

	[tests/functional/zvol/zvol_swap]
	tests = ['zvol_swap_001_pos', 'zvol_swap_002_pos']
	tags = ['functional', 'zvol', 'zvol_swap']

	[tests/functional/zpool_influxdb]
	tests = ['zpool_influxdb']
	tags = ['functional', 'zpool_influxdb']
	diff --git a/sys/contrib/openzfs/tests/test-runner/bin/test-runner.py.in b/sys/contrib/openzfs/tests/test-runner/bin/test-runner.py.in
	index bbabf247c1dc..d32e05c45392 100755
	--- a/sys/contrib/openzfs/tests/test-runner/bin/test-runner.py.in
	+++ b/sys/contrib/openzfs/tests/test-runner/bin/test-runner.py.in
	@@ -1,1056 +1,1107 @@
	#!/usr/bin/env @PYTHON_SHEBANG@

	#
	# This file and its contents are supplied under the terms of the
	# Common Development and Distribution License ("CDDL"), version 1.0.
	# You may only use this file in accordance with the terms of version
	# 1.0 of the CDDL.
	#
	# A full copy of the text of the CDDL should have accompanied this
	# source. A copy of the CDDL is also available via the Internet at
	# http://www.illumos.org/license/CDDL.
	#

	#
	# Copyright (c) 2012, 2018 by Delphix. All rights reserved.
	# Copyright (c) 2019 Datto Inc.
	#
	# This script must remain compatible with Python 2.6+ and Python 3.4+.
	#

	# some python 2.7 system don't have a configparser shim
	try:
	import configparser
	except ImportError:
	import ConfigParser as configparser

	import os
	import sys
	import ctypes
	+import re

	from datetime import datetime
	from optparse import OptionParser
	from pwd import getpwnam
	from pwd import getpwuid
	from select import select
	from subprocess import PIPE
	from subprocess import Popen
	from threading import Timer
	from time import time

	BASEDIR = '/var/tmp/test_results'
	TESTDIR = '/usr/share/zfs/'
	KILL = 'kill'
	TRUE = 'true'
	SUDO = 'sudo'
	LOG_FILE = 'LOG_FILE'
	LOG_OUT = 'LOG_OUT'
	LOG_ERR = 'LOG_ERR'
	LOG_FILE_OBJ = None

	# some python 2.7 system don't have a concept of monotonic time
	CLOCK_MONOTONIC_RAW = 4 # see <linux/time.h>


	class timespec(ctypes.Structure):
	_fields_ = [
	('tv_sec', ctypes.c_long),
	('tv_nsec', ctypes.c_long)
	]


	librt = ctypes.CDLL('librt.so.1', use_errno=True)
	clock_gettime = librt.clock_gettime
	clock_gettime.argtypes = [ctypes.c_int, ctypes.POINTER(timespec)]


	def monotonic_time():
	t = timespec()
	if clock_gettime(CLOCK_MONOTONIC_RAW, ctypes.pointer(t)) != 0:
	errno_ = ctypes.get_errno()
	raise OSError(errno_, os.strerror(errno_))
	return t.tv_sec + t.tv_nsec * 1e-9


	class Result(object):
	total = 0
	runresults = {'PASS': 0, 'FAIL': 0, 'SKIP': 0, 'KILLED': 0, 'RERAN': 0}

	def __init__(self):
	self.starttime = None
	self.returncode = None
	self.runtime = ''
	self.stdout = []
	self.stderr = []
	self.result = ''

	def done(self, proc, killed, reran):
	"""
	Finalize the results of this Cmd.
	"""
	Result.total += 1
	m, s = divmod(monotonic_time() - self.starttime, 60)
	self.runtime = '%02d:%02d' % (m, s)
	self.returncode = proc.returncode
	if reran is True:
	Result.runresults['RERAN'] += 1
	if killed:
	self.result = 'KILLED'
	Result.runresults['KILLED'] += 1
	elif self.returncode == 0:
	self.result = 'PASS'
	Result.runresults['PASS'] += 1
	elif self.returncode == 4:
	self.result = 'SKIP'
	Result.runresults['SKIP'] += 1
	elif self.returncode != 0:
	self.result = 'FAIL'
	Result.runresults['FAIL'] += 1


	class Output(object):
	"""
	This class is a slightly modified version of the 'Stream' class found
	here: http://goo.gl/aSGfv
	"""
	def __init__(self, stream):
	self.stream = stream
	self._buf = b''
	self.lines = []

	def fileno(self):
	return self.stream.fileno()

	def read(self, drain=0):
	"""
	Read from the file descriptor. If 'drain' set, read until EOF.
	"""
	while self._read() is not None:
	if not drain:
	break

	def _read(self):
	"""
	Read up to 4k of data from this output stream. Collect the output
	up to the last newline, and append it to any leftover data from a
	previous call. The lines are stored as a (timestamp, data) tuple
	for easy sorting/merging later.
	"""
	fd = self.fileno()
	buf = os.read(fd, 4096)
	if not buf:
	return None
	if b'\n' not in buf:
	self._buf += buf
	return []

	buf = self._buf + buf
	tmp, rest = buf.rsplit(b'\n', 1)
	self._buf = rest
	now = datetime.now()
	rows = tmp.split(b'\n')
	self.lines += [(now, r) for r in rows]


	class Cmd(object):
	verified_users = []

	def __init__(self, pathname, identifier=None, outputdir=None,
	timeout=None, user=None, tags=None):
	self.pathname = pathname
	self.identifier = identifier
	self.outputdir = outputdir or 'BASEDIR'
	"""
	The timeout for tests is measured in wall-clock time
	"""
	self.timeout = timeout
	self.user = user or ''
	self.killed = False
	self.reran = None
	self.result = Result()

	if self.timeout is None:
	self.timeout = 60

	def __str__(self):
	return '''\
	Pathname: %s
	Identifier: %s
	Outputdir: %s
	Timeout: %d
	User: %s
	''' % (self.pathname, self.identifier, self.outputdir, self.timeout, self.user)

	def kill_cmd(self, proc, keyboard_interrupt=False):
	"""
	Kill a running command due to timeout, or ^C from the keyboard. If
	sudo is required, this user was verified previously.
	"""
	self.killed = True
	do_sudo = len(self.user) != 0
	signal = '-TERM'

	cmd = [SUDO, KILL, signal, str(proc.pid)]
	if not do_sudo:
	del cmd[0]

	try:
	kp = Popen(cmd)
	kp.wait()
	except Exception:
	pass

	"""
	If this is not a user-initiated kill and the test has not been
	reran before we consider if the test needs to be reran:
	If the test has spent some time hibernating and didn't run the whole
	length of time before being timed out we will rerun the test.
	"""
	if keyboard_interrupt is False and self.reran is None:
	runtime = monotonic_time() - self.result.starttime
	if int(self.timeout) > runtime:
	self.killed = False
	self.reran = False
	self.run(False)
	self.reran = True

	def update_cmd_privs(self, cmd, user):
	"""
	If a user has been specified to run this Cmd and we're not already
	running as that user, prepend the appropriate sudo command to run
	as that user.
	"""
	me = getpwuid(os.getuid())

	if not user or user is me:
	if os.path.isfile(cmd+'.ksh') and os.access(cmd+'.ksh', os.X_OK):
	cmd += '.ksh'
	if os.path.isfile(cmd+'.sh') and os.access(cmd+'.sh', os.X_OK):
	cmd += '.sh'
	return cmd

	if not os.path.isfile(cmd):
	if os.path.isfile(cmd+'.ksh') and os.access(cmd+'.ksh', os.X_OK):
	cmd += '.ksh'
	if os.path.isfile(cmd+'.sh') and os.access(cmd+'.sh', os.X_OK):
	cmd += '.sh'

	ret = '%s -E -u %s %s' % (SUDO, user, cmd)
	return ret.split(' ')

	def collect_output(self, proc):
	"""
	Read from stdout/stderr as data becomes available, until the
	process is no longer running. Return the lines from the stdout and
	stderr Output objects.
	"""
	out = Output(proc.stdout)
	err = Output(proc.stderr)
	res = []
	while proc.returncode is None:
	proc.poll()
	res = select([out, err], [], [], .1)
	for fd in res[0]:
	fd.read()
	for fd in res[0]:
	fd.read(drain=1)

	return out.lines, err.lines

	def run(self, dryrun):
	"""
	This is the main function that runs each individual test.
	Determine whether or not the command requires sudo, and modify it
	if needed. Run the command, and update the result object.
	"""
	if dryrun is True:
	print(self)
	return

	privcmd = self.update_cmd_privs(self.pathname, self.user)
	try:
	old = os.umask(0)
	if not os.path.isdir(self.outputdir):
	os.makedirs(self.outputdir, mode=0o777)
	os.umask(old)
	except OSError as e:
	fail('%s' % e)

	self.result.starttime = monotonic_time()
	proc = Popen(privcmd, stdout=PIPE, stderr=PIPE)
	# Allow a special timeout value of 0 to mean infinity
	if int(self.timeout) == 0:
	self.timeout = sys.maxsize
	t = Timer(int(self.timeout), self.kill_cmd, [proc])

	try:
	t.start()
	self.result.stdout, self.result.stderr = self.collect_output(proc)
	except KeyboardInterrupt:
	self.kill_cmd(proc, True)
	fail('\nRun terminated at user request.')
	finally:
	t.cancel()

	if self.reran is not False:
	self.result.done(proc, self.killed, self.reran)

	def skip(self):
	"""
	Initialize enough of the test result that we can log a skipped
	command.
	"""
	Result.total += 1
	Result.runresults['SKIP'] += 1
	self.result.stdout = self.result.stderr = []
	self.result.starttime = monotonic_time()
	m, s = divmod(monotonic_time() - self.result.starttime, 60)
	self.result.runtime = '%02d:%02d' % (m, s)
	self.result.result = 'SKIP'

	def log(self, options, suppress_console=False):
	"""
	This function is responsible for writing all output. This includes
	the console output, the logfile of all results (with timestamped
	merged stdout and stderr), and for each test, the unmodified
	stdout/stderr/merged in its own file.
	"""

	logname = getpwuid(os.getuid()).pw_name
	rer = ''
	if self.reran is True:
	rer = ' (RERAN)'
	user = ' (run as %s)' % (self.user if len(self.user) else logname)
	if self.identifier:
	msga = 'Test (%s): %s%s ' % (self.identifier, self.pathname, user)
	else:
	msga = 'Test: %s%s ' % (self.pathname, user)
	msgb = '[%s] [%s]%s\n' % (self.result.runtime, self.result.result, rer)
	pad = ' ' * (80 - (len(msga) + len(msgb)))
	result_line = msga + pad + msgb

	# The result line is always written to the log file. If -q was
	# specified only failures are written to the console, otherwise
	# the result line is written to the console. The console output
	# may be suppressed by calling log() with suppress_console=True.
	write_log(bytearray(result_line, encoding='utf-8'), LOG_FILE)
	if not suppress_console:
	if not options.quiet:
	write_log(result_line, LOG_OUT)
	elif options.quiet and self.result.result != 'PASS':
	write_log(result_line, LOG_OUT)

	lines = sorted(self.result.stdout + self.result.stderr,
	key=lambda x: x[0])

	# Write timestamped output (stdout and stderr) to the logfile
	for dt, line in lines:
	timestamp = bytearray(dt.strftime("%H:%M:%S.%f ")[:11],
	encoding='utf-8')
	write_log(b'%s %s\n' % (timestamp, line), LOG_FILE)

	# Write the separate stdout/stderr/merged files, if the data exists
	if len(self.result.stdout):
	with open(os.path.join(self.outputdir, 'stdout'), 'wb') as out:
	for _, line in self.result.stdout:
	os.write(out.fileno(), b'%s\n' % line)
	if len(self.result.stderr):
	with open(os.path.join(self.outputdir, 'stderr'), 'wb') as err:
	for _, line in self.result.stderr:
	os.write(err.fileno(), b'%s\n' % line)
	if len(self.result.stdout) and len(self.result.stderr):
	with open(os.path.join(self.outputdir, 'merged'), 'wb') as merged:
	for _, line in lines:
	os.write(merged.fileno(), b'%s\n' % line)


	class Test(Cmd):
	props = ['outputdir', 'timeout', 'user', 'pre', 'pre_user', 'post',
	'post_user', 'failsafe', 'failsafe_user', 'tags']

	def __init__(self, pathname,
	pre=None, pre_user=None, post=None, post_user=None,
	failsafe=None, failsafe_user=None, tags=None, **kwargs):
	super(Test, self).__init__(pathname, **kwargs)
	self.pre = pre or ''
	self.pre_user = pre_user or ''
	self.post = post or ''
	self.post_user = post_user or ''
	self.failsafe = failsafe or ''
	self.failsafe_user = failsafe_user or ''
	self.tags = tags or []

	def __str__(self):
	post_user = pre_user = failsafe_user = ''
	if len(self.pre_user):
	pre_user = ' (as %s)' % (self.pre_user)
	if len(self.post_user):
	post_user = ' (as %s)' % (self.post_user)
	if len(self.failsafe_user):
	failsafe_user = ' (as %s)' % (self.failsafe_user)
	return '''\
	Pathname: %s
	Identifier: %s
	Outputdir: %s
	Timeout: %d
	User: %s
	Pre: %s%s
	Post: %s%s
	Failsafe: %s%s
	Tags: %s
	''' % (self.pathname, self.identifier, self.outputdir, self.timeout, self.user,
	self.pre, pre_user, self.post, post_user, self.failsafe,
	failsafe_user, self.tags)

	def verify(self):
	"""
	Check the pre/post/failsafe scripts, user and Test. Omit the Test from
	this run if there are any problems.
	"""
	files = [self.pre, self.pathname, self.post, self.failsafe]
	users = [self.pre_user, self.user, self.post_user, self.failsafe_user]

	for f in [f for f in files if len(f)]:
	if not verify_file(f):
	write_log("Warning: Test '%s' not added to this run because"
	" it failed verification.\n" % f, LOG_ERR)
	return False

	for user in [user for user in users if len(user)]:
	if not verify_user(user):
	write_log("Not adding Test '%s' to this run.\n" %
	self.pathname, LOG_ERR)
	return False

	return True

	def run(self, options):
	"""
	Create Cmd instances for the pre/post/failsafe scripts. If the pre
	script doesn't pass, skip this Test. Run the post script regardless.
	If the Test is killed, also run the failsafe script.
	"""
	odir = os.path.join(self.outputdir, os.path.basename(self.pre))
	pretest = Cmd(self.pre, identifier=self.identifier, outputdir=odir,
	timeout=self.timeout, user=self.pre_user)
	test = Cmd(self.pathname, identifier=self.identifier,
	outputdir=self.outputdir, timeout=self.timeout,
	user=self.user)
	odir = os.path.join(self.outputdir, os.path.basename(self.failsafe))
	failsafe = Cmd(self.failsafe, identifier=self.identifier,
	outputdir=odir, timeout=self.timeout,
	user=self.failsafe_user)
	odir = os.path.join(self.outputdir, os.path.basename(self.post))
	posttest = Cmd(self.post, identifier=self.identifier, outputdir=odir,
	timeout=self.timeout, user=self.post_user)

	cont = True
	if len(pretest.pathname):
	pretest.run(options.dryrun)
	cont = pretest.result.result == 'PASS'
	pretest.log(options)

	if cont:
	test.run(options.dryrun)
	if test.result.result == 'KILLED' and len(failsafe.pathname):
	failsafe.run(options.dryrun)
	failsafe.log(options, suppress_console=True)
	else:
	test.skip()

	test.log(options)

	if len(posttest.pathname):
	posttest.run(options.dryrun)
	posttest.log(options)


	class TestGroup(Test):
	props = Test.props + ['tests']

	def __init__(self, pathname, tests=None, **kwargs):
	super(TestGroup, self).__init__(pathname, **kwargs)
	self.tests = tests or []

	def __str__(self):
	post_user = pre_user = failsafe_user = ''
	if len(self.pre_user):
	pre_user = ' (as %s)' % (self.pre_user)
	if len(self.post_user):
	post_user = ' (as %s)' % (self.post_user)
	if len(self.failsafe_user):
	failsafe_user = ' (as %s)' % (self.failsafe_user)
	return '''\
	Pathname: %s
	Identifier: %s
	Outputdir: %s
	Tests: %s
	Timeout: %s
	User: %s
	Pre: %s%s
	Post: %s%s
	Failsafe: %s%s
	Tags: %s
	''' % (self.pathname, self.identifier, self.outputdir, self.tests,
	self.timeout, self.user, self.pre, pre_user, self.post, post_user,
	self.failsafe, failsafe_user, self.tags)

	+ def filter(self, keeplist):
	+ self.tests = [x for x in self.tests if x in keeplist]
	+
	def verify(self):
	"""
	Check the pre/post/failsafe scripts, user and tests in this TestGroup.
	Omit the TestGroup entirely, or simply delete the relevant tests in the
	group, if that's all that's required.
	"""
	# If the pre/post/failsafe scripts are relative pathnames, convert to
	# absolute, so they stand a chance of passing verification.
	if len(self.pre) and not os.path.isabs(self.pre):
	self.pre = os.path.join(self.pathname, self.pre)
	if len(self.post) and not os.path.isabs(self.post):
	self.post = os.path.join(self.pathname, self.post)
	if len(self.failsafe) and not os.path.isabs(self.failsafe):
	self.post = os.path.join(self.pathname, self.post)

	auxfiles = [self.pre, self.post, self.failsafe]
	users = [self.pre_user, self.user, self.post_user, self.failsafe_user]

	for f in [f for f in auxfiles if len(f)]:
	if f != self.failsafe and self.pathname != os.path.dirname(f):
	write_log("Warning: TestGroup '%s' not added to this run. "
	"Auxiliary script '%s' exists in a different "
	"directory.\n" % (self.pathname, f), LOG_ERR)
	return False

	if not verify_file(f):
	write_log("Warning: TestGroup '%s' not added to this run. "
	"Auxiliary script '%s' failed verification.\n" %
	(self.pathname, f), LOG_ERR)
	return False

	for user in [user for user in users if len(user)]:
	if not verify_user(user):
	write_log("Not adding TestGroup '%s' to this run.\n" %
	self.pathname, LOG_ERR)
	return False

	# If one of the tests is invalid, delete it, log it, and drive on.
	for test in self.tests:
	if not verify_file(os.path.join(self.pathname, test)):
	del self.tests[self.tests.index(test)]
	write_log("Warning: Test '%s' removed from TestGroup '%s' "
	"because it failed verification.\n" %
	(test, self.pathname), LOG_ERR)

	return len(self.tests) != 0

	def run(self, options):
	"""
	Create Cmd instances for the pre/post/failsafe scripts. If the pre
	script doesn't pass, skip all the tests in this TestGroup. Run the
	post script regardless. Run the failsafe script when a test is killed.
	"""
	# tags assigned to this test group also include the test names
	if options.tags and not set(self.tags).intersection(set(options.tags)):
	return

	odir = os.path.join(self.outputdir, os.path.basename(self.pre))
	pretest = Cmd(self.pre, outputdir=odir, timeout=self.timeout,
	user=self.pre_user, identifier=self.identifier)
	odir = os.path.join(self.outputdir, os.path.basename(self.post))
	posttest = Cmd(self.post, outputdir=odir, timeout=self.timeout,
	user=self.post_user, identifier=self.identifier)

	cont = True
	if len(pretest.pathname):
	pretest.run(options.dryrun)
	cont = pretest.result.result == 'PASS'
	pretest.log(options)

	for fname in self.tests:
	odir = os.path.join(self.outputdir, fname)
	test = Cmd(os.path.join(self.pathname, fname), outputdir=odir,
	timeout=self.timeout, user=self.user,
	identifier=self.identifier)
	odir = os.path.join(odir, os.path.basename(self.failsafe))
	failsafe = Cmd(self.failsafe, outputdir=odir, timeout=self.timeout,
	user=self.failsafe_user, identifier=self.identifier)
	if cont:
	test.run(options.dryrun)
	if test.result.result == 'KILLED' and len(failsafe.pathname):
	failsafe.run(options.dryrun)
	failsafe.log(options, suppress_console=True)
	else:
	test.skip()

	test.log(options)

	if len(posttest.pathname):
	posttest.run(options.dryrun)
	posttest.log(options)


	class TestRun(object):
	props = ['quiet', 'outputdir']

	def __init__(self, options):
	self.tests = {}
	self.testgroups = {}
	self.starttime = time()
	self.timestamp = datetime.now().strftime('%Y%m%dT%H%M%S')
	self.outputdir = os.path.join(options.outputdir, self.timestamp)
	self.setup_logging(options)
	self.defaults = [
	('outputdir', BASEDIR),
	('quiet', False),
	('timeout', 60),
	('user', ''),
	('pre', ''),
	('pre_user', ''),
	('post', ''),
	('post_user', ''),
	('failsafe', ''),
	('failsafe_user', ''),
	('tags', [])
	]

	def __str__(self):
	s = 'TestRun:\n outputdir: %s\n' % self.outputdir
	s += 'TESTS:\n'
	for key in sorted(self.tests.keys()):
	s += '%s%s' % (self.tests[key].__str__(), '\n')
	s += 'TESTGROUPS:\n'
	for key in sorted(self.testgroups.keys()):
	s += '%s%s' % (self.testgroups[key].__str__(), '\n')
	return s

	def addtest(self, pathname, options):
	"""
	Create a new Test, and apply any properties that were passed in
	from the command line. If it passes verification, add it to the
	TestRun.
	"""
	test = Test(pathname)
	for prop in Test.props:
	setattr(test, prop, getattr(options, prop))

	if test.verify():
	self.tests[pathname] = test

	def addtestgroup(self, dirname, filenames, options):
	"""
	Create a new TestGroup, and apply any properties that were passed
	in from the command line. If it passes verification, add it to the
	TestRun.
	"""
	if dirname not in self.testgroups:
	testgroup = TestGroup(dirname)
	for prop in Test.props:
	setattr(testgroup, prop, getattr(options, prop))

	# Prevent pre/post/failsafe scripts from running as regular tests
	for f in [testgroup.pre, testgroup.post, testgroup.failsafe]:
	if f in filenames:
	del filenames[filenames.index(f)]

	self.testgroups[dirname] = testgroup
	self.testgroups[dirname].tests = sorted(filenames)

	testgroup.verify()

	+ def filter(self, keeplist):
	+ for group in list(self.testgroups.keys()):
	+ if group not in keeplist:
	+ del self.testgroups[group]
	+ continue
	+
	+ g = self.testgroups[group]
	+
	+ if g.pre and os.path.basename(g.pre) in keeplist[group]:
	+ continue
	+
	+ g.filter(keeplist[group])
	+
	+ for test in list(self.tests.keys()):
	+ directory, base = os.path.split(test)
	+ if directory not in keeplist or base not in keeplist[directory]:
	+ del self.tests[test]
	+
	def read(self, options):
	"""
	Read in the specified runfiles, and apply the TestRun properties
	listed in the 'DEFAULT' section to our TestRun. Then read each
	section, and apply the appropriate properties to the Test or
	TestGroup. Properties from individual sections override those set
	in the 'DEFAULT' section. If the Test or TestGroup passes
	verification, add it to the TestRun.
	"""
	config = configparser.RawConfigParser()
	parsed = config.read(options.runfiles)
	failed = options.runfiles - set(parsed)
	if len(failed):
	files = ' '.join(sorted(failed))
	fail("Couldn't read config files: %s" % files)

	for opt in TestRun.props:
	if config.has_option('DEFAULT', opt):
	setattr(self, opt, config.get('DEFAULT', opt))
	self.outputdir = os.path.join(self.outputdir, self.timestamp)

	testdir = options.testdir

	for section in config.sections():
	if 'tests' in config.options(section):
	parts = section.split(':', 1)
	sectiondir = parts[0]
	identifier = parts[1] if len(parts) == 2 else None
	if os.path.isdir(sectiondir):
	pathname = sectiondir
	elif os.path.isdir(os.path.join(testdir, sectiondir)):
	pathname = os.path.join(testdir, sectiondir)
	else:
	pathname = sectiondir

	testgroup = TestGroup(os.path.abspath(pathname),
	identifier=identifier)
	for prop in TestGroup.props:
	for sect in ['DEFAULT', section]:
	if config.has_option(sect, prop):
	if prop == 'tags':
	setattr(testgroup, prop,
	eval(config.get(sect, prop)))
	elif prop == 'failsafe':
	failsafe = config.get(sect, prop)
	setattr(testgroup, prop,
	os.path.join(testdir, failsafe))
	else:
	setattr(testgroup, prop,
	config.get(sect, prop))

	# Repopulate tests using eval to convert the string to a list
	testgroup.tests = eval(config.get(section, 'tests'))

	if testgroup.verify():
	self.testgroups[section] = testgroup
	else:
	test = Test(section)
	for prop in Test.props:
	for sect in ['DEFAULT', section]:
	if config.has_option(sect, prop):
	if prop == 'failsafe':
	failsafe = config.get(sect, prop)
	setattr(test, prop,
	os.path.join(testdir, failsafe))
	else:
	setattr(test, prop, config.get(sect, prop))

	if test.verify():
	self.tests[section] = test

	def write(self, options):
	"""
	Create a configuration file for editing and later use. The
	'DEFAULT' section of the config file is created from the
	properties that were specified on the command line. Tests are
	simply added as sections that inherit everything from the
	'DEFAULT' section. TestGroups are the same, except they get an
	option including all the tests to run in that directory.
	"""

	defaults = dict([(prop, getattr(options, prop)) for prop, _ in
	self.defaults])
	config = configparser.RawConfigParser(defaults)

	for test in sorted(self.tests.keys()):
	config.add_section(test)
	+ for prop in Test.props:
	+ if prop not in self.props:
	+ config.set(test, prop,
	+ getattr(self.tests[test], prop))

	for testgroup in sorted(self.testgroups.keys()):
	config.add_section(testgroup)
	config.set(testgroup, 'tests', self.testgroups[testgroup].tests)
	+ for prop in TestGroup.props:
	+ if prop not in self.props:
	+ config.set(testgroup, prop,
	+ getattr(self.testgroups[testgroup], prop))

	try:
	with open(options.template, 'w') as f:
	return config.write(f)
	except IOError:
	fail('Could not open \'%s\' for writing.' % options.template)

	def complete_outputdirs(self):
	"""
	Collect all the pathnames for Tests, and TestGroups. Work
	backwards one pathname component at a time, to create a unique
	directory name in which to deposit test output. Tests will be able
	to write output files directly in the newly modified outputdir.
	TestGroups will be able to create one subdirectory per test in the
	outputdir, and are guaranteed uniqueness because a group can only
	contain files in one directory. Pre and post tests will create a
	directory rooted at the outputdir of the Test or TestGroup in
	question for their output. Failsafe scripts will create a directory
	rooted at the outputdir of each Test for their output.
	"""
	done = False
	components = 0
	tmp_dict = dict(list(self.tests.items()) +
	list(self.testgroups.items()))
	total = len(tmp_dict)
	base = self.outputdir

	while not done:
	paths = []
	components -= 1
	for testfile in list(tmp_dict.keys()):
	uniq = '/'.join(testfile.split('/')[components:]).lstrip('/')
	if uniq not in paths:
	paths.append(uniq)
	tmp_dict[testfile].outputdir = os.path.join(base, uniq)
	else:
	break
	done = total == len(paths)

	def setup_logging(self, options):
	"""
	This function creates the output directory and gets a file object
	for the logfile. This function must be called before write_log()
	can be used.
	"""
	if options.dryrun is True:
	return

	global LOG_FILE_OBJ
	- if options.cmd != 'wrconfig':
	+ if not options.template:
	try:
	old = os.umask(0)
	os.makedirs(self.outputdir, mode=0o777)
	os.umask(old)
	filename = os.path.join(self.outputdir, 'log')
	LOG_FILE_OBJ = open(filename, buffering=0, mode='wb')
	except OSError as e:
	fail('%s' % e)

	def run(self, options):
	"""
	Walk through all the Tests and TestGroups, calling run().
	"""
	try:
	os.chdir(self.outputdir)
	except OSError:
	fail('Could not change to directory %s' % self.outputdir)
	# make a symlink to the output for the currently running test
	logsymlink = os.path.join(self.outputdir, '../current')
	if os.path.islink(logsymlink):
	os.unlink(logsymlink)
	if not os.path.exists(logsymlink):
	os.symlink(self.outputdir, logsymlink)
	else:
	write_log('Could not make a symlink to directory %s\n' %
	self.outputdir, LOG_ERR)
	iteration = 0
	while iteration < options.iterations:
	for test in sorted(self.tests.keys()):
	self.tests[test].run(options)
	for testgroup in sorted(self.testgroups.keys()):
	self.testgroups[testgroup].run(options)
	iteration += 1

	def summary(self):
	if Result.total == 0:
	return 2

	print('\nResults Summary')
	for key in list(Result.runresults.keys()):
	if Result.runresults[key] != 0:
	print('%s\t% 4d' % (key, Result.runresults[key]))

	m, s = divmod(time() - self.starttime, 60)
	h, m = divmod(m, 60)
	print('\nRunning Time:\t%02d:%02d:%02d' % (h, m, s))
	print('Percent passed:\t%.1f%%' % ((float(Result.runresults['PASS']) /
	float(Result.total)) * 100))
	print('Log directory:\t%s' % self.outputdir)

	if Result.runresults['FAIL'] > 0:
	return 1

	if Result.runresults['KILLED'] > 0:
	return 1

	if Result.runresults['RERAN'] > 0:
	return 3

	return 0


	def write_log(msg, target):
	"""
	Write the provided message to standard out, standard error or
	the logfile. If specifying LOG_FILE, then `msg` must be a bytes
	like object. This way we can still handle output from tests that
	may be in unexpected encodings.
	"""
	if target == LOG_OUT:
	os.write(sys.stdout.fileno(), bytearray(msg, encoding='utf-8'))
	elif target == LOG_ERR:
	os.write(sys.stderr.fileno(), bytearray(msg, encoding='utf-8'))
	elif target == LOG_FILE:
	os.write(LOG_FILE_OBJ.fileno(), msg)
	else:
	fail('log_msg called with unknown target "%s"' % target)


	def verify_file(pathname):
	"""
	Verify that the supplied pathname is an executable regular file.
	"""
	if os.path.isdir(pathname) or os.path.islink(pathname):
	return False

	for ext in '', '.ksh', '.sh':
	script_path = pathname + ext
	if os.path.isfile(script_path) and os.access(script_path, os.X_OK):
	return True

	return False


	def verify_user(user):
	"""
	Verify that the specified user exists on this system, and can execute
	sudo without being prompted for a password.
	"""
	testcmd = [SUDO, '-n', '-u', user, TRUE]

	if user in Cmd.verified_users:
	return True

	try:
	getpwnam(user)
	except KeyError:
	write_log("Warning: user '%s' does not exist.\n" % user,
	LOG_ERR)
	return False

	p = Popen(testcmd)
	p.wait()
	if p.returncode != 0:
	write_log("Warning: user '%s' cannot use passwordless sudo.\n" % user,
	LOG_ERR)
	return False
	else:
	Cmd.verified_users.append(user)

	return True


	def find_tests(testrun, options):
	"""
	For the given list of pathnames, add files as Tests. For directories,
	if do_groups is True, add the directory as a TestGroup. If False,
	recursively search for executable files.
	"""

	for p in sorted(options.pathnames):
	if os.path.isdir(p):
	for dirname, _, filenames in os.walk(p):
	if options.do_groups:
	testrun.addtestgroup(dirname, filenames, options)
	else:
	for f in sorted(filenames):
	testrun.addtest(os.path.join(dirname, f), options)
	else:
	testrun.addtest(p, options)


	+def filter_tests(testrun, options):
	+ try:
	+ fh = open(options.logfile, "r")
	+ except Exception as e:
	+ fail('%s' % e)
	+
	+ failed = {}
	+ while True:
	+ line = fh.readline()
	+ if not line:
	+ break
	+ m = re.match(r'Test: .(tests/.)/(\S+).*\[FAIL\]', line)
	+ if not m:
	+ continue
	+ group, test = m.group(1, 2)
	+ try:
	+ failed[group].append(test)
	+ except KeyError:
	+ failed[group] = [test]
	+ fh.close()
	+
	+ testrun.filter(failed)
	+
	+
	def fail(retstr, ret=1):
	print('%s: %s' % (sys.argv[0], retstr))
	exit(ret)


	def options_cb(option, opt_str, value, parser):
	- path_options = ['outputdir', 'template', 'testdir']
	-
	- if option.dest == 'runfiles' and '-w' in parser.rargs or \
	- option.dest == 'template' and '-c' in parser.rargs:
	- fail('-c and -w are mutually exclusive.')
	+ path_options = ['outputdir', 'template', 'testdir', 'logfile']

	if opt_str in parser.rargs:
	fail('%s may only be specified once.' % opt_str)

	if option.dest == 'runfiles':
	parser.values.cmd = 'rdconfig'
	value = set(os.path.abspath(p) for p in value.split(','))
	- if option.dest == 'template':
	- parser.values.cmd = 'wrconfig'
	if option.dest == 'tags':
	value = [x.strip() for x in value.split(',')]

	if option.dest in path_options:
	setattr(parser.values, option.dest, os.path.abspath(value))
	else:
	setattr(parser.values, option.dest, value)


	def parse_args():
	parser = OptionParser()
	parser.add_option('-c', action='callback', callback=options_cb,
	type='string', dest='runfiles', metavar='runfiles',
	help='Specify tests to run via config files.')
	parser.add_option('-d', action='store_true', default=False, dest='dryrun',
	help='Dry run. Print tests, but take no other action.')
	+ parser.add_option('-l', action='callback', callback=options_cb,
	+ default=None, dest='logfile', metavar='logfile',
	+ type='string',
	+ help='Read logfile and re-run tests which failed.')
	parser.add_option('-g', action='store_true', default=False,
	dest='do_groups', help='Make directories TestGroups.')
	parser.add_option('-o', action='callback', callback=options_cb,
	default=BASEDIR, dest='outputdir', type='string',
	metavar='outputdir', help='Specify an output directory.')
	parser.add_option('-i', action='callback', callback=options_cb,
	default=TESTDIR, dest='testdir', type='string',
	metavar='testdir', help='Specify a test directory.')
	parser.add_option('-p', action='callback', callback=options_cb,
	default='', dest='pre', metavar='script',
	type='string', help='Specify a pre script.')
	parser.add_option('-P', action='callback', callback=options_cb,
	default='', dest='post', metavar='script',
	type='string', help='Specify a post script.')
	parser.add_option('-q', action='store_true', default=False, dest='quiet',
	help='Silence on the console during a test run.')
	parser.add_option('-s', action='callback', callback=options_cb,
	default='', dest='failsafe', metavar='script',
	type='string', help='Specify a failsafe script.')
	parser.add_option('-S', action='callback', callback=options_cb,
	default='', dest='failsafe_user',
	metavar='failsafe_user', type='string',
	help='Specify a user to execute the failsafe script.')
	parser.add_option('-t', action='callback', callback=options_cb, default=60,
	dest='timeout', metavar='seconds', type='int',
	help='Timeout (in seconds) for an individual test.')
	parser.add_option('-u', action='callback', callback=options_cb,
	default='', dest='user', metavar='user', type='string',
	help='Specify a different user name to run as.')
	parser.add_option('-w', action='callback', callback=options_cb,
	default=None, dest='template', metavar='template',
	type='string', help='Create a new config file.')
	parser.add_option('-x', action='callback', callback=options_cb, default='',
	dest='pre_user', metavar='pre_user', type='string',
	help='Specify a user to execute the pre script.')
	parser.add_option('-X', action='callback', callback=options_cb, default='',
	dest='post_user', metavar='post_user', type='string',
	help='Specify a user to execute the post script.')
	parser.add_option('-T', action='callback', callback=options_cb, default='',
	dest='tags', metavar='tags', type='string',
	help='Specify tags to execute specific test groups.')
	parser.add_option('-I', action='callback', callback=options_cb, default=1,
	dest='iterations', metavar='iterations', type='int',
	help='Number of times to run the test run.')
	(options, pathnames) = parser.parse_args()

	- if not options.runfiles and not options.template:
	- options.cmd = 'runtests'
	-
	if options.runfiles and len(pathnames):
	fail('Extraneous arguments.')

	options.pathnames = [os.path.abspath(path) for path in pathnames]

	return options


	def main():
	options = parse_args()
	+
	testrun = TestRun(options)

	- if options.cmd == 'runtests':
	- find_tests(testrun, options)
	- elif options.cmd == 'rdconfig':
	+ if options.runfiles:
	testrun.read(options)
	- elif options.cmd == 'wrconfig':
	+ else:
	find_tests(testrun, options)
	+
	+ if options.logfile:
	+ filter_tests(testrun, options)
	+
	+ if options.template:
	testrun.write(options)
	exit(0)
	- else:
	- fail('Unknown command specified')

	testrun.complete_outputdirs()
	testrun.run(options)
	exit(testrun.summary())


	if __name__ == '__main__':
	main()
	diff --git a/sys/contrib/openzfs/tests/test-runner/bin/zts-report.py.in b/sys/contrib/openzfs/tests/test-runner/bin/zts-report.py.in
	index d04663705309..e14472e2f4f5 100755
	--- a/sys/contrib/openzfs/tests/test-runner/bin/zts-report.py.in
	+++ b/sys/contrib/openzfs/tests/test-runner/bin/zts-report.py.in
	@@ -1,459 +1,490 @@
	#!/usr/bin/env @PYTHON_SHEBANG@

	#
	# This file and its contents are supplied under the terms of the
	# Common Development and Distribution License ("CDDL"), version 1.0.
	# You may only use this file in accordance with the terms of version
	# 1.0 of the CDDL.
	#
	# A full copy of the text of the CDDL should have accompanied this
	# source. A copy of the CDDL is also available via the Internet at
	# http://www.illumos.org/license/CDDL.
	#

	#
	# Copyright (c) 2017 by Delphix. All rights reserved.
	# Copyright (c) 2018 by Lawrence Livermore National Security, LLC.
	#
	# This script must remain compatible with Python 2.6+ and Python 3.4+.
	#

	import os
	import re
	import sys
	+import argparse

	#
	# This script parses the stdout of zfstest, which has this format:
	#
	# Test: /path/to/testa (run as root) [00:00] [PASS]
	# Test: /path/to/testb (run as jkennedy) [00:00] [PASS]
	# Test: /path/to/testc (run as root) [00:00] [FAIL]
	# [...many more results...]
	#
	# Results Summary
	# FAIL 22
	# SKIP 32
	# PASS 1156
	#
	# Running Time: 02:50:31
	# Percent passed: 95.5%
	# Log directory: /var/tmp/test_results/20180615T205926
	#

	#
	# Common generic reasons for a test or test group to be skipped.
	#
	# Some test cases are known to fail in ways which are not harmful or dangerous.
	# In these cases simply mark the test as a known failure until it can be
	# updated and the issue resolved. Note that it's preferable to open a unique
	# issue on the GitHub issue tracker for each test case failure.
	#
	known_reason = 'Known issue'

	#
	# Some tests require that a test user be able to execute the zfs utilities.
	# This may not be possible when testing in-tree due to the default permissions
	# on the user's home directory. When testing this can be resolved by granting
	# group read access.
	#
	# chmod 0750 $HOME
	#
	exec_reason = 'Test user execute permissions required for utilities'

	#
	# Some tests require a minimum python version of 3.5 and will be skipped when
	# the default system version is too old. There may also be tests which require
	# additional python modules be installed, for example python-cffi is required
	# by the pyzfs tests.
	#
	python_reason = 'Python v3.5 or newer required'
	python_deps_reason = 'Python modules missing: python-cffi'

	#
	# Some tests require the O_TMPFILE flag which was first introduced in the
	# 3.11 kernel.
	#
	tmpfile_reason = 'Kernel O_TMPFILE support required'

	#
	# Some tests require the statx(2) system call on Linux which was first
	# introduced in the 4.11 kernel.
	#
	statx_reason = 'Kernel statx(2) system call required on Linux'

	#
	# Some tests require that the NFS client and server utilities be installed.
	#
	share_reason = 'NFS client and server utilities required'

	#
	# Some tests require that the lsattr utility support the project id feature.
	#
	project_id_reason = 'lsattr with set/show project ID required'

	#
	# Some tests require that the kernel support user namespaces.
	#
	user_ns_reason = 'Kernel user namespace support required'

	#
	# Some rewind tests can fail since nothing guarantees that old MOS blocks
	# are not overwritten. Snapshots protect datasets and data files but not
	# the MOS. Reasonable efforts are made in the test case to increase the
	# odds that some txgs will have their MOS data left untouched, but it is
	# never a sure thing.
	#
	rewind_reason = 'Arbitrary pool rewind is not guaranteed'

	#
	# Some tests may by structured in a way that relies on exact knowledge
	# of how much free space in available in a pool. These tests cannot be
	# made completely reliable because the internal details of how free space
	# is managed are not exposed to user space.
	#
	enospc_reason = 'Exact free space reporting is not guaranteed'

	#
	# Some tests require a minimum version of the fio benchmark utility.
	# Older distributions such as CentOS 6.x only provide fio-2.0.13.
	#
	fio_reason = 'Fio v2.3 or newer required'

	#
	# Some tests require that the DISKS provided support the discard operation.
	# Normally this is not an issue because loop back devices are used for DISKS
	# and they support discard (TRIM/UNMAP).
	#
	trim_reason = 'DISKS must support discard (TRIM/UNMAP)'

	#
	# Some tests on FreeBSD require the fspacectl(2) system call and the
	# truncate(1) utility supporting the -d option. The system call was first
	# introduced in FreeBSD version 1400032.
	#
	fspacectl_reason = 'fspacectl(2) and truncate -d support required'

	#
	# Some tests are not applicable to a platform or need to be updated to operate
	# in the manor required by the platform. Any tests which are skipped for this
	# reason will be suppressed in the final analysis output.
	#
	na_reason = "Not applicable"

	#
	# Some test cases doesn't have all requirements to run on Github actions CI.
	#
	ci_reason = 'CI runner doesn\'t have all requirements'

	summary = {
	'total': float(0),
	'passed': float(0),
	'logfile': "Could not determine logfile location."
	}

	#
	# These tests are known to fail, thus we use this list to prevent these
	# failures from failing the job as a whole; only unexpected failures
	# bubble up to cause this script to exit with a non-zero exit status.
	#
	# Format: { 'test-name': ['expected result', 'issue-number \| reason'] }
	#
	# For each known failure it is recommended to link to a GitHub issue by
	# setting the reason to the issue number. Alternately, one of the generic
	# reasons listed above can be used.
	#
	known = {
	'casenorm/mixed_none_lookup_ci': ['FAIL', '7633'],
	'casenorm/mixed_formd_lookup_ci': ['FAIL', '7633'],
	'cli_root/zfs_unshare/zfs_unshare_002_pos': ['SKIP', na_reason],
	'cli_root/zfs_unshare/zfs_unshare_006_pos': ['SKIP', na_reason],
	+ 'cli_root/zpool_import/import_rewind_device_replaced':
	+ ['FAIL', rewind_reason],
	'cli_user/misc/zfs_share_001_neg': ['SKIP', na_reason],
	'cli_user/misc/zfs_unshare_001_neg': ['SKIP', na_reason],
	'privilege/setup': ['SKIP', na_reason],
	'refreserv/refreserv_004_pos': ['FAIL', known_reason],
	'rootpool/setup': ['SKIP', na_reason],
	'rsend/rsend_008_pos': ['SKIP', '6066'],
	'vdev_zaps/vdev_zaps_007_pos': ['FAIL', known_reason],
	}

	if sys.platform.startswith('freebsd'):
	known.update({
	'cli_root/zpool_wait/zpool_wait_trim_basic': ['SKIP', trim_reason],
	'cli_root/zpool_wait/zpool_wait_trim_cancel': ['SKIP', trim_reason],
	'cli_root/zpool_wait/zpool_wait_trim_flag': ['SKIP', trim_reason],
	'link_count/link_count_001': ['SKIP', na_reason],
	})
	elif sys.platform.startswith('linux'):
	known.update({
	'casenorm/mixed_formd_lookup': ['FAIL', '7633'],
	'casenorm/mixed_formd_delete': ['FAIL', '7633'],
	'casenorm/sensitive_formd_lookup': ['FAIL', '7633'],
	'casenorm/sensitive_formd_delete': ['FAIL', '7633'],
	'removal/removal_with_zdb': ['SKIP', known_reason],
	})


	#
	# These tests may occasionally fail or be skipped. We want there failures
	# to be reported but only unexpected failures should bubble up to cause
	# this script to exit with a non-zero exit status.
	#
	# Format: { 'test-name': ['expected result', 'issue-number \| reason'] }
	#
	# For each known failure it is recommended to link to a GitHub issue by
	# setting the reason to the issue number. Alternately, one of the generic
	# reasons listed above can be used.
	#
	maybe = {
	'chattr/setup': ['SKIP', exec_reason],
	'crtime/crtime_001_pos': ['SKIP', statx_reason],
	'cli_root/zdb/zdb_006_pos': ['FAIL', known_reason],
	'cli_root/zfs_destroy/zfs_destroy_dev_removal_condense':
	['FAIL', known_reason],
	'cli_root/zfs_get/zfs_get_004_pos': ['FAIL', known_reason],
	'cli_root/zfs_get/zfs_get_009_pos': ['SKIP', '5479'],
	'cli_root/zfs_rollback/zfs_rollback_001_pos': ['FAIL', known_reason],
	'cli_root/zfs_rollback/zfs_rollback_002_pos': ['FAIL', known_reason],
	'cli_root/zfs_share/setup': ['SKIP', share_reason],
	'cli_root/zfs_snapshot/zfs_snapshot_002_neg': ['FAIL', known_reason],
	'cli_root/zfs_unshare/setup': ['SKIP', share_reason],
	'cli_root/zpool_add/zpool_add_004_pos': ['FAIL', known_reason],
	'cli_root/zpool_destroy/zpool_destroy_001_pos': ['SKIP', '6145'],
	- 'cli_root/zpool_import/import_rewind_device_replaced':
	- ['FAIL', rewind_reason],
	'cli_root/zpool_import/import_rewind_config_changed':
	['FAIL', rewind_reason],
	'cli_root/zpool_import/zpool_import_missing_003_pos': ['SKIP', '6839'],
	'cli_root/zpool_initialize/zpool_initialize_import_export':
	['FAIL', '11948'],
	'cli_root/zpool_labelclear/zpool_labelclear_removed':
	['FAIL', known_reason],
	'cli_root/zpool_trim/setup': ['SKIP', trim_reason],
	'cli_root/zpool_upgrade/zpool_upgrade_004_pos': ['FAIL', '6141'],
	'delegate/setup': ['SKIP', exec_reason],
	'fallocate/fallocate_punch-hole': ['SKIP', fspacectl_reason],
	'history/history_004_pos': ['FAIL', '7026'],
	'history/history_005_neg': ['FAIL', '6680'],
	'history/history_006_neg': ['FAIL', '5657'],
	'history/history_008_pos': ['FAIL', known_reason],
	'history/history_010_pos': ['SKIP', exec_reason],
	'io/mmap': ['SKIP', fio_reason],
	'largest_pool/largest_pool_001_pos': ['FAIL', known_reason],
	'mmp/mmp_on_uberblocks': ['FAIL', known_reason],
	'pyzfs/pyzfs_unittest': ['SKIP', python_deps_reason],
	'no_space/enospc_002_pos': ['FAIL', enospc_reason],
	'pool_checkpoint/checkpoint_discard_busy': ['FAIL', '11946'],
	'projectquota/setup': ['SKIP', exec_reason],
	'redundancy/redundancy_004_neg': ['FAIL', '7290'],
	'redundancy/redundancy_draid_spare3': ['SKIP', known_reason],
	'removal/removal_condense_export': ['FAIL', known_reason],
	'reservation/reservation_008_pos': ['FAIL', '7741'],
	'reservation/reservation_018_pos': ['FAIL', '5642'],
	'rsend/rsend_019_pos': ['FAIL', '6086'],
	'rsend/rsend_020_pos': ['FAIL', '6446'],
	'rsend/rsend_021_pos': ['FAIL', '6446'],
	'rsend/rsend_024_pos': ['FAIL', '5665'],
	'rsend/send-c_volume': ['FAIL', '6087'],
	'rsend/send_partial_dataset': ['FAIL', known_reason],
	'snapshot/clone_001_pos': ['FAIL', known_reason],
	'snapshot/snapshot_009_pos': ['FAIL', '7961'],
	'snapshot/snapshot_010_pos': ['FAIL', '7961'],
	'snapused/snapused_004_pos': ['FAIL', '5513'],
	'tmpfile/setup': ['SKIP', tmpfile_reason],
	'threadsappend/threadsappend_001_pos': ['FAIL', '6136'],
	'trim/setup': ['SKIP', trim_reason],
	'upgrade/upgrade_projectquota_001_pos': ['SKIP', project_id_reason],
	'user_namespace/setup': ['SKIP', user_ns_reason],
	'userquota/setup': ['SKIP', exec_reason],
	'vdev_zaps/vdev_zaps_004_pos': ['FAIL', '6935'],
	'zvol/zvol_ENOSPC/zvol_ENOSPC_001_pos': ['FAIL', '5848'],
	'pam/setup': ['SKIP', "pamtester might be not available"],
	}

	if sys.platform.startswith('freebsd'):
	maybe.update({
	'cli_root/zfs_copies/zfs_copies_002_pos': ['FAIL', known_reason],
	'cli_root/zfs_inherit/zfs_inherit_001_neg': ['FAIL', known_reason],
	'cli_root/zfs_receive/receive-o-x_props_override':
	['FAIL', known_reason],
	'cli_root/zfs_share/zfs_share_011_pos': ['FAIL', known_reason],
	'cli_root/zfs_share/zfs_share_concurrent_shares':
	['FAIL', known_reason],
	'cli_root/zpool_import/zpool_import_012_pos': ['FAIL', known_reason],
	'delegate/zfs_allow_003_pos': ['FAIL', known_reason],
	'inheritance/inherit_001_pos': ['FAIL', '11829'],
	'resilver/resilver_restart_001': ['FAIL', known_reason],
	'pool_checkpoint/checkpoint_big_rewind': ['FAIL', '12622'],
	'pool_checkpoint/checkpoint_indirect': ['FAIL', '12623'],
	})
	elif sys.platform.startswith('linux'):
	maybe.update({
	'alloc_class/alloc_class_009_pos': ['FAIL', known_reason],
	'alloc_class/alloc_class_010_pos': ['FAIL', known_reason],
	'alloc_class/alloc_class_011_neg': ['FAIL', known_reason],
	'alloc_class/alloc_class_012_pos': ['FAIL', known_reason],
	'alloc_class/alloc_class_013_pos': ['FAIL', '11888'],
	- 'cli_root/zfs_copies/zfs_copies_003_pos': ['FAIL', '12301'],
	'cli_root/zfs_rename/zfs_rename_002_pos': ['FAIL', known_reason],
	'cli_root/zpool_expand/zpool_expand_001_pos': ['FAIL', known_reason],
	'cli_root/zpool_expand/zpool_expand_005_pos': ['FAIL', known_reason],
	'cli_root/zpool_reopen/zpool_reopen_003_pos': ['FAIL', known_reason],
	'fault/auto_spare_shared': ['FAIL', '11889'],
	'io/io_uring': ['SKIP', 'io_uring support required'],
	'limits/filesystem_limit': ['SKIP', known_reason],
	'limits/snapshot_limit': ['SKIP', known_reason],
	'mmp/mmp_active_import': ['FAIL', known_reason],
	'mmp/mmp_exported_import': ['FAIL', known_reason],
	'mmp/mmp_inactive_import': ['FAIL', known_reason],
	'refreserv/refreserv_raidz': ['FAIL', known_reason],
	'rsend/rsend_007_pos': ['FAIL', known_reason],
	'rsend/rsend_010_pos': ['FAIL', known_reason],
	'rsend/rsend_011_pos': ['FAIL', known_reason],
	'snapshot/rollback_003_pos': ['FAIL', known_reason],
	'zvol/zvol_misc/zvol_misc_snapdev': ['FAIL', '12621'],
	'zvol/zvol_misc/zvol_misc_volmode': ['FAIL', known_reason],
	})


	# Not all Github actions runners have scsi_debug module, so we may skip
	# some tests which use it.
	if os.environ.get('CI') == 'true':
	known.update({
	'cli_root/zpool_expand/zpool_expand_001_pos': ['SKIP', ci_reason],
	'cli_root/zpool_expand/zpool_expand_003_neg': ['SKIP', ci_reason],
	'cli_root/zpool_expand/zpool_expand_005_pos': ['SKIP', ci_reason],
	'cli_root/zpool_reopen/setup': ['SKIP', ci_reason],
	'cli_root/zpool_reopen/zpool_reopen_001_pos': ['SKIP', ci_reason],
	'cli_root/zpool_reopen/zpool_reopen_002_pos': ['SKIP', ci_reason],
	'cli_root/zpool_reopen/zpool_reopen_003_pos': ['SKIP', ci_reason],
	'cli_root/zpool_reopen/zpool_reopen_004_pos': ['SKIP', ci_reason],
	'cli_root/zpool_reopen/zpool_reopen_005_pos': ['SKIP', ci_reason],
	'cli_root/zpool_reopen/zpool_reopen_006_neg': ['SKIP', ci_reason],
	'cli_root/zpool_reopen/zpool_reopen_007_pos': ['SKIP', ci_reason],
	'cli_root/zpool_split/zpool_split_wholedisk': ['SKIP', ci_reason],
	'fault/auto_offline_001_pos': ['SKIP', ci_reason],
	'fault/auto_online_001_pos': ['SKIP', ci_reason],
	'fault/auto_online_002_pos': ['SKIP', ci_reason],
	'fault/auto_replace_001_pos': ['SKIP', ci_reason],
	'fault/auto_spare_ashift': ['SKIP', ci_reason],
	'fault/auto_spare_shared': ['SKIP', ci_reason],
	'procfs/pool_state': ['SKIP', ci_reason],
	})

	maybe.update({
	'events/events_002_pos': ['FAIL', '11546'],
	})


	def usage(s):
	print(s)
	sys.exit(1)


	def process_results(pathname):
	try:
	f = open(pathname)
	except IOError as e:
	print('Error opening file: %s' % e)
	sys.exit(1)

	prefix = '/zfs-tests/tests/functional/'
	pattern = \
	r'^Test(?:\s+$\S+$)?:' + \
	r'\s\S%s(\S+)\s$run as (\S+)$\s\[(\S+)\]\s*\[(\S+)\]' \
	% prefix
	pattern_log = r'^\sLog directory:\s(\S*)'

	d = {}
	for line in f.readlines():
	m = re.match(pattern, line)
	if m and len(m.groups()) == 4:
	summary['total'] += 1
	if m.group(4) == "PASS":
	summary['passed'] += 1
	d[m.group(1)] = m.group(4)
	continue

	m = re.match(pattern_log, line)
	if m:
	summary['logfile'] = m.group(1)

	return d


	+class ListMaybesAction(argparse.Action):
	+ def __init__(self,
	+ option_strings,
	+ dest="SUPPRESS",
	+ default="SUPPRESS",
	+ help="list flaky tests and exit"):
	+ super(ListMaybesAction, self).__init__(
	+ option_strings=option_strings,
	+ dest=dest,
	+ default=default,
	+ nargs=0,
	+ help=help)
	+
	+ def __call__(self, parser, namespace, values, option_string=None):
	+ for test in maybe:
	+ print(test)
	+ sys.exit(0)
	+
	+
	if __name__ == "__main__":
	- if len(sys.argv) != 2:
	- usage('usage: %s <pathname>' % sys.argv[0])
	- results = process_results(sys.argv[1])
	+ parser = argparse.ArgumentParser(description='Analyze ZTS logs')
	+ parser.add_argument('logfile')
	+ parser.add_argument('--list-maybes', action=ListMaybesAction)
	+ parser.add_argument('--no-maybes', action='store_false', dest='maybes')
	+ args = parser.parse_args()
	+
	+ results = process_results(args.logfile)

	if summary['total'] == 0:
	print("\n\nNo test results were found.")
	print("Log directory: %s" % summary['logfile'])
	sys.exit(0)

	expected = []
	unexpected = []
	+ all_maybes = True

	for test in list(results.keys()):
	if results[test] == "PASS":
	continue

	setup = test.replace(os.path.basename(test), "setup")
	if results[test] == "SKIP" and test != setup:
	if setup in known and known[setup][0] == "SKIP":
	continue
	if setup in maybe and maybe[setup][0] == "SKIP":
	continue

	- if ((test not in known or results[test] not in known[test][0]) and
	- (test not in maybe or results[test] not in maybe[test][0])):
	- unexpected.append(test)
	- else:
	+ if (test in known and results[test] in known[test][0]):
	expected.append(test)
	+ elif test in maybe and results[test] in maybe[test][0]:
	+ if results[test] == 'SKIP' or args.maybes:
	+ expected.append(test)
	+ elif not args.maybes:
	+ unexpected.append(test)
	+ else:
	+ unexpected.append(test)
	+ all_maybes = False

	print("\nTests with results other than PASS that are expected:")
	for test in sorted(expected):
	issue_url = 'https://github.com/openzfs/zfs/issues/'

	# Include the reason why the result is expected, given the following:
	# 1. Suppress test results which set the "Not applicable" reason.
	# 2. Numerical reasons are assumed to be GitHub issue numbers.
	# 3. When an entire test group is skipped only report the setup reason.
	if test in known:
	if known[test][1] == na_reason:
	continue
	elif known[test][1].isdigit():
	expect = issue_url + known[test][1]
	else:
	expect = known[test][1]
	elif test in maybe:
	if maybe[test][1].isdigit():
	expect = issue_url + maybe[test][1]
	else:
	expect = maybe[test][1]
	elif setup in known and known[setup][0] == "SKIP" and setup != test:
	continue
	elif setup in maybe and maybe[setup][0] == "SKIP" and setup != test:
	continue
	else:
	expect = "UNKNOWN REASON"
	print(" %s %s (%s)" % (results[test], test, expect))

	print("\nTests with result of PASS that are unexpected:")
	for test in sorted(known.keys()):
	# We probably should not be silently ignoring the case
	# where "test" is not in "results".
	if test not in results or results[test] != "PASS":
	continue
	print(" %s %s (expected %s)" % (results[test], test,
	known[test][0]))

	print("\nTests with results other than PASS that are unexpected:")
	for test in sorted(unexpected):
	expect = "PASS" if test not in known else known[test][0]
	print(" %s %s (expected %s)" % (results[test], test, expect))

	if len(unexpected) == 0:
	sys.exit(0)
	+ elif not args.maybes and all_maybes:
	+ sys.exit(2)
	else:
	sys.exit(1)
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/cmd/libzfs_input_check/libzfs_input_check.c b/sys/contrib/openzfs/tests/zfs-tests/cmd/libzfs_input_check/libzfs_input_check.c
	index 0e552c2680a0..fba18068cfc8 100644
	--- a/sys/contrib/openzfs/tests/zfs-tests/cmd/libzfs_input_check/libzfs_input_check.c
	+++ b/sys/contrib/openzfs/tests/zfs-tests/cmd/libzfs_input_check/libzfs_input_check.c
	@@ -1,1063 +1,1063 @@
	/*
	* CDDL HEADER START
	*
	* This file and its contents are supplied under the terms of the
	* Common Development and Distribution License ("CDDL"), version 1.0.
	* You may only use this file in accordance with the terms of version
	* 1.0 of the CDDL.
	*
	* A full copy of the text of the CDDL should have accompanied this
	* source. A copy of the CDDL is also available via the Internet at
	* http://www.illumos.org/license/CDDL.
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2018 by Delphix. All rights reserved.
	*/

	#include <stdio.h>
	#include <stdlib.h>
	#include <string.h>
	#include <strings.h>
	#include <libzfs_core.h>
	#include <libzutil.h>

	#include <sys/nvpair.h>
	#include <sys/vdev_impl.h>
	#include <sys/zfs_ioctl.h>
	#include <sys/zfs_bootenv.h>

	/*
	* Test the nvpair inputs for the non-legacy zfs ioctl commands.
	*/

	boolean_t unexpected_failures;
	int zfs_fd;
	const char *active_test;

	/*
	* Tracks which zfs_ioc_t commands were tested
	*/
	boolean_t ioc_tested[ZFS_IOC_LAST - ZFS_IOC_FIRST];

	/*
	* Legacy ioctls that are skipped (for now)
	*/
	static unsigned ioc_skip[] = {
	ZFS_IOC_POOL_CREATE,
	ZFS_IOC_POOL_DESTROY,
	ZFS_IOC_POOL_IMPORT,
	ZFS_IOC_POOL_EXPORT,
	ZFS_IOC_POOL_CONFIGS,
	ZFS_IOC_POOL_STATS,
	ZFS_IOC_POOL_TRYIMPORT,
	ZFS_IOC_POOL_SCAN,
	ZFS_IOC_POOL_FREEZE,
	ZFS_IOC_POOL_UPGRADE,
	ZFS_IOC_POOL_GET_HISTORY,

	ZFS_IOC_VDEV_ADD,
	ZFS_IOC_VDEV_REMOVE,
	ZFS_IOC_VDEV_SET_STATE,
	ZFS_IOC_VDEV_ATTACH,
	ZFS_IOC_VDEV_DETACH,
	ZFS_IOC_VDEV_SETPATH,
	ZFS_IOC_VDEV_SETFRU,

	ZFS_IOC_OBJSET_STATS,
	ZFS_IOC_OBJSET_ZPLPROPS,
	ZFS_IOC_DATASET_LIST_NEXT,
	ZFS_IOC_SNAPSHOT_LIST_NEXT,
	ZFS_IOC_SET_PROP,
	ZFS_IOC_DESTROY,
	ZFS_IOC_RENAME,
	ZFS_IOC_RECV,
	ZFS_IOC_SEND,
	ZFS_IOC_INJECT_FAULT,
	ZFS_IOC_CLEAR_FAULT,
	ZFS_IOC_INJECT_LIST_NEXT,
	ZFS_IOC_ERROR_LOG,
	ZFS_IOC_CLEAR,
	ZFS_IOC_PROMOTE,
	ZFS_IOC_DSOBJ_TO_DSNAME,
	ZFS_IOC_OBJ_TO_PATH,
	ZFS_IOC_POOL_SET_PROPS,
	ZFS_IOC_POOL_GET_PROPS,
	ZFS_IOC_SET_FSACL,
	ZFS_IOC_GET_FSACL,
	ZFS_IOC_SHARE,
	ZFS_IOC_INHERIT_PROP,
	ZFS_IOC_SMB_ACL,
	ZFS_IOC_USERSPACE_ONE,
	ZFS_IOC_USERSPACE_MANY,
	ZFS_IOC_USERSPACE_UPGRADE,
	ZFS_IOC_OBJSET_RECVD_PROPS,
	ZFS_IOC_VDEV_SPLIT,
	ZFS_IOC_NEXT_OBJ,
	ZFS_IOC_DIFF,
	ZFS_IOC_TMP_SNAPSHOT,
	ZFS_IOC_OBJ_TO_STATS,
	ZFS_IOC_SPACE_WRITTEN,
	ZFS_IOC_POOL_REGUID,
	ZFS_IOC_SEND_PROGRESS,
	ZFS_IOC_EVENTS_NEXT,
	ZFS_IOC_EVENTS_CLEAR,
	ZFS_IOC_EVENTS_SEEK,
	ZFS_IOC_NEXTBOOT,
	ZFS_IOC_JAIL,
	ZFS_IOC_UNJAIL,
	};


	#define IOC_INPUT_TEST(ioc, name, req, opt, err) \
	IOC_INPUT_TEST_IMPL(ioc, name, req, opt, err, B_FALSE)

	#define IOC_INPUT_TEST_WILD(ioc, name, req, opt, err) \
	IOC_INPUT_TEST_IMPL(ioc, name, req, opt, err, B_TRUE)

	#define IOC_INPUT_TEST_IMPL(ioc, name, req, opt, err, wild) \
	do { \
	active_test = __func__ + 5; \
	ioc_tested[ioc - ZFS_IOC_FIRST] = B_TRUE; \
	lzc_ioctl_test(ioc, name, req, opt, err, wild); \
	} while (0)

	/*
	* run a zfs ioctl command, verify expected results and log failures
	*/
	static void
	lzc_ioctl_run(zfs_ioc_t ioc, const char name, nvlist_t innvl, int expected)
	{
	zfs_cmd_t zc = {"\0"};
	char *packed = NULL;
	const char *variant;
	size_t size = 0;
	int error = 0;

	switch (expected) {
	case ZFS_ERR_IOC_ARG_UNAVAIL:
	variant = "unsupported input";
	break;
	case ZFS_ERR_IOC_ARG_REQUIRED:
	variant = "missing input";
	break;
	case ZFS_ERR_IOC_ARG_BADTYPE:
	variant = "invalid input type";
	break;
	default:
	variant = "valid input";
	break;
	}

	packed = fnvlist_pack(innvl, &size);
	(void) strlcpy(zc.zc_name, name, sizeof (zc.zc_name));
	zc.zc_name[sizeof (zc.zc_name) - 1] = '\0';
	zc.zc_nvlist_src = (uint64_t)(uintptr_t)packed;
	zc.zc_nvlist_src_size = size;
	zc.zc_nvlist_dst_size = MAX(size * 2, 128 * 1024);
	zc.zc_nvlist_dst = (uint64_t)(uintptr_t)malloc(zc.zc_nvlist_dst_size);

	if (lzc_ioctl_fd(zfs_fd, ioc, &zc) != 0)
	error = errno;

	if (error != expected) {
	unexpected_failures = B_TRUE;
	(void) fprintf(stderr, "%s: Unexpected result with %s, "
	"error %d (expecting %d)\n",
	active_test, variant, error, expected);
	}

	fnvlist_pack_free(packed, size);
	free((void *)(uintptr_t)zc.zc_nvlist_dst);
	}

	/*
	* Test each ioc for the following ioctl input errors:
	* ZFS_ERR_IOC_ARG_UNAVAIL an input argument is not supported by kernel
	* ZFS_ERR_IOC_ARG_REQUIRED a required input argument is missing
	* ZFS_ERR_IOC_ARG_BADTYPE an input argument has an invalid type
	*/
	static int
	lzc_ioctl_test(zfs_ioc_t ioc, const char name, nvlist_t required,
	nvlist_t *optional, int expected_error, boolean_t wildcard)
	{
	nvlist_t *input = fnvlist_alloc();
	nvlist_t *future = fnvlist_alloc();
	int error = 0;

	if (required != NULL) {
	for (nvpair_t *pair = nvlist_next_nvpair(required, NULL);
	pair != NULL; pair = nvlist_next_nvpair(required, pair)) {
	fnvlist_add_nvpair(input, pair);
	}
	}
	if (optional != NULL) {
	for (nvpair_t *pair = nvlist_next_nvpair(optional, NULL);
	pair != NULL; pair = nvlist_next_nvpair(optional, pair)) {
	fnvlist_add_nvpair(input, pair);
	}
	}

	/*
	* Generic input run with 'optional' nvlist pair
	*/
	if (!wildcard)
	fnvlist_add_nvlist(input, "optional", future);
	lzc_ioctl_run(ioc, name, input, expected_error);
	if (!wildcard)
	fnvlist_remove(input, "optional");

	/*
	* Bogus input value
	*/
	if (!wildcard) {
	fnvlist_add_string(input, "bogus_input", "bogus");
	lzc_ioctl_run(ioc, name, input, ZFS_ERR_IOC_ARG_UNAVAIL);
	fnvlist_remove(input, "bogus_input");
	}

	/*
	* Missing required inputs
	*/
	if (required != NULL) {
	nvlist_t *empty = fnvlist_alloc();
	lzc_ioctl_run(ioc, name, empty, ZFS_ERR_IOC_ARG_REQUIRED);
	nvlist_free(empty);
	}

	/*
	* Wrong nvpair type
	*/
	if (required != NULL \|\| optional != NULL) {
	/*
	* switch the type of one of the input pairs
	*/
	for (nvpair_t *pair = nvlist_next_nvpair(input, NULL);
	pair != NULL; pair = nvlist_next_nvpair(input, pair)) {
	char pname[MAXNAMELEN];
	data_type_t ptype;

	strlcpy(pname, nvpair_name(pair), sizeof (pname));
	pname[sizeof (pname) - 1] = '\0';
	ptype = nvpair_type(pair);
	fnvlist_remove_nvpair(input, pair);

	switch (ptype) {
	case DATA_TYPE_STRING:
	fnvlist_add_uint64(input, pname, 42);
	break;
	default:
	fnvlist_add_string(input, pname, "bogus");
	break;
	}
	}
	lzc_ioctl_run(ioc, name, input, ZFS_ERR_IOC_ARG_BADTYPE);
	}

	nvlist_free(future);
	nvlist_free(input);

	return (error);
	}

	static void
	test_pool_sync(const char *pool)
	{
	nvlist_t *required = fnvlist_alloc();

	fnvlist_add_boolean_value(required, "force", B_TRUE);

	IOC_INPUT_TEST(ZFS_IOC_POOL_SYNC, pool, required, NULL, 0);

	nvlist_free(required);
	}

	static void
	test_pool_reopen(const char *pool)
	{
	nvlist_t *optional = fnvlist_alloc();

	fnvlist_add_boolean_value(optional, "scrub_restart", B_FALSE);

	IOC_INPUT_TEST(ZFS_IOC_POOL_REOPEN, pool, NULL, optional, 0);

	nvlist_free(optional);
	}

	static void
	test_pool_checkpoint(const char *pool)
	{
	IOC_INPUT_TEST(ZFS_IOC_POOL_CHECKPOINT, pool, NULL, NULL, 0);
	}

	static void
	test_pool_discard_checkpoint(const char *pool)
	{
	int err = lzc_pool_checkpoint(pool);
	if (err == 0 \|\| err == ZFS_ERR_CHECKPOINT_EXISTS)
	IOC_INPUT_TEST(ZFS_IOC_POOL_DISCARD_CHECKPOINT, pool, NULL,
	NULL, 0);
	}

	static void
	test_log_history(const char *pool)
	{
	nvlist_t *required = fnvlist_alloc();

	fnvlist_add_string(required, "message", "input check");

	IOC_INPUT_TEST(ZFS_IOC_LOG_HISTORY, pool, required, NULL, 0);

	nvlist_free(required);
	}

	static void
	test_create(const char *pool)
	{
	char dataset[MAXNAMELEN + 32];

	(void) snprintf(dataset, sizeof (dataset), "%s/create-fs", pool);

	nvlist_t *required = fnvlist_alloc();
	nvlist_t *optional = fnvlist_alloc();
	nvlist_t *props = fnvlist_alloc();

	fnvlist_add_int32(required, "type", DMU_OST_ZFS);
	fnvlist_add_uint64(props, "recordsize", 8192);
	fnvlist_add_nvlist(optional, "props", props);

	IOC_INPUT_TEST(ZFS_IOC_CREATE, dataset, required, optional, 0);

	nvlist_free(required);
	nvlist_free(optional);
	}

	static void
	test_snapshot(const char pool, const char snapshot)
	{
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *optional = fnvlist_alloc();
	nvlist_t *snaps = fnvlist_alloc();
	nvlist_t *props = fnvlist_alloc();

	fnvlist_add_boolean(snaps, snapshot);
	fnvlist_add_nvlist(required, "snaps", snaps);

	fnvlist_add_string(props, "org.openzfs:launch", "September 17th, 2013");
	fnvlist_add_nvlist(optional, "props", props);

	IOC_INPUT_TEST(ZFS_IOC_SNAPSHOT, pool, required, optional, 0);

	nvlist_free(props);
	nvlist_free(snaps);
	nvlist_free(optional);
	nvlist_free(required);
	}

	static void
	test_space_snaps(const char *snapshot)
	{
	nvlist_t *required = fnvlist_alloc();
	fnvlist_add_string(required, "firstsnap", snapshot);

	IOC_INPUT_TEST(ZFS_IOC_SPACE_SNAPS, snapshot, required, NULL, 0);

	nvlist_free(required);
	}

	static void
	test_destroy_snaps(const char pool, const char snapshot)
	{
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *snaps = fnvlist_alloc();

	fnvlist_add_boolean(snaps, snapshot);
	fnvlist_add_nvlist(required, "snaps", snaps);

	IOC_INPUT_TEST(ZFS_IOC_DESTROY_SNAPS, pool, required, NULL, 0);

	nvlist_free(snaps);
	nvlist_free(required);
	}


	static void
	test_bookmark(const char pool, const char snapshot, const char *bookmark)
	{
	nvlist_t *required = fnvlist_alloc();

	fnvlist_add_string(required, bookmark, snapshot);

	IOC_INPUT_TEST_WILD(ZFS_IOC_BOOKMARK, pool, required, NULL, 0);

	nvlist_free(required);
	}

	static void
	test_get_bookmarks(const char *dataset)
	{
	nvlist_t *optional = fnvlist_alloc();

	fnvlist_add_boolean(optional, "guid");
	fnvlist_add_boolean(optional, "createtxg");
	fnvlist_add_boolean(optional, "creation");

	IOC_INPUT_TEST_WILD(ZFS_IOC_GET_BOOKMARKS, dataset, NULL, optional, 0);

	nvlist_free(optional);
	}

	static void
	test_destroy_bookmarks(const char pool, const char bookmark)
	{
	nvlist_t *required = fnvlist_alloc();

	fnvlist_add_boolean(required, bookmark);

	IOC_INPUT_TEST_WILD(ZFS_IOC_DESTROY_BOOKMARKS, pool, required, NULL, 0);

	nvlist_free(required);
	}

	static void
	test_clone(const char snapshot, const char clone)
	{
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *optional = fnvlist_alloc();
	nvlist_t *props = fnvlist_alloc();

	fnvlist_add_string(required, "origin", snapshot);

	IOC_INPUT_TEST(ZFS_IOC_CLONE, clone, required, NULL, 0);

	nvlist_free(props);
	nvlist_free(optional);
	nvlist_free(required);
	}

	static void
	test_rollback(const char dataset, const char snapshot)
	{
	nvlist_t *optional = fnvlist_alloc();

	fnvlist_add_string(optional, "target", snapshot);

	IOC_INPUT_TEST(ZFS_IOC_ROLLBACK, dataset, NULL, optional, B_FALSE);

	nvlist_free(optional);
	}

	static void
	test_hold(const char pool, const char snapshot)
	{
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *optional = fnvlist_alloc();
	nvlist_t *holds = fnvlist_alloc();

	fnvlist_add_string(holds, snapshot, "libzfs_check_hold");
	fnvlist_add_nvlist(required, "holds", holds);
	fnvlist_add_int32(optional, "cleanup_fd", zfs_fd);

	IOC_INPUT_TEST(ZFS_IOC_HOLD, pool, required, optional, 0);

	nvlist_free(holds);
	nvlist_free(optional);
	nvlist_free(required);
	}

	static void
	test_get_holds(const char *snapshot)
	{
	IOC_INPUT_TEST(ZFS_IOC_GET_HOLDS, snapshot, NULL, NULL, 0);
	}

	static void
	test_release(const char pool, const char snapshot)
	{
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *release = fnvlist_alloc();

	fnvlist_add_boolean(release, "libzfs_check_hold");
	fnvlist_add_nvlist(required, snapshot, release);

	IOC_INPUT_TEST_WILD(ZFS_IOC_RELEASE, pool, required, NULL, 0);

	nvlist_free(release);
	nvlist_free(required);
	}


	static void
	test_send_new(const char *snapshot, int fd)
	{
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *optional = fnvlist_alloc();

	fnvlist_add_int32(required, "fd", fd);

	fnvlist_add_boolean(optional, "largeblockok");
	fnvlist_add_boolean(optional, "embedok");
	fnvlist_add_boolean(optional, "compressok");
	fnvlist_add_boolean(optional, "rawok");

	/*
	* TODO - Resumable send is harder to set up. So we currently
	* ignore testing for that variant.
	*/
	#if 0
	fnvlist_add_string(optional, "fromsnap", from);
	fnvlist_add_uint64(optional, "resume_object", resumeobj);
	fnvlist_add_uint64(optional, "resume_offset", offset);
	fnvlist_add_boolean(optional, "savedok");
	#endif
	IOC_INPUT_TEST(ZFS_IOC_SEND_NEW, snapshot, required, optional, 0);

	nvlist_free(optional);
	nvlist_free(required);
	}

	static void
	test_recv_new(const char *dataset, int fd)
	{
	dmu_replay_record_t drr = { 0 };
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *optional = fnvlist_alloc();
	nvlist_t *props = fnvlist_alloc();
	char snapshot[MAXNAMELEN + 32];
	ssize_t count;

	int cleanup_fd = open(ZFS_DEV, O_RDWR);

	(void) snprintf(snapshot, sizeof (snapshot), "%s@replicant", dataset);

	count = pread(fd, &drr, sizeof (drr), 0);
	if (count != sizeof (drr)) {
	(void) fprintf(stderr, "could not read stream: %s\n",
	strerror(errno));
	}

	fnvlist_add_string(required, "snapname", snapshot);
	fnvlist_add_byte_array(required, "begin_record", (uchar_t *)&drr,
	sizeof (drr));
	fnvlist_add_int32(required, "input_fd", fd);

	fnvlist_add_string(props, "org.openzfs:launch", "September 17th, 2013");
	fnvlist_add_nvlist(optional, "localprops", props);
	fnvlist_add_boolean(optional, "force");
	fnvlist_add_int32(optional, "cleanup_fd", cleanup_fd);

	/*
	* TODO - Resumable receive is harder to set up. So we currently
	* ignore testing for one.
	*/
	#if 0
	fnvlist_add_nvlist(optional, "props", recvdprops);
	fnvlist_add_string(optional, "origin", origin);
	fnvlist_add_boolean(optional, "resumable");
	fnvlist_add_uint64(optional, "action_handle", *action_handle);
	#endif
	IOC_INPUT_TEST(ZFS_IOC_RECV_NEW, dataset, required, optional,
	ZFS_ERR_STREAM_TRUNCATED);

	nvlist_free(props);
	nvlist_free(optional);
	nvlist_free(required);

	(void) close(cleanup_fd);
	}

	static void
	test_send_space(const char snapshot1, const char snapshot2)
	{
	nvlist_t *optional = fnvlist_alloc();

	fnvlist_add_string(optional, "from", snapshot1);
	fnvlist_add_boolean(optional, "largeblockok");
	fnvlist_add_boolean(optional, "embedok");
	fnvlist_add_boolean(optional, "compressok");
	fnvlist_add_boolean(optional, "rawok");

	IOC_INPUT_TEST(ZFS_IOC_SEND_SPACE, snapshot2, NULL, optional, 0);

	nvlist_free(optional);
	}

	static void
	test_remap(const char *dataset)
	{
	IOC_INPUT_TEST(ZFS_IOC_REMAP, dataset, NULL, NULL, 0);
	}

	static void
	test_channel_program(const char *pool)
	{
	const char *program =
	"arg = ...\n"
	"argv = arg[\"argv\"]\n"
	"return argv[1]";
	char *const argv[1] = { "Hello World!" };
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *optional = fnvlist_alloc();
	nvlist_t *args = fnvlist_alloc();

	fnvlist_add_string(required, "program", program);
	- fnvlist_add_string_array(args, "argv", argv, 1);
	+ fnvlist_add_string_array(args, "argv", (const char * const *)argv, 1);
	fnvlist_add_nvlist(required, "arg", args);

	fnvlist_add_boolean_value(optional, "sync", B_TRUE);
	fnvlist_add_uint64(optional, "instrlimit", 1000 * 1000);
	fnvlist_add_uint64(optional, "memlimit", 8192 * 1024);

	IOC_INPUT_TEST(ZFS_IOC_CHANNEL_PROGRAM, pool, required, optional, 0);

	nvlist_free(args);
	nvlist_free(optional);
	nvlist_free(required);
	}

	#define WRAPPING_KEY_LEN 32

	static void
	test_load_key(const char *dataset)
	{
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *optional = fnvlist_alloc();
	nvlist_t *hidden = fnvlist_alloc();
	uint8_t keydata[WRAPPING_KEY_LEN] = {0};

	fnvlist_add_uint8_array(hidden, "wkeydata", keydata, sizeof (keydata));
	fnvlist_add_nvlist(required, "hidden_args", hidden);
	fnvlist_add_boolean(optional, "noop");

	IOC_INPUT_TEST(ZFS_IOC_LOAD_KEY, dataset, required, optional, EINVAL);
	nvlist_free(hidden);
	nvlist_free(optional);
	nvlist_free(required);
	}

	static void
	test_change_key(const char *dataset)
	{
	IOC_INPUT_TEST(ZFS_IOC_CHANGE_KEY, dataset, NULL, NULL, EINVAL);
	}

	static void
	test_unload_key(const char *dataset)
	{
	IOC_INPUT_TEST(ZFS_IOC_UNLOAD_KEY, dataset, NULL, NULL, EACCES);
	}

	static void
	test_vdev_initialize(const char *pool)
	{
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *vdev_guids = fnvlist_alloc();

	fnvlist_add_uint64(vdev_guids, "path", 0xdeadbeefdeadbeef);
	fnvlist_add_uint64(required, ZPOOL_INITIALIZE_COMMAND,
	POOL_INITIALIZE_START);
	fnvlist_add_nvlist(required, ZPOOL_INITIALIZE_VDEVS, vdev_guids);

	IOC_INPUT_TEST(ZFS_IOC_POOL_INITIALIZE, pool, required, NULL, EINVAL);
	nvlist_free(vdev_guids);
	nvlist_free(required);
	}

	static void
	test_vdev_trim(const char *pool)
	{
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *optional = fnvlist_alloc();
	nvlist_t *vdev_guids = fnvlist_alloc();

	fnvlist_add_uint64(vdev_guids, "path", 0xdeadbeefdeadbeef);
	fnvlist_add_uint64(required, ZPOOL_TRIM_COMMAND, POOL_TRIM_START);
	fnvlist_add_nvlist(required, ZPOOL_TRIM_VDEVS, vdev_guids);
	fnvlist_add_uint64(optional, ZPOOL_TRIM_RATE, 1ULL << 30);
	fnvlist_add_boolean_value(optional, ZPOOL_TRIM_SECURE, B_TRUE);

	IOC_INPUT_TEST(ZFS_IOC_POOL_TRIM, pool, required, optional, EINVAL);
	nvlist_free(vdev_guids);
	nvlist_free(optional);
	nvlist_free(required);
	}

	static int
	zfs_destroy(const char *dataset)
	{
	zfs_cmd_t zc = {"\0"};
	int err;

	(void) strlcpy(zc.zc_name, dataset, sizeof (zc.zc_name));
	zc.zc_name[sizeof (zc.zc_name) - 1] = '\0';
	err = lzc_ioctl_fd(zfs_fd, ZFS_IOC_DESTROY, &zc);

	return (err == 0 ? 0 : errno);
	}

	static void
	test_redact(const char snapshot1, const char snapshot2)
	{
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *snapnv = fnvlist_alloc();
	char bookmark[MAXNAMELEN + 32];

	fnvlist_add_string(required, "bookname", "testbookmark");
	fnvlist_add_boolean(snapnv, snapshot2);
	fnvlist_add_nvlist(required, "snapnv", snapnv);

	IOC_INPUT_TEST(ZFS_IOC_REDACT, snapshot1, required, NULL, 0);

	nvlist_free(snapnv);
	nvlist_free(required);

	strlcpy(bookmark, snapshot1, sizeof (bookmark));
	*strchr(bookmark, '@') = '\0';
	strlcat(bookmark, "#testbookmark", sizeof (bookmark) -
	strlen(bookmark));
	zfs_destroy(bookmark);
	}

	static void
	test_get_bookmark_props(const char *bookmark)
	{
	IOC_INPUT_TEST(ZFS_IOC_GET_BOOKMARK_PROPS, bookmark, NULL, NULL, 0);
	}

	static void
	test_wait(const char *pool)
	{
	nvlist_t *required = fnvlist_alloc();
	nvlist_t *optional = fnvlist_alloc();

	fnvlist_add_int32(required, "wait_activity", 2);
	fnvlist_add_uint64(optional, "wait_tag", 0xdeadbeefdeadbeef);

	IOC_INPUT_TEST(ZFS_IOC_WAIT, pool, required, optional, EINVAL);

	nvlist_free(required);
	nvlist_free(optional);
	}

	static void
	test_wait_fs(const char *dataset)
	{
	nvlist_t *required = fnvlist_alloc();

	fnvlist_add_int32(required, "wait_activity", 2);

	IOC_INPUT_TEST(ZFS_IOC_WAIT_FS, dataset, required, NULL, EINVAL);

	nvlist_free(required);
	}

	static void
	test_get_bootenv(const char *pool)
	{
	IOC_INPUT_TEST(ZFS_IOC_GET_BOOTENV, pool, NULL, NULL, 0);
	}

	static void
	test_set_bootenv(const char *pool)
	{
	nvlist_t *required = fnvlist_alloc();

	fnvlist_add_uint64(required, "version", VB_RAW);
	fnvlist_add_string(required, GRUB_ENVMAP, "test");

	IOC_INPUT_TEST_WILD(ZFS_IOC_SET_BOOTENV, pool, required, NULL, 0);

	nvlist_free(required);
	}

	static void
	zfs_ioc_input_tests(const char *pool)
	{
	char filepath[] = "/tmp/ioc_test_file_XXXXXX";
	char dataset[ZFS_MAX_DATASET_NAME_LEN];
	char snapbase[ZFS_MAX_DATASET_NAME_LEN + 32];
	char snapshot[ZFS_MAX_DATASET_NAME_LEN + 32];
	char bookmark[ZFS_MAX_DATASET_NAME_LEN + 32];
	char backup[ZFS_MAX_DATASET_NAME_LEN];
	char clone[ZFS_MAX_DATASET_NAME_LEN];
	char clonesnap[ZFS_MAX_DATASET_NAME_LEN + 32];
	int tmpfd, err;

	/*
	* Setup names and create a working dataset
	*/
	(void) snprintf(dataset, sizeof (dataset), "%s/test-fs", pool);
	(void) snprintf(snapbase, sizeof (snapbase), "%s@snapbase", dataset);
	(void) snprintf(snapshot, sizeof (snapshot), "%s@snapshot", dataset);
	(void) snprintf(bookmark, sizeof (bookmark), "%s#bookmark", dataset);
	(void) snprintf(clone, sizeof (clone), "%s/test-fs-clone", pool);
	(void) snprintf(clonesnap, sizeof (clonesnap), "%s@snap", clone);
	(void) snprintf(backup, sizeof (backup), "%s/backup", pool);

	err = lzc_create(dataset, LZC_DATSET_TYPE_ZFS, NULL, NULL, -1);
	if (err) {
	(void) fprintf(stderr, "could not create '%s': %s\n",
	dataset, strerror(errno));
	exit(2);
	}

	tmpfd = mkstemp(filepath);
	if (tmpfd < 0) {
	(void) fprintf(stderr, "could not create '%s': %s\n",
	filepath, strerror(errno));
	exit(2);
	}

	/*
	* run a test for each ioctl
	* Note that some test build on previous test operations
	*/
	test_pool_sync(pool);
	test_pool_reopen(pool);
	test_pool_checkpoint(pool);
	test_pool_discard_checkpoint(pool);
	test_log_history(pool);

	test_create(dataset);
	test_snapshot(pool, snapbase);
	test_snapshot(pool, snapshot);

	test_space_snaps(snapshot);
	test_send_space(snapbase, snapshot);
	test_send_new(snapshot, tmpfd);
	test_recv_new(backup, tmpfd);

	test_bookmark(pool, snapshot, bookmark);
	test_get_bookmarks(dataset);
	test_get_bookmark_props(bookmark);
	test_destroy_bookmarks(pool, bookmark);

	test_hold(pool, snapshot);
	test_get_holds(snapshot);
	test_release(pool, snapshot);

	test_clone(snapshot, clone);
	test_snapshot(pool, clonesnap);
	test_redact(snapshot, clonesnap);
	zfs_destroy(clonesnap);
	zfs_destroy(clone);

	test_rollback(dataset, snapshot);
	test_destroy_snaps(pool, snapshot);
	test_destroy_snaps(pool, snapbase);

	test_remap(dataset);
	test_channel_program(pool);

	test_load_key(dataset);
	test_change_key(dataset);
	test_unload_key(dataset);

	test_vdev_initialize(pool);
	test_vdev_trim(pool);

	test_wait(pool);
	test_wait_fs(dataset);

	test_set_bootenv(pool);
	test_get_bootenv(pool);

	/*
	* cleanup
	*/
	zfs_cmd_t zc = {"\0"};

	nvlist_t *snaps = fnvlist_alloc();
	fnvlist_add_boolean(snaps, snapshot);
	(void) lzc_destroy_snaps(snaps, B_FALSE, NULL);
	nvlist_free(snaps);

	(void) zfs_destroy(dataset);
	(void) zfs_destroy(backup);

	(void) close(tmpfd);
	(void) unlink(filepath);

	/*
	* All the unused slots should yield ZFS_ERR_IOC_CMD_UNAVAIL
	*/
	for (int i = 0; i < ARRAY_SIZE(ioc_skip); i++) {
	if (ioc_tested[ioc_skip[i] - ZFS_IOC_FIRST])
	(void) fprintf(stderr, "cmd %d tested, not skipped!\n",
	(int)(ioc_skip[i] - ZFS_IOC_FIRST));

	ioc_tested[ioc_skip[i] - ZFS_IOC_FIRST] = B_TRUE;
	}

	(void) strlcpy(zc.zc_name, pool, sizeof (zc.zc_name));
	zc.zc_name[sizeof (zc.zc_name) - 1] = '\0';

	for (unsigned ioc = ZFS_IOC_FIRST; ioc < ZFS_IOC_LAST; ioc++) {
	unsigned cmd = ioc - ZFS_IOC_FIRST;

	if (ioc_tested[cmd])
	continue;

	if (lzc_ioctl_fd(zfs_fd, ioc, &zc) != 0 &&
	errno != ZFS_ERR_IOC_CMD_UNAVAIL) {
	(void) fprintf(stderr, "cmd %d is missing a test case "
	"(%d)\n", cmd, errno);
	}
	}
	}

	enum zfs_ioc_ref {
	#ifdef __FreeBSD__
	ZFS_IOC_BASE = 0,
	#else
	ZFS_IOC_BASE = ('Z' << 8),
	#endif
	ZFS_IOC_PLATFORM_BASE = ZFS_IOC_BASE + 0x80,
	};

	/*
	* Canonical reference check of /dev/zfs ioctl numbers.
	* These cannot change and new ioctl numbers must be appended.
	*/
	static boolean_t
	validate_ioc_values(void)
	{
	boolean_t result = B_TRUE;

	#define CHECK(expr) do { \
	if (!(expr)) { \
	result = B_FALSE; \
	fprintf(stderr, "(%s) === FALSE\n", #expr); \
	} \
	} while (0)

	CHECK(ZFS_IOC_BASE + 0 == ZFS_IOC_POOL_CREATE);
	CHECK(ZFS_IOC_BASE + 1 == ZFS_IOC_POOL_DESTROY);
	CHECK(ZFS_IOC_BASE + 2 == ZFS_IOC_POOL_IMPORT);
	CHECK(ZFS_IOC_BASE + 3 == ZFS_IOC_POOL_EXPORT);
	CHECK(ZFS_IOC_BASE + 4 == ZFS_IOC_POOL_CONFIGS);
	CHECK(ZFS_IOC_BASE + 5 == ZFS_IOC_POOL_STATS);
	CHECK(ZFS_IOC_BASE + 6 == ZFS_IOC_POOL_TRYIMPORT);
	CHECK(ZFS_IOC_BASE + 7 == ZFS_IOC_POOL_SCAN);
	CHECK(ZFS_IOC_BASE + 8 == ZFS_IOC_POOL_FREEZE);
	CHECK(ZFS_IOC_BASE + 9 == ZFS_IOC_POOL_UPGRADE);
	CHECK(ZFS_IOC_BASE + 10 == ZFS_IOC_POOL_GET_HISTORY);
	CHECK(ZFS_IOC_BASE + 11 == ZFS_IOC_VDEV_ADD);
	CHECK(ZFS_IOC_BASE + 12 == ZFS_IOC_VDEV_REMOVE);
	CHECK(ZFS_IOC_BASE + 13 == ZFS_IOC_VDEV_SET_STATE);
	CHECK(ZFS_IOC_BASE + 14 == ZFS_IOC_VDEV_ATTACH);
	CHECK(ZFS_IOC_BASE + 15 == ZFS_IOC_VDEV_DETACH);
	CHECK(ZFS_IOC_BASE + 16 == ZFS_IOC_VDEV_SETPATH);
	CHECK(ZFS_IOC_BASE + 17 == ZFS_IOC_VDEV_SETFRU);
	CHECK(ZFS_IOC_BASE + 18 == ZFS_IOC_OBJSET_STATS);
	CHECK(ZFS_IOC_BASE + 19 == ZFS_IOC_OBJSET_ZPLPROPS);
	CHECK(ZFS_IOC_BASE + 20 == ZFS_IOC_DATASET_LIST_NEXT);
	CHECK(ZFS_IOC_BASE + 21 == ZFS_IOC_SNAPSHOT_LIST_NEXT);
	CHECK(ZFS_IOC_BASE + 22 == ZFS_IOC_SET_PROP);
	CHECK(ZFS_IOC_BASE + 23 == ZFS_IOC_CREATE);
	CHECK(ZFS_IOC_BASE + 24 == ZFS_IOC_DESTROY);
	CHECK(ZFS_IOC_BASE + 25 == ZFS_IOC_ROLLBACK);
	CHECK(ZFS_IOC_BASE + 26 == ZFS_IOC_RENAME);
	CHECK(ZFS_IOC_BASE + 27 == ZFS_IOC_RECV);
	CHECK(ZFS_IOC_BASE + 28 == ZFS_IOC_SEND);
	CHECK(ZFS_IOC_BASE + 29 == ZFS_IOC_INJECT_FAULT);
	CHECK(ZFS_IOC_BASE + 30 == ZFS_IOC_CLEAR_FAULT);
	CHECK(ZFS_IOC_BASE + 31 == ZFS_IOC_INJECT_LIST_NEXT);
	CHECK(ZFS_IOC_BASE + 32 == ZFS_IOC_ERROR_LOG);
	CHECK(ZFS_IOC_BASE + 33 == ZFS_IOC_CLEAR);
	CHECK(ZFS_IOC_BASE + 34 == ZFS_IOC_PROMOTE);
	CHECK(ZFS_IOC_BASE + 35 == ZFS_IOC_SNAPSHOT);
	CHECK(ZFS_IOC_BASE + 36 == ZFS_IOC_DSOBJ_TO_DSNAME);
	CHECK(ZFS_IOC_BASE + 37 == ZFS_IOC_OBJ_TO_PATH);
	CHECK(ZFS_IOC_BASE + 38 == ZFS_IOC_POOL_SET_PROPS);
	CHECK(ZFS_IOC_BASE + 39 == ZFS_IOC_POOL_GET_PROPS);
	CHECK(ZFS_IOC_BASE + 40 == ZFS_IOC_SET_FSACL);
	CHECK(ZFS_IOC_BASE + 41 == ZFS_IOC_GET_FSACL);
	CHECK(ZFS_IOC_BASE + 42 == ZFS_IOC_SHARE);
	CHECK(ZFS_IOC_BASE + 43 == ZFS_IOC_INHERIT_PROP);
	CHECK(ZFS_IOC_BASE + 44 == ZFS_IOC_SMB_ACL);
	CHECK(ZFS_IOC_BASE + 45 == ZFS_IOC_USERSPACE_ONE);
	CHECK(ZFS_IOC_BASE + 46 == ZFS_IOC_USERSPACE_MANY);
	CHECK(ZFS_IOC_BASE + 47 == ZFS_IOC_USERSPACE_UPGRADE);
	CHECK(ZFS_IOC_BASE + 48 == ZFS_IOC_HOLD);
	CHECK(ZFS_IOC_BASE + 49 == ZFS_IOC_RELEASE);
	CHECK(ZFS_IOC_BASE + 50 == ZFS_IOC_GET_HOLDS);
	CHECK(ZFS_IOC_BASE + 51 == ZFS_IOC_OBJSET_RECVD_PROPS);
	CHECK(ZFS_IOC_BASE + 52 == ZFS_IOC_VDEV_SPLIT);
	CHECK(ZFS_IOC_BASE + 53 == ZFS_IOC_NEXT_OBJ);
	CHECK(ZFS_IOC_BASE + 54 == ZFS_IOC_DIFF);
	CHECK(ZFS_IOC_BASE + 55 == ZFS_IOC_TMP_SNAPSHOT);
	CHECK(ZFS_IOC_BASE + 56 == ZFS_IOC_OBJ_TO_STATS);
	CHECK(ZFS_IOC_BASE + 57 == ZFS_IOC_SPACE_WRITTEN);
	CHECK(ZFS_IOC_BASE + 58 == ZFS_IOC_SPACE_SNAPS);
	CHECK(ZFS_IOC_BASE + 59 == ZFS_IOC_DESTROY_SNAPS);
	CHECK(ZFS_IOC_BASE + 60 == ZFS_IOC_POOL_REGUID);
	CHECK(ZFS_IOC_BASE + 61 == ZFS_IOC_POOL_REOPEN);
	CHECK(ZFS_IOC_BASE + 62 == ZFS_IOC_SEND_PROGRESS);
	CHECK(ZFS_IOC_BASE + 63 == ZFS_IOC_LOG_HISTORY);
	CHECK(ZFS_IOC_BASE + 64 == ZFS_IOC_SEND_NEW);
	CHECK(ZFS_IOC_BASE + 65 == ZFS_IOC_SEND_SPACE);
	CHECK(ZFS_IOC_BASE + 66 == ZFS_IOC_CLONE);
	CHECK(ZFS_IOC_BASE + 67 == ZFS_IOC_BOOKMARK);
	CHECK(ZFS_IOC_BASE + 68 == ZFS_IOC_GET_BOOKMARKS);
	CHECK(ZFS_IOC_BASE + 69 == ZFS_IOC_DESTROY_BOOKMARKS);
	CHECK(ZFS_IOC_BASE + 70 == ZFS_IOC_RECV_NEW);
	CHECK(ZFS_IOC_BASE + 71 == ZFS_IOC_POOL_SYNC);
	CHECK(ZFS_IOC_BASE + 72 == ZFS_IOC_CHANNEL_PROGRAM);
	CHECK(ZFS_IOC_BASE + 73 == ZFS_IOC_LOAD_KEY);
	CHECK(ZFS_IOC_BASE + 74 == ZFS_IOC_UNLOAD_KEY);
	CHECK(ZFS_IOC_BASE + 75 == ZFS_IOC_CHANGE_KEY);
	CHECK(ZFS_IOC_BASE + 76 == ZFS_IOC_REMAP);
	CHECK(ZFS_IOC_BASE + 77 == ZFS_IOC_POOL_CHECKPOINT);
	CHECK(ZFS_IOC_BASE + 78 == ZFS_IOC_POOL_DISCARD_CHECKPOINT);
	CHECK(ZFS_IOC_BASE + 79 == ZFS_IOC_POOL_INITIALIZE);
	CHECK(ZFS_IOC_BASE + 80 == ZFS_IOC_POOL_TRIM);
	CHECK(ZFS_IOC_BASE + 81 == ZFS_IOC_REDACT);
	CHECK(ZFS_IOC_BASE + 82 == ZFS_IOC_GET_BOOKMARK_PROPS);
	CHECK(ZFS_IOC_BASE + 83 == ZFS_IOC_WAIT);
	CHECK(ZFS_IOC_BASE + 84 == ZFS_IOC_WAIT_FS);
	CHECK(ZFS_IOC_PLATFORM_BASE + 1 == ZFS_IOC_EVENTS_NEXT);
	CHECK(ZFS_IOC_PLATFORM_BASE + 2 == ZFS_IOC_EVENTS_CLEAR);
	CHECK(ZFS_IOC_PLATFORM_BASE + 3 == ZFS_IOC_EVENTS_SEEK);
	CHECK(ZFS_IOC_PLATFORM_BASE + 4 == ZFS_IOC_NEXTBOOT);
	CHECK(ZFS_IOC_PLATFORM_BASE + 5 == ZFS_IOC_JAIL);
	CHECK(ZFS_IOC_PLATFORM_BASE + 6 == ZFS_IOC_UNJAIL);
	CHECK(ZFS_IOC_PLATFORM_BASE + 7 == ZFS_IOC_SET_BOOTENV);
	CHECK(ZFS_IOC_PLATFORM_BASE + 8 == ZFS_IOC_GET_BOOTENV);

	#undef CHECK

	return (result);
	}

	int
	main(int argc, const char *argv[])
	{
	if (argc != 2) {
	(void) fprintf(stderr, "usage: %s <pool>\n", argv[0]);
	exit(2);
	}

	if (!validate_ioc_values()) {
	(void) fprintf(stderr, "WARNING: zfs_ioc_t has binary "
	"incompatible command values\n");
	exit(3);
	}

	(void) libzfs_core_init();
	zfs_fd = open(ZFS_DEV, O_RDWR);
	if (zfs_fd < 0) {
	(void) fprintf(stderr, "open: %s\n", strerror(errno));
	libzfs_core_fini();
	exit(2);
	}

	zfs_ioc_input_tests(argv[1]);

	(void) close(zfs_fd);
	libzfs_core_fini();

	return (unexpected_failures);
	}
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/cmd/nvlist_to_lua/nvlist_to_lua.c b/sys/contrib/openzfs/tests/zfs-tests/cmd/nvlist_to_lua/nvlist_to_lua.c
	index b130d667f231..3072ca97939c 100644
	--- a/sys/contrib/openzfs/tests/zfs-tests/cmd/nvlist_to_lua/nvlist_to_lua.c
	+++ b/sys/contrib/openzfs/tests/zfs-tests/cmd/nvlist_to_lua/nvlist_to_lua.c
	@@ -1,305 +1,305 @@
	/*
	* CDDL HEADER START
	*
	* This file and its contents are supplied under the terms of the
	* Common Development and Distribution License ("CDDL"), version 1.0.
	* You may only use this file in accordance with the terms of version
	* 1.0 of the CDDL.
	*
	* A full copy of the text of the CDDL should have accompanied this
	* source. A copy of the CDDL is also available via the Internet at
	* http://www.illumos.org/license/CDDL.
	*
	* CDDL HEADER END
	*/

	/*
	* Copyright (c) 2016 by Delphix. All rights reserved.
	*/

	#include <stdio.h>
	#include <stdlib.h>
	#include <string.h>
	#include <strings.h>
	#include <libzfs_core.h>
	#include <sys/nvpair.h>

	nvlist_t *nvl;
	const char *pool;
	boolean_t unexpected_failures;

	static boolean_t
	nvlist_equal(nvlist_t nvla, nvlist_t nvlb)
	{
	if (fnvlist_num_pairs(nvla) != fnvlist_num_pairs(nvlb))
	return (B_FALSE);
	/*
	* The nvlists have the same number of pairs and keys are unique, so
	* if every key in A is also in B and assigned to the same value, the
	* lists are identical.
	*/
	for (nvpair_t *pair = nvlist_next_nvpair(nvla, NULL);
	pair != NULL; pair = nvlist_next_nvpair(nvla, pair)) {
	char *key = nvpair_name(pair);

	if (!nvlist_exists(nvlb, key))
	return (B_FALSE);

	if (nvpair_type(pair) !=
	nvpair_type(fnvlist_lookup_nvpair(nvlb, key)))
	return (B_FALSE);

	switch (nvpair_type(pair)) {
	case DATA_TYPE_BOOLEAN_VALUE:
	if (fnvpair_value_boolean_value(pair) !=
	fnvlist_lookup_boolean_value(nvlb, key)) {
	return (B_FALSE);
	}
	break;
	case DATA_TYPE_STRING:
	if (strcmp(fnvpair_value_string(pair),
	fnvlist_lookup_string(nvlb, key))) {
	return (B_FALSE);
	}
	break;
	case DATA_TYPE_INT64:
	if (fnvpair_value_int64(pair) !=
	fnvlist_lookup_int64(nvlb, key)) {
	return (B_FALSE);
	}
	break;
	case DATA_TYPE_NVLIST:
	if (!nvlist_equal(fnvpair_value_nvlist(pair),
	fnvlist_lookup_nvlist(nvlb, key))) {
	return (B_FALSE);
	}
	break;
	default:
	(void) printf("Unexpected type for nvlist_equal\n");
	return (B_FALSE);
	}
	}
	return (B_TRUE);
	}

	static void
	test(const char *testname, boolean_t expect_success, boolean_t expect_match)
	{
	char *progstr = "input = ...; return {output=input}";

	nvlist_t *outnvl;

	(void) printf("\nrunning test '%s'; input:\n", testname);
	dump_nvlist(nvl, 4);

	int err = lzc_channel_program(pool, progstr,
	10 * 1000 * 1000, 10 * 1024 * 1024, nvl, &outnvl);

	(void) printf("lzc_channel_program returned %u\n", err);
	dump_nvlist(outnvl, 5);

	if (err == 0 && expect_match) {
	/*
	* Verify that outnvl is the same as input nvl, if we expect
	* them to be. The input and output will never match if the
	* input contains an array (since arrays are converted to lua
	* tables), so this is only asserted for some test cases.
	*/
	nvlist_t *real_outnvl = fnvlist_lookup_nvlist(outnvl, "return");
	real_outnvl = fnvlist_lookup_nvlist(real_outnvl, "output");
	if (!nvlist_equal(nvl, real_outnvl)) {
	unexpected_failures = B_TRUE;
	(void) printf("unexpected input/output mismatch for "
	"case: %s\n", testname);
	}
	}
	if (err != 0 && expect_success) {
	unexpected_failures = B_TRUE;
	(void) printf("unexpected FAIL of case: %s\n", testname);
	}

	fnvlist_free(nvl);
	nvl = fnvlist_alloc();
	}

	static void
	run_tests(void)
	{
	const char *key = "key";

	/* Note: maximum nvlist key length is 32KB */
	int len = 1024 * 31;
	char *bigstring = malloc(len);
	for (int i = 0; i < len; i++)
	bigstring[i] = 'a' + i % 26;
	bigstring[len - 1] = '\0';

	nvl = fnvlist_alloc();

	fnvlist_add_boolean(nvl, key);
	test("boolean", B_TRUE, B_FALSE);

	fnvlist_add_boolean_value(nvl, key, B_TRUE);
	test("boolean_value", B_FALSE, B_FALSE);

	fnvlist_add_byte(nvl, key, 1);
	test("byte", B_FALSE, B_FALSE);

	fnvlist_add_int8(nvl, key, 1);
	test("int8", B_FALSE, B_FALSE);

	fnvlist_add_uint8(nvl, key, 1);
	test("uint8", B_FALSE, B_FALSE);

	fnvlist_add_int16(nvl, key, 1);
	test("int16", B_FALSE, B_FALSE);

	fnvlist_add_uint16(nvl, key, 1);
	test("uint16", B_FALSE, B_FALSE);

	fnvlist_add_int32(nvl, key, 1);
	test("int32", B_FALSE, B_FALSE);

	fnvlist_add_uint32(nvl, key, 1);
	test("uint32", B_FALSE, B_FALSE);

	fnvlist_add_int64(nvl, key, 1);
	test("int64", B_TRUE, B_TRUE);

	fnvlist_add_uint64(nvl, key, 1);
	test("uint64", B_FALSE, B_FALSE);

	fnvlist_add_string(nvl, key, "1");
	test("string", B_TRUE, B_TRUE);


	{
	nvlist_t *val = fnvlist_alloc();
	fnvlist_add_string(val, "subkey", "subvalue");
	fnvlist_add_nvlist(nvl, key, val);
	fnvlist_free(val);
	test("nvlist", B_TRUE, B_TRUE);
	}
	{
	boolean_t val[2] = { B_FALSE, B_TRUE };
	fnvlist_add_boolean_array(nvl, key, val, 2);
	test("boolean_array", B_FALSE, B_FALSE);
	}
	{
	uchar_t val[2] = { 0, 1 };
	fnvlist_add_byte_array(nvl, key, val, 2);
	test("byte_array", B_FALSE, B_FALSE);
	}
	{
	int8_t val[2] = { 0, 1 };
	fnvlist_add_int8_array(nvl, key, val, 2);
	test("int8_array", B_FALSE, B_FALSE);
	}
	{
	uint8_t val[2] = { 0, 1 };
	fnvlist_add_uint8_array(nvl, key, val, 2);
	test("uint8_array", B_FALSE, B_FALSE);
	}
	{
	int16_t val[2] = { 0, 1 };
	fnvlist_add_int16_array(nvl, key, val, 2);
	test("int16_array", B_FALSE, B_FALSE);
	}
	{
	uint16_t val[2] = { 0, 1 };
	fnvlist_add_uint16_array(nvl, key, val, 2);
	test("uint16_array", B_FALSE, B_FALSE);
	}
	{
	int32_t val[2] = { 0, 1 };
	fnvlist_add_int32_array(nvl, key, val, 2);
	test("int32_array", B_FALSE, B_FALSE);
	}
	{
	uint32_t val[2] = { 0, 1 };
	fnvlist_add_uint32_array(nvl, key, val, 2);
	test("uint32_array", B_FALSE, B_FALSE);
	}
	{
	int64_t val[2] = { 0, 1 };
	fnvlist_add_int64_array(nvl, key, val, 2);
	test("int64_array", B_TRUE, B_FALSE);
	}
	{
	uint64_t val[2] = { 0, 1 };
	fnvlist_add_uint64_array(nvl, key, val, 2);
	test("uint64_array", B_FALSE, B_FALSE);
	}
	{
	char *const val[2] = { "0", "1" };
	- fnvlist_add_string_array(nvl, key, val, 2);
	+ fnvlist_add_string_array(nvl, key, (const char **)val, 2);
	test("string_array", B_TRUE, B_FALSE);
	}
	{
	nvlist_t *val[2];
	val[0] = fnvlist_alloc();
	fnvlist_add_string(val[0], "subkey", "subvalue");
	val[1] = fnvlist_alloc();
	fnvlist_add_string(val[1], "subkey2", "subvalue2");
	- fnvlist_add_nvlist_array(nvl, key, val, 2);
	+ fnvlist_add_nvlist_array(nvl, key, (const nvlist_t **)val, 2);
	fnvlist_free(val[0]);
	fnvlist_free(val[1]);
	test("nvlist_array", B_FALSE, B_FALSE);
	}
	{
	fnvlist_add_string(nvl, bigstring, "1");
	test("large_key", B_TRUE, B_TRUE);
	}
	{
	fnvlist_add_string(nvl, key, bigstring);
	test("large_value", B_TRUE, B_TRUE);
	}
	{
	for (int i = 0; i < 1024; i++) {
	char buf[32];
	(void) snprintf(buf, sizeof (buf), "key-%u", i);
	fnvlist_add_int64(nvl, buf, i);
	}
	test("many_keys", B_TRUE, B_TRUE);
	}
	#ifndef __sparc__
	{
	for (int i = 0; i < 10; i++) {
	nvlist_t *newval = fnvlist_alloc();
	fnvlist_add_nvlist(newval, "key", nvl);
	fnvlist_free(nvl);
	nvl = newval;
	}
	test("deeply_nested_pos", B_TRUE, B_TRUE);
	}
	{
	for (int i = 0; i < 90; i++) {
	nvlist_t *newval = fnvlist_alloc();
	fnvlist_add_nvlist(newval, "key", nvl);
	fnvlist_free(nvl);
	nvl = newval;
	}
	test("deeply_nested_neg", B_FALSE, B_FALSE);
	}
	#endif
	free(bigstring);
	fnvlist_free(nvl);
	}

	int
	main(int argc, const char *argv[])
	{
	(void) libzfs_core_init();

	if (argc != 2) {
	(void) printf("usage: %s <pool>\n",
	argv[0]);
	exit(2);
	}
	pool = argv[1];

	run_tests();

	libzfs_core_fini();
	return (unexpected_failures);
	}
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/Makefile.am b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/Makefile.am
	index db90e058559d..bfb01dcb8f86 100644
	--- a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/Makefile.am
	+++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/Makefile.am
	@@ -1,17 +1,18 @@
	include $(top_srcdir)/config/Rules.am

	pkgdatadir = $(datadir)/@PACKAGE@/zfs-tests/tests/functional/cli_root/zfs_diff

	dist_pkgdata_SCRIPTS = \
	cleanup.ksh \
	setup.ksh \
	zfs_diff_changes.ksh \
	zfs_diff_cliargs.ksh \
	zfs_diff_encrypted.ksh \
	+ zfs_diff_mangle.ksh \
	zfs_diff_timestamp.ksh \
	zfs_diff_types.ksh

	pkgexecdir = $(datadir)/@PACKAGE@/zfs-tests/tests/functional/cli_root/zfs_diff

	pkgexec_PROGRAMS = socket
	socket_SOURCES = socket.c
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/socket.c b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/socket.c
	index a8c814e7b5b6..be4bf31dde9f 100644
	--- a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/socket.c
	+++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/socket.c
	@@ -1,58 +1,57 @@
	/*
	* This file and its contents are supplied under the terms of the
	* Common Development and Distribution License ("CDDL"), version 1.0.
	* You may only use this file in accordance with the terms of version
	* 1.0 of the CDDL.
	*
	* A full copy of the text of the CDDL should have accompanied this
	* source. A copy of the CDDL is also available via the Internet at
	* http://www.illumos.org/license/CDDL.
	*/

	/*
	* Copyright 2017, loli10K <ezomori.nozomu@gmail.com>. All rights reserved.
	*/

	#include <fcntl.h>
	#include <sys/un.h>
	#include <sys/socket.h>
	#include <sys/stat.h>
	#include <sys/types.h>
	#include <unistd.h>
	#include <errno.h>
	#include <stdio.h>
	#include <stdlib.h>
	#include <string.h>

	-/* ARGSUSED */
	int
	main(int argc, char *argv[])
	{
	struct sockaddr_un sock;
	int fd;
	char *path;
	size_t size;
	if (argc != 2) {
	fprintf(stderr, "usage: %s /path/to/socket\n", argv[0]);
	exit(1);
	}
	path = argv[1];
	size = sizeof (sock.sun_path);
	strncpy(sock.sun_path, (char *)path, size - 1);
	sock.sun_path[size - 1] = '\0';

	sock.sun_family = AF_UNIX;
	if ((fd = socket(AF_UNIX, SOCK_DGRAM, 0)) == -1) {
	perror("socket");
	return (1);
	}
	if (bind(fd, (struct sockaddr *)&sock, sizeof (struct sockaddr_un))) {
	perror("bind");
	return (1);
	}
	if (close(fd)) {
	perror("close");
	return (1);
	}
	return (0);
	}
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/zfs_diff_cliargs.ksh b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/zfs_diff_cliargs.ksh
	index 7063bbe9ce6a..67eb18fa4a5d 100755
	--- a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/zfs_diff_cliargs.ksh
	+++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/zfs_diff_cliargs.ksh
	@@ -1,78 +1,78 @@
	#!/bin/ksh -p
	#
	# This file and its contents are supplied under the terms of the
	# Common Development and Distribution License ("CDDL"), version 1.0.
	# You may only use this file in accordance with the terms of version
	# 1.0 of the CDDL.
	#
	# A full copy of the text of the CDDL should have accompanied this
	# source. A copy of the CDDL is also available via the Internet at
	# http://www.illumos.org/license/CDDL.
	#

	#
	# Copyright 2017, loli10K <ezomori.nozomu@gmail.com>. All rights reserved.
	#

	. $STF_SUITE/include/libtest.shlib

	#
	# DESCRIPTION:
	# 'zfs diff' should only work with supported options.
	#
	# STRATEGY:
	# 1. Create two snapshots
	# 2. Verify every supported option is accepted
	# 3. Verify supported options raise an error with unsupported arguments
	# 4. Verify other unsupported options raise an error
	#

	verify_runnable "both"

	function cleanup
	{
	for snap in $TESTSNAP1 $TESTSNAP2; do
	snapexists "$snap" && destroy_dataset "$snap"
	done
	}

	log_assert "'zfs diff' should only work with supported options."
	log_onexit cleanup

	-typeset goodopts=("" "-F" "-H" "-t" "-FH" "-Ft" "-Ht" "-FHt")
	-typeset badopts=("-f" "-h" "-h" "-T" "-Fx" "-Ho" "-tT" "-")
	+typeset goodopts=("" "-h" "-t" "-th" "-H" "-Hh" "-Ht" "-Hth" "-F" "-Fh" "-Ft" "-Fth" "-FH" "-FHh" "-FHt" "-FHth")
	+typeset badopts=("-f" "-T" "-Fx" "-Ho" "-tT" "-")

	DATASET="$TESTPOOL/$TESTFS"
	TESTSNAP1="$DATASET@snap1"
	TESTSNAP2="$DATASET@snap2"

	# 1. Create two snapshots
	log_must zfs snapshot "$TESTSNAP1"
	log_must zfs snapshot "$TESTSNAP2"

	# 2. Verify every supported option is accepted
	for opt in ${goodopts[@]}
	do
	log_must zfs diff $opt "$TESTSNAP1"
	log_must zfs diff $opt "$TESTSNAP1" "$DATASET"
	log_must zfs diff $opt "$TESTSNAP1" "$TESTSNAP2"
	done

	# 3. Verify supported options raise an error with unsupported arguments
	for opt in ${goodopts[@]}
	do
	log_mustnot zfs diff $opt
	log_mustnot zfs diff $opt "$DATASET"
	log_mustnot zfs diff $opt "$DATASET@noexists"
	log_mustnot zfs diff $opt "$DATASET" "$TESTSNAP1"
	log_mustnot zfs diff $opt "$TESTSNAP2" "$TESTSNAP1"
	done

	# 4. Verify other unsupported options raise an error
	for opt in ${badopts[@]}
	do
	log_mustnot zfs diff $opt "$TESTSNAP1" "$DATASET"
	log_mustnot zfs diff $opt "$TESTSNAP1" "$TESTSNAP2"
	done

	log_pass "'zfs diff' only works with supported options."
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/zfs_diff_mangle.ksh b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/zfs_diff_mangle.ksh
	new file mode 100755
	index 000000000000..ffce9f06848f
	--- /dev/null
	+++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/cli_root/zfs_diff/zfs_diff_mangle.ksh
	@@ -0,0 +1,48 @@
	+#!/bin/ksh -p
	+#
	+# This file and its contents are supplied under the terms of the
	+# Common Development and Distribution License ("CDDL"), version 1.0.
	+# You may only use this file in accordance with the terms of version
	+# 1.0 of the CDDL.
	+#
	+# A full copy of the text of the CDDL should have accompanied this
	+# source. A copy of the CDDL is also available via the Internet at
	+# http://www.illumos.org/license/CDDL.
	+#
	+
	+. $STF_SUITE/include/libtest.shlib
	+
	+#
	+# DESCRIPTION:
	+# 'zfs diff' escapes filenames as expected, 'zfs diff -h' doesn't
	+#
	+# STRATEGY:
	+# 1. Prepare a dataset
	+# 2. Create some files
	+# 3. verify 'zfs diff' mangles them and 'zfs diff -h' doesn't
	+#
	+
	+verify_runnable "both"
	+
	+function cleanup
	+{
	+ log_must zfs destroy -r "$DATASET"
	+}
	+
	+log_assert "'zfs diff' mangles filenames, 'zfs diff -h' doesn't"
	+log_onexit cleanup
	+
	+DATASET="$TESTPOOL/$TESTFS/fs"
	+TESTSNAP1="$DATASET@snap1"
	+
	+# 1. Prepare a dataset
	+log_must zfs create "$DATASET"
	+MNTPOINT="$(get_prop mountpoint "$DATASET")"
	+log_must zfs snapshot "$TESTSNAP1"
	+
	+printf '%c\t'"$MNTPOINT/"'%s\n' M '' + 'śmieszny żupan' + 'достопримечательности' \| sort > "$MNTPOINT/śmieszny żupan"
	+printf '%c\t'"$MNTPOINT/"'%s\n' M '' + '\0305\0233mieszny\0040\0305\0274upan' + '\0320\0264\0320\0276\0321\0201\0321\0202\0320\0276\0320\0277\0321\0200\0320\0270\0320\0274\0320\0265\0321\0207\0320\0260\0321\0202\0320\0265\0320\0273\0321\0214\0320\0275\0320\0276\0321\0201\0321\0202\0320\0270' \| sort > "$MNTPOINT/достопримечательности"
	+log_must diff -u <(zfs diff -h "$TESTSNAP1" \| grep -vF '<xattrdir>' \| sort) "$MNTPOINT/śmieszny żupan"
	+log_must diff -u <(zfs diff "$TESTSNAP1" \| grep -vF '<xattrdir>' \| sort) "$MNTPOINT/достопримечательности"
	+
	+log_pass "'zfs diff' mangles filenames, 'zfs diff -h' doesn't"
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/Makefile.am b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/Makefile.am
	index 4d9ae17084bc..be881faccf3b 100644
	--- a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/Makefile.am
	+++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/Makefile.am
	@@ -1,7 +1,8 @@
	pkgdatadir = $(datadir)/@PACKAGE@/zfs-tests/tests/functional/pam
	dist_pkgdata_SCRIPTS = \
	setup.ksh \
	cleanup.ksh \
	pam_basic.ksh \
	pam_nounmount.ksh \
	+ pam_short_password.ksh \
	utilities.kshlib
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/cleanup.ksh b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/cleanup.ksh
	index 62131c6d68a7..e41622d771b4 100755
	--- a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/cleanup.ksh
	+++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/cleanup.ksh
	@@ -1,32 +1,33 @@
	#!/bin/ksh -p
	#
	# CDDL HEADER START
	#
	# The contents of this file are subject to the terms of the
	# Common Development and Distribution License (the "License").
	# You may not use this file except in compliance with the License.
	#
	# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	# or http://www.opensolaris.org/os/licensing.
	# See the License for the specific language governing permissions
	# and limitations under the License.
	#
	# When distributing Covered Code, include this CDDL HEADER in each
	# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	# If applicable, add the following below this CDDL HEADER, with the
	# fields enclosed by brackets "[]" replaced with your own identifying
	# information: Portions Copyright [yyyy] [name of copyright owner]
	#
	# CDDL HEADER END
	#

	. $STF_SUITE/tests/functional/pam/utilities.kshlib

	+rmconfig
	destroy_pool $TESTPOOL
	del_user ${username}
	del_group pamtestgroup

	rm -rf "$runstatedir"
	for dir in $TESTDIRS; do
	rm -rf $dir
	done
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/pam_basic.ksh b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/pam_basic.ksh
	index 96ac59453694..f146a6e5f157 100755
	--- a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/pam_basic.ksh
	+++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/pam_basic.ksh
	@@ -1,49 +1,49 @@
	#!/bin/ksh -p
	#
	# CDDL HEADER START
	#
	# The contents of this file are subject to the terms of the
	# Common Development and Distribution License (the "License").
	# You may not use this file except in compliance with the License.
	#
	# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	# or http://www.opensolaris.org/os/licensing.
	# See the License for the specific language governing permissions
	# and limitations under the License.
	#
	# When distributing Covered Code, include this CDDL HEADER in each
	# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	# If applicable, add the following below this CDDL HEADER, with the
	# fields enclosed by brackets "[]" replaced with your own identifying
	# information: Portions Copyright [yyyy] [name of copyright owner]
	#
	# CDDL HEADER END
	#

	. $STF_SUITE/tests/functional/pam/utilities.kshlib

	log_mustnot ismounted "$TESTPOOL/pam/${username}"
	keystatus unavailable

	genconfig "homes=$TESTPOOL/pam runstatedir=${runstatedir}"
	-echo "testpass" \| pamtester pam_zfs_key_test ${username} open_session
	+echo "testpass" \| pamtester ${pamservice} ${username} open_session
	references 1
	log_must ismounted "$TESTPOOL/pam/${username}"
	keystatus available

	-echo "testpass" \| pamtester pam_zfs_key_test ${username} open_session
	+echo "testpass" \| pamtester ${pamservice} ${username} open_session
	references 2
	log_must ismounted "$TESTPOOL/pam/${username}"
	keystatus available

	-log_must pamtester pam_zfs_key_test ${username} close_session
	+log_must pamtester ${pamservice} ${username} close_session
	references 1
	log_must ismounted "$TESTPOOL/pam/${username}"
	keystatus available

	-log_must pamtester pam_zfs_key_test ${username} close_session
	+log_must pamtester ${pamservice} ${username} close_session
	references 0
	log_mustnot ismounted "$TESTPOOL/pam/${username}"
	keystatus unavailable

	log_pass "done."
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/pam_nounmount.ksh b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/pam_nounmount.ksh
	index 8179f398dfaf..eb9976f2f776 100755
	--- a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/pam_nounmount.ksh
	+++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/pam_nounmount.ksh
	@@ -1,51 +1,51 @@
	#!/bin/ksh -p
	#
	# CDDL HEADER START
	#
	# The contents of this file are subject to the terms of the
	# Common Development and Distribution License (the "License").
	# You may not use this file except in compliance with the License.
	#
	# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	# or http://www.opensolaris.org/os/licensing.
	# See the License for the specific language governing permissions
	# and limitations under the License.
	#
	# When distributing Covered Code, include this CDDL HEADER in each
	# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	# If applicable, add the following below this CDDL HEADER, with the
	# fields enclosed by brackets "[]" replaced with your own identifying
	# information: Portions Copyright [yyyy] [name of copyright owner]
	#
	# CDDL HEADER END
	#

	. $STF_SUITE/tests/functional/pam/utilities.kshlib

	log_mustnot ismounted "$TESTPOOL/pam/${username}"
	keystatus unavailable

	genconfig "homes=$TESTPOOL/pam runstatedir=${runstatedir} nounmount"
	-echo "testpass" \| pamtester pam_zfs_key_test ${username} open_session
	+echo "testpass" \| pamtester ${pamservice} ${username} open_session
	references 1
	log_must ismounted "$TESTPOOL/pam/${username}"
	keystatus available

	-echo "testpass" \| pamtester pam_zfs_key_test ${username} open_session
	+echo "testpass" \| pamtester ${pamservice} ${username} open_session
	references 2
	keystatus available
	log_must ismounted "$TESTPOOL/pam/${username}"

	-log_must pamtester pam_zfs_key_test ${username} close_session
	+log_must pamtester ${pamservice} ${username} close_session
	references 1
	keystatus available
	log_must ismounted "$TESTPOOL/pam/${username}"

	-log_must pamtester pam_zfs_key_test ${username} close_session
	+log_must pamtester ${pamservice} ${username} close_session
	references 0
	keystatus available
	log_must ismounted "$TESTPOOL/pam/${username}"
	log_must zfs unmount "$TESTPOOL/pam/${username}"
	log_must zfs unload-key "$TESTPOOL/pam/${username}"

	log_pass "done."
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/pam_short_password.ksh b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/pam_short_password.ksh
	new file mode 100755
	index 000000000000..1f72c9468a46
	--- /dev/null
	+++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/pam_short_password.ksh
	@@ -0,0 +1,84 @@
	+#!/bin/ksh -p
	+#
	+# CDDL HEADER START
	+#
	+# The contents of this file are subject to the terms of the
	+# Common Development and Distribution License (the "License").
	+# You may not use this file except in compliance with the License.
	+#
	+# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	+# or http://www.opensolaris.org/os/licensing.
	+# See the License for the specific language governing permissions
	+# and limitations under the License.
	+#
	+# When distributing Covered Code, include this CDDL HEADER in each
	+# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	+# If applicable, add the following below this CDDL HEADER, with the
	+# fields enclosed by brackets "[]" replaced with your own identifying
	+# information: Portions Copyright [yyyy] [name of copyright owner]
	+#
	+# CDDL HEADER END
	+#
	+
	+#
	+# Copyright 2021 Attila Fülöp <attila@fueloep.org>
	+#
	+
	+
	+. $STF_SUITE/tests/functional/pam/utilities.kshlib
	+
	+if [[ -z pamservice ]]; then
	+ pamservice=pam_zfs_key_test
	+fi
	+
	+# DESCRIPTION:
	+# If we set the encryption passphrase for a dataset via pam_zfs_key, a minimal
	+# passphrase length isn't enforced. This leads to a non-loadable key if
	+# `zfs load-key` enforces a minimal length. Make sure this isn't the case.
	+
	+log_mustnot ismounted "$TESTPOOL/pam/${username}"
	+keystatus unavailable
	+
	+genconfig "homes=$TESTPOOL/pam runstatedir=${runstatedir}"
	+
	+# Load keys and mount userdir.
	+echo "testpass" \| pamtester ${pamservice} ${username} open_session
	+references 1
	+log_must ismounted "$TESTPOOL/pam/${username}"
	+keystatus available
	+
	+# Change user and dataset password to short one.
	+printf "short\nshort\n" \| pamtester ${pamservice} ${username} chauthtok
	+
	+# Unmount and unload key.
	+log_must pamtester ${pamservice} ${username} close_session
	+references 0
	+log_mustnot ismounted "$TESTPOOL/pam/${username}"
	+keystatus unavailable
	+
	+# Check if password change succeeded.
	+echo "testpass" \| pamtester ${pamservice} ${username} open_session
	+references 1
	+log_mustnot ismounted "$TESTPOOL/pam/${username}"
	+keystatus unavailable
	+log_must pamtester ${pamservice} ${username} close_session
	+references 0
	+
	+echo "short" \| pamtester ${pamservice} ${username} open_session
	+references 1
	+log_must ismounted "$TESTPOOL/pam/${username}"
	+keystatus available
	+
	+
	+# Finally check if `zfs load-key` succeeds with the short password.
	+log_must pamtester ${pamservice} ${username} close_session
	+references 0
	+log_mustnot ismounted "$TESTPOOL/pam/${username}"
	+keystatus unavailable
	+
	+echo "short" \| zfs load-key "$TESTPOOL/pam/${username}"
	+keystatus available
	+zfs unload-key "$TESTPOOL/pam/${username}"
	+keystatus unavailable
	+
	+log_pass "done."
	diff --git a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/utilities.kshlib b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/utilities.kshlib
	index ef80f5a4f12e..d328034300b1 100644
	--- a/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/utilities.kshlib
	+++ b/sys/contrib/openzfs/tests/zfs-tests/tests/functional/pam/utilities.kshlib
	@@ -1,40 +1,47 @@
	#!/bin/ksh -p
	#
	# CDDL HEADER START
	#
	# The contents of this file are subject to the terms of the
	# Common Development and Distribution License (the "License").
	# You may not use this file except in compliance with the License.
	#
	# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	# or http://www.opensolaris.org/os/licensing.
	# See the License for the specific language governing permissions
	# and limitations under the License.
	#
	# When distributing Covered Code, include this CDDL HEADER in each
	# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	# If applicable, add the following below this CDDL HEADER, with the
	# fields enclosed by brackets "[]" replaced with your own identifying
	# information: Portions Copyright [yyyy] [name of copyright owner]
	#
	# CDDL HEADER END
	#

	. $STF_SUITE/include/libtest.shlib

	username="pamtestuser"
	runstatedir="${TESTDIR}_run"
	+pamservice="pam_zfs_key_test"
	+pamconfig="/etc/pam.d/${pamservice}"
	+
	function keystatus {
	log_must [ "$(zfs list -Ho keystatus "$TESTPOOL/pam/${username}")" == "$1" ]
	}

	function genconfig {
	for i in password auth session; do
	printf "%s\trequired\tpam_permit.so\n%s\toptional\tpam_zfs_key.so\t%s\n" "$i" "$i" "$1"
	- done > /etc/pam.d/pam_zfs_key_test
	+ done > "${pamconfig}"
	+}
	+
	+function rmconfig {
	+ log_must rm "${pamconfig}"
	}

	function references {
	log_must [ "$(cat "${runstatedir}/$(id -u ${username})")" == "$1" ]
	}

	diff --git a/sys/modules/zfs/zfs_config.h b/sys/modules/zfs/zfs_config.h
	index a680b5f49c92..eec0dc9bb9d0 100644
	--- a/sys/modules/zfs/zfs_config.h
	+++ b/sys/modules/zfs/zfs_config.h
	@@ -1,864 +1,885 @@
	/*
	* $FreeBSD$
	*/

	/* zfs_config.h. Generated from zfs_config.h.in by configure. */
	/* zfs_config.h.in. Generated from configure.ac by autoheader. */

	/* Define to 1 if translation of program messages to the user's native
	language is requested. */
	/* #undef ENABLE_NLS */

	/* bio_end_io_t wants 1 arg */
	/* #undef HAVE_1ARG_BIO_END_IO_T */

	/* lookup_bdev() wants 1 arg */
	/* #undef HAVE_1ARG_LOOKUP_BDEV */

	/* submit_bio() wants 1 arg */
	/* #undef HAVE_1ARG_SUBMIT_BIO */

	/* bdi_setup_and_register() wants 2 args */
	/* #undef HAVE_2ARGS_BDI_SETUP_AND_REGISTER */

	/* vfs_getattr wants 2 args */
	/* #undef HAVE_2ARGS_VFS_GETATTR */

	/* zlib_deflate_workspacesize() wants 2 args */
	/* #undef HAVE_2ARGS_ZLIB_DEFLATE_WORKSPACESIZE */

	/* bdi_setup_and_register() wants 3 args */
	/* #undef HAVE_3ARGS_BDI_SETUP_AND_REGISTER */

	/* vfs_getattr wants 3 args */
	/* #undef HAVE_3ARGS_VFS_GETATTR */

	/* vfs_getattr wants 4 args */
	/* #undef HAVE_4ARGS_VFS_GETATTR */

	/* kernel has access_ok with 'type' parameter */
	/* #undef HAVE_ACCESS_OK_TYPE */

	/* posix_acl has refcount_t */
	/* #undef HAVE_ACL_REFCOUNT */

	/* Define if host toolchain supports AES */
	#define HAVE_AES 1

	#ifdef __amd64__
	#ifndef RESCUE
	/* Define if host toolchain supports AVX */
	#define HAVE_AVX 1
	#endif

	/* Define if host toolchain supports AVX2 */
	#define HAVE_AVX2 1

	/* Define if host toolchain supports AVX512BW */
	#define HAVE_AVX512BW 1

	/* Define if host toolchain supports AVX512CD */
	#define HAVE_AVX512CD 1

	/* Define if host toolchain supports AVX512DQ */
	#define HAVE_AVX512DQ 1

	/* Define if host toolchain supports AVX512ER */
	#define HAVE_AVX512ER 1

	/* Define if host toolchain supports AVX512F */
	#define HAVE_AVX512F 1

	/* Define if host toolchain supports AVX512IFMA */
	#define HAVE_AVX512IFMA 1

	/* Define if host toolchain supports AVX512PF */
	#define HAVE_AVX512PF 1

	/* Define if host toolchain supports AVX512VBMI */
	#define HAVE_AVX512VBMI 1

	/* Define if host toolchain supports AVX512VL */
	#define HAVE_AVX512VL 1
	#endif

	/* bdev_check_media_change() exists */
	/* #undef HAVE_BDEV_CHECK_MEDIA_CHANGE */

	+/* block_device_operations->submit_bio() returns void */
	+/* #undef HAVE_BDEV_SUBMIT_BIO_RETURNS_VOID */
	+
	/* bdev_whole() is available */
	/* #undef HAVE_BDEV_WHOLE */

	/* bio->bi_bdev->bd_disk exists */
	/* #undef HAVE_BIO_BDEV_DISK */

	/* bio->bi_opf is defined */
	/* #undef HAVE_BIO_BI_OPF */

	/* bio->bi_status exists */
	/* #undef HAVE_BIO_BI_STATUS */

	/* bio has bi_iter */
	/* #undef HAVE_BIO_BVEC_ITER */

	/* bio__io_acct() available /
	/* #undef HAVE_BIO_IO_ACCT */

	/* bio_max_segs() is implemented */
	/* #undef HAVE_BIO_MAX_SEGS */

	/* bio_set_dev() is available */
	/* #undef HAVE_BIO_SET_DEV */

	/* bio_set_dev() GPL-only */
	/* #undef HAVE_BIO_SET_DEV_GPL_ONLY */

	+/* bio_set_dev() is a macro */
	+/* #undef HAVE_BIO_SET_DEV_MACRO */
	+
	/* bio_set_op_attrs is available */
	/* #undef HAVE_BIO_SET_OP_ATTRS */

	+/* blkdev_get_by_path() handles ERESTARTSYS */
	+/* #undef HAVE_BLKDEV_GET_ERESTARTSYS */
	+
	/* blkdev_reread_part() exists */
	/* #undef HAVE_BLKDEV_REREAD_PART */

	/* blkg_tryget() is available */
	/* #undef HAVE_BLKG_TRYGET */

	/* blkg_tryget() GPL-only */
	/* #undef HAVE_BLKG_TRYGET_GPL_ONLY */

	/* blk_alloc_disk() exists */
	/* #undef HAVE_BLK_ALLOC_DISK */

	/* blk_alloc_queue() expects request function */
	/* #undef HAVE_BLK_ALLOC_QUEUE_REQUEST_FN */

	/* blk_alloc_queue_rh() expects request function */
	/* #undef HAVE_BLK_ALLOC_QUEUE_REQUEST_FN_RH */

	/* blk queue backing_dev_info is dynamic */
	/* #undef HAVE_BLK_QUEUE_BDI_DYNAMIC */

	/* blk_queue_flag_clear() exists */
	/* #undef HAVE_BLK_QUEUE_FLAG_CLEAR */

	/* blk_queue_flag_set() exists */
	/* #undef HAVE_BLK_QUEUE_FLAG_SET */

	/* blk_queue_flush() is available */
	/* #undef HAVE_BLK_QUEUE_FLUSH */

	/* blk_queue_flush() is GPL-only */
	/* #undef HAVE_BLK_QUEUE_FLUSH_GPL_ONLY */

	/* blk_queue_secdiscard() is available */
	/* #undef HAVE_BLK_QUEUE_SECDISCARD */

	/* blk_queue_secure_erase() is available */
	/* #undef HAVE_BLK_QUEUE_SECURE_ERASE */

	/* blk_queue_update_readahead() exists */
	/* #undef HAVE_BLK_QUEUE_UPDATE_READAHEAD */

	/* blk_queue_write_cache() exists */
	/* #undef HAVE_BLK_QUEUE_WRITE_CACHE */

	/* blk_queue_write_cache() is GPL-only */
	/* #undef HAVE_BLK_QUEUE_WRITE_CACHE_GPL_ONLY */

	/* Define if revalidate_disk() in block_device_operations */
	/* #undef HAVE_BLOCK_DEVICE_OPERATIONS_REVALIDATE_DISK */

	/* Define to 1 if you have the Mac OS X function CFLocaleCopyCurrent in the
	CoreFoundation framework. */
	/* #undef HAVE_CFLOCALECOPYCURRENT */

	/* Define to 1 if you have the Mac OS X function
	CFLocaleCopyPreferredLanguages in the CoreFoundation framework. */
	/* #undef HAVE_CFLOCALECOPYPREFERREDLANGUAGES */

	/* Define to 1 if you have the Mac OS X function CFPreferencesCopyAppValue in
	the CoreFoundation framework. */
	/* #undef HAVE_CFPREFERENCESCOPYAPPVALUE */

	/* check_disk_change() exists */
	/* #undef HAVE_CHECK_DISK_CHANGE */

	/* clear_inode() is available */
	/* #undef HAVE_CLEAR_INODE */

	/* dentry uses const struct dentry_operations */
	/* #undef HAVE_CONST_DENTRY_OPERATIONS */

	/* copy_from_iter() is available */
	/* #undef HAVE_COPY_FROM_ITER */

	/* copy_to_iter() is available */
	/* #undef HAVE_COPY_TO_ITER */

	/* yes */
	/* #undef HAVE_CPU_HOTPLUG */

	/* current_time() exists */
	/* #undef HAVE_CURRENT_TIME */

	/* Define if the GNU dcgettext() function is already present or preinstalled.
	*/
	/* #undef HAVE_DCGETTEXT */

	/* DECLARE_EVENT_CLASS() is available */
	/* #undef HAVE_DECLARE_EVENT_CLASS */

	/* lookup_bdev() wants dev_t arg */
	/* #undef HAVE_DEVT_LOOKUP_BDEV */

	/* sops->dirty_inode() wants flags */
	/* #undef HAVE_DIRTY_INODE_WITH_FLAGS */

	/* disk__io_acct() available /
	/* #undef HAVE_DISK_IO_ACCT */

	/* disk_update_readahead() exists */
	/* #undef HAVE_DISK_UPDATE_READAHEAD */

	/* Define to 1 if you have the <dlfcn.h> header file. */
	#define HAVE_DLFCN_H 1

	/* d_make_root() is available */
	/* #undef HAVE_D_MAKE_ROOT */

	/* d_prune_aliases() is available */
	/* #undef HAVE_D_PRUNE_ALIASES */

	/* dops->d_revalidate() operation takes nameidata */
	/* #undef HAVE_D_REVALIDATE_NAMEIDATA */

	/* eops->encode_fh() wants child and parent inodes */
	/* #undef HAVE_ENCODE_FH_WITH_INODE */

	/* sops->evict_inode() exists */
	/* #undef HAVE_EVICT_INODE */

	/* fops->aio_fsync() exists */
	/* #undef HAVE_FILE_AIO_FSYNC */

	/* file_dentry() is available */
	/* #undef HAVE_FILE_DENTRY */

	/* file_inode() is available */
	/* #undef HAVE_FILE_INODE */

	/* iops->follow_link() cookie */
	/* #undef HAVE_FOLLOW_LINK_COOKIE */

	/* iops->follow_link() nameidata */
	/* #undef HAVE_FOLLOW_LINK_NAMEIDATA */

	/* fops->fsync() with range */
	/* #undef HAVE_FSYNC_RANGE */

	/* fops->fsync() without dentry */
	/* #undef HAVE_FSYNC_WITHOUT_DENTRY */

	/* generic_fillattr requires struct user_namespace* */
	/* #undef HAVE_GENERIC_FILLATTR_USERNS */

	/* generic__io_acct() 3 arg available /
	/* #undef HAVE_GENERIC_IO_ACCT_3ARG */

	/* generic__io_acct() 4 arg available /
	/* #undef HAVE_GENERIC_IO_ACCT_4ARG */

	/* generic_readlink is global */
	/* #undef HAVE_GENERIC_READLINK */

	/* generic_setxattr() exists */
	/* #undef HAVE_GENERIC_SETXATTR */

	/* generic_write_checks() takes kiocb */
	/* #undef HAVE_GENERIC_WRITE_CHECKS_KIOCB */

	/* Define if the GNU gettext() function is already present or preinstalled. */
	/* #undef HAVE_GETTEXT */

	/* iops->get_acl() exists */
	/* #undef HAVE_GET_ACL */

	/* iops->get_acl() takes rcu */
	/* #undef HAVE_GET_ACL_RCU */

	/* iops->get_link() cookie */
	/* #undef HAVE_GET_LINK_COOKIE */

	/* iops->get_link() delayed */
	/* #undef HAVE_GET_LINK_DELAYED */

	/* group_info->gid exists */
	/* #undef HAVE_GROUP_INFO_GID */

	/* has_capability() is available */
	/* #undef HAVE_HAS_CAPABILITY */

	/* Define if you have the iconv() function and it works. */
	#define HAVE_ICONV 1

	/* yes */
	/* #undef HAVE_INODE_LOCK_SHARED */

	/* inode_owner_or_capable() exists */
	/* #undef HAVE_INODE_OWNER_OR_CAPABLE */

	/* inode_owner_or_capable() takes user_ns */
	/* #undef HAVE_INODE_OWNER_OR_CAPABLE_IDMAPPED */

	/* inode_set_flags() exists */
	/* #undef HAVE_INODE_SET_FLAGS */

	/* inode_set_iversion() exists */
	/* #undef HAVE_INODE_SET_IVERSION */

	/* inode->i_time's are timespec64 /
	/* #undef HAVE_INODE_TIMESPEC64_TIMES */

	/* timestamp_truncate() exists */
	/* #undef HAVE_INODE_TIMESTAMP_TRUNCATE */

	/* Define to 1 if you have the <inttypes.h> header file. */
	#define HAVE_INTTYPES_H 1

	/* in_compat_syscall() is available */
	/* #undef HAVE_IN_COMPAT_SYSCALL */

	/* iops->create() takes struct user_namespace* */
	/* #undef HAVE_IOPS_CREATE_USERNS */

	/* iops->mkdir() takes struct user_namespace* */
	/* #undef HAVE_IOPS_MKDIR_USERNS */

	/* iops->mknod() takes struct user_namespace* */
	/* #undef HAVE_IOPS_MKNOD_USERNS */

	/* iops->rename() takes struct user_namespace* */
	/* #undef HAVE_IOPS_RENAME_USERNS */

	/* iops->symlink() takes struct user_namespace* */
	/* #undef HAVE_IOPS_SYMLINK_USERNS */

	/* iov_iter_advance() is available */
	/* #undef HAVE_IOV_ITER_ADVANCE */

	/* iov_iter_count() is available */
	/* #undef HAVE_IOV_ITER_COUNT */

	/* iov_iter_fault_in_readable() is available */
	/* #undef HAVE_IOV_ITER_FAULT_IN_READABLE */

	/* iov_iter_revert() is available */
	/* #undef HAVE_IOV_ITER_REVERT */

	+/* iov_iter_type() is available */
	+/* #undef HAVE_IOV_ITER_TYPE */
	+
	/* iov_iter types are available */
	/* #undef HAVE_IOV_ITER_TYPES */

	/* yes */
	/* #undef HAVE_IO_SCHEDULE_TIMEOUT */

	/* Define to 1 if you have the `issetugid' function. */
	#define HAVE_ISSETUGID 1

	/* kernel has kernel_fpu_* functions */
	/* #undef HAVE_KERNEL_FPU */

	/* kernel has asm/fpu/api.h */
	/* #undef HAVE_KERNEL_FPU_API_HEADER */

	/* kernel fpu internal */
	/* #undef HAVE_KERNEL_FPU_INTERNAL */

	+/* kernel has asm/fpu/xcr.h */
	+/* #undef HAVE_KERNEL_FPU_XCR_HEADER */
	+
	/* uncached_acl_sentinel() exists */
	/* #undef HAVE_KERNEL_GET_ACL_HANDLE_CACHE */

	/* kernel does stack verification */
	/* #undef HAVE_KERNEL_OBJTOOL */

	/* kernel has linux/objtool.h */
	/* #undef HAVE_KERNEL_OBJTOOL_HEADER */

	/* kernel_read() take loff_t pointer */
	/* #undef HAVE_KERNEL_READ_PPOS */

	/* timer_list.function gets a timer_list */
	/* #undef HAVE_KERNEL_TIMER_FUNCTION_TIMER_LIST */

	/* struct timer_list has a flags member */
	/* #undef HAVE_KERNEL_TIMER_LIST_FLAGS */

	/* timer_setup() is available */
	/* #undef HAVE_KERNEL_TIMER_SETUP */

	/* kernel_write() take loff_t pointer */
	/* #undef HAVE_KERNEL_WRITE_PPOS */

	/* kmem_cache_create_usercopy() exists */
	/* #undef HAVE_KMEM_CACHE_CREATE_USERCOPY */

	/* kstrtoul() exists */
	/* #undef HAVE_KSTRTOUL */

	/* ktime_get_coarse_real_ts64() exists */
	/* #undef HAVE_KTIME_GET_COARSE_REAL_TS64 */

	/* ktime_get_raw_ts64() exists */
	/* #undef HAVE_KTIME_GET_RAW_TS64 */

	/* kvmalloc exists */
	/* #undef HAVE_KVMALLOC */

	/* Define if you have [aio] */
	/* #undef HAVE_LIBAIO */

	/* Define if you have [blkid] */
	/* #undef HAVE_LIBBLKID */

	/* Define if you have [crypto] */
	#define HAVE_LIBCRYPTO 1

	/* Define if you have [tirpc] */
	/* #undef HAVE_LIBTIRPC */

	/* Define if you have [udev] */
	/* #undef HAVE_LIBUDEV */

	/* Define if you have [uuid] */
	/* #undef HAVE_LIBUUID */

	+/* linux/blk-cgroup.h exists */
	+/* #undef HAVE_LINUX_BLK_CGROUP_HEADER */
	+
	/* lseek_execute() is available */
	/* #undef HAVE_LSEEK_EXECUTE */

	/* makedev() is declared in sys/mkdev.h */
	/* #undef HAVE_MAKEDEV_IN_MKDEV */

	/* makedev() is declared in sys/sysmacros.h */
	/* #undef HAVE_MAKEDEV_IN_SYSMACROS */

	/* Noting that make_request_fn() returns blk_qc_t */
	/* #undef HAVE_MAKE_REQUEST_FN_RET_QC */

	/* Noting that make_request_fn() returns void */
	/* #undef HAVE_MAKE_REQUEST_FN_RET_VOID */

	/* Define to 1 if you have the <memory.h> header file. */
	#define HAVE_MEMORY_H 1

	/* iops->mkdir() takes umode_t */
	/* #undef HAVE_MKDIR_UMODE_T */

	/* Define to 1 if you have the `mlockall' function. */
	#define HAVE_MLOCKALL 1

	/* lookup_bdev() wants mode arg */
	/* #undef HAVE_MODE_LOOKUP_BDEV */

	/* Define if host toolchain supports MOVBE */
	#define HAVE_MOVBE 1

	/* new_sync_read()/new_sync_write() are available */
	/* #undef HAVE_NEW_SYNC_READ */

	+/* folio_wait_bit() exists */
	+/* #undef HAVE_PAGEMAP_FOLIO_WAIT_BIT */
	+
	/* iops->getattr() takes a path */
	/* #undef HAVE_PATH_IOPS_GETATTR */

	/* Define if host toolchain supports PCLMULQDQ */
	#define HAVE_PCLMULQDQ 1

	/* percpu_counter_add_batch() is defined */
	/* #undef HAVE_PERCPU_COUNTER_ADD_BATCH */

	/* percpu_counter_init() wants gfp_t */
	/* #undef HAVE_PERCPU_COUNTER_INIT_WITH_GFP */

	/* posix_acl_chmod() exists */
	/* #undef HAVE_POSIX_ACL_CHMOD */

	/* posix_acl_from_xattr() needs user_ns */
	/* #undef HAVE_POSIX_ACL_FROM_XATTR_USERNS */

	/* posix_acl_release() is available */
	/* #undef HAVE_POSIX_ACL_RELEASE */

	/* posix_acl_release() is GPL-only */
	/* #undef HAVE_POSIX_ACL_RELEASE_GPL_ONLY */

	/* posix_acl_valid() wants user namespace */
	/* #undef HAVE_POSIX_ACL_VALID_WITH_NS */

	/* proc_ops structure exists */
	/* #undef HAVE_PROC_OPS_STRUCT */

	/* iops->put_link() cookie */
	/* #undef HAVE_PUT_LINK_COOKIE */

	/* iops->put_link() delayed */
	/* #undef HAVE_PUT_LINK_DELAYED */

	/* iops->put_link() nameidata */
	/* #undef HAVE_PUT_LINK_NAMEIDATA */

	/* If available, contains the Python version number currently in use. */
	#define HAVE_PYTHON "3.7"

	/* qat is enabled and existed */
	/* #undef HAVE_QAT */

	/* iops->rename() wants flags */
	/* #undef HAVE_RENAME_WANTS_FLAGS */

	/* REQ_DISCARD is defined */
	/* #undef HAVE_REQ_DISCARD */

	/* REQ_FLUSH is defined */
	/* #undef HAVE_REQ_FLUSH */

	/* REQ_OP_DISCARD is defined */
	/* #undef HAVE_REQ_OP_DISCARD */

	/* REQ_OP_FLUSH is defined */
	/* #undef HAVE_REQ_OP_FLUSH */

	/* REQ_OP_SECURE_ERASE is defined */
	/* #undef HAVE_REQ_OP_SECURE_ERASE */

	/* REQ_PREFLUSH is defined */
	/* #undef HAVE_REQ_PREFLUSH */

	/* revalidate_disk() is available */
	/* #undef HAVE_REVALIDATE_DISK */

	/* revalidate_disk_size() is available */
	/* #undef HAVE_REVALIDATE_DISK_SIZE */

	/* struct rw_semaphore has member activity */
	/* #undef HAVE_RWSEM_ACTIVITY */

	/* struct rw_semaphore has atomic_long_t member count */
	/* #undef HAVE_RWSEM_ATOMIC_LONG_COUNT */

	/* linux/sched/signal.h exists */
	/* #undef HAVE_SCHED_SIGNAL_HEADER */

	/* Define to 1 if you have the <security/pam_modules.h> header file. */
	#define HAVE_SECURITY_PAM_MODULES_H 1

	/* setattr_prepare() is available, doesn't accept user_namespace */
	/* #undef HAVE_SETATTR_PREPARE_NO_USERNS */

	/* setattr_prepare() accepts user_namespace */
	/* #undef HAVE_SETATTR_PREPARE_USERNS */

	/* iops->set_acl() exists, takes 3 args */
	/* #undef HAVE_SET_ACL */

	/* iops->set_acl() takes 4 args */
	/* #undef HAVE_SET_ACL_USERNS */

	/* set_cached_acl() is usable */
	/* #undef HAVE_SET_CACHED_ACL_USABLE */

	/* set_special_state() exists */
	/* #undef HAVE_SET_SPECIAL_STATE */

	/* struct shrink_control exists */
	/* #undef HAVE_SHRINK_CONTROL_STRUCT */

	/* kernel_siginfo_t exists */
	/* #undef HAVE_SIGINFO */

	/* signal_stop() exists */
	/* #undef HAVE_SIGNAL_STOP */

	/* new shrinker callback wants 2 args */
	/* #undef HAVE_SINGLE_SHRINKER_CALLBACK */

	/* ->count_objects exists */
	/* #undef HAVE_SPLIT_SHRINKER_CALLBACK */

	#if defined(__amd64__) \|\| defined(__i386__)
	/* Define if host toolchain supports SSE */
	#define HAVE_SSE 1

	/* Define if host toolchain supports SSE2 */
	#define HAVE_SSE2 1

	/* Define if host toolchain supports SSE3 */
	#define HAVE_SSE3 1

	/* Define if host toolchain supports SSE4.1 */
	#define HAVE_SSE4_1 1

	/* Define if host toolchain supports SSE4.2 */
	#define HAVE_SSE4_2 1

	/* Define if host toolchain supports SSSE3 */
	#define HAVE_SSSE3 1
	#endif

	/* STACK_FRAME_NON_STANDARD is defined */
	/* #undef HAVE_STACK_FRAME_NON_STANDARD */

	/* standalone <linux/stdarg.h> exists */
	/* #undef HAVE_STANDALONE_LINUX_STDARG */

	/* Define to 1 if you have the <stdint.h> header file. */
	#define HAVE_STDINT_H 1

	/* Define to 1 if you have the <stdlib.h> header file. */
	#define HAVE_STDLIB_H 1

	/* Define to 1 if you have the <strings.h> header file. */
	#define HAVE_STRINGS_H 1

	/* Define to 1 if you have the <string.h> header file. */
	#define HAVE_STRING_H 1

	/* Define to 1 if you have the `strlcat' function. */
	#define HAVE_STRLCAT 1

	/* Define to 1 if you have the `strlcpy' function. */
	#define HAVE_STRLCPY 1

	/* submit_bio is member of struct block_device_operations */
	/* #undef HAVE_SUBMIT_BIO_IN_BLOCK_DEVICE_OPERATIONS */

	/* super_setup_bdi_name() exits */
	/* #undef HAVE_SUPER_SETUP_BDI_NAME */

	/* super_block->s_user_ns exists */
	/* #undef HAVE_SUPER_USER_NS */

	/* Define to 1 if you have the <sys/stat.h> header file. */
	#define HAVE_SYS_STAT_H 1

	/* Define to 1 if you have the <sys/types.h> header file. */
	#define HAVE_SYS_TYPES_H 1

	/* i_op->tmpfile() exists */
	/* #undef HAVE_TMPFILE */

	/* i_op->tmpfile() has userns */
	/* #undef HAVE_TMPFILE_USERNS */

	/* totalhigh_pages() exists */
	/* #undef HAVE_TOTALHIGH_PAGES */

	/* kernel has totalram_pages() */
	/* #undef HAVE_TOTALRAM_PAGES_FUNC */

	/* Define to 1 if you have the `udev_device_get_is_initialized' function. */
	/* #undef HAVE_UDEV_DEVICE_GET_IS_INITIALIZED */

	/* kernel has __kernel_fpu_* functions */
	/* #undef HAVE_UNDERSCORE_KERNEL_FPU */

	/* Define to 1 if you have the <unistd.h> header file. */
	#define HAVE_UNISTD_H 1

	/* iops->getattr() takes struct user_namespace* */
	/* #undef HAVE_USERNS_IOPS_GETATTR */

	/* iops->getattr() takes a vfsmount */
	/* #undef HAVE_VFSMOUNT_IOPS_GETATTR */

	/* aops->direct_IO() uses iovec */
	/* #undef HAVE_VFS_DIRECT_IO_IOVEC */

	/* aops->direct_IO() uses iov_iter without rw */
	/* #undef HAVE_VFS_DIRECT_IO_ITER */

	/* aops->direct_IO() uses iov_iter with offset */
	/* #undef HAVE_VFS_DIRECT_IO_ITER_OFFSET */

	/* aops->direct_IO() uses iov_iter with rw and offset */
	/* #undef HAVE_VFS_DIRECT_IO_ITER_RW_OFFSET */

	/* All required iov_iter interfaces are available */
	/* #undef HAVE_VFS_IOV_ITER */

	/* fops->iterate() is available */
	/* #undef HAVE_VFS_ITERATE */

	/* fops->iterate_shared() is available */
	/* #undef HAVE_VFS_ITERATE_SHARED */

	/* fops->readdir() is available */
	/* #undef HAVE_VFS_READDIR */

	/* fops->read/write_iter() are available */
	/* #undef HAVE_VFS_RW_ITERATE */

	/* __set_page_dirty_nobuffers exists */
	/* #undef HAVE_VFS_SET_PAGE_DIRTY_NOBUFFERS */

	/* __vmalloc page flags exists */
	/* #undef HAVE_VMALLOC_PAGE_KERNEL */

	/* yes */
	/* #undef HAVE_WAIT_ON_BIT_ACTION */

	/* wait_queue_entry_t exists */
	/* #undef HAVE_WAIT_QUEUE_ENTRY_T */

	/* wq_head->head and wq_entry->entry exist */
	/* #undef HAVE_WAIT_QUEUE_HEAD_ENTRY */

	/* xattr_handler->get() wants dentry */
	/* #undef HAVE_XATTR_GET_DENTRY */

	/* xattr_handler->get() wants both dentry and inode */
	/* #undef HAVE_XATTR_GET_DENTRY_INODE */

	/* xattr_handler->get() wants xattr_handler */
	/* #undef HAVE_XATTR_GET_HANDLER */

	/* xattr_handler has name */
	/* #undef HAVE_XATTR_HANDLER_NAME */

	/* xattr_handler->list() wants dentry */
	/* #undef HAVE_XATTR_LIST_DENTRY */

	/* xattr_handler->list() wants xattr_handler */
	/* #undef HAVE_XATTR_LIST_HANDLER */

	/* xattr_handler->list() wants simple */
	/* #undef HAVE_XATTR_LIST_SIMPLE */

	/* xattr_handler->set() wants dentry */
	/* #undef HAVE_XATTR_SET_DENTRY */

	/* xattr_handler->set() wants both dentry and inode */
	/* #undef HAVE_XATTR_SET_DENTRY_INODE */

	/* xattr_handler->set() wants xattr_handler */
	/* #undef HAVE_XATTR_SET_HANDLER */

	/* xattr_handler->set() takes user_namespace */
	/* #undef HAVE_XATTR_SET_USERNS */

	/* Define if you have [z] */
	#define HAVE_ZLIB 1

	/* __posix_acl_chmod() exists */
	/* #undef HAVE___POSIX_ACL_CHMOD */

	/* kernel exports FPU functions */
	/* #undef KERNEL_EXPORTS_X86_FPU */

	/* TBD: fetch(3) support */
	#if 0
	/* whether the chosen libfetch is to be loaded at run-time */
	#define LIBFETCH_DYNAMIC 1

	/* libfetch is fetch(3) */
	#define LIBFETCH_IS_FETCH 1

	/* libfetch is libcurl */
	#define LIBFETCH_IS_LIBCURL 0

	/* soname of chosen libfetch */
	#define LIBFETCH_SONAME "libfetch.so.6"
	#endif

	/* Define to the sub-directory where libtool stores uninstalled libraries. */
	#define LT_OBJDIR ".libs/"

	/* make_request_fn() return type */
	/* #undef MAKE_REQUEST_FN_RET */

	/* hardened module_param_call */
	/* #undef MODULE_PARAM_CALL_CONST */

	/* struct shrink_control has nid */
	/* #undef SHRINK_CONTROL_HAS_NID */

	/* Defined for legacy compatibility. */
	#define SPL_META_ALIAS ZFS_META_ALIAS

	/* Defined for legacy compatibility. */
	#define SPL_META_RELEASE ZFS_META_RELEASE

	/* Defined for legacy compatibility. */
	#define SPL_META_VERSION ZFS_META_VERSION

	/* True if ZFS is to be compiled for a FreeBSD system */
	#define SYSTEM_FREEBSD 1

	/* True if ZFS is to be compiled for a Linux system */
	/* #undef SYSTEM_LINUX */

	/* zfs debugging enabled */
	/* #undef ZFS_DEBUG */

	/* /dev/zfs minor */
	/* #undef ZFS_DEVICE_MINOR */

	/* enum node_stat_item contains NR_FILE_PAGES */
	/* #undef ZFS_ENUM_NODE_STAT_ITEM_NR_FILE_PAGES */

	/* enum node_stat_item contains NR_INACTIVE_ANON */
	/* #undef ZFS_ENUM_NODE_STAT_ITEM_NR_INACTIVE_ANON */

	/* enum node_stat_item contains NR_INACTIVE_FILE */
	/* #undef ZFS_ENUM_NODE_STAT_ITEM_NR_INACTIVE_FILE */

	/* enum zone_stat_item contains NR_FILE_PAGES */
	/* #undef ZFS_ENUM_ZONE_STAT_ITEM_NR_FILE_PAGES */

	/* enum zone_stat_item contains NR_INACTIVE_ANON */
	/* #undef ZFS_ENUM_ZONE_STAT_ITEM_NR_INACTIVE_ANON */

	/* enum zone_stat_item contains NR_INACTIVE_FILE */
	/* #undef ZFS_ENUM_ZONE_STAT_ITEM_NR_INACTIVE_FILE */

	/* global_node_page_state() exists */
	/* #undef ZFS_GLOBAL_NODE_PAGE_STATE */

	/* global_zone_page_state() exists */
	/* #undef ZFS_GLOBAL_ZONE_PAGE_STATE */

	/* Define to 1 if GPL-only symbols can be used */
	/* #undef ZFS_IS_GPL_COMPATIBLE */

	/* Define the project alias string. */
	-#define ZFS_META_ALIAS "zfs-2.1.99-FreeBSD_g269b5dadc"
	+#define ZFS_META_ALIAS "zfs-2.1.99-FreeBSD_gf291fa658"

	/* Define the project author. */
	#define ZFS_META_AUTHOR "OpenZFS"

	/* Define the project release date. */
	/* #undef ZFS_META_DATA */

	/* Define the maximum compatible kernel version. */
	-#define ZFS_META_KVER_MAX "5.14"
	+#define ZFS_META_KVER_MAX "5.15"

	/* Define the minimum compatible kernel version. */
	#define ZFS_META_KVER_MIN "3.10"

	/* Define the project license. */
	#define ZFS_META_LICENSE "CDDL"

	/* Define the libtool library 'age' version information. */
	/* #undef ZFS_META_LT_AGE */

	/* Define the libtool library 'current' version information. */
	/* #undef ZFS_META_LT_CURRENT */

	/* Define the libtool library 'revision' version information. */
	/* #undef ZFS_META_LT_REVISION */

	/* Define the project name. */
	#define ZFS_META_NAME "zfs"

	/* Define the project release. */
	-#define ZFS_META_RELEASE "FreeBSD_g269b5dadc"
	+#define ZFS_META_RELEASE "FreeBSD_gf291fa658"

	/* Define the project version. */
	#define ZFS_META_VERSION "2.1.99"

	/* count is located in percpu_ref.data */
	/* #undef ZFS_PERCPU_REF_COUNT_IN_DATA */

	diff --git a/sys/modules/zfs/zfs_gitrev.h b/sys/modules/zfs/zfs_gitrev.h
	index d8d9df0110ad..e3c1dae475d4 100644
	--- a/sys/modules/zfs/zfs_gitrev.h
	+++ b/sys/modules/zfs/zfs_gitrev.h
	@@ -1,5 +1,5 @@
	/*
	* $FreeBSD$
	*/

	-#define ZFS_META_GITREV "zfs-2.1.99-530-g269b5dadc"
	+#define ZFS_META_GITREV "zfs-2.1.99-577-gf291fa658"

File Metadata

Mime Type: application/octet-stream
Expires: Wed, Apr 9, 8:22 AM (1 d, 23 h)
Storage Engine: chunks
Storage Format: Chunks
Storage Handle: tbAWZYMgqCov
Default Alt Text: (7 MB)

No OneTemporaryActions

View Options

File Metadata

Event Timeline

No OneTemporary
Actions